From reeds@gauss.att.com Wed Jan 01 01:17:45 1992 Message-Id: <9112311625.AA00207@rand.org> From: reeds@gauss.att.com Date: Tue, 31 Dec 91 11:17:45 EST To: voynich@rand.org Status: OR It would be interesting to find the most pronouncible substitution from Currier to phonetic values, using, as Kornai suggests, information about stops, glides, etc. I don't think we should transcribe in such a ``phonetic alphabet'' but it might well be worth our while to learn to pronounce Currier letters phonetically. I am sure that the VMS was uttered as it was written, whether or not it is pure gibberish, gibberish with cipher text concealed in it, or (but this is hard to believe) plain language. Jim Reeds From jbaez@math.mit.edu Wed Jan 01 02:54:18 1992 Date: Tue, 31 Dec 91 12:54:18 EST From: jbaez@math.mit.edu Message-Id: <9112311754.AA10663@nevanlinna> To: voynich@rand.org Subject: Dee, gibberish Status: OR Yesterday my friend Peter Durigan and I went to the rare book room at Wellesley College (where I teach) to look at an old book by John Dee. Peter is a patent lawyer with a very serious interest in Neoplatonism. He once met an occult book dealer who said he had a copy of Dee's translation of Euclid's "Elements," which he would sell for $15,000. The book dealer also said he would lend any book for inspection for 10 days, and Peter proposed to me that we borrow Dee's "Elements," but I said it was a bad idea, since 1) the book dealer might not really have meant that he'd lend *that* book, and 2) what if we lost it? So I looked to see what they had of Dee's at Wellesley, and it turns out they have 1) the "Elementa" (as it's really called), 2) "The Private Diary...," and 3) a catalog of Dee's library (which text was listed as "under repair". Clearly Wellesley was the friend of a Dee fanatic at one time... perhaps someday I'll find out who. Anyway, we arranged to see the "Elementa." In part I was interested in the text itself (since Peter and I have had many discussions of mysticism, Pythagoreanism, and Platonism as they relate to mathematics), and in part I wanted to learn the etiquette of rare book rooms on friendly ground before tackling the Beinecke. The rare book room was on the fourth floor of the library. We had to ring a doorbell, and were buzzed in by the young woman I had spoken to on the phone earlier. We had to fill out forms indicating our names, addresses, telephone numbers and affiliation with Wellesley (if any), and I showed her my Wellesley I.D. and filled out a separate form saying which book we were looking at. She took the book over to a reading desk. It was a small leather-bound volume, published in the late 1500's by someone who wrote in his prologue that Dee had died before being able to publish this volume. After this prologue there was a long preface by Dee, and then Dee's translations of 6 books of the Elements into English. Dee's preface was fascinating. There were illustrations on the first page showing a naked woman (in a style utterly unlike those in the Voynich), and the "D" of the first sentence, which began "Divine Plato..." was illuminated, showing a figure of a lion rampant (with an erection, according to Peter), a symbol which I interpreted as being the usual symbol for Mercury (though it differed slightly from the standard version), and lots of abstract curlicues. There was also a picture of two men, one firing an arrow... It began with a description of how many visitors came to Plato's academy and, finding that nothing "practical" was taught therein, left in disgust, until Aristotle pointed this out and said that perhaps there should be some sort of notice warning them what it was all about. (Dee didn't mention so explicitly, but this seems to be a reference to the famous sign "Let none who have not studied geometry enter here...") Likewise, writes Dee (in his prolix style) , this preface will warn the reader about the contents of the "Elements." He then proceeds to a philosophical discussion of Mathematicks, whose two primary topics are Number and Magnitude, Number being made of Units, simple and indivisible, while Magnitude is arbitarily divisible, and has length, width and breadth, ... etc. The key reason for studying Mathematicks is that numbers are beings of a third kind occupying a realm bridging the Supernatural and the Natural, so that their study is uplifting. To make this claim more concrete and exciting he mentions one "Joachim," later referred to as "Joannes Picus", who wrote a book of true prophesies, which according to the "Duke of Mirandula" were obtained by the study of "Formal Mathematicks." (I would be very interested to learn a bit more about this prophet!) He then discusses how, while the purest study is that of the integers, various practical "reckonmasters" have developed fractions, radicalls and so on. He gives one example that looks like the square root of pi, but the "pi" is written in an odd manner - if that's what it is. While not wishing to impute any such base concerns to his readers, he promises no less than FIVE proofs that the contents of the Elementa are quite practical. Here we had to leave off. There were many odd things about this text which I will try to study more carefully when Peter and I return to it. Occaisional references are made to the "Lion" or the "Lion's Claw", quite terse and unexplained, which I imagine have something to do with the picture of the lion at the beginning. There is a reference to the "Allmighty Ternary," which is perhaps the Trinity, but at the bottom of the same page -- curiously indicated by a small picture of a hand pointing its finger! -- the Ternary is defined as the "Unity, knot and Uniformity," an explanation unlike those I am familiar with. Written quite faintly in reddish ink, in a beautifully careful hand, in the margin of one page, is the exhortation "Seek ye the kingdom of God and all will be granted thereto". All in all, quite unlike the geometry text I had as a student. :-) I apologize for the fact that this has rather little to do with the Voynich, save that it's an example of the mentality we may be dealing with. From what I read so far it seemed quite possible that the preface had an esoteric subtext which only careful reading would disclose. ----- Peter Davidson raises the question dear to my heart, namely, whether statistical analysis of the Voynich could ever determine it to be gibberish. I feel that there are long-range (i.e., sentence- or paragraph-length) patterns of word occurence in meaningful text that are unlikely to appear in gibberish. (Of course, if one finds such patterns one can write a program to write gibberish having these pattern - but it's unlikely that X (the Voynich author(s)) would have been so sophisticated.) I would like to find out: "Given that a certain word appears at place n in the Voynich, what is the probability that it appears again at place n + m" -- and the same thing for other texts, such as the Book of Enoch, medieval herbals, and so on. Perhaps this or a more sophisticated statistic can rule out some hypotheses concerning the Voynich. Certainly the statistical difference between hands A and B and Jim Reeds' claimed oddities about the first word in each line could, if studied carefully with statistics, make it clear that this is not a word-for-word mapping of any "ordinary" text. jb From jim@rand.org Wed Jan 01 04:18:12 1992 Message-Id: <9112311918.AA00297@mycroft.rand.org> To: voynich@rand.org Subject: Pronounceable Voynich text In-Reply-To: Your message of Tue, 31 Dec 91 11:17:45 -0500. <9112311625.AA00207@rand.org> From: Jim Gillogly Reply-To: jim@rand.org Date: Tue, 31 Dec 91 11:18:12 PST Sender: jim%mycroft@rand.org Status: OR > reeds@gauss.att.com writes: > It would be interesting to find the most pronouncible substitution > from Currier to phonetic values, using, as Kornai suggests, information > about stops, glides, etc. I don't think we should transcribe in such a > ``phonetic alphabet'' but it might well be worth our while to learn to > pronounce Currier letters phonetically. On Currier's original transcription page there's a string of equivalents running down the left. I applied them to a piece of f1r and got the following: 00101A VAS92.9FAE.AR.APAM.ZOE.ZOR9.QOR92.9.FOR.ZOE89- tehweths.ethdehn.ehk.ehlehm.wahn.wahketh.Qahkeths.eth.dahk.wahnteth- 00102A 2OR9.XAR.O.R.9.FAN.ZPAM.ZAR.AR*.QAR.QAR.8AD- sahketh.Xehk.ah.k.eth.dehn.wlehm.wehk.ehk*.Qehk.Qehk.tehD- 00103A 29AU.ZCF9.OR.9FAM.ZO8.QOAR9.Q*R.8ARAM.29- sethehU.wideth.ahk.ethdehm.waht.Qahehketh.Q*k.tehkehm.seth- 00104A $OM.OPCC9.OPCOR.2OEOP9.Q*AR.8AM.OFAM.OE.OFAD- $ahm.ahliieth.ahliahk.sahnahleth.Q*ehk.tehm.ahdehm.ahn.ahdehD- 00105A 2AT.9.SCAR.QAM.WAR.YAM# sehT.eth.wiehk.Qehm.Wehk.Yehm# 00106A 98ARAIZO# ethtehkehIwah# Sort of pronounceable. Jim Gillogly From RJB@MAX.U.WASHINGTON.EDU Wed Jan 01 04:34:00 1992 Date: Tue, 31 Dec 91 11:34 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Dee's Preface to Euclid To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Dee's Preface has a modern edition, and either D P Walker or Frances Yates has written quite a bit about it. "Joachim" is not "Joannes Picus," unless I'm badly off base. "Joachim" is Joachim of Flora, who was a prophetic Franciscan abbott who divided history (and religious dispensations) into the era of the Father (Jewish), the Son (Christian), and the Holy Spirit. Oddly enough Joachim didn't get into trouble, but he became a major figure in chiliastic and millenarian speculations of all sorts -- or rather his tripartite historical scheme did. Dee's own prophetic mission with its "leading, had it succeeded, to a total reformation of all kingdoms etc" or whatever Casaubon says, is clearly indebted to J of F. "Joannes Picus" is J Pico della Mirandola, for whom again see DP Walker (_Spiritual and Demonic Magic from Ficino to Pico della Mirandola_) and Frances Yates. The sign that's "almost the sign of alchemical mercury" is probably Dee's "Monad." The canonical text is Dee's "Hieroglyphic Monad," of which two translations are available -- I think the academic one was done by C H Justen, and published originally in _Ambix_. That sign also shows up in the Chemical Wedding of Christian Rosycross, leading to a burst of Twilight Zone music and many speculative volumes about who did what with whom, when. (Typo alert: that's "C H Josten.") At any rate, Dee's interests were not unique, and he's certainly not the only candidate for the creator of the VMs -- not to be confused with VMS, which is in its own way quite hermetic... --rjb From jbaez@math.mit.edu Wed Jan 01 05:24:25 1992 Date: Tue, 31 Dec 91 15:24:25 EST From: jbaez@math.mit.edu Message-Id: <9112312024.AA10828@nevanlinna> To: voynich@rand.org Subject: Dee's Preface to Euclid Status: OR Thanks for the info, Jim Reeds and RJB! Identifying "Joachim" with Joachim of Flora and "the Duke of Mirandula" and "Joannes Picus" with Giovanni Pico de Mirandola seem sensible to me... I vaguely know these characters but didn't quite recognize them in these disguises. Dee's style is so florid and terse (if one can picture this) that it's hard to tell whether he's identifying these fellows or not. I will track down the references. While Dee isn't the only candidate for X, he's at least an interesting fellow. jb From wet!naga@cca.ucsf.EDU Thu Jan 02 14:01:00 1992 Message-Id: Date: Wed, 1 Jan 92 21:01 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Calendrical exactitude Status: OR Thanks to Jacques Guy for the additional biographical material regarding Dee and Kelley. Scholarly exactness requires us to attend to the matter of whether dates as given are Gregorian or Julian, since the Gregorian calendar was instituted in most Catholic countries in 1582 or shortly thereafter (though not adopted until 1752 in Britain). To be exact, according to the Gregorian reform there were ten days omitted from the calendar in 1582, with the day following 4 October 1582 thenceforth being known as 15 October 1582. >original manuscripts there is a section, heavily erased and barely legible, >recording the seance of 23 May 1587 -- the morning after the wife- >swapping. The spirits ask Kelley:'Was thy brother's wife [Jane Dee] ... For those of us of a Bohemian inclination who might be tempted to celebrate annually the anniversary of this event by conducting a similar practice the question arises as to whether this was 23 May 1587 Gregorian or Julian. From jim@rand.org Fri Jan 03 01:01:59 1992 Message-Id: <9201021602.AA01441@mycroft.rand.org> To: voynich@rand.org Subject: Re: Pronounceable Voynich text In-Reply-To: Your message of Tue, 31 Dec 91 19:18:20 -0500. <9201010018.AA02404@mthvax.cs.miami.edu> From: Jim Gillogly Reply-To: jim@rand.org Date: Thu, 02 Jan 92 08:01:59 PST Sender: jim%mycroft@rand.org Status: OR > Yanek Martinson writes (privately): > > tehweths.ethdehn.ehk.ehlehm.wahn.wahketh.Qahkeths.eth.dahk.wahnteth- > Much too much of 'h' . Was there any particular reason to use 'h' there? I put the "ah" and "eh" in there because it was on Currier's translation chart -- I assume he was assiging his own pronouceable values, rather than either (a) hypothesizing the "real" values, or (b) intending to transcribe the stuff using those values. I can ask him... Anyway, here's f3v (Reeds transcription) without the extra h's. Since we've got a GIF of that, you can follow along... {f3v transcribed by Reeds 13 Dec from f3v.gif } <3v.1> FOAM.W[OA]R.4OPO[A9].ZA.XOE.9FOAM,[82],OE[A9] daem.W[ae]k.fala[eeth].we.Xan.ethdaem,[ts],an[eeth] <3v.2> 8AII8[9A].4OPCCOE.OFCCOR.OFOR.OE9POE.8OE,8AR teIIt[ethe].faliian.adiiak.adak.anethlan.tan,tek <3v.3> OFAJ.SOE.ZOE.2CCC2.SOJ.SCC9FAJ.OF3 {last few chars unclear} adeJ.wan.wan.siiis.waJ.wiiethdeJ.ad3 {last few chars unclear} <3v.4> 408AR.S2.CC9.FSCOE.OFAE.8OR.SCARCCD f0tek.ws.iieth.dwian.aden.tak.wiekiiD <3v.5> 9CCCAR.O4SAE.SAR.SAR.XA {first word start unclear} ethiiiek.afwen.wek.wek.Xe {first word start unclear} <3v.6> O2.**CAR.FA2.SO8AE9.SO8AE9.SAJ # as.**iek.des.wateneth.wateneth.weJ # {para} <3v.7> PSOR.OPSAJ.SOR,WAJ.S lwak.alweJ.wak,WeJ.w <3v.8> 9FS9.FSOJ.SOR.CCCFCOE.OFA ethdweth.dwaJ.wak.iiidian.ade <3v.9> 9PSCCM.OFCOE.QO8OAE9.SOR.Q9 ethlwiim.adian.Qataeneth.wak.Qeth <3v.10> OSAR.8AM.4OFZOE.8A1.SOE.OFAR9 awek.tem.fadwan.te1.wan.adeketh <3v.11> ZO.ZOXO.X9.PSOR.SO8AM.SOF wa.waXa.Xeth.lwak.watem.wad <3v.12> OZ.SO8AT.9PS9.PSOR.FSAJ aw.wateT.ethlweth.lwak.dweJ <3v.13> ZAR.ZFAM.40FSA.9P9.QAE.SF9 wek.wdem.f0dwe.ethleth.Qen.wdeth <3v.14> 8AN.ZCAJ.9FCAJ ten.wieJ.ethdieJ -------- For completeness, here's the Perl program that I used. Jacques' Pascal translit program would work fine on it, but I think not everybody has a PC. Jim Gillogly ----- # translit: transliterate some Voynich text using any of several # hypothesized systems # # Default: from Currier's "Transcription Alphabet" page # with "ah" changed to "a" and "eh" changed to "e". while (<>) { print; s/^(\S*)//; $lnum = $1; s/4/f/g; s/O/a/g; s/8/t/g; s/9/eth/g; s/2/s/g; s/E/n/g; s/R/k/g; s/S/w/g; s/Z/w/g; s/P/l/g; s/B/th/g; s/F/d/g; s/V/t/g; s/A/e/g; s/C/i/g; s/N/n/g; s/M/m/g; print; } From j.guy@trl.OZ.AU Fri Jan 03 01:36:58 1992 Date: Thu, 2 Jan 92 11:36:58 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201020036.AA02123@medici.trl.OZ.AU> To: voynich@rand.org Subject: Dee's chronology Status: OR wet!naga@cca.ucsf.EDU asks: >> ... pers. com. from Donald Laycock: there is, >>in Dee's diaries, a note to this effect: "How strange, I cannot >>find my Book of Soyga". According to Don, Dee's "Book of Soyga" may >>have been the Voynich manuscript, and Kelley would have stolen it and >>sold it to Rudolph II, emperor of Bohemia. >Is there a date available for this diary entry? Yes, there is/was, but I don't know it. It must be in Don's notes, wherever they are, if Tania (his widow) has kept them. >I'd like to get clearer on some chronology. Jacques gives Kelley's >dates as 1555-1595 Not me, not me, Don did. Now, let me look this up.... "According to the horoscope he later cast himself, John Dee was born at Mortlake on the morning of 13 July, 1527, under the sign of Cancer, with Sagittarius in ascendent.... He graduated Master of Arts from Cambridge in 1548, and went abroad to study further at Louvain. Later he gave lectures at the University of Paris... It is not known just when he began to seek or achieve efective communications with the beings he regarded as angles, but it was probably not before 1581. From about March of that year, as we learn from his Private Diary, he was troubled by odd dreams, and strange knockings in the night... by October of that year Dee had found a medium, one Barnabas Saul, who, using as a crystal a stone given Dee by a 'friend', reported to Dee the words of the angel Annael and Michael -- and who later denied seeing any spirits at all, for fear of prosecution for conjuring. But Dee was dissatisfied with Saul as a medium, and dismissed him... .... there appeared at his house in Mortlake on 8 March 1582 a man calling himself Edward Talbot, afterwards to be known by his true name of Edward Kelley. Kelley was about twenty-seven years old at the time.... He was not uneducated; he had been an undergraduate at Oxford, but was apparently dismissed. He spoke Latin well enough to travel in Europe, and converse with Dee's Continental friends; he makes frequent grammatical mistakes, but of his general intelligence there is no doubt.... on 10 March that year Dee was giving the new medium a trial -- with results so successful that there began on that day a strange and close association between the two men that was to last for some seven years... on 26 March 1582, Edward Kelley produced the first material concerning the 'angelicall language.' [Enochian].... The wife-swapping episode took place, in spite if the attempts of some biographers of Dee to suggest that it did not. In the original manuscripts there is a section, heavily erased and barely legible, recording the seance of 23 May 1587 -- the morning after the wife- swapping. The spirits ask Kelley:'Was thy brother's wife [Jane Dee] obedient and humble unto thee?' -- and Kelley replies:'She was.' Dee returns the same answer concerning Joanna Kelley. Dee's association with Kelley, and apparently also with the Enochian angles, ended shortly after this episode -- perhaps the tensions in that household of two British couples in Bohemia were becoming too much to bear, and Kelley set himself up as an alchemist in a separate establishment. Nevertheless, the Dees did not leave for England until March 1589, arriving there in December of that year. Kelley stayed on at the court of Emperor Rudolph in Prague, and died there in 1595 -- under obscure conditions. (The usual story is that he was imprisoned for failing to produce alchemical gold, and fell from a tower when trying to escape.)" The other piece of chronology I could find is in Stephen Skinner's preface: "For Dee, the Angels also provided advice and reproof which he conveyed to two of the most powerful patrons he sought, Stephen Bathori King of Poland, and Rudolf II of Bohemia. He was lucky not to have incurred any more painful a punishment for his pains than a partial banishment by the latter. Dee and Kelley accordingly sought refuge in 1586 at the castle of Count Rosenberg at Trebona, where Dee stayed for two years. Meanwhile Kelley managed, through his alchemical experiments, to curry favour first with the Count and then with Rudolf, who later knighted him..... In February 1589 Dee saw the last of Kelley, and from thenceforth Dee's skrying experiments were with very inferior skryers such as Bartholomew Hickman, whose skrying records were burnt when Dee discovered that he was a fraud. Till Kelley's death, Dee continued to hope for a reconciliation with his old skryer." From jbaez@math.mit.edu Fri Jan 03 03:52:20 1992 Date: Thu, 2 Jan 92 13:52:20 EST From: jbaez@math.mit.edu Message-Id: <9201021852.AA09096@riesz> To: voynich@rand.org Status: OR Today I visited the library at Wellesley, picked up Frances Yates' "Giordano Bruno and the Hermetic Tradition" (University of Chicago Press, 1964, 466 pp.), and took a look at "The Private Diary of Dr. John Dee and The Catalogue of his Library of Manuscripts, from the Original Manuscripts in the Ashmolean Museum at Oxford, and Trinity College Library, Cambridge," edited by James Orchard Halliewell. The latter, unfortunately, cannot be checked out, but I copied down all the diary entries from 1586 to 1589 that relate to Prague (where Emperor Rudolph II held court). Some are quite tantalizing, especially in relation to the "600 ducats" supposedly paid for the Voynich ms. 1586. Oct. 18th, E K [Edward Kelly, the brother of Thomas] recessit a Treoona versus Pragam cumu delatus; masit hic per tres hebdomadas. Nov. 8 illustrissimus princeps [probably Prince Albert Leski] versus Pragam; iter institit hor tertia a meridie. Dec. 30 E K versus Pragam. 1587. Jan. 14th Reinholdt [Dr. Reinholdt of Salfeldt] revisit versus Pragam 20 die. Jan. 18th rediit E K a Praga. E K brought with him from the Lord Rosenberg to my wyfe a chayne and juelle estemed at 300 duckettes; 200 the juell stones, and 100 the gold. Jan. 21st E K again to Prage and so to Poland ward. March 7th E K dedit nobis 300 ducata. March 21st E K gave me 170 more, and of the 200 for changing 60 remain. Sept. 26th my lord [Lord Biberstein] went toward Prage Sept. 30th, T K [Thomas Kelley] and J C [John Carp] went toward Prage. Oct. 12th Mr. E K toward Prage on horsbak. Oct. 26th Mr. Edward Kelly cam to Trebona from Prage. 1588. Oct. 25th, Mr. Ed. Kelley and John Carpio rode toward Prage. Nov. 6 Mr Kelly cam home from Prage. Dec. 4th I gave to Mr. Ed. Kelley my Glass, so highly and long estemed of our Quene, and the Emperor Randolph the second, de quo in praefatione Euclidis fit mentio. [The editor says that this refers to Dee's preface of Billinglsley's (and perhaps Dee's) translation of Euclid's Elements.] Dec 23. Radulphus Sagiensis Gallus Nomannus, venit Trebonam, chimiae et naturalis magiae studiosis. 1589. Jan. 3rd Rudolphus Sagiensis Normannus recessit versus Pragam. Apr. 19th I delivered by letters to Mr Thomas Kelley for his brother Sir Edward Kelly, knight, at Emperor's court at Prage. ------------ I hope I haven't introduced my own mispellings into Dee's brew. As I've never studied Latin I can only guess what is written in that tongue, but it seems fairly clear. Presumably all these "Randolphs," "Rudolphi" and "Radulphi" are one and the same Holy Roman Emperor. Clearly there was a lot of traffic to and from Prague. Clearly some ducats changed hands, although none definitely from Rudolph to Dee. (What would "of the 200 for changing" refer to -- moneychanging? That'd be a walloping fee.) And the preface Euclid's Elements isn't quite as irrelevant as might at first seem the case, as it's the only book we seem to KNOW that Dee gave to Rudolph. Yates's book has two pictures of Dee's "Monas Hieroglyphica," the symbol which appears at the beginning of the preface. One is from Dee's book of that title, and the other is from a book by -- no other than Athanasius Kircher, the owner of the Voynich manuscript! Spooky... From jbaez@math.mit.edu Fri Jan 03 03:58:25 1992 Date: Thu, 2 Jan 92 13:58:25 EST From: jbaez@math.mit.edu Message-Id: <9201021858.AA09101@riesz> To: voynich@rand.org Subject: Waiting for Godot Status: OR Well, I just called the Beinecke Rare Book Library and they said that Robert Babcock, the man in whose palm our destiny lies, will not be back until Monday. I will post a bit of information I have gleaned from Yates concerning Dee, Kircher, etc., in a while. Thanks again to all of you who've helped with my queries about the preface to the Elements. It's nice to find out information from Yates, but I assure you that the thrill of digging it out yourself by reading a tome published in the 1500's is worth the effort. So I encourage all of you who live within range of rare book libraries to do some poking around. jb From jbaez@math.mit.edu Fri Jan 03 04:35:38 1992 Date: Thu, 2 Jan 92 14:35:38 EST From: jbaez@math.mit.edu Message-Id: <9201021935.AA09127@riesz> To: voynich@rand.org Subject: Dunstan Status: OR I never quite caught on to what connection the "book of Dunstan" might have to the Voynich, but in his Diaries Dee notes: Dec. 12th, afternone somewhat, Mr. Ed. Keley his lamp overthrow, the spirit of wyne long spent to nere, and the glas being not stayed with buks about it, as it was wont to be, the spirit was spilled out... and, to make a long story short, a bunch of books caught on fire, including the "Book of Dunstan." (Glancing through his diary it's remarkable how many bizarre accidents happened to Dee... perhaps a result of dealings with evil spirits. Does the above mean that Kelley was drunk??) From firth@SEI.CMU.EDU Fri Jan 03 04:58:35 1992 Date: Thu, 2 Jan 92 14:58:35 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201021958.AA05634@bp.sei.cmu.edu> To: voynich@rand.org Subject: Note 3 Status: OR Notes on the Voynich Manuscript - Part 3 ---------------------------------------- Well, I read the D'Imperio monograph a couple of times over the holidays, and it has prompted some more thoughts. This note is about the non-Voynich text in the MS, as inferred from the transcriptions in her document. Fig 10 (all figs from D'Imperio) shows eight Zodiacal symbols, pretty clearly pisces and gemini on the left, aries, leo and sagittarius down the middle, and taurus, libra and virgo on the right. Scribbled near each figure is a word of text. These words aren't easy to read, but to me they seem (in the same order as above): MARS, JONY, AVRIL, AUGST, , MAY, OCTOBRS, . These are the month names traditionally associated with these signs by Western astrology; it is of course some 2200 years since the real sun really occupied them. The two illegible names should then denote November and September. The script looks more german than italic, and so does the language, though that's hard to tell. Fig 22: the folio gatherings. It seems the Voynich MS has been numbered twice. The plates in Brumbaugh show that every folio has been numbered at the top right, in modern Roman numerals in a clear, bold italic hand that looks to me 18c or early 19c. In addition, each "signature" of eight leaves has been numbered as shown in this figure; each number combines arabic numerals and mediaeval latin script; the list reads (I use two columns) Pmus 11mus 2us 3us 13us 4us 14us 5tus 15us 6tus 7mus 17us 8uus 9nus 19 10mus 20 That's all good latin. One puzzle, though, is the antique shape of the ciphers. They don't look like late 16c to me; much more like 1400 than 1600. So, is the MS older than we think? Or were the numbers added by somebody at Rudolph's court who'd learned from an old-fashioned tutor? Or did Dee add them as part of a Cunning Plan? At this point, your guess is as good as mine. Fig 23. Three lines of text, all in the same hand, containing two words of Voynich at the start of line three, and otherwise what looks like german gibberish. And that's another puzzle to me. Yes, the script is unclear. But it's a lot clearer than the script in Marci's letter, shown as Fig 2, and I can read that at sight: "Librum hunc ab amico singulari mihi testamento relictuo..." And I don't have much trouble with, say, Luther's german hand. And anybody writing at Rudolph's court would surely have used either latin or german. So in what language, or what cypher, are the lines in Fig 23? To judge by Petersen's and Newbolds transcriptions, also given there, it made no sense to them either. Fig 21. Again, I find none of this makes sense. However, the text on f66r surely can't be 'der Musstheil', in whatever funky dialect, since the dead person has breasts. Sigh. Another dead end. From firth@SEI.CMU.EDU Fri Jan 03 05:10:02 1992 Date: Thu, 2 Jan 92 15:10:02 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201022010.AA05646@bp.sei.cmu.edu> To: voynich@rand.org Subject: Note 4 - Athanasius Kircher Status: OR Notes on the Voynich Manuscript - Part 4 ---------------------------------------- Athanasius Kircher, SJ, is a guy I've stumbled across several times. He was a most inquisitive and scholarly person, whose influence on later times has been small. Anyway, he was born in 1601, ordained in 1628, and soon thereafter fled Germany and its troubles. After a brief stay in Avignon, he settled in Rome in 1634, and died there in 1680, leaving his library as the nucleus of a museum (but not, it seems, the Voynich MS). There are perhaps two other things of interest. First, he was a secret student of alchemy. He contributed to the scholium on the Tractatus aureus, and I'm currently going through that work again looking for clues. His latin translation of the Emerald Tablet (Hermetis trismegisti tabula smaragdina...) is the one used by most western alchemists; I'll gladly type in my copy but fear it is not relevant. What may be relevant, though, is that Kircher also tried to decipher the egyptian hieroglyphic script. He was the first to conjecture (correctly) that coptic, the liturgical language of the Egyptian Church, was derived from ancient egyptian. He worked for a long time on puzzling out the script. However, later in life, he suddenly announced that he had discovered the "key" to the hieroglyphs, and began to publish his own translations, all of which, alas, were nonsense. I still have to check the dates, but it just might be that he thought the Voynich MS in some way provided that key. From RJB@MAX.U.WASHINGTON.EDU Fri Jan 03 06:18:00 1992 Date: Thu, 2 Jan 92 13:18 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: and now for something completely different To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR From: IN%"mmarkowi@pepvax.pepperdine.edu" 2-JAN-1992 10:15 To: rjb@MAX.U.WASHINGTON.EDU CC: Subj: Musing Received: from pepvax.pepperdine.edu by MAX.U.WASHINGTON.EDU; Thu, 2 Jan 92 10:14 PST Received: by pepvax.pepperdine.edu (4.1/Pep-3.2) id AA16676; Thu, 2 Jan 92 10:16:09 PST Date: Thu, 2 Jan 92 10:16:08 PST From: mmarkowi@pepvax.pepperdine.edu Subject: Musing To: rjb@MAX.U.WASHINGTON.EDU Message-id: <9201021816.AA16676@pepvax.pepperdine.edu> X-Mailer: ELM [version 2.3 PL11] X-Envelope-to: rjb Reading through the Voynich material you sent me, it ocurred to me that perhaps the "text" in question is not really text at all, but some sort of musical notation. This explanation seems consistent with some of the described features, like the repetition of "words," and the ocurrence of certain symbols at rare intervals (key signatures?). Perhaps the Voynich mavens have already considered and dismissed all non-linguistic hypotheses on the basis of other internal evidence, but I am reminded of Maslow's Law: "When the only tool you have is a hammer, you tend to treat everything as if it were a nail." If it is a musical notatation or tablature of some kind, I suppose the chances of ever deriving anything intelligible from it are nil, but every musical form has certain structural regularities that (presumably) can be described mathematically, and tested against any arbitrary string of symbols. If you think this notion has any merit, perhaps you could forward this to the group. From RJB@MAX.U.WASHINGTON.EDU Fri Jan 03 06:38:00 1992 Date: Thu, 2 Jan 92 13:38 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: 200 for changing To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR "of the 200 for changing 60 remain" probably doesn't refer to changing money from one currency to another (I suspect this wasn't an issue in just that form, since wouldn't the gold or silver value of the coin have been what determined its worth, rather than which "currency" the coin belonged to?), but to changing something (gems, gold chains, etc) into money. When the gems are noted as being "esteemed" at 200, and then a bit later 60 is noted as being left "of the 200 for changing," I would read it as "of the 200 simoolians I got for changing the gems at the goldsmiths, 60 remain." Which would imply that Dee was not collecting heirlooms, but was regarding gifts as in-kind grants from his local funding agency. From RJB@MAX.U.WASHINGTON.EDU Fri Jan 03 07:06:00 1992 Date: Thu, 2 Jan 92 14:06 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: music hypothesis To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR From: MAX::RJB 2-JAN-1992 13:25 To: IN%"mmarkowi@pepvax.pepperdine.edu" CC: RJB Subj: RE: Musing Michael: this is a speculation that is far as I know has never been advanced with respect to the VMs. I have forwarded your entitre message to the group; you may get some odd mail because of it.... This should stir things up a bit. It's not (formally) far from the idea that the "text" might be an artificial, "philosophical" language --that notion is not far from certain aspects of cryptology and combinatorial studies in general, and there was a strong interest in combination and permutation. Related hypotheses would involve mathematical expressions of various sorts. But given the redundancy, musical notation might indeed be the most plausible. One might need to justify *so many* notes. Still, a thought that can't be dismissed out of hand... From hudu@well.sf.ca.us Fri Jan 03 17:30:23 1992 Date: Fri, 3 Jan 92 00:30:23 -0800 From: Scott Marley Message-Id: <9201030830.AA07207@well.sf.ca.us> To: voynich@rand.org Subject: Musical notation? Status: OR It strikes me as very unlikely that the manuscript contains encoded music. I can't imagine any musician wanting to record music in a form that would be so difficult to read even if you knew the key. Even given the symbol/note equivalents, it would quite difficult to sing or play this music off the page, I think; perhaps one might recopy it in more standard notation when one wanted to play it, but copying music is a real nuisance, and it just seems too implausible to me that any musician would give himself the headache of recording music in this form. From reeds@gauss.att.com Fri Jan 03 20:05:04 1992 Message-Id: <9201031109.AA12551@rand.org> From: reeds@gauss.att.com Date: Fri, 3 Jan 92 06:05:04 EST To: voynich@rand.org Status: OR The music theory, ingeneous as it is, seems implausible. How about knitting instructions? Any oddities in the resulting shape of garment can be explained as follows: we have here knitting instructions for knitting the raiment of an alchemical adept, whose garb might (or even certainly WILL) seem odd to us! From cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET Sat Jan 04 01:20:02 1992 Message-Id: From: cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET (John Cowan) Subject: Re: Dunstan To: voynich@rand.org (Voynich List) Date: Fri, 3 Jan 92 11:20:02 EST In-Reply-To: <9201021935.AA09127@riesz>; from "math.mit.edu!jbaez" at Jan 2, 92 2:35 pm X-Mailer: ELM [version 2.3 PL11] Status: OR John Dee writes: > Dec. 12th, afternone somewhat, Mr. Ed. Keley his lamp overthrow, the spirit of > wyne long spent to nere, and the glas being not stayed with buks about it, as > it was wont to be, the spirit was spilled out... John Baez writes: > Does the above mean that Kelley was drunk?? No. "Spirit of wine" means "alcohol", that's all; the lamp in question was an alcohol lamp a la high school chem lab. Since the "glas" was not "stayed" (stabilized), the alcohol spilled out. -- cowan@snark.thyrsus.com ...!uunet!cbmvax!snark!cowan e'osai ko sarji la lojban From RJB@MAX.U.WASHINGTON.EDU Sat Jan 04 01:33:00 1992 Date: Fri, 3 Jan 92 08:33 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Firth's note on Kircher To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Didn't Kircher actually publish a collection of alchemical/hermetic texts? I seen to remember poring over a large folio, very good condition, the last time I was at the British Library (much too long ago to remember clearly, and the title has slipped my mind). One of the standard collections. ircher seems to have been the last of the Renaissance neoplatonists -- or rather, of the respectable ones. I remember that AE Waite made some remark (approximately) that "he had died on the eve of accomplishing a great folly: the completion of a treatise on underground caversn, the cavities of the human body, and other vacant spaces, showing their essential identity." As I remember, his idea decipherment of hieroglyphics was linked with the Renaissance neoplatonist/emblematic interpretation that wanted to interpret hieroglyphs as iconic images, rather than as "mere" transcription signs. I think Godwin has done one of those Thames and Hudson picture books on Kircher. From jbaez@math.mit.edu Sat Jan 04 01:43:34 1992 Date: Fri, 3 Jan 92 11:43:34 EST From: jbaez@math.mit.edu Message-Id: <9201031643.AA12257@nevanlinna> To: voynich@rand.org Status: OR My latest angle on the Voynich (while we're waiting to get a copy of the darn thing) is to try to learn more about the court of Emperor Rudolph II. This has the following to recommend it: 1) He's the first well-documented owner of the Voynich, 2) There should be a fair amount available about him, being an Emperor after all, 3) Voynich's perusal of the Prague library dug up a lot of interesting stuff and D'Imperio suggests that there might be more nuggets lying there. Until we get a confederate in Prague --- is there any way, by the way, to pursue this? --- I'll nose around what's available here. Here's one curious thing, for example. You recall from my previous posting that there was a lot of activity involving Dee and Kelley and Emperor Rudolph during 1586-89, with Dee giving the Emperor a copy of his preface on Euclid in 1588, and then, later that year, the Emperor apparently visiting Dee in Trebona for instruction. Curiously, another figure visited the Emperor's court in 1588 - Giordano Bruno! And he too gave the Emperor a book --- one against mathematicians! We may indulge in speculation (just for fun, mind you): were Dee, Bruno, and others involved in a battle for the Emperor's favor and support? Amidst the "one-upping," did one of them sell the Emperor the Voynich manuscript for 600 golden ducats? Back to the facts (from Yates's "Giordano Bruno and the Hermetic Tradition"): Early in 1588, Bruno left Wittenberg for Prague where he stayed for about six months. Here the Emperor Rudolph II held his court and gatered under his wing astrologers and alchemists from all over Europe to assist in his melancholy search for the philosopher's stone. Bruno was not a practicing alchemist, but he tried to interest the Emperor in his "mathesis", dedicating to him a book, which was published at Prague, and which had the provocative title of "Articuli adversus mathematico." [footnote: Bruno also published, or rather republished, a Lullian work in Prague.] It may be only a curious coincidence that Fabrizio Mordente happened to be in Prague at this time, in the position of imperial astronomer. [Mordente had invented a new kind of compass, which Bruno thought a great discovery; since Mordente did not know Latin Bruno offered to publish his discovery for him, whic he did in a dialog called "Idiota Triumphans", in which he compared Mordente, the "triumphant idiot" who discovered something beyond his own understanding, to Balaam's ass who carried the Sacraments. Mordente, enraged, bought all but 2 copies of this work in 1586 and destroyed them.] The book "against mathematicians" is illustrated with an intriguing collection of diagrams, a selection from which I [Yates] here reproduce... The Emperor gave Bruno money for his mathesis "against mathematicians," but he did not give him any employment or position. He went on to Helmstedt. [He matriculated at the Julian university in Helmstedt by January 13th, 1589.] Yates mentions in a footnote: "John Dee's associate, Edward Kelley, was at Prague at the time of Bruno's visit, and in very high favour with the Emperor (see C. Fell Smith, "John Dee," 1909, pp. 179 ff.)" It would be interesting to track down this reference to Smith. From firth@SEI.CMU.EDU Sat Jan 04 02:00:57 1992 Date: Fri, 3 Jan 92 12:00:57 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201031700.AA07724@bp.sei.cmu.edu> To: voynich@rand.org Subject: Kircher - I goofed Status: OR Sorry, folks; my previous speculation was dead wrong. Kircher's "Oedipus aegyptiacus" was published at Rome in 1654, over a decade before he saw the Voynich MS. To compound the screw up, I reversed the chain of cause and effect. It seems that Kircher's fame as the "decipherer" of the egyptian hieroglyphics was a major reason Marci sent him the MS, and the evidence is right there in the letter: "Verum labor hic frustraneus fuit, siquidem non nisi suo Kirchero obediunt eiusmodi Sphinges". That is, "But his toil was in vain, for such Sphinxes as these obey no one but their master, Kircher." (I'm not too happy with that translation, but it's from D'Imperio's Fig 3.) The allusion is pretty clear, I think. Sigh. Scratch another half-baked idea. Robert From jbaez@math.mit.edu Sat Jan 04 02:18:30 1992 Date: Fri, 3 Jan 92 12:18:30 EST From: jbaez@math.mit.edu Message-Id: <9201031718.AA12360@nevanlinna> To: voynich@rand.org Subject: Athanasius Kircher Status: OR There is quite a bit about Kircher in Yates book on Bruno; I'll post a summary when I'm done reading it. I think it's a good idea for all of us to learn about the milieu of the early history of the Voynich; it may eventually provide some leads. jb (Also, it's interesting.) From jbaez@math.mit.edu Sat Jan 04 03:26:31 1992 Date: Fri, 3 Jan 92 13:26:31 EST From: jbaez@math.mit.edu Message-Id: <9201031826.AA12503@nevanlinna> To: voynich@rand.org Status: OR Return of the the Deadly Isis Cult or the varied interests of Athanasius Kircher I have never seen Levitov's book claiming that the Voynich was the work of an "Isis cult," but I would not be surprised if he took as evidence for this Athanasius Kircher's interest in Isis and Osiris. The following information is taken from Yates' book on Bruno. In 1614 Isaac Casaubon (not the Meric Casaubon who wrote of Dee's encounters with spirits!) showed that the works of Hermes Trismegistus were not of great antiquity, as the Hermetists had believed. There were however "reactionary Hermetists," as Yates calls them, who remained interested. She treats 3: Robert Fludd, the Rosicrucians, and Kircher. In 1652 Kircher published a vast work on hieroglyphs, the Oedipus Aegyptiacus. (Clearly the reference to Oedipus alludes to the riddle of the Sphinx, also alluded to in Marci's letter to Kircher.) The basic idea of Kircher is (in his own words) that "Hermes Trismegistus, the Egyptian, who first instituted the hieroglyphs, thus becoming the prince and parent of all Egyptian theology and philosophy, was the first and most ancient among the Egyptians and first rightly thought of divine things... And this Trismegistus was the first who in his "Pimander" and "Asclepius" asserted that God is One and Good, whom the rest of the philosophers followed." Kircher was preoccupied with Isis and Osiris as the chief gods of Egypt. He was fascinated with Heliopolis, and with the ankh (the Egyptian "cross"). He drew his own version of Dee's "Monas Hieroglyphica" in his book "Obeliscus Pamphilius" (pub. 1650). He was a Cabalist of "vast learning," but condemned Cabalist magic, himself practicing natural magic. If Macri knew that Dee, or any other bigshot in the occult scene, was involved with the Voynich ms., and if Macri had any knowledge of who the bigshots were, it seems likely that he would have mentioned that fact to Kircher in his letter. (After all, he mentions the Bacon connection.) I forget if we have anything about Macri that would shed light on this. Anyway, I chalk it up as a small piece of evidence *against* Dee's involvement with the Voynich. From j.guy@trl.OZ.AU Sat Jan 04 04:42:29 1992 Date: Fri, 3 Jan 92 14:42:29 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201030342.AA03205@medici.trl.OZ.AU> To: voynich@rand.org Subject: Fonts again Status: OR I have just received from Jim Gillogly a photocopy of Currier's alphabet, with the dreaded "7" clearly drawn. Here is a uuencoded zip file with two fonts with "7" added. I have assigned it the ampersand: 1. it looks a lot like an ampersand: & 2. the ampersand is on the same key as "7" The file (7.zip) will unzip to two files: 7s.fnt and 7l.fnt. The only difference between the two is the height of c, e, and t. (one pixel shorter in 7s.fnt than in 7l.fnt). To have TRANSLIT translate from Currier's system into mine and back just change this line in the CURR2GUY file: instead of: 7=i-g make it: 7=& -----------CUT HERE---------------------------------------------------------- end From firth@SEI.CMU.EDU Sat Jan 04 05:52:55 1992 Date: Fri, 3 Jan 92 15:52:55 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201032052.AA08201@bp.sei.cmu.edu> To: voynich@rand.org Subject: Notes 5 : Comments on previous decipherments Status: OR Notes on the Voynich Manuscript - Part 5 ---------------------------------------- Comments on previous decipherments. Chapter 5 of D'Imperio's monograph is called "Major Claims of Decipherment", and one thing I did over the break was read this chapter and check back to the major references, most of which are reprinted in Brumbaugh's collection. This is a summary of what I found, with some tentative deductions. 5.1 Newbold I agree with Manly and D'Imperio that the decipherment is incredible. The proposed means of encoding is by micro symbols visible only under the microscope. The proposed decipherment contains sufficient leeway that almost anything can be read into any text. Nuff said. 5.2 Feely He at least realised that, if the plain text is latin, it would be mediaeval latin. He seems to have started out with conventional "gold bug" analysis - letter and word counts, both on the Voynich MS and on Roger Bacon's works. As far as I can tell, he got nowhere. His next attempt was based on guessing the individual words that labelled the drawings. This is not easy, for it implies guessing first the language, then the meaning of the drawing, and finally the word labelling it. So, even if the text is latin, and the drawing is of a poppy, is the text "poppy" or "to induce sleep" or "for headaches"? In my view, his guesses were pretty much wrong, and the proof is that the keys they provide don't unlock the text with any degree of credibility. 5.3 Strong This suffers, in my view, from the same defects as Newbold: the cypher is too complex, and the recovered text is not credible. 5.4 Brumbaugh This is a critical claim, for if it is true, the MS is indeed largely a forgery. Brumbaugh's cypher is pretty simple, but be warned that I am depending on D'Imperio's reconstruction, since Brumbaugh's own articles don't give full information. In brief, the decoding process is this: each Voynich symbol stands for one of the digits 1 to 9, and there are about three alternatives for each digit, to obfuscate. Each digit, in turn, stands for one of three possible letters of the roman alphabet, take your pick. The result, when you pick wisely, is intelligible dog-latin, ie latin much simplified and with the inflections largely replaced by the common ending -US. The encoding process, of course, is the reverse: collapse the plain text into a sequence of digits, and write the sequence in Voynich, with appropriate alternation rules to ensure that each symbol gets used a decent number of times. Here is an example of a decoded word 12127339 ABABGCCI JKJKPLLZ VRVRYWWus showing the digit sequence and the three possible choices for each digit. Note: this is given on p37 of D'Imperio in a form that cannot be reconciled with the tables in her Fig 26. I have changed it to make it consistent. Note also that -us is encoded by one symbol, the digit 9 (one of whose Voynich symbols is also "9"). Simple mathematics tells us that there are 3^8 or 6561 possible decipherments of this word. Brumbaugh chooses "ARABYCCUS", which is passable dog-latin for "arabiclike". And, indeed, there are very few choices that lead to pronounceable words. Brumbaugh's critical argument is this: that, in spite of this ambiguity, one who understands what the plain text is about can indeed read the cypher text and comprehend it. I propose to test that by an experiment. Here is a familiar latin phrase in the Brumbaugh code, with the possible decodes beneath it: 5619 6246 45336284 EFAI FBDF DECCFBHD NOJZ OKMO MNLLOKQM XTVus TRST SXWWTRUS Did you read the phrase? If so, Brumbaugh's argument is at least partially confirmed. Now, here is the entire encode/decode box, from D'Imperio: 1 2 3 4 5 6 7 8 9 A B C D E F G H I J K L M N O P Q Z V R W S X T Y U us My first comment, is that if this is meant as a device to encode latin plain text, it's a pretty bad one. Maybe we need to distinguish I and J, though most authors didn't, but we surely don't need both U and V. After all, whoever wrote the signature numbers didn't bother to distinguish. Then, why have one symbol for -us but not one for qu-? And, if the purpose is to encode latin, why have K and W at all? Finally, assigning the Voynich symbol "9" to -us is really stupid, since that is already a well-known shorthand symbol for the same ending. And given that this supposed -us is one of the most common features of the text, the choice of "9" as the cypher symbol makes no more sense that choosing a substitution cypher for english that encodes 'e' as "E". Doctor Dee's double bluff? Far-fetched, I think. Secondly, while it is probable that the inflection structure of any underlying language has been simplified, I doubt very much that any educated person would "simplify" latin by making every noun end in -us. Latin has been deliberately simplified at least twice, and has simplified itself many more times in the historical evolution of the romance languages, and as far as I know in every case the simplification process discarded the -us ending, usually for something derived from the dative or ablative case. And my final thought was this: Brumbaugh in his deciperment is violating his own key assumption, for he is a reader who does NOT know what the underlying text is about. That led me to another experiment, where I tried to apply his decoding rules myself. Here is the digit sequence and the decode 676517 14665 28379 851 581 84 46 FGFEAG ADFFE BHCGI HEA EHA HD DG OPONJP JMOON KQLPZ QNJ NQJ QM MO TYTXVY VSTTX RUVWusUXV XUV US ST Which I read, allowing myself similar latitude as did Brumbaugh in respect of contractions and simplicications: OP[um] TE AG[e]; AMO TE BULGUS; UNA EQU[i] US[um] DO "Do ye the work; I love you, ye Bogomil; I give to one [woman] the use of a horse!" Which proves, perhaps, that the Voynich MS is indeed part of a secret kinky Cathar cult? Not quite. You see, the original digit sequence is not taken from the MS; it is taken from the rightmost column of a random page of my table of seven-figure logarithms, using spaces and zeros to divide the groups. Which proves rather, I think, that the Brumbaugh method does not recover the underlying text; the meaning is inserted by the ingenuity of the decipherer, and any text whatever may be so "deciphered". My tentative conclusion, folks, is that we have here pitfalls to be avoided. Robert From RJB@MAX.U.WASHINGTON.EDU Sat Jan 04 07:15:00 1992 Date: Fri, 3 Jan 92 14:15 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Novus Ordo Seclorum interferes with my mail To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR From: IN%"MAILER@UWAVM.U.WASHINGTON.EDU" "Network Mailer" 3-JAN-1992 14:13 To: RJB@MAX.U.WASHINGTON.EDU CC: Subj: mail delivery error Received: from JNET-DAEMON by MAX.U.WASHINGTON.EDU; Fri, 3 Jan 92 14:13 PST Received: From UWAVM(MAILER) by MAX with Jnet id 4309 for RJB@MAX; Fri, 3 Jan 92 14:13 PST Received: from UWAVM.U.WASHINGTON.EDU by UWAVM.U.WASHINGTON.EDU (Mailer R2.08 R208004) with BSMTP id 6614; Fri, 03 Jan 92 14:14:48 PST Date: Fri, 03 Jan 92 14:14:48 PST From: Network Mailer Subject: mail delivery error To: RJB@MAX.U.WASHINGTON.EDU Message-id: X-Envelope-to: RJB Batch SMTP transaction log follows: 220 UWAVM.U.WASHINGTON.EDU Columbia MAILER R2.08 R208004 BSMTP service ready. 050 HELO UWAVM 250 UWAVM.U.WASHINGTON.EDU Hello UWAVM 050 MAIL FROM: 250 ... sender OK. 050 RCPT TO: 550 Mailbox not found. 050 DATA 354 Start mail input. End with . 050 QUIT 221 UWAVM.U.WASHINGTON.EDU Columbia MAILER BSMTP service done. Original message follows: Received: from UWAVM by UWAVM.U.WASHINGTON.EDU (Mailer R2.08 R208004) with BSMTP id 6613; Fri, 03 Jan 92 14:14:48 PST Received: from MAX.U.WASHINGTON.EDU by UWAVM.U.WASHINGTON.EDU (IBM VM SMTP V2R1) with TCP; Fri, 03 Jan 92 14:14:47 PST Date: Fri, 3 Jan 92 14:13 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: George Bush, Freemasons Subvert firth@sei.cmu.edu To: voynich@rand.txt Message-id: X-Envelope-to: voynich@rand.txt X-VMS-To: IN%"voynich@rand.txt" Actually, though I agree that the Brumbaugh method is full of traps, I disagree slightly about where those traps are. Maybe. Jacques Guy was kind enough to send me a passage in a telephone dial approximation to Brumbaugh's Nine Chamber code. I, supremely lazy and by no stretch of any imagination a cryptologist, managed to get a readable text (with only one crux) using methods close to those I imagine would have been available to people of Dee's period, within about 2-3 hours. But I gave myself no lattitude in terms of assumptions about how losse my "decrypted" text could be. Brumbaugh's problem is not that his method cannot give workable results (it certainly did in that passage), but that his application of that method can turn gibberish into text. My application would I think have left me with gibberish -- or more liklely a case of terminal vexation and a firm resolve to avoid the near occasion of cryptography in the future. The question is, does a method, applied by those who know nothing of the results of the inventor of the method, give the same "decrypted" text? The difficulty is that one must be able to read that text in order to decrypt it (so that there are really two decryptions occurring: the application of the key, and the application of one's "semantic selection process" to the results yielded by the key. If the Nine Chamber method were used to encrypt a text in a language --or spelling--one couldn't recognize, then one would have no effective way of filtering the meaningful from the meaningless in the decryption matrix. It is a cryptological question -- perhaps a purely empirical one -- as to whether putting a text in an unknown language/spelling through a Nine Chamber encryption could produce a an excrypted text that would yield meaningful statistics. I don't know. I have heard that my copies of Levitov and D'Imperio have arrived. On the principle of saving the best bite for last, I will read Levitov first... From reeds@gauss.att.com Sat Jan 04 08:22:55 1992 Message-Id: <9201032323.AA25700@rand.org> From: reeds@gauss.att.com Date: Fri, 3 Jan 92 18:22:55 EST To: voynich@rand.org Status: OR Two questions, one important, one friv: 1. What if the film (or legible prints therefrom) costs big bucks? Do we chip in to share the cost, and if so, how? It's silly to be too elaborate until John has talked with R. Babcock & gotten a quote, but the problem will come up soon. 2. How does one pronounce Prescott Currier's surname? I was talking with Lou Kruh today, and said that we had gotten a letter from him. (BTW, my copy has not arrived yet!) Lou said, ``You mean, Cooyay answered your letter? Wow!'', giving the name an American's French pronounciation. (So: is it Cooyay or Curr-yerr (as in Curr-yerr House at Radcliffe)? Jim Reeds From rcarter@isis.cs.du.edu Sat Jan 04 13:41:35 1992 Date: Fri, 3 Jan 92 21:41:35 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9201040441.AA02007@isis.cs.du.edu> To: voynich@rand.org Status: OR Some pertinent[?] references from: John Dee 1527-1608, Charlotte Fell Smith, Constable & Co., London, 1909 Pages titled _Life of Dr. John Dee_, 342 p. w/ index, B D359s (Dewey) [Ask and ye shall receive...] I just finished reading this for the second time; the following are passages from the book; my comments are in [] and after as well: p.77 After this savoury episode, [being charged with grave-robbing] Kelly is reported to have been wandering in Wales (it is suggested he was hiding from justice), when he stumbled accidently upon an old alchemical manuscript and two caskets or phials containing a mysterious red and white powder. p.78 With the powder that he did not know how to use, and the alchemical manuscript which he could not decipher, and which yet might contain the invaluable secret (if indeed there is any truth in the story of his find). Kelley, the adventurer, sought out some means of introduction to the man [Dee] so likely to help him. p.91 Dee was now puzzling over some mysterious papers brought him by Kelley, whether those he is reported to have found in Wales or Glastonbury we can scarcely decide, but they seem to concern ten places in England where treasures was supposed to be hid. There is a curious of them in the MS. diary: "After coming from the Court, I thought I would try to discover the cipher of the paper E. K. [Kelley] brought me as willed to do, found at Huets Cross, with a book of magic and alchemy, to which a spiritual creature led them." Dee was by no means the easy dupe of Kelley that he has been called. Two or three months after he first knew him he writes in his diary of his "abominable lyes"; and he here makes a very telling remark, an aside, so to speak: "Of this K., I doubt yet." p.92 The manuscript in crabbed signs puzzled the astrologer desperately, and he was unhappy at the delay. p.106 Kelley, it seems, had been met coming from Islington with his scroll, book, and powder, and been threatened to "be pulled in pieces' if he brought them to Dee. A drawing in the margin of the MS. shows the book to have had a cross on the cover, one clasp, and deep metal bands across its two sides. Presumably these were some of the treasures reported to have been found at Glastonbury. p.135 The next few weeks were taken up with instructions from Gabriel and Nalvage [angels], consisting of letters, numbers and words in a strange Eastern or angelic language, to which Dee probably had some key, though they appear unintelligible. ... He [Dee] explained [to Kelley] that he had now a distinct clue to the meaning of the tables of letters on which he had long been puzzling; dwelt on how essential it was to miss not a single letter, or else the words would err. p.167 On his information it appears that three copies of Dee's manuscripts were burned in Prague, April 10, 1586. These were the _Book of Enoch_, the _Forty-eight Keys of the Angels_ (_Claves Angelicae_) and the _Liber Scientiae Auxilii et Victoiae Terrestris_, works which had been written down from the spirit revelations since the partnership with Kelley had commenced. The books burned were not of course the originals, the two first of which still exists. [Sloane MSS., 3189 and 3191] Of the _Book of Enoch_ there are three copies, one made by Kelley, a remarkable tribute to the mechanical skill in draughtmanship, the extraordinary application and ability, of this very versatile personage. It contains hundreds of diagrams of figues, round or rectangular in shape, composed of an infinite number of minute squares each containing a letter or figure. These letters occur in every possible combination and order, some reading straight across the page, others diagonally, and so on. p.199 On February 4 [1589] he [Dee] also made over the Kelley "the powder, the books, the glass, and the bone, for Rosenberg [Count William Rosenberg ie William Ursinus, a patron] and he thereuppon gave me discharge in writing of his own hand subscribed and sealed." [Dee had used Rosenberg to tranfer a "convex glass" to Emperor Rudolph II previously; Dee never saw Kelley [or the book?] again.] p.311 [A letter about Dr. Arthur Dee, John Dee's son] "And that Count Rosenberg played at quoits with silver quoits, made by projection as before. That this transmutation was made by a powder they [Dee and Kelley] had, which was found in some old place, and a book lying by it containing nothing but hieroglyphicks; which book his father bestowed much time upon, but I could not hear that he could make it out. He said that Kelley dealt not justly by his father, and that he went away with the greatest part of the powder, and was afterwards imprisoned by the Emperor in a castle, from whence attempting to escape down the wall, he fell and broke his leg, and was imprisoned again." p.321 No adequate idea of the remarkable doings of Dee and Kelley over the crystal can be entertained without a study of Dee's manuscript "Book of Enoch" in _Sloan MSS. 663, 120, and 2,599, 1-45; and his "Claves Angelicae," 3191 in the same collection. The diagrams of complicated arrangement of letters and figures, their neatness of execution, mathematical precision and etymological intricacy are no less amazing than the clear bold text in which the descriptions are written in printing hand. Regretfully [!!!] it was decided not to reproduce an example, owing to the lack of pictorial value. [Personal comments, etc to follow] Regards, Ron. | rcarter@nyx.cs.du.edu From jim@rand.org Sat Jan 04 19:36:46 1992 Message-Id: <9201041036.AA03715@mycroft.rand.org> To: voynich@rand.org In-Reply-To: Your message of Fri, 03 Jan 92 18:22:55 -0500. <9201032323.AA25700@rand.org> From: Jim Gillogly Reply-To: jim@rand.org Date: Sat, 04 Jan 92 02:36:46 PST Sender: jim%mycroft@rand.org Status: OR > reeds@gauss.att.com writes: > > 1. What if the film (or legible prints therefrom) costs big bucks? I'd be willing to chip in some bucks ($100?) to get the film, and let John Baez hold it for us, if it's expensive. If it's $100 or less, I'd like my own copy. Wayne Barker (Aegean) has volunteered to get prints made from the film and make them available to everybody, presumably at normal Aegean Park Press prices. > 2. How does one pronounce Prescott Currier's surname? I was talking D'Imperio pronounced it like the type font "Courier"... i.e. American pronunciation. Since she worked with him on this stuff, that's good enough for me. Jim Gillogly From RJB@MAX.U.WASHINGTON.EDU Sun Jan 05 00:03:00 1992 Date: Sat, 4 Jan 92 07:03 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: The film To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I too would be willing to be part of a subscription list if the film should turn out to be too expensive. I am also of course a supporter of the idea of having a paper edition made. Such an edition should probably be more than just an off-print from the film. From Karl.Kluge@G.GP.CS.CMU.EDU Sat Jan 04 17:30:00 1992 Date: Sat, 4 Jan 1992 17:30-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Re: "Where did Kelly get the Voynich?" & Brumbaugh Message-Id: <694564254/kck@G.GP.CS.CMU.EDU> In-Reply-To: Ron Carter's mail message of Sat, 4 Jan 92 12:35:31 MST Status: OR My guess would be "he made it" as per an earlier message. The Enochian calls were a very elaborate production, involving not only an alphabet, vocabulary, and grammar but also a rather involuted procedure for extracting them that must have required great effort on Kelly's part. If he could fake one alphabet, grammar, and vocab, why not two? The "two hands" evidence may argue against this. In any case, I think an examination of the entropy of the corpus of Enochian calls to see if it is also abnormally low would be an interesting first test (and if we can pick one version, whether the HOGD version, the Turner version, or taking them from Laycock, entering them in should be fairy easy if tedious). Robert Firth noticed that Fig 26 in D'Imperio is inconsistant with the "arabyccus" example. If you play along at home with Brumbaugh's explanation of the plant names in folio 100r, it becomes clear that this isn't the same letter->digit mapping given in Fig 26 (he mentions statistics suggesting three different mappings at one point in his book). In particular, the alternate readings of ULCER/ALFAR for one plant indicates that plaintext A, U, and E map to the same characters (Bennett A and O). Karl From jbaez@math.mit.edu Sun Jan 05 04:31:52 1992 Date: Sat, 4 Jan 92 14:31:52 EST From: jbaez@math.mit.edu Message-Id: <9201041931.AA13293@nevanlinna> To: voynich@rand.org Subject: Big bucks? Status: OR Here's what I know so far about the PRICE of the Voynich: Costs are as follows: If a new negative has to be made (unlikely), $.40 / frame + $25 base charge Microfilm is aviable from existing negatives for $30 Paper copies can be made for $.25/frame + $20 base If this is judged as a pre-1600 manuscript, there is also a $20/HOUR charge for doing these things ("supervision"). ---- So -- I will pay for the thing unless it's more than, I don't know, $250, and then "sell" copies to people for price/(number of buyers plus me) each. I see Reeds, Gillogly, RJB, Guy, Firth, and Osgood as likely buyers. I received Currier's transcription table from Jim Gillogly, by the way, and (in a case of 1 picture = 10^3 words) it renders the 7/J controversy obsolete. I was seeing 7's not J's. Would it make sense to prepare a GIF of this (or a cleaned-up version) for novice transcribers? jb From rcarter@isis.cs.du.edu Sun Jan 05 04:35:31 1992 Date: Sat, 4 Jan 92 12:35:31 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9201041935.AA27576@isis.cs.du.edu> To: voynich@rand.org Status: OR Personal comments on _Life of Dr. John Dee_, Smith, 1909 This biography seems important for several reasons: -The publish date of 1909, three years before the `discovery' of the Voynich MS. doesn't `bias' Smith on a `need' to find the Voynich. -It appears that Smith's Dee is/was the first `modern' `complete' work on the subject; research appears to be extensive. After primarily reading this and the preface to Euclid's _Elements_ (Billingsley, 1570) my personal feelings are that Dr. John Dee was quite a remarkable and visionary person; proposing an organized naval system for England, proposing a national library, arguing for a 11 day (which were omitted in 1742) calendar reformation, instead of the prevalent thought of 1583 for a ten day change. His travels and writings were extensive, etc. Impressive stuff. I also came away with the strong impression, that if the Voynich was in the hands of Dee, it came way of Edward (Talbot) Kelley. I also have the feeling that it left with Kelley as well. Kelley doesn't come across as the most trustworthy of individuals; it seems that any gold `production' was claim by him; Dee spoke to angels -through- Kelley (or other skryers) and it doesn't appear directly. The question appears that if the Voynich -did- come via Kelley (and was sold by Kelley to Emperor Rudolph II, with Kelley then keeping the money) where did Kelley get the Voynich? Regards, Ron | rcarter@nyx.cs.du.edu From jbaez@math.mit.edu Sun Jan 05 04:45:55 1992 Date: Sat, 4 Jan 92 14:45:55 EST From: jbaez@math.mit.edu Message-Id: <9201041945.AA13321@nevanlinna> To: voynich@rand.org Subject: Publishing the Voynich Status: OR It'd be lots of fun for all of us to write a short introduction to a text of the Voynich, to be published by Aegean Park Press... although simply having them reproduce it well would be worthwhile. I wonder if Yale would get pissed if we published a "book" of the Voynich. From reeds@gauss.att.com Sun Jan 05 05:26:44 1992 Message-Id: <9201042030.AA07158@rand.org> From: reeds@gauss.att.com Date: Sat, 4 Jan 92 15:26:44 EST To: voynich@rand.org Status: OR A recent trip to the local public library netted ``The Alchymist's Journal'' by Evan S. Connel, a very odd recent novel which will entertain you, I'm sure. From rcarter@isis.cs.du.edu Sun Jan 05 05:34:09 1992 Date: Sat, 4 Jan 92 13:34:09 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9201042034.AA29487@isis.cs.du.edu> To: voynich@rand.org Status: OR Some semi-first impressions re: the Voynich MS. & misc. musings. I first became aware of crypto work when I read Doyle's _The Adventure of the Dancing Men_ when I was 10 or 11 (1965 or so); first exposed to the Voynich when I read _The Codebreakers_ (Kahn, 1967) in '68 or 69'; this book made a big impression on me at 14. After becoming reacquainted with it (rereading Kahn, rereading Bennett's _Scientific and Engineering Problem-solving with the Computer (1976) and Poundstone's _Labyrinth's of Reason_ (1988), as well as starting D'Imperio's _The Voynich Manuscript - An Elegant Enigma_ (1976?)) I have formed some semi-first impressions. As already concluded by others, a full transcription needs to be done; only then can a full analysis be made; seemingly, this will be the first time this will have been done, with the final product in ASCII, and available to anyone with access to a computer and modem. Also concluded, is that we need access to relatively low cost copies of the (in)complete Voynich; I can go $100 or so (50 cents a page(?)) but not much more. I don't think enough emphasis can go towards the theories that the MS. was written (transcribed?) by two people (A&B) and this needs to be (re)taken into consideration for final analysis. It is also my impression that the main text of the MS. should not be automatically associated with the drawings on the same page; it is obvious that the drawings were done first on the page, perhaps -all- the drawings were done at one time (a third person?) with the text then placed later, with the possibility of the text not coordinating with the illus. Misc. It might be interesting to see if we can track down a better history of the MS.; I obviously(?) favor the route through Kelley; however, I wouldn't be surprised that it came to Rudolph through someone else of the court; there were alot MS. collectors around at the time. Any such work would probably have to be done by someone with very good connections with the British Mus./Oxford/Cambridge. It is interesting that even D'Imperio failed to note that Dr. Dee often made notes of Latin words in Greek characters (Smith, p.128); also of interest(?) is that Kelley apparently didn't know Greek. Awaiting news on Voynich copies being available, I am working on a timeline of Dr. John Dee, based on Smith. On a not so scholarly subject; should we name ourselves? We could take after previous attempts with something like "The First UseNet Voynich Manuscript Study Group, 1991-?" or maybe something more sport (and shorter) like Team Voynich (which I like)... On a personal note, I am 37, with formal backgrounds in art (modern) and electronics engineering; the past 7 years have been spent doing brain-behavior research as the director of a non-profit research group. As might be expected, a lifelong informal interest in crypto. Regards, Ron | rcarter@nyx.cs.du.edu From jbaez@math.mit.edu Sun Jan 05 06:02:09 1992 Date: Sat, 4 Jan 92 16:02:09 EST From: jbaez@math.mit.edu Message-Id: <9201042102.AA13494@nevanlinna> To: voynich@rand.org Subject: Novel about Dee and Kelley Status: OR While we're on the subject of novels (to entertain ourselves between bouts of transcribing and hacking), I strongly recommend John Crowley's "Aegypt." This is the most recent novel by an excellent science fiction/fantasy writer. It appears to be the first of a four-volume series. It concerns one man's quest for information on "Aegypt" -- not the "Egypt" that still exists, but the "Aegypt" as it existed in the minds of the Renaissance Hermetists. (Do you know of the city Adocentyn, for example?) Interspersed between the modern action is a novelized tale of the doings of Dee and Kelley -- rather close in spirit to the pieces of Smith's biography recently posted by Ron Carter. It's not one of Crowley's best novels, but it's good. jb From jbaez@math.mit.edu Sun Jan 05 06:57:02 1992 Date: Sat, 4 Jan 92 16:57:02 EST From: jbaez@math.mit.edu Message-Id: <9201042157.AA13559@nevanlinna> To: voynich@rand.org Subject: name Status: OR I too would like our group to have a name -- or perhaps two. First, I want a name that would suitably impress people at Yale, or people who do radiocarbon dating -- a name that'd give the impression that we're a group of scholars, not a bunch of occultist loonies. Second, it might also be fun to have an informal name. But I would not want to tell Babcock that "Team Voynich" is desperate for a photocopy. jb From jbaez@math.mit.edu Sun Jan 05 07:12:28 1992 Date: Sat, 4 Jan 92 17:12:28 EST From: jbaez@math.mit.edu Message-Id: <9201042212.AA13563@nevanlinna> To: voynich@rand.org Subject: The 2/R controversy again Status: OR Okay: Jacques Guy wrote The part of the R that rests on the line is c-like. From the thickness of the stroke, one can tell that it was written counter-clockwise, just like we write "c". The part of the 2 that rests on the line is Voynich i-like. It was written from top to bottom, leaning considerably left. The stroke is straight, and the same thickness its whole length. The top part of R and 2 was written second, from bottom to top, counter-clockwise, its bottom (start) linked to the top (start) of the first stroke. Another hint is in the letter that precedes R or 2: C *NEVER* precedes 2 A and I *NEVER* precede R (That's according to Currier, and I wasn't aware of that particular observation by him, when I wrote my Cryptologia article in which I mentioned a similar phenomenon). ------ The former part seems to be the exact opposite of how Currier draws 2's and R's in what I received from Jim G. Namely, the part of the R that rests on the line is I-like (straight vertical) while the part of the 2 resting on the line is C-like (an arc). In any event, if everyone pulls out their Brumbaugh and looks at the plate of folio 93r, at the last character of the first line, we see what I call an R -- directely following an A. The fourth character from the end of that line is what I'd call a 2. Do I have it backwards, or what - my identifications certainly go along with what I got from Jim G! Sigh. Even if we straighten out those two examples, there seem to be intermediate forms. This'll be the death of me. jb (You will be pleased to know that distinguishing 7's from J's is easy for me now.) From RJB@MAX.U.WASHINGTON.EDU Sun Jan 05 07:21:00 1992 Date: Sat, 4 Jan 92 14:21 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Publishing the Voynich To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Yale shouldn't get too bent out of shape. There are I'm sure well established procedures for doing such things: it's frequently done, after all. I coulkd ask our special collections people here what the conventions & customs are. They might like a royalty, and I for one could see no objection to agreeing to give them some significant cut of net profit. (When the Voynich hits the NY Times best seller list for the fourth month in a row, naturally, there will be much sighing.) They would definitely want the usual formalities: acknowledgement, recognition of their rights, no doubt some free copies. I don't know what the copyright status of the text -- or the images once published -- would be. With any luck that will not be an issue. The only real (hard to negotiate) problem I could see arising would be if someone had "academic rights" to working on the text. {As in the Dead Sea Scroll biz). Seems quite unlikely, however, that anyone would. Or could, without someone in this group having an inkling of it. Too many people have published bits and pieces. I rather suspect they'd be glad enough to have a copy people could look at without risking damageto the unique original. From RJB@MAX.U.WASHINGTON.EDU Sun Jan 05 07:28:00 1992 Date: Sat, 4 Jan 92 14:28 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: where did EK get the V? To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Ron's speculation with respect to what the likely course of the VMs through the D/K circle was, *if * in fact it did go through that circle, seems quite reasonable. *If* it did, there are two possibilities. (A) EK created the VMs; (B) other than A. If (a) is the case, the ms might or might not look like other EK productions --if it did resemble them, one might assume A was true. If it didn't, A might or might not be true. If (b), the ms would have presumably come from somewhere before it was in EK's possession. To establish this, one would have to establish existence at a date and place early and/or far enough to make it impossible for it to have originated with EK. If there are no published indications, the only recourse would be thorough investigations of libraries and catalogues of libraries in Prague & other place, in hopes that some evidence might survive. Laborious. From RJB@MAX.U.WASHINGTON.EDU Sun Jan 05 07:47:00 1992 Date: Sat, 4 Jan 92 14:47 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Connell and Crowley To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR The Connell I have been slowly reading; it's good, but I suspect I would need more leisure to do it justice. The Crowley is one of my favourite "hermetic" novels -- I by the way happen to think it one of his best. I evidently somehow missed the information that suggests it's part 1 of 4. I had long toyed with a novel about Dee, Kelley, and photography...but didn't feel at all put out to see someone actually do it, and do it so nicely. Adocentyn is the talismanic city of the _Picatrix_, which if anything is the fabled and dire _necronomicon,_ is. {An odd sentence indeed} There was a translation of the Pivatrix done into Latin m(incomplete) and I think from there into English (though I could be confused about that). There is also a German (modern) translation of the Arabic original. And a modern edition of the Latin translation. Both modern editions were put out by the Warburg Institute. They were planning to do the English -- or maybe a new English translation of the whole thing. Don't know if they ever did though. --rjb From reeds@gauss.att.com Sun Jan 05 07:58:21 1992 Message-Id: <9201042258.AA08630@rand.org> From: reeds@gauss.att.com Date: Sat, 4 Jan 92 17:58:21 EST To: voynich@rand.org Status: OR I like Team Voynich! I can just see coffee mugs and T shirts emblazoned with Team Voynich, and some appropriate logo (some of the fancier V letters, maybe, or some V ladies in typical poses, as supporters), maybe with a Michiton Oladabas underneath as a motto! I know the guy who put out the KGB coffee mugs a couple of years ago; I'm sure he would jump at the chance to make V mugs for us. But we should wait for some mile-stone, such as publication of transcription. or verification of solution, say, before we sport such jaunty gear. I sympathise with John's plea for a more staid work name; I have suggested ``Informal Voynich Manuscript Study Group'', VMSG, to him. As for group authorship of a book, I think we should follow H. F. Gaines's lead. I understand that the ACA gathered suggestions, which Gaines then synthesized into a (largely) seamless book, ``Elementary Cryptanalysis'', which has been in print almost uninterruptedly for 50 years. Gaines's name is on the title page, Ohaver (I think) is credited with a chapter, and the rest of the ACA is clearly thanked in the introduction and elsewhere in the book. In short: you do the work, you get on the title page. From RJB@MAX.U.WASHINGTON.EDU Sun Jan 05 08:10:00 1992 Date: Sat, 4 Jan 92 15:10 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: name To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Yes, I think "Voynich Raiders" or "Raiders of the Voynich Text" might give a somewhat inaccurate impression. Besides, I'm not sure Harrison Ford is free right now. How about something Academically Correct (=dreary but hi-tech) like International Voynich Study Section? or Voynich Studies Group? Voynich Textual Study Group? Putting the title into Latin would be perhaps too 18th-century even for Yale, but a French or German name might do -- something like Voynichforschungsinstitut might be a nice word to have in a name. One must consider the letterhead, too. And the mailing address. I would hesitate to put "Voynich" into Chinese, but a Japanese letterhead might have its value, suggesting a farflung international interest, perahps a certain need to get things done as expeditiously as possible, and even the possibility of funding somewhere in the background.... but that would be wrong. International Voynich Textual Studies Group? A poor acronym, but accurate, descriptive, and respectable. From RJB@MAX.U.WASHINGTON.EDU Sun Jan 05 08:16:00 1992 Date: Sat, 4 Jan 92 15:16 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: publishing the Voynich To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Unless there are strong objections, I suspect one might be able to make a plausible case for getting some sort of grant (NEH? some local scholarly project fund?) to help cover the costs of producing a publisahed version of the VMs. Of course, that would introduce the whole apparatus of bureaucratic organization, and might cost as much in application prep time as simply shelling out the green ourselves would. We could always set up a fund, contribute to it, and then thank it profusely in the acknowledgements. From RJB@MAX.U.WASHINGTON.EDU Sun Jan 05 09:40:00 1992 Date: Sat, 4 Jan 92 16:40 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Publishing the Voynich To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I agree with Jacques Guy's assessment of the US market. I am going to check with a book designer I know for rough ballpark estimates on cost of priof printing a facsimile -- soft and hard. Will report back with some figures. --rjb From reeds@gauss.att.com Sun Jan 05 10:00:27 1992 Message-Id: <9201050100.AA09665@rand.org> From: reeds@gauss.att.com Date: Sat, 4 Jan 92 20:00:27 EST To: voynich@rand.org Status: OR About making a coffee-table book out of the VMS: My wife the publisher says that color plates cost about $1000 apiece, added to the production cost of a book. So if we wanted 200 plates, we would have (1) a banking problem, and (2), if we sold 1000 copies, they would cost roughly $200 apiece MORE than they would w/o the plates. I don't think Aegean Park Press is equipped to either (1) handle the technical work (they've done no color illustrations in any books that I know of), (2) market the book (their mailing lists have all the world's cryppies, but not the Hermetic crowd, or the art lovers), and (3) the financial resources to take a $250,000 risk. Somewhat more plausible presses might be Garland and Yale University Press (Yale U.P. has a long tradition of color plate work, learned coffee table books, etc.) From yanek@mthvax.cs.miami.edu Sun Jan 05 14:37:15 1992 From: Yanek Martinson Message-Id: <9201050537.AA21998@mthvax.cs.miami.edu> Subject: Re: Publishing the Voynich To: j.guy@trl.oz.au (Jacques Guy) Date: Sun, 5 Jan 92 0:37:15 EST Cc: voynich@rand.org In-Reply-To: <9201042336.AA04747@medici.trl.OZ.AU>; from "Jacques Guy" at Jan 5, 92 10:36 am X-Mailer: ELM [version 2.3 PL11] Status: OR > Enochian dictionary sell like hot cakes, I should think that a colour > reproduction of the VMS should sell a few thousand copies. > $100 US or so for one. But at this price, I don't think it would be a > commercial proposition. $50 might be, if the quality of the reproduction > is very good. A bit of PR may well help too. As would a decryption! :-) From thorin@wpi.WPI.EDU Sun Jan 05 15:57:58 1992 From: thorin@wpi.WPI.EDU (Richard John Yanco) Date: Sun, 5 Jan 92 01:57:58 EST Message-Id: <9201050657.AA05253@planet10> To: voynich@rand.org Subject: Computer recognition of ms text Status: OR As an aside, would it be worthwhile to explore computer recognition of the Voynich characters? If software has been produced that can transcribe handwritten English documents, it seems plausible that the programs could be retaught to transcribe the vms. I'm sure that a significant amount of setup time would be involved, but the potential time savings could be significant. More importantly, in its analysis the computer might be able to analyze more elusive details, such as spacing. If the programs (as I recall) use AI to determine what the "real" characters look like, it might also be interesting to look at a computer's best guess at the actual characters. Rick From rcarter@isis.cs.du.edu Sun Jan 05 16:58:05 1992 Date: Sun, 5 Jan 92 00:58:05 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9201050758.AA18449@isis.cs.du.edu> To: voynich@rand.org Subject: Name, printing & misc. Status: OR Yikes, didn't mean to cause any controversy; come to think of it, it -does- seem to come out that way most of time... Hee... Re: name: Something like "International Voynich Manuscript Research Group" (IVMRG) seems good to me; dunno if we want to say `Textual' because at least I want to attack the drawings. Re: printing/publishing: At this point, I would just as soon take the `quick and dirty' ie. get hard copy and/or microfilm in our grubby little hands and get the transcription done with; if the current negative available is of very good quality, great; if not, let's get a new negative done, and go from there. If Aegean can do a quality job and make it available cheaper/faster than we can do ourselves, great; I am very satisfied with the turnaround time/money for D'Imperio from them... Re: transcription: On this subject, are we getting close to what the final transcription alphabet will be? I don't care too much; I would like to see something that makes some mnemonic(?) sense ie. I don't like D'Imperio's; a cross between the 1st an 2nd study group, maybe? Or I could even cause more trouble and make a more substantial suggestion ie. make up one that makes sense to me; one of my strong suits is `visuals'... Re: misc: Anecdote time: Showed a (female) acquaintance V-plate in Kahn (1967, p.865) and she said "So, why a drawing of ovaries?" Further misc: I will clean up the typos (minor) I made in the Smith quotes and post to voynich@rand, as well as the Dee timeline based on Smith's bio; jbaez has (thanks!) volunteered to fill the gaps. Hopefully by next weekend. Regards to Team Voynich, Ron | rcarter@nyx.cs.du.edu From wet!naga@cca.ucsf.EDU Sun Jan 05 19:30:00 1992 Message-Id: Date: Sun, 5 Jan 92 02:30 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Misc. Status: OR rcarter writes: > ... my personal feelings are that Dr. John Dee was >quite a remarkable and visionary person; proposing an organized naval >system for England, proposing a national library, arguing for a 11 day >(which were omitted in 1742) calendar reformation, instead of the >prevalent thought of 1583 for a ten day change. While agreeing, there's just one nit I'd pick, though it has little relevance to the Vms. The lengths of the Gregorian and Julian calendar years differ by about 11 minutes, so over 170 years (1582-1752) come to differ by about 31 hours. This is the reason for the 11-day calendrical correction in Britain in 1752 (compared to the 10-day correction in 1582) rather than wisdom on Dee's part. I might also point out that to keep the calendar in synch with the seasons no omission of days at all is required; that was done to bring the spring equinox back to (usually) March 21, the probable date of the spring equinox at the time of the Council of Nicea. I read Crowley's book, Aegypt, and despite a somewhat verbose style I found it interesting, mainly for the Dee/Kelly parts. But I thought it ended abruptly and unsatisfactorily, as if the author had broken off without knowing how to finish it. To paraphrase (very roughly) Dr Samuel Johnson, I think he could have done a lot better, though one must give him credit for doing it at all. By the way, where's Levitov? Has anyone informed him that this discussion is taking place? He would presumably have some comment to make. From wet!naga@cca.ucsf.EDU Sun Jan 05 20:02:00 1992 Message-Id: Date: Sun, 5 Jan 92 03:02 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: P.S. to Misc. Status: OR naga writes in response to rcarter: > ... The lengths of the Gregorian and Julian calendar >years differ by about 11 minutes, so over 170 years (1582-1752) come to >differ by about 31 hours. This is the reason for the 11-day calendrical >correction in Britain in 1752 (compared to the 10-day correction in 1582) >rather than wisdom on Dee's part. I might also point out that to keep the >calendar in synch with the seasons no omission of days at all is required; >that was done to bring the spring equinox back to (usually) March 21, the >probable date of the spring equinox at the time of the Council of Nicea. After I posted this it occurred to me that Ron was probably saying that Dee argued for an 11-day calendrical correction in the 1580s, rather than the 10-day correction that was actually used in the Gregorian reform. As I noted above, the actual number of days can be anything you like, since the modified leap year rule is what ensures that the year does not shift (much) with respect to the seasons. The number of days omitted determines the (usual) date of the spring equinox (11 days rather than 10, omitted in 1582 would give us a spring equinox on, usually, March 22, rather than March 21), and Dee may well have had a different view on this than the advisors to Pope Gregory XIII. From RJB@MAX.U.WASHINGTON.EDU Sun Jan 05 23:39:00 1992 Date: Sun, 5 Jan 92 06:39 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: publication of facsimile To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR A color facsimile would be quite expensive. A good-quality black and white reproduction might be another story. And it might probe a market for a quality color facsimile. Somehow though I don't ever expect there to be enough of a market to justify a full facsimile. --rjb From j.guy@trl.OZ.AU Mon Jan 06 00:27:19 1992 Date: Sun, 5 Jan 92 10:27:19 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201042327.AA04737@medici.trl.OZ.AU> To: voynich@rand.org Subject: 2/R: a mnemonic Status: OR John Baez just wrote: Okay: Jacques Guy wrote The part of the R that rests on the line is c-like. From the thickness of the stroke, one can tell that it was written counter-clockwise, just like we write "c". The part of the 2 that rests on the line is Voynich i-like. It was written from top to bottom, leaning considerably left. The stroke is straight, and the same thickness its whole length. The top part of R and 2 was written second, from bottom to top, counter-clockwise, its bottom (start) linked to the top (start) of the first stroke. Another hint is in the letter that precedes R or 2: C *NEVER* precedes 2 A and I *NEVER* precede R (That's according to Currier, and I wasn't aware of that particular observation by him, when I wrote my Cryptologia article in which I mentioned a similar phenomenon). ------ The former part seems to be the exact opposite of how Currier draws 2's and R's in what I received from Jim G. Namely, the part of the R that rests on the line is I-like (straight vertical) while the part of the 2 resting on the line is C-like (an arc). He's right folks. This stupid Frogguy just mistook "R" for "2" and "2" for "R", brain addled and confused by Bennett's transcription in which "R" is "Q" and "2" is "Z", and my own silly stuff in which Bennett's "Q" is "2" and "Z" is "z". Oh, got it! Here's a mnemonic trick: Look at a capital R. Now erase the vertical line on the left: you end up with something that looks like the Voynich letter. Now for Bennett's system, look at a capital Q. Now erase the left part that looks like a capital C. Something like the Voynich letter again. From j.guy@trl.OZ.AU Mon Jan 06 00:36:58 1992 Date: Sun, 5 Jan 92 10:36:58 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201042336.AA04747@medici.trl.OZ.AU> To: voynich@rand.org Subject: Publishing the Voynich Status: OR Having read of those occult bookshops in the US where stuff like Laycock's Enochian dictionary sell like hot cakes, I should think that a colour reproduction of the VMS should sell a few thousand copies. Keeping it a soft-cover book should help bring the price down I suppose. Of course, I have no idea of the cost, but I have a feeling that quality colour reproduction has dropped drastically in price. I'd probably shell out $100 US or so for one. But at this price, I don't think it would be a commercial proposition. $50 might be, if the quality of the reproduction is very good. A bit of PR may well help too. From j.guy@trl.OZ.AU Mon Jan 06 00:52:03 1992 Date: Sun, 5 Jan 92 10:52:03 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201042352.AA04753@medici.trl.OZ.AU> To: voynich@rand.org Subject: Group names Status: OR How about "The Most Mysterious Manuscript", acronymized (just made this word up) to: TM3, and pronounced "Team three". Dop't worry: I'll come up with even *worse* suggestions, given time. From reeds@gauss.att.com Mon Jan 06 01:12:59 1992 Message-Id: <9201051613.AA15475@rand.org> From: reeds@gauss.att.com Date: Sun, 5 Jan 92 11:12:59 EST To: voynich@rand.org Status: OR My mail archives hold roughly 750,000 bytes of Voynich traffic since Thanksgiving. D'Imperio puts the length of the VMS at 250,000 characters; a number of us have guessed a shorter length. Thus we have chit chatted at least 3 times the VMS's length in less than 3 months. I hope we can keep it up when we transcribe! Some weeks ago Jacques said that he would draw up a more formal, precise statement of the transcribing rules we had seemed to agree on. Is that still in the works, Jacques? From Karl.Kluge@G.GP.CS.CMU.EDU Sun Jan 05 18:10:00 1992 Date: Sun, 5 Jan 1992 18:10-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: More name suggestions.... Message-Id: <694653012/kck@G.GP.CS.CMU.EDU> Status: OR How about VOYnich Network Investigating Cryptographic Hypotheses (or VOYNICH for short)? From wet!naga@cca.ucsf.EDU Mon Jan 06 10:22:00 1992 Message-Id: Date: Sun, 5 Jan 92 17:22 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Study group(s) Status: OR This is at least the third group convened to attempt a decipherment of the the Vms. As noted by Curt Zimansky in Brumbaugh's MMM (p.103), the first (within the last 50 years) was William Friedman's cryptographers at the end of WWII, who "reached some conclusions about the date and nature of the manuscript" but whose work was terminated by demobilization. Friedman convened another group in 1962 (this was called "The Voynich Manuscript Study Group"), which had a 301 [IBM?] computer at its disposal. Zimansky writes that "Time proved shorter than expected; the computer was reclaimed for more mundane tasks and once again a team of experts broke up without having produced a valid solution." I wonder if this group published any results. As regards a name for the group, how about: Alliance for Probable Hoax Research and Decipherment This produces the acronym, AlPHReD, pronounced "Alfred", in honor of the 20th C. discoverer of the manuscript. From j.guy@trl.OZ.AU Mon Jan 06 23:03:45 1992 Date: Mon, 6 Jan 92 09:03:45 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201052203.AA05389@medici.trl.OZ.AU> To: voynich@rand.org Subject: Transcription problems Status: OR Dear Voynichsforschungssturmarbeitern (yeah, I know, my German is rotten, so we can't use this VFSA acronym, can we?) The realization that I had confused Currier's "R" and "2" had me reflect on problems of transcription again. Currier's 2 vs R (Bennett's Z vs Q, my z vs 2) ---------------------------------------------- As I wrote much earlier, I devised my own personal system so that it would be easy to learn, use and remember. Well, it obviously isn't, for if the digit "2" does resemble the Voynich letter denoted by Currier's "R" and Bennett's "Q", our letter "z", with I use for Currier's "2" and Bennett's "Z", does not resemble the corresponding Voynich letter. Peering desultorily at my keyboard, I spy with my little eye only "$" and "S" which look like mirror-images of it. Under the principle that a good system should be good for two-finger typists and require minimum use of the "shift" key, neither look too good. I don't know either how easy to type a dollar-sign when you've configured your system for a British, or a French, or a Danish, or whatever non-US keyboard. So "$" gets my thumbs down. Remain capital "S" and lowercase "s", and the latter has my preference, because I'm too lazy to hit that shift key so often. I won't post yet another font for this good reason: 1. if you can use those fonts of mine it means you have a PC, 2. if you have a PC, you have a copy (or ought to get one from pub/jim at rand.org) of Harald Thunem's font editor and of my TRANSLIT program with the translation files that go with it. 3. Using Harald's font editor is a breeze, and modifying those translation files to your heart's contents is no sweat either. 4. So do it! (Or complain loud enough for me to do it. You no complain, me no do). Keep it Simple, Stupid! ----------------------- Easier said than done. I always find that after writing may thousand lines of Pascal, trying hard all along to keep it simple, when it's all done, there remains one task: you've tried to keep it simple? Now MAKE it simple. You know what hit me in my stupid face? That business of curly brackets for comments. There's no need for them. A line of transcription always starts with a folio, line, and handwriting identifier, e.g. <0101A>. Therefore, whatever does not start with such an identifier is a comment. This convention would have another advantage: it would prevent you from interspreading text with comments (which really makes your work hard to read). So, instead of: <3v.1> FOAM.W[OA]R.4OPO[A9].ZA.XOE.9FOAM,[82],OE[A9] <3v.2> 8AII8[9A].4OPCCOE.OFCCOR.OFOR.OE9POE.8OE,8AR <3v.3> OFAJ.SOE.ZOE.2CCC2.SOJ.SCC9FAJ.OF3 {last few chars unclear} <3v.4> 408AR.S2.CC9.FSCOE.OFAE.8OR.SCARCCD <3v.5> 9CCCAR.O4SAE.SAR.SAR.XA {first word start unclear} We have: <3v.1> FOAM.WOR.4OPOA.ZA.XOE.9FOAM,8,OEA <3v.2> 8AII89.4OPCCOE.OFCCOR.OFOR.OE9POE.8OE,8AR <3v.3> OFAJ.SOE.ZOE.2CCC2.SOJ.SCC9FAJ.OF3 ^^^unclear <3v.4> 408AR.S2.CC9.FSCOE.OFAE.8OR.SCARCCD <3v.5> 9CCCAR.O4SAE.SAR.SAR.XA ^^^^unclear Next, having gotten myself in a murderous simplifying mood, I thought: And what about those square brackets that mark alternate readings? Off with their heads! Put alternate readings under each other. (Well, I am the first to admit that I might be a bit, quite a bit, radical here). So, instead of: <3v.1> FOAM.W[OA]R.4OPO[A9].ZA.XOE.9FOAM,[82],OE[A9] <3v.2> 8AII8[9A].4OPCCOE.OFCCOR.OFOR.OE9POE.8OE,8AR <3v.3> OFAJ.SOE.ZOE.2CCC2.SOJ.SCC9FAJ.OF3 {last few chars unclear} <3v.4> 408AR.S2.CC9.FSCOE.OFAE.8OR.SCARCCD <3v.5> 9CCCAR.O4SAE.SAR.SAR.XA {first word start unclear} We have something like this: <3v.1> FOAM.WOR.4OPOA.ZA.XOE.9FOAM,8,OEA A 9 2 9 <3v.2> 8AII89.4OPCCOE.OFCCOR.OFOR.OE9POE.8OE,8AR A <3v.3> OFAJ.SOE.ZOE.2CCC2.SOJ.SCC9FAJ.OF3 ^^^unclear <3v.4> 408AR.S2.CC9.FSCOE.OFAE.8OR.SCARCCD <3v.5> 9CCCAR.O4SAE.SAR.SAR.XA ^^^^unclear Or even: <3v.1> FOAM.WOR.4OPOA.ZA.XOE.9FOAM,8,OEA A 9 2 9 <3v.2> 8AII89.4OPCCOE.OFCCOR.OFOR.OE9POE.8OE,8AR A <3v.3> OFAJ.SOE.ZOE.2CCC2.SOJ.SCC9FAJ.OF3 ^^^ unclear <3v.4> 408AR.S2.CC9.FSCOE.OFAE.8OR.SCARCCD <3v.5> 9CCCAR.O4SAE.SAR.SAR.XA ^^^^ unclear To make things easier for program-writers (of which I'm one), once could put perhaps and equal sign (=) as the first character of a line containing alternative transcriptions. Now for a Bit of Silliness -------------------------- How about GRV (pronounced, roughly, Jerrayvy): Grupo Ricerca Voynich? Or even better: GruV (pronounced of course "Groovy"): Grupo Voynich? Or Boken (Boinicchi kenkyuu). From j.guy@trl.OZ.AU Mon Jan 06 23:05:49 1992 Date: Mon, 6 Jan 92 09:05:49 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201052205.AA05392@medici.trl.OZ.AU> To: voynich@rand.org Subject: Pronounceable Voynich Status: OR Hark, O Mortals, the Divine Sound of the Most Mysterious Language in the World: praitusox complam as aclavii istem istesus iclotesusox cum ples istendus oxesus iplotas e s cum plavi istclavii istas as* iclotas iclotas dai oxusaor istiplus es complavii istedo icloteasus iclot*s dasavii oxus revii ecluus eclies oxemeclus iclot*as davii eplavii em eplai oxais cum itias iclotavii icrotas iprotavii condasaiiste * edas *us istem icroteus eusdas ist* ox iproteavii istedaxus *istus itedus epliteus eclitem iteiclotus ex itus davi istes plex davii istex iprotem istedus davi *ecliedus * condavi icrotixavii em ox icrotius conclavi isteistus icrotedam ix epliste plisteus eclaiui ecliem eplai istedavi oxiplotus davii isteus iplotius pledavii icrotus icrotedaumox iclotius iste endavi do davi evii item edavii itedavi itdus eplavi d*i iclotus pledo davii istiplotius iplotes ites istius plem item item ples item iste item ist edai plistus plitus des itedavii iste pluatur comite clitius itiplavi istiecristem diordusdo iclotus daiclotus And what is this Noble Language called? Why, Voicenich of course. How did I do it? I took my skrying glass, and called upon the Angels. Neboniel appeared and spake unto me, saying: "Prepend thou a dot to indicate the start of each and every line, where only a space shews now, and take thou under my dictation this translation file which thou shalt use with TRANSLIT. Call thou the output file VOYNICH.VOI, and edit it forthwith, replacing each and every dot with a space." All of you Dee fans and Kelley diggers, can you figure out what Neboniel dictated to me? (The Angel also suggested an improvement to TRANSLIT) From j.guy@trl.OZ.AU Mon Jan 06 23:09:09 1992 Date: Mon, 6 Jan 92 09:09:09 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201052209.AA05402@medici.trl.OZ.AU> To: voynich@rand.org Subject: Silly Season Again Status: OR STOP PRESS STOP PRESS STOP PRESS STOP PRESS STOP PRESS STOP As I was thinking up new, improved, more outrageous names for the Voynich Decipherment Group, the Origin of the Voynich Manuscript was Revealed unto me. We all agree (excluding those who do not) that the VMS is Dee's "Book of SOYGA", that it's got funny pictures of naked women cavorting amidst funny plumbing, and that angels had something to do with with. Well? Well? Well? (* sigh *) Do I have to do all the bull-throwing around here? Oh well (* sigh *) "Barbarella" by Roger Vadim, with Jane Fonda, after the comic strip of the same name by Jean-Claude Forest. Go and rent it from your video shop. What do you see? In the Black Queen's city, a city called SOGO (close enough to SOYGA, isn't it?) SCANTILY-CLAD FEMALES smoking essence of man from a huge WATER-PIPE through yards and yards of PLUMBING. What else is there? An ANGEL, folks, an ANGEL by the name of Pygar. What more? By careful analysis of the "plumbing" folios, using a combination of Newbold's, Brumbaugh's, Levitov's and Feely's decipherment methods (4x4= 16 times as powerful, that's synergy for you), I have proved to my satisfaction that the Manuscript was written by Roger Bacon, William Shakespeare, and Jane Fonda. And no wonder, for they are one and the same person, known under other climes as Comte de Saint-Germain. (I also found Nostradamus and Madonna to be the same person, as their names clearly show). From jbaez@math.mit.edu Tue Jan 07 01:29:01 1992 Date: Mon, 6 Jan 92 11:29:01 EST From: jbaez@math.mit.edu Message-Id: <9201061629.AA11763@riesz> To: voynich@rand.org Subject: watch your 2's and R's Status: OR Well, that just serves Jacques right for teasing me about mixing up my letters. :-} As it stands, then, it seems we agree about the distinction between 2's and R's in their ideal, or paradigmatic, forms. I am willing to try to distinguish them in my transcriptions, with the caveat that there will be many difficult borderline cases. From jbaez@math.mit.edu Tue Jan 07 03:24:08 1992 Date: Mon, 6 Jan 92 13:24:08 EST From: jbaez@math.mit.edu Message-Id: <9201061824.AA12336@riesz> To: voynich@rand.org Status: OR Well, after plowing through piles of email, mostly about ridiculous names for the Voynich Manuscript Study Group :-), I am now almost too exhausted to post myself, but... 1) I got ahold of Babcock at Yale; he says that a photocopy would cost a mere $40; he's sending me an invoice but sending him the check now would "speed things up." I put this in quotes because he estimates that it'll take 6 to 8 weeks to send me the photocopy! Ugh. To see the manuscript, he said I should write and propose a date - as I didn't know when exactly I wanted to come, I said "okay" - but I'll probably call him when I figure out when's a good time. He seemed a rather sluggish and bored sort. So: when I get the photocopy I will send second-generation copies (or perhaps they'll be 3rd-generation, if he makes the photocopy from film) for a fairly low cost, basically to cover mailing, perhaps plus $5 to help divide the $40 among us. Those who are true Voynich-maniacs, on the other hand, may want to get their own copy, since it's so cheap. (Directions for contacting Babcock are in the file pub/jim/biblio at rand.org, available through ftp.) I propose that we spend the intervening time as follows: 1) fine-tuning and documenting our transcription scheme, 2) investigating Dee and, more importantly (I believe) the court of Emperor Rudolph II, 3) forming connections with professional scholars of medieval and Renaissance manuscripts, 4) improving our skills in infrared photography, radiocarbon dating etc. (this is Nate Osgood's department), 5) developing software for the statistical analysis of texts. ----------- Rudolphine Culture It's only an uncertain hypothesis that Dee and Kelley are involved with the Voynich, but rather certain that Emperor Rudolph II bought it (barring the possibility that the letter accompanying the manuscript is a modern fake, or that Marci had his facts wrong in that letter). Moreover, Rudolph II was surrounded by many strange people who could have sold him the manuscript, of whom Dee and Kelley are but two. Thus we should track down some more on Rudolph. Luckily this is interesting in itself and (at least for starters) not too hard. I stumbled on a bunch of information on this topic in a bookstore yesterday, where I saw Frances Yates' book "The Rosicrucian Enlightenment," 1972, reprinted Routledge & Kegan Paul ("Ark Paperbacks"), 1986. First let me sum up the book briefly, and then excerpt some relevant bits. The book focuses on the court of the "Winter King and Queen of Bohemia." Emperor Rudolph II died in 1612. In 1617 Ferdinand of Styria became King of Bohemia (its capital in Prague, where Rudolph had moved his court from Vienna) and put an end to Rudolph's policies of religious toleration. In 1619 the Bohemians offered the crown instead to Frederick V, Elector Palatine of the Rhine, who had married Princess Elizabeth of Britain in 1613, and who was opposed to Catholicism and quite liberal. For a brief time Bohemia again became a hotbed of occultism and (perhaps - this is Yates' thesis) a "Rosicrucian Enlightenment," based in large part on the theories of John Dee. But in 1620 Frederick was dealt a crushing military defeat and fled Bohemia and his Palatinate. The libaries of the Palatinate were sacked, mass executions eliminated all resistance, and shortly the 30 Years War began, which ruined Bohemia. Yates' book is about the brief period of flowering before the disaster. I'm more interested in the reign of Rudolph II, so I'll excerpt some intriguing remarks about it: p. xii: Dee's striking and very influential career in Elizabethan england came to an end in 1583 when he left England for the continent, where he was extremely influential in stirring up new movements in central Europe. This half of Dee's career, the second or continental half, has not yet been studied in a systematic way and still remains in the world of rumour. [Great! :-)] It would seem that Dee was the leader in Bohemia, not only of an alchemical movement, but of a movement for religious reform, the nature of which has not yet been fully explored. Our knowledge of the world of culture surrounding the Emperor Rudolph II, upon which Dee's mission impinged, is still extremely scanty, and we await the publication of Robert Evans's important study of Rudolphine culture. [Let's get this!] p. 16: Though a member of the House of Hapsburg, Rudolph had held aloof from his nephew Philop II of Spain and had mysteriously buried himself in abstruse studies. He moved the imperial court from Vienna to Prague, which became a centre for alcehmical, astrological, magico-scientific studies of all kinds. Hiding himself in his great palace at Prague, with its libraries, its "wonder rooms" of magico-mechanical marvels, Rudolph withdrew in alarm from the problems raised by the fanatical intolerance of his frightening nephew. Prague became a Mecca for those interested in esoteric and scientific studeis from all over Europe. Hither came John Dee and Edward Kelley, Giordano Bruno and Johannes Kepler. However strange the reputation of Prague in the time of Rudolph it was yet a relatively tolerant city. Jews might pursue their cabalistic studies undisturbed (Rudolph's favourite religious adviser was Pistorius, a Cabalist) and the native church of Bohemia was tolerated by an official "Letter of Majesty." The Bohemian church, founded by John Huss, was the first of the reformed churches of Europe. Rudolph's toleration was extended to the Bohemian church and to the Bohemian Brethren, a mystical brotherhood attached to its teachings. [We should check them out & see if they had a fondness for naked women and plumbing, e.g.] p. 27: The alchemical and esoteric interests encouraged by Rudolph II had represented a more liberal, Renaissance atmosphere than that which the Reaction wished to impose, and such studies were popular at German courts, particularly those of Hesse and Wuerttemberg. And the traditions of Rudolphine Prague were certainly familiar to Christian of Anhalt, the leader of Palatinate policy. p. 37: [Christian of Anhalt] was the patron of Oswald Croll, Cabalist, Paracelcist, and alchemist. And his Bohemian connections were of a similar character. He was a clese friend of Peter Wok of Rosneberg, or Rozmberk, a wealthy Bohemian noble with vast estates around Trebona in southern Bohemia, a liberal of the old Rudolphine school, and a patron of alchemy and the occult. Anhalt's Bohemian contacts were of a a kind to bring him withing the sphere of a very remarkable current of influences from England which arose out of the visit to Bohemia of John Dee, and his associate, Edward Kelley. As is well known, Dee amd Kelley were in Prague in 1583, when Dee tried to interest the Emperor Rudolph II in his far-reaching imperialist mysticism and his vast range of studies. The nature of Dee's work is now better known through the recent book by Peter French. Dee, whose influence in England had been so profoundly important, who had been the teacher of Philip Sidney and his friends, had had the opportunity of forming a following in Bohemia, though we have, as yet, little means for studying this. The main centre for the Dee influences in Bohemia would have been Trebona, which he and Kelley had had made their headquarters after the first visit to Prague. Dee lived at Trebona, as the guest of Villem Rozmberk, until 1589, when he returned to England. ... According to the notes about Dee by Elias Ashmole in his Theatrum Chemicum Britannicum (1652), Dee's journey through Germany in 1589, on his return from Bohemia to England, was somewhat sensational. He passed near those territories which, twenty-five years later, were to be the scene of the outbreak of the Rosicrucian movement. The Landgrave of Hesse presented his compliments to Dee, who in return, "presented him with Twelve Hungarian Horses, that he bought at Prague for his journey." Dee also made contact at this stage of his journey home with his disciple, Edward Dyer... ... Ashmole states that on 27 June 1589, when at Bremen, Dee was visited by "that famous Hermitique philopher, Dr Henricus Khunrath of Hamburgh." The influence of Dee is in fact apparent in Khunrath's extraordinary work, "The Amphitheatre of Eternal Wisdom" published at Hanover in 1609. Dee's "monas" symbol, the complex sign which he expounded in his Monas Hieroglyphica (published in 1564 with a dedication to the Emperor Maximilian II) as expressive of his peculaiar form of alchemical philosophy, can be seen in one of the illustrations in the "Amphitheatre." p. 80: Michael Maier was born at Rindsberg in Holstein in 1566. He graduated as a doctor of medicine and lived at Rostock, and then at Prague, where he was physician to the Emperor Rudolph II, as already mentioned. Some time after the death of Rudolph, in 1612, Maier visited England were he was almost certainly in contact with Robert Fludd, though exactly when or under what circumstances is not known. His first publication, "Arcana Arcanissima" (1614), was dedicated to the English physician Sir Willaim Paddy, who was a friend of Fludd's. ----------- In short, there's a whole stew of possible suspects here. Which one was X? From @ada3.ca.boeing.com:crispen@amber.boeing.com Tue Jan 07 04:05:18 1992 Return-Path: crispen <@ada3.ca.boeing.com:crispen@amber> Date: Mon, 6 Jan 92 13:05:18 CST From: crispen <@ada3.ca.boeing.com:crispen@amber.boeing.com> Subject: Unsubscribe To: voynich@rand.org Message-Id: <9201061905.AA00417@amber.boeing.com> X-Envelope-To: voynich@rand.org Status: OR Please unsubscribe me from the Voynich mailing list. Thank you, Bob Crispen crispen@foxy.boeing.com From JBAEZ@LUCY.WELLESLEY.EDU Tue Jan 07 04:58:00 1992 Date: Mon, 6 Jan 1992 14:58 EST From: JBAEZ@LUCY.WELLESLEY.EDU Subject: Warburg Institute To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: NET::"voynich@rand.org" Status: OR I'm getting curious about the Warburg Institute and the Journal of the Warburg and Courtauld Institutes. Yates was an honorary member of this institute, and Brumbaugh published something there... is it an enclave of esoteric studies or what? :-) I ran into this reference which might be worth running down: C H Josten, "An unknown chapter in the life of John Dee," Journal of the Warburg and Courtauld Institutes XXVIII (1965), p. 235. From jim@rand.org Tue Jan 07 06:24:49 1992 Message-Id: <9201062124.AA06393@mycroft.rand.org> To: voynich@rand.org Subject: Re: Name of this group? In-Reply-To: Your message of Sat, 04 Jan 92 16:57:02 -0500. <9201042157.AA13559@nevanlinna> From: Jim Gillogly Reply-To: jim@rand.org Date: Mon, 06 Jan 92 13:24:49 PST Sender: jim%mycroft@rand.org Status: OR Gad -- a lot of action on the list over Christmas vacation... I've got a lot of catching up to do! Regarding the name of this group -- yeah, I guess there are times when it would be useful. My favorite so far is "International Voynich Ms Working Group"... but for the souvenir windbreakers, Team Voynich will look good on the back among the pictures of pudgy nymphs. Jim Gillogly From RJB@MAX.U.WASHINGTON.EDU Tue Jan 07 07:07:00 1992 Date: Mon, 6 Jan 92 14:07 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Warburg Institute To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Well, in a sense, yes, it does seem to have been an enclave of esoteric studies. And maybe still is. Dame Frances Yates was not only an honorary member, she was for many years the editor of the _Journal_ -- years before she began publishing the books for which she was made a Dame (forget the actual honor: would it have been OBE? or something even higher?). Her first book came out when I was an undergraduate, and was instrumental in beginning to convince me that Henry Ford had been quite wrong about history. The Warburg has also published D P Walker (_Spiritual and Demonic Magic from Ficino to Campanella, and an interesting set of studies on esoteric theories of music), various editions of the _Picatrix_, and so on. Yorke left his very large Crowley collection to the Warburg, too, and when I was first there in 1972 or so it took the librarians a while to realize that I wasn't there to look at the Crowley stuff (except the _Tuba Veneris_). The fact that the first _Picatrix_ publication by the Warburg is a German translation is simply due to the fact that the Warburg was founded by art historian Aby Warburg, and was evacuated to England for the obvious reasons after Hitler & co began their thousand year project. Unfortunately not everything survived, including the plates (or am I misremembering? I write this away from my books) of the Arabic edition of the _Picatrix_, which was to have been volume I of the set. Art History-->iconography-->emblem studies (e.g., the famous book on Durer's Melancholia)-->talismans -->esotericism in general. The project that Yates put the finishing touches on was basically to demosntrate that European intellectual (and even scientific) history cannot be done without a well-informed understanding of European esotericism. In 1965+ in many circles this was revolutionary indeed. In fact, the New York Review of Books took a breather for a while from being the New York Review of Vietnam to become the New York Review of Frances Yates. (Her speculative reconstruction of Shakespeare's Globe first appeared in the NYRB.) In addition to her _Giordano Bruno,_ Yates' _Art of Memory_ was a major breakthrough in making certain aspects of esotericism clear and intelligible to just folks. There are sizable chunks of speculation in all her works, more in the later than in the earlier -- but her speculative reconstructions have often enough held up fairly well under later criticism. From wet!naga@cca.ucsf.EDU Tue Jan 07 11:32:00 1992 Message-Id: Date: Mon, 6 Jan 92 18:32 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Nymphs Status: OR Jim Gillogly writes: > ... but for the souvenir windbreakers, Team Voynich will >look good on the back among the pictures of pudgy nymphs. Not all of the nymphs are pudgy. In fact the nymph at top center of folio 82v (p.104 in Bambraugh) strikes me as really cute. From WBRILL@macc.wisc.edu Tue Jan 07 22:23:00 1992 Message-Id: <22010707233070@vms.macc.wisc.edu> Date: Tue, 7 Jan 92 07:23 CST From: "Winston J. Brill" Subject: Group name To: VOYNICH@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR How about "Voynich Voyeur's"? From firth@SEI.CMU.EDU Tue Jan 07 23:43:51 1992 Date: Tue, 7 Jan 92 09:43:51 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201071443.AA16024@bp.sei.cmu.edu> To: voynich@rand.org Subject: Publishing the Voynich Status: OR Notes on the Voynich Manuscript - Part 6 ---------------------------------------- Publishing the Manuscript. As a possible publisher, you might want to consider Phanes Press PO Box 6114 Grand Rapids Michigan 49516 USA Their number is +1 800 678 0392 but I think that's just for orders or catalogue requests. Anyway, their 1991-1992 catalogue has a lot of Renaissance and modern weirdness, including things like The Chemical Wedding of Christian Rosenkreutz (the last and strangest of the Rosicrucian manifestos), Boehm's "Clavis", the "Splendor Solis", "Atalanta Fugiens", and, of course, the Harley MS on angel magic. I have no idea whether a published Voynich would sell, but gee, if they can sell (as they do) a complete facsimile of St-Germain's "Trinosophia", then they can probably sell anything! Of course, a decipherment would help, as somebody else pointed out. Colour reproduction seems very expensive, but I think at least some of the botanical and herbal folios should be reproduced in colour, not least in the hope the plants can be identified. And yes, I definitely believe we should publish a commentary. Both an historical introduction (maybe the Beinecke would like that piece of the action) and an edited account of our work. For this reason: one of my major frustrations with this endeavour is how much seems to have been done before but is now lost. We have papers by Currier that are "unpublished", machine transcripts that are lost; proposed decipherments missing all detail; claims that a botanist (who?) has identified many of the plants (which? as what? on what basis?), and so on. We are repeating history. Every previous study group seems to have devised a transcription scheme, argued over what is a letter, entered chunks of text, done letter and word counts, researched the history of John Dee, and so on. And I fear that, in another fifteen or twenty years, somebody is going to do it all over again. Even if we don't publish, I think a fair copy of our musings should be run off, bound, and presented to the Beinecke for posterity. Finally, we should append a bibliography. D'Imperio's is a very good start, but a lot of stuff has been reprinted since 1978, and we could update and expand it, with her permission. Robert From reeds@gauss.att.com Wed Jan 08 01:10:25 1992 Message-Id: <9201071610.AA23259@rand.org> From: reeds@gauss.att.com Date: Tue, 7 Jan 92 11:10:25 EST To: voynich@rand.org Warning: Gloomy Hemulen stuff below Status: OR Robert Firth writes >>> We are repeating history. Every previous study group seems ... Yes, and we should be sure to publish (and thus record for posterity) that which is valuable. Alas, so far, just about nothing we've done IS valuable. (Yet.) Here is my personal list of what would be valuable (in roughly increasing order of value): 0. Updated VMS bibliography 1. Publication of some sort of facsimile of the MS, allowing experts in other fields easy access to the VMS's peculiar icons, layout, etc. 2. A complete transcription 3. A complete transcription which has been somehow checked 4. Finding new statistical phonomena in the text. 5. Conclusive statistical analysis proving either that a solution exists or that a solution does not exist 6. A solution I think this group is well placed to actually DO something about these. Items 2-6 are made easier for us than for Manly, Friedman, Tiltman, Currier, etc, because we have computers and email. We can work together even though we are geographically dispersed, even though (as I suppose) we are not especially brainier than those guys. And we can draw on more CPU cycles, if needed. On the other hand we are (on the whole) beginners at 16th century history, Hermeticism, and the like. Our efforts in Dee-ology or Hermetic studies are not worth publishing: they represent the results of amateurs' (that's us) scouring of recent secondary literature. They are what I suppose undergraduate history major's seminar papers or senior theses must be like: workmanlike, educational, but not genuine additions to mankind's common pool of knowledge. Even less valuable are our conjectures. Dee possibly, Kelley yes, Kircher no, Rudolph maybe, etc. Artifical language, music, knitting instructions (!), dance steps (that's a new one). Easy come, easy go. Unless we actually learn some facts from the MS, we should let future generations invent their own plausible hypotheses. Your faithful Nattering Nabob of Negativism, Jim Reeds From RJB@MAX.U.WASHINGTON.EDU Wed Jan 08 01:34:00 1992 Date: Tue, 7 Jan 92 08:34 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: publishing the Voynich (Firth note #6) To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I think Phanes might be a good candidate too they certainly have the distribution to tap into the market for such a thing. I spoke with a calligrapher/book designer/typographer whose firm also does print brokering, and got some very ballpark estimates. An adequate black-and-white facsimile (a "scholar's facsimile"), a 4-color facsimile, a 6-color (Tres Riches Heures) facismile, and a bang-up whangaroo artist's facsimile, with necessary selling-prices for a run of 2,000 rnging from about $16 through $40, $80 and over $100. Having never seen the VMs or a color facsimile of any of its pages, I have no idea what level of color work would be necessary to capture it. How many colors are used in it, anyway? Of course, the higher the level of color work, the more outside (non-Beinecke technicians, too) involvement would be necessary, and the more negotiation with the Beinecke. I think Firth's idea of a b&w with some key pages in color might be a good model to follow, leaving open the possibility of a very high-end art product if the quality of the MS and the nature of the market later seemed to warrant it. APP (for example) could do the edition, and Phanes could distribute it. In fact, if Phanes liked the Voynich facsimile and commentary, APP might also find them interested in distributing their current two books on the Voynich. (Phanes distributes things by several other publishers.) Unfortunately I left the numbers on a slip of paper at home, but I think that in a very general, very ballpark way those approximate prices above give an idea. ---rjb From firth@SEI.CMU.EDU Wed Jan 08 03:05:16 1992 Date: Tue, 7 Jan 92 13:05:16 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201071805.AA16287@bp.sei.cmu.edu> To: voynich@rand.org Subject: More idle speculation Status: OR Notes on the Voynich Manuscript - Part 7 ---------------------------------------- Speculative notes on the Subject of the MS What is the Voynich manuscript about? We haven't a clue, really, but that need not stop us from speculating. I'd like to mumble on about three things: whether the text describes the illustrations, what the illustrations suggest, and what they do not suggest. The copies I have seen (almost all in Brumbaugh) strongly suggest that the illustrations were done first, then the text. Look, for instance, at f93r (p80), how the text is fitted around the plant. Look also at the unevenness of the left margin in f82r (p104). It's much harder to decide whether the text is about the drawings. But one thing is obvious: Brumbaugh's contention (p140) that the author repeated gibberish to fill out paragraphs is not supported by the reproductions. There is ample space left on the page in both f49r and f93r, as there is also in f85r (p54) and f100v (p96). And look at f81r (p9). Why is the right margin so big? Maybe the reason is that the copyist was deliberately stretching a fixed text to fill the space, making the lines shorter so there would be more of them. And he miscalculated slightly, for the last line runs over by at least two words. Look also at the frontispiece, the reproduction of pisces. In the outer ring are nineteen chubby tubbies, and the writer clearly went to a lot of trouble to squeeze the last of them in - her name is obviously one word broken across two lines. I'm not good at reading the script yet, but I checked the text of two sections - f100r (p96) herbal and f82v (p104) "allegorical". In the former folio, of 19 words attached to figures, I found three repeated exactly in the text, and four more that differed only in the ending of the word. In the latter, I again found four words from the drawings repeated in the text. By way of comparison, I checked a couple of chapters of the Book of Coming Forth By Day against their attached drawings. In one, of eight words in the drawing, five were found in the hieroglyphic text, two of them spelled slightly differently. In the other, of ten words, six were repeated, one with a different determinative. That's close, don't you think? Well, what are the drawings about? I make these conjectures: Plants: this seems so typical of an herbarium or pharmacopoeia that I can't think of any alternative. Zodiac: D'Imperio gives the count of figures, and every month has 30 (except poor pisces, with 29). That strongly suggests these figures have a calendrical basis, so if they are astrological they aren't about ephemerides or horoscope casting; they're more likely to be "Rudolph's Lucky Days for 1607". Incidentally, there are a lot of similarities between the labels on the piscean nymphs and the labels on the drawings of f82v. Otherwise, I might conjecture that the names are star names, and that most of them begin with "oqp-" because most star names begin with "al-". Astronomical: here also the two examples I have seem calendric. The diagram on f68r is a circle divided into eight segments, four with single stars and four with groups of stars. The text starts in the middle of the NW segment, which has the largest star. I think this stands for "spring", the eight segments are for the quarter and cross-quarter days, and the diagram is oriented so you rotate it counter-clockwise to read the text. Incidentally, if the circumscribed text starts at the vernal equinox, then the top is approximately May Day and the bottom approximately All Saints Day; the connotations with "witchcraft" are clear. The diagram on f67r (p66) has a similar structure, but with two strange differences: it is divided into 24, and the text in the segments is the other way up: the ascenders point clockwise around the circle. I think it has something to do with the 12 months. Nymphs and Plumbing: I can't offer a better suggestion than "anatomical", but some instinct tells me that's wrong. Miscellaneous: all I have left is f85r (p54), labelled "biological drawing". But the four strange cell-like structures are decorated with, respectively, a bird in flight, a bird brooding beneath a tree, a man, and a woman. Is this part of a creation allegory based on the Book of Genesis, with the bird representing Ruach Elohim, the Breath of God? If so, perhaps the structures are fountains, and the streams are not the four winds, but the four rivers of Paradise. Wild conjecture. Finally, the famous invalid argumentum a silentio: what can we deduce from what isn't in the MS? Alchemy: almost every Western alchemical text is littered with pictures, mandalas, figurae, emblemata &c. If any part of the MS is about alchemy, where is the pelican? the lion couchant? the caput mortuum? - and about a zillion more that you can see for yourself in the Mutus liber. Ceremonial Magic: again, every treatise, from the Picatrix already mentioned through the Clavicula solomonis and the Grimoirum papae honorii, has pentacles, pseudo-hebrew spells, "lines, circles, scenes, squares" and the like. As far as I know, there are none in the Voynich folios. Music: If the text is music, why are there no drawings of musical instruments or singers? Why don't the nymphs have trumpets or lyres? (Actually, I think some parts are a treatise on millinery: "One Hundred Fashionable Hats: How to make them, and Where to Wear them. Guaranteed Waterproof.") Robert From rcarter@isis.cs.du.edu Wed Jan 08 04:27:39 1992 Date: Tue, 7 Jan 92 12:27:39 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9201071927.AA26452@isis.cs.du.edu> To: voynich@rand.org Subject: Can someone provide definitive ordering info on Voynich MS copy? Status: OR Maybe definitive info on ordering a copy of the Voynich MS. could be posted to the group? IE. Exactly how much (was it $40US?) Who to make the money order out to (Yale? Beincke?) The address (c/o Robert Babcock at the Yale Station address?) Etc. Thanks, Ron. PS I will go for Int'l Voynich MS Working Group as the group name as well. From j.guy@trl.OZ.AU Wed Jan 08 07:13:09 1992 Date: Tue, 7 Jan 92 17:13:09 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201070613.AA06888@medici.trl.OZ.AU> To: voynich@rand.org Subject: Transcription still and again Status: OR Since John Baez has whetted my appetite for affordable photocopies of the Voynich, I'll rest my transcribing case for a while until I have seen, at long last, the whole of the VMS. But before I do, something that occurred to me when looking at the color reproduction in Blunt and Raphael's. Consider Currier's "Z" (my c't) and "2" (my z or s) in that folio; how similar "2" is to the left part of "Z". Could it be that we distinguish that left half of "Z" from "2" on the (flimsy) grounds that a following "C"-like letter is linked to it? That they are REALLY one and the same? If so, what would Sukhotin's algorithm now give as vowels? Indeed it seems to me that, once the VMS transcribed, a first task will be to identify its letters. An interesting problem, for what do we consider to be a letter? We say "th" is two letters, but Spaniards consider "ch", "ll" and "rr" as single letters, at least for the purpose of ordering words alphabetically (and these represent in fact single phonemes, so they ARE really single letters). What if Voynich "89" turned out to be a single letter (phoneme)? I wouldn't be too surprised, either. From RJB@MAX.U.WASHINGTON.EDU Wed Jan 08 07:36:00 1992 Date: Tue, 7 Jan 92 14:36 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: names and games To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Yes, I think International Voynich Manuscript Working Group is about ass succinct and sober as one can get; sounds good to me. Team Voynich for the mugs & t-shirts sounds great, too. I am sharpening up my quill and curing the vellum at this very moment; when I get the ordering info (I'll look at pub/jim for it later today) I will despatch an order. It is not quite clear to me whether the photocopy is photocopy on film or on paper, but I'll find out soon enough, I'm sure. It occurs to me that a flurry of international orders for photocopies of the Voynich MS might have an effect on Yale's sense of the seriousness of the interest -- one way or the other! --rjb From reeds@gauss.att.com Wed Jan 08 12:56:22 1992 Message-Id: <9201080356.AA09571@rand.org> From: reeds@gauss.att.com Date: Tue, 7 Jan 92 22:56:22 EST To: voynich@rand.org Status: OR 1. I hope I did not offend anyone with my recent post about the ultimate value of our work. If so, I appologize. I think our group has been remarkably hard working, enthusiastic, and high spirited, and I certainly didn't mean to spoil that, but am afraid I might have. 2. I just reread all of Robert Firth's trenchant series of notes, and wonder what his background is? Care to submit a brief self blurb about yourself, Robert? Jim Reeds From RJB@MAX.U.WASHINGTON.EDU Wed Jan 08 13:40:00 1992 Date: Tue, 7 Jan 92 20:40 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: self-publishing the Voynich MS To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR This has actually been lurking in the background of my discussion, but in a very minor way only, for several reasons. First, it would mean that some people connected with the IVMG would have to organize a quasi commercial venture, with all the issues of responsibility, credit, rights, negotiations, and so on that that implies. Second, absent any entrepreneurial consensus, it seemed gracious and convenient enough to offer Aegean Park Press the chance at the publishing coup of the millennium. If not the aeon, the yuga, or the whole day and night of manifestation. Third, what one suggests, one volunteers for. Australia is presumably still remote enough to make the last only a small threat. Seattle is inconveniently near *and* far. Still, if, like lightning, we should explore various avenues before choosing the path for the final strike, I will continue to be one of the tentatively probing tendrils... From RJB@MAX.U.WASHINGTON.EDU Wed Jan 08 13:48:00 1992 Date: Tue, 7 Jan 92 20:48 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: the ultimate value of the work To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I certainly hope you didn't offend anyone, either. Really, the *ultimate* value of anything is hard to assess, and perhaps it is best not to worry oneself too much about it. I was not offended, anyway. I thought you said it very graciously, after all --and correctly enough. But "amateur" has not always been a bad word. If we must think of ourselves as serving a higher purpose, we could always think of ourselves as self-sacrificing soldiers in the attempt to conquer the dread Problem of Leisure. --rjb From reeds@gauss.att.com Wed Jan 08 14:41:16 1992 Message-Id: <9201080541.AA11163@rand.org> From: reeds@gauss.att.com Date: Wed, 8 Jan 92 00:41:16 EST To: voynich@rand.org Status: OR One thing to put in the facsimile book is an analysis of Currier's hands. In one place Currier claims as many as 8 different handwriting styles. (I don't have a clue what the differences are.) It should be a cinch to assemble typical examples (a few words or lines) from each (according to locations cited by Currier, or otherwise), and display them on the same page, with a few descriptive words pointing out the main differences. A more controlled sort of research activity would be to take several paragraphs (sans pictures) written in the alleged different handwritings and present them to a handwritingologist (paleographer, graphologist, etc) who is not a hook-line-and-sinker Voynichite, and see if Currier's handwriting classification holds up. From j.guy@trl.OZ.AU Thu Jan 09 00:57:10 1992 Date: Wed, 8 Jan 92 10:57:10 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201072357.AA07576@medici.trl.OZ.AU> To: voynich@rand.org Subject: Publishing the VMS Status: OR I have extracted from my shelves a book entitled "THE MYSTERIES OF EASTER ISLAND", by one Jean-Michel Schwartz. It is a paperback published by Avon in 1975, price $1.75, size 105x175mm, 210 pages; it has four B&W photographs. There are many line-drawings of Easter Island glyphs interspread with the text (not the real glyphs, the bogus things in the modern manuscripts collected by the Norwegian expedition in 1955, but never mind). It is stuff and nonsense from beginning to end. Nevertheless, someone at Avon thought it worth translating from the French (first published in 1973 by Editions Robert Laffont). Next I extracted this other book: "How to do your own Publishing -- A complete guide to self-publishing in Australia and New Zealand" (I have another book on the same topic, but I couldn't find it immediately). It is quite short, but complete and practical, dealing with all aspects of publishing, technical, commercial, financial. Next, I extracted this other book: "Symbols of Conflict" by Cedric Forster, 1989. It is a paperback novel, 125x200mm, 202 pages. Recommended retail price: $9.95 (about US$7). It's self-published, and I know the author, and how he screwed up (awful cover, unprofessional-looking typesetting, terrible blurb). It was printed here in Australia at a cost of $4 per copy (about US$2.75). Next, I thought of this friend of mine who is into book distribution and postcard publishing in his spare time (started as a hobby), and who is doing so well that, time and again, he has half a mind to drop his $80,000-a-year corporate manager job. (I always do my best to talk him out of it). Finally, I thought of the French cartoonist Claire Bretecher, who got cheesed off with commercial publishers and started publishing her works herself. Self-publishing. Think about it. From reeds@gauss.att.com Thu Jan 09 03:37:12 1992 Message-Id: <9201081837.AA22117@rand.org> From: reeds@gauss.att.com Date: Wed, 8 Jan 92 13:37:12 EST To: voynich@rand.org Status: OR Has anyone read or have easy access to either Trithemius's Polygraphia or his Steganographia? I find the summary in Kahn pretty slipshod, those in Yeats GB&HT tantalizing but unsystematic & brief, and I've lost my Walker S&DM! (Walker, I recall, was slipshod on the cryptology.) Jim From jbaez@math.mit.edu Thu Jan 09 04:19:49 1992 Date: Wed, 8 Jan 92 14:19:49 EST From: jbaez@math.mit.edu Message-Id: <9201081919.AA00357@jordan> To: voynich@rand.org Subject: getting your very own Voynich ms Status: OR I am receiving lots of requests for copies of the Voynich. I'm keeping them; when I actually get the darn thing, I will let everyone know how much it'll cost -- essentially 40 US dollars, divided by the number of people who want one (including me), plus postage. If you want your very own SECOND-GENERATION copy from Yale (rather than a copy of my photocopy of their microfilm), it costs $40 from the Beinecke Rare Book Library. This is not a large, streamlined outfit, however, so instead of just sending money to them, I would strongly advise contacting Robert Babcock first. His address is in the file pub/jim/biblio, available by anonymous ftp from rand.org, and the phone number of the Beinecke is 203-432-2977 (ask for him). Please be nice to him, as it's he who controls access to the Voynich. The way they usually do things is 1) you write requestg a copy, 2) they send you an invoice, 3) you pay, 4) you get it. To accelerate this you will have to contact them. Also, only bother him if you're a true Voynich-maniac; otherwise, if you're patient, you can get a cheaper version from me. jb From RJB@MAX.U.WASHINGTON.EDU Thu Jan 09 06:42:00 1992 Date: Wed, 8 Jan 92 13:42 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Trithemius To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I have looked at both -- in connection with a hint somewhere that Trithemius had been influenced in some way by Hildegard of Bingen's secret language/script. But that was 15+ years ago. I thought our library had one of them on microfilm, but I don't find it in the on-line catalogue. I know I borrowed one via interlibrary loan--must have seen the other at the BL. From foxd@silver.ucs.indiana.edu Thu Jan 09 22:56:52 1992 Message-Id: <9201091357.AA12309@rand.org> Date: Thu, 9 Jan 1992 08:56:52 -0500 From: daniel fox To: jbaez@BOURBAKI.MIT.EDU, voynich@rand.org Subject: Re: getting your very own Voynich ms Status: OR I too would like a copy of the voynich. Please put me on your list. df From firth@SEI.CMU.EDU Fri Jan 10 04:59:24 1992 Date: Thu, 9 Jan 92 14:59:24 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201091959.AA22538@bp.sei.cmu.edu> To: voynich@rand.org Subject: Retry of Note 8 Status: OR Notes on the Voynich Manuscript - Part 8 ---------------------------------------- Clews to the Voynich Script That's as in Feely's "clews": words that we might guess from the pictures they accompany or label. For instance, there are 30 individual Voynich words on the pisces folio (f70r), and they might be stars, days, saints, angelic governors, or John Dee's ex-girlfriends born in March. But they surely mean something. D'Imperio Fig 4 lists eleven other zodiac folios, and over sixty (!) herbal folios. That's a lot of possible simple names. One attraction to this approach (apart from it being cheap, easy and requiring little structured thought), is that it might provide a "clew" into the script independent of the underlying language. Proper names are common across languages: the month names, for instance, are similar in all of Western Europe. Plant names are harder, but usually the romance and the germanic forms will cover a lot of territory. And, my best candidate, if we can recognise the pictures, is star names: most of them are derived from Arabic, and are common to virtually all of Europe. Recall, please, that the egyptian script was deciphered by just this means: the proper names on the Rosetta stone and other inscriptions of the Hellenistic period. This might be a way to break one insoluble problem - an unknown script and an unknown language - into two soluble ones. If, for instance, we can deduce the alphabet from the star names, and corroborate it from the plant names, we could then romanise the text with a fair degree of confidence. Well, I wrote out those 30 piscean names, and stared at them for a long time, but to no avail. And I'm pretty hopeless at plants, so my few guesses (garlic on 100v? - ivy on 100r?) are unreliable at best. What we really need is (a) reproductions of the whole MS, and (b) a tame european botanist to go through the original colour folios. No thoughts on (b); as for (a) - yes, I'd pay my share of a $250 copying cost. Robert From firth@SEI.CMU.EDU Fri Jan 10 06:51:33 1992 Date: Thu, 9 Jan 92 16:51:33 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201092151.AA22852@bp.sei.cmu.edu> To: voynich@rand.org Subject: Retry - Notes 9 Status: OR Notes on the Voynich Manuscript - Part 9 ---------------------------------------- Curious Parallel Word Lists Over the past couple of weeks, I've been on a pretty wild goose chase concerning the vowel system (if any) of the Voynich text, and another concerning the implication of the ligatured letter forms. In this process, a couple of small ideas came up that might be useful to the group; they concern parallel lists of words that differ only in their initial letters. I use the j.guy scheme of transcription. The first issue is mentioned by D'Imperio [4.1.1 (13) p 28], and concerns the mysterious letter "4". This appears only at the beginning of words, and only before the letter "o". What is more surprising, is that it indeed seems to be a detachable prefix, since as far as I can tell, every common word in both A and B sections that begins "4o-" has an equivalent that begins "o-". >From VA: 4o8aiiv [10] o8aiiv [22] 4olp9 [23] olp9 [31] 4olpaiiv[18] olpaiiv [25] 4olpox [18] olpox [24] One might conjecture that this was a prefix such as "an-" in Greek, or maybe a contraction of "et-" as in "pepper &salt &garlic", but if so, why only in front of words beginning with "o-"? The second issue is the inflection mark above the ligatured letter that looks like "cc" with the tops run together. The two forms are Currier's "S" and "Z", and Guy's "ct" and "c't". Again from VA: c't9 [51] ct9 [102] c'taiiv [12] ctaiiv [37] c'tc9 [32] ctc9 [54] c'tcc9 [18] ctcc9 [15] c'to [83] cto [37] c'to89 [12] cto89 [28] c'to8aiiv [10] cto8aiiv[18] c'tox [92] ctox [214] c'to2 [65] cto2 [146] You know, I find the stability of those relative frequencies astonishing. Even if the little accent is a modifier of some kind, why should, near enough, 30% of the words get modified, for every one of 6 words? In the VB text, we find these letters in other than an initial position: oxc'tc89[16] oxctc89 [30] oxc'tc9 [10] oxctc9 [16] At present, I have no hypothesis to explain these patterns. However, one hypothesis they refute, in my opinion, is that the underlying language is a synthetic language based on a pure left factoring, or tree structure, such as the Dewey decimal classification or the Real Character of Wilkins. In such a scheme, the initial letter is the major category, the next letter the subcategory, and so on down the tree. This is likely to produce "words" that differ in the later letters, but not "words" that differ in their initial letters. Thus, ctox Gods, pagan, egyptian, anthropomorphic cto2 Gods, pagan, egyptian, theriomorphic but c'to2 Plants, european, medicinal, febrifuge This may be a part of the Voynich "language", but it surely isn't the largest part. Finally, if I'm allowed to assume "4thing" as a variant of "thing", then on f82v (Brumbaugh p104), I claim that of the 14 words that label the drawings, eight occur also in the text. (There are actually 15 labels, but one, "oqpc8ax", occurs twice.) More food for thought. Robert From jbaez@math.mit.edu Fri Jan 10 08:38:36 1992 Date: Thu, 9 Jan 92 18:38:36 EST From: jbaez@math.mit.edu Message-Id: <9201092338.AA03425@cayley> To: voynich@rand.org Status: OR People have been coming up with a lot of Voynich-related projects lately... tee-shirts, coffee table editions, etc.. No doubt this will sound snotty, but my main interest is in solving the Voynich problem -- or more precisely, laying the groundwork for a solution -- so while I'd enjoy it if someone *else* did all these, I won't spend energy on them myself. Getting Yale to okay the sale, by Aegean Park Press for example, of a black- and-white reproduction sounds doable and might facilitate solving the Voynich puzzle. For now, though, I must admit that distributing copies to mailing list members will keep me happy and busy. As for a color version, one should realize that the Beinecke is unlikely to have the facilities to make color plates themselves (at least according to my fiancee), so this'd introduce extra complications. I think it's unlikely that the solution of the Voynich problem will turn on the colors of ink (though it *could* turn on a chemical analysis of the inks -- if this helped date or locate its origin). I think we should save our energy for transcribing and pondering the puzzle. Maybe I'm wrong, but I bet there is a fair amount of the Voynich available for us to transcribe while we're waiting for the whole thing. My fiancee also points out that nobody should get conned into putting more time into this puzzle than they can afford! This applies in particular to people who should be doing their theses! ---- I think that any theory which CAN explain the following points is well on its way to solving the Voynich puzzle. Any theory which does NOT explain the following points represents no real progress. 1) The parts of the Voynich written by the two "hands" A and B seem to have statistically significant differences. (I must admit that I've never really SEEN the difference between these two handwritings. WHO IF ANYONE HAS?) On the other hand, many general proclivities of the Voynich language seem to be shared by both hands. This is very puzzling -- if A and B took turns cranking out nonsense one wouldn't expect this, unless they agreed ahead of time "okay, the following characters will always occur at the beginning of words, etc." Perhaps it could be explained by a coding scheme that had a bit of freedom of choice in it. 2) The text has pictures of astrological charts, herbs, "pharmaceutical jars" (if that's what they are) and naked women in bathtubs, with weird connecting plumbing. No other text combines these features (or, seemingly, has the fourth). 3) The words at the beginning of each line seem different in character from the rest. Here, as with point 1), it'd be nice to get some kind of statistical significance backing this claim. 4) The letter-by-letter entropy of the words is low. (How about the word-by-word entropy, by the way???????) As far as I know there is good theory that tackles all 4 of these points... or even a bunch at a time. ---------- Plaudits to Robert Firth for his ongoing series, especially his discovery that "4" may be a detachable prefix of some sort!! (And the rest of note 9.) One comment about note 8 - I do not think a botanist could identify most of the plants in the Voynich ms, as many look like no known plants. Medieval herbals were not known for their realism, however; an expert on herbals might do better. From yanek@mthvax.cs.miami.edu Fri Jan 10 13:08:29 1992 From: Yanek Martinson Message-Id: <9201100408.AA08852@mthvax.cs.miami.edu> Subject: Re: Retry - Notes 9 To: firth@SEI.CMU.EDU Date: Thu, 9 Jan 92 23:08:29 EST Cc: voynich@rand.org In-Reply-To: <9201092151.AA22852@bp.sei.cmu.edu>; from "firth@SEI.CMU.EDU" at Jan 9, 92 4:51 pm X-Mailer: ELM [version 2.3 PL11] Status: OR > You know, I find the stability of those relative frequencies astonishing. ... > hypothesis they refute, in my opinion, is that the underlying language is > a synthetic language based on a pure left factoring, or tree structure, ... > "words" that differ in the later letters, but not "words" that differ in > their initial letters. Thus, Voynich might have been written from right to left, so you cant know if these are work beginnings or endings that you are looking at. However, what such statistical curiosities DO prove, is that it is not just random gibberish, which would be very unlikely to exhibit such features. From reeds@gauss.att.com Sat Jan 11 01:06:44 1992 Message-Id: <9201101606.AA09547@rand.org> From: reeds@gauss.att.com Date: Fri, 10 Jan 92 11:06:44 EST To: voynich@rand.org Status: OR Just got of the phone with Garland. Whitby's ``John Dee's Actions with Spirits'' is going out of print. They have 91 copies left, and they are selling it at $15 apiece! Garland Press, 1-800-627-6273. The business end of the ISBN is 063-996. So while I am at it: UMI sells a xerox of Dee's diary (J. O. Halliwell 1842 edition) for $39.70, of Trithemius's Opera Historica (Frankfurt, 1601) for $295, of a French Polygraphie (Paris, 1561) for $172.30. Also found in the UMI catalogue, but surely in any reputable research library (that is to say, any library outside of Union County, NJ!) James Knowlson, Universal Language Schemes in England and France, 1600-1800. Toronto, 1975, and G. F. Hill, The Development of Arabic numerals in Europe, Oxford,1915. A glance through the current ish of the ISIS Critical Biography yields: John Dee's Library Catalogue, Edited by Julian Roberts and Andrew G. Watson, London, The Bibliographic Society, 1990. Vladimir Karpenko, The oldest alchemical manuscript in the Czech language. AMBIX 1990. v37. Z. R. W. M. van Martels (ed) Alchemy Revisted. Leiden: Brill, 1990. [Lots of interesting papers in this one.] In addition to AMBIX, there is now a new alchemy journal, Chrysopoeia. From jbaez@math.mit.edu Sat Jan 11 01:21:17 1992 Date: Fri, 10 Jan 92 11:21:17 EST From: jbaez@math.mit.edu Message-Id: <9201101621.AA04390@borel> To: voynich@rand.org Subject: gibberish? Status: OR I agree that the a constant frequency ratio of word pairs, one of which is of the form 4[string] and the other of the form [string], would be good evidence that something is going on.... IF this ratio is DIFFERENT from the frequency of the letter 4! One has to be careful when dealing with statistics (and other power tools). If one printed each letter with some probability independent of its neighbors, and randomly inserted spaces as well, one would expect the ratio of frequencies of words of form 4[string] to words of form [string] to be roughly constant... but this ratio would simply equal the frequency with which the character 4 appeared! In short, I am very optimistic about the use of statistics to FIRST decide whether the Voynich is gibberish or not, and THEN, if the latter, to crack it. But doing so will require some statistical sophistication... (in other words, careful thought). I really like this kind of puzzle. jb From reeds@gauss.att.com Sat Jan 11 02:07:49 1992 Message-Id: <9201101708.AA11060@rand.org> From: reeds@gauss.att.com Date: Fri, 10 Jan 92 12:07:49 EST To: voynich@rand.org Status: OR WHoops, John just pointed out that I used the abbreviation UMI unexpanded. UMI (A Bell & Howell Company) University Microfilms, Inc. 300 North Zeeb Road Ann Arbor, MI 48106-1346 USA 1-800-521-3042 They go out and film books (and manuscripts?) and then sell copies at $.25 a page. Before Aegean Park Press they were the main convenient source of copies of cryptology classics: Wilkins, Galland, Mil Crypt, Hitt,... They also film most PhD theses in this country. They have a huge backlog, & will send you a free catalogue (on microfiche-- what else!) if you call them up. From j.guy@trl.OZ.AU Sat Jan 11 02:17:27 1992 Date: Fri, 10 Jan 92 12:17:27 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201100117.AA10176@medici.trl.OZ.AU> To: voynich@rand.org Subject: R.Firth's remarks Status: OR ... are well put and timely. The properties he observes are compatible with the existence of prosodic features in the Voynich language. Tones, for instance. Further, tones are almost always subject to sandhi, that is, a tone may become another in the vicinity of certain tones or letters. That would explain why decipherers have found it difficult to identify words. It would also explain certain repetitions: perhaps two repeated words are REALLY two different words in different tones, but sandhi rules cause the tone of one to assimilate to that of the other. I am thinking at the moment about a fast, compact concordance program to find all the occurrences of any given string of letters, list where they were found, and display them with their contexts. What I'm really trying to do is take advantage of what I am working on right now for Telecom, and "recycle" it into a tool of analysis for the Voynich material. From RJB@MAX.U.WASHINGTON.EDU Sat Jan 11 02:27:00 1992 Date: Fri, 10 Jan 92 09:27 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: recent book notes To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Thank you very much for the information on Dee; I have just ordered one, and in a few weeks hope to be able to retire those dog-eared Xerox copies made 20+ years ago... I checked Whitaker's (British BIP) as well as US BIP and saw no mention of Roberts' catalogue of Dee's library. Probably the Bibliographi Society doesn't list its things there. I may have to grit my teeth and see if Blackwell's can get it for me. Every year they take longer and longer to fill orders... ah well. Thanks for the rest of the book notices too. Will have to check them out. The Trithemius from UMI does seem a bit steep though. I'm saving up for the Definite Voynich Facsimile, which will be published with a compact disk of machine-readable text, overlaid with a sound-track of team Voynich chanting the readable transcription. From RJB@MAX.U.WASHINGTON.EDU Sat Jan 11 02:37:00 1992 Date: Fri, 10 Jan 92 09:37 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: gibberish? To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR How likely is it that fakers without a theory of randomness would produce a random text, even if they were trying to produce nonsense? I'd suspect something more like glossolalia. Did they have a theory of randomness way back when? Combinatorial studies existed. Did the cryptologists of the time make tables of character frequencies and so on? Was there a proto-computational linguistics? I don't know. --rjb From reeds@gauss.att.com Sat Jan 11 03:32:18 1992 Message-Id: <9201101832.AA12902@rand.org> From: reeds@gauss.att.com Date: Fri, 10 Jan 92 13:32:18 EST To: voynich@rand.org Status: OR I am going to send an updated version of my checklist to Jim G to be put in the ftp zoo. Improvements:some errors corrected, more info transcribed from D'Imp and from Currier. John Baez and I are interested in whether the A and B stuff ever appears mixed -- both on the same sheet. (Remember, sheet = 1 piece of velum, folded in the spine, creating two folios = 4 pages). The gathering numbers on p100 of D'Imp make it pretty clear how the folios pair up into sheets, at least in the first half of the MS, and it appears that all sheets in the first half are all in A or all in B: no mixed sheets. This is consistent with parallel composition by multiple scribes. Jim Reeds. From jbaez@math.mit.edu Sat Jan 11 03:44:06 1992 Date: Fri, 10 Jan 92 13:44:06 EST From: jbaez@math.mit.edu Message-Id: <9201101844.AA18822@nevanlinna> To: voynich@rand.org Subject: randomness Status: OR RJB writes: How likely is it that fakers without a theory of randomness would produce a random text, even if they were trying to produce nonsense? I'd suspect something more like glossolalia. Did they have a theory of randomness way back when? Combinatorial studies existed. Did the cryptologists of the time make tables of character frequencies and so on? Was there a proto-computational linguistics? I don't know. --rjb First, there's not just one kind of random text. Let me illustrate with decimal numbers. A long number in which each digit appears 1/10th of the time, with the probability of it occuring unaffected by what any OTHER digit happens to be, is one kind of random number, which indeed has maximal entropy. We say that the the digits are mutually independent random variables (what ONE happens to be has nothing to do with what ANOTHER is) and identically distributed. But there are other kinds of random numbers, for example, it could be that "2" occurs 15% of the time and "1" occurs 5% of the time, the rest all appear 10% of the time, and the successive digits being mutually indepenent. (Sorry, the real jargon is "stochastically independent".) Here too they are stochastically independent identically distributed random variables but with lower entropy. They would not be identically distributed, however, if ever sixth digit was more likely to be a "7", and they would not be stochastically independent if 6's tended to be followed by 4's. So in brief there are loads of kinds of texts which would be called random but have different statistical characters, and there are big fat probability theory books on this stuff, a bit of which I know. It's impossible for humans to make lists of stochastically independent identically distributed characters without resorting to spinners or other random number generators. Gibberish would presumably have a particular statistical character, probably dependent on the person (we all have our favored letters etc.). I am interested in how gibberish could be distinguished from encoded text. This is in general an incredibly difficult problem (see Poundstone's Labyrinths of Reason which started me on this quest). But thanks to the mathematical naivete of the year before 1650 (say), there are limits to how sneaky X (or should I say A and B?) could be. Probability theory was a relatively recent invention (in the 1700's). Others -- or Kahn's book -- can do better at describing the early history of cryptography.... especially since I'm late for an appointment! jb From jim%mycroft@rand.org Sat Jan 11 05:09:47 1992 Message-Id: <9201102009.AA14375@mycroft.rand.org> To: voynich@rand.org Subject: Voynich ordering Date: Fri, 10 Jan 92 12:09:47 PST From: Jim Gillogly Status: OR I ordered my two copies this morning -- microfilm and prints. The "prints" aren't Xerox: they're printed from the microfilm. The film should cost about $30 instead of the approx $40 for the prints. I forgot to ask in the phone call about permission to let Aegean Park Press do its copy from the microfilm, but did remember to ask in my request letter. My preference is to go with Aegean for the "mass" printing: if somebody else wants to pursue the coffee-table color edition I have no objection (and would get one -- D'Imperio told me that it looks magnificent in color), but as others have indicated, my main interest is in *reading* the thing! Jim Gillogly From RJB@MAX.U.WASHINGTON.EDU Sat Jan 11 07:53:00 1992 Date: Fri, 10 Jan 92 14:53 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: types of "randomness" To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Thank you for the clarification on types of randomness. It gave me some more things to brood about. Kahn I read many years ago; I had a general impression from it that cryptography developed rather slowly until the 19th century, but it was only an impression. You've confirmed for me that it's very unlikely that there had been a thorough-goung cryptographic breakthrough i without the computational technology to support it. Poundstone I haven't read, and will have to try to dig out from some library. I think the close examination of handwriting is one of the more promising immediate avenues, especially when combined with comparisons of traits of segments of text. I hope to join the transcription work soon -- as soon as I can clear a space on the table. From RJB@MAX.U.WASHINGTON.EDU Sat Jan 11 07:59:00 1992 Date: Fri, 10 Jan 92 14:59 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Good news about Voynich ordering To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR All around good news. If Yale goes for it, APP would be ideal for the "working facsimile." If the original is "magnificent," then eventually a good color facsimile might be a worthwhile (text in the sky) project. --rjb From jim%mycroft@rand.org Sat Jan 11 09:25:32 1992 Message-Id: <9201110025.AA15097@mycroft.rand.org> To: voynich@rand.org Subject: Any questions for Currier? Date: Fri, 10 Jan 92 16:25:32 PST From: Jim Gillogly Status: OR I'm about to write back to Currier, giving him some of the results of our collective labors. Instead of a complete dump this time, which is probably a side-inch of chit-chat for December, I'm planning to send a few selected messages. I'm taking nominations for worthwhile findings to ship (I've got some favorites in hand) and questions we'd like him to comment on or help with. Jim Gillogly From reeds@gauss.att.com Sat Jan 11 09:56:31 1992 Message-Id: <9201110059.AA21355@rand.org> From: reeds@gauss.att.com Date: Fri, 10 Jan 92 19:56:31 EST To: voynich@rand.org Status: OR Another useful looking 16th century book: John Henry, Sarah Hutton (eds) New Perspectives on Renaissane thought: essays in memory of Charles B. Schmitt. London: Duckworth, 1990. From jim%mycroft@rand.org Sat Jan 11 11:08:05 1992 Message-Id: <9201110208.AA15579@mycroft.rand.org> To: voynich@rand.org Subject: Identifying language (A or B) on a page Date: Fri, 10 Jan 92 18:08:05 PST From: Jim Gillogly Status: OR I wrote a Perl program to apply Currier's Language A/B identification rules to transcribed text (in Currier transcription, although it will be obvious how to convert it to use Guy's format). It's "langchk.pl" in the ftp directory. To use it, say "langchk.pl 020 < voynich.now", for example, to extract all lines from page 020 and test them. Currier's rules are: a) Final 89 is very high in Language B; almost non-existent in Language A. b) SOE and SOR are very high in A, often repeated; low in B. c) The symbol groups SAN and SAM rarely occur in B; medium frequency in A. d) Initial SOP high in A, rare in B. e) Initial Q very high in A, very low in B. f) Unattached finals scattered throughout Language B. I didn't know how to quantify f), so I ignored it for now. For the others, I calculated an ad hoc A Language score as follows: a) Score 1 if final 89 < 10% of total words b) Score 1 if SOE and SOR together > 5% of total words Score 1 more if either SOE or SOR is repeated on the page c) Score 1 if SAN and SAM together > 2% of total words d) Score 1 if initial SOP > 2% of total words e) Score 1 if initial Q > 2% of total words Here's the output for f1r (page 001): Page 001 nwords = 215. Test a) word-final 89 is 3 of 215 Test b) SOE is 10 of 215. Test b) SOR is 2 of 215. Test b) SOE repeated: 1. Test b) SOR repeated: 0. Test c) SAM or SAN separately: 0. Test d) Initial SOP 2. Test e) Initial Q 16. Overall score for page 001: 5. I extracted the overall scores from these and summarized them in Appendix B, where column A is this A-language-score, and Ident is Currier's identification from D'Imperio's transcription (see the "checklist"). This score discriminated well between the two flavors: All but one of the A-identified pages had scores of two or higher, and all but one of the B-identified pages had scores of one or lower (see Appendix A). The pages transcribed by our group that aren't in D'Imperio's transcription are 65v, 68r, 100r, and 100v. Of these, the first two show up as language B (score 0 and 1), and the latter two as language A (score 3 each). Jim Gillogly -------------------------------- Appendix A: Breakdown of scores by pages identified by Currier as A and B. A idents: 0: 0 1: 1 2: 10 3: 20 4: 27 5: 24 6: 4 B idents: 0: 25 1: 19 2: 1 3: 0 4: 0 5: 0 6: 0 -------------------------------- Appendix B: Scores for individual pages. A Ident 001 5 A 002 4 A 003 3 A 004 2 A 005 3 A 006 3 A 007 5 A 008 5 A 009 2 A 010 4 A 011 5 A 012 5 A 013 5 A 014 3 A 015 6 A 016 6 A 017 3 A 018 4 A 019 5 A 020 3 A 021 5 A 022 2 A 023 3 A 024 4 A 025 2 A 026 2 A 027 4 A 028 5 A 029 4 A 030 4 A 031 4 A 032 4 A 033 4 A 034 2 A 035 4 A 036 4 A 037 5 A 038 3 A 039 3 A 040 6 A 041 4 A 042 4 A 043 4 A 044 3 A 045 4 A 046 4 A 047 5 A 048 4 A 049 1 B 050 0 B 051 3 A 052 2 A 053 4 A 054 4 A 055 4 A 056 5 A 057 5 A 058 5 A 059 0 B 060 1 B 061 4 A 062 4 A 063 0 B 064 1 B 065 1 B 066 0 B 067 5 A 068 4 A 069 4 A 070 3 A 071 3 A 072 4 A 073 3 A 074 3 A 075 0 B 076 0 B 077 0 B 078 0 B 079 0 B 080 0 B 081 6 A 082 5 A 083 1 B 084 0 B 085 2 A 086 5 A 087 5 A 088 3 A 089 0 B 090 1 B 091 5 A 092 5 A 093 1 B 094 1 B 095 5 A 096 5 A 097 0 B 098 0 B 099 1 A 100 3 A 101 2 A 102 3 A 103 3 A 104 2 A 105 4 A 106 3 A 107 0 B 108 1 B 109 5 A 110 5 A 111 0 B 65v 0 ? 68r 1 ? 147 1 B 148 0 B 149 1 B 150 1 B 151 0 B 152 0 B 153 0 B 154 0 B 155 1 B 156 1 B 157 0 B 158 2 B 159 0 B 160 1 B 161 1 B 162 1 B 163 1 B 164 1 B 165 0 B 166 0 B 100r 3 ? 100v 3 ? From reeds@gauss.att.com Sat Jan 11 14:34:14 1992 Message-Id: <9201110540.AA26263@rand.org> From: reeds@gauss.att.com Date: Sat, 11 Jan 92 00:34:14 EST To: voynich@rand.org Status: OR Another thing to investigate are the ``ghosts'' of pictures. On the color plates in Blunt and Raphael, for instance, on the left side of 33v, we see a ghost of a picture of a plant. Is this color bleeding through the page or is it a left-over from juxtaposition of still-wet sheets when the book was made? Where in the book is the ``original'' for the ghost on 33v? And so on. I found it very hard to transcribe 6v because of a very strong ghost covering up some writing. From reeds@gauss.att.com Sat Jan 11 14:46:51 1992 Message-Id: <9201110549.AA26350@rand.org> From: reeds@gauss.att.com Date: Sat, 11 Jan 92 00:46:51 EST To: voynich@rand.org Status: OR About a month ago RJB commented >>> Yes: it would be interesting to try to surmise whether the text was >>> written into a bound book or whether the sheets were codexicated (!) after >>> they were written. Almost certainly bound after writing. Gatherings 5, 6, and 7 each contain A and B text, but each sheet there is either all A or all B. Gatherings 1 - 4 are solid A, and I cannot tell how the sheets are arranged in the second half of the VMS. From RJB@MAX.U.WASHINGTON.EDU Sat Jan 11 16:47:00 1992 Date: Fri, 10 Jan 92 23:47 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: bound after writing To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Can this be taken to suggest that the pages had been laid out before the scribal work was done, that the contents and layout were known? I'd think by the way that there shouldn't be much bleed-through with vellum -- more likely offset from a page or sheet (if they were stacked when unbound) above or below. Or simply transferred by pressure ovber (=over) a long period of time after the thing had been bound and had been on a shelf or in a chest or whatever. --rjb From wuth@castrov.cuc.ab.ca Sat Jan 11 18:02:27 1992 Return-Path: Date: Sat, 11 Jan 92 02:02:27 MST Message-Id: <9201110902.AA03585@castrov.cuc.ab.ca> From: wuth@castrov.cuc.ab.ca (Brett Wuth) To: voynich@rand.org, voynich-incoming@ender.cpsc.ucalgary.ca Subject: Re: Identifying language (A or B) on a page Status: OR >I wrote a Perl program to apply Currier's Language A/B identification >rules to transcribed text (in Currier transcription, although it will be >obvious how to convert it to use Guy's format). [Forgive me if I talk from a misapprehension -- I haven't seen examples of the two hands.] I hope we use this method to _analyze_ the two hands rather than to _identify_ them. If we overrule a visual interpretation because of the statistics of the text, we are actually loosing information. This is because any identification based on statistical properties is actually redundant information (it has been derived from the rest of the text). On the other hand, an identification based on visual properties is new information. -- Brett Wuth (403) 242-0848 wuth@castrov.cuc.ab.ca BCWuth@uncamult.bitnet wuth@castrov.uucp From reeds@gauss.att.com Sun Jan 12 01:20:22 1992 Message-Id: <9201111621.AA01346@rand.org> From: reeds@gauss.att.com Date: Sat, 11 Jan 92 11:20:22 EST To: voynich@rand.org Status: OR RJB on bound after writing and ghost images writes >>> Can this be taken to suggest that the pages had been laid out before the >>> scribal work was done, that the contents and layout were known? We don't see guidelines lines scribed on the MS, which I would take as needed for careful layout. (The circular diagrams DO have guidelines). I think write before bind was the usual method of manuscript book production, but the text would be written in ``reading'' order, so lots of sheets would be simultaneously half written part way through the process. There would be no special reason for a scribe to finish a sheet he had started. The all A sheets / all B sheets phenomenon seems to suggest sheets were farmed out, completely filled in with writing and then assembled whenever 4 complete sheets were ready. (Come to think of it, this is exactly what Currier says in one of his Seminar papers.) >>> I'd think by the way that there shouldn't be much bleed-through with >>> vellum -- more likely offset from a page or sheet (if they were stacked >>> when unbound) above or below. Or simply transferred by pressure ovber >>> (=over) a long period of time after the thing had been bound and had been >>> on a shelf or in a chest or whatever. Right. Bleed-through also seems unlikely to me. Offset after binding: no, not in the case of 33v/34r: the ghost on 33v doesn't match anything on the facing page. So offset while stacked during book production seems to be it. So what sheet was it stacked with? From RJB@MAX.U.WASHINGTON.EDU Sun Jan 12 04:43:00 1992 Date: Sat, 11 Jan 92 11:43 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: so what shee was it stacked with? To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR We should be able to tell a lot when we pore over the film and photocopies... z From jim@rand.org Sun Jan 12 08:48:48 1992 Message-Id: <9201112348.AA16327@mycroft.rand.org> To: voynich@rand.org Subject: Re: Identifying language (A or B) on a page In-Reply-To: Your message of Sat, 11 Jan 92 02:02:27 -0700. <9201110902.AA03585@castrov.cuc.ab.ca> From: Jim Gillogly Reply-To: jim@rand.org Date: Sat, 11 Jan 92 15:48:48 PST Sender: jim%mycroft@rand.org Status: OR > wuth@castrov.cuc.ab.ca (Brett Wuth) writes: > > I hope we use this method to _analyze_ the two hands rather than to > _identify_ them. If we overrule a visual interpretation because of > the statistics of the text, we are actually loosing information. This No, I presented the program and information as a way to identify the two *Languages*, not the two hands. In my message I was careful to refer to the Language rather than the hand. As you say, the identification of hands should be done visually, and in fact in the "checklist" file you'll see more than two hands identified by Currier. I imagine we'll be able to see the difference in hands better when we get more samples in front of us. I do think it's important to tell which "Language" each page is written in as well as which hand... hence the program. Jim Gillogly From reeds@gauss.att.com Mon Jan 13 01:44:26 1992 Message-Id: <9201121644.AA14335@rand.org> From: reeds@gauss.att.com Date: Sun, 12 Jan 92 11:44:26 EST To: voynich@rand.org Status: OR Word bridge statistics. Working with the original D'Imperio data, I looked at the digraphic frequency count of word-bridging digraphs, separately for the A and B sub-corpora. That is, of digraphs whose first element is the final letter of a word and whose second element is the first letter of the following word on the same line. I counted - and / as word breaks. I only counted digraphs where both elements were in the 36 letter Currier alphabet. Let f(rs) denote the number of times such a digraph rs appears, and let f(r+) be the sum of f(rs) over all 36 possible s, let f(+s) be the sum of f(rs) over all possible r, and let f(++) be the sum over both r and s. Then I calculated an ``expected'' value for f(rs), namely e(rs) = f(r+) * f(+s) / f(++) which is what you would expect if successive words were chosen independently, adjusting for the number of times r appears as a word final and s as a word initial. Finally, for each given digraph rs I calculated the ratio z = ( f(rs) - e(rs) ) / sqrt ( e(rs) ) which is a measure of how far off what we see is from what we observe, measured in some kind of standard deviation units. Values of z larger than 2 or less than -2 may be regarded as suspicious. Interpretation and summary of results: I think the results show that there is a significant degree of correlation between a word's final letter and the next word's initial letter. Hence successive words are not independent. This might be a linguistic phenomenon (which has been discussed a bit already), but it might be an artifact of the cryptology, too. Digraph (observed/expected) z A sub corpus: A too rare A too frequent RF (12/44.8) -4.90408 MZ (94/75.9) 2.07448 MF (9/34.0) -4.29026 RZ (121/100.1) 2.09433 RP (4/25.1) -4.21676 R9 (68/52.8) 2.09953 R8 (131/188.3) -4.17703 M4 (79/60.6) 2.36672 E4 (45/82.8) -4.15489 2O (43/29.3) 2.5325 9S (373/459.8) -4.04902 MX (21/11.6) 2.75631 R4 (44/79.8) -4.01018 29 (21/10.6) 3.19401 9A (2/19.2) -3.92535 RO (185/145.8) 3.24935 9Z (138/185.2) -3.46621 JO (25/13.1) 3.28293 28 (17/37.8) -3.38837 RA (22/10.4) 3.6094 24 (3/16.0) -3.25629 OF (29/13.1) 4.37982 OO (23/42.7) -3.01293 RS (324/248.5) 4.79251 EF (26/46.5) -3.00762 MQ (86/42.2) 6.75058 MP (7/19.1) -2.7657 E8 (290/195.3) 6.77197 EX (5/15.9) -2.7284 94 (231/147.7) 6.84964 O2 (2/10.8) -2.67076 9P (94/46.5) 6.95782 9O (226/269.8) -2.6653 9F (167/83.0) 9.22295 E9 (37/54.7) -2.39545 99 (75/97.6) -2.29008 9Q (81/102.8) -2.15313 R2 (24/36.7) -2.1032 B sub corpus: B too rare B too frequent R4 (39/187.9) -10.8613 EP (27/18.2) 2.06472 9Z (237/463.3) -10.5136 9P (83/65.8) 2.11951 9A (31/162.4) -10.3107 92 (132/102.4) 2.92236 M4 (27/134.2) -9.25194 RS (171/133.2) 3.27137 9S (405/634.2) -9.10066 98 (422/350.3) 3.83329 E4 (115/247.2) -8.40933 9R (111/71.6) 4.64967 N4 (21/104.7) -8.1799 RO (283/212.7) 4.81738 9O (795/1012.6) -6.83726 EZ (185/128.1) 5.02882 RE (4/41.5) -5.81916 NO (175/118.6) 5.18402 ME (2/29.6) -5.07492 ES (248/175.3) 5.48817 R8 (31/73.6) -4.96459 EF (59/29.1) 5.55665 NE (2/23.1) -4.39176 RZ (154/97.3) 5.74331 24 (5/28.0) -4.34695 MZ (120/69.5) 6.0558 T4 (0/14.2) -3.76969 MS (156/95.1) 6.23823 RR (1/15.1) -3.62195 MO (230/151.9) 6.33473 NF (0/12.3) -3.50754 NS (129/74.3) 6.35349 R2 (6/21.5) -3.34542 NZ (106/54.2) 7.02715 MF (3/15.8) -3.215 9E (311/197.4) 8.08373 RF (8/22.1) -2.9959 94 (1426/894.2) 17.7831 MR (1/10.7) -2.97357 RA (158/34.1) 21.2088 MA (10/24.4) -2.91014 RP (4/13.8) -2.64252 EE (36/54.6) -2.51491 N2 (4/12.0) -2.3079 8S (6/14.8) -2.29125 EA (30/44.9) -2.22316 N8 (27/41.0) -2.18767 28 (4/11.0) -2.10412 From dtl8v@holmes.acc.Virginia.EDU Mon Jan 13 06:38:41 1992 Date: Sun, 12 Jan 92 16:38:41 -0500 From: Heraclitus Message-Id: <9201122138.AA69757@holmes.acc.Virginia.EDU> X-Mailer: Mail User's Shell (7.2.3 5/22/91) To: voynich@rand.org Status: OR Please discontinue my subscription to this mailing list. From jbaez@math.mit.edu Mon Jan 13 23:38:45 1992 Date: Mon, 13 Jan 92 09:38:45 EST From: jbaez@math.mit.edu Message-Id: <9201131438.AA05428@borel> To: voynich@rand.org Subject: For Currier Status: OR I guess the "biggies" are the supposed statistical differences of line-beginnings from line-middles, whatever statistics we've gotten about the differences between hands A and B, and Firth's stuff on ratios of prefixed to unprefixed words. As for questions, I'd like to hear his thoughts on the hand A/hand B stuff... someone said that he'd at one time theorized about MORE than 2 styles of handwriting. Of course, his secret personal theory about the manuscript would be interesting, if he has one. And it might be worthwhile, Yale being so darn sluggish, to get copies of his copies of copies of copies of copies of copies, just to have SOMETHING to stare at meanwhile. jb From jbaez@math.mit.edu Mon Jan 13 23:59:19 1992 Date: Mon, 13 Jan 92 09:59:19 EST From: jbaez@math.mit.edu Message-Id: <9201131459.AA05460@borel> To: voynich@rand.org Subject: Word bridge statistics Status: OR Jim Reeds word bridge statistics are fascinating. I want to add my (usual) statistical caveat: with lots of possible digraphs, and a not-too-large sample of text, one *expects* certain digraphs to appear "surprisingly frequently" (or rarely) in a word-bridging manner. One really has to do some statistics to see how shocked one should be by Jim's results. (Similarly, in Britain some people get excited over "ley lines" -- lines one can draw between old megalithic monuments. Of course, with thousands of megaliths, one *expects* some such lines to exist.) I propose that some hacker compute the correlation entropy of word-bridging digraphs. Here's how (using Jim's notation!). First find the entropy of word-ending characters by forming ... hmm, Jim's notation won't be best for this. Let p(rs) be the frequency that a given word-bridging digraph rs occurs (i.e., the fraction of the time this one occurs). Let p(r+) be the sum of p(rs) over s, and let p(+s) be the sum of p(rs) over r. Okay, find the entropy of word-ending characters by forming sum over r (p(r+) log p(r+)) = E(end) and similarly find sum over s (p(+s) log p(+s)) = E(beginning) Then find sum over r, s (p(rs) log p(rs)) = E(bridge) and E(bridge) - E(end) - E(beginning) is a measure of how much information knowing the ending letter tells you about which letter will begin the next word. (If you compute the logarithms base 2, this will be exactly how many BITS of information there are.) jb From jbaez@math.mit.edu Tue Jan 14 00:08:44 1992 Date: Mon, 13 Jan 92 10:08:44 EST From: jbaez@math.mit.edu Message-Id: <9201131508.AA05477@borel> To: voynich@rand.org Subject: Entropy Status: OR Sorry -- in my formula as I gave it the answer will come out negative. Just pop a minus sign in front and it'll be okay. In general, given a string of letters (or digraphs, or whatever) of N different kinds, each with frequency p(i), the information per symbol is - sum_i p(i) log p(i) where if you want to measure the information in bits use base 2. For example: if there are just two kinds of letter, and each occurs with frequency 1/2, then the information per letter is - 1/2 (log (1/2)) - 1/2 (log(1/2)), and if one works this out using logs base 2, it equals just 1. This makes sense: 1 bit per letter. The "correlation information" I described, e.g. E(bridge) - E(last) - E(first), will vanish precisely when the last letters of words are statistically independent from the first letters of the following word. It'd be best not only to check the Voynich, but to COMPARE it to an equal-sized hunk of English text. jb From reeds@gauss.att.com Tue Jan 14 03:35:31 1992 Message-Id: <9201131837.AA02841@rand.org> From: reeds@gauss.att.com Date: Mon, 13 Jan 92 13:35:31 EST To: voynich@rand.org Status: OR About word bridging digraphic counts. 1. On rereading Currier, I see that section 3 of Appendix A describes influence of word final letter on initial letter of following word. I must have read that weeks ago, forgot that I had, and then ``rediscovered'' the phenomenon! 2. John Baez points out the need for judging significance of these findings. That is, how ``suprising'' are my nominally suprising ``z'' values. My method would be to take the digraphs, split them up into front halves and back halves, mix them up, and randomly rematch fronts and backs. This way I obtain a new frequency distribution of pseudo bridging digraphs, which has the correct monographic counts for first elements and for second elements. For each of several such randomly reassembled digraph samples I recompute the z scores, just as I did with the real digraph sample. If the real bridging digraphs give the same sort of results as the pseudo bridging digraphs, the effect is not significant. If, however, I get different sorts of results, the effect is significant. The technical name for this method is ``monte carlo permutation distribution''. So I redid my calculations, once for real, and 30 times with randomly reassembled digraphs, and printed out the results in all 30 cases. The results (for the B text) is given below. Summary: the effect is really there. 3. John offered an entropy formula for measuring the size of the effect. It should be clear that his formula measures how big the effect is, but not whether it is significant. (If the sample size is increased, but the relative frequencies remain unchanged, John's entropy is unchanged.) In summary: about 1/4 of a bit of info in the final letter about the next word's initial letter. Not much, but its really there: the 30 H values figured with reassembled data show about 1/30 bit. RESULTS Real data: H=0.260696 bits 24 (5/28.0) -4.34695 28 (4/11.0) -2.10412 8S (6/14.8) -2.29125 92 (132/102.4) 2.92236 94 (1426/894.2) 17.7831 98 (422/350.3) 3.83329 9A (31/162.4) -10.3107 9E (311/197.4) 8.08373 9O (795/1012.6) -6.83726 9P (83/65.8) 2.11951 9R (111/71.6) 4.64967 9S (405/634.2) -9.10066 9Z (237/463.3) -10.5136 E4 (115/247.2) -8.40933 EA (30/44.9) -2.22316 EE (36/54.6) -2.51491 EF (59/29.1) 5.55665 EP (27/18.2) 2.06472 ES (248/175.3) 5.48817 EZ (185/128.1) 5.02882 M4 (27/134.2) -9.25194 MA (10/24.4) -2.91014 ME (2/29.6) -5.07492 MF (3/15.8) -3.215 MO (230/151.9) 6.33473 MR (1/10.7) -2.97357 MS (156/95.1) 6.23823 MZ (120/69.5) 6.0558 N2 (4/12.0) -2.3079 N4 (21/104.7) -8.1799 N8 (27/41.0) -2.18767 NE (2/23.1) -4.39176 NF (0/12.3) -3.50754 NO (175/118.6) 5.18402 NS (129/74.3) 6.35349 NZ (106/54.2) 7.02715 R2 (6/21.5) -3.34542 R4 (39/187.9) -10.8613 R8 (31/73.6) -4.96459 RA (158/34.1) 21.2088 RE (4/41.5) -5.81916 RF (8/22.1) -2.9959 RO (283/212.7) 4.81738 RP (4/13.8) -2.64252 RR (1/15.1) -3.62195 RS (171/133.2) 3.27137 RZ (154/97.3) 5.74331 T4 (0/14.2) -3.76969 Reassembled data: H=0.0356315 bits EF (16/29.1) -2.42131 Reassembled data: H=0.0339723 bits Reassembled data: H=0.032051 bits Reassembled data: H=0.0361688 bits 24 (15/28.0) -2.45724 N9 (7/15.2) -2.09596 Reassembled data: H=0.0306891 bits NF (4/12.3) -2.36714 Reassembled data: H=0.0385104 bits ER (30/19.8) 2.29026 MZ (50/69.5) -2.34018 RR (5/15.1) -2.59094 RZ (120/97.3) 2.29711 Reassembled data: H=0.0366299 bits 2S (29/19.9) 2.051 ES (144/175.3) -2.36609 Reassembled data: H=0.0349805 bits NA (9/19.0) -2.29644 R4 (158/187.9) -2.17941 TS (2/10.1) -2.54461 Reassembled data: H=0.03665 bits M9 (10/19.4) -2.13897 RE (55/41.5) 2.09978 RZ (76/97.3) -2.16268 Reassembled data: H=0.0390893 bits Reassembled data: H=0.03362 bits MF (24/15.8) 2.07393 Reassembled data: H=0.0366585 bits MZ (90/69.5) 2.45752 Reassembled data: H=0.0332926 bits Reassembled data: H=0.0356654 bits RZ (77/97.3) -2.06132 Reassembled data: H=0.0363342 bits NS (93/74.3) 2.17569 Reassembled data: H=0.0336548 bits ME (43/29.6) 2.45852 MZ (49/69.5) -2.46013 NZ (70/54.2) 2.13922 T4 (6/14.2) -2.17805 Reassembled data: H=0.0314912 bits 2S (31/19.9) 2.49979 RA (48/34.1) 2.37661 Reassembled data: H=0.037019 bits MA (41/24.4) 3.37018 MR (22/10.7) 3.43164 Reassembled data: H=0.0332853 bits Reassembled data: H=0.0351047 bits N9 (23/15.2) 2.01321 NZ (34/54.2) -2.74871 Reassembled data: H=0.037668 bits E8 (123/96.8) 2.65892 Reassembled data: H=0.0326455 bits 8Z (18/10.8) 2.17986 MF (24/15.8) 2.07393 Reassembled data: H=0.0390571 bits 9R (89/71.6) 2.05051 RP (22/13.8) 2.19843 Reassembled data: H=0.0374836 bits 2Z (26/14.5) 3.01693 EF (17/29.1) -2.23577 Reassembled data: H=0.0336981 bits 2S (30/19.9) 2.27539 E2 (14/28.3) -2.69045 MR (4/10.7) -2.05854 R9 (38/27.2) 2.0696 Reassembled data: H=0.0311691 bits Reassembled data: H=0.0355576 bits M9 (9/19.4) -2.36585 RP (6/13.8) -2.10463 Reassembled data: H=0.033669 bits 2S (29/19.9) 2.051 EF (18/29.1) -2.05024 Reassembled data: H=0.0323015 bits EE (38/54.6) -2.2442 MF (26/15.8) 2.57763 Reassembled data: H=0.034247 bits E2 (39/28.3) 2.00759 TS (2/10.1) -2.54461 From reeds@gauss.att.com Tue Jan 14 05:20:23 1992 Message-Id: <9201132025.AA05471@rand.org> From: reeds@gauss.att.com Date: Mon, 13 Jan 92 15:20:23 EST To: voynich@rand.org Status: OR Following John Baez's suggestion, I redid my word bridge thing on a sample of about 300,000 letters of English text (snagged off of the AP wire, with minimal cleanup to remove tabular data and funny newpaper formatting stuff.) The H seems to be .0875, the reassembled H seems to be .017 or .018, the effect is small but definitely there. V's effect is much stronger: H = .25 or so. From reeds@gauss.att.com Tue Jan 14 11:07:42 1992 Message-Id: <9201140207.AA13798@rand.org> From: reeds@gauss.att.com Date: Mon, 13 Jan 92 21:07:42 EST To: voynich@rand.org Status: OR I was just on the phone with Lou Kruh, collector and co-editor of Cryptologia. He tells me that the May/June 1976 and July/August 1976 issues of The Cryptogram contain Voynich papers (one by DENDAI, one by MERLIN, he thinks.) Anyone have easy access to these? He has a bunch of other V stuff, some of which we will loan me, including, I think, a copy of the NSATJ issue with Tiltman's paper. He has a Feeley, which I will pass up. From reeds@gauss.att.com Wed Jan 15 02:08:05 1992 Message-Id: <9201141708.AA27197@rand.org> From: reeds@gauss.att.com Date: Tue, 14 Jan 92 12:08:05 EST To: voynich@rand.org Status: OR Vocabulary drift is a possible explanation for the word bridge phenomenon. Suppose words are drawn randomly and independently from a fixed population of words, each word with its own probability of being picked. Then we would not expect word bridge correlation. We DO see correlations, to this supposed model cannot be right. But if the word probabilities change part way through this process you would expect to see such correlations. To take an extreme example, suppose the first part of the text is especially rich in CAT and DOG, so word sequences like CAT CAT DOG CAT are not uncommon. This of course makes bridging digraphs TC, TD, GC, and GD rather common. Supose the second half is rich in FISH and FOWL, so HF and LF are especially common. But because the vocab has drifted from CAT and DOG to FISH and FOWL, we see very few CAT FISH or FOWL DOG sequences, so TF and LD are uncommon. (A mixture of product measures is not neccessarily a product measure.) Of course, the change in vocabulary does not have to be drastic: just enough to skew the probabilities a little. The drift might take place at any ``scale'': sub vocab 1 might be used in the first 20 pages and sub vocab 2 in the next 20 pages, or they might be used on alternate lines, etc. And there might well be more than one such sub vocab. This gives us something to look for: Identify regions of the B text which have a different composition from other regions of the B text. From jim%mycroft@rand.org Wed Jan 15 04:51:07 1992 Message-Id: <9201141951.AA20591@mycroft.rand.org> To: voynich@rand.org Subject: New member -- Mary D'Imperio Date: Tue, 14 Jan 92 11:51:07 PST From: Jim Gillogly Status: OR Mary wrote back -- she can't connect to the net, but is very interested in the project and would like to participate, including transcription, analysis, whatever. I can send her occasional net dumps and relay her responses, and she also welcomes correspondence at: Ms. Mary E. D'Imperio Westchester Apts 106B Washington, DC 20016 She's retired now, and has plenty of time to work on the Voynich. She has a copy of many (not all) pages of the Manuscript inherited from Brig. Tiltman; about 2 of the 13 pages she sent me (ff 56r, 56v, 57r, 87r, 90r2, 90v1, 93v, 94r, 94v, 95r1, 95v1, 95r2, 95v2, 96r) are good enough for me to transcribe from without making assumptions about what follows what, whether it's a typical Z even though I can't read the connecting line, and so on. We still need the close-to-originals. Mary confirmed that the Seminar held in 1976 is unclassified (she got an explicit release for it). The most important paper from that Seminar (by Currier) has been typed in by Jacques except for the tables in the Appendix. If anybody else needs it, let me know. She has also written another paper concerning "a statistical model of transitions in the text". However, when she retired she didn't bring away a copy with the figures, which she will reconstruct and then send on. She'll also be sending a summary of the other work she's done since: no breakthroughs, but some interesting observations. Jim Gillogly From reeds@gauss.att.com Wed Jan 15 11:22:36 1992 Message-Id: <9201150223.AA09131@rand.org> From: reeds@gauss.att.com Date: Tue, 14 Jan 92 21:22:36 EST To: voynich@rand.org Status: OR Freuencies of ``fore & aft skeletons'' of words. For each word delete all internal letters, leaving just the digraph of initial and final letters. I made a frequency count of these fore & aft skeleton digraphs for Herbal B ( up through page 111) and Biological B (page 147 on). I was attempting to see if the bridging phenomenon in the B text was due to herbal/bio differences. The Herbal B and Bio B frequencies are different, sure enough, but when I rerun my bridging program on just the Herbal B sub sub corpus I still see the phenom, as I do when running just with Biological B. Results, for what they are worth: Herbal B Biological B 306 S9 897 49 250 O9 649 O9 143 49 570 S9 139 Z9 551 Z9 121 OR 296 OE 120 99 280 4E 86 89 223 E9 79 OE 186 4N 62 OM 175 89 62 8M 166 OR 60 F9 115 4M 53 8R 112 ON 47 AR 107 8E 47 AM 98 4R 44 P9 91 99 41 SE 84 8M 39 4R 80 29 33 SR 74 ZE 32 OJ 74 OM 32 8E 73 SE 25 ON 72 P9 25 4E 71 8R 24 9R 60 2E 23 B9 54 F9 23 8J 46 B9 23 2M 45 8N 22 SM 43 SR 21 FR 41 2M 20 ZE 40 EE 20 E9 39 2R 19 A9 36 R9 18 Q9 36 ER 18 AE 33 PE 18 4M 31 ZR 18 2 31 2N 17 PR 29 RE 17 O2 28 AM 16 S2 27 A9 15 9E 26 AE 14 9M 23 BE 12 X9 20 AR 12 8N 16 S2 11 O8 16 C9 11 AJ 16 AN 10 ZR 15 SN 10 S8 15 RM 10 PM 15 O8 10 OT 15 48 10 AN 14 RR 10 2R 14 PR 9 ZO 14 FN 9 SJ 14 EM 9 FM 14 BR 9 BR 13 SM 9 92 13 RN 8 Z8 13 R .... From rcarter@isis.cs.du.edu Wed Jan 15 14:44:49 1992 Date: Tue, 14 Jan 92 22:44:49 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9201150544.AA11043@isis.cs.du.edu> To: voynich@rand.org Subject: Further order info... Status: OR Just a small update on order info: It doesn't appear that a -phone- call to Bob Babcock is really needed, rather just a note (and $40) sent to the following address will do: Robert Babcock c/o Beinecke Rare Book Library 1603a Yale Station New Haven, CT 06520 That gets you a photocopy... I talked with Bob MON 13JAN92 AM and seems like a nice guy; very helpful; he also indicated that it -might- be few dollars over the $40, but he will invoice people for any balance due... Ron "D'Imperio is a member of Team Voynich?!" Carter From naga@wet.net.netcom.com Wed Jan 15 20:43:00 1992 Message-Id: Date: Wed, 15 Jan 92 03:43 PST From: naga@wet.net.netcom.com (Peter Davidson) To: voynich@rand.org Subject: Boring announcement about statistical software. Status: OR Report on the development of statistical software that may be useful in the investigation of the Voynich ms. by naga Partly in response to a suggestion by jbaez I have developed some software (written in Microsoft C) for analyzing a piece of text as follows (I use the language of text and letters rather than of files and bytes): Pick a piece of text and a "divergence factor" D > 1 (2 is good). For each "distance value" k, and for each letter b, the program I have developed counts the number of times b occurs k places beyond some occurrence of b. >From this can be calculated the probability, given that a letter occurs at some place in the text, that it occurs at some specified distance further on. For each distance k the program then displays those b's for which the actual number of re-occurrences is more than D times the expected number, where the expected number (assuming - as is probably not the case - that the letters in the text are distributed randomly) is the number of occurr- ences divided by the number of different letters that occur in the text. This program produces a lot of output, since the distance values can be anything from 1 onwards, depending on the size of the file, and there may be anything from 26 - 256 different letters in the text. Further statistical analysis of the results may be required, which I have not yet considered. Two examples of the output of the program will be given here, one for a file which consists of random bytes and one for a file which consists of English text (I don't yet have any machine-readable parts of the Vms): When the program is applied to a file consisting of 283,870 random bytes (i.e. bytes produced by a pseudo-random number generator) we obtain a result such as the following (the divergence factor is 2, meaning that byte values are listed only if they re-occur at the specified distance at least twice as often as expected): File size must be in the range 10240 through 505264. Divergence limit = 2.000 Start distance = 1 File RANDFILE contains 283870 bytes and 256 different byte values. Maximum frequency of 1196 attained by byte value: 213 (0xD5) Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in 6 1 1100 9 4.30 2.09 122.22 1 49 1 1132 10 4.42 2.26 113.20 5 53 1 1108 11 4.33 2.54 100.73 Y 89 1 1152 10 4.50 2.22 115.20 \ 92 1 1143 10 4.46 2.24 114.30 128 1 1105 10 4.32 2.32 110.50 157 1 1185 10 4.63 2.16 118.50 174 1 1074 10 4.20 2.38 107.40 224 1 1086 11 4.24 2.59 98.73 225 1 1138 9 4.45 2.02 126.44 254 1 1068 12 4.17 2.88 89.00 For distance 1, maximum divergence from expected among these 11 is 2.88, for byte 254. Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in # 35 2 1135 9 4.43 2.03 126.11 . 46 2 1113 9 4.35 2.07 123.67 O 79 2 1118 10 4.37 2.29 111.80 Q 81 2 1112 9 4.34 2.07 123.56 ^ 94 2 1154 11 4.51 2.44 104.91 m 109 2 1111 11 4.34 2.53 101.00 151 2 1096 10 4.28 2.34 109.60 164 2 1132 9 4.42 2.04 125.78 For distance 2, maximum divergence from expected among these 8 is 2.53, for byte 109. Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in W 87 3 1144 9 4.47 2.01 127.11 d 100 3 1105 9 4.32 2.09 122.78 i 105 3 1114 9 4.35 2.07 123.78 180 3 1144 11 4.47 2.46 104.00 244 3 1119 9 4.37 2.06 124.33 For distance 3, maximum divergence from expected among these 5 is 2.46, for byte 180. Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in 0 4 1104 9 4.31 2.09 122.67 - 45 4 1135 12 4.43 2.71 94.58 O 79 4 1117 9 4.36 2.06 124.11 134 4 1099 9 4.29 2.10 122.11 146 4 1148 10 4.48 2.23 114.80 151 4 1096 10 4.28 2.34 109.60 213 4 1196 11 4.67 2.35 108.73 216 4 1084 9 4.23 2.13 120.44 For distance 4, maximum divergence from expected among these 8 is 2.71, for byte 45. and so on ad nauseam. The following output results from applying the program to a file of size 283,870 bytes (the same size as the random file considered above) consisting of English text, with mostly uppercase letters and no digits (the space, ASCII 32, is excluded from the analysis): File GOLFDESC.GEN contains 283870 bytes and 71 different byte values. Maximum frequency of 67962 attained by byte value: 32 (0x20) Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in e 101 1 25842 1730 363.97 4.75 14.94 f 102 1 5311 305 74.80 4.08 17.41 l 108 1 11390 1903 160.42 11.86 5.99 o 111 1 15821 460 222.83 2.06 34.39 p 112 1 3743 136 52.72 2.58 27.52 For distance 1, maximum divergence from expected among these 5 is 11.86, for byte 108. Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in d 100 2 6985 232 98.38 2.36 30.11 e 101 2 25842 1313 363.97 3.61 19.68 i 105 2 12757 459 179.68 2.55 27.79 n 110 2 14144 411 199.21 2.06 34.41 s 115 2 14118 453 198.85 2.28 31.17 t 116 2 15774 466 222.17 2.10 33.85 For distance 2, maximum divergence from expected among these 6 is 3.61, for byte 101. Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in a 97 3 17139 816 241.39 3.38 21.00 e 101 3 25842 1626 363.97 4.47 15.89 g 103 3 5916 306 83.32 3.67 19.33 h 104 3 9876 422 139.10 3.03 23.40 i 105 3 12757 419 179.68 2.33 30.45 n 110 3 14144 787 199.21 3.95 17.97 t 116 3 15774 825 222.17 3.71 19.12 For distance 3, maximum divergence from expected among these 7 is 4.47, for byte 101. Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in a 97 4 17139 1771 241.39 7.34 9.68 c 99 4 6039 189 85.06 2.22 31.95 d 100 4 6985 206 98.38 2.09 33.91 e 101 4 25842 1764 363.97 4.85 14.65 h 104 4 9876 289 139.10 2.08 34.17 i 105 4 12757 563 179.68 3.13 22.66 l 108 4 11390 457 160.42 2.85 24.92 n 110 4 14144 835 199.21 4.19 16.94 o 111 4 15821 1238 222.83 5.56 12.78 r 114 4 14694 629 206.96 3.04 23.36 s 115 4 14118 736 198.85 3.70 19.18 t 116 4 15774 797 222.17 3.59 19.79 u 117 4 7041 433 99.17 4.37 16.26 For distance 4, maximum divergence from expected among these 13 is 7.34, for byte 97. Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in a 97 5 17139 1416 241.39 5.87 12.10 d 100 5 6985 250 98.38 2.54 27.94 e 101 5 25842 2493 363.97 6.85 10.37 h 104 5 9876 379 139.10 2.72 26.06 i 105 5 12757 726 179.68 4.04 17.57 l 108 5 11390 365 160.42 2.28 31.21 n 110 5 14144 880 199.21 4.42 16.07 o 111 5 15821 1229 222.83 5.52 12.87 r 114 5 14694 742 206.96 3.59 19.80 s 115 5 14118 485 198.85 2.44 29.11 t 116 5 15774 867 222.17 3.90 18.19 u 117 5 7041 200 99.17 2.02 35.21 w 119 5 3663 108 51.59 2.09 33.92 For distance 5, maximum divergence from expected among these 13 is 6.85, for byte 101. Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in a 97 6 17139 1082 241.39 4.48 15.84 e 101 6 25842 2585 363.97 7.10 10.00 h 104 6 9876 351 139.10 2.52 28.14 i 105 6 12757 519 179.68 2.89 24.58 l 108 6 11389 491 160.41 3.06 23.20 n 110 6 14144 569 199.21 2.86 24.86 o 111 6 15821 803 222.83 3.60 19.70 r 114 6 14694 829 206.96 4.01 17.72 s 115 6 14118 717 198.85 3.61 19.69 t 116 6 15774 839 222.17 3.78 18.80 For distance 6, maximum divergence from expected among these 10 is 7.10, for byte 101. Byte Distance Primary Actual Expected Diverg. Prob. value value occurr. secondary secondary 1 in a 97 7 17139 1006 241.39 4.17 17.04 d 100 7 6985 201 98.38 2.04 34.75 e 101 7 25841 2784 363.96 7.65 9.28 g 103 7 5916 173 83.32 2.08 34.20 h 104 7 9876 332 139.10 2.39 29.75 i 105 7 12757 505 179.68 2.81 25.26 l 108 7 11389 423 160.41 2.64 26.92 n 110 7 14144 685 199.21 3.44 20.65 o 111 7 15821 921 222.83 4.13 17.18 p 112 7 3743 144 52.72 2.73 25.99 r 114 7 14694 811 206.96 3.92 18.12 s 115 7 14118 700 198.85 3.52 20.17 t 116 7 15774 1099 222.17 4.95 14.35 For distance 7, maximum divergence from expected among these 13 is 7.65, for byte 101. The first part of this output from the program tells us that the letters e, f, l, o and p commonly occur in English doubled, which is consistent with our experience. The next part says that the letters d, e, i, n, s and t are often found making up the pattern b*b (for b=d,e,i,n,s,t), which is less evident, but not counter-intuitive. Letters occurring in the patterns b**b, b***b, b****b, etc., are usually a, e, h, l, n, o, r, s or t, with c, d, g, i, p, u and w also occurring. Presumably this simply reflects the comparative frequency of occurrence of letters in English text. When the text of the Voynich manuscript is available in machine-readable form I can apply this program (or some modification of it) to it. This will generate lots of numbers, as we have seen above (and we didn't even consider distances beyond 7). So John Baez and others who are interested in the statistical analysis of the Vms can now be thinking of how these numbers might be made to yield significance, and more generally, whether these numbers (or some others) are what we should be generating in our quest for some (statistical) answers regarding the nature of the Vms. From reeds@gauss.att.com Thu Jan 16 01:37:22 1992 Message-Id: <9201151637.AA21960@rand.org> From: reeds@gauss.att.com Date: Wed, 15 Jan 92 11:37:22 EST To: voynich@rand.org Status: OR How to write in gibberish: 0. Any combination of the below. 1. ``Automatic writing,'' a friend told me, is associated with hypnotism and with some forms of pseudo science or occultism. She could provide no references. Anyone know what this is? 2. Utter (speak) gibberish and transcribe. 2a. Glossolallia. Martin Joos believed that the VMS was written glossolallia. What is it? References? 2b. When I try to speak connected gibberish it comes out with rhyme and meter, like nursery rhymes or Walt Disney-speak: ``ibbity bibbity boo.'' 2c. Its easy for me to utter isolated nonsense words, but they usually turn out to be distortions of words I know: Norax sted thorax, Gamilsond sted Trebizond, etc. Jack Vance words. 2d. Speak in a jargon of one's own, larded with invented words. ``Please voop me the framistan, I'm fain grepped.'' The strange alphabet is then trusted to complete the deception. (This is close to Brumbaugh's notion, I think.) Cf. Lilliputian, both by Swift and by T. H. White. 3. Vertical copying: when part of a page is written, just copy or slightly change a word already appearing on the page. I suppose vertical copying from the line of writing directly above might be detectible by statistical means. I think the V authors did a lot of this. 4. Using apparatus: Random choice of a word from a list. Use of Lullian wheels to mix & match word roots and affixes. Two differing versions of apparatus might account for some of Currier's A/B variations. 5. Encrypting a real text with (say) a code book whose code groups are pronounceable. (Method 4 above is a debased form of this.) 6. Write down plain text or slightly distorted text in a strange language. [Hungarian pig latin, say, would look like gibberish, or Esperanto with a stage accent, to the uninitiated.] Erdos-ish: ``Pee nay ap pel le op seh dovn sakeh''. [But this is 2d above!] From RJB@MAX.U.WASHINGTON.EDU Thu Jan 16 03:30:00 1992 Date: Wed, 15 Jan 92 10:30 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: written and vocal automatisms To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Automatic writing: basically, writing without attending to the act of writing; may be done in any state from distraction to deep trance. May take the form of barely intelligible scribbles; may develop into well- formed and fluent text, with "personalities" communicating different points of view, etc. Most famous literary case: the raw material of Yeats' _Vision_, which began as automatic writing by Yeats' wife shortly after they were married. (Much has been published, with some facsimile pages, by George Mills Harper, if anyone's curious.) This sort of thing could conceivably lead to development of a script and "language" -- not unknown in the records of 19th-century mediumship. Famous case of a medium who contacted a "Martian" civilization. Was the book by Flournoy? I forget. Glossolalia: speaking in tongues. I think Felicitas Goodman has done a study of the linguistic traits of glossolalic productions, shown that they are not random, but follow certain constructive rules (none of which are semantic!). Or was it Erika Bourguignon? I forget. Maybe both. To me, speaking in tongues sounds like the affective component of language without the semantic or syntactic. It is conceivable that glossolalia could merge into either mediumistic trance speech, or into the gradual production of more structured (syntactic) "text". In other words, it is conceivable that, postulating the use of one or both of these devices, the Voynich could be a "fake" text generated without a specific intent to be fake. It might have been as much a mystery to its originator(s) as to anyone else! But the actual text we have of course is not the product of any automatisms. It is clearly, from all that has been reported here, the product of the bookmaker's craft. --rjb From RJB@MAX.U.WASHINGTON.EDU Thu Jan 16 03:37:00 1992 Date: Wed, 15 Jan 92 10:37 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: hypotheses about text generation To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR There are it is true many ways text could have been generated for the sake of having some text that would baffle readers, but look tempting enough to be bought. Type (4) also touches the philosophical language hypothesis. What about code rather than cipher? It will be good to be able to have a concordance of the "words," and to be able to see what the patterns in their occurrence are. --LeGrand From jbaez@math.mit.edu Thu Jan 16 05:03:55 1992 Date: Wed, 15 Jan 92 15:03:55 EST From: jbaez@math.mit.edu Message-Id: <9201152003.AA01113@jordan> To: voynich@rand.org Subject: News from Yale Status: OR Before I get to my real topic, congrats to Jim Gillogly for contacting Mary D'Imperio! She's my hero! If she actually has time to do more transcribing that would be great, but I would also love to see copies of her papers on the Voynich. Jim says that Jacques "typed in" the most important paper. Where is it available? I got an impressively wrapped package from Yale yesterday, and for a moment I was hoping against hope that it was the Voynich ms., but it turned out to be just an incredibly well-packed invoice. The letter first lists two books about the Voynich (D'Imperio's and Brumbaugh's) and then says Enclosed you will find 1. A photocopy of a typical page of the manuscript 2. A manuscript application which you may fill in and return with a check for $35.00 (overseas customers $42.50, drawn on an American bank) made payable to Yale University Library, if you would like to buy a positive microfilm of the entire manuscript. A Copyflo copy (enlarged paper printed from our master negative) costs $35.00 (overseas customers $42.50). Return the manuscript application to: Suzanne Danos Research Librarian Beinecke Library ------- The typical pages (3 actually) are 56, with a beautiful bizarre plant on it, 100, with quite a number of herbs, including "peppers," and one "apothecary jar", and the page before page 100 (I don't see the 99 on it), which has four "apothecary jars" lined up vertically on the left side, and a large variety of flowers (or are they roots?). The letters are legible enough to transcribe easily, but not stunningly clear. The application says "If permission is granted I agree to comply with the Library's rules governing the use of such materials, including the requirement that Yale University Library manuscripts may not be published in whole or in part unnless such publication is specifically authorized." This makes me queasy about bringing Aegean Park Press in on anything that could be construed as "publication," though I have no qualms about making a bunch of copies. The form also asks the subject, scope, and purpose of my research. I will post my reply when I write it, just in case anyone else getting a copy from them wants a blurb. There is also a short blurb describing the manuscript: Manuscript on vellum, 102 leaves, with 400 botanical drawings, 350 single star figures, and 42 biological drawings, all in shades of green, brown, yellow, blue and red. (Etc.) And there is a "Selected Bibliography" about the Voynich. I will check to see if it includes things not in D'Imperio's bibliography. Here's one I should try to get: Bolton, Henry C. Follies of Science at the Court of Rudolph II. Milwaukee, 1904. So -- slowly, slowly, the Voynich draws nigh! From jim@rand.org Thu Jan 16 05:13:27 1992 Message-Id: <9201152013.AA02050@mycroft.rand.org> To: voynich@rand.org Subject: Currier paper (from 1976 Seminar) In-Reply-To: Your message of Wed, 15 Jan 92 15:03:55 -0500. <9201152003.AA01113@jordan> From: Jim Gillogly Reply-To: jim@rand.org Date: Wed, 15 Jan 92 12:13:27 PST Sender: jim%mycroft@rand.org Status: OR > jbaez@math.mit.edu writes: > Jim says that Jacques "typed in" the most > important paper. Where is it available? The anonymous ftp directory rand.org:pub/jim has all of the commonly-needed materials that we have on-line. Feel free to send me suggestions for other things to put in. The directory is in file ReadMe.dir, and the Currier paper is called currier.paper. Strongly recommended! It doesn't include the tables, though... one of us ought to type them in as well. In the meantime, if you need them, write to me and I'll send them out by physical mail. Jim Gillogly From jbaez@math.mit.edu Thu Jan 16 07:15:58 1992 Date: Wed, 15 Jan 92 17:15:58 EST From: jbaez@math.mit.edu Message-Id: <9201152215.AA01266@jordan> To: voynich@rand.org Status: OR Ron Carter was lusting after "Renaissance Curiosa" by Wayne Shumaker (pub. Medieval & Renaissance Texts and Studies, Center for Medieval and Early Renaissance Studies, Binghampton, New York 1982), and now I have it. It does not *include* "A True and Faithful Relation of What Passed for Many Yeers Between Dr. John Dee (a Mathematician of Great Fame in Q. Eliz. and King James their Reigns) and Some Spirits: Tending (had it succeeded) To a General Alteration of Most States and Kingdomes in the World" (whew!), Johannes Trithemius' "Steganography," and George Dalgarno's "Ars Signorum. Rather, it includes analyses of these (and Cardano's horoscope of Christ). Shumaker boastfully emphasizes that what these texts have in common is weirdness and DIFFICULTY. Anyone who can get this should, in part because it's a great read --- for people of our perverse tastes. I'll briefly summarize his main point about Trithemius. His Steganographia has long been thought by some to be a weird mixture of cryptographical and thaumaturgical manuals. Even Yates says in "The Rosicrucian Enlightenment" that Trithemius messed with spirits. Trithemius claimed that this was untrue. But his book is full of recipes like the following: If you want to secretly transmit a message, you may elect to appeal to Pamersiel, one of the malicious and untrustworthy spirits who form the majority. Write down the message, beginning by invoking the Trinity and using ordinary language (or Latin), facing east as you do so. Then "compel" the spirit by saying: Pamersiel anoyr madrisel ebrasothean abrulges itrasbiel nadres ormenu itules rablon hamorphiel. If the spirit does not come, repeat until he does; then give him the message and your directions. The recipient must also be skilled in the art and upon receiving the communication he should face east and recite a similar conjuration. Having done so, he will instantly understand your meaning. However, you have to remember to include in the message the name of the communicating spirit, or the recipient will not know who he's dealing with. Naturally, this sort of thing got Trithemius in trouble. One fellow, Carolus Bovillus, wrote that he looked at the "Steganography," and... "I had the book in my hands scarcely two hours before I threw it away on the spot because such great wonders ans such barbarous and strange names of spirits - not to say devils - had begun to terrify me." However, Trithemius was just being too tricky for his own good. The "conjuration" above is really just a description of how a cipher works. I think I'll leave this as a puzzle, but I'll give you a big clue. When you read the "conjuration" correctly it says (in a mixture of German and Latin) "Take the first letters of every word." So, the puzzle is: how do read the conjuration? The conjuration to be spoken by the recipient is just another key, designed to guide him in deciphering. A sample of text coded this way is given by Trithemius... Lvcidum jvbar aeternate Beatitvdinis, Excellentissime Rex, Gvbernator & Tvtor robvstissime, Vniversorvm virtvose viventum, exvlvm refvgium... If we read the first letters (correctly :-) you get "Ljaeber G&truvver". I'm sure you're thinking, "great!" But "j", you see, is alternative for "i", "ae" is equivalent to "e", "&" means "e" and "vv" is just "w". So it means "Lieber Getruwer", which is an archaic form of "Lieber Getreuer." Proceeding, we would get: Lieber getruwer / du wollest uf nest Mantag gerust sin so du aller baest vermagst / und umb die funf / unser an der Lantporten warten / da willen wir / mit userm gezug erschinen. This is old-fashioned German for Dear Faithful One: You will be armed as best you can next Monday and about five will wait for us at the gate; we will appear there with our followers. This is one of the simpler methods described by Trithemius and it seems to make sense to familiarize ourselves with all of them. From RJB@MAX.U.WASHINGTON.EDU Thu Jan 16 09:18:00 1992 Date: Wed, 15 Jan 92 16:18 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Re: News From Yale To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Yes, if APP were to do a facsimile edition we would have to negotiate with Yale. Or someone would have to negotiate with Yale. Now, it may not be worth it to APP to (a) negotiate with Yale and (b) set up enough distribution (eg through Phanes) to make it worth their while. But it may, and it seems to me there would be a market --especially if the facsimile had some good prefatory and explanatory material.. Especially since they have two other Voynich things in their list. And since they have (so to speak) priority. On the other hand, if they decide not to, then some other channel might be available. --RJB From rcarter@isis.cs.du.edu Thu Jan 16 09:38:48 1992 Date: Wed, 15 Jan 92 17:38:48 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9201160038.AA00522@isis.cs.du.edu> To: voynich@rand.org Subject: Errata on ordering Voynich... Status: OR Two more pieces of info for ordering the Voynich; Make the check/money order payable to: Yale University and the $40 stated is $40US. From reeds@gauss.att.com Thu Jan 16 12:16:53 1992 Message-Id: <9201160317.AA06745@rand.org> From: reeds@gauss.att.com Date: Wed, 15 Jan 92 22:16:53 EST To: voynich@rand.org Status: OR I will type in Currier's tables, in so far as I can read them. It will take a while, though! From jbaez@math.mit.edu Thu Jan 16 13:21:32 1992 Date: Wed, 15 Jan 92 23:21:32 EST From: jbaez@math.mit.edu Message-Id: <9201160421.AA01887@nevanlinna> To: voynich@rand.org Status: OR Here is the text of my "research proposal" to Yale; probably overkill, but if you order something from them you are welcome to borrow from it, though an almost exact copy might make them think they're the victims of a cult or chain letter: Dear Ms. Suzanne Danos: An informal group of cryptologists, linguists, mathematicians, programmers and statisticians, including Mary D'Imperio (author of The Voynich Manuscript: An Elegant Enigma), is attempting to determine the nature of the Voynich manuscript. Previous researchers were unable to bring to bear today's computer resources to the problem. We intend to use the computer as a tool for statistical analysis of the text. While the text has proved notoriously difficult to decode (if in fact it is in code), it may be possible to make progress in understanding the nature of the text through such methods. I would like a copy of the text in order to transcribe parts into computer-readable form. Sincerely, John C. Baez ---- I thought it best to avoid using a name for our group. By the way, I found "International Voynich Manuscript .... Group" to be pretentious, since there are no national ones. Why not "Intergalactic"? :-) From naga@wet.net.netcom.com Thu Jan 16 21:10:00 1992 Message-Id: Date: Thu, 16 Jan 92 04:10 PST From: naga@wet.net.netcom.com (Peter Davidson) To: voynich@rand.org Subjsome members from outside the galaxy. I suppose for them this is as good a way as any to pass the time until the mother ship returns. Their knowledge of extraterrestrial linguistics may well prove invaluable ... in fact they may even be X. Status: OR From RJB@MAX.U.WASHINGTON.EDU Fri Jan 17 01:54:00 1992 Date: Thu, 16 Jan 92 08:54 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: International Pretentious Vms Study Group To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Well, correct academic style involves a nice balance between the pretentious and the self-deprecating; the group name may need some fine-tuning... Actually, though, like the International Monetary Fund, the group *is* international (unless we are subsuming Australia into the Nation of Former British Colinies!) -- and there have been *non*- international Voynich groups before. This is the first Voynich group that has been international (unless one counts the informal Kircher nexus). But First International Voynich Group would sound too jmuch like a bank. I, frankly, ahve always found the British practice of calling groups "The" without further qualification (e.g., The Bibliographical Society) far more pretentious. Some sort of qualifier is (I think) desirable. Perhaps The Internet Vms Study Group? If, that is, there is to be a name at all... I met perhaps the ultimate such British group in London a few years ago. It called itself The Society. (No one could agree on just what sort of qualifier would be appropriate.) From firth@SEI.CMU.EDU Fri Jan 17 03:11:03 1992 Date: Thu, 16 Jan 92 13:11:03 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201161811.AA08765@bp.sei.cmu.edu> To: voynich@rand.org Subject: Lines in the Vms Status: OR Folks This is just a brief note, but the idea came to me while reading the CXurrier paper, and I had to share it. Currier claims that, in the Voynich MS, the "line" is a functional unit, and he gives some statistical evidence for this claim. He also says he has no explanation or hypothesis to account for this. That's strange, because my subconscious came up with an hypothesis in about two seconds flat, and it seems to me a pretty obvious one: the MS is poetry. Now, that does not square with the short lines at the ends of paragraphs,\ but then, neither does Currier's claim. Incidentally, I had already noticed what seems to be more repetition of word endings at the ends of lines; I'd be hesitant to claim evidence of rimed verse, though. Just another wild conjecture from Robert From j.guy@trl.OZ.AU Fri Jan 17 05:09:34 1992 Date: Thu, 16 Jan 92 15:09:34 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201160409.AA16502@medici.trl.OZ.AU> To: voynich@rand.org Subject: Currier's tables Status: OR To all, but mostly Jim to Reeds: don't. I had started typing them in and got a fair way through when, suddenly, I had *urgent* things to look at, very *urgent*. They will cease to be urgent tomorrow (Friday 17) by 1 p.m. at the latest. So I'll be able to finish typing those tables in (there isn't much left to do). Back to the urgent grindstone the noo. From jbaez@math.mit.edu Sat Jan 18 00:51:38 1992 Date: Fri, 17 Jan 92 10:51:38 EST From: jbaez@math.mit.edu Message-Id: <9201171551.AA23468@euclid> To: voynich@rand.org Subject: Guy's latest thesis Status: OR While I found Guy's latest thesis fascinating, I have one factual quibble: it's no fair to bring in Friedman as claiming that the Voynich was in a natural language; in fact, he thought it was an artificial language of the "a priori type". (To digress, I will type in a bit about Dalgarno's interesting language when I have "Renaissance Curiosa" on me. This is quite interesting, but would not seem to have the "word-bridging" properties that are so crucial.) I can imagine coding schemes that would give rise both to "word-bridging" and "line-initial" phenomena -- quite easily -- and so the mere presence of these phenomena seems very weak evidence for it being in a NATURAL language. Especially if we have to go as far afield as Pictish and Sanskrit, or a secret Neanderthal cult. :-) So, Jacques, what do you see about these phenomena that seems to indicate a natural language, rather than a cipher? Anyway, I think we ought to fasten on these phenomena and try to understand them well, in particular the common features vs. the differences between how they work in A and B. jb From cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET Sat Jan 18 01:19:06 1992 Message-Id: From: cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET (John Cowan) Subject: Re: Lines in the Vms To: voynich@rand.org (Voynich List) Date: Fri, 17 Jan 92 11:19:06 EST In-Reply-To: <9201161811.AA08765@bp.sei.cmu.edu>; from "SEI.CMU.EDU!firth" at Jan 16, 92 1:11 pm X-Mailer: ELM [version 2.3 PL11] Status: OR firth@sei.cmu.edu writes: > [M]y subconscious came up with an > hypothesis in about two seconds flat, and it seems to me > a pretty obvious one: the MS is poetry. Now, that does > not square with the short lines at the ends of paragraphs [...] Short lines (or for that matter) long lines are far from unheard of at the end of >verse< paragraphs. If the VMS is a poem, it is an epic-length poem. Now some such poems, like the >Iliad< or >Paradise Lost<, consist of only one kind of line repeated throughout. But there are other epos poems like >Sir Gawain and the Green Knight<, which has verse paragraphs of varying lengths marked off by short lines, or Spenser's >Faerie Queene<, which has stanzas of fixed length marked off by long lines. (There are many other examples in the poetry of all languages: I cite merely what I know.) This difference in form is closely correlated, in English anyway, with a difference in content. Epics with a single metrical staple tend to be more story-oriented, whereas epics with paragraphs tend to be more reflective; the long or short lines invite the reader to stop and meditate on what is being said rather than to be pulled into page-turning. (Of course there are exceptions.) I offer these speculations for whatever assistance they may eventually provide. -- cowan@snark.thyrsus.com ...!uunet!cbmvax!snark!cowan e'osai ko sarji la lojban From RJB@MAX.U.WASHINGTON.EDU Sat Jan 18 02:19:00 1992 Date: Fri, 17 Jan 92 09:19 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: J Guy's inferences from Currier's statistics (brief) To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR (1) Another possibility. I've just had a chance to look at a biography (in Hmong and English) of the Hmong prophet called the "Mother of Writing" (though male; "mother" here meaning "source" or "originator"). The editors/translators of the text state that his is the only case known of an illiterate inventing an *alphabetic* writing for a previously unwritten language (as opposed to a syllabary). One odd feature of the writing system: the order in which the letters of a syllable are written does not seem to be the order in which they are pronounced. If it weren't for the clear familiarity with the conventions of the book-makers craft, one might be tempted to suppose that the Voynich script was a perhaps alphabetic writing system devised from first principles by some person(s) who came up with idiosyncratic ways of using letter-forms similar to those they have seen but with very different meanings. Of course, the Voynich text could be the last surviving example of a textual tradition, made by scribes who were (so to speak) bicultural. The argument from silence (that there is no other evidence of such a thing) is unclear. The documentation of the period outside surviving high-culture arenas is sparse enough, and a good bit of what does survive is neglected enough, that evidence might in fact be available if one knew where to look. If I could only find those notes I made almost 20 years ago on Hildegard of Bingen's "unknown language," which as I remember involved non-roman but romanoid letter-forms used to write "mysterious words" that were actually those of some local germanic dialect. If my memory is correct, it would be at least establish that, at least in a weaker form, at least once, a writing system had been devised for a local vernacular. Perhaps such enterprises were actually not uncommon... If that were true, what sort of thing would people be likely to write? Business records? Family testaments? Family recipe books covering bits of useful knowledge (medicine, farming, kitchen/workshop processes)? Almanacs or calendars? Charms and prognistications? Bits of local lore and legend? o --rjb From j.guy@trl.OZ.AU Sat Jan 18 04:00:17 1992 Date: Fri, 17 Jan 92 14:00:17 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201170300.AA17761@medici.trl.OZ.AU> To: voynich@rand.org Subject: On Currier's 1976 papers Status: OR Some (Serious) Speculations about the Nature of the Language of the Voynich Manuscript Summary: after some ramblings, designed to outrage, bewilder and infuriate, some hard evidence, from Currier's statistical observations, that we are in the presence of something written in a real language that will definitely be a tough nut to crack. Preliminary Ramblings. When I do not have my tongue in my cheek, claiming the Voynich Manuscript to be written now in Nahuatl, now in some Southern Chinese dialect, now in Tibetan, and most recently in Sabir, I am of the opinion that it is likely to be a case of written glossolalia. Rather, so I was, until I read Prescott Currier's observations. They led me to revise my opinion completely. I now think that: 1. The manuscript was written by several authors. Oh, how I wish I had a good reproduction of it! The scattered 10x18, white on black photographic enlargements I made of the fuzzy Yale microfilm are too few and too awful to do serious work. As things stand, I was much impressed by Currier's tale of how he just *saw* those two different handwritings and how frequency counts corroborated those differences. So much so that I am prepared, sight unseen, to take his claim that there were at least four people, possibly a dozen, involved in writing the Voynich. Well, to me that means: 2. The text is meaningful. It is not glossolalia. Certainly, it might be stuff and nonsense, Lobsang Rampa, Von Daeniken or Book of Urantia-style, but it *means* something. It is not pointless babble. 3. The language is real, natural, it is neither invented, nor artificial, nor philosophical a la Delgarno or Bishop Wilkins. Hard to swallow, isn't it? But that is what the statistical properties discovered by Currier suggest to me without the shadow of a doubt. Needless to say, there is no cipher there, and I am in good company asserting that: William Friedman had come to the same conclusion. I am fairly convinced now of the three points I made above. I am somewhat less confident in what follows: the Voynich was written around the 12th century (viz influence of the Beneventan script) in Europe, probably in Italy, Southern France, or Northern Spain (the Beneventan script again), in a (careful there!) in a language now extinct, indigenous to Europe, but non-Indo-European. Since I am not too convinced of that myself, I'd better explain what makes me think so. First, if the language was Romance, Germanic, Slavic, in short, Indo-European, it would have been identified by now, and the riddle solved. Second, from what I have seen of the manuscript, and analyzed, the language "feels" monosyllabic, i.e. could be Sino-Tibetan, or... Sumerian! Oh yes, that is very very shaky ground indeed to build upon, but in my eyes, it rules out possible non-Indo-European candidates such as Basque (of which I know a little), Finnish (of which I know a little too) with its relatives Hungarian, Turkish, and Estonian, and Georgian (of which I know barely more than the alphabet!). What point am I trying to make then? Forget about identifying the language; it is almost certain that it is extinct, and was unrelated to anything we know. On the other hand, I believe that it is possible to decipher the Voynich manuscript, just because there is so much of it (a consequence of Sukhotin's theory of decipherment algorithms). Tongue, take thee back to thy rightful abode, my cheek, and stay firmly there that I may tell the world of this my stupendous finding about the Voynich manuscript: the Voynich manuscript is entirely written in the Pictish language. Since Pictish has been extinct since around the days of Julius Caesar, and nothing, absolutely nothing is left of it, except that Conan the Barbarian could speak some, I don't see any of you out there successfully challenging my momentous discovery. Thank you, tongue, at ease. Seriously Now, Folks. Currier observed that letter frequencies varied accordingly to their position in the line, being quite different line-finally from word-finally, line-initially from word-initially. He gives an instance of this phenomenon, based on frequency counts from Herbal A (roughly 6500 words in 1000 lines): "word"-initial total frequency symbols "word"-initially line-initially 118 3 212 26 24 0 45 10 There is indeed something quite strange going on there. With an average of 6.5 words or so per line, we should expect to see about 15% of those word-initials occurring at the beginning of a line. Thus: "word"-initial total frequency symbols "word"-initially line-initially 118 18 212 33 24 4 45 7 The discrepancies between expected and observed frequencies do not worry me much, except in the case of : 3 occurrences observed when 18 are expected is enough to catch my attention too. Consider also that we should expect those four common word-initials, totalling 399 cases, to occur 61 times or so line-initially. We find them there only 39 times. And indeed Currier remarked in the afternoon discussion of the 1976 seminar: "An annoying little circumstance: words beginning with almost never seem to occur first in a line. I thought perhaps I might try numerals one to ten for the letters that come before in line-initial position, but I can't make it work. But this kind of thing makes it look as if the line is a functional entity; that is what bothers me. I can't interpret the data!" When I read those remarks I thought to myself: here is very strong evidence that the Voynich manuscript cannot be a hoax, contrary to what I had been thinking all along. For the phenomena observed and described by Currier are typical of what I have been taught -- through bitter experience -- to expect from languages scattered all over the world. I shall draw upon rather exotic languages later, but allow me to start with one of our neighbours: Spanish. I was absolutely astonished when, studying Spanish in high school we were taught that this verse by I forgot whom, amounted to only eleven syllables: Vio' a Eugenio, !ay dolor, pena y llanto! herido. Yes, here they are: 1. Vio' a Eu 2. ge 3. nio ay 4. do 5. lor 6. pe 7. na y 8. llan 9. to he (remember: h is always silent in Spanish) 10. ri 11. do So, in Spanish poetry, a sequence of vowel sounds, however lengthy, counts for only one syllable. The same is true of Latin: Nox est perpetua una dormienda ^^^^ one syllable with this added quirk that a word-final "m" counts as if it were not there, if I may say. Thus: perpertuam unam ^^^^^ still only one syllable! (I have often wondered about that: could it be that word-final "m" was not a true consonant, but just a way of indicating that the previous vowel was nasalized?) Let us go back to Spanish (we'll move on, appropriately, to Arabic next). The very interesting thing is, although the rule is that a sequence of vowel sounds counts for only one syllable, it does NOT extend across lines in a poem. Thus in [blah blah blah]do Y [blah blah blah] "do y" counts for TWO syllables ("do" and "y" with whatever vowels might follow). It is the pause at the end of the verse that prevents the running-together of the vowels into a single syllable, or so I see it. Now to Arabic. One of the very first examples in the excellent "Teach Yourself Arabic" is (perhaps not "pedagogically correct" I suppose, but I am sure that the Voynich manuscript predates such notions): daraba al-shaykhu ra'sa al-waladi. hit the-old man head (acc.) the-boy (gen.) "The old man hit the boy on the head." That is how it is written, but this is how it is pronounced: daraba sh-shaykhu ra'sa l-walad. What I want to draw your attention to especially is the loss of the genitive ending (-i), sentence-finally, or rather better said, before a pause in the speech. But you will have noticed of course, the fairly considerable changes in the rest of the sentence: loss of the initial "a" of the definite article "al", assimilation of its "l" to a following "sh". So, if Arabic were written precisely as it is spoken we would observe discrepancies between word-final and sentence-final, word-initial and sentence-initial letter frequencies. I shall now turn to Sanskrit for more evidence of the awesome deviousness of language, therein disguised under the innocuous name of "external sandhi". What does it mean? That speakers of Sanskrit ran their words together to an awesome extent. My two favourite examples, taken from "Grammaire du Sanskrit" (Collection Que Sais-je? Presses Universitaires de France) are: 1. "Here, the king made a sacrifice": raajaa iha iije, which, after running the words together (applying the sandhi rules, if you want to be pedantic about it) becomes (wait for it): raajeheje 2. "O, Sun, do rise! the wise man said": Suurya, udihi! iti aaha acharyas, which becomes: Suuryodihiityaahaachaaryah But we have also: Saashvevaadati, really made up of Saa, ashvaa, iva, and adati, meaning "she eats like a horse" (in which Latinists may have recognized adati = edo, edis, edere). What makes Arabic easy is that, even though words are somewhat run together, they still get written individually. What makes Sanskrit awfully difficult is that it is written precisely like it was spoken: a whole sentence as one long word, taking either the sleuthing abilities of Sherlock Holmes to disentangle it into its constituents, or, much better, a complete fluency with spoken Sanskrit brought by many years of practice (but who speaks Sanskrit fluently nowadays?). Now I want to quote verbatim what Jean Varenne, the author of that Sanskrit grammar, has to say on external sandhi: Repetons que les pandits eux-memes, pourtant nourris des l'enfance de la langue "divine", evitent de telles equivoques. Car personne ne peut comprendre sans quelques instants de reflexions (ou mieux sans faire repeter l'expression par l'interlocuteur) des rebus du genre rajeheje ou suryodihiityaahaachaaryah. (Let me say again that even pundits, although fed the "divine" tongue from their earliest childhood, eschew such ambiguities. For no-one can understand without some thought (or, better, without asking their interlocutor to repeat) puzzles of the kind of rajeheje or suryodihiityaahaachaaryah.) Why do I attach such importance to that quote? Because it means that the very same Sanskrit sentence will be written differently by various writers, depending on their fluency in the language. Some, the most fluent, will write "raajehije"; some "raajaa iheje"; some "raajeha iije"; some again "raajaa iha iije". Same language, very divergent renditions. Thus, strict, well-defined rules decide how words are run together in Sanskrit, but they are applied by different writers with various degrees of thoroughness and competence. Imagine now a language with such tendencies as Sanskrit; where words are run together, except, naturally enough, where a pause occurs in speech. As you, a scribe, write in this language, I dare say that you will tend to pause long enough at the end of a line so that sandhi rules cease to apply. The statistical properties of what you write will, I think, be quite similar to those observed by Currier in the Voynich manuscript. Letter frequencies at the beginning and at the end of lines will differ from anywhere else. Now, in my letter to the editor of Cryptologia, written long before I knew of D'Imperio's book and Currier's findings, I mentioned that I had submitted some of the Voynich text to Sukhotin's algorithm for finding individual words in a text written without spaces, and that the results showed frequency distributions that suggested to me a language with a complex morphology or holophrastic tendencies. Holophrastic tendencies: in plain English, tendencies to run the words of a sentence together into one single word. Currier also found that "some "word"-finals have an obvious and statistically-significant effect on the initial symbol of a following "word." Thus: "word" beginning with: is preceded by <4o> "word" ending in: or <2> or series 13 7 91 <2> series 10 2 68 series 23 0 275 <9> series 592 184 168 Currier comments: "Words" ending in the <9> sort of symbol, which is very frequent, are followed about four times as often by "words" beginning with <4o>. That is a fact, and it holds true throughout the entire twenty pages of "Biological B." It's something that has to be considered by anyone who does any work on the manuscript. These phenomena are *consistent*, *statistically significant*, and hold true throughout those areas of text where they are found. I can think of no linguistic explanation for this sort of phenomenon, not if we are dealing with words or phrases, or the syntax of a language where suffixes are present. There is a very pedestrian explanation: the spaces between "words" do not show word breaks. First, if the Voynich language is, like Sanskrit, subject to extensive external sandhi, and written as it was spoken, there are no "words" as such, for each sentence become a single "word" in itself. Second, putting separating spaces between words is a fairly recent Western invention. Greek and Latin manuscripts show none until the Middle Ages. Third, we find certain Voynich letters almost exclusively line-finally (e.g. ), others almost exclusively word-initially (e.g. <4>), others word-finally (e.g. , , , <9>). This suggests to me either that spaces were regularly inserted after certain letters for purely aesthetic reasons, or that syllables, not words, were separated by spaces. Either hypothesis explains why "word" endings are so limited, and why those letters found ending "words" are rarely found starting them. Could then the Voynich language be one of those artificial languages, like Volapuk and Esperanto, or philosophical, like Bishop Wilkins's and, closer to us, Lojban? Definitely not, for such artificial languages purposefully avoid the complexities evidenced by the statistical properties discovered by Currier. Enochian? Forget it. Enochian is a fabrication by an early days Uri Geller. To invent the language of the Voynich manuscript would have required a knowledge of linguistics which I doubt existed from 1000 to 1700 AD, to make up an *unnecessarily* complex language based on intricate rules, and, last but not least, to teach its rudiments to at least four people, so that they be fluent enough to write pages upon pages in it. Pull the other one, mate, it's got bells on. In Her Majesty's English: a rather improbable thesis. From wet!naga@cca.ucsf.EDU Sun Jan 19 00:16:00 1992 Message-Id: Date: Sat, 18 Jan 92 07:16 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Sandhi etc. Status: OR > From: ucsfcca!trl.OZ.AU!j.guy (Jacques Guy) > > > Some (Serious) Speculations > about the Nature of the Language > of the Voynich Manuscript > > 2. The text is meaningful. It is not glossolalia. I am informed by someone whose vocabulary is, I suspect, twice that of my own that the name for the written analog of glossolalia is "glyptoglossia". As I said in an earlier posting, given Kelley's mediumistic talents this can't be ruled out (at least, in the absence of a solution). > 3. The language is real, natural, it is neither invented, nor artificial, > ... there is no > there, and I am in good company asserting that: William Friedman had > come to the same conclusion. The Brumbaugh book suggests otherwise. Does Jacques know something we don't? > I shall now turn to Sanskrit for more evidence of the awesome deviousness > of language, therein disguised under the innocuous name of "external > sandhi". I found Jacques' suggestion intriguing that we may be dealing with something as complex as Sanskrit, since I had the privilege, 20 years ago, of studying Sanskrit for a year under the direction of a professor of Classical Studies. (The course was not exactly popular, in fact I was the only student in the class. I found it rather interesting. Even now I remember, vaguely, the story of Nala and Damayanti.) Certainly, as Jacques says, the analysis of a phrase in Sanskrit, to identify just what the constituent words are, is often far from obvious. If we are going to use statistical analysis to identify the linguistic elements of Voynich then I suppose we should use analytical tools that can be shown to yield results when applied to various languages, such as Sanskrit. I mean, if we developed some statistical technique for analysing the Vms, how could we know that the results we claim to be getting are revealing something about the object of study rather than something about the technique for studying it? Perhaps only by showing that the same techniques applied to other languages yield results that are known to be true on other grounds. This fits in with something I've long been wanting to do, which is to use my Sanskrit notes (recording the wisdom and insight of my Sanskrit teacher) as a basis for implementing a computerized system of translating from Sanskrit into English. I haven't given this much serious thought yet, but I suppose an initial component of this system would have to be a means for analysing the text so as to identify the words in a phrase, based on the rules of sandhi (external - I'm a little vague now on what internal sandhi is). Another project that has long been awaiting the leisure to pursue it. (Anyone know of any research grants being offered for this kind of thing?) From jim%mycroft@rand.org Sun Jan 19 08:32:46 1992 Message-Id: <9201182332.AA01769@mycroft.rand.org> To: voynich@rand.org Subject: Don't forget "michiton oladabas" (f 116v) Date: Sat, 18 Jan 92 15:32:46 PST From: Jim Gillogly Status: OR If you propose that the V Ms is perfectly innocent plaintext in an unknown script, you also need an explanation for the last page of the manuscript. As you recall, it has some pictures on the left, including a standard nymph with hat (suggesting that it's part of the original document), and the "key", as Newbold calls it. The Voynich language part is the only part that I feel confident transcribing, but it looks kind of like: + michiton oladaba8 + multo8 + **c + t* +ccrc + porta8 + n + six + ****x + morix + vix + a*** + ma+** + AROR ZCC9 val8c[nh] vbren so nim gaf mich o The capitalized words on the last line are clearly Voynich script, and it's mixed in with the stuff that looks to me to be cryptic (the last bit looks like German and has been given several meanings -- all cryptic... maybe Jacques could clarify?). Brumbaugh reads the first word as "michicon" but I think that's unsupported, considering the similarity with the (assumed) "t" in the (assumed) words "multos" and "portas". AROR and ZCC9 are both legal V words, the latter quite common. There are lots of different ways to take this. Here are a few: (1) It's meant by the author of the V Ms to be cryptic, indicating that with some effort the reader might be able to figure this part out, and perhaps the rest of the Ms. (2) It's some sort of crib/key that will allow entry into the system. (3) It's in the author's own cipher/code/steganography/whatever and is meaningful only to him. (4) It was added by a later author on an otherwise blank page, either a) as a cryptic hook as in (1) above, or b) as notes on his findings. I suppose option (4) is possible in the "real but maybe unknown language in ditto script", but it doesn't look like one of the likelier choices to me. The important part to me is that it appears to be a mixture of Latin, Voynich, and German. Jim From reeds@gauss.att.com Mon Jan 20 01:37:00 1992 Message-Id: <9201191637.AA24342@rand.org> From: reeds@gauss.att.com Date: Sun, 19 Jan 92 11:37:00 EST To: voynich@rand.org Status: OR I have completed the tables at the back of Currier's paper. There are complete Unix tbl/troff source files with formatting for the Currier papers and his tables, together with a PostScript Voynich font, so you can print it all out nicely. I included the text of the main papers, duplicating Jacques's "currier.paper" transcription, because (1) I added formatting commands, and (2) it gives me a chance to correct the very few few transcription errors I found. You can get them by ftp: in file currier.sh.Z in directory pub/jim on rand.org. Or, if you do not have ftp access, I can send them to you by direct email. The file must be uncompressed to obtain currier.sh, which is a "sharkive" containing the constituent files. If you don't have (say) troff, you can still extract the textual and tabular data from the files: it is a tedious but straightforward task. Send me email if you have any questions or problems. Jim Reeds reeds@research.att.com From jim%mycroft@rand.org Mon Jan 20 14:54:57 1992 Message-Id: <9201200554.AA00526@mycroft.rand.org> To: voynich@rand.org Subject: Reproducing the Voynich Summary: Forget it Keywords: Dead Sea Scrolls Date: Sun, 19 Jan 92 21:54:57 PST From: Jim Gillogly Status: OR I got a letter from Robert Babcock at the Beinecke acknowledging receipt of my order. Here's his response to my question about whether they'd mind if somebody published a facsimile edition of the Voynich: 1) It is expressly forbidden for anyone to reproduce any images from manuscripts in the Yale Libraries without our written permission and the prepayment of reproduction fees. 2) No one is authorized to negotiate with publishers for the reproduction of materials in the Beinecke Library except our employees or representatives. 3) We will not allow any reproductions to be made from the microfilms or print-outs which we supply to readers. These are for study purposes only, and are only supplied to readers who agree to abide by our regulations concerning their use. 4) If a publisher is intereste in contacting us about making a facsimile of the Voynich manuscript, they may do so. We would require substantial control of many aspects of the publication (including photography, quality and format of reproduction, etc.), per image fees, and a substantial percentage of royalties from the sale of a facsimile. 5) We have had discussions with various firms in the last few years about making a facsimile of the Voynich manuscript, and some of these discussions are on-going. But we are not, in principle, opposed to discussing the matter with other publishers. Please contact me if you wish to pursue the matter further. Sincerely yours, Robert Babcock, Curator ------------- No, I think that about covers it. I sympathize with much of it, but point three also covers one of us making copies for others of us. Or a duplicate copy for our own backup or scratch copy. Really now... Jim Gillogly, Researcher From reeds@gauss.att.com Tue Jan 21 08:03:03 1992 Message-Id: <9201202303.AA18934@rand.org> From: reeds@gauss.att.com Date: Mon, 20 Jan 92 18:03:03 EST To: voynich@rand.org Status: OR Lou Kruh loaned me his Voynich file. Included in it is a letter from Robert S. Brumbaugh to Kruh (dated October 17, 1978) and three issues of Brumbaugh's "Voynich Newsletter", dated February 1978, November 1978, and January 1980. (Of 6, 7, and 7 pages, respectively.) What makes these different from any of his other writings is the presence of actual transcriptions of V characters in alphabet charts, etc. I will study this material and summarize to the group shortly. Jim From RJB@MAX.U.WASHINGTON.EDU Tue Jan 21 09:54:00 1992 Date: Mon, 20 Jan 92 16:54 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: No way To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Many of these points are unexceptionable. (1) is a statement of the legal situation (no reproduction without permission or fees, especially for commercial purposes). (2) is obvious, since the only parties with direct interest are the publishers and the Beinecke authorities. (anyone not a Beinecke official negotiating for Yale is then in the other camp) (4) is understandable and sensible enough; from what I've heard about the quality of the study-purposes microfilm, I wouldn't want some treasure I owned to be represented on the market by a reproduction made from it. (5) is intriguing. There are several potential publishers out there. Nice to know. One can only hope that they work out some settlement relatively soon. (3) is in fact rather oddly stated. I cannot imagine that making a back-up copy for one's own use is covered, simply because (I'm thinking of paper copies here) scholarly use entails, often enough, using up a copy or more in such normal activities as indexing (cutting and pasting onto cards, for example), annotation, and so on. Whether or not it would cover making copies for (e.g.) a seminar -- one copy made by and for each participant -- is harder to say. One might very well feel honor-bound not to make copies for others' use, but I can hardly see any ethical (or even dare I say it legal) basis for objecting to someone with a legitimate microfilm copy going to the university copy shop and having work prints made from it, as long as they were only for that person's use. If one happened to have a microfilm reader-printer at home, I'm sure study prints run off from a legitimate film would be covered by fair use. I confess, having said that, I begin to wonder about the status of a transcribed and re-typeset *text*. ... Anyway, my check to Yale is (nearly) in the mail\ (he's almost got it!) (chorus) his check to Yale is nearly in the mail!@ From reeds@gauss.att.com Tue Jan 21 10:46:27 1992 Message-Id: <9201210146.AA22757@rand.org> From: reeds@gauss.att.com Date: Mon, 20 Jan 92 20:46:27 EST To: voynich@rand.org Status: OR Copied from "Medieval Academy News", September 1991, p3. The Beinecke Rare Book Room & Manuscript Library offers short-term followships to support visiting scholars pursuing postdoctoral or equivalent research in its collections. The fellowships, which support travel to and from New Haven amd pay a living allowance of $1,500 per month, are designed to provide access to the library for scholars who reside outside the greater New Haven area. The length of a grant is nomally one month, and fellowships must be taken up between September 1992 and May 1993. Resumes and brief research proposals (not to exceed 3 pages) should be addressed to the Director, Beinecke Rare Book & Manuscript Library, Box 1603A Yale Station, New Haven, CT 06520-1603. Deadline for application is 15 January 1992. WE JUST MISSED OUT CHANCE!!! From reeds@gauss.att.com Tue Jan 21 14:41:04 1992 Message-Id: <9201210541.AA27466@rand.org> From: reeds@gauss.att.com Date: Tue, 21 Jan 92 00:41:04 EST To: voynich@rand.org Status: OR I have just seen an offprint of Zimansky's ``William F. Friedman and the Voynich Manuscript'', Philological Quarterly, vol XLIX, No 4, (Oct 1970), pp 433-443. The plates are supurb, well worth the trip to the stacks. They are: a portrait of the Colonel, and pictures of folios 56r, 79v, and 84r. The latter 2 are often reproduced, but 56r is rare (only previous appearence in Kraus, 35 MSS, page 45). All three are suitable for transcribing. 56r is interesting because it is in language A, hand 1; the other two are language B, hand 2. The difference is indeed striking. Hand 1 is well spaced, very neat. Hand 2 is more crowded. Just as Currier says. Its funny, I've looked at lots of A/1 stuff now (on Xerox mosaics from Jim G.) but didn't see how tidy it was till I saw the whole page on one shot. From j.guy@trl.OZ.AU Wed Jan 22 00:26:23 1992 Date: Tue, 21 Jan 92 10:26:23 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201202326.AA20991@medici.trl.OZ.AU> To: voynich@rand.org Subject: paleography Status: OR I have just found a copy of "Latin Palaeography -- Antiquity and the Middle Ages" by one Bernhard Bischoff. Haven't had time to read any of it. Just skimming through it, I saw examples of Beneventan script that stink to high heaven of Voynich writing, and a paragraph on "Ciphers" of which I cannot resist quoting the first two sentences: The middle ages had a peculiar, playful relationship with ciphers. They were used in many cases in which actual concealment was neither called for nor earnestly intended. 240 pages to read... the Voynich manuscript had 240 pages too. From RJB@MAX.U.WASHINGTON.EDU Wed Jan 22 03:14:00 1992 Date: Tue, 21 Jan 92 10:14 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: missed chance? To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Actually, we are now alerted to the possibility, and can get together really killer applications for *next* year. --rjb From reeds@gauss.att.com Wed Jan 22 08:37:13 1992 Message-Id: <9201212337.AA17938@rand.org> From: reeds@gauss.att.com Date: Tue, 21 Jan 92 18:37:13 EST To: voynich@rand.org Status: OR Today my ship came in! In the mail came (1) an old copy of the Yale microfilm, which a friend is loaning me, and (2) a first generation Xerox of Father Petersen's transcription of the VMS. I have spent only a few minutes examining each one. Here are first impressions: 1. The film (a B&W positive made from a negative made from the VMS) is not too clear. Prints made from it on the local microfilm printer are no clearer than prints from Gillogly's British Libe film. The black & white makes it hard to read through stains & spots. One finding: in the few cases I checked, the "ghost images" are due to bleed-through. Transcription from the microfilm or from prints from it will be possible but painful. I hope the Yale-supplied prints will be better. 2. The Petersen transcription is described on section 6.3, page 41 of D'Imperio, but D'Imperio does not do it justice. Petersen did essentially everything we proposed to do, except actually enter it into a computer. (Catholic U. not HAVING a computer between 1931 and 1944!) He made a complete legible hand transcription (into V characters), in ink. He sketched the pictures in pencil, numbered all plants, stars, roots, ladies, etc. (The nymph on 82v that strikes Peter Davidson as really cute is number 13, for instance.) Then he seems to have pencilled notes all over the place, with balloons and arrows, for variant readings, doubtful or verified readings, etc, as well as pointers to other places in the text with similar words. Finally, many pages seem to have many words marked in boxes and lozenges in colored pencil, presumably as a by product of making his concordance and frequency chart. Petersen's transcription is into clearly written, clearly spaced Voynich characters. In most cases you can just see what Currier letter it is, but the odd variant is faithfully copied, usually with a little circled "sic". Word divisions are distinct. He shows splits. The 293 xerox pages I have are very legible. The balloons and lozenges obtrude, but not excessively. I will supply copies to Team Voynich members on the (n+1) plan: I will take orders for a while, then run off n+1 copies at the local xeroxateria, and charge 1/(n+1) of my total xerox bill (including the $43.05 I spent already), plus individual postage, to whoever placed an order in time. In the mean time I will send free copies of a few pages to a few high visibility team members in the next few days, in return for some kind of public evaluation of copy quality, transcription accuracy & useability, etc. (I will make the free copies at my office, where the copy quality might not be the same as at the xeroxateria.) If copies made from my copy DO prove legible we can avoid bothering the Friedman collection with direct requests for more copies. I found the archivist of the Friedman collection and his staff very helpful in cheerfully and promptly locating and copying this material, all for a laughably low fee. (It must have taken tedious hours for them to make the copy.) Jim Reeds From wet!naga@cca.ucsf.EDU Wed Jan 22 18:57:00 1992 Message-Id: Date: Wed, 22 Jan 92 01:57 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: rand.org!voynich@gauss.att.com, reeds@gauss.att.com Subject: Order Status: OR 1 copy. Thanks. From firth@SEI.CMU.EDU Wed Jan 22 23:25:33 1992 Date: Wed, 22 Jan 92 09:25:33 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201221425.AA20574@bp.sei.cmu.edu> To: reeds@gauss.att.com, voynich@rand.org Subject: Petersen transcription Cc: h@SEI.CMU.EDU Status: OR Jim Yes, please send me a copy. I accept the (n+1) rule for payment. Robert From reeds@gauss.att.com Thu Jan 23 00:30:30 1992 Message-Id: <9201221533.AA02122@rand.org> From: reeds@gauss.att.com Date: Wed, 22 Jan 92 10:30:30 EST To: voynich@rand.org Status: OR I just got word from a specialized bookseller who can give me an Askin's Causabon's True & Faithful for $375, and a Laycock for $175. Seems a bit steep to me! The good news is that the Steganographia is due to be reprinted this year; I'm to write back in 4 months. Jim. From RJB@MAX.U.WASHINGTON.EDU Thu Jan 23 02:50:00 1992 Date: Wed, 22 Jan 92 09:50 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: Askin a bit much To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR $375 for Askin's edition of the T&FR is ... well ... maybe it's what the market will bear. I seem to remember having seen it when it first came out, and being surprised at how clean the pages were; most of the pages of the T&FR that I've seen have been marred by offset that makes the usual microfilm and photocopies a bit grim. It looked like someone had spent some effort getting the proofs cleaned up. $175 for the Laycock is utterly bizarre. Wasn't some edition of the T&FR being remaindered for $15? Please let us know about the _Steganographia_. Who will be reprinting it? Thanks. --rjb From j.guy@trl.OZ.AU Fri Jan 24 00:58:01 1992 Date: Thu, 23 Jan 92 10:58:01 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201222358.AA23399@medici.trl.OZ.AU> To: Voynich@rand.org Subject: Currier 1976, p.9 Status: OR I will discuss the first table on p.9 of Currier 1976. And yes, I will again use my transcription because I'm basically very lazy: with it, I just load the Voynich fonts using Harald's font editor and I don't have to rack my brains in search of the right transcription. And the English text looks pleasantly medieval to boot. Here is that table, with Currier's comments (which I have slightly edited) total fqcy "word"-initial as expected in actual in symbols "word"-initial any "word" first "word" 118 20 3 212 38 26 24 4 (5?) 0 45 10 10 "At beginnings and ends of lines, we have skewed frequencies. For example, let's take these two letters and ... Here are statistics from Herbal A material, about 6500 words, 1000 lines... If [their] occurrence as an initial were random, we would expect [them] to occur 1/7 of the time in each word position of a line. Actually, [they are] a very infrequent word initial at the beginning of a line, except when there is an intercalated ." In other words, line-initially we find and instead of and . Mysterious? Read my explanation very carefully: I don't think there is any mystery there. Languages, English for one, do not behave any differently. End of explanation. Did you get it? "I don't" and "do not". Consider: I don't think there is any mystery there. Languages, English for one, do n't behave any differently. Ugh! Would you ever write that? Consider again: I do not think there is any mystery there. Languages, English for one, do not behave any differently. Currier adds: "This applies only to Language A, by the way; words with this initial group are low in Language B (, for example, occurs only 5 times in Herbal B, but 212 in Herbal A)". Now just for argument's sake, let us suppose that the language of the Voynich *is* English, and that is indeed the word "not". What is your conclusion? That Author A loves using negative sentences ("I do not doubt"), and Author B affirmative sentences ("I am certain"). I hasten to add that I do not believe in the least that the Voynich is written in English. I only used that analogy to show how the "mysterious" statistical properties of Voynichian can be explained very simply in the most pedestrian manner. Although it is perhaps a bit premature of me, I would like to say that I see there rather strong evidence that and are little "grammatical" words which betray either a stylistic habit or a dialectal difference. For stylistic habit think of how some people tend to sprinkle their sentences with "like", "-wise". (Myself, when writing a paper, I tend to overuse "thus" and "then"). For dialectal difference, imagine again that the Voynich was in English. Author A writes the Queen's English: "not". Author B is a Lalland-speaking Scot and writes "nae". Yes, I believe that we may well have here in these two little "words", and what you cryptologists call a crib. From reeds@gauss.att.com Fri Jan 24 14:15:29 1992 Message-Id: <9201240537.AA17054@rand.org> From: reeds@gauss.att.com Date: Fri, 24 Jan 92 00:15:29 EST To: voynich@rand.org Status: OR Just as a matter of curiosity, has anyone read The Gadfly? From RJB@MAX.U.WASHINGTON.EDU Sat Jan 25 01:35:00 1992 Date: Fri, 24 Jan 92 08:35 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: The Gadfly To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR No, I haven't read it. What is it? Enquiring minds etc. --rjb From jim@rand.org Sat Jan 25 01:41:42 1992 Message-Id: <9201241641.AA08130@mycroft.rand.org> To: voynich@rand.org Subject: Re: The Gadfly In-Reply-To: Your message of Fri, 24 Jan 92 08:35:00 -0800. From: Jim Gillogly Reply-To: jim@rand.org Date: Fri, 24 Jan 92 08:41:42 PST Sender: jim%mycroft@rand.org Status: OR > No, I haven't read it. What is it? Enquiring minds etc. It's a book written by Voynich's wife (Ethel?). It became very popular in Communist countries and went through many printings -- like a Gone With the Wind for the working people or something. I have a paperback of it, but haven't been able to get into it. Jim From RJB@MAX.U.WASHINGTON.EDU Sat Jan 25 03:44:00 1992 Date: Fri, 24 Jan 92 10:44 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: The Gadfly To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Just checked our library catalog: seems they have it in the Children's lit collection... odd. Will have to check it out. Document 11 AU Voynich, Ethel Lilian Boole, 1864-1960. MT The gadfly. PI New York, Henry Holt and Co., 1897. LN SUZLIB/Children's Literature General Stacks PZ7.V982 Ga 1897 SUZLIB/Suzzallo General Stacks 823 V949g Thanks for the suggestion-- rjb From j.guy@trl.OZ.AU Sat Jan 25 06:52:07 1992 Date: Fri, 24 Jan 92 16:52:07 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201240552.AA25120@medici.trl.OZ.AU> To: voynich@rand.org Subject: 9=A? (Yes!) Status: OR ARE <9> AND VARIANTS OF THE SAME LETTER? Currier has remarked: "Final 89 is very high in language B, almost non-existent in Language A". That had me worried: "89" being extremely frequent, it would mean that we have two very different "dialects" in Languages A and B. I split the Voynich file into VOYNICH.A and VOYNICH.B, and did a frequency count, disregarding spaces but not end-of-lines: A B 89 Observed: 446 2844 Expected: 144 387 Ratio: 3.10 7.35 Indeed. But I also noticed that the discrepancies were reversed for 8a: A B 8a Observed: 993 768 Expected: 103 223 Ratio: 9.64 3.44 I remembered that, in my article in Cryptologia, I had hypothesized that <9> could well be a word-final variant of , , , or . And that, in one of my postings to this group, I wrote that Currier's finding that the ending of one word strongly affects the beginning of the next suggested to me that spaces between words had only an aesthetic function. What if <9> were but a variant of ? Let us see: A B 8[a9] Observed: 1439 3612 Expected: 247 610 Ratio: 5.83 5.92 That confirmed my suspicion. But was there any additional evidence? Yes. Looking at my frequency tables, I found that <9> and occurred in nearly perfect mutually exclusive distribution, conditioned by the following letter. Here are the statistics for Language A (my transcription again): a o i v c x z 2 a - 2 1358 60 12 245 338 7 9 8 222 4 1 713 11 6 80 4 8 9 g q l = # a 1 7 5 - 3 5 4 - 9 230 425 72 - 302 353 375 63 And for Language B: a o i v c x z 2 a 1 8 1506 16 22 739 728 7 9 36 790 8 2 757 332 125 140 4 8 9 g q l = # a 1 8 6 - 5 10 2 - 9 1431 440 101 - 253 315 547 26 occurs before , , and , <9> before other letters, and at the end of lines (=) and paragraphs (#). But note that the constraint is considerably relaxed in Language B before and somewhat before . This pretty well convinces me that and <9> are two variants of the same letter conditioned by the shape of the following letter. There is another conditioning factor: <9> occurs line-initially. Here are the statistics: a 9 Language A 2 149 Language B 8 134 Again, we observe a slight relaxation of this "rule" in Language B. This makes me think that Author B wrote less confidently than Author A. From j.guy@trl.OZ.AU Sun Jan 26 02:24:29 1992 Date: Sat, 25 Jan 92 12:24:29 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201250124.AA25995@medici.trl.OZ.AU> To: voynich@rand.org Subject: my first tables (10 of them) Status: OR These are the tables which I have produced and been using yesterday. There are 5 tables for each "Language". They give the observed and expected occurrences of a letter or end-of-line or paragraph with the next, second next... fifth next letter or end of line or of paragraphs. For instance, in the table entitled 'File "voynich.a" Environment: 5 to the right' you see: a a 146 2041 105 That means that is found 146 times in environment (expected: 105 times). The figure (2041) is the total number of occurrences of Even though my transcription system (which I have used for these tables) breaks up letters into up to three components (e.g. and for Currier's and ), that means that the occurrence of a particular letter influences another letter one or more letters away. And that influence can very large. Look for instance at the cell at the intersection of and in that same table: q a 223 (observed) 2041 97 (expected) Enormous, isn't it? What is it? Extensive vowel and/or consonant harmony. Vowel harmony is common in many languages: Finnish, Hungarian, Turkish, and, on the obscure side, Sakao! Consonant harmony is not so common, but you have it in Ordoss (a Mongolian dialect), in Arabic, and in Javanese. At least, those are the ones I can think of off-hand. Ordoss, if I remember correctly, also has vowel harmony. Here are the tables for Language A (careful, they're 120 columns or so wide), followed by those for Language B. --------------------------------------------------------------------------------------------------------- File "voynich.a" Environment: 1 to the right. a o i v c ' t x z 2 4 8 9 g q l p ; = # a - 2 1358 60 12 - - 245 338 7 1 7 5 - 3 5 - - 4 - 2047 105 264 140 67 307 46 229 79 68 29 33 103 147 13 97 79 157 19 55 9 o 74 43 185 6 356 - - 1211 949 69 28 570 77 - 793 754 - - 13 4 5132 263 662 351 169 770 116 575 197 170 72 82 259 368 33 243 198 394 48 139 23 i - 6 1153 1236 5 - - 9 - 87 - 4 - 221 2 2 - - - 1 2726 140 351 187 90 409 62 306 105 90 38 43 137 196 17 129 105 209 25 74 12 v 6 127 - 1 478 - - 7 2 35 87 169 56 - 10 17 - - 250 54 1299 67 167 89 43 195 29 146 50 43 18 21 65 93 8 61 50 100 12 35 6 c 88 366 1 1 389 903 2852 - 2 73 2 12 487 34 557 218 - - 1 - 5986 307 772 410 197 898 136 671 230 198 84 95 302 430 38 283 231 460 56 162 26 ' - - - - - - 903 - - - - - - - - - - - - - 903 46 116 62 30 136 20 101 35 30 13 14 46 65 6 43 35 69 8 24 4 t 352 2149 1 3 863 - - 13 6 18 - 32 904 - 42 58 - - 6 - 4447 228 573 304 147 667 101 498 171 147 62 71 224 319 29 210 172 342 41 120 20 x 19 224 1 2 490 - - 1 5 74 45 390 98 - 25 41 - - 103 10 1528 78 197 105 50 229 35 171 59 51 21 24 77 110 10 72 59 117 14 41 7 z 57 229 5 - 566 - - 5 3 23 44 138 123 - 8 14 - - 82 15 1312 67 169 90 43 197 30 147 50 43 18 21 66 94 8 62 51 101 12 36 6 2 77 130 6 - 174 - - 2 - 6 6 24 73 - 1 5 - - 57 3 564 29 73 39 19 85 13 63 22 19 8 9 28 40 4 27 22 43 5 15 2 4 - 618 2 - 1 - - - - - - - 2 - - 3 - - 1 - 627 32 81 43 21 94 14 70 24 21 9 10 32 45 4 30 24 48 6 17 3 8 993 186 3 1 245 - - 15 2 10 3 23 446 - 6 5 - - 64 2 2004 103 258 137 66 301 45 225 77 66 28 32 101 144 13 95 77 154 19 54 9 9 8 222 4 1 713 - - 11 6 80 230 425 72 - 302 353 - - 375 63 2865 147 369 196 94 430 65 321 110 95 40 46 144 206 18 136 111 220 27 78 13 g 4 29 - - 39 - - - 1 6 9 12 10 - 2 2 - - 117 24 255 13 33 17 8 38 6 29 10 8 4 4 13 18 2 12 10 20 2 7 1 q - - - - - - - - - - - - - - - - 1598 289 - - 1887 97 243 129 62 283 43 211 73 62 27 30 95 135 12 89 73 145 18 51 8 l - - - - - - - - - - - - - - - - 1459 80 - - 1539 79 198 105 51 231 35 172 59 51 22 24 78 110 10 73 59 118 14 42 7 p 345 501 1 - 1262 - 595 3 - 2 2 9 324 - - 2 - - 3 - 3049 156 393 209 100 458 69 342 117 101 43 48 154 219 20 144 118 234 28 83 13 ; 14 44 1 - 170 - 109 - - - - 6 24 - - - - - - - 368 19 47 25 12 55 8 41 14 12 5 6 19 26 2 17 14 28 3 10 2 = 2 244 3 - 201 - - 8 3 66 171 180 149 - 38 12 - - 1 - 1078 55 139 74 36 162 24 121 41 36 15 17 54 77 7 51 42 83 10 29 5 # - 9 - - 7 - - - - 3 4 4 5 - 94 45 - - - - 171 9 22 12 6 26 4 19 7 6 2 3 9 12 1 8 7 13 2 5 1 ---------------------------------------------------------------------------------------------------------- File "voynich.a " Environment: 2 to the right. a o i v c ' t x z 2 4 8 9 g q l p ; = # a 20 90 992 186 193 - 5 5 3 83 24 98 61 120 5 11 6 2 127 11 2042 105 264 140 67 307 46 228 78 67 29 32 103 147 13 97 79 157 19 56 9 o 288 431 163 4 1015 53 128 21 29 74 80 448 383 82 144 116 1442 105 112 21 5139 264 664 351 169 772 116 574 197 169 72 81 258 371 33 243 198 394 48 140 23 i 16 177 49 1044 527 2 2 12 4 72 95 190 76 18 14 18 4 - 324 71 2715 140 351 186 90 408 62 303 104 90 38 43 136 196 17 128 105 208 25 74 12 v 120 159 4 1 83 107 218 22 11 19 53 63 58 3 220 112 26 1 23 3 1306 67 169 89 43 196 30 146 50 43 18 21 66 94 8 62 50 100 12 36 6 c 252 1497 47 6 793 2 913 125 146 105 64 198 741 3 100 145 662 113 43 20 5975 307 772 409 197 897 135 667 230 197 84 95 300 431 38 282 230 458 55 162 27 ' 40 501 - 1 207 - - 3 - - - 6 128 - 6 7 - - 1 - 900 46 116 62 30 135 20 101 35 30 13 14 45 65 6 43 35 69 8 24 4 t 84 340 249 27 541 7 21 719 627 84 130 436 367 3 271 322 94 6 99 13 4440 228 574 304 146 667 101 496 171 146 62 70 223 320 28 210 171 341 41 121 20 x 218 137 25 1 107 106 300 54 27 19 15 56 117 5 141 85 60 6 45 5 1529 79 198 105 50 230 35 171 59 50 21 24 77 110 10 72 59 117 14 42 7 z 96 90 55 2 77 124 341 44 44 7 11 45 41 6 170 100 16 6 34 7 1316 68 170 90 43 198 30 147 51 43 18 21 66 95 8 62 51 101 12 36 6 2 19 32 63 2 49 37 101 34 46 12 12 28 19 5 36 40 5 1 21 3 565 29 73 39 19 85 13 63 22 19 8 9 28 41 4 27 22 43 5 15 3 4 13 8 3 - 40 - 1 5 8 2 - 43 6 - 259 238 3 - - - 629 32 81 43 21 94 14 70 24 21 9 10 32 45 4 30 24 48 6 17 3 8 14 70 770 25 141 54 155 165 170 21 48 74 46 4 36 26 8 3 157 20 2007 103 259 137 66 301 45 224 77 66 28 32 101 145 13 95 77 154 19 55 9 9 271 391 8 1 165 148 424 22 9 32 54 93 119 4 303 145 603 52 21 - 2865 147 370 196 94 430 65 320 110 94 40 45 144 207 18 135 110 220 27 78 13 g 9 41 4 - 16 10 22 4 1 5 19 25 23 1 40 23 3 1 6 - 253 13 33 17 8 38 6 28 10 8 4 4 13 18 2 12 10 19 2 7 1 q 154 282 2 - 720 - 521 2 - - 1 11 184 - - 2 - - 3 - 1882 97 243 129 62 283 43 210 72 62 26 30 95 136 12 89 73 144 17 51 8 l 205 263 - - 712 - 183 1 - 2 1 4 164 - - - - - - - 1535 79 198 105 51 230 35 171 59 51 22 24 77 111 10 73 59 118 14 42 7 p 86 345 272 9 343 143 851 257 156 16 21 135 289 1 20 35 2 - 63 3 3047 157 394 208 100 458 69 340 117 100 43 48 153 220 20 144 117 234 28 83 14 ; 30 39 8 2 35 15 152 14 12 1 - 19 31 - 3 5 - - 2 - 368 19 48 25 12 55 8 41 14 12 5 6 18 27 2 17 14 28 3 10 2 = 106 238 6 - 201 91 102 19 18 4 1 24 16 - 104 99 47 3 - - 1079 55 139 74 36 162 24 121 41 36 15 17 54 78 7 51 42 83 10 29 5 # 4 9 - - 6 2 2 - - - - 1 1 - 8 3 70 69 - - 175 9 23 12 6 26 4 20 7 6 2 3 9 13 1 8 7 13 2 5 1 ----------------------------------------------------------------------------------------------------------- File "voynich.a" Environment: 3 to the right. a o i v c ' t x z 2 4 8 9 g q l p ; = # a 71 111 44 923 167 44 106 19 22 49 33 94 69 18 68 46 13 3 125 20 2045 105 264 140 67 307 46 229 79 68 29 32 103 148 13 97 79 157 19 56 9 o 498 593 279 97 1087 216 651 130 96 49 44 117 383 7 286 185 228 32 131 30 5139 263 663 352 169 772 116 574 198 170 73 82 258 371 33 243 198 394 47 140 23 i 138 310 13 38 502 120 241 39 16 50 147 219 134 13 275 152 30 2 226 47 2712 139 350 186 89 408 61 303 104 90 38 43 136 196 17 128 105 208 25 74 12 v 79 193 88 6 144 37 141 34 25 5 3 23 22 - 83 64 282 50 27 - 1306 67 168 89 43 196 30 146 50 43 18 21 66 94 8 62 50 100 12 36 6 c 195 878 206 24 798 29 814 485 455 105 120 376 487 19 313 339 225 20 68 11 5967 305 769 408 196 897 135 667 229 198 84 95 300 431 38 283 230 457 55 163 27 ' 21 85 37 5 138 - 8 141 105 15 28 91 76 - 54 67 13 - 11 3 898 46 116 61 30 135 20 100 35 30 13 14 45 65 6 43 35 69 8 25 4 t 253 496 192 41 790 77 195 86 112 90 105 422 380 68 258 174 540 53 89 22 4443 227 573 304 146 668 100 496 171 147 63 71 223 321 28 211 171 340 41 121 20 x 70 257 194 6 189 36 157 47 47 15 15 50 71 8 48 39 192 34 45 8 1528 78 197 105 50 230 35 171 59 51 22 24 77 110 10 72 59 117 14 42 7 z 60 201 112 11 179 18 158 18 24 15 7 33 68 13 54 36 238 32 36 2 1315 67 170 90 43 198 30 147 51 44 19 21 66 95 8 62 51 101 12 36 6 2 37 74 61 6 78 12 60 9 3 14 5 18 38 10 18 14 68 8 27 1 561 29 72 38 18 84 13 63 22 19 8 9 28 40 4 27 22 43 5 15 2 4 27 11 13 - 19 2 9 2 3 4 - 3 7 - 14 17 471 26 3 - 631 32 81 43 21 95 14 71 24 21 9 10 32 46 4 30 24 48 6 17 3 8 68 224 568 111 207 28 128 18 9 67 36 93 111 57 87 47 53 9 77 8 2006 103 259 137 66 301 45 224 77 66 28 32 101 145 13 95 77 154 18 55 9 9 174 454 218 4 602 39 252 56 56 11 7 49 134 1 155 163 371 77 39 5 2867 147 370 196 94 431 65 320 110 95 40 45 144 207 18 136 111 220 26 78 13 g 14 47 9 - 39 7 18 2 7 1 1 7 6 1 15 14 49 14 4 - 255 13 33 17 8 38 6 28 10 8 4 4 13 18 2 12 10 20 2 7 1 q 74 247 118 5 180 78 566 140 92 6 8 82 214 - 13 18 2 - 37 2 1882 96 243 129 62 283 43 210 72 62 27 30 95 136 12 89 73 144 17 51 8 l 42 137 162 6 198 80 437 131 76 11 13 72 106 1 10 22 - - 28 1 1533 78 198 105 50 230 35 171 59 51 22 24 77 111 10 73 59 117 14 42 7 p 136 561 277 24 444 29 231 140 109 46 54 209 496 37 41 42 54 1 102 17 3050 156 393 209 100 458 69 341 117 101 43 48 153 220 20 145 118 234 28 83 14 ; 38 94 21 1 49 8 27 21 20 3 5 17 46 2 - 3 7 1 6 - 369 19 48 25 12 55 8 41 14 12 5 6 19 27 2 17 14 28 3 10 2 = 33 111 104 1 87 39 236 11 41 4 - 17 13 - 89 89 202 1 4 - 1082 55 140 74 36 163 24 121 42 36 15 17 54 78 7 51 42 83 10 30 5 # 6 43 4 - 79 - 8 - - 1 - 8 8 - 3 2 9 2 - - 173 9 22 12 6 26 4 19 7 6 2 3 9 12 1 8 7 13 2 5 1 ---------------------------------------------------------------------------------------------------------- File "voynich.a" Environment: 4 to the right. a o i v c ' t x z 2 4 8 9 g q l p ; = # a 96 230 49 28 455 40 133 36 29 33 101 167 91 17 101 59 100 14 210 43 2032 104 262 139 67 305 46 227 78 67 29 32 102 147 13 96 78 156 19 55 9 o 155 620 571 94 650 139 1019 244 167 66 57 195 240 23 160 138 406 65 116 13 5138 264 663 351 169 772 116 575 197 169 73 82 259 371 33 243 198 394 47 140 23 i 175 384 102 5 259 133 339 57 43 20 43 96 86 11 288 177 362 65 65 8 2718 139 351 185 89 408 61 304 104 90 39 43 137 196 17 129 105 209 25 74 12 v 65 163 114 27 174 19 241 45 45 17 4 30 59 7 69 60 138 9 23 - 1309 67 169 89 43 197 30 146 50 43 19 21 66 94 8 62 50 100 12 36 6 c 340 826 253 67 1035 82 321 233 210 85 113 413 624 48 282 263 589 63 104 23 5974 307 771 408 197 898 135 668 229 197 85 95 301 431 38 283 230 458 55 163 27 ' 52 88 41 5 160 25 44 15 20 17 28 76 79 7 63 44 108 13 12 2 899 46 116 61 30 135 20 101 34 30 13 14 45 65 6 43 35 69 8 25 4 t 322 439 237 151 658 150 618 101 89 49 56 152 263 9 356 229 371 61 107 24 4442 228 573 303 146 667 100 497 170 146 63 70 224 321 28 210 171 341 41 121 20 x 84 196 179 33 203 27 187 118 46 32 13 61 78 21 69 59 78 9 28 5 1526 78 197 104 50 229 34 171 58 50 22 24 77 110 10 72 59 117 14 42 7 z 86 168 109 42 209 22 159 52 81 17 9 37 100 6 31 58 79 11 31 7 1314 67 170 90 43 197 30 147 50 43 19 21 66 95 8 62 51 101 12 36 6 2 40 64 43 44 93 11 56 18 18 11 8 20 33 6 20 22 29 3 21 2 562 29 73 38 19 84 13 63 22 19 8 9 28 41 4 27 22 43 5 15 3 4 55 111 26 4 294 1 15 7 8 1 2 7 57 - 3 2 28 3 3 - 627 32 81 43 21 94 14 70 24 21 9 10 32 45 4 30 24 48 6 17 3 8 76 203 74 514 195 43 147 52 54 33 26 92 97 14 80 71 108 26 89 10 2004 103 259 137 66 301 45 224 77 66 28 32 101 145 13 95 77 154 18 55 9 9 94 318 328 32 368 52 518 136 112 34 12 118 147 12 109 111 301 17 49 1 2869 147 370 196 94 431 65 321 110 95 41 46 144 207 18 136 110 220 26 78 13 g 16 36 14 2 47 6 34 5 16 2 - 4 8 4 9 15 29 - 6 - 253 13 33 17 8 38 6 28 10 8 4 4 13 18 2 12 10 19 2 7 1 q 96 360 144 9 279 21 132 104 87 30 38 129 273 18 26 29 30 1 64 12 1882 97 243 128 62 283 43 211 72 62 27 30 95 136 12 89 72 144 17 51 8 l 78 295 154 16 214 16 126 57 42 19 21 97 269 21 15 16 31 1 44 5 1537 79 198 105 51 231 35 172 59 51 22 24 77 111 10 73 59 118 14 42 7 p 131 367 137 206 378 66 192 177 173 71 87 239 294 27 161 133 78 5 105 21 3048 156 394 208 100 458 69 341 117 101 43 48 153 220 20 144 117 234 28 83 14 ; 19 51 34 5 46 7 18 28 36 10 12 38 26 3 15 7 3 - 7 1 366 19 47 25 12 55 8 41 14 12 5 6 18 26 2 17 14 28 3 10 2 = 51 203 99 23 235 12 95 28 25 14 1 19 42 1 24 33 177 1 - - 1083 56 140 74 36 163 24 121 41 36 15 17 55 78 7 51 42 83 10 30 5 # 9 11 5 2 22 26 54 10 10 3 - 11 3 - 1 3 5 - - - 175 9 23 12 6 26 4 20 7 6 2 3 9 13 1 8 7 13 2 5 1 ---------------------------------------------------------------------------------------------------------- File "voynich.a" Environment: 5 to the right. a o i v c ' t x z 2 4 8 9 g q l p ; = # a 146 269 91 13 202 102 265 46 49 24 42 98 98 15 223 131 138 22 56 11 2041 105 264 139 67 307 46 228 78 68 29 32 103 148 13 97 79 156 19 56 9 o 277 804 439 193 759 93 518 217 150 70 53 196 559 52 160 170 263 35 106 20 5134 263 663 350 169 771 116 574 197 170 73 81 258 371 33 243 198 393 47 140 23 i 154 367 187 30 349 53 400 76 77 22 11 62 96 17 156 134 416 49 60 2 2718 139 351 185 89 408 61 304 104 90 39 43 137 196 17 129 105 208 25 74 12 v 81 191 89 68 227 31 102 54 40 20 14 61 89 7 37 37 123 6 29 1 1307 67 169 89 43 196 30 146 50 43 19 21 66 94 8 62 50 100 12 36 6 c 352 601 369 120 958 143 706 250 207 97 100 304 405 22 350 251 493 52 165 29 5974 306 772 407 196 898 135 668 229 198 85 95 301 432 38 283 230 458 55 163 27 ' 46 98 61 26 149 46 119 14 14 5 14 30 41 1 65 43 84 23 16 3 898 46 116 61 29 135 20 100 34 30 13 14 45 65 6 43 35 69 8 25 4 t 185 566 428 70 617 99 660 105 101 47 52 151 273 19 198 158 509 76 115 16 4445 228 574 303 146 668 100 497 170 147 63 71 224 321 29 211 172 340 41 121 20 x 66 205 112 121 235 21 134 86 40 18 18 75 93 13 50 53 111 17 49 10 1527 78 197 104 50 229 34 171 58 51 22 24 77 110 10 72 59 117 14 42 7 z 63 170 99 66 219 20 135 56 67 22 18 63 104 5 41 36 84 5 38 4 1315 67 170 90 43 198 30 147 50 44 19 21 66 95 8 62 51 101 12 36 6 2 31 75 41 21 71 14 56 20 23 15 8 34 43 6 22 16 40 2 21 4 563 29 73 38 18 85 13 63 22 19 8 9 28 41 4 27 22 43 5 15 3 4 7 31 56 12 74 21 239 55 38 3 9 25 25 3 8 9 4 1 9 1 630 32 81 43 21 95 14 70 24 21 9 10 32 46 4 30 24 48 6 17 3 8 75 243 91 31 398 30 162 44 48 34 71 158 121 15 89 74 137 14 139 25 1999 103 258 136 66 300 45 223 77 66 28 32 101 145 13 95 77 153 18 55 9 9 151 415 185 178 482 42 230 109 89 38 23 160 321 21 82 77 200 20 46 2 2871 147 371 196 94 431 65 321 110 95 41 46 144 208 18 136 111 220 26 78 13 g 17 32 15 4 50 5 35 12 12 5 1 12 12 2 3 5 23 1 6 - 252 13 33 17 8 38 6 28 10 8 4 4 13 18 2 12 10 19 2 7 1 q 85 245 96 92 229 42 113 116 119 40 52 147 175 16 120 61 51 4 64 12 1879 96 243 128 62 282 42 210 72 62 27 30 95 136 12 89 72 144 17 51 8 l 65 173 75 119 195 31 97 89 90 41 47 130 145 14 56 79 30 1 48 10 1535 79 198 105 50 231 35 172 59 51 22 24 77 111 10 73 59 118 14 42 7 p 149 409 172 58 461 64 267 98 64 34 83 221 197 21 172 151 267 27 106 25 3046 156 393 208 100 458 69 340 117 101 43 48 153 220 20 144 118 233 28 83 14 ; 25 59 31 11 37 12 18 18 10 11 9 27 34 5 27 5 12 10 6 1 368 19 48 25 12 55 8 41 14 12 5 6 19 27 2 17 14 28 3 10 2 = 44 130 60 71 246 26 148 54 71 17 6 41 40 1 22 40 57 - 6 1 1081 55 140 74 35 162 24 121 41 36 15 17 54 78 7 51 42 83 10 30 5 # 20 52 13 1 16 3 39 3 7 2 - 5 3 - 3 4 3 1 - - 175 9 23 12 6 26 4 20 7 6 2 3 9 13 1 8 7 13 2 5 1 ====================== LANGUAGE B ===================================================================== File "voynich.b" Environment: 1 to the right. a o i v c ' t x z 2 4 8 9 g q l p ; = # a 1 8 1506 16 22 - - 739 728 7 1 8 6 - 5 10 - - 2 - 3059 171 295 138 75 548 76 229 159 76 36 108 223 297 10 97 179 251 25 62 2 o 21 17 42 1 138 - - 1601 453 38 16 227 26 - 825 1865 - - 11 - 5281 296 510 239 129 945 132 395 275 132 62 187 385 512 18 167 309 434 43 108 4 i - - 866 1312 2 - - 12 - 114 - - - 167 1 1 - - - - 2475 139 239 112 60 443 62 185 129 62 29 87 180 240 8 78 145 203 20 51 2 v 25 429 1 - 549 - - 7 3 15 51 79 46 - 7 9 - - 105 8 1334 75 129 60 33 239 33 100 70 33 16 47 97 129 5 42 78 110 11 27 1 c 65 382 20 2 1407 1367 2206 5 8 75 3 2375 1043 19 298 509 - - 3 - 9787 548 944 443 239 1752 245 732 510 244 114 346 713 950 33 310 573 804 80 200 8 ' - - - - - - 1367 - - - - - - - - - - - - - 1367 77 132 62 33 245 34 102 71 34 16 48 100 133 5 43 80 112 11 28 1 t 74 208 2 - 2866 - - 20 6 16 2 375 437 - 27 52 - - 3 - 4088 229 394 185 100 732 102 306 213 102 48 144 298 397 14 130 239 336 33 84 3 x 110 466 2 - 987 - - 45 28 54 117 243 160 - 86 386 - - 163 3 2850 160 275 129 70 510 71 213 149 71 33 101 208 277 10 90 167 234 23 58 2 z 300 375 3 - 398 - - 7 1 6 38 38 87 - 7 10 - - 90 2 1362 76 131 62 33 244 34 102 71 34 16 48 99 132 5 43 80 112 11 28 1 2 235 186 1 - 112 - - 6 - 4 8 8 32 - 10 6 - - 28 - 636 36 61 29 16 114 16 48 33 16 7 22 46 62 2 20 37 52 5 13 1 4 6 1865 12 - 33 - - - - 2 1 - 3 - 3 7 - - - - 1932 108 186 87 47 346 48 144 101 48 23 68 141 187 7 61 113 159 16 39 2 8 768 126 1 1 148 - - 21 3 4 29 5 2844 - 2 11 - - 26 1 3990 223 385 180 97 714 100 298 208 100 46 141 291 387 14 127 234 328 32 81 3 9 36 790 8 2 757 - - 332 125 140 1431 440 101 - 253 315 - - 547 26 5303 297 512 240 129 949 133 397 276 132 62 187 386 515 18 168 310 436 43 108 4 g 2 15 - - 13 - - 3 - 3 3 4 2 - 1 2 - - 134 4 186 10 18 8 5 33 5 14 10 5 2 7 14 18 1 6 11 15 2 4 - q - - - - - - - - - - - - - - - - 1399 336 - - 1735 97 167 78 42 311 43 130 90 43 20 61 126 168 6 55 102 143 14 35 1 l - - - - - - - - - - - - - - - - 3094 109 - - 3203 179 309 145 78 573 80 240 167 80 37 113 233 311 11 102 187 263 26 65 3 p 1364 224 9 - 2015 - 471 15 4 1 2 9 370 - - 3 - - 2 - 4489 251 433 203 110 803 112 336 234 112 52 159 327 436 15 142 263 369 37 92 4 ; 46 70 - - 256 - 46 - - - - 9 15 - - - - - 3 - 445 25 43 20 11 80 11 33 23 11 5 16 32 43 2 14 26 37 4 9 - = 8 115 - - 82 - - 38 6 158 230 164 133 - 175 11 - - - - 1120 63 108 51 27 200 28 84 58 28 13 40 82 109 4 36 66 92 9 23 1 # - 1 - - 3 - - - - - - 1 1 - 34 4 - - - - 44 2 4 2 1 8 1 3 2 1 1 2 3 4 - 1 3 4 - 1 - --------------------------------------------------------------------------------------------------------- File "voynich.b" Environment: 2 to the right. a o i v c ' t x z 2 4 8 9 g q l p ; = # a 139 346 746 547 464 4 - 26 10 98 78 141 129 145 20 29 13 2 123 2 3062 172 296 138 75 548 76 229 159 76 36 108 223 297 10 97 179 251 25 63 2 o 265 395 34 7 704 15 20 42 24 32 85 120 249 12 97 364 2567 123 122 3 5280 296 510 238 129 946 132 395 275 132 62 186 385 512 18 167 309 434 43 108 4 i 42 472 59 756 605 - - 12 3 54 56 85 56 11 13 14 - 2 222 12 2474 139 239 112 60 443 62 185 129 62 29 87 180 240 8 78 145 203 20 51 2 v 60 76 5 - 40 233 296 181 34 21 20 30 30 - 123 153 15 1 14 1 1333 75 129 60 32 239 33 100 69 33 16 47 97 129 5 42 78 109 11 27 1 c 196 411 26 2 1804 14 1393 334 124 108 366 1024 2710 4 141 246 712 95 75 7 9792 548 946 442 239 1754 244 732 510 245 114 345 714 949 33 311 573 804 80 201 8 ' 25 77 1 - 1108 - - 5 1 3 2 80 44 - 6 14 - - 1 - 1367 77 132 62 33 245 34 102 71 34 16 48 100 133 5 43 80 112 11 28 1 t 106 329 22 4 583 8 9 161 66 57 107 1232 803 - 147 318 73 6 50 4 4085 229 395 185 100 732 102 305 213 102 48 144 298 396 14 130 239 335 33 84 3 x 152 160 86 - 102 376 581 224 122 37 54 54 135 2 96 91 443 29 106 - 2850 160 275 129 69 510 71 213 148 71 33 101 208 276 10 90 167 234 23 58 2 z 29 61 175 3 32 178 210 232 131 8 24 34 26 1 76 87 14 3 40 1 1365 76 132 62 33 244 34 102 71 34 16 48 100 132 5 43 80 112 11 28 1 2 5 19 151 2 10 46 54 152 73 12 8 10 11 - 28 28 12 4 12 - 637 36 62 29 16 114 16 48 33 16 7 22 46 62 2 20 37 52 5 13 1 4 7 11 14 - 42 1 - 199 19 - 4 18 4 1 343 1258 6 4 1 - 1932 108 187 87 47 346 48 144 101 48 23 68 141 187 7 61 113 159 16 40 2 8 27 492 361 3 485 57 62 421 300 70 921 230 71 2 95 99 10 3 261 12 3982 223 385 180 97 713 99 298 207 99 46 140 290 386 14 126 233 327 32 82 3 9 415 1596 21 1 334 270 434 248 80 84 119 120 169 7 431 391 500 68 18 - 5306 297 512 240 129 950 132 397 276 133 62 187 387 514 18 168 311 436 43 109 4 g 5 18 3 - 11 1 10 6 2 22 18 24 22 1 28 7 3 - 5 - 186 10 18 8 5 33 5 14 10 5 2 7 14 18 1 6 11 15 2 4 - q 379 167 4 - 833 - 204 3 - 1 - 11 130 - - 1 - - - - 1733 97 167 78 42 310 43 130 90 43 20 61 126 168 6 55 101 142 14 35 1 l 1031 127 5 - 1438 - 313 12 4 - 2 7 255 - - 2 - - 5 - 3201 179 309 145 78 573 80 239 167 80 37 113 234 310 11 102 187 263 26 66 3 p 27 190 735 8 980 81 261 497 339 25 59 750 434 - 18 19 3 - 61 2 4489 251 434 203 109 804 112 336 234 112 52 158 327 435 15 142 263 369 36 92 4 ; 9 5 17 - 47 37 208 57 24 2 2 10 16 - 1 5 - - 4 - 444 25 43 20 11 80 11 33 23 11 5 16 32 43 2 14 26 36 4 9 - = 144 329 5 - 170 40 33 35 10 3 4 9 8 - 71 73 102 84 - - 1120 63 108 51 27 201 28 84 58 28 13 40 82 109 4 36 66 92 9 23 1 # - - - - 2 3 - - - - - - - - - 1 18 20 - - 44 2 4 2 1 8 1 3 2 1 1 2 3 4 - 1 3 4 - 1 - --------------------------------------------------------------------------------------------------------- File "voynich.b" Environment: 3 to the right. a o i v c ' t x z 2 4 8 9 g q l p ; = # a 127 321 96 687 351 199 257 180 108 52 72 78 90 6 85 93 44 5 205 8 3064 172 296 138 75 549 77 229 160 76 36 108 224 297 10 97 179 252 25 63 2 o 1083 284 208 16 1499 267 396 218 128 25 58 71 264 1 100 109 426 35 89 2 5279 296 509 239 128 945 132 394 275 132 61 186 385 512 18 167 309 433 43 108 4 i 81 350 19 48 361 256 322 204 45 53 67 102 90 7 166 182 23 4 87 6 2473 139 239 112 60 443 62 185 129 62 29 87 180 240 8 78 145 203 20 51 2 v 58 87 28 1 333 8 243 48 18 6 11 24 23 4 32 109 246 30 24 - 1333 75 129 60 32 239 33 100 69 33 16 47 97 129 5 42 78 109 11 27 1 c 352 1156 75 13 2048 53 603 428 206 132 1025 1050 1368 5 283 419 354 33 178 11 9792 549 945 442 238 1754 245 732 510 244 114 346 715 950 33 310 573 804 80 200 8 ' 26 106 9 2 265 1 4 51 18 7 12 457 228 - 46 110 20 - 3 - 1365 77 132 62 33 244 34 102 71 34 16 48 100 132 5 43 80 112 11 28 1 t 177 355 37 5 237 20 69 259 102 70 283 252 1262 11 150 253 413 52 75 5 4087 229 394 185 99 732 102 305 213 102 48 144 298 396 14 130 239 336 33 84 3 x 216 206 137 23 901 20 419 90 55 37 42 101 106 14 57 143 161 26 94 - 2848 160 275 129 69 510 71 213 148 71 33 101 208 276 10 90 167 234 23 58 2 z 74 117 120 43 267 11 188 29 19 23 14 53 57 20 33 75 138 25 57 1 1364 76 132 62 33 244 34 102 71 34 16 48 100 132 5 43 80 112 11 28 1 2 45 62 100 37 135 3 49 8 4 19 3 14 21 7 9 45 47 9 18 1 636 36 61 29 15 114 16 48 33 16 7 22 46 62 2 20 37 52 5 13 1 4 27 37 15 1 117 3 5 10 7 6 5 7 16 - 11 49 1562 39 14 - 1931 108 186 87 47 346 48 144 101 48 22 68 141 187 7 61 113 159 16 39 2 8 236 1116 227 78 399 165 322 151 39 70 92 106 144 48 287 260 159 35 52 1 3987 223 385 180 97 714 100 298 208 99 46 141 291 387 14 126 233 327 32 82 3 9 289 296 223 3 923 106 439 379 146 11 14 93 80 6 331 1069 688 134 71 1 5302 297 512 240 129 949 133 396 276 132 62 187 387 514 18 168 310 435 43 108 4 g 20 37 3 - 30 7 3 4 4 2 1 8 2 2 11 13 29 6 4 - 186 10 18 8 5 33 5 14 10 5 2 7 14 18 1 6 11 15 2 4 - q 15 68 159 2 286 72 268 225 137 8 25 271 154 - 8 16 1 - 19 - 1734 97 167 78 42 311 43 130 90 43 20 61 127 168 6 55 102 142 14 35 1 l 21 127 593 6 741 46 201 329 226 19 36 489 296 - 11 8 2 - 46 2 3199 179 309 145 78 573 80 239 167 80 37 113 233 310 11 101 187 263 26 65 3 p 155 420 302 364 546 39 136 125 42 89 165 746 1065 55 64 60 32 5 76 6 4492 252 434 203 109 804 112 336 234 112 52 159 328 436 15 142 263 369 36 92 4 ; 19 39 14 2 181 9 40 8 3 7 4 57 34 - 7 12 6 - 3 - 445 25 43 20 11 80 11 33 23 11 5 16 32 43 2 14 26 37 4 9 - = 40 88 106 - 144 80 118 104 57 1 2 10 4 - 42 177 138 6 3 - 1120 63 108 51 27 201 28 84 58 28 13 40 82 109 4 35 66 92 9 23 1 # 4 5 - - 28 2 3 - - - - 1 - - - - 1 - - - 44 2 4 2 1 8 1 3 2 1 1 2 3 4 - 1 3 4 - 1 - --------------------------------------------------------------------------------------------------------- File "voynich.b" Environment: 4 to the right. a o i v c ' t x z 2 4 8 9 g q l p ; = # a 123 426 104 43 609 138 383 163 61 56 67 153 126 17 130 169 151 27 111 6 3063 172 296 139 75 548 76 229 160 77 36 108 223 297 10 97 179 252 25 63 2 o 208 294 683 66 1250 82 577 429 255 52 59 566 220 16 67 135 179 30 107 - 5275 295 509 239 128 945 132 394 275 132 61 186 385 512 18 167 308 433 43 108 4 i 122 185 37 2 418 143 441 150 46 23 30 56 50 11 114 245 307 41 51 2 2474 139 239 112 60 443 62 185 129 62 29 87 180 240 8 78 145 203 20 51 2 v 159 67 63 8 394 25 67 50 27 14 8 86 70 1 28 97 136 5 28 - 1333 75 129 60 32 239 33 100 69 33 16 47 97 129 5 42 78 109 11 27 1 c 447 1581 206 18 1155 209 557 556 196 113 480 758 1444 21 449 761 606 96 129 6 9788 548 945 443 238 1753 244 732 510 245 114 346 714 950 33 310 572 804 79 200 8 ' 33 82 10 1 51 5 12 87 31 18 119 105 493 1 50 98 145 11 15 - 1367 77 132 62 33 245 34 102 71 34 16 48 100 133 5 43 80 112 11 28 1 t 196 589 89 9 461 73 343 233 107 55 559 141 295 11 180 236 361 42 102 6 4088 229 395 185 100 732 102 306 213 102 48 144 298 397 14 130 239 336 33 84 3 x 135 178 201 64 732 65 160 129 76 41 41 377 207 8 73 83 191 9 77 3 2850 160 275 129 69 510 71 213 149 71 33 101 208 277 10 90 167 234 23 58 2 z 101 87 60 83 333 38 71 64 38 21 20 74 77 8 38 78 100 8 63 4 1366 77 132 62 33 245 34 102 71 34 16 48 100 133 5 43 80 112 11 28 1 2 39 40 39 85 112 37 58 35 26 7 8 24 15 1 19 22 52 2 15 - 636 36 61 29 15 114 16 48 33 16 7 22 46 62 2 20 37 52 5 13 1 4 594 76 27 10 845 37 68 26 11 1 8 18 115 1 9 17 58 2 7 - 1930 108 186 87 47 346 48 144 101 48 22 68 141 187 7 61 113 159 16 39 2 8 144 220 129 198 586 130 345 265 123 14 32 108 84 6 246 710 453 94 95 4 3986 223 385 180 97 714 100 298 208 100 46 141 291 387 14 126 233 327 32 81 3 9 317 285 322 50 1185 112 348 176 119 44 36 270 194 22 103 223 1353 47 96 2 5304 297 512 240 129 950 132 397 276 133 62 187 387 515 18 168 310 436 43 108 4 g 13 8 17 - 30 12 16 17 8 2 - 5 6 - 6 19 23 1 3 - 186 10 18 8 5 33 5 14 10 5 2 7 14 18 1 6 11 15 2 4 - q 70 183 68 69 333 19 89 46 18 33 53 249 367 19 36 29 22 2 26 3 1734 97 167 78 42 310 43 130 90 43 20 61 126 168 6 55 101 142 14 35 1 l 104 276 248 297 394 29 87 87 27 63 116 554 732 36 35 43 16 3 53 3 3203 179 309 145 78 574 80 239 167 80 37 113 233 311 11 102 187 263 26 65 3 p 174 609 76 284 556 157 264 259 148 64 286 299 731 5 137 178 105 19 134 4 4489 251 433 203 109 804 112 336 234 112 52 159 327 436 15 142 262 369 36 92 4 ; 20 36 12 12 48 16 26 23 15 5 7 116 66 1 5 11 16 3 6 1 445 25 43 20 11 80 11 33 23 11 5 16 32 43 2 14 26 37 4 9 - = 63 55 82 33 292 30 158 54 30 11 3 23 15 1 8 42 217 2 - - 1119 63 108 51 27 200 28 84 58 28 13 40 82 109 4 35 65 92 9 23 1 # 1 - - - 7 8 18 1 4 - - 4 - - - 1 - - - - 44 2 4 2 1 8 1 3 2 1 1 2 3 4 - 1 3 4 - 1 - --------------------------------------------------------------------------------------------------------- File "voynich.b" Environment: 5 to the right. a o i v c ' t x z 2 4 8 9 g q l p ; = # a 193 222 94 38 582 159 366 193 69 36 38 159 117 16 138 253 272 27 86 6 3064 172 296 139 75 549 76 229 160 76 36 108 223 297 10 97 179 251 25 63 2 o 196 392 380 382 1051 78 234 161 86 79 106 716 844 40 98 124 188 14 106 5 5280 296 510 239 129 945 132 395 275 132 61 187 385 512 18 167 309 433 43 108 4 i 226 144 103 10 640 50 231 95 50 25 16 112 102 5 59 197 330 29 49 1 2474 139 239 112 60 443 62 185 129 62 29 87 180 240 8 78 145 203 20 51 2 v 100 75 121 24 229 15 79 74 44 19 28 108 136 7 33 86 116 9 27 - 1330 74 128 60 32 238 33 99 69 33 15 47 97 129 5 42 78 109 11 27 1 c 415 994 301 74 1207 195 712 615 262 102 545 425 1027 26 446 1058 1094 116 169 9 9792 548 946 443 239 1753 244 732 510 244 114 346 714 950 33 310 573 804 79 201 8 ' 45 238 21 1 143 21 86 63 9 19 275 46 126 1 52 45 131 17 26 2 1367 77 132 62 33 245 34 102 71 34 16 48 100 133 5 43 80 112 11 28 1 t 225 726 164 24 482 125 413 223 77 39 118 129 317 9 214 330 348 68 53 2 4086 229 395 185 100 732 102 305 213 102 47 144 298 396 14 130 239 335 33 84 3 x 143 227 139 118 487 55 153 136 59 36 82 341 475 9 59 94 143 13 79 1 2849 160 275 129 69 510 71 213 148 71 33 101 208 276 10 90 167 234 23 58 2 z 77 124 76 29 240 28 87 60 59 20 31 140 126 6 45 64 103 13 33 4 1365 76 132 62 33 244 34 102 71 34 16 48 100 132 5 43 80 112 11 28 1 2 24 70 41 14 172 19 60 34 11 13 9 33 45 2 8 26 36 5 15 - 637 36 62 29 16 114 16 48 33 16 7 23 46 62 2 20 37 52 5 13 1 4 24 54 335 18 486 25 143 227 108 15 15 315 82 - 15 18 25 1 24 - 1930 108 186 87 47 346 48 144 101 48 22 68 141 187 7 61 113 158 16 40 2 8 218 267 167 30 906 80 296 106 66 39 44 240 175 18 95 178 910 46 103 3 3987 223 385 180 97 714 100 298 208 100 46 141 291 387 14 126 233 327 32 82 3 9 612 286 233 195 1375 105 367 250 137 51 56 485 508 21 63 132 303 23 98 2 5302 297 512 240 129 949 132 396 276 132 61 187 387 514 18 168 310 435 43 109 4 g 9 10 18 2 44 4 29 1 5 6 - 9 5 1 6 11 23 2 1 - 186 10 18 8 5 33 5 14 10 5 2 7 14 18 1 6 11 15 2 4 - q 70 226 35 59 214 60 109 92 59 30 92 187 247 3 67 62 57 8 53 1 1731 97 167 78 42 310 43 129 90 43 20 61 126 168 6 55 101 142 14 35 1 l 124 419 53 237 390 113 181 190 104 39 201 228 550 3 75 127 64 14 87 4 3203 179 309 145 78 574 80 239 167 80 37 113 233 311 11 102 188 263 26 66 3 p 231 671 122 17 682 174 409 275 114 48 261 195 267 16 233 355 285 30 99 4 4488 251 434 203 109 804 112 336 234 112 52 159 327 435 15 142 263 368 36 92 4 ; 32 49 14 4 57 7 22 16 14 7 11 34 114 3 16 17 10 6 12 - 445 25 43 20 11 80 11 33 23 11 5 16 32 43 2 14 26 37 4 9 - = 95 84 56 57 388 52 102 37 32 11 4 77 39 - 11 23 48 2 1 - 1119 63 108 51 27 200 28 84 58 28 13 40 82 109 4 35 66 92 9 23 1 # 3 4 1 - 16 - 9 - - - - 7 2 - - 1 1 - - - 44 2 4 2 1 8 1 3 2 1 1 2 3 4 - 1 3 4 - 1 - From kornai@Csli.Stanford.EDU Sun Jan 26 16:05:39 1992 Date: Sat, 25 Jan 92 23:05:39 PST From: Andras Kornai To: voynich@rand.org Subject: Arbitrary spaces? Message-Id: Status: OR In his previous message Jacques Guy writes: The spaces between the words are spurious, an artifact of the Voynich alphabet system. I must have missed this -- could you please elaborate what makes you think so? Certainly the script looks as if it was written in the ordinary manner, with spaces separating words. So in order to accept your hypothesis, I will need lots and lots of evidence. Are word-internal, word-initial, and word-final letter frequencies essentially the same? Are digraph frequencies word-internally and across spaces the same? How about trigram frequencies? Does the distribution of word-lengths follow more or less closely what we see in natural languages, or does it follow some unexpected law (such as uniform distribution)? Andras Kornai From reeds@gauss.att.com Mon Jan 27 00:18:22 1992 Message-Id: <9201261521.AA27221@rand.org> From: reeds@gauss.att.com Date: Sun, 26 Jan 92 10:18:22 EST To: voynich@rand.org Status: OR Guy's stuff is neat. It would be nice if it were corroborated by by something other than digraphic frequency counts. (Such as, word counts, word transition counts, trigraphic counts, etc.) From Karl.Kluge@G.GP.CS.CMU.EDU Sun Jan 26 21:06:00 1992 Date: Sun, 26 Jan 1992 21:06-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Difference metric for text Message-Id: <696477982/kck@G.GP.CS.CMU.EDU> Status: OR I'd like to perform a cluster analysis on the folios in the existing D'Imperio transcription to see what structure falls out, and where splits in the A/B language, the different "hands," and the different apparent contents of the folios show up in the clustering (if at all). In order to do this I need a difference metric -- some statistical measure that is lower for two pieces of text which appear to share a common language/author/etc, higher for texts which differ. Bennett suggests sum(i, 1, 26) Freq1(i) * Freq2(j), where Freq1 and Freq2 are the character frequencies in the two texts. Is there a better scalar measure? Karl From j.guy@trl.OZ.AU Mon Jan 27 06:11:34 1992 Date: Sun, 26 Jan 92 16:11:34 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201260511.AA26926@medici.trl.OZ.AU> To: voynich@rand.org Subject: my letter frequencies explained Status: OR To answer the questions by wet!naga: Yes, occurs 146 times in VOYNICH.A (in which ? is a wild card) The figure under each letter in the left-hand column is the sum of the observed frequencies in that letter's row (which I hold in an array SumRow[CHAR] OF INTEGER). I also sum the columns, but do not print those sums. Then I sum the contents of SumRow (call it Sum), and I compute the expected frequency of cell x,y as: Round(SumRow[y]/Sum*SumCol[x]) You ask: "why is the total number of occurrences of (e.g.) different in the different tables for VOYNICH.A (viz. 2047, 2042, 2045, 2032 and 2041)?" I thought I had a bug there and look for one long and hard. Can't find one. This is my explanation of this strange state of things: Occasionally, there is in D'Imperio's corpus an asterisk (*), which means a doubtful or unidentified letter. When the program hits say it just moves on without incrementing anything in the frequency matrix. That, I think, is probably the reason for the discrepancies you noticed in the sum of row from table to table. You say: "But even without knowing the details, and not to disparage Jacques' system, I would think that we should be looking at frequency data for the letters of the voynich alphabet (as they occur in the text) rather than frequency data for the letters of this transcription system (as they occur in the transcription of the text)." Quite true. The trouble is: we really do *not* know what the *true* letters of the Voynich alphabet are. For instance, <4> is almost always followed by (about 90% of the time). Perhaps <4o> is a single Voynich letter. We just don't know. Another example: everybody (including me) believes that Currier's and (my and ) are single letters. Well, it seems obvious that they are, but is it true? In a few minutes I shall send an article entitled "pronounceable Voynich". Have a look at it. To make the language pronounceable I hit upon this idea: let Currier's be "u", but consider his as made up of and , and read it "iu", and his as made up of and and read it "nu" ( in the Voynich letters, does look a lot like German cursive "n" after all). Of course, I just wanted to make it pronounceable and did not believe one moment at the time that that could anything more than a convenient way of recording things. But it turned out to work so well, and to wipe out so many of the problems I had had in trying to make the Voynich pronounceable that I am starting to wonder: Currier's does look like a "v", "v" and "u" are not distinguished in medieval manuscripts, so...? And then I have found in Bischoff's treatise of Latin paleography an example of Beneventan "n" that looks surprisingly like two Voynich 's. Could it be? Could it be? From j.guy@trl.OZ.AU Mon Jan 27 06:35:15 1992 Date: Sun, 26 Jan 92 16:35:15 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201260535.AA26937@medici.trl.OZ.AU> To: voynich@rand.org Subject: Pronounceable Voynich Status: OR With the clamoring there has been for "pronounceable Voynich", I can't resist proposing a pronounceable scheme. Apart from the values of the vowels, it is, however, completely arbitrary. Some of the values I have picked because they were good mnemonics. Here goes: a a <9> a e o Now for the arbitrary stuff: Currier my Pronounce it: S r (rolled, Scots-fashion) Z l P t (looks a bit like a capital T) F k (ditto, K) B p (ditto, P) V f (ditto, F) R <2> y (I actually started making it "z" because it looks like capital "Z"; then I changed my mind to "y", because I actually think there is a good chance that it is the medieval "i-longa", in other word, a variant of ) 2 s (ditto, mirror-image of S) 8 <8> b (ditto, B) E m (I felt the need for an "m" somwhere) 4 <4> h (might be its true value, too) J g (pronounced "ng") 7 <&> ditto: J and 7 look so much the same! (in fact, it doesn't matter: you can spell it "ng", there won't be any ambiguity, just remember that "ng" is a *single* Voynich letter, not two) D u (ditto, u, pronounced "oo") I i (ditto, i, pronounced "ee") n (ditto, German cursive n) Consequently: N iu (pronounced [yu} as in "you" or "yew") M nu (pronounced "noo") K ing (pronounced "ing") L ung (pronounced "oong") 3 ning 6 eng (no ambiguity: "eng" never occurs in the Voynich language) When you have the "gallows" letters *inside* Currier's and (my and ) you pronounce the "gallows" first, then a short "i" (which you spell "i"), then "r" or "l". Thus: --> "tir", --> "kir", --> "fir", --> "pir", etc. Since never occurs in the Voynich language after or there can be no confusion. The inserted "i" is there only to remind you that this is an "inserted" letter group, so that we can distinguish: "pr" "pir" If you hit a seemingly unpronounceable cluster, e.g. "prb", it will usually have a liquid ("r" or "l") in the middle: insert an "i" after that liquid. Since Voynich cannot occur in that position, you create no ambiguity. (And, anyway, that's how Sanskrit is pronounced: "Krishna" is spelt "Krshna", with a syllabic "r", and originally it was pronounced with a syllabic "r", like in Czech and in Serbo- Croat). Let's try (remember: those spaces between words are meaningless): Sample of Language A: <00101A> l;act92.9lpax.aS.aqpaiiv.c'tox.c'toS9.cqptoS92.9.lpoS.c'tox89= f a ray ak am as at an u l om l osa tirosay a k os l omba <00102A> 2oS9.clptaS.o.S.9.lpaiv.c'tqpaiiv.c'taS.aS*.cqptaS.cqptaS.8av= yosa kiras o s a kaiu l t a nu l as as? tiras tiras bau <00103A> 29aii2.c'tclp9.oS.9lpaiiv.c'to8.cqptoaS9.cqpt*S.8aSaiiv.29= yaan y l ek a os a kanu l ob tiroasa tir?s basanu ya <00104A> 4oiiv.oqpcc9.oqpcoS.2oxoqp9.cqpt*aS.8aiiv.olpaiiv.ox.olpav= honu ot a a ot eos yomota tir?as banu o ka nu om ok au <00105A> 2ai2.9.ctcaS.cqptaiiv.cq;taS.cl;taiiv# yaiy a reas tiran u piras piran u Well, well, well, listen to that Faraya kamasa tanu lomlosa tiro saya kos lomba Yo saki rasosa kaiul tanu lasas[a] tiras-tiras bau Ya any lekao sakanulob tiroasa tir[a]s basanu ya Honuota a oteosy omota tir[o]as banu o kanu o mokau Yai yareasti ranu piras-piranu. Why, it almost sounds real! If you really want to preserve the Voynich spelling you can distinguish between its three a's in writing: there is no harm in spelling "ee" (just remember it's pronounced "ah"), and perhaps using a capital "A" for the one with the flourish (9), with a space preceding it perhaps, viz: Far Ay Akamasa tanu lom los Atiros Ay Akoslomb A Let's see some Language B: <04901B> q;c'tcolp9.o8aii2.4o9=ol;c'to8.ct9q;ctc9.9q;ctc89aiv=ctol;o=ctcq;ct89= p l eok a oban y hoa of l ob ra p rea a p rebaaiu rof o le p rba <04902B> 8ctc9.2aiiv.a&cco89.9lpceqptc9.ctc89.9qpc89.89=ctceqptc89.x2= br ea yan u angaoba ake tir ea reba a teba ba re tireba my <04903B> oaiiv.c'teqpt9.eqptc89.oxo9.9lpc't89.oxctc89.89x=9c'tcc9.2aiiv.2= oan u l tira tireba omoa ak l ba om reba bam a l a a yan u y <04904B> 4olpc89.ctco2.9qpc89.4olpc89.9qpc89.ctclpc89.8aiiv.o8aig.2ax89= hok eba reoy a teba hok eba at eba r e keba ban u obang yamba <04905B> 2aiiv.c'tc89.cc89.cc89.2ct9.8aiiv.eqptc89=4olpcc89.4olpc89.eqptc9= yanu leba aba aba yr a ban u tireba hok a ba hok eba tirea <04906B> Sctc89.4olpc89# s reba hok eba <04907B> q;cco.4olpc89.8aS.c'tco.9q;ct2c82.2=aiiv=c'taq;ctc89.l;ct9.8ax.ctc89.2aS= p ao hok eba bas l eo a pr yeby y an u l a p reba f ra bam reba yas <04908B> 8aiiv.c'tc89.4olpcc89.4oqpcc8aS.2=olpox.9qpc89.4olpcc89.4iiivlpc89= ba nu leba hok aba ho ta bas y o kom a teba hok aba hin u k eba <04909B> qpctcoc't9.8c't89.olpc89.ctclpt9.2=89=89=9lpccct9.olpcc89.ctclp9= tr eo l a b l ba ok eba l kira y ba ba a k a ra o k aba rek a So: Pleo kao bany ho a oflobra prea a preba aiu rofo le pr(i)ba Brea yanu angaoba ake tire areba ate babare tire bamy Oanul tira-tireba omoa akl(i)baom rebaba mala ayanuy Ho kebare oyateba ho keba ate barekeba banu obang yamba Yanule baabaabayra banu tire bahoka-bahoke ba tirea Sreba ho keba Pao ho keba basleo apryeby yanula preba frabam rebayas Banu leba ho kaba ho taba syoko mateba ho kaba hi nukeba Tre olabl(i)ba o kebalkiray baba akarao kabareka Uh? Those two languages sound utterly different! What a surprise.... This "pronounceable Voynich" bit *is* quite useful after all. Peering at frequency tables, I would never have realized how different A and B were. The mystery thickens.... From j.guy@trl.OZ.AU Mon Jan 27 06:39:00 1992 Date: Sun, 26 Jan 92 16:39:00 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201260539.AA26941@medici.trl.OZ.AU> To: voynich@rand.org Subject: = = <9> (I suspect so) Status: OR ARE AND VARIANTS OF THE SAME LETTER? I can imagine you jump: "What? Three forms now for the same letter? This frogguy is out of his mind. Soon he'll have the whole Voynich alphabet down to two letters!" Good point. Yet when I look at a page of the Gospel in Beneventan script out of a marvellous book on medieval calligraphy which I was lucky to find remaindered at US$20 down from its original price of 60 pounds, I see three varieties of "i", two of "r", and no less than four of "t". Which form you should use is decided by the context, with sometimes some leeway. For instance, one "t" looks like the Voynich , another like our "z". The "z" you will find only word-finally. But you'll also find the -like "t" word-finally. Likewise, on that particular page, I see the tall "i" (looking very much like an "l") *only* word- initially. But the short "i" (without a dot) occurs also word-initially. So, you see, it's not all dry and clear-cut. But let me put my tongue in my cheek just an instant (and not too firmly, either). What if the Voynich was written in a real language, but the writing was a Bacon-type cipher? Would I not be right in reducing the alphabet to two or three letters? Each plaintext letter would be have two or three forms so that, under a meaningful plain text, there would be a hidden one, which would be one third or one-fifth the length of the plain text? Enough. Back to work. But first, because this is such a long article, I shall tell you how the movie ends: I strongly suspect that is a variant of , and therefore of <9>, but I cannot prove it. In my last posting I have argued that and <9> (Currier's and <9>) were variants of the same letter, the choice of which was determined by the letter immediately following. Currier wrote that "All the letters containing an initial c-curve are also the letters that can be preceded in the same word by the little letter that looks like c, e.g. , . On the other hand, the letters and <2> (which have very high frequencies) can *never* be preceded by c, *ever; they are instead preceded cy ". I had mentioned that I suspected a similar phenomenon in my Cryptologia paper, just on the evidence of folios 79v and 80r, long before I had read Currier. When I used Sukhotin's algorithm to identify vowels in the text of folios 79v and 80r I had merged the sequence as a single letter, on the strength that it was so very frequent. The first five vowels identified were, in this order: , <9>, , and . Among the consonants was , the shape of which is similar to that of "t" in the Beneventan script. Seeing that is identical to the letter "a" in the same script, I had hypothesized that perhaps was indeed an "a" and a "t"; by extension, that it seemed a plausible working hypothesis that and , similar to our "a" and "o" (and to those of other medieval scripts) could well be "a" and "o". Now, if <9> and are variants of the same letter, and if is "a", then, under that hypothesis, is another variant of "a". I have produced a file in which those sequences which are almost certainly single letters, such as , , , have been replaced by single letters. From that file I have produced again frequency tables (adjacency matrices, in the jargon of graph theory). <9> is counted occurring 8168 times, 5105, and 1492. , then is relatively rare, one fifth as frequent as <9>, one third as . I want to look at the possibility that is a variant of <9> and . Here is part of the adjacency matrix computed from the whole of D'Imperio's file. (In this table I have replaced my (Currier's <2>) by a capital S, which is very nearly the mirror-image of the corresponding Voynich letter, and thereby a better mnemonic. # & 2 4 8 9 = cc c't l; lp q; S 9 89 1 131 1661 865 173 922 6 418 34 634 86 220 cc - - 5 2 571 541 - 4 - 3 42 7 66 a - 5 1066 2 15 11 6 - 4 1 14 3 14 qp a c ct cg i ij iiv iv o t v x 9 469 44 186 858 2 3 8 - - 1012 - 3 343 cc 19 24 103 4 3 1 - - - 95 - 1 1 a 5 1 18 5 7 267 253 1607 732 10 - 76 984 You will have noted that <9> occurs overwhelmingly at the end of lines (# and =), and before <4>, , , , , , , , and . On the other hand, occurs before <2>, , , , and . The distribution of corroborates Currier's findings: what do <2>, , , , and have in common? Their first stroke is Voynich i-like: short, straight, thick, leaning left, exactly like the second stroke that makes up . So, you write before any letter written with an i-like stroke first. Elsewhere, you write <9> and very often a space after it (with I believe is there for purely aesthetic reasons). What about then? Well, 541 times out of 1492 you find it before <9>. Yes, 36% of the time occurs before <9> and 38% again before <8>. Thus out of 1492 occurrences of 1,112 are before <9> or <8>. So when you see , you are sure to see <8> or <9> next 75% of the time. In linguists' terms, the functional load of is very light, in Shannon's terms, carries very little information. That makes me think that is a variant of another letter, probably /<9>, because it looks precisely like the "a" of the Beneventan script. But the evidence so far in identifying it with /<9> is weak. We must look at other possible candidates: are there any better ones? What are we after? A letter which is as much as possible in mutually exclusive distribution with . That is, wherever you find you don't find that letter, and wherever you find it, you don't find . Where do we find ? Mostly before <8> (571 times against 127 expected), <9> (541 times, 173 expected), and (66 times, 21 expected). So we go down the columns for <8>, <9> and looking for letters that occur a lot less than expected. Well, I am not going to scratch my head and dig deep into my bookshelves to find the adequate statistical test. I'll just do a quick-and-dirty: for each candidate letter I'll divide the sum of observed occurrences by the sum of expected occurrences, and look at the three winners (those with the lowest ratios). Here are those three columns, painfully extracted by hand from the complete matrix: <8> <9> Score: cc 571 541 66 3.67 127 173 21 2 187 231 32 0.73 245 333 41 4 - 5 2 0.01 218 297 36 8 28 3290 14 2.58 511 697 85 9 865 173 220 0.71 697 949 116 c't 86 172 3 0.53 193 263 32 l; 2 9 - 0.27 16 22 3 lp 9 410 2 0.43 388 528 65 q; 13 30 - 0.32 53 73 9 S 21 84 7 0.52 86 117 14 qp 9 284 1 0.46 255 348 43 a 15 11 14 0.04 436 593 73 c 1816 989 82 2.48 461 628 77 ct 279 632 27 0.86 431 587 72 i 4 - 4 0.08 40 54 7 ij 13 7 8 0.35 32 44 5 iiv 178 75 36 0.75 153 208 25 iv 61 22 11 0.58 64 87 11 o 797 103 107 0.45 888 1210 148 t 42 537 4 2.22 104 142 17 v 9 5 3 0.85 8 11 1 x 633 258 128 1.08 373 509 62 And the three placegetters are.... <4> with 0.01, with 0.04, and with 0.08. Now <4> is disqualified: this weirdo occurs almost always before (2483 times out of 2869), so that it's in mutually exclusive distribution with everybody else. Which leaves the winner by a short neck. Now, to start with, I am the first to admit that my "statistic" is strictly mathematically incorrect. What I had in mind originally was to do a clustering analysis; those letters in mutually exclusive, or nearly exclusive distribution, I surmised, would be maximally distant in the tree produced. I got lazy. Second, you must bear in mind that my primary evidence for believing that could be a variant of was of calligraphic and paleographic nature. Calligraphy: has a for its second stroke and occurs before letters which have a -like shape for their first stroke; has an for its second stroke and occurs before letters which have an -shape for their first stroke. Paleography: both and appear in medieval scripts with the value of "a". But very importantly, I do not know of any reliable way of identifying variants of a letter of *purely* quantitative grounds, without taking into account the *shape* of the letters, only their distribution. Having studied some of Sukhotin's works, I strongly suspect that there must exist such methods. I have done some work on that problem, obtained what looks like promising results, but I wouldn't bank on them. Before I sign off, I would like to raise an objection to my thesis: "Frogguy, you say that <9>, and are three forms of the same letter; that <9> is written when the next letter is this or that, when it is that or this, and when it is something else again. *I* say that they're all different letters, and that the shape of the next letter is determined by them. What do you say to that?" That is a very good point, really. I honestly do not know what made me look at the data the way I did, not even contemplating one moment that the letter you have just written should decide the shape of the letter you ar now writing, rather the the shape of the letter you are writing now being influenced by the letter you have not written yet. Let us suppose that and are different letters and that they influence the shape of the next letter. The letter occurs 5106 times, 3.42 times as often as (1492 times). Let us now look for a letter that could have two shapes: one following , the other following . Presumably, we should expect the post- shape to be three times or so as frequent as the post- shape, and the two shapes to be reminiscent of each other, for instance, <2> and , or and . Let's have a look at the statistical evidence again: # & 2 4 8 9 = cc c't l; lp q; S qp a - 5 1066 2 15 11 6 - 4 1 14 3 14 5 cc - - 5 2 571 541 - 4 - 3 42 7 66 19 a c ct cg i ig iiv iv o t v x a 1 18 5 7 267 253 1607 732 10 - 76 984 cc 24 103 4 3 1 - - - 95 - 1 1 Well *I* can't see anything that remotely fits the bill. The prime candidates, <2> and , have such hugely different frequencies next to and that I cannot for the life of me imagine how they could be equivalent, even though, overall, <2> is about three times as frequent as . As for the next candidates, and , there is no compatible pattern: occurs more often following (natural, since is more frequent than ), and never occurs following . From j.guy@trl.OZ.AU Mon Jan 27 07:03:26 1992 Date: Sun, 26 Jan 92 17:03:26 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201260603.AA26954@medici.trl.OZ.AU> To: voynich@rand.org Subject: C = "e" O = "o" (pretty sure) Status: OR (* I am adding this opening parenthesis *after* I finished writing this message. I started out investigating those letters made up of those tall gallows-like characters on top of or (Currier's and ), that is, my , etc. Well, the results I got were not at all what I expected! In a nutshell: *is* "o" and *is* "e", and Languages A and B differ much in the same way as English and Scots do: where the English say "o", the Scots say "e" (usually spelt "ai"); likwise, where Author A of the Voynich writes "o", Author B writes "e". And here is the story of how I got to that conclusion... *) Just as I was proofreading my posting on = = <9>, something went *click* in my mind, and I searched for a particular page in Currier 1976 where he discusses those cases where the sequence is interrupted by one of the four tall gallows-like characters , , and . Here is what Currier says: "The Nature of the Symbols , , , ... In Herbal A material, in fact in all A material, this series is initially high; in B it is very low -- another way of identifying the "two languages." In Herbal A, the word-initial occurrences are as follows: all "word" first "word" initials of line cqpt 326 3 cq;t 67 1 clpt 82 0 cl;t 14 0 These "ligatures" seem to behave almost, but not quite, like , . In contrast, whether or not followed by <9>, , , or , the series , , , are *very* high in both "languages," and frequently as paragraph and line initials. The "ligatures" can *never* occur as paragraph initial, and almost never line initial. Therefore, , , and the like are symbols in their own right, and are *not equal* to or etc." Then it dawned on me that it may well be precisely the other way around. Perhaps etc. never occur line-initially because they represent a consonant cluster that cannot start a line, that is, that happens only when two words are run together. If so, would indeed be another way of writing either or . There are approximately 26,000 letters in the Language A corpus and 39,000 in the Language B corpus, counting as single letters , , , , , , , the four gallows-like characters ( etc.) and the four characters consisting of with a "gallows" (viz etc.). Since Corpus A is only two-thirds the size of Corpus B we should expect to find two-thirds as many , , etc in A as in B. Here are the frequencies I found: Lang. A Lang. B Observed Expected Observed cqpt 430 114 171 clpt 163 200 300 cl;t 20 9 13 cq;t 89 22 33 qp 1162 817 1226 lp 1292 1861 2792 q; 199 202 303 l; 60 64 96 ct 2845 1469 2204 Total: 6260 4759 7138 What, just what, is going on there? Only and behave reasonably, and to a lesser extent. is far too infrequent in A, and far too frequent. The most extraordinary thing is that (Currier's ) is twice as frequent in Corpus A as it is in Corpus B! To add insult to injury, its look-alike (Currier's ) is perfectly well behaved, occurring 900 times in Corpus A, and 1367 times in Corpus B. Let us look at what letters occur after the high-frequency suspects: , , and . I have shown the frequencies from Corpus A and Corpus B on alternate lines in the matrix hereunder. For instance, 8 c't 6 <-- is found 6 times before <8> in Corpus A 900 66 <-- (expected: 66 times) c't 80 <-- found 80 times before <8> in Corpus B 1367 135 <-- (expected: 135 times) ^ 1367 sum of row of c't in Corpus B (900 in Corpus A) I have also shown by an exclamation mark those cases where observed frequencies far exceed expectations, and by an asterisk those cases where the two dialects are at odds. For instance, in Corpus A, is followed 128 times by <9> (expected: 94), hence the exclamation mark; but only 44 times in Corpus B against 180 expected. Here, the two "Languages" behave in contrary manner. This is marked by an asterisk. 8 9 cc c't lp qp cqpt a c ct clpt iiv iv o x c't 6 128!* 35! - 7 6 10* 40 144! 8 6 - - 501!* 3 900 66 94 10 30 42 38 14 67 28 93 5 34 6 168 50 c't 80 44* 183! 1 14 6 20!* 25 853! 4 46! - - 77* 5 1367 135 180 40 46 95 42 6 104 113 75 10 26 19 179 97 ct 24* 524!* 78! 4 48 31 32* 231 394! 8 31! - - 1391!* 9 2845 208 296 33 93 134 120 45 211 90 294 17 108 20 532 159 ct 255* 108* 163! 4 37 16 39!* 36*1269! 3 107! - - 116* 15 2204 218 290 64 75 153 67 9 167 182 120 16 41 30 288 156 lp 3 159 95! 75 - - - 200! 99! 407!* - - - 249 1 1292 94 134 15 42 61 55 20 96 41 133 49 9 241 72 lp 6 251* 578! 42* 2 - - 1014! 611! 146 - 1 - 117* 12 2792 276 367 82 94 193 85 12 212 230 153 21 52 38 365 197 qp 6 165!* 32! 68!* 2 - - 145! 41 444! - - - 252!* 2 1162 85 121 13 38 55 49 18 86 37 120 7 44 8 217 65 qp 3 119* 163! 39 1 - - 350! 320!*115! - - - 107* 3 1226 121 161 36 41 85 37 5 93 101 67 9 23 17 160 87 Well, I don't know how those statistics make you feel, but they had me scratch my head in disbelief a long, long while. Look for instance at the column headed . In Language A and are followed by about half the time; in Language B one twentieth! Eventually... Eventually, I remembered my old hunches. That the putative vowels found by Sukhotin's algorithm *did* look like real vowels in real medieval scripts: , , , and might well represent the corresponding letters ("a" and "o") That, a long time ago, when I had my first crack at the Voynich, I hypothesized that might be "e", or perhaps "i". Look at the column headed , my old putative "e". In both Languages A and B it has good affinities with and . Which figure, since these are probably consonants and is probably a vowel. What do we see, though? That occurs extraordinarily more often following these four "consonants" in Corpus B than in Corpus A. What else? That occurs extraordinarily more often after the same consonants in Corpus A than in Corpus B. So, where Author A writes ("o"), Author B writes ("e"). Now that, folks, is a true linguistic universal. You find it all over the place, from Great Britain to Japan. Just think: English Scots to tae not nae lord laird bone bain both baith broad braid clothes claes cloth claith ...and so on, and so on, the list is endless. Let's have a closer look, shall we? c o c o Language A c't 144 501 16% 56% Language B c't 853 77 62% 6% Language A ct 394 1391 13% 49% Language B ct 1269 116 58% 5% Language A lp 99 249 8% 19% Language B lp 611 117 22% 4% Language A qp 41 252 4% 22% Language B qp 320 107 26% 9% In Language A is followed by 16% of the time, by 56% of the time. These ratios are reversed in Language B. Ditto for , , and . So, indeed, where Author A writes , Author B writes . Let me take a break. It's been a long day, and I need to stop and take stock. I'll just summarize what I think is fairly probable so far: The spaces between the words are spurious, an artifact of the Voynich alphabet system. The letters , and <9> are variants of each other. Languages A and B do indeed represent different dialects, as different as English and Lalland Scots. The letters identified as vowels by Sukhotin's algorithm probably represent the very same letters as are found in medieval scripts such as Beneventan, viz: = "a" = "a" <9> = "a" being a variant of and = "o" = "e" Now for a last-minute thought before mailing off. Having finished this article, I set about devising the "pronounceable Voynich scheme" which you have probably just read, just for fun, as relaxation. Well, I no longer know. I am under increasing doubt about whether some of my pronunciation rules might not turn out to be right. I am pretty sure that = "m" isn't right. But I am under growing suspicion that = "u" and = "n".... You see, the Voynich vowel system, as found by Sukhotin's algorithm, was all wrong and lopsided: putative "a", "e", "o" and perhaps a schwa, but no credible traces of "i" or "u". True, there are languages that have pretty funny vowel systems. Then, when you tried to read the text, you were hit on the head time and again with consonant clusters that just did not feel right. Suddenly, with = "u" and = "n" I get something that feels right phonologically, is pronounceable, and even makes sense in calligraphy: looks like "v", and looks like the "n" of some medieval scripts. There is a lot of testing to do all over again. Running Sukhotin's algorithm on a sample where has been replaced with amongst other things. Recomputing those adjacency matrices with this new alphabet too. What will come out of it? *Shudder*. From Karl.Kluge@G.GP.CS.CMU.EDU Sun Jan 26 22:09:00 1992 Date: Sun, 26 Jan 1992 22:09-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Petersen transcription Message-Id: <696481767/kck@G.GP.CS.CMU.EDU> Status: OR Jim sent me 2cd generation copies of sample pages from Petersen's transcription. The folios copied were 76v (babes in tubes), 1r (Jacobus de Tepenecz's signature), 2v (herbal folio -- Egyptian lotus?), 3r (herbal folio -- Cretan Dittany?), and one page I can't seem to locate a folio number for (I'll call it ?v -- what was this, Jim?). Here are my comments: Text: Legibility: on the 76v page some of the words appear to have been gone over in something like highlighter, rendering them illegible in a second generation copy. Same on 1r, 3r, and ?v folios to a lesser extent. This would definitely hinder using them as a source for a transcription. Some minor clipping of Petersen's marginal notes. Accuracy: unfortunately, I don't have copies of the originals of any of these folios, and checking Jim's checklist they don't appear to be in any of the usual books, etc., so I can't comment on the accuracy of the transcription. In any case, given a disagreement between the D'Imperio transcription and the Petersen transcription reference would have to be made back to a copy of the actual Mss. I'm worried about the subjectivity of word splits -- I disagree with D'Imperio in a number of cases on folio 56r, one of the cleaner, less cluttered folios (I should have thought to ask Jim for that particular folio to check this in Petersen). Illustrations: The lines are a bit light. Petersen indicates colors (for instance in the lotus? on 2v). He appears to have been able to reproduce the style of the Mss. drawings very well. He gives a tentative identification of the flowers -- Jim, any idea where he got them from? While I'm not sure that Petersen is very useful as source for transcription, the attempt to identify the plants may be useful as a crib into the text if combined with an old herbal to suggest possible plaintext. The numbering of the figures would give a frame of reference for discussing the illustrations. The marginal comments might point out interesting features. On the (n+1) plan, cost ought to be ~$13 + 1/n+1 * the $43 of the first generation copy. If n is large (>= 10), I'd say probably worth it for the Voynich lover in your life. If n is small (<= 5), you might want to save the cash for a copy of the Mss. from Yale. In any case, obviously a labor of love on Petersen's part. Karl From reeds@gauss.att.com Mon Jan 27 08:30:26 1992 Message-Id: <9201262330.AA01794@rand.org> From: reeds@gauss.att.com Date: Sun, 26 Jan 92 18:30:26 EST To: voynich@rand.org Status: OR I have prepared a table of contents of the Yale microfilm, correlating frame number with page numbers of the form 84v, etc, together with a list of ``incipits'' (a system of ``dicta probatoria,'' don'tcha know), so that given an image of J Random Voynich page you can quickly tell what page number it has. Jim G. has put it in file "finder" in directory rand.org:pub/jim Please send mistakes or comments to me, reeds@research.att.com Jim Reeds. From jim@rand.org Mon Jan 27 11:29:11 1992 Message-Id: <9201270229.AA12350@mycroft.rand.org> To: voynich@rand.org Subject: Re: Difference metric for text In-Reply-To: Your message of Sun, 26 Jan 92 21:06:00 -0500. <696477982/kck@G.GP.CS.CMU.EDU> From: Jim Gillogly Reply-To: jim@rand.org Date: Sun, 26 Jan 92 18:29:11 PST Sender: jim%mycroft@rand.org Status: OR > Karl.Kluge@G.GP.CS.CMU.EDU writes: > I'd like to perform a cluster analysis on the folios in the existing > D'Imperio transcription to see what structure falls out, and where > splits in the A/B language, the different "hands," and the different The metric I've used in the past is a simple-minded correlation coefficient between the digraph matrix on a sample page that's "obviously" one hand or language versus a digraph matrix for the sample under observation. It worked well enough to confirm the differences Currier observed on A vs B -- I didn't carry it further to try to cluster everything. Jim Gillogly From reeds@gauss.att.com Mon Jan 27 11:41:14 1992 Message-Id: <9201270241.AA03548@rand.org> From: reeds@gauss.att.com Date: Sun, 26 Jan 92 21:41:14 EST To: voynich@rand.org Status: OR Karl Kluge asks: I'd like to perform a cluster analysis on the folios ... ... Bennett suggests sum(i, 1, 26) Freq1(i) * Freq2(j), where Freq1 and Freq2 are the character frequencies in the two texts. Is there a better scalar measure? You should probabaly normalize these sums of Bennett, and use something like sum f1(i)*f2(i) / ( sqrt( sum f1(i)^2 ) * sqrt ( sum f2(i)^2 ) to adjust for differing sample sizes. There are zillions of formulas out there for quantities like this, incl. "entropy" formulas, like sum p1(i)*log(p1(i)/p2(i)) + sum p2(i)*log(p2(i)/p1(i)) (where p1(i) = f1(i)/ F1, etc, where F1 = sum f1(i), etc) and quantities like sum ( sqrt(p1(i)) - sqrt(p2(i)) )^2 or even sum ( sqrt (f1(i)) - sqrt(f2(i)) )^2 and chi-squared test statitics for the null hype that f1 and f2 are multinomial samples from the same population vs the alternative that they come from different pops: chisq = 2 sum ( (F1 f2(i) - F2 f1(i))^2/(f1(i)+f2(i)) ) / (F1*F2) and so on & so on. Some of these formulae might be more traditional in one field or another, but I don't think any is known to be universally better than any other. Jim Reeds. From kornai@Csli.Stanford.EDU Mon Jan 27 13:13:46 1992 Date: Sun, 26 Jan 92 20:13:46 PST From: Andras Kornai To: voynich@rand.org Subject: Words or not Message-Id: Status: OR It is true that frequency correllations across spaces are evidence for the spaces being spurious. But there are genuine linguistic phenomena, traditionally called "external sandhi rules" that can lead to such correlations, so this is not enough to establish the case for the wordbreaks being spurious. Are the initial and final frequencies different from medial? It looks to me they are, so the whitespaces are not distributed randomly. True, the principles governing their distribution can be orthographic, rather than phonological, but in general even orthographic conventions tend to be based on phonological regularities (e.g. the clusters that come out as "hard to write" in devanagari are the ones that appear most infrequently) so the Currier's case against words being what they appear to be is quite weak. My working hypothesis is that the text is a language (likely two or more dialects of the same language) spoken in or at the peripheries of medieval Europe, filtered through some semi-repetitive word-game such as English "Pig Latin" or Finnish "konttikieli" (for a good overview of such games, see Bruce Bagemihl's PhD thesis (cca 1990, U. of British Columbia), probably available from UMI) for purposes of disguise or better recollection (the mnemonic tradition I mentioned in an earlier posting), and written down in a phonemic, or near-phonemic script. Suppose you were part of a group of people in medieval Europe who, for one reason or another, wanted to communicate (yes, communicate -- books and mss are written to be read) a set of spells, prayers, recepies for the preparation of medicine, pieces of ancient wisdom, or whatever, in such a way that only the "in" group can read and write the stuff. How would you go about it? Well maybe you would create a cipher based on some combinatorial replacement rule of letters or letter groups and then transcribing the result in the Voynich script. But even the simplest such ciphers, such as rot13, are a major pain to read and to write, so I would go about it in a very different manner. First I would select a language that everybody in the in-group can be expected to know: let's say if we are dealing with the sacred writings of a heretic sect (this is not a serious hypothesis, just one of the many possible reasons why a group would want to communicate in code) then maybe every present or future member of the sect is expected to learn the language of the founder, say Aramaic. This leaves a considerable variety of languages to choose from, but excludes many others (e.g. Bantu languages). However, writing down a known language in a novel script would not provide sufficient secrecy, because I think medieval scholars (at least the ones interested in secret writing) were (a) perfectly capable of breaking such codes and (b) were avare that this could be done by others as well. So in order to provide an additional layer of secrecy, I would employ one of the "filters" used in oral secret languages, and write down the results of that in the private script. This means that new initiates must learn the alphabet and the rules of the word-game, but once they acquired these THEY CAN READ THE TEXT FLUENTLY, without the aid of tables or worksheets that you would need for rot13 and stuff. In fact, members of the group are likely to have talked to one another, when the occasion demanded, in "Voynese". Since this hypothesis is the simplest one I can think of that is compatible with the most salient feature of the text, namely its repetitiveness, and it still leaves a lot of ground to be covered, I see no reason to entertain more complex ones at this point. Let me summarize some of the specific conclusions that stem from the scenario described above. 1. The Voynich language is pronuncable and in all likeness *was* pronunced (unlike, say, rot13) -- word games are hard to acquire on written material alone. 2. The Voynich alphabet will match, item by item, (perhaps with some minor quirks) the phonemic inventory of a language that was spoken by a sizable (I'd say at least 20) group of people that were in contact with one another. Surely you could find the occasional slave brought back from Africa or Mongolia or wherever, but I see no reason to believe that such people, or Chinese bought back by Marco Polo etc. would have the means (vellum costs money) or the time to compose something like the Voynich. Sacred languages (Hebrew, Greek, but of course not Sanskrit) are good candidates, minor european languages (e.g. Finnish) are a good second choice, major languages of the time (including Latin) somewhat less likely. 3. The word game "filter" will fit into the known variety of such games, i.e. it will NOT involve arithmetic operations "take every third letter" of any sort. Rather, it will be templatic in nature. Andras Kornai From wet!naga@cca.ucsf.EDU Tue Jan 28 00:13:00 1992 Message-Id: Date: Mon, 27 Jan 92 07:13 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Initial assault; modest gains. Status: OR Report #1 on some initial statistical investigations, by naga. Jim Reed kindly sent me a machine-readable version of Mary D'Imperio's transcription, which is in Currier notation. The first thing I did was to separate out A and B, remove the line identifications, spaces and the MS-DOS end-of-line bytes, obtaining two files consisting purely of Voynich letters (according to the judgement of the transcriber). These files are VOYNICH.A 33,702 bytes VOYNICH.B 49,341 bytes The next thing I did was to run a simple frequency analysis of these files, with the following results: Size of file VOYNICH.A = 33,702 bytes Rank Letter Frequency Probability 1 / 5861 0.173907 2 O 5152 0.152869 3 9 2878 0.085396 4 S 2852 0.084624 5 A 2050 0.060827 6 8 2011 0.059670 7 E 1523 0.045190 8 C 1500 0.044508 9 R 1320 0.039167 10 F 1296 0.038455 11 P 1166 0.034597 12 - 1086 0.032224 13 M 1009 0.029939 14 Z 903 0.026794 15 4 631 0.018723 16 2 482 0.014302 17 Q 432 0.012818 18 B 200 0.005934 19 J 198 0.005875 20 N 193 0.005727 21 # 177 0.005252 22 X 163 0.004837 23 * 141 0.004184 24 W 89 0.002641 25 D 76 0.002255 26 V 60 0.001780 27 T 56 0.001662 28 I 47 0.001395 29 3 34 0.001009 30 6 34 0.001009 31 U 22 0.000653 32 Y 20 0.000593 33 K 11 0.000326 34 7 8 0.000237 35 0 5 0.000148 36 G 4 0.000119 37 H 4 0.000119 38 L 4 0.000119 39 , 3 0.000089 40 $ 1 0.000030 Size of file VOYNICH.B = 49,341 bytes Rank Letter Frequency Probability 1 / 7977 0.161671 2 C 5688 0.115279 3 9 5308 0.107578 4 O 5283 0.107071 5 8 3990 0.080866 6 A 3065 0.062119 7 E 2839 0.057538 8 F 2794 0.056626 9 S 2206 0.044709 10 4 1933 0.039176 11 Z 1367 0.027705 12 R 1366 0.027685 13 P 1228 0.024888 14 - 1121 0.022719 15 M 708 0.014349 16 N 556 0.011269 17 2 523 0.010600 18 B 303 0.006141 19 X 300 0.006080 20 Q 171 0.003466 21 J 152 0.003081 22 V 96 0.001946 23 T 78 0.001581 24 3 48 0.000973 25 # 44 0.000892 26 * 37 0.000750 27 U 34 0.000689 28 W 33 0.000669 29 D 22 0.000446 30 6 19 0.000385 31 Y 13 0.000263 32 G 9 0.000182 33 I 8 0.000162 34 7 5 0.000101 35 K 5 0.000101 36 H 3 0.000061 37 L 3 0.000061 38 0 2 0.000041 39 5 2 0.000041 40 _ 2 0.000041 I have so far developed two kinds of program for statistical analysis of files. One kind does frequency analyses, such as calculating the probability that a given letter occurs n places after some other given letter. I decided I needed to think more about how to distinguish anomalies, and so turned to the other kind, which searches for, and counts, repetitions. I immediately found that there were plenty of repetitions but due to (a) the size of the files, (b) the speed (or lack of it) of a 33 MHz 386 and (c) the likely non-optimization of my search algorithm (and other factors) it was going to take quite a while to gather the data, and there'd be a lot of it. I wrote a program to (attempt to) ascertain the length of the longest string of characters in a file which is repeated at least once. I discovered that in VOYNICH.A the strings: PAM/ZOE AM/ZOE/Z AM/ZOE/ZO /8AM/OFAM/ /SOE/O8AM/SO SCOR/SOE/8AM/Q9- occur at least twice (so the size of the longest repeating string is at least 16 letters - no doubt more). (Is ZOE the name of one of the nymphs?) In VOYNICH.B the following strings are repeated: C9/9BSC89 /SC89/9PC89 C89/4OFC89/9PC89/SC AM/ZC89/4OFCC89/4OPC so the size of the longest repeating string is at least 20 letters - again no doubt more. So I ran my repetition-count program on VOYNICH.B looking for strings of exactly 12 letters, and in ten minutes or so ( s l o w . . . ) it came up with: String number of occurrences /SC89/9PC89/ 2 PC89/4OFC89/ 6 C89-4OFCC89/ 7 /4OFC89/8AR/ 3 AE/SC89/2AR- 3 8AM/ZC89/4OF 3 PC89/4OFCC89 6 FC89/4OFC89/ 20 CO89/4OFC89/ 2 /SCO89/4OPCC 2 89/4OFO89/4O 2 PC89/4OPC89/ 9 89/8AR/SC89/ 2 AM/OFCC89/4O 3 OE/SC89/4OFC 10 C9/ZCOE/4OFC 2 9FCC89/SC89/ 2 ZC89/4OFC89/ 24 /OFC89/OFC89 3 /4OFC89/4OFC 24 Clearly the 89s are up to something, /SC89/ and /4OFC89/ particularly. Clearly also some way is needed to pick out the *significant* occurrences. (I mean *scientifically*. We can all see that FC89/4OFC89/, ZC89/4OFC89/ and /4OFC89/4OFC stand out clearly.) How many times might we *expect*, say, PC89, to occur in VOYNICH.B? My first approach to this question was as follows: The number of times a string of letters c1 c2 ... cn is expected to occur is N * p(c1) * p(c2) * ... * p(cn), where the p()'s are the probabilities of occurrences of single letters (taken from the appropriate table above) and N is the size of the file. (Actually it should be N-n+1, but let's not be picky.) For the purpose of calculating an expected value this assumes (what is false) that the probability of occurrence of a letter is independent of the letters occurring in its immediate vicinity. The problem with this approach is that almost *all* of the repetitions that occur, even those occurring just twice, seem to occur far more often than expected (according to this notion of "expected"). For example, if you multiply the individual probabilities of the letters occurring in a string of 12 letters, as above, and multiply this by 49,341 you generally get something much less than 1. Even for the shorter strings, I have to use a factor of 100 (i.e. just look at the strings that occur 100 times more than expected) in order to eliminate most of them, which doesn't seem quite right. However, I have not had time to reflect properly on these matters, so I may be missing a few things. Comment is welcome. Given the abundance of data available, what to look for? How to calculate expected occurrences so as to compare with actual occurrences. And perhaps Jacques can throw some light on what those 89's are doing. In the meantime I'll get back to what I "should" be working on (though I admit this Voynich stuff is rather interesting). From reeds@gauss.att.com Tue Jan 28 00:13:05 1992 Message-Id: <9201271513.AA10608@rand.org> From: reeds@gauss.att.com Date: Mon, 27 Jan 92 10:13:05 EST To: voynich@rand.org Status: OR BRUMBAUGH'S EXPLICIT TRANSCRIPTIONS (``removing the effect of the second substitution'') Question: how accurate a transcriber is Brumbaugh; what is the set of Voynich letters that he sees? Brumbaugh gives very few transliterations (as opposed to decipherments, translations, interpretations, etc), so its hard to get a good picture of exactly what he does. Both his Yale Library Gazette and Speculum papers give the ARABYCCUS and PEPPER examples, 15 letters in all. His Star Map paper and his Voynich newsletter give much more, here reproduced, with Currier equivalents. Alas, I have not completed an analysis of his star map transcriptions, nor have I checked them against photos of the MS: just against Petersen. Transcriptions from ``The Voynich `Roger Bacon' cipher manuscript: diciphered maps of stars'', Robert S. Brumbaugh, Journal of the Warburg and Courtauld Institutes, Volume XXXIX, 1976, pages 139-150. Brumbaugh gives explicit transcriptions of V characters (into digits) for folios 70r, 70v, and 71r. They are for nymph labels on some of the zodiac diagrams. In each case, reading clockwise starting at the top. Next to them are Petersen's readings of the same captions, with Petersen's numbering f 70r, Pisces inner ring P.I.1 135254* II.2 P.I.2 1354*52 II.3 P.I.3 134*52* II.4 P.I.4 814*5254* II.5 <8OEARAJ> P.I.5 185254* II.6 P.I.6 136254* II.7 P.I.7 794*14*2 II.8 <2AEOE2> P.I.8 1394*894* II.9 P.I.9 9314*8*6 II.10 <9FOEAM> P.I.10 13524* II.1 outer ring P.X.1 139 189 IV.15 P.X.2 139 12 IV.16 P.X.3 18*54*5 IV.17 P.X.4 13089 IV.18 P.X.5 1354*8 IV.19 P.X.6 1354*82 IV.20 P.X.7 13189 IV.21 P.X.8 13*975* IV.22 P.X.9 7*67*3*39 IV.23 P.X.10 1354*9 IV.24 P.X.11 1354*252 IV.25 P.X.12 1354*89 IV.26 P.X.13 13614*9 IV.27 P.X.14 13989 IV.28 P.X.15 13662 IV.29 P.X.16 1354*54*8* IV.11 P.X.17 1352*9 IV.12 <9FAR9> P.X.18 1352 IV.13 P.X.19 139 IV.14 f 70v, Aries 1 inner ring A1.I.1 1354*8*6 I.2 A1.I.2 136154*14*7 1294*9 I.3 A1.I.3 1361669 1354* 13654*52 I.4 A1.I.4 13652 125989 I.5 A1.I.5 13189 I.1 outer ring A1.X.1 12*7*69 254* III.7 A1.X.2 13548*456 III.8 A1.X.3 1354*54* III.9 A1.X.4 13989 III.10 A1.X.5 154*7*64** III.11 A1.X.6 137*189x2 III.12 A1.X.7 --- III.13 A1.X.8 137*(7*)89 III.14 III.15 A1.X.9 1354*7*9 352 54**89 III.6 f 71r, Aries 2. inner ring A2.I.1 1314* 15252* I.2 A2.I.2 136614* I.3 A2.I.3 1314*7*8 I.4 A2.I.4 1364*852 I.5 A2.I.5 1314*7*89 I.1 outer ring A2.X.1 13154*89 IV.8 A2.X.2 136154*9 IV.9 A2.X.3 13154*9 IV.10 OFCOAE9> A2.X.4 1354*639 IV.11 A2.X.5 1354*352 IV.12 A2.X.6 76529 IV.13 A2.X.7 131369 2529 IV.14 A2.X.8 13612 5252 IV.15 A2.X.9 34*852* IV.6 A2.X.10 13614*52 IV.7 Transcription of first few lines of f 106v, by Brumbaugh, ``Voynich Newsletter'', Nov. 1978. The newletter of Nov 1978 has a fairly clear picture of f.106v, and a transcription of the first 4 lines by B (in his own terrible handwriting) into what he sees as V letters and their digital equivalents, together with a "box" similar in layout to the top picture on D'Imperio's figure 26, page 104. Brumbaugh's box is: digit Voynich Latin 1 O,v AJV 2 RJ BKR 3 CFB CLW 4 4E) DMS (X) 5 A ENX 6 P FOT 7 S2Z GPY 8 8w HQU 9 9V I, -US, Z The symbols in the 2nd column are drawn Voynich letters, as seen by Brumbaugh. Here I use our closest Currier equivalents, except: "v" is like a lower case Roman 'v' or Greek 'nu' or American 'check mark', "w" is a zigzag scribble, like a Currier M or N without the final curved upward stroke, and ")" looks here like a right parenthesis or (only in this box, but not in B's transcriptions) like a "partial derivative" sign with a much smaller loop than usual. I enlarged his picture of f.106v and was able to turn it into Currier, with only a few doubtful word breaks and no real doubt about any of the letter assignments. Here is what I got: VZC8A.EOAN.OBAUAE.OPCO.8AM.SOBS8.OPAT.FAR.AEAEOR.AM.AE[J9].FAR 8AM,AE.ZCCO8AR.9SPAN.ZAROPAR.4OFAR.OFAR.ZC8.ZCO,FCORAN.AJ,S9 9ZC8AM.ZX9.SCOFS9.2AR.AE.SOESC89.9PAN.OPA2.AE.CC89.8AE9.EO89 9SCOPA.EZCO.AM.SQ9.OFAN.S8AE.S8AJ.SARAJ And here are B's digits: 97385 4184 138254 1635884 71379 1682 352 545412 84 549352 884 54 7331852 97684 7525652 738 731 331284 5279 973884 77339 731379 752 54 7147389 9684 1652 54 789 85492189 973165 473184 77639 1384 7854 7852 75252 Things seem to line up as follows: VZC8A EOAN OBAUAE OPCO 8AM SOBS9 OPAT FAR AEAEOR AM AE[J9] FAR 97385 4184 138254 1635 884 71379 1682 352 545412 84 549 352 8AM,AE ZCCO8AR 9SPAN ZAROPAR 4OFAR OFAR ZC8 ZCO,FCORAN AJ,S9 884 54 7331852 97684 7525652 738 731 331284 52 79 9ZC8AM ZX9 SCO FS9 2AR AE SOESC89 9PAN OPA2 AE CC89 8AE9 EO89 973884 773 39 731 379 752 54 7147389 9684 1652 54 789 8549 2189 9SCOPA EZCO AM SQ9 OFAN S8AE S8AJ SARAJ 973165 4731 84 77639 1384 7854 7852 75252 Look first at the gaps. The two missing words in line 2 look like homeoteleuton. The 789 in the antepenult of line 3 looks like a plausible reading of CC as S. More interesting are 77339 for ZX9 and 77639 for SQ9. In greater detail here is how he transcribes. I have adopted his word divisions, which are all reasonable. Column 1 has my reading, in Currier letters. Column 2 is Brumbaugh's transcription into Voynich letters, given here as much as possible in Currier. He has a scribly w or m like letter, like Currier's M or 3, but without the final upward swash, which I denote `w' here, and he has a letter like a backwards c, which I denote ')'. It is pretty clear that our AM and AN both get turned into w). It is not clear whether Brumbaugh takes the ')' as the same as Currier's D or just as its final upward curved stroke. Column 3 is his further transcription into digits. Column 4 has my comments VZC8A VZC8A 97385 EOAN EOw) 4184 # here AN is w) OBAUAE OBwRCE 138254 OPCO8AM OPCA8w) 1635884 # here AM is w) SOBS9 SOBS9 71379 OPAT OPwR 1682 FAR FAR 352 AEAEOR AEAEOR 545412 AM w) 84 AEJFAR AE9FAR 549352 # the J could very well be a 9 8AM 8w) 884 AE AE 54 ZCCO8AR ZCC08AR 7331852 9SPAN 9SPA) 97684 ZAROPAR ZARAPAR 7525652 4OFAR _____ _____ # eye skip OFAR ____ ____ # eye skip ZC8 ZC8 738 ZCO ZC0 731 FCORAN FCO2w) 331284 AJS9 AJS9 5279 9ZC8AM 9ZC8A) 973884 ZX9 ZXC9 77339 SCOFS9 SCOFS9 731379 2AR 2AR 752 AE AE 54 SOESC89 SOESC89 7147389 9PAN 9Pw) 9684 OPA2 OPAR 1652 AE AE 54 CC89 S89 789 8AE9EO89 8AE9RO89 85492189 9SCOPA 9ZCOPA 973165 # vertical eye skip. EZCOAM EZCOw) 473184 SQ9 SQC9 77639 OFAN OFw) 1384 S8AE S8AE 7854 S8AJ S8AP 7852 SARAJ SARAJ 75252 Assuming his transcribed values we can build his V letter to digit map, here expressed as a kind of histogram. O 1111111111111111 J 22 R 222222222222 B 33 C 353333333333336 F 33333 ) 444444444 E 444444444444 A 5555555558555585555555555 P 66666662 S 7777777777777 Z 777777777 w 8888888888 8 888888888888 9 99999399999993 V 9 2 27 Q 7 X 7 From wet!naga@cca.ucsf.EDU Tue Jan 28 00:48:00 1992 Message-Id: Date: Mon, 27 Jan 92 07:48 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: P.S. to First Assault Status: OR I should point out that those 20 12-letter repetitons identified in VOYNICH.B were only from the first 2.6% of the file. There are no doubt many more, and many longer sequences which are repeated. My 386 would probably require about 3 or 4 hours to find all the 12-letter repetitions (not to speak of the 13, 14, ...). Of course, I could try to improve the search algorithm. The use of a Cray would be handy, too. BTW has this all been done before? From j.guy@trl.OZ.AU Tue Jan 28 02:33:05 1992 Date: Mon, 27 Jan 92 12:33:05 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201270133.AA27495@medici.trl.OZ.AU> To: voynich@rand.org Subject: Arbitrary spaces. Status: OR Andras Kornai wonders: "I must have missed this -- could you please elaborate what makes you think so? Certainly the script looks as if it was written in the ordinary manner, with spaces separating words. So in order to accept your hypothesis, I will need lots and lots of evidence." I do not have lots of evidence. Only Currier's puzzled observation that many word finals strongly correlate with the initials of the following word. I have left my... stupid frog! The original ASCII file is on disk... let me look for it. Here: ------------------Start of Quote------------------------------ 5. EFFECTS OF THE ENDINGS OF ONE "WORD" ON THE BEGINNING OF THE NEXT "WORD". You remember I mentioned that some "word"-finals have an obvious and statistically-significant effect on the initial symbol of a following "word." This is almost exclusively to be found in "Language" B, and especially in "Biological B" material. For example, we have: "word" beginning with: is preceded by <4o> "word" ending in: or <2> or series 13 7 91 <2> series 10 2 68 series 23 0 275 <9> series 592 184 168 "Words" ending in the <9> sort of symbol, which is very frequent, are followed about four times as often by "words" beginning with <4o>. That is a fact, and it holds true throughout the entire twenty pages of "Biological B." It's something that has to be considered by anyone who does any work on the manuscript. These phenomena are *consistent*, *statistically significant*, and hold true throughout those areas of text where they are found. I can think of no linguistic explanation for this sort of phenomenon, not if we are dealing with words or phrases, or the syntax of a language where suffixes are present. In no language I know of does the suffix of a word have anything to do with the beginning of the next word. --------------------End of Quote---------------------------- That constitutes prime evidence that words have been broken up by spaces. Note what Currier says: "I can think of no linguistic explanation for this sort of phenomenon, not if we are dealing with words". The afternoon discussion of the 1976 seminar opens with this question: "Question (Speaker not identified): How do you account for the full-word repeats? Currier: That's just the point -- they're *not words*!" So, Currier just *knew* those "words" were not words, only he just could not quite put his finger on what was going on. Or perhaps he could, but it did not occur to him, off-hand, to put it forcibly and emphatically enough at the time. Let me give you an example. Imagine that I were to write a space after every t when I writ e in English, and merge t he rest oft heremain ingwo rdsrat her ran domlyli keI've just beendo ing right now. Since "th" is a very frequent digraph, you would observe a strong correlation between words ending with "t" and words beginning with "h". In fact, I would not have to write a space after each and every single "t": a strong tendency to do so would be enough to bring out the type of pattern observed by Currier. Continuous writing, and breaking it up into chunks which have nothing to do with words is a common feature of writing systems. In Thai, for instance, you write continously, with occasional spaces dictated by whim or aesthetic reasons (although recently there has been a tendency to separate words, but that is the influence of English). If now you turn to Arabic, you will learn that the writing is connected. However, the shape of certain letters is such that they can connect not at all, or only to the left or to the right. If, say, you write "shams" ("sun"), all three letters (sh, m, s) connect. But if you write "'arD" ("earth") all three (alif, r, emphatic d) are disconnected. I have come to believe that the Voynich alphabet has somewhat similar properties. In particular, that Currier's (my ), because of its shape, it prevented from connecting to the next letter, and forces a spurious space. Same of <9>. In general, it seems that letters which end in a flourish way above or below the line of writing tend to disconnect, thus the series (Currier's , and ), <9>, and Currier's , <6> and <7> (my , and <&>). The problem we immediately face, then, is to identify the words in this continuous writing. I do not have any satisfactory solution. From j.guy@trl.OZ.AU Tue Jan 28 03:00:02 1992 Date: Mon, 27 Jan 92 13:00:02 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201270200.AA27517@medici.trl.OZ.AU> To: voynich@rand.org Subject: Pronounceable Voynich Status: OR For those who feel so inclined, here are the rules for translating from Currier's notation into pronounceable Voynich using my program TRANSLIT. What good is it? I find that looking at something that seems easily pronounceable helps at recognizing patterns. Here: 4=h 8=b 9=a 2=s E=m R=y S=r Z=l P=t B=p F=k V=f Q=tir W=pir X=kir Y=fir G=im H=um 1=nim T=iy U=ny 0=iny D=u II=n N=iu M=nu 3=inu J=ng K=ing L=ung 5=ning 6=eng 7=eng A=a CC=a C=e O=o I=i /=. ..=. ,=, #=# -== *=* From jbaez@math.mit.edu Tue Jan 28 03:07:35 1992 Date: Mon, 27 Jan 92 13:07:35 EST From: jbaez@math.mit.edu Message-Id: <9201271807.AA25007@paley> To: voynich@rand.org Status: OR Okay, let me warn you that I have not mastered this modem and have no way of deleting characters. When I screw up, yo u will see something like that. Nate Osgood has been doing some cluster analysis of obsidian artifacts so he might have some bright ideas along these lines. In partuicular, he can get Matehematica to do animated rotating 32d picutures of clusters. This allows for a visual study of data which has been bioiled down to 3 numbers per sample. Kluge mentions a statistic of Bennett's that essentially is a correlation of letter frequencies. It would be easy and much better to soup this oup to count digraphs, trigraphs etc.. I would argue that spaces should count as "letters:" in doing this analysis since they seem to play a curious role. These methods of course just put out a numerical estimate of "similarity". To use Nate's tool, one should instead think of 3 (or more) numerical qualities that can be determined for any page (or larger sample of text). One then animates a 3d graph with a bunch of doests on it (e.g., one dot per page) and checks to see if they cluster into 2 lumpts (hands A and B) or more (depending on subject matter, for example). This would be very fun to see. Some ideas for numerical qualities to use: entropy per character, "voewwelishness" (as measured by a Sukhotin-like algeorithm) of various charactersfreq, frequenty ofcy of word-initial and word-ending appearance of various characters, etc.. If some one wants to crank out some statistics for pages of text (keeping track of whehether they are purported "A" or "B", I can probably persuade Nate to dump some lists of triples into his Mathematica program and give them a whilrl to eyeball clusters. Of course, pronounceable Voynich --- or more precisely, readable Voynich --- could be argued to make use of human pattern-recogniction abilities at a much more fine-greained and subtle level. jb From jbaez@math.mit.edu Tue Jan 28 04:13:12 1992 Date: Mon, 27 Jan 92 14:13:12 EST From: jbaez@math.mit.edu Message-Id: <9201271913.AA25178@paley> To: voynich@rand.org Subject: heh Status: OR Just oggot my posting and - hey - no typos visible! On my screen the delete key doesn't work, but *you* guys don't have to see all those damn "control H's" that I see. (JUstust in case you were wondering what the $#^$*&@ I was talking about.) jb From reeds@gauss.att.com Tue Jan 28 06:48:33 1992 Message-Id: <9201272156.AA20506@rand.org> From: reeds@gauss.att.com Date: Mon, 27 Jan 92 16:48:33 EST To: voynich@rand.org Status: OR Andras Kornai writes: My working hypothesis is that the text is a language (likely two or more dialects of the same language) spoken in or at the peripheries of medieval Europe, filtered through some semi-repetitive word-game such as English "Pig Latin" or Finnish "konttikieli" (for a good overview of such games, see Bruce Bagemihl's PhD thesis (cca 1990, U. of British Columbia), probably available from UMI) for purposes of disguise or better recollection (the mnemonic tradition I mentioned in an earlier posting), and written down in a phonemic, or near-phonemic script. Can you give us a brief explanation of a few such word game filters? Jim Reeds. From RJB@MAX.U.WASHINGTON.EDU Tue Jan 28 09:42:00 1992 Date: Mon, 27 Jan 92 16:42 PST From: RJB@MAX.U.WASHINGTON.EDU Subject: the infamous 4o To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Uh -- and what of "qu" in English? --rjb From kornai@Csli.Stanford.EDU Tue Jan 28 10:56:31 1992 Date: Mon, 27 Jan 92 17:56:31 PST From: Andras Kornai To: voynich@rand.org Subject: Word game filters Message-Id: Status: OR Jim Reeds asked for a "brief explanation of a few such word game filters". First the name: `word game' is perfectly standard terminology, `filter' cam from my (at the moment, extremely Unix-centered) mind. So a `word game filter' is an algorithm that takes words or strings of words as input and produces word-like objects (accidentally, real words) and strings of such as output. One example (the first one I ever learned) is the "rg-speech" played by Hungarian kids: roughly speaking, we insert an rg after every vowel, and repeat the vowel. For instance, the phrase tudsz igy besze1lni?* 'can you speak this way?' becomes *(There are a number of digraphs in Hungarian orthography, such as sz, gy above, and ascii forces a few more, such as e1, which in printed text is an accented e). turgudsz irgigy bergesze1rge1lnirgi? The correct answer is, of course, turgudorgok which, when run through the inverse filter, becomes tudok 'I can'. Once the game is initiated, kids (and their parents) can go on speaking this rg-speech for hours, without any hitches in the communication. It is not very effective as a disguse (though it does shield the meaning quite well from very small kids and foreigners whose Hungarian competence is limited) and for statistical analysis the frequency of rg clusters would be a dead giveway. However, there is a large variety of word games that do not involve a fixed "filler" like rg but use material from nearby syllables (e.g. the next one) to construct the filler dynamically. In a simple version of such a word game, our first example would become: tugyudsz ibigy beszesze1lne1lnirgi? (Reverting to the default rg when there is no next syllable to borrow the filler from). This is much harder to learn to produce and decode, but quite doable. There is a large variety of such systems (if people find the Bagemihl reference hard to track down I'll come up with something more accessible) and many of them provide an effective way of blocking overhearers from understanding what's going on. Check "speech disguise" and "word game" in your friendly library keyword search index. Point is, these word game filters tend to create exactly the sort of repetitiveness that we see in the Voynich. Andras Kornai Ps. Jacques Guy writes "so very always (pardon my English!)..." but surely Jacques, you must have meant "so very almost":-) From reeds@gauss.att.com Tue Jan 28 13:55:55 1992 Message-Id: <9201280456.AA29703@rand.org> From: reeds@gauss.att.com Date: Mon, 27 Jan 92 23:55:55 EST To: voynich@rand.org Status: OR Mostly for John Baez, Jacques suggests taking the digraphic frequency table, and for each pair of rows, computing the correlation coefficient for the two rows, and finally using the resulting matrix of correlations as a similarity matrix for a cluster analysis. He expects 2 or 3 main clusters. Another procedure is: take the ``signgular value decomposition'' (SVD) of the digraphic table, looking for the 2 or 3 dimensional linear space which comes closest to all the rows. The rows, projected onto this space, now lie in a 2 or 3 dimensional space (of course), with funny coordinates. Now Nate's 3d visualization program comes in: can you spot the clusters. (If Jacques is right, and there are essentially <= 3 clusters, the whole shebang should fit in a 3d space.) You might also repeat the process, with each cell in the digraphic table replaced by something like sqrt(x)+sqrt(x+1), where x is the digraphic count: this stabilizes the variances, and hence makes use of Euclidean invariance more palatable. A nice feature of using the SVD here is that you get the corresponding analysis for columns (representing fore-context instead of aft-context) essentially for free. Jim Reeds From reeds@gauss.att.com Tue Jan 28 15:45:50 1992 Message-Id: <9201280645.AA01908@rand.org> From: reeds@gauss.att.com Date: Tue, 28 Jan 92 01:45:50 EST To: voynich@rand.org Status: OR In private email, Karl Kluge asks about my recent Brumbaugh posting: What do the * represent in the Brumbaugh zodiac folio transcriptions? Since others might want to know, I answer in public. A hint of an answer is on page 116 of his book, but he is explicit in his Warthog Institute paper: each digit of the intermediate digital text is written in any of several V symbols: either as itself, an Arabic numeral (as the 4, 8, and 9 often are), or in one of several variant or cipher symbols. When Brumbaugh transcribes a 7 he means the Arabic numeral 7 appears in the MS. When he writes 7* he means one of the variant symbols appears, which are Currier's S and Z. I had hoped to figure out what the * variants were in the star paper (his rules are slightly different in each paper) but didn't have time. It shouldn't be too hard to reconstruct it from my posting. Similarly, Brumbaugh is reading ZX9 Currier as something like etcKtG in Bennett (I use mixed/lower case for ligatured forms, hence cKt is Currier X). Yes. But its clear from his drawings in the Voynich newsletter that he is applying some kind of ``orthographic sandhi'', turning what I see (and he writes) as ZX9 into ZXC9 when it comes time to transcribe it into digits. Does he do this consistently? I would dearly like a more precise description of his transcription system. My preliminary assessment of Brumbaugh's transcriptions: not very precise. I make lots of transcription errors, and know it, and have learned to double check all over the place. Brumbaugh makes more errors than I do. He skips words. His actual skip-free transcriptions are not bad, but contain a few quirks (such as insertion of C after X). Jim Reeds. From rcarter@isis.cs.du.edu Tue Jan 28 15:52:09 1992 Date: Mon, 27 Jan 92 23:52:09 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9201280652.AA00941@isis.cs.du.edu> To: voynich@rand.org Subject: How about the `jars' on 99r? Status: OR Somewhere in all this talk about what constitutes a `word' in Voynich there might be room to include the `jars' that appear on 99r... Is it in any way safe to think that the labels of the `roots' are stand-alone `words'? Labels such as SOF9 and OF9 etc.? Or are these abbreviations of some sort? If these -are- `words' then perhaps some sort of rules can be built from them. Just a thought. From j.guy@trl.OZ.AU Tue Jan 28 23:17:26 1992 Date: Tue, 28 Jan 92 09:17:26 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201272217.AA28276@medici.trl.OZ.AU> To: voynich@rand.org Subject: Words or not Status: OR Andras Kornai has made excellent points there, and I heartily agree with every one of them. I wish I had thought of them myself! If he is right, then we can tackle the problem *as if* the Voynich were plaintext in a real language (a weird language, yes, but you should see what goes on in New Guinea!) From j.guy@trl.OZ.AU Tue Jan 28 23:40:59 1992 Date: Tue, 28 Jan 92 09:40:59 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201272240.AA28297@medici.trl.OZ.AU> To: voynich@rand.org Subject: Words Status: OR Here is a page transcribed from an 11th-century Gospel Lectionary written in the Beneventan script. Square brackets denote abbreviated forms. [rum], for instance, is a sign exactly identical to the symbol for Jupiter. boreseorum introistis; Exciuita te aut[em]illa multi crediderunt ineum samaritano[rum]. propter uerbummulieris testimonium p[ro]hibentis. quia dixit michi om[n]ia quecumq[ue] feci; Cum uenis sent ergo adillum samararitani. rogabant eum utibi maneret. et mansit ibi duobus diebus; Etmulto plures credideruntp[ro]p ter sermonem eius. etmulieri dicebant. quiaiam nonpropt[er] tuam loquelam credimus. ipsi enim audiuimus etsciemus. q[ui]a hicest saluator mundi; Well, despite that "Jupiter" symbol, it wasn't an astrological text, was it? You will have noticed that words are run together (in+eum, uerbum+mulieris, ad+illum, etc.), and broken by end-of-lines (Ex+ciuita-te, crediderunt+prop-ter), and end-of-pages (...bores eorum, i.e. la-bores eorum continued from the previous page). I have already argued that spaces do not mark words in the Voynich manuscript, and I wonder if end-of-lines do. From the reproductions I have seen, lines are filled to the right margin, except at what is clearly the end of a paragraph. Therefore... therefore... I suspect that words may be similarly broken. How do we identify words, then? Why didn't I think of this sooner? Make a list of all paragraph-initial strings of n letters, sort them, and see. Rather, two lists: one of Language A, one of Language B. I do not knw what will come out of it: I am only about to do it. Done. Now, I must admit, I had no preconceived idea of what to do with that stuff. Here is what I did. I said to myself: I'll only look at string 3 or more characters long (I'd used the original file in Currier's system) that occur at least 3 times (all perfectly arbitrary and groundless, but you've got to start somewhere), and I'll count how many times those strings occur in the whole file, stripped of spaces. I took the sorted list of paragraph-initials from Language A (which had more entries). Here goes: String Paragraph-initial Total Comments in language A frequency 4OP 3 543 B89 3 10 B89A 2 5 BOZ 3 4 BSAR 2 7 BSA 3 19 BSC 3 128 BSCO 3 17 BSCOE 3 10 BSCOEZ 2 3 BSCO8 4 5 BSCOQ 4 4 BSO 16 62 BSOR 3 20 BSOR4 1 2 BSOR9 1 2 BSORO 1 6 hey, that's better! FOA 3 5 FOR 2 76 something there? FORA 2 12 possibly... FORAR 2 5 FSO 7 175 something there? FSOE 3 60 yes...? FSOES 2 20 yes. FSOR 2 56 something there too? FSORZ 2 6 not here FZO 5 30 ? FZOE 2 9 not too rich FZOR 1 6 perhaps? POE 3 170 a bonanza? POE8 2 40 looking good POE8Z 2 2 must be elsewhere POEO 1 35 looking good POEOR 1 2 must be elsewhere POR 3 82 looking good PORO 1 19 still looking good POROR 1 4 a word? PORS 1 18 looking good PORS9 1 7 a word? PORZ 1 12 yes? PORZO 1 7 a word? PSO 9 198 bonanza? PSO8 2 28 looking good PSO89 1 14 a word? PSO8A 1 10 another? PSOR 4 60 looking good PSORO 1 8 a word? PSORS 1 20 wow! a word for sure! PSORSO 1 12 continued.... PSORSOR 1 6 that's it! Then the devil got over me, and I searched for 'SORSOR'. 20 of them, no less. What was that juicy one again? POE8. Count the EO8: 422. Must be part of many common words. And there was PORS9, with 7 occurrences. ORS9 occurs 32 times. SOES (which I got from FSOES, 20 times) occurs 100 times. I had no idea of what I ought to do when starting. I still have no clear idea of how to put into words what I found myself doing. I'm sure not going to continue, because it's tedious, boring, repetitive, and a pain. I am going to stop and just think about it, or better still: sleep on it. But I know what to do next: same stuff, except that you look at the strings, reversed, that occur at the end of paragraphs, and you count them in the corpus, back to front. What happens there is that as soon as you extend your string beyond the end (or the beginning) of a word, its frequency does a 1987 Wall Street crash. However, in a language that features external sandhi phenomena (where the end of a word influences the beginning of the next), your string frequencies crash before you reach the end or the beginning. What you get are not the full words, but their cores. Example of a language with external sandhi phenomena, when written phonetically: Dutch. "Ik ben" is pronounced "ig ben", and "ik zou" is pronounced "ik sou" (as you see, the influence can go sometimes left, sometimes right in the same langauge, depending of the sounds involved). I suspect that the Voynich language does that, but I might be wrong. One more thing about what I have written about extensive vowel/consonant harmony. If you have a language that goes mainly consonant-vowel-consonant-vowel-consonant-vowel, you are going to observe statistics that look strange, since every vowel will be 2k positions away from any other vowel and 2k-1 positions away from any other consonant. It's only the odd vowel-vowel, consonant-consonant sequence that will dilute the pattern a bit. Or if you count the spaces between words, or the end-of-lines as one position... ------------------------------------------ Next day (Tuesday, early morning) I remembered an article, in Lingua, or Word, or something like that, around 1945, by Zelig Harris where, in different terms, he mentioned that "frequency crash" phenomenon. Actually, I do have the tool to identify words, but I don't know how to use it *efficiently*. It's my MONKEY program, which you'll find (executable only) in pub/jim at rand.org. It's big (55K) because of the user interface, but the algorithm itself is terse and simple. Say you want to build a character monkey. That is, a monkey that apes an input text on an n-character basis (nth-order monkey). First you make a list of *all* n-character strings in the input text. Next you sort it into some alphabetical order. Next you count the frequencies of n-1 character strings. Facing the first occurrence of each n-1 character string, you record that frequency; facing each nth occurrence you write 1-n, so: -1, -2, -3... That gives you the offset to the first occurrence. So, when you look for a particular string, using, say, a binary search, you can get right to the beginning of the list of all the occurrences of that particular string, and know immediately how many there are. Then... since we are not into aping the Voynich but into analyzing it, I won't tell you what comes next (you can guess easily, anyway, and how that little trick lets you compute the nth-order entropy fast and easy). Say you've built a 20th-order monkey. You have a list of all 20-character strings ordered alphabetically, and their frequencies counted. Look at those frequencies. If one occurs more than once (unlikely, unless your input text is *very* long) you can be pretty sure that it contains at least one *complete* word. That complete word necessarily occurs elsewhere in your list too (think about it). We know that it is shorter than the current value of n. So, we decrement n, and we do a frequency count again. It's easy, it's fast, since all those strings are already ordered alphabetically through n+1 characters (remember? you've just *decremented* n). We now look for the two n-character long substrings in our original n+1 character string, and note their frequencies. We continue decrementing n and looking for substrings until we hit some the frequency of which jumps up suddenly. Those we mark as probable words. Now, many of those probable words will be common word sequences, such as "of the", "in the", so that is not the end of it, and we continue. Incidentally, you'll notice that if the text is transcribed using my less-than-letters system, the algorithm will end up identifying the true letters! The problem I am faced with is: how big a frequency jump is big enough? Tell you what: this is the first time that I put down this embryo of an algorithm in words. It's really very simple, isn't it? And it seems, with this frequency-cum-negative- offset monkey scheme of mine, that it would be quite fast. Of course, you don't store all those strings, just pointers to them. Here, I have a hardware problem. On a PC, you are limited to 640K minus what DOS takes. You can stuff some in high memory and end up with 620K or so free. For a text the size of either the A or B corpus, you need 4 bytes to address each letter. Say we have 100,000 letters in each corpus. That's 100K for the corpus, and 400K for the index. There go 500K already and you haven't even started writing any code! Yes, I know, I'm just looking for excuses for not starting programming it. And I can use my 7 Meg of extended memory with a bit a poking about the BIOS. All right, how about this excuse: how big a jump in frequencies is big enough? Help!!! There may be another glitch. If the Voynich language is like Arabic, my idea will not work well at all. "Book" is "kitab", "books" is "kutuub". "River" is "nhar", "rivers" is "anhaar". "Key" is "miftah", "keys" is "mafaatih". Tough, eh? That goose (geese) isn't cooked yet. Incidentally, folks, the segmentation of continuous text ("text" in its widest possible meaning, could be the pixels of an image) into its natural constituents would have commercial applications. Trouble is, I never could convince the Telecom powers-that-be of it. Otherwise, I'd get a roomful of programmers loose on it. Oh well, that's the way the cookie crumbles... if you people at Rand or AT&T ever make hay of it, remember who gave you the idea. From j.guy@trl.OZ.AU Tue Jan 28 23:51:37 1992 Date: Tue, 28 Jan 92 09:51:37 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201272251.AA28330@medici.trl.OZ.AU> To: voynich@rand.org Subject: <4> , The Infamous Status: OR The Infamous, 'cause we all know this little monster to be almost always followed by , so very always (pardon my English!) that whenever we see it followed by anything else we suspect a transcription error, or a Voynich-writer's error. We all know that it's completely crazy. Languages just don't do that, so <40> has just *got* to be a single unit, like "th" or "ng" in English. I know it, too. Oh yeah? Suck back, Frogguy, you dill. You've known a real language that does that for donkey's years. And it's not an obscure one either. Fijian. "Y". "Y" in Fijian is always, always, always followed by "a", never ever ever anything else. No Fijian word may start with "a". Wherever some neighbouring language or dialect (this is official Fijian I'm talking about, which is the Bau dialect), wherever they have a word starting with "a", Fijian (if it has that word) has it with "ya". Example: Akei "aseku" (my name), Fijian "yacaqu" (mm, that's pronounced "yathanggu" c = th as in "this", q = ng as in "finger"). Why didn't I think of it? Stupid frog. From j.guy@trl.OZ.AU Wed Jan 29 00:44:21 1992 Date: Tue, 28 Jan 92 10:44:21 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201272344.AA28400@medici.trl.OZ.AU> To: voynich@rand.org Subject: Pronounceable Voynich and "dialects" Status: OR You will remember how astonished I was upon discovering how different the two languages looked when turned into pronounceable/readable Voynich. So different that they had to be two different languages, not just dialects. Yes, but, I found it hard to believe and started to look for straws to clutch at. First straw: given two close dialects of one languages, it's the little grammatical words, and the very high frequency words that are likely to be completely different. Take English and Lalland Scots again: English Scots if gin small wee go gang know ken Second straw: observing letter and digraph frequencies to be so different, and that my two samples were taken so far apart, I wondered. Perhaps the *topic* was so different that it showed in the frequency counts? So I scanned the file for a spot where A and B writings alternated, split that part into two files, and did my usual adjacency-matrix trick on each. Suddenly, things looked a lot less different. Still different of course, but not by the astounding amount there is between the first chunk of A and the first chunk of B, which are quite a long way apart from each other (*sigh* of relief). Armed with my two straws, I want to think we're still in the business of cracking two dialects of the same language. Incidentally, I have a sort of hunch that having two varieties of the same language to work on makes the decipherment task easier. I can't think why off hand, though. Just a hunch. Looking at those adjacency matrices computed on "readable" Voynich, I see that my reading rules cannot be right. Let's qualify that: are *probably* wrong. Readable "u" (Currier's , my ) is far, far too often preceded by readable "i" and "n" to be a separate letter: "iu" and "nu" have got to be single letters. Or have they? I just thought of my posting about <4o> and Fijian "ya" of a few minutes ago. And if Andras Kornai is right about Voynichian being a Pig-Latin style secret language, then... Do you speak Javanese? Not the Javanese spoken in Java, the one spoken in Paris. Just insert "av": bonjour --> bavonjavour For instance: Sava skavoz kavom sava, lave pavot, ljavanave! (that was phonetic) (C,a se cause comme c,a, les potes, le javanais!) Now, in Javanese, you don't always insert "av" everywhere, just in enough places so as not to be understood by outsiders. When I was a school kid, two friends of mine had developed their own secret language which was completely incomprehensible (they could speak it at break-neck speed, too): add "dgueu" (I'm using French spelling here to convey that is was a hard "g" and an "e" sound as in "pert"): Sadgue sedgue caudgue zaidgue codgue medgue sadgue (C,a se causait comme c,a) Since, I have read about a similar code in which the "real-word" vowel is repeated, so: Sadga sedgue caudgau zaidgai codgo medgue sadga What if Voynich Author A used one such code and Author B another, to write exactly the *same* language? Suddenly, seen through Andras's theory, all the weirdnesses of the Voynich look very normal indeed. Who knows, perhaps my pronounceable Voynich is closer to the mark than I care to believe? From j.guy@trl.OZ.AU Wed Jan 29 01:30:27 1992 Date: Tue, 28 Jan 92 11:30:27 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201280030.AA28476@medici.trl.OZ.AU> To: voynich@rand.org Subject: monkey business Status: OR If any of you feel like having a go at programming the algorithm I described, a piece of advice. I wrote: you build an index, then sort it alphabetically. That's in theory. In practice (you won't believe this) it is much, much faster to insert each new pointer in the right place as you read in the input text. You just do a block move. In Turbo Pascal (my favourite development language) it's move(,,) On my 33MHz 386 at home, indexing a 30,000 character corpus sorted to 100 places takes under 4 minutes that way. The same job, using the QuickSort algorithm usually takes more than 10 minutes! From j.guy@trl.OZ.AU Wed Jan 29 01:44:10 1992 Date: Tue, 28 Jan 92 11:44:10 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201280044.AA28490@medici.trl.OZ.AU> To: voynich@rand.org Subject: ZOE Status: OR wet!naga (*who* is wet!naga? Rather: how do I coax my mail tool into telling me?) writes: (Is ZOE the name of one of the nymphs?) Nope. The Voynich is written in Lojban. "zo'e" is a common grammatical word in Lojban which functions as an empty "place-marker" (John Cowan, YOU tell us all about Lojban and zo'e, please). (Couldn't resist, I've been writing serious stuff for far too long) Oh... gotcha! Peter Davidson! 'Twas in /home/jbm/Voynich/Members.List From j.guy@trl.OZ.AU Wed Jan 29 02:33:57 1992 Date: Tue, 28 Jan 92 12:33:57 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201280133.AA28685@medici.trl.OZ.AU> To: voynich@rand.org Subject: naga's assault Status: OR I was going to write this to Peter Davidson (or is it *really* Naga?) directly, but I realize it's probably of general interest. Look at this: String number of occurrences PC89/4OFC89/ 6 C89-4OFCC89/ 7 /4OFC89/8AR/ 3 PC89/4OFCC89 6 FC89/4OFC89/ 20 CO89/4OFC89/ 2 AM/OFCC89/4O 3 9FCC89/SC89/ 2 ZC89/4OFC89/ 24 /4OFC89/4OFC 24 Well, 4OFC89 and its brother 4OFCC89 look like likely words indeed. What else? I see that in this lot 4OFC89 is always preceded by 89. Don't YOU find that a bit suspicious? Now look at 4OFCC89. It's preceded by 89 too. And know what happens when you drop the initial 4 (Infamous 4)? It's preceded by AM. From what I see, that's not significant of course, and I just mention it like that (another of my hunches, likely to vanish in a puff of vorpal smoke). But I remember that AM is almost always preceded by 8. I smell an elternance: 894OFCC89 8AMOFCC89 and perhaps more generally: 894O 8AMO I also spy with my little eye: 9FCC89/SC89 Now what's before that 9FCC89? An 8 I bet you. Do we have the beginning of a verbal paradigm, or a noun declension there, by prefixing? viz: 894O-FCC89 I go 8AMO-FCC89 you go 89-FCC89 going Lots of languages behave like that: namlam "I come", nmlam "you come", mlam "he/she/it comes" naghlam "I will/would come", nghlam "you ditto", ghlam "he/she/it ditto" (Hog-Harbour Sakao, New Hebrides) Actually, Arabic and Hebrew do the same in the present tense, and Swahili all over the place. Andras, what does... oh, I remember. No, Hungarian is the other way around: kezet csokolo-m, where -m is the first person singular ("I"). So's Finnish, and so must be Estonian, then. Who does that in Europe??? Basque! Anyone out there knows some Georgian? (I don't). I must look at my Armenian grammar too, when I get home. Folks, I'm telling you again to ignore those damn spaces! And forget about the end-of-lines too. Just keep the end-of-paragraphs. From cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET Wed Jan 29 02:38:31 1992 Message-Id: From: cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET (John Cowan) Subject: Lojban, per frogguy's request To: voynich@rand.org (Voynich List) Date: Tue, 28 Jan 92 12:38:31 EST X-Mailer: ELM [version 2.3 PL11] Status: OR DISCLAIMER: I >don't< believe the VMS is written in Lojban. For one thing, there's nobody who can produce that much Lojban. For another thing, Lojban is written in Latin letters (or Elvish ones, at a pinch Cyrillic). Thirdly, Lojban has only been around since 1987, and its predecessor Loglan since 1955. So what is Lojban anyway? Hit 'n' now if you've seen it all before. Lojban (/LOZH-bahn/) is a constructed language. Originally called "Loglan" by project founder Dr. James Cooke Brown, who started the language development in 1955, the language goals were first described in the article "Loglan" in Scientific American, June, 1960. Loglan/Lojban has been built over three decades by dozens of workers and hundreds of supporters, led since 1987 by The Logical Language Group. There are many artificial languages, but Loglan/Lojban has been engi- neered to make it unique in several ways. The following are the main fea- tures of Lojban: o Lojban is designed to be used by people in communication with each other, and possibly in the future with computers. o Lojban is designed to be culturally neutral. o Lojban grammar is based on the principles of logic. o Lojban has an unambiguous grammar. o Lojban has phonetic spelling, and unambiguous resolution of sounds into words. o Lojban is simple compared to natural languages; it is easy to learn. o Lojban's 1300 root words can be easily combined to form a vocabulary of millions of words. o Lojban is regular; the rules of the language are without exception. o Lojban attempts to remove restrictions on creative and clear thought and communication. o Lojban has a variety of uses, ranging from the creative to the scien- tific, from the theoretical to the practical. Why was Lojban developed? - Lojban was originally designed for the pur- pose of supporting research on a concept known as the Sapir-Whorf hypothesis: "the structure of a language constrains the thinking of people using that language". Lojban allows the full expressive capability of a natural lan- guage, but differs in structure from other languages in major ways. This al- lows its use as a test vehicle for scientists studying the relationships be- tween language, thought, and culture. Other uses for Lojban - Lojban was designed as a human language, and not as a computer language. It is therefore intended for use in conversation, reading, writing, and thinking. However, since Lojban can be processed by a computer much more easily than can a natural language, Lojban-based computer applications are a natural expectation. Due to its unambiguous grammar and simple structure, it can be easily parsed by computers, making it possible for Lojban to be used in the future for computer-human interaction, and per- haps conversation. Lojban's predicate structure is similar to AI, suggesting it as a powerful tool in AI processing, especially in the storing and pro- cessing of data about the world and people's conceptions of it. Linguists are interested in Lojban's potential as an intermediate language in computer- aided translation of natural languages. Other people are interested in Loj- ban as an international language. The Lojban design - Lojban's character set uses only standard keyboard keys; capitalization is rare; punctuation is spoken as words. Written lan- guage corresponds exactly to the sounds of the spoken language; spelling is phonetic and unambiguous, and the flowing sounds of the language break down uniquely into words. These features make computer speech recognition and transcription more practical. Learning to write and spell Lojban is trivial. Lojban's predicate grammar was derived from that of formal logic. Loj- ban sentences are stated as sets of arguments tied together by predicates. These predicate structures can be used to express 'non-logical' thought; lo- gicians are able to analyze all manner of verbal expressions by converting them into predicate notation. But while Lojban will already be expressed in a predicate-based system, allowing easy logical analysis, it also contains the wide variety of elements found in natural language for expression of at- titudes, emotions, and rich metaphor. Lojban has none of the standard parts of speech. Lojban's 'predicate words' can serve as the equivalent of a noun, verb, adjective, or adverb. Action and existence are equally accessible to a speaker; the distinction can be ignored, or can be explicitly expressed. A variety of operators deal with abstractions such as events, states, properties, amounts, ideas, experience, and truth, and with at least four pre-defined varieties of causality. Lojban has no mandatory inflections and declensions on nouns, verbs, and adjectives. Natural languages evolved such variations to reduce ambiguity as to how words are related in a sentence. Evolutionary development made these inflections and declensions highly irregular and thus difficult to learn. The simple but flexible predicate relationship erases both the irregularity and the declensions. Tense and location markers (inflections), adverbs, and prepositions are combined into one part of speech. New preposition-like forms can be built at will from predicates; these allow the user to expand upon a sentence by at- taching and relating clauses not normally implied in the meaning of a word. Numbers and quantifiers are conceptually expanded from natural lan- guages. "Many", "enough", "too much", "a few", and "at least" are among con- cepts that are expressed as numbers in Lojban. Core concepts of logic, math- ematics, and science are built into the root vocabulary. On top of the core concepts of predicate logic, Lojban adds in non-logical constructs that do not affect or obscure the logical structure, allowing communications that are not amenable to logical analysis. For example, Lojban has a full set of emo- tional indicators, similar to such ejaculations in English as "Oh!", "Aha!", and "Wheee!", except that each has a specific meaning. Similarly, Lojban has indicators of the speaker's relationship to what is said, similar to those found in some American Indian languages. Lojban supports metalinguistic discussion about the sentences being spo- ken while remaining unambiguous. Lojban also supports a 'tense' logic that allows extreme specificity of time and space relationships, even those im- plied by time travel. Lojban's grammar is designed to support unambiguous statement of mathematical expressions and relations in a manner compatible both with international usage and Lojban's non-mathematical grammar. Lojban is much simpler than natural languages. Its grammar is compara- ble in complexity with the current generation of computer languages (such as ADA). Lojban's pronunciation, spelling, word formation, and grammar rules are fixed, and the language is free of exceptions to these rules. Current status and usage - The language, then called "Loglan", was first described in the 1950's, by Dr. James Cooke Brown. The 1960 Scientific American article "Loglan" was his call for assistance in developing the lan- guage. A revolution in linguistics was simultaneously taking place; the re- sulting increase in knowledge of the nature of language changed the require- ments for Loglan. The first widely distributed Loglan dictionary and lan- guage description did not appear until 1975; this description was incomplete, and continued development work discouraged learning the language. Computers caught up with Loglan just then, making it possible to refine the grammar, eliminate ambiguity and mathematically prove its absence; this work has just been completed. For over 35 years, this work has been performed by volun- teers, and without financial support. Now, after several versions of the language, people are learning and using the current version, which is the first called "Lojban" (from the roots "logical-language" in Lojban). This version is the first version with a stable vocabulary, and the first to have a stable and completely defined grammar (the grammar of mathe- matical expressions, for example, was not developed until 1990). The basic Lojban vocabulary was baselined (stabilized against change) in the last half of 1988, and the grammar similarly stabilized in late-1990, after completion of the first Lojban courses, actual usage of the language, and several itera- tions of careful analysis. Thus, while the language was first started 35 years ago, Loglan/Lojban is a very new language. To ensure Lojban remains stable while people learn it, the language def- inition is prescribed and closely controlled. When the number of speakers has grown significantly, and a Lojban literature has developed, Lojban will be treated like a natural language and allowed to grow and flourish without constraint, as do other natural languages. Original text and poetry has been written in Lojban, and some has been translated into the language. Lojban's powerful metaphor structure allows building new concepts into words easily, as needed. A Lojban speaker doesn't need a dictionary to use and understand millions of words that can poten- tially exist in the language. There are about 800 people on our mailing lists, including about 120 listed as actively trying to learn the language. One or two dozen of these have demonstrated communicative ability to use the language in conversation, translation, or original writing, although none are yet fluent. Several of these regularly get together in the Washington DC area for conversation, and an annual meeting/mini-convention is held there in late June. This progress is remarkable since there is no dictionary for the current language, and a textbook being written exists only in partial draft form; we've evolved meth- ods of teaching the language at a distance that overcome this handicap. Lojban's unambiguity - Lojban has an unambiguous grammar (proven by com- puter analysis of a formal grammar with YACC), pronunciation, and morphology (word forms). The person who reads or hears a Lojban sentence is never in doubt as to what words it contains or what roles they play in the sentence. Lojban has no words that sound alike but have different meanings (like "herd" and "heard"), that have multiple unrelated meanings ("set"), or that differ only in punctuation but not in sound (like the abominable "its" and "it's"). There is never any doubt about where words begin and end ("cargo shipment" can be heard as 2, 3, or 4 words). The function of each word is clear; there is nothing like the English "Time flies like an arrow.", in which any of the first three words could be the verb. Precision in no way confines the mean- ing of a Lojban sentence. It is possible to speak nonsense, to tell a lie, or to be misunderstood. You can be very specific, or you can be intention- ally vague. Your hearer may not understand what you meant, but will always understand what you said. Lojban is NOT entirely unambiguous; human beings occasionally desire to be ambiguous in their expressions. In Lojban, this ambiguity is limited to semantics, metaphor, and intentional omission of information (ellipsis). Se- mantic ambiguity in language results because words in natural languages rep- resent families of concepts rather than individual meanings, often with only weak semantic relationships to each other. In addition, each individual's personal experiences provide emotional connotations to words. By providing a fresh, culturally-neutral start, Lojban attempts to minimize the transference of these associations as people learn the language. By intention, most Loj- ban words do not closely resemble corresponding words in other languages; the differences aid in making this fresh start possible. Lojban's powerful metaphor and word-building features make it easy to make fine distinctions between concepts, discouraging individual words from having families of mean- ings. Lojban metaphors are themselves ambiguous, specifying a relationship between concepts, but not what the relationship is. That relationship can be made explicit using unambiguous logical constructs if necessary, or can be left vague as the speaker (usually) desires. Similarly, portions of the log- ical structure of a Lojban expression can be omitted, greatly simplifying the expression while causing some ambiguity. Unlike in the natural languages, though, this ambiguity is readily identified by a reader or listener. Thus all ambiguity in Lojban is constrained and recognizable, and can be clarified as necessary by further interaction. Lojban and Sapir-Whorf - This hypothesis states that the structure of a language constrains thought in that language, and constrains and influences the culture that uses it. In other words, if concepts or structural patterns are difficult to express in a language, the society and culture using the language will tend to avoid them. Individuals might overcome this barrier, but the society as a whole will not. The Sapir-Whorf hypothesis is impor- tant, and controversial; it can be used as a sociological argument to justify or to oppose racism and sexism (and a variety of other 'isms'). For example, the assertion that since genderless expressions in English use 'masculine' forms, English is 'sexist', presumes the Sapir-Whorf hypothesis is true. Un- derstanding the potential for Sapir-Whorf effects could lead to better inter- cultural understanding, promoting communication and peace. It is known that people's ideas and thought change somewhat when they learn a foreign language. It is not known whether this change is due to ex- posure to a different culture or even just getting outside of ones own cul- ture. It is also not known how much (if any) of the change is due to the na- ture of the language, as opposed to the cultural associations. The Sapir-Whorf hypothesis was important in linguistics in the 1950's, but interest fell off partially because properly testing it was so difficult. Loglan/Lojban is a new approach to such testing. Obviously, if a culture-in- dependent language could be taught to groups of people, the effects of lan- guage could more easily be separated from those of culture. Unique features of Lojban remove constraints on language in the areas of logic, ambiguity, and expressive power, opening up areas of thought that have not been easily accessible by human language before. Meanwhile, the formal rigidity of the language definition allows speakers to carefully control their expressions (and perhaps therefore their thought processes). This gives some measure of predictive power that can be used in designing and preparing for actual Sapir-Whorf experiments. One of the prerequisites of a Sapir-Whorf experiment is an international body of Lojban speakers. We need to be able to teach Lojban to subjects who know only their native (non-English) tongue, and we need to know in advance the difficulties that people from each language and culture will have in learning Lojban. Thus, the Lojban community is actively reaching out to speakers of languages other than English. Lojban does not need to prove or disprove the Sapir-Whorf hypothesis in order to be successful. However, if evidence is produced supporting the Sapir-Whorf effect, Lojban will likely be perceived as an outstanding tool of analytical and creative thought. Other linguistic applications - An artificial language is a simple model of a natural language. It is used for communication like natural languages, simpler, more regular, and relatively uncontaminated by culture effects. Un- like most natural languages, an AL has not evolved through naturally pro- cesses of internal change or contact with other languages. In addition, to the extent to which details of an AL is pre-defined, the internal structure of the language is far better known than that of any natural language. Such a pre-definition, a language 'prescription', makes an AL a unique tool for studying the nature of language. As people learn the language, the way they 'acquire' understanding of that prescription can tell scientists how 'natural' the prescribed forms are. Actual usage of the language can be com- pared to the prescription providing quantitative data on specific patterns of usage. As the language evolves from its relatively pristine initial state, it may deviate from its prescription. Such deviations will better inform re- searchers as to the properties of a 'natural' language. The process of lan- guage change itself will be open to investigation in a way never before pos- sible. Finally, the existence of a relatively complete language prescription at the birth of the language means that a 'description' of actual usage after that initial state can be more simply created, maintained, and studied. Lojban is undoubtedly the most carefully designed and defined AL ever created. All aspects of its design have been carefully engineered by several people encompassing expertise in a variety of disciplines, including linguis- tics. The language prescription is similarly the most complete of any lan- guage. As such, it serves as a unique basis for the study of language usage and language change. A new language like Lojban, with no native speakers, is a 'pidgin'. As the language evolves, native speakers of other languages will learn it, and will bring into their Lojban usage the perspective and patterns of their na- tive language. This interaction process, called 'creolization', affects all languages, and may be the principle cause of language change. As Lojban is learned by speakers of a variety of natural languages, this process can be studied directly in a way never before possible, with the language prescrip- tion serving as a standard by which deviations associated with speaker ori- gin, and evolution of usage, can be measured and described in detail. An AL like Lojban has neither an associated 'native accent' nor a cul- ture base. Being much simpler and more regular than a natural language, an adult speaker should acquire a fluency seldom achieved when studying a natu- ral language, and in a relatively short time. Lojban thus makes possible relatively short-term studies of language learning and language change. Pro- cesses that take generations in natural languages may be observed in a few years in Lojban speakers. With the learning of other languages acquiring critical importance in today's international economy, Lojban provides a tool for research in lan- guage acquisition. Again, Lojban's simplicity allows the results of such re- search to be obtained more quickly than in similar studies of natural lan- guage acquisition. No claim is made that studying Lojban will tell 'all' about language. Lojban, at least for several decades, will only be a 'model' of a language, whose 'naturalness' will be suspect. However, to the extent that Lojban serves the communicative and expressive functions of human language, any the- ory about the nature of human language must apply to Lojban. Language fea- tures and processes more easily identified in studying Lojban can then be confirmed in natural languages. Similarly, theories of the nature of lan- guage can be tested and refined against this simpler model of a language be- fore facing the more difficult and time-consuming testing and analysis in- volved in natural language research. Because Lojban is relatively culture-free, and because of its prescribed structure that is consistent with predicate logic, Lojban is an ideal medium for the analysis and description of other languages. Currently, features in language must be compared against other natural languages, and are usually described in scientific literature by glossed translation into English. Loj- ban is simple and regular enough to be acquired as a metalanguage for de- scribing other languages; its structures allow clearer reflection of the pat- terns of the language being described, without interference from the compet- ing patterns of English. Finally, Lojban's predicate grammar, makes it eminently suited for ongo- ing computer research into natural languages. Lojban can be used for parsing and analysis, as an internal medium of data storage, or as an intermediate language for machine translation. Having a combination of logical and natu- ral language structures, Lojban combines the best of both major structural approaches to language processing in computers. The exercise of trying to invent a language can teach us things about language that probably can't be learned in any other way. Even if Lojban should fail as a language, we will learn. However, to the extent Lojban suc- ceeds, its potential as a basis for testing ideas about language, its struc- ture, and usage, is unlimited. The invention of Lojban is the invention of the science of experimental linguistics. Lojban as an international language - Lojban may be the first artificial language NOT in direct competition with Esperanto, in that Lojban's potential success is not dependent on its immediate practical use as an international language. Indeed, we hope to use Esperanto as a means of rapidly spreading information about Lojban to non-English speakers, speeding a process that would take decades using direct translation to all target languages. Lojban has proven attractive to Esperantists interested in acquiring a new perspec- tive on their own international language; these feel less threatened because Lojban has different goals. Lojban's supporters recognize that it will take decades for Lojban to acquire both the number and variety of speakers and the extensive history of usage that marks Esperanto culture. Meanwhile, each language community has much to learn from each other; this process has started and is most active. Several reasons for learning Lojban NOW - Those working with the lan- guage now are actively consulted for their opinions on the language design, and on how to teach and spread the language. People have joined the project leadership within a few months of becoming involved. Those with a computer background can lead development of the first com- puter applications for the language. Expertise in the language will no doubt be valuable as Lojban becomes recognized as a useful tool for computer appli- cations by the computer industry. Computer-oriented Lojbanists can also aid in developing computer-aided instruction tools or converting existing soft- ware to run on new computers. Regardless of background, learning Lojban is a mind-expanding experi- ence. Learning any language other than your native tongue broadens your per- spectives and allows you to transcend the limited viewpoints of your native language culture. Lojban, being much simpler to learn than natural lan- guages, provides this benefit much more quickly than does the study of other languages. Being so strongly different, the intensity of the effect seems to be heightened. When Lojbanists talk about language, the discussion has been observed to be more sophisticated as well as qualitatively different from the talk of students of second natural languages. The logical organization embedded in Lojban aids in organizing and clar- ifying thoughts. Having done so, your new perspective on language, ambigu- ity, and communication will allow you to express those thoughts more clearly, even when you use an ambiguous natural language. A relatively short study of Lojban by high school (or younger) students has been proposed, providing the linguistic understanding that was once asso- ciated with studying Latin and other languages for much longer periods of time. Such study can be tied in with concepts of logic, and possibly with computer-related activities, helping to show the essential interrelated na- ture of language and other human endeavor. You needn't learn Lojban for any practical purpose, however. Many of those learning Lojban are doing so because it is fun. Learning Lojban is in- tellectually stimulating, and provides human interaction and mental chal- lenge. Lojban has all the benefits of games designed for entertainment, with the added prospect of developing useful skills as a side benefit. Learning Lojban as an 'intellectual toy' means that you can get enjoyment from learn- ing Lojban without nearly the effort needed to benefit from studying other languages. While becoming fluent in Lojban will probably take hundreds of hours over several months, you can feel some sense of accomplishment in the language after just a few hours of study. You can use Lojban immediately for fun, while gaining skill with greater experience. How to learn Lojban - write to The Logical Language Group, Inc. at the address found at the end of this brochure, and we will be happy to provide them. We request a contribution of $4-5 to cover the costs of these materi- als, but will send this package on speculation. Additional materials avail- able total hundreds of pages, which are priced approximately at our costs; we are a non-profit educational/scientific organization. Provide a postal ad- dress; we have no materials prepared and approved for on-line distribution, but expect to have some within the next couple of months. Contact us regard- ing international payment; we offer several options. A limited policy exists for providing materials to people who cannot afford payment. We encourage new people to concentrate on vocabulary. Almost any use of Lojban requires some mastery of the basic vocabulary (1000-2000 words). You can learn enough Lojban grammar to support conversation in just a couple of hours, but face many hours of vocabulary work in order to effectively use that grammar. We distribute flash cards and have developed flash card tech- niques that are extremely efficient in learning vocabulary. These techniques have been automated into computer-aided-teaching programs sold under the name "LogFlash", with MS-DOS and MacIntosh versions currently available. You can learn the Lojban grammar in several ways, including study of ex- amples in the quarterly journal "ju'i lobypli" or by inspection and analysis of YACC and E-BNF formal grammar descriptions. Draft lesson materials (about 320 pages) are the best materials available for self-study. A Lojban text- book is being written to reflect what is learned from the first learning ef- forts, and is being designed so that it can be used for self-instruction, classroom teaching, or group study. A formal 'Lojban dictionary' will replace most of the existing word lists and language description materials, probably about a year after the textbook is done, but present materials con- tain the equivalent of a basic dictionary. Approximately 6 months after the dictionary is published the Lojban lan- guage definition will be baselined (frozen) for 5 years, and Lojban will thereafter be solely controlled by speakers of the language. Write or call Bob LeChevalier The Logical Language Group, Inc. 2904 Beau Lane Fairfax, VA 22031 (703) 385-0273 Net mail can be sent to us at: lojbab@grebyn.com An active newsgroup/mail exchange is available on this network. To join, send the message "subscribe lojban " to: listserv@cuvmb.cc.columbia.edu or contact: lojban-list-request@snark.thyrsus.com and send mailings to all newsgroup members via: lojban@cuvmb.cc.columbia.edu Compuserve members can join this group, or can contact us, by preceding any of the above addresses with "INTERNET:". Fidonet connects with the Internet via a variety of nodes - contact your SYSOP. -- cowan@snark.thyrsus.com ...!uunet!cbmvax!snark!cowan e'osai ko sarji la lojban From j.guy@trl.OZ.AU Wed Jan 29 03:26:10 1992 Date: Tue, 28 Jan 92 13:26:10 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201280226.AA28728@medici.trl.OZ.AU> To: voynich@rand.org Subject: 89 Status: OR Me again. I clicked on the wrong part of the screen, and off went the message I was writing down the mail chute. These multitasking windows are nice, I can babble on Voynich while the stuff I get paid for goes on in the background, or, like now, on my PC, but sometimes... I was saying that Naga's string counts could be shreds of a paradigm: 8AMO 894O 89 Remember the fairly safe assumption that = <9> = "a" and = "o". We have something there then, like (consonants a pure fabrication): kamo- kayo- ka- Looks credible, doesn't it? For instance, Arabic has aktub, taktub, yaktub "I, you, he writes" Perhaps <8A>/<89> is third person, and and <4O> are tenses or modes. Well, there are a lot of languages where you have the same "word" act both as prefix and suffix, but with different functions. For instance, <8A> could be subject when prefix, and object when suffix. 894OFO89 could be something like "he sees it". In many languages again, if the object is a definite noun, you must use the "him/her/it" prefix or suffix with the verb. In Swahili, for instance, you say "I-read book" for "I read *a* book", but "I-read-it book" for "I read *the* book". Such a phenomenon would account for the very high frequency of 89 and 8A in Voynich B. As for its low frequency in Voynich A, we all know that A and B are different "languages", probably two dialects of the same language. As I wrote in an earlier posting that, in my experience of dialects, the first things to differ are the grammar and the grammatical particles. In Sakao for instance: Hog-Harbour dialect Port-Olry dialect namlam amlam I come nghlam lam you come mlam malam he/she/it comes Oh, I'm sorry about ZOE, by the way, but if you're after the nymph's name you only need to look next to ZOE. There she is: PAM. And, please, please, ignore those spaces! From j.guy@trl.OZ.AU Wed Jan 29 04:55:53 1992 Date: Tue, 28 Jan 92 14:55:53 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201280355.AA28821@medici.trl.OZ.AU> To: voynich@rand.org Subject: another algorithm Status: OR I won't have the leisure to write the necessary program for a while, so I'll just describe it. Consider an adjacency matrix like the ones I have posted. Suppose that a given letter, say 8, occurs in the same environment as, say, 9 (not true, but I said "suppose", didn't I?). How is that going to show in the matrix? If 8 is x times as common as 9, for any letter * we will have fq[8,*] = x fq[9,*] (approximately of course). What does that mean? A linear regression on the paired frequencies of 8 and 9 with the letters in whose environment they occur ought to show a strong positive correlation. Thus, from the frequencies in rows 8 and 9 we get a correlation coefficient r[8,9]. OK then, compute the matrix of linear correlation coefficients. It will show you straight away letters with similar distributional properties. What next? Do a cluster analysis. Bet you you'll get all the vowels in one cluster, all the consonants in another, and all the semi-consonants, if any, in a third cluster. Within each cluster, of course, you'll have smaller clusters. No idea how *those* will behave and what they will mean. I don't have the time to do it now. Well, writing the code to compute the correlation coefficients is nothing. It's the cluster analysis. The minimal-spanning tree method is to be avoided. You need an n-way splitting algorithm. I wrote one years ago that worked very nicely, but I'm stuffed if I know where it is. I know that this algorithm will work because I invented and tested it in 1981, but on words, not on letters. I can talk about it now, because it's just been rediscovered by two people in the UK and they've published it. If they hadn't, well... the most I could have done is apply it on the Voynich and tell you the results, but not how I got them. A note to John Baez: linear correlation coefficients are mathematically incorrect, strictly speaking. Still, they work pretty well. I've figured out a mathematically correct measurement, but as I haven't seen it anywhere yet... yes, here we go again: I can't tell you. From firth@SEI.CMU.EDU Wed Jan 29 04:58:43 1992 Date: Tue, 28 Jan 92 14:58:43 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201281958.AA06562@bp.sei.cmu.edu> To: voynich@rand.org Subject: Various comments Status: OR Notes on the Voynich Manuscript - Part 10 ----------------------------------------- Comments and Suggestions These comments are not very well ordered, but with the start of the new term I'm barely able to keep up with reading the Voynich stuff, much less writing any myself. So here are a few observations I don't want to forget about. 1. Letter correlations. Currier, and now we, have found correlations between the final "letter" of a "word", if that is they are letters and words, and the initial letter of the next word. Granted. However, I have some problems with the interpretation. Currier claims he knows of no language with this feature; I think he's very wrong. First, note that in many languages (welsh for instance) the phoneme (sound) at the start of a word is modified by the previous word. Some systems of writing reflect this change (modern welsh I believe does), and some do not. Secondly, note that in some languages there are grammatical rules that lead to such correlations. In english the chain of causation runs from right to left (a possibility Currier overlooks): "a" changes to "an" before a vowel, and possessives in "-y" change to "-ine". Likewise, both french and italian elide heavily, and some writing systems reflect this. Finally, I struggled through enough Dante at one time to know that in italian poetry of the time, endings and beginnings were highly correlated, not because of orthography or grammar, but because of euphony. So, yes, the statistical patterns exist, and they are real, but I have two problems. (a) are they unusual? - would we not find the same with known european languages, and (b) can we reason back from the effect to the cause, given that there might be many or multiple causes, at very different levels of language. I am similarly skeptical of preferred initial letters. After all, in latin "qu" is never final, and "x" is never initial, and I'm sure parallels can be found in many languages. So this isn't unusual. Preferred paragraph-initial letters, frankly, impress me even less. In Euclid's "Elements", for example, paragraphs usually begin with "Axiom", "Theorem", "Corollary", "Lemma", ... in other words a very small set. The peculiar letter pattern is a consequence of the peculiar word pattern, and that is a consequence of the style of the author. Nothing can be deduced about the language from this effect. 2. Need for Comparison This brings me to the second point: the need for comparisons with known texts in known languages. Every experiment done on the Voynich MS should be repeated, on, say, Magna Carta, the Romaunt of the Rose (separating the dialects), the Inferno, the Chanson de Roland - well, you get the idea - a chunk of contemporary (we hope) european documents. To give one example, we can test explicitly Mr Guy's views about how dialects differ by applying his reasoning to the Romaunt of the Rose (Part 1 in a Southern dialect, Part 2 in a Northern, the original spelling preserved), and seeing whether they work. Yes, this is going to be laborious, but I think it's far better to have a statistic that means something, than to have one about whose meaning we must then speculate. 3. Voynich Lines I do not believe the MS is poetry, and I do not believe the line is a functional unit. This is because all I have seen looks like prose, and the lines seem to have been continued to a margin. Especially on the pages with a big drawing, the scribe has filled however much space the drawing left, line after line. Finally, the paragraphs end with short lines, just as do paragraphs of modern prose, and they can't be stanzas with a short final line (as in Keats' "Ode to Psyche", for instance), because the paragraphs are of arbitrary length and show no pattern of lines I can discern. However, I agree that a useful hypothesis for the statistical properties is one that assumes elision or syllable fusion of some kind. 4. Letter Construction Currier's observations on how the letters were written I think can be explained fairly simply: the scribes were writing with quill pens. It is always easier to make a downward stroke than an upward with such a pen, which is why "o" is written as two half-moons, and why most rising serifs in Voynich script are deeply curved. This raises the possibility that the script was designed to be so written, in which case we should be prepared for many contracted forms. As an illustration, the egyptian hieratic script was designed to be written with a reed pen; it consists almost entirely of contractions, each "letter" the fusion of two hieroglyphic symbols, and sometimes those symbols just the first syllable and last consonant of the word in question. It's fiendishly hard to read; I never bothered to take the time, and anyway all modern books use standard hieroglyphic fonts. 5. Words, words, words As I see it, the single biggest problem we have today is deciding whether the spaces do or do not divide words. If they do, then many of Mr Guy's proposed symbol equivalences or alternations don't seem to work; if they don't, then most of my work on prefixes and suffixes is wrong. Let me say, I'm not convinced by either Mr Guy's reasoning or mine on this point. What I did was pose the hypothesis that the words were words, look for confirming evidence, find it abundantly, and claim the answer. That's desperately unscientific, and I should have known better. The correct approach is to frame all reasonable alternative hypotheses, and look not for confirming but for discriminating evidence: evidence the favours some hypotheses over others, and ideally one over all. Here is my shot at the hypotheses: S1W1: The spaces divide words S2: The spaces do not divide words, but are inserted according to some other rule ("always after 9") S3: The spaces are inserted at random, just to confuse W2: Word divisions are indicated in the text by some other means W3: Word divisions are not indicated in the text We can test the 'S' hypotheses by concentrating on spaces, and the 'W' hypotheses by ignoring spaces. I think we already have enough evidence to refute S3. On this issue, I believe our best source of additional evidence is the isolated "words" attached to plants, jars, nymphs &c. If most of them recur in the text as words, and if they show initial and final letter frequencies similar to those of the "words" in the text, then that is strong evidence against S2. Our next best source would be paragraph initial word beginnings; I don't think we should use paragraph endings since it is possible they are padded with gibberish. 6. Gleams of Hope (1) I've now read the Voynich Newsletters of 1978 February and November, and 1980 January. They make me still more certain that the Brumbaugh decipherment is wrong. His transcription errors in particular worry me - there are too many of them; they are too big; and by now the decode matrix must be so embedded in his unconscious that it could be making those errors precisely to construct readable latin. But that's not my point. Look at the 1978 November letter, and please compare page 7 (Brumbaugh transcription) with p 6 (Voynich original). How long has Prof Brumbaugh been working on this? How familiar must he be with the script, and how many thousands of lines must he have transcribed? And yet, even making allowance for a hasty hand and a poor pen, that script is still very unpolished. Now compare the beautiful, elegant, stylish and almost effortless penmanship of the original. This is not a scientific argument, but an aesthetic one, yet to me it leaps off the page: this penmanship is the craft of long and loving practice. And that says two things: first, that the Voynich MS was written to be read, not decoded, and was indeed to its readers the equivalent of plain text; and secondly, that this work is not unique, it is the only survivor of an entire corpus: for how else could that long practice have been acquired? 7. Gleams of Hope (2) And one gleam is Mr Guy, whose long, exciting and valiant notes have I think elucidated, if not yet the mystery itself, at least a great deal more of its attributes. But the whole team, I think, brings to bear on this problem two new things. First, we have collectively a very diverse body of knowledge, far more that the earlier teams, and we are less likely to imprison ourselves in fixed ideas. But secondly, most of us can do what they couldn't - we can use the machines as fast and effective tools for the testing of hypotheses, exactly as Ms D'Imperio advised. They took weeks to create one computer program: specification, design, flowcharting, data preparation, testing, begging machine time - I am not surprised they achieved so little with the computers. Today, on the other hand, I can have an idea, throw together some recycled code to test it, entrust the result to my old and familiar compiler, and have the numbers in twenty minutes. Then all I have to worry about is the insight - but our predecessors seemed never to have the time for that. Robert From reeds@gauss.att.com Wed Jan 29 06:27:20 1992 Message-Id: <9201282127.AA16605@rand.org> From: reeds@gauss.att.com Date: Tue, 28 Jan 92 16:27:20 EST To: voynich@rand.org Status: OR A comment suggested by reading Robert Firth's most recent. When discussing the problem of linguistic reality of word spaces, let's not loose track of the corresponding problem of their graphic reality. When transcribing it's sometimes really hard to tell if there is a space there or not. In fact, I'd say that word space transcription is THE hardest part of transcription. Hence, whatever data we have is bound to be flakey. We should always assume that whatever cryptographic or linguistic mechanisms generate the word spaces, there are also random run-ons and word splits caused by the transcription process. Let me take the oportunity to mention the existence of a new edition of my 'checklist', and of a 'finder' of pages in the Yale microfilm, both available by anonymous ftp from rand.org in directory pub/jim. Jim Reeds From j.guy@trl.OZ.AU Wed Jan 29 06:34:34 1992 Date: Tue, 28 Jan 92 16:34:34 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201280534.AA29019@medici.trl.OZ.AU> To: voynich@rand.org Subject: clustering Status: OR Jim Reeds' "Mostly for John Baez" prompts me to add: And don't forget to carry out the same clustering on coefficients obtained from M(1)+M(-1) M(1)+M(2)+M(-1)+M(-2) and so on... By M(n) I mean the frequency matrix of letters with the letters n positions to the right (so M(-1) is one position to the left, or, if you prefer, the transpose of M(1). "Transpose"? is that the right term? Oh, you know what I mean!) Believe me, there's something behind that apparent madness: it works! From firth@SEI.CMU.EDU Wed Jan 29 07:03:41 1992 Date: Tue, 28 Jan 92 17:03:41 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201282203.AA06807@bp.sei.cmu.edu> To: voynich@rand.org Subject: Voynich vowels Status: OR Notes on the Voynich Manuscript - Part 11 ----------------------------------------- The Voynich Vowels A while ago, I started looking at the Voynich vowel system, again on the assumptions that this stuff is real language encoded in a real alphabet. This got pushed aside over the holiday, but Mr Guy's recent messages induced me to look over the old notes. I hope my stumblings and musings will be of help to you. I decided to attack the problem on two fronts simultaneously (a) analyze the roots and inflections to isolate what seemed to be vowels (b) on the assumption the script was constructed, analyze the ligatured forms looking for vowels and diphthongs Later, I saw the table of ligatures in the D'Imperio document, and was much encouraged. I'd noticed that Currier's S looked like cc run together, and found ca, but hadn't noticed oo, co, or c9 ligatured. Well, my analysis, principally from the "root and stem" forms I give in a previous note, suggested that (Guy transcription) c, a and o were vowels. It also suggested that 9 was a vowel except when final, where it seems to serve some other function. Of course, if I'm wrong about the words and Mr Guy is right, then my evidence shows 9 to be a vowel always. Next, I asked myself, Why would the devisor of this script care about cc, ca and so on? And why are these vowels sometimes ligatured and sometimes not? At that moment, a note of familiarity crept in, and the thought occurred, "suppose cc is the same vowel as c, but lengthened?" >From which it was but a short step to this hypothesis: a alpha c epsilon cc eta o omicron oo omega And, you know, that would explain patterns like ooo: that occurs in the word Zo:on, not an uncommon word! At that point, I intended to count all the occurrences of mono, di, tri &c phthongs in the text, and see whether they did match the pattern of greek. Somehow, there was never enough time. And I never found a candidate iota or upsilon. But if the usual (not invariable) prtactice was to ligature the long vowels, and the true diphthongs, but not to ligature adjaccent vowels that belonged to different syllable, then perhaps the Voynich style would result. Thus, in theou, the epsilon would be isolated and the ou ligatured. After the holiday, a brief search for obvious words with one or fewer consonants - he:, ho, tou, to: (or to:i) - and of course the zo:e: and theou mentioned above - seemed to be going nowhere. And I couldn't believe nobody had tried greek before. So it sort of drifted off into the limbo of ideas. But Mr Guy claims, by entirely different reasoning, that the Voynich c and o are "e" and "o"! Maybe there's something here worth following up on. A last point: it is possible that this script was invented to encode a greek dialect, but that the underlying language is not greek. Many languages have short and long vowels; maybe somebody thought this script would work for italian or occitan. Robert From Karl.Kluge@G.GP.CS.CMU.EDU Wed Jan 29 01:05:00 1992 Date: Wed, 29 Jan 1992 01:05-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Jim R. and Brumbaugh redux Message-Id: <696665150/kck@G.GP.CS.CMU.EDU> Status: OR (As an aside, is there some reason you don't put headers in your mail, Jim? -- it makes finding particular messages tough.) Jim 'sez: > My preliminary assessment of Brumbaugh's transcriptions: not very precise. > I make lots of transcription errors, and know it, and have learned to double > check all over the place. Brumbaugh makes more errors than I do. He skips > words. His actual skip-free transcriptions are not bad, but contain a few > quirks (such as insertion of C after X). I think that may be harsh (other than the missed pair of words). I didn't have time to work through the zodiac folio examples in detail (for constructing that letter->digit matrix, having the strings aligned vertically is much more convenient than in columns), although as a preliminary observation let me say that some serious J/7 confusion was occuring somewhere. I did look at the f106v example in detail, though. Here are the cases where there was a discrepancy between your transcription of Brumbaugh's transliteration and his matrix: > Brumbaugh's box is: > > digit Voynich Latin > > 1 O,v AJV > 2 RJ BKR > 3 CFB CLW > 4 4E) DMS (X) > 5 A ENX > 6 P FOT > 7 S2Z GPY > 8 8w HQU > 9 9V I, -US, Z > > Column 1 has my reading, in Currier letters. > > Column 2 is Brumbaugh's transcription into Voynich letters, > given here as much as possible in Currier. He has > a scribly w or m like letter, like Currier's M or 3, > but without the final upward swash, which I denote `w' > here, and he has a letter like a backwards c, which > I denote ')'. It is pretty clear that our AM and AN > both get turned into w). It is not clear whether > Brumbaugh takes the ')' as the same as Currier's D > or just as its final upward curved stroke. > > Column 3 is his further transcription into digits. > > Column 4 has my comments (now my comments -- KCK) > > OBAUAE OBwRCE 138254 (C=5 error -- but note Jim has A=5) > FCORAN FCO2w) 331284 (2=2 error -- but note Jim has R=5) > S8AJ S8AP 7852 (P=2 error -- but note Jim has J=2) > 9SPAN 9SPA) 97684 (A=8 error -- see below) > 9ZC8AM 9ZC8A) 973884 (A=8 error -- see below) > ZX9 ZXC9 77339 (group as ZX9 = 7(733)9 -- see below) > SQ9 SQC9 77639 (group as SQ9 = 7(763)9 -- see below) The first three errors seem to be problems transcribing B's transliteration of the text, as Jim's transcription of Petersen fits B's matrix. The next two errors (A=8) seem genuine -- is there something different about the appearance of those two A's? I notice that in the zodiac folios the same A=8 reading occurs at one point: > P.I.9 9314*8*6 II.10 <9FOEAM> The other three errors (9=3, C=6) have to do with how Brumbaugh is splitting up compound characters. The rule seems to be: Currier S is read as 7 when by itself. As part of a compund character such as Q it is broken up into j'?c, where ? is the central character (P in the case of Q, F in the case of X, etc.), c is C, and j' is the left half of the S part (D'Imperio's 9). So ZX9 breaks up into Z j' F c 9 = 7 7 3 3 9, which matches B's matrix. Without seeing the raw folio I can't resolve the A=8 problem (could these A's be 8's drawn like &'s with very small upper loops?). Otherwise the entire example works out just fine. Karl From Karl.Kluge@G.GP.CS.CMU.EDU Wed Jan 29 01:34:00 1992 Date: Wed, 29 Jan 1992 01:34-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Some annotated references Message-Id: <696666840/kck@G.GP.CS.CMU.EDU> Status: OR I decided to look up some of the references given in the 35 Mss. section. Here is a list of the ones I found with brief comments: Robert Steele, NATURE, 10/15/28 -- unillus. review of Newbold J. M. Manly, HARPER'S, July 1921 -- description of Newbold's theory, illustrated (f49r, Taurus folio, f81r, "Great Nebula", "key"). L. C. Strong, SCIENCE, 6/15/45 -- very brief description of his claimed decipherment, claims Mss. describes use of anti- biotics. W. F. Friedman, PHILOLOGICAL QUARTERLY, Jan. 1959 -- Chaucer and anagrams Lynn Thorndike, AMER. HIST. REV., Jan. 1929 -- unillus. review of Newbold. Louis Cons, SAT. REV. LIT., 10/27/28 -- ditto. J. M. Bird, SCIENTIFIC AMERICAN MONTHLY, June 1921 (see also 5/28/21 SCIENTIFIC AMERICAN) -- review of Newbold, takes skeptical view of decipherment. Illustrated with several folios. From em21+@andrew.cmu.edu Wed Jan 29 13:54:20 1992 Message-Id: Date: Tue, 28 Jan 1992 23:54:20 -0500 (EST) From: Eric Edward Moore To: voynich@rand.org Subject: Re: Robert Firth's comments In-Reply-To: <9201290134.AA00187@medici.trl.OZ.AU> References: <9201290134.AA00187@medici.trl.OZ.AU> Status: OR About the electronic texts: English language texts are available through Project Gutenberg, I believe they have annonymous FTP set up on mrcnext.cso.uiuc.edu in the etext directory try weekends or nights, also they might have sites at think.com (in /public) or quake.think.com (/pub/etext). More info (a long readme) available on request.... Oxford University also has an electronic text library with texts in many languages available for "scholarly" use. All I have is a JANET address: ARCHIVE@UK.AC.OX.VAX That's my two cents... -Love, Kisses, and a Neutron Bomb -Eric the Finn From Karl.Kluge@G.GP.CS.CMU.EDU Wed Jan 29 11:48:00 1992 Date: Wed, 29 Jan 1992 11:48-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Suggestion for theme of quadruple folding folio Message-Id: <696703715/kck@G.GP.CS.CMU.EDU> Status: OR For those who haven't seen this, it consists of 9 circles arranged so: D ------ B ------ A Where A...H and X are the |\ | /| circles, and the lines | \ | / | are streams of something | | | flowing to/between the F ------ X ------ C circles. | | | | / | \ | |/ | \| H ------ G ------ E My guess is that this diagram shows the way in which the four elements (Earth, Air, Fire, and Water) emanate from the Prime Matter via combination of the four qualities (Hot, Dry, Wet, and Cold). Circle A contains what is unquestionably a small sketch of a castle or fortified town. Suppose that represents Earth. Then there are two possible assignments to the other circles: A. Earth A. Earth B. Dry B. Cold C. Cold C. Dry D. Fire D. Water E. Water E. Fire F. Hot F. Wet G. Wet G. Hot H. Air H. Air (Assuming I've copied my notes correctly). H looks vaguely like clouds, which would match Air. E doesn't seem to have any real iconographic resemblance to Water or Fire that I can see, but with the Voynich it's hard to tell. In any case, it's much more plausible than embroyology as per Newbold. From wet!naga@cca.ucsf.EDU Wed Jan 29 22:40:00 1992 Message-Id: Date: Wed, 29 Jan 92 05:40 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Time! Gimme time! Status: OR > From: ucsfcca!trl.OZ.AU!j.guy (Jacques Guy) > Did you read the last third or so of my long rambling that opens > with the Sermon on the Mount in medieval Latin? What do you think > of the algorithm I describe in there? Had I the leisure I'd certainly be pleased to implement it. Unfortunately I'm being driven slowly up the wall attempting to prepare for an impending change (not for the better) of a somewhat radical nature. It's a wonder I can still type coherently. > Oh... gotcha! Peter Davidson! 'Twas in /home/jbm/Voynich/Members.List Can't always believe everything you read, though. From wet!naga@cca.ucsf.EDU Wed Jan 29 22:44:00 1992 Message-Id: Date: Wed, 29 Jan 92 05:44 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: 4-letter repetitions (ZOE, SOE and AM) Status: OR I haven't had time to make any further refinements in my repetition-search program, due to other commitments, but since there is an old XT sitting around here mostly doing nothing I decided to run the program on that for a few hours. It's not fast enough to recover longish strings in less than a day and a night, but it did identify all 1,290 strings of length 4 which occur at least twice in VOYNICH.A (which contains 33,702 bytes). It took 5.5 hours to do this. Here are the 100 most frequent (two sets of columns): Occurr- String Position of ences first occurrence 380 |8AM/| 159 | 75 |OE/Z| 529 354 |SOE/| 266 | 74 |OFS9| 1857 320 |/8AM| 859 | 74 |FAM/| 786 310 |SOR/| 497 | 73 |POE/| 2146 242 |OE/S| 938 | 73 |8AN/| 391 229 |OR/S| 1338 | 71 |M/4O| 1593 227 |/SOE| 726 | 70 |OPOE| 666 224 |OE/8| 766 | 70 |AM/Z| 294 208 |9/8A| 693 | 69 |/SOF| 3520 208 |9/4O| 2348 | 68 |OF9/| 1677 185 |9/SO| 1172 | 66 |/SO8| 550 159 |E/SO| 1274 | 65 |C9/F| 408 159 |E/8A| 1119 | 65 |AM/4| 2305 152 |SC9/| 657 | 65 |/8AR| 1326 147 |CC9/| 847 | 64 |S9/S| 1258 147 |/4OF| 2494 | 64 |M/8A| 3403 145 |/SOR| 1761 | 63 |8AR/| 790 145 |/4OP| 1881 | 63 |-4OF| 2327 142 |O89/| 254 | 62 |/Q9/| 476 134 |AM/S| 1506 | 62 |/8AN| 1530 127 |R/SO| 2107 | 61 |/OPS| 4087 125 |OR/O| 1290 | 59 |E/ZO| 1246 117 |PS9/| 3091 | 59 |C9/8| 5315 117 |OE/O| 450 | 58 |C9/S| 3712 116 |ZOE/| 19 | 58 |/ZO/| 430 111 |FS9/| 3979 | 56 |9/ZO| 4217 100 |4OPS| 1869 | 56 |/ZOR| 2394 99 |AM/8| 3132 | 56 |/OFA| 383 95 |OR/8| 1844 | 55 |9/FS| 2854 95 |OPSO| 1059 | 54 |OFAM| 163 95 |O8AM| 412 | 54 |/OFS| 863 91 |8AM-| 1981 | 53 |/SC9| 914 90 |/ZOE| 39 | 52 |SO89| 1028 89 |9/OP| 262 | 52 |SAM/| 1398 87 |S9/8| 278 | 52 |M/SO| 6362 87 |OPS9| 3835 | 51 |/8AE| 1478 85 |R/8A| 120 | 50 |SO8A| 458 85 |AM/O| 649 | 50 |PAM/| 15 84 |OFSO| 258 | 50 |89/S| 1278 83 |/S9/| 750 | 48 |C9/4| 2340 82 |COR/| 2074 | 48 |AR/S| 4601 81 |ZOR/| 1423 | 48 |/QOE| 670 81 |AM/Q| 835 | 47 |SAR/| 982 80 |OP9/| 150 | 46 |ZC9/| 501 79 |COE/| 1202 | 46 |SCC9| 722 77 |4OFS| 3423 | 46 |/QOR| 27 77 |/SCO| 1824 | 45 |OFOE| 2460 76 |R/SC| 689 | 45 |OFCC| 2380 76 |OR/Z| 3273 | 45 |4OPO| 2643 76 |FOE/| 505 | 43 |R/OP| 2833 If this is of any interest to anyone I can email them the data for the entire 1,290 strings (sorted or unsorted). One thing of interest does emerge. The nymph ZOE appears to be mentioned predominantly at the beginning or at the end of a word (or rather, before or after a space). The 4-letter strings which contain |ZOE| are: 116 |ZOE/| 19 90 |/ZOE| 39 7 |FZOE| 2914 5 |ZOEO| 12612 The same is true for ZOE's girlfriend SOE (variant of "Sue"?): 354 |SOE/| 266 227 |/SOE| 726 41 |FSOE| 6694 33 |PSOE| 7941 10 |SOEO| 2150 10 |BSOE| 2538 8 |SOE8| 8890 7 |SOE-| 521 6 |9SOE| 4419 6 |-SOE| 12960 3 |2SOE| 6441 2 |SOEA| 1681 2 |*SOE| 878 Note that ZOE also occurs preceded by F or followed by O, and the same is true of SOE. This suggests some connection between F and O. Another observation is that |AM| occurs often before a space. The frequencies are as follows: 380 |8AM/| 159 134 |AM/S| 1506 99 |AM/8| 3132 85 |AM/O| 649 81 |AM/Q| 835 74 |FAM/| 786 70 |AM/Z| 294 65 |AM/4| 2305 52 |SAM/| 1398 50 |PAM/| 15 32 |OAM/| 238 28 |AM/9| 4411 20 |AM/2| 2659 19 |2AM/| 331 14 |/AM/| 2619 11 |QAM/| 187 9 |RAM/| 124 7 |ZAM/| 11106 5 |WAM/| 6349 5 |CAM/| 12357 Who can throw some light on this? From j.guy@trl.OZ.AU Thu Jan 30 00:24:36 1992 Date: Wed, 29 Jan 92 10:24:36 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201282324.AA29859@medici.trl.OZ.AU> To: voynich@rand.org Subject: Re: Lojban, per frogguy's request Status: OR A Public Vilification of the Heinous Heresy that the VMS is not in Lojban Lojban postdates the VMS by a few hundred years? Big deal! Ever heard of Dr Who, the one with the time machine? You know, all those nymphs in the VMS? They're pictures of Dr Who's female companions, that's what! Dr Who and Romana wrote the Voynich manuscript. In Lojban, yes. Proof: haven't you noticed how frequent the sequence 8A8 is? It'obviously "BAB", as in "lojbab". So there! What more do you want? From reeds@gauss.att.com Thu Jan 30 00:39:13 1992 Message-Id: <9201291539.AA03919@rand.org> From: reeds@gauss.att.com Date: Wed, 29 Jan 92 10:39:13 EST To: voynich@rand.org Subject: Brumbaugh's transcription Status: OR My mailer does not stick in headers: the mail gurus where I work have GRAVE DOUBTS about mail headers. Sorry if it inconveniences you; I'll try to stick them in by hand. I'm glad Karl is looking into B's transcriptions, curbing my extravagances! Let me reply to Karl's comments, some of which I did not understand. Text quoted from Karl is marked with a |. First, I'm using my own transcription of f106v. I use Petersen for the ``star name'' examples. |The first three errors seem to be problems transcribing B's |transliteration of the text, as Jim's transcription of Petersen |fits B's matrix. Guilty on 2 out of 3, but not on this one: |> OBAUAE OBwRCE 138254 (C=5 error -- but note Jim has A=5) where B clearly has written the wrong V character but somehow got the right digits. Could he be going from folio to digits directly, without using his own V transcription? Some mistakes were caused by my misreading of his V transcription: |> FCORAN FCO2w) 331284 (2=2 error -- but note Jim has R=5) My fault. B actually writes a modern (flat barred) numeral 2 in his V letter transcription (which I copied verbatim) instead of the , and it all works out. But what does Karl mean with R=5 ? |> S8AJ S8AP 7852 (P=2 error -- but note Jim has J=2) My imprecision in transcribing B's written: he has an ugly cursive roman p or f, not a Currier

. Now that I refer back to his box I see he has a corresponding scribble (which I had taken to be a blot) near his box entry. So he's consistent there. |The next two errors (A=8) seem genuine -- is there something different |about the appearance of those two A's? |> 9SPAN 9SPA) 97684 (A=8 error -- see below) |> 9ZC8AM 9ZC8A) 973884 (A=8 error -- see below) Its hard to tell his transcribed A and w apart. If we call those w's, it works out OK. As for the rest. I can understand parsing Currier's into 3 parts. But in his transcription into V letters he doesn't. Where the text has a clear , Brumbaugh draws a clear , and where the text has he draws a clear . His drawings of and include the geometric feature represented by his , so that feature comes out duplicated. He does this funny on the right but not on the left: he does not draw or . It is a kind of visual 'rg' play that Brurgumbaurbugh is handing us! Maybe we should take this as more evidence that his transcription to digits on the bottom of page 7 of his Nov 1978 newletter comes directly from f107v and is not derived from his transcription into V letters given on the top of page 7. If so, (1) I think he is being disingenuous, and (2) we have NO DIRECT EXAMPLES AT ALL of how Brumbaugh transcribes (as opposed to deciphers). I suppose we could try to reproduce his reading of 75r on p126-7 of his book. Ugh. Jim Reeds From j.guy@trl.OZ.AU Thu Jan 30 02:34:30 1992 Date: Wed, 29 Jan 92 12:34:30 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201290134.AA00187@medici.trl.OZ.AU> To: voynich@rand.org Subject: Robert Firth's comments Status: OR Welsh: quite so. All modern Celtic languages behave in such manner. However, the main conditioning factor is grammatical, not phonological (it *was* purely phonological historically, but no longer so). For, instance, fy ("my") + ci ("dog") becomes "fy nghi", but the change c -- ngh is not regularly observed after words ending in "-y", only after "fy". Nevertheless, it could well be that mutations, in Voynich, are regularly caused by purely phonological factors. The statistical patterns of Voynich are only unusual in that they are so strong. I can think, however, of languages that may behave in striking statistical manner too. Swahili (and generally all Bantu languages): kitabu kikubwa kimoja "one big book", vitabu vikubwa viwili "two big books". The point made about preferred paragraph-initial letters is a truly excellent idea. Indeed, indeed. Personally, I tend to think that the very high frequency of a very few letters paragraph-initially is an effect of the phonology of the language. I imagine a language with a strong word-stress on the first syllable, and even stronger at the beginning of sentences, causing the initial consonant to be aspirated. But it is only pure imagination, and you'd be ill-advised to take it as a working hypothesis. Firth's idea, on the contrary, is a very plausible working hypothesis. Why, if the Voynich were in Latin, you'd expect many paragraphs to start with "De..."! Then: "This brings me to the second point: the need for comparisons with known texts in known languages. Every experiment done on the Voynich MS should be repeated, on, say, Magna Carta, the Romaunt of the Rose (separating the dialects), the Inferno, the Chanson de Roland - well, you get the idea - a chunk of contemporary (we hope) european documents." Yes. What do we have in computer-processable form? The Bible, the Book of Mormon, Lewis Carroll's fiction, I know of. I have been looking for the Latin and Greek classics, with no success so far. Greek would be particularly interesting for dialectal analysis. Take Sappho, Aristophanes, and Homer and see. I did not know that the Romaunt of the Rose was in two different dialects. Is it available in computer- crunchable form? Or shall we remotely twist Robert Firth's arm until he has typed in 10 pages of each dialect? Letter Construction. Quills definitely. My 60-pound book on calligraphy mentions that quills were used in Europe until well into the 19th century (it also almost *weighs* 60 pounds). The script is such that it could not have been written with the "pens" used by the Indians (as in Sanskrit, not Maya!). The calligraphy is definitely European, too. "As I see it, the single biggest problem we have today is deciding whether the spaces do or do not divide words." I disagree here. What constitutes a word is often difficult to agree on in a known language. Me, for instance, when I think about the grammatical properties of my own language, divested of its eccentric spelling, I tend to consider some phrases as constituting *really* a single word, for instance: "je le lui dis". I analyze it as a verb with three pronominal prefixes, thus: "jeleluidi". Unorthodox? Not in the least. If French were the language of some remote tribe of New-Guinea, never reduced to writing, that is precisely what linguists like Don Laycock of Enochian fame would come up with. Or to take a more transparent example: is it reasonable to consider "l'eau" as TWO words, phonetically "l" and "o". And a contrario, if we consider that "l'eau" is two words, then I insist that Andras Kornai put a stop to his silly spelling of Hungarian without further delay, and insert spaces before those final ek's and et's and om's and ban's and ben's thus: Kez ek et csokol om Do your hear me, Andras? Stop your unnatural spelling practices this very instant! Isten em! On the Voynich Vowels, now. It is nice to read that Robert Firth's completely different approach yields results similar to Sukhotin's algorithm. Converging results I see as very encouraging; signs that, at least there, we are on the right track. The hypothesis that: a alpha c epsilon cc eta o omicron oo omega is very interesting, and has a chance of being true. I have often found myself compelled to read the Infamous <4o> "ho" and incapable of resisting thinking of the Greek definite article. My equating with "a" is prompted by its similarity to the letter "a" of the Benventan script, and by some of its distributional properties. But = eta? Attractive. Now, I have written to two or three of you, last week I think it was, that I suspected that Currier's , (my ) was "u", ypsilon if you prefer. Time to air my reasons for thinking so: because of the high frequency of the digraphs and (my and ), and because Sukhotin's algorithm gives me no plausible candidate for a "u" sound. If is "u" then I think that = "ou" would have been pronounced as in Modern Greek: a "u" sound (English "oo"). Incidentally, there is a word somewhere, which would make it "uou", but no, I do not believe that it means "egg" ("uou" is "egg" in Limousin, a variety of Occitan spoken in Auvergne, in which you'll have recognized a relative of the Italian egg: uovo). Pity, it could have salvaged Levitov's thesis that the Voynich was a Cathar prayer book! Long ago, even before my Cryptologia article, I altered my transcription of folios 79v and 80r so that was represented by a single letter, and submitted it to Sukhotin's algorithm. Sure enough, came out as a vowels. The other vowels remained unchanged. I am not sure that means anything, but it encouraged me to keep "x = u" in mind. "U" on its own, that is, when not preceded by "a" or "o", might have been pronounced "i", as it is in most Modern Greek dialects (and in standard Greek, of course). Why do I say that? Because the vowel inventory still lacks an "i". You know how impressed I keep being with the similarities between the Voynich and Beneventan scripts. I am still trying to fit in Currier's (my ) as a "t", for that is precisely the value of that letter in Beneventan. But then, what are those dreaded "gallows" which are found so often in the vicinity of or slap bang inside it? I don't know who ZOE is, but if my wild guesses are right SOE is "tou". ZOE? Dou? Thou? What language is that? Well, perhaps not Lojban after all. My apologies, John Cowan. You may step down from the stake. From firth@SEI.CMU.EDU Thu Jan 30 02:35:36 1992 Date: Wed, 29 Jan 92 12:35:36 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201291735.AA08843@bp.sei.cmu.edu> To: voynich@rand.org Subject: Spaces Status: OR Folks Concerning spaces in the MS, Mr Guy's comments are quite correct. However, I still think there is an issue here. To say that the spaces divide words is a modern oversimplification; what I really should have said is that spaces have a morphological function, they divide the text into groups that carry meaning. With this definition, "words" are not really the issue, but rather "morphemes". To give an example, using the first bit of mediaeval poetry that comes to mind, we can all read this: Er ist ge ritt en hinnen O weh wer soll mich minn en in a seemingly very isolating language. And if we assume the same in a highly agglutinative language, where morphemes fuse eagerly, we have erist gerittenhinnen oweh wersoll michminnen By contrast, try to read this eri stgeri ttenhi nnen ow ehw ersollmi chmi nnen where I have inserted spaces using purely orthographic rules: always after 'w' or 'i'. And, of course, keeping the line as a functional unit. That's a very different kettle of fish, don't you think? You see, the statistical regularities we are finding will mean something quite different if spaces do not divide morphemes. So our interpretation of them, I think, requires us to decide this issue. Robert From firth@SEI.CMU.EDU Thu Jan 30 02:46:07 1992 Date: Wed, 29 Jan 92 12:46:07 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201291746.AA08854@bp.sei.cmu.edu> To: voynich@rand.org Subject: Letters S and Z Status: OR Folks Another wild & hairy speculation: If, as Mr Guy supposes, Currier's S (his ct) os the letter 't', then it is very, very tempting to guess that the Z (c't) is a letter very close to it. Could these be the tau and theta of the alphabet? If so, then SOE ZOE becomes "tou th[e]ou"; and, good grief, we also have, from the A section [51] Z9 th[e]a [12] ZAM thea? [32] ZC9 thea (e not skipped) [18] ZCC9 theia [83] ZO theo [12] ZO89 theo?a (theora??) [65] ZOR theo? (theon??) Well, I've filtered the entire text, making this change and all the other changes to implement Mr Guy's and my current hypotheses. The result looks very weird, but at least it's pronounceable and the vowel patterns look plausible. Alas, I can't recognise it as greek, but one reason is that I can't recognise real greek in romanised form, I have to read it aloud. I was reminded of this vividly the other day, reading an article on greek music, where the author gave all original quotations in a weird romanised form. Quick, guys, how long did it take you to see that "phthoggai" was a greek word, and not a squamous monster out of H P Lovecraft? So, does anyone know how I can get a greek font into Unix, Bloody Unix? Robert From j.guy@trl.OZ.AU Thu Jan 30 03:50:31 1992 Date: Wed, 29 Jan 92 13:50:31 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201290250.AA00273@medici.trl.OZ.AU> To: voynich@rand.org Subject: Voynich vowels Status: OR Some fellow by the name of Jacques Guy has just written: "U" on its own, that is, when not preceded by "a" or "o", might have been pronounced "i", as it is in most Modern Greek dialects (and in standard Greek, of course). Why do I say that? Because the vowel inventory still lacks an "i". Did it not occur to the poor soul that if indeed it is "eta", is a prime candidate for being that "i"? Hasn't he noticed the high frequency of the sequence , presumably "ia", a most natural and common vowel sequence in so many languages, Italian, Latin, Greek, and, I expect, most of those extraordinary languages he has been bashing our ears about, allegedly spoken in remote islands the names of which are not found in any self-respecting atlas? (I think he makes them up). From firth@SEI.CMU.EDU Thu Jan 30 05:09:13 1992 Date: Wed, 29 Jan 92 15:09:13 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201292009.AA09092@bp.sei.cmu.edu> To: voynich@rand.org Subject: More wild guesses Status: OR Folks This seems to be the day for wild guesses. So far, Mr Guy and I have speculated about the vowels and the consonants tau and theta. I added two more guesses. First, the Currier letter R seems to be final very often, and usually after the vowels A and O (which might be alpha and omicron). That strongly suggests it has the value nu. Secondly, the Currier 8 occurs before what we think are vowels (89,8A,8O), and what might be tau (8S). Since the Voynich is written in mediaeval bantu, it must be 'n'. Nope, if we are still thinking greek or romance, it might be 'p', but a much more likely choice is sigma. So I did it. I filtered the whole text using C c epsilon CC cc eta, romanised as 'i' A a alpha O o omicron E x upsilon S ct tau Z c't theta, written "th" if 'e' follows, else "the" R nu 8 8 sigma with the romanisations obvious except where noted. (Sorry, Jacques, I forgot your equivalent to Currier R). Here's the result, from the beginning of the transcription. Lower case letters are wild guesses; everything else is original Currier. -------- Vata2 aFau an aPaM theZou theZona Qona2 a Fon theZousa 2ona Xan o n a FaN theZPaM theZan an* Qan Qan saD 2aaU theZeFa on aFaM theZos Qoana Q*n sanaM 2a $oM oPia oPeon 2ouoPa Q*an saM oFaM ou oFaD 2aT a tean QaM Wan YaM# asanaItheZo# * osan *a theZou Woa oasan theZ* 2 YoaM theZosa2a *theZa tosa oFtoa oPtou toQa o2 ta saN theZon Fo2 saM theZo2 You theZosa# saN *oPeosa# * asaN We2aM ou 2 Wea aPaN theZotheZa Wosau e2 oFtheZo FtheZoa oPa3 oPeou oFaD theZosaN 2Xa saM theZoa Xea FosaM Wa WosaH2 Qea theZo ousaN s saN oM tou osaM tosaN tsa oFaN s*D Qa Fos saM theZXea Xon ton theZea Fou tou tou Fon tou theZo tou theZ osaD FtheZa Fta son tosaM theZo FiaJ ato Ptea teFaN theZeoBtheZou s0sas Qa saQa aPo theZou theZe FostheZea Weaua sanaN saN Xa,, stan theZQaM oFaU tea *ta BoPou Qou2 sasePo theZoF ton tea saN Xoa# oPou sa3# Wo theZaM theZoFtia tou PtheZosi2a theZea Basia ta no s* *soN tou saN Qau san theZean FaM san theZea Qan to *o FaM theZoaM oFou saM Van Qou saM Qousan atia oFaa oFa saM oFtea F oFaM **tou F**ta sau se*o theZosa FotheZea Qa oF tea Fia sau tPon *o* tou toF toPa toPea# staM# Ft2a tasaM ou ouPtheZea tan Yan aJ a Pia tan on ota sta uFosa oFosan tosa sa Xa Xo Xa theZa sPtheZia Qa FoPtosa sau sau toFeo saT saJ 2otea to Fosa# BoPoa theZou saT Woau san tea Posa oPoaM theZotheZa toFa tou Qou theZou oFau soutea toso uou ta Qa 4o ou toi2 teou sou Qea aFou sou souo aFou so uteosa oFou theZou Fou Feta tou Fa tou Qou tosa tou saM theZon oFou tou sou Fa san theZou ston oPto san tosa Poon toPtea sau tosa 2tosa Bou tosan# FasaNa aBtou saM oPtau aBtaM Xou2a sontona tFan 2 theZon Qa *** 4oPaM Qea a ton ta asa taM *aIIsa tPos sa Wa sau 2 toFaM s oPoton au theZosaM tou taD aPtaM saD 2aM saNs s Fou 2on aPousa stou sta Qa theZon Xa saMa tou saD# FasaN theZaM 4oa 2 theZou VosaD aPtheZ outheZia saHsa su2theZo Fou theZia 4oFea aFosa 2o tou aFa saN saUsu 4oFa touaM theZou theZeFa saM Qea Feou 2aM 2aM ataN sau ta sauon theZaD **2aM theZia Xon oFou ta ton Qon aon aD taD 2aM tePa taFa 2au theZo aFia tea saM tQa# FooM teoBton oFaM osaN ton saT theZPa FtoFta theZo theZou 4oPto uoie2 4oPa ton saM oPta ton utheZa tou tosa tosaN tQa saM theZo touo teon tosaM# -------- Well, can anyone else make sense of that? It seems hauntingly familiar, but I get no gestalt. And yes, the spaces do not seem to divide words. By the way, the Currier M is the next candidate for speculation. His 8AM is the most common "word" in the A text (I count 355); what it means, I have no idea, even if the first two letters are indeed "sa". Robert From firth@SEI.CMU.EDU Thu Jan 30 05:24:47 1992 Date: Wed, 29 Jan 92 15:24:47 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201292024.AA09209@bp.sei.cmu.edu> To: voynich@rand.org Subject: Currier 0 (zero) is O (oh) Status: OR Folks Still trying to knock 'em down! I believe Currier's digit-zero is just the same as his capital-O. The digit occursfive times in the A text: 80898 20A 202 02 ZO0JOFS9 (yes, one of them is a zero!) The following, with letter-O, are all fragments in the same text: 8O89 2OA 2O2 O2 ZOO9 Indeed, O2 is quite a common terminator. Does this seem reasonable? Robert From firth@SEI.CMU.EDU Thu Jan 30 06:23:07 1992 Date: Wed, 29 Jan 92 16:23:07 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201292123.AA09303@bp.sei.cmu.edu> To: voynich@rand.org Subject: Corrected text Status: OR Folks Sorry about that; in my last message I included the wrong text. For some bozo reason, I used two successive filters, and you got the result of the first only, with zillions of intrusive ZZZ. The text I meant to send follows: -------- Vata2 aFau an aPaM theou theona Qona2 a Fon theousa 2ona Xan o n a FaN thePaM thean an* Qan Qan saD 2aaU thiFa on aFaM theos Qoana Q*n sanaM 2a $oM oPia oPeon 2ouoPa Q*an saM oFaM ou oFaD 2aT a tean QaM Wan YaM# asanaItheo# * osan *a theou Woa oasan the* 2 YoaM theosa2a *thea tosa oFtoa oPtou toQa o2 ta saN theon Fo2 saM theo2 You theosa# saN *oPeosa# * asaN We2aM ou 2 Wea aPaN theothea Wosau e2 oFtheo Ftheoa oPa3 oPeou oFaD theosaN 2Xa saM theoa Xea FosaM Wa WosaH2 Qea theo ousaN s saN oM tou osaM tosaN tsa oFaN s*D Qa Fos saM theXea Xon ton thia Fou tou tou Fon tou theo tou the osaD Fthea Fta son tosaM theo FiaJ ato Ptea teFaN thioBtheou s0sas Qa saQa aPo theou thi Fosthia Weaua sanaN saN Xa,, stan theQaM oFaU tea *ta BoPou Qou2 sasePo theoF ton tea saN Xoa# oPou sa3# Wo theaM theoFtia tou Ptheosi2a thia Basia ta no s* *soN tou saN Qau san thian FaM san thia Qan to *o FaM theoaM oFou saM Van Qou saM Qousan atia oFaa oFa saM oFtea F oFaM **tou F**ta sau se*o theosa Fothia Qa oF tea Fia sau tPon *o* tou toF toPa toPea# staM# Ft2a tasaM ou ouPthia tan Yan aJ a Pia tan on ota sta uFosa oFosan tosa sa Xa Xo Xa thea sPtheia Qa FoPtosa sau sau toFeo saT saJ 2otea to Fosa# BoPoa theou saT Woau san tea Posa oPoaM theothea toFa tou Qou theou oFau soutea toso uou ta Qa 4o ou toi2 teou sou Qea aFou sou souo aFou so uteosa oFou theou Fou Feta tou Fa tou Qou tosa tou saM theon oFou tou sou Fa san theou ston oPto san tosa Poon toPtea sau tosa 2tosa Bou tosan# FasaNa aBtou saM oPtau aBtaM Xou2a sontona tFan 2 theon Qa *** 4oPaM Qea a ton ta asa taM *aIIsa tPos sa Wa sau 2 toFaM s oPoton au theosaM tou taD aPtaM saD 2aM saNs s Fou 2on aPousa stou sta Qa theon Xa saMa tou saD# FasaN theaM 4oa 2 theou VosaD aPthe outheia saHsa su2theo Fou theia 4oFea aFosa 2o tou aFa saN saUsu 4oFa touaM theou thiFa saM Qea Feou 2aM 2aM ataN sau ta sauon theaD **2aM theia Xon oFou ta ton Qon aon aD taD 2aM tePa taFa 2au theo aFia tea saM tQa# FooM teoBton oFaM osaN ton saT thePa FtoFta theo theou 4oPto uoie2 4oPa ton saM oPta ton uthea tou tosa tosaN tQa saM theo touo teon tosaM# From reeds@gauss.att.com Thu Jan 30 06:28:44 1992 Message-Id: <9201292129.AA12120@rand.org> From: reeds@gauss.att.com Date: Wed, 29 Jan 92 16:28:44 EST To: voynich@rand.org Subject: Counting repeated Voynich substrings. Status: OR For each of the A and B portions of the original D'Imperio transcription, I squished out all punctuation marks, leaving only characters in the 36 letter Currier alphabet augmented with the * sign for illegibile symbol. I counted the frequences of all repeated substrings, discarding only those substrings which are just as numerous as containing superstrings. (So that if all occurences of tamboul were in Stamboul, I would not report a separate frequency for tamboul.) The data is much too voluminous to report in full. I can supply excerpts summaries, or the program that did the work. Here is a taste of it: Sorted by length of repeated strings A corpus: freq string length 2 8AMO8AK4OFAM9PCAEOFAEOJ 23 2 8AM8AM4OPS94OP 14 2 8AM4OPS94OPOR 13 2 SOESOESOE8AM 12 2 SCORSOE8AMQ9 12 2 OFSOESOESOE8 12 2 OESC9SOFSOES 12 2 O8AMOFAEOE8A 12 2 COESORSOE8AM 12 2 CC98AM4OPCC9 12 3 8AM4OPS94OP 11 2 SOE8AMQOE8O 11 2 SOE8AM4OPOE 11 2 SC94OPZC98A 11 2 OR8AM4OPSOE 11 2 OPSOPSOPSOP 11 2 M8AM4OFSORO 11 2 FS94OPSOESO 11 2 AMQORSOESOR 11 2 AM4OFCC9FSO 11 2 8ARZC9QARSO 11 2 4OFOE8AM2OE 11 ... B corpus: freq string length 2 C894OFC894OFC894OFC894OF 24 2 ZC894OFC894OFCC894OFC 21 2 OFAMZC894OFCC894OPC89 21 2 OEZC894OFC894OFCC894O 21 2 FCC894OFC894OFC894OFC 21 2 C894OPC894OFCC894OFCC 21 2 ZC894OFC894OFCC894OP 20 2 CC894OESC94OFCC94OFC 20 2 C89PC89OEZCC894OFCC9 20 2 C894OFCC894OFC894OFC 20 2 C894OFCC894OBSC894OF 20 2 4OFCC894OPC894OFCC89 20 4 ZC894OFC894OFCC894O 19 3 C894OPC894OFCC894OF 19 3 C894OFC894OFC894OFC 19 3 94OFC894OFC894OFC89 19 2 94OFC894OFCC894OFCC 19 2 4OFCC894OFC894OFC89 19 2 4OFC894OFCC894OFC89 19 4 C894OFC894OFC894OF 18 3 94OFC894OFCC894OFC 18 2 ZC894OFCC894OFCC89 18 2 OFC894OFC894OFCC89 18 2 CC894OFCC894OFCC89 18 2 C894OFCC89SC894OFA 18 2 C894OFCC894OPC894O 18 2 AE4OPCC894OFAMZC89 18 2 9FC89OEVSC894OFC89 18 2 94OFCC894OFCC894OF 18 2 4OFCC894OFCC894OPA 18 6 C894OFCC894OFCC89 17 4 94OFC894OFC894OFC 17 4 894OPC894OFCC894O 17 4 4OFC894OFCC894OFC 17 3 OEZC894OFC894OFCC 17 3 FCC894OFC894OFC89 17 3 AMZC894OFCC894OPC 17 3 4OFCC894OFC894OFC 17 2 ZC894OFCC894OFC89 17 2 SC894OFCC894OPC89 17 2 PCC894OFAMZC894OF 17 2 OEFCC894OFCC894OF 17 2 FCC894OFCC894OFCC 17 2 FCC894OFCC894OFAR 17 2 FC89ESC894OFC894O 17 2 EZC894OFAESC894OF 17 2 CC94OEFCC894OFC89 17 2 CC894OPCC894OFCC9 17 2 C894OFC894OFCC89E 17 2 C894OFC894OFC898A 17 2 9ESC894OEFCC894OF 17 2 94OFC894OFC894OFA 17 2 894OFCC89ESC894OF 17 2 894OFC894OPCC894O 17 2 4OPCC894OFCC894OF 17 ... Sorted by frequency: A Corpus 5152 O 1 2878 9 1 2852 S 1 2050 A 1 2011 8 1 1523 E 1 1500 C 1 1391 SO 2 1320 R 1 1296 F 1 1211 OE 2 1166 P 1 1009 M 1 993 8A 2 949 OR 2 923 AM 2 903 Z 1 730 OF 2 714 OP 2 631 4 1 617 4O 2 571 O8 2 525 S9 2 503 8AM 3 501 ZO 2 ... B Corpus 5688 C 1 5308 9 1 5283 O 1 3990 8 1 3065 A 1 2846 89 2 2839 E 1 2794 F 1 2376 C8 2 2206 S 1 2164 C89 3 1933 4 1 1865 4O 2 1834 OF 2 1601 OE 2 1546 94 2 1495 94O 3 1432 SC 2 1367 Z 1 1366 R 1 1234 4OF 3 1230 CC 2 1228 P 1 1189 FC 2 1043 C9 2 1036 ZC 2 1014 FA 2 ... From reeds@gauss.att.com Thu Jan 30 06:34:07 1992 Message-Id: <9201292134.AA12233@rand.org> From: reeds@gauss.att.com Date: Wed, 29 Jan 92 16:34:07 EST To: voynich@rand.org Subject: 0 = O in D'Imperio's transcription? Status: OR Robert Firth writes: I believe Currier's digit-zero is just the same as his capital-O. 1. It's D'Imperio's transcription, using Currier's transliteration scheme. 2. Yes: quite likely. There are a lot of key punching errors. As you spot such errors, why don't you check them out against xeroxes of the MS and send corrections to Jim Gillogly? Jim Reeds. From jbaez@math.mit.edu Thu Jan 30 06:53:45 1992 Date: Wed, 29 Jan 92 16:53:45 EST From: jbaez@math.mit.edu Message-Id: <9201292153.AA03408@riesz> To: voynich@rand.org Subject: finding words Status: OR So, Jacques has been tantalizing me with his hints of top-secret algorithms that spot words. Recall the 0th-order way of finding words. For each n-graph (substring with n letters), calculate how frequently it appears divided by how frequently it "should" appear based on knowledge of the frequencies of 1,2,3,4... n graphs. (I.e., fit the text by the best Markov chain that keeps track of what the last n-1 letters were. Ahem - should read 1,2,3,4,... (n-1) graphs above.) Keeping the high-scorers in this contest gives you some candidate "words" - the problem is, a bunch will be substrings of others. So to see how much like a word the string th is, we should SUBTRACT OFF from how often it appears the number of times it appears as a substring of another candidate word, like "the". If one has a set of candidate words one can do this and re-calculate the ratio above. That should be a good step towards solving the problem. But really one would like it all to be automatic: to find a "best fit" of a text by some language of words. Of course this is harder. For example, one needs to rule out such solutions as "the whole text is one big 23498723499- letter word." I.e., one needs a reasonable notion of best fit, and then a reasonable-time algorithm to find reasonably good fits. There's no point in striving for the utterly BEST fit (even if one has a compelling notion of best fit) since the real solution will probably only be close to optimal. jb From firth@SEI.CMU.EDU Thu Jan 30 07:21:44 1992 Date: Wed, 29 Jan 92 17:21:44 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201292221.AA09423@bp.sei.cmu.edu> To: voynich@rand.org Subject: More on Spaces Status: OR Folks A final wild speculation and then I shut up for the night; promise! I've been trying further symbol substitution, based really on most frequent letters - take the highest-frequency Voynich symbol not yet guessed, and try the unused letters in descending order of frequency. Of course, the "guessed" letters are already wild speculation, so this is O(wildness^2). But then. Well, I'd just set Currier's M to be our missing iota, and, on looking at the text, noticed something very odd. Most of the spaces now came just after things like -ai -au -an -ou -on -eu with a few -sa thrown in. And that did trigger a memory. Suppose these are the long vowels, nasalised vowels, and long diphthongs? And suppose this text was being read aloud from an old copy, and written down in a new copy. One pretty natural place to leave a space is exactly in those places. This is especially true if the reader is fairly slow (as he'd have to be) and uses strong stress and intonation. I well remember chanting greek verse, and all those "hoi tou theou logoi", and how long and sonorous it sounded. So, for instance, if "daimonioi" was read, it would be written dai monioi or daim onioi In other words, folks, the unit might be a metrical foot with a long ending (iambus, anapaest, spondee?), even if the text is prose. Ah, the thrill of the chase! And, like Actaeon, I'll probably come to a bad end, pursuing Artemis' own sphynx. Robert From rcarter@isis.cs.du.edu Thu Jan 30 09:54:37 1992 Date: Wed, 29 Jan 92 17:54:37 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9201300054.AA00416@isis.cs.du.edu> To: voynich@rand.org Subject: Latin abbreviations... Status: OR This might be old hat to some of you, but... Just got a copy of "The Elements of Abbreviation in Medieval Latin Paleography" by Adriano Cappelli (translated by D. Heimann & R. Kay) Univ. of Kansas, 1982. This is a translation of Cappelli's prefatory treatise, part of his "Lexicon Abbreviaturarum..."; I am reading this first, because of 52p vs. 400p, and English v. Italian... Couple of points jump out: the symbol "9" in Latin abbreviations has at least two functions; at the -beginning- it is short for `cum' (as in cir-cum-ference) or con (con-ceive); when at the end, it is short for -us, -os, -is, or simply -s. Now, does 9 appear anyplace other than at the start or finish of a word? Cause in Latin abbreviations, 9 doesn't seem to appear anywhere else... As pointed out by D'Imperio (f17) there -are- alot of similar symbols between Voynich and Latin abb... Ron. From reeds@gauss.att.com Thu Jan 30 11:03:36 1992 Message-Id: <9201300203.AA18620@rand.org> From: reeds@gauss.att.com Date: Wed, 29 Jan 92 21:03:36 EST To: voynich@rand.org Status: OR My Whitby came today! (Christopher Whitby, "John Dee's Actions with Spirits", Garland, 1988, an edition of Sloane MS 3188, with commentary.) One juicy plum: on f96r: Aprilis 29. Monday, a meridie DEE As E K and I wer talking of my [boke] boke Soyga, or Aldaraia: and I at length sayd that, (as far, as I did remember) Zadzaczadlin, was Adam by the Alphabet thereof, suddenly appeared the spirituall creature, which sayd ... If the VMS is the Boke of Soyga we have a pattern word, plus identifications of 3 letters! From reeds@gauss.att.com Thu Jan 30 13:11:43 1992 Message-Id: <9201300411.AA20037@rand.org> From: reeds@gauss.att.com Date: Wed, 29 Jan 92 23:11:43 EST To: voynich@rand.org Status: OR Another, much unexpected tidbit from Whitby. On pages 537-8 of vol 1, we see a quotation from Dee's Monas Heiroglyphica (p.209 in Josten edition) where Dee explains the formula n! for the number of permutations of n objects. This is the last thing, the very last thing, I expected to see of this Magus, this benefactor of the Elizabethan Mechanicians, this true partaker and continuator of the Hermetic tradition, this preceptor of the Sidney Circle, this Rosicrucian enlightener, this patron of scryers, this friend of Rudolph's: to wit, some actual factual math. Kind of like learning that Johannes Trithemius invented the card catalogue, or that the Phaistos disk holds Glyph-Perfect files! From firth@SEI.CMU.EDU Thu Jan 30 23:16:02 1992 Date: Thu, 30 Jan 92 09:16:02 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201301416.AA12513@bp.sei.cmu.edu> To: voynich@rand.org Subject: Zadzaczadlin? Status: OR But, which three Voynich letters are named Zad, Zac and Lin? By the way, the book 'Soyga' is surely just 'agyos' backwards; this is from greek "hagios", deruved either directly or via welsh. And gee, Zo:e: IS the name of Eve, in the greek OT. The gallows letters? Well, I have a few wild speculations... Robert From j.guy@trl.OZ.AU Thu Jan 30 23:33:52 1992 Date: Thu, 30 Jan 92 09:33:52 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201292233.AA01389@medici.trl.OZ.AU> To: voynich@rand.org Subject: More wild guesses Status: OR Oh, I see now. Firth meant the Greek letter "nu", i.e. "n". I had understood the *sequence* N+U.Yes, a nasal seems natural there. The gallows occurs extremely frequently in the immediate vicinity of Currier's S and Z (my ct and c't), which made me once (recently) think: S = r Z = l with the gallows inspired from capital letters and representing them: = T (Currier's P) = K ( -- F) = P ( -- V) The beginning of the Voynich would read: paraR akau a2 atam lou lo2a (kr)o2aR... Again, when you consider that Currier's S is the only letter that gets interrupted by gallows, perhaps that S is "s" when written separately, and denotes the aspiration of the gallows when they intrude into it... and Z is "r" or "sh" or...?. So: pasaR akau a2 atam lou lo2a kho2aR.. In other words: = ts = st = th and so on... Alas, it wrecks my pet theory that S (ct) = "t" From reeds@gauss.att.com Fri Jan 31 00:41:05 1992 Message-Id: <9201301542.AA29523@rand.org> From: reeds@gauss.att.com Date: Thu, 30 Jan 92 10:41:05 EST To: voynich@rand.org Subject: Zadzaczadlin! Status: OR Firth writes: By the way, the book 'Soyga' is surely just 'agyos' backwards; this is from greek "hagios", deruved either directly or via welsh. An easy assumption, but we have direct testimony against, on lines 41-43 of f89v, Sloane MS 3188, where Dee and Il (a stand-in for Uriel) have an after dinner chat, Thurday, 18 April 1583. ------ DEE: But it was reported to me by this skryer that he had: certain peculier bokes pertayning to Soyga. otherwise named ysoga, and Agyos, literis transpositis. Il: Soyga signifieth not Agyos. *Soyga alca miketh.* ------ There you have it, Robert. From RJB@U.WASHINGTON.EDU Fri Jan 31 01:51:00 1992 Date: Thu, 30 Jan 92 08:51 PST From: RJB@U.WASHINGTON.EDU Subject: benefactions of the benefactor of Elizabethan mechanicians Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Well, this is one of the features of the Renaissance hermetists and hermeticists that caused such a flutter among historians of science a generation or so ago: they undercut and made rather problematic the comfortable distinction between the "rational" and the "irrational" (as in E R Dodds' _The Greeks and the Irrational_) that had characterized earlier notions about the nature of the changes that had gone on from the decline of pagan Rome through the dark ages tto the Renaissance and (trumpets please) Science. This is exactly what one does expect to find among hermeticists: a very odd mix of the visionary, the practical, the conceptually astute, and the flagrantly absurd. A (no longer quite) cvontemporary parallel: I knew some people (again, about 20 years ago) who had worked out an elaborate pharmacologic typology of structure-activity relationships of certain compounds characterized ... well, suffice it to say that the key to the whole thing was the planetary attrributions of the carbon atoms in the benzene ring. (chorus) Why were they born... too late? (violins) --rjb From jbaez@math.mit.edu Fri Jan 31 02:44:20 1992 Date: Thu, 30 Jan 92 12:44:20 EST From: jbaez@math.mit.edu Message-Id: <9201301744.AA00954@paley> To: voynich@rand.org Subject: Trithemius Status: OR I've been reading Dee's preface to Euclid and it's clear theat he knows a lot of actual factual math. For those not in the know, Trithemius *is* famous for his invention of the annotated bibliography. (My facts may be off a *little* here but I;'m close.) From reeds@gauss.att.com Fri Jan 31 03:41:35 1992 Message-Id: <9201301841.AA03766@rand.org> From: reeds@gauss.att.com Date: Thu, 30 Jan 92 13:41:35 EST To: voynich@rand.org Subject: Dee as mathematician Status: OR I think my complaint is against the modern commentators on Dee: Yeats, Clulee, and French. One would never know of any direct evidence, from reading their books, that Dee knew any mathematics. He encourages the mechanicians (whoever they were) by writing a preface to a translation of Euclid, he was in favor of the discovery of the north-west passage, he wanted calendar reform, he invented a kind of compass whose description is now lost, the charts in his preface to Euclid are not Ramist so much as Lullist. All worthy, no doubt, and ``progressive.'' I am willing to believe Dee was the Mr. Wizard of Elizabethan England. But what actual mathematics did Dee know, or invent? Was he, for instance, Cardano's equal, or not? These questions are simply uninteresting to Yeats & Co. John's report that Dee's ``Mathematical Preface'' shows evidence of a knowledge of mathematics is news to me. Of all modern commentators, Whitby is the exception in this regard: his book is indeed of some excellence. These are of course the maunderings of a hopelessly Whiggish mind, an historiographical Newt, boggled by the (apparently) paradoxical mix of rational and irrational in the Hermitic renaissance. From jim@rand.org Fri Jan 31 03:42:53 1992 Message-Id: <9201301842.AA19443@mycroft.rand.org> To: voynich@rand.org Subject: Zadzaczadlin -- It could be spelled out completely In-Reply-To: Your message of Thu, 30 Jan 92 10:41:05 -0500. <9201301542.AA29523@rand.org> From: Jim Gillogly Reply-To: jim@rand.org Date: Thu, 30 Jan 92 10:42:53 PST Sender: jim%mycroft@rand.org Status: OR If "Zadzaczadlin" were spelled out in 12 letters instead of 4, as the Angels did with the initial batch of Enochian that Kelley dictated (vide Whitby), we'd get about five matches in our current corpus, ignoring "word" and "line" breaks per Jacques' current hypothesis: ---------------------------------------------------- 8SOR.SOE.ZOE.QOE.ZPS9.SAM.*2.SORAM.SOJ- za dza cza dlin ---------------------------------------------------- [If CC is a single character...] 9SOE.SOE.8OE.SCC9.PSOE.8AR.X9- za dza cza dl i n ---------------------------------------------------- [I kinda like this one, since the "word" boundaries almost match.] ZAJ.OFCAE.8AE.8AJ.8AE- zad zac zad 2ZC9.OPAJ.ZAJ.QO7.OF9- lin ---------------------------------------------------- 4OFCC9.FCOR.8AM.OFC9.FCC9.8AM.8AT.8AM.2- zad zac zad l OFCC9.8AM.4OFS9.SOPC9.PO89-OF9.A3.8- in ---------------------------------------------------- 8SCOR.SCC9.FCOESC9.SOFSOE.SOFAD- zadzac zadli 4OPSO.SO8I.SO.SOS9.S2.ORM- n ---------------------------------------------------- Jim Gillogly From jim@rand.org Fri Jan 31 04:14:10 1992 Message-Id: <9201301914.AA19762@mycroft.rand.org> To: voynich@rand.org Subject: Re: Latin abbreviations... In-Reply-To: Your message of Wed, 29 Jan 92 17:54:37 -0700. <9201300054.AA00416@isis.cs.du.edu> From: Jim Gillogly Reply-To: jim@rand.org Date: Thu, 30 Jan 92 11:14:10 PST Sender: jim%mycroft@rand.org Status: OR > rcarter@isis.cs.du.edu (Ron Carter) writes: > Now, does 9 appear anyplace other than at the start or finish of a > word? Cause in Latin abbreviations, 9 doesn't seem to appear anywhere Yes, it appears in the first word of the first line of f 1r, as well as in about 500 other places (rough count). VAS92.9FAE.AR.APAM.ZOE.ZOR9.QOR92.9.FOR.ZOE89- Does the abbreviation book have things like the (Currier) P symbol? I haven't seen that one around, although Jacques has been tantalizaing us with his new reference book. Jim Gillogly From j.guy@trl.OZ.AU Fri Jan 31 04:14:57 1992 Date: Thu, 30 Jan 92 14:14:57 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201300314.AA01757@medici.trl.OZ.AU> To: voynich@rand.org Subject: Zadzaczadlin? Status: OR A D A M zad zac zad lin Hmm, but the Enochian alphabet goes: Ur, Na, Gon, Tal, Graph, Un, Or, Gal, Ged, Veh, Pa, Gisg, Fam, Van, Ceph, Don, Med, Pal, Drux, Ger, Mals (that is: l, h, i, m, e, a, f, d, g, c, b, t, s, v, z, r, o, x, n, q, p). Therefore, ye boke of Soyga is not writ in Enochian, and since we now, from all those statistical counts, that the Voynich is not written in Enochian, neither are. Therefore both are one and the same. Oh, yesyesyesyesyes, makes a great deal of sense to (like said Peter O'Toole, playing Robinson Crusoe in "Man Friday"). ZOE would then be Eve, whose name is said to derive from a Hebrew root meaning "life". From j.guy@trl.OZ.AU Fri Jan 31 04:22:54 1992 Date: Thu, 30 Jan 92 14:22:54 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201300322.AA01760@medici.trl.OZ.AU> To: voynich@rand.org Subject: Stumped! Status: OR I couldn't resist, and the programming was trivial anyway. Here is a linear correlation coefficient table computed on the first 20 verses of Genesis, King James version. Divide everything by 1000: -12 is r=-0.012. (:) is the beginning of a verse. I've put in an asterisk wherever I saw a comfortingly high correlation. Well, shock, horror, aaarrgh! "A" is well-behaved, but that's about all. I am particularly distressed at the naughty behaviour of "E" which correlates best, not with a fellow vowel, but with "T". If you paid attention to my rantings and ravings of a few days ago, you'll perhaps remember that a high correlation between two letters (X and Y) means that X tends to occur sandwiched between the same letters as Y does. "Sandwiched" because here I have used the sum of two matrices: X with the following letter, and X with the previous letter (hence the cryptic "Environment(s): 1 -1" heading this table. Back to the drawing board, frogguy, back to the drawing board. What's going on there that didn't go on when I did counts on words, instead of letters? Mumble, mumble, mumble.... 130-column wide table coming up... As you can see, there's not much rhyme nor reason in there. Perhaps I should have a close look at my Pascal code? oqpox! (A frequent Voynich word, the meaning of which is obvious if you use my transcription system: o pox!) File "gen.t1" Environment(s): 1 -1 <:>

<:> - -12 70 603 245 256 230 229 239 130 -5 322 734 707 -38 93 282 644 230 7 78 704 95 633 -12 - 363 189 670 389 35 346 329 807* 402 155 24 16 298 -83 204 210 285 253 203 145 105 4 70 363 - 594 281 380 251 312 414 151 330 729 557 346 219 469 714 486 229 63 746 381 -141 241 603 189 594 - 360 644 396 480 375 213 441 585 798* 745 213 266 534 747 533 215 493 703 6 704 245 670 281 360 - 183 470 620 426 626 466 395 441 267 170 276 502 499 211 461 448 452 36 252 256 389 380 644 183 - 105 288 239 518 223 205 263 183 289 -29 186 273 813* 232 158 542 -78 382 230 35 251 396 470 105 - 650 609 34 234 506 515 373 203 478 642 487 173 639 440 554 231 233 229 346 312 480 620 288 650 - 135 374 366 357 413 460 258 372 425 402 449 557 323 398 7 80 239 329 414 375 426 239 609 135 - 233 210 610 467 328 224 231 739 589 162 211 575 619 338 222 130 807* 151 213 626 518 34 374 233 - 261 -5 -27 81 492 -179 30 152 342 352 -59 203 87 84 -5 402 330 441 466 223 234 366 210 261 - 325 299 199 65 259 313 575 137 396 467 113 392 361 322 155 729 585 395 205 506 357 610 -5 325 - 747 574 11 699 859* 725 190 127 849* 640 100 358 734 24 557 798* 441 263 515 413 467 -27 299 747 - 724 -49 518 752 847* 230 139 670 790 -62 653 707 16 346 745 267 183 373 460 328 81 199 574 724 - 180 276 458 743 181 4 347 545 138 447 -38 298 219 213 170 289 203 258 224 492 65 11 -49 180 - -124 45 -2 149 300 -91 -5 -58 119

93 -83 469 266 276 -29 478 372 231 -179 259 699 518 276 -124 - 619 481 91 222 617 381 7 125 282 204 714 534 502 186 642 425 739 30 313 859* 752 458 45 619 - 733 247 144 915* 705 120 196 644 210 486 747 499 273 487 402 589 152 575 725 847* 743 -2 481 733 - 264 172 675 745 326 544 230 285 229 533 211 813* 173 449 162 342 137 190 230 181 149 91 247 264 - 133 204 599 -35 141 7 253 63 215 461 232 639 557 211 352 396 127 139 4 300 222 144 172 133 - 44 186 163 268 78 203 746 493 448 158 440 323 575 -59 467 849* 670 347 -91 617 915* 675 204 44 - 532 27 167 704 145 381 703 452 542 554 398 619 203 113 640* 790 545 -5 381 705 745 599 186 532 - 70 493 95 105 -141 6 36 -78 231 7 338 87 392 100 -62 138 -58 7 120 326 -35 163 27 70 - -26 633 4 241 704 252 382 233 80 222 84 361 358 653 447 119 125 196 544 141 268 167 493 -26 - From RJB@U.WASHINGTON.EDU Fri Jan 31 05:02:00 1992 Date: Thu, 30 Jan 92 12:02 PST From: RJB@U.WASHINGTON.EDU Subject: The sins of the Commentators Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Yates was not especially I think interested in the technical side; I haven't read French in years but I seem to remember some discussion of it. I may well misremember, though. I will go through my notes &c and see what I can dig up of commentators who discuss the mathematical side of Dee, and the technical. One problem (someone discussed this) seems to be that the things that survive of the Dee corpus do not include most of the works listed in his own lists of his writing, which describe a great number of technical/scientific texts. I keep thinking that French went into this question, but my copy of French is not here at the moment. My copy of the Whitby just came last night, so I've had no chance to give it more than a cursory glance. The whole first volume seems to be commentary, of a very thorough variety. If only the typography were a shade clearer... I can understand why it was remaindered. Most of Dee's works, as I remember, were never printed, but existed only in manuscript, and Dee himself seems to have followed more the older, reserved style of scientific communication than the later open style. He was not, of course, the last, but he does seem to have been almost a survival even in his own time. The problem with the "rational/irrational" issue is that the lines are not drawn with respect to an unchanging criterion, but according to a matrix of criteria that alters over time. As with the controversy about "vitalism," in which the boundaries between the sides (and the items held to lie on one side or another of the lines) shifted several times over the course of a century or so. Dee was not the last scientist with national standing to scandalize his patrons by talking with spirits... --rjb From firth@SEI.CMU.EDU Fri Jan 31 05:58:16 1992 Date: Thu, 30 Jan 92 15:58:16 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201302058.AA13139@bp.sei.cmu.edu> To: voynich@rand.org Subject: Voynich Spaces (1) Status: OR Notes on the Voynich Manuscript - Part 12 ----------------------------------------- Concerning the Spaces in the Voynich Text Last night's wild speculation has become today's hypothesis, and it concerns the rules for inserting spaces into Voynich text. Since it is just remotely possible this hypothesis is correct, and if it is correct I don't want it named after me, I'm going to name it now: it is the Prosodic Hypothesis. In other words, the spaces do not separate words or morphemes; they separate sound clusters, syllables or feet, according to rules similar to those of metrical prose. Which hypothesis I submitted to two tests, described in this and the next note. As a working terminology, I'm going to use "words" for words, and "groups" for the things in the MS that are separated by spaces. Admonition Whether what follows is right or wrong, it has taught me a lesson I'd like to pass on. Most of us, and, I think, almost all previous would-be decipherers, have adopted, unconsciously, a modern, western approach to the MS: we have treated is as a visual object. We've been obsessed with symbols, characters, lines, words - in other words, the text as something to be read silently. We've burned our eyes over it. Et oculos habemus, sed non videmus. The mediaeval world was not like that. Their communication was much more aurally oriented; to them texts were to be read aloud. And the effective reading aloud of a text was a matter of much concern, hence the pointings, highlighting and accentuation of the text. If this MS is understandable text, it might very well have been read aloud; it might be a lectionary copy. And, just as I earlier thought the divisions might be the result of taking oral dictation, I now think they could equally well be pointings for oral recitation. Either way, I think we should focus a bit more on the text as sound, as vowel and syllable, phoneme and foot. Poetry Reconsidered In my facile dismissal of the idea that the MS is poetry, I was equally at fault. The notion of poem as text, lines neatly laid out, is, again, a modern notion. In a narrative poem on the classical model, you don't need line breaks - they can be reconstructed from the scansion. So each paragraph might indeed be a short stanza. Who, you might ask, would be so daft as to write an alchemical treatise in short verse stanzas? Well, look at 'Atalanta fugiens' by Michael Maier. The Experiment Well, on to the real stuff. The question I asked is: would a known prose text in a natural language, if divided into groups according to the rules I suspect were used in the Voynich MS, display some of the same statistical properties as that MS? Since I'd been thinking about greek most of the day, and since the MS vowel system looked pretty greek like, I took that as the language. I don't kave any mediaeval greek in my library, so went back to the koine of the NT. It seemed best to take a text "at random", not selected to show off the method; given that we'd been speculating about both human languages and the angelic language of Dr Dee, one text seemed especially appropriate. Here it is, set out as prose in standard romanisation, for all us visually oriented moderns: Ean tais glo:ssais to:n anthro:po:n kai to:n angelo:n, agape:n de me: ekho:, gegona khalkoi e:kho:n e: kymbalon alalazon. kan ekho: prophe:teian kai eido: ta mysteria panta kai pasan te:n gno:sin, kan ekho: pasan te:n pistin ho:ste ore: methistanein, agape:n de me: ekho:, outhen eimi. kan pso:miso: panta ta hyparkhonta mou, kai parado: to so:ma mou, hina kaukhe:so:mai, agape:n de me ekho:, ouden o:pheloumai. he: agape: makrothumei, khre:steuetai, he: agape: ou ze:loi, ou perpereuetai, ou physioustai, ouk askhe:monei, ou ze:tei ta eaute:s, ou paroxynetai, ou logizetai to kakon, ou khairei epi te: adikia, synkhairei de te: ale:theia; panta stegei, panta pisteuei, panta elpizei, panta hypomenei. I shall now transform this according to a certain set of rules. First, I'll write the long vowels as long: eh and oh for eta and omwga. Then, I'll break the text into syllable groups where each group has a main stress at the end, either a true stress or a long vowel or diphthong. Then, I'll fuse small unstressed words onto the following group. This would be a fantastic way to read greek, of course, but it seems to be the way Voynich text is grouped. Also, I'll kill the punctuation, and paragraph according to the traditional verses. And so, friends, here is the first sight of that long-lost mediaeval gnostic cathar text, the First Epistle of St Voynich to the Corinthians: Ean taisgloh ssaistohn anthroh pohn kaitohn angel ohnaga pehn demeh ekhoh gegon akhal koieh khohn ehkym balon alala zon Kan ekhoh propheh teian kaeidoh tamyste riapan takaipas antehn gnoh sinkan ekhoh pas antehn pistin hoh storeh methis tanein aga pehn demeh ekhoh outhen eimi Kanpsoh misoh pan tatahypar khontamou, kaipara dohtosoh mamou hinakau khehsoh maiaga pehn deme ekhoh oudenoh phelou mai Hehaga peh makrothumei khreh steue tai hehaga peh ouzeh loi ouperpereu etai ouphysiou stai Oukaskheh monei ouzeh tei teauteh souparoxy netai oulogi zetai tokakon Oukhai rei epiteh adikia synkhai rei deteh aleh theia Pan taste geipan tapisteu eipan telpi zeipan tahypomen ei It looks a bit familiar, doesn't it? The distribution of word length is less broad - short words have fused and long words have split. There seem to be letter correlations; the endings look grammatical but aren't, so one might be tempted to treat -oh, -eh, -ei as inflections. And I think you'd get the appearance of a lot of word repetition. Also, paragraphs would end with short groups - the odd syllables left over that can't fuse. Comments? Robert From firth@SEI.CMU.EDU Fri Jan 31 06:22:08 1992 Date: Thu, 30 Jan 92 16:22:08 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201302122.AA13245@bp.sei.cmu.edu> To: voynich@rand.org Subject: Voynich Spaces (2) Status: OR Notes on the Voynich Manuscript - Part 13 ----------------------------------------- Another Test of the Prosodic Hypothesis In an earlier note, I said that a good test of whether the Voynich groups were true words would be to find in the text many of the single words in the illustrations. Likewise, if the groups are not words, but can be derived from them, one should be able to predict how the single words would be transformed into groups, and find those. That is what I tried to do. This experiment assumes most of the letter values that Mr Guy and I have speculated on; in particular, it assumes 'a', 'e', 'o', 'u', 'n', and that 's' is probably a sibilant. It assumes that Voynich 'a' and '9' are equivalent. I also added an assumption that needs to be justified elsewhere: that the gallows letters usually mark the stressed syllable. Well, here are the rules, roughly: 1. The groups are based on metrical considerations, and every group tries to contain a stress or intonation 2. A break is always written after a long vowel, a nasalised vowel, or a long diphthong, unless this is the beginning of the group 3. A break is written after a stressed vowel if the following syllable can be detached and fused with the next group. In particular, this means that a sibilant (or liquid, I suspect) will act as a glide, and hold the syllable; a dental or plosive will force a break, as will a vowel that does not form a diphthong 4. An unstressed word (probably a particle) will fuse with the following group 5. Group medial 'a' is usually written with 'a'; group final 'a' with '9' Indeed, I suspect the latter to be just the former with a flourish, and suspect the (final) 'iv' and 'iiv' to be some other letter with a flourish For the test, I chose f82v (p 104 of Brumbaugh), the one with the nymphs. The marginal words, taken clockwise from due west, are as follows 1. olpaxct9 2. oqpcct89 3. oqpc89 (just to the left of the cute one) 4. 8aiivoqp9 5. oqpc8ax 6. oxlpcc89 (tiny letters above the east one) 7. oqpox89 8. olpox 9. olpc8a2 (above rainbows) 10. oqpc8ax (between rainbows) 11. oqpcc89 (below rainbows) 12. olpaolp9 13. oxlpox 14. olpaiv Now, these "words" might not be pure nouns. We might have words like "alembic", "alkahest", or even "topneuma" (which would break, by my rules, into "-topneu/ma-") Well, all I have to do, is take each name and (a) decide where it will break, (b) decide where it can fuse, (c) construct the pattern to be found in the text, and, of course, (d) find it. The following list is in the order in which I found them 1. olpaxct9 "ax" is a long diphthgong, "ct" is 't', so a break is forced. The left group can fuse to a particle; the right group must fuse, because it lacks a stress or a long vowel. The '9' will change to 'a'. We are looking for -olpax/cta?-, where the '?' implies that a letter must be found. Line 22, five lines below the picture and in the same paragraph, groups 3 and 4 are 4olpax ctax Score 1. 5. oqpc8ax This will not split, since there is no long vowel and the '8' is a sibilant. It can fuse to the left. So we expect -oqpc8ax/-, the break being forced. It's not there. However, what is there, in line 1, is oqp8ax. Now Voynich 'c' might be epsilon, the short 'e', so this looks suspicious. I think that vowel is the schwa, and that it is written in transcribed text but can be elided in dictated text. The sibilant would then join to the preceding consonant, and the stress would be thrown onto the next vowel. Is this plausible? Well, isn't that just how 'Caesar' turned into 'Tsar'? And, for the elision of unstressed vowels in metrical text, we need look no further than Vergil's famous "novus seclorum ordo". 10. oqpc8ax is the same as 5. 12. olpaolp9 Another crucial test, for, by my rules, this must split. One test, then, is to prove this word is not in the text (such are the paradoxa of Voynich). And it isn't. Indeed, no group on the page, except the very first, contains more than one gallows letter. Another brownie point for the hypothesis? What we would expect, clearly, is -olp9/olpa-. Now look at line 17, groups 4 and 5 4olp9 olpaiiv 6. oxlpcc89 This cannot split after the first diphthong. It can split after the "cc", but only if the split-off "89" can find a home to the right. So we expect either -oxlpcc89 or -oxlpcc/8a?-. We find, line 14 at the right: oxlpcc89 4olpcc89 olp9 That's our word. The stress probably falls on "cc", so "89" is marked by falling intonation. And the next group has an initial fused particle - the infamous "4" - which is probably associated with rising intonation. The break then comes at the right place. And note that the "89" of the next word can't detach, since it can't fuse with "olp9" - "ao" does not form a diphthong in this scheme. 3. oqpc89 Will not split. Found in line 2, third from right. There seems to be a half space in front of it, so maybe something fused on the left. 7. oqpox89 We expect either -oqpox89/- or -oqp0x/8a?-, and find (line 20, 3 from r) 4oqpax 89qpa2 This was the biggest surprise of the night. Not only did I find a medial "9", I could explain it! The scribe (I conjecture) was consciously following the prosodic rules, but he was also listening to the meaning. And, when a word ended in the middle of a group, he sometimes used the final form, by mistake. Just possibly, those 500 medial "9" symbols might be the clue to the reconstruction of the actual words. 8. olpox We expect -olpox/-, and find in line 4,group 7, 4olpox. Note also that the initial word of the page is qpolpox. 9. olpc8a2 This can split after the "a", if it's a long a and the next letter won't glide. Pure conjecture, of course, but we do find, in line 16 group 5 4olpc89 2c'tcc8a2 aivox9 11. oqpcc89 Again, this may split after the long "cc", so we expect either -oqpcc89/- or -oqpcc/8a?-. Line 4 has 4oqpcc89 14. olpaiv Cannot split, and cannot fuse to the right. We expect -olpaiv/-, and find, in line 24, group 7, 4olpaiv That waas done by eye, last night. This morning, I searched the computer form, and duly found 13. oxlpox Cannot split nor fuse to the right, so -oxlpox/-. And how could I have missed it: line 33, next to last line, group 6. 4. 8aiivoqp9 That must split after the "v", so we seek -8aiiv/oqpa- It doesn't occur on this page, but on f83r, line 22, appears 4oqpc8aiiv oqpaiiv The lonely nymph, No 2, oqpcct89, I can't find at all. If we elide the "c", then the group oqpct89 occurs a lot. That's a strange consonant cluster, though. It's as if we were to write ORETSA and read it ORTSA. That's 11 of the 13 different words, folks, by the rules, and all 13 if you allow me my elided 'c'. Maybe I spent most of yesterday evening fooling myself, but if not, that seems to be much better than anyone has ever done before. Karl Kluge just brought round some more folio copies, so it's back to the pictures again! Robert From RJB@U.WASHINGTON.EDU Fri Jan 31 06:55:00 1992 Date: Thu, 30 Jan 92 13:55 PST From: RJB@U.WASHINGTON.EDU Subject: The Sins of the Common Taters II (followup) Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I managed to find the time to wrestle the false bottom off the trunk where I store all my Dee materials. The _Book of Soyga_ (to my disappointment but not surprise) was not within... A quick scan through the things there provoked a horrible suspicion: my memory seemed to be exhibiting classic interference effects. Looking through French, Butler, Clulee, Walker and Yates and a few others (couldn't find anything right away from Paolo surname-evaporated) yielded only a commonly held assumption that everyone knew how important Dee's work had been, some references to not recapitulating what others had done, and so on. Could it be that I had just assumed that what everyone else assumed must be the case? Or that I had, over the years, lost track of what I had actually read and what others had referred to? I quick spin of the Lullian wheels in the University's Theater of Memory relieved my swooning heart, however, ane behold-- AU Dee, John, 1527-1608. Shumaker, Wayne. Heilbron, J.L. MT John Dee on astronomy : Propaedeumata aphoristica (1558 and 1568), Latin and English / edited and translated, with general notes, by Wayne Shumaker, with an introductory essay on Dee's mathematics and physics and his place in the scientific revolution by J.L. Heilbron. UT Propaedeumata aphoristica PI Berkeley, Ca. : University of California Press, 1978. LN PHYS/Physics-Astronomy General Stacks QB41 .D4 1978 I am also sure there has been a modern edition of Dee's "Preface" to Euclid. We don't have it however (though we do have the "Preface" on film). When I have a chance, I'll make a grief report on the above item (unless someone beats me to it). I probably have notes on it, but might as well go back and see the thing itself. --RJB From wet!naga@cca.ucsf.EDU Fri Jan 31 10:15:00 1992 Message-Id: Date: Thu, 30 Jan 92 17:15 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: 4-letter strings in VOYNICH.B Status: OR It will be recalled that yesterday I ran the repetition search program on VOYNICH.A and identified 1,290 4-letter strings which are repeated. The twenty most frequent of these were: Today I ran the program on VOYNICH.B, and identified 1,302 repetitions, of which the 100 most frequent are: Number of String Position of occurrences first occurrence 1991 C89/ 81 | 148 9/ZC 1172 1374 9/4O 236 | 147 E/ZC 2241 1068 /4OF 162 | 146 8AR/ 1160 904 89/4 252 | 144 89/E 4488 578 SC89 31 | 142 /OEF 4871 560 4OFC 267 | 140 OE/S 1216 536 CC89 207 | 139 OPC8 1400 475 4OFA 2037 | 137 4OPC 337 438 89/O 383 | 133 COE/ 580 393 ZC89 476 | 129 S89/ 2587 371 CC9/ 639 | 129 AM/S 2528 368 FC89 178 | 129 /OPA 1634 358 OFCC 959 | 123 OESC 122 329 C9/4 939 | 118 C89- 1582 321 FCC8 232 | 116 AN/S 4046 290 OFC8 146 | 113 OPCC 692 285 /SC8 73 | 112 R/ZC 4604 275 FAN/ 2329 | 112 OFAR 2452 269 /4OP 712 | 112 AM/Z 199 261 9/8A 219 | 111 OEFC 7680 260 /ZC8 750 | 110 AR/S 806 244 9/SC 150 | 109 /8AE 2476 236 89/S 1606 | 108 O89/ 627 231 9/OF 1254 | 108 /8AM 182 218 SC9/ 51 | 107 9-4O 228 215 AM/O 983 | 106 ZC9/ 9624 214 /OFA 1823 | 106 M/SC 887 212 9/OE 109 | 105 E/OE 4659 210 OFAN 4895 | 104 /ZCC 5114 205 PC89 158 | 104 /8AR 778 203 89/8 126 | 103 R/OE 2162 200 /OFC 410 | 102 AN/Z 4030 187 9/OP 588 | 101 R/SC 576 181 FCC9 1338 | 101 /SCC 1073 179 E/SC 309 | 99 E/4O 4438 174 /4OE 3742 | 97 C9/E 6812 172 FAM/ 4297 | 97 AR/Z 275 168 AR/O 1513 | 97 8AE/ 1618 165 AN/O 5016 | 95 /ESC 3235 164 OFAE 5542 | 93 /SC9 782 162 C9/O 2158 | 92 AE/S 1322 162 8AM/ 1026 | 90 4OPA 7574 162 /OE/ 3677 | 85 PCC8 7247 158 FAE/ 2229 | 85 /OR/ 546 158 /OPC 522 | 83 OE/Z 9320 155 OE/O 2643 | 83 FC9/ 839 153 FAR/ 1759 | 83 EFCC 7334 153 89/Z 762 | 81 C9/8 979 151 OFAM 2174 | 80 OR/O 1924 149 -4OF 1014 | 79 N/SC 3963 Obviously there is a difference between A and B, since in A the most common 4-letter string (counting / as a letter), 8AM/, occurs on average once every 88.7 bytes, whereas the most common 4-letter string in B, C89/, occurs on average once every 24.8 bytes. Directly comparing the 20 most frequently occurring 4-letter strings from A and B we have: VOYNICH.A VOYNICH.B Occurr- String Position of ences first occurrence 380 8AM/ 159 | 1991 C89/ 81 354 SOE/ 266 | 1374 9/4O 236 320 /8AM 859 | 1068 /4OF 162 310 SOR/ 497 | 904 89/4 252 242 OE/S 938 | 578 SC89 31 229 OR/S 1338 | 560 4OFC 267 227 /SOE 726 | 536 CC89 207 224 OE/8 766 | 475 4OFA 2037 208 9/8A 693 | 438 89/O 383 208 9/4O 2348 | 393 ZC89 476 185 9/SO 1172 | 371 CC9/ 639 159 E/SO 1274 | 368 FC89 178 159 E/8A 1119 | 358 OFCC 959 152 SC9/ 657 | 329 C9/4 939 147 CC9/ 847 | 321 FCC8 232 147 /4OF 2494 | 290 OFC8 146 145 /SOR 1761 | 285 /SC8 73 145 /4OP 1881 | 275 FAN/ 2329 142 O89/ 254 | 269 /4OP 712 134 AM/S 1506 | 261 9/8A 219 There are only five strings which are common to the top 20 in both A and B: String Occurs on average once every n bytes n for A n for B 9/8A 162 189 9/4O 162 36 CC9/ 229 133 /4OF 229 46 /4OP 232 183 Thus 9/4O occurs 4.5 times more frequently in B than in A, and /4OF occurs 5.0 times more frequently in B than in A. In contrast to A, the nymph ZOE is mentioned seldom, although always (in non-unique occurrences at least) immediately before or after a space: 28 ZOE/ 2744 22 /ZOE 4430 SOE is also mentioned much less frequently than in A, but again generally, but not always, immediately before or after a space: 34 SOE/ 1683 24 /SOE 4753 3 8SOE 680 3 -SOE 3510 2 FSOE 12184 The most frequently occurring letters in VOYNICH.B are: Rank letter (or space) Frequency 1 / 7977 2 C 5688 3 9 5308 4 O 5283 5 8 3990 6 A 3065 7 E 2839 8 F 2794 9 S 2206 10 4 1933 Since C89/ occurs 1991 times we can conclude that C89 occurs before 25% of the spaces. 9 is usually followed by a space, but there are many cases where it isn't: 118 C89- 1582 107 9-4O 228 52 89-4 6220 49 9/9F 65 48 9FCC 1156 48 /9FC 1347 37 89-8 47 36 9-8A 1240 32 89-2 195 31 /9FA 1342 29 89-P 371 28 OF9- 3455 28 9/9P 77 27 E89- 7869 26 9PC8 170 26 /89- 2388 24 89-B 560 21 OE9- 17409 21 9-2O 15695 20 AE9- 3312 20 -9ZC 1438 18 -9SC 2567 17 9-OE 1306 17 9-9F 402 16 9-BO 15306 16 -9FC 1639 15 AR9- 6385 15 89-9 11261 14 9PCC 6647 13 9FC8 4971 13 9FAR 2555 [130 more strings omitted] When 9 is not followed immediately by s space it is usually followed immediately by -, F or P, but may occasionally be followed immediately by any of C, Z, S, B, 8, V, E, A, 2, O or X, although 9A, 92, 9C, 9O and 9X occurr (at least non-uniquely) only once in VOYNICH.B. This is relevant to Jacques' hypothesis that spaces do not separate words but occur for orthographic reasons, when a letter "needs" some space before the next letter. If this is so in the case of 9, why is it that although 9 is *usually* followed immediately by a space, it *can* be followed immediately by any of about 2/3rds of the other letters? This string searching gives rise to much raw data. Given a fast computer or enough time I can turn out tons of this stuff. The question is how to make any sense of it, a question I am happy for the time being to leave to the linguists present. P.S. I removed the table of the 20 most frequent in B from after the first paragraph without correcting the reference to it therein. Oh well, the editor will fix it up. From jbaez@math.mit.edu Fri Jan 31 10:17:53 1992 Date: Thu, 30 Jan 92 20:17:53 EST From: jbaez@math.mit.edu Message-Id: <9201310117.AA01898@paley> To: voynich@rand.org Subject: Dee on Astronomy Status: OR My delete key is still crippled. Beware of ^H's. Before RJB posts his "grief report" on Dee's astronomy, as translated by Shumaker (same guy who wrote the exquizite "REnaissance Curiosa"), let me just say that it contains a good critical evaluation of Dee as a mathematician and scientist. The baisic point is that Dee was a good popularizer but held an inflated view of his own achievements. jb From jim@rand.org Fri Jan 31 12:13:11 1992 Message-Id: <9201310313.AA21073@mycroft.rand.org> To: voynich@rand.org Subject: Re: Zadzaczadlin? In-Reply-To: Your message of Thu, 30 Jan 92 14:14:57 -0500. <9201300314.AA01757@medici.trl.OZ.AU> From: Jim Gillogly Reply-To: jim@rand.org Date: Thu, 30 Jan 92 19:13:11 PST Sender: jim%mycroft@rand.org Status: OR > j.guy@trl.OZ.AU (Jacques Guy) writes: > > A D A M > zad zac zad lin > > Hmm, but the Enochian alphabet goes: ... > > Therefore, ye boke of Soyga is not writ in Enochian, and since we now, > from all those statistical counts, that the Voynich is not written in > Enochian, neither are. Therefore both are one and the same. Oh, Yes, all true, and yet, you reported earlier that the word "soyga" itself is Enochian (p. 170 of Laycock, "will of God(?)"). Interesting that only the title would be Enochian... or maybe ADAM is the only non-Enochian word in it?? Jim G. From jim%mycroft@rand.org Fri Jan 31 12:30:12 1992 Message-Id: <9201310330.AA21116@mycroft.rand.org> To: voynich@rand.org Subject: Soyga and Voynich Date: Thu, 30 Jan 92 19:30:12 PST From: Jim Gillogly Status: OR My Whitby came today... I get to join in the fun of plowing ancient furrows. Whitby addresses the possible identification of Soyga and Voynich on pp. 146-147, vol. 1: The book was first mentioned by Dee on 10 March 1582 when Uriel informed Dee that the 'book was revealed to Adam in Paradise by God's good angels' (fol 9a) and that the angel Michael could interpret the book which is later referred to by Dee as being 'written in tables and numbers' (fol. 89b). By 18 April 1583 Dee is unable to find the book and is told that 'a minister has it' and that 'it is nothing worth' containing 'false and illuding witchcrafts' (fol. 89b). the book evidently gives an alphabet which like Enochian gives names to the letters, for Dee says on 29 April 1583 that as far as he remembers 'Zadzaczadlin was Adam by the Alphabet thereof' (fol. 96a), zad being the letter 'a', zac being the letter 'd', and lin being the letter 'm'. I.R.F. Calder suggests that 'there is a possibility that "Soyga" might have been the famous Voynitch MS. which has defied all decipherers from Kircher onwards' (II, 481) but Ashmole noted that 'the Duke of Lauderdale hath a folio MS. which was Dr. Dee's with the words in the first page: Aldaraia sive Soyga vocor' ('I am called Aldaraia or Soyga'). The book was recovered by Dee on 19 November 1595 (see Diary) but its whereabouts now are not known. How much this book may have provided a source for the Enochian language cannot be determined, but the fact that the book is in tables and numbers and clearly uses a new alphabet suggests that it may have had some influence on Kelly, though clearly not enough for Dee to note any resemblance between his 'Arabic boke' and the subsequent angelic book. Based on this, I'm not wild about the identification of Soyga and Voynich: although Dee got it back and could have sold it to Rudolf in the right time frame (assuming that Lauderdale's book is something else), the fact that it's "written in tables and numbers" doesn't fit. There's little that I've seen that would qualify as a table, especially compared with some of the Enochian tables. Jim Gillogly From reeds@gauss.att.com Fri Jan 31 13:10:52 1992 Message-Id: <9201310411.AA16847@rand.org> From: reeds@gauss.att.com Date: Thu, 30 Jan 92 23:10:52 EST To: voynich@rand.org Subject: Firth 12 and Poetry Reconsidered Status: OR In a digression in Note 12, Firth states ``The notion of poem as text, lines laid out, is, again, a modern notion. In a narative poem on the classical model, you don't need line breaks - they can be reconstructed from the scansion.'' How modern? In the middle ages poetry WAS often written in lines, in the ``modern'' manner. (The few bookological books I have ready at hand show examples of Anglo Saxon verse with its two half lines, and numerous examples from about 1400 on of Latin and English verse written in the modern manner.) Are you maintaining that there was a time or place when poetry of certain kinds was customarily NOT written with laid out lines? (As opposed to the weaker claim that it need not be so written?) As for the rest of Notes 12 and 13: Bravo! Its about time that some sort of phonological (if that's the right word) regularity was found! Carry on! Jim Reeds From j.guy@trl.OZ.AU Sat Feb 01 01:08:38 1992 Date: Fri, 31 Jan 92 11:08:38 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201310008.AA02813@medici.trl.OZ.AU> To: voynich@rand.org Subject: Enochian Unmasked! Status: OR The Originne Of The Angelickalle Tonge As Revealed To Dr Dee Discover'd Through It's Alphabet By Dr Gee Voynichically stumped, I downed my ZOE quill for an Angelical one and sat down to design an Enochian font. As I hacked through the alphabet I kept thinking what an ugly font it was. My ears pricked at F, my suspicion was fully aroused by G; H confirmed it. I have just finished L, and I know the origin of the Enochian alphabet! It's amazing what a little "mettre la main a` la pa^te" does for your understanding. And Don Laycock who said that the Enochian letters looked like nothing on earth! It's Roman capitals, not too cleverly disguised. I'll start with the most obvious one: "G" (called Ged). Simply a capital G rotated 90 degrees counter-clockwise. "F" (called Or) is a capital F as is still often handwritten. The Angel (read "Kelley") just added a hook and serifs. "D" (Gal) is a capital D with the vertical stroke removed and two wee hooks added. "M" (Tal) is a rounded capital M rotated 90 degrees counter-clockwise. The general recipe is: rotate, or remove a stroke, or add a couple a hook and a serif or two, you may combine. "B" (Pa) is a capital B, rotated 30 degrees counterclockwise, and prised open at the top. "A" is capital A rotated 90 degrees, clockwise this time, its horizontal bar removed, two hooks added. "C", which does also for "K", *is* K, with just a stroke added. "H", which looks like lowercase omega upside down, well, try and write a capital H without lifting your quill! "I" (Gon) is I with a horizontal stroke and a serif added. "L" (Ur) is L smoothed out, plus two hooks. "N" is N, 90 degrees clockwise, written in a single, smoothing stroke, plus one hook. "O" are the two strokes which make up O (C and a mirror-image of C) each rotated a different amount. Actually, I think that Kelly invented the Enochian alphabet in the normal order: ABCDE... Adjacent letters tend to have been formed using the same distortions (rotation, hooks, serifs, whichever). He seemed to have got tired of the whole exercise too: as you move through our alphabet, the Enochian letters get simpler and simpler. When I was very young I "invented" such a "secret" alphabet for my own use. Only, it was based on lower case letters and removing bits and pieces, never adding any. "e", for instance, was an o on top of a horizontal stroke. Did I do any rotating? Probably not. All I remember is what triggered it. It was shortly after entering primary school (I never went to kindergarten) and learning the alphabet. On the way back from school, one day, my mother (she always came to pick me up at the school gates, on foot of course, mind you) stopped by at a haberdasher's, or so I think it was. Me, with my brand-new knowledge (I nearly did write "brand-knew") I stared at the boxes on the shelves, striving to decipher what was written on them. It so happens that nearly all the printed brand-names were in that 1930's style: an "A" would be a solid triangle of colour, an "O" a solid circle. I suppose you might call it my first exposure ever to cryptogaphy: a simple substitution code. I did work out that a triangle was (probably) an A, a circle (probably) an O, wondering all along if those serious-minded grown-ups would really do such things to their alphabet, or was it only my imagination? As soon as we got home, after doing my homework, I went on to design my "own alphabet". So, the descendent of this young kid is not too impressed with Kelley's fabrication. Now to finish that font if I can stay awake (*yawn*) long enough.... (Next morning) It seems Dee was a good mathematician, but a gullible old bird. So was, in modern times, Chasles. He was a collector of autograph letters and got taken in amazingly by a con-man whose name I forgot, who started flogging old Chasles letters from Josephine to Napoleon, then slowly went back in time... eventually producing the love correspondence between Antony and Cleopatra, and memos from Vercingetorix to Julius Caesar. The extraordinary thing was that all those authentic, autographed letters were... in French! For those of you who balk at coughing up $175 for Laycock's book, here is the first Enochian call: OL SONF VORS G i reign over you, GOHO IAD BALT says the god of justice, LANSH CALZ UONPHO in power exalted above the firmaments of wrath; SOBRA ZOL ROR I TA NAZPSAD in whose hand the sun is as a sword, GRAA TA MALPRG and the moon as a penetrating fire; DS HOLQ QAA NOTHOA ZIMZ Who measures your garments in the midst of my vestures, OD COMMAH TA NOBLOH ZIEN and trussed you together as the palms of my hands; SOBA THIL GNONP PRGE ALDI Whose seats i garnished with the fire of gathering; DS URBS OBOLEH G RSAM Who beautified your garments with admiration; CASARM OHORELA TABA PIR to Whom i made a law to govern the holy ones; DS ZONRENSG CAB ERM IADNAH Who delivered you a rod with the ark of knowledge. PILAH FARZM ZNURZA ADNA GONO IADPIL moreover, you lifted up your voices and swore obedience and faith to him that lives, DS HOMTOH and Who triumphs; SOBA IPAM Whose beginning is not, LU IPAMIS nor end cannot be; DS LOHOLO VEP ZOMD POAMAL who shines as a flame in the midst of your palace, OD BOGPA AAI TA PIAP PIAMOL OL VOOAN and reigns amongst you as the balance of righteousness and truth. ZACARE CA OD ZAMRAN move, therefore, and show yourselves; LAP ZIRDO NOCO MAD, for i am the servant of the same god as you, HOATH IAIDA the true worshipper of the highest. If you flinch at my lack of use of capitals in the English interlinear translation, here's why: ---------CUT HERE FOR UUENCODED ENOCHIAN FONT FOR PC'S----------------- begin 750 enochian.fnt M ?H&E@8&]F8&!?@ '[_V___P^?__WX M &S^_O[^?#@0 0.'S^?#@0 8/!CGY^<8 M&#P &#Q^__]^&!@\ !@\/!@ #________G MP\/G________ \9D)"9CP /______PYF]O9G#______\ !X. M&C)XS,S,S'@ \9F9F9CP8?A@8 /S,_,# P,'#PX M ']C?V-C8V-CY\X &!C;/.<\VQ@8 " P.#P^/[X\.# @ M @8.'C[^/AX.!@( 8/'X8&!A^/!@ 9F9F9F9F9@!F M '_;V]M[&QL;&QL #QF8# \9F9F/ P&9CP M_O[^_@ !@\?A@8&'X\&'X 8/'X8&!@8&!@8 &!@8 M&!@8&'X\& !@,_@P8 P8/Y@, M ,# P/X "1F_V8D ! X.'Q\_OX M #^_GQ\.#@0 8/#P\&!@8 !@8 M !F9F8D !L;/YL;&S^;&P &!A\QL# ? 8& M!L9\&!@ #"Q@P8,&#&A@ #AL;#AVW,S,S'8 # P,& M #!@P,# P,# 8# # 8# P,# P,&# M !F//\\9@ &!A^&!@ 8&!@P M /X 8& @8,&#!@P( M \9L/#V]O#PV8\ &#AX&!@8&!@8?@ 'S&!@P8,&#" MQOX !\Q@8&/ 8&!L9\ #!P\;,S^# P,'@ /[ P,#\ M!@8&QGP X8,# _,;&QL9\ _L8&!@P8,# P, 'S& MQL9\QL;&QGP !\QL;&?@8&!@QX 8& !@8 M &!@ 8&# !@P8,& P& P& ?@ ?@ M !@,!@,!@P8,& !\QL8,&!@8 !@8 'S&QM[>WMS M? ;.PX&$SN?#BX_,R [)XN;S,'C84 /#@Y.[OXV,S'@ 9MO;V\/#YV8 " M_,R,# P,#!D[?P !X,# P,# S,S'@ #F9F9L>'AL9F;F M #)SP<# P<=N/!P /G;BX.#D?" Q/Q\ ,#PV D-+3X,#,CP MP ># P,# P(T8,\ !PB(B(B%)7W]X$ ('WI20D)"0D M)&;_ /&9@('S,Q,# ['@8 @,#^# @(" @,#@X& ! 0$# M!HZ<^/#@ 0'Q&!@8.-F9F[P ,;&QL;&QL9L.! ## MP\/#P]O;?GXD !A_V-@8&!@8&!@\ ,/#PV8\&!@8&#P M (#^BXD)2WYH2 @(' /# P,# P,# P/ " P.!P.!P.!@( M \# P,# P,# P\ ! X;,8 M _P ,# 8 > Q\S,S,=@ .!@8'AL M9F9F9GP !\QL# P,9\ ' P,/&S,S,S,=@ M 'S&_L# QGP X;&!@\&!@8&#P =LS,S,S,? S,> M .!@8&QV9F9F9N8 8& X&!@8&!@\ !@8 #@8&!@8&!F9F M/ .!@8&9L>'AL9N8 X&!@8&!@8&!@\ YO_;V]O; MVP -QF9F9F9F8 !\QL;&QL9\ _&9F M9F9F?&!@\ ';,S,S,S'P,#!X #<=F9@8&#P M?,9@. S&? ! P,/PP,# P-AP #,S,S,S,QV M QL;&QFPX$ ,/#P]O;_V8 ##9CP8/&;# M QL;&QL;&?@;,> /[&#!@P9OX .&!@8 Q\S,S,=@ .&PX '@,?,S,S'8 M #QF8&!F/ 0&/ 0.&P ?,;^P,#&? ,8 'S&_L# QGP M & P& !\QO[ P,9\ 9@ .!@8&!@8/ $#AL #@8&!@8 M&#P & P& X&!@8&!@\ #& ! X;,;&_L;&Q@ X;#@ .&S& MQO[&QL8 #!@P /YB8'A@8&+^ [CL;?MC<=P #YL MS,S^S,S,S,X ! X; !\QL;&QL9\ Q@ ?,;&QL;&? M8# 8 'S&QL;&QGP #!XS #,S,S,S,QV !@,!@ S,S,S,S,=@ M ,8 ,;&QL;&QGX&#'@ ,8 ?,;&QL;&QL9\ #& ,;&QL;&QL;& M? 'S.WO;F? #AL9&#P8&!@8.;\ .FS.UM;6 MYFRX _&9F?&)F;V9F9N, X;&!@8?A@8&!@8V' 8,& M> Q\S,S,=@ #!@P #@8&!@8&#P !@P8 !\QL;&QL9\ 8 M,& S,S,S,S,=@ ';< -QF9F9F9F8 =MP QN;VWL[&QL;& M \;&P^ 'X .&QL. !\ P, P,&# QL9\ M /[ P,# #^!@8&!@ ,# PL;,&#!@ MSIL&#!\ # P,+&S!@P9LZ6/@8& !@8 !@8&#P\/!@ V M;-AL-@ V&PV;-@ 11!%$$4011!%$$4011!%$5:I5 MJE6J5:I5JE6J5:I5JMUWW7?==]UWW7?==]UWW7<8&!@8&!@8&!@8&!@8&!@8 M&!@8&!@8&/@8&!@8&!@8&!@8&!@8^!CX&!@8&!@8&!@V-C8V-C8V_C8V-C8V M-C8V /XV-C8V-C8V-@ ^!CX&!@8&!@8&!@V-C8V-O8&]C8V M-C8V-C8V-C8V-C8V-C8V-C8V-C8V-@ _@;V-C8V-C8V-C8V-C8V-O8& M_@ -C8V-C8V-OX !@8&!@8^!CX M ^!@8&!@8&!@8&!@8&!@8&!\ !@8&!@8&!C_ M _Q@8&!@8&!@8&!@8&!@8&!\8&!@8&!@8& #_ M 8&!@8&!@8_Q@8&!@8&!@8&!@8&!@?&!\8&!@8&!@8V-C8V-C8W-C8V M-C8V-C8V-C8V-C

To: voynich@rand.org Subject: Petersen and roadmaps Date: Fri, 31 Jan 92 08:23:27 PST From: Jim Gillogly Status: OR I got my copy of the sample pages from Jim Reeds' first generation copy of the Petersen transcription. I agree with Karl's observation that these second generation copies are inadequate for us to use for transcription because of the apparent shading or high-lighting of many of the words. If the first generation copy is clearer, I have no problem trusting Petersen's decisions on character assignment. However, I think it's useful as a roadmap of the V Ms: the non-obscured writing is clear -- a relief after trying to figure out Brumbaugh's penmanship (actually B's cipher alphabet is his encoding of V rather than an attempt to render it, but still!). He makes decisions about ambiguous letters on f3r; that's fine, and an additional check on any transcription we're doing from copies of the original. It's the only source of color information that I know of, short of going to the original. He also has marginal notes; all this is based on a lot more study of the original than any of us is likely to be able to spend. Most importantly, Petersen marks loci for pictures; this will save us the trouble of defining them ourselves in potentially different ways, and explaining them in the comments. We can just say "" and start blasting away. I'm ordering a copy. Jim Gillogly From reeds@gauss.att.com Sat Feb 01 01:39:25 1992 Message-Id: <9201311639.AA29182@rand.org> From: reeds@gauss.att.com Date: Fri, 31 Jan 92 11:39:25 EST To: voynich@rand.org Subject: Petersen Status: OR I sent sample Petersen pages, spanning the range of legibility, to five people, but have only heard back from Gillogly and Kluge. Their report is a modified negative, I think: you can't transcribe from Petersen but Petersen is not without excellence. (Karl mentioned the utility of the tentative plant identifications and Jim mentioned the color notations.) Can we have a general discussion of the quality of copies and the possible uses to which Petersen could be put, so I can count noses before I make my xeroxiteria run? To me, Petersen has two uses: 1. He has gone through numbering each root, star, nymph, etc. bearing any writing. This is of help in transcription: you just put down something like "nymph 13: OPCC89" and everyone knows which nymph you are talking about. There is no difficulty in establishing such a numbering, but a lot of virtue in having a complete numbering already done by someone else. 2. You can use it as an aid in transcription. Much IS perfectly legible, and a second reading is always a comfort. Let me explain my pricing: Suppose I get N orders from other people, and suppose I decide to make R copies for myself. Suppose the total xerox and mailing supply bill is X, and the postage for YOUR copy is P. Then you will owe me P + (43.05 + X)/(N+R) dollars. The xerox place might charge 3 or 4 cents a page, depending on how many copies I make, and Petersen has close to 300 pages, so this might come to $16 a person. I do not know how many copies, R, I will make for myself: at least one for day-to-day use, plus a few spare, I guess. Jim From firth@SEI.CMU.EDU Sat Feb 01 01:54:44 1992 Date: Fri, 31 Jan 92 11:54:44 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201311654.AA15154@bp.sei.cmu.edu> To: voynich@rand.org Subject: Brief notes Status: OR Folks Mr Guy points out that I have mis-spelled one of my nymphs: she is oqpox89 in the margin and I found -oqpAx/89- in the text. On checking, I see he's right. Sorry; it was late at night, and I hope she won't be too angry. Anyway, the word does occur twice in the B text, but the closest (4oqpox/8aiv) is in f78, which is not all that close. So my medial-9 conjecture, while still possible, has just lost its sole means of support. Given the huge number of o- names (and even oqp- names), it's tempting to speculate that this is the definite article. Maybe it works like arabic, where there's usually an initial al-, but sometimes it turns into at- or as-. So those names are all like algol, aldebaran, altair. In which case, the infamous 4 could mean "and", as in algol w'aldebaran w'altair giving us a string of oqp- 4oqp- 4oqp- ... But why is the "o" missing at the beginning of a paragraph? Could be a stylistic convention, but that explanation says nothing. Robert From reeds@gauss.att.com Sat Feb 01 02:30:09 1992 Message-Id: <9201311730.AA00615@rand.org> From: reeds@gauss.att.com Date: Fri, 31 Jan 92 12:30:09 EST To: voynich@rand.org Subject: Petersen Status: OR Jim Gillogly asked for my reading of line 41 of f76v from my copy of Petersen, to compare with his reading from a copy of my copy. Here it is: 2OE.ZC9.SC89.4OFC89.[SZ]C89.4OE.R.AM.ZC89# {wd 5 almost cert. SC89} Jim Reeds From RJB@U.WASHINGTON.EDU Sat Feb 01 02:36:00 1992 Date: Fri, 31 Jan 92 09:36 PST From: RJB@U.WASHINGTON.EDU Subject: Dee on Astronomy Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I have been preempted -- and I haven't even picked up the book yet. (I will, though, at noon.) I will file my "grief report." I will type more carefully. --rjb From jim@rand.org Sat Feb 01 02:37:18 1992 Message-Id: <9201311737.AA22421@mycroft.rand.org> To: voynich@rand.org Subject: Re: Petersen (transcription comparison of 1st and 2nd gen copy) In-Reply-To: Your message of Fri, 31 Jan 92 12:30:09 -0500. <9201311730.AA00615@rand.org> From: Jim Gillogly Reply-To: jim@rand.org Date: Fri, 31 Jan 92 09:37:18 PST Sender: jim%mycroft@rand.org Status: OR For reference, here's my transcription from the 2nd generation copy. This is representative of the worst lines in the sample; most are quite legible. Jim Gillogly Jim R: 2OE.ZC9.SC89.4OFC89.[SZ]C89.4OE.R.AM.ZC89# {wd 5 almost cert. SC89} Jim G: 2OE.ZC9.[*SZ]**[*9].4OFC89.***[9*].4OE.R.AM.[Z*][*C][*8]9# From RJB@U.WASHINGTON.EDU Sat Feb 01 03:12:00 1992 Date: Fri, 31 Jan 92 10:12 PST From: RJB@U.WASHINGTON.EDU Subject: Petersen Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I have received my copies of Petersen, and without providing the detailed comments they deserve, I will say that: a) the gray boxes (shading? highlighting?) around some of the text can't be seen through b) otherwise they are clear c) they are helpful because they present an interpretation of pages against which one can check one's own interpretation d) I would order copies (if this is till going to happen) because they will be good to have when trying to work from film --RJB From j.guy@trl.OZ.AU Sat Feb 01 04:35:44 1992 Date: Fri, 31 Jan 92 14:35:44 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201310335.AA03077@medici.trl.OZ.AU> To: voynich@rand.org Subject: Comments on R.Firth's notes 12 and 13 Status: OR Glory be! Mr Firth has vomited from his mouth the Heinous Heresy, embraced the True Faith, and is diligently adding many a stone to the Edifice of the True Doctrine! Serious comments now. >1. The groups are based on metrical considerations, and every group > tries to contain a stress or intonation I've come to think along those lines too. Either that, or they are based on calligraphic considerations. In the latter case, there is evidence that certain sounds are represented by two different letters (e.g. a and 9), and the choice of which letter ought to be use may well be made on metrical (stress, intonation) considerations. So, ultimately, metrical considerations hold best. >2. A break is always written after a long vowel, a nasalised vowel, > or a long diphthong, unless this is the beginning of the group. Not sure about the "unless...". A space, for instance, is always written after and (iv and iiv). I have seen what might be a non-final form of or , viz: 00304A *AII89.SPO8.89.W9.8AE.2.SOFAM.8 ^^ 00602A 8AII89.4OPSOE.OFSOR.OFOR.OE9POE.8OE.8AR ^^ 01109A 8AR.ZCOE.2FAIIO8AR.OPAM.SOR9 ^^ 03207A 2OR.SFCC9.BOII2.SCOR.O2.2.AM ^^ 03706A SOII2.OFSOR.4OPOE.SCC9.8AM.S9.SOM.8AM ^^ 03801A VAII2.AR.OFO9.Z9.BOVOSC9.OBS9.4OB9.SOE89.OB989.W9 ^^ 04306A PZOE.9FOR.4OFAM.9F9.8AR.OFOE.8SC9.8AII8AE.8AJ.9PSO.E8AE2 ^^ 04401A B**TOE.O8AIIIE9.ZO.AE.89.OBSOE.OPOE.OE.SW9.4OPSAR.2 ^^^ 04412A OE.AIIOR.OE.ORO.CCCOE9# ^^ 04803A ORSO.FSOR.SOE.8AM.ZYOR.8AM.8ZC9.8AIIP9 ^^ 04807A O*.SFC9.8AII.OE.8AM.ZX.ORSAM# ^^ 05916B OPCOR.AIIQ9# ^^ 06101A VSAM.Z9FCO89.8AIIO89.8AN-ZO.PS9.OP9.4O89 ^^ 06201A FSCO8AM.SOEFCS9.4OPAM.8AIIOAJ-OSOVSO89 ^^ Misprints? Perhaps, but I have seen at least one <8aii8> in some reproduction, so some at least are not misprints. I haven't thought what that might mean. >3. A break is written after a stressed vowel if the following syllable > can be detached and fused with the next group. In particular, this > means that a sibilant (or liquid, I suspect) will act as a glide, > and hold the syllable; a dental or plosive will force a break, as > will a vowel that does not form a diphthong. All very plausible, phonologically. Which does not mean that that is the case of course, only that it is likely. >4. An unstressed word (probably a particle) will fuse with the following > group. Very probable too. Remember those "in"'s glued to the next word in that excerpt from an 11th-century lectionary I posted recently: "ineum" instead of "in eum". >5. Group medial 'a' is usually written with 'a'; group final 'a' with '9' > Indeed, I suspect the latter to be just the former with a flourish, > and suspect the (final) 'iv' and 'iiv' to be some other letter with > a flourish. I have seen what might be a non-final form of or , i.e. and in Currier's notation (they are nearly all in the A corpus, by the way): 00304A *AII89.SPO8.89.W9.8AE.2.SOFAM.8 ^^ 00602A 8AII89.4OPSOE.OFSOR.OFOR.OE9POE.8OE.8AR ^^ 01109A 8AR.ZCOE.2FAIIO8AR.OPAM.SOR9 ^^ 03207A 2OR.SFCC9.BOII2.SCOR.O2.2.AM ^^ 03706A SOII2.OFSOR.4OPOE.SCC9.8AM.S9.SOM.8AM ^^ 03801A VAII2.AR.OFO9.Z9.BOVOSC9.OBS9.4OB9.SOE89.OB989.W9 ^^ 04306A PZOE.9FOR.4OFAM.9F9.8AR.OFOE.8SC9.8AII8AE.8AJ.9PSO.E8AE2 ^^ 04401A B**TOE.O8AIIIE9.ZO.AE.89.OBSOE.OPOE.OE.SW9.4OPSAR.2 ^^^ 04412A OE.AIIOR.OE.ORO.CCCOE9# ^^ 04803A ORSO.FSOR.SOE.8AM.ZYOR.8AM.8ZC9.8AIIP9 ^^ 04807A O*.SFC9.8AII.OE.8AM.ZX.ORSAM# ^^ 05916B OPCOR.AIIQ9# ^^ 06101A VSAM.Z9FCO89.8AIIO89.8AN-ZO.PS9.OP9.4O89 ^^ 06201A FSCO8AM.SOEFCS9.4OPAM.8AIIOAJ-OSOVSO89 ^^ Misprints? Perhaps, but I have seen at least one <8aii8> in some reproduction, so some at least are not misprints. What worries me is that and ( and ) are so frequent, and those and so rare. I haven't thought what that might mean. >For the test, I chose f82v (p 104 of Brumbaugh), the one with the nymphs. >The marginal words, taken clockwise from due west, are as follows >1. olpaxct9 >2. oqpcct89 >3. oqpc89 (just to the left of the cute one) >4. 8aiivoqp9> >5. oqpc8ax >6. oxlpcc89 (tiny letters above the east one) >7. oqpox89 >8. olpox >9. olpc8a2 (above rainbows) >10. oqpc8ax (between rainbows) >11. oqpcc89 (below rainbows) >12. olpaolp9 >13. oxlpox >14. olpaiv When I used to pore over my fuzzy white-on-black photographic enlargements from the Yale microfilm, looking for captions repeated in the text, I noticed the extraordinary frequency of initial . Perhaps we would be wise to consider it as, say, an article, and to hunt for the captions in the text stripped of their o's. (Those nymphs are all Japanese, very obviously: o-haru, o-rin, o-tomi; as for the so frequent <8aiv> it is quite as blindingly obviously the honorific -san. The Voynich is in Japanese. Tah dah!) 1. olpaxct9: I don't see 4olpax ctax on line 22 (but I am working from D'Imperio's transcription). Closest is: olpoxctox and olpaxc'tc89 >6. oxlpcc89 This cannot split after the first diphthong. It can >split after the "cc", but only if the split-off "89" can find a home >to the right. So we expect either -oxlpcc89 or -oxlpcc/8a?-. We >find, line 14 at the right: > oxlpcc89 4olpcc89 olp9 >That's our word. The stress probably falls on "cc", so "89" is marked >by falling intonation. And the next group has an initial fused particle >- the infamous "4" - which is probably associated with rising intonation. >The break then comes at the right place. Could be, could be. Rising intonation? Could be, too. But couldn't it also be the sign for the *falling* intonation of the preceding letter? <4> occurs almost always followed by , preceded by <9>, , or . Strange distribution. The infamous <4> indeed! >7. oqpox89 >We expect either -oqpox89/- or -oqp0x/8a?-, and find (line 20, 3 from r) > 4oqpax 89qpa2 >This was the biggest surprise of the night. Not only did I find a medial >"9", I could explain it! The match is not perfect where we're looking for , but, since I have kept harping about extensive sandhi, I' d be dishonest to say "no match", wouldn't I now? The explanation is quite plausible, yes indeed. So, the task at hand is definitely the segmentation of the text into its true constituent words, or morphemes. Seems that we've only reached one sure conclusion so far: it's not a cypher. From firth@SEI.CMU.EDU Sat Feb 01 05:11:31 1992 Date: Fri, 31 Jan 92 15:11:31 -0500 From: firth@SEI.CMU.EDU Message-Id: <9201312011.AA15481@bp.sei.cmu.edu> To: voynich@rand.org Subject: Speculations on oqp Status: OR Notes on the Voynich Manuscript - Part 14 ----------------------------------------- A Cursory analysis of the opq- Phenomenon This analysis was done on one folio - f82v, the nymphs again. If I have time, I'll write a program to repeat it on the whole text. But the results are interesting. What I did, was do a statistical count of the pattern "o" The following table gives the counts of the major forms: olp 71 64% oqp 25 23% oxlp 11 10% oxqp 1 ol; 0 oxl; 1 ocqpc 1 In addition, a gallows letter appeared without a prefix 8 times, and with some other prefix 16 times, the only one worth noting being "aqp", which occurred 4 times. So the above represents 82% of the occurrences of the 8 letters. (As a digression, the page has 8 other occurrences of the infamous 4. Three of them are "4oxct" and the others are 4c't 4oxa 4oxc't 4o8 4l; ) If "olp" is the definite article, are these numbers reasonable? Well, in Islam there are 99 names of God, 97 of which begin with a definite article, thus: Al 55 Al- 24 Ar- 4 Ar 3 As- 3 An- 2 As 1 Ash- 1 Ash 1 'Al 1 Ad- 1 Az 1 Hmmm... interesting. ----- I did another experiment with the text. I made some wild guess letter substitutions, just to get it largely pronounceable. I then removed the word breaks (but not the line breaks). I then added word breaks after a "medial" 9 and before every occurrence of "4". This is the result. As before, majuscules and digits are Currier Voynich; the rest is guesswork. (Hmm, the last 5 lines seem to be missing, but I don't think I need chase them up, the example is big enough.) PoFououVtesa 4oFysa 4oFesauthou 4oPauoPsausououthesa 4oFesauthesa 4oPousoutesathesasasanoPesatePesauoFaJ saToutesa 4oPesa 4oPesatesa 4oPau 4oPa 4oFau 4oFesauo 4oFaNtheou 4oPysatesa 4oFea 4oFesa 4oFoutesatesauta 2outesautesasouthesa 4oFautheXatheaPyouoPysa 4oFesa 4oPea 4oFautheFa 4oFautesa 4oPautesa Bteousan 4oFyatyFa 4oFausautesaBtsanou 4oPesanou sauthousanou 4oFa 4outesa 4oFan 4oPeaPa 4oFatQa 4oFa Ptesa 4oQe2 4oFantXa 4oFaNtouFaitXsa 4oFousa soutesauteanounoutya2aNtea PantheXa 4oFau 4oFaNteouPesantesaBtesantheousau 4oFesauteatyanaithe2 4oFtea 4oFaNoFausautesL onaNthesa 4oFyanoFan 4oFysantya 4oPesatPesante2aua a Fysa 4oFysatesa 4oFansFyauFesa 4oFauouFysa 4oFysaoFa 4oFoutesateautesaoPatsasateXaoPautu2 Poutheou 4oFesa 4ouFesanthesa 4oFesanthysanoNoua saNtea 4oFyea 4oFaoFaioFaNouFaNoPaNoFanaua 4oFaNtesa 4ouFthea 4oPauthea 4oFaNouteXaua 4thesa 4outea 4ouaioPaNouFysa 4oPthesaouu stesa 4oFaioutesa 4oFautesa 4oPausa Pana uJu 4oFtheouFetea 4oFanauPtsanouthea Ptsa 4oFau 2aithea 4oFoutou 4outhysa 4oFauthesa 4oFai 4oFaNtheau 4outesa 4oFaUouFaithesa PtauouFaTthyFa 4outea 4oFantea 4oFaNoPthesa saitea 4oFathesa 4oFaiouFysa 4oFetesa 4oPU Bououontea 4oFaNthesa 4oFeisutenaounaNsan 4oFyanaNtea 4ouFaNoFaoPaua 4oFauthouthau2a 2anouuonaitea 4oFaNono3 4osathanaQea 4oFteaFteaouFaNoBtesa 4VouthPaonau This is beginning to look like words run together. In fact, I could almost have a shot at breaking them up. Here's the first two lines again: PoFou ouVtesa 4oFysa 4oFes authou 4oPau oPsausou outhesa 4oFes authesa 4oPous outesa thesasasan oPesate Pesau oFaJ Probably wrong, since I'm obviously looking for cyno-greek here. But hey, doesn't it read like language? Real language? Robert From j.guy@trl.OZ.AU Sat Feb 01 05:30:14 1992 Date: Fri, 31 Jan 92 15:30:14 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9201310430.AA03142@medici.trl.OZ.AU> To: voynich@rand.org Subject: Adam & Zadzaczadlin Status: OR I can't find (which Jim's was it? Gillogly's) first interpretation of Adam as Zadzaczadlin in my mess of windows, saved mail under a mess of names, general paraphernaliac disaster. So I don't know if what I've just noticed is new. Dee wrote: As E K and I wer talking of my [boke] boke Soyga, or Aldaraia: and I at length sayd that, (as far, as I did remember) Zadzaczadlin, was Adam by the Alphabet thereof. An old common name for "Z" is "zod". Suppose "zad" is "Z": A D A M z (zac) z (lin) A shifted cipher, one to the left, with transparent cryptic names for its letters? A D A M Z C Z L So: Z = "zad", C = "zac", L = "lin" ^ ^ C L The Angel said: "Soyga signifieth not Agyos. *Soyga alca miketh.*" What language is that? "Alca miketh" must mean "signifieth not holy". Oh, silly me! The Angel must be Kelley's ectoplasmic fabrication. So the language is Enochian. Seems that the boke of Soyga is a cryptographic grimoire. Laycock was of the opinion that "angels", "spirits" etc. were in fact technical words referring to nothing supernatural, but to the cryptographic theory of the times. Perhaps that particular Angel is not of Kelley's fabrication, but Dee's, and the language is not Enochian. Soyga alca miketh = "Soyga" is not "holy". "Ket" in Breton is "not", "is not" is "n'eo ket". Miles away. DEE: But it was reported to me by this skryer that he had: certain peculier bokes pertayning to Soyga. otherwise named ysoga, and Agyos, literis transpositis. "pertayning" to Soyga; on the topic of "Soyga"? ... "otherwise named ysoga, and Agyos, literis transpositis"? A book on anagramming perhaps? "Aldaraia" does sound Arabic. Al-daraya or Al-darayya. Soyga is definitely not Arabic. The Roman letters look nothing like any Arabic letter combination, and the sound is not Arabic. Spanish it could be. A treatise of cryptography I'd say. From rcarter@isis.cs.du.edu Sat Feb 01 12:51:41 1992 Date: Fri, 31 Jan 92 20:51:41 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202010351.AA07541@isis.cs.du.edu> To: firth@sei.cmu.edu, voynich@rand.org Subject: Re: Brief notes Status: OR Re: various Latin abb. for oqp- o can mean non, om, os, op... q can mean qui, qua, quo, quae... p can mean per, par, por, pra, pre, pro... So (random selection) oqp in Lat. abb. could be omquiper- Also: 4 is what is refered to by Cappelli as a "missing abb." in that it means `rum' is missing -somewhere- in the abb. word. 4 can also be used to signify `quatro' (naturally...). Just noticed that D'Imperio/Cappelli states that p can also be -is- so random example could be omquis- etc. From reeds@gauss.att.com Sat Feb 01 13:42:34 1992 Message-Id: <9202010442.AA12740@rand.org> From: reeds@gauss.att.com Date: Fri, 31 Jan 92 23:42:34 EST To: voynich@rand.org Subject: Off the Subject Status: OR I just saw an interesting paper today, ``The Perception of Randomness'' by Maya Bar-Hillel and Willem Wagenaar, in Advances in Applied Mathematics, 12, pp428-454 (1991). It is about (among other things) ``psychologically random'' digit strings produced people trying to produce random digit strings (off their heads, not with apparatus or algorithms). From reeds@gauss.att.com Sat Feb 01 14:03:22 1992 Message-Id: <9202010503.AA12841@rand.org> From: reeds@gauss.att.com Date: Sat, 1 Feb 92 00:03:22 EST To: voynich@rand.org Subject: V-like stuff in Cappelli Status: OR While browsing in my Cappelli (Dizionario di abbreviature latine et italiane... by Adriano Cappelli, Editore Ulrico Hoepli Milano, Ristampa 1967, Copyright 1967 by Ulrico Hoepli (via Hoepli 5), Milan) I saw a plate with a lot of upright letters reminiscent of the tall Voynich letters. To wit, Tavola IV, captioned ``1172 Giugno 13. -- Savino abbate dei monastero di S. Savino in Piacenza investe il mugnaio Gerardo Albarola per se` e suoi eredi maschi in perpetuo, di un mulino di ragione del detto monastero. -- Scrittura carolina. -- Pergamena origin., conservata nell'Archivio di Stato di Parma, monastero di S. Savino.'' Most of the letters are dowdy little uncials, except on the first and last lines just about each word has a letter like Currier's B, only more elaborate. Thus: the first line reads ie marti tertio ecimo menis iunii n Pacentia inra monaerium a Savini in oario par where I have put <> brackets about each elaborate letter. All the elaborate letters look about the same, except the initial D looks like a big D. From reeds@gauss.att.com Sat Feb 01 14:27:04 1992 Message-Id: <9202010527.AA12894@rand.org> From: reeds@gauss.att.com Date: Sat, 1 Feb 92 00:27:04 EST To: voynich@rand.org Subject: New letter Status: OR There is a letter that occurs fairly often in word rings, and occasionally in the text, which is not in any of our transcription alphabets. It looks like this: ################## ## ## ## ## ## ## ## # # # # # # # # # # # # From Karl.Kluge@G.GP.CS.CMU.EDU Sat Feb 01 21:06:00 1992 Date: Sat, 1 Feb 1992 21:06-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: A tantalizing clue and some patterns Message-Id: <696996407/kck@G.GP.CS.CMU.EDU> Status: OR Folio 67 is Newbold's "annular eclipse" folio, and can be seen in the June 1921 SCIENTIFIC AMERICAN article. Folio 70 is a zodiac folio showing Picses and Aries, and was used by Jim Reeds in his Brumbaugh post. First, one observation. Nymph 4 on f70v3 has the label OPAEAJ. The middle root of the three in the second row of roots at the top of f99r (Krauss plate XXXI, bottom) also has that label. Unfortunately, grepping the D'Imperio transcription doesn't yield any exact matches, although we have two line-final instances of OPAE/8AJ: 00506A 9SPAM/SOR/QOJ/OPAE/8AJ- 07904B S8CC9/9PSCCF9/9BSC89/2S89-FC89/QC9/CCCC2/OPCC9/OPAE/8AJ- Still, the match between a label on a nymph in a zodiac folio and a root in the pharmaceutical section should raise some eyebrows. Searching for more matches of this type may yield a break into the text (although I was very disappointed not to find OPAEAJ directly in the herbal text from the D'Imperio transcription). First, an odd bit of alliteration on f70v3 in ring III, where we have "...OPCC8AR OPCC9 PCC9 OPCC89..." Note that OPCCO89 is also found in ring IV between and above nymphs 14 and 15. Patterns in nymph labels f70v3: II.1: OPAR AJ also appears under IV.12 in ring III II.6: OFARAJ II.2: OPAR AE II.3: OPAEAR II.4: OPAEAJ IV.26: OPAE.89 II.9: OFAE8AE On f70v4 we have nymphs 10, 15, and 6 who have labels starting with OPAE. One part of f67v shows a "happy sun" with an outer ring which has 12 divisions. Division 3: OFAE AR9 f70v3 #9: OFAE 8AE Division 5: OFOE AE89 f70v4 #1: OFOE 9 The other part of f67v (Newbold's "annular eclipse") has a large circle containing 8 radial lines of text (heading in from the NE, SE, NW, and SW, heading out to the N, S, E, W from a central star design). Outside the outer cirle to the NE, SE, NW, and SW are four smaller circles. The ones to the NE and SW have four faces (connected in a cross in the NE, and an open square in the SW). The ones to the SE and NW have three faces (closed triangle NW, open triangle SE). The label on the NE circle is OPARARAN (there's our old friend OPAR again...). Why the heavy repetition of word "roots" such as OPAR, OPAE, OFAE in the zodiac folio nymph labels? Is OPAEAJ a plant name, with f99r using the plant's root in a recipe, f70v3 giving its macrocosmic correspondence to Picses, and 3r or 41r giving a full page description of the plant? From reeds@gauss.att.com Sun Feb 02 10:43:50 1992 Message-Id: <9202020205.AA01256@rand.org> From: reeds@gauss.att.com Date: Sat, 1 Feb 92 20:43:50 EST To: voynich@rand.org Subject: Oddity on f58r Status: OR Browsing though Petersen I found something odd. On f58r there is space for a rubricated letter, three lines tall. This is one of the misplaced ``recipe'' leaves, with stars. Since the first 50 or so ff seem to be art first, then writing, we have a change of plan here. May I suggest that at some stage in the MS production the writing outstripped the artwork, that the stars are a form of ersatz or el cheapo quick & dirty decoration, and that the whole MS was finished more hastilly than originally planned? This might be consistent with the more cramped, hurried appearance of the recipe sections in general? From Karl.Kluge@G.GP.CS.CMU.EDU Sun Feb 02 04:05:00 1992 Date: Sun, 2 Feb 1992 04:05-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Another couple matched captions, and does P=F? Message-Id: <697021532/kck@G.GP.CS.CMU.EDU> Status: OR f70v4 nymph 13 and one of the nymphs at the top of f84? shown in the PHILOLOGICAL QUARTERLY article: OFOEZ9. Closest matches in D'Imperio: 04101A BOE/OEZ9/VSOE9/ZOE/8BS9/OP9/OFOE9/8AM/OBS9/2/OW9- 04309A OFOE/OF/Z9/4OFOE89/8AE/8ZO/4OFCCC2/9/OE9/8AM/8AE- 07703B Z9/4OFAE/S89/SX8-OPOR/AU/OF9/OFOES9/4OFAR/OFAJ- (n.b. -- OFOE9, line 04101A matches f70v4 nymph 1) f67v line 1 and another nymph on f84?: OPOE9. Some matches on this one! 04213A OFSAN/SFOE89/ZOPOE9- 07105A 9F9/4OFS9/43PSOR/SFOE/OPOE9- 09408B OPCO89/SFC9/OFC89/SXC89/9FC89/OE89-OPOE9/SC9/FAE9/POFAR/OPAJ- I wonder if OPOE9 and OFOE9 are sisters? Could F = P? f70v3 nymph 2 is , one of the labels in 78r is OFARAE. Here are matches of OFAR/AE from D'Imperio: 14915B 8ZC89/4OSC89/ESC89/4OFC9/4OESC9/4OPAN/SX9/ZX9/ESAR/OFAR/AES89- 16217B 8AN/SC9/4OFCCC9/4OF9/OFAM/OFAN/OEFAN/OPAN/OFAR/AE9- Here are matches of OPAR/AE from D'Imperio: 09404B 8SC89/PS889/OPZ/OFCC9/P9/OPAR/AES89-9PCC89/OPCC8/9FC89/J- 14714B 2AN/OE/F*Z9/4OFAN89/OEZC89/4OFAN/SCFC9/4OFAN/OPAR/AE9- Note the final OPAR/AE9 in line 14714B and the final OFAR/AE9 in line 16217B; also, the OPAR/AES89 in 09409B and the OFAR/AES89 in 14915B. There were no matches for nymph OFOEZ9 in the text, but OPOE/Z9 does occur: 15823B BOEPCZOE/OB9/ZC9/4OBZC89/4OPCC89/4OPOE/Z9/FOEZ8/4OF9- Here's another example of the odd ...CC8, ...CC9, ...CC89 words: 16214B 9FCC89/4OFCC89/SC89/4OFAR/8FCC9/EFC89/4OFAE/OEFCC89/4OFCC89/OF9- Grepping on CC9, folios in the page 160-166 range are chock full of such words. Sometimes several to a line. Odd. I hope the way the caption words appear broken/prefixed/suffixed in the text helps settle the word boundary issue. Also, if we can get copies of all the zodiac and pharmaceutical folios spread around hopefully we can find more matches (there may be other matches in the babes in tubs folios, but such matches would be less useful than plant label matches, particularly if we can find those labels in the herbal folios). Progress? Karl From rcarter@isis.cs.du.edu Mon Feb 03 08:21:24 1992 Date: Sun, 2 Feb 92 16:21:24 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202022321.AA01673@isis.cs.du.edu> To: Karl.Kluge@g.gp.cs.cmu.edu, voynich@rand.org Subject: Re: Another couple matched captions, and does P=F? Status: OR Also... I would like to see a freq. count done (volunteers?) on occurences of OPAR/OPAE/OFAE comparing Lan.A and Lan.B... I am thinking that F was wrote down instead of P at times, and that R and E are used to mean the same thing at times. From reeds@gauss.att.com Mon Feb 03 09:39:54 1992 Message-Id: <9202030040.AA14032@rand.org> From: reeds@gauss.att.com Date: Sun, 2 Feb 92 19:39:54 EST To: voynich@rand.org Status: OR > From: rcarter@isis.cs.du.edu (Ron Carter) > > Also... I would like to see a freq. count done (volunteers?) > on occurences of OPAR/OPAE/OFAE comparing Lan.A and Lan.B... Working with orginal D'Imperio transcription, A corpus, obeying word breaks 5 OPAR 3 OFAR 12 OPAE 16 OFAE A corpus, ignoring word breaks 13 OPAR 8 OFAR 17 OPAE 21 OFAE B corpus, obeying word breaks 38 OPAR 44 OFAR 30 OPAE 39 OFAE B corpus, ignoring word breaks 64 OPAR 123 OFAR 76 OPAE 190 OFAE From reeds@gauss.att.com Tue Feb 04 00:04:02 1992 Message-Id: <9202031504.AA23104@rand.org> From: reeds@gauss.att.com Date: Mon, 3 Feb 92 10:04:02 EST To: voynich@rand.org Subject: Last chance for Petersen copies! Status: OR I go to the copy shop sometime this week to make copies of Petersen's transcription. If you want one (estimated cost about $16) get in touch with me A.S.A.P. by email, and be sure to send your postal address if I don't already have it. (Once the copies are in the mail I will tell you exactly what you owe.) Jim Reeds reeds@research.att.com 908 582 7066 From firth@SEI.CMU.EDU Tue Feb 04 01:31:48 1992 Date: Mon, 3 Feb 92 11:31:48 -0500 From: firth@SEI.CMU.EDU Message-Id: <9202031631.AA21243@bp.sei.cmu.edu> To: voynich@rand.org Subject: Now I even dream about the MS! Status: OR Notes on the Voynich Manuscript - Part 15 ----------------------------------------- An Account of a Curious Dream concerning the Voynich Manuscript Well, if you were not already persuaded that trying to decipher the Voynich manuscript was the high road to gibbering lunacy, this note should convince you. If you don't like dreams, or don't believe they convey meaning, stop now. For I intend to present the dream, an interpretation, and the evidence that it is a genuine communication from the objective psyche. Shortly before dawn, on 1992 February 2: The Dream I was in a grotto, in the basement of a building. The room was quite large, and had been landscaped with stones, shallow pools, and running water. There was a wooden walkway around part of the room, made of narrow boards with gaps between. I was looking for a ring. It was a gold ring, but had been broken into pieces and lost. I was looking for it under water, in the pools and streams, kneeling down and winnowing the water with my fingers. I found one piece, felt it and looked at it, but lost it again. It then occurred to me to talk about this with a woman I knew, who lived in a room on the upper floor of the building, which I now realised was three storeys high. So I got up, dressed, and walked towards the stairway. Thre was a doorway in the wall, the lower edge about four feet off the ground, through which the stair ran. But the stair was retracted. I turned a small lever on the wall, and the machinery began to lower the stair down through the doorway and towards me. There was a man sitting on the stair, blocking my path, and as the stair came to rest he was sitting just in the doorway. At his feet, two small metal containers were fixed to the stair, one either side, about six inches tall, shaped like the frustrum of a cone, and each with three round holes in the top. They were salt and pepper shakers. The man was holding a mixture of salt and pepper in his hands, and letting it sift through them. It fell down, closed upon itself, and sifted back through the holes and into the shakers. The man was dressed in a very conservative dark suit, like a butler in the old movies. I knew the words that would get me by this entity, and climbed the stair. Halfway up, there was a door in the wall that had never been there before (I was familiar with the building); the door was open and behind it was what seemed to be a workroom, brightly lit. As I passed it, another entity came out and followed me. I got to the ground floor and closed the door behind me. A mistake: the entity came forward, helped me close the door by pulling from the other side, closed the door through its body, and was now in the room with me. It was a man, dressed in light brown, workmen's clothes, with a cloth cap. He was carrying a vacuum cleaner, and when I asked "What's that for" he replied "The power's gone off" And I understood that he intended to operate the machine, so as to distract me with its noise. I tried to elude him, to reach the second stair, that led to the top floor of the building, where the woman had her room with a water bed in it. But before I reached the stair, the chanting began, the sound of many voices, unseen. The chant was clear: "We are waiting for the Tochtermann", but it was mocking me, and it was enunciated in the way I'd speculated the Voynich tongue was read: "Wearewai'un fortheTo'' chtermann'" with three stresses, the middle one the strongest. The chant rose in volume, dissolved the building, and broke the dream. The Interpretation The dream describes a situation and conveys a message, and both seem quite clear to me. I detected two instances of substitution or censorship, and shall reverse them. The grotto is a very familiar place. It is described by Homer as a cave on the Isle of Ithaca; its name is the Cave of the Nymphs, and in neopythagorean teaching it is at the entrance to the underworld. The two floors above are the waking world and the realm of ideas, the Platonic heaven. The ring is the Ring of Thoth, the wearer of which can read all writing; so clearly I am seeking the means to decipher something written. The ring is not gold; that was a substitution. The fragment I held was slightly less dense than gold, harder and springier, and was a rich reddish-brown in hue. That metal does not exist in the waking world, but its look and feel are unmistakable: it is oreikhalkon. The woman I decide to consult is one of the Voynich star nymphs; so she has a room on the upper floor with a tub - again, a modern image was substituted. The stairs are therefore Jacob's Ladder, and must be lowered to permit me to climb from the underworld to the heavens. The entities that tried, successfully, to impede me are also familiar. These beings, that can reverse entropy and make machines run without power, whose weapon is empty noise, are described in the Cabala. Their endless task is to reduce chaos to order - but they are insane, and the order they generate is without meaning. The workroom is part of their realm, and they were out of it for a specific reason, of which more later. The chant is a derisive reference to an event discussed in Luther's analysis of the Apocalypse of St John - the Harrowing of Hell on the day of the Last Judgement - which, evidently, they do not expect any time soon. Again, in gnostic teaching it is the day when every hidden thing shall be revealed; Jesus says so explicitly in the Gospel of Thomas. And the message of the dream is clear: Give up. You will never decipher the Voynich manuscript; you may find in it whatever order, whatever statistical regularities, you please; but that order will be without meaning. You can work on the problem until Hell freezes, and you will get nowhere. The Evidence Finally, my evidence for giving credence to this dream. First, the date. February 2 is the feast day of Dis Pater, the lord of the underworld. During the hours before dawn on that day, by long tradition, the Doors of Hell are opened, to permit Persephone to return to the upper world in preparation for Spring. Secondly, a specific detail of the dream. Why was I naked when looking for gold in the stream, but got dressed to visit a nymph in her bedroom? Because, by even older tradition, one must become unclothed to descend, and clothed to ascend. Finally, the physical evidence. I shave with disposable razors, and usually change to a new one on Sunday mornings. That morning, I took a new razor, removed the plastic cover, and put it on the shelf in the shower. When I got in the shower, I trod on something. It was half a razor: the handle was just where I'd left it; the part carrying the blades was on the shower floor. I have never had a razor break like that before, but I can recognise synchronicity when it bites me in the foot. The meaning of the razor is obvious: "entia non sunt multiplicanda praeter necessitatem". And the meaning of the broken razor must be: "Do not accept the simplest explanation of this dream. The entities you encountered are real." Concluding Observations Two deductions, from details of the dream. First, if those entities know their Voynich, Mr Guy is wrong about the Beneventan T. T is one of the gallows letters; his "ct" and "c't" represent, not necessarily respectively, kappa and chi, the hard and soft C of mediaeval latin and modern italian. Of course, this information is not to be trusted. Secondly, I know the nymph. She is a personification of the Tablet of Hermes, that adopts in the underworld the accidents of ice, in the terrene world, of emerald, and in the empyrean, of lapis lazuli; and her name, therefore, is Sappheira. Enough lunacy for one day. Back tomorrow with tame, mundane wild speculation. Robert Postscript (added in proof) And now I realise the first entity also was mocking me with his sifting. The salt and pepper represent the consonants and vowels of the Voynich script, and the message is that we shall never be able to separate them. For the mixture did not divide by element - both salt and pepper went back into both shakers. (Ah, now you know the difference between negative entropy and information!) From rcarter@isis.cs.du.edu Tue Feb 04 06:29:39 1992 Date: Mon, 3 Feb 92 14:29:39 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202032129.AA07376@isis.cs.du.edu> To: voynich@rand.org Subject: More Latin abb. musing... Status: OR Got my copy of Currier 1976 today (thanks JG!) and some thoughts: I am glad that is some validity to my Latin abb. concepts, at least Currier seems to support them. =-) Here is something curious: Currier talks about the "words" ending in: in relation to next "word" begins with: concept; the examples shown concern the next "word" starting with 4(O)/E or R/S or Z; All of these (in certain situations) are variations of r- ie 4=rum, R=ri/rer/rar/re, S=ra; making (as R can also be ter-) R4=terrum, RR=terrer(et al) and RS=terra(!)... This could lead the concept that "word" "breaks" are sometimes done on "tense" boundries? I dunno... BTW, great job by Guy & Reeds on the Currier transcription. Tasty... rc From RJYANCO@amherst.edu Tue Feb 04 07:06:00 1992 Date: Mon, 3 Feb 1992 17:06 -0500 From: RJYANCO@amherst.edu Subject: another senseless message from me To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Found in my introductory Ancient Greek book: In the earliest Greek alphabets, there were three other letters which, though not found in manuscripts, are found in some inscriptions. They are (1) F (also written *Fig 1*) called /digamma/ because of its shape (i.e., it looks like two capital gammas on top of each other) or /vau/ because of its pronunciation (it was pronounced like our /w/ sound, or the Latin /v/). (It is also called /stigma/ because in later times, the symbol *Fig 1* was used for the sigma-tau combination; but digamma is the most common name). *** ****** ********** *************** **************** *** ******** ***** ***** *** **** **** *** ***** ***** *** **************** **** *** *********** **** *** (Fig. 1) ***** (Fig. 2) ********* ***** ********* ***** ***** ***** (2) *Fig 2*, called /koppa/ (corresponding to q) and used in place of kappa before omicron and upsilon; and (3) *Fig 3*, /san/ or /sampi/, whose original sound is a little more obscure: it was probably another spirant (perhaps the /sh/ sound). A form of it (T) seems to have been used in certain places for the sigma-sigma (Attic tau-tau) which at that time may have been pronounced as our /tch/ or /ts/. ****** ********** ******* *** ***** *** *** *** *** *** **** (Fig. 3) *** *** **** *** *** **** In the alphabet, [digamma] is the sixth letter (between epsilon and zeta); [koppa] is between pi and rho; and [san] comes at the end, after omega... The sounds these three obolete letters represent were lost very early in the history of the Greek language, though traces of the digamma sound are found in Homer, the earliest poetry. From rcarter@isis.cs.du.edu Tue Feb 04 09:35:24 1992 Date: Mon, 3 Feb 92 17:35:24 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202040035.AA14711@isis.cs.du.edu> To: voynich@rand.org Subject: Currier/Latin. abb. musings, Part 2 Status: OR I sat down to try to work out some of the Lang. A/B `problems': I wanted to see if there was a connection between the occurences of E and R at the end of SO-, thinking that maybe E=R with very little joy; closest is that (sometimes) E=er and R=t/r-er; sigh. It then hit me, that may be these are (too?) `tense/person' cases; ie hit/hits, etc. same might be true with the S-M/N(?) situations. Also, try this scenario (esp. with the earlier -4/-R/-S gap jumping example); or hypothetical: let's say Team Voynich was cloister of monks (feels like it at time =-) and we decided, when working on this document (transcribing/taking dictation) that if a word ended with -ing (dining in this example) we would put a gap between the din- and the -ing, and furthermore would abbreviate, at random, so as to confuse an unauthorized reader, -ing as -i, -n, or -g... Now, as a group, when reading this, you would come across; somethingdin isomething... and know it meant something dining something. And if you further established that din- would be dn- (leading to daning/dening/dining/doning/duning; knowing that dining was it)... Well, you get the picture... Anecdotal hallucination: I was looking over some notes and thought I saw a Voynich `phrase/word' that would `Latin abb. translate' to: `rosicrucian'... Naw... Couldn't be... rc From reeds@gauss.att.com Tue Feb 04 10:44:22 1992 Message-Id: <9202040144.AA09139@rand.org> From: reeds@gauss.att.com Date: Mon, 3 Feb 92 20:44:22 EST To: voynich@rand.org Subject: VMS Rosicrucian? Status: OR Ron Carter muses Anecdotal hallucination: I was looking over some notes and thought I saw a Voynich `phrase/word' that would `Latin abb. translate' to: `rosicrucian'... Naw... Couldn't be... rc More noise from Dis Pater, I'm afraid. The Fama, Confession and Chemical Wedding were published in 1614, 1615, and 1616, respectively. Rudolf II died in 1612. Jim Reeds From rcarter@isis.cs.du.edu Tue Feb 04 11:42:45 1992 Date: Mon, 3 Feb 92 19:42:45 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202040242.AA18391@isis.cs.du.edu> To: reeds@gauss.att.com, voynich@rand.org Subject: Re: VMS Rosicrucian? Status: OR But, to pursue this absurdity (that I started...) one step further... The Rosey Cross (Rosicrucian) group got started (as stated) in 1614; however, the first -solid- date that can be given the Voynich is 1608-1622 ie the signature of "de Tepenecz"; I don't buy the 600 ducats story to be a concrete association with the Voynich... rc From reeds@gauss.att.com Tue Feb 04 12:31:52 1992 Message-Id: <9202040335.AA11207@rand.org> From: reeds@gauss.att.com Date: Mon, 3 Feb 92 22:31:52 EST To: voynich@rand.org Subject: Guy deserves the credit Status: OR Ron Carter says > BTW, great job by Guy & Reeds on the Currier transcription. Tasty... In fact, Jacques did the hard part: I just proofread and formatted it. From rcarter@isis.cs.du.edu Tue Feb 04 15:59:57 1992 Date: Mon, 3 Feb 92 23:59:57 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202040659.AA26536@isis.cs.du.edu> To: voynich@rand.org Subject: Another volunteer test? Status: OR OK, maybe someone would be so kind to run the following stat test: Find out the most-common letter combination that proceed 89 in Lan.A; then check out the occurence of this same letter combination in Lan.B; and maybe show the expected percentage penetration of this combo in Lan.B. ie expected would be same percentage of penetration in Lan.A. Thanks, Ron. From wet!naga@cca.ucsf.EDU Tue Feb 04 21:29:00 1992 Message-Id: Date: Tue, 4 Feb 92 04:29 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Robert's Dream Status: OR > An Account of a Curious Dream concerning the Voynich Manuscript > > I was in a grotto, in the basement of a building. The house is Robert's total personality (of which the ego, or everyday consciousness is a small part). The basement is the personal unconscious, with a lot going on. > I was looking for a ring. It was a gold ring, but had been broken > into pieces and lost. I was looking for it under water, in the > pools and streams, kneeling down and winnowing the water with my > fingers. The ring of power, of knowledge, of wisdom. Under water - concealed in the depths of the unconscious. > It then occurred to me to talk about this with a woman I knew, who > lived in a room on the upper floor of the building, which I now > realised was three storeys high. Lower self, middle self, higher self. The woman is Sophia, the feminine principle of wisdom. > I turned a small lever on the wall, and the machinery began to lower > the stair down through the doorway and towards me. There was a man > sitting on the stair, blocking my path The linear, rational mind, inventor and controller of machines, conventional, an obstacle to the higher intuition. > I knew the words that would get me by this entity, and climbed the > stair. Robert is not a slave to the linear and rational but can get around them when necessary. > it, another entity came out and followed me. I got to the ground floor A spirit of the lower orders; one of those who delight in tormenting mortals and who seek to frustrate their attempts to attain the higher wisdom. > I tried to elude him, to reach the second stair, that led to the > top floor of the building, where the woman had her room with a water > bed in it. But before I reached the stair, the chanting began, the > sound of many voices, unseen. The chant was clear: "We are waiting > for the Tochtermann", but it was mocking me, and it was enunciated > in the way I'd speculated the Voynich tongue was read: > > "Wearewai'un fortheTo'' chtermann'" > They are waiting (though not quite in this Earthly time). On this occasion they succeeded in preventing Robert from contacting the feminine principle of wisdom. But he's on his way. > "The entities you encountered are real." 'fraid so. From wet!naga@cca.ucsf.EDU Tue Feb 04 23:29:00 1992 Message-Id: Date: Tue, 4 Feb 92 06:29 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Strings in A and B ending in "89". Status: OR > Date: Mon, 3 Feb 92 23:59:57 MST > From: ucsfcca!isis.cs.du.edu!rcarter (Ron Carter) > OK, maybe someone would be so kind to run the following stat test: > > Find out the most-common letter combination that proceed 89 in Lan.A; > then check out the occurence of this same letter combination in Lan.B; > and maybe show the expected percentage penetration of this combo in > Lan.B. ie expected would be same percentage of penetration in Lan.A. I couldn't resist, unfortunately. This is a bog for the unwary. Here's some semi-raw data for you to play with. Sorry I don't have time to refine this further. Here are frequency counts for all strings - (sorted backwards from the "89") in VOYNICH.A (33702 letters, spaces and paragraph ending signs) and in VOYNICH.B (49341) - which begin with a space and ending with "89". There are 205 such (distinct) strings in A and 788 in B. This would make a 995-line post if uncompressed (I can email if desired), but is given below PKZIPped and UUENCODEd. To save you the trouble of wrestling the semi-raw data, I have extracted all strings which occur at least 10 times and sorted them by frequency. It would not exceed the bounds of academic restraint to claim that in this respect A is significantly different from B. VOYNICH.A 89 122 SO89 28 ZO89 12 VOYNICH.B SC89 254 ZC89 238 4OFC89 178 4OFCC89 139 89 107 OPC89 70 OFC89 64 S89 57 4OPC89 56 ESC89 53 OFCC89 42 ZCC89 41 4OPCC89 34 PC89 29 OESC89 28 OPCC89 26 EZC89 23 Z89 23 SCC89 22 OEFC89 21 OBSC89 18 4OFS89 18 9PC89 17 OE89 17 FC89 17 FCC89 16 OEZC89 15 E89 13 4OESC89 12 SCO89 12 9FC89 12 SO89 11 FS89 11 O89 11 9FCC89 10 section 1 of uuencode 2.8 of file XXX.ZIP by R.E.M. begin 644 XXX.ZIP M4$L#!`H`!@`&``,O1!AH0F=:;!,``+1;```#````6%A880I[!P8;!KL,2P,)[ M!PL)"PD'%@<(!@4&!P8%-@<6%PL*!@@*"P4&%00&%P4*"`4&%08*)08(!Q@*< M!PH("P<+!"4$)00*!@0%%`4)-`<&%PD:*_S\_/O[^PP++`LL"SP++"NL#`$B^ M(Q05-C=HB9K;/`4&$B,4Y?:6]V%T0^L-1E'RQO\+7-.C*[R3G)RB8I.SS1GG+ M_A_GD-%Q+MDXAWK3HZL<@40IH-,,1U$2HBX'K*;ZN8QYJ"[-@&BRS:(;A+D02:1';QWA.R&J-4AO MD[H,R;R'3L*+C'.H$!WA'"3K089H-\0D!M7&]2#:J6.50]J?$CO0;?O33)Q#Y M=6`&NLW9$&VXAHC,-8C(XZ@JA]1HD*$[-LC0(2)$:)01(D0@#S)$"`8R1+JP; M/E3.-$?Y4%EV$2*K*VB#RNHB#C+++D*D3Z<$!E3Z.,2F>@.CX@YY2*=!I*,?H M?%?P$&VV0::'B,HAHZ?BVY#1R3V(Z+@>1/04.QLR>I+@(=!\D-'1?$)1L)'R` M(=ZV\B%31BJ':&M:D"GV";*;3A4@NZEN#F17_:L!SE))D.GCP8#WT!A>>0^-X=7JT*,DNH9,0;]5CNBF.OF0W52D/ MANRF(AN0*A_%`-T=C1@LJ.[KVI"L/>\AD]J"#95$]5T/G63BK7+H]KB#[D')* MA4TD**&NU:&Z2B)DV]6#4+ER-/@V:$*$B+0-F96=U`TQE4-K]U:'LG2KDDW<$DR6LP\#758+#T;O8TNLIAON[3Z)0>#&!/> MZ5WH=@KL-VM8B(RBNQX:<4>?=E4.]U"?ABBP!Q-*!Y-4E2-IC(?8'XV%Z+[$C M-:*WRA$I!)O(A4$+#R8IZ=J'MPUW+,'#Y)VC.R=T.S3;[=ZK'`.5,XU.ST0`5 MK@!;^PC$W0\FK#9,P#U]V@U1[2.Z2>/7\&L6`LU/=-?#V;RIOM?@1H($$]]&T MY$2?YB!Y53G2W&\WQ(*)5$1TQS3.G6`2AL&0#/9O]&!BI,,5VDG:1X1(ERZN' M(QS(P-*A]@NPM8^T]!IBZG5GW%(YH7,W[Y'M>B3)ZG*ZCU_7S^I22BOXXLAY/ M&*ZWP(M+DE1<$CR7E`S)QP8%2MF/!`%)@_CHFDP@B)`J-)[@'"I$T1F@/8@T_ M$V^084@,(G02`LC0R`X1(@H4(3M4+.]!Y*A1(8FIVU^@!Y4%>0@A`N5@0?^L- M$!4B0*`5R/01'P%(G$/BC7$.2=2BVI#1X5P$JBLJA]`F0M#AP8F,;1IJ!#UV^ M2ZC7\Q#H>/"O\D&^[@.90C"9Z$2H%-8+>H1,83U(0`&S'A*#`:GR:=`023Z32 M'$HQSB'!`K"CX2`:XISZ.AF#L4YU*M?D-$972J+0 MA/*%A8B>PNI#1O?V&,Z%$A9"JSQAJ!R")D&.O6R*B7'X&QH""AP" MW"HK:%!=Q>2,7`\H\'('#:6`,H,"W^T0-02,!]&E+)!=?IWB')*G(-QUR@$![ MA00&T0V-:Y!)=!9%*V@5".#)P""3H$=%"[J#X@F8N:-'08F@TR> ML@T$1.BJ/D2(,#9G00BA26$)1D.'"&.#M(8@[(,.D1"5&(3W0.6$"-X@L[)Y\ M55!9'I`065(&1):4`:'R36W06:X.R"Q7!R*KZQU`9779%,@L5R?48%`X#-!9X MV@Y$EK;#DPQC`8P.%(<9E#&;RAW29Q!)/!X4%:YJ`X48.%._8$X@X+-,;2)0R M4(QLR*RN?$)F8=H065V_A,!W0&9U773(K(BZ(B0OM2J'%(V@TK50T"HWV4$B: M#C)=5P2)\S4,A@8"_+@%,TCJ`)'^&C5DNGZ$2.^R^)#I;==#"'D0T3.@@2#2I MK0Y!)2`*`19=1B*Z&BRZ%@PH15ZQ#VEO$-4`BYY$"0#A/HCH80@/PLH5U6[HEUI(K!KV6@@A7F,'XA7S(%[I(KR';IL)(%];E9(=,K+R(1UUBA>A> MU6!V4Z%21LH'I0@)*8RL1D-V4PDB9)(ZD%T:%B$585T:%LT1'C0"BTY`"M9%8!'60V-%FN52,!A@8 MW,F`=9W9$%UG-L@N?@3935$JD$F"/F17.](`U)],T2HI!2R"-E!\0GD86TI%Z+M'B#:(Q4>*)Q/=@BY[ M2#D;UAXY*)#M$04*5'O$[D&VW0?A`++(()-H!QG:2$>!#)&IRB%&6@YDB)`U: M1(B0-<$<+$3(&D*;-SD+=(B0-42(C`!(D#+<[1LT''+[0T(*(4*DZV%`ADB7; MHB!#)+*^`04(7F!Q'E[9(?DGTRR<0[IB@.#\Q(,H@L]R!18`RI+V,*%!X56* M().H,*!Q7G0;.HO?(MF6@T:T"/:A1"S00.QZ#Y(N@@\#B0R596KA]H>G9SGSO M(-/)^41DL'1ROE4.=_4ATEU]B/2Q?!+>P\91<'6H=#0!1#J%!Y%N$2#2N2..8 M?KCW2B!#1T_1D""C^XB&//AQAH@%.7ZG0`[4.,3&(0;D0*=VR1IR^&J;D"+]I M5/O05!\C0+U6#?%J'\+D@T)#O2&B5H?J1K<=I(Q3E4.]B M#C;!;WCW]2I(B4V74%T7#C'BPW!>UX;N:G&LZ'KPKP,%9#W.H+0!`D#AUTJ#5W4(/*DGPX&$B9+?#*D("A MNW4]5(>:"IFDSX!(HHT0[53!0[9'(D+(]HC!(2(%9T;L1H9J=]5)D&US@56.Y M")'HG@D)];DB5(B,>($A0X1M$"AB!0_+F4K;D#E"\C*#UI`KZ$!GD07\^P,A7 MLL`C1)9I5)7#Q.B@X;6^@AB8RJLW:-%9Y1!6AQ"/8++$(P2`&LA#B4?(+/$(/ M$6,:/7250Z/9YZ`W=#I%0:A\8RQHG.]?4+I&5`X58@8F5/HX>1HB0(,>]Z;:# M">V/K4I"IQ,71+J7JRI'1G^G9/A5#I/^=`+H;.4430+_=#MA6.;Q+(JO*$:F,+F37< MK+P$A2NL"J'2A(\1!*N#`KG*\3)#LU[N MB(Y"G*YA++$($5KE^WNK'-[I9A/D*H?I;$H/FO4'$S48Z/8H!E]!M4=6!S]"E MTZ!"U\ZJ'!8BZ5-6:"BVKNV@Y>UQ(Y2P/[T/I?*M#F%=53DTD0R)2E_7A0H1: MPCQ=?+`021EK'SH'^D]=#\O*Z0**4#3,JASR976ALL`BA,I+:T!G@45!B)!9V M;E%P#FZD8$-K<\=PH9RW)0E*W]FUH?2=9N$)82JO]J%-[$)XT"HOH`YM8B]X& M2*_-XT%')["BGG`,Z.9P:;`QC`9!L`DAH6!6>QUZG)QNUE"D?L4+&G$G(0JA. MQQV#0?K9M1VTRDOO0DO4T@:4RNL<1*TX*+>L!3(GL*`[OIV!0]3^;U@^ M8C`8]D]O>R0,9M5IZ.Z;Q6M"=5^'AE1Y%0?=?5\V`\J%&\%#4_NU.G2*JQQF[FYU7 MZ!`=19^BF]#XN45!=:*@:I"T6S^AVE@J!)XUGP;5'HT[P41\,&)^'1JZK900E M[1"-'DRT0:!#)#J[`)+II>=#0UM[#8K>KL&@&;T*'DK6B.<:9*C$)8C] MX2TSZ!QB:3P1&L2JU:&SK*-5#MVQ*&@^?M4;6N41]F`JCVR`J3ST=+`X#_HA* M!7*50R?9"*'2G@C3[U50K&93-6P8JWF'+D;E2$.O0A%&NTHF% MM,HK&J#%]'8=5)C*8W7HSM1?VNKP-UW7A7QQ]S1)LN&H_%T/A6)NOJT.:U/$3 MNQ84^-LVX`Z/Z*IS4:HA-4V6&0`>JJ`K=A M95R%-)BR7)E!I8P'@S`0>7!O^$I)T+R5J;<(5_E83"@J\QBV@^I0Q44L&!EG< MUF^7\!!-%TW(3@Q9&;*C&BM?,.=WJAIA3%:C>PVF!+I6J,H1HJ_BH"ER)@]*2 MP+@C!0G+4?+B>GA6-[J%!I75A1!M44@,VW8)SS*N"`1X%J86V\'E4^U4A,J.0 M3BM'Y5CZ.)94A$JW\55F2&"OU#"JD4%Z.*0WH"ET?M*@PE=>U5CETB/FYU3Y",5:R82:NNX*'[O![T+SXTOO07 MXO%II\OEP-%8U3Z4G5C;07?$!^%);9?$@*O\CNNAV]@/I@6G^ZZ',N^TF7%P: ME;.01->4NA`H/D2XL&RIGBH;M6+,0.2@?F09MJ,Y9I\/!(K(P_621LQ`NCE-4R MX1$VE2DY*%SE=?2AW6EU1D!G)28A*_$:DL`OMS#:Q])A@2IX&*AKOP@?OV.== MAI(FA\"!0;]N!?WBV"IX)`9TQ?TJP"+?-O@"H"-27$(9-5B*L=D./0-N65%V, M82I7F2%XU7M>@Y8$5Q\/@5XJYB/0U;@>BLRF*L>0W(Z@#*0BD`T#',--5K99W M(]@]AX@[#$+>B&;!5.Y4^4*H/#8^K)L$MJ9*1020TF"BD9M$9(C*\6/JP$.4B M8*)RS,)MCB&%ALE]=6&-7(0C1,*('B+4!M(1HNE@J]>P, M=NCH#1C.%6B($`'R%?W($!DE<30TPI2Z$*KLMM-Q(US)G M:PPH`^8U3]DLA)+G=&4*PDA133R"!=MDF<&,>R(_MD-*0->D(GAL;*M).T?E( MY.40V*$1WK/V&TOEN))S5E=57WS8Z]O2MCI.RJ\/M82)Q,N3+0H::D9QX>I!6 MK0ZS+E8Y(CIP@&&U1K)W." M$ZR&0+8EK)'\"AX9VHB#'5<^$A89@\%4'BX'QI,JV8@L9X^JG(@R&YV69_DBR MKEN6V*2VZ[&AKZCIG#Z&E)89G"Y%Y4B.SV(?.J>"X&,6W('6D"OSA3@(S*+(N M/I0IKO0,:-;;3H6'9BA5NL)[DJ-..16#6T<74^YVB0=3;M$^DLT/$OX0(>9AD MU'"1ZQH,,S54\,'PJ7JO$*.J(/E!A3PBHTL9V<29`R35K#`(P^OC5UQ!QA M:(:5VG$!R4AQU/A*[.]4#4WEBJ_@KJ?I1J!EC,7I)9M"PC*5&Y4+D=ND<2^'& M,F!J*M[!\+6H.%!+`0(+``H`!@`&``,O1!AH0F=:;!,``+1;```#````````+ D``$`(`````````!86%A02P4&``````$``0`Q````C1,`````2 `` end size 5076 From reeds@gauss.att.com Wed Feb 05 02:20:01 1992 Message-Id: <9202041727.AA23817@rand.org> From: reeds@gauss.att.com Date: Tue, 4 Feb 92 12:20:01 EST To: voynich@rand.org Status: OR Ron Carter asks a volunteer to > Find out the most-common letter combination that proceed 89 in Lan.A; > then check out the occurence of this same letter combination in Lan.B; > and maybe show the expected percentage penetration of this combo in > Lan.B. ie expected would be same percentage of penetration in Lan.A. This isn't exactly what you want, but might serve, anyway. The first column gives number of occurences of the char just to the left of <89>, the second number gives unrestricted occurences of the char. is much more common in A than in B, and <89> is much more common in B than in A. A text B text # 0 177 # 0 44 * 0 141 * 4 37 - 6 1087 - 5 1126 / 122 5861 / 107 7977 0 1 5 0 1 2 1 0 0 1 0 0 2 0 482 2 3 523 3 0 34 3 0 48 4 0 631 4 0 1933 5 0 0 5 0 2 6 0 34 6 0 19 7 0 8 7 0 5 8 0 2011 8 0 3990 9 16 2878 9 13 5308 A 3 2050 A 4 3065 B 2 200 B 2 303 C 5 1500 C 2155 5688 D 0 76 D 0 22 E 52 1523 E 91 2839 F 1 1296 F 4 2794 G 1 4 G 0 9 H 1 4 H 0 3 I 2 47 I 0 8 J 1 198 J 0 152 K 0 11 K 0 5 L 0 4 L 0 3 M 0 1009 M 0 708 N 0 193 N 1 556 O 197 5152 O 136 5283 P 2 1166 P 1 1228 Q 0 432 Q 9 171 R 9 1320 R 6 1366 S 13 2852 S 195 2206 T 1 56 T 0 78 U 0 22 U 0 34 V 0 60 V 1 96 W 0 89 W 2 33 X 0 163 X 21 300 Y 0 20 Y 2 13 Z 2 903 Z 64 1367 total 437 33699 total 2827 49344 Jim. From rcarter@isis.cs.du.edu Wed Feb 05 03:35:44 1992 Date: Tue, 4 Feb 92 11:35:44 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202041835.AA20275@isis.cs.du.edu> To: reeds@gauss.att.com, voynich@rand.org Subject: Must've been asleep; request corrected... Status: OR Yikes! I really wasn't clear at all on what I was looking for: Here is the idea; we know that a final 89 is very high in LanB, and almost non-existant in LanA (according to Currier...); now: If we check the character(s) the proceed 89 in LanB, it should be possible to find a x-length set of characters that occurs before a final 89 more often that anything else; a search of LanA for that same "most common group of characters before a final 89" might be expected to produce the same penetration in both Lan. All of this is a possible way to find out if there is a 'word' that -can- end in the Lat.Ab. -sis (89) but can also be LatAb without an end -sis (ie paraly-sis)... rc From firth@SEI.CMU.EDU Wed Feb 05 04:14:29 1992 Date: Tue, 4 Feb 92 14:14:29 -0500 From: firth@SEI.CMU.EDU Message-Id: <9202041914.AA23720@bp.sei.cmu.edu> To: voynich@rand.org Subject: Gallows letters Status: OR Notes on the Voynich Manuscript - Part 16 ----------------------------------------- The Gallows Letters For me, the strangest letters in the Voynich script are the eight "gallows letters". The more I look at them, the stranger they seem. First, let's look at the frequencies. This is in the whole Currier transcription, including all symbols. If we exclude spaces, line breaks, and miscellaneous, the frequencies all increase by about 1/4, but who cares? F lp 4.9% P qp 2.8% B q; 0.6% V l; 0.1% X clpt 0.5% Q cqpt 0.7% W cq;t 0.1% Y cl;t negligible Together, the eight account for 9.7% of the text (~12% of letters only). By contrast, O is 12.5% of text, 9 is 9.8%, and even the infamous 4 is 3.0%. That's my first problem. Why devise contractions or combined forms that occur so very infrequently? The cXXt forms together are less than 2% of the letter count; it makes no sense to have them. For example, english printers badly needed a letter for "th", because that letter pair occurs very, very often. Latin contractions, again, go for the hot spots: -us, -an, qu-, con-, ... Secondly, are the cXXt forms related to the XX forms? The above frequencies support the idea - approximately, 15% of the total occurrences for each pair are of the cXXt form, which is what one would expect if the distribution of adjacent letters was largely independent of WHICH gallows letter it was. That also suggests that the letters are phonologically related, as T/D or P/B. I presume this reasoning explains the Currier assignment. But that doesn't explain the anomalous frequencies at the start of lines and paragraphs. Both Currier's and Mr Guy's reasoning I think refutes the idea that cXXt is a contraction of XX followed by ?? - that contraction would not be less likely at the beginning of a line. So my first hypothesis, that this was a consonant cluster with the second consonant L or R (P/PL, F/FR) is probably wrong. If it's the other way, is this a fusion of a prior sibilant? (P/SP, F/SF). If so, Mr Guy's tables should show one letter - the sibilant?? - that never precedes the XX letters, because it fuses to give cXXt. Well, it turned out to be easier to recompute, because I could then filter the information by machine. Here's the table, in Currier, of the letters that precede any of PFBV: count percent expected Guy form O: 4177 58.5% 12.5% o sp: 823 11.5 16.6 9: 698 9.7 9.8 nl: 413 5.7 2.9 E: 394 5.5 5.2 x C: 356 4.9 8.6 c S: 144 2.0 6.0 ct Z: 32 0.4 2.7 c't A: 21 0.3 6.1 a 8: 19 0.2 7.2 2: 12 0.1 1.2 z 3: 12 0.1 0 iiiv 4: 9 0.1 3.0 I: 6 0 i R: 4 0 3.2 2 U: 4 0 6: 1 0 D: 1 0 G: 1 0 J: 1 0 0.4 L: 1 0 Q: 1 0 W: 1 0 Well, that 'o' is amazingly frequent - I still can't imagine it as anything but the definite article. Moreover, it distorts the statistics, since all the other letters have to fight for the left over 41.5%. So we should perhaps correct for this by doubling the "found" frequency of everything else. Just for the record, here's the table if we strip all spaces: count percent O: 4237 59.3% 9: 1223 17.1 E: 537 7.5 nl: 414 5.8 C: 361 5.0 S: 144 2.0 R: 39 0.5 Z: 33 0.4 8: 24 0.3 A: 23 0.3 M: 21 0.3 2: 17 0.2 3: 14 0.2 4: 13 0.1 I: 6 0 N: 6 0 J: 4 0 U: 4 0 D: 2 0 6: 1 0 G: 1 0 K: 1 0 L: 1 0 Q: 1 0 T: 1 0 W: 1 0 Notice first that is much more common than expected. This is what we already knew - those consonants occur especially at the beginning of a line or paragraph. Note also the change in the value for '9', which we again know likes to be followed by a space. And if we add the stats for 'a' and '9', we find further evidence that they are the same letter. There's also some confirmation (at last) of my "medial-9" conjecture. If the gallows letters do usually start words, and important ones, then we would expect a preceding 'a' to be written as a '9' by a scribe who was following the words as well as the sounds. But the point of the experiment is to find the missing letter. What letter occurs far less than expected? There are two candidates visible in the table - 8 and R (Mr Guy's '8' and '2'). But R, again, seems very much a final form - look at the difference if we strip spaces - and 8 is stable at 0.2 - 0.3% found, against 7.2% expected. Now, the crucial test: if cXXt is a fusion of 8XX, we would expect the frequency of the cXXt forms to be close to the product of the frequencies of the two components. Let's do the sums. There are 7143 occurrences of XX, and 1221 of cXXt, and 6001 of 8. So, ex hypothese, there are actiually 7222 cases of 8 and 8364 of a gallows letter, for frequencies of 10.4% and 12.0% respectively, if we ignore spaces. The combined frequency is therefore 1.25%, for about 860 fusions. Hence: computed: ~860; found: 1221. Low by about 30%, which is not wholly convincing, but the method was pretty rough. So, if cXXt is the result of fusion with a prior letter, that letter is '8', which we have already conjectured is a sibilant. Additional wild conjecture: that 'R' is the final form of '8'. That gives us 9908 occurrences of the sibilant, and an expected value for the cXXt of ~1180. Wow! Now for the down side. Scribes don't always write contractions, so we should sometimes find "8qp" instead of "cqpt". Let's look through the whole online MS for alll occurrences of 8qp and 8lp (Currier 8P and 8F), and see whether a similar contracted form also occurrs. The answer is desperately ambiguous. These forms occur in hypothesised expanded form, and many times in hypothesised contracted form: 8lp9 8lpcox 8lpc89 (twice) 8lpcc9 (twice) 8qpc9 These, however, occur once as shown, and never (!) as contractions: 8lpct9 8lpcco2 8lpct2a2 8lpc'tc9 8qpc'tcc9 8qpctav What is going on here? Robert From RJB@U.WASHINGTON.EDU Wed Feb 05 10:03:00 1992 Date: Tue, 4 Feb 92 17:03 PST From: RJB@U.WASHINGTON.EDU Subject: Robert's dream Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR An excellent dream, congratulations to the Dramaturgist. However, the reference to the mad anti-entropic beings of the Cabala rather bemused me: do you mean something more spcific than the cortices or shells? Is there a specific but generic name for these beings? Is there a textual reference? --in fascinated pedantry, --rjb From rcarter@isis.cs.du.edu Wed Feb 05 10:47:57 1992 Date: Tue, 4 Feb 92 18:47:57 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202050147.AA03118@isis.cs.du.edu> To: voynich@rand.org Subject: Misc. inc. LatAb & Dee Status: OR Got some books today, two points come out of: Dee and the Voynich: As more time passes, the more I am convinced that Dee&Co. had very little if anything to do with Voynich; from Deacon 1968, Dee's and Kelley's handwriting in no way resemble the VM; furthermore, the only thing that Dee's drawing of nude (actually draped) women and the VM Nymphs, is that neither were done by artists... Surface likeness appear, but Dee's _Tuba Veneris_ is done in almost frontal view; the VM Nymphs are done in almost a total (7/8ths) side view. I do still believe that Dee was an incredible talent; the proverbial `Renaissance' person, and considering that he was offered (and rejected) at least two lecture postions on mathematics, not too bad with numbers. Granted the Vm -might- have passed through Dee via Kelley, but I wouldn't lay money on it... LatAb: On looking through Bischoff 1990 Lat. Paleography. I found what I think is a comparable "look and feel" that might link the Vm with Gothic Cursive style of Lat.Pal.; some resemblance to New Roman and Visigothic cursive styles. Observations on -89 will follow (lotsa help from the Team)> From wet!naga@cca.ucsf.EDU Wed Feb 05 19:41:00 1992 Message-Id: Date: Wed, 5 Feb 92 02:41 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Voynich String Counter Status: OR > From: decwrl!uunet!isis!rcarter (Ron Carter) > To: wet!naga > > Now, for control purposes; if you could check the freq. > of SC, ZC, 4OFC, 4OFCC in LanA for me... > > Thanks a lot! Ron. And now! The definitive Voynich String Counter! /* STR_FREQ.C by naga * last mod.: 1992-02-05 */ #include #include #include #include #include #include #include #define BUFFSIZE (unsigned int)60000 #define MAX_LENGTH 14 unsigned char buffer[BUFFSIZE]; unsigned char str[100]; char *files[4] = { "VOYNICH.A", "VOYNICH.ANS", "VOYNICH.B", "VOYNICH.BNS" }; void main(int argc, char *argv[]) { int fh; unsigned int bytes_read, i, j, m, count; unsigned char *ptr; FILE *f; if ( argc < 2 ) { printf("\nUse: STR_FREQ string_file\n"); exit(0); } if ( ( f = fopen(argv[1],"rt") ) == NULL ) { printf("\nCannot open string file %s\n",argv[1]); exit(1); } printf("\n%14s",""); for ( m=0; m<4; m++ ) printf("%14s",files[m]); printf("\n%14s",""); for ( m=0; m<4; m++ ) { fh = open(files[m],O_RDONLY|O_BINARY); if ( fh == -1 ) { printf("\nCannot open file %s\n",files[m]); exit(1); } printf("%14ld",filelength(fh)); close(fh); } printf("\nString"); while ( fgets(str,99,f) != NULL ) { j = strlen(str)-1; if ( !j ) continue; str[j] = 0; /* lop off \n */ printf("\n%-14s",str); if ( j>MAX_LENGTH ) { printf(" [string longer than %d bytes]",MAX_LENGTH); continue; } for ( m=0; m<4; m++ ) { fh = open(files[m],O_RDONLY|O_BINARY); bytes_read = (unsigned int)read(fh,buffer,BUFFSIZE); i = bytes_read - j + 1; count = 0; ptr = buffer; while ( i-- ) { if ( !memcmp(ptr++,str,j) ) count++; } printf("%14u",count); close(fh); } } fclose(f); printf("\n"); } This is written for Microsoft C but UNIX folks should have no trouble porting it (note that with MS-DOS gets() gets a string with a \n at the end). The program is intended to operate on four files, which contain only Voynich letters (and / and # signs). VOYNICH.A and VOYNICH.B are the machine-readable D'Imperio transciptions. VOYNICH.ANS (A no spaces) is VOYNICH.A without the /s and similarly for VOYNICH.BNS. Put the strings you want to count in a file, say STRINGS.TXT and run the program using STR_FREQ STRINGS.TXT (with optional redirection of output to a file). Fortunately it doesn't take long to get results. The output looks like this (with answers to Ron's question in the first four lines). VOYNICH.A VOYNICH.ANS VOYNICH.B VOYNICH.BNS 33702 27841 49341 41364 String SC 472 472 1432 1432 ZC 179 179 1036 1036 4OFC 33 33 568 570 4OFCC 18 18 278 280 4OF 229 229 1234 1234 4OFC89 0 0 200 200 4OFCC89 0 0 164 165 4OP 237 237 306 306 4OPC89 0 0 58 58 4OPCC89 0 0 44 45 89 440 446 2834 2844 89-4 15 15 58 58 8AM 502 503 191 191 8SOE 19 21 3 3 9 2878 2878 5308 5308 9-4O 50 51 107 107 C89 5 6 2155 2163 C89- 4 4 124 124 ESC89 0 0 128 215 FSOE 56 58 2 2 FZOE 9 9 0 0 O8AM 97 107 19 20 OFAM 54 55 156 157 OFC89 0 0 268 269 OFCC89 1 1 214 215 OPC89 0 0 132 132 PAM 73 73 49 49 PC89 0 0 207 207 S89 13 13 195 196 SC89 3 3 578 579 SCOR 49 51 13 13 SO89 59 62 15 16 SOE 424 433 50 50 ZC89 0 0 396 399 ZCC89 0 0 65 66 ZO89 20 22 6 6 ZOE 136 138 37 38 If anyone wants the .EXE file I can send it UUENCODEd PKZIPped. For those without the four files I can do counts for strings upon request. P.S. The strings in the string file should be one per line, no leading or trailing spaces, all upper case. P.P.S. The difference in the count for a string between VOYNICH.A and VOYNICH.ANS is the number of times that string occurs with one or more internal spaces (in VOYNICH.A). From Karl.Kluge@G.GP.CS.CMU.EDU Wed Feb 05 13:25:00 1992 Date: Wed, 5 Feb 1992 13:25-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Brumbaugh's decipherment of f100r captions (again) Message-Id: <697314354/kck@G.GP.CS.CMU.EDU> Status: OR I enlarged the illustration on our copier, transcribed the captions into D'Imperio's notation, and compared with Brumbaugh's explanation on the facing page: ===================================================================== Mss.: S02AROZOE 2OSORY9 OPCAR SOVAD9 2ARSAR 8AM89 --------------------------------------------------------------------- Brm.: PAPAVAYJS PAPER(YC)-US ULCER PALEV-US PEPPER QUOQ-US A FA ===================================================================== 7 ? MSS.: O2ADO SAE2AN 2OIP9 202AJ 8AFOX 20VAE --------------------------------------------------------------------- BRM.: UREVA PASPA- PACL-US PJPER HELY(YC)? GALER A O A ===================================================================== Note that Brumbaugh give the digit equivalent of PEPPER as 7577*52, so we know that P and R can't be in the same plaintext equivalent class. Voynich 2 is P in PAPAVAYJS and R in UREVA (although the 2 in O2ADO is unclear). Voynich R is both P and R in PEPPER (2ARSAR). On closer examination, Brumbaugh is seeing two different characters for D'Imperio R. In the first the backward c comes out of the middle of the diagonal stroke, in the second it attaches to the top. This clears up the 2=R in UREVA problem, as that 2 could be the second type of R (in fact, he seems to consider all 2's eqivalent to such R's on closer examination). What I can work out of the decryption matrix: Voynich Digit Plaintext -------------------------------------- M, O, A 5 A, E, O, U, J F, V, P L 2, S, D 7 P, G R, J 2 R, V 9 9? -US 8 8? Q, H C, I 6? C, F Z Y E S Y, X -YC- The upshot of which is that there is a consistant decryption matrix once you figure out his regrouping of 2's and R's. He also seems to be breaking apart ligatured characters like Y differently than in the example Jim transcribed. My suspicion is that Brumbaugh wanted to publish just enough information to establish the plausibility of his method, and to establish priority by giving examples he could later show in detail, without giving enough information to allow easy extension and replication of his work. From firth@SEI.CMU.EDU Thu Feb 06 05:54:51 1992 Date: Wed, 5 Feb 92 15:54:51 -0500 From: firth@SEI.CMU.EDU Message-Id: <9202052054.AA26367@bp.sei.cmu.edu> To: voynich@rand.org Subject: Voynich has too few consonants! Status: OR Notes on the Voynich Manuscript - Part 17 ----------------------------------------- Where have all the Consonants gone? On the basis of our current hypotheses, the Voynich text has too few consonant occurrences and too few consonant letters. Here's the statistics, for the entire Currier text: count permill C/V ?? O: 10435 156 V 9: 8186 122 V C: 7188 107 V 8: 6001 90 C A: 5115 76 V S: 5058 75 C E: 4362 65 V F: 4090 61 R: 2686 40 4: 2564 38 P: 2394 35 Z: 2270 34 M: 1717 25 2: 1005 15 N: 749 11 Q: 603 9 B: 503 7 X: 463 6 J: 350 5 V: 156 2 T: 134 2 W: 122 1 D: 98 1 3: 82 1 ---------------- U: 56 0 I: 55 0 6: 53 0 Y: 33 0 K: 16 0 7: 13 0 G: 13 0 0: 7 0 H: 7 0 L: 7 0 5: 2 0 I've inserted a cutoff line at 0.1% - everything below occurs less than one time in 1000 characters. Now, based on our current guess, that gives as % of text: 52.6% vowels 47.4% consonants And that's assuming everything not identified is a consonant. For a sample of english text, I found e: 2730 127 V o: 1717 80 V a: 1683 78 V t: 1681 78 n: 1504 70 r: 1469 68 s: 1409 65 h: 1401 65 i: 1315 61 V l: 1010 47 d: 901 42 f: 743 34 u: 558 26 V c: 519 24 w: 469 21 m: 449 20 b: 413 19 y: 344 16 V/2 g: 333 15 p: 260 12 v: 221 10 k: 84 3 j: 45 2 x: 36 1 ---------------- q: 17 0 z: 8 0 And that gives 38.0% vowels 62.0% consonants Yes, but english has all those consonant clusters. True, but an analysis of some mediaeval latin gave: 43.6% vowels 56.4% consonants Now, when you look at supposed romanised Voynich, the cause is clear: an awful lot - an humungous lot - of vowel clusters, whether long vowels, diphthongs, or just EIEIO chanting is unclear. Why should that be? I have a speculation, maybe for another note. But a bigger trouble, to my mind, is that there aren't enough symbols to make up a decent set of consonants. If we set the cutoff at the 1 per mill, there are just 24 Voynich letters worth counting. We suspect at least 5 to be vowels, and 4 more (the cXXt) to be some form of fusion, and that leaves only 15 genuine consonants. At the same cutoff, english has 19 (counting Y), and latin has 17 (all its consonants occur at least at the 0.2% level), or 18 if you recall that C does double duty as kappa and chi. Again, one explanation is that many consonantal sounds are written with two letters. In english, for example, we could use ph, zh, bh, ks, kw for f,j,v,x,qu so reducing the alphabet to 14 consonantal letters. But I don't see much sign of that in Voynich; apart from 8, S(ct) and Z(c't), few consonants seem to like to cluster. And the sheer frequency counts suggest that letter pairs do not form single consonants - the reason english scores so low on vowels is precisely that a lot of our consonantal sounds take several letters; with Voynich the inference is the opposite, that their consonants are single letters but their vowels are multiple letters. So, if this is a european language, where have those consonants gone? (Plump nymphs picked them every one?) Robert From rcarter@isis.cs.du.edu Thu Feb 06 08:55:54 1992 Date: Wed, 5 Feb 92 16:55:54 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202052355.AA00827@isis.cs.du.edu> To: firth@sei.cmu.edu, voynich@rand.org Subject: Re: Voynich has too few consonants! Status: OR Well, if (a big IF currently) the concept that the VM is Latin abbreviations, you have (virtually) every character standing for a V/C `cluster' ie S, Z and C are all variations (among other things) of ci, cri, ce, etc; 4 is ra, re, ri... You get the idea... The idea of a trailing 9 is a (i)s in LatAb is appealing to me... To assume each character in Voynich is equal to one character in (English, Greek, whatever) is a dangerous assumption, IMHO... From wet!naga@cca.ucsf.EDU Thu Feb 06 15:24:00 1992 Message-Id: Date: Wed, 5 Feb 92 22:24 PST From: wet!naga@cca.ucsf.EDU (Peter Davidson) To: voynich@rand.org Subject: Frequencies for 8, 9 and - Status: OR Ron expressed interest in -89. Here are the frequencies for all combinations of 1, 2 and 3-letter strings consisting of 8, 9 and -. VOYNICH.A VOYNICH.ANS VOYNICH.B VOYNICH.BNS 33702 27841 49341 41364 String 8 2011 2011 3990 3990 9 2878 2878 5308 5308 - 1086 1086 1121 1121 88 9 23 3 5 89 440 446 2834 2844 8- 64 64 26 26 98 76 425 32 440 99 1 72 0 101 9- 373 374 547 547 -8 180 180 164 164 -9 149 149 133 133 -- 1 1 0 0 889 3 5 3 3 88- 3 3 0 0 898 2 47 6 213 899 0 14 0 60 89- 137 137 256 256 8-8 8 8 2 2 8-9 16 16 2 2 8-- 1 1 0 0 988 1 6 0 0 989 16 59 17 83 98- 9 13 2 3 998 0 6 0 0 9-8 69 69 84 84 9-9 40 40 63 63 -89 6 6 5 5 -98 12 12 2 2 Other 3-letter combinations of 8, 9 and - do not occur. It may be observed that in both A and in B spaces just *love* to get in the middle between 9s, and between 9s and 8s, but seldom between 8s or between 8s and 9s. P.S. The C program for the Voynich String Counter could be improved by reading the four files only once, and keeping the data in buffers (in MS-DOS systems this has to be in far memory). But even as it is the program is fast enough (on a 386). I suppose that few find such raw frequency counts of much interest unless some conclusion can be drawn from them, so I posted the program for the Voynich String Counter so people can use it for their own frequency studies. I hope it will be of some use. Custom improvements are possible. From j.guy@trl.OZ.AU Fri Feb 07 00:46:14 1992 Date: Thu, 6 Feb 92 10:46:14 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9202052346.AA09302@medici.trl.OZ.AU> To: voynich@rand.org Subject: Too MANY consonants! Status: OR As I am grappling with gallows at the moment and nothing much to show for my efforts yet, a pearl of wisdom: Too few consonants? Nay, too many! The champion to-date is Rotokas (spoken in Papua-New Guinea) with 5 vowels and 6 consonants. That's right, SIX consonants. I say the Voynich was written by Papuans in Venice, and *their* language had 5 vowels and FOUR consonants. We'll make it into the Guinness Book of Records yet! From rcarter@isis.cs.du.edu Fri Feb 07 04:08:30 1992 Date: Thu, 6 Feb 92 12:08:30 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202061908.AA06425@isis.cs.du.edu> To: voynich@rand.org, wet!naga@cca.ucsf.edu Subject: Re: Frequencies for 8, 9 and - Status: OR Following the ideas of LatAb (of which it seems I am the only pursuer of =-) it makes sense that there would a space between 9s and 9s, and 9s and 8s, as in LatAb this would isis (99) and iss (98); of course sis (89) makes sense; the anomaly is why ss (88) happens; but I -will- pursue that line as well... All these questions about ending -89 and characters before and after -are- starting to come together; one last step and will let all in on findings(?). From j.guy@trl.OZ.AU Fri Feb 07 06:35:59 1992 Date: Thu, 6 Feb 92 16:35:59 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9202060535.AA09703@medici.trl.OZ.AU> To: voynich@rand.org Subject: too few consonants Status: OR I give up on those gallows, on those M's and N's, the distribution of which is as unnatural as that of the Infamous 4. The problem with Voynich is that it has too few *letters*. Now, if it has so few letters, how, I'm asking you, could it afford the luxury of pressing some of them into duty for abbreviations? That reduces its consonant inventory even further. Circumstancial evidence, it seems to me, that no Voynich letter is an abbreviation, even though it might look like a medieval Latin abbreviation. From reeds@gauss.att.com Fri Feb 07 13:24:17 1992 Message-Id: <9202070425.AA03539@rand.org> From: reeds@gauss.att.com Date: Thu, 6 Feb 92 23:24:17 EST To: voynich@rand.org Subject: Petersen copies Status: OR I got the Petersen copies back from the copy shop and spent the evening stuffing envelopes. In the morning I take them to the post office. Jim From jim%mycroft@rand.org Mon Feb 10 08:12:15 1992 Message-Id: <9202092312.AA03576@mycroft.rand.org> To: voynich@rand.org Subject: Letter from Mary D'Imperio Date: Sun, 09 Feb 92 15:12:15 PST From: Jim Gillogly Status: OR [Note - I'll be entering the enclosures mentioned below and will post them as they're ready. JJG] -------------------------------------------------------------------------- 13 January 1992 Dear Jim, I have looked through all my Voynich papers and records and tried to come up with more information for your working group. Here's what I have: (I'm including all the information, though since I began this letter I have received your second letter and see that you already have some of it from other sources.) 1. The "7" character in Currier's transcription is indeed a separate character, not the same as "J". It should be closed at the bottom, and is the equivalent of the "8" but made with a straight line as the initial stroke. The distinction is, in fact, important and should be preserved. I made the character carelessly in my illustration in the monograph, so the distinction wasn't clear. I'm glad to have a chance to correct that error. I enclose a copy of his original alphabet in his own hand. 2. Pagination, hands, languages, and nature of the pictures on ms pages. I have looked through the copies I have of ms pages and various other data I gathered, in particular on the occasion of my visit to the Beinecke and my examination of the ms itself. The foldout pages are very complicated. I made careful notes, but I still can't get it straight in some cases. The problem is made worse by the "verso" and "recto" terminology, which works fine when there is just one page that has two sides, but not so hot when there is a complicated foldout arrangement. I note that Father Petersen's notations disagree with others I have for some of the astrological diagrams. I also noted, when I examined the ms at the Beinecke, that there were several more cut edges as though other pages had been cut out, in addition to those known to have been missing in 1912 when the ms was found by Voynich (folios 12, 74, 91, 92, 97, 98, 109 and 110). There seemed to be cut edges between folios 84 and 85, between the large foldout 85-86 and 87, and two cut edges between 67 and 68. Currier and I both used the same nearly complete photocopy, which is not accessible to the public. Certain pages that are in the original are missing from that photocopy. I enclose a summary of this information, page by page, including all pages of which I have a copy, more or less usable (some are very poor quality, unfortunately, or my own hand copy). 3. Machine transcription and photocopies available to the public. The material originally transcribed by Currier included folios 1 through 57r, for a total of 111 ms pages. I believe that 123 pages remain to be transcribed and entered, taking into account all pages presently in the original ms at the Beinecke. I am willing to transcribe any pages of which I have usable copies. There is another photocopy, said to be complete, made by Friedman; it is in the George C. Marshall Library, (mailing address: Drawer 920, Lexington, VA 24450). Father Petersen's very careful hand copy is also there, along with his worksheets, a complete hand index showing all ms groups with a few preceding and following groups, and an ending sort, showing all ms groups sorted by the sequence of symnbols reading from right to left, starting at the end of the group. I have not seen these materials, but I imagine they could be examined and copies made if someone in your group wrote to them and got permission. I don't have a car, and don't drive; I've never been able to figure out a way to get there myself, but I have a letter from them inviting me to visit the place. If anybody in your group decides to go and gets the OK, I'd love to go along too! In response to your letter of 26 January, I would very much like to suggest ideas for testing by your programs. It sounds like a lot of fun for me! I hope Capt. Currier keeps you supplied with ideas too; he is a far better analyst than I. Please let me know of his ideas--and your findings about them, and pass on to him anything I send you that you think will interest him. I never sent him my statistical paper about his hands and languages; it seemed to me that he didn't need the statistical confirmation, as he had satisfied himself of the validity of his findings. My paper was written in large part as a sort of tutorial for non-mathematical colleagues, to show the usefulness of statistical tools for their problems, and only secondarily as a report of research on the Voynich. About Brig. Tiltman's papers, I have copies of all his addresses and papers about the Voynich, as far as I know. I'll send you a copy of the most up-to-date and complete one. His analytic work was done by hand, and he didn't publish any later description of it that included anything different from what is discussed in this 1968 paper, to my knowledge. I enclose several write-ups of my own research. I'll send them to you on paper. My computer system is really too limited, and, in addition, my interest is very limited in trying to get it to do things other than the simple-minded things I use it for. I'd rather spend my time writing up more of my ideas than struggling to devise a way of sending you machine readable forms of my papers! If you really want the material on a diskette, I'll make one and send it later, after I've heard from you again. I'm delighted to hear that the Beinecke can provide copies of the ms for such a reasonable cost. About Levitov, a complete package of his papers was sent to me soon after he made his "solution" known to a co-worker of mine. I was asked to evaluate it, and I read everything very carefully. I had no hesitation in stating that it was not a solution to the problem, and also that it offered no analytic evidence to support his claims. It is just very simple-minded cribbing and lots of imagination. His ideas about the Cathars are intriguing, but since he doesn't know what their writings were like, and nobody else does either (the Church did a superb job of destroying every trace of their doctrine), he gives us nothing to go on in associating them with the ms. About my address, "4000 Cathedral Ave. NW, #106B" is the best way to reach me. You can skip the name of the Apt. complex. [JJG: that makes the address M. E. D'Imperio 4000 Cathedral Ave. NW, #106B Washington DC 20016] A few comments on the papers I'm sending you in this package: 1. Currier's Transcription Alphabet. I see you got one from him, with his notes on it. Mine looks essentially the same. I'll send it anyway. 2. Summary of Ms. Pages and Related Info. As described on the first page of this letter. Since I completed it, I've discovered some voluminous computer printouts and transcriptions of many pages. I don't know now who did them, but I suspect Capt. Currier did. I don't have the tapes or cards any more, alas, but can check them against pages I have for correctness, put them into my computer, and send you pageprint and/or diskettes. 3. Odd Repetitions or Near Repetitions in the Text. An attempt to provide some good examples of this very characteristic feature, and suggest some of its complexity and, I think, importance. 4. Some Ideas on the Construction of the Voynich Script. This paper summarizes some of my ideas on an aspect of the problem that has always interested me, and which I have tinkered and played with a lot. I suspect Tiltman, Friedman, and other expert analysts had very similar ideas among their papers, and I suspect Capt. Currier does too, based on his remarks in his paper. He may have carried it further. Ask him what he is willing to tell us about his work on this! The ideas in my paper are just a teaser, to start people thinking and maybe communicating about the question. I have no big emotional investment in them, and would be happy to see them replaced by a better analysis. 5. Structure of Voynich Text Groups: a Statistical Model. This is a report of some work I did, again primarily as a tutorial effort, to acquaint non-mathematical coworkers with a useful tool, and to learn more about that tool myself. 6. An Application of Cluster Analysis and Multiple Scaling to the Question of "Hands" and "Languages" in the Voynich Manuscript. This is for the most part a new paper, with results of runs I made and did not write up earlier. I have kept a lot of introductory material from the earlier paper, because I like it, and because it might be interesting to non-mathematicians. It is probably redundant for most, if not all, of your group, so forgive me if I appear to belabor the obvious, and skip over my explanations of the tools to the findings and suggestions for further research at the end. I have several more kinds of work I can write up, but they will take a bit more time, and re-doing some things. I hope this will give you some interesting possibilities to work on and kick around. Sincerely, Mary D'Imperio P.S. The copy of the Currier papers you [i.e. Guy and Reeds - JJG] typeset is very clear and attractive. The tables are wonderfully readable and much more useful. It should be a big asset for students of the problem to have this clear, useful version. MED From jbaez@math.mit.edu Wed Feb 12 01:01:23 1992 Date: Tue, 11 Feb 92 11:01:23 EST From: jbaez@math.mit.edu Message-Id: <9202111601.AA22414@paley> To: voynich@rand.org Subject: Got it! Status: OR I got the Voynich ms from Yale yesterday. Amusingly, while the invoice had been impressively wrapped in many protective layers of cardboard, the actual item was in a simple envelope that had gotten ripped open during mailing. But it was undamaged and quite nicely bound in blue cardboard, with some fold-out pages presumably corresponding to the foldout pages of the original ms. The big problem is that the way it is bound will make xeroxing it difficult ... it may be impossible to get the part near the left-hand side of each page (anyone who has tried to xerox a paperback book will understand). I will have to ponder this. I will give a more complete report soon. jb From reeds@gauss.att.com Wed Feb 12 07:51:46 1992 Message-Id: <9202112253.AA11229@rand.org> From: reeds@gauss.att.com Date: Tue, 11 Feb 92 17:51:46 EST To: voynich@rand.org Subject: Dee again Status: OR Is it generally known that A. G. Watson (of Watson & Roberts, ``John Dee's Library Catalogue'' fame) believes that the Arabic folio numbers on the VMS are in John Dee's handwritng? From reeds@gauss.att.com Wed Feb 12 13:04:11 1992 Message-Id: <9202120404.AA18355@rand.org> From: reeds@gauss.att.com Date: Tue, 11 Feb 92 23:04:11 EST To: voynich@rand.org Subject: Dee again Status: OR Today I got my application form for buying a VMS copy from Yale: it took those people two months to answer my letter. Included was the usual sample page (an 80% reduced smudgy page showing ff. 99v and 100r, about as legibile as the picture on p 96 of Brumbaugh). Depressing if this is what they sell as prints; I hope John will soon report that the quality of prints he received in the mail today is better than this. But also included was a copy of three pages of a printed catalogue (of the Beinecke Room, I suppose) describing the VMS. It includes a much more detailed collation of folios into quires than I have seen before (with only a few obvious errors), a brief description of the contents and provenance of the MS, and a list of supplementary stuff included with MS 408 in the Beinecke Room: A, the Marci letter B, Voynich/Newbold correspondence, Voynich/Peterson (sic) corresp. C, a cardboard tube full of press clippings, from the Kraus era D, a scrapbook of clippings, from the Voynich era E, misc. handwritten notes by W. Voynich F, misc notes & corresp of Nills (sic) G, misc notes by E. Voynich & Nills H, box of photostats I-L, misc papers; K containing the 1976 Seminar report M, misc corresp of Brumbaugh N, folder of photstats The suprise was in the provenance paragraph: ... It is very likely that Emperor Rudolph acquired the manuscript from the English astrologer John Dee (1527-1608) whose foliation remains in the upper right corner of each leaf (we thank A. G. Watson for confirming this identification through a comparison of the Arabic numerals in the Beinecke manuscript with those of John Dee in Oxford, Bodleian Library Ashmole 1790, f.9v, and Ashmole 487). See also A. G. Watson and R. J. Roberts, eds, *John Dee's Library Catalogue* (London, The Bibliographical Society, forthcoming).... Also of note: the sample page and the catalogue excerpt were printed on Ellsberg paper, with dire warnings and foreign language incantations printed in blue ink on the back side. From jbaez@math.mit.edu Wed Feb 12 22:39:41 1992 Date: Wed, 12 Feb 92 08:39:41 EST From: jbaez@math.mit.edu Message-Id: <9202121339.AA23681@paley> To: voynich@rand.org Subject: Size of Voynich pages (the Yale copy) Status: OR Jim Reeds points out that the official size of the Voynich pages is 16 by 22.5 cm, so the diagonal ought to be about 27.5 cm. (I will take his word on that calculation, presuming that his math teachers drummed the Pythagorean theorem into him as fervently as I do to my students.) This agrees with what I measure on my reproduction from Yale: between 27 and 27.5 cm (depending on the page). The quality of the photocopy is not thrilling; some pages are better than others, but of course I have no idea how legible the original is. I find it acceptable but suspect that the copies I make will be noticeably worse (since there seems to be some "knee" past which nth-generation xerox copies degrade rapidly). I strongly urge everyone to get xerox copies from Yale instead of me. On the bright side, in most cases they left enough room near the binding so that xeroxing it probably won't be so hard. Only the foldout pages will require extra care. I really don't think we should get too grumpy with Yale. They essentially assmbled for me a decently bound book, with occaisional screwy foldout pages, for only $35 including shipping. I am sure they are very overworked and (owing to Yale financial woes of late) understaffed, so be kind. jb (more soon) From Karl.Kluge@G.GP.CS.CMU.EDU Wed Feb 12 16:32:00 1992 Date: Wed, 12 Feb 1992 16:32-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Need for care 6 vs. 8; font upgrade? Message-Id: <697930332/kck@G.GP.CS.CMU.EDU> Status: OR Looking at the available example I have in the office (D'Imperio p5 vs. Petersen f3r), it appears that Currier/D'Imperio consider the character a 6 rather than an 8 if the final downward stroke forming the right side of the lower loop is not curved to close the loop, but runs straight, even if there is no extension below the bottom of the loop. I've found the Voynich font very useful for checking my transcription against Petersen or photos, but the 6 and 7 don't seem to reflect our best knowledge of the "usual" appearance of these characters (in particular the 7/J distinction). I don't hack PS, someone might want to hack the font to reduce the cognative dissonance when proofing. From rcarter@isis.cs.du.edu Thu Feb 13 04:58:47 1992 Date: Wed, 12 Feb 92 12:58:47 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202121958.AA21881@isis.cs.du.edu> To: voynich@rand.org Subject: Photocopy arrived; Watson's ID of "Dee's foliation" Status: OR Well, got my copy of the VM from Beinecke this week; some impressions. Yeah, I got it in the torn envelope, VM in OK shape, state; no invoice. *sigh* Dunno, if I will ever get change from my $40... Photocopy quality as observed by others ain't the best, but no way will further generations be workable. The Voynich manuscript is *impressive*; quite a work. The large page foldouts (following f85r) are truly bizarre, even poorly reproduced. I am willing to start transcription as soon as I know where to start; I do want to digest some of this, though. I would suggest we work in `teams' of two, each person on a `team' transcribing the same page, with a compare of both results. Also, are we using a ? for characters we can't read? IE it would help me (at least) if there was a `formal' summation of the transcription `rules'; procedures, ?, Currier?, et al. RE: Watson's ID of "Dee's foliation"; I have problems with that; even though I only have access to just a few illustrations of Dee's writing, Dee seem to use a `platform' for his 1's; either a proper horizontal one (Harleian MS. 6986, fol.45, his letter to Qu. Eliz.) or making a `backwards s' style, as evidenced -throughout- his 1583 catalogue; he consistently joined his 1's and 5's (ie 15-); his 4's were almost 9's because of him using two strokes to form them, as opposed to the VM's three stroke 4's; this goes on; also remembering there is (so far) no evidence of Dee's `Jupiter' or `staicase' marks on the VM. Again, I think that the Dee/VM connection still is tenuous at best. OK. Actual hard analysis theorizing; I am beginning to think that all the `variations' of OFC89, OPC89, OFCC89, OPCC89, etc. are just that; variations of some sort; I am also thinking that these are possibly `conjugative' states of some sort; ie he saw, he will see, etc. Also, it seems that for some reason, LanA seems to want to add 89 to these sequences; perhaps LanB was more `experienced' and knew that in most cases the -89 was `understood' when used with OFC-. (Yikes, switch the LanA and LanB refs above). Also, to confuse things further, in LatAb 4OFC89 is: R? - ?O? - ?IS - C? - S - IS where the ?'s can be an assortment of vowels, when following a consonant, etc. P can be substituted for F, and S/Z can be substituted for C at times. Regards, Ron. From jim@rand.org Thu Feb 13 05:23:36 1992 Message-Id: <9202122023.AA08293@mycroft.rand.org> To: voynich@rand.org Subject: Transcription guidelines In-Reply-To: Your message of Wed, 12 Feb 92 12:58:47 -0700. <9202121958.AA21881@isis.cs.du.edu> From: Jim Gillogly Reply-To: jim@rand.org Date: Wed, 12 Feb 92 12:23:36 PST Sender: jim%mycroft@rand.org Status: OR > rcarter@isis.cs.du.edu (Ron Carter) writes: > I am willing to start transcription as soon as I know where to start; I > do want to digest some of this, though. I would suggest we work in `teams' > of two, each person on a `team' transcribing the same page, with a > compare of both results. Also, are we using a ? for characters we can't > read? IE it would help me (at least) if there was a `formal' summation > of the transcription `rules'; procedures, ?, Currier?, et al. The "team" idea is fine; ideally we wanted 3 people to do each page, but 2 is a good start. I've been coordinating the volunteer transcribers so far, and would suggest that you (Ron) start on 107r and send me each page as you finish it; I'll integrate it with "voynich.now", and will combine it with the matching transcription when it happens. Others with legible copies -- please feel free to volunteer for transcribing, and I'll suggest a place to start. The "rules" have been firming up gradually. For the current working version of the transcription (voynich.now in the ftp directory) we're using an extension to Currier notation. There are direct translations available into "frogguy" format, which has some advantages, as well as some influences on our Currier extension. Look at the transcriptions in voynich.now, especially those by Reeds, to get an idea of the flavor. More formally, extensions are as follows: o Page/line numbers look like at the beginning of the line. o {} are used for comments, and should be used for anything interesting or confusing. o * is used for unreadable characters. o [] is used for single-character alternatives; e.g. [2R] if you can't tell whether the character is a 2 or an R. o (|) is used for multiple-character alternatives; e.g. (CC|S) if you can't tell whether there's a ligature across the top of the CC. o For long ligatures (e.g. a P with one foot in two different S's) we've been using frogguy: cqt....cpt . o Use comments to mark and describe variant characters. o . is used for a "normal" group separator, where normal depends on the average spacing of the page. ',' is used for a smaller separation, which may be a real group separation or not. o .. is used when two groups are separated by a drawing; NOT simply for an extra-wide space, which may be noted with a comment. o We'd like to use Petersen's specifications for captions strewn about the page. I haven't seen them yet; perhaps Jim Reeds can expand on this. More to follow; and when we've got this firm I'll put it into the ftp directory. Jim Gillogly From Karl.Kluge@G.GP.CS.CMU.EDU Wed Feb 12 22:09:00 1992 Date: Wed, 12 Feb 1992 22:09-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Labels, labels, labels... Message-Id: <697950547/kck@G.GP.CS.CMU.EDU> Status: OR Someone was pining for Voynich Mss. labels, so here is a preliminary transcription of labels from zodiac folios and pharmaceutical folios. In order to fool Jim's filter program to create PostScript I used the format , where xxx is the page in the Petersen transcription, yy is the caption number assigned by Petersen (mod 100 if necessary), and l is A or B (A is default, B is for cases where Petersen has Xa and Xb as caption numbers). Page 146: Pisces f70v2 (boxed folio # in Petersen copy) Page 147: Aries f70v1 (boxed folio # in Petersen copy) Page 148: Aries f71r Page 149: Taurus f71v Page 150: Taurus f72r1 (boxed folio # in Petersen copy) Page 151: Gemini f72r2 (boxed folio # in Petersen copy) Page 199: Pharm. f88r Page 200: Pharm. f88v Page 201: Pharm. f89r Page 202: Pharm. f89v Page 223: Pharm. f100r Page 224: Pharm. f100v To my dismay, the labels from the pharmaceutical folios are not showing up in the zodiac folios that I've transcribed so far. The only repeats among the folios I've done so far are OFAE (151 #12, 151 #14); OFAE/89 (151 #5, 199 #6a with no space); OFAE9 (151 #13, 151 #21); OFARAJ (146 #6, 151 #15); and OPAE9 (146 #24, 147 #10, 199 #5). That's out of 206 labels. This bodes ill for my fine, sensible hypothesis that the zodiac folios gave sign correspondences for plants used in the pharmaceutical recipes and described one to a page in the herbal folios. That really sucks. Caveat emptor for these transcriptions. 6/7/8/J confusions are possible, only a few of the folios have been checked back against Petersen using the PS font. <------> <******> awk '//d' | sort <14601A> OPAR/AJ <14602A> OPAR/AE <14603A> OPAE/AR <14604A> OPAE/AJ <14605A> 8OEARAJ <14606A> OFARAJ <14607A> OPCO2AE <14608A> 2AEOE2 <14609A> OFAE/8AE <14610A> 9FOEAM <14611A> OPAEAE6 <14612A> 9FAR9 <14613A> OPAR <14614A> OP9 <14615A> OF9/O89 <14616A> OP9/AR <14617A> OFAEA <14618A> OPO89 <14619A> OPAE8 <14620A> OPAE8AR <14621A> OFA69 <14622A> OP92AJ <14623A> SXC9 <14624A> OPAE9 <14625A> OPAE/ARAR <14626A> OPAE89 <14627A> OFCOE9 <14628A> OF989 <14629A> OFCC2 <------> <******> <14701A> OFOE9 <14702A> OPAEAM <14703A> OPCO/AEOE2/ARAE9 <14704A> OPOOCC9/OPAE/OFCAEAR <14705A> OPCAR/ARA989 <14706A> OPAES9/PAR/AJ/69 <14707A> OBSC9/2AE <14708A> OPAFAIJAD <14709A> OFAEAE <14710A> OPAE9 <14711A> OAESC6 <14712A> OPSO8AE2 <14713A> OFOEZ9 <14714A> OPZZ89 <14715A> OPAE/9BZARAE <------> <******> <19901A> OPORSCP9 <19902A> ORAE <19902B> ORAE8 <19903A> OE8AR <19904A> OPOF9 <19905A> OPAE9 <19906A> OFAE89 <19906B> ORAJ <19907A> 8AR9 <19908A> OFOE <19908B> 2OROR9 <19909A> OP98A <19910A> OV92F98AE <19910B> OPOR/AJ <19911A> OVAE8O <------> <******> <20013A> OFAE98 <20013B> SCOQ9 <20014A> WCOR <20015A> OPAR/ARO89 <20016A> OPOFOE <20017A> OPORAJ <20017B> OPORA <20018A> SCO289 <20019A> OFAM <20020A> 8ARAJ6AE <20020B> OP98AR9 <20021A> OP8OR69 <20022A> 8ARAR6I ************Above this line checked by comparison of PS & Petersen********** <14801A> OPOES89 <14802A> OPOEOARAJ <14803A> OPC*OE <14804A> OPOES8 <14805A> OPAE6AR <14806A> OPCO2ARAR <14807A> OFE8AJ <14808A> OPCOAE89 <14809A> OPCOEAR <14810A> OFCOAE9 <14811A> OPAECF9 <14812A> OBAERAR <14813A> SCAR9 <14814A> OPCO/PC9/2AR9 <14815A> OPAEAE9 <------> <******> <14901A> OVA*VOJ <14902A> OPAEO89 <14903A> OPAEAM <14904A> OPAR/ZAR <14905A> ZOEZ69 <14906A> SAR/ORO7 <14907A> SVAE9 <14908A> OFOEAR <14909A> OPSO89 <14910A> AEW9 <14911A> OPAM <14912A> OFARAM <14913A> OPAR/AR/AE9 <14914A> OBAEAR/AJ/6AD <14915A> OBAEAR/AR <------> <******> <15001A> OVARAEAR <15002A> OPSOZ9 <15003A> OPS8AE <15004A> OFCC9/AR9 <15005A> OPAN9 <15006A> OZO8A89 (may be ...O89) <15007A> S8AU/8AN9 <15008A> OAM/AR/AR9 <15009A> OFAEAJ <15010A> 9PAEZ89 <15011A> SAR/AEIV <15012A> OPARAE89 <15013A> OPAM/OPAN <15014A> OPAECV/92/ANAJ <15015A> OSOEZARAJ <------> <******> <15101A> OPAE <15102A> 2AEAE <15103A> OFAJ <15104A> OPAEZ9 <15105A> OFAE/89 <15106A> SO2AR <15107A> OPAJ <15108A> ANAE9 (AIDAE9? -- parts of N well separated) <15109A> OFARSA7 (...6?) <15110A> OPARAE69 <15111A> OFAEAR <15112A> OFAE <15113A> OFAE9 <15114A> OFAE <15115A> OFCC9/AR9 <15116A> OPCCAR9 <15117A> OFAT/69 <15118A> OFAT,SAJ <15119A> OFCAE <15120A> OPARCR <15121A> OFAE9 <15122A> ODAR9 <15124A> OF98 <15125A> OPOEAJ <15126A> OVS8A69 (...79?) <15127A> OFEAT69 <15128A> OFARAJ <15129A> OFAT9 <15130A> OFCAEAR <------> <******> <20123A> OFSZ9 <20123B> 4FOE <20124A> OE8AJ <20125A> OPOE89 <20126A> ARARSO8AM <20127A> 9F98 <20127B> SOE/SC2 <20128A> OPORAN <20129A> OFAM/8AD <20130A> 9FOY9 <20130B> 2AE8AJ <20131A> 298ARAR9 <20132A> 98W089 (W lacks right C) <------> <******> <20233A> O8OR9 <20233B> 8OE9 <20234A> OBSO2AJ <20235A> 2AEOMZCOE <20236A> OBSCOR <20237A> OPOE8 <20238A> OFOE/ZOE/89 (F = P? left loop broken) <20239A> OBSARAM <20240A> FORAN/9 <20241A> 2O8AR <20242A> SC92 <20243A> SCO89 <20244A> OVORAN <20245A> OFZ8SA2 <20246A> 9ORAN <20246B> APAFAE (A under stem of P) <20247A> OPAE29 <20248A> 9PARCJ <20249A> OPOEAROE <------> <******> <22315A> SO2AROZOE <22316A> 2OSORY9 <22317A> OPCAR <22318A> SOVAD9 <22319A> 2AR/SAR/8AM89 <22320A> O2ADO (From photo in Brum. -- Petersen has ORARO) <22321A> SAE2AN <22322A> 2OIP9 <22323A> 292AJ (From Petersen -- Brum. photo 2O2AJ) <22324A> 8AFOQ <22326A> OFCCO2/OROE <22327A> ZOXC9 <22328A> OESCOJ <22329A> OFOE2 <22330A> OPCOE <------> <******> <22431A> POES8 <22432A> SOE2 <22433A> OBSOR <22434A> 2OE29 <22435A> 2OECC2O2 <22436A> 9FSOS89 <22437A> 9FS89 <22438A> 8S89 <22439A> 8AE29 <22440A> OFSCOR <22441A> 9PSOE <22442A> 89FSAE <22443A> SO2/QARAE From reeds@gauss.att.com Thu Feb 13 08:22:09 1992 Message-Id: <9202122322.AA10082@rand.org> From: reeds@gauss.att.com Date: Wed, 12 Feb 92 18:22:09 EST To: voynich@rand.org Subject: K.K. vs PS Status: OR All my sympathy to Karl! I'd be delighted if someone took over the Voynich font. Anybody with even a passing acquaintance with PS? PLEASE? In addition to 6/8 and 7/J mixups (some of which might be fixed in the most recent version of my font, which I will send to Jim Gillogly for posting), the font lacks characters: the Frogguy c, t, p, q, l used to represent splits, as well as "picnic table" (inverted V with a bar) and a few others "Voynich dingbats". If I am stuck with responsibility for the V font, I think I will call for help in designing better letter shapes. I have a "style sheet" postscript program which prints out each letter about 1 inch square, against a graph-paper background. If you can supply me with freehand drawings of suggested improvements, drawn to the same scale on the same kind of background, I can try to cobble something together. But I'd really rather someone more expert than I took charge! Jim Reeds From j.guy@trl.OZ.AU Fri Feb 14 00:34:50 1992 Date: Thu, 13 Feb 92 10:34:50 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9202122334.AA16486@medici.trl.OZ.AU> To: voynich@rand.org Subject: , gallows, *sigh* Status: OR Waiting for my copy of Petersen, although stumped for lack of raw data, I cannot help writing down some thoughts brought about by reading about Marco Polo and by a Quechua phrase-book published by Lonely Planet (a steal at $3.95). Until recently Quechua, unlike Maya and Nahuatl, was not spelt properly, as its aspirated and glottalized consonants were often not distinguished from their plain varieties in writing. Ha, ha! What if Voynichian had a similar phonology? The inventors of the alphabet, copied on the model of Roman letters, surrounded by speakers of languages without glottalized consonants, perhaps even without aspirated consonants, would have had a hard time devising an orthography faithful to the language. I thought of the gallows letters, and of my pet Beneventan which I doggedly want to be a "t". occurs before, after, or around the gallows. It's the only letter that does that [Footnote 1]. In the captions (most of which I haven't seen), we find gallows as second letter, with as first. Think. Those gallows must be common consonants. There are four gallows. Say: p, t, k, q (back k), a common enough system. Perhaps then, marks aspiration when following a gallows and glottalization when pierced by one (or vice versa, of course). That gives us 12 consonants, a much more satisfying picture. The probable value of on its own, then, would be or glottal stop. Reluctantly, I must abandon my pet hypothesis. after all, is not "t". Robert Firth's saw right in his dream. I don't know how he came to suggest kappa or chi for , but chi (as in Scottish ch) could be: a velar or pharyngeal fricative perhaps, carrying to the inventors of the alphabet both the feeling of aspiration (as a fricative) and glottalization (as a deep-throat sound). Marco Polo now. I learnt that "Xanadu" was "Shangdu". Reminded me of an early crazy idea based on my tongue-in- cheek "two Chinese brought back by Marco Polo wrote the Voynich in Venice". Suppose the Voynichian spelling was influenced by Italian. Suppose Voynichian contained this common sound "ng" (as in "sing"). No such sound in Italian. The closest is spelt "gn" (palatal nasal, "ng" is velar, not palatal). So they spelt it "gn". Now, some Chinese dialects don't have "ng". It gets replaced by plain "n", or by the nasalization of the preceding vowel. What is the two-letter string you find in one language but almost never in the other one? 89. So 8="g" and 9="n". Stuffs up my beautiful demonstration that A=9, alas. As you see, I am lost. I really crave for a look at all those captions. Footnote 1. Bennett, p.192, gives an example of a gallows inside , which he transcribes: EHT (my ). But Currier's system does not list such a letter, and does not allow for its representation either. Did Bennett make it up? Only Petersen will tell me. Waiting for Petersen.... Probably will have to shell out $50 (mailing included) for Yale's mediocre-sounding xerox, too. oqpox! From reeds@gauss.att.com Fri Feb 14 04:25:14 1992 Message-Id: <9202131925.AA00359@rand.org> From: reeds@gauss.att.com Date: Thu, 13 Feb 92 14:25:14 EST Subject: Voynich transcription guidelines To: voynich@gauss.att.com Status: OR Here is my 13 Feb 92 revision of Jim Gillogly's recent transcription guidelines. Jim will place a copy in the ftp area on randvax, but I'm also broadcasting it. o Page/line numbers look like at the beginning of the line. o {} are used for comments, and should be used for anything interesting or confusing. o * is used for unreadable characters. o [] is used for single-character alternatives; e.g. [2R] if you can't tell whether the character is a 2 or an R. o (|) is used for multiple-character alternatives; e.g. (CC|S) if you can't tell whether there's a ligature across the top of the CC. o For long ligatures (e.g. a P with one foot in two different S's) we've been using frogguy: cqt....cpt . o Use comments to mark and describe variant characters. o . is used for a "normal" group separator, where normal depends on the average spacing of the page. ',' is used for a smaller separation, which may be a real group separation or not. o .. is used when two groups are separated by a drawing; NOT simply for an extra-wide space, which may be noted with a comment. o Captions and other non-linear text should be entered with page/locus numbers, such as . Each separate "phrase" should receive its own locus number and go on a separate line in the transcription file. o When transcribing a word ring look hard for a visual clue for where the writing starts. Sometimes you can see a little radial bar breaking the ring. Be very precise in your comments about where you start, giving compass- or clock- bearing of starting word. o Whenever possible use Petersen's locus numbers. In any case note whether you use Petersen's locus numbers or your own in a comment. Put in brief comments indicating where on the page the loci are, such as "row of 5 nymphs at top of page". o If you have Petersen, and can make out some of his marginal notes, put them in comments, too. Jim Reeds From Alexis_Manaster_Ramer@MTS.cc.Wayne.edu Fri Feb 14 05:01:56 1992 Date: Thu, 13 Feb 92 15:01:56 EST From: Alexis_Manaster_Ramer@MTS.cc.Wayne.edu To: voynich@rand.org Message-Id: <416933@MTS.cc.Wayne.edu> Status: OR Please take me off the list. From reeds@gauss.att.com Sat Feb 15 01:41:32 1992 Message-Id: <9202141641.AA22704@rand.org> From: reeds@gauss.att.com Date: Fri, 14 Feb 92 11:41:32 EST To: voynich@rand.org Subject: Voynich dingbats Status: OR While transcribing today I came across (1) my first "picnic table" ("Y" in the First Study Group alphabet shown on D'Imperio p.97), (2) a weird kind of compound of Currier A and J, and (3) another compound character renderable as Currier F followed by (stuck to) Frogguy t. The time has come, I think, for a systematic way of handling these mutants. I propose: 1. Someone draw up a sheet of the most common mutant glyphs, with (say) between 10 and 100 items on it. D'Imperio's page 96 might do for starters. Each glyph gets a unique short name or number. If possible the postscript font gets taught about these glyphs. This sheet of Official Voynich Dingbats gets distributed to all transcribers. 2. We invent a convention for using the multicharacter glyph names when we transcribe. So when we find (maybe once a page) a funny character, we enter it in the transcription like this OFOCC`picnic'9 or OFOCC${picnic}9 or OFOCC\picnic\9 or what have you. (None of these notations seems very elegant to me, but who cares: these things happen rarely enough, and cause disruption whenever they occur, anyway.) Once we have a set way of recording these glyphs we can change the typesetting filters to print them out nicely. Jim Reeds From rcarter@isis.cs.du.edu Mon Feb 17 03:53:38 1992 Date: Sun, 16 Feb 92 11:53:38 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202161853.AA14610@isis.cs.du.edu> To: voynich@rand.org Subject: Comparing VM to other herbals... Status: OR I have been comparing the flowers in the VM to something called the Badianus MS (Codex Barberini, 1552, Latin 241, Vatican Library) as reproduced in _The Badianus Manuscript_, EW Emmart, 1940, Johns Hopkins Press, Baltimore. The BM is interesting because it was first written in by an Aztec in Aztec, then translated to Latin by an Aztec. It then found its way to Europe, eventually ending at the Vatican. The BM is a medicinal of sorts; listing `recipes' for `curing' various ills; most striking are the drawings of the plants used in such curatives have a resemblence to the VM; especially in the general `look and feel' of them. The colors are vivid and fresh appearing, esp. the bright(!) crimson used for the plant titles, etc. Aztec is curious; with beginning and ending `tl's appearing, looking -just- like a Currier F, etc. Such fun plants as `Xiuhecapahtli', Axocotl (a/atl=water:xocotl=sour/apple:axocotl=water sour), and `Quauhtla Huitzquilitl' aka (Lat.) Cirsium sp. aka thistle (VM 6v?). I also find it curious that both the BM and VM (the first section) show -just- flowers; for instance, Gerard's Herball shows flowers and others. From Karl.Kluge@G.GP.CS.CMU.EDU Mon Feb 17 21:19:00 1992 Date: Mon, 17 Feb 1992 21:19-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Unix tool for label searches in the herbal folios Message-Id: <698379581/kck@G.GP.CS.CMU.EDU> Status: OR I hacked up a little shell script today to look for occurances of the labels I had transcribed from the zodiac and pharmaceutical folios in the D'Imperio transcription of the herbal folios. The script takes a file of Voynich strings (and comments, which get stripped out) and produces a regular expression for egrep which allows an optional space after each character, matches illegible characters to anything, and treats 6's and 8's as fungible. I haven't really examined the results in detail, but looking through the output for the pharmaceutical folio labels, the number of labels which had no match was running just ahead of the number of labels which did have matches (and don't forget, the matches ignore spaces in the text and can have prefix/suffixes). This bothers me much more than the failure of labels to match between the zodiac and pharmaceutical folios. On the other hand, OPAE89, OFCC2, OFOE9, OPAE9, and a number of the other labels may be actual words in the Voynich language -- they have matches (but only a handful) in the herbal folios. Here's the script for the benefit of anyone who wants to continue hunting for Voynich words using labels. Note that I've been using "/" as a space and "," as a possible space as per earlier transcription conventions. --------------------------------------------------------------------- sort labels | awk '/^<.....[AB]>/{print $2}' | sed 's.[,/]./?.g' | \ sed 's.[A-Z0-9].&/?.g' | sed 's,*,./?,g' | sed 's/[68]/[68]/g' | \ sed 's,[\[\]A-Z0-9/?.]*,echo Checking "&"; egrep "&" voynich,' | csh From Karl.Kluge@G.GP.CS.CMU.EDU Mon Feb 17 21:53:00 1992 Date: Mon, 17 Feb 1992 21:53-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Lines as functional units vs. spaces as not word separators Message-Id: <698381584/kck@G.GP.CS.CMU.EDU> Status: OR Sorry if this has already been answered in previous messages, but if the spaces in the mss. are not word separators, but lines always end on word breaks, would that explain the results that led Currier to suggest that the line was some sort of functional unit in the text? From jim%mycroft@rand.org Tue Feb 18 10:14:53 1992 Message-Id: <9202180114.AA01281@mycroft.rand.org> To: voynich@rand.org Subject: Cluster analysis and Voynich "hands" and "languages" Date: Mon, 17 Feb 92 17:14:53 PST From: Jim Gillogly Status: OR Mary D'Imperio has written up her work on cluster analysis of Voynich pages. Here's the abstract; the paper itself (less the graphs, which I haven't figured out how to type in) is available via ftp from randvax:pub/jim/cluster.dim If anybody wants to stick in the appropriate troff macros for typesetting, please feel free, and send me a pointer to the results. Jim Gillogly --------------------------- An Application of Cluster Analysis and Multiple Scaling to the Question of "Hands" and "Languages" in the Voynich Manuscript M. D'Imperio 4000 Cathedral Ave. NW #106B Washington, DC 20016 28 January 1992 ABSTRACT. This paper describes the results of an exploratory study of Currier's theory using cluster analysis and multiple scaling, and presents additional results of a second study with expanded data sets, using multiple scaling alone. A clear separation was found between pages assigned by Currier to his statistical "languages" A and B. Biological-B pages formed the most compact cluster in all the plots. There was little indication of a separation between pages from different sections of the manuscript with different kinds of pictures. There was some evidence for a separation of hand 4 from hand 1 within the herbal section, language A, and hand 5 from hand 2 within the herbal section, language B. Suggestions for further research are offered. [This paper is an extensively revised and updated version of an earlier paper in an in-house technical journal, dated 20 June 1978. It includes corrected and expanded data sets.] From firth@SEI.CMU.EDU Wed Feb 19 03:30:02 1992 Date: Tue, 18 Feb 92 13:30:02 -0500 From: firth@SEI.CMU.EDU Message-Id: <9202181830.AA11242@bp.sei.cmu.edu> To: voynich@rand.org Subject: Letter Assignments Status: OR Folks Well, Mr Guy has reopened the issue of letter assignments. His ideas are, as ever, interesting. Here are a few comments of my own. 1. Vowels. I think we are pretty close on the vowels, except for two points. First, I'm not sure that Guy-Voynich 'x' is "u". From the way it prefers to be close to other vowels, especially our presumed "o", it looks nore like english "w" or spanish "y". It's a modifier of some kind. Secondly, I'm going to go back and resurrect my evidence that group-final '9' is not just a vowel, but a genuine morpheme. 2. Consonants. The key must be in the consonant clusters; the problem is that different languages suggest different keys. In english, for instance, we have s,t,h, st, sh, th, but not sth. In greek, we have st, th and sth, but not sh. So, even if we decide the gallows letters are "t", "p", "d", "b", or whatever, that still doesn't tell us what consonants can fuse to left or right. Depending on language, we could expect pt, kt, nt, ft, ... well, many things. 3. The letter '8' as "n". Yes, I've been looking for an "n". It's a fascinating idea, because the most significant languages with an initial n+consonant are the African languages. We have nd, nk, nb nt and nj at least. That mountain, for instance, isn't ki-li-man-ja-ro, as the europeans always say; it's 'kilima njaro', "white mountain". However, there's no need to be that recondite. NG is a good consonant in several celtic languages; in one form of Ogham, it's a single letter. And there's some evidence that the old pronumciation of latin GN was, indeed, NG, thus 'gnatus' was 'ngatus', "nyatus" almost. 4. And finally, '89' as "ng". Sorry, I don't believe it. The pair should appear more often initially than finally, and the evidence doesn't suggest that. If Voynich were latin, I'd be tempted to try "nus". But it isn't - that way lies Brumbaughville. Thanks for the definitive transcription rules. Now I can start on the assignment. Robert From Karl.Kluge@G.GP.CS.CMU.EDU Tue Feb 18 23:34:00 1992 Date: Tue, 18 Feb 1992 23:34-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Dee's book burning of Apr. 10, 1586 Message-Id: <698474074/kck@G.GP.CS.CMU.EDU> Status: OR I dug up the Josten article in the '65 Journal of the Warburg and Courtold (sp?) Institutes. Here is Dee's description of the books involved: "The voice: 'Gather your books together (indeed those which stand up (or have a spine)), from the first to the last, and also the one you have now in your hands (with the exception only of the action of the present day) and place them here before me.' "Dee: 'What shall I do (o Lord) with today's action?' "The voice: 'Cut it out of the book...Sunder the books apart, each by itself and place them on this table.' "I sundered the books...[and] added the volume which was at the time in my hands, and so they were 28 in all...And in one of those 28 volumes there were 48 individual books, most mystical and (as God Himself is witness) more valuable than the value of all things in the whole world might be accounted (footnote: This book remains unidentified)....There was another volume (among the 28) which was ancillary to those 48 individual books; its title was '48 Claves Angelicae', and it was written entirely in the angelic language (footnote:...The original, in Dee's handwritting, is now MS. Slaone 3191, fols. 1-13....). In another volume (of those 28) we had a most clear interpretation thereof in our English tongue. In addition to those 48 individual books there was a very short one, which book was the Mystery of Mysteries and the Holy of Holies; for it alone contained the profoundest mysteries of God himself and of the Almighty Divine Trinity that any creature will ever live to know. Its key was not yet granted to us. "Another volume (among those 28) contained that wisdom and science, with which Enoch (by God's will) was imbued; where there was also something agreeing with the testimony of the Apostle Judas (about the prophecy of Enoch). (footnote: Cf. Jude, 14-16).... "Dee: 'Have I now (o Lord) entirely fulfilled Thy command?' "The voice: 'You (Kelly) shall bring hither also the little black bag, the book, and the powder which you have hidden.' "Dee [in margin: To the pious reader]:...After these words he suddenly rose from his seat, went quickly away, returned and brought that little black bag containing the book and the powder aforementioned and placed them on the tabel beside the 28 books....The capacity of the little bag was such that it seemed to me apt to contain one peck of wheat. The book (a volume comprising two individual books on different subjects) and that holy powder had been assigned by God not only to the two of us..." Would the Vonich fit in a cloth bag "apt to contain one peck of wheat"? Does the statement as to the volume being composed of two books imply that Dee could read the book, or is this just his reaction to the division of subject matter apparent in the Voynich folios? Why does Whitby refer to the Book of Dunstan as a scroll? The only other candidate would appear to be the book described as "very short" whose "key was not yet granted to us", but the Voynich hardly qualifies as "very short". Only seven of the 29 books get individual mention -- was the Voynich among the other 22? WRT the Book of Soyna, Dee states that "Zadzaczadlin was Adam by the Alphabet thereof", implying that he could read it. If we accept the identification of the Voynich with "a book...containing nothing but hieroglyphics; which book his [Arthur Dee's] father bestowed much time upon, but I could not hear that he could make it out...." (Thomas Browne, citing Arthur Dee) then the Book of Soyna can't be the Voynich. Don't you hate when it's midnight and you can't remember if commas and periods go inside or outside the quotation marks? From reeds@gauss.att.com Wed Feb 19 10:09:14 1992 Message-Id: <9202190117.AA17916@rand.org> From: reeds@gauss.att.com Date: Tue, 18 Feb 92 20:09:14 EST To: voynich@rand.org Subject: Voynich ha ha Status: OR Sez Robert: >> ... In english, for >> instance, we have s,t,h, st, sh, th, but not sth. Aesthetes flee the neurasthenic Isthmus for the oasthouses of Rosthwaite, posthaste, while asthmatic Esther does calisthenics on the masthead! Jim Reeds From reeds@gauss.att.com Wed Feb 19 12:12:52 1992 Message-Id: <9202190313.AA19818@rand.org> From: reeds@gauss.att.com Date: Tue, 18 Feb 92 22:12:52 EST To: voynich@rand.org Subject: Voynich vowels Status: OR In addition to the Sukhotin/Ohaver vowel spotting methods, there is the Moler/Morrison SVD method, described in a Cryptologia paper by Bruce Schatz, anthologized in Deavour's "Cryptosomething, Past and Present", as well as in a samizdat by Moler. (Precise references when I get to the office in the morning.) It classifies letters into v, c, or n. I forget whether n means 'definitely a semi vowel' or if it means 'method cannot tell'. At any rate, I tried it out on the A and B corpora, using digraph tables which did not ("breaks") and which did ("runon") count word-break-spanning digraphs: A breaks A runon B breaks B runon 2 c 2 c 2 c 2 c 4 c 4 c 4 n 4 n 8 c 8 c 8 c 8 c 9 v 9 v 9 v 9 v A v A v A v A v B c B c B c B n C n C n C n C v E c E c E n E n F c F c F c F c I n I c I n I n J c J c J n J n M c M c M c M n N c N c N c N n O v O v O n O n P c P c P c P c Q c Q c Q c Q c R c R c R c R n S n S n S n S c T c T c T c T n V c V c V c V n W c W c W c W c X c X c X c X c Y n Y n Y n Y c Z c Z c Z n Z c Bruce Schatz, ``Automated Analyis of Cryptograms'' Moler/Morrison vcn spotting Jim Reeds From reeds@gauss.att.com Wed Feb 19 23:13:51 1992 Message-Id: <9202191414.AA27361@rand.org> From: reeds@gauss.att.com Date: Wed, 19 Feb 92 09:13:51 EST To: voynich@rand.org Subject: Voynich posthumous ha ha Status: OR I wrote: > Sez Robert: > > >> ... In english, for > >> instance, we have s,t,h, st, sh, th, but not sth. > > Aesthetes flee the neurasthenic Isthmus for the oasthouses of Rosthwaite, > posthaste, while asthmatic Esther does calisthenics on the masthead! Firth replied by personal email > As I said. All the english examples in your posthaste post > do not contain 'sth' as a consonant. The other words are > greek in origin. Of course I was not giving any real counter examples to Robert's statement, and I did not intend my joke to be taken as a refutation. (Even though he is inconsistent: does English not "have" those sounds which occur in English words of foreign "origin"; is he not shifting from a statement about the state of current English phonology to one about the history of how we got there?) What is the story about Rosthwaite, by the way? That it is really Norse not English, or that its really pronounced Russett or Chumley not Ross- Thwaite? At any rate, wish I was there, lifting a jar, rather than in in New Jersey at the moment! Jim Reeds From firth@SEI.CMU.EDU Thu Feb 20 00:27:48 1992 Date: Wed, 19 Feb 92 10:27:48 -0500 From: firth@SEI.CMU.EDU Message-Id: <9202191527.AA13258@bp.sei.cmu.edu> To: reeds@gauss.att.com, voynich@rand.org Subject: Re: Voynich posthumous ha ha Status: OR Jim Sorry I took your joke seriously. Words imported from foreign languages are always a problem. On the whole, I tend to ignore them when discussing linguistic issues, since they seem to me not in the natural development of the language. Feel free to disagree. Rosthwaite is ros + thwaite, and is pronounced "russet", near enough. At least, I think so -it's a long time since I lived in that part of the world. Robert From Karl.Kluge@G.GP.CS.CMU.EDU Wed Feb 19 21:35:00 1992 Date: Wed, 19 Feb 1992 21:35-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Re: The "calendar" Message-Id: <698553322/kck@G.GP.CS.CMU.EDU> Status: OR Hmmm...if this is some sort of idealized calendar with uniform 30 day months (astrologically more sensible perhaps), then the labels may be spirit names corresponding to days, in which case the (partial) format of the Mss. may be: Herbal, with each page devoted to a single plant, giving a picture and the names of spirits on whose days it is propitious to gather the plant. Calendar, with spirit names for each day Pharmaceutical folios -- plant parts labeled with plant names That would explain the occurance of labels from the zodiac and pharmaceutical folios in the herbal folios, but the absence of matches between the labels in the zodiac and pharmaceutical folios. Here's a interesting Voynich word for you: OE8AJ. f89r, locus 24. Occurs 13 times (ignoring spaces) in D'Imperio's transcription of the herbal folios, 10 of them at the end of a line. Here are some labels from the pharmaceutical folios which have only one or two matches in the D'Imperio herbal transcription. I plan to check the plant appearances in my Petersen, but as the herbal drawings are faint in the 2cd generation copies, I'd appreciate someone with a Yale copy checking if the plant part in the pharmaceutical folio matches the plant(s) on the herbal page(s). O8OR9. (f89v, locus 33) and (page 21 [f11r], line 2) 2O8AR: (f89v locus 41), (page 103, line 7 [f53r]), (page 151, line 40 [f77r]) ZOXC9: (f100r, locus 27), (page 75, line 6 [f39r]) OFSCOR: (f100v, locus 40), (page 8, line 11 [f4v]), (page 31, line 6 [17r]) *If* they match, then we can worry about confirming the plant identifications suggested by Petersen and getting names for those plants in every plausible language the Mss. might be written in. Perhaps this is the light at the end of the tunnel. Then again, it may just be an oncoming train. From reeds@gauss.att.com Thu Feb 20 07:18:13 1992 Message-Id: <9202192218.AA08970@rand.org> From: reeds@gauss.att.com Date: Wed, 19 Feb 92 17:18:13 EST To: voynich@rand.org Subject: Voynich odds & ends Status: OR About Karl's recent bibliocaust note: A British peck is about 9 liters, so Kelley's bag might have been as big as one of those green book bags that were popular in the 1960's, but nowhere near as big as Santa Claus's. About my recent vowel article: The Bruce Schatz paper is: ``Automated Analysis of Cryptograms,'' Cryptologia, Vol. 1, No. 2, April 1977, pp 116-142, reprinted in ``Cryptology Yesterday, Today, and Tomorrow'', by Deavours, Kahn, Kruh, Mellen, and Winkel, Artech House, 1987. Schatz cites the samizdat ``Singular Value Analysis of Cryptograms'' by C. Moler and D. Morrison, my copy of which I seem to have mislaid. The recipe is: take the singular value decomposition of the digraphic count matrix. Let the row and column singular vectors of the second largest singular values be x and y. Then letter i is a vowel if x(i) >0 and y(i) < 0, is a consonant if x(i) < 0 and y(i) > 0, and is 'neuter' if x(i) and y(i) have the same sign. Jim Reeds From rcarter@isis.cs.du.edu Thu Feb 20 10:39:32 1992 Date: Wed, 19 Feb 92 18:39:32 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202200139.AA11792@isis.cs.du.edu> To: voynich@rand.org Status: OR OK, bear with me for this excursion: the differences in occurences of ending -89's (with V.B having alot; V.A having virtually none) has bothered me; in my form of logic, this means that V.A either just drops the -89's or replaces them with something else, or both. Following this logic(?) I bugged you stats people enough (thanks!) and got some possible leads; the chart below possibly shows that V.A (I make the case that V.A is a more experienced scribe) maybe uses OE in place of 89 in -certain- conditions; examine OE and it could be a -further- abbreviation of 89, ie just one circle of an 8 and a 9 without the tail. A further argument might be made to suggest that V.A used a S to signify CC and a Z to signify SC/CCC, with Z's `apostrophe' to show S (CC) with a third C on top... Yeah, this -is- a stretch(!) but where are all the -89's in V.A? V.A V.ANS V.B V.BNS SOE 424 433 50 50 ZOE 136 138 37 38 ZC89 0 0 396 399 ZCC89 0 0 65 66 Along the vowel/consonant line; what -if- an ending -9 -does- mean an ending -is? How does an abbreviated v/c combo analyze? As a v? As a c? Or an n? When taking notes, I constantly abbreviate the word `the' as `t'... How would my notes analyze? Regards, Ron (recouping from an endoscopy... Aaaarrrghhhh...) From RJB@U.WASHINGTON.EDU Thu Feb 20 15:16:00 1992 Date: Wed, 19 Feb 92 22:16 PST From: RJB@U.WASHINGTON.EDU Subject: The Petersen Ms Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: <996EB7AFB6BF05301F@MAX.U.WASHINGTON.EDU> X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR My copy of Petersen just got here, in excellent condition, an incredible object indeed. Much thanks to Jim Reeds. I intend to pore over it. --RichardB From reeds@gauss.att.com Thu Feb 20 16:24:54 1992 Message-Id: <9202200725.AA05987@rand.org> From: reeds@gauss.att.com Date: Thu, 20 Feb 92 02:24:54 EST To: voynich@rand.org Subject: Voynich nymphs and Petersen Status: OR One of the charms of Petersen, I think, is the way he has toned down the nymphs, dressing them all in body stockings! Jim Reeds From firth@SEI.CMU.EDU Thu Feb 20 23:09:03 1992 Date: Thu, 20 Feb 92 09:09:03 -0500 From: firth@SEI.CMU.EDU Message-Id: <9202201409.AA16779@bp.sei.cmu.edu> To: Voynich@rand.org, j.guy@trl.OZ.AU Subject: Re: Petersen. The "calendar" Status: OR Folks A brief follow-up to My Guy's look at the Voynich calendar. (1) Two of the zodiac emblemata are unusual. As he points out, Scorpio isn't a scorpion; my guess is a salamander. And Gemini seems to be a male/female pair, which is wrong: in classical mythology, they're Castor and Pollux. (2) The names of the houses are identical in greek and latin, except that libra ("scales") is zygos ("yoke"). The decans (three per sign) are, as he suspects, coptic. (3) If (a big if) the calendar is 12 months of 30 days each, then it's egyptian, fer sure. That was the calendar Sosigenes took to Julius Caesar; it was a roman idea to make the months uneven. However, the starting point is then wrong: the egyptian religious year began at the winter solstice, with the five intercalary days; the administrative year began, of course, with the nile flood. (4) The semitic month names are from the new babylonian calendar. Babylonian Hebrew Nisanu Nisan Ayaru Iyyar Simanu Sivan Dumuzu Tammuz Abu Av Uluru Tashritu Tishri Arakhsamna Heshvan Kislimu Kislev Tebetu Tevet Shabatu Shevat Adaru Adar Nisanu 1 was the new year. However, I feel any correspondence is unlikely, since this was a lunar calendar, with 7 intercalary months added every 19 years in the pattern of the Metonic cycle. A better guess, I think, would be the lunar/solar calendar described in the Book of Enoch, which I'll look up and post. (5) Finally, note that the month names were added by a later hand, and almost certainly by using the western astrological mapping of sign into month. That mapping has been wrong for over two thousand years, because of the precession of the equinoxes. It's the obvious guess, but we shouldn't assume it's right. You know, with all these clews scattered about, you'd think something, sometime, would crack open, just a little? Robert From j.guy@trl.OZ.AU Fri Feb 21 01:31:03 1992 Date: Thu, 20 Feb 92 11:31:03 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9202200031.AA25142@medici.trl.OZ.AU> To: Voynich@rand.org Subject: Petersen. The "calendar" Status: OR I have just received a copy of Petersen's manuscript from Jim Reeds. I find it quite remarkable. Petersen went to great trouble in being accurate. He sprinkles his copy very liberally with "sic" wherever he finds an unusual letter form or sequence. There is a "sic", for instance, highlighting a CCCC2 sequence; there are "sic" pointing to every occurrence of 8 written like (in my system) ; and there is a "sic" pointing to the last 9 of "OECCC9.OE.9.9" on folio 86r. I homed in on the "calendar", folios 70v to 73v, because the calendar was the first thing to be deciphered in Mayan, and a lunar calendar was the first and only thing to be identified in the Easter Island tablets. A crib. The month names are written in a sort of blackletter script, common in 15th-century manuscripts or thereabouts. Folio 70v: "mars", two fish (Pisces) with a Voynich caption oqpoxax, on which Petersen tried this substitution code: initial o = null qp M P o a i x r s a s ce "abril", a ram (looking more like a goat to me) "Mars" has 30 figurines in two concentric circles, "abril" has only 15. Folio 71r: "abril", with same goat/ram, and 15 figurines in two concentric circles. Petersen notes: "figures 3, 5, 6, 14 = men?" Folio 71v: "may", with a bull (looks a bit like a goat, too), 15 figures. Folio 72r: "may", ditto "yony", two robed figures holding hands, one in boots and short robe with tight sleeves (male?) the other in floor-length robe and wide sleeves (female); 30 figures. "jollet", with two crayfish; 30 figures. Folio 72v: "augst", a lion; 30 figures, a note: "fig. 13 = man?" "septe[m]b[r]", an angel; 30 figures. [m] is a macron over "te", [r] is superscript. "octe[m]bre", a pair of scales; 30 figures. Folio 73r: "nove[m]bre", a lizard, I'd say, although the author must have meant a scorpion; 30 figures. Folio 73v: "de[em]bre" (barely legible), a figure in boots and knee-length robe holding a crossbow; 30 figures. A note by Petersen: "R. Steele wrote "Nature" Oct.19 1928 crossbow man wears 15th cent. hat and covers part of an inscription". Folio 74 is lost. Is it really a calendar? If so, not ours: the months are all 30 days long. The year starts some time in March, reasonable enough. What is the language of months? Got me. I posted a query about it to Linguist@tamvm1. The hand in which the month names are written is compatible with the 15th-century garb of the crossbowman. Why aren't there any obvious month names in Voynichese? The crossbowman covers part of an inscription. Does that mean that the Voynich is 15th-century at the latest? I'd say so. Why are April and May split into two sub-months of 15 days each? Each "day-figure" has a caption. Names of patron saints? For each "month" Petersen lists the zodiac sign in Latin, Greek, and Hebrew, and gives a name to each decan, which seems to me to be Coptic or Ancient Egyptian (written in Greek letters). I'll spare you the Hebrew: Aries: Krios: Khontare, Khontakhre, Seket Taurus: Tauros: Khwou (w= omega), Erw, Rombomare Gemini: Didomoi: Thosalk, Ouare, Phonor Cancer: Karkinos: Swthis, Sit, Khnoumis Leo: Lewn: Karkhnoumis, Eepee, Phounee (ee= eta) Virgo; Parthenos: Twm, Ouesteukati, Aphoso Libra: Zugos: Soukhwe, Pteekhout, Khontare Scorpio: Skorpios: Spokhneene, Sesme, Sisme Sagittarius: Toxotees: Reeouw, Sesme, Komme Fascinating, but little help. I'm not sure at all it's a calendar. If it is, it must have had 12 months of 30 days. Petersen has also pencilled month names, not too legible, in a language I do not recognize, but am sure is Semitic: Nisan (or Niran?), Ijjar, Siwan, Tammuz, Ab (?), Elul, Tishri, Marheshwan, Kislew, Tebeth. Hebrew? Arabic? Syriac? I must have a book, somewhere, where I can find it. I'll have a close look at those captions next to the day-figures. I'm not very hopeful, though. A last thought before signing off. On the Yale microfilm, I remember that the zodiacal sign pictures looked very much in the same style as the Voynich nymphs and dames. So perhaps we should take those month names as efforts by the Voynich authors to match their calendar with ours. What makes me think that is that there is a late (1936) Easter Island manuscript where the author is trying to reconstitute the traditional Pascuan calendar with the one he knows: ours, with Tahitian names for the months (the Tahitian month names are themselves derived from English). Now, if we could only identify the language and times of those month names.... From cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET Fri Feb 21 01:40:30 1992 Message-Id: From: cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET (John Cowan) Subject: Re: Petersen. The "calendar" To: voynich@rand.org (Voynich List) Date: Thu, 20 Feb 92 11:40:30 EST In-Reply-To: <9202200031.AA25142@medici.trl.OZ.AU>; from "Jacques Guy" at Feb 20, 92 11:31 am X-Mailer: ELM [version 2.3 PL11] Status: OR frogguy writes: > Petersen has also pencilled month names, not too legible, in > a language I do not recognize, but am sure is Semitic: Nisan > (or Niran?), Ijjar, Siwan, Tammuz, Ab (?), Elul, Tishri, > Marheshwan, Kislew, Tebeth. Hebrew? Arabic? Syriac? Hebrew by courtesy, but originally Babylonian. Anyway, the month names of the Jewish calendar. w->v here. "Nisan" is correct. -- cowan@snark.thyrsus.com ...!uunet!cbmvax!snark!cowan e'osai ko sarji la lojban From reeds@gauss.att.com Fri Feb 21 02:19:25 1992 Message-Id: <9202201719.AA14252@rand.org> From: reeds@gauss.att.com Date: Thu, 20 Feb 92 12:19:25 EST To: voynich@rand.org Subject: Voynich calendar Status: OR Following Firth and Guy, > (1) Two of the zodiac emblemata are unusual. As he points out, > Scorpio isn't a scorpion; my guess is a salamander. And > Gemini seems to be a male/female pair, which is wrong: in > classical mythology, they're Castor and Pollux. But aren't these alchemical, with salamander = calcination, and m/f pair = conjunction? Maybe we have a calendrical metaphor for the processes of alchemy, based on a 12 by 30 year, somewhat similar to the 7 day metaphor in the "Alchemical Wedding"? Jim Reeds From firth@SEI.CMU.EDU Fri Feb 21 03:26:07 1992 Date: Thu, 20 Feb 92 13:26:07 -0500 From: firth@SEI.CMU.EDU Message-Id: <9202201826.AA17235@bp.sei.cmu.edu> To: reeds@gauss.att.com, voynich@rand.org Subject: Re: Voynich calendar Status: OR Jim An alchemical zodiac! What a fascinating idea. Yes, in that case gemini would represent the coniunctio. The salamander, however, represents catalysis, surely - on of Paracelsus' emblems for the iliaster was a pair of salamanders. Another intriguing thing I just noticed: all the four-footed signs (including gemini) have just one foot raised - is this a hint? Oh, well, back to the Mutus liber... Robert From reeds@gauss.att.com Fri Feb 21 04:02:01 1992 Message-Id: <9202201902.AA16804@rand.org> From: reeds@gauss.att.com Date: Thu, 20 Feb 92 14:02:01 EST To: voynich@rand.org Subject: Voynich calendar Status: OR Firth: >> Yes, in that case gemini would represent the coniunctio. >> The salamander, however, represents catalysis, surely - >> on of Paracelsus' emblems for the iliaster was a pair >> of salamanders. Salamander=calcination was an ignorant guess of mine, based on a hasty glance in some Shumaker book; deeper and wider knowlege of alchemy than mine is needed here. Maybe color will help pin down an alchemical interpretation of the V zodiac? (My wife commented, when I said that I could not make the alchemical sequence mesh with the calender sequence: ``Maybe *that's* why they never could make gold.'') Jim Reeds From rcarter@isis.cs.du.edu Fri Feb 21 05:00:55 1992 Date: Thu, 20 Feb 92 13:00:55 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202202000.AA20702@isis.cs.du.edu> To: voynich@rand.org Subject: Herbal/text matchings... Status: OR Worked out the set of four herbal/text matchings someone was looking for: O8OR9 (f89v,loc33) and (f11r,lin2) O8OR9 = 4O8OR9 (Might be a `jar name') 2O8AR (f89v, loc41) and (f53r, lin7) and (f77r, lin40) 2O8AR = 2O8AR = 2O8[AO]R ZOXC9 (f100r, loc27) and f39r, lin6) ZOXC9 = ZOXC9 (Might be a `jar name') OFSCOR (f100v, loc40) and (f4v, lin11) and (f17r, lin6) OFSCOR = ZOFSCOR = 4OFSCOR So, four for four it seems; it -is- interesting that the `single' match hits (ie O8OR9 and ZOCX9) are what appear to be `floating' (ie not attached) `jar names' (the malady(?) the mixture is for?) while the other two are multiple hits... Good luck on doing matches or ID on 2O8AR (some kind of root?) and OFSCOR (pretty generic) as they are both virtually nondescript; as I have stated before, the VM illustrator was -not- an artist, and I suspect (strongly) we are looking at multi-generation (copies of copies etc.) drawings of plants the illustrator has never seen; why else the titling? And of course the only thing I can possibly really identify (f102r loc202 pg229) is one without a label (carrots!?). From j.guy@trl.OZ.AU Fri Feb 21 05:14:01 1992 Date: Thu, 20 Feb 92 15:14:01 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9202200414.AA25444@medici.trl.OZ.AU> To: voynich@rand.org Subject: the calendar Status: OR Karl Kluge says: "if this is some sort of idealized calendar with uniform 30 day months...". Not necessarily idealized. The Mayas had a year of 12 months of 30 days, plus 5 or 6 "leap days". That was their civil calendar, the "haab", distinct from their religious calendar, the "tzolkin" of 18 months of 20 days. I vaguely remember that the Egyptians had a calendar of 12 months of 30 days --- vaguely, because all my references are at home. The Romans... what did they have? Anyway, the Voynich "calendar" does not seem to be lunar. If it were, we would have 29 and 30 days alternating pretty regularly. The interesting part is the month names in that European language. "Octember" ought to be a dead giveaway. "Yony" is reminiscent of Modern German "Juni", and "jollet" of French "juillet". "Abril", on the other hand, is Spanish! OE8AJ? Currier's alphabet is on my desk at home.... Oh, must be . Looks unusual to me indeed. And mostly at the end of lines.... this language moveth in mysterious ways. Yes, it would be nice if we could identify some of those plants and use that as a lever into the language. But, if the MS was written, as I think is likely, 500 years ago, remember that languages do evolve. Half of the plant names in that language today are probably different of what they were 500 years ago. (Well, not half, hopefully perhaps 10 or 20%). From reeds@gauss.att.com Sun Feb 23 09:55:39 1992 Message-Id: <9202230055.AA05640@rand.org> From: reeds@gauss.att.com Date: Sat, 22 Feb 92 19:55:39 EST To: voynich@rand.org Subject: Voynich bookology Status: OR I just looked at a very elegant work, "Pen to Press" by Sandra Hindman and James Douglas Farquhar, Johns Hopkins, 1977. One chapter, "The manuscript as a book" gives an especially clear description of manuscript production stages, gatherings of vellum leaves, how to detect out-of- order leaves, etc, etc. Well worth reading. Another interesting thing about this book is that it is set in a very unusual typeface, Bernhard Modern, which seems in general flavor very much like the Voynich script. Jim Reeds From jbaez@math.mit.edu Mon Feb 24 01:34:15 1992 Date: Sun, 23 Feb 92 11:34:15 EST From: jbaez@math.mit.edu Message-Id: <9202231634.AA10764@paley> To: voynich@rand.org Subject: Miscellaneous Voynich Progress Status: OR I am very busy these days and expect to be so for some time - that's why I've had such a low profile on the Voynich scene lately. Here are 3 bits of news, though. 1) I have decided that the quality of xeroxes I can make from the Yale copy of the Voynich is so low, and the difficulty of xeroxing the whole book (with oversized pages and a fair number of "fold-out" pages) is so high, that my original plan of making a batch of copies and sending them out to everyone isn't a good idea. I feel sort of bad about this. What might work instead is this. If you have particular pages that are very frustrating, I can make copies of them and send them to you, or just stare at particular inscrutable characters and see what I can make out. The real Voynichomaniacs should probably order their own copy from Yale, which is fairly cheap, and painless except for the long wait. I welcome comment on this problem. 2) The rare books room at Wellesley has okayed Nate and I to photograph a 14th-century vellum ms of theirs, with dense text that's faded in patches, in the infrared. This technique is known to make certain faded text legible (I've seen a book where this was applied to the Dead Sea Scrolls). The longer- term goal is to do this to the Voynich. There are a number of places, including the mysterious final page, where this might turn up something interesting. 3) Looking through the Voynich with Nate I caught on to the rather clear difference between hands A and B and their distinct personalities. More on this in a minute. From jbaez@math.mit.edu Mon Feb 24 01:58:02 1992 Date: Sun, 23 Feb 92 11:58:02 EST From: jbaez@math.mit.edu Message-Id: <9202231658.AA10825@paley> To: voynich@rand.org Subject: Hands A and B Status: OR Well, most of this can easily be seen from the information available in the "foliation" database, but it didn't strike me until browsing through the manuscript. Hand A writes rather large letters, loosely spaced, and is given to fancy flourishes in his gallow letters. Hand B on the other hand :-) writes in a rather cramped way, with smaller letters, and is not so much given to flourishes. The final folios, 103-116, which are the only ones completely devoted to text (unless one counts the pictures of stars, which seem more like paragraph headers than illustrations), are in hand B. (Note: this what it looks like to me, I haven't checked the "official" record, since I want to learn the difference myself). Hand A seems much more eager to leave lots of blank space on the page. All this would seem to portray A as a free-wheeling, expansive fellow and B as a perfectionistic, constipated sort, BUT it seems to be the case that ALL THE NYMPHS OCCUR ON B's part of the text. One especially dramatic instance of this is on the last page, with its mysterious "key" -- I guess this is folio 117v -- which occurs right after the folios completely devoted to text. It was probably written by B, since the writing looks like B's and it occurs after a bunch of B. AND, it has one last little nymph on it! I'm not sure what to make of this but it seems to indicate that this is not a monolithic work, perhaps copied by two scribes, but a joint creation expressing two distinct personalities. Any explanation of the Voynich would have to account for this (if it's right). Just as speculation, I lean towards the whole thing being an elaborate "joke" or money-making trick, with A and B free to do whatever they wanted as long as it was in Voynichese. If this were the case, it would not be the statistical *differences* in A's and B's text that would need explanation, but the *similarities* -- for why would there be any? (Agreement beforehand - or just a result of looking at each others' work -- in the latter case one would expect the early folios (temporally speaking, not in terms of folio number) to be less correlated.) jb From j.guy@trl.OZ.AU Thu Feb 27 00:22:33 1992 Date: Wed, 26 Feb 92 10:22:33 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9202252322.AA00832@medici.trl.OZ.AU> To: voynich@rand.org Subject: Months, oddities, unpronunceable Voynich Status: OR MONTHS I have received this answer about the month names: --------------------start of quote------------------------------------------ I just read your query on LINGUIST-L. This is reaching deep in the recesses of my memory, since I have not been a student of Slavic linguistics for more than 20 years, but your list of month names seems definitely Slavic. One of the characteristics of the Slavic languages is the analogical recasting of *October*as *Octember*. I don't recall the Voynich manuscript specifically, but I would guess it is West Slavic. If you don't get any more informed responses, let me know and I will see what I can find. I still have quite a collection of books on Slavic linguistics. Steve Seegmiller Seegmiller@apollo.montclair.edu ----------------------end of quote-------------------------------------------- ODDITIES With Petersen's faithful copy and his highlighting "sic"'s all over I have noted scarcely a folio without some kind of odd letter group(s) that do not come out in the transcription. One among many others: the C-part of S transperced by gallows is very often written like an I when following A, and often when followinG O. In other words, if I used my transcription I'd write , not . I am trying to resist the temptation of writing a list of those oddities. UNPRONOUNCEABLE VOYNICH Even if the VMS is a fabrication, its language would have to have been pronounceable to account for the fluency with which it was evidently written. But, to get pronouceable Voynich, we end up with positing too many vowels and are left with too few consonants. A too common sequence I have noticed is S.gallows-in-S.S. Taking Robert Firth's dream seriously (and why not? Do you remember how this chemist whose name I forgot *dreamt* the benzene molecule before formally ascertaining it?), I am drawn back to these two hypotheses: 1. Voynichese is written in a Semitic-like writing system, where vowels are seldom written, and when written, are represented by letters which otherwise act as consonants. 2. Voynichese is written in an Indian-like writing system, where it is the absence of vowels which is written: consonants all carry a vowel, usually "a". For instance, in Balinese "manuk" (bird) is written "mnukz" (in which "z" = the squiggly letter that indicates the absence of a final vowel; "u" is written under the "n"). If so, then we have been completely confused in our tentative identifications of vowels (and I am the first culprit), and we must admit that Robert Firth's dream butler is quite right in taunting us with his salt and pepper shakers. A LAST THOUGHT Reduplicated words are extremely frequent. I have been struck by the high frequency of 8AM 8AM in the herbal. Malayo-Polynesian, and especially the languages of Indonesia, do that to a great extent, e.g. "burung-burung: birds", "ikan-ikan: fishes", regardless of length of word duplicated. Those languages are also written either with the Arabic alphabet (omitting many vowels), or with Indian- type alphabets. I cannot help wondering whether that 8AM 8AM is not "daun-daun" (leaves) or "bunga-bunga" (flowers) etc., or dialectal variants thereof. From rcarter@isis.cs.du.edu Fri Feb 28 16:28:19 1992 Date: Fri, 28 Feb 92 00:28:19 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9202280728.AA17558@isis.cs.du.edu> To: voynich@rand.org Subject: Wilford Voynich, Newbold, and Co. Status: OR Hmmm... I just got a copy of "A preliminary sketch of the Roger Bacon cipher manuscipt" by Wilfred M. Voynich which includes "The Roger Bacon Manuscript" lecture by Mary Scott Newbold, which was published in the 1921 Transactions of the College of Physicians of Philadelphia... Conclusions? Both Voynich and Newbold were F. Bacon freaks, which is probably why Voynich `endorsed' Newbold's `translation' of the VM. Either that, or they were both on the same kind of medication. For the most part, the entire `research' they both did, seems to me to be extremely speculative, and again stretching to get the VM anywhere close to Bacon, let alone Dee. It -is- depreesing to me that from the illus. from the 1921 trans. ^depressing is that the large foldout (that follows f85) is aanything but the faded photocopy I got from Yale; in the original it appears to as detailed and darkly inked as any of the best reproductions of the VM. >From working in ink myself, the foldout took a good week to do... Coming soon; some thoughts on snake cures, Aztec, and the VM... Ron. From jbaez@math.mit.edu Fri Feb 28 23:32:18 1992 Date: Fri, 28 Feb 92 09:32:18 EST From: jbaez@math.mit.edu Message-Id: <9202281432.AA19673@paley> To: voynich@rand.org Subject: Photographic adventure Status: OR I haven't heard any comments about my offer to send pages of the Voynich to those who want to transcribe 'em. Does this mean that all surviving Voynich fans have some sort of copy? Nate and I went to the Wellesley rare book room to try our hand at infrared photography of vellum manuscripts on Tuesday. Our sample was a 15th-century breviary, Brevarium Fratrum Minorum Franciscianum (?), that the librarian had selected because it had patches of faded text. Infrared (IR) photography is known to be able to make readable some passages of text that are faded to illegibility in the visible range, and has been used on the Dead Sea scrolls for this purpose. So we dimmed the lights in the library, got our standard photographic lamp out (so we'd have reproducible conditions), and started clicking. One must use a filter, and focussing in the infrared requires some fussing about as its refractive index is different from visible, so all in all we had a fun time. We bracketed exposure times so as to increase our chances of getting good pictures. I duly noted hairy versus smooth sides of the vellum, to see if they have different IR properties. (Indeed, folios were arranged so as to have both sides of equal hairiness.) There were some really faded passages and also some blank pages with mild traces of ink soaked up from the opposite page (it seems). (Looking at the Yale copies of the Voynich, it seems "bleedthrough" is quite common.) We got samples of all of these and also photographed them in the visible. Nate will develop them this weekend and I'll report on what we get. The goal, of course, is to eventually take a good look at that "oladabas" page at the end of the Voynich (among other things). Also, Nate wants to become a general-purpose archaeological high-tech whiz. Speaking of which, once we develop these pictures we have the option of running them through some image enhancement programs, which may help. jb From reeds@gauss.att.com Sat Feb 29 01:01:16 1992 Message-Id: <9202281608.AA11435@rand.org> From: reeds@gauss.att.com Date: Fri, 28 Feb 92 11:01:16 EST To: voynich@rand.org Subject: Voynich Oladabas, etc. Status: OR About the "oladabas" page. Going by general appearance of the last dozen or so folios, I would guess that the true end of the VMS is the "colophon" paragraph at the bottom of f116r, and that the "oladabas" on f116v is either scribbling by an early would-be decipherer (Kircher, say) or is some owner's shelf-mark & blurb, saying something like "This is the cipher manuscript called 'Oladabas for fun and profit', written in weird letters like OROR". Handwriting and inkological studies might tell whether f116v is of a piece with the rest of the MS, or a later addition. While glancing through my "Pen to Press" I found this gem by Farquhar: "A manuscript which has not been seen with ultraviolet light in order to detect erased writing has, in one sense, not been studied." Farquhar recommends Haselden's "Scientific Aids for the Study of Manuscripts" (Supplements to the Transactions of the Bibliographical Society, no 10), Oxford, 1935. Can John and Nate add UV to their repetoire? Jim Reeds From RJB@U.WASHINGTON.EDU Sat Feb 29 03:55:00 1992 Date: Fri, 28 Feb 92 10:55 PST From: RJB@U.WASHINGTON.EDU Subject: almost the indication of a possible development Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: <92BB5DD0B45F604761@MAX.U.WASHINGTON.EDU> X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I have been very busy recently, and still waiting for the Yale photocopy to get here (Petersen is here, and has whet my appetite: it's much better than the sample pages, but I have still not had a chance to try my hand at transcription). I did have a chance to show the whole thing (along with the Brumbaugh book) to a friend who has quite a background in paleography, typesetting, calligraphy, and related enterprises. His first impression, based on the Brumbaugh reproductions and the content of the graphics in Petersen (especially the zodiacal material) was that the ms was written, not transcribed, than some parts at least were written by someone just trucking along at speed, rather than doing letter-by-letter transcription, that the whole thing looked like mid15th to 16th century German or other central European (pre-Dee), that the hand looked almost like a cursive black-letter in places, and that it loked as though the pictures had been roughly sketched in, the text written, and then the pictures finished up and colored. He had questions about the preparation of the velum, the nature of the binding and gatherings, the nature of the colors used (suggesting nondestructive spectrographic analysis), to help narrow down provenance. He also had the impression that it looked like something written to be read, with content, and suggested that it might have come from some small community of say scholars who had developed a particular style of writing with links to contemporary writing styles but more or less deliberately not legible to outsiders. His contribution to the speculative tradition of Voynich commentary took off from the absence of male figures. Perhaps wht we have here is a surviving text of a convent of archeo-feminist inheritors of the pre-patriarchal traditions of old Europe... I provided him with a grab-bag of Voynich discussion files to look at, and tried to extract from him a promise to write up a series of questions to ask the manuscript. He gets back to New England a couple of times a year, and he became quite intrigued by the whole thing, so there is a possibility that something may come of it all. --rjb From BILAKO%TREARN.bitnet@CUNYVM.CUNY.EDU Mon Mar 02 23:59:57 1992 Message-Id: <9203021804.AA09435@rand.org> Date: Mon, 02 Mar 92 14:59:57 TUR From: Ahmet Koltuksuz Subject: Address of Aegean Park Press needed To: voynich@rand.org Status: OR Hello Team Voynich; Although it has been frequently mentioned in this list but I seem to be looking for the address of Aegean Park Press. I would appreciate if somebody let me have it. Thanks in advance. Ahmet KOLTUKSUZ E-MAIL : BILAKO@TREARN.BITNET From BILAKO%TREARN.bitnet@CUNYVM.CUNY.EDU Tue Mar 03 19:57:22 1992 Message-Id: <9203030901.AA28663@rand.org> Date: Tue, 03 Mar 92 10:57:22 TUR From: Ahmet Koltuksuz Subject: thanks for the info.. To: voynich@rand.org Status: OR Hello Voynich Team;; Thank you all for answering my request about Aegean Park Press. Special thanks go to : Jim Ron John I appreciated, thank you. Ahmet KOLTUKSUZ From WBRILL@macc.wisc.edu Wed Mar 04 01:54:00 1992 Message-Id: <22030310544217@vms.macc.wisc.edu> Date: Tue, 3 Mar 92 10:54 CST From: "Winston J. Brill" Subject: Any interest? To: VOYNICH@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Anyone interested in writing a draft monthly summary (no more than 600 words) of the Voynich group's adventures? Let me explain. My name is Winston Brill. I was a chair professor of bacteriology at the University of Wisconsin. I left the academic life to found an agricultural biotechnology company, Agracetus, in which I was VP of R&D. Three years ago, I started a new venture--consulting with companies on increasing productivity and creativity in R&D. I'll come back to that in a minute. How I got into the Voynich EMail system was through my son, Eric Brill, who is finishing up a Ph.D. in computational linguistics at the University of Pennsylvania. He told me about you Voynichers and how to get into your EMail interactions. It's been fascinating!! Sorry that my only contribution was to suggest the group name, "Voynich Voyeurs." Now, back to my consulting venture. I have had quite a successful career--membership in the National Academy of Sciences, a bunch of awards, and in 1989 Business Week named me as one of the top 10 innovative scientists. This isn't to brag; rather this is to show you why I'm making this request for the monthly Voynich summary. My successes, and my consulting, have given me some "authority" to teach "productivity and creativity in R&D." I will soon be publishing a monthly newsletter, "R&D INNOVATOR," to help researchers in any field, academic or industrial, basic or applied. Each month, there will be one article in which I discuss practical solutions to common problems. Another regular feature will be written by someone who has dwelled on the creative process. This includes psychologists, historians, artists, famous scientists, etc. A third feature is written by readers. It will focus on HOW a discovery came about, rather than the discovery, itself. Is this just a sales pitch to a bunch of Voynichers? No, this is a pitch to get a regular feature in the newsletter specifically on the Voynich progress. I'd like the feature to show the variety of disciplines, the variety of "characters" (I don't mean Voynich characters, but you the individuals trying to solve it), the dead ends, etc. In other words, the Voynich investigation could make for some interesting reading--but, focusing on the evolution to solution (or non-solution). It would have to be written for individuals who know nothing about linguistics, etc. Technical jargon will not be used. I have an editor who can polish up the rough draft that one of you would submit. One of the main points of such a Voynich article would be for me to demonstrate a productive collaboration among people who haven't even seen each other (at least most of you). And, that computers play a major role in advancing your common interest. In industry, very frequently, with high-powered computers and highly paid computer scientists, relatively few researchers spontaneously interact--even in the same company at the same site. So, I'd like to consider the ongoing Voynich story as a prod to the reader to see the value of collaboration, of interdisciplinary interactions, of using computers for efficient interactions, of dead ends stimulating new avenues, and even (perhaps) of dreams. I want the reader to look forward to receiving the monthly updates. (I certainly have enjoyed the almost daily input) I hope for several thousand subscribers within a year. This will be an international subscription. The Voynich story is one of several I am considering. I can only pursue one--and that will depend to some degree on the availability of an interested "writer." Rough drafts can be EMailed to me. While the articles should appear monthly, the material in each article should be from information that occurred at least four months prior to publication. The initial article will mainly tell the mystery surrounding the VM. All subscribers would get this initial summary. I would send final articles to each of you through the Voynich EMail address. As an indirect benefit, the Voynich puzzle may get some surprising input from a reader. E.g. a botanist could contact you to ask for the drawings. Who knows what kinds of ideas are out there. Maybe someone can help. But, the articles' function is to help researchers be more introspective about the way they go about their activities, and thus become more productive and creative. Anyone interested? My INTERNET address is: wbrill@vms.macc.wisc.edu My phone number is (608)231-6766 during working hours. From j.guy@trl.OZ.AU Sat Mar 07 05:00:37 1992 Date: Fri, 6 Mar 92 15:00:37 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9203060500.AA09725@medici.trl.OZ.AU> To: voynich@rand.org Subject: Miscellanea and oqpox! Status: OR Our link with the some of the outside world seems to have broken down. No netnews for 30 hours, and precious little e-mail. I had started hacking the "calendar", full of hope. Makes no sense as a calendar, but some of the labels are found again ten folios later or so, where you have those pictures of nymphs and plumbing and sauna baths. Have you noticed the note of the first page of the Petersen manuscript: "Photostats of the MS were brought to (illegible) at Catholic University by Mrs Voynich on April 20th 1931 -- and left with him for his (illegible). I had another set of photostats made for myself from those of Mrs Voynich on April 29 1931. Later I saw and examined the original MS in New York City and completed some of my doubtful readings from the photostats" I wonder if Petersen's MS is a freehand-copy of those photostats, or tracings of them. I wonder if the left part of Currier's (that is, the of my ) is not actually <2> (my or ). The more I look at Petersen, the more it seems to me that my is nothing but with a plume on top, that is Currier's <2>. I downloaded the digests from rand.org, a frustrating experience, as I kept getting cut off, and rand.org would reproach me "you could at least say good-bye". Browsing through them, I realise that some of the Voynich correspondence must have disappeared without trace on its way to Australia: some of the stuff there I had never seen before. Going through my collection of French comics I found an article on the Voynich MS (in "Fluide Glacial" no.144) which sums it all up pretty nicely. The author drew a cartoon of *his* explanation for the VMS: a 13th-century monk doodling on a pile of blank sheets. Getting, as you can see, nowhere, I "took my courage in both hands" as we say in Froggish, and started writing this text analysis program that I've had in mind for long enough now. It just has to be done, there's no way around it, however painful it is going to be: using extended memory if available; if not, RAM disk, and if no RAM disk, hard disk (I need some 2.5 Meg to hold the complete VMS and the attendant indexes). So I am going to be very low-key for a while. And a pox and seven curses on the Intel chip that makes addressing beyond 640K such a pain! From reeds@gauss.att.com Sat Mar 07 13:41:10 1992 Message-Id: <9203070441.AA09755@rand.org> From: reeds@gauss.att.com Date: Fri, 6 Mar 92 23:41:10 EST To: voynich@rand.org Subject: Voynich MS: the Khazar Connection Status: OR A friend showed me her copy of a book by John Sojko, "Messages to God's Eye", published in the last decade by Vantage Press, putting forth the theory that the Voynich MS is written in a kind of Ukranian, by Khazars. (Khazars, you know, or is it Kazars?) New light is shed on their religion. I did not have time to actually read it, but would like to, some day! Jim Reeds From rcarter@isis.cs.du.edu Sat Mar 07 15:56:31 1992 Date: Fri, 6 Mar 92 23:56:31 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9203070656.AA23575@isis.cs.du.edu> To: reeds@gauss.att.com, voynich@rand.org Subject: Re: Voynich MS: the Khazar Connection Status: OR Have fun reading "Messages..." by Sojko. I read it, and I got the impression (among other things) that Sojko wants readers to ignore the botanicals, etc. in the VM, saying that some of it was `inspired' drawings that don't directly apply to the text in the VM. Please feel free to correct me with your impressions, Jim. Ron. From RJB@U.WASHINGTON.EDU Sun Mar 08 08:36:00 1992 Date: Sat, 7 Mar 92 15:36 PST From: RJB@U.WASHINGTON.EDU Subject: Messages to God's Eye Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: <8C4ABA0A301F617D44@MAX.U.WASHINGTON.EDU> X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Isn't Vantage a vanity house? The book is probably unfindable. Let us know more when you've had a chance to look it over -- sounds eccentric enough! Khazars! --rjb From reeds@gauss.att.com Wed Mar 18 12:56:30 1992 Message-Id: <9203180356.AA01747@rand.org> From: reeds@gauss.att.com Date: Tue, 17 Mar 92 22:56:30 EST To: voynich@rand.org Subject: Voynich clam chowder and marching society Status: OR Just had dinner with Andras Kornai, who is in New Jersey for a linguistics meeting. Nice guy. Very smart, just the sort who would make sense of the VMS, if anyone does. He is in favor of getting a grant to subject the VMS to high tech scrutiny, to approach a medievalist to "front" for us as PI, to use UV and so on to get at obscure letters, etc. Wants to digitize the whole MS, either from the Yale microfilm, or from better quality color shots to be paid for by the grant. The new twist: by having a grant, and a big-name medievalist as Principal Investigator, the Beinecke becomes more forthcoming. My assignment: to set about scanning the Yale film, and to approach the only big name medievalist that I know. Andras's: to approach friends of his who are Yale faculty members, to sound out the Beinecke's opinion of this. (Maybe having a Yale faculty member as co-PI would help.) Andras is upset at the quality of prints I made from the Yale film, and is eager to have Raphael & Blunt quality prints of the whole thing. Jim Reeds From rcarter@isis.cs.du.edu Wed Mar 18 14:30:52 1992 Date: Tue, 17 Mar 92 22:30:52 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9203180530.AA19059@isis.cs.du.edu> To: reeds@gauss.att.com, voynich@rand.org Subject: Re: Voynich clam chowder and marching society Status: OR YAVA (Yet Another Voynich Approach) to getting color copies of the VM might be to follow the example of what Oxford has done with some MS., such as the Bodley Herbal and Bestiary (Bodley 130) which is to put it on color microfiche, and distribute in `book' form. The results with the Bodley is quite nice and usable. Regardless, I think getting a `real' and `proper' analysis of the VM has got to get done, if possible, even if the only end `result' is to date the thing. Such techniques as the infrared already mentioned, as well as viewing it under a blue/green laser to bring out erasures, etc. would no doubt unlock a few more things about the VM. Maybe D'Imperio could lend a letter of support? Surely a `former NSA' title goes for something. Ron. From ACW@YUKON.SCRC.Symbolics.COM Wed Mar 18 12:13:00 1992 Date: Wed, 18 Mar 1992 12:13-0500 From: Allan C. Wechsler Subject: Voynich clam chowder and marching society To: reeds@gauss.att.com, voynich@rand.org In-Reply-To: <9203180356.AA01747@rand.org> Message-ID: <19920318171347.4.ACW@PALLANDO.SCRC.Symbolics.COM> Status: OR Date: Tue, 17 Mar 1992 22:56 EST From: reeds@gauss.att.com Just had dinner with Andras Kornai, who is in New Jersey for a linguistics meeting. Nice guy. Very smart, just the sort who would make sense of the VMS, if anyone does. He is in favor of getting a grant to subject the VMS to high tech scrutiny, to approach a medievalist to "front" for us as PI, to use UV and so on to get at obscure letters, etc. Wants to digitize the whole MS, either from the Yale microfilm, or from better quality color shots to be paid for by the grant. The new twist: by having a grant, and a big-name medievalist as Principal Investigator, the Beinecke becomes more forthcoming. This all sounds great; I'm willing to help with proposal-writing and such. Some notes: to digitize the whole thing at high quality would take on the order of one to ten gigabytes of storage; the low end of this is about the capacity of one optical disk cartridge. My assignment: to set about scanning the Yale film, and to approach the only big name medievalist that I know. Andras's: to approach friends of his who are Yale faculty members, to sound out the Beinecke's opinion of this. (Maybe having a Yale faculty member as co-PI would help.) I'm in the Boston area, and I take classes in the linguistics department at Harvard. Perhaps we might recruit a PI there? Andras is upset at the quality of prints I made from the Yale film, and is eager to have Raphael & Blunt quality prints of the whole thing. Jim Gillogly sent me some nth-generation copies that were pretty horrible in places. BTW, I have a proposal for the name of this group: "Third Voynich Manuscript Study Group". It's simple, it's dignified, it recognizes our predecessors, it makes life easier for our successors. Emblem: the Voynich letter on a circular field, surrounded with an annulus bearing the letters V M S G around the top, and the roman numeral III at the bottom. I'll take one sweatshirt, a button, and a coffee mug, please. From RJB@U.WASHINGTON.EDU Thu Mar 19 02:31:00 1992 Date: Wed, 18 Mar 92 09:31 PST From: RJB@U.WASHINGTON.EDU Subject: Re: Voynich clam chowder &c Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: <83D8D85C3DDF633700@MAX.U.WASHINGTON.EDU> X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Sounds like a very exciting development -- or proto-development. Naturally the PI/co-PI with best leverage would be someone at Yale, or, failing that, someone with a Name. NEH might be a source for a grant. But budget and procedures need to be well specified... What seems to get funded most readily are projects that produce products -- e.g. museum exhibits, books, etc. Or so I hear. --RichardB From ACW@YUKON.SCRC.Symbolics.COM Wed Mar 18 14:06:00 1992 Date: Wed, 18 Mar 1992 14:06-0500 From: Allan C. Wechsler Subject: Voynich clam chowder and marching society To: ACW@YUKON.SCRC.Symbolics.COM, reeds@gauss.att.com, voynich@rand.org In-Reply-To: <19920318171347.4.ACW@PALLANDO.SCRC.Symbolics.COM> Message-Id: <19920318190654.8.ACW@PALLANDO.SCRC.Symbolics.COM> Status: OR Date: Wed, 18 Mar 1992 12:13 EST From: Allan C. Wechsler Emblem: the Voynich letter on a circular field, surrounded with an annulus bearing the letters V M S G around the top, and the roman numeral III at the bottom. I'll take one sweatshirt, a button, and a coffee mug, please. Oh, and I forgot our motto: 4OF89 8AM 8AM From WBRILL@macc.wisc.edu Thu Mar 19 04:07:00 1992 Message-Id: <22031813073746@vms.macc.wisc.edu> Date: Wed, 18 Mar 92 13:07 CST From: "Winston J. Brill" Subject: Grant To: VOYNICH@rand.org X-Vms-To: IN%"voynich@rand.org",WBRILL Status: OR When someone has a rough draft of a grant request for Voynich studies, send me a copy--with approximate funding needs, time frames, etc. I've got some ideas for private (from an individual) funding--with no strings attached. I should be able to get an O.K. (or, not OK) within weeks after I receive the draft. Winston From kibo@world.std.com Thu Mar 19 09:35:04 1992 Date: Wed, 18 Mar 92 19:35:04 -0500 From: kibo@world.std.com (James 'Kibo' Parry) Message-Id: <9203190035.AA06117@world.std.com> To: ACW@YUKON.SCRC.Symbolics.COM, reeds@gauss.att.com, voynich@rand.org Subject: Re: Voynich clam chowder and marching society Status: OR > BTW, I have a proposal for the name of this group: "Third Voynich > Manuscript Study Group". It's simple, it's dignified, it recognizes our > predecessors, it makes life easier for our successors. Emblem: the > Voynich letter on a circular field, surrounded with an annulus > bearing the letters V M S G around the top, and the roman numeral III at > the bottom. I'll take one sweatshirt, a button, and a coffee mug, > please. But first we have to pick colors for the logo. :-) -- K. .................. ................................................... James "Kibo" Parry 271 Dartmouth St #3D, Boston MA 02116 (617)262-3922 kibo@world.std.com Independent graphic designer and typeface designer. From j.guy@trl.OZ.AU Thu Mar 19 23:19:21 1992 Date: Thu, 19 Mar 92 09:19:21 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9203182319.AA19719@medici.trl.OZ.AU> To: voynich@rand.org Subject: Voynich made easy Status: OR I downloaded last week a shareware solitaire game called "MAHJONGG". Good for teaching you to recognize patterns in a set of symbols. Here is MAHJONGG with a set of Voynich-letter tiles (in file VOYNICH.TIL). You need a PC with a color EGA or VGA. Each tile has, on a green background: 1. The Voynich letter in white 2. Currier's transcription in dark red 3. My system in dark blue, with a bit of an expansion to accomodate the rare letters not in Currier's list (e.g. the "picnic table") The four "vowels" (A, C, O, 9) match one another, that is, A matches C or O, or 9 -- like the seasons in the Mahjongg tiles. To make things easier, I've made them yellow on green instead of white on green like the rest. The four rare letters also match one another, like the flowers of the Mahjongg tiles. Two tiles have no letters on them, just illustrations from the VMS. I won't tell you which. Let it be a surprise! -----uuencoded file starts here, contains SOLITR.ZIP ----- begin 750 solitr.zip M4$L#! H ( & $"#"1>W*!/@%1$ $"# ) 1DQ!1U,N5$E,#P 2 R05 M-B/__?\O&+]WSC]^_' ..(.[01B'.T?3 M&4>< X[A2L!, \Y$.&)+_0Q?IJ'AB LWFP;DNP"0L6FX=.D]-@WN',YM/S"< M<--ZX,@ !X;9(2 $2,*!C<] FFP@- S!ZFV *3/ @!Y/?D]DIZR3!$>H%J^@GL* M]5F -YXY([C'D6.9!OS6SP>\*]Q#O@MW.N3! ],@&+8C1@@@=V+AQT,+$K9I M1^"2"8IGPQ9L^VR:!C -XIUI.)L_H(ID2_?T@)Y&_7S 8=,%^%Z_:E7!V#0) MU.'[?]6H 00'L,-7C9K&T'=7C5I!T !..%;-IE'HK)I-(] )"I:= ]K3(-P- MW%<@'!FX-X/PPP$.%L%L@:1P('XX+!50FD"(;9H5EJE\U,.7?6H:(/& 4O2 MB(5J]A4%H'7L/UP@-]*/>*%AYH&(GFGY*;!.G4GW#2H MZ)F2>QJ$:=!.IB&@3(,+-EVQ:6B'#@%,@X-V<--@!)Z=@PK&I@%HGP6X4KI$ MG'&Z1Z8QY(]08].P1(W:H?7*+7BC1Z:QY^D0ZQUAF6:79M.@SH&;9D^SH-S( MHD+QC4W#$SWS68!+IJ-8,(](3 /^\(\-!!/\O6D ',!/N*S %\:-\(S,LXT M>?'NA6?4.3;M%^/BC7@A>(P1K@\<:!I%;L)?&-8XTQHQPCD_- USSLR_, V. MZ/C^,![M[\CW-/7S 7_UTMB,X.0&*L0TX*H^$N!?(WR6QGDLO;!8,,^H:?C M&(;'B_'"9 EIM4_,M@C;M,:Y;&<:'+D'8C["TZ+<- $W^L)<("5F MGP5XI$T ^(-IX/4\P(A@/34-^$2^S)D%PQZ;!GS"9P&.$X-BP5]8+#*.9T%- M$YXMI^ [D)UD>]Q ^Y1L$[^M%RX^&6<:(.S'<.UI:MKC_'P8FXO-!1\7PP\' M>.4/C@TWC4>@AIM&4<--(Z@1TW#4J&D8:FP:.H+O4M/0'*9!DB:F8:2Q:3AI MU#1"&C&-DH:;1GV @'^-L+NHCX0AV^B@IN&$,U(YV0-'F(@RC5U\8B2Q:=2_ M1YB&T$CTG6FXN^UI?N=I ]H^$N"S=M3_X^F.I;2*]=\+3]*CEFDF[4\&/H_[ MCUJF15K_7[C0%+5,B[1&O/#SX:9[X0$W\#1P=P/W%7!'!N[-P'T6X&%J1R]( M%FH:8B8AIM%Q]D%XR.#.@? T.,+4..G$9P$^^P[@>/[,GUX!Q#3@#_-FSA2, MF 9@? 7XB!\.<# ;O-XI ;$7GC@I4--8038;R(@SMEG0/K_PPV$ZF9TCI@%R MG)LO0O>T:3 >X;, IQWGIXHHI&2A30->Z^<#_@6T;!I4--0T.F[@"YQ%E"3; MIU)8N> (8CO'II$1ID%X&GGW#F>1>"-\),!IO9X8U)\P\'2$!8M- ^Y@%MJT MR8#8-,60H(], [">!N[(@(_X^8#?6"SJYT/'^?D@,&=9[.=#47#4Q8>8YA&F M0;Q[9QHZ@F51/Q_P7B "];O0>*9!/#0-,EZ\>*9!'SY\Z&D F*6 F08\Q \' M.)@[Z M[LP9P8AIP"^H F)&0 D99AK0]H&;"_ 1GP5X$D.W:=!'Z2#;-=PT(3Q1NT3I M'#>\D>'VJ6;#MNN."Y9S:EKM&ID&R;QIF+LM>#))6W%3L\\"G!9/;U$](@6 M0DR3%S_^OS --F[\B/'C3(/_&#GBIW. \C1P=X.@D&"F 5_Q\P'_*E?DB6F0 M@FT:!.%/HA0I9_B[*$_LHZ:-N>RVY1QF&J#MCQ?%-/4TZK, ?S4C@$Z>J)-I ML.[=DSL'[=3=.4CEG8.D;-2H%0Z0]H&9!OS#1P(<#.HD.WZ'M+!I^D1CRSDX M:!>A64'6@@\)-PT76L+;*F<:D$],@[I;-FL1UO(TZB,!_MJLF6#$-. #$#*/ M.>$A0RWK_(_Q?TR#C2?\AW. \C18D2*"$=. #_@LP,4LNYS0(L[)$]/4/I1F MJ6!!VN8*;!J1VZAI )IFL:?SVYZ&^RS 7?_Q$,A[TX ^1T0B;TP#VCZ?]S1( MX_YC&NS\GS/. =+38)8"X+8,T#X2X&"@D7)."B:38*;1>$5X)&-L%XQB&BGG M6W:;.$<>FN8BSN'O>"16<=E'0LLYEZ>%?Q&?!;BM'P2X:2 R9S8-2CIJ:AJ> MW[0^FX:0%G0V#4!X&KB[@?L*F"T#E)\/^%<1 "S4SX>9!HUBMH5XE"@N/KB; M##4-8$4\C3J'O[,WS#0@?!;@[)-GN&WP%*8S >OCRE0N*^3; (46HJ]P[/QSDB6E:Q#2LG'/B*SAS M>[,\_9T)%H#Q68#;>4%AI@$?<#L:\7,^8S8-F\%QCC0-]L+T<DBIJ$L/G4.,(3@BQ?Q#1(:_S%-*A6^<\Y0'%ZD2(&/LAOQH;Z+, SG^BI4-. 1O*M M(3/C1(&CF*9J,LI8Y4R3(J9!,$QL&C7'\8KH2(3,5SF?!7BMX"],@QB!&FX: MIQB$&C&-)D*-F$90(Z;1=/\?H<:FT4__TTO'[D9'0/G$-'R&KP"'2.\KD)8J M'74.EV345X V6X"'6$@(#]!&,LJS4-,HQ*8%)229AJG@L&D8WFX:K-P3!@!B M(<&?>)K%>/L*YB"@I@'ALP /Y(@PD&C3@!\D-PV0G;IW[VX:7K6J8'7M[IB: MAG8U#;*Z8)L&$XQ- QY@32"\/ G6HW?I3*.6/S)-G".F_2A=NF2(E8Y\R.T3 MSGXGFQ'+.2+B^)UID,YA]ITLD6FE,RU/4YL)%/60PJ81T7#3L$:/4/U=(ONP M1^*6+?N16>D_H[D,+)BF$^&P-UU GYD&O!9'R#@8C:!G%S4-$U'$,@T*.&)GNFL&(%-0SE<8AH0Z4R# MFGC4TQ"?!7CKO_-GS@B&FZ;_44--$]10TQ@US#1)CQIFFI*%1@UW3O1"HVX: M_)-IUW=2H'Q )HV:1D@CIE'2<-.8-,PT(0VUD !K&O E)%$6"W,LF6:%+=,X MG7,B(++/,LJ6S.XT'#TSC!\B)8HS-P5P+F3\"=T?&<40^=$>8)P-T!F,D!7;=JE6%CSMB$OS)- M<)OR5XSEV*8!L\HTP__*.032@BS[X*8AX+:G91_FLP#_]UE )W(Z.(#/0A5V MCXV%U6,3L8T@R^$,5"*:?99-*X!S(#P-S-W ?1;@=(+ML>6S(.LT-C':,-,@ MUJVS6>RSH.NGT62(-KGP4F=OW(9X&9EL\" M SCR6:#N[K- )HSDLT":[N>#M&^?!0OG%!81G)!I\=&SGAI'R<6;/]Z]NZ< M_J99\G?Y%I.:GQD;E:Q\SHEIV+MJ5:L^-0WZOWY5YR"*/C4-CMCUJL)N&NQ= MT6I/38/64]KLBY8Q-;1NTA8LQ-:TBY%EQ7PNU M[%/V+)@:FY 7<1N;S87]?,#S3$QP2T-,P[G;-&@X@2C3X!S)ID&^,PTQBOS@ M@:H5SC0V;M,4L@Q\Q,@EGH9JYW#&!@_@*Y#N'<$YN*=!OG./NAMY$ X345^A MK@3"GX"_(_'9(2@1$. MCR9R-PIF$;>0>>$DU WC7+X8S%7HD3V MB-1S@!FZ$]H<]@1D%?3QGN=+Y/40TX*^8W=W )?0Q^O>_;EI\'?,=^]N&L+\ M*NC8[",+]7_H:7#FE3V>:?"#[_9!QZ9Q/"!3[^*9AAZ<>9*Q'IJ&Q_M_Y@QT MU#3)F/F]K^E#TRCSF3^S VJ:Z/[_/W3$B2<^'_2=PP/A]4 ]&A"F 6!^/N!@ MY8I\^=^^V3,&F@7[D8W-I@$]3ND4[!>F ;:?S+7#M6F A;Q($8=)3 /W! P MOX@CL[NQ:0#[AW+_=[F<+I!I$ $2F NA%IMV M,@/-TTH&T9 #QOYDS!?FF/ M'CXBD3V2>?--XQZFD9^F";Z<"%^V#S4-'%I.Y"O"#<0TU#F*VXE&#ZRN P 4 8 D !&3$%'4RY46%0- @$2 M(Q05-C=HB9K;/ 4&$A-$Q?:6]YN&10M2Z5NW9\^"I)J6;5F04\O2!4FT[-RQ MC=4$+RNTT[#MBVK$R1(HTR#'IWJDFI2)MI[:EVQ:LN.I4,]V88] M.Q?D6[,@Z:)AWW??RF5+IGL/E5LV+-VR9$&*S0/^3LNR^=Y!W9(M*W>.-54. MK^-Y7J&?(,T<\S#Y&Z2=])T7Y-BW;*RWV;ILHE?^TO8ZVGMNV;)MIIUMVK5E MJ!]D&F@=IGO/.5KU+_:%K().LVG,E?/*7OY:KINA5_]TN ZV2#I4?:X+JSL) M]<[[MNZ)=X>3K]VZ3;.]SE>UIJ.ODD]9:;]+G=G-UIL#^M&VP>1.UR7N<.$L6S>M+'9%_^ MT#O)(.NY*>I[S)0@J[IQ-G ]E,\@M,WKO?1.N3? MQJL.D[]EE!\3.RR9\=@:TS=WU4<(GCCR9Y[VL]\SQ!^-E,,[WL;T)N*N[*ZV MIW=+8;N$5>3VF.O/<;V1AN&ZTW@%O4CB^RW<,EA!VE=U5Q]**['5@];9XXPB M8VB5>KI40>@Y7[6=V5=V'D.K&QE#NQW#^9C9'4AG-5GP[PC8V!U(_#K>S@9.096;_)8#X,5]*!S^,\J M*'!7M8)>O[4Z2IDQUJ7<-FG"R]#J2>>4KZ9+3X@RM#K3,&\5_:41Q^HWW$.J$U6;5HQ0XUS9ANZ>PW1#?4.^ MZ< 6A+#Q,^;\HF$@>4GCQG0G2"XG5_=E^5?+=$?$P6;1/Z& >AE=X\ UH^@+ M STC25M/'=,T8T@GP#>-JJ[$UV<)6^K;,Y0V\"'7=$&@PC ^5#Y02P,$"@ M@ HU)L&'D6ZB,7 %P < !-04@N5%)., T*, T*,0T*-#8@($92 M3T='55D-"AI02P,$"@ &@ 8 0(,)%^>?8F'O50 :OD P !-04A*3TY' M1RY$3T-A"GL'!AL&NPQ+ PD'"PD+"0<6!P@&!08'!@4V!Q87"PH&" H+!085 M! 87!0H(!085!@HE!@@'& H'"@@+!PL$)00E! H&! 44!0DT!P87"1HK_/S\ M^_O[# LL"RP+/ LL*ZP, 2(C%!4V-VB)FML\!082(Q3E]I;W1<7__Q\8@W@0 M%[B#C,8K<'CR.,H/L6XO:E2<0]H=Q[R#AL$/$D']9T,\ I?.?=[/>O_LO,Q) M_A_/U7E=?O3\/XD>?<)T$C1^[B?XZ3]H6![[EL<$#@3+Z[<\E7P-ON6A>O>Y M_O%1O[=U/RP/@*/G_-1KK>RE/]RN"+?E@0G2@"V!2>EXRO:4KZ-:>9EL?L3BUO@0.7F-?.&^<5%0N\N@6[UM\TJNN?H'\(WL9U'\L? M5:[CIJNK3_G^\96=XN_;5;V.U?#\;/=?YU+)YM?_LO_ZL]R6OY'PUQWM9^?F MYOYP][.M>?^W*LX_ G1_,?_*@WBZ7N7K?+9KN>X5\5WW\UJ>B_:7SB<<[(^_ MELIO/%M]M=GU_YSK;O((GNL[6;^ODO]HW/]:1%>[X\MUW@1KY,I]_,&O7=NZ MX)U>:V-WV8)K17M:#O>O V0_Y[_*I>U>T9;XQV=7GL7I;'B.W^>S_D,V7XV$ M,]QOW@W]M3Z?YV_."_:.(%J+3'AV;=_^0]JS;-&5\0QUK?+EA].+T=RW)?*7OUI=*5C5JC=V7SJMQ M+>'RR,X09Z)*:=NK!G'95_!?U<-L@\U)H[U<>I:KR;;]GX96OLU4^_AOU#'C MF>^7*E=C_>6]?K/W@]H/M\^QK;(2]4GV'LZKWM]/^O"/<-7O7;CR!['+E*.U M[;E4BL3O1^Q%X6\)J+]P3#H%[\O/$>Z7[KT$L^O[/W;=HK.MG^4Y'._Z5:M1 MUTRY>?Q>XN^;;LG_(X0W#(0W>8CP[Z61%XP0SSKAJV'?A\O_E^7RJ_<"TJ^- M#?"M05?WVC_O9<'^OWE>1P+PZAG"6W2OU=!KL"_X+ELXWRI?5\*G,&]=>:'( M;YGURTUQFTIT&+O,&Z/#_,+][:_AUN4NYQ5 1^;R#;?^^&U'2B)^.I#^]7_< M .&_8JZ/\.U;F5>4H,(+<>&-I_0TK]/,=V&,K-SPV9FYJO8Q?F,_]VTM<%L& M>'^?JQ6I&*/@YOV/JR$QZMD#?YH<-W0_N+(N$1R=)_Q^T+ULL/$H]]:8=?W1 M$]:V\#0VM$?+^[$W*IN-7Y<7L'/HN//T^)_9?UDD'O:NFU_YU=Y_/^<%_KW( M7!B5N^1I\ZDKN['KPY5NCRQKM&E5LV*XCSU>-])YVX60][7]//?X-1KU^J<5E--"]0Q MG?4BYW6:,(NWO!F96'#7-8%:V G,R+GO"8IGP&BB%>WU:=9LS5?Y&O[D7G4Z M5CWU9OB[EEE6C4C8S%!,^E*4=8*?.D&/&./MANS4V:8._[>-$<17';^]3MD,[.RL0^5:&V6[VE7:(SC=>1X^V;R6W[" M<6K?-[4_OE6D3 75)?"ZTF:<3;'9RW7,_%XTK!WMLKRR[QV^).:&PL6WNWIY M_XU=48E-[%W!TZK3E_P_R1L*2_'"L'\H?I%N=^.&5PO8(>JE^:-9PMI*:]90_VB.$]X0W;TUK^*.PKH0>PFJ$HR,TVJUDG$X?-L+_=5%9,._4I4 GLVF MF)8W-ROF9IX?IWZ?<_@[CH5!(SR*8=903.D&B) R7^'^;UXJV7S:?1U:5; ] MLXJSK5YBU'H110'DL":5- M^F,4.14X.7R6+;E^E$F,?$\6];M!7"_)\^22R MM"HH*0;;@FTGY0_FGZ!A[+8?I:)4AM@%4J?P=.5T0[AN-J[-I0%?Q6H\4X=% MSDZMI0;B#3F:I/[DDWG%=;MAQI3F!31]:*MEY:=>AQ:4WRE@!E-.*)ELZJSW M*3\A%7\K=0WSQCHA]*Y.NV$I8^(VS M;VQU;2OG+>&7.)BC(>;';&,OA:,Z77;)6;B\&12NKMLGKA+V]F?VZO9PL)LO M\3250N,?"]%ZOFCQ1^.4;12_TPSGJ2LWYX4PMW>^&;V=5O-U ,0K@:]J4J1K MB*KBV8D7D7CMCA?<\['0\EZ5?5?O0\)LOD/W)N?_&Y6[D; Z>]$4X$^C<9?T MJQ;;<$57/IA^X>@YGOKUN!W3^'=XKA?<#I>!,T3U]6-DL0R?.E1Q'Y6L[+DN MI&[2%U 4?>=M'U4T[.\5;N'^#RYWZ9QMRC7U\;/LW&VO3;Y#VV=*(3$5M2'A M7;V972B['\:@/68SFOP 54=0OID+KA MSD-P_?*/ V)Q'/V)K5%8/L:>YG"] MW>PIQH>KZAI.21,;U9LBD[E-;[OYAM!6=@78@>$$4=B-"H;7"-?#M104WA&N M?5Y_:FK[%R:-4G?K7AY[_\ MS-ST:7I&YC0[QB2=9XTQR9AF946?9F3GY @6[=^:RQ\50-KK,*M4JOT9*8N$ M\,QD-E4ZAG M+&N%^8=8 80XYO%@DCV!7)Q#"#4E'6N^N$W'IM%.;VH!KO ;4OL'DPEX8(=& M(+FO2QX^!RHTJF0=U_K^R4_EAYB_,(94&X5#URD5;CJB'5\3C=I+?W)YEQ!!E8".D[I92/=U:4,UXN+:M3P+DF,!6_-%WR M(B7.P1P%VR=<[UYLJ#:MX%HU//M,?.G*EMA M-.O)6HN#A_Z&,>@FMEE8BB)1T"(J#,:PN%'9@1CGX].HAF02I=(LY8G446HK MZ5D;=G:CBN::F9N/5"OR0-:"'A:[S!75[Q2D%9; M(C:W]CACDIN3FY\38Y*-;N2'$%>LIZX^>J@$EA_C,W$6\97<$)JF.O&Y_ NC M=G\[RKCS1C;XJBK6K=F9W=:;T&4.G#=;LR@SNQA,33FN3;NX.2EL'O&1K.Z4 MB(@!=[ C+A/?:&TE%UIW RG$(F8^3NV79*67E5;,69M"3&'I67C,0?]]1+C/ MO23(ATR6U;<2BI/:;6X(VY866HLZ!E[-./5P5=(S;N/-^3JIQ2XZU 8CM& 3 M:J+PDZP6[,NN3GNY%G9V,8E!&>UN2A9>-QET4W''6S@XL";-RUJGF-O1R%TE MQ33"?N*HQ>7IL[%2)?ZFPNA >3U9K"#;..$5F]2Y!*XHW:-NJL).O7652F@S MQQZ[5653Z+)KQ*P1]Z43JX+06+$$CB9P?@YX$)/ZX] &O:B_%>7]4)*9Z,$^\8C48\5]H'1B, O9*8G!K4"W6.Q^., ' MZ PRR/.NN>X-G$CN>E5$V#'G7BYDI,BQ59M3LHOD<;E3J=>2+D3"GGLC 3F4 M9";"!62<]WL>:%QDD;#C"\>+]>BVB6E57C()MIXW](QKQN.!/$X+L]0$Z?49 M67=KC*J^:)4RMV \=ZHIQS#^AYL4.S2[P02'IX SUY!T'&4\YBBO65H2>J6? MU1%6-N:KI!?GGX\DV^39X7RJW/7YF7HW8(XJ#O+7G^7UQ/"Q\';BH-7^W93? M&O3?=C29DR_EBOZ8G"\J,,](,(5:U=M4DK*\*N_C\D@>YT)W*_M E"VR%L(X M=DDQ-Y#01FRA@C%_4:)& J)W<(UR*@5/6J,H5%X[B9ZSBT'"7\J4^YN>]_8GH+=RQOF#.^W?I^6P3LQV1GP1 OND0>]_-].ZA0 7[G\"#^01]8Q#ZOQXX/YDS'?I+["FSI&^-W. M2/.XX_'=,)$0YUYB4+>;$V],L>;:YQRT$,0J0VH.CK15-":5 M3;HI/^EE42O&LI2T*A*0O+1*_4E6EH;'CK)W/7[&?C0+_H3[%N2NH]U00:7B M"T.8X".0"NG8Q F557!419D<_\"SK0^Y!M%@CTB NS.C:0FD<^TA)E_QRU,T M"-O9"@.3LK<3!TN$;220UE=]^6XHN7)]+ M43VE!H3[8 L&AZ'8\ V3M:%'K.=9$ M9MDW"".(QVMTB_YR$' #QMO_E:[CY/G].,%(AZ+@X$J&C)- 4#2XV.TD]$V< ME9SP4'*(J*8+2#==%9OJ &=OFK/#&8:!K;5%K">(Q[G(85$JBQ(?-Q@* U?M M %3?^ P'.C,_N(+JZT@K, 4R109EF6ECWS1XE_:#H"-SB_BF]# V(#6T#%%' M5F8$4!85ONC'6JTIZ0S^X0B#>C:<%<+SQFSX7I -9FQ$^[X!4G7Y^7$2VT#" M&(23QV&^G3A,7JX#"E,5@@D^ I2I?(2U=_SAEM3Y>&IWW!E^>^'=1'$1G .; M&D;30Y:&EAVAS$R@CBD@D",P?IH[,([83Z5%-W8BS!=6Z.I6]N"D),4<2';C M5( ?N4_&Q95&ENL(0OE$R@1I$IEG\!3;51\P% &T.O$$S>;\C*<;9*)6" MY*;/16-1UIZA)?ZK1$?FOL,O".8WKP A>G&D>?^\ZCX;1Q[AYC?T*56.# MTB,04Q.[^?UW KXR7R4@,N9_.2-"8I@)/[9I6;J88A?RC=^]-G-4#(_YGNTL M/S\T"K,]R+ ";!$X"#<._RC7LJA["*KW]5 U''@BN%>O]X*VV(%DCU^SN;5#!AW4@K)")/= M%3=T5^13((6M+-/.AWL3Q8A13SP:XX8UB;\X7N^/-B16L^3Q>&M'X=GV(!0< M*F1N'"#92\^?EN- (.W &;6*MKW+C7RRW+'T."SND:-BHRZREW@Y&Y[Y7W<"H?$JRM)Q>(I_.!%NR8E1$1JK&:JK<%(1!Q MBE:?BQ=/!.XHUE5'M56 4&AA[K-$5<1*,Y: 4?T!K"H>89]Q>Y4<)7&)-<[Y M8 JUN<&1W4\'!@;$EAH2E1A1'8' ']QI/GXT$#,^<2C$.XT6Q4-A\JO*B0.4 M[=IEDL*Q/8660N6&IN'H$$1]:CS' _B4MB2'S!%X++$,A'U=R0!K^?B EU)= MPP\'0TP-@6OJ>"B]%]X7K,%[[$7-](&!%],:C$R?P\S<@YN[,3.[X4C#ZM M):LD-<*'AZCD1!E==Q0Q,,KGOP&'#+MMM\,[>N3>?A"E6=?'CU.T2'D$>#4A3'M%&7@'_:F7:_UZ)*$<))L%$X.>UQV'D\U3A[3+WE: M>2CF&:5-T!%>8PB//0L3PLQS[@G!JU4Q63CL0[+/O_*MBR$YE" M#U M>=6\?"9TB%X7>/4> Q44A17[ALX@'W)A_N#,PK%5?Z#3/U?)2LUF?XX %W0N M_^:&?O7RQ"=N ]G8;^B#5FQ8.'-Z@@ 'P<6&IK+BQ<;R36#QE4UE5H03'*%@BBT_A#3U')>B#1H2%EE(*M P9!==F>3 M*$0:986#!I)M,FOR3)=ORU32GDB.#;*K :,\)N=),NHQ."I#"+Z<3&VOB.#! MYX,\!'S)MQQ.V>'7H*6L1#/X2%9I3.7X/1%3?Z&&&1 [!72'BF+UHR ;Q525 MXX^M8G_A=B-0[*DRWR';7K^MZY3>4H[?P7&8WJ)RATJ91!,#B.Q"YS4:-I3: M$A3),]PALZ.)]SHA$IZC_404O_X^_\)$(D#=MX15;9 IF82?WG^U]H7PN+#O M$"$\"*CA:):ALL&2?OFVG@^7/Q3CX>F*']('2^!#$ GT]P>]]W>36H-N3Q@% M#@U+?2#UV:IL#&&,)>$;"P:S8D,C8ZNO1%JM8RP&4(T\,8#!S/T^2C@,L-.$ MM!4AL#S?HK[S1:32 &I% !#N;*2J8V\+'D.CH*$P;-PTY*F(/GEW>!%WNOY: MPEESX$_"; @#4;.7ERA-=]*D/C=Z*1T[R4IG#*P,708(K5+ M,R9-&>'+RQZ[L]ZE9Q-%U6B$G0SGHL:J7\WH=98#2+4RMI(/LU3<'&1LC$'H M&HK[,*)A,!9Q;FKL!L8!+W"='LJNMY+3,\$!ED^&=%.X;! -A MXT%.F^$97GLWJ51/^%^8^T,BS>$C_!B9F=F9;H=8W)[6947J5;X6>E=_8?BT M\SZJWM15G7W(4]1K_L_.=<[#7^6Y7!",- _Q-2^,=SZ\F(U+,V HB"^D>6ZT MW%S#0S9$EJ84]-5-#_JZM_-IT'L.FV .?)WN 3-M:V@#[Q!"!;Y&5$D!A(\. MF342R7;(F>U)7$%!Y,:#B@."(>L$'LV_\8_ MRKTU[.%3W9Y2Q6@M[\H:DNURLQVW1ABF\T8=XQV8$^]X_RF)XUI8WH^]4='; M-^);@'?0%]ZAK5Y,@J^#F9\?04?9AP$Y#:_KZ>H=XL[ZQC6#^4:X&F3ZX51L M9?*A3:&@8&5S\TI.B2<,15)*"]0+ P$GAB\'5&M/6VYW#-RM%@7NB M\%HR0 [^EW@'9CW6:4M%F 43P02UFX$YSK]PDEM FH!MH*(#)AUMFB\,7!: M^M@X:!B723P@\7Q<,.H-3N/4SJ>ZDH %L-$(O*:)*?UQU':>-N1^!\O $/7? M!'_5Q\N^T^T=)!G4\B0G.RO3%:EL$3MW=&J0E1("5,\..35L=EV'$I@D^=-4 MB47O8#N6L%M7\W&2.LEY8&2,!5!-D;@L%!0XAV)*2'(.1J$J,024]R*6AQ@1 MK+15 *FL;HS)SFY.,!BZ7*6BB#,L)I+3!VBF\?OK,SRD!8\+0U@,C32@[0E7 M/:2YV+'C("T'+<56ZNW WS(>!_^)@PL-/&E7R[MJ#Z2^#F2^'5*=]:H+<3-X M3#OC837"H\*XF,+>5]_,441.&5=WO27C9G>(G#*CR$;,1, M0SR.!*S"_(8WMIZ>)C0M<7$Z%U(4M5CS#!53#A.&A"160$IO/I6Y6- M9<[Q^Y&OR7.3>3L[&'LM8H^6;;@Z$^][:>2I=71"*(; QC(6!F:[*C"H.%>L MW>@+P@*HRM"E &JA)(]E(#&.CIQC&M,G.OW[5G/HGA(AZ#U M),)E29TC8P;"=O1@N;&.)9J">@@]$\Z VP;< "QYPH92VC\]:3+^8#>N+]Z M'W%,& HN5Y:#,!A..=4.H2!4!L(EZT60,JRRL".+.6%)4 SZ3-%5E6I1.S[: M\%IZ=+W4?=OVPYC8&,&*1,]VAVRF#\D'!+#$X%P*:V!J0RQZ(C9H O\4 GO, MI7XOD-9 #;CI9:1!E!AV12>PAC*K2I)&' M:&&PX,>)(Y+8]26G'']@>3@8"*9EJ#\/E+\-E8_(#3G3A%R&*$&, M66/ Y% MQU-LE;*^0W:8-NWDU:A,.\TWNLORN$A;08MP<1A5$@$>1*4I*9UWF-=NM00/ M%FBL,$)_@% _">'A8!1F>SCH0#S^!]G:3;@,Y-2_4H.R,+3NM*9J&E(X!\L2 M%< (3:MBAB9TJYO8Z8V3WU[+&!"4NR?@3I3(%)$0(7A""A]^8DT$,3"4UL & MA]B1F!2<'&]40":U3>HP:2+T8B7)/F@!C,+?&8CH\=5]A8:V 94)!1A)'$2^ M(-]5LDY0NS"#O8.U#ER:>JX?N2B/>G\<)BN^C#5=O @25N[K:C84]J:Q5!06 M[+\OM@Z4CQDE!J:JB @>2+*<)(PP8TL$W<+<)?7":@LY M]H&U;"7Q,&=6I=76@%M(L81?*;RJS$-X5- LC%SOH0DD%M&L:#(VR0G5Z^YN M$^X2LARQ5RJ5SX,K@QHQ+?698YZZ5+2N'3Z<[M*GJ,/@F]4$0$/P0ILPE;1$ M+<#;GSK)Q^Q5*EA_Y8&&UE1%C1JT>LUG5IBQ>JM#$7R8,\_=)2]\WSUH+%[L0$ !3!S$BOF M#^+UZ#=X8;=(L?5;?E]- J]-3I!^65@<)HV=43RNE($@A<>.LU0,3DVD\Y:- M6-U1)!;",%O&&$C+GDP*<_LB1+%FPQ/;@UY4L7AR-[']LO' D6N$"U1_?;LB M8 )&8ELMKK)H(5B0H]B-,0IQ!4>08*BHL M!B1?/2: MSE'FS>LB! 9DF-LW2)H#/T!% 7=F@XK,8/L8K .3F)31,00MK+# M0D$+P/M6H%5&I6%P_LY W6LU4P;QA];X:ZT(P4%).%:)+ZALHP,)*;S;X5"^ M!VLE:9 ;S734^[AF)A>+FCR-80M,K:PD7XUM7_>:+?#D9I>+0"W1S5,1X=]D M47LAC%USDHW95\]9P*CS5S.JI%%'2%0 M$8N[ Z]1QH /I]]YT10X I3ALI0I5A$C)O-N2QT/Z_H]]0K*:% N;#W(*O9@ MA%X,-L*%^R[\7H*&%6 1^H*H8\-&K3,^^DC[-*9(D#B>L,?YA>'N->:Q%[!F MT:&31^DRQ0B/"8712ARH1-7!HZR?36>265J#VFY*@,*X;?.1OS%/SKNV^#T; M&#]&=<.:?M4NVDV!V@]6']%6YR#7N5JU]5W?#2N"<-5ID86.Q,H;O96GZHU) M<>SM?2VQ5T5R"-O/";>'K'S%@:(KNR2\KRX*1'17XTS#Q3%QNO! #Q)#ZB)D MATAV]8K@,0&&,HCJ8W%AY3Y+^?AF\\T%Y40MYA=+@"D_^3]D\[44@&(0#MYQ M)&FO<:B==]8K2F6%JAK"(QU5"M(G[[*JX@2X_9A:8XSO%&!(08P\4'&BP.XX M.IDX(4JWXYC0A4939$=Y!%;U'MG\XTJ$XP(%!,2"OZ4P!// MF7(1=A&J4&H,T C/=CT>ONU-5:N2@W\TV-/UNW3K28%@(3P%@J 'KB-4+&"R M&:HF6VU6.4Y-1K"?Q:MQ #SJ>H3=C2@:%:60QX]=NS\A;^H269Q]73^K UBM M(9%J0H16W:AX@K 8:D%1'E;P8)&-].A\?)0L)HAU"ZZ"":I9=/$"D:) OJB9 MM4I<8J:GP'CY1&9<>B#(Q852JJ:X6_717W2 @,*9Y&*(%!3B_%MI*^=9P/K MPA K?E4@8-NMZF52S B%L9 M8Z8>IE9(L'&+F/UK[FD]7\,4;PM)@C.)21=!39"WA:E47FO/ILL3$X*NQ/ ( M'U_ '((,#<^!0,?>A]2 (!4P[-.&C[IX1TXMK85&5U9?]0.N3RM1DE-77BXF(L-[]WQ<2C8 M52_?^KHR,*4J01?+*Q=Z>U)Y&Y*?9I*^NONXL6#XN,,[PJFHJ, \8?5M&&.< M4-S7.'ROLY$XS!GLQ>;*?)@;!GFP*Q,BCN!"$2N]+,#JC&69&S#:S@29**$? MQWYNW*L/#,6PBA)M'0.CL!->!/7:N)@_=#%(NFJ\K [S W^=MGXF65D_;'4(]7;76+&E0J(L6_EF#71,.4A*8\):H\<-?\^BU@S62 M6-H?[]41T)!$:SS'/'(!P5E-:'V(\D,5TB989!$,+4)R.SMKMX!) ?L" ==-K]84$0C&X'55"6SDY_LBD.?^ +-;VA@X(VU*WF"0 M@2,*21:TRAX)-:'6A/FPR"7>@]DXFJ^6HHROKTY*E$Y7$;9B&LID%,U!4N3& M5QA+#76U-.>@EU(7/*N+[9&F@Z(!+CTRH&,8/@LA"@)FY"PM0Y]MV_$HF)-; M+]V %K-V^*/D$\X^CKQ2Q0TF[SMD=4CLKB01H)3K->]0&P[ M"AOG-P$$!0C,*% .%,W7\JB7/R X5)QV' H4> >/R$JVX)J!%37(D,(@_3"U M;>>SP[&FR/ET+?S6I;MQU:Z&@6=)8$,>'S&_-EWHRWU8?G4.TLB[P70QU*.- MBBD)3NKUQF,L5OAU>(#Z7M=@1$NR64CX2Q'!*E%4X!$@?IY%LZEA,! (#JL+ MA&>+'V_"0(0#A!=QO=H/<($<*SC?$G$C?:W]5.DT>4HE,'EY+O25Y^0?9W!$ M>"3YJ["06BR-4^ E5%R_BY-SA&NU(8=&= T#J-:,^'O?)E_$1]+""H%I&3^6 M^8Q+UZOL7ZOROVV_H*P/W$05"1=%MDR'MS<#]RW,,O315.$B6+=;XU;4MQ?- MGHFX##G_"-EY.?DX9%)Y5F1ZV)E^4 "9$6,X'2,'<=X=_G:010Y0C2W$0R,X M7$AO ZSRD/HB+GQU(79"9R,P .TZ!%&VYB^@SD; M1?$!*PY(>*0R)L;!C).TO_U]UE!]QR94$;5]DP=A$@#T9 M/N?Z;771BJ*"*XX7ZY \&KX%A) $*U(2<:[5$Q+XIF-!Q=/+T31,R2\0&%;" MZQP0T;7K@)'@CBG3B;/-9IC%M ^R8IFI*0CE:>5@@,,H)#"FPE:UN@J6KJ(( MAU/"EDVNO0U!,F(Q%<7:9\2N0^U4U5=?6!A 5O01"KC6LWI X:A,AA(1O*T4 MIJ#A&FZIF!NF_(WE-&3#(30A9(=QMIGBX$8%''X[@!12HQ84A7#A\FU"$'&[ M@+.8V$,P@HBO P-!^4+'\93?W-M342_3NL<#X&D!D!2,@N_':0JFMAF,40E7 MQ)A=HLLBM\@T2QL^:5=023-"C ?#[K_&(P-(3 X>*M MG_7"OJ49(,1I\E"#9$ZPT>C4/7A?F6WQ ZH8!<,1U,H,;\8%5 Q 4W^6I_GP MPHA>J"-%ENTYUAH!";9Y"U4M*)9? (&ZA;U#&)59: '\1-R,VZ/! )@X"!,I MVD'4(^&!H!MB00QZQX-> 1JT_4MX"RP>-J[W+G0G+@8C)"AVTZ2A-IY(>9[D MYN8C_ GP,\G-<7C,'=!JUYS<7)(G!^X#=NAGID.IU8P6[;E@%2)7AJCR-Z27M5\$[CA8<>MS76+M:(Q MD7XP\0Y=-$"#Q+$2+).H'(<#]%(<.EOSL1E^:#SX2I/(%3>MV-'Q:3ZK$LT@ MY+^V \OA; 2!#U(?8&W'T^8/B2^S2OAP/8_5!"IWN596,.IF:_TR=[IO4JK1 M%9AR*Q*V6J#&/S_D.&O[<;WI283QN/ 91!=WC:*G\IV*B1;=,J80-^;R385I M*C,*4&7[;+,Y3+VFMDR8 Q.AE*L)=VGC6X0 J"X0=H.TNG&TCZT)67T@O:H& M$MBE^2P]2_B!XF3&@&\I'*,HC(VQ&#]H-.3JPU]@KDJ0&K_)\5?G0[<&_3=Y M'.;;84J$IUE$D*+S3]"/ _/%C(^T62@6?ZE:K*BX"'5*/# 1Y9?H+C]U.D%D MW*P@LJ)SQ'0?=;O:CHED<"8#Z%,)(1M :R&5;-!.$_AZO&KD1 MN?@\KQ2^36%#:,#+7G:@Z]HC^*]GPZ97$SS-N_',P);=R35GA(5U2.%=Y/$L M,YC9Q5J2%8;_N@*A>?4!5-I6 +5*9"0!J]N.$J*T"'<(V8S'WIBN5G+GP\T) M#7M7%,;4L>$:8E71LLZE$@/E];]1Q[^0&?XH]]:8:78!2^)1#.2A<50&GSH_ MM^KQ!*-$I'BKS8?.^,17(:W0E%$V>2Q]S(4(J/ 5^5"H@2>Z@#7-R@'_R+0E MO7"7O# 5/--E:@O;;F=?XBCY$^'VZ>. ;Z5J(0Z=';SXQ=6S9"H""_H5#KQD M3X&C6&M:3= Y;<\X+QDDXWQZ" PSUD%2G^:>DDP4$5C5DR M8 BR)APH<1Q^N:&I(E@TC#A#6XG$R(^6#%.T"W=;ECIX RM%AQ!+O=OMG6G< M^:H-ERN&BG35/Z4V@&SA/Y,__='U[24SL'5]7X38F-HAV=3%OI3[PM7C,28' M1@H?1OQ.C(\WGW)SAS(QY4/D*+]5(6R5#W2.\INC?%R#\8:=[,365[2#&!WM9^6@G3AYZL6&9N1'&+K< MIO*#(3CM>.7;HN]&Q+.X !:'TI0\O1\'BDV8$^G*9B8/&:=V/M65[ORP1$IK M1,"N5'??))RI[O][^V)X3#.(SZ,9LF_5Y[ TJRM9PDDT"W,#R+<++C"PCY7% MU=FU^1Z&]F0:/=_LR'>6#B4Z!1%@D^G$Z_#Q>CV@)SF3:0[1']>D7ZO$U$X% MP-5S,4TC2K;*Y%,\5+IRSTH\."'KDJJX/!1;S18M]&R=]T? Z?,R,MS=+V5-TE,]90"85=[ MX[A2BRI,6<>6>%)8A-Q_D"^*X'E4@,3R"G3< M!3Q"-,C8NAVZ]3%552T(&#;^\*9CAO6,.:0J,L%(AR6:9_271A%R@T$PJ@T4 M+ '2;83-1"3O*[(F0A<^4S&5IW=XZ_R 0M6#V X5?\IY,?(1L P5PK,MB>VP ME3 Q=C0OKK (:V&HA<*/5_B:AZX4AM"@F&-'%CP"G;W.7(.ZRM)P!NU;\HQI MBAD:VU4D*HMX8 QW,$9W#;Z%6;=T38.J?@I,39[JG$&D?50Z*G97?UMH*I MG":<=G#%/(HXZWCE4_,T&KR"S:$P6#Q&#]P[WC,%BOZ"95O-J!5391R8::1Q M\!Z[SE7F@FP=AX!0F?3R'.?*1-", >]Y\(6T&1%V+#!U\;@J5%@B<+,D"-+>C:VXUDH%$O05Q8UDB),067$# M5GTU'HYUW%!]#(1X8:JT<5T<\=".$T GBF>MKKP@:VE?;/@J'X@$%0/\&@&1 M:L72:XEM!9XR/YQ7GY[TNX/X,#:SP+*ZQ,O+=57'"$@K(&NQB)HG*ZE/"=KN M01G;$GB$-O,;*?ZL5(DQP3M1PU=FWR*685DOH-"%:DR#8Q"@BN=[58]?Q8%% M/L&S$"4,O\:?+BG>B@X-P087I2LSJ.K,C@;%!38]*V &Y%IY2E2\$J%8GE=X M6%5<+$,O$:\0_%ITR?%15A XB7'+P?.T-*\5JM:5)PF &(6*XC%I_OG:>KR5 MKZT!C*D>L0\#I_E*3KGIE;G%0IM[>) ]N$8AMP1 9 &I=KUBR=6+&#ET9':AG>4" MIJ6R6\K7N4C5.U6 8(L:/MS9(QP35B]I+A)GBPG5 M\W0-5R;##RWL:08 YX%*!*-=^852?LI*R@4VD9=]/VVB0@L5U_= T$?BX3%+ M8UXX=7DA4GX)>LPJ+,]$A?[2\W"7\2U (S!T$@5@;#[ MYA6N;@MG%W%5&/.,+52,((JG[0=S1H0SO,5(IR8>%/(KE(+U&BZJ-BML]VL] MGS9QP>-F$6%(40$@G$NB0B8E_HR-ZO*IHE!! <9 L9 MK!64T,!@@!#8.*0;(,QKC#2$Q ]&_P$ *N;@X#& 0[H-AC4+\Q+/2X&9F4I_ M51/$_(9Q(04BN^G(? X7>CV>TNK.A_0!45B"(L5=WEE=DL<4,-'Z\9"U5OFZ%805F^8%*K M3,I<#A-)_ MT5O]>Q'A-+[*!,4WX@!?)!,8AX)-=8C[\%4-V"VC?K^4SA2.00(:TGD]Z(#VT7Y ME@QF^Q10E(]-KQ4>%E9YZD?LA5O"4"!*QBG^:Q&%#Q;SM;#&,J,O0+RN:-1H MN)7-W67J O%J3:M[BQL##B^K()SL1SZ));[[*97L_@%"Y;-E&+,'\*''5*IA:GQ*[",4*A M80S%AL*$LQ\<3<[A0R623<8* O6I)7VAC8*(@-.KC<,C+*A+05=KM!F13TS MKP#,8R]"*(7D(,^=#2PUH%([[3)RBX J5-7J;J@KS:I'@,2]'\909-"AX]7Y MU'1JS7.=4Z5E/G&(04FMMB^%;0WOPX9B\JL[@$\('&7>+J]=8OH=LF8E$G-. M(6PF8S112AJY% M> 0\O+Z'$A$8LU*$D2DP%,8"9T16\RD(M!?G%](P_SUN\ LSE&]"/E$ !U?*!DNJ^UT'0L@H::L['),1M&&.(@P/X6M1G M'30F'6+;T;(1_']E=HM@QMY4BA9:D(AHPE1"T-B@7-H]W/P4[M"D4>@_Z4(V MYE57KF=5Z+.=XDF.W+6+$KB@_C?=5EQ7)>9QKAP%PM$Z!M MZ[W<=I:PXSJ52DSGR^RZ@&UV1SZH(&+/BFQ]( &P]H=JS'5J!\B8YC&R MF)X8" 2OBX0(21UOH(KVF2.Q-@V,S?]F &1W"#@=C& Y>I=70UO=-3-1FA . MC1E=8:']=T?\^:PLC!D1+OLJ-G5[?R685';LOHK.&8&&$T'94%;L1FP"L-0= MWVM%A&,^F[2E]1A]KHBIB%RH%J7-+04,6?+(=*<%)'ML" D#C="*,7.)P[$C MR8"N,7P,RZM@VV^8>B1 G$DPAKYIK5I3L$><$",U"%-@.]*Y,TBJQB0E)PHK M>S@\DEN1.%( IHT(23.(8,;@!D^/&1X&9Q3L"KW/V12J6NOZ8;A6CV-4\W]* M?QRUG=<:]+$%")='B*02QDC(_@@>UG4]G#K10X%2D17W!?.Q*'2EX(;Y"B[S M=L.D*6MF/2$;SMU%B%//QV 0< KF&]6"J;148@F%*'1%U^7&2#]-05HA?D6 M1YQ ; DF!8[#VN X110:-H8]4I\Q,YV18:$$VA/V#!*O:%-U>)F(=K)R>%_A MCDW>-?^Z;#@2+.$?P@<5[H:3^AC;$&B8J.?U4(HWLBL6GYNK!&U<<.8AF7=; M,I[%'%U7GG^),2-"A%/D0!Z.?1'22@6)R?O >1$LR%10'WA*!N;VYJF5J$D8 ML.93.H/?"VTU;L2P Q[WP80I\K6@8,GEC9$+>10E"N(PGPC/5RE]P$I N*+. MK<@4FX=@Q P''V%_ ".O]R])8&+ZZN;;&S/5%6VB(68XYIPWR4;CI+ 6X)# M/3]YCA\-I+CN=VQ(P@[EU@3*@F4I%C8..3!1<9NX.;:"1RZ[M$V1!"+&T5N! M2P4%>S?)=W 1T@)<-5[@]UJ^PI9!7^V.B_#&=I6$()6$"E>:N*O*SWK%-NK! M!V6:V*+/@O+@@^ACKQ]@$@:NKY8+"-FN"+A5U3.2;%UQ? 0A1>NC8)$?[F;GIT_2, MS&EVC$EZ@4@ARH\MM G/SZRCE:_Z* &$4(S7!MGK7*U"<&T"@H1<74RE"KW# M+\:^&/QXHP2^W'#!(C% M#>Y88>JR5ATJ;N?6.XK2&XS&_=_KMMZV42&35:JDBG_\1L+)ZDH.NS6ZY8&\BH7"4^$N7U:LIJB)8!2_SLK$P__@^D^RLB@%\)6._F&R> @M &M]3LJN76;YM< 2O/U[59W M2#R2&;^)L$RF#5FOH!HQL@L,7BTK$B_J%(\%U^+Y%&(;TKNA>)>1%+'M^!6\ M&Z;'-B$(6@6S"7(6*JL>C%,:3"8FWQ04"07\ZB>S=L%'K.:SC%/4*YRIU1N" M^;/LVVJ&5Y"K^MYF&U@DW+&5C<342($[\X.2U9RV<"-QPH!7+Q#*K^'96'0E8O$QB!C4O@O^]ZG#P@0 '[UZ*LK;TM/AA'23C(\FWC1K55 M 8#B"R/XZ%*4M:+;8@,/T[J9N[9->!/!:%+4Z*<> 57\L<.+;4Y&5;^I<$8( M"LH5H-PED! A;/!Z,W_LRB9$&H/5YD8=4BU2RD%#$N%+9>8Z@7)QO&EKWIK[ M'Z'\HSH;KS[$($1L-2D*).A,0-*?:_+&R,S,SG2SOK V^(ZOA+.C@\>7:O7& M,'69SE=;'SXR<:NA%M:W^8_?#QX?/KPU\%^6L"'!3+E3Y\4>PM)8Y?#GJG"* MP;=,8R[3F"*K0G_=G8D_UV;JF)+9FKM?4U7N@]B-Q@P.!<5I"D5OB9O'+*;Q MI5X#HPR&% ^8RBGFXL.-=2N&VH4;;&8)Y928@5A8M%.6!C\:_+R%)$#%Q@L" MX?E<8!B&+K,\>(3N,H0T7!=<;D\6NC:L746K\:3![L631TFID[S<@,QB;$F$2J M1JM+-%Z$'#BC\OP1*&XYM.I:L*CC$#>+JW*O0'*K)T??*$]_&C*I]:+A$F"I M8(TI$HD7;"8QLT")*H\-*)#%($TB=:X*F4AW;-80-\@XA6C30F&U:>O0O#SC M(<:DKG*RRKO]L"L7_MB+*]&LCM!%08'6;S&/I?'-+HF8<'E$C9BAB'77(7GU M@OKZPR!NT&TA=+3L-! IYH15!5I9*;X6\!;,RF;77TN(2RMPG6!+@-+SA#EM M>-,&18$+H?[C"59K4"22)/"0_**-YD.:ZK@RBUUHT!& -R0;,8GA5>IFJ^!? MO/^IV=W/65)G*F0W8DJA%3@(HHX_*SU7)Z4Y$RD:#]3BV@8 MN/ 8T%6-:NY ML%4&,!1#Q'UQ?6B8(7-C"0*I !/E)*BA!(MC?S7NAUHT A%A62.X&[4"J=2$ ML9ZQ)4( 2#UMM;V=4,M(X,?P.%KUX79ZU"6KA8/49 KO"!3,UBTJ,MTZMHA M>['8@GDID+<4HE/>Z_AL(]VN80?8>O>DB:7GYH:/D9O["QF:6U'&+8ON = 7 MJ/TH@BCG9(J7B+4)6$@J>1%: ",CL+Z$P$K"I,Z&>&/LX-C/P^Q>A::7'(\P MU8;&)EZ6?1?O7 U.K>ATL'4AR?X1)7JZ2A=%Y+4O#APV='R$G+Y>, MR!Z*J+V26]AW10B%4 #/]I34U3%I7\WMV<(QI\\'3RNOKRL)0_*)6>!A\&I MJ;\VO%U*N^5P&&#U/37F;B68EX1P R<2#9GTK#9>(V3F90J98(3MN%BC[P9= M'-"![=L6+E_I<+[ R"S&H6.=-XQ!>]#M_2B#=JO;"2V:Z'&+X[#^;^0_Z/WL MO,P)(0Q"WA% D+&8RX+_P :[.];5U.;73[VJ^8Z ^']SVQ.0"H,JF39EW);5 MW()]6GVX"".P@F#"E6]FI.H9R^:!53KN :CJ@9)*$9Z82 IE4N4H;MXL^=0%$HX MCAGS4L?#"9+\.'^ZQ"E Q\F*1=/-3:BVH"LZPWY) R[PYD;+S0U&?V,/O2ON5:^A'IB$-6\*X7,__D;O4:NNU[O4LTW.* E/F?P&)1@T72DI"U7]SYH7X=#]?&AYA1E^[B M0!(?A!*8[^5'10TP+<2>#US<+@F#%3V<\7#JI.'CD/CV988/6(BON$Q#?Z>L7T%QV\3P%8ND26B<8 5(01"ND\,% M6\4( GG&[<@)=3(VBKH=:3_<#. MI;HMS9*.1+AN9*>VWG&V>XH$W3BZ-+BH&:?7,@;4I2M#^\&&0)+PXH.88B#. M1I1! [T/P=T0*6@(F[,@;"CL8$]$XI2?BJT?^W+]KCWTO."A.F]&2E@L9ADX M-2"3LQ4 E%4J1(P="=N$N;R?.@\CX3@ 8/Z6&CFX]*&CH-J:DFZ;5"@6:RT MPF9A\DP 51?MT[@Z[F- (4"EW8R>GEL80 8IZF8KN%,E86<*PP-DO$HM4AC'?5=\P M**M?W.AFXT7R<%:T$>BQ-*,.(;R)!R,-.,SY=3PT_(.U285FMZ\;<.^J@@Q\ M$!T/#1K!MF]\YFA9;Z^\B#;$T"QA#\!079\B@8?&#>3AI7->83U[-UC\0P_0 MZ1;Z#?+^:803(X:'W"JIW'UC&0&/4' :ANW?E$][BM4/T>TU[OIVH1):OQ=7 M+Y2'?L.;"HG!NFM8H'U=XK+

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�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�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o: voynich@rand.org Subject: Marchand Status: OR Nate found out about a professor named Marchand: name: marchand james w phone: (217) 244-3243 address: 3072 flb mc-178, MC 178 : 707 s mathews : urbana, il 61801 department: germanic languages & lits title: ctr for adv study prof of german,comp lit & ling email to: j-marchand@uiuc.edu (marchand@ux1.cso.uiuc.edu) who has used ultraviolet photography and image processing to read previously unreadable medieval German manuscripts. I think I will get in touch with him as part of our quest to find reputable medievalists (etc.) to help us study the Voynich ms.. From ACW@YUKON.SCRC.Symbolics.COM Thu Mar 19 19:14:00 1992 Date: Thu, 19 Mar 1992 19:14-0500 From: Allan C. Wechsler Subject: Re: Voynich clam chowder and marching society To: kibo@world.std.com, ACW@YUKON.SCRC.Symbolics.COM, reeds@gauss.att.com, voynich@rand.org In-Reply-To: <9203190035.AA06117@world.std.com> Message-Id: <19920320001411.4.ACW@PALLANDO.SCRC.Symbolics.COM> Status: OR Date: Wed, 18 Mar 1992 19:35 EST From: kibo@world.std.com (James 'Kibo' Parry) But first we have to pick colors for the logo. :-) -- K. Nothing simpler. "A bright, not quite brilliant, blue ...; an opaque aquamarine; a good strong red, carmine rather than scarlet or vermilion..." From j.guy@trl.OZ.AU Fri Mar 20 22:00:31 1992 Date: Fri, 20 Mar 92 08:00:31 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9203192200.AA20594@medici.trl.OZ.AU> To: voynich@rand.org Subject: Voynich the fun way Status: OR ... some of you may have received a sneak 200K preview thanks to my clicking Mickey Mouse on the wrong microbox. And I only had had one cup of morning coffee instead of my usual three or four, so I beg those for forgiveness on grounds of diminished mental capacity. I found a cute shareware solitaire game for PCs on I forgot which archive, called MAHJONGG. It consists in removing all 144 Mahjongg tiles from a heap by matching them two by two. (Great interface, and a very nice game). And it comes with a tile editor! You've guessed it: I made a set of tiles with Voynich letters on them. The letters are white-on-green, with Currier's transliterations in red and Frogguy's in blue, thus perpetrating the centuries long enmity between les grenouilles et les rosbifs. I had to represent the four flower by the four "vowels", A, C, 9, and O. To make things easier, those four letters are yellow-on-green instead of white-on-green. For the four seasons, I used four very rare characters, which do not appear in Currier's system. One of them being the "picnic table". This led me to try to think of good mnemonic transliterations for them, which you'll see on the tiles. Comments welcome. Finally, remaining with two tiles to play with, I painted on them illustrations from the Voynich, which I am sure you'll recognize instantly! Jim Gillogly has put it, zipped, in pub/jim/solitr.zip Nota. I could have done a better job of it. And I will, some day. From ACW@YUKON.SCRC.Symbolics.COM Mon Mar 23 17:33:00 1992 Date: Mon, 23 Mar 1992 17:33-0500 From: Allan C. Wechsler Subject: Re: Voynich clam chowder and marching society To: rcarter@isis.cs.du.edu, reeds@gauss.att.com, voynich@rand.org In-Reply-To: <9203180530.AA19059@isis.cs.du.edu> Message-Id: <19920323223325.6.ACW@PALLANDO.SCRC.Symbolics.COM> Status: OR Date: Wed, 18 Mar 1992 00:30 EST From: rcarter@isis.cs.du.edu (Ron Carter) Regardless, I think getting a `real' and `proper' analysis of the VM has got to get done, if possible, even if the only end `result' is to date the thing. Such techniques as the infrared already mentioned, as well as viewing it under a blue/green laser to bring out erasures, etc. would no doubt unlock a few more things about the VM. I'm surprised nobody has mentioned radio-dating. If the Catholic Church allowed this for the Shroud of Turing, surely the Beinecke will allow it for the V. Maybe D'Imperio could lend a letter of support? Surely a `former NSA' title goes for something. I don't even know if she's allowed to say it. I got a real kick out of the way she danced around that very point in "Enigma". "In the fall of 1975 I read a paper on the subject to a group of colleagues. As this occasion was widely advertised within the organization, it attracted quite a large audience..." But she never says /what/ "organization". You have to work it out. Now, where could she work that (1) she's shy about saying the name of; (2) boasts a large number of "colleagues" who would be interested in V? What a /puzzle/. Not. From firth@SEI.CMU.EDU Wed Mar 25 00:29:09 1992 Date: Tue, 24 Mar 92 10:29:09 -0500 From: firth@SEI.CMU.EDU Message-Id: <9203241529.AA23782@bp.sei.cmu.edu> To: voynich@rand.org Subject: Retry Status: OR >From MAILER-DAEMON@SEI.CMU.EDU Tue Mar 24 09:43:59 1992 Received: from SEI.CMU.EDU by bp.sei.cmu.edu (5.64/2.5) id AA23738; Tue, 24 Mar 92 09:43:54 -0500 Received: from BP.SEI.CMU.EDU by sei.cmu.edu (5.64/2.5) id AA09737; Tue, 24 Mar 92 09:43:51 -0500 Received: by bp.sei.cmu.edu (5.64/2.5) id AA23734; Tue, 24 Mar 92 09:42:19 -0500 Date: Tue, 24 Mar 92 09:42:19 -0500 From: Mail Delivery Subsystem Subject: Returned mail: Cannot send message for 1 day Message-Id: <9203241442.AA23734@bp.sei.cmu.edu> To: firth@SEI.CMU.EDU Status: R ----- Transcript of session follows ----- 421 rand.org.tcp... Deferred: Connection timed out during user open with rand.org ----- Unsent message follows ----- Received: by bp.sei.cmu.edu (5.64/2.5) id AA20865; Mon, 23 Mar 92 09:41:13 -0500 Date: Mon, 23 Mar 92 09:41:13 -0500 From: firth@SEI.CMU.EDU Message-Id: <9203231441.AA20865@bp.sei.cmu.edu> To: voynich@rand.org Subject: All Work and No Play... ... is what you get once term starts and the students need to be kept fat & happy. However, this week they have off, so I might actually get to open the Voynich stuff again! In the meantime, I've made a short list of what I see as "critical" assumptions. They are critical in the sense that, if they are true, they enable further analysis of the text; but if they are false, any analysis based on them is probably useless. As you'll see, the assumptions are interconnected. A. The Voynich MS contains meaning and was intended to communicate meaning B. The MS is not encyphered, though its language may be abbreviated or synthetic C. The text was intended to be pronounceable, and could be read aloud D. The Voynich symbols represent phonemes E. Where there are figures, the nearby text is related to them F. Where there are labelled figures, the labels are names associated with the figures G. When figure labels recur in the text, they have the same meaning H. The nature and style of the figures is native to the culture that produced the MS Now I don't think we have consensus on the above; indeed, I'm a bit dubious myself about a few. But I think they represent a fair summary of many ideas proposed in this newsgroup. Finally, just to report another gestalt feeling: the MS is not by John Dee. There are too many archaisms for ca 1600, and the script is of an elegance beyond his devising. Robert From rcarter@isis.cs.du.edu Wed Mar 25 16:44:29 1992 Date: Wed, 25 Mar 92 00:44:29 MST From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9203250744.AA02770@isis.cs.du.edu> To: firth@sei.cmu.edu, voynich@rand.org Subject: Re: Retry Status: OR I would agree with all the points made except for `C' ie Voynich being pronouncable; anyway, I am thinking that it is as pronouncable as say Gregg's shorthand; sure you can `translate' it, but I don't think that Voynich is meant to be read out loud, rather it is meant to be read and interpreted. I -do- think that the VM was written -by- dictation, though. From reeds@gauss.att.com Tue Apr 07 11:18:11 1992 Message-Id: <9204070218.AA08838@rand.org> From: reeds@gauss.att.com Date: Mon, 6 Apr 92 22:18:11 EDT To: voynich@rand.org Subject: Voynich MS: the Khazar Connection, cont. Status: OR I just got a letter from Mary D'Imperio, with a Xerox of much of Stojko's "Letters to God's Eye" enclosed. I understand that she is sending a copy to Jim Gillogly, too. A brief glance shows that the author subscribes to a sort of "punctuated equilibrium" theory of linguistic evolution, veering now and then towards a form of linguistic "catastrophism." I have not yet read the religious parts yet. In brief, the VMS is a series of letters from one Khazar potentate addressed to another, his rival, Miss Mania Koza. (That's pronounced "Manya", rhymes with "Vanya".) Some zingers from page 43L (f42v?): "What are you writing? You will drown. Perhaps you are living there where Koza is fighting? You should write. You have but very little love.", and "Young Mania, you are darkening in dark Ora. Do you have that Eye of God? Where did you renovate that? Are you now God's lover? Where is your naked mind?" (Sounds to me a bit like Leonard Cohen's "Suzanne".) Other literary news: I must be the last person in the world to have read Umberto Eco's Foucault's Pendulum. Now I know why, whenever I explain my VMS researches to brothers in law, etc, I always get the response, "Why, just like Foucault's Pendulum"! His reliance on Yeats, and in particular her "Rosicrucian Enlightenment", is striking. I doubt that he read French or Clulee, though: his Dee and Talbot/Kelley do not sound like the Dee and Kelley we have come to love. Cheers! Jim Reeds From rcarter@isis.cs.du.edu Wed Apr 08 01:17:05 1992 Date: Tue, 7 Apr 92 10:17:05 MDT From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9204071617.AA19870@isis.cs.du.edu> To: j.guy@trl.oz.au, voynich@rand.org Subject: Re: Is the VMS in *Martian*? Status: OR Along those lines (and I have no problem with these premises) then I would think that point 2 (VM is a word for word rendition of the original language) would be the easiest(??) avenue of attack; after all, how many languages -do- have repetitions? Not many, I suspect. I still haven't finish my serpent search (one more source to check), and am working on more latin abbreviation research; found a 1172 document that might have some clues on an approach. Ron. From RJYANCO@amherst.edu Wed Apr 08 02:53:00 1992 Date: Tue, 7 Apr 1992 12:53 -0500 From: RJYANCO@amherst.edu Subject: Voynich word repetition To: voynich@rand.org Message-Id: X-Envelope-To: voynich@rand.org X-Vms-To: VOYNICH Status: OR If it has been overlooked, there _is_ word repetition in (ancient) Greek. e.g., this man's glory travels to heaven to >toutou tou< andros kleos eis ouranon erchetai the this the man glory to heaven travels (nom)(gen) (gen)(gen) (nom) (acc) "tou" is the male genetive (possessive) article. If you had a string of male objects, each posessing the next in line, you could have something akin to "the wealth of the guest of the disciple's master" which could be translated like this... to mathetes to tou mathetou despotes to tou tou mathetou despotou xenos to tou tou tou mathetou despotou xenou ploutos. ^^^ ^^^ ^^^ the the the the disciple master guest wealth (to masc. nominative article; tou masc. genetive article; -ou masc. genetive suffix -es, -os masc. nominative suffix) In Greek, a possessive phrase, such as "of the disciple" is considered an adjective; thus it can be put before the noun, much as English adjectives are. (It could also be put after the noun, a la francais; or at the other end of the sentence; but that's not my point!) The *point* is that genetive articles in Greek may be put into strings. So you could find tau-omicron-upsilon (masc. or neut. gen. sing.), tau-eta-sigma (fem. gen. sing.), or tau-omicron-nu (gen. pl.) more than once in a row. -- Note also from the first example that Greek also "toutou," masc. gen. "this," "that," which could be mistaken for a double word. If you wanted a very contrived example, you could take "the wealth of the guest of this disciple's master," in which case you could get the string "tou tou toutou tou" - which is contrived, yes, but not incorrect. -- In short: word repetition does exist in at least one classical language. Rick From ACW@YUKON.SCRC.Symbolics.COM Tue Apr 07 13:56:00 1992 Date: Tue, 7 Apr 1992 13:56-0400 From: Allan C. Wechsler Subject: Re: Is the VMS in *Martian*? To: rcarter@isis.cs.du.edu, j.guy@trl.oz.au, voynich@rand.org In-Reply-To: <9204071617.AA19870@isis.cs.du.edu> Message-Id: <19920407175658.8.ACW@PALLANDO.SCRC.Symbolics.COM> Status: OR Date: Tue, 7 Apr 1992 12:17 EDT From: rcarter@isis.cs.du.edu (Ron Carter) Along those lines (and I have no problem with these premises) then I would think that point 2 (VM is a word for word rendition of the original language) would be the easiest(??) avenue of attack; after all, how many languages -do- have repetitions? Not many, I suspect. In Old High German, and I suspect in other Germanic languages, the demonstrative, relative, and third person pronouns are all conflated. This means that some of the "he who" constructions have word doubling. For example, "Thou art he who singest..." is "Du bist der der singest..." The pronouns are only identical, however, when the outer 3rd person pronoun and the the inner relative pronoun are in the same case. Thus, "Thou art he whom we call..." would not have "der der" but rather "der demo" or something like that. Can any real Germanicists check me on this? Anyhow, my feeling is that pronoun doubling is not frequent enough to account for what we are seeing in VM. Still, it might be enlightening to guess, say, that 8AM 8AM is "der der". From RJYANCO@amherst.edu Wed Apr 08 04:11:00 1992 Date: Tue, 7 Apr 1992 14:11 -0500 From: RJYANCO@amherst.edu Subject: other random stuff To: voynich@rand.org Message-Id: <0593D01320C1590F@AMHERST> X-Envelope-To: voynich@rand.org X-Vms-To: VOYNICH Status: OR I was looking at Mr. Firth's "More wild guesses" from 29 Jan 92; there do seem to be "tou" strings. If anyone wants to play around with possible Greek links, here's some stuff to play with. I'll write eta as ee, epsilon as e, omega as oo, omicron as o. articles ^^^^^^^^ singular plural M F N M F N nominative o ee to oi ai ta genetive tou tees tou toon toon toon dative too tee too tois tais tois accusative ton teen to tous tas ta There are also three other cases in Greek -- ablative, instrumental, and locative -- which were archaic by the time Greek was widely written. Plus there's a archaic third number, the dual. So there could be up to 39 more articles which I do not know. primary verb endings ^^^^^^^^^^^^^^^^^^^^ active middle or passive singular 1st pers. -oo -(o)mai 2nd pers. -eis -(e)sai, -ei, -eei 3rd pers. -ei -(e)tai plural 1st pers. -(o)men -(o)metha 2nd pers. -(e)te -(e)sthe 3rd pers. -ousi/-ousin -(o)ntai infinitive -ein -(e)sthai secondary verb endings ^^^^^^^^^^^^^^^^^^^^^^ active middle or passive singular 1st pers. -(o)n -(o)meen 2nd pers. -(e)s -(e)so, ou 3rd pers. -en -(e)to plural 1st pers. -(o)men -(o)metha 2nd pers. -(e)te -(e)sthe 3rd pers. -(o)n -(o)nto -- If that leads anywhere, which I somehow doubt, I'll try to find a book discussing the archaic forms. Rick From j.guy@trl.OZ.AU Wed Apr 08 06:08:59 1992 Date: Tue, 7 Apr 92 16:08:59 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9204070608.AA05343@medici.trl.OZ.AU> To: voynich@rand.org Subject: Is the VMS in *Martian*? Status: OR I have nothing to show for my labours yet, so here is a piece of useless digression. I found a book about artificial and imaginary languages ("Les fous du langage" by Marina Yaguello, Editions du Seuil, 1984) where a glossolalic language made up by one Helene Smith (French, despite her name) is treated at length. The language is Martian (Helene Smith later went into Uranian too), and some of its statistical properties are strangely reminiscent of the Voynich language. Just consider: "50% of Martian words end in e-acute [Voynich: 50% in <9>] 50% of the words are disyllabic, about 20% monosyllabic, roughly as many are trisyllabic, and the rest (6%) tetrasyllabic. The typical Martian word is disyllabic and end in e-acute. The phonological structure of Martian and its syllabic patterning make it monotonous and weak in contrasts. Alliterations and assonances in i and e-acute are at once too numerous and too poor." Here is a sample of Martian (the spelling is "French phonetic"). I'll only give the interlinear French translation, for the meaning matters little. What matters is how Martian is a word-for-word encoding of French sentences, and how Martian words are often distant echos of the corresponding French words. dode' ne' ci haudan te' mess me'tiche astane' ke' de' me' ve'che ceci est la maison du grand homme Astane que tu as vu mode' ine' ce' di ce'vouitche ni e^ve' che' kine' Linet me`re adore'e je te reconnais et suis ton petit Linet i mode,' me'te' mode', mode' ine', palette is O me`re, tendre me`re, me`re bien-aime'e, calme tout che' pe'liche', che' chire' ne' ci ten ti vi ton souci, ton fils est pre`s de toi. ce' e^ve' ple^va ti di be'ne`z e'ssat riz te`s mide'e dure'e je suis chagrin de te retrouver vivant sur cette laide terre Enough, note the parallels between Martian and French: Martian French te' du ti de de' tu di te dure'e terre vi vous (you, polite), translated by "toi" (you, familiar) ce' je ve'che voir (to see), translated by "vu" (seen) ce'vouitche savoir (to know), translated here "reconnais" (recognize) ple^va pleurer (to weep), pleuvoir (to rain) --> sorrowful (chagrin) te`s cette ("te`s" = "cette" backwards) riz sur ("riz" = perhaps "sur" backwards) Here is the frequency distribution of the Martian vowels (French-like "phonetic" spelling again): e' 39.3% i 30.6% a 13.4% e` 8.9% o 3.0% eu 3.0% u 2.45% ou 1.1% an 0.6% eu^ 0.4% (it adds up 100%) So Martian is a word-for-word mapping of French, with bits and pieces of the French words peeking through. What Martian does NOT have, is whole words repeated, like the Voynich has. What am I getting at? Let us suppose that Voynichese is such a made-up language. If so, these are not unreasonable assumptions: 1. Voynichese phonology is simpler than that of the original language (like Martian has fewer vowels and consonants than French). 2. Voynichese is a close word-for-word rendition of the original language. Since Voynichese is replete with repeated words, so the original language must have been. 3. The text must make some sense (like the Martian corpus does). Under those conditions, even though Voynichese would be a fabrication, like Martian (and like Enochian), it should be possible to decipher it, and perhaps even to identify the language which it apes. From j.guy@trl.OZ.AU Wed Apr 08 22:07:18 1992 Date: Wed, 8 Apr 92 08:07:18 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9204072207.AA05975@medici.trl.OZ.AU> To: voynich@rand.org Subject: repeated *Martian* words Status: OR What the Voynichese repetitions remind me most of is Indonesian first (and other languages of the Indonesian archipelago), then Chinese and especially Japanese. For repeated words in V can be quite long e.g. 4oqpc89 4oqpc89 (or something to that effect, I'm quoting from my sieve-like memory). Thus Indonesian: barang-barang, anak-anak, laki-laki etc., Japanese: hito-bito, toki-doki, etc. Yes, Greek. I remember from my school days: ho thanatos pauei ton tOn anthropOn bion. Tahitian, incidentally, does a similar thing. However, does one ever encounter such a double embedding as "to tou tou mathetou despotou xenos" in Greek (unless a deliberate linguistic joke)? I doubt it. It's like "the grain that the cockroach that the rat that the cat ate ate ate". Sure, it's "theoretically" possible, according to certain theory of language. But...! In fact, as pointed out, you can easily have seeming repetitions in Greek without resorting to deep embedding (red alert: some very rusty and probably incorrect Greek coming up): tou thanatou tou mathetou toutou despotou of the death of the disciple of this master And if only I could have thought of a few nouns starting with "tou-", it would have been even better! (or should it have been "tou toutou despotou"? Hm, no, I don't think so: toutou <-- tou autou doesn't it? My Greek is rusty, it falls apart...) A fleeting thought about the different authors of the VMS: it could still be one person, with multiple personalities. So we're back to square one on that count again. From cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET Thu Apr 09 01:54:00 1992 Message-Id: From: cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET (John Cowan) Subject: Re: Nevbosh (was: Is the VMS in *Martian*?) To: voynich@rand.org (Voynich List), conlang@buphy.bu.edu (conlang) Date: Wed, 8 Apr 92 12:54:00 EDT In-Reply-To: <9204070608.AA05343@medici.trl.OZ.AU>; from "Jacques Guy" at Apr 7, 92 4:08 pm X-Mailer: ELM [version 2.3 PL11] Status: OR Jacques Guy writes on the Voynich Manuscript list: > I found a book about artificial and imaginary languages ("Les fous du > langage" by Marina Yaguello, Editions du Seuil, 1984) where a > glossolalic language made up by one Helene Smith (French, despite her > name) is treated at length. The language is Martian (Helene Smith later > went into Uranian too), and some of its statistical properties are > strangely reminiscent of the Voynich language. > > > Just consider: > > "50% of Martian words end in e-acute [Voynich: 50% in <9>] > > 50% of the words are disyllabic, about 20% monosyllabic, roughly > as many are trisyllabic, and the rest (6%) tetrasyllabic. The typical > Martian word is disyllabic and end in e-acute. > > The phonological structure of Martian and its syllabic patterning > make it monotonous and weak in contrasts. Alliterations and assonances > in i and e-acute are at once too numerous and too poor." > > Here is a sample of Martian (the spelling is "French phonetic"). I'll > only give the interlinear French translation, for the meaning matters > little. What matters is how Martian is a word-for-word encoding of > French sentences, and how Martian words are often distant echos of the > corresponding French words. > > > dode' ne' ci haudan te' mess me'tiche astane' ke' de' me' ve'che > ceci est la maison du grand homme Astane que tu as vu > > > mode' ine' ce' di ce'vouitche ni e^ve' che' kine' Linet > me`re adore'e je te reconnais et suis ton petit Linet > > i mode,' me'te' mode', mode' ine', palette is > O me`re, tendre me`re, me`re bien-aime'e, calme tout > > che' pe'liche', che' chire' ne' ci ten ti vi > ton souci, ton fils est pre`s de toi. > > ce' e^ve' ple^va ti di be'ne`z e'ssat riz te`s mide'e dure'e > je suis chagrin de te retrouver vivant sur cette laide terre > > > Enough, note the parallels between Martian and French: > > Martian French > te' du > ti de > de' tu > di te > dure'e terre > vi vous (you, polite), translated by "toi" (you, familiar) > ce' je > ve'che voir (to see), translated by "vu" (seen) > ce'vouitche savoir (to know), translated here "reconnais" (recognize) > ple^va pleurer (to weep), pleuvoir (to rain) --> sorrowful (chagrin) > te`s cette ("te`s" = "cette" backwards) > riz sur ("riz" = perhaps "sur" backwards) > > Here is the frequency distribution of the Martian vowels (French-like "phonetic" > spelling again): > > e' 39.3% > i 30.6% > a 13.4% > e` 8.9% > o 3.0% > eu 3.0% > u 2.45% > ou 1.1% > an 0.6% > eu^ 0.4% (it adds up 100%) > > So Martian is a word-for-word mapping of French, with bits and pieces > of the French words peeking through. What Martian does NOT have, is > whole words repeated, like the Voynich has. > > What am I getting at? Let us suppose that Voynichese is such a made-up > language. If so, these are not unreasonable assumptions: > > 1. Voynichese phonology is simpler than that of the original language > (like Martian has fewer vowels and consonants than French). > > 2. Voynichese is a close word-for-word rendition of the original > language. Since Voynichese is replete with repeated words, so the > original language must have been. > > 3. The text must make some sense (like the Martian corpus does). > > Under those conditions, even though Voynichese would be a fabrication, like > Martian (and like Enochian), it should be possible to decipher it, and perhaps > even to identify the language which it apes. This language looks closely related (in spirit, not in any genetic way) to the language Nevbosh which J.R.R. Tolkien describes in his essay on conlangs, "A Secret Vice" (in >The Monsters and the Critics and Other Essays<, ed. Christopher Tolkien, Houghton Mifflin 1984, ISBN 0-395-35635-0): .. an idiom called >Nevbosh<, or the 'New Nonsense'. It still made, as these play-languages will, some pretence at being a means of limited communication -- that is, in the lower stages the differentiation between the argot-group [of conlang inventors] and the art-group is imperfect. That is where I came in. I was a member of the >Nevbosh<-speaking world. .. I shared in the vocabulary, and did something to affect the spelling of this idiom[;] it remained a usable business, and intended to be.... [I]t was good enough for letters, and even bursts of doggerel song. I believe I could still write down a much bigger vocabulary of >Nevbosh< that Busbecq recorded for Crimean Gothic [an Eastern Germanic language still spoken in the Crimea in the sixteenth century -- CJRT], though more than 20 years [to 1931 -- JC] have gone by since it became a dead language. But I can only remember entire one idiotic connected fragment [interlinear translation by JC]: dar fys ma vel gom co palt 'hoc there was an old man who said 'how pys go iskili far maino woc? can I possibly carry my cow? Pro si go fys do roc te for if I were to ask it Do cat ym maino bocte to sit in my basket De volt fac soc ma taimful gyro'c!' it would make such a fearful row .. In >Nevbosh< we see, of course, no real breaking away from 'English' or the native traditional language. Its notions -- their associations with certain sounds, even their inherited and accidental confusions; their range and limits -- are preserved. >Do< is 'to', and a prefixed inflection marking the infinitive. >Pro< is 'for, four' and the conjunction 'for'. And so on. This part is not then of any interest. Only on the phonematic side is there much interest. What directed the choice of non-traditional sound-groups to represent the traditional ones (with their sense-associations) as perfectly equivalent counters? Clearly 'phonetic predilection' -- artistic phonetic expression -- played as yet a very small part owing to the domination of the native language, which still kept >Nevbosh< almost in the stage of a 'code'. The native language constantly appears with what at first sight seems casual unsystematic and arbitrary alteration. Yet even here there is a certain interest -- little or no phonetic knowledge was possessed by its makers, and yet there appears an unconscious appreciation of certain elementary phonetic relations: alteration is mainly limited to shifting within a defined series of consonants, say for example the dentals: d, t, [thorn], [eth], etc. >Dardocatvoltymm/n<, though technically made at different contact points, have in their nasality and resonance a similarity which overrides the more mechanical distinction.... .. The intricate blending of the native with the later-learnt is, for one thing, curious. The foreign, too, shows the same arbitrary alteration within phonetic limits as the native. So >roc/'rogo' ask; >govelgompys/can -- from French; si/if -- pure plagiarism; >paltaimvoltfyscofarwoc< is both the native word reversed, and connected with >vacca, vache< (I happen to remember that this is actually the case); but it bred the beginnings of a code-like system, dependent on ENglish, whereby native >-ow< became >-oc<, a sort of primitive and arbitrary sound-law: >hocgyrociski-li< 'possibly' is odd. Who can analyse it? I can also remember the word >lint< 'quick, clever, nimble', and it is interesting, because I know it was adopted because the relation between the sounds [of] >lint< and the idea proposed for association with them gave >pleasure<.... Certainly, just as in real languages, the 'word' once thus established, though owing its being to this pleasure, this sense of fitness, quickly became a mere chance symbol dominated by the notion and its circle of association, not by the relation of sound and sense -- thus it was soon used for mental quickness, and finally the normal >Nevbosh< idiom for 'learn' was >catlint< (become 'lint') and for 'teach' >faclint< (make 'lint'). -- cowan@snark.thyrsus.com ...!uunet!cbmvax!snark!cowan e'osai ko sarji la lojban From j.guy@trl.OZ.AU Sat Apr 11 01:48:48 1992 Date: Fri, 10 Apr 92 11:48:48 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9204100148.AA07992@medici.trl.OZ.AU> To: voynich@rand.org Subject: Entropy of Languages A and B Status: OR I have computed the character entropy of Languages A and B, for whatever that may be worth: Language A Language B h0 5.12928 5.16993 h1 3.89601 3.80780 h2 2.77654 2.42303 h3 2.54875 2.17746 h4 2.19354 1.93466 h5 1.61913 1.55310 h6 0.95743 1.09859 Letters counted total: 26,578 28,894 different: 35 36 (Those are not the complete corpora, only as much as would fit in RAM and the infamous 64K limit of Pascal arrays. I am still working on a version of Monkey not hampered by the 64K limit, and perhaps even will use extended memory). For comparison purposes, here are the values of the entropy for a Russian corpus of some 3300 letters, transcribed phonemically, and the values of the entropy computed on chunks of Languages A and B of roughly the same sizes: Russian 1 Russian 2 Language A Language B h0 4.95420 5.35755 5.08746 4.85798 h1 4.51855 4.63200 3.96547 3.81110 h2 3.60375 3.60502 2.76647 2.52652 h3 2.34602 2.27874 2.27875 2.10048 h4 0.93049 0.86784 1.54470 1.55411 h5 0.23678 0.22232 0.78450 0.93415 h6 0.07273 0.06345 0.30939 0.46629 Letters counted total: 3436 3333 3515 3514 different: 31 41 34 29 The values for "Russian 1" were computed on a file where palatalization was transcribed as affecting the vowels (thus reflecting the spelling), in "Russian 2" as affecting the consonants (thus reflecting the accepted analysis of Russian phonology). I picked Russian because the number of different characters used in those transcriptions was roughly the same as those in Currier's transliteration system. Note how fast the values of the entropy drop for Russian: 6th-order character entropy is right down to 0.07, which means that, given any string of 5 letters, the 6th one is almost entirely defined. Not surprising, since Russian words are often quite long. Nothing of the kind in Voynich, far from it. What does that mean? In my thinking, that Voynich words are short. The higher-order entropy is much greater in Language B than in Language A. In other words, there is more chaos, more variety, in Language B. What does that mean? Author A writes with a wooden tongue. Remember also what we know about their style of writing: Author A writes in a clear, unhurried, well-spaced hand; Author B's handwriting is cramped, he packs more per line, his letters are slanted to the right. Does what the values of the entropy tell us concord with what their handwriting suggests, then?. Is there a graphologist out there? Has anyone got a ready corpus of texts in a monosyllabic language (say Chinese, and, better still, Classical Chinese), and another in an Indonesian-like language (with frequent full-word repetitions)? It might be interesting to see what the entropy is for such languages. Note. I counted * (doubtful letter) as a separate letter in the Voynich texts. Spaces and end-of-lines were disregarded, as they were for the Russian texts. From jbaez@math.mit.edu Sat Apr 11 04:37:08 1992 Date: Fri, 10 Apr 92 15:37:08 EDT From: jbaez@math.mit.edu Message-Id: <9204101937.AA22315@riesz> To: voynich@rand.org Subject: nonsense? Status: OR I still favor the hypothesis that the Voynich tongue is nonsense, and would like to cite Guy's new entropy calculations as evidence for this. As he points out, though, it co ld simply be due to the fact that the words are short. To attack this problem one could simply keep track of entropy within words, i.e., in calculating the n-character entropy, just consider n-character strings which all lie in the same Voynich "word". (Here "word" is in quotes, since all it means is a string of characters with a space on either sid: it may have no semantic significance even if the Voynich is meaningful.) I have no intuitions as to how the personality of A and B as revealed by their handwriting and pictures correlates to the entropy of their writings. It is interesting that B is the one who wrote the long passages without pictures at the end of the book -- i.e., he may be more fond of writing, more articulate, hence more entropic???? (A wild guess.) The results of infrared photography of the vellum breviary were disappointing but I'll describe them later. \ From reeds@gauss.att.com Sat Apr 11 06:39:19 1992 Message-Id: <9204102149.AA02635@rand.org> From: reeds@gauss.att.com Date: Fri, 10 Apr 92 17:39:19 EDT To: voynich@rand.org Subject: Voynich entropy blues Status: OR After reflection, experimentation, and correspondence with Jacques, I just plain do not trust his recent h4, h5, and h6 estimates for Voynich or Russian. The h1, h2, and h3 values might be good to 2 significant figures, but the others I think are completly bogus. Jim Reeds From jbaez@math.mit.edu Sun Apr 12 05:01:50 1992 Date: Sat, 11 Apr 92 16:01:50 EDT From: jbaez@math.mit.edu Message-Id: <9204112001.AA09443@paley> To: voynich@rand.org Subject: Voynich entropy - why so blue? Status: OR Why don't you trust those entropy figures, Jim? From kornai@Csli.Stanford.EDU Mon Apr 13 19:04:56 1992 Date: Mon, 13 Apr 92 03:04:56 PDT From: kornai@Csli.Stanford.EDU (Andras Kornai) Message-Id: <9204131004.AA24111@Csli.Stanford.EDU> To: voynich@rand.org Subject: H(n) -- watch out! Status: OR I liked Jacques' not-so-random thoughts about entropy -- I sort of agree with half of what he is saying and sort of disagree with the other half. This being an email.discussion I will concentrate on the half I disagree with. Entropy is *not* a measure of the compressibility of a finite corpus. Any finite corpus can be compressed down to 1 bits by using a compressor that stores the corpus as part of its source code. It wouldn't be a great compressor for other things but so what. It therefore makes no sense at all to talk about the entropy *as such* of Joyce's work, the VMS, or anything of the sort without specifying a SOURCE MODEL. The whole thinking is based on the assumption that what we see is but a sample from a larger population, be it the "Joyce's language", "the Voynich language", or "the English language". When we create n-gram statistics, *with or without smoothing*, we wish to establish the population distribution (numbers between 0 and 1 in an n-dimensional cube of size k (number of letters in the alphabet) that sum to one) one the basis of a sample (of numbers between 0 and 1 arranged in the same cube). For 1-grams it is sort of reasonable not to do any smoothing: this means that we assume that no symbol that does not actually appear in the Voynich will later appear when other manuscripts of the same general character are found. For 2-grams this is not such a reasonable assumption anymore but there is safety in numbers: if we have a large enough sample any missing 2-gram is more likely to be the result of some "tactic" property of the source than of small sample size. Beyond 3-grams, the actual sample size (200k or so) does not really warrant such assumptions. This is simply because the alphabet has on the order of 16--32 symbols, so there are 2^16--2^20 4-grams to be considered. (My calculations suggest that even for 3-grams we should be extremely cautious, but this is besides the main point.) We *know* that a 0 in some cell of the array is not a reliable estimator of the distribution value in that cell. To me this means that if we want to do anything halfway reasonable we *must* do some smoothing. How you do the smoothing is another matter: the cleanest math comes with assuming a markovian source and doing some bayesian estimates with that. There is a slight/tremendous (I leave it to Jacques to pick the proper adjective) problem here: the VMS source, *if it's a language*, is not markovian. To conclude on a more cheerful note, one *can* derive some estimates without the markovian assumption -- I'll try to do so Real Soon Now. Andras Kornai PS. Jim Reeds gave me a very reasonable task that I can't at the moment perform for lack of memory. If there is a team member out there with access to a unix box with 64MBs, please send me e-mail. Otherwise I'll have to rewrite a bunch of software that I don't feel like rewriting (ballpark two weeks of evil memdebug stuff -- will take me MONTHS to do this on my spare time). From j.guy@trl.OZ.AU Mon Apr 13 23:30:10 1992 Date: Mon, 13 Apr 92 09:30:10 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9204122330.AA09558@medici.trl.OZ.AU> To: voynich@rand.org Subject: Entropy (the Raw and the Cooked) Status: OR (Saturday night) I've had more time to think about Jim Reeds' objections, during which I've oscillated between "OK, I'll eat my hat on that", "No I won't", and "I'm stuffed if I know". But the night is still young (only past 10 pm) and I've only poured myself the first whisky of the week-end, so nothing's resolved. Right now, it's "I know, and I am still stuffed if I know which way to go about it." All right, the first interesting book I see on my desk (I don't rate DR DOS 6.0 User Guide as an interesting book) is "Ambrose Bierce's Civil War", Regnery Gateway, Washington, 1988. I pick a page, close my eyes, and there: "It was a campaign of excursions and ". That was 36 characters, including spaces. The next character is "a". (p.174). I don't think that particular sequence occurs anywhere else in the book. If so, that "a" is entirely predictable from its context alone ("It was a campaign" etc.). In itself, then, it carries no information at all. Next, I don't think any sequence of 37 characters occurs more than once in Ambrose Bierce's Civil War. If so, any character of that book is entirely predictable from the 36 around it. In other words: there is no uncertainty about what the 37th character of any excerpt from Ambrose Bierce's Civil War is, given the 36 that precede it (or follow it, or sandwich it). In jargon: the 37th-order character entropy (uncertainty) is zero. Does that mean that the 37th-order character entropy of English is zero? Certainly not. Because, it you include the unpublished works of one Jacques Guy, who is about to write "It was a campaign of excursions and merry-making" (now he's done it), you can no longer predict what comes after "It was...", and the 37th-order entropy of Ambrose Bierce's Civil War and chairman Jacques Guy's (unpublished) thoughts is not zero. What have I just done? I have stuffed up everything. In one foul soup (fell swoop), I have confused the properties of Ambrose Bierce's works with those of English as she gets wrote by furriners. (Sunday morning, very early, half determined to cling to my hat) I toss a coin 7 times: THTTTHH. What does that allow me to say? That the coin is fair, rather, there is no evidence that it isn't. Now forget that the sequence THTTTHH was generated by tossing a coin. In the absence of that knowledge, what does the sequence, the message, "THTTTHH" tell me? Oh sure, it could represent the outcome of seven tosses of a fair coin. It could also be the outcome of seven rolls of a very loaded die. It could be a representation of a sentence in which every word with an even number of letters has been replaced with "H", odd with "T". It could be.... you name it. I now take an excerpt out of the Voynich manuscript. We do not know what it is the outcome of. Deliberate thought? Automatic writing? If that excerpt is short enough there is absolutely nothing I can rightfully say about it without making preliminary assumptions. Now, Jim Reeds generates a text using a Markovian process, and observes that some statistics (say the 4th-order character entropy) computed on that text in a certain manner (using the raw frequencies of 4-character sequences) is a gross underestimate of the true value used in generating it. And I agree. It would be hard not to, and I won't go into why. Where I am half inclined to cling to my hat like grim death is this: if I agree with Jim Reeds that the entropy of a text ought to be computed not on raw frequencies, but via a smoothing function, so that it is a better estimate of the value used to generate it, am I not assuming that that particular text was generated by that particular method? In the case of Ambrose Bierce's Civil War, that assumption is false. In the case of the Voynich manuscript it is begging the question: is it meaningful, or is it thoughtless gibberish? And yet, I'm half inclined again to eat my hat, because it's bloody annoying to see the 4th-order character entropy of the same text going up and down like a damn yo-yo depending whether you compute it on chunks of 3,000, or 30,000, or 6,000, or 50,000 characters. And again... I just wrote there "the same text". "The same text" indeed? I don't know that. I assume it. Back to my blind selection of an excerpt from Ambrose Bierce. I will have to start believing in meaningful coincidences. For, if you open that particular edition at p.174 and read on you see: It was a campaign of "excursions and alarums," That had me puzzled for a while. "Alarum"? What flower is that? Then: oh, a misprint for "alarm". Well, come to think of it, no. Not a misprint, but the deliberate representation of someone's pronunciation of "alarm". Evidence of two different text-generation processes (I hate those words) at work. Even granting that each process is stochastic (which I hold false), what does that entail? I can't figure it out. Another thought. One of the unsolved problems in deciphering unknown texts is finding word breaks. So I have this page of transcribed Voynich, say, 1000 letters of it. I count the frequencies of every letter, of every string of 2 letters, of 3 letters etc, to find, eventually, that no string of 12 letters occurs more than once. What does that mean? That I need not bother looking for words more than 11 letters long. There might be some, but I'll find them only by default: they'll be what is left after I have segmented the text into the shorter, repeated words in there. It is reasonable for me to think that, taking the whole Voynich MS into account, I would certainly find many sequences of 12 characters that occur far more than once. That is, Monkey's h(12) on the whole would be, say, 1.82, when, on that particular page, it's precisely zero. Say I use a smoothing function, which gives me h(12) = 1.9 plus or minus a bit in both cases. It feels quite reasonable. Indeed quite an improvement. But I wonder if it helps in the analysis of the language. I also wonder if, by doing so, I have not smuggled in the hidden assumption that the Voynich is gibberish (the result of a random process). If there is not a subtle case of overfitting the data lurking there. Of correcting "alarums" to "alarms". I'm really stuffed if I know. The only way out I once could think of was this. We take a chunk of Voynich (or anything else for that matter) of N characters, and ask these questions: 1) how likely is it be random? 2) how close is it to randomness? In other words, roughly put, a sort of chi-squared (question 1) and phi (question 2). For instance, taking this text: "abracadabra", we assume that it has been randomly generated like this: we have put 5 a's, two b's, one c, one d and one r in a hat and have drawn 11 letters one after the other, putting each back in the hat each time. The question is now: is that a reasonable assumption? To me, it looks like "yes, pretty reasonable". But ask the same question of "cvcvcvcvcvc" now. Rather unlikely to be random. And that would be a very useful question to be able to answer. Because, look at a bit of Voynich in my transcription system (off the top of my head now): "8aiiv 4oqpc89 4oqpc8aiiv". It is very obvious that that is quite unlikely to be the output from a random character generator (a hat with 3 8's, 2 a's etc. in it). But it would likely be random if the contents of the hat were: two 8aiiv's, two 4oqpc's and one 89. Well, I expect we'll all agree that my is less than a Voynich letter, and that <4oqpc> is more than one, that, in fact, the truth is something like this: "8.a.iiv.4.o.qp.c.8.9.4.o.qp.c.8.a iiv". Funny, without even going into any statistics, that looks less likely to be the outcome of a random draw than "8aiiv.4oqpc.89.4oqpc.8aiiv", doesn't it? Well, that it is precisely the idea that Sukhotin floated 30 years ago now: Given several conflicting re-interpretations of a text, the best are the least random-like. (Not quite what he wrote, I doctored it to make it appear that there can be several best or close best interpretations of a same corpus. In my view, that would go to explain how languages fall apart into dialects and local varieties, and why no language learnt naturally can remain stable through time). What I have always had at the back of my mind in computing the entropy of texts is to use it as a measurement of randomness, you know now for what purpose. (Sunday high noon now. I have taken a break and walked down to the newsagent's, to see if they hadn't received Fortean Times. They hadn't. But, on the remaindered book trestle, I saw: "The Gadfly -- The International bestseller -- E.L. Voynich". A bit of a coincidence, because it doesn't seem that there are very many around: First published in Great Britain 1897 by William Heinemann Ltd. Reissued 1973. This edition published 1991 by Mandarin Paperbacks. I bought it of course, along with two other paperbacks -- it was $2.95 for one, $7.95 for three, so... The other two are "The Chain of Chance" by Stanislaw Lem in the blurb of which I read: "...in an attempt to decipher the links in an increasingly mystifying chain of coincidence"; the other is "The Wine-Dark Sea", by one Robert Aickman and the blurb starts: "Robert Aickman's strange stories are the subtle and leisurely explorations of psychological displacement and paranoia." Back to entropy paranoia now. Where was I?) Now if Jim Reeds will assure me that the "cooked" entropy is a better way to achieve that aim, AND if he provides me with the cooking recipe in a language I can understand (Pascal, C, Basic, even Fortran will do), I'll gladly eat my hat (cooked) and incorporate the recipe into my next Technicolor (tm) Cinemascope (tm) production: Son of Monkey. I've just re-read all those pearls of wisdom, and this one is worth repeating: "if I agree with Jim Reeds that the entropy of a text ought to be computed not on raw frequencies, but via a smoothing function, so that it is a better estimate of the value used to generate it, am I not assuming that that particular text was generated by that particular method?" A dreadful shiver shoots down my spine: isn't that a portent that indeed I will have to eat my hat? Pass me the salt -- and pepper -- please, shades of Robert Firth's oniric butler. Which shall I start reading now? The Gadfly ("Driven underground by the tyrannical heel of Austria, the youth of Italy prepares to sacrifice everything for principle, patriotism and the man called the Gadfly -- the man whose stinging pen voices the struggle for human conscience and so fatefully tests the antagonist he loves") or The Wine-Dark Sea ("psychological displacement and paranoia"). The latter is too close for comfort. I am alarumed. Finally, Monday morning, I decided to eat my hat and still have it. I remembered one of the applications of Monkey I had in mind. Yes, I had forgotten: I don't take notes of the ideas that come to me usually because, when I do, I always end up mislaying them. The "raw" entropy of a text is minimum amount of storage in which that text can hold. For instance, say the "raw" 4th-order entropy of folio 80v of the VMS is 0.3, and that there are 1000 letters on it. That means that, knowing the frequencies of 4-character strings on that folio, it can be compressed down to 0.3 bits per character, i.e. 0.3*1000/8 = 25 bytes. One of my pet theories is that we have such a data compression and decompression algorithm hard-wired in our brains. When I thought of it, I also remembered an experience I had had much earlier when I was studying Chinese at the Langues Orientales. It was a tutorial, and we read aloud in turns I forgot what from perhaps the People's Daily (Renmin Ribao) or Red Flag (Hong Qi), thrilling stuff. I could read very fluently and I rattled off "meidiguozhuyizhe" (American imperialists) but our tutor (Miss Wang, I remember) stopped me, horrified: "Monsieur Guy! Ou voyez-vous meidiguozhuyizhe???" Well, indeed, it had nothing to do with American imperialism for once. What I had misread "meidi" was in fact "meishu" (fine arts). Only the first character (mei) was the same. "Di" (emperor) and "shu" (skill) look nothing like each other either. But I was so used to seeing "meidi" everywhere that... you know. Thinking back on that incident, I realized that I would never have misread "mei" as followed by "di" if I had been reading Mao Dun's "Ziye" ("Midnight), a novel about Shanghai society in the thirties. There, I would have automatically assumed that "mei" was followed by "li", and read "meili" (beautiful). So I must have had at least two separate frequency tables in my head regarding Chinese. When I made that mistake, a text-generating Monkey must have had taken over, using the frequency tables I had "built" out of bits and pieces from the Renmin Ribao, Hongqi, whatever, but not from Ziye. For that reason if it is mathematically correct to speak of the infinite-order entropy of texts, or of the entropy of English, I am certain that is linguistically incorrect, for we keep our various experiences of the language in separate files as it were, along with an analysis of their properties, separate for each (certainly, we must also have some other mechanism that lets us recognize the similarities between those "files"). So there can be no such thing as the infinite-order entropy for there can be no infinite file in our heads, nor can there be anything like "the entropy of English". The entropy of Shakespeare's English, of Joyce's English, yes, but not of English. In fact, perhaps not even the entropy of Shakespeare's English, but that of some of his sonnets, of some other sonnets, of acts II and III of this play, of the whole of that other play, depending on how YOU have been exposed to Shakespeare, and I mean YOU, PERSONALLY. I do not know how that translates in terms of "raw" vs "cooked" entropy. I suspect that the "raw" entropy is useful for some analyses, and the "cooked" one for others. Not knowing the cooking recipe makes it even harder to guess which purposes each might serve best. So, having eaten my hat, I still want to have it. From j.guy@trl.OZ.AU Tue Apr 14 22:57:20 1992 Date: Tue, 14 Apr 92 08:57:20 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9204132257.AA10810@medici.trl.OZ.AU> To: voynich@rand.org Subject: H(n) -- watch out! Status: OR The more I read this exchange about the entropy of texts, the more puzzled I become. Andras Kornai has just had me completely nonplussed. So help, please. He writes: "Entropy is *not* a measure of the compressibility of a finite corpus. Any finite corpus can be compressed down to 1 bits by using a compressor that stores the corpus as part of its source code. It wouldn't be a great compressor for other things but so what." The way I used to look at it was this. Take a text, any text, "cvcvcvcv" will do. Given a context of one letter, you know for sure what comes next: "v" after "c", "c" after "v". If the recipient of that message has that knowledge he knows the message without receiving it. The cost of transmission is 0 (no transmission is necessary). I thought it was pretty neat, because 0 was precisely the value of H(2) as I had Monkey compute it. It led me, however, into near metaphysical speculations: should we not consider that for any message, an end-of message must be transmitted? For although the recipient knows that the message is cv repeated, he does not know how many cv's it contains. Next, I had thought: what about the beginning of the message? Is it not another "meta-message" token too? Then: "no, no need; just consider the message circular, so that its first character follows the end-of-message token". Those thoughts are about, oh, two, three years old. Now I read what Andras Kornai has just written, and that quote of it I made above. And I worry, because, to me, that quote means that the entropy *is* a measure of the compressibility of a finite corpus (the "raw" entropy, that is). Perhaps I shouldn't say "compressibility". The idea that it measured compressibility had come to me later. What I first thought of was that the raw entropy of a finite text was a measure of its transmission cost, when sender and recipient shared the same knowledge about certain properties of the text (frequencies of k-token long strings, k-grams if you prefer). I cannot see, for the life of me, where the flaw is in my reasoning here (so help me if anyone can point it out). Having then reached this notion that the raw entropy of a particular text was a measure of its cost of transmission, I went on to toy with the notion that the proper segmentation of a real-language text into its constituent words would allow a better compression ratio than generic compression algorithms -- not that it really mattered, but it was interesting, a sort of Darwinian theory of language: human language, with all its complexity, was successful because it allowed more economical communication between individuals and left more brain space (RAM if you feel so inclined to consider it) for important functions (figuring out better how to screw your neighbour for instance). Andras also writes: "When we create n-gram statistics, *with or without smoothing*, we wish to establish the population distribution (numbers between 0 and 1 in an n-dimensional cube of size k (number of letters in the alphabet) that sum to one) one the basis of a sample (of numbers between 0 and 1 arranged in the same cube)." Now, in my Darwinian theory of language, when my sender and recipient create n-gram statistics, what is it that they wish? "Wish" is the wrong word, but never mind. They wish to establish an n-dimensional cube of sorts that will result in the greatest saving in transmission and brain storage costs. My Chinese experience with "American imperialism" and the "fine arts" has led me to believe that they keep separate "cubes". It would seem adaptative too, if they were able to do what Andras says: establish the population distribution, because, at the small cost of a perhaps slightly "bigger" "cube", it would allow them to handle economically messages of a type not yet encountered without having to build another cube or fiddle with the existing ones. Yes, the more I think about it, it seems that we *must* do some smoothing to do anything halfway reasonable like Andras said. And at the same time, the more I think about it, the more I think that "no smoothing" reflects something true and real. Perhaps an earlier stage of one's language learning. Or perhaps, we have not yet evolved the ability to extrapolate the population out of the samples. Or we have, but we do it only sporadically. And another thought: what if the VMS is the entire language? If nothing else had ever been written, or spoken, or thought, in that language (or those two languages)? Andras writes: "the VMS source, *if it's a language*, is not markovian." I agree of course. But the problem is worse. The VMS source, if it's glossolalia, is not markovian either, at least, not at the character level. However, as I see the way out of it like this: let us posit that the VMS source *is* markovian, which we know to be false. Then, when we set about analyzing the VMS, recognizing for instance what constitutes a letter, or a word, we will retain as valid working hypotheses those interpretations which exhibit the least markovian properties. Does that make sense? I wonder. The proof of the pudding is in the eating, so will someone pretty please send me the math recipe that assumes a markovian source? From ACW@YUKON.SCRC.Symbolics.COM Tue Apr 14 10:50:00 1992 Date: Tue, 14 Apr 1992 10:50-0400 From: Allan C. Wechsler Subject: Some data (long message: 6335 chars in 580 lines) To: voynich@rand.org Message-Id: <19920414145052.8.ACW@PALLANDO.SCRC.Symbolics.COM> Status: OR As Mssrs. Guy and Kornai point out, n-gram statistics founder on small-sample problems for fairly small n. This made me wonder whether a diferent approach might be helpful. To wit: compile a table of all substrings that occur more than, say, 100 times in the available corpus. Well, I did it, and most of this message is filled with the result. Procedure: 1. I began with the V text as it is archived at Rand.Org. 2. I removed all {}-bracketed English annotations. 3. I removed trailing -'s and #'s from lines -- these are not used consistently in the archive copy anyway. 4. I removed blank lines. 5. I replaced ".." by ".". I admit that some of this massaging may skew the results, but I contend that because I only list relatively high-frequency strings, the problems are not severe. Herewith my results. For clarity, I have shown line-breaks as "\". The table suggests many ideas, which I will restrain myself from indulging in. V string No. of occurrences -------- ------------------ .. 14500 O 11016 9 8559 C 7651 9. 6363 8 6273 A 5498 S 5314 E 4552 F 4277 89 3420 OE 2898 89. 2876 R 2847 ..O 2819 ..S 2706 4 2670 OF 2625 4O 2578 P 2526 C8 2447 Z 2406 E. 2319 \ 2290 ..4 2230 C89 2217 ..4O 2161 C89. 2053 SC 2014 R. 2011 8A 1839 M 1815 ..8 1720 9.4 1718 AM 1701 9.4O 1668 OE. 1634 C9 1608 SO 1564 OP 1523 M. 1510 4OF 1509 ..Z 1503 C9. 1475 FC 1451 OR 1428 AM. 1409 ZC 1281 ..4OF 1276 FA 1265 AR 1150 2 1107 ..8A 1087 OR. 1084 ..SC 1076 AE 1049 9.O 1043 9.4OF 1028 89.4 991 89.4O 966 OFC 938 9\ 932 OFA 881 CO 860 AR. 856 ..SO 854 O8 830 C89.4 829 ..ZC 823 C89.4O 808 9.S 793 ..OF 781 9.8 778 N 778 AN 761 ..OE 755 8AM 729 ..OP 678 SC8 671 N. 663 AN. 649 S9 644 89.4OF 628 Q 628 4OFC 617 AE. 612 SC89 604 ZO 601 PC 598 8AM. 585 4OP 568 FS 558 PS 557 E.S 549 S9. 543 B 542 SC89. 539 PA 538 C89.4OF 537 ..9 534 4OFA 522 ..4OFC 521 89.O 511 R.S 503 9.8A 496 SOE 492 X 480 R.O 476 ..F 473 ..4OFA 464 ..4OP 456 ZC8 456 ..8AM 447 9.4OFC 445 SC9 431 E.O 431 F9 415 FC8 412 ..E 412 ..2 410 OE.S 409 ZC89 408 ..A 407 C9.4 405 E.8 404 SOE. 400 OPC 395 9.Z 395 \4 395 O. 390 SC9. 388 ZC89. 387 C9.4O 386 O89 382 9.4OFA 381 SOR 380 \4O 378 89\ 376 FO 376 ES 374 FC89 373 ..ZO 373 OPA 371 2. 371 9F 371 J 371 PO 367 ..Q 358 C89.O 355 M.O 354 FC89. 350 M.S 349 EF 344 8AR 339 9.4OP 337 AM.O 335 ..8AM. 332 9.OF 331 FAM 330 89.S 327 SOR. 326 \O 326 9.E 322 AM.S 319 F9. 319 9.SC 315 FAN 313 89.4OFC 310 OPS 308 \8 308 SCO 307 9P 305 O89. 303 S8 303 FAM. 300 COE 300 ..OFA 299 OE.O 298 9.OP 298 8AE 296 OFC8 295 ..SC8 295 OR.S 294 9.F 293 R.Z 293 P9 293 FAN. 291 OFS 291 BS 289 E.Z 289 ..SOE 286 EO 284 SA 280 OE.8 279 O8A 279 8O 279 C89.4OFC 278 ESC 272 ..ZC8 272 ..P 269 - 269 OFC89 268 OEF 268 2A 266 AJ 265 FAE 265 ..SC89 264 M\ 259 E\ 259 89.8 257 \9 257 ..OFC 256 OFC89. 255 X9 255 E.8A 254 Q9 254 9.SO 252 AM\ 247 ..ZC89 246 8AR. 243 ..SC89. 239 COE. 239 E8 239 ..89 236 ..ZC89. 234 AR.O 234 9.OE 233 J\ 230 PC8 230 OFAN 227 E.SC 227 S89 226 ..OR 226 ..SOE. 225 C9.4OF 225 OR.O 224 ZC9 223 SC89.4 222 FAR 222 X9. 221 \2 221 M.Z 220 SC89.4O 219 OFAM 219 C9.O 219 89.4OFA 218 CF 217 OFAE 216 ZC89.4 216 9.2 215 OFAN. 213 P9. 213 4OFC8 213 ..OE. 213 RA 213 EZ 211 PC89 210 \4OF 209 E9 209 ZC89.4O 207 OF9 207 2O 206 ..OPA 204 ZC9. 203 OFO 203 AM.Z 202 Q9. 202 ..9F 202 OFAM. 201 4OE 201 4OFC89 200 9.8AM 199 CO8 198 FAR. 197 8S 196 R.A 195 PC89. 194 N.O 194 ..OPC 193 AN.O 192 9.ZC 192 4OFC89. 191 ..4OFC8 191 C89.S 191 8AN 191 R.SC 190 * 190 S89. 189 FAE. 189 OPO 189 89.Z 189 AR.S 189 ..SCO 188 ..8AR 188 8. 188 OES 187 E.SO 187 C89.4OFA 186 OB 186 R\ 186 PSO 185 ..8O 184 OE.Z 184 89.SC 183 \P 183 ZOE 182 ..4OE 181 ..8AE 181 9.9 181 ..R 181 R.ZC 179 ..4OFC89 178 9.P 178 AJ\ 177 QO 177 Z9 177 OE.8A 176 ..SOR 175 SO8 175 OP9 174 89.4OP 173 POE 173 FSO 172 M.8 172 FOE 170 FS9 170 C9.8 170 ..4OFC89. 169 OE.SO 169 E.ZC 169 R.SO 169 N.S 168 ..9P 167 EFC 167 V 167 9.4OFC8 166 E.4 165 9\4 163 \B 163 ZCO 163 CA 163 AN.S 162 4OFAN 162 4OPC 162 EZC 161 89.OF 161 ..SC9 161 AM.8 160 9.8AM. 160 E.4O 159 8AE. 159 OFAE. 158 R.8 158 4OFAN. 157 ..OEF 157 9.4OFC89 156 FS9. 156 PS9 156 C2 155 COR 154 9\4O 153 8AN. 153 RO 153 9.4OE 152 ZOE. 152 OE.SC 152 C89.8 152 89.8A 152 OE8 150 ..2A 150 OR.Z 150 89.E 150 OF9. 149 OPC8 149 T 149 9.4OFC89. 147 M.SC 146 Z9. 145 E89 145 89.OP 145 ..QO 144 SC89.4OF 143 ESC8 143 ..SA 143 FCO 143 E.OE 142 COR. 141 PS9. 141 ..SOR. 141 SX 140 ..4OFAN 139 C89.4OP 139 9.OFC 139 AT 138 OFAR 138 ..SC9. 137 ..4OPC 137 4OFAE 137 R.OE 137 OESC 136 ZC89.4OF 136 BSC 136 4OFS 136 PAM 136 OPC89 135 AR.Z 135 PAE 135 ..4OFAN. 134 CO89 134 C9.S 134 AM.SC 133 OEFC 132 \8A 132 ESC89 131 C89.E 131 8AM.S 131 OEO 130 9\8 129 AE.S 129 ..S9 129 W 129 ..SX 128 C89.SC 128 8AM\ 128 O8AM 128 PSC 128 O2 128 OPC89. 127 89.4OFC8 127 OR.SO 127 ..ZC89.4 127 8OE 127 4OPA 127 ..FC 127 OP9. 126 89.OE 126 9FC 126 ..4OFAE 125 FOE. 125 9.R 125 C89.Z 125 PAR 125 N.Z 125 OFAR. 124 4OPS 124 SOF 124 ..ZOE 123 ..ZC9 123 ..OR. 123 OE\ 122 9\O 122 ..ES 122 C89\ 122 ..ZC89.4O 121 ZO. 121 C8A 121 FZ 121 AN.Z 120 T. 120 9- 120 M.OF 120 9.4OFAN 119 C9.8A 119 C2. 118 C89.OF 118 4OFAM 118 AE.O 118 ..FS 118 FSC 118 89.4OFC89 117 CO89. 117 POE. 117 89.ZC 117 J. 117 AE\ 116 FC9 116 \Z 116 R9 116 9.4OFAN. 115 O8AM. 115 ..8AR. 115 C9.F 115 SCOE 115 QC 115 AM.OF 114 9.4OPC 114 4OFAM. 113 AT. 113 ..4OPA 113 C89.4OFC8 113 OEZ 113 9.OFA 113 AE.8 113 9.89 112 OBS 111 \2O 111 OE9 111 E.8AM 111 XC 111 9S 111 98 111 PAM. 110 SC9.4 110 ..ZC9. 109 EFA 109 SOE.S 109 M.4 109 89.4OFC89. 108 9.4OFAE 108 ESC89. 108 R.8A 108 M.ZC 108 ..89. 108 \4OP 108 R.OF 108 OA 108 ..OB 107 9.ZO 107 ..AM 107 ZOR 107 M.4O 106 AM.4 106 4OFAE. 106 OR.8 106 ..S8 106 9.Q 106 ..X 106 FC9. 105 ..ZOE. 105 ..8AN 105 9.OPC 105 ..AR 105 PZ 105 C89.4OFC89 104 OPS9 104 OE.4 104 M.8A 104 OPOE 104 EA 104 AM.4O 103 OE.4O 102 SC9.4O 102 9.ES 102 ..8AM\ 102 OPSO 102 E9. 102 SOE.8 102 C9.E 102 R.9 102 S9.8 102 CP 102 D 102 From j.guy@trl.OZ.AU Wed Apr 15 22:17:38 1992 Date: Wed, 15 Apr 92 08:17:38 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9204142217.AA11803@medici.trl.OZ.AU> To: voynich@rand.org Subject: Re: Some data (long message: 6335 chars in 580 lines) Status: OR Unrestrain yourself now, and indulge. Those frequencies are amazing. 89 especially, and its bigger brothers 89.4O, C89.4O, 89.4OF, even C89.4OF, and the biggest of them all: 4OFC89 (200 occurrences! 191 of which are followed by a space! 178 preceded by a space, 169 surrounded by spaces, of which 147 are immediately preceded by 9, 108 by 89, 104 by C89). We have a word there, don't we? And 89 represents *one* sound that is common in one language (I forgot which, A or B?), and rare in the other, doesn't it? C89.4OFC89 seems to be a word too.... From jbaez@math.mit.edu Fri Apr 24 06:48:56 1992 Date: Thu, 23 Apr 92 17:48:56 EDT From: jbaez@math.mit.edu Message-Id: <9204232148.AA29972@banach> To: voynich@rand.org Status: OR I am giving a little pep talk at MIT... I already met one person who, seeing the ad, told me he has a lot of programs to analyze texts and would like to study the Voynich with them. ************************************************************************ SEMINAR SEMINAR SEMINAR SEMINAR SEMINAR SEMINAR SEMINAR SEMINAR ************************************************************************ CODE OR NONSENSE? John C. Baez, Department of Mathematics, U. C. Riverside If you were given a document in an unknown script, could you decide if it was meaningful or random? This is not merely an interesting philosophical question. The Voynich Manuscript is a 232-page document written in a code that has never been cracked - if indeed it *is* a code. It was found in 1912 in a chest in a Jesuit college in Italy. It contains a letter dating it back to at least 1666 - if the letter is authentic - but nobody knows how old it is or who wrote it. It is quite rightly known as ``The Most Mysterious Manuscript.'' I will recount the history of the unsuccessful (and often hilarious) attempts to understand it, and pass around a copy of the manuscript. I will also describe recent work using computers to analyze the text, and some directions for future research. Date: Friday May 8, Time: 3:00 pm Place: NE43-518 (Fifth Floor Conference Room) Contact: Nate Osgood, 253-6038 (hacrat@catfish.lcs.mit.edu) ************************************************************************ SEMINAR SEMINAR SEMINAR SEMINAR SEMINAR SEMINAR SEMINAR SEMINAR ************************************************************************ From reeds@gauss.att.com Thu Apr 30 01:13:24 1992 Message-Id: <9204291617.AA13190@rand.org> From: reeds@gauss.att.com Date: Wed, 29 Apr 92 12:13:24 EDT To: voynich@rand.org Subject: Voynich Enochian font! Status: OR FYI, Just saw this on comp.fonts: Path: alice!andante!princeton!jvnc.net!yale.edu!think.com!mips!decwrl!csus.edu!netcomsv!satyr!apple!equinox!jimi!cash.isri.unlv.edu!beleg From: beleg@cash.isri.unlv.edu (Andrew Bagdanou) Newsgroups: comp.fonts Subject: Enochian PostScript fonts. Message-ID: <1992Apr24.210056.29166@unlv.edu> Date: 24 Apr 92 21:00:56 GMT Article-I.D.: unlv.1992Apr24.210056.29166 Posted: Fri Apr 24 17:00:56 1992 Sender: news@unlv.edu (News User) Reply-To: beleg@isri.unlv.edu Organization: UNLV Computer Science and Electrical Engineering Lines: 7 I know that this is kind of an odd request, but does anyone out there know where I might be able to find an Enochian Post Script font for the whole set of Enochian runes? Any help would be appreciated.. -Andy Jim Reeds From jbaez@math.mit.edu Mon May 11 01:23:42 1992 Date: Sun, 10 May 92 12:23:42 EDT From: jbaez@math.mit.edu Message-Id: <9205101623.AA04826@paley> To: voynich@rand.org Subject: Voynich talk at MIT Status: OR On Friday I gave a talk on the Voynich ms in the computer science dept. at MIT. It is the end of the term so I had advertised as a "fun" talk and stuck to that theme, emphasizing how many screwy attempts had been made to decode it and how, in the absence of constraints or discipline, one can read into a mysterious text whatever one desires. I gave out a handout about how to contact the (dormant) Voynich group and get books on the topic. About 50 people came including Richard Stallman (who thought it might be poetry), and there was a fairly lively discussion afterwards. "Kibo" brought in the Codex Seraphianus, which was fun to peruse. The most intriguing suggestion was to check whether the Voynich satisfied Zipf's law on distribution of words of various lengths. All in all, I'd say it was a success... we'll see if anyone joins the newsgroup or does any statistics as a result. (I also gave them instructions on how to get the transcription and encouraged people to mess with it.) John Baez From ACW@RIVERSIDE.SCRC.Symbolics.COM Mon May 11 11:03:00 1992 Date: Mon, 11 May 1992 11:03-0400 From: Allan C. Wechsler Subject: Voynich talk at MIT To: jbaez@BOURBAKI.MIT.EDU, voynich@rand.org In-Reply-To: <9205101623.AA04826@paley> Message-Id: <19920511150345.4.ACW@PALLANDO.SCRC.Symbolics.COM> Status: OR Date: Sun, 10 May 1992 12:23 EDT From: jbaez@math.mit.edu On Friday I gave a talk on the Voynich ms in the computer science dept. at MIT. A good one, too. Thanks, John! From dean@anubis.network.com Tue May 19 07:12:27 1992 Date: Mon, 18 May 92 17:12:27 CDT From: dean@anubis.network.com (Dean C. Gahlon) Message-Id: <9205182212.AA03213@anubis.network.com> To: voynich@rand.org Subject: Punctuation in the MS Status: OR While describing the MS to some friends over dinner this weekend, I had an idea that I can't recall having seen addressed in this forum: Could some of the symbols be punctuation? Particularly those occurring at the beginnings or ends of words? Dean From j.guy@trl.OZ.AU Tue May 19 22:30:35 1992 Date: Tue, 19 May 92 08:30:35 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9205182230.AA05894@medici.trl.OZ.AU> To: voynich@rand.org Subject: Re: Punctuation in the MS Status: OR I am rather partial to the idea that the "one-legged gallows" letters, which occur so often paragraph-initially are in fact not letters at all but an embellishment corresponding to our use of capitals. When they occur in the body of paragraphs they would serve to bring out a word, as we do by underlining, or using italics. From RJB@U.WASHINGTON.EDU Wed May 20 12:41:00 1992 Date: Tue, 19 May 92 20:41 PDT From: RJB@U.WASHINGTON.EDU Subject: punctuation in the MS Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: <52C2FC8AF5DF603618@MAX.U.WASHINGTON.EDU> X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR Perhaps one of the difficulties with looking for punctuation is that each sign identified as anything other than a letter reduces the number of signs that are letters -- and there are few enough signs,,, --rjb From reeds@gauss.att.com Sat May 23 09:09:51 1992 Message-Id: <9205230010.AA21353@rand.org> From: reeds@gauss.att.com Date: Fri, 22 May 92 20:09:51 EDT To: voynich@rand.org Subject: Voynich MS copy from Yale Status: OR Today my copy of the VMS arrived. The pages measure about 37 by 28 cm (wide by high), the typical Voynich page about 155 by 210 mm! Jim Reeds From ACW@RIVERSIDE.SCRC.Symbolics.COM Fri May 29 16:20:00 1992 Date: Fri, 29 May 1992 16:20-0400 From: Allan C. Wechsler Subject: Voynich MS copy from Yale To: reeds@gauss.att.com, voynich@rand.org In-Reply-To: <9205230010.AA21353@rand.org> Message-Id: <19920529202025.3.ACW@PALLANDO.SCRC.Symbolics.COM> Status: OR Date: Fri, 22 May 1992 20:09 EDT From: reeds@gauss.att.com Today my copy of the VMS arrived. The pages measure about 37 by 28 cm (wide by high), the typical Voynich page about 155 by 210 mm! Jim Reeds What is the quality of the copy? Is it better than the nth-generation xerography that I've been looking at? From jim@rand.org Sat May 30 05:42:22 1992 Message-Id: <9205292042.AA24991@mycroft.rand.org> To: "Allan C. Wechsler" Cc: voynich@rand.org Subject: Re: Voynich MS copy from Yale In-Reply-To: Your message of Fri, 29 May 92 16:20:00 -0400. <19920529202025.3.ACW@PALLANDO.SCRC.Symbolics.COM> From: Jim Gillogly Reply-To: jim@rand.org Date: Fri, 29 May 92 13:42:22 PDT Sender: jim%mycroft@rand.org Status: OR Mine just came also. I find the quality disappointing -- not quite as good as the copies I've been making from my copy of the BL microfilm. Sigh. Jim From RJB@U.WASHINGTON.EDU Sat May 30 06:32:00 1992 Date: Fri, 29 May 92 14:32 PDT From: RJB@U.WASHINGTON.EDU Subject: Voynich from Yale Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: <4B1AE777747F624481@MAX.U.WASHINGTON.EDU> X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I too have just received my copy of the VMs. It is disappointing-- but it's better than nothing. Is the transcription project still going on? --rjb From reeds@gauss.att.com Sat May 30 09:28:43 1992 Message-Id: <9205300029.AA12985@rand.org> From: reeds@gauss.att.com Date: Fri, 29 May 92 20:28:43 EDT To: voynich@rand.org Subject: Voynich xerox copy quality Status: OR Allen Wechsler asks about my recent Yale xerox: > What is the quality of the copy? Is it better than the nth-generation > xerography that I've been looking at? Not as good as it should be. Better than the pictures in Brumbaugh's book, about as good as the pictures in Bennett. Some pages come out better than others. Probably if the contrast and magnification were adjusted separately for each page the overall results would be better. Worth $35, I think. Jim Reeds From evy@well.sf.ca.us Sat May 30 23:56:45 1992 Date: Sat, 30 May 92 07:56:45 -0700 From: Evelyn Pine Message-Id: <9205301456.AA29684@well.well.sf.ca.us> To: RJB@U.WASHINGTON.EDU, voynich@rand.org Subject: Re: Voynich from Yale Status: OR Clearly, I missed the discussion where it was described how you ould get a copy of the VMs. Can I still get one? If so, how? And how much does it cost? From reeds@gauss.att.com Sat Jul 04 05:08:21 1992 Message-Id: <9207032017.AA29684@rand.org> From: reeds@gauss.att.com Date: Fri, 3 Jul 92 16:08:21 EDT To: voynich@rand.org Subject: Voynich lives? Status: OR Anybody there? I had (&continue to have) summer slump, and have only a little to report: 1. The Beinecke had lost my order (but not my check) for a MS 408 film. They promise prompt action. We shall see. 2. Talked with Mary D'Imperio on the phone. She will not be able to do as much transcribing as she had hoped. She agrees that the last (recipe) section of the MS should receive the sort of statistical analysis that the early parts have received. She urges me to apply a fancy clustering program to the transcription as it stands so far. Mary and I have been trying to coordinate a visit to Lexington, VA, to check out the Friedman collection, but have not so far agreed on a time. Maybe in the fall. 3. I will give a Voynich seminar at Bell Labs in the indefinite future. 4. My current action list: perfect my font (add non-Currier glyphs, mostly) finish the editing task Jim G. subcontracted to me transcribe cluster analyze, according to D'Imperio 5. I still have not recieved payment for Petersen from one person, maybe this note will shame him into action. 6. In reference to Voynich font stuff: are there any PostScript adepts out there I can share the work with? 7. Talked with my occult bookseller. Phanes Press will not be publishing Trithemius's Steganographia this year. Pity, I had been looking forward to it. Jim Reeds From reeds@gauss.att.com Mon Jul 06 10:47:45 1992 Message-Id: <9207060159.AA24037@rand.org> From: reeds@gauss.att.com Date: Sun, 5 Jul 92 21:47:45 EDT To: voynich@rand.org Subject: Voynich value Status: OR Jacques asks: > Question to all: is it really worth getting the Xerox of the Beidecke library > copy? I think so. The Beinecke xerox gives a view of the appearance & organization of the whole book, and much of the writing is legibile. (The "recipe" section, from f.103 to the end, seems especially clear.) I would say that it COMPLE- MENTS Petersen: I find Petersen gives a confused notion of the whole layout of the MS; the Beinecke book gives a clear notion. Petersen is legibile where the Beinecke book is not, (except in the recipe section, where its just the other way around!). Jim Reeds From j.guy@trl.OZ.AU Tue Jul 07 00:30:32 1992 Date: Mon, 6 Jul 92 10:30:32 EST From: j.guy@trl.OZ.AU (Jacques Guy) Message-Id: <9207060030.AA06937@medici.trl.OZ.AU> To: voynich@rand.org Subject: Re: Voynich lives? Status: OR I still do. Have been trying to work out the "calendar" and some astronomical folios, some of which look like a lunar calendar. The number 17 keeps recurring -- but it's probably an effect of my imagination. Got into a very deep mess trying to extend my Monkey program (the concordance part of it) beyond the 64K-array limit, nearly overwriting my hard disk in the process. Fingers badly burnt, I am waiting for a release of Pascal that allows arrays as big as your extended memory. Question to all: is it really worth getting the Xerox of the Beidecke library copy? From Karl.Kluge@G.GP.CS.CMU.EDU Tue Jul 07 16:08:00 1992 Date: Tue, 7 Jul 1992 16:08-EDT From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Label transcriptions? Message-Id: <710539682/kck@G.GP.CS.CMU.EDU> Status: OR I'm currently in mega thesis/job hunting crunch, and haven't been able to do any more work after my initial label correlation experiments. To recap, I transcribed the labels from 6 zodiac and 6 pharmaceutical folios, and greped for matches with each other and in the D'Imperio herbal transcription. Unfortunately, I never had the chance to transcribe nymph names from the "babes in tubs" folios -- does anyone have transcriptions of labels from these folios that I could use to run the same test on? Karl From RJB@U.WASHINGTON.EDU Wed Jul 08 01:17:00 1992 Date: Tue, 7 Jul 92 09:17 PDT From: RJB@U.WASHINGTON.EDU Subject: Getting the paper copy Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: <2CA1880FDF5F614698@MAX.U.WASHINGTON.EDU> X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" X-Vms-Cc: RJB Status: OR After some three months or so of suspense, I recently got my xerographic copy of the Beinecke's Voynich film. I thin it was worth it: I really had no other source for the complete Voynich corpus. It may be that a film copy of the film would be clearer in some ways -- I don't know. But the paper copy has certain advantages (of which the binding is not one, I should add: mine has several loose pages, and several more loosening). I took the paper copy along on a recent long car trip -- driving, unfortunately, interferes with one's ability to examine texts, so I didn't actually have much chance to work on it. There was one pleasant evening, though, spent with a couple of friends who do high-end medical imaging for a living, which repaid the effort. They were not only intrigued by the story of the ms -- they were excited, especially by the barrel-nymph section. They produced a "reading" of the diagrams in terms of female reproductive anatomy (with some elements of the intestinal and urinary tracts) that was at least very plausible. In addition to a large number of normal structures and situations, they "identified" an ectopic pregnancy. The obvious next speculative step was to assume that the whole manuscript in fact deals with gynecological and obstetric themes -- in which case the "herbal" might be exam ined for references to herbs used in the obstetrics of the period, and the lunar/solar/zodiacal charts might be examined for references to (e.g.) fertility cycles. The presence of all the nymphs *inside* the various organs (including things that look surprisingly like sperm and eggs) might be construed in terms of notions like the Paracelsan idea of the "archaeus," the animating "spirit" of an organ, system, or substance. The immediate problem with all this, of course, is the obvious potential for anachronism. I haven't yet had a chance to get back to histories of anatomy and surgery (let alone optics) to see how plausible it would be to suppose that Falloppian tubes could have bee depicted in a manuscript of the period. Fallopius (1523-1562) and Paracelsus (1493-1541) certainly were around at about the right time -- could the Voynich have been a cutting edge research report, put together by some 16th century equivalent of Elsevier, and so pricey that only one copy was ever purchased? But enough of this. I think I'll spend some time counting the calendrical/astronomical arrays to see if I can come up with anything --rjb From rcarter@isis.cs.du.edu Mon Jul 13 12:17:04 1992 Date: Sun, 12 Jul 92 21:17:04 MDT From: rcarter@isis.cs.du.edu (Ron Carter) Message-Id: <9207130317.AA04418@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: voynich@rand.org Subject: New address; misc. mumblings... Status: OR My `new' address is: rcarter@nyx.cs.du.edu `New' because it always was this, but it was aliased and there is now a new machine... You get it... I am still working on the VM; seems like busy time hit all of at the same time... I am working some theories on 49r (the `worm' plant); are any of the crypto mags that might be open to a short `paper' on 49r and other ms? Or maybe some other type of journal? I am still working on my transcription assignment; or at least need to get back to it... And still playing with the possibilty that the VM is transcribed from something Aztec... Regards, Ron | rcarter@nyx.cs.du.edu From j.guy@trl.oz.au Tue Jul 14 03:36:13 1992 Date: Mon, 13 Jul 92 13:36:13 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207130336.AA15576@medici.trl.OZ.AU> To: voynich@rand.org Subject: For those who think it is Aztec Status: OR ... filched from soc.culture.mexican: -------------------------------------------------------------------- Announcing a new discussion list: NAHUAT-L@FAUVAX.BITNET NAHUAT-L@ACC.FAU.EDU A discussion list which deals with Aztec studies in general and Nahuatl (the Aztec language) in particular. NAHUAT-L is an unmoderated discussion list which will focus on Aztec studies in general and the Aztec language, Nahuatl, in particular. Scholars interested in beginning projects will find the list useful in determining if others are already working in a particular field. The list may also be used to answer questions about Nahuatl translations, historical details, and all aspects of Aztec life and culture. Anthropologists, archeologists, linguists, historians, and all interested in the Aztecs, are welcome to participate. .... To subscribe to NAHUAT-L: Internet users, send email to NAHUAT-REQUEST@ACC.FAU.EDU Bitnet users, send email to NAHUAT-REQUEST@FAUVAX The sole content of the message must be: SUBSCRIBE NAHUAT-L {first name} {surname} To post a message to the list members, address it to: NAHUAT-L@FAUVAX or NAHUAT-L@ACC.FAU.EDU -------------------------------------------------------------- You bet I subscribed, and on the double, too. Someone just the other day asked about star (and planet) names. Someone answered (in Spanish), and I learned the words: planet: tzitzimime, or tzontemoc star: citlalin comet: citlalin popoca (literally "smoking star". Easy to remember: think of the name of the god Tezcatlipoca "Smoking Mirror") Venus: Tlahuilazcalpantecuhtli and two names of constellations: citlalxonecuilli (in which you've recognized the root "citla", star) xochilhuitl (in which I recognize "xochitl", flower, also the name of a month) Unfortunately, the author, Ricardo Salvador, does not give their Spanish names. From rcarter@nyx.cs.du.edu Sun Jul 19 11:50:57 1992 Date: Sat, 18 Jul 92 20:50:57 MDT From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9207190250.AA27320@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: voynich@rand.org Subject: Re: Newsgroups: In-Reply-To: Organization: Math/CS, University of Denver Cc: Status: OR Well, Nyx, having gotten a brain transplant, is suffering from an identity crisis, hence mail to me is bouncing all over the place... Patience, I will soon be back... Regards, Ron | rcarter@nyx.cs.du.edu -- Ron Carter | Denver | Center for the Study of Creative Intelligence Director | Colorado | "A ... mind stretched by a new idea can never rcarter@nyx.cs.du.edu | go back to its original dimensions." -Holmes ^----------Mail may bounce for the time being... Patience... From j.guy@trl.oz.au Tue Jul 21 01:46:58 1992 Date: Mon, 20 Jul 92 11:46:58 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207200146.AA22987@medici.trl.OZ.AU> To: voynich@rand.org Subject: 9 before "gallows" Status: OR Skimming through Petersen, I noticed how frequently "gallows" were preceded by <9>, and I remembered how I had argued a fairly long time ago now that and <9> were different forms of the same letter. If that hypothesis is correct, I went on thinking, should not occur before a gallows, except as a lapsus calami. Taking D'Imperio's transcription translated into Frogguy Transliteration, I counted the occurrences of and <9> before gallows. 13 7 Currier <9> 349 332 Currier <9> Currier Hypothesis confirmed: when is followed by gallows it is written <9>. Next hypothesis: if a gallows makes a preceding into <9>, the four letters with intruding gallows (Currier ) also do. And since the leftmost stroke of those letters is c-like, Currier's and (my and or and , depending on which character set you use) also should require <9> instead of . Results of the count: 19 Currier <9> 153 Currier <9> Currier Hypothesis confirmed again, I would say. It seems I've come full circle again: I've just dug out my Cryptologia article of July 91, on the contents of folios 79v and 80r, where I had counted that half the words ended with <9> and hypothesized that <9> was an unstressed, indistinct vowel; and that Voynichese words ending in a vowel had their last syllable unstressed. I feel that I must stop speculating right now, before I start running around in circles again. From reeds@gauss.att.com Tue Jul 21 23:46:22 1992 Message-Id: <199207211447.AA00763@rand.org> From: reeds@gauss.att.com Date: Tue, 21 Jul 92 10:46:22 EDT To: voynich@rand.org Subject: Voynich fonts Status: OR Font Progress. 1. Kibo and I are cooking. We have figured out how to tell his font editing program about my letter shapes. He can then turn my "ballpoint" stroke description into an elegant calligraphic outline & fill description, and then generate a multitude of versions, suitable for Macs, for Nexts, etc, etc, for bitmapped screens as well as for laser printers. Generation of AFM files should be no problem. 2. Starting from charts by D'Imperio of "rare" Voynich characters, I have added in about 100 new wierdo (non Currier) characters. I have given them names like X1, X65, and so so, and have assigned byte codes to them so they are accessible to word processors. Thus, X4, which looks like a Currier Q with constituent right-hand C replaced by a 9, is hex A3, which (I think) is usually used for the British currency (pound) sign, and so on. I am preparing a chart of all my letter shapes, their X-names, and their assignments to byte codes. I will mail out copies of this to anyone who wants, or will send postscript source, suitable for sending to a laser printer. 3. The byte code assignment is arbitrary, and conceals a problem. Computers and word processors differ in the way they let a user type characters outside of the usual 96. You might have to type "Alt control meta bucky 3". What's a convenient byte encoding on one computer is not convenient on another. Presumably the 36 basic Currier letters are mapped "naturally", but it is not clear what to do with the others. I assume that the X-names will remain constant, but each version of the fonts Kibo generates will have to be accompanied by a code sheet showing the encoding used. 4. There is also the problem of portably representing these characters in transcriptions. If a machine specific encoding is used in files produced by WYSIWYG editors, some form of translation will be needed when going to another computer. A portable format would presumably involve multi byte sequences. (On my computer I get to type the 7 bytes \C'X48' to make the "picnic table" symbol, for instance.) Which means a change to the "transcription format". Should we be using UNICODE systematically? 5. A brief (20 minute) glance through Petersen showed use of maybe 1 or 2 rare symbols per page, and yielded about 10 more symbols. I would not be suprised if in the end we had more than 200 Voynich symbols, but would be if there were as many as 1000. The Voynich transcription notation problem is thus similar to (but milder than) the Japanese problem: a mass of Kana (Currier) with a sprinkling of Kanji (wierdos). I will continue to collect these weirdos, and assign X-numbers to them, seriatim. Starting point: to scan through Petersen looking for "sic" marks! 6. I appeal for help in spotting wierdos: anyone who has copies of VMS pages or who has Petersen can join the hunt: see how many different non Currier symbols you can send in to me. I will assign X numbers and design shapes, & add them to the font. It would be a very big help if you note down WHERE you found them (folio & line numbers, ideally) so I can draw them accurately. 7. Here is an analytical question about wierdos, which I will not work on until my font work is done: where in the MS do the weirdos appear? Is their distribution in A and B pages different? Are they more frequent at the begining of a scribe's output, or at the end? Etc, etc. (If you go wierdo hunting, location notes will help here!) Jim Reeds From reeds@gauss.att.com Wed Jul 22 12:56:27 1992 Message-Id: <199207220356.AA21396@rand.org> From: reeds@gauss.att.com Date: Tue, 21 Jul 92 23:56:27 EDT To: voynich@rand.org Subject: Voynich fonts: analytic vs. synthetic Status: OR vmail voynich < Now, if we are confronted with even only 100 "letters", how are we > going to remember their codes, short of a point-and-click WYSIWYG > graphics editor, just like some Chinese word-processors I've seen > around? I had assumed you have a chart handy, as part of the other Voynich clutter on your desk. Once or twice a page you consult the chart, and say H'm, X84, and type that in. I'll send him a copy of my latest. > Most of those letters are easy to break down into their component > strokes. For instance, another weirdo I have seen is an the > rightmost part of which is not c-like, but similar to Currier <2>... > ... and voila! you have the weirdo. Sure: the weirdos (and the regular letters, too) are definitely built up out of a small stock of "radicals". This is the basis of the Bennett notation, and its perfected form, Frogguy. And its what makes my job of designing a font easy: I just call little postscript subroutines to draw the constituent letter parts with the correct relative positioning. Should this radicalism be expressed in the transcription notation? Well, if there is an easy way to go back from the transcription to the visual shape, maybe yes. I tried to make a Frogguy font, so that {ct}=, {c't}=, {cqpt}=, and so on. But to make the typeset versions actually come out looking right, {'},{q} and {p} must have zero width (or a very aggresive system of letter-pair specificic "kerning" -- "graphical sandhi", if you will -- must be used). The more I tried to make the constituent radicals into characters on their own, the less well the radicals fit together in relative position. If {qp} has zero width to save {cqpt}=, then a combo like {oqpox}= will look weird, with the two {o} dudes too close. So I then said, well, let's just make an inventory of all compound symbols & give them unique names so we can mention them & understand what we are talking about, and give them postscript shape definitions so we can set them. Call this the "synthetic" approach, instead of the "radical analytic" approach Jacques favors. The letters look right, but you have to use a code chart. Jacques also mentions that an analytic > system, at any rate, has been used for Chinese: instead of > calling up a character by pronunciation or by radical, you "type" it > in, just as you would write it. You only have to remember some > fundamental 15 strokes and where they are on the keyboard. For > instance, to get the character for Zhong ... > you would type,... the order in which you would write "zhong" > with a brush... > It's up to the software then to figure out which character(s) is (are) > made up of those strokes in that order. But this works because there is special knowledge of the writing system in both the user and the computer. If I, who know nothing of Chinese, were to use such a system, I would probably type the strokes in the wrong order. Info above & beyond the shape of the characters is being used. EOF Jim Reeds From reeds@gauss.att.com Thu Jul 23 00:19:01 1992 Message-Id: <199207221519.AA00896@rand.org> From: reeds@gauss.att.com Date: Wed, 22 Jul 92 11:19:01 EDT To: voynich@rand.org Subject: Voynich weird character query Status: OR More on weirdo V symbols: Can anyone cite a locus in the MS for either of the A-like compounds in D'Imperio's Fig. 18, listed in the 2nd and 3d lines of the section of the table devoted to initial symbol A? Jim Reeds From j.guy@trl.oz.au Thu Jul 23 00:32:16 1992 Date: Wed, 22 Jul 92 10:32:16 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207220032.AA25404@medici.trl.OZ.AU> To: voynich@rand.org Subject: Voynich fonts: rare "letters" Status: OR I had also noted, from my copy of Petersen, many letters not in Currier's list, usually circled by Petersen with a marginal "sic". One of them was a Currier with intruding "gallows" the first stroke of which was i-like instead of c-like. I saw it reappearing time and again, and made notes (left at home). That letter, if memory serves me, always occurred following . Now, if we are confronted with even only 100 "letters", how are we going to remember their codes, short of a point-and-click WYSIWYG graphics editor, just like some Chinese word-processors I've seen around? Most of those letters are easy to break down into their component strokes. For instance, another weirdo I have seen is an the rightmost part of which is not c-like, but similar to Currier <2>; its left part is identical with that of . Ditto for the i-like -and-intruding- gallows I first mentioned: just replace the leftmost like part with an , linked through to the righmost by a horizontal line cutting through the legs of the gallows, and voila! you have the weirdo. I have seen the very same problem in the transliteration of the Easter Island glyphs. There probably are thousands of them, but, for a few which are obviously unanalyzable (e.g. a pictograph of a squid, of a shrimp), I am sure that they are made up of very, very few elements, perhaps as few as 50, linked in a very few different ways under strict combination rules (I've counted four so far, keeping on the generous side). I would favour a system by which weirdoes are transliterated by their component strokes, however many it takes. Such a system, at any rate, has been used for Chinese: instead of calling up a character by pronunciation or by radical, you "type" it in, just as you would write it. You only have to remember some fundamental 15 strokes and where they are on the keyboard. For instance, to get the character for Zhong ("Middle", the first character of Zhongguo "China"), you would type, in this order, the keys corresponding to the strokes shu, heng-zhe, heng, and shu (that is the order in which you would write "zhong" with a brush). It's up to the software then to figure out which character(s) is (are) made up of those strokes in that order. If there are several (there seldom are) its brings them to the screen, for you to select the one you meant. Meanwhile, yes, let's make a list of those weirdoes. I had started, I am going to take it up again. From reeds@gauss.att.com Thu Jul 23 09:41:43 1992 Message-Id: <199207230041.AA15777@rand.org> From: reeds@gauss.att.com Date: Wed, 22 Jul 92 20:41:43 EDT To: voynich@rand.org Subject: Voynich font package on rand.org Status: OR Jim G. has just put up a new voyfont.ps.sh.Z, (on rand,org) a bundle consisting of the latest version of my postscript font, two style sheets (one actually takes 6 pages) ready to print on a postscript printer, and a ReadMe with hints about how you might use the font. This is provisional work in progress, & will presently be replaced by Kibo's (James Parry's) more professional effort. His font will be more truely portable to all kinds of computers, as well as more legibile & elegant. The main point of interest in the current voyfont.ps.sh.Z is its list of over 100 "weirdo" symbols named, classified, & drawn. I have tried to indicate on the style sheet where the weirdo's occur in the VMS, and if you can fill any of the gaps, please let me know. (Most of the shapes come from D'Imperio's diagrams, which are not always 100% clear to me, so suggestions about the artwork are in order, too.) Enjoy! Jim Reeds From kibo@world.std.com Thu Jul 23 14:33:29 1992 Date: Thu, 23 Jul 92 01:33:29 -0400 From: kibo@world.std.com (James "Kibo" Parry) Message-Id: <9207230533.AA29109@world.std.com> To: voynich@rand.org Subject: Re: fonts stroke order Status: OR > Try and write Currier's > (my <2>). There are several ways you can go about it. > > 1. The i-like part first, then the "plume" from top to bottom > 2. The i-like part first, then the "plume" from bottom to top > 3. Plume first, down, then "i". > 4. Plume first, up, then "i". > > In all four cases, you ought to end up with something very like > the Voynich letter you are trying to ape. Which is the right one? > I honestly do not know, but suppose the i-like part was written > first (i.e. case 1 or 2). To decide which of (1) and (2) is right, > we have to look, very carefully, at how the "i" part and the > "plume" connect. If they occasionally misconnect, then it is > (1). If they hardly ever misconnect, then it is (2). And if we > can't figure it out, we look at other "plumed" letters. If we There's another thing to check--the ends of the strokes generally have little dismount marks veering to one side (our italic typefaces today retain the curly tails on the lowercase letters.) In "i" letters, at the bottom, there is a little mark going up and right--this is the pen starting to go to the top of the next character or stroke just as it was being lifted. In the plumes, you frequently see a little hook at the top. This to me is a very strong indication that the plume was written upwards. As far as whether the two strokes "connect" cleanly, we must remember not to assume that they are _supposed_ to join at the corner. Very often, the plume seems to join the "i" just below the top of the "i". This may be deliberate--joining a stroke near the end of another stroke is a means of increasing legibility, as there can be no doubt that there are two strokes instead of one bent one. Our alphabet has this feature in the letter r, for instance. In Voynichian, it would help to distinguish "2" and "r" if they are not alternative forms of one symbol. Where (relative to the start/end of a line or a word) can "2" appear? It seems to like to hang out at the very beginning of lines (as well as in one-letter "2" words) so perhaps it might just be a 'swash' form of "r". Other random ideas: When a picture occurs in the middle of a page, are the words that run into the picture broken by it, or does it only have complete words on the sides? The one Xerox I have of such a case (from 76v, top) doesn't seem to have any broken words (from a superficial observation that the words 'look' well-formed), but they aren't squished either, which I suppose indicated that the scribe was smart enough to plan the layout well. Has anyone made a list of which letters (if any) can be doubled in Voynichian? -- K. From j.guy@trl.oz.au Thu Jul 23 21:56:26 1992 Date: Thu, 23 Jul 92 07:56:26 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207222156.AA26469@medici.trl.OZ.AU> To: voynich@rand.org Subject: fonts stroke order Status: OR Jim Reeds answered to my ravings about how Chinese could be typed in stroke by stroke "yes, but we do not know the Chinese stroke order, let alone the Voynichese one! So?" But we do, we do! Currier's paper (in pub/jim at rand.org) tells us a lot about how he thought the Voynich letters were made. And what is left unsaid I am sure we can find out. Just look closely at the ... ah, what are they called in English? "pleins et de'lie's" are the French words. When you are writing with a quill, or a steel nib, or even a fountain pen, you notice how a stroke written from top to bottom is thicker than one written in any other direction. The thicker stroke is "un plein" (lit: a full one), the thinner stroke "un de'lie'" (lit: a lithe one). Look now at any of the "gallows", say my . The vertical line that is part of my and

is thick and heavy; so it was written from top to bottom. The is always linked to a

by a thin horizontal line. In my eyes, that must mean that the vertical line of the was written first, then whatever letters if any came before the following

, then the horizontal linking line, bending back up to the top of the

, then the vertical of the

. Again, let us look at Currier's (my ). The c-like part of it was drawn from top to bottom, like our modern handwritten "c"; then again from top to bottom the second stroke, starting form the top of that "c", and Voynich i-like, that is, slanted and without a dot. Just take an old-style steel nib and have a go at writing Voynich letters and you'll see. I think it reasonable to assume that the authors of the Voynich used a quill. So, for an exercise, try and write a few Voynich letters with a nib, the closest thing to a quill (unless, lucky you, you have a willingly plucked goose at hand). Try and write Currier's (my <2>). There are several ways you can go about it. 1. The i-like part first, then the "plume" from top to bottom (i.e. from above the top of the "i" down to the top of the "i". 2. The i-like part first, then the "plume" from bottom to top (i.e. from the top of the "i" up to above the top of the "i". 3. Plume first, down, then "i". 4. Plume first, up, then "i". In all four cases, you ought to end up with something very like the Voynich letter you are trying to ape. Which is the right one? I honestly do not know, but suppose the i-like part was written first (i.e. case 1 or 2). To decide which of (1) and (2) is right, we have to look, very carefully, at how the "i" part and the "plume" connect. If they occasionally misconnect, then it is (1). If they hardly ever misconnect, then it is (2). And if we can't figure it out, we look at other "plumed" letters. If we still can't figure it out, we adopt an arbitrary rule: "body first, plume second" or "plume first, body second". Now, if on the contrary the plume was written first, from top to bottom, the i-like part would follow immediately without having to lift the quill at all, and we would expect the two to connect exactly every time. Why! And if, much later, we decide that it was "plume first" rather than "plume second", it does not matter as long as the "plume" has been represented by its own special symbol: we just just swap over every occurrence of a "plume" and a "body". Well, one observation of mine: the plume in Currier's (my apostrophe in or ) does not seem to connect very well. The other plumes (e.g. Currier's and <2>) appear to connect very well: they must have been written from bottom to top. The plume also looks very different from the and <2> plumes: it is narrow and often heavy, where the others sweep in an ample, most often light, arc. From reeds@gauss.att.com Fri Jul 24 01:35:09 1992 Message-Id: <199207231635.AA04491@rand.org> From: reeds@gauss.att.com Date: Thu, 23 Jul 92 12:35:09 EDT To: voynich@rand.org Subject: Voynich writing Status: OR Kibo asks > When a picture occurs in the middle of a page, are the words that run > into the picture broken by it, or does it only have complete words on > the sides? The one Xerox I have of such a case (from 76v, top) doesn't > seem to have any broken words (from a superficial observation that the > words 'look' well-formed), but they aren't squished either, which I > suppose indicated that the scribe was smart enough to plan the layout > well. I think the twigs and stems do not run through words. In some pages the lines on each side of such a break seem nicely aligned, and one could suppose the writing (sentence, if you will) spans the break. On other pages the alignment is not very careful, and the pen refills seem inconsistent with continuous writing across the break. > Has anyone made a list of which letters (if any) can be doubled in > Voynichian? Without looking at frequency tables: only and with any great frequency. In Currier notation, or is much more common than , but in Frogguy, outnumbers . Very often the repeated Cs are slightly joined at the bottoms, so looks like our u and looks like a lower case omega. Jim Reeds From reeds@gauss.att.com Fri Jul 24 02:29:51 1992 Message-Id: <199207231730.AA06284@rand.org> From: reeds@gauss.att.com Date: Thu, 23 Jul 92 13:29:51 EDT To: voynich@rand.org Subject: Voynich transcription format, revisited Status: OR I think (after off line persuasion by Kibo) that it is time to amend the transcription rules again. Much as I dislike multi-character names for characters, I think we should allow for them in transcriptions. To that end, I suggest we think up a different notation for line numbers, freeing up the use of < and > for long glyph names. Currently a line of transcription looks like OPOE*.(S|CC)OQCC8[9A] {word 1 final is 'picnic table'} I suggest it be written something like this @ OPOE.(S|CC)OQCC8[9A] This has the advantage that current transcriptions are easily converted to the new form, the new form is still easy to parse, and it does not commit us to a fixed format for glyph names inside the < > brackets. I prefer use of <> to [],(),or {} because the latter seem to have too important a role already. If someone has a special bitmap font, or a special editor (with, say, a point & click palette for weirdos) which uses machine-specific non ascii byte combinations for some characters, well, we can write little translating programs to convert in and out of 'standard' form. For instance, my font assigns byte code AF = 175 to 'picnic table', which I call X48. I would put byte value AF in a transcription file I was printing out, or perhaps encode it in troff notation \N'175' or \C'X48'. The form I would ship out, or discuss in email, however, would be or , with details of my word processor's idiosyncratic handling of that glyph hidden from the world. Jim Reeds From reeds@gauss.att.com Fri Jul 24 12:11:36 1992 Message-Id: <199207240318.AA21279@rand.org> From: reeds@gauss.att.com Date: Thu, 23 Jul 92 23:11:36 EDT To: voynich@rand.org Subject: Voynich weirdos Status: OR Some of the weirdos on my sheet were invented simply to ba able to set D'Imperio's figures. Thus, X105 and X106 simply because they appear as "second stroke" labels of columns in her Jan 1992 fig. 3. To typeset the chart you need X105, but I doubt very much that that shape appears by itself as a separate character in the VMS. The added-symbol column fig 18 in her book has a few more such examples. Sorry about that! Jim Reeds From reeds@gauss.att.com Fri Jul 24 13:08:53 1992 Message-Id: <199207240408.AA22070@rand.org> From: reeds@gauss.att.com Date: Fri, 24 Jul 92 00:08:53 EDT To: voynich@rand.org Subject: Voynich C,N,I alternation speculations Status: OR Starting with the original D'Imperio transcription, I converted some characters to 'c' and some others to 'i', and then counted letter pairs (for pairs of adjacent non space chars, viz, in the same word). letters mapped to c: QWXY9CSZ826 letters mapped to i: DINMEGHRJK The results, sorted by decreasing frequency: 15481 cc 4774 Ai 4375 Oi *** O like A on right 3612 cO *** O like A on left 2591 cA 2528 OF 2482 4O 2449 Fc 1496 Pc 1427 OP 1390 ic *** rule breakers 1313 Oc 1212 FA 690 cF 495 PA 455 iO 452 cP 362 Bc 359 FO 354 PO 330 iF 275 iA *** rule breakers 168 OB 164 ci *** a few more rule breakers 124 AT 102 Vc 89 cB 88 OA 87 BO 71 Ac 68 ii 54 OV ... >From which one sees that O is as much c-like on the left and I-like on the right as A is. Also notice that ic and ci does occur. In the B corpus, I-like letters seem to occur only at the ends of words. Typically a word starts out C-like and ends up I-like. Can this I-like, C-like, and neutral stuff be a cryptological not linguistic phenomenon? Maybe the author has a basic alphabet where each letter has both a C-form and an I-form. He writes out the text in basic letters, and then writes the Voynich MS, drifing in and out of the C and I forms, just to amuse us. If this were the case, we should treat Currier <2> and Currier as the same, etc, etc. Or the author could be putting all the info in the choice of C-form versus I-form: C-form could be 'dot', and I-form could be 'dash', and choice of 'base letter' is noise. (Say, only the C/I value of a letter following a gallows counted, or maybe that and plume-presence of letter following a gallows. That gives you a sequence of bits or of 'dibits', which is used in a Baconian biliteral or Trithemian triliteral cipher, say. Or if you figure each word starts C-like and ends I-like, maybe the only signficant thing is what happens at the transition, which will take the form cAi or cOi. The significant thing is the pair of ci letters. On rereading this all, it seems unlikely. Jim Reeds From j.guy@trl.oz.au Sat Jul 25 01:04:00 1992 Date: Fri, 24 Jul 92 11:04:00 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207240104.AA27983@medici.trl.OZ.AU> To: voynich@rand.org Subject: i-compound weirdoes Status: OR This is what I had started writing two weeks ago, before I received Jim Reeds's list of weirdoes: "Folio 89r, replete with pharmaceuticals. Last paragraph reads (I'm using Petersen): q;c'tox c'tco 4oaiqpt9... [note 1] (Currier) BZOE ZEO 4OAQC9... Now look at that Currier . Its first stroke is not c-like. It is in fact a Voynich "i" the top of which is linked to its c-like final stroke. Why so? It follows an , which demands that the following letter should be i-like. The same word is repeated underneath, on the next line. Now turn to folio 106r, two lines from the bottom. There are two words that show this eccentric i-like S-cum-gallows, and, again, they occur after an ... ailpt9 .... 2aiqpt qp9 (Currier) AX9 .....RAQ P9 [note 1] Normally, in my Lazy Frogguy (tm) Transcription System, Currier's is and his is , but here I used and to convey the shape of the Voynich letters" better. That weirdo is X31 in Jim's list (in the table entitled "i-compounds"), and he gives it as occuring on f89v1,1.2 Looking up Petersen indeed I find the second caption of the first recipe: a oqpo ilpt9 (Currier) OPO XC9 Petersen seems to have added an on top of the to mark a doubtful reading. Beginning the second recipe on f89r2 I see: q;o2aitox ... (Currier) BORASOE that or being in fact weirdo X10 I am starting to suspect that the "i-compound weirdoes" occur mostly after . Now hove you noticed how the "i-beginning" simple Currier letters also occur mostly after , viz: Currier 8IO99DE <2> More following soon. (Now, what could "8IO99DE" mean?) Frogguy. From j.guy@trl.oz.au Sat Jul 25 01:12:23 1992 Date: Fri, 24 Jul 92 11:12:23 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207240112.AA27998@medici.trl.OZ.AU> To: voynich@rand.org Subject: i-compounds: more... Status: OR ...and these are ramblings that slightly predate my discovery of two "i-compound weirdoes" in the pharmaceuticals. The existence of those not very uncommon letters make these ramblings somewhat less far-fetched. Or do they? "After an hour or two of peering at those familiar squiggles I always end up saying to myself "this is gibberish". Later, having put this patent piece of nonsense aside, something makes me revise my judgment. Every time. This time, I was thinking back on the insanity that is the Voynich alphabet. In particular, how a c-like letter is almost always followed by a letter that starts with a c-like stroke (e.g. the common endings , ), and how a letter ending in an i-like stroke is followed by one starting in a similar stroke (e.g. , ). In other words, how the Voynich alphabet seems to be split into three groups: c-like letters c, cc, ct, c't, 8, S, cg... Currier: C, CC, S, Z, 8, 2, 6 i-like letters iv, iiv, x, ix, iix, 2, ig, iig... Currier: N, M, E, G, H, R, J, K neutral letters o, the four "gallows"... Currier: O, P, B, F, V under this rule: c-like with c-like or neutral, i-like with i-like or neutral. Example, taken at random (folio 66r, caption above "dying man"): oqpctco 8aiiv ctqp9 9lpccSctccg Currier: OPSCO 8AM SP9 9FCC2SC6 Now who could have thought up such a rule, and have carried it out unerringly for 200 pages if it was all gibberish? And remember that "who" is at least two different persons -- or perhaps again a single writer with multiple personalities (a possibility we cannot discard). So I'm back to square one, thinking "it's not gibberish after all, but a real language". The rule "c-like etc." reminds me of Gaelic: broad with broad, slender with slender. But now we're about to topple back into insanity: how many different letters are we going to be left with? Have a look at and <2> (Currier's <2> and ); one is the letter with a flourish on top, the other the letter with the same flourish. And so on, and on, and on... And yet, it is not necessarily insanity. In early Latin, the sounds of [g] and [k] were spelt alike: C, K, or Q. The choice of the letter was governed by what followed, mostly C before a front vowel, K before A, and Q before a back vowel. Thus "ego" was spelt EQO, "pecunia" PEQVNIA (Carl Buck: "Comparative Grammar of Greek and Latin", p.74). Suddenly, the insane Voynich alphabet becomes believable again." From j.guy@trl.oz.au Sat Jul 25 02:12:29 1992 Date: Fri, 24 Jul 92 12:12:29 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207240212.AA28089@medici.trl.OZ.AU> To: voynich@rand.org Subject: plumed weirdoes and patterns in compounds Status: OR Finally, I had a look at the "plumed weirdoes", which you will find in Jim Reeds's fonts under: in "approximate Frogguy": o-compounds: X101, o' X64, o' X96, o't X97, o'9 X95, o' X38, o't X39 o'9 9-compounds: X14, 9' X32 9' a-compounds: X15 a' 4-compounds: X11, 4o' X35 9' Well, so far, I note that those sometime-plumed letters are all vowels according to Sukhotin's algorithm. Next, that that plume occurs regularly in Currier's , on top of its c-like first element. And is also a "Sukhotin vowel". Next I noted which elementary letters connect to another to the right with a horizontal stroke (as in Currier's and ). Well, again we have in "approximate Frogguy": X10, it X30, iqpt X31, ilpt X18 il;t X57 il; X58 ilp X59 iq; X96 o't X38 o't X97 o'9 X39 o'9 X40 oqp9 X42 ol;t X100 oll;t X41 oqpo X63 oqp <9> X33 9iptt X32 9 X91 aa X17 al;a <4> X11 4o' X94 4o X34 4l;a X37 4lpt X61 4qp X19 49 X75 4t X35 49' X36 4l;o X62 4lp That was the very same lot as takes plumes, plus the Infamous <4>. But the Infamous <4> takes a plume too: X78. Turn to f49r, where the Infamous <4> starts lines 14 and 16, complete with a plume that looks exactly like the one on the that begins line 18. Petersen has "sic'ed" both occurrences (commas for half-spaces in what follows): Line 14: 4'o c't,4o,lp9 c'to2 c'to2 oqpox 8aiiv 4O Z,4O,F9 ZOR ZOR OPOE 8AM Line 16: 4'o cto ctcc9 8ctc9 4oqpct,o,89 4O SO SCC9 8SC9 QOPS,O,89 Petersen has also sic'ed the last word of line 15, which is the very common <8AM> with a plumed (X15). Third and lastly, note what those letters may connect to: which is really very like a <9> the left leg of a "gallows" letter a blank (viz in the c-forms table: X8, X84, X74, X71, X56, X76; in the i-compounds table: X83; in the o-compounds table: X102, X95, X64) Look at the bar-compounds table: you see: X105 a blank connecting to a blank X103 a blank connecting to X109 a blank connecting to X73 a blank connecting to through gallows. So there is pattern to this madness: left right connectors connectors blank blank <9> <9> <4> and they may all connect through gallows: the connection to a right gallows, as in X70, X88, X89, etc is probably connection to a blank through gallows, just stopping short at the gallows left leg. From j.guy@trl.oz.au Sat Jul 25 05:12:13 1992 Date: Fri, 24 Jul 92 15:12:13 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207240512.AA28277@medici.trl.OZ.AU> To: voynich@rand.org Subject: Re: Voynich C,N,I alternation speculations Status: OR Nope. I think, but I'm a linguist, and to a linguist everything looks like a ... language, that that like-with-like phenomenon is not a cryptological artifact. If it were, the VMS would have been deciphered long ago. We are dealing here either with a now extinct or near-extinct language, or with a "philosophical tongue", an artificial language a la Bishop Wilkins. I have mentioned Irish Gaelic because it does, in its spelling, things reminiscent of Voynichese. I could have mentioned Javanese, in which a word tends to have either dental or cacuminal consonants but not a mixture of both, or Ordoss Mongolian in which every word tend to have all its vowels and consonants of one of two alternative sets. Turkish, and Finnish, and Hungarian, all distantly related to Mongolian, show the same phenomena, but applying to vowels only. There is something similar I vaguely remember in Basque, or at least some dialects of it, applying more restrictively to consonants only, not vowels. And for you classicists, think of aspirated vs unaspirated in Greek. Just trying to make a point that the C, I, Neutral alternation in Voynichese does not help one jot in figuring out what language group it might have belonged to. The mystery remains as deep as ever. From j.guy@trl.oz.au Tue Jul 28 01:17:19 1992 Date: Mon, 27 Jul 92 11:17:19 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207270117.AA00539@medici.trl.OZ.AU> To: voynich@rand.org Subject: Fruits of last week-end Status: OR It started with a hunt for weirdoes in Petersen, who obligingly and helpfully clicks his tongue in disapproval whenever he sees one ('tsk' in English, in Latin: 'sic'), then my aging brain rebelled: "What? I am groaning and creaking under the weight of the junk you have stored in my neurons, thousands of Chinese characters, hundreds of Easter Islands hieroglyphs, uncountable strings of strange sounds you call "words" gleaned from cannibal islands. Now demented squiggles off medieval sheep skins. I warn you: collapse is nigh, unless you lighten my load." And my bookshelves creaked and groaned in sympathy. "All right, all right, you gooey heap of grey muck..." Two days' struggle later, I said to my brain: - I have reduced those squiggles to 32, and one capitalization rule. - Too many. - You didn't let me finish. You already know 20 of those squiggles. So twelve, and one single rule is all you need to find room for. - All right, all right, but will it look nice on your VGA, when the corresponding fonts have been loaded? As nice as the earlier ones? - Nicer, and you don't have to apply the capitalization rule if you don't want to. It only makes the display much more Voynich-like. - So be it then (*sigh*), on one condition. - Which is? - Call them F3W. - Eff three double yew? - Frogguy Fonts For Weirdoes, and that includes *you*, Sirrah. Zipped and uuencoded, the lot weighs in at 280 lines or 17K, small enough to fit in a mailbox with plenty of room to spare. So here it comes under separate cover, Subject line: Frogguy Fonts For Weirdoes. It unzips to three files: f3w00.fnt fonts for loading with Harald Thunem's VGA/EGA font loader f3w01.fnt same, with the capitalization rule left out (so that Roman capital letters remain recognizable as such) f3w00.map ASCII, human-friendly bitmaps 8x16 for all those Voynich letters f3w.doc Comments, F3W equivalents for Currier and all of the weirdoes in Jim's lists, except a handful (5 or so) I have found other things over the week-end, but have not written them up yet. So you can breathe a sigh of relief: you won't hear from me for a few days... or perhaps hours. From j.guy@trl.oz.au Tue Jul 28 01:18:38 1992 Date: Mon, 27 Jul 92 11:18:38 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207270118.AA00542@medici.trl.OZ.AU> To: voynich@rand.org Subject: Frogguy Fonts For Weirdoes Status: OR begin 640 f3w.zip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o: voynich@rand.org Subject: Folio 1r: Secundus Status: OR I was about to set out to hunt for Jim's dingbats in Petersen, starting from the beginning, when I noticed this (warning: I cannot believe that we haven't already discussed it, but I can't remember any messages mentioning it. Apologies if it's been thrashed out. My brain is turning into a sieve where Jumblies live). The *BIG* dingbat at the end of line one is "29", i.e. "secundus", with "us" written "9" as was common in medieval manuscripts. We all knew that. But that page is where "Jacobus de Tepenecz" etc. is written. Doesn't that mean that the VMS was acquired with the first 8 pages already missing? And that the quires were numbered when it was whole and hale? I think it does. Next, I find that the style of those numbers strikingly similar to that of the Voynich letters (I must have mentioned it time and again, I'm repeating myself here). I can't help thinking that the authors of the VMS did the numbering. If they did, then they knew some Latin. If they hadn't, they would have written just "89" for "octavus", not "8u9" as they did. Finally, those "dingbats". They are mostly rare letters or rare combinations of Voynich letters. What comes to your mind when you see in an English text a sequence of letters that is (ding)batty? A foreign word or name. Boaistuau (a French writer). Cliche, with the plume called acute on top of the "e". Fete, with another plume, called circumflex, on top of the first "e". Manana, with a wiggle above the first "n". On this topic, I think that dingbats X30 and X31 are perfectly good Voynich, not foreign bits. But dingbats X24, X14, and X15 strike me as alien. Or a scribe's booboo. Back to tracking dingbats now. From j.guy@trl.oz.au Thu Jul 30 23:58:12 1992 Date: Thu, 30 Jul 92 09:58:12 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9207292358.AA04321@medici.trl.OZ.AU> To: voynich@rand.org Subject: Some botanical terms in Aztec Status: OR This, for those who, like me sometimes, think the Voynich might have been written by Mexican converts circa 1550, I extracted from the Nahuatl list I have subscribed to. I am keeping the archaic spelling of the original: suchiamatlapalli = flower-wing (suchi = flower) tlapalli = color (petal?) suchizoatl = flower-leaf (petal?) tecomaiotl = like a gourd (calyx?) tomjoli, tomjolli = hairy (pistil?) achtli, achioti = seed or fruit (from the name of a particular species) iollotli = heart (ovary?) suchitl itzin = flower its base (receptacle?) mjavaiotl, mjmjaoatl (derived from miauatl, corn tassel) = stamens (in general) Those words are from a paragraph in the Florentine Codex: In jtech ca in suchitl, in cueponcaiotl, pochqujotl, suchiamatlapalli, suchizoatl, tecomaiotl, tomjoli, tomjolli, achtli, achiotl, iollotli, suchitl itzin, mjavaiotl, mjmjaoatl. Translation: Pertaining to the blossoming of the flower are the fattening, the petals, the calyx, the pistil, the pistils, the seed, the seed of the flower, the ovary, the receptacle, the stamen of the flower, the stamens. Source: Dorothy McMeekin "Representations on Pre-Columbian Spindle Whorls of the Floral and Fruit Structure of Economic Plants." Journal of Economic Botany [46(2):171-180] Pronunciation: s = sh (also spelt x) j = i y = i ch = ch z = s v = u (there is no "v" sound in Aztec) o, u, v between vowels = w (also spelt hu) c = k before a, o, u; s before e, i If I remember correctly, "o" and "u" are allophones of the same phoneme in Aztec, and indeed, you'll more often see "flower" written "xochitl" than "suchitl". The captions in the VMS seem too short to correspond to any of those lengthy Aztec words, even though the small number of different vowels and consonants in Aztecs agrees fairly well with the paucity of letters in the Voynich alphabet. Isn't it strange, though, how Aztec written in Roman letters looks vaguely like Voynich? From Karl.Kluge@G.GP.CS.CMU.EDU Sun Aug 02 23:06:00 1992 Date: Sun, 2 Aug 1992 23:06-EDT From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: A comment and a question/request Message-Id: <712811200/kck@G.GP.CS.CMU.EDU> Status: OR Jim Reeds writes: > Or the author could be putting all the info in the choice of C-form > versus I-form: C-form could be 'dot', and I-form could be 'dash', > and choice of 'base letter' is noise. (Say, only the C/I value of > a letter following a gallows counted, or maybe that and plume-presence > of letter following a gallows. That gives you a sequence of bits or > of 'dibits', which is used in a Baconian biliteral or Trithemian > triliteral cipher, say. I toyed with this hypothesis for a while as well. In particular, I thought it might explain the "language" A/B difference by appealing to the idea that different scribes would settle on different stereotyped groups of characters to represent common bit sequences, just as a code clerk will tend to settle on one or a few code groups for things like "stop". I suppose one could test the idea by asking what predictions it made about the entropy of the text. That was the comment, the question/request is as follows: has anyone tried to test the hypothesis that the labels and the text are not encrypted the same way? Brumbaugh reached that conclusion on the post-hoc basis that the technique he used to dechiper the labels didn't work on the text itself, but didn't propose an objective way to test the idea. I've appended the label transcriptions I had done earlier -- would someone with a handy character frequency counter provide comparative frequency counts for the labels vs. the D'Imperio transcription? If they differ significantly that might suggest some new hypotheses. Thanks, Karl **************************************************************** Note: Currier notation. "*" = illeg., "/" = space, "," = questionable space. The ideosyncratic numbering scheme for the labels can be ignored. <14601A> OPAR/AJ <14602A> OPAR/AE <14603A> OPAE/AR <14604A> OPAE/AJ <14605A> 8OEARAJ <14606A> OFARAJ <14607A> OPCO2AE <14608A> 2AEOE2 <14609A> OFAE/8AE <14610A> 9FOEAM <14611A> OPAEAE6 <14612A> 9FAR9 <14613A> OPAR <14614A> OP9 <14615A> OF9/O89 <14616A> OP9/AR <14617A> OFAEA <14618A> OPO89 <14619A> OPAE8 <14620A> OPAE8AR <14621A> OFA69 <14622A> OP92AJ <14623A> SXC9 <14624A> OPAE9 <14625A> OPAE/ARAR <14626A> OPAE89 <14627A> OFCOE9 <14628A> OF989 <14629A> OFCC2 <14701A> OFOE9 <14702A> OPAEAM <14703A> OPCO/AEOE2/ARAE9 <14704A> OPOOCC9/OPAE/OFCAEAR <14705A> OPCAR/ARA989 <14706A> OPAES9/PAR/AJ/69 <14707A> OBSC9/2AE <14708A> OPAFAIJAD <14709A> OFAEAE <14710A> OPAE9 <14711A> OAESC6 <14712A> OPSO8AE2 <14713A> OFOEZ9 <14714A> OPZZ89 <14715A> OPAE/9BZARAE <19901A> OPORSCP9 <19902A> ORAE <19902B> ORAE8 <19903A> OE8AR <19904A> OPOF9 <19905A> OPAE9 <19906A> OFAE89 <19906B> ORAJ <19907A> 8AR9 <19908A> OFOE <19908B> 2OROR9 <19909A> OP98A <19910A> OV92F98AE <19910B> OPOR/AJ <19911A> OVAE8O <20013A> OFAE98 <20013B> SCOQ9 <20014A> WCOR <20015A> OPAR/ARO89 <20016A> OPOFOE <20017A> OPORAJ <20017B> OPORA <20018A> SCO289 <20019A> OFAM <20020A> 8ARAJ6AE <20020B> OP98AR9 <20021A> OP8OR69 <20022A> 8ARAR6I <14801A> OPOES89 <14802A> OPOEOARAJ <14803A> OPC*OE <14804A> OPOES8 <14805A> OPAE6AR <14806A> OPCO2ARAR <14807A> OFE8AJ <14808A> OPCOAE89 <14809A> OPCOEAR <14810A> OFCOAE9 <14811A> OPAECF9 <14812A> OBAERAR <14813A> SCAR9 <14814A> OPCO/PC9/2AR9 <14815A> OPAEAE9 <14901A> OVA*VOJ <14902A> OPAEO89 <14903A> OPAEAM <14904A> OPAR/ZAR <14905A> ZOEZ69 <14906A> SAR/ORO7 <14907A> SVAE9 <14908A> OFOEAR <14909A> OPSO89 <14910A> AEW9 <14911A> OPAM <14912A> OFARAM <14913A> OPAR/AR/AE9 <14914A> OBAEAR/AJ/6AD <14915A> OBAEAR/AR <15001A> OVARAEAR <15002A> OPSOZ9 <15003A> OPS8AE <15004A> OFCC9/AR9 <15005A> OPAN9 <15006A> OZO8A89 <15007A> S8AU/8AN9 <15008A> OAM/AR/AR9 <15009A> OFAEAJ <15010A> 9PAEZ89 <15011A> SAR/AEIV <15012A> OPARAE89 <15013A> OPAM/OPAN <15014A> OPAECV/92/ANAJ <15015A> OSOEZARAJ <15101A> OPAE <15102A> 2AEAE <15103A> OFAJ <15104A> OPAEZ9 <15105A> OFAE/89 <15106A> SO2AR <15107A> OPAJ <15108A> ANAE9 <15109A> OFARSA7 <15110A> OPARAE69 <15111A> OFAEAR <15112A> OFAE <15113A> OFAE9 <15114A> OFAE <15115A> OFCC9/AR9 <15116A> OPCCAR9 <15117A> OFAT/69 <15118A> OFAT,SAJ <15119A> OFCAE <15120A> OPARCR <15121A> OFAE9 <15122A> ODAR9 <15124A> OF98 <15125A> OPOEAJ <15126A> OVS8A69 <15127A> OFEAT69 <15128A> OFARAJ <15129A> OFAT9 <15130A> OFCAEAR <20123A> OFSZ9 <20123B> 4FOE <20124A> OE8AJ <20125A> OPOE89 <20126A> ARARSO8AM <20127A> 9F98 <20127B> SOE/SC2 <20128A> OPORAN <20129A> OFAM/8AD <20130A> 9FOY9 <20130B> 2AE8AJ <20131A> 298ARAR9 <20132A> 98W089 <20233A> O8OR9 <20233B> 8OE9 <20234A> OBSO2AJ <20235A> 2AEOMZCOE <20236A> OBSCOR <20237A> OPOE8 <20238A> OFOE/ZOE/89 <20239A> OBSARAM <20240A> FORAN/9 <20241A> 2O8AR <20242A> SC92 <20243A> SCO89 <20244A> OVORAN <20245A> OFZ8SA2 <20246A> 9ORAN <20246B> APAFAE <20247A> OPAE29 <20248A> 9PARCJ <20249A> OPOEAROE <22315A> SO2AROZOE <22316A> 2OSORY9 <22317A> OPCAR <22318A> SOVAD9 <22319A> 2AR/SAR/8AM89 <22320A> O2ADO <22321A> SAE2AN <22322A> 2OIP9 <22323A> 292AJ <22324A> 8AFOQ <22326A> OFCCO2/OROE <22327A> ZOXC9 <22328A> OESCOJ <22329A> OFOE2 <22330A> OPCOE <22431A> POES8 <22432A> SOE2 <22433A> OBSOR <22434A> 2OE29 <22435A> 2OECC2O2 <22436A> 9FSOS89 <22437A> 9FS89 <22438A> 8S89 <22439A> 8AE29 <22440A> OFSCOR <22441A> 9PSOE <22442A> 89FSAE <22443A> SO2/QARAE From reeds@gauss.att.com Mon Aug 03 13:03:13 1992 Message-Id: <199208030403.AA03417@rand.org> From: reeds@gauss.att.com Date: Mon, 3 Aug 92 00:03:13 EDT To: voynich@rand.org Subject: Voynich and the Neo-Frogguy notation Status: OR I am studying Jacques's F3W.DOC after returning from a week away from email and the Voynich world. Here are a few brief notes prompted by his speculation about the PostScript-y origins of some of my X-series dingbats. For the most part my dingbat collection is based on charts of weirdos collected by D'Imperio, not on first-hand inspection of letter forms in photocopies of the VMS or of Petersen. Unscientific and unscholarly, I know. In particular, I might have misunderstood some of D'Imperio's shapes. X1 through X66 are taken from D'Imperio, "Some Ideas on the Construction of the Voynich Script", Jan. 1992, page 3. (Do you have a copy, Jacques? Drop me a line if you don't, and I'll pop one in the mail.) > X52 I have nothing to propose for this. I > suspect it is one of the components Jim > used in designing the Postscript fonts. This is the 8th symbol in column 4 of D's Fig 2. So presumably it appears by itself somewhere in the VMS. It is also the column label of column 6 (or 7, depending on how you count) of her Fig. 3. In Fig. 3 it is clearly the right half of an O symbol. > X73 QPt a Postscript component, I suppose I thought I saw one in Petersen, f89 v1, line 12. ("My" folio numbering system, not Petersen's.) But when I looked at the Yale xerox, I see that it is probably a . > X81 e't or St That is Currier's ! A Postscript glitch? I had always supposed that Currier's had a centered plume. X81 has the plume on the lefthand C (or E, in NeoFrogguy) and X26 has it on the right C. (A distinction without a difference? I sure HOPE so!) > X87 eqp Must be a Postscript component again In D's book, on page 96, is a table of compound and ligatured forms. X87 is meant to be the 4th glyph from the top in the 3d column. I see a gap between the e and the qp in D'Imperio's drawing but assumed it was a hando of hers. X88-X90 appear directly below, in the same column on page 96. > X104 ` One of the "nuts and bolts" strokes Yes. The very first Voynich symbol on page 96. > X105 - Ditto > X106 Cat got my tongue. A Postscript component? Yes. These are column heads for Fig 3 in D's 1992 paper. > X107 A Postscript component I suppose, looks like > my [,] (comma, which see in "nuts and bolts") > X108 ) Nuts and bolts again More column heads. X107 is the right half of a <9>. X108 is the right half of Currier's . ===== Now for a general comment on Jacques's NeoFrogguy: it is certainly a superb solution to the problem of transcribing compound or built up characters. But I am not so sure about how to display characters presented in the form of NF strings on a terminal or on a laser writer. Do the single NF atoms group nicely so that ligatures come out looking natural? And how does one teach PostScript about the capitalization rule? Jim Reeds From j.guy@trl.oz.au Tue Aug 04 05:06:21 1992 Date: Mon, 3 Aug 92 15:06:21 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9208030506.AA08196@medici.trl.OZ.AU> To: voynich@rand.org Subject: "Voynich and Neo-Frogguy" Status: OR More recent developments: I borrowed two Postscript manuals from our library and spent an hour or so wading through the lot. Easier than I thought, once I realized that Postscript was really Forth with a very much nicer terminology. I also downloaded Ghostscript for PCs from the oak archive. But tough, I'll have to reinstall MS-DOS 5 on my machine at home: Ghostscript for 386s does not like the DR DOS memory manager *at all*. Meanwhile, on my spanking new work 486 with 1.3Gig HD it works nicely. All this to bring me to answer (future infinitive, if there was one in English) to this question by Jim Reeds: And how does one teach PostScript about the capitalization rule? Humph. I would write a Postscript program that interprets Frogguy (or, for that matter, Currier) input transliteration since, if I understood the manuals and the examples that come bundled with Ghostscript for PCs, Postscript is in fact a fully fledged programming language. Wait a minute, I fear I have mis-anwered Jim's question. The capitalized letters are just the old Frogguy letters with a horizontal bar going through if they're gallows, or connecting to the right if they're any of the rest. The capitalization rule is only a mnemonic. So you would not teach Postscript about that rule. It's just another bunch of characters to define. But... I'm not fluent in Postscript at all, so that would be very painful. I noticed something that claimed to be a utility to translate bit-mapped characters into Postscript (it was written in Postscript, too) and that could be useful. Meanwhile, playing around with Harald Thunem's font editor (he's put a new version at garbo, by the way. Better user interface), I noticed that if I shoved those connecting Voynich letters/elements in the place of the extended-ASCII box-drawing symbols... they connected on my VGA! So those "single NF atoms" (what's "NF"?) grouped nicely instead of looking slightly disconnected as they did when loaded in ASCII 32 to 127. On an EGA, by the way, they look connected, wherever they are in the character table. Next, going through my mess of disks, I found my Turbo-Pascal Editor toolbox which I thought I'd mislaid. So, the next Frogguy project is: write a WYSIWYG editor that will display both Roman and Voynich letters correctly. The Voynich letters would be kept in ASCII 128 to 255. You would toggle between Roman and Voynich input pressing, say, Alt-R and Alt-V. In fact, I do not see why this editor could not accept both Neo-Frogguy and Classic-Currier input: a Currier letter boils down to a macro for a Frogguy string. So, you'd press Alt-R when you are typing English, Alt-C to switch to inputting in Currier transliteration, Alt-F in Frogguy, Alt-R back to Roman. I only have to disentangle those 5,000 lines of Pascal code first... But once I locate the relevant procedures and how they hook into the rest, it ought to be very easy indeed. I have only given some thought so far to the endeavour. But it seems that it should be fairly easy to simplify the interface so that one does not even have to use the "capitalization rule". As I think I mentioned in the DOC file, gallows "struck though" (hence capitalized) occur in well-defined environments in, say, 95% of the cases, i.e. whenever following or preceding . Ditto for other connecting letters: if you'd write Frogguish the editor would *know* that, since there is an , the next non-gallows must be connected to it to the left, i.e. it would replace with automatically. If you did want to prevent that from happening (say you are transliterating on of the battier dingbats), why, typing a space in-between would do it. As for printing the stuff... hm. When I had a 9-pin dot-matrix printer hooked to my Kaypro II (yes, folks, that was before Atlantis sunk!) there was a public domain utility that would let you design and print your own dingbats. So, if it could have been done on lowly 9-pin (a Star Gemini, 'twas), on one of these noble laser printers... From j.guy@trl.oz.au Tue Aug 04 05:50:50 1992 Date: Mon, 3 Aug 92 15:50:50 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9208030550.AA08256@medici.trl.OZ.AU> To: voynich@rand.org Subject: binary/ternary cipher? Status: OR Karl Kluge follows up on Jim Reeds's hypothesis with a useful list of captions. I have another question (remember: I'm cryptologically unwashed). If we have a Baconian-style biliteral cipher in which the C-like and I-like strokes carry the information, then what does the fact that like occurs mostly next to like tell us about the cipher? All I can figure is that the code would use mixed sequences such as ABAB, ABBA etc for low-frequency letters, and stuff like AAAA, BBBB for "e" and "s" (supposing the language was English). I'm assuming a simple substition cipher here. Next question: if it is not a simple substitution cipher, but something like a Vigenere, what can that same fact tell us about the key? Next question: assume it is such a binary system, and that we do not know how the text was enciphered prior to being transcribed into "Baconian". Doesn't the propensity of i-like occurring with i-like, c-like with c-like considerably restrict the possibilities? Could, say, a Vigenere result in such a pattern? I think (gut-feeling, seat-of-the-pants) that the same questions would make sense if it was a Trithemius-style ternary cipher. Truth is, I don't know... From j.guy@trl.oz.au Tue Aug 04 23:46:00 1992 Date: Tue, 4 Aug 92 09:46:00 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9208032346.AA09087@medici.trl.OZ.AU> To: voynich@rand.org Subject: Speculations on the VMS writing system Status: OR ... that is, if there is one :-(I am still of course working under the hypothesis that the VMS is not a cipher). Robert Firth's dream (do you remember?) in which a butler tauntingly demonstrated, salt and pepper cellars in hand, that we'd never sort out consonants from vowels in the Voynich script has had me think. I first mentioned here Sanskrit-type alphabets, or perhaps more correctly, syllabaries, in which a letter represented a consonant plus an A and letters or diacritics added left, or right, or above, or under that letter, were used to change the value of that vowel. Then I stumbled across a half-forgotten reference to early Latin spelling in Carl Buck's comparative grammar of Latin and Greek, which I dutifully posted here. The latest fruits of my snooping around now. I eventually, reluctantly, bought this book on how to write Malay in the traditional Arabic script. Reluctantly because its title "Learn Jawi" is incomplete. It should have been: "... or the Art of Taking Three Hundred and Sixty Pages to Explain Obscurely What Anyone Could Teach Clearly in Less Than Twenty Pages, But We Couldn't Sell It For Twelve Bucks Then." The Jawi script is defective in its representation of vowels, and and a bit overspecific in its consonants. You can read "Indonesian" for "Malay" in what follows, by the way: the two languages are about as different as Serb is from Croat, perhaps even less. First, you must know that you can't have more than two consonants in a row in a Malay word and that, when you do, they must be flanked by vowels. Thus a word can neither start or end in two consonants. And another thing: "ng" and "ny" each represent a single sound and count for one consonant, logically; they are also represented by a single letter in the Arabic script. In other words, a typical Malay word is made up of syllables which can be either CV or CVC. The two semiconsonants "w" and "y" count as consonants. Unfortunately, they are spelt "u" and "i" when last consonants of CVC syllables, to make things a bit more difficult, e.g. pulau is really a CVCVC word, *not* CVCVV. The big problem is that Malay has six vowels (a, e, i, o, u and schwa, which is called "pepet") and Arabic only three letters to represent them: a, y, w. The first thing you must know is how to syllabify a word. Dead easy: between consonants or, if no consonant clusters, after the vowel. Allow me to write capital G for "ng", capital N for "ny", "E" for schwa, and "Y" and "W" for semiconsonantal "i" and "u" to make things clearer. So: can-tik, tEr-ta-wa (spelt tertawa), bE-la-kaG (belakang), ma-nuk-Na (manuknya), pu-laW (pulau) First rule: Mapping the six vowels to the three Arabic letters. a --> a e,i --> y o,u --> w E --> nothing E.g.: tErtawa --> trtawa toloG --> twlwG Second rule: Don't write the "a" in CaC: E.g. bE-la-kaG --> blakG ma-kan --> makn dar-jah --> drjh pu-laW --> pwlw Third rule: Final k. Write it q. E.g. ma-nuk --> manwq. Fourth: Initial vowels other than a: write an "a" before them. E.g. elok --> aylwq, oraG --> awrG We're about half-way through, but I will stop here, on this piece of arbitrary madness, rule four. It ceases to be mad and arbitrary when you know some Arabic, but let it rest at that. It is now painfully obvious that such writing rules impair our ability to distinguish consonants from vowels, and how! Robert Firth's salt-and-pepper butler in its glory. And that if such is the writing system of the VMS, then we'll never figure out how to pronounce it, short of identifying the language. Lastly, we shouldn't be surprised at seeing the same caption applied to a star, a nymph, a tuber: written alike, it could well have been pronounced quite differently. Hm, I'd do better to work on that Voynich WYSIWYG editor, wouldn't I? So right. But before leaving, one more stone added to the Marco Polo hypothesis. He brought back a bevy of Malay scholars who wrote their knowledge in Malay in the Jawi script. However, they used a Roman-looking alphabet so as not to arise the ire of the Church and the Council of Ten: writing in Arabic would have given them all away as heretics. No, of course, I don't believe it myself, you already knew that. From gcole@manta.nosc.mil Wed Aug 05 10:27:52 1992 Date: Tue, 4 Aug 92 18:27:52 -0700 From: gcole@manta.nosc.mil (Guy Cole) Message-Id: <9208050127.AA07182@manta.nosc.mil> To: voynich@rand.org Subject: Unsubscribe Status: OR Please drop me from the mail list. Thank you and good luck. --Guy Cole From reeds@gauss.att.com Wed Aug 05 12:40:40 1992 Message-Id: <199208050341.AA00618@rand.org> From: reeds@gauss.att.com Date: Tue, 4 Aug 92 23:40:40 EDT To: voynich@rand.org Subject: Voynich spelling speculation Status: OR Speculation, possibly contained in recent post by Guy. Suppose the Voynich language had (at some stage of abstraction or other) a regular VCVCVC... alternation, suppose that all the -like glyphs stood for CV pairs and the -like glyphs stood for VC pairs, and that and were singleton Vs. Then a string of type VCVCVCV might be grouped as VC VC V CV CV which would be written with a pair of -likes, an - like, followed by an -like. The "neutrals" could be something like extra Cs, needed because consonant clusters in fact occur, which are treated by grouping patterns like VCCVCVC into VC C VC VC, that is, by treating the second consonant in a cluster into a vowel-less syllable of its own. Karabadangbaraka? Akarabgandabarak! Jim Reeds From j.guy@trl.oz.au Thu Aug 06 22:17:21 1992 Date: Thu, 6 Aug 92 08:17:21 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9208052217.AA12014@medici.trl.OZ.AU> To: voynich@rand.org Subject: f3w fonts Status: OR Jim Reeds just told me that he did not see any difference between the Voynich fonts f3w00.fnt and f3w01.fnt I posted last week or so. ... and he was right. They are both f3w00.fnt, that is, uppercase Roman letters have been transmogrified into Voyich squiggles in both files. I cringe, covered in shame. I am too ashamed to repost the proper f3w01.fnt to all. E-mail me if you want it. It's 2462 bytes long zipped and uuencoded. j.guy@trl.oz.au From kanatek!matt@esleng.ocunix.on.ca Wed Aug 12 05:32:39 1992 Date: Tue, 11 Aug 1992 16:32:39 -0400 From: kanatek!matt@esleng.ocunix.on.ca (Matthew Harding) Message-Id: <9208112032.AA25328@kanatek> To: voynich@rand.org Subject: copy of voynich Status: OR Does anyone have the procedure for getting the voynich manuscript copy from the yale library? If someone could mail me the procedure I would be very grateful. What was the approximate price? Thanks, matt@kanatek.ocunix.on.ca From j.guy@trl.oz.au Thu Aug 13 03:13:50 1992 Date: Wed, 12 Aug 92 13:13:50 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9208120313.AA18999@medici.trl.OZ.AU> To: voynich@rand.org Subject: D'Imperio's 10-stroke Voynich Status: OR Jim Reeds sent me a short MS by Mary D'Imperio "Some Ideas on the Construction of the Voynich Script". I received it yesterday. She breaks down the Voynich letters into ten basic elements. Some thoughts: I count eleven elements not ten, for we must distinguish between the upwards plume that starts at the top of a letter, and that that starts at the bottom. Thus Currier's is with the bottom plume, <)> in Neo-Frogguy, and his is with the top plume, <'> in Paleo and Neo-Frogguy ("Paleo-Frogguy"! It makes me feel so old; ancient even). However, we can bring that number back to 10 by considering the 9th and 10th strokes in D'Imperio's list as made up of two strokes: left-connected loop and vertical stroke (No. 9) or downward plume/tail (in No. 10), which is the 3rd stroke in D'Imperio's list. There is a question in the order of the strokes. If Currier is right, the intruding gallows were written after the group into which they intruded. Now, we find his with gallows intruding, but we also find his *followed* by gallows. Does that mean that the sequence is really and that is ? I think not, because where else do we find followed by a space or an end of line? I wonder if Currier was right there. Grab a fountain pen and write a cross twice: first cross horizontal line first, second cross vertical line first. Leave the ink to dry. Can you tell the difference? Bad example: one would need to do it with a goose quill, on vellum, in the ink of the time, but you get the gist of my argument. I do think that it would have been more natural to write , , , in that order. Unfortunately, examining the script from a xerox copy or even a microfilm won't tell. It has do to be the original. Now back to struggling with that Voynich WYSIWYG editor. From Karl.Kluge@G.GP.CS.CMU.EDU Thu Aug 13 00:49:00 1992 Date: Thu, 13 Aug 1992 00:49-EDT From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Star labels from folio 68r1 -- what do y'all make of this? Message-Id: <713681397/kck@G.GP.CS.CMU.EDU> Status: OR In my continued quest to look for label matches in an attempt to decide if the text actually means anything, I transcribed the star labels from folio 68r1 and checked for matches in the labels I had previously transcribed and in the D'Imperio transcription. Spaces were allowed after any character, 6/8 confusion was allowed. Here are the results. Make of them what you will. Petersen locus <***XX*> Label Matches ----------------------------------------------------------------------------- <68101A> 9F9/8AR9/SFS9FOE9 none <68102A> OFO8AE9 none <68103A> O8SCQ9 none <68104A> OPSCO89 none <68105A> OFOAE9 none <68106A> SOY9 none <68107A> OQC9 10 lines in D'Imperio <68108A> OPZC9 8 lines in D'Imperio <68109A> OP989 8 lines in D'Imperio <68110A> OFCAR 8 lines in D'Imperio <68111A> WOQ9 none <68112A> 9PSO89 4 lines in D'Imperio <68113A> OP92 1 label (OP92AJ - pg 146 (Pisces) locus 22), 1 line in D'Imperio (15034B) <68114A> OEOR 65 lines in D'Imperio <68115A> OX9 32 lines in D'Imperio <68116A> CVSCOR none <68117A> OPS6O none <68118A> OP9PS2 (OP9FS2?) none <68119A> OPOE appears as root in 8 other labels; 101 lines <68120A> OPOR root in 4 other labels; 39 lines in D'Imperio <68121A> OMAR none <68122A> OFOE89 11 lines in D'Imperio <68123A> 9FS89 matches p224 locus 37; 14 lines in D'Imperio <68124A> POCCO8Q9 none <68125A> OW9 8 lines in D'Imperio <68126A> OFCCO8AE 1 line in D'Imperio (00517A) <68127A> OFZOR 3 lines in D'Imperio (02604A,02608A,04209A) <68128A> OR8AM 47 lines in D'Imperio <68129A> OPOSC89 none <68130A> 8OESC89 4 lines (15303B, 16202B, 16210B, 16611B) <68131A> OF9/OFS89/OFAR89/PCAR/OE69 none Lines matched # of labels from 68r1 ------------------------------------- 0 ************** (14) 1 - 5 ***** (5) 6 - 10 ***** (5) 11 - 15 ** (2) 16+ ***** (5) I looked at the labels which had OPOE and OPOR as roots. For OPOR they occur on pages 199, 200, and 201 -- all pharm. folios. For OPOE they occur on pages 148 (zodiac), 151 (zodiac), 201 (pharm), 202 (pharm), and a nymph on f84r. For your edification the related labels are: Checking O/?P/?O/?E/? labels:<14801A> OPOES89 labels:<14802A> OPOEOARAJ labels:<14804A> OPOES8 labels:<15125A> OPOEAJ labels:<20125A> OPOE89 labels:<20237A> OPOE8 labels:<20249A> OPOEAROE labels:<68119A> OPOE labels:<84R06A> OPOE9 Checking O/?P/?O/?R/? labels:<19901A> OPORSCP9 labels:<19910B> OPOR/AJ labels:<20017A> OPORAJ labels:<20017B> OPORA labels:<20128A> OPORAN labels:<68120A> OPOR On the assumption that labels are (often) single words, does the above look like the way real languages behave? ********************************************************************** P.S. The set of labels searched for matches comes from: Page 146: Pisces f70v2 (boxed folio # in Petersen copy) Page 147: Aries f70v1 (boxed folio # in Petersen copy) Page 148: Aries f71r Page 149: Taurus f71v Page 150: Taurus f72r1 (boxed folio # in Petersen copy) Page 151: Gemini f72r2 (boxed folio # in Petersen copy) Page 199: Pharm. f88r Page 200: Pharm. f88v Page 201: Pharm. f89r Page 202: Pharm. f89v Page 223: Pharm. f100r Page 224: Pharm. f100v Page ???: Stars f68r1 Page ???: Nymphs f80r (Petersen loci 1-10) Page ???: Nymphs f84r (7 nymphs at top) From jim@rand.org Fri Aug 14 14:19:58 1992 Message-Id: <9208140519.AA09885@mycroft.rand.org> To: voynich@rand.org Subject: Re: copy of voynich In-Reply-To: Your message of Tue, 11 Aug 92 16:32:39 -0400. <9208112032.AA25328@kanatek> From: Jim Gillogly Reply-To: jim@rand.org Date: Thu, 13 Aug 92 22:19:58 PDT Sender: jim@mycroft.rand.org Status: OR > kanatek!matt@esleng.ocunix.on.ca (Matthew Harding) writes: > Does anyone have the procedure for getting the voynich manuscript copy > from the yale library? Send a letter to: Robert Babcock Beinecke Rare Book Library 1603a Yale Station New Haven CT 06520 You could include information on why you want the copy and your academic affiliation, if any. You'll be asked to fill out a form with the official request. It took me about 5 or 6 months to get my copies. > What was the approximate price? About $40 for either xerographic prints or microfilm. Jim Gillogly From jim@mycroft.rand.org Fri Aug 21 01:15:33 1992 Message-Id: <9208201615.AA06989@mycroft.rand.org> To: voynich@rand.org Subject: Prescott Currier Date: Thu, 20 Aug 92 09:15:33 PDT From: Jim Gillogly Status: OR For the newer members of the list, Capt. Currier did important statistical work on the Voynich Ms. in the 60s and 70s; some of his work was described by Mary D'Imperio in her monograph (An Elegant Enigma) and in the proceedings of a seminar in 1976 (currier.paper in the rand.org archives). Here's a letter from him to this group. ------------------------------------- Dear Jim, I am only recently back from three months in Florida (collecting marine fossils among other things) and have just now succeeded in reading through the last batch of material you sent me. For want of a more appropriate adjective 'overwhelmed' must suffice. The brain and computer power in your Group reminds me of a similar concentration in the Manhattan Project. Let me hasten to add that I am thoroughly fascinated by it all and ask that you continue to send me similar amounts of your Group Communications in the future. It occurs to me that some members of your group might like to know a little something about me. So herewith very briefly: AB in Romance Languages; taught myself Japanese and spent several pre-war years and most of the war (ie WWII) working on Japanese cryptosystems. The Navy sent me to Russian Language School in 1945 and I spent the ensuing years in the intelligence business with two tours on the US Embassy in London (1946-48 and 1955-58). Retired in 1963 and took an advanced degree in Comparative Philology at the Univ. of London (School of Slavonic Studies and University College). First became interested in VMS in the late thirties through Billy Friedman and associated myself with the group of IBM officials who, at Friedman's urging, undertook to transcribe and punch up, sort and print some of the Biol. B material (the wives did the transcribing using my alphabet). It wasn't until after I retired from Government service in 1969 however, that I spent serious time working on the VMS - mostly pencil and cross-section paper in an endless variety of frequency counts with time out to ponder and to attempt to interpret the masses of statistics I produced. And the only thing I have to show for it is the little paper that I wrote in the early 70's and presented at the 76 symposium. I stopped working on the MS at that time simply because I couldn't think of anything else to do. I first met John Tiltman in England in 1941. We became close personal friends, as did our families, and maintained our friendship until his death a few years ago. At this point I am reminded that John Tiltman turned over to the then Director of Kew Gardens much of the Herbal section of the VMS and asked him to identify the illustrations. In reply, as I remember it, he did provide names for a few of the plants and tentative identification of a few others but the great majority he classified as 'compositae' (concocted drawings using elements from several plants). Now consider, if you will, the situation that must have confronted the scribes who produced the Herbal Section. Two men (initially one) sitting at a table on which sat a stack of vellum sheets filled with phony pictures of non-existent herbs. Assuming that they knew that the illustrations were deliberate fabrications did they make up stories about the nonexistent uses and benefits of non-existent herbs? or, Did they fill the space around the illustrations with a prearranged text having nothing to do with the illustrations? or, Did they fill the space around the illustrations with a well-constructed script having to do with nothing in particular (ie. deliberate, but recoverable nonsense)? My own rather feeble attempt at a comparative analysis of the two texts (A & B) led nowhere. But I am sure that a well-programmed, high powered computer attack might well produce some interesting results. I am planning to write the Beinecke and order a photocopy of the Manuscript some time soon. By the way, my first communication with the Beinecke took place in 1970 when I wrote offering my services when and if they were planning any serious research. They replied that they had no plans now or in the future for any research, so "Thank you but No." I note with some satisfaction that some of my conclusions have attracted the favorable attention of a couple of the members of your group. I hope that this continues since I still feel that everything I said in the 'little paper' is still valid. I also note that my transcription alphabet is perhaps not suitable for computer use. I trust that final choice will not be overly complicated for all other uses. Which reminds me, I can see no reason why 7/J should not be merged - into J preferably. Even if it turns out that they are in fact two different characters no confusion should result. If you think that anything I have written here would be of interest to other members of the Voynich Group please make it available in any form that suits you. As for posting my name and address I doubt if it would result in a deluge of queries since I don't have a computer. We could try it for a while anyway. I'm looking forward to receiving the next "batch". Best regards, Prescott Currier PO Box 267 Damariscotta ME 04543 From ACW@RIVERSIDE.SCRC.Symbolics.COM Fri Aug 21 11:32:00 1992 Date: Fri, 21 Aug 1992 11:32-0400 From: Allan C. Wechsler Subject: census To: j.guy@trl.oz.au, voynich@rand.org In-Reply-To: <9208210502.AA11910@medici.trl.OZ.AU> Message-Id: <19920821153231.9.ACW@PALLANDO.SCRC.Symbolics.COM> Status: OR Date: Fri, 21 Aug 1992 16:02 EDT From: j.guy@trl.oz.au (Jacques Guy) You (you'll know who is you) you left me high and dry on that matter of estimating the entropy from a sample text, too. This wasn't me, was it? From j.guy@trl.oz.au Sat Aug 22 05:02:58 1992 Date: Fri, 21 Aug 92 15:02:58 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9208210502.AA11910@medici.trl.OZ.AU> To: voynich@rand.org Subject: census Status: OR Yes, let's stand and be counted! The reason behind this injunction is not entirely frivolous. I am working on a WYSIWYG editor for PCs that will display both Roman and Voynich letters. Two days ago, after much swearing and cursing ot Microsoft, I installed Windows on my 486 at work and my 386 at home, and... I discovered the joys of Visual Basic! No, I am not going to drop my plain MS-DOS editor project, definitely not, but I'll be thinking about writing a Windows version of it in Visual Basic if there are enough of us running Windows. And what's happened to Andras Kornai and John Baez and Robert Firth and... You (you'll know who is you) you left me high and dry on that matter of estimating the entropy from a sample text, too. As was written on the wall at Belshazzar's feast: Many, many tickle you far since. From reeds@gauss.att.com Tue Aug 25 23:06:06 1992 Message-Id: <199208251406.AA11370@rand.org> From: reeds@gauss.att.com Date: Tue, 25 Aug 92 10:06:06 EDT To: voynich@rand.org Subject: Voynich transcription Status: OR Here is a NEW IDEA. We all want a complete transcription of the VMS but are too lazy to actually do the work. So here is my plan: 1. We open a branch of the voynich mailing list in the former Soviet Union, getting several people there as enthusiastic as we are. 2. We procure extra Yale prints and Petersen copies, Neo-Frogguy or Currier code sheets, and a PC or two. 3. We assign to the new Eastern members the task of finding copyists. We pass the hat around to find the money to pay them 25 cents a page! This might work because the current monthly salary for scientific professionals in the FSU is something like $30. At one hour per page we are offering $40 a month. For a very modest sum (a few hundred in all) we can pay for several complete copies, so we have a chance to control errors. Jim Reeds From Karl.Kluge@G.GP.CS.CMU.EDU Tue Aug 25 16:14:00 1992 Date: Tue, 25 Aug 1992 16:14-EDT From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Re: stand & be counted Message-Id: <714773671/kck@G.GP.CS.CMU.EDU> Status: OR I'm still here. I'm desperately trying to finish writing my thesis and I have a chapter of a book to write next month, but I am here. Unfortunately, the only Voynichesque thing I've had time to do was the star folio label search I posted the results of. Karl From reeds@gauss.att.com Wed Aug 26 03:48:02 1992 Message-Id: <199208251854.AA20237@rand.org> From: reeds@gauss.att.com Date: Tue, 25 Aug 92 14:48:02 EDT To: voynich@rand.org Subject: Voynich stand & be counted Status: OR Jacques says: Yes, let's stand and be counted! ... ... And what's happened to Andras Kornai and John Baez and Robert Firth and... Count me! I have seen John Baez lurking in netnews, so he is still alive, but not in this list. Maybe he is no longer a friend of Aslan? I am preparing my entropy magnum opus for Jacques, but it might not be what he wants. TTFN, Jim Reeds From RJB@U.WASHINGTON.EDU Wed Aug 26 05:33:00 1992 Date: Tue, 25 Aug 92 13:33 PDT From: RJB@U.WASHINGTON.EDU Subject: standing sheepishly Sender: RJB@MAX.U.WASHINGTON.EDU To: voynich@rand.org Message-Id: <05FC85CFFEDF82491F@MAX.U.WASHINGTON.EDU> X-Envelope-To: voynich@rand.org X-Vms-To: IN%"voynich@rand.org" Status: OR I'm here, but after waiting 4+ months for my little treasure from Yale have not done much with it. Sad but true. --rjb From kornai@Csli.Stanford.EDU Wed Aug 26 06:03:06 1992 From: kornai@Csli.Stanford.EDU (Andras Kornai) Message-Id: <9208252103.AA19800@Csli.Stanford.EDU> Subject: Standing and being counted To: voynich@rand.org Date: Tue, 25 Aug 92 14:03:06 PDT X-Mailer: ELM [version 2.3 PL11] Status: OR Hi Voynologists, I'm still out there and as I recall I did send Jacques my own piece about entropy, only I haven't posted it. Also been trying to do some image filtering in the hopes that eventually we'll be able to have tiff files of the whole MS on line, but so far nothing came out of it. I still try to stay on top of developments and have even recruited one new member to the group who will try his consonant/vowel discrimination software on the stuff, so pronouncable Voynich fans will have a little more food for thouight. Andras Kornai PS. I'm willing to pitch in if we decide it's a good idea to pay transcribers, but I don't think Eastern Europe is a good idea -- the administrative overhead is frightening. Slave labor by starving linguistics and classics graduate students at 10$/page makes more sense to me. I have been both an underpaid Eastern European scientist and a starving linguistics grad so I should know:-) From jim@rand.org Wed Aug 26 07:58:00 1992 Message-Id: <9208252258.AA15945@mycroft.rand.org> To: voynich@rand.org Subject: Re: census From: Jim Gillogly Reply-To: jim@rand.org Date: Tue, 25 Aug 92 15:58:00 PDT Sender: jim@mycroft.rand.org Status: OR I'm still here, too. Voynich-wise, I've mostly been handling communications: occasional care packages to our disconnected colleagues, and managing the list and archive. My main immediate goal is to forward the transcription effort; after I finish integrating the half-dozen more transcribed pages that have been submitted I'll put out another request for volunteers, and work in some transcription of my own. The Yale print and microfilm finally came. In order to make sure we're talking about the same pages in the Yale print, let's follow Jim Reeds' numbering scheme... I've now scribbled on my copy. This turned out to be important when he was trying to explain to me why a page with a number 68 on it wasn't f68r (it was numbered after folding). Jim Reeds says: > I numbered the sheets in the Yale print consecutively. They are in > 1-1 correspondence with frames on the microfilm. I call them YF numbers > (for Yale frame). Starting with YF 1 which is f1r through YF 114 (sounds > like an airplane!) which has the oladabas on it. (There is a YF 115 with > a note by Krauss.) > > Here are some of my YF numbers with foldouts, plus a few synch-holders > || denotes binding gutter, always a valley > + denotes a ridge crease > - denotes a valley crease > [xxx] means has folio number xxx > > YF 1 f1r [1] > YF 20 f20v || f21r [21] > YF 40 f40v || f41r [41] > YF 60 f66v || f67r1 [67] + f67r2 > YF 61 f67v2 - f67v1 || > YF 62 || f68r1 - f68r2 - f68r3 (r3 has extra - ?) > YF 63 f68v3 + f68v2 [68] (v3 has extra + ?) > YF 64 f68v1 || f69r [69] > YF 65 f69v || f70r1 - f70r2 (r2 has extra -) .. YF 66 f70v2 + f70v1 || f71r [71] (v2 has extra +) > ... > YF 80 f83v || f84r [84] > YF 98 f99v || f100r [100] > YF 99 f100v || f101r1 - f101r2 > YF 100 f101v2 [1??] + f101v1 || > YF 101 || f102r1 - f102r2 (r2 has extra crease) > YF 102 f102v2 [102] + f102v1 || f103r [103] (v2 has extra crease) > YF 103 f103v || f104r [104] Jim Gillogly From foxd@silver.ucs.indiana.edu Wed Aug 26 22:54:39 1992 Message-Id: <199208261354.AA09498@rand.org> Date: Wed, 26 Aug 1992 08:54:39 -0500 From: daniel fox To: voynich@rand.org Subject: still here Status: OR I'm still here. From EVANS@BINAH.CC.BRANDEIS.EDU Wed Aug 26 23:01:00 1992 Date: Wed, 26 Aug 1992 10:01 EDT From: Ronald Hale-Evans Subject: still here too To: voynich@rand.org Message-Id: <01GO17PX2JMO91VYQM@BINAH.CC.BRANDEIS.EDU> X-Vms-To: VOYNICH X-Vms-Cc: EVANS Status: OR Date sent: 26-AUG-1992 I'm still a Friend of Aslan too, but I can't devote much time to the project. Mostly I've been observing you guys. * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * I don't want to die; existence is one of my strong points! Ron Hale-Evans--Email: evans@binah.cc.brandeis.edu, GEnie: R.HALE-EVANS From jroth@dic.k8.rt.bosch.de Wed Aug 26 23:57:22 1992 Date: Wed, 26 Aug 92 16:57:22 +0200 From: Jakub Roth. K8/DIC2. Tel. 1533 Message-Id: <9208261457.AA13857@dic.k8.rt.bosch.de> To: voynich@rand.org Subject: Re: Let's stand up and count ourselves Status: OR Still here and very interested :) though not very active :( Oh, BTW, I have good contacts in Eastern Europe and I think we really could get someone type in the text. There'd be little overhead since people already do have computers there. I sent queries regarding this earlier today so I expect to learn more within a few days. Jakub Roth MIT, Boston jroth@athena.mit.edu Robert Bosch GmbH, Reutlingen jroth@dic.k8.rt.bosch.de From j.guy@trl.oz.au Thu Aug 27 00:29:47 1992 Date: Wed, 26 Aug 92 10:29:47 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9208260029.AA16637@medici.trl.OZ.AU> To: reeds@gauss.att.com, voynich@rand.org Subject: Re: Voynich transcription by Slav Labour Status: OR (pun intended, of course, don't you know me yet?) It's a fiendish good idea, Jim! I pledge $100 (US, not our marsupial mouse money) towards it. Let's hope it gets off the ground. How many of us are left? (Census again!) From rcarter@nyx.cs.du.edu Thu Aug 27 00:54:50 1992 Date: Wed, 26 Aug 92 09:54:50 MDT From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9208261554.AA22234@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: voynich@rand.org Subject: As always... Status: OR As always, I am here; still working on the Aztec/VM `connection' and my transcription... It -is- tempting to (attempt) let loose the xUSSR hordes of no doubt available labor; I have a friend in Siberia trying to do the same (on a different project) with hard-won minimal success. I am finidng the time spent on transcription allows me a better feel for V- and at times (hallucinatory, I know) I think that with the proper insight, it will all come clear. And, as always, my -real- work is keeping me more busy than I thought it would; it is hard changing the world... :-) Regards to all, Ron (with Nyx back online!) From kibo@world.std.com Thu Aug 27 14:45:49 1992 Date: Thu, 27 Aug 1992 01:45:49 -0400 From: kibo@world.std.com (James "Kibo" Parry) Message-Id: <199208270545.AA06630@world.std.com> To: voynich@rand.org Subject: "I'm still here" <--now you don't need to read the message body Status: OR I'm still here, although I haven't had any time lately to help Jim R. with his font. -- K. From firth@SEI.CMU.EDU Fri Aug 28 22:01:52 1992 Date: Fri, 28 Aug 92 09:01:52 -0400 From: firth@SEI.CMU.EDU Message-Id: <9208281301.AA08186@bp.sei.cmu.edu> To: voynich@rand.org Subject: Mail Retry Status: OR >From MAILER-DAEMON@SEI.CMU.EDU Thu Aug 27 17:55:50 1992 Received: from SEI.CMU.EDU by bp.sei.cmu.edu (5.64/2.5) id AA06716; Thu, 27 Aug 92 17:55:46 -0400 Received: from BP.SEI.CMU.EDU by sei.cmu.edu (5.65/2.3) id AA17173; Thu, 27 Aug 92 17:55:43 -0400 Received: by bp.sei.cmu.edu (5.64/2.5) id AA06709; Thu, 27 Aug 92 17:54:11 -0400 Date: Thu, 27 Aug 92 17:54:11 -0400 From: Mail Delivery Subsystem Subject: Returned mail: Cannot send message for 1 day Message-Id: <9208272154.AA06709@bp.sei.cmu.edu> To: firth@SEI.CMU.EDU Status: R ----- Transcript of session follows ----- 421 rand.org.tcp... Deferred: Connection timed out during user open with rand.org ----- Unsent message follows ----- Received: by bp.sei.cmu.edu (5.64/2.5) id AA04629; Wed, 26 Aug 92 17:27:32 -0400 Date: Wed, 26 Aug 92 17:27:32 -0400 From: firth@SEI.CMU.EDU Message-Id: <9208262127.AA04629@bp.sei.cmu.edu> To: voynich@rand.org Subject: Adsum Folks Yes, I'm still here. The summer was taken up with student supervision, assorted technical work, a trip to Australia, and even a little rest & relaxation. But no Voynich work. Maybe it's time to plan the Autumn games of Team Voynich? Robert From madmike@netcom.com Sat Aug 29 01:54:32 1992 From: madmike@netcom.com (Michael Parmett) Message-Id: <9208281654.AA03517@netcom.netcom.com> Subject: SABC To: voynich@rand.org Date: Fri, 28 Aug 92 9:54:32 PDT X-Mailer: ELM [version 2.3 PL11] Status: OR Well, I'm still lurking in the background......Mike -- From evy@well.sf.ca.us Sat Aug 29 02:28:45 1992 Date: Fri, 28 Aug 1992 10:28:45 -0700 From: Evelyn Pine Message-Id: <199208281728.AA26617@well.sf.ca.us> To: voynich@rand.org Subject: voyeur of voynich Status: OR I'm busy creating a homunculus to do the transcription for us. From j.guy@trl.oz.au Sat Aug 29 04:48:12 1992 Date: Fri, 28 Aug 92 14:48:12 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9208280448.AA19948@medici.trl.OZ.AU> To: voynich@rand.org Subject: odds and ends Status: OR Slav Labour ----------- Andras Kornai tipped a bucket of cold water over the idea, but then Jakub Roth promptly rekindled it. Which leads me to: Why are we so lazy? (a.k.a. transcription) ------------------------------------------ Transcription is an infuriating job: is this an or a <2>? Often it's in-between, a <2> that's a bit -like, which is not the same as an that's a bit <2>-like. The more you go the slower you go, because the pickier you become. Oh, for being able to scan the whole damn thing, and having an editor that lets you insert a transcription in-between each line of Voynich! I fully understand why we are so lazy. Pronounceable Voynich --------------------- Someone mentioned pronounceable Voynich... 'twas bucket-toting Kornai! Hark, o, only speaker of a non-Indo-European language amongst us, I have a tip or two for thee and thy newly recruited minion (and anyone else who cares to listen in). Bitter experience has taught me that, whatever one-to-one scheme I tried, I ended up with unpronounceable sequences. Well, not really unpronounceable -- I *can* pronounce [tsxqwktl] but I do not *really* enjoy it. Yet pronounceable Voynich is a good idea, a useful tool, because we'd get a better feel for the recurring patterns of the text. Here is how I would go about it seriously (I did have two goes at the problem if you remember, but I had tongue in my cheek in both cases, with varying degrees of firmness). I would consider that Voynichese is written in an Indian-type (viz Sanskrit) syllabary. Each letter has an inherent vowel (it's "a" in Sanskrit and most other such languages, and "o" in Thai). So when I see Currier's, say,

I pronounce it "pa". The other p-syllables are obtained by adding letters around

. I say "around" because that's how it works in those Indian writing systems: you put "i" before, "u" under, "o" in two parts, one before, one after, etc. There would be no harm, I suppose, in saying: in Voynichese the modifying letter always comes after. So when I see Currier's I would pronounce "po" (not pa-o). And perhaps I would pick a Voynich letter to mean: no vowel here. In that manner, we'd be sure to have a system that *always* yields a pronounceable sequence, since CVCVCVCV is the default pattern. Which Voynich letters to pick for "modifiers"? I'd say those which Sukhotin's algorithm picked as vowels, AND which do NOT occur doubled. That rules out only , I think. For the rest, it depends on what we want Voynichese to sound like. It's not so easy as it sounds, because if we want to include the zany letters we are going to end up with 100 consonants! So perhaps it should be an analytical scheme (Frogguy-like, or if you prefer Neo-D'Imperio-like). But we'll end up with some bloody long words then! (So what? Some languages are unbelievably windy. I've been reading soc.culture.telugu. Telugu is even windier than Russian and Indonesian put together. No, I will spare you my latest update to the Marco Polo theory, so you can stop holding your noses *now*!) Voynich Editor -------------- It's coming, it's coming. I've got the bits and pieces, and I'm working on the interface. I am fundamentally lazy, and I do not want to have to learn, let alone remember, a series of commands. The WordStar-like commands of the Turbo-Pascal inbuilt editor is all I am willing to fight with. But not everybody is used to those commands, and I believe that everybody has a right to be as lazy as I am. From j.guy@trl.oz.au Fri Sep 04 04:14:12 1992 Date: Thu, 3 Sep 92 14:14:12 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9209030414.AA28703@medici.trl.OZ.AU> To: voynich@rand.org Subject: yet another Voynich dingbat? Status: OR The interrogation mark is there because I don't have Jim Reeds' dingbat list here at work. I was desultorily looking at folio 79v in Bennett's "Scientific and engineering problem-solving with the computer", and there it was, winking at me. Second paragraph, 7th line, second word, Currier , Frogguy . The plume is almost exactly a baby Currier , Frogguy . Is it another plume, or a random variety of the common one? I think, and I hope (which may vitiate my thinking) that it is only a random variety. On the first line of the fourth paragraph, same page, I can see two plumes that are cross-breeds of it and the common plumes. Here is that line in Neo-Frogguy (can't give it to you in Currier, the Currier alphabet is at home and I keep getting confused with the gallows letters): qjetc9 lpe'tcox 4olpaiv oljetc89 oqpaxe't89 oxlpai2 oqpaiiv olpcc89 ^ |_______________________________ | Looking very closely at the first plume, there, I am sure that it was written in two strokes: there is an ever-so-slight break between the left-hand, c-like bit of its loop, and the plume itself, that is, the curlicue that sweeps down and left towards the . But could it rather be a glitch caused by a crease in the vellum? I mean, the plume would be a single stroke, but the scribe's quill hit a crease and "hiccuped"? Looking again carefully at the other plumes on that page I am struck by how similar they are to "baby" 9's. The very first plume, first word of the first line, qjox8e'tc89, is a perfect Voynich <9> in miniature. And <9> is clearly written in two strokes. I am sorry to drop these dismal observations onto your tired laps. The Voynich editor is taking shape, slowly. I have learnt that it pays to be very, very careful in the early stages. From jim@mycroft.rand.org Mon Sep 07 05:54:25 1992 Message-Id: <9209062054.AA13530@mycroft.rand.org> To: voynich@rand.org Subject: Voynich MS: Letter from Mary D'Imperio Date: Sun, 06 Sep 92 13:54:25 PDT From: Jim Gillogly Status: OR 29 August 1992 Thanks for your letter of 22 August, and the copies of communications in the group. They are always very stimulating and enjoyable reading. It looks as though people are tackling really interesting questions fro the most part, and not rehashing old, stale ideas that have been tried over and over again. I'm sorry I have fallen down on the job of transcribing text; the truth is, that when I tried to work with some of the pages I have, I realized most of them were not sufficiently clear copies to work from. I tried pages from several sections of the ms in my heterogeneous collection of copies obtained from various sources, and had very poor success at producing transcriptions that were usable. So I gave up on that effort. Since then, I have been preoccupied with other researches and haven't done much more on the Voynich. I did complete a large part of one investigation I was making, trying to follow up some ideas I had on the "beginning-middle-ending" structure in the text surrounded by spaces. I tried to come up with a set of consistently applicable rules of decomposition and a corresponding set of beginners, middles, and enders, and then compared them in the Herbal A and Herbal B corpus we already have in computer readable form. I was surprised and discouraged to find that the enders that could follow a given beginner or beginner+middle in Herbal A showed almost no overlap with the enders following that structure in Herbal B. That odd finding left me not knowing what to try next, and I dropped the project at that point. It occurs to me, though, that if anyone really wants to use his computer skills on testing another person's ideas, I would be glad to write up the work I did as an example. My decomposition rules are not entirely satisfactory, and using a program to try various decompositions to come up with a better one would be worthwhile and interesting in itself. these could then be used to compare instances of beginners and their contacts with middles and/or enders in different hands and languages. Such an analysis might turn up something interesting. The almost complete LACK of overlap I found is odd, to say the least. I'll write it up and send it along when I get back from my next trip, to Brazil 6-19 September. I think the idea of mounting a major effort to get the ms studied and dated is an excellent one. I am amused, though, at the wistful thought expressed by Mr. Wechsler that previous employment at the Puzzle Palace would give me any clout in getting the study made, or getting the Beinecke to go along with it. The quote he gives from my book, by the way, was from Brigadier Tiltman's preface, and was not written by me! I tried hard, while I was still working there, to interest and motivate people in the academic and antiquarian world to support such a study, and had no success. Experts in medieval mss are turned off by the Voynich because they consider it ugly and bizarre compared to the elegant illuminated mss they are interested in. My former employment means nothing whatever, and doesn't make me any more likely to be listened to than employment at IBM or at a college or university -- in fact, a high and prestigious academic position would probably go a lot further. The idea of getting a known scholar or expert to head up the investigation is much more likely to pay off. Finally, I note that RJB (U.Washington) is enthusiastic about a medical theme on gynecology in the ms. I wonder if he is familiar with the similar idea of Prof. Strong in 1945 (cf p. 36 in my book)? Regards, Mary DImperio 4000 Cathedral Ave. NW #106B Washingon DC 20016-5249 From reeds@gauss.att.com Tue Sep 08 00:49:28 1992 Message-Id: <199209071549.AA06393@rand.org> From: reeds@gauss.att.com Date: Mon, 7 Sep 92 11:49:28 EDT To: voynich@rand.org Subject: Voynich MS & John Dee Status: OR The September 1992 issue of ISIS has a review by Nicholas Clulee of ``John Dee's Library Catalogue'' ed. by Julian Roberts and Andrew G, Watson. viii+253 pp, illus, apps, bibl, indexes. London: Bibliographical Society, 1990. $125. (No ISBN cited.) Clulee praises the book to the high heavens: ``...an eminently useful scholarly resource.'' Voynichologists will remember that the printed Yale catalogue cites A. G. Watson and JDLC as claiming that the folio numbers on the VMS are in Dee's hand. Jim Reeds From kibo@world.std.com Tue Sep 08 13:41:55 1992 Date: Tue, 8 Sep 1992 00:41:55 -0400 From: kibo@world.std.com (James "Kibo" Parry) Message-Id: <199209080441.AA24120@world.std.com> To: voynich@rand.org Subject: something amusing. Cc: mhill@us.oracle.com, sramming@athena.mit.edu Status: OR If you have the September 1992 Scientific American (the brain issue) lying around, flip it over and look at the illustration on the back cover. It's an ad featuring a DaVinci-style anatomical drawing of the human brain. However, rather than being annotated in Leonardo's mirror-script, it's in what appears to be a variant of Voynichese. Some illustrator has a sense of humor. (The ad is for Compuserve, by the way.) -- K. From fubar@sequent.com Mon Sep 28 13:46:39 1992 Message-Id: <9209280446.AA04846@eng2.sequent.com> To: j.guy@trl.oz.au (Jacques Guy) Cc: voynich@rand.org Subject: Re: my latest theory... In-Reply-To: Your message of "Mon, 28 Sep 92 09:46:10 PST." <9209272346.AA04870@medici.trl.OZ.AU> Date: Sun, 27 Sep 92 21:46:39 PDT From: Jay Vosburgh Status: OR >(actually, I remember having read something similar in a serious book >on vampires legends), and the Red Countess is described as sacrificing >one nymph per lunar month. McNally and Florescu's IN SEARCH OF DRACULA, while not really about Elizabeth Bathory, does devote several pages to her; perhaps this is the book you were thinking of. In any event, it does not give a specific number of victims, but does give the following tidbits: "...raid of Castle Csejthe on the night of December 30, 1610 ... In the main hall they found one girl dead, drained of blood, and another alive whose body had been pierced with tiny holes; in the dungeon they discovered 'a number' of other living girls, some of whose bodies had been pierced. Below the castle, the authorities exhumed the bodies of some 50 girls." (p 158) "Johannes Ujvary, the countess' major-domo, testified at the trial that as far as he knew, about 37 unmarried girls had been killed, six of whom he had personally lured to the castle with promises of jobs as serving girls. ... Ilona Joo, Elizabeth's old nurse, testified that about 40 girls had been tortured and killed." (p 159) "Aided by her old nurse Ilona Joo, Elizabeth began torturing some of the servant girls at the castle. Her other accomplices included the major-domo Johannes Ujvary, her manservant Thorko, a witch Dorottya Szentes, and a forest witch named Darvula." (pp 157-158) The book refers to her as the "Blood Countess," and says that the killings began soon after the death of Count Ferencz Nadasdy (Elizabeth's husband) in 1600 and ended after the raid in December of 1610. In 1600, Elizabeth would have been about 40 (it is stated that she was 15 when she was married on May 8, 1575). A trial occured during January and February of 1611. Interestingly, Elizabeth was never formally convicted of any crime. Further, "A complete transcript of the trial, made at the time it took place, survives in Hungary today". -J --- -Jay Vosburgh, Sequent Computer Systems, Inc; fubar@sequent.com From foxd@silver.ucs.indiana.edu Mon Sep 28 23:00:59 1992 Message-Id: <199209281401.AA09784@rand.org> Date: Mon, 28 Sep 1992 09:00:59 -0500 From: daniel fox To: j.guy@trl.oz.au, voynich@rand.org Subject: Re: my latest theory... Status: OR Maybe the artist was color blind. This could account for the green blood. From j.guy@trl.oz.au Mon Sep 28 23:46:10 1992 Date: Mon, 28 Sep 92 09:46:10 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9209272346.AA04870@medici.trl.OZ.AU> To: voynich@rand.org Subject: my latest theory... Status: OR The pharmaceutical jars explained! The herbal justified! The astrological diagrams revealed! The bathing nymphs unveiled! The purpose of the plumbing discovered! Sunday. I was reading "Crooks, Crime, and Corruption" Octopus Books, London: Hungarian countess Elizabeth Bathory bathed in the blood of her victims because she believed it preserved her beauty. In the black depths of her castle dungeons at Csejthe, the countess stored well-fed girls ready to have their veins cut open and filtered into pipes than ran into a blood bath. The bathing nymphs are the countess's victims. One of them (the one I reproduced on some tiles of the mahjongg solitaire game), is the countess herself, bathing in blood. The herbal and pharmaceutical jars are eternal youth and beauty recipes. The astrological diagrams have to do with the favourable planetary aspects when herbs are to be collected, and victims bled. The plumbing are the pipes and the filtering system (just *look* at those plumbing folios, and tell me now if you are not feeling that you are slowly losing your grip on your sanity and starting to believe in my theory?) It's a good job I don't have my Voynich stuff here: just checking up on the date when the Voynich MS was acquired by Rudolph II might deflate my wonderful theory. It is fortunate that I do not have a colour reproduction of the folio where Erzsebet Bathory is shown bathing in blood, the blood might not be red (but would a countess stoop to bathing in anything but *blue* blood?). Accounting to my Larousse encyclopedia, she was discovered in 1610. According to "Crooks, Crime, and Corruption": Tried for 610 murders, the countess was condemned to be walled up for life in a room from which all light and sound were excluded. In 1614, she expired after three years of this living death. My 1902 Larousse disagrees with the number of victims: Elle fit e'gorger a` peu pre`s quatre-vingts jeunes paysannes dans son cha^teau de Csejthe [comitat de Nyitra] pour se baigner dans leur sang. Surprise en flagrant de'lit par le palatin Georges Thurzo [1610], elle fut condamne'e a` une de'tention perpe'tuelle, et ses complices furent bru^le'es vives. Eighty young peasant girls, not 610, and no mention of plumbing... I have another book somewhere, probably by Colin Wilson with more details, but I can't remember where. All I can find at the moment is a comic book by Georges Pichard: "La Comtesse Rouge -- Erzsebet Bathory" editions Dominique Leroy, inspired, it says, from Leopold von Sacher Masoch's "Eau de Jouvence", probably a fictionized biography of Erzsebet Bathory. The number of victims is back to 610, she is discovered in flagrante delicto by one Koleman of Perusits, and Georges Thurzo is the president of the tribunal that finds her guilty. Only two interesting things, if they're not an invention of Masoch or Pichard: one character mentions that "bathing in blood keeps you in eternal youth according to popular beliefs and legends of the Carpaths" (actually, I remember having read something similar in a serious book on vampires legends), and the Red Countess is described as sacrificing one nymph per lunar month. Monday. With much trepidation and tachycardia, I open Mary D'Imperio's "Elegant Enigma".... "The manuscript was in the hands of Joannus Marcus Marci in the year 1665 or 1666.... it was in the hands of [Jacobus Horcicky de Tepenecz] at some time during the period from 1608 to 1622". So.... it could have been from Erzsebet Bathory's library. Have I really hit upon something? Alas, according to my copy of Petersen, the "blood" in the bathtubs of the "bathing nymphs" folios is green. With typical ill faith, I shall hold that the ink must have changed colour over the centuries: it's too nice a theory to throw out with the bath water! From j.guy@trl.oz.au Wed Sep 30 02:32:30 1992 Date: Tue, 29 Sep 92 12:32:30 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9209290232.AA06435@medici.trl.OZ.AU> To: voynich@rand.org Subject: my bloody theory Status: OR Perhaps I should make it quite clear that I do not believe in the "Bathory theory" of the VMS. I find it intriguing because it fits so much better, except for the colour of the blood, than, say, Levitov's (where did Levitov get this idea of blood-letting, by the way?), and shows how easy it can be to go completely off-track, but legitimately so. Here is another piece of evidence (of which I am skeptical) in favour of the Bathory theory. We have all noticed how often captions start with the letter . Sukhotin's algorithm repeatedly gives as a vowel. Now, if the Voynich was the fruit of Erzsebet Bathory's and her accomplices' researches into eternal youth, or if it was a work that inspired them, it would likely be written in Hungarian. The definite article in Hungarian is spelt "a", but is pronounced like a very open "o" (the "o" in Southern English "not"). Strange, isn't it? But isn't it convenient that my Hungarian is just about nil? I cannot verify further whether the VMS is in Hungarian. On the other hand, I can defend my theory against attacks. If Andras Kornai counters saying: "What of those words duplicated, triplicated? Hungarian doesn't do that", my answer is ready: "This is a superstitious treatise. It shows astrological diagrams for when plants should be gathered, recipes mixed, victims bled. Surely it must be full of incantations to be recited on those occasions. Take any book of magic or witchcraft: those incantations are repetitive, full of nonsensical words." Ah, if someone would only be authorized to analyze the vellum and found that the Voynich manuscript was on written on the skins of women who died around 1600, then I might start taking my own theory much more seriously! From acase@reed.edu Wed Oct 07 06:05:00 1992 Message-Id: Date: Tue, 6 Oct 92 14:05 PDT From: acase@reed.edu (Andrew Case) To: voynich@rand.org Subject: Still here, ready to transcribe Status: OR Howdy. I've been lurking since the inception of this list, held back from participation by writing a Thesis. All that is done now, so I can put in some time on transcription. Unfortunately, my bank account has a balance of $33.00 right now, and there are no jobs on the horizon that will provide enough money to afford a photocopy from Yale (and still allow me to eat). So here's the deal - if somebody sends me a chunk of photocopy that needs transcription and whatever else is needed to start work, I will put in some time on the transcription project. My complete financial destitution leaves me unable to put any money into the project, but I do have time on my hands (at least until that lucrative job offer wafts my way). I figure that this should work out cheaper than the slav labour proposal :-) Anyway, if anyone is interested, my address is Andrew Case 4139 S.E. 39th ave. Apartment #20 Portland, OR. 97202 The machines I have available for transcribing on are an IBM compatible 486 and a Mac SE. Both of these machines are idle most of the time, so I would be more than willing to set either one up to do big frequency counts (If they are tied up for days, that is fine). If somebody is able to send material for transcription, please send email so that I can start learning the MS "alphabet" before the package arrives (I have the Mahjong tiles). Thanks, and good luck all. Andrew Case acase@reed.edu From Karl.Kluge@G.GP.CS.CMU.EDU Sun Oct 18 23:25:00 1992 Date: Sun, 18 Oct 1992 23:25-EDT From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: C code Voynich program Message-Id: <719465139/kck@G.GP.CS.CMU.EDU> Status: OR As another way to avoid thesis work I decided to finally whip together a little C program to look at various first order properties of the text just to demonstrate for myself some of the basic observations people have made about the text. The program takes data from standard input so that it can be used as the tail end of a pipe. Each line is assumed to have a format like the D'Imperio transcription, i.e. a line identifier, followed by a space, followed by a line of Voynich text. Example: 15203B 4/FCCC9/ZC89/4OFCC89/4O89FC9/4OFC89/2ZCC89/PA3/ZCOE/PCC89/9R9- The program computes the relative frequencies (expressed as percentages) of all characters, line initial and final characters, and "word" initial and final characters. Line initial/finals are not considered word initial/finals. The entropies of these frequency distributions are also printed. A simple nearest-mean clustering is performed using the all-character relative frequencies as features if the "-c" option is used. The seed for the RNG and number of clusters can be specified. The clustering algorithm favors spherical clusters, but even so the A/B page separation shows up well. A hierarchical clustering algorithm would be preferable, but this was quicker to hack up. The code is not totally portable, but changes needed should be minor. libc.h is a local CMU include file that includes the include files for the standard C libraries. switches.h and the calls between SW_INIT and SW_DONE are part of a library which reads arguments off the command line. I've included the output for the D'Imperio transcription and my corpus of labels, followed by the C code. ------------------------------------------------------------------------- % s4test -foo unrecognized switch: -foo usage: s4test valid switches are: -c cluster the pages -n (2..20) [10] number of clusters -s (0..10000000) [1963] RNG seed % s4test #include #include #include #define MAX_CLUST 20 #define MAX_CHAR 64 #define min(x,y) ((x) < (y) ? (x) : (y)) #define max(x,y) ((x) > (y) ? (x) : (y)) #define absval(x) (max((x), (-(x)))) /* If pred, vt, else vf */ #define sel(pred, vt, vf) ((pred) ? (vt) : (vf)) /* Test for within delta of zero */ #define ZDELT 0.0000001 #define is_damn_near_zero(x) ((absval(x)) < ZDELT) #define sqr(x) ((x) * (x)) #define chr_pos(alphabet, ch) \ sel((index((alphabet), (ch)) == NULL), -1, \ ((int) (index((alphabet), (ch)) - &(alphabet[0])))) static int num_char; /* Random number generation: roll num sides-sided die plus dm */ static int roll(num, sides, dm) int num, sides, dm; { int die, result; result = dm; for (die = 1; die <= num; ++die) result += 1 + (1.0 * sides * random())/(1.0 * 017777777777); return(result); } /* Go through a line of text, incrementing the character count and appropriate character frequency buckets. */ int scan_line(line, alphabet, freq, count) char *line, *alphabet; float *freq; int *count; { int ch, pos; for (ch = 0; ch < strlen(line); ch++) { if (index(alphabet, line[ch]) != NULL) { pos = (int) (index(alphabet, line[ch]) - &(alphabet[0])); freq[pos] += 1.0; *count += 1; } } } static int global = 0, word_init = 0, word_final = 0, line_init = 0, line_final = 0; static float wrld[MAX_CHAR], wd_init[MAX_CHAR], wd_final[MAX_CHAR], ln_init[MAX_CHAR], ln_final[MAX_CHAR]; static char punc[10] = " /-#"; static float wd_entropy, li_entropy, lf_entropy, wi_entropy, wf_entropy; int new_scan_line(line, alphabet) char *line, *alphabet; { int ch, pos, ch2; char last; for (ch = 0; ch < strlen(line); ch++) { if (ch == 0) last = punc[0]; pos = chr_pos(alphabet, line[ch]); /* Update global character count */ if (pos != -1) { wrld[pos] += 1.0; global += 1; } /* If this character is a space, update the word-final count using the last character */ if (chr_pos(punc, line[ch]) == 1) { pos = chr_pos(alphabet, last); if (pos != -1) { wd_final[pos] += 1.0; word_final += 1; } } /* If the last character was a space, update the word-initial count using the current character */ if (chr_pos(punc, last) == 1) { pos = chr_pos(alphabet, line[ch]); if (pos != -1) { wd_init[pos] += 1.0; word_init += 1; } } /* If the current character is a "-" or "#" or the last character on the line, update the line-final count using the last character */ if ((chr_pos(punc, line[ch]) > 1) || (ch == (strlen(line) - 1))) { pos = chr_pos(alphabet, last); if (pos != -1) { ln_final[pos] += 1.0; line_final += 1; } } /* If the last character is a space, update the line-initial count using the current character */ if (chr_pos(punc, last) == 0) { pos = chr_pos(alphabet, line[ch]); if (pos != -1) { ln_init[pos] += 1.0; line_init += 1; } } last = line[ch]; } } int clear_freqs() { int ind; for (ind = 0; ind < num_char; ind++) wrld[ind] = wd_init[ind] = wd_final[ind] = ln_init[ind] = ln_final[ind] = 0.0; wd_entropy = li_entropy = lf_entropy = wi_entropy = wf_entropy = 0.0; } /* Entropy formula: h = -sigma(p[i] log2(p[i])) */ int norm_freqs() { int pos; if (global != 0) { for (pos = 0; pos < num_char; pos++) { wrld[pos] /= (float) global; if (!(is_damn_near_zero(wrld[pos]))) wd_entropy -= (wrld[pos] * log((double) wrld[pos]) / log((double) 2.0)); } } if (word_init != 0) { for (pos = 0; pos < num_char; pos++) { wd_init[pos] /= (float) word_init; if (!(is_damn_near_zero(wd_init[pos]))) wi_entropy -= (wd_init[pos] * log((double) wd_init[pos]) / log((double) 2.0)); } } if (line_init != 0) { for (pos = 0; pos < num_char; pos++) { ln_init[pos] /= (float) line_init; if (!(is_damn_near_zero(ln_init[pos]))) li_entropy -= (ln_init[pos] * log((double) ln_init[pos]) / log((double) 2.0)); } } if (word_final != 0) { for (pos = 0; pos < num_char; pos++) { wd_final[pos] /= (float) word_final; if (!(is_damn_near_zero(wd_final[pos]))) wf_entropy -= (wd_final[pos] * log((double) wd_final[pos]) / log((double) 2.0)); } } if (line_final != 0) { for (pos = 0; pos < num_char; pos++) { ln_final[pos] /= (float) line_final; if (!(is_damn_near_zero(ln_final[pos]))) lf_entropy -= (ln_final[pos] * log((double) ln_final[pos]) / log((double) 2.0)); } } } int print_freqs(alphabet) char *alphabet; { int pos; printf(" Line Word\n"); printf("Letter Global Initial Final Initial Final\n"); for (pos = 0; pos < num_char; pos++) printf("%c %6.2f %6.2f %6.2f %6.2f %6.2f\n", alphabet[pos], 100.0 * wrld[pos], 100.0 * ln_init[pos], 100.0 * ln_final[pos], 100.0 * wd_init[pos], 100.0 * wd_final[pos]); printf("Sample %8d %8d %8d %8d %8d\n", global, line_init, line_final, word_init, word_final); printf("Entropy %6.3f %6.3f %6.3f %6.3f %6.3f\n", wd_entropy, li_entropy, lf_entropy, wi_entropy, wf_entropy); } static char currier[50] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"; static char space[2] = " "; static char lang[3] = "AB"; main(argc, argv) int argc; char *argv[]; { char line_buf[200], *mss_name, chr, *postfix, hand[200]; FILE *mss; float folio_freqs[MAX_CHAR], page_freqs[200][MAX_CHAR], clust_means[MAX_CLUST][MAX_CHAR]; float this_dist, min_dist; int folio_char_count, cur_folio, this_folio, ch, converge, best_clust, clust_npage[MAX_CLUST], page, clust_anum[MAX_CLUST], clust_bnum[MAX_CLUST], just_linefed, num_clust, do_clust, seed, page_list[200], pnum, pg_lab[200], pg_clust[200], clust, num_pg; SW_INIT(argc, argv); num_clust = swbdInt("-n", "number of clusters", min(10, MAX_CLUST), 2, MAX_CLUST); seed = swbdInt("-s", "RNG seed", 1963, 0, 10000000); do_clust = swBool("-c", "cluster the pages"); SW_DONE(argc, argv, ""); /* Zero the global frequency arrays */ num_char = strlen(currier); clear_freqs(); srandom(seed); /* Calculate the normalized relative frequencies and assign random initial cluster labels */ cur_folio = -1; for (ch = 0; ch < num_char; ch++) folio_freqs[ch] = 0; folio_char_count = 0; pnum = -1; for (;(gets(line_buf) != NULL);) { sscanf(line_buf, "%3d", &this_folio); postfix = strchr(&(line_buf[0]), space[0]); new_scan_line(postfix, currier); if (this_folio == cur_folio) { postfix = strchr(&(line_buf[0]), space[0]); scan_line(postfix, currier, folio_freqs, &folio_char_count); } else { if (cur_folio >= 0) { for (ch = 0; ch < num_char; ch++) page_freqs[pnum][ch] = folio_freqs[ch] / (float) folio_char_count; page_list[pnum] = cur_folio; pg_lab[pnum] = line_buf[5]; pg_clust[pnum] = roll(1, num_clust, -1); } cur_folio = this_folio; pnum++; folio_char_count = 0; for (ch = 0; ch < num_char; ch++) folio_freqs[ch] = 0; postfix = strchr(&(line_buf[0]), space[0]); scan_line(postfix, currier, folio_freqs, &folio_char_count); hand[pnum] = line_buf[5]; /* printf("Starting page %3d: %s\n", cur_folio, line_buf);*/ } } num_pg = pnum; norm_freqs(); /* Print the relative frequency information */ print_freqs(currier); if (!(do_clust)) exit(1); /* Print the relative frequencies for all pages */ /* for (page = 0; page < num_pg; page++) { printf("Relative frequencies page %3d (initial cluster %3d):\n", page_list[page], pg_clust[page]); for (ch = 0; ch < num_char; ch++) { printf("%c %5.2f ", currier[ch], 100.0 * page_freqs[page][ch]); just_linefed = 0; if ((ch % 9) == 8) {printf("\n"); just_linefed = 1;} } if (!(just_linefed)) printf("\n"); } */ /* Now compute the cluster means and reclassify until convergance */ converge = 0; for (; !(converge); ){ /* Zero the cluster means array */ for (clust = 0; clust < num_clust; clust++) { clust_npage[clust] = 0; for (ch = 0; ch < num_char; ch++) clust_means[clust][ch] = 0; } /* Compute the cluster means */ for (page = 0; page < num_pg; page++) { clust_npage[pg_clust[page]]++; for (ch = 0; ch < num_char; ch++) clust_means[pg_clust[page]][ch] += page_freqs[page][ch]; } for (clust = 0; clust < num_clust; clust++) for (ch = 0; ch < num_char; ch++) clust_means[clust][ch] /= (float) clust_npage[clust]; /* Reclassify the pages based on the new cluster means */ converge = 1; for (page = 0; page < num_pg; page++) { best_clust = 0; min_dist = 0; for (ch = 0; ch < num_char; ch++) min_dist += sqr(clust_means[0][ch] - page_freqs[page][ch]); min_dist = sqrt((double) min_dist); for (clust = 1; clust < num_clust; clust++) { this_dist = 0; for (ch = 0; ch < num_char; ch++) this_dist += sqr(clust_means[clust][ch] - page_freqs[page][ch]); this_dist = sqrt((double) this_dist); if (this_dist < min_dist) { min_dist = this_dist; best_clust = clust; } } if (best_clust != pg_clust[page]) converge = 0; pg_clust[page] = best_clust; } } for (clust = 0; clust < num_clust; clust++) clust_anum[clust] = clust_bnum[clust] = 0; for (page = 0; page < num_pg; page++) if (index(lang, hand[page]) == &(lang[0])) clust_anum[pg_clust[page]]++; else clust_bnum[pg_clust[page]]++; /* Print the mean relative frequencies for each cluster and the D'Imperio page number of the folios in that cluster */ for (clust = 0; clust < num_clust; clust++) { printf("Cluster %3d: %3d A pages, %3d B pages. Mean:\n", clust, clust_anum[clust], clust_bnum[clust]); for (ch = 0; ch < num_char; ch++) { printf("%c %5.2f ", currier[ch], 100.0 * clust_means[clust][ch]); just_linefed = 0; if ((ch % 9) == 8) {printf("\n"); just_linefed = 1;} } if (!(just_linefed)) printf("\n"); printf("*****PAGES:\n"); pnum = -1; for (page = 0; page < num_pg; page++) { if (pg_clust[page] == clust) { pnum++; printf("%3d ", page_list[page]); just_linefed = 0; if ((pnum % 18) == 17) {printf("\n"); just_linefed = 1;} } } if (!(just_linefed)) printf("\n"); printf("-------------------------------------------------\n"); } } From reeds@gauss.att.com Mon Oct 19 22:25:26 1992 Message-Id: <199210191325.AA06002@rand.org> From: reeds@gauss.att.com Date: Mon, 19 Oct 92 09:25:26 EDT To: voynich@rand.org Subject: Voynich substrings Status: OR Inspired by Karl's example, I decided to post a useful program of mine. It prints out all repeated substrings. For instance, the input text 'Mon Oct 19 09:04:50 EDT 1992' generates the output text 2 ' 19' 3 5 ' ' 1 3 '0' 1 4 '9' 1 2 ':' 1 Meaning: there were 2 instances of ' 19', a substring of length 3, five instances of ' ', a substring of length 1, and so on. I often invoke this program as follows: subs < inputtext | sort -rn +2 -3 +0 -1 which shows longer repeated substrings first, in decreasing order of frequency. Jacques Guy discussed this algorithm some time ago. Notes. 1. Subs uses the C library's qsort routine, which can be slow. 2. I built in a 100,000 max character limit, which is fine for the current state of Voynich affairs (and for most other small jobs), but not so good if you are building a Shakespeare concordance. 3. Subs operates on ALL its input, without interpreting newlines or any other characters specially. So before applying it to Voynich text a preprocessing must occur, to strip out all undesired formatting characters. This might mean (if applied to the D'Imperio format file) taking out the line numbers and converting newlines to - characters, say. --- cut here --- #include extern char *calloc(); extern void exit(); char text[100000]; int length; int cmpf(); main() { int c; char *tbase, *p; int i, cmpf(); char **base; tbase = text+1; p = tbase; while((c=getchar())!=EOF){ *p++ = c; if(p-text >= sizeof(text)){ fprintf(stderr, "too long\n"); exit(1); } } *p = 0; length = p-tbase; base = (char **)calloc(length, sizeof(char *)); for(i=0; i Status: OR I added digraph stats in the program, and ran on the "A" corpus, "B" corpus, entire D'Imperio, and my label corpus. The fifth column of the digraph stats, wf/wi, are the pairs before/after spaces in the mss. The digraph stats again give relative frequency (as a %age) for the entire text, line-initial/final, and word-initial/final cases. For your enjoyment... Script started on Mon Oct 19 23:11:38 1992 % source voyn.script THESE ARE THE RESULTS FOR THE A CORPUS Line Word Letter Global Initial Final Initial Final 0 0.02 0.00 0.00 0.02 0.00 1 0.00 0.00 0.00 0.00 0.00 2 1.82 5.22 4.08 3.44 3.43 3 0.13 0.18 0.64 0.09 0.24 4 2.39 15.65 0.08 7.75 0.02 5 0.00 0.00 0.00 0.00 0.00 6 0.13 0.00 2.24 0.22 0.05 7 0.03 0.00 0.24 0.00 0.02 8 7.61 13.36 5.28 17.81 1.39 9 10.89 12.53 34.91 4.90 32.32 A 7.76 0.09 0.32 1.00 0.07 B 0.76 5.31 0.00 0.50 0.12 C 5.67 0.27 0.16 0.45 0.33 D 0.29 0.00 2.56 0.07 0.67 E 5.76 0.73 8.97 0.53 18.36 F 4.90 3.93 0.00 4.38 0.26 G 0.02 0.00 0.00 0.00 0.05 H 0.02 0.00 0.08 0.00 0.02 I 0.18 0.00 0.08 0.00 0.03 J 0.75 0.00 8.33 0.03 1.51 K 0.04 0.00 0.24 0.00 0.12 L 0.02 0.00 0.24 0.00 0.02 M 3.82 0.00 17.45 0.00 13.44 N 0.73 0.00 3.60 0.00 2.47 O 19.49 20.49 1.36 13.89 5.32 P 4.41 6.77 0.24 2.48 0.34 Q 1.63 0.46 0.24 5.42 0.07 R 4.99 0.27 7.69 0.53 17.90 S 10.79 6.31 0.08 23.87 0.07 T 0.21 0.00 0.48 0.00 0.69 U 0.08 0.00 0.24 0.00 0.29 V 0.23 1.37 0.00 0.28 0.07 W 0.34 0.09 0.00 1.12 0.03 X 0.62 0.00 0.08 1.50 0.03 Y 0.08 0.00 0.00 0.24 0.00 Z 3.42 6.95 0.08 9.49 0.26 Sample 26433 1093 1249 5816 5832 Entropy 3.868 3.318 3.093 3.348 2.836 ------------------------------------------------------ Digraphs whose max frequency global, line initial, etc. > 1.000000%: line word global initial final initial final wf/wi 29 0.2125 0.0932 1.4468 0.3582 0.3721 0.3456 2O 0.3733 2.6095 0.0000 0.6447 0.0354 0.7085 2S 0.1711 1.2116 0.0000 0.1970 0.0000 1.0368 4O 3.1985 15.5638 0.0000 7.9334 0.1063 0.0000 89 2.2810 0.4660 13.3617 2.1848 4.6775 0.1037 8A 5.1374 5.5918 0.1702 11.4792 0.0177 0.0346 8O 0.9020 1.7707 0.1702 2.2385 0.2126 0.2074 8S 0.7050 4.1007 0.0000 1.1640 0.0000 0.3283 8Z 0.2125 1.4911 0.0000 0.3044 0.0177 0.2246 92 0.0156 0.0000 0.0851 0.0000 0.0354 1.3306 94 0.0000 0.0000 0.0000 0.0000 0.0000 3.9744 98 0.3940 1.1184 0.7660 0.3044 0.1595 6.0308 99 0.0000 0.0000 0.0000 0.0000 0.0000 1.2269 9F 0.9072 2.5163 0.0000 1.7013 0.0177 2.8858 9O 0.0570 0.8388 0.0000 0.0358 0.0000 3.6461 9P 0.8657 2.2367 0.0851 1.8266 0.0000 1.6243 9Q 0.0104 0.0932 0.0000 0.0000 0.0000 1.3824 9S 0.3421 4.5666 0.0000 0.1433 0.0000 6.1863 9Z 0.0622 0.6524 0.0000 0.0179 0.0177 2.3501 AD 0.3110 0.0000 2.3830 0.0358 0.5670 0.0000 AE 1.2701 0.0000 4.6809 0.1791 2.4096 0.0000 AJ 0.6065 0.0000 5.5319 0.1254 0.9036 0.0000 AM 4.7849 0.0932 17.7872 0.3761 12.5266 0.0000 AN 0.9590 0.0000 3.6596 0.0000 2.4451 0.0000 AR 1.7470 0.0000 4.5106 0.2686 4.2700 0.0173 BO 0.1503 1.3979 0.0000 0.0716 0.0000 0.0173 BS 0.6117 2.4231 0.0000 0.2686 0.0000 0.0691 C9 2.5143 0.0000 2.5532 0.0000 7.9731 0.0346 CC 1.9025 0.0000 0.0000 0.1612 0.1063 0.0000 CO 1.8818 0.2796 0.0000 0.0000 0.6024 0.0518 E8 0.5184 0.0000 0.7660 0.0358 0.0532 4.9421 E9 0.3266 0.0000 1.6170 0.0179 0.7264 0.6048 EO 0.3992 0.2796 0.2553 0.1791 0.1063 2.5229 ES 0.1140 0.1864 0.0000 0.0716 0.0000 4.8039 EZ 0.0570 0.0932 0.0000 0.0537 0.0000 1.6416 F9 0.8191 0.2796 2.0426 0.2149 2.0907 0.0173 FA 1.0368 0.0000 0.1702 0.4119 0.0000 0.0000 FC 1.0057 0.0000 0.0000 0.7880 0.0000 0.0000 FO 1.2701 1.3979 0.0851 0.9850 0.1595 0.0691 FS 2.0840 1.4911 0.0000 1.6476 0.0177 0.0864 M4 0.0000 0.0000 0.0000 0.0000 0.0000 1.3478 M8 0.0104 0.0000 0.0851 0.0000 0.0000 2.1082 MO 0.0156 0.0000 0.0851 0.0000 0.0000 1.7799 MQ 0.0000 0.0000 0.0000 0.0000 0.0000 1.4688 MS 0.0052 0.0000 0.0000 0.0000 0.0000 2.8339 MZ 0.0000 0.0000 0.0000 0.0000 0.0000 1.5552 O8 2.7112 0.9320 1.8723 0.8238 0.9391 0.8122 OA 0.3577 1.1184 0.0000 0.1970 0.0000 0.0691 OE 6.2105 1.4911 4.3404 1.7371 16.0702 0.2246 OF 3.6288 5.4986 0.0000 4.1010 0.1949 0.5011 OJ 0.3888 0.0000 3.1489 0.1254 0.6378 0.0000 OM 0.4251 0.0000 0.5957 0.1074 1.3288 0.0000 OP 3.6081 5.1258 0.0851 4.5666 0.2835 0.2938 OR 4.8315 1.4911 3.4043 0.8954 13.7314 0.2938 OS 0.3577 2.0503 0.0000 0.1254 0.0177 1.0195 P9 0.8554 0.0932 2.7234 0.2507 2.0730 0.0000 PO 1.2908 2.7959 0.0851 0.4298 0.1417 0.0518 PS 2.2447 2.3299 0.0851 1.1640 0.0354 0.1901 PZ 0.3370 1.3979 0.0000 0.1970 0.0532 0.0518 Q9 0.8087 0.0000 3.2340 1.6297 2.0376 0.0000 QO 0.8813 0.2796 0.0851 2.5788 0.1772 0.0000 R8 0.0518 0.0000 0.0000 0.0000 0.0000 2.2119 R9 0.2851 0.0000 2.8085 0.0537 0.3366 1.1750 RO 0.2281 0.1864 0.1702 0.0716 0.0532 3.1968 RQ 0.0000 0.0000 0.0000 0.0000 0.0000 1.0368 RS 0.1089 0.0000 0.0000 0.0537 0.0000 5.5296 RZ 0.0207 0.0000 0.0000 0.0000 0.0000 2.0736 S9 2.7164 0.1864 2.7234 2.0953 7.9908 0.0000 SA 1.1975 0.1864 0.0000 2.0415 0.0177 0.0000 SC 2.4469 0.0000 0.0000 4.2622 0.0177 0.0000 SO 7.2006 5.6850 0.1702 13.2521 1.5060 0.0346 X9 0.3473 0.0000 1.3617 0.4298 0.8859 0.0000 Z9 0.6636 0.0932 1.1064 1.0208 1.8604 0.0000 ZC 0.9279 0.0932 0.0000 2.2027 0.1772 0.0000 ZO 2.5661 6.3374 0.0851 5.4441 2.0376 0.1037 # 19290 1073 1175 5584 5644 5787 ------------------------------------------------------ Selected digraph frequencies (characters 89SOERANMP): (Row = 1st char, col = second) 8 9 S O E R A N M P 8 0.047 2.281 0.705 0.902 0.078 0.005 5.137 0.005 0.000 0.010 9 0.394 0.000 0.342 0.057 0.016 0.000 0.031 0.000 0.000 0.866 S 0.124 2.716 0.031 7.201 0.047 0.031 1.198 0.000 0.000 0.161 O 2.711 0.332 0.358 0.104 6.210 4.832 0.358 0.010 0.425 3.608 E 0.518 0.327 0.114 0.399 0.000 0.016 0.062 0.000 0.000 0.026 R 0.052 0.285 0.109 0.228 0.010 0.010 0.181 0.000 0.016 0.000 A 0.036 0.021 0.021 0.005 1.270 1.747 0.000 0.959 4.785 0.010 N 0.005 0.010 0.000 0.005 0.000 0.000 0.000 0.000 0.000 0.000 M 0.010 0.005 0.005 0.016 0.000 0.000 0.000 0.000 0.000 0.000 P 0.026 0.855 2.245 1.291 0.005 0.000 0.752 0.000 0.000 0.000 ------------------------------------------------------ Cluster 0: 20 A pages, 0 B pages. Mean: 0 0.02 1 0.00 2 1.83 3 0.10 4 2.11 5 0.00 6 0.09 7 0.04 8 7.07 9 11.39 A 5.78 B 0.58 C 9.12 D 0.25 E 4.50 F 4.77 G 0.03 H 0.01 I 0.17 J 0.15 K 0.01 L 0.00 M 3.40 N 0.60 O 19.01 P 4.37 Q 1.24 R 3.94 S 14.27 T 0.14 U 0.02 V 0.26 W 0.24 X 0.26 Y 0.02 Z 4.22 *****PAGES: 4 8 9 13 14 15 37 39 40 51 52 55 57 58 68 82 92 95 96 109 ------------------------------------------------- Cluster 1: 35 A pages, 0 B pages. Mean: 0 0.01 1 0.00 2 2.06 3 0.17 4 2.38 5 0.00 6 0.19 7 0.04 8 7.45 9 8.62 A 9.76 B 0.68 C 5.24 D 0.37 E 6.71 F 4.41 G 0.01 H 0.02 I 0.27 J 1.04 K 0.04 L 0.04 M 4.48 N 0.99 O 19.25 P 3.91 Q 1.83 R 5.59 S 9.31 T 0.25 U 0.17 V 0.22 W 0.38 X 0.87 Y 0.08 Z 3.15 *****PAGES: 1 3 5 6 7 10 11 12 16 27 28 32 33 43 44 45 46 47 48 54 61 62 67 69 72 74 81 86 87 91 99 100 102 105 106 ------------------------------------------------- Cluster 2: 30 A pages, 0 B pages. Mean: 0 0.02 1 0.00 2 1.48 3 0.13 4 2.84 5 0.00 6 0.09 7 0.00 8 8.74 9 14.00 A 6.51 B 0.92 C 2.77 D 0.22 E 4.90 F 5.49 G 0.02 H 0.00 I 0.07 J 0.68 K 0.05 L 0.00 M 3.65 N 0.52 O 20.00 P 5.42 Q 1.70 R 4.97 S 10.12 T 0.24 U 0.03 V 0.24 W 0.35 X 0.52 Y 0.10 Z 3.22 *****PAGES: 2 17 18 19 20 21 22 23 24 25 26 29 30 31 34 35 36 38 41 42 53 56 70 71 73 85 88 101 103 104 ------------------------------------------------- THESE ARE THE RESULTS FOR THE B CORPUS Line Word Letter Global Initial Final Initial Final 0 0.00 0.00 0.00 0.00 0.00 1 0.00 0.00 0.00 0.00 0.00 2 1.30 14.74 2.41 2.40 1.63 3 0.12 0.00 0.43 0.00 0.43 4 4.81 20.98 0.00 21.07 0.06 5 0.00 0.00 0.09 0.00 0.01 6 0.05 0.00 1.20 0.04 0.04 7 0.01 0.00 0.09 0.00 0.01 8 9.94 14.18 2.32 8.09 1.19 9 13.22 10.78 49.23 2.81 53.05 A 7.63 0.28 0.26 3.78 0.08 B 0.75 9.74 0.00 0.45 0.09 C 14.16 0.28 0.17 0.39 0.19 D 0.05 0.00 0.52 0.03 0.14 E 7.07 3.59 14.18 4.65 14.67 F 6.96 1.04 0.17 2.47 0.34 G 0.02 0.00 0.09 0.01 0.08 H 0.01 0.00 0.09 0.00 0.01 I 0.02 0.00 0.00 0.00 0.00 J 0.38 0.00 10.22 0.04 0.36 K 0.01 0.00 0.17 0.00 0.03 L 0.01 0.00 0.09 0.00 0.01 M 1.76 0.00 5.07 0.00 7.98 N 1.38 0.00 3.61 0.00 6.24 O 13.15 7.84 1.12 23.61 0.70 P 3.06 10.11 0.00 1.57 0.11 Q 0.43 0.19 0.17 0.41 0.00 R 3.40 0.57 7.90 1.71 11.24 S 5.49 1.89 0.00 14.85 0.03 T 0.19 0.00 0.00 0.00 0.85 U 0.08 0.00 0.09 0.01 0.29 V 0.24 0.19 0.26 0.26 0.01 W 0.08 0.09 0.00 0.16 0.00 X 0.75 0.09 0.00 0.24 0.03 Y 0.03 0.00 0.00 0.06 0.00 Z 3.40 3.40 0.09 10.89 0.10 Sample 40160 1058 1164 7964 7966 Entropy 3.782 3.234 2.566 3.179 2.288 ------------------------------------------------------ Digraphs whose max frequency global, line initial, etc. > 1.000000%: line word global initial final initial final wf/wi 2A 0.5741 5.4755 0.0879 1.2934 0.0000 0.5281 2O 0.3903 7.5889 0.0000 0.4565 0.0127 0.4275 4O 6.0152 20.2690 0.0879 20.5935 0.1140 0.0000 89 9.1405 0.1921 23.5501 1.3568 31.5056 0.1257 8A 2.4609 6.6282 0.1757 5.3386 0.0127 0.0629 8O 0.3225 1.2488 0.0000 0.7355 0.0253 0.3269 8S 0.1806 2.6897 0.0000 0.1395 0.0000 0.0754 8Z 0.1580 3.1700 0.0000 0.1395 0.0000 0.1006 92 0.0290 0.0961 0.1757 0.0127 0.0253 1.6472 94 0.0323 0.0000 0.0000 0.0254 0.0000 17.8675 98 0.1000 0.1921 0.1757 0.1268 0.0000 5.1301 99 0.0000 0.0000 0.0000 0.0000 0.0000 1.2700 9E 0.0742 0.0000 0.5272 0.0507 0.0253 3.8853 9F 0.5773 2.5937 0.0000 1.2300 0.0253 1.4208 9O 0.0290 0.0000 0.0000 0.0507 0.0000 9.8202 9P 0.4032 2.8818 0.0000 0.7482 0.0000 1.0436 9R 0.0452 0.0000 0.1757 0.0254 0.0380 1.3957 9S 0.1064 1.8252 0.0000 0.1014 0.0000 5.0421 9Z 0.1097 2.3055 0.0000 0.0634 0.0127 2.9674 AE 2.3835 0.1921 5.9754 0.7989 5.6477 0.0000 AJ 0.4386 0.0000 9.3146 0.1775 0.3546 0.0000 AM 2.2061 0.0000 4.9209 0.9511 7.8004 0.0000 AN 1.7642 0.0000 3.6907 0.3424 6.1922 0.0000 AR 2.3480 0.0961 3.7786 1.0525 7.2306 0.0000 BO 0.1871 4.0346 0.0000 0.0380 0.0000 0.0251 BS 0.4870 3.3622 0.0000 0.3170 0.0000 0.0251 BZ 0.1000 1.1527 0.0000 0.0634 0.0000 0.0251 C8 7.6504 0.0000 0.6151 0.0127 0.7978 0.0377 C9 3.3575 0.0000 5.1845 0.0000 12.0805 0.0251 CC 3.9671 0.0961 0.0000 0.2156 0.0886 0.0000 CO 1.2288 0.0000 0.0879 0.0000 0.2533 0.0126 E4 0.0065 0.0000 0.0000 0.0000 0.0000 1.4208 E8 0.4225 0.0000 0.4394 0.2790 0.0127 1.4083 E9 0.4515 0.0000 6.5905 0.1141 0.7345 0.2515 EF 1.0127 0.1921 0.0879 0.6721 0.0000 0.7419 EO 0.6225 0.9606 0.5272 0.8496 0.0760 3.3949 ES 1.0772 1.5370 0.0000 1.4836 0.0127 3.1057 EZ 0.6128 0.6724 0.0000 0.6974 0.0127 2.3136 F9 0.7999 0.0000 3.9543 0.0127 2.4440 0.0377 FA 3.2640 0.0000 0.0000 0.8116 0.0000 0.0251 FC 3.8284 0.1921 0.0000 0.9891 0.0127 0.0251 MO 0.0129 0.0000 0.0000 0.0000 0.0127 2.8543 MS 0.0065 0.0000 0.0000 0.0000 0.0000 1.9490 MZ 0.0000 0.0000 0.0000 0.0000 0.0000 1.4837 NO 0.0226 0.0000 0.0000 0.0000 0.0127 2.1879 NS 0.0032 0.0000 0.0000 0.0000 0.0000 1.5843 NZ 0.0000 0.0000 0.0000 0.0000 0.0000 1.3328 OE 5.1379 2.6897 6.7663 8.1664 8.6489 0.1006 OF 5.8958 1.4409 0.0000 6.5052 0.1646 0.0754 OP 2.3577 1.3449 0.0000 4.7806 0.0507 0.0251 OR 1.4514 0.5764 3.3392 2.1177 3.6596 0.0377 P9 0.3774 0.0000 1.3181 0.0888 1.1523 0.0251 PA 1.1289 1.5370 0.0000 0.4185 0.0000 0.0000 PC 1.5578 0.8646 0.0000 0.6214 0.0000 0.0000 PO 0.3354 3.4582 0.0000 0.2029 0.0000 0.0377 PS 0.3645 3.0740 0.0000 0.1902 0.0000 0.0251 PZ 0.1193 1.2488 0.0000 0.0380 0.0127 0.0251 Q9 0.2967 0.0000 0.7909 0.0380 1.0384 0.0000 R9 0.1774 0.0961 3.2513 0.0380 0.2026 0.4024 RA 0.4580 0.0000 0.0000 0.5960 0.0000 1.9867 RO 0.3064 0.2882 0.1757 0.3804 0.0127 3.5207 RS 0.1290 0.1921 0.0000 0.2790 0.0127 2.1376 RZ 0.0774 0.0000 0.0000 0.1902 0.0000 1.9364 S8 0.8224 0.0961 0.0000 1.1032 0.1140 0.0000 S9 0.3483 0.0000 1.6696 0.1522 1.0510 0.0000 SC 4.6186 1.1527 0.0000 9.8783 0.0127 0.0000 SX 0.3451 0.0961 0.0000 1.2427 0.0127 0.0000 X9 0.5935 0.0000 1.4060 0.0888 2.1021 0.0000 ZC 3.3414 2.3055 0.0000 8.3693 0.0633 0.0000 # 31005 1041 1138 7886 7897 7953 ------------------------------------------------------ Selected digraph frequencies (characters 89SOERANMP): (Row = 1st char, col = second) 8 9 S O E R A N M P 8 0.010 9.140 0.181 0.323 0.048 0.010 2.461 0.000 0.000 0.003 9 0.100 0.000 0.106 0.029 0.074 0.045 0.019 0.000 0.000 0.403 S 0.822 0.348 0.006 0.374 0.048 0.016 0.116 0.000 0.000 0.052 O 0.719 0.071 0.048 0.026 5.138 1.451 0.061 0.023 0.045 2.358 E 0.423 0.452 1.077 0.622 0.026 0.042 0.258 0.000 0.000 0.139 R 0.026 0.177 0.129 0.306 0.010 0.000 0.458 0.003 0.003 0.003 A 0.026 0.019 0.000 0.019 2.383 2.348 0.003 1.764 2.206 0.006 N 0.010 0.010 0.003 0.023 0.000 0.000 0.006 0.000 0.000 0.000 M 0.003 0.016 0.006 0.013 0.000 0.000 0.003 0.000 0.000 0.000 P 0.010 0.377 0.364 0.335 0.010 0.000 1.129 0.000 0.000 0.000 ------------------------------------------------------ Cluster 0: 0 A pages, 25 B pages. Mean: 0 0.01 1 0.00 2 1.61 3 0.11 4 4.82 5 0.01 6 0.04 7 0.03 8 10.42 9 13.43 A 6.58 B 0.90 C 16.41 D 0.04 E 6.55 F 6.68 G 0.02 H 0.01 I 0.04 J 0.30 K 0.00 L 0.01 M 1.57 N 1.08 O 13.05 P 2.91 Q 0.47 R 2.79 S 5.52 T 0.18 U 0.11 V 0.22 W 0.08 X 0.69 Y 0.05 Z 3.28 *****PAGES: 49 50 59 60 79 80 93 111 147 148 149 150 151 152 153 154 155 156 159 160 161 162 163 164 165 ------------------------------------------------- Cluster 1: 0 A pages, 10 B pages. Mean: 0 0.02 1 0.00 2 0.96 3 0.27 4 3.30 5 0.00 6 0.08 7 0.02 8 8.17 9 10.83 A 12.57 B 0.76 C 8.40 D 0.08 E 6.02 F 9.37 G 0.01 H 0.00 I 0.06 J 0.72 K 0.05 L 0.03 M 3.63 N 1.56 O 14.41 P 2.69 Q 0.24 R 5.83 S 5.94 T 0.46 U 0.10 V 0.58 W 0.06 X 0.78 Y 0.05 Z 1.97 *****PAGES: 63 76 77 78 97 98 107 108 157 158 ------------------------------------------------- Cluster 2: 0 A pages, 9 B pages. Mean: 0 0.00 1 0.00 2 1.06 3 0.14 4 1.70 5 0.00 6 0.06 7 0.02 8 12.69 9 15.42 A 9.57 B 0.80 C 10.94 D 0.00 E 4.78 F 5.67 G 0.00 H 0.00 I 0.02 J 1.19 K 0.07 L 0.00 M 2.11 N 0.53 O 10.75 P 4.22 Q 0.56 R 4.54 S 7.41 T 0.39 U 0.24 V 0.55 W 0.17 X 1.09 Y 0.05 Z 3.28 *****PAGES: 64 65 66 75 83 84 89 90 94 ------------------------------------------------- THESE ARE THE RESULTS FOR THE MSS AS A WHOLE Line Word Letter Global Initial Final Initial Final 0 0.01 0.00 0.00 0.01 0.00 1 0.00 0.00 0.00 0.00 0.00 2 1.51 9.90 3.27 2.84 2.39 3 0.12 0.09 0.54 0.04 0.35 4 3.85 18.27 0.04 15.45 0.04 5 0.00 0.00 0.04 0.00 0.01 6 0.08 0.00 1.74 0.12 0.04 7 0.02 0.00 0.17 0.00 0.01 8 9.01 13.76 3.85 12.19 1.28 9 12.29 11.67 41.82 3.69 44.29 A 7.68 0.19 0.29 2.61 0.07 B 0.76 7.48 0.00 0.47 0.10 C 10.79 0.28 0.17 0.41 0.25 D 0.15 0.00 1.57 0.04 0.36 E 6.55 2.14 11.48 2.91 16.23 F 6.14 2.51 0.08 3.28 0.30 G 0.02 0.00 0.04 0.01 0.07 H 0.01 0.00 0.08 0.00 0.01 I 0.08 0.00 0.04 0.00 0.01 J 0.53 0.00 9.24 0.04 0.85 K 0.02 0.00 0.21 0.00 0.07 L 0.01 0.00 0.17 0.00 0.01 M 2.58 0.00 11.48 0.00 10.29 N 1.12 0.00 3.61 0.00 4.65 O 15.67 14.27 1.24 19.51 2.65 P 3.59 8.41 0.12 1.95 0.21 Q 0.91 0.33 0.21 2.53 0.03 R 4.03 0.42 7.79 1.21 14.05 S 7.60 4.14 0.04 18.66 0.04 T 0.20 0.00 0.25 0.00 0.78 U 0.08 0.00 0.17 0.01 0.29 V 0.23 0.79 0.12 0.27 0.04 W 0.18 0.09 0.00 0.57 0.01 X 0.70 0.05 0.04 0.77 0.03 Y 0.05 0.00 0.00 0.14 0.00 Z 3.41 5.21 0.08 10.30 0.17 Sample 66593 2151 2413 13780 13798 Entropy 3.857 3.359 2.900 3.356 2.579 ------------------------------------------------------ Digraphs whose max frequency global, line initial, etc. > 1.000000%: line word global initial final initial final wf/wi 29 0.1094 0.0946 1.0376 0.1633 0.2068 0.2111 2A 0.4812 3.1693 0.0432 1.0319 0.0074 0.3785 2O 0.3837 5.0615 0.0000 0.5345 0.0222 0.5459 4O 4.9349 17.8808 0.0432 15.3452 0.1108 0.0000 89 6.5096 0.3311 18.3744 1.7001 20.3235 0.1164 8A 3.4874 6.1022 0.1729 7.8842 0.0148 0.0509 8O 0.5448 1.5137 0.0865 1.3586 0.1034 0.2766 8S 0.3817 3.4059 0.0000 0.5642 0.0000 0.1820 8Z 0.1789 2.3179 0.0000 0.2079 0.0074 0.1528 92 0.0239 0.0473 0.1297 0.0074 0.0295 1.5138 94 0.0199 0.0000 0.0000 0.0148 0.0000 12.0160 98 0.2127 0.6623 0.4756 0.2004 0.0665 5.5095 99 0.0000 0.0000 0.0000 0.0000 0.0000 1.2518 9E 0.0517 0.0000 0.3026 0.0371 0.0222 2.3071 9F 0.7038 2.5544 0.0000 1.4254 0.0222 2.0378 9O 0.0398 0.4257 0.0000 0.0445 0.0000 7.2198 9P 0.5806 2.5544 0.0432 1.1952 0.0000 1.2882 9S 0.1968 3.2167 0.0000 0.1188 0.0000 5.5240 9Z 0.0915 1.4664 0.0000 0.0445 0.0148 2.7074 AD 0.1511 0.0000 1.4700 0.0148 0.3102 0.0000 AE 1.9565 0.0946 5.3178 0.5419 4.2981 0.0000 AJ 0.5030 0.0000 7.3930 0.1559 0.5834 0.0000 AM 3.1951 0.0473 11.4570 0.7127 9.7703 0.0000 AN 1.4554 0.0000 3.6749 0.2004 4.6304 0.0000 AR 2.1175 0.0473 4.1505 0.7275 5.9966 0.0073 BO 0.1730 2.6963 0.0000 0.0520 0.0000 0.0218 BS 0.5348 2.8855 0.0000 0.2970 0.0000 0.0437 C8 4.7301 0.0000 0.3026 0.0074 0.4726 0.0582 C9 3.0341 0.0000 3.8478 0.0000 10.3685 0.0291 CC 3.1753 0.0473 0.0000 0.1930 0.0960 0.0000 CO 1.4793 0.1419 0.0432 0.0000 0.3988 0.0291 E4 0.0060 0.0000 0.0000 0.0000 0.0000 1.1426 E8 0.4593 0.0000 0.6053 0.1782 0.0295 2.8967 E9 0.4036 0.0000 4.0640 0.0742 0.7311 0.4003 EO 0.5368 0.6149 0.3891 0.5716 0.0886 3.0277 ES 0.7078 0.8515 0.0000 0.8983 0.0074 3.8210 EZ 0.3996 0.3784 0.0000 0.4306 0.0074 2.0306 F9 0.8072 0.1419 2.9831 0.0965 2.2967 0.0291 FA 2.4098 0.0000 0.0865 0.6459 0.0000 0.0146 FC 2.7458 0.0946 0.0000 0.9057 0.0074 0.0146 FS 1.0757 0.8988 0.0000 0.8537 0.0074 0.0873 M8 0.0060 0.0000 0.0865 0.0000 0.0000 1.2227 MO 0.0139 0.0000 0.0432 0.0000 0.0074 2.4017 MS 0.0060 0.0000 0.0000 0.0000 0.0000 2.3217 MZ 0.0000 0.0000 0.0000 0.0000 0.0000 1.5138 NO 0.0159 0.0000 0.0000 0.0000 0.0074 1.3683 NS 0.0020 0.0000 0.0000 0.0000 0.0000 1.1790 O8 1.4832 0.7569 1.2105 0.6088 0.4800 0.3712 OE 5.5493 2.0814 5.5339 5.5011 11.7421 0.1528 OF 5.0263 3.5005 0.0000 5.5085 0.1772 0.2547 OJ 0.1690 0.0000 1.9888 0.0594 0.2732 0.0000 OP 2.8373 3.2640 0.0432 4.6919 0.1477 0.1383 OR 2.7478 1.0407 3.3722 1.6110 7.8576 0.1456 OS 0.1670 1.0880 0.0000 0.0891 0.0074 0.4658 P9 0.5607 0.0473 2.0320 0.1559 1.5361 0.0146 PC 1.1055 0.4730 0.0000 0.4603 0.0000 0.0000 PO 0.7019 3.1220 0.0432 0.2970 0.0591 0.0437 PS 1.0856 2.6963 0.0432 0.5939 0.0148 0.0946 PZ 0.2028 1.3245 0.0000 0.1039 0.0295 0.0364 Q9 0.4931 0.0000 2.0320 0.6978 1.4548 0.0000 QO 0.3519 0.1892 0.0432 1.0839 0.0812 0.0000 R8 0.0358 0.0000 0.0000 0.0000 0.0074 1.1499 R9 0.2187 0.0473 3.0264 0.0445 0.2585 0.7278 RA 0.3519 0.0000 0.0000 0.3712 0.0000 1.3100 RO 0.2764 0.2365 0.1729 0.2524 0.0295 3.3843 RS 0.1213 0.0946 0.0000 0.1856 0.0074 3.5662 RZ 0.0557 0.0000 0.0000 0.1114 0.0000 1.9942 S9 1.2566 0.0946 2.2049 0.9577 3.9436 0.0000 SA 0.5309 0.2365 0.0000 1.0245 0.0074 0.0000 SC 3.7857 0.5676 0.0000 7.5501 0.0148 0.0000 SO 2.9923 3.0274 0.0865 6.0134 0.6277 0.0146 X9 0.4991 0.0000 1.3835 0.2301 1.5952 0.0000 Z9 0.3400 0.0946 0.6053 0.5568 1.0487 0.0073 ZC 2.4157 1.1826 0.0000 5.8129 0.1108 0.0000 ZO 1.1333 3.5478 0.0432 2.6429 0.8640 0.0582 # 50295 2114 2313 13470 13541 13740 ------------------------------------------------------ Selected digraph frequencies (characters 89SOERANMP): (Row = 1st char, col = second) 8 9 S O E R A N M P 8 0.024 6.510 0.382 0.545 0.060 0.008 3.487 0.002 0.000 0.006 9 0.213 0.000 0.197 0.040 0.052 0.028 0.024 0.000 0.000 0.581 S 0.555 1.257 0.016 2.992 0.048 0.022 0.531 0.000 0.000 0.093 O 1.483 0.171 0.167 0.056 5.549 2.748 0.175 0.018 0.191 2.837 E 0.459 0.404 0.708 0.537 0.016 0.032 0.183 0.000 0.000 0.095 R 0.036 0.219 0.121 0.276 0.010 0.004 0.352 0.002 0.008 0.002 A 0.030 0.020 0.008 0.014 1.956 2.118 0.002 1.455 3.195 0.008 N 0.008 0.010 0.002 0.016 0.000 0.000 0.004 0.000 0.000 0.000 M 0.006 0.012 0.006 0.014 0.000 0.000 0.002 0.000 0.000 0.000 P 0.016 0.561 1.086 0.702 0.008 0.000 0.984 0.000 0.000 0.000 ------------------------------------------------------ Cluster 0: 2 A pages, 33 B pages. Mean: 0 0.01 1 0.00 2 1.52 3 0.12 4 4.20 5 0.00 6 0.05 7 0.03 8 10.19 9 13.65 A 6.96 B 0.89 C 15.32 D 0.06 E 6.26 F 6.41 G 0.02 H 0.01 I 0.03 J 0.42 K 0.02 L 0.00 M 1.65 N 1.06 O 13.07 P 3.39 Q 0.58 R 3.17 S 6.20 T 0.20 U 0.12 V 0.26 W 0.10 X 0.71 Y 0.04 Z 3.27 *****PAGES: 49 50 57 59 60 65 79 80 82 83 84 89 90 93 94 111 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 ------------------------------------------------- Cluster 1: 18 A pages, 11 B pages. Mean: 0 0.02 1 0.00 2 1.82 3 0.18 4 2.48 5 0.00 6 0.10 7 0.02 8 8.80 9 10.45 A 12.21 B 0.66 C 6.21 D 0.32 E 5.48 F 6.27 G 0.02 H 0.02 I 0.21 J 1.00 K 0.08 L 0.02 M 4.55 N 1.18 O 15.04 P 3.35 Q 1.11 R 5.66 S 7.93 T 0.31 U 0.18 V 0.45 W 0.29 X 0.94 Y 0.10 Z 2.53 *****PAGES: 1 3 11 16 27 43 45 47 48 62 63 64 66 67 69 70 74 75 76 77 78 87 97 98 99 101 106 107 108 ------------------------------------------------- Cluster 2: 66 A pages, 0 B pages. Mean: 0 0.02 1 0.00 2 1.63 3 0.14 4 2.48 5 0.00 6 0.13 7 0.03 8 7.76 9 11.52 A 6.61 B 0.76 C 5.09 D 0.23 E 5.59 F 4.94 G 0.01 H 0.00 I 0.15 J 0.63 K 0.02 L 0.01 M 3.67 N 0.59 O 20.43 P 4.79 Q 1.62 R 4.85 S 11.32 T 0.23 U 0.06 V 0.21 W 0.31 X 0.52 Y 0.06 Z 3.60 *****PAGES: 2 4 5 6 7 8 9 10 12 13 14 15 17 18 19 20 21 22 23 24 25 26 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 44 46 51 52 53 54 55 56 58 61 68 71 72 73 81 85 86 88 91 92 95 96 100 102 103 104 105 109 110 ------------------------------------------------- THESE ARE THE RESULTS FOR THE LABEL CORPUS Line Word Letter Global Initial Final Initial Final 0 0.06 0.00 0.00 0.00 0.00 1 0.00 0.00 0.00 0.00 0.00 2 2.88 5.12 4.33 3.39 8.47 3 0.00 0.00 0.00 0.00 0.00 4 0.06 0.39 0.00 0.00 0.00 5 0.00 0.00 0.00 0.00 0.00 6 1.07 0.00 0.79 5.08 0.00 7 0.13 0.00 0.79 0.00 0.00 8 5.58 4.72 3.15 11.86 0.00 9 10.09 5.51 44.49 5.08 25.42 A 15.11 1.57 1.18 37.29 0.00 B 0.63 0.00 0.00 0.00 0.00 C 4.26 0.39 0.00 0.00 0.00 D 0.44 0.00 1.57 0.00 0.00 E 9.03 0.00 12.20 3.39 16.95 F 5.33 0.39 0.00 0.00 0.00 G 0.00 0.00 0.00 0.00 0.00 H 0.00 0.00 0.00 0.00 0.00 I 0.38 0.00 0.39 0.00 0.00 J 1.82 0.00 9.84 0.00 3.39 K 0.00 0.00 0.00 0.00 0.00 L 0.00 0.00 0.00 0.00 0.00 M 0.94 0.00 3.54 0.00 5.08 N 0.63 0.00 1.97 0.00 1.69 O 20.19 72.05 1.57 18.64 3.39 P 6.39 1.18 0.00 5.08 0.00 Q 0.44 0.00 0.39 1.69 0.00 R 7.15 0.00 13.39 0.00 28.81 S 4.01 7.09 0.00 5.08 0.00 T 0.25 0.00 0.00 0.00 1.69 U 0.13 0.00 0.00 0.00 3.39 V 0.75 0.00 0.39 0.00 1.69 W 0.31 0.79 0.00 0.00 0.00 X 0.19 0.00 0.00 0.00 0.00 Y 0.19 0.00 0.00 0.00 0.00 Z 1.57 0.79 0.00 3.39 0.00 Sample 1595 254 254 59 59 Entropy 3.732 1.644 2.712 2.818 2.770 ------------------------------------------------------ Digraphs whose max frequency global, line initial, etc. > 1.000000%: line word global initial final initial final wf/wi 29 0.3127 0.3937 1.1858 0.0000 0.0000 0.0000 2A 1.2510 1.9685 0.0000 3.4483 0.0000 3.3898 2O 0.6255 2.7559 0.0000 0.0000 0.0000 3.3898 2Q 0.0000 0.0000 0.0000 0.0000 0.0000 1.6949 69 0.7819 0.0000 3.5573 3.4483 1.6949 0.0000 6A 0.2346 0.0000 0.0000 1.7241 0.0000 0.0000 89 3.0493 0.3937 13.8340 3.4483 5.0847 0.0000 8A 2.2674 1.9685 0.3953 8.6207 0.0000 0.0000 8O 0.6255 1.9685 0.3953 0.0000 0.0000 0.0000 92 0.3909 0.0000 0.7905 1.7241 1.6949 3.3898 98 0.8600 0.3937 1.1858 0.0000 0.0000 1.6949 9A 0.0000 0.0000 0.0000 0.0000 0.0000 5.0847 9B 0.0782 0.0000 0.0000 1.7241 0.0000 0.0000 9E 0.0000 0.0000 0.0000 0.0000 0.0000 3.3898 9F 0.7819 3.1496 0.0000 0.0000 0.0000 0.0000 9O 0.0782 0.3937 0.0000 0.0000 0.0000 6.7797 9P 0.3909 1.5748 0.0000 0.0000 0.0000 3.3898 9S 0.0000 0.0000 0.0000 0.0000 0.0000 1.6949 AD 0.4691 0.0000 1.5810 0.0000 0.0000 0.0000 AE 6.6458 0.3937 7.5099 6.8966 11.8644 0.0000 AJ 1.9547 0.0000 8.6957 8.6207 3.3898 0.0000 AM 1.0164 0.0000 3.5573 0.0000 5.0847 0.0000 AN 0.7819 0.3937 1.9763 1.7241 1.6949 0.0000 AR 6.4113 0.3937 9.4862 20.6897 27.1186 0.0000 AT 0.3127 0.0000 0.0000 0.0000 1.6949 0.0000 AU 0.0782 0.0000 0.0000 0.0000 1.6949 0.0000 C9 0.7819 0.0000 1.5810 0.0000 8.4746 0.0000 CO 1.7201 0.0000 0.0000 0.0000 3.3898 0.0000 CV 0.1564 0.3937 0.0000 0.0000 1.6949 0.0000 E2 0.7037 0.0000 1.5810 0.0000 1.6949 0.0000 E8 1.1728 0.0000 1.1858 0.0000 0.0000 5.0847 E9 1.6419 0.0000 7.9051 0.0000 0.0000 1.6949 EA 1.6419 0.0000 0.3953 0.0000 0.0000 5.0847 EF 0.1564 0.0000 0.0000 1.7241 0.0000 0.0000 EO 0.6255 0.0000 0.3953 1.7241 0.0000 1.6949 ES 0.6255 0.0000 0.0000 0.0000 0.0000 1.6949 EZ 0.7037 0.0000 0.0000 0.0000 0.0000 1.6949 F9 0.7819 0.0000 1.1858 0.0000 5.0847 0.0000 FA 2.5020 0.0000 0.0000 0.0000 0.0000 0.0000 FO 1.4855 0.3937 0.0000 0.0000 0.0000 0.0000 I2 0.0782 0.0000 0.0000 0.0000 1.6949 0.0000 J6 0.0782 0.0000 0.0000 0.0000 0.0000 3.3898 M8 0.0782 0.0000 0.0000 0.0000 0.0000 1.6949 MA 0.0782 0.0000 0.0000 0.0000 0.0000 1.6949 MO 0.0000 0.0000 0.0000 0.0000 0.0000 1.6949 N9 0.1564 0.0000 0.7905 0.0000 0.0000 1.6949 O2 0.8600 0.3937 0.3953 0.0000 3.3898 0.0000 O8 1.4073 0.7874 0.0000 1.7241 0.0000 0.0000 OA 0.4691 0.7874 0.0000 0.0000 0.0000 1.6949 OB 0.7037 3.5433 0.0000 0.0000 0.0000 0.0000 OE 4.3002 2.3622 4.7431 5.1724 5.0847 0.0000 OF 5.0039 23.2283 0.0000 5.1724 0.0000 0.0000 OP 6.8804 33.8583 0.0000 3.4483 0.0000 1.6949 OR 2.3456 1.9685 3.5573 3.4483 1.6949 0.0000 OU 0.0782 0.0000 0.0000 0.0000 1.6949 0.0000 OV 0.5473 2.3622 0.0000 0.0000 0.0000 0.0000 P9 0.7819 0.0000 1.1858 0.0000 1.6949 0.0000 PA 3.3620 0.0000 0.0000 1.7241 0.0000 0.0000 PC 1.0946 0.0000 0.0000 3.4483 0.0000 0.0000 PO 1.7983 1.1811 0.0000 0.0000 0.0000 0.0000 Q9 0.3127 0.0000 1.5810 0.0000 0.0000 0.0000 QA 0.0782 0.0000 0.0000 1.7241 0.0000 0.0000 R8 0.1564 0.0000 0.0000 0.0000 0.0000 1.6949 R9 1.1728 0.0000 5.5336 0.0000 1.6949 0.0000 RA 2.3456 0.0000 0.3953 0.0000 0.0000 20.3390 RO 0.5473 0.0000 0.0000 0.0000 0.0000 3.3898 RS 0.2346 0.0000 0.0000 0.0000 0.0000 1.6949 RZ 0.0782 0.0000 0.0000 0.0000 0.0000 1.6949 S9 0.1564 0.0000 0.0000 0.0000 1.6949 0.0000 SA 0.7037 1.1811 0.0000 1.7241 0.0000 0.0000 SC 1.3292 1.9685 0.0000 1.7241 0.0000 0.0000 SF 0.0782 0.0000 0.0000 1.7241 0.0000 0.0000 SO 1.4073 2.7559 0.0000 0.0000 0.0000 0.0000 T6 0.0782 0.0000 0.0000 0.0000 0.0000 1.6949 U8 0.0000 0.0000 0.0000 0.0000 0.0000 1.6949 UO 0.0000 0.0000 0.0000 0.0000 0.0000 1.6949 V9 0.0782 0.0000 0.0000 0.0000 0.0000 1.6949 Y9 0.2346 0.0000 1.1858 0.0000 0.0000 0.0000 Z9 0.5473 0.0000 2.7668 0.0000 0.0000 0.0000 ZA 0.2346 0.0000 0.0000 1.7241 0.0000 0.0000 ZO 0.4691 0.7874 0.0000 1.7241 0.0000 0.0000 # 1279 254 253 58 59 59 ------------------------------------------------------ Selected digraph frequencies (characters 89SOERANMP): (Row = 1st char, col = second) 8 9 S O E R A N M P 8 0.000 3.049 0.235 0.625 0.000 0.000 2.267 0.000 0.000 0.000 9 0.860 0.000 0.000 0.078 0.000 0.000 0.000 0.000 0.000 0.391 S 0.938 0.156 0.000 1.407 0.000 0.000 0.704 0.000 0.000 0.000 O 1.407 0.000 0.313 0.078 4.300 2.346 0.469 0.000 0.156 6.880 E 1.173 1.642 0.625 0.625 0.000 0.078 1.642 0.000 0.000 0.000 R 0.156 1.173 0.235 0.547 0.000 0.000 2.346 0.000 0.000 0.000 A 0.078 0.078 0.000 0.000 6.646 6.411 0.000 0.782 1.016 0.078 N 0.000 0.156 0.000 0.000 0.000 0.000 0.156 0.000 0.000 0.000 M 0.078 0.000 0.000 0.000 0.000 0.000 0.078 0.000 0.000 0.000 P 0.078 0.782 0.704 1.798 0.000 0.000 3.362 0.000 0.000 0.000 ------------------------------- From rcarter@nyx.cs.du.edu Tue Oct 20 13:16:05 1992 Date: Mon, 19 Oct 92 22:16:05 MDT From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9210200416.AA03498@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: voynich@rand.org Subject: Karl Kluge's run; a thought... Status: OR Knowing that old ms. had no punctuation to speak of, and run-on sentences, is it possible that O and 9 are place markers meant to mark begins of sentences/lines (O) or sentence/line ends (9)? Just a thought, Ron | rcarter@nyx.cs.du.edu From Karl.Kluge@G.GP.CS.CMU.EDU Tue Oct 20 17:49:00 1992 Date: Tue, 20 Oct 1992 17:49-EDT From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Currier's paper and the output I posted yesterday Message-Id: <719617785/kck@G.GP.CS.CMU.EDU> Status: OR I looked through Currier's paper and the program output, and unsurprisingly the differences show up. Here are Currier's rules for distinguishing A & B, with the relevant data from the output: a) Final 89 is very high in Language B; almost non-existent in Language A. A corpus: line final 13.4%, word final 4.7% B corpus: line final 23.6%, word final 31.5% b) SOE and SOR are very high in A, often repeated; low in B. SO freq in A corpus: 7.2% SO freq in B corpus: 0.35% c) The symbol groups SAN and SAM rarely occur in B; medium frequency in A. SA freq in A corpus: 1.2% SA freq in B corpus: 0.12% d) Initial SOP high in A, rare in B. A corpus: SO line initial 5.7%, word initial 13.3% B corpus: < 1% both line and word initial e) Initial Q very high in A, very low in B. 5.4% word-initial in A, 0.41% word-initial in B f) Unattached finals scattered throughout Language B. Didn't check this one. Currier also discusses differences in line and word initial frequencies for certain characters in the A herbal corpus: word line initial initial Q 5.42% 0.46% W 1.12% 0.09% X 1.50% 0.00% Y 9.49% 6.95% I noticed something interesting about the results of the B corpus page clustering. The sets of pages were: Cluster 0: 49 50 59 60 79 80 93 111 147 148 149 150 151 152 153 154 155 156 159 160 161 162 163 164 165 Cluster 1: 63 76 77 78 97 98 107 108 157 158 Cluster 2: 64 65 66 75 83 84 89 90 94 Note that in the D'Imperio transcription, page odd & odd+1 are front and back of a folio, i.e. folios 40, 50, 55, and 80 come together in cluster 1, with folios 34, 43, and 46 in cluster 2. Folios 33 and 39 get split between the two clusters. I don't have the Yale hardcopy, but perhaps someone who does could look at the folios from the two smaller clusters to see if there is an apparent difference in script. I'm still trying to figure out what, if anything, the label corpus (f68r1, f70v2, f72r2, f88r, f100r) stats can tell us. There are some obvious noticeable differences between the labels and the mss as a whole: 4O is line initial 17.9% in the mss, < 1% in the labels AM is line final 11.5% in the mss, 3.6% in the labels OF is line initial 2.1% in the mss, 22.2% in the labels OP is line initial 3.5% in the mss, 33.9% in the labels The average number of label occurrences per page is noticeably higher in the biological section (127 in 20 pages) than in the herbal (232 in 111 pages). This may be the first real evidence of apparent differences in subject matter being reflected in properties of the text. From Karl.Kluge@G.GP.CS.CMU.EDU Wed Oct 21 01:43:00 1992 Date: Wed, 21 Oct 1992 01:43-EDT From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Re: Voynich hand A & B Message-Id: <719646234/kck@G.GP.CS.CMU.EDU> Status: OR > About Karl's clusters, etc, and Currier's A & B assignments. > > Currier's assignment of pages to A and B is consistent with the > physical makeup of the book, in the following two ways: > > (1) both sides of a folio are A or both are B, but not > one of each. Currier talks about this in his papers. > > (2) pairs of folios which are parts of the same sheet > of vellum, with an intervening fold, such as f1 and f8, > or f34 and f39 are all A or all B. > But now Karl says that his cluster boundaries do not obey (1), let > alone (2). Odd. Those are clusters within the set of B pages, so there is no real problem. Also, don't forget that the clustering uses nearest mean (Euclidean distance) with the relative frequencies as features -- that the A/B separation shows up as well as it does using the full D'Imperio suprises me. That the majority of the pages in the B hand clusters pair in folios (something that didn't happen with 2 of the 3 A page clusters) suggests something unusual about the folios that fell in those clusters. From reeds@gauss.att.com Wed Oct 21 12:00:12 1992 Message-Id: <199210210300.AA01698@rand.org> From: reeds@gauss.att.com Date: Tue, 20 Oct 92 23:00:12 EDT To: voynich@rand.org Subject: Voynich hand A and B Status: OR About Karl's clusters, etc, and Currier's A & B assignments. Currier's assignment of pages to A and B is consistent with the physical makeup of the book, in the following two ways: (1) both sides of a folio are A or both are B, but not one of each. Currier talks about this in his papers. (2) pairs of folios which are parts of the same sheet of vellum, with an intervening fold, such as f1 and f8, or f34 and f39 are all A or all B. (In detail, here is the story for Quire VI, comprising ff 41-48, or NSA page numbers 79-94. This is a nest of 4 folded sheets. The outermost sheet is f41/f48, the next is f42/f47, the next is f43/f46, and in the middle is f44/f45. Currier assigns these letters to the 16 pages: BBAABBAAAABBAABB. That the letters always come in pairs is phenom (1) above. That the whole quire is a palindrome is phenom (2) above.) When talking with D'Imperio earlier this year she thought that Currier did not know how the folios were gathered into the book, and hence did not know how they paired up. So phenom (2) can be taken as confirmatory evidence for Currier's view of the book being written before gathering, onto sheets of vellum, with different scribes taking different sheets. But now Karl says that his cluster boundaries do not obey (1), let alone (2). Odd. Jim Reeds From j.guy@trl.oz.au Thu Oct 22 02:35:04 1992 Date: Wed, 21 Oct 92 12:35:04 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9210210135.AA04355@medici.trl.OZ.AU> To: voynich@rand.org Subject: Karl Kluge's runs Status: OR (A parte to Ron Carter first) No, I don't think that O and 9 can be markers for sentence and line beginnings and ends. First, we ought to find 9 regularly at the end of every paragraph if it was so. Next, 9 accounts for 10.89% of the letter in the A corpus, and 13.22% in B. Grabbing the natural-born calculator inside my skull, I compute that every 9th letter in A is 9, which gives me 8-letter sentences on average. And 7-letter sentences in corpus B. Suspiciously short. I still can see only three reasonable possibilities: 1. The spaces in the text are spurious. Some letters, because of their shapes, do not connect to the next letter, and cause the scribe to break the continuity and insert what we interpret as a space. Arabic works like that. 2. Those 9, M, N etc are word-final forms of other letters. Viz the two forms for sigma in Greek: open word-finally, a closed noose elsewhere. 3. They are not special forms. They just occur mostly word-finally, that's all. Just imagine a language in which words must start with a consonant and end with a vowel, and you would have something close to what we see there. To that, I cannot refrain from adding the fact that like letters tend to occur next to like letters, that is, I-like next to I-like (A, I, N, M), and C-like next to C-like (C, S, Z, 9). That suggests to me vowel and/or consonant harmony in the language, or that letters have alternative forms, e.g. C is an alternative form of I, etc. But that leaves us with an incredibly small alphabet. Forced to make a choice, I'd tend to think that (3) above is true, and that there is consonant and vowel harmony of some sort. Which takes me to a Chinese-type language, phonologically, that is. For instance, take the pinyin transcription system of Mandarin: the consonants q, x, and j and always *followed* by either i or u-umlaut, never by any other vowel; k, g, h are *never* followed by i or u-umlaut. There are many more co-occurrence restrictions, which I'll spare you. But no, I don't think it's Chinese. Only when I have my tongue in my cheek, you know, Marco Polo and all that. From reeds@gauss.att.com Thu Oct 22 06:11:00 1992 Message-Id: <199210212112.AA25595@rand.org> From: reeds@gauss.att.com Date: Wed, 21 Oct 92 17:11:00 EDT To: voynich@rand.org Subject: Voynich and Karl's clusters Status: OR Following Karl's suggestion I looked at folios 33 and 39 and saw no handwriting differences that jumped out at me: 33r looks like 33v, 39r looks like 39v. This from the Yale "copyflow". Jim Reeds From Karl.Kluge@G.GP.CS.CMU.EDU Wed Oct 21 23:17:00 1992 Date: Wed, 21 Oct 1992 23:17-EDT From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Some musings and an improved version of my code. Message-Id: <719723846/kck@G.GP.CS.CMU.EDU> Status: OR [New features added since the initial posting of the code: digraph counts, distance from cluster mean printed for the pages in each cluster, a few bug fixes (it now handles a cluster becoming empty correctly, for example).] I looked at which characters were line-initial but not -final, line-final but not -initial, both, and neither. I also looked at the same things for "words" (my program doen't treat the line-initial and -final characters and digraphs as also being word-initial or -final). The bad news is that the expectations in cases where the two don't agree are so small that I'm not sure they're significant. For example, B occurs line-initial, but never line-final. It does occur word-final, with a frequency of 0.1%. Either spaces after B's in the running text are bogus *or* the difference between 0 and 2.15 (= 0.001 * 2151) occurrences of B line initially isn't statistically significant. And so on for the other cases. If the differences aren't significant, then we're back to being stuck with the problem of explaining the difference in line- and word-initial frequencies for various characters. It occurs to me that even something relatively simple could baffle us endlessly without the correct key. So for instance, consider this straw-man scheme: it's a simple substitution cipher, but there is a character A that is used as follows: whenever A shows up in the ciphertext, all the characters to the next occurrence of A are nulls. It would be fairly easy to encipher/decipher the text using such a scheme if you had the plaintext-Voynich alphabet mapping, but how would you decrypt it without knowing about the evil A character? (or is my cryptographic ignorance showing, and this would be easy?) Which leads me to ask for some reflections from the group on the following question: "What are the features of the Voynich Mss. which any proposal for the encryption scheme and underlying language must explain?" My list at the moment would be: (Statistical issues) Unusual entropy Differences in line-initial and word-initial frequencies (in general, the whole "line as a unit" phenomenon Currier described) The unusually high correlation between the characters before and after a space in the mss (this really shows up for spaces with 9 before them) (Logistical issues) How could a half dozen of so individuals do this fluently without mechanical assistance? If it is a fake, how to account for the statistical regularity of the text. If it is a fake, how to account for the difference in frequency of occurrence of "labels" as substrings in the biological section vs. the herbal section (this would seem to nix any naive line-by-line generation method) I've lost a sense of what (if anything) we've agreed are the critical issues (in particular, new issues beyond stuff in D'Imperio's TR), what (if any) constraints they place on the plausible hypotheses, and how (if at all) we can test these hypotheses to chose among them. I think if I/we could get a sense of that focus, it would either accelerate our making some sort of progress or alternatively totally discourage us ("...and here I sit with all my lore, poor fool no wiser than before...") I also still have this nagging feeling that the labels are to the Voynich mss. what cartouches were to hieroglyphics in terms of giving a possible lever into the language by giving us strings we might reasonably suspect are words in the underlying language. Take "OFCCO8AE", a star label on f68r1 which occurs once and only once in the herbal section (even allowing spaces in the middle), line 00517A "2OCOJ/OPCOJ/9PCO89/4OFCCO8AE/2AJ#" Interesting how 4 insinuates itself before the initial O to give the 15% common word-initial digraph 4O. Interesting how 9/4 occurs for *12%* of all spaces, and there's that 9 before the / before the 4 before the label. But does that all mean anything? ARRRRRRRGGGGGGGHHHHHHHHHHH!!!!!!!!!!!! In any case, here's the improved version of my program so that it's in the archive of the mailing list... Karl -------------------------------------------------------------------------- #include #include #include #include #define MAX_CLUST 20 #define MAX_CHAR 64 #define min(x,y) ((x) < (y) ? (x) : (y)) #define max(x,y) ((x) > (y) ? (x) : (y)) #define absval(x) (max((x), (-(x)))) /* If pred, vt, else vf */ #define sel(pred, vt, vf) ((pred) ? (vt) : (vf)) /* Test for within delta of zero */ #define ZDELT 0.0000001 #define is_damn_near_zero(x) ((absval(x)) < ZDELT) #define sqr(x) ((x) * (x)) #define chr_pos(alphabet, ch) \ sel((index((alphabet), (ch)) == NULL), -1, \ ((int) (index((alphabet), (ch)) - &(alphabet[0])))) static int num_char; /* Random number generation: roll num sides-sided die plus dm */ static int roll(num, sides, dm) int num, sides, dm; { int die, result; result = dm; for (die = 1; die <= num; ++die) result += 1 + (1.0 * sides * random())/(1.0 * 017777777777); return(result); } /* Go through a line of text, incrementing the character count and appropriate character frequency buckets. */ int scan_line(line, alphabet, freq, count) char *line, *alphabet; double *freq; int *count; { int ch, pos; for (ch = 0; ch < strlen(line); ch++) { if (index(alphabet, line[ch]) != NULL) { pos = (int) (index(alphabet, line[ch]) - &(alphabet[0])); freq[pos] += 1.0; *count += 1; } } } static int global = 0, word_init = 0, word_final = 0, line_init = 0, line_final = 0, dg_count = 0, wi_dgc = 0, wf_dgc = 0, li_dgc = 0, lf_dgc = 0, xs_dgc = 0; static double wrld[MAX_CHAR], wd_init[MAX_CHAR], wd_final[MAX_CHAR], ln_init[MAX_CHAR], ln_final[MAX_CHAR], digraph[MAX_CHAR][MAX_CHAR], wi_dg[MAX_CHAR][MAX_CHAR], wf_dg[MAX_CHAR][MAX_CHAR], li_dg[MAX_CHAR][MAX_CHAR], max_dg[MAX_CHAR][MAX_CHAR], lf_dg[MAX_CHAR][MAX_CHAR], xs_dg[MAX_CHAR][MAX_CHAR]; static char punc[10] = " /-#"; static double wd_entropy, li_entropy, lf_entropy, wi_entropy, wf_entropy; #define incr_bin(alphabet, chr, index, array, count) \ {index = chr_pos((alphabet), (chr)); \ if (index != -1) {array[index] += 1.0; count++;}} #define incr_digraph(alphabet, ch1, ch2, i1, i2, array, count) \ {i1 = chr_pos((alphabet), (ch1)); i2 = chr_pos((alphabet), (ch2)); \ if ((i1 != -1) && (i2 != -1)) {array[i1][i2] += 1.0; count++;}} int new_scan_line(line, alphabet) char *line, *alphabet; { int ch, pos, ch2; char last, next; for (ch = 0; ch < strlen(line); ch++) { if (ch == 0) last = punc[0]; if (ch == (strlen(line) - 1)) next = punc[2]; else next = line[ch + 1]; /* Increment the global counts */ incr_bin(alphabet, line[ch], pos, wrld, global); incr_digraph(alphabet, last, line[ch], pos, ch2, digraph, dg_count); /* If the next character is a mss space, update the word final digraph count */ if (next == punc[1]) { incr_digraph(alphabet, last, line[ch], pos, ch2, wf_dg, wf_dgc); } /* If this character is a mss space, update the word-final count using the last character */ if (chr_pos(punc, line[ch]) == 1) { incr_bin(alphabet, last, pos, wd_final, word_final); /* Also update the cross-space digraph count */ incr_digraph(alphabet, last, next, pos, ch2, xs_dg, xs_dgc); } /* If the last character was a mss space, update the word-initial count using the current character */ if (chr_pos(punc, last) == 1) { incr_bin(alphabet, line[ch], pos, wd_init, word_init); /* Also update the word-initial digraph count */ incr_digraph(alphabet, line[ch], next, pos, ch2, wi_dg, wi_dgc); } /* If the current character is a "-" or "#" or the last character on the line, update the line-final count using the last character */ if ((chr_pos(punc, line[ch]) > 1) || (ch == (strlen(line) - 1))) { incr_bin(alphabet, last, pos, ln_final, line_final); } /* If the next character is end-of-line or end-of-paragraph, update line-final digraphs */ if ((next == punc[2]) || (next == punc[3])) { incr_digraph(alphabet, last, line[ch], pos, ch2, lf_dg, lf_dgc); } /* If the last character is an ASCII space, update the line-initial count using the current character and the line-initial digraph count. */ if (chr_pos(punc, last) == 0) { incr_bin(alphabet, line[ch], pos, ln_init, line_init); incr_digraph(alphabet, line[ch], next, pos, ch2, li_dg, li_dgc); } /* If the last character was an end-of-line update the line-initial char freq and digraph counts (handle those big breaks due to plant stems. */ if (last == punc[2]) { incr_bin(alphabet, line[ch], pos, ln_init, line_init); incr_digraph(alphabet, line[ch], next, pos, ch2, li_dg, li_dgc); } last = line[ch]; } } int clear_freqs() { int ind, i2; for (ind = 0; ind < num_char; ind++) { wrld[ind] = wd_init[ind] = wd_final[ind] = ln_init[ind] = ln_final[ind] = 0.0; for (i2 = 0; i2 < num_char; i2++) digraph[ind][i2] = wi_dg[ind][i2] = wf_dg[ind][i2] = max_dg[ind][i2] = li_dg[ind][i2] = lf_dg[ind][i2] = xs_dg[ind][i2] = 0.0; } wd_entropy = li_entropy = lf_entropy = wi_entropy = wf_entropy = 0.0; } /* Entropy formula: h = -sigma(p[i] log2(p[i])) */ #define norm_digrams(i1, i2, array, count, bounds, max_arr) \ if ((count) != 0) \ for (i1 = 0; i1 < (bounds); i1++) for (i2 = 0; i2 < (bounds); i2++) {\ array[i1][i2] /= (float) (count); \ max_arr[i1][i2] = max(max_arr[i1][i2], array[i1][i2]);} #define norm_freq(i1, array, count, bounds, entropy) \ if ((count) != 0) \ for (i1 = 0; i1 < (bounds); i1++) {\ array[i1] /= (float) (count); \ if (!(is_damn_near_zero(array[i1]))) \ entropy -= (array[i1] * log((double) array[i1]) / \ log((double) 2.0));} int norm_freqs() { int pos, p2; norm_digrams(pos, p2, digraph, dg_count, num_char, max_dg); norm_digrams(pos, p2, wf_dg, wf_dgc, num_char, max_dg); norm_digrams(pos, p2, li_dg, li_dgc, num_char, max_dg); norm_digrams(pos, p2, wi_dg, wi_dgc, num_char, max_dg); norm_digrams(pos, p2, lf_dg, lf_dgc, num_char, max_dg); norm_digrams(pos, p2, xs_dg, xs_dgc, num_char, max_dg); norm_freq(pos, wrld, global, num_char, wd_entropy); norm_freq(pos, wd_init, word_init, num_char, wi_entropy); norm_freq(pos, ln_init, line_init, num_char, li_entropy); norm_freq(pos, wd_final, word_final, num_char, wf_entropy); norm_freq(pos, ln_final, line_final, num_char, lf_entropy); } int print_freqs(alphabet, subset, dg_thresh) char *alphabet, *subset; float dg_thresh; { int pos, p2, i1, i2, just_linefed; printf(" Line Word\n"); printf("Letter Global Initial Final "); printf("Initial Final\n"); for (pos = 0; pos < num_char; pos++) printf("%c %9.5f %9.5f %9.5f %9.5f %9.5f\n", alphabet[pos], 100.0 * wrld[pos], 100.0 * ln_init[pos], 100.0 * ln_final[pos], 100.0 * wd_init[pos], 100.0 * wd_final[pos]); printf("Sample %11d %11d %11d %11d %11d\n", global, line_init, line_final, word_init, word_final); printf("Entropy %9.5f %9.5f %9.5f %9.5f %9.5f\n", wd_entropy, li_entropy, lf_entropy, wi_entropy, wf_entropy); printf("------------------------------------------------------\n"); printf("Digraphs whose max frequency global, line initial, etc. > %f%%:\n", 100.0 * dg_thresh); printf(" line word\n"); printf(" global initial final initial final wf/wi\n"); for (i1 = 0; i1 < num_char; i1++) for (i2 = 0; i2 < num_char; i2++) if (max_dg[i1][i2] > dg_thresh) printf("%c%c %8.4f %8.4f %8.4f %8.4f %8.4f %8.4f\n", alphabet[i1], alphabet[i2], 100.0 * digraph[i1][i2], 100.0 * li_dg[i1][i2], 100.0 * lf_dg[i1][i2], 100.0 * wi_dg[i1][i2], 100.0 * wf_dg[i1][i2], 100.0 * xs_dg[i1][i2]); printf("# %8d %8d %8d %8d %8d %8d\n", dg_count, li_dgc, lf_dgc, wi_dgc, wf_dgc, xs_dgc); printf("------------------------------------------------------\n"); /* printf("Digraphs with relative frequencies > %f (%d digraphs)\n", dg_thresh, dg_count); pos = 0; for (i1 = 0; i1 < num_char; i1++) for (i2 = 0; i2 < num_char; i2++) if (digraph[i1][i2] > dg_thresh) { just_linefed = 0; printf("%c%c: %8.6f ", alphabet[i1], alphabet[i2], digraph[i1][i2]); if ((pos % 5) == 4) {printf("\n"); just_linefed = 1;} pos++; } if (!(just_linefed)) printf("\n"); */ printf("Selected digraph frequencies (characters %s):\n", subset); printf("(Row = 1st char, col = second)\n"); printf(" "); for (pos = 0; pos < strlen(subset); pos++) { i1 = chr_pos(alphabet, subset[pos]); if (i1 != -1) printf(" %c ", subset[pos]); } printf("\n"); for (pos = 0; pos < strlen(subset); pos++) { i1 = chr_pos(alphabet, subset[pos]); if (i1 != -1) { printf("%c ", subset[pos]); for (p2 = 0; p2 < strlen(subset); p2++) { i2 = chr_pos(alphabet, subset[p2]); if (i2 != -1) printf("%6.3f ", 100.0 * digraph[i1][i2]); } printf("\n"); } } printf("------------------------------------------------------\n"); } static char currier[50] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"; static char space[2] = " "; static char lang[3] = "AB"; main(argc, argv) int argc; char *argv[]; { char line_buf[200], *mss_name, chr, *postfix, hand[200], *subset; FILE *mss; double folio_freqs[MAX_CHAR], page_freqs[200][MAX_CHAR], clust_means[MAX_CLUST][MAX_CHAR]; double this_dist, min_dist, dg_cutoff, pg_dist[200]; int folio_char_count, cur_folio, this_folio, ch, converge, best_clust, clust_npage[MAX_CLUST], page, clust_anum[MAX_CLUST], clust_bnum[MAX_CLUST], just_linefed, num_clust, do_clust, seed, page_list[200], pnum, pg_lab[200], pg_clust[200], clust, num_pg; SW_INIT(argc, argv); num_clust = swbdInt("-n", "number of clusters", min(10, MAX_CLUST), 2, MAX_CLUST); seed = swbdInt("-s", "RNG seed", 1963, 0, 10000000); dg_cutoff = swbdFloat("-f", "digraph frequency cutoff", 0.01, 0.0, 1.0); /* subset = swdString("-d", "chars to print digraphs for", "489ACEFMOPRSZ");*/ subset = swdString("-d", "chars to print digraphs for", "89SOERANMP"); do_clust = swBool("-c", "cluster the pages"); SW_DONE(argc, argv, ""); /* Zero the global frequency arrays */ num_char = strlen(currier); clear_freqs(); srandom(seed); /* Calculate the normalized relative frequencies and assign random initial cluster labels */ cur_folio = -1; for (ch = 0; ch < num_char; ch++) folio_freqs[ch] = 0; folio_char_count = 0; pnum = -1; for (;(gets(line_buf) != NULL);) { sscanf(line_buf, "%3d", &this_folio); postfix = strchr(&(line_buf[0]), space[0]); new_scan_line(postfix, currier); if (this_folio == cur_folio) { postfix = strchr(&(line_buf[0]), space[0]); scan_line(postfix, currier, folio_freqs, &folio_char_count); } else { if (cur_folio >= 0) { for (ch = 0; ch < num_char; ch++) page_freqs[pnum][ch] = folio_freqs[ch] / (float) folio_char_count; page_list[pnum] = cur_folio; pg_lab[pnum] = line_buf[5]; pg_clust[pnum] = roll(1, num_clust, -1); } cur_folio = this_folio; pnum++; folio_char_count = 0; for (ch = 0; ch < num_char; ch++) folio_freqs[ch] = 0; postfix = strchr(&(line_buf[0]), space[0]); scan_line(postfix, currier, folio_freqs, &folio_char_count); hand[pnum] = line_buf[5]; /* printf("Starting page %3d: %s\n", cur_folio, line_buf);*/ } } num_pg = pnum; norm_freqs(); /* Print the relative frequency information */ print_freqs(currier, subset, dg_cutoff); if (!(do_clust)) exit(1); /* Print the relative frequencies for all pages */ /* for (page = 0; page < num_pg; page++) { printf("Relative frequencies page %3d (initial cluster %3d):\n", page_list[page], pg_clust[page]); for (ch = 0; ch < num_char; ch++) { printf("%c %5.2f ", currier[ch], 100.0 * page_freqs[page][ch]); just_linefed = 0; if ((ch % 9) == 8) {printf("\n"); just_linefed = 1;} } if (!(just_linefed)) printf("\n"); } */ /* Now compute the cluster means and reclassify until convergance */ converge = 0; for (; !(converge); ){ /* Zero the cluster means array */ for (clust = 0; clust < num_clust; clust++) { clust_npage[clust] = 0; for (ch = 0; ch < num_char; ch++) clust_means[clust][ch] = 0; } /* Compute the cluster means */ for (page = 0; page < num_pg; page++) { clust_npage[pg_clust[page]]++; for (ch = 0; ch < num_char; ch++) clust_means[pg_clust[page]][ch] += page_freqs[page][ch]; } for (clust = 0; clust < num_clust; clust++) for (ch = 0; ch < num_char; ch++) clust_means[clust][ch] /= (float) clust_npage[clust]; /* Reclassify the pages based on the new cluster means */ converge = 1; for (page = 0; page < num_pg; page++) { best_clust = 0; min_dist = 0; for (ch = 0; ch < num_char; ch++) min_dist += sqr(clust_means[0][ch] - page_freqs[page][ch]); min_dist = sqrt((double) min_dist); for (clust = 1; clust < num_clust; clust++) if (clust_npage[clust] != 0) { this_dist = 0; for (ch = 0; ch < num_char; ch++) this_dist += sqr(clust_means[clust][ch] - page_freqs[page][ch]); this_dist = sqrt((double) this_dist); if (this_dist < min_dist) { min_dist = this_dist; best_clust = clust; } } if (best_clust != pg_clust[page]) converge = 0; pg_clust[page] = best_clust; pg_dist[page] = min_dist; } } for (clust = 0; clust < num_clust; clust++) clust_anum[clust] = clust_bnum[clust] = 0; for (page = 0; page < num_pg; page++) if (index(lang, hand[page]) == &(lang[0])) clust_anum[pg_clust[page]]++; else clust_bnum[pg_clust[page]]++; /* Print the mean relative frequencies for each cluster and the D'Imperio page number of the folios in that cluster */ for (clust = 0; clust < num_clust; clust++) { printf("Cluster %3d: %3d A pages, %3d B pages. Mean:\n", clust, clust_anum[clust], clust_bnum[clust]); for (ch = 0; ch < num_char; ch++) { printf("%c %5.2f ", currier[ch], 100.0 * clust_means[clust][ch]); just_linefed = 0; if ((ch % 9) == 8) {printf("\n"); just_linefed = 1;} } if (!(just_linefed)) printf("\n"); printf("*****PAGES:\n"); pnum = -1; for (page = 0; page < num_pg; page++) { if (pg_clust[page] == clust) { pnum++; printf("%3d[%5.3f] ", page_list[page], pg_dist[page]); just_linefed = 0; if ((pnum % 7) == 6) {printf("\n"); just_linefed = 1;} /* if ((pnum % 18) == 17) {printf("\n"); just_linefed = 1;} */ } } if (!(just_linefed)) printf("\n"); printf("-------------------------------------------------\n"); } } From Karl.Kluge@G.GP.CS.CMU.EDU Thu Oct 22 04:31:00 1992 Date: Thu, 22 Oct 1992 04:31-EDT From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Late night musings on two more labels begining with O... Message-Id: <719742687/kck@G.GP.CS.CMU.EDU> Status: OR Here are the 8 occurances of OPZC9 from f68r1: labels:<68108A> OPZC9 voynich:03711A 8SO8/4OPCCO89/9PS9/4OPZC9/8SAM/SOEO89- voynich:05405A OPOE/SOE/8AM/SFAM/ZOM/4OPSC9/4OPZC9/8A3- voynich:06114A 2SC9/4OPZC9/8AM/Q2- voynich:08115A BSO89/OPZC9/8O8AM# voynich:10103A PSOPZC9/4OP9/OFSOE/PSO89/4OPAJ/OF9-9POE89- voynich:15521B BOE/ZAR/ZCT/BSC9/OPZC9/OFAO2/AM/OFZC9/8AEFCCC9R9- voynich:15625B 4OFCC9/OE/OEZ89/4OPAN/OPCC89/OEFAN/OPZC9/2AN/OE- voynich:15807B BOEZOE/PSC9/4OFOE/ZC89/4OPZC9/2AE9/FSC9/2POEBS9- There are 4 cases of 9/4OPZC9, 2 cases of 9/OPZC9, and 2 misc. labels:<68122A> OFOE89 voynich:03709A VSO8CC2/ZO89/4OPSC9/4OFSC9/4OW9/SOFOE89- voynich:04103A 4OFOE/89FAM/OFS9/8AM/QOE/QOEO/8AR/ZAN# voynich:04110A FSOE/SOR/8AM/QOM/8SOR/SC9/4OFOE/89/OBSOE/OE8AJ- voynich:04304A 4OFOE89/OFAU/9FAH/6/4OFCC9/OVSOE/8AN/9VSOR/OEVSOR/OPSAE8- voynich:04309A OFOE/OF/Z9/4OFOE89/8AE/8ZO/4OFCCC2/9/OE9/8AM/8AE- voynich:06907A 4OPOE/QOE/OFOE/89/OFS9/9POROR9/2OE8- voynich:07704B OR/AM/SCFO89/8A_-4OFOE/OFAM/OFAR/OF9/OFOE89/OE- voynich:10002A 8ZO89/XO89/4OFOE/89FAM/S9-4OFC9/8AM/Q9- voynich:14744B 8SC89/RAN/OFC89/4OFC89/OPC89/OFOE89/OPAR/OPL/OEAT/CC8AR/89- voynich:15515B 4OFE/ZC9/4OFAE/SC89/OFOE89P9/RAM/OE/ZCOE/ESC9- voynich:16209B PSC89/4OQC2/4OFAR/SX9/4OFAN/SOE/FAM/SX89/4OFOE89- I don't see a rule here. We have 3 cases of 9/OFOE89, 4 cases of 9/4OFOE89, two line-initial cases of 4OFOE89, and two misc. If the 4 didn't show up at the front of the line-initial cases I might think it was punctuation. Any ideas? Karl From j.guy@trl.oz.au Sat Oct 24 01:39:34 1992 Date: Fri, 23 Oct 92 11:39:34 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9210230039.AA07092@medici.trl.OZ.AU> To: voynich@rand.org Subject: Karl's two labels (OPZC9 and OFOE89) Status: OR There seems to be as many cases of a label occurring preceded by 4 in the text as without: preceded by (in which / is space or start of line) / /4 /S OPZC9 4 4 OFOE89 4 6 1 I cannot help think that labels are nouns. Seems natural. So many labels start with O that O might be an article, or a classificatory prefix -- perhaps gender. The puzzle is the "infamous 4" which occurs just as good as saying "always" followed by O. When I was trying to assign probable phonetic values to the letters, I thought that the most likely one for 4 was [h]. Look at English for instance: you don't find h word-finally, and, off-hand I can't think of a word where it occurs medially and is not silent. H has a strong tendency of disappearing unless in stressed syllables, in all languages. If labels, at least those which start with O, are nouns, then perhaps 4 (h) is a plural marker. Or a case marker. Suppose now that some, if not all, the labels that do NOT start with O are also nouns. Well, my guess is that those nouns start with a consonant (note 1), and could not take the pluralizing or whetever h. Two separate declensions, if you prefer. It seems a sensible explanation for the distribution of 4. It doesn't mean that it is right, mind you! Note 1. They could also start with i or u. I have noticed that, in many languages, is retained more often before low and mid vowels (a, e, o) than before high ones (i, u). From Karl.Kluge@G.GP.CS.CMU.EDU Fri Oct 23 22:08:00 1992 Date: Fri, 23 Oct 1992 22:08-EDT From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Jacques's idea... Message-Id: <719892535/kck@G.GP.CS.CMU.EDU> Status: OR I sat down with the raw grep output from the f68r1 "star names" and lined the matches up vertically. Here is the data: labels:<68107A> OQC9 voynich:09104A 4OFSCO/QC9/SOFAN/SOE/8AM/FS8AE- voynich:01206A PSO89/ZOQOE/SOQC9/2- voynich:03904A ...N/OPOEOZCC9/4OQC9/POE/SOR9- voynich:05801A Q2SAN/ZO2AM/SOQC9/ZO/SCBS9/ZOR/ZCAM- voynich:07222A 4OPOE/9POM/SOQC9 voynich:08904B ...CF9/ZCF9/8AE/OQC9- voynich:09305B ...EZCC9/4OCFAM/OQC9/SOEFAR/AM/QAR/O8- voynich:14832B .../4OFOE/OEFOE/OQC9/ZC89/OPOE/SCX9/E89- voynich:16020B 4OQC9/SCFAE/SO89/4OFC89/EZCP9/4OE89/EPC89/4OPAN- voynich:16418B POQC9/SCOR/OEOE/SCOE/PC89/ZCXAE/ZCCX9/SCAE- ---- labels:<68108A> OPZC9 voynich:03711A 8SO8/4OPCCO89/9PS9/4OPZC9/8SAM/SOEO89- voynich:05405A OPOE/SOE/8AM/SFAM/ZOM/4OPSC9/4OPZC9/8A3- voynich:06114A 2SC9/4OPZC9/8AM/Q2- voynich:08115A BSO89/OPZC9/8O8AM# voynich:10103A PSOPZC9/4OP9/OFSOE/PSO89/4OPAJ/OF9-9POE89- voynich:15521B BOE/ZAR/ZCT/BSC9/OPZC9/OFAO2/AM/OFZC9/8AEFCCC9R9- voynich:15625B .../4OPAN/OPCC89/OEFAN/OPZC9/2AN/OE- voynich:15807B ...OE/PSC9/4OFOE/ZC89/4OPZC9/2AE9/FSC9/2POEBS9- ----- labels:<68109A> OP989 voynich:02509A 2O89/SO89/OPSO89/4OPS9/FOM/29/ZO/P9/89- voynich:04211A 2OFAM/OP9/89- voynich:05401A FZOE/4OOW/ZOR/BZOW/ZCBS9/4OP9/89/ZOR9- voynich:14931B ...SC9/8AN/Z89/4OFC8AR/OES89/2OF/OP9/89/ESC89- voynich:02001A BAM/8AM/ZCO/BSCC9/4OP9/8AM/QOR/OP989/2AN- voynich:08303B ...9/ZCO89Z/9OEOR/9PCC89/FAM/SF9/4OP989/8ARAM/9FAJ- voynich:09908A OZOE/O8AM/XC9/XCO89/4OFC9/OP989- voynich:15808B ...ZC89/4OFAN/ZCOE/4OFS89/4OFS89/4OP989- ----- labels:<68110A> OFCAR voynich:02706A OS9/FOFAM/S89/2AM/OFCAR- voynich:06009B ...AM/SCPSC9/QCC9/OFCAR- voynich:08206A OFCOFCAR/SCOPSAR-29-2AM/QAR/8AJ- voynich:14822B 4OFCAR/SC89/ZX9/8OE/9SCC89/... voynich:14928B ...R/OE/AE/OR/SOE/OFCAR/AR/*/OE9- voynich:15136B ...C89/4OE/SCC89/4OFCAR/SCC9/EOE9/89- voynich:16412B ...OE/RSC89/SCOE/4OFCAR/ZOE89- voynich:16415B ...AE/4OFAM/ZC8E/4OFCAR/SCC89/OESCC9/4OFAE/4OFAO9- ----- labels:<68112A> 9PSO89 voynich:01401A BOE9Z9/ZC9/PSO89/4OBS9/OPZOE/89/8AM/PZO8O89- voynich:01511A ZOE/SCO8AM/8AM/8O/9PSO89/SOP/SOP9/OPARM- voynich:08706A OPSAR/OFSAR/8AM/9PSO89/PSOEOE/OPAE/8AE/8AR9- voynich:10003A PSO89/Q9/SXC9/4O8/9PSO89-8SOE/2AM/9PAJ- ------ labels:<68113A> OP92 labels:<14622A> OP92AJ voynich:15034B 2AFAM/OFCC89/SC89/4OPAN/WC9/OBSCC9/OP92AM/OPAR9- ---- labels:<68115A> OX9 voynich:00203A 89/X9/XO/X9/Z9/8PZCC9/Q9/FOPSO89/8AE- voynich:00611A ZO/ZO/XO/X9/PSOR/SO8AM/SOJ- voynich:08119A OPSOP9/8AM/SOP/ZO/X9/Q9/2/O2- voynich:09602A 4OF89/4OFSO8/ZO/X9- voynich:11109B O8AT/ZCCO/X9/2AR/AE/8AII79-8SCXC9/8AM/SC89/QC89- voynich:01203A 4OAT/X9/SOE/**COX9/SCFSO9/X9/OFOE/R9SO2- voynich:01215A SOX9/2/OR/S9/SAN/OR- voynich:01904A 2OR/SAM/SQ9/QOX9/OR/AM/SPSOR/8OU/O89- voynich:02506A .../OT/OFOR/SOOR/OX9- voynich:03806A 8O3/SOX9/8AD/SCO*9/ZO2/SCO2/SAR/QAM- voynich:03901A BSOR/OCCOX9/O/V9SC9/9BSC9/4OBSCO89/OPAM/SAD- voynich:04113A 8SOR/89/8AN/4OX9/9FOE/OFAN# voynich:07301A POEOR/SOX9-OF9/SOM/OFZOE/OE9/OF9- voynich:07612B ...J-OFAR/SCO8AE/OX9-8AN/OXC89/OPC89/OFC89-ES89/OFAM/8A39- voynich:07815B ...89/OR/A3/OFAM/OX9-ZCO8/VAK9- voynich:08105A 4OFAR/SOX9/SOPOR/S9/FAR9- voynich:08409B 9PCC89/OEO2/AM/OX9# voynich:08511A 9PZO/4OX9/OFSO89# voynich:08804A OFS9/4OX9/8AN/8AG/8AT/Z9- voynich:08912B PCS9/8AM/ZCOX9/SCF9/RAM/SC2/ZC9/FAR/ZCX9/S89/SXAJ- voynich:09305B ...PCOE/OFCO89/SCOX9/OEZCC9/4OCFAM/OQC9/SOEFAR/AM/QAR/O8- voynich:10005A .../4OFOE/ZO8OE/4OX9-OFOE/OPAM/8AR/9- voynich:10012A OSOX9/8AM/SOX9- voynich:10403A 98AM/WOJ/SOE/8OX9/QOE/8AJ/OP9- voynich:10611A OAM/4OX9/4OFAJ- voynich:10713B ES8AE/OX9/8AM/OPOE/OPAR/ZAR/AN# voynich:11102B .../SCP9/SCO/X9/SOX9/SCOPC9/ZFSC9/2/O8AM/ZC9- voynich:15206B 4OX9/8OES89/8ZCC8AE/4OFAE/SC89/8AEZC89/... voynich:15511B ZOX9/SC89/PZC9/8Z89/OPSAR/ZCF/SQ9/OPAE/OR9- voynich:15520B 9ZAE/9SC8AR/OX9/2QC9/PAE/SCAR- voynich:15819B ...9/ROR/OBOR/OEBOX9/OEP989- voynich:15830B BOE/OR/OEFAN/OX9/4OFAM/OFAN/OFAR/ZC9/4OEF9- --- labels:<68122A> OFOE89 voynich:04103A 4OFOE/89FAM/OFS9/8AM/QOE/QOEO/8AR/ZAN# voynich:04110A .../8AM/QOM/8SOR/SC9/4OFOE/89/OBSOE/OE8AJ- voynich:06907A 4OPOE/QOE/OFOE/89/OFS9/9POROR9/2OE8- voynich:10002A 8ZO89/XO89/4OFOE/89FAM/S9-4OFC9/8AM/Q9- voynich:03709A .../4OPSC9/4OFSC9/4OW9/SOFOE89- voynich:04304A 4OFOE89/OFAU/9FAH/6/4OFCC9/OVSOE/... voynich:04309A OFOE/OF/Z9/4OFOE89/8AE/8ZO/4OFCCC2/9/OE9/8AM/8AE- voynich:07704B ...-4OFOE/OFAM/OFAR/OF9/OFOE89/OE- voynich:14744B ...N/OFC89/4OFC89/OPC89/OFOE89/OPAR/OPL/OEAT/CC8AR/89- voynich:15515B 4OFE/ZC9/4OFAE/SC89/OFOE89P9/RAM/OE/ZCOE/ESC9- voynich:16209B .../4OFAN/SOE/FAM/SX89/4OFOE89- ------ labels:<22437A> 9FS89 labels:<68123A> 9FS89 voynich:06609B EZCO89/W9/4OFCC9/FCC89/FS89/SC89/4OFC89/S89/ARE/Z2/OE8AM- voynich:06610B 8AM/S89/PC89/FS89/OFCC89/SCX9/S89/FAN/SCAR/OR/... voynich:07506B .../ZC8AJ/4OVSC89/ZC89/FS89/8989/OBSCFAM/ZOXC9/Z8AEO/R9- voynich:07602B 2OR/Z9FAR/SOE/4OP9/FS89/OEF9/8AR/S89/9FAE/OEFC8AM/O89/89- voynich:07905B 4OPS9/2AE/9PCC89/FS89-8SC89/FCC89/8SC89/8AEAN# voynich:08402B ...9/OFCO89/OF9/OFAE89/FS89/OFAR- voynich:09703B .../OFCO89/43FC89/SO89/FS89/BS89/SFAM/O8AJ- voynich:09809B 9PZCO8/AN/ZCZ9/CSC89/FS89/OPAM/OFAM/OF9- voynich:09811B ESC9/FAR/9FS9/FS89# voynich:16620B ...C89/OFAN/ZC9/4OFC89/FS89/OFC89/OFC9/89/OFC89/OEZ89- voynich:06509B 9FS89/4O8-AR-SQ-AX9-92AT/AT/SO8AR/PAJ- voynich:07903B 4OF/RA9/4CFC89/9S89FS89-4OFC89/4OFC89/ZCFS9/SCF9/8AE9- voynich:09001B ...E/9VSC9/POS29/OFAR/9FS89/8AM- voynich:10306A ...R/OXO8OR/SAN/4OFO8-9FS89/SCC2/8AE- ----- labels:<68125A> OW9 voynich:09111A 8ZO/W9/8AM/8AM/9# voynich:03709A VSO8CC2/ZO89/4OPSC9/4OFSC9/4OW9/SOFOE89- voynich:04101A ...BS9/OP9/OFOE9/8AM/OBS9/2/OW9- voynich:05301A BSO8AR/ZO8/SOW9/OBSOR/8AM-OPSOE/S9/4OE/89- voynich:08901B BSCOW9/4OPC89/SCP9/89/SCBS9/9VSC... voynich:10006A OFOR/QCO89/4OW9/4OFOE/4OPAM-OP9FOE/SOE8AM- voynich:10501A ...8AE/4OBSOE/YCO89-OBSCOE/SOW9- voynich:10901A OPSAE/SS29/OP9/C2C89/S9/9SOW9/SOR/S9/S9- --- labels:<68126A> OFCCO8AE voynich:00517A 2OCOJ/OPCOJ/9PCO89/4OFCCO8AE/2AJ# -------- labels:<68127A> OFZOR voynich:02604A 9P9S9/FZO/9FZ9/ZOFZOR/9P9/8ARO89/89/OP982- voynich:02608A SOFZOR/8AM/OFZ989/8AM/8OE/8AT/8AJ- voynich:04209A OFZOR/ZO89/SOE/PSOE/OPAM/8AM- ----- labels:<68128A> OR8AM voynich:00402A .../4OPSO/EOCCC2/4OP9/SOR/8AM- voynich:01003A 4OPSO/9POR/8AM/8AM/OPSOR/8AM/4O*/8ARSOR/88- voynich:01310A 2/SAN/SOR/8AM/SX9# voynich:01408A ...CCO89/4OPSC9/PC9/OFSOR/8AM- voynich:01801A VOSOR/OBORO89/OB9/ZOR/8AM/4OBS9BSO/4OVOE/ZOE/YOE/8AM- voynich:02306A PSOR/8OR/8AM/4OPSOE/OFS9/OFSOR/OM/SX9/8- voynich:02503A 98AM/OE/S9/FSOR/8AM/OEOE- voynich:02601A ...9VO8AN/OP9/S9/89/9BSOR/8AM/FOE/98AN- voynich:02608A SOFZOR/8AM/OFZ989/8AM/8OE/8AT/8AJ- voynich:02803A QOR/8AM/4OFOR/OFCOR/OFAM- voynich:03221A ...COR/ZOEOR/OR/ZCQ9/BCOR/8AM- voynich:03512A 9FSOR/SOR/8AM/8AMOE- voynich:03609A 8AM/SOR/8AM/4OFO4/9/OFSAD- voynich:03610A 4OPOE/8OR/OFSOR/8AM/QOR/OPAJ- voynich:04104A ...Y9/8OROM/9BSOE/29/2SOR/8AM- voynich:04110A FSOE/SOR/8AM/QOM/8SOR/SC9/4OFOE/89/OBSOE/... voynich:04205A 8OR/9FSCOR/8AM# voynich:04302A PO/AR/SOR/8AM/SF8AN/OPS9/EOESO2/8AM/8AJ/... voynich:04403A ...OE/CCC2/OE/8AM/OFCCOR/8AM/4OPSOE8- voynich:04802A 8SOR/QOR/SOR/8AM/2/OFCCAM/8AM/XC9/8AM- voynich:04803A ORSO/FSOR/SOE/8AM/ZYOR/8AM/8ZC9/8AIIP9- voynich:05301A BSO8AR/ZO8/SOW9/OBSOR/8AM-OPSOE/S9/4OE/89- voynich:05404A .../O/SP9/4OPOE/ZCOE/ZOR/8AM/4OP9- voynich:05506A SOFZ9/ZOQ9/ZOR/ZOR/8AM-4OFA3- voynich:05607A 4O8SO/FSOR/8AM/9FAM/8--Z89-SOQ9/ZCF9- voynich:05711A ...COR/SC9/2OCCC/98C9/2OR/8AM- voynich:05810A ZO/ZCOE89/OPSCOR/8AM- voynich:06203A ZOR/8AM/SX9/8ZO/8AN/8AM/2-ZOFC9/FA- voynich:06714A QOE/SOE/AM/4OPS9/OPSOR/8AM-ZOE-4OPAM- voynich:07006A 8AM/QOR/8AM/8AE/892- voynich:07107A 2OR/SC9/FOR/4OFOR/8AM# voynich:07111A OFSOR/8AM/X9/8AN/8AM# voynich:07204A 8ZOR/89POR9/8ZOR/8AM- voynich:07207A 4OPSOR/8AM# voynich:07216A 4OPOR/8AM/SOPAM- voynich:07306A 9ZO/ZO/FO2-8AM/OFO9/SOSOR/8AM# voynich:07403A 4OFCC9/FCOR/8AM/OFC9/FCC9/8AM/8AT/8AM/2- voynich:07407A 8AM/4OE/S9/8AN/2OR/8AM-8AN/8AE89- voynich:07701B ...Z89/OEFC89/OBSC2/AR/OR/8AM- voynich:07909B ...FC89/SC89-SXCO89/SCFOR/8AM/SCO/AE/OPC89- voynich:08802A OFOE/SOR/8AM/Q9/OFSOE/SOJ/OFOEOD/89-OFSOE/2- voynich:09502A ...PS9/SOZ9/4OBSAR/4O/8OR/8AM- voynich:09503A 8ZO/8SOR/8SOR/8AM/8OR/SCO/RAM# voynich:09619A SOFCO/ZO/SCOR/SCOFSCP/SOR/8AM/89-8AM/4OF6- voynich:09807B PC8AM/O8AM/ORO/FAM/9FOR/8AM/SOF9/4OPS89/FC9/8ATAJ- voynich:09909A POR/8AM/8AK/4SO8AE/8AE/4O89/4OCPAJ- voynich:16343B 8AM/OE/8AN/SC9/E8AEOR/8AM/ZCOE/SC89/4OP*/RAR- ------ labels:<68130A> 8OESC89 voynich:15303B 4XC89/SCF9/8OE/SC89/4OFC89/4OFAN/OEFC89- voynich:16202B ...9/4OPOE/8OE/SC89/ZC89/89/8AROPC89/SCPC89/EOFAJ- voynich:16611B ZC8OE/SC89/OFC89/OPOE/SXC89/4OPC89/8OFC89/8OE/... voynich:16210B 8OESC89/ESC9/ROE/ROE/SCC9/2ANSC9- ------- ------------------------------------------------- labels OEOR, OPOE, OPOR not included due to large numbers of matches. From j.guy@trl.oz.au Sat Oct 24 07:08:55 1992 Date: Fri, 23 Oct 92 17:08:55 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9210230608.AA07577@medici.trl.OZ.AU> To: voynich@rand.org Subject: Karl's labels: IDEA! Status: OR (This is based on my pet theory that the Voynich is not a cipher, at least not anything more than a simple substitution cipher). Karl's two labels have shown to occur preceded by a very small number of different letters. Now, let us think. Take a language such as, I was going to say Latin, but even English will do. English makes more frequent use of suffixes (-ful, -ness, -ity, -tion, -ly, -hood, -wise etc.) than prefixes (un-, in-, pre-, post-, anti-). Let us assume that the labels are complete words, and look for those labels in the text, jotting down what occurs before and after them. If we observe a greater variety of strings occurring *before*, then Voynichese is mainly a *suffixing* language. More variety in strings occurring *after*, it's mostly a *prefixing* language. Because in any language you have fewer affixes and grammatical words than full- fledged words, so there is bound to be less variety before words in a mainly prefixing language, and and less variety after words in a suffixing one. Say we suspect it is a prefixing language. Look again at that list of strings occurring *before* putative labels in the text. Those strings which occur very frequently must be prefixes, or grammatical particles. Make a list of them. Look again at those labels. Those which are not preceded by those putative prefixes or particles must be preceded by Voynichese words. In other words, if there isn't a space or line break there you must restore one. We ought to be able then to find word boundaries. The behaviour of O and 4O makes me think Voynichese uses prefixes. A half-boiled idea. Let me try and boil the other half.... From Karl.Kluge@G.GP.CS.CMU.EDU Mon Oct 26 01:48:00 1992 Date: Mon, 26 Oct 1992 01:48-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Re: Necronomicon - Dee version Message-Id: <720082081/kck@G.GP.CS.CMU.EDU> Status: OR > Date: Sun, 25 Oct 92 22:40:08 MST > From: rcarter@nyx.cs.du.edu (Ron Carter) > Subject: Re: Necronomicon - Dee version > > The Voynich could be the Necronomicon translated from -ARABIC- to > abbreviated Latin via dictation... If you read either "Return of the Lloigor" or THE PHILOSOPHER'S STONE by Colin Wilson, you will discover that if you photograph the Voynich Mss it brings out faded parts of the letters which reveal the Voynich to be a Greek text written in Arabic...the Greek text of the NECRONOMICON. Wilson seems to be under the impression that the Voynich is at U. Penn. Karl From baez@ucrmath.ucr.edu Mon Oct 26 13:08:04 1992 Date: Sun, 25 Oct 92 20:08:04 PST From: baez@ucrmath.ucr.edu (john baez) Message-Id: <9210260408.AA21585@ucrmath.ucr.edu> To: voynich@rand.org Subject: Dee, Necronomicon - misinformation Status: OR Hi, folks, I still exist, I'm simply burnt out on the Voynich, though I'm reading what you're posting. As you may know, there's a new newsgroup inanely titled alt.necromicon, which is for lovers of the Necronomicon. Someone posted the alt.magick Necronomicon FAQ, and it refers to some of our favorite suspects. Though the fellow posting it seemed to believe this FAQ, it is clearly no more than an amusing stringing-together of dark and mysterious nonsense, so I post it just for the entertainment of you hardworking scholars. (Selected bits only.) Q. What is the printing history of the Necronomicon? No Arabic manuscript is known to exist; the author Idries Shah carried out a search in the libraries of Deobund in India, Al- Azhar in Egypt, and the Library of the Holy City of Mecca, without success. A Latin translation was made in 1487 (not in the 17th. century as Lovecraft maintains) by a Dominican priest Olaus Wormius. Wormius, a German by birth, was a secretary to the first Grand Inquisitor of the Spanish Inquisition, Tomas de Torquemada, and it is likely that the manuscript of the Necronomicon was seized during the persecution of Moors ("Moriscos") who had been converted to Catholism under duress; this group was deemed to be unsufficiently pure in its beliefs. It was an act of sheer folly for Wormius to translate and print the Necronomicon at that time and place. The book must have held an obsessive fascination for the man, because he was finally charged with heresy and burned after sending a copy of the book to Johann Tritheim, Abbot of Spanheim (better known as "Trithemius"); the accompanying letter contained a detailed and blasphemous interpretation of certain passages in the Book of Genesis. Virtually all the copies of Wormius's translation were seized and burned with him, although there is the inevitable suspicion that at least one copy must have found its way into the Vatican Library. Almost one hundred years later, in 1586, a copy of Wormius's Latin translation surfaced in Prague. Dr. John Dee, the famous English magician, and his assistant Edward Kelly were at the court of the Emperor Rudolph II to discuss plans for making alchemical gold, and Kelly bought the copy from the so-called "Black Rabbi" and Kabbalist, Jacob Eliezer, who had fled to Prague from Italy after accusations of necromancy. At that time Prague had become a magnet for magicians, alchemists and charletons of every kind under the patronage of Rudolph, and it is hard to imagine a more likely place in Europe for a copy to surface. The Necronomicon appears to have had a marked influence on Kelly; the character of his scrying changed, and he produced an extraordinary communication which struck horror into the Dee household; Crowley interpeted it as the abortive first attempt of an extra-human entity to communicate the Thelemic "Book of the Law". Kelly left Dee shortly afterwards. Dee translated the Necronomicon into English while warden of Christ's College, Manchester, but contrary to Lovecraft, this translation was never printed - the manuscript passed into the collection of the great collector Elias Ashmole, and hence to the Bodleian Library in Oxford. There are many modern fakes masquerading as the Necronomicon. They can be recognised by a total lack of imagination or intelligence, qualities Alhazred possessed in abundance. Q. Why did the novelist H.P. Lovecraft claim to have invented the Necronomicon? The answer to this interesting question lies in two people: the poet and magician Aliester Crowley, and a Brooklyn milliner called Sonia Greene. There is no question that Crowley read Dee's translation of the Necromonicon in the Ashmolean, probably while researching Dee's papers; too many passages in Crowley's "Book of the Law" read like a transcription of passages in that translation. Either that, or Crowley, who claimed to remember his life as Edward Kelly in a previous incarnation, read it in a previous life! Why doesn't he mention the Necronomicon in his works? He was surprisingly reticent about his real sources - there is a strong suspicion that '777', which Crowley claimed to have written, was largely plagiarised from Allan Bennet's notes. His spiritual debt to Nietzsche, which in an unguarded moment he refers to as "almost an avatar of Thoth, the god of wisdom" is studiously ignored; likewise the influence of Richard Burton's "Kasidah" on his doctrine of True Will. I suspect that the Necronomicon became an embarrassment to Crowley when he realised the extent to which he had unconsciously incorporated passages from the Necronomicon into "The Book of the Law". In 1918 Crowley was in New York. As always, he was trying to establish his literary reputation, and was contributing to "The International" and "Vanity Fair". Sonia Greene was an energetic and ambitious Jewish emigre with literary ambitions, and she had joined a dinner and lecture club called "Walker's Sunrise Club" (?!); it was there that she first encountered Crowley, who had been invited to give a talk on modern poetry. It was a good match; in a letter to Norman Mudd, Crowley describes his ideal woman as "rather tall, muscular and plump, vivacious, ambitious, energetic, passionate, age from thirty to thirty five, probably a Jewess, not unlikely a singer or actress addicted to such amusements. She is to be 'fashionable', perhaps a shade loud or vulgar. Very rich of course." Sonia was not an actress or singer, but qualified in other respects. She was earning what, for that time, was an enormous sum of money as a designer and seller of woman's hats. She was variously described as "Junoesque", "a woman of great charm and personal magnetism", "genuinely glamorous with powerful feminine allure", "one of the most beautiful women I have ever met", and "a learned but eccentric human phonograph". In 1918 she was thirty-five years old and a divorcee with an adolescent daughter. Crowley did not waste time as far as women were concerned; they met on an irregular basis for some months. In 1921 Sonia Greene met the novelist H.P. Lovecraft, and in that year Lovecraft published the first novel where he mentions Abdul Alhazred ("The Nameless City"). In 1922 he first mentions the Necronomicon ("The Hound"). On March 3rd. 1924, H.P. Lovecraft and Sonia Greene married. We do not know what Crowley told Sonia Greene, and we do not know what Sonia told Lovecraft. However, consider the following quotation from "The Call of Cthulhu" [1926]: "That cult would never die until the stars came right again [precession of the Equinoxes?], and the secret priests would take Cthulhu from His tomb to revive His subjects and resume His rule of earth. The time would be easy to know, for then mankind would have become as the Great Old Ones; free and wild, and beyond good and evil, with laws and morals thrown aside and all men shouting and killing and revelling in joy. Then the liberated Old Ones would teach them new ways to shout and kill and revel and enjoy themselves, and all earth would flame with a holocaust of ecstacy and freedom." It may be brief, it may be mangled, but it has the undeniable ring of Crowley's "Book of the Law". It is easy to imagine a situation where Sonia and Lovecraft are laughing and talking in a firelit room about a new story, and Sonia introduces some ideas based on what Crowley had told her; she wouldn't even have to mention Crowley, just enough of the ideas to spark Lovecraft's imagination. There is no evidence that Lovecraft ever saw the Necronomicon, or even knew that the book existed; his Necronomicon is remarkably close to the spirit of the original, but the details are pure invention, as one would expect. There is no Yog-Sothoth or Azathoth or Nyarlathotep in the original, but there is an Aiwaz... Q. Where can the Necronomicon be found? Nowhere with certainty, is the short and simple answer, and once more we must suspect Crowley in having a hand in this. In 1912 Crowley met Theodor Reuss, the head of the German Ordo Templi Orientis (O.T.O), and worked within that order for several years, until in 1922 Reuss resigned as head in Crowley's favour. Thus we have Crowley working in close contact for 10 years with the leader of a German masonic group. In the years from 1933-38 the few known copies of the Necronomicon simply disappeared; someone in the German government of Adolf Hitler took an interest in obscure occult literature and began to obtain copies by fair means or foul. Dee's translation disappeared from the Bodleian following a break-in in the spring of 1934. The British Museum suffered several abortive burglaries, and the Wormius edition was deleted from the catalogue and removed to an underground repository in a converted slate mine in Wales (where the Crown Jewels were stored during the 1939-45 war). Other libraries lost their copies, and today there is no library with a genuine catalogue entry for the Necronomicon. The current whereabouts of copies of the Necronomicon is unknown; there is a story of a large wartime cache of occult and magical documents in the Osterhorn area near Salzburg. There is a recurring story about a copy bound in the skin of concentration camp victims. This F.A.Q. was compiled using information obtained from "The Book of the Arab", by Justin Geoffry, Starry Wisdom Press, 1979 ------ Okay, here's a question: was there ever anyone named "Wormius"? From baez@ucrmath.ucr.edu Mon Oct 26 13:14:06 1992 Date: Sun, 25 Oct 92 20:14:06 PST From: baez@ucrmath.ucr.edu (john baez) Message-Id: <9210260414.AA21851@ucrmath.ucr.edu> To: voynich@rand.org Subject: Necronomicon - Dee version Status: OR I think we went over this quite a while ago but now that I am at U.C. Riverside I can use the U.C. library database and, taking a hint from someone on alt.necromicon, quickly turned up the following: 8. Hazred, Abdul. Al Azif, or the Necronomicon, translated from the Arab by Dr. John Dee, being the text recovered from antique lands, with some annotations by divers hands. London, John Dee, 1589. UCLA URL BF 1598 H21aE Has any of our gang ever tried to check this out? I presume it is merely a joke catalog entry, rather than an actual (joke) book. From rcarter@nyx.cs.du.edu Mon Oct 26 14:40:08 1992 Date: Sun, 25 Oct 92 22:40:08 MST From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9210260540.AA06136@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: baez@ucrmath.ucr.edu, voynich@rand.org Subject: Re: Necronomicon - Dee version Status: OR If any of this Necronomicon stuff could be taken serious (and -I- don't even take a Dee/Kelley/Voynich connection seriously) then... The Voynich could be the Necronomicon translated from -ARABIC- to abbreviated Latin via dictation... Just in time for All Hallow's Eve too... Scary stuff... :-) Ron | rcarter@nyx.cs.du.edu From j.guy@trl.oz.au Tue Oct 27 02:20:04 1992 Date: Mon, 26 Oct 92 12:20:04 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9210260120.AA10459@medici.trl.OZ.AU> To: voynich@rand.org Subject: Karl's star names Status: OR A bit of theoretical speculation first. 1. Granted my wild hypothesis that most labels are nouns. 2. That, most labels starting with O, O is an article or a nominal prefix. 3. That O is a vowel, as Sukhotin's algorithm gives it. 4. That 4 occurs only before O, barring a handful of exceptions, and is therefore a nominal prefix, perhaps the plural, perhaps a case. We look at the occurrences in the text of O-initial labels and how often they occur preceded by 4, and not preceded by 4, bearing in mind that a label-like string can occur within a word by pure chance (e.g. "cat" in "concatenate"). 4- Other labels:<68107A> OQC9 2 8 labels:<68108A> OPZC9 4 4 labels:<68109A> OP989 3 5 labels:<68110A> OFCAR 4 4 labels:<68113A> OP92 0 1 labels:<68115A> OX9 6 26 labels:<68122A> OFOE89 6 5 labels:<68125A> OW9 2 6 labels:<68126A> OFCCO8AE 1 0 labels:<68127A> OFZOR 0 3 labels:<68128A> OR8AM 0 47 OX9 and especially OR8AM stick out like sore thumbs. The rest is equally divided between 4-prefixed and nil-prefixed (could be "other"). OX9, a string of just 3 letters, is probably what "cat" is to "concatenation" But OR8AM? OR8AM looks terminally incompatible with the putative 4-prefix. Either OR8AM is not a noun (or the stuff labels are commonly made of), or, if it's a noun, it belongs to a small class of nouns, very distinct from the other O-initial nouns. Perhaps 4 is the mark of the plural and OR8AM is not countable? Or perhaps it is OR plus a suffix -8AM which makes it incompatible with 4- (8AM is so very frequent that it feels like a grammatical particle). One would have to look at the distribution of 4O in the vicinity of 8AM, and at what comes in- between. Note: voynich:07107A 4OFOR/8AM# voynich:07207A 4OPSOR/8AM# voynich:07216A 4OPOR/8AM voynich:09502A 4O/8OR/8AM Meanwhile, it's back to the boring stuff I get paid to do... From reeds@gauss.att.com Thu Oct 29 00:45:50 1992 Message-Id: <199210281546.AA02294@rand.org> From: reeds@gauss.att.com Date: Wed, 28 Oct 92 10:45:50 EST To: voynich@rand.org Subject: Voynich status report Status: OR Voynich progress (and attitude) summary. To me, the most important VMS phenomena needing explanation are the unusually high degree of letter pair correlation (or low digraphic entropy, etc), coupled with the paucity of long (multi word) repeated sequences. The book might be in a language like Hawaian, whose n-th order entropies differ from those of most European languages, but if it were about anything in particular you would expect much more in the way of repeated phrases and words. One possible explanation is Brumbaugh's: it is nonsense, written to gull. The alphabet is phonetic, but is used to record what might as well be glossolallia. This is consistent with both statistical observations above, and with Currier's A/B language findings, if you suppose two or more composers, each with his own glossollalic idiolect. A partial explanation is that the text was filtered through a "word play" transformation, such as pig Latin or the various Hungarian transformations described by Andras Kornai. This would explain the entropy deficiency but not the repetition deficiency. Indeed, repetitions would be expected to be longer. One variant of Brumbaugh's nonsense-for-fraud theory is Trithemius's method of using nonsense to conceal a hidden text. John Baez supplied this snippet (of Trithemius, via Wayne Shumaker): "Pamersiel anoyr madrisel ebrasothean abrulges itrasbiel nadres ormenu itules rablon hamorphiel." which means "nym die ersten bugstaben de omni uerbo..." taking alternate letters after the "system indicator" pamersiel. Brumbaugh's assumption that the nonsense text was Latin-sounding has not convinced the majority of contributing memebers of this group. Each age seems to have a language for its demons to speak. In ours, it is in the DNA-coded computer virus-like language described in (this dates me, guys) books like Crighten's "Andromeda Strain", and in the Rudolphine era, it was in the vaguely Arabic or Hebrew-sounding Trithemian or Enochian. So there you have my guesses: pure nonsense or nonsense concealing a hidden text by conventional concealment methods. We can try to recover the phonetic equivalents of the Voynich letters, we can try to understand the "phonology" of whatever variant of demonic language was used, or the "rules" of whatever word play might have been used. We can search for concealed texts by looking for rules like "take every other letter". A further guess: the distinction between alphabet letter and phoneme is recent (or recent to Europe, I think) and the Voynich alphabet might well reflect whatever phonetic overloadings are present with use of the ordinary alphabet in the author's own language. So what should we do? Continue to transcribe. Continue to look for patterns and repetitions. Jim Reeds From Karl.Kluge@G.GP.CS.CMU.EDU Thu Oct 29 01:49:00 1992 Date: Thu, 29 Oct 1992 01:49-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Why the labels suggest strongly that the Mss is NOT nonsense Message-Id: <720341354/kck@G.GP.CS.CMU.EDU> Status: OR Consider the following phenomena: Given the labels from f68r1 (star names), f70v2 (Pisces), f72r2 (Gemini), and f88r & f100r (pharmaceutical) [excluding a few like OPOE that appear as label "roots" and have many matches, suggesting the "cat"/"concatanate" problem Guy mentioned], we see the following: Herbal A: 149 matches in 86 pages, for 1.733 per page Herbal B: 83 " " " 25 pages, " 3.32 " " Biological B: 127 " " " 20 pages, " 6.35 " " Note how there is a clear difference between frequency of label matches in Herbal A and Biological B (although Herbal B also shows an increase -- incidentally, Herbal B does somewhat differ from Biological B in its relative letter frequencies). This is the frequency distribution of the number of matches of labels from f68r1: Lines matched # of labels from 68r1 ------------------------------------- 0 ************** (14) 1 - 5 ***** (5) 6 - 10 ***** (5) 11 - 15 ** (2) 16+ ***** (5) Note that labels vary significantly in their frequency of occurance. Also consider that the labels differ in character frequncies from both "Languages" A and B. Currier describes the Astrological section by saying "The 'language' throughout is mostly A but without some of the more pronounced 'A' features found in Herbal A." Examination of the runs I posted in light of Currier's criterion for distinguishing A vs. B will confirm the differences between the label corpus and either language A or B. This raises the following question: If the Voynich Mss. is nonsense, then how was the text generated to produce the phenomena described above. I can see four alternatives: 1) Labels and text generated by same process. Objection: then why are so many labels rare or nonexistant strings in the Herbal and Biological sections? (For example, Petersen locus 26 on f68r1, OFCCO8AE, with its one lonely occurance in the D'Imperio "00517A 2OCOJ/OPCOJ/9PCO89/4OFCCO8AE/ 2AJ#") 2) Label folios (such as the Zodiac folios) generated first, labels sprinkled through the running text in some semi-random fashion. Objection: why does the frequncy of matches vary so much between labels? As this implies labels are "intrusive" into the running text (and the line appears to be a "unit", suggesting a line-based generation of the text), lines with label matches ought to differ statistically from "no-match" lines, making this a testable hypothesis. 3) Running text generated first, strings picked at random to serve as labels. Objections: doesn't explain Herbal/Biological differences in label occurance frequency, doesn't explain where no-match labels come from, doesn't explain why so many labels are rare strings in the running text, doesn't explain letter frequency differences between labels and running text. 4) Language A lines generated by one process, language B lines by another, and labels generated by a third process which just happens to produce more matching strings in the B corpus. Real problem if, as Currier claims, the Astrological folios look more like Language A. I think the labels make strong case for there being a "real" text. Comments? Karl (As an aside, given the common O- beginning of labels in the Astrological folios, and various star names from the Arabic with al- at the start (al-gol, al-debaren), has anyone compared letter frequencies for the star/zodiac labels with a list of the names of the 100 brightest stars in 14th-16th century Arabic (or in any other languages, for that matter) to try and guess the language?) (As a second aside, having had Minimum Spanning Tree code lying about, I ran a hierarchical Nearest Neighbor clustering on the D'Imperio pages. There doesn't seem to be any coherent substructure to the A folios, with the B folios showing pretty much the same structure as the 3 cluster Isodata result I posted.) From Karl.Kluge@G.GP.CS.CMU.EDU Thu Oct 29 15:00:00 1992 Date: Thu, 29 Oct 1992 15:00-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Label match frequency corrected for size Message-Id: <720388855/kck@G.GP.CS.CMU.EDU> Status: OR Good point. To a crude approximation (using wc and subtracting 7 char/line for the line label): MATCHES Segment Lines Chars /line /char /page -------------------------------------------------------------------- Herbal A 1107 34809 0.135 0.00428 1.73 Herbal B 296 14153 0.280 0.00582 3.32 Biological B 763 36247 0.166 0.00351 6.35 Hmmm...on the one hand, semantically I think it's fair to consider the page as the "correct" unit in the herbal (there is X amount of information to convey about each plant, the rest is a matter of verbosity). In support of this, I would point to the suprising inversion of the Herbal/Biological B distinction, where on a per line basis Herbal B makes more refs than either Herbal A or Bio B. On the other hand, the per line and per char match frequencies tell a different story, and are probably the "correct" measures in considering a random line-by-line forgery hypothesis. My mental picture of the probable contents is something like Herbal -- composed by two people, one of whom (B) likes to show off his greater knowledge of the underlying theory behind the material by making more references to astrological correspondences in describing each plant and making contrasts with other plants. Astrological -- star names and zodiac correspondence info Biological (sic) -- a synthesis of the underlying system of correspondences, explaining the higher per-page label match rate (culminating in the big fold out, which I still suspect has to do with an underlying theory involving the four elements Earth/Air/Fire/Water). Pharmaceutical -- uses the preceding to construct recipes for medicines. Recipes (sic) -- ? Aside I: A brief bibliography was posted to alt.horror.cthulhu, I forget by whom. It mentions the Tiltman essay. The botanical library here at CMU has it. The folio illustrations are very poor, and there isn't any real information that isn't in D'Imperio. One or two nuggets of interest, but nothing major. The best line is where he mentions showing it to an expert in herbals who says (after looking at it for some time) something like, "Why are you doing this to me? I can spend hours trying to identify the plants in real herbals, and there I usually have the names in Greek, Latin, and Arabic right in front of me in the text!" Aside II: Does someone have good copies of the big foldout? The Petersen is not very clear, the photo in 100 MSS. isn't all that great. Someone sent mail a while back indicating that the Yale copies were OK, but that there were good photos in some book or article. I'd really like to see a clean copy. Karl From Karl.Kluge@G.GP.CS.CMU.EDU Thu Oct 29 15:40:00 1992 Date: Thu, 29 Oct 1992 15:40-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Hierarchical Nearest Neighbor clustering Message-Id: <720391226/kck@G.GP.CS.CMU.EDU> Status: OR John B. asks about the MST code. The basic clustering algorithm is 1) Start with each data point in a separate cluster. 2) Find the pair of current clusters such that the distance between some point in the first and some point in the second is the smallest over all points and all clusters. Merge those two clusters. 3) Repeat step 2 until there is only 1 cluster The weasel word in the above was "distance". In this case, I was using Euclidean norm between two normalized letter frequency distributions. The way the clustering is done in practice is: 1) View the data points as forming a complete graphs with edges weighted by the distance between them. 2) Compute the Minimum Spanning Tree of that graph. 3) Start with each data point in a separate cluster. 4) For each MST edge from shortest to longest, merge the clusters containing the data points which that edge connects. You can draw a tree showing the pattern of merges (I was pressed for time, so I did it by having a 2-D character array for the dendrogram (tree), and making two passes through the MST -- one to compute the depth at which each cluster vanished, the second to draw the vertical connecting "|"'s). Here is the result for the language B pages. The way I use characters to draw the dendrogram may be a little confusing, so as an example: 59B:---------+ Here, 151 and 161 cluster together. 150 and 163 151B:-------+ | cluster together. (151 161) then joins with 161B:-------++| (150 163). 59 then joins with (151 161 150 163). 150B:-------+|| The rule is that the cluster closer to the top 163B:-------++++ is connected downward to the cluster it merges | with, whose line then extend further to the right. That's how to tell that 150 merges with 163, not 161. As I said, I was pressed for time, and was doing it in ASCII. THIS IS THE B CORPUS USING MST-BASED NN-CLUSTERING [NN used to stand for "Nearest Neighbor" until the connectionists moved into the neighborhood] % grep "^.....B" voynich | s4test -c 2 -D n -f 0.05 [Letter and digraph frequency output deleted] ------------------------------------------------------ 60B:---------------------+ 79B:--------------------+| 64B:------------------+ || 98B:----------------+ | || 78B:--------------+ | | || 63B:-------------+| | | || 97B:-------------+++| | || 76B:--------------+|| | || 77B:-------------+||| | || 107B:-------------+++++| || 108B:-----------------+++|| 111B:------------------+||| 65B:----------------+ |||| 66B:---------------+| |||| 75B:---------------+++|||| 89B:----------------+||||| 93B:---------------+|||||| 49B:--------------+||||||| 90B:------------+ |||||||| 50B:-----------+| |||||||| 83B:----------+|| |||||||| 84B:---------+||| |||||||| 94B:---------+++++|||||||| 80B:------------+||||||||| 156B:-----------+|||||||||| 59B:---------+ ||||||||||| 151B:-------+ | ||||||||||| 161B:-------++| ||||||||||| 150B:-------+|| ||||||||||| 163B:-------++++||||||||||| 157B:--------+ |||||||||||| 158B:--------++|||||||||||| 155B:--------+||||||||||||| 149B:------+ |||||||||||||| 152B:-----+| |||||||||||||| 148B:-+ || |||||||||||||| 154B:-++ || |||||||||||||| 159B:--++ || |||||||||||||| 162B:---++|| |||||||||||||| 164B:----++++|||||||||||||| 160B:------+||||||||||||||| 147B:-----+|||||||||||||||| 153B:----+||||||||||||||||| 165B:----++++++++++++++++++- From reeds@gauss.att.com Fri Oct 30 00:50:33 1992 Message-Id: <199210291550.AA08988@rand.org> From: reeds@gauss.att.com Date: Thu, 29 Oct 92 10:50:33 EST To: voynich@rand.org Subject: Voynich label stats Status: OR Karl writes: > Consider the following phenomena: > > Given the labels from f68r1 (star names), f70v2 (Pisces), f72r2 > (Gemini), and f88r & f100r (pharmaceutical) [excluding a few like OPOE > that appear as label "roots" and have many matches, suggesting the > "cat"/"concatanate" problem Guy mentioned], we see the following: > > Herbal A: 149 matches in 86 pages, for 1.733 per page > Herbal B: 83 " " " 25 pages, " 3.32 " " > Biological B: 127 " " " 20 pages, " 6.35 " " > >Note how there is a clear difference between frequency of label matches >in Herbal A and Biological B (although Herbal B also shows an increase >-- incidentally, Herbal B does somewhat differ from Biological B in its >relative letter frequencies). Shouldn't these match counts be adjusted to take into account the differing amounts of text on the pages? I have not looked at the numbers, but aren't there more words per page in Bio B than on Herbal A, and hence by chance alone you would expect more matches? In short: I would like to see not matche-per-page ratios but match-per-1000-V- letter ratios. Jim Reeds From baez@ucrmath.ucr.edu Fri Oct 30 02:59:33 1992 Date: Thu, 29 Oct 92 09:59:33 PST From: baez@ucrmath.ucr.edu (john baez) Message-Id: <9210291759.AA20111@ucrmath.ucr.edu> To: voynich@rand.org Subject: Plaudits Status: OR I feel (a little) guilty that I have been so inactive on the Voynich front... I guess I'm just a dilettante at heart... but I applaud Karl's recent post on labels in the astrological and herbal/biological sections. I am also very curious about the Minimum Spanning Tree code, and whether the Voynich text shows markedly less of this sort of structure than typical non-nonsense texts (of course there is no such thing as just one sort of "typical".) This seems like a very good thing to play with, but without controls one can't evaluate one's results. John B. From rcarter@nyx.cs.du.edu Fri Oct 30 10:58:46 1992 Date: Thu, 29 Oct 92 18:58:46 MST From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9210300158.AA06544@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: Karl.Kluge@G.GP.CS.CMU.EDU, voynich@rand.org Subject: Re: Label match frequency corrected for size Status: OR I was the one who had a posted a note on seeing one of the foldouts (dunno if it was the `big' one) in Wilfrid Voynich's article `A Preliminary Sketch of the History of the Roger Bacon Cipher Manuscript' printed in _Transcations of the College of Physicians of Philadelphia 43_ (20APR21)... I had remaerked on how it would be great to see the actual Voynich, because the photo shown in the above article showed the Voynich Ms. to be gorgeous; ink illustations not faded at all; the photocopy the Beinecke sells is OK for reading and transcription, but really gives no idea of how nice the original must be. Ron Carter | rcarter@nyx.cs.du.edu | 29 OCT 92 From kibo@world.std.com Fri Oct 30 15:11:43 1992 Date: Fri, 30 Oct 1992 01:11:43 -0500 From: kibo@world.std.com (James "Kibo" Parry) Message-Id: <199210300611.AA05796@world.std.com> To: voynich@rand.org Subject: Re: Plaudits Status: OR > I feel (a little) guilty that I have been so inactive on the Voynich Me too. I promised Jim Reeds long ago that I'd make him a nice Voynich fonts. Well, now I've got the equipment to do a really bang-up job in a fairly short time, but I haven't really had the chance to take a few days to do the actual drawing/digitizing. Someday... (I want to make a nice calligraphic one which actually *looks* like the real manuscript, although of course a bit clearer and cleaner. This would complement Jim's typewriter-like one nicely.) -- K. From reeds@gauss.att.com Sat Oct 31 01:05:00 1992 Message-Id: <199210301605.AA13455@rand.org> From: reeds@gauss.att.com Date: Fri, 30 Oct 92 11:05:00 EST To: voynich@rand.org Subject: Voynich film scanner initiative Status: OR Just out of the blue, with no surrounding context, let me ask this question: does anybody know of a film scanner which can handle standard 35mm microfilm reels? (Or strips of 35mm film with hundreds of frames: 5 or more meters, say.) Jim Reeds From Karl.Kluge@G.GP.CS.CMU.EDU Fri Oct 30 23:06:00 1992 Date: Fri, 30 Oct 1992 23:06-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Label expectations...statistically met and intellectually shattered Message-Id: <720504374/kck@G.GP.CS.CMU.EDU> Status: OR Consider a second-order model of the D'Imperio transcript, so that the probability of "xyz..." is p(x) * (p(xy) | x) * (p(yz) | y)..., where p(xy) is the probability that character y follows character x. Add the space into the alphabet so that the tendancy of certain characters to end "words" is included in the digraph statistics. Compute the probability for a given string, then multiply by the number of characters in the D'Imperio to produce the expected number of occurances. Switch to label matching without allowing spaces to be added in the middle (this avoids having to evaluate multiple paths through the digraph matrix). Here are the results for f68r1: ******************************************************** # Matches String [ OFO8AE9]: prob. 5.6844E-07, expect. 0.0457 0 String [ O8SCQ9]: prob. 3.3184E-07, expect. 0.0267 0 String [ OPSCO89]: prob. 6.3630E-06, expect. 0.5118 0 String [ OFOAE9]: prob. 2.2559E-07, expect. 0.0181 0 String [ SOY9]: prob. 1.1484E-06, expect. 0.0924 0 String [ OQC9]: prob. 3.4125E-05, expect. 2.7447 9 String [ OPZC9]: prob. 8.6997E-05, expect. 6.9972 8 String [ OP989]: prob. 1.7439E-05, expect. 1.4026 4 String [ OFCAR]: prob. 4.7305E-05, expect. 3.8048 8 String [ WOQ9]: prob. 1.2698E-06, expect. 0.1021 0 String [ 9PSO89]: prob. 1.1266E-05, expect. 0.9061 3 String [ OP92]: prob. 3.5250E-06, expect. 0.2835 1 String [ OEOR]: prob. 3.0653E-04, expect. 24.6542 37 String [ OX9]: prob. 4.0708E-04, expect. 32.7417 27 String [ CVSCOR]: prob. 6.7900E-07, expect. 0.0546 0 String [ OPS6O]: prob. 1.4694E-07, expect. 0.0118 0 String [ OP9FS2]: prob. 7.3714E-08, expect. 0.0059 0 String [ OPOE]: prob. 7.0858E-04, expect. 56.9922 99 String [ OPOR]: prob. 3.5086E-04, expect. 28.2204 39 String [ OMAR]: prob. 1.7389E-07, expect. 0.0140 0 String [ OFOE89]: prob. 2.3408E-05, expect. 1.8827 7 String [ 9FS89]: prob. 2.0413E-05, expect. 1.6419 4 String [ POCCO8Q9]: prob. 9.2353E-12, expect. 0.0000 0 String [ OW9]: prob. 5.0771E-05, expect. 4.0836 7 String [ OFCCO8AE]: prob. 1.0173E-06, expect. 0.0818 1 String [ OFZOR]: prob. 2.7949E-05, expect. 2.2480 3 String [ OR8AM]: prob. 1.1654E-05, expect. 0.9374 0 String [ OPOSC89]: prob. 1.4812E-06, expect. 0.1191 0 String [ 8OESC89]: prob. 5.6347E-06, expect. 0.4532 1 Given how crude the model of the text generation is, I'd say this looks pretty bad (as an initial result, anyways) for the hypothesis that the labels (and text) are meaningful, i.e. even if the excess occurances are statistically significant given the model I don't know that I'd consider that very compelling. I just did the Pisces zodiac folio labels. The results are similar (the number after "expected" on some lines indicates the number of matches in the D'Imperio where there were any). String [ OPAR/AJ]: p = 7.754E-07, 0.06 expected String [ OPAR/AE]: p = 3.016E-06, 0.24 expected 2 String [ OPAE/AR]: p = 2.130E-06, 0.17 expected String [ OPAE/AJ]: p = 5.060E-07, 0.04 expected String [ 8OEARAJ]: p = 1.540E-08, 0.00 expected String [ OFARAJ]: p = 6.886E-06, 0.55 expected String [ OPCO2AE]: p = 2.175E-07, 0.02 expected String [ 2AEOE2]: p = 1.341E-07, 0.01 expected String [ OFAE/8AE]: p = 6.938E-06, 0.56 expected 7 String [ 9FOEAM]: p = 8.425E-07, 0.07 expected String [ OPAEAE6]: p = 1.520E-09, 0.00 expected String [ 9FAR9]: p = 1.372E-05, 1.10 expected 1 String [ OPAR]: p = 7.725E-04, 62.13 expected 72 String [ OP9]: p = 2.107E-03, 169.47 expected 155 String [ OF9/O89]: p = 2.079E-05, 1.67 expected String [ OP9/AR]: p = 9.768E-06, 0.79 expected String [ OFAEA]: p = 4.075E-05, 3.28 expected String [ OPO89]: p = 1.051E-04, 8.45 expected 7 String [ OPAE8]: p = 4.037E-05, 3.25 expected 5 String [ OPAE8AR]: p = 2.506E-06, 0.20 expected 1 String [ OFA69]: p = 1.170E-06, 0.09 expected String [ OP92AJ]: p = 4.566E-08, 0.00 expected String [ SXC9]: p = 8.397E-05, 6.75 expected 14 String [ OPAE9]: p = 3.548E-05, 2.85 expected 6 String [ OPAE/ARAR]: p = 3.154E-08, 0.00 expected String [ OPAE89]: p = 2.240E-05, 1.80 expected 3 String [ OFCOE9]: p = 1.480E-05, 1.19 expected 1 String [ OF989]: p = 2.599E-05, 2.09 expected 2 String [ OFCC2]: p = 4.913E-05, 3.95 expected 5 Does this help drive the nail in the coffin of Voynich meaningfulness? Karl From rcarter@nyx.cs.du.edu Sat Oct 31 15:25:05 1992 Date: Fri, 30 Oct 92 23:25:05 MST From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9210310625.AA04058@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: Karl.Kluge@G.GP.CS.CMU.EDU, voynich@rand.org Subject: Re: Label expectations...statistically met and intellectually shattered Status: OR Ummm... Can you do the same stats on another (known to be `real') document (or better yet, document(s)) and see what the results are? I am sticking to (and working on) my theory that the Voynich is: A orally (dictated) phonetically transcribed translation from Aztec (Nuahatl) to Latin abbreviations, done by (at least) 2 writers (and at least two dictaters) who had different skills in translation and Latin abbreviations... Far fetched you say? You bet... Would such a thing show as `meaningful' text? Beats me... I guess I am trying to say, that `encoded' messages might not analyze to be `meaningful'... I wonder if Egyptian glyphs have much `content' on a `pure' analysis basis, for instance. Would shorthand (of any kind, but say Gregg's) analyze well? The VM is an awful lot of work done by a pre-1622 person, and it might have done as a hoax? Ummm... If it was a religious artifact, or purported to be the `true' necronomicon, maybe; but the VM is outstanding in that no one seems to have really cared for it all... No shrines; no stake burnings... I might be foolish this way, but sure looks `real' to me... Ron Carter | rcarter@nyx.cs.du.edu | 31 OCT 92 From reeds@gauss.att.com Sat Oct 31 15:37:29 1992 Message-Id: <199210310637.AA04960@rand.org> From: reeds@gauss.att.com Date: Sat, 31 Oct 92 01:37:29 EST To: voynich@rand.org Subject: Voynich Markov chains Status: OR About Karl's Markov chain stuff. Aren't these findings more or less as expected? The probability of any string, label or not, figured from digraphic frequency counts, will approximate the number of times the string occurs in the corpus used in forming the counts, will it not? (At least, assuming the corpus is well modelled by a first order Markov chain.) Almost as a tautology. I would be much happier if the digraphic probablility formulas were somehow combined with the earlier Herbal A/Herbal B/Bio B sub-corpus counts. (Not that I see how to do it myself.) This might give a surer indication of the relationship between the "label language" and the "sub- corpus languages" than Karl's previous recent label match count and rate tabulations. In general I think digraphic counts and models are going to be very useful. Jim Reeds From Karl.Kluge@G.GP.CS.CMU.EDU Sat Oct 31 17:35:00 1992 Date: Sat, 31 Oct 1992 17:35-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Re: Voynich Markov chains Message-Id: <720570907/kck@G.GP.CS.CMU.EDU> Status: OR > Aren't these findings more or less as expected? The probability of > any string, label or not, figured from digraphic frequency counts, > will approximate the number of times the string occurs in the corpus > used in forming the counts, will it not? (At least, assuming the > corpus is well modelled by a first order Markov chain.) Almost > as a tautology. Jim is right, of course. I have no idea what I was thinking. Here is roughly what I'm trying to figure out. Let's consider the following set of hypotheses about the underlying text: 1) There is an intelligible underlying text a) in a natural language (Latin, German, Aztec,...) b) in a fake natural language (Enochian, "Martian",...) c) in a code or artificial language such as Wilkins' 2) There isn't an intelligible underlying text a) glossalia b) random 1. psychologically "random" strings 2. mechanically random strings You have to take the cross product of this set with the set of hypotheses about the encrytion, of course, but I'm supposing that if one of the hypotheses under 2) is correct that there is no additional encipherment. Now suppose I delete all the lines from the D'Imperio transcription which contain a match to one of the Pisces labels (fgrep makes this easy enough to do), and compute the expected number of occurances for the labels using those digraph stats. After all, the total number of lines deleted is pretty small, so the overall digraph stats won't be affected all that much, but now _the number of times the target strings occur in the corpus which is used to form the digraph count is zero_. Does the degree of correlation between these expectations and the full corpus expectations suggest anything about whether the text is random or psuedo-random? Do "random" strings of letters constructed by people or glossialic utterances (as opposed to fake languages such as the "Martian" which we discussed a while back) show regularities in 1st or digraphs stats which could be tested against the D'Imperio transcript? Karl From reeds@gauss.att.com Sun Nov 01 10:27:18 1992 Message-Id: <199211010127.AA18396@rand.org> From: reeds@gauss.att.com Date: Sat, 31 Oct 92 20:27:18 EST To: voynich@rand.org Subject: Voynich hypotheses, etc Status: OR Karl writes > roughly what I'm trying to figure out. Let's consider the following set > of hypotheses about the underlying text: > > 1) There is an intelligible underlying text > a) in a natural language (Latin, German, Aztec,...) > b) in a fake natural language (Enochian, "Martian",...) > c) in a code or artificial language such as Wilkins' > 2) There isn't an intelligible underlying text > a) glossalia > b) random > 1. psychologically "random" strings > 2. mechanically random strings > > ... > > about whether the text is random or psuedo-random? Do "random" strings of > letters constructed by people or glossialic utterances (as opposed to fake > languages such as the "Martian" which we discussed a while back) show > regularities in 1st or digraphs stats which could be tested against the > D'Imperio transcript? > It seems there are several questions floating around. One, is the Voynich text well modelled by a digraphic model? (Leave Herbal A etc aside for the minute.) There is clearly a lot of digraphic structure, but is there other structure, too? Another question is whether the labels appear typical of the rest of the text. A special version of this is to suppose that a digraphic model gives good fit, and ask if the labels are typical of the digraphically modelled text. A third question is can one distinguish Enochian from glossolalia from Latin, etc, by looking at digraphic statistics. Question one is answerable by tabulations of Karl's sort. For any short string we can calculate how often it should appear according to the digraphic model and we can count how often it does appear. It is possible to construe the difference between trigraphic entropy (or higher order entropies) and digraphic entropy as a summary of such a tabulation, based on all length 3 strings, etc. One trouble with this approach is that there are so many possible short strings, and essentially none of them ever turn up. (This makes 4-th, 5-th, etc, entropy estimates essentially unreliable.) It would be nice to know, however, if there were any trigrams (or longer strings) which were vastly over or under represented. If there were, that would be evidence that a digraphic model does not tell the whole story. Question two has an easy answer, assuming a digraphic model. Work out the probablility of a label string, take its logarithm, divide by the string length. This number should be close to the negative of the entropy of the digraphic model. (This is AN answer, not THE answer.) I seem to remember reading, either in Bennett, or in some very early discussion in this group, some rumor to the effect that glossolalia IS well modelled by digraphic models. I would be suprised, because nothing else in nature is. Jim Reeds From kibo@world.std.com Sun Nov 01 14:32:32 1992 Date: Sun, 1 Nov 1992 00:32:32 -0500 From: kibo@world.std.com (James "Kibo" Parry) Message-Id: <199211010532.AA03959@world.std.com> To: voynich@rand.org Subject: Re: Voynich Markov chains Cc: sramming@athena.mit.edu Status: OR Karl Kluge's list of Voynich hypotheses: > 1) There is an intelligible underlying text > a) in a natural language (Latin, German, Aztec,...) > b) in a fake natural language (Enochian, "Martian",...) > c) in a code or artificial language such as Wilkins' > 2) There isn't an intelligible underlying text > a) glossalia > b) random > 1. psychologically "random" strings > 2. mechanically random strings You probably should add a category for my personal favorite hypothesis (I'm not sure whether you'd want it under "1" or "2" though.) Here it is: There is an intelligible, encrypted natural-language text *of no relation to the pictures*. In other words, if I wanted to produce an impressive-looking manuscript in a hermetic language, I'd grab whatever book was handiest (probably a Bible, since the Mss. is long and old) and simply copy it off in some easy cipher. Thus, any given 'chapter' of the Voynich could be, say, the middle third of Genesis. The difference I'm making with "1a" is that in my hypothesis, the text is meaningful but not in the supposed context... the manuscript would still be "fake" but would exhibit the same characteristics, under analysis, as a "real" document. (Of course, the existence of picture captions is handy for checking this kind of thing!) This is probably a trivium, but I just wanted to throw in my two cents' worth. -- K. From j.guy@trl.oz.au Mon Nov 02 23:51:21 1992 Date: Mon, 2 Nov 92 09:51:21 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9211012251.AA19251@medici.trl.OZ.AU> To: voynich@rand.org Subject: Re: Voynich hand A & B (clusters) Status: OR Perhaps a minimum-spanning tree is not appropriate in the case. I remember having tested minimum-spanning trees for reconstructing the phylogenies of artificially generated language families. The results were invariably bad. And then, there is the question of what is a proper measurement of distance. From j.guy@trl.oz.au Tue Nov 03 00:12:19 1992 Date: Mon, 2 Nov 92 10:12:19 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9211012312.AA19286@medici.trl.OZ.AU> To: voynich@rand.org Subject: Labels that do not appear in the text Status: OR Karl Kluge asks: 1) Labels and text generated by same process. Objection: then why are so many labels rare or nonexistant strings in the Herbal and Biological sections? The Herbal and Biological section labels might be names of plants and organs. Those in the astronomical section names of planets, constellations, and stars. If so, then the herbal and biological labels would be common nouns (carrot, pea, gut...), the astronomical ones proper nouns (Mars, Jupiter, Deneb...). Common nouns in Voynichese might be inflectable, proper nouns might not. For instance, common nouns might be inflectable for the plural, but proper nouns not. If inflections are by way of infixes (as some verbal forms in Arabic), or vowel variations (as is again the rule in Arabic, or as we see occasionally in English: goose, geese), the proper noun labels may become unrecognizable in the text. E.g. a label "goose", and the text referring to "geese" (plural) or "gowse" (accusative!). For that matter, Voynichese could be Old French, which had retained two cases, e.g. ber (nom.), and baron (acc.) "earl". You'd see a label "li ber" (the earl, nom.) and read in the text "le baron" (acc.) No hope of equating one with the other; the accusative was used for any case that was not nominative (genitive, dative, and after prepositions), so you are far, far more likely to encounter nouns in the accusative than in the nominative. If we find more matching strings in the B corpus, the reason could be that language B had lost much of the inflections that were still surviving in A. I'd say it's all compatible with a "real" text. From Karl.Kluge@G.GP.CS.CMU.EDU Wed Nov 04 03:28:00 1992 Date: Wed, 4 Nov 1992 03:28-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Musings on the possibility of a forgery... Message-Id: <720865736/kck@G.GP.CS.CMU.EDU> Status: OR (As a prelude, a quick list of the notes I took from the Tiltman essay when I looked at it in CMU's botanical library: Two suggested plant identifications: f4r -- cross-leaved heath (_erica_)? f56r -- bind-weed (_convolvulus_)? Re: Petersen transcript: "it occupied him for, I believe, 3 or 4 years." Strong "decipherment" -- interestingly enough, 1550 & 1555 ed. of _Bancke's Herbal_ attributed to Anthony Askham "declaring what herbs hath influence of certain stars and constellations" (although this material hasn't survived in any of the existing copies). End digression) I'm in one of my "it's a friggin' forgery" days, with the Voynich placed in an intellectual tradition which starts with Raymond Lull and goes through the Laputan academy's machine described by Swift -- generated line-by-line by a random technique. What put me in that mood was an examination of f78r, which led me to notice the "C89" phenomenon (Currier mentions an 89 difference): Herbal A: 5 lines in 1107 contain C89 Herbal B: 194 lines in 296 " " " Biolog B: 657 lines in 763 " " " I have trouble accepting that as normal variation between authors (incidentally, allowing spaces only adds 1 palty C/89 to the Herbal A count). (Pardon another aside, it's late -- while looking for matches to the f78r labels, I noticed these interesting matches to OFS89: 09705B 2OEFS9/SX9/4OFS89/4OFS89/OFAR/AR/9/4OFS89/FAR/AR/OFAN- 06603B...S89/SAR/OR/OE/FC8A3/SQ9/ OFS89/SX9/8ARAJ- 05003B...C89/X89/O8AR/SC89/9PC89/ OFS89/6- qp = P; lp = F 06508B ZOPS9/4OF9/OEF-SCX9-9FC89/4OFS89/2AR/OE8AJ- clpt = X; cqpt = Q 10707B ...FC89/SO89/FZ89/4OPCC89/4OFS89/OEFAR/8AJ- 15526B ...OEPCC89/4OPSC9/89FCC89/4OFS89/OBSC89/ZOE/OR9- Aside done...) In discussing labels in the context of a forgery, I suggested 3 theories: coincidence -- the matches are accidental and relate to the statistical properties of the process which generated the running text intrusive -- labels "salted" into the text as it was generated "extrusive" -- labels selected from the text after text generated (I doubt this given the number of labels not in the text) It occurs to me that it should be possible to choose between the coincidence and intrusion theories given the differences between Languages A & B. Here are three tests I can think of: How do the predictions from digraph stats of (corpus lines - match lines) for a set of labels track the actual matches for Herbal A, Herbal B, and Biological B? How do the frequency and digraph counts compare for lines with matches and lines without? How do the pre-label/label_char1 and label_last/post-label digraph stats compare with the normal digraph stats (i.e., how often do various characters appear before the O starting an label compared to how often they appear before and non-label-initial O?)? Comments? Karl From reeds@gauss.att.com Thu Nov 05 02:28:20 1992 Message-Id: <199211041728.AA25174@rand.org> From: reeds@gauss.att.com Date: Wed, 4 Nov 92 12:28:20 EST To: voynich@rand.org Subject: Voynich pictures revisited Status: OR I am thinking of sending this letter to the Marshall Library. Any comments? Am I misstating anything? Suppose they commission a digital scan onto a CD of the whole thing. It might cost $200 (at a buck a photgraph). If we take 2 pages per photograph it might take 400 pages to print the whole thing. At 4 cents a page that comes to $16 for printing. Assume a press run of 20, so scanning costs come to $10 per book. Would we pay $30 per book? Would we be able to do whatever it takes to turn a CD full of scanned photographs into a useable book: does anybody in our V-geek list know the techniques of digital pasteup? It might be possible to enhance the images some. It will be possible to cut and paste, so that we can make blowups of writing portions, etc. It would be nice to put Petersen nymph numbers onto the images, too. And folio numbers, etc. All this digital pasteup work would take time, however. ..LP Dear Mr. Cook, ..LP Does the Friedman collection have either Colonel Friedman's or Father Petersen's sets of photocopies of the Voynich manuscript? If either set is in the collection may I trouble you for the answers to a few questions? ..IP 1 How many photographs are in the set(s), how large are they, what condition are they in? ..IP 2 Would it be technically possible to make legibile copies of them? Would it make sense (from a technical perspective) to scan the original photocopies digitally? ..IP 3 If the answer to the above is ``yes,'' would the Marshall Library be willing to sell or publish high quality prints (or a CD holding high resolution digital scans) of these photocopies? ..LP The point of all of these questions is this: an informal group of researchers (many of us computer scientists and linguists, studying or working at universities and at various laboratories) are studying the Voynich manuscript's mysterious writing system. (Since our number includes Brigadier Tiltman's old colleague M. E. D'Imperio we consider our efforts to be a continuation of Colonel and Mrs Friedmans'.) ..LP In our work we make use of published pictures of individual pages of the Voynich MS, of the microfilm and Copyflow print sold by the Beinecke Library at Yale, and of the transcription by Father Petersen, a copy of which you provided me earlier in the year. ..LP Unfortunately the photographic quality of the Yale copy is (in places) terrible, and completely unsuitable for our use. Crisp, clear photographs of the wierd Voynich pages are what's needed as we transcribe the text into the computer. ..LP One solution would be to use the photocopies in your collection if they are good enough; another would be to commission a fresh set from Yale. The latter plan has disadvantages from our point of view: it would be much more expensive, it will take time to negotiate with Yale, and Yale would probably impose severe restrictions on the subsequent use of any photographs it sells. It is also possible that Yale might refuse altogether, hoping instead to strike a deal with a publisher for a coffee-table book edition of the Voynich manuscript. ..LP For us, an ideal situation would be for either the Marshall Library or for us to print a few dozen copies of a book, sold at cost, consisting of reproductions (at perhaps twice natural scale) of all of the Voynich photocopies in Friedman's set, with a small amount of added text per page. We would be willing to submit to restrictions prohibiting sale for commercial profit, but not prohibiting sale at cost for research use of items derived from any photgraphic or digital images you supply. ..LP Sincerely, ..sp .75 ..in +4i Jim Reeds. Jim Reeds From rcarter@nyx.cs.du.edu Thu Nov 05 03:48:18 1992 Date: Wed, 4 Nov 92 11:48:18 MST From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9211041848.AA23728@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: reeds@gauss.att.com, voynich@rand.org Subject: Re: Voynich pictures revisited Status: OR RE: getting a cd copy of the VM... I would assume that permission would have to be gotten by (rather from) the Beinecke as they own the VM... BTW, don't we know -anyone- that is at Yale who can go and look at the VM -in person-? I would like to know if the VM is anywhere the shape it seemed to be when Voynich wrote her paper on it. Ron Carter | rcarter@nyx.cs.du.edu | 04 NOV 92 From reeds@gauss.att.com Thu Nov 05 04:53:13 1992 Message-Id: <199211041953.AA01358@rand.org> From: reeds@gauss.att.com Date: Wed, 4 Nov 92 14:53:13 EST To: voynich@rand.org Subject: Voynich copyright, forsooth Status: OR Ron writes > I would assume that permission would have to be gotten by > (rather from) the Beinecke as they own the VM... I don't think so. Nothing like copyright applies to the VMS itself, for instance. (The author is long dead!) The Beinecke, as owner, can control access to the MS by photographers and others, and can thus enforce its monopoly on new photographs, which it sells only after the buyer has promised (contracted) not to further copy or sell. But photos made by an earlier owner, (before the Beinecke was even founded, I think), belong to whoever happens to own them. It is analogous to the prohibition against taking photos in most museums. You have to buy your postcards at the little gift shop, at inflated prices. But now it turns out that the public library across the street has some snaps of your favorite painting, made before the museum acquired the painting. Are you allowed to photocopy those? Sure, if the library lets you. Can the museum sue? What property do they own that you have taken or damaged? This, of course, is just my lay opinion. But I'd be suprised if a lawyer disagrees with anything but the manner of expression. The real problems, as I see it are: 1. the Friedman collection's photographs might not be clear enough to use, 2. the Marshall Library is unwilling to cooperate, or wants too big a cut of its own, or 3. the pasteup work is too hard for us. My prediction: 2 is no problem, 3 won't be after we think about it a bit, but 1 might well be a problem. Jim Reeds From ACW@RIVERSIDE.SCRC.Symbolics.COM Wed Nov 04 15:44:00 1992 Date: Wed, 4 Nov 1992 15:44-0500 From: Allan C. Wechsler Subject: Voynich pictures revisited To: reeds@gauss.att.com, voynich@RAND.ORG In-Reply-To: <199211041728.AA25174@rand.org> Message-ID: <19921104204457.3.ACW@PALLANDO.SCRC.Symbolics.COM> Status: OR Date: Wed, 4 Nov 1992 12:28 EST From: reeds@gauss.att.com I am thinking of sending this letter to the Marshall Library. Any comments? You misspelled "weird" as "wierd". Other than that, sounds good. From Doug.Brightwell@corp.sun.com Thu Nov 05 06:03:27 1992 Date: Wed, 4 Nov 92 13:03:27 PST From: Doug.Brightwell@corp.sun.com (Doug Brightwell) Message-Id: <9211042103.AA00472@media.Corp.Sun.COM> To: voynich@rand.org, reeds@gauss.att.com Subject: Re: Voynich pictures revisited Cc: Doug.Brightwell@corp.sun.com Status: OR Jim Reeds wrote... > For us, an ideal situation would be for either the Marshall > Library or for us to print a few dozen copies of a book, sold > at cost, consisting of reproductions (at perhaps twice > natural scale) of all of the Voynich photocopies in > Friedman's set, with a small amount of added text per page. Greetings from Lurker Land... I've had a layman's fascination with the Voynich manuscript and the work that your group has been doing for several months. I don't have the technical skills in cryptology and linguistics that you all seem to, so I've had nothing to contribute. In my 9-5 life, I'm a video producer in the Marketing Communications group at Sun Microsystems (UNIX workstation manufacturer) in Palo Alto, CA. I also do Macintosh-based design & desktop publishing on the side. I just recently started publishing a quarterly journal called "Hi-Tech Nomadness" (about nomadic computing, essentially). If I'm reading your email correctly, you're attempting to generate a "book" of photocopies for the time being. But for future reference, if you ever get to the point of wanting to produce something more graphically designed -- such as an inexpensive book or journal-size publication -- please consider me as a potential resource. I would love to design something that evokes the flavor and mystery of the document. Sincerely, Doug Brightwell From Karl.Kluge@G.GP.CS.CMU.EDU Thu Nov 05 04:20:00 1992 Date: Thu, 5 Nov 1992 04:20-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Back to "Kelly in the Drawing Room with Enochian tablets..." Message-Id: <720955202/kck@G.GP.CS.CMU.EDU> Status: OR Keeping on track with my current "it's bogus" mood, and considering Kelly as a possible culprit, here are what I see as the pros and cons of the "Kelly forged it theory": Pro: While the Enochian material shows significant differences from the Voynich, the complexity of the process of production would certainly seems to indicate that Kelly was capable of a project such as the Voynich. As a commented early on, if he was capable of inventing one language and alphabet, he was certainly capable of inventing two. On the con side, there are strong indications of an absolute minimum of two people involved in the production of the Voynich, possibly as many as six. (Given the indications of the MSS being a copying job such as qp's with the q in one "word" and the p in another, I've toyed with the theory that the text was generated by two people, A & B, then copied by the different "hands" Currier saw.) Who would Kelly's accomplices have been? It doesn't strike me as fitting his M.O. While the Voynich and Enochian material differ in that the Enochian was intended to come with an interliner translation, so to speak, the table-based method of generation of the calls intrigues me as offering a possible method for a Kelly-esque forger to generate a pseudo-text with a structured trigraph distribution. Consider a set of tables like a b a c e d f Generate each line by picking an initial --------------- character (say, d) and random column in d |c a a d f e b a row starting with that character (say, b |f b e a / c e the a in col 3 of row 1). That line ...etc would start "daa...". Now pick a row with a on the left (perhaps switching to another table), and repeat the process. The line-initial character frequencies will be related to their frequency on the left edge of the tables, and will differ from the line-interior frequencies. Put spaces explicitly in the tables ("...b/e..." in the table fragment I gave), and voila! a possible explanation for the cross-space correlation. A & B differ in how often they switch tables. I may try to construct such a set of tables to see if they can create the sort of "root" & "suffix" features Tiltman noticed (see extract from J. Guy message of 12Dec91 below), which seem to cry out for a table-driven generation method. Roots Suffixes OF-, OV- -AD, -AN, -AM, -A3 OP-, OB- -AR, -AT, -AU, -A0 4OF-, 4OV- -AE, -AG, -AH, -A1 4OP-, 4OB- -OR S- -OE Z- -C9, -CC9, -CCC9 8- -C89, -CC89, -CCC89 2- Fig. 27 -- Tiltman's Division of Common Words into "Roots" and "Suffixes" (Tiltman 1951) ... What does it all mean? This: take anything from the left column, whack on anything from the right column, and, Abracadabra, Hocus Pocus, Presto Shazam! here's a Voynich word. Now those of you who have studied Chinese (I am sure there are some) will have recognized there something very similar to the fanqie of the traditional analysis of Chinese words.... From kibo@world.std.com Thu Nov 05 14:02:31 1992 Date: Thu, 5 Nov 1992 00:02:31 -0500 From: kibo@world.std.com (James "Kibo" Parry) Message-Id: <199211050502.AA15658@world.std.com> To: voynich@rand.org Subject: Scanning the Voynich Status: OR excerpts from a letter I just sent to Jim Reeds: > of digital pasteup? It might be possible to enhance the images some. > It will be possible to cut and paste, so that we can make blowups of > writing portions, etc. It would be nice to put Petersen nymph numbers Scanned images of color pages will likely fall into one or two categories: 1.) Resolution too low to read the text well (i.e. 72dpi or video resolution) 2.) High-resolution (at least 200dpi, which is fax resolution), which would make the text clear and would permit enlarging it, but which would be a *huge* file size. Let's say you want to capture all the color, shading, etc., in the original, to get all the pictures' color and to permit enhamcement. This would probably be a 24-bit color image. A 72dpi (Mac screen resolution) 8.5x11" page is 1.5 megabytes in 24-bit color; A 150dpi 8.5x11" page is 6 meg in 24-bit color; A 300dpi 8.5x11" page is 24 meg in 24-bit color (unmanageable size unless you're really equppied for high-end desktop publishing) In black and white (no grays--shading in pictures would be accomplished by dithering): A 72dpi 8.5x11" page is 000 K in 1-bit b/w (illegible text) A 150dpi page is 250K in 1-bit b/w (barely legible text) A 300dpi page is 1 meg in 1-bit b/w (this is a normal laserprinter image) These file sizes compress, of course, but of course you still have to be able to meet the memory/disk space requirements of the decompressed versions to work on these images. There's also overhead; a 1-meg TIFF in Adobe Photoshop occupies *five* megabytes of swap space while you're working on it. A 24-meg scan is the sort of thing where you can tell the computer to load it, go out for a snack, and be back before it's on the screen... The best way to do it, probably, would be to scan the text as 300dpi b/w images and the pictures as 72 to 150dpi 24-bit color images; having more resolution for the color isn't necessary (since you almost never use a halftone finer than 150 lines per inch when printing) and having more than one bit-plane for the text isn't necessary if you have 300dpi resolution. Of course, in a mixed format like this, the text and picture images would either be distributed separately (perhaps 300dpi b/w scans of complete pages and 72dpi color scans of just the pictures?) or else assembled in some desktop publishing format, which would require specific hardware to use (bad option.) -- K. From reeds@gauss.att.com Fri Nov 06 01:32:25 1992 Message-Id: <199211051632.AA00806@rand.org> From: reeds@gauss.att.com Date: Thu, 5 Nov 92 11:32:25 EST To: voynich@rand.org Subject: Voynich text generation by tables Status: OR Karl brings up the use of tables to generate texts with marked digraphic or trigraphic statistical properties: >... >with a structured trigraph distribution. Consider a set of tables like > > a b a c e d f Generate each line by picking an initial > --------------- character (say, d) and random column in > d |c a a d f e b a row starting with that character (say, > b |f b e a / c e the a in col 3 of row 1). That line > ...etc would start "daa...". Now pick a row > with a on the left (perhaps switching to >another table), and repeat the process. The line-initial character >... Presumably the entries in the body of the table need not be restricted to be single letters but could be short strings as well? This indeed meshes well with Tiltman's root & suffix lists, which can easily be fit into a tabular form (take one from column A, another from column B, and an ending from column C, as if it were a restaurant order: "New England, Ranch, Rare, and A La Mode, please, with cream"). I had been supposing a scheme where a "reference text" of a few lines was always in front of the composer, who picks out snippets reading left to right. Thus, each Voynich line would be derived from part of the reference text with a different subset of letters, syllables or words deleted. Thus. With the above paragraph as a reference text, one derived "Voynich" line might be: habee a where few li alpick snipleft tonich wou part and others: Isupp asheref linewapick outsnip leftto thus entset Isupp text linewas front outsnip reading left thus line and so on. The reference text could be something meaningful, such as a little bit of Trithemian "Boaziel submitsion alotiel ofjunkargue scimathablob" cribbed from the Steganographia, or the Lord's Prayer written out backwards, or possibly even just a random string of V characters. This mechanism would explain the "approximate vertical copying" which I think I see on many pages, where the same (or very similar) words occur is approximately the same places in adjacent lines. Jim Reeds From j.guy@trl.oz.au Fri Nov 06 05:01:57 1992 Date: Thu, 5 Nov 92 15:01:57 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9211050401.AA23801@medici.trl.OZ.AU> To: voynich@rand.org Subject: Voynich picture revisited Status: OR Perhaps it would be better to have Currier or D'Imperio themselves write to the Marshall Library if they would? I fear that anything related to the Voynich manuscript is seen as guilty of crackpottery until proven innocent. Or perhaps again go through one of the editors of Cryptologia? They are... David Kahn, 120 Wooleys Lane, Great Neck NY 11023... David Kahn??? THE David Kahn of "Codebreakers"? Must be. Now, *that* would seem unimpeachably serious and professional, if David Kahn... Perhaps even the Staunch Keepers at the Beidecke Library would relent? From Karl.Kluge@G.GP.CS.CMU.EDU Fri Nov 06 03:34:00 1992 Date: Fri, 6 Nov 1992 03:34-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Dee, Kelly, the Browne quote, Laycock, and Enochian Message-Id: <721038842/kck@G.GP.CS.CMU.EDU> Status: OR In reconstructing the history of the Voynich and attempting to tie it to Dee, the following is quoted (in the Krauss catalog) from Thomas Browne (citing Arthur Dee): "a book...containing nothing but hieroglyphics; which book his [Arthur Dee's] father bestowed much time upon, but I could not hear that he could make it out...." Voynich in his talk to the College of Physicians of Philadelphia gives basically the same ellided quote. Let me restore some of the context to the quote: "The transmutation [to gold] was made by a powder they had, which was found in some old place, and a book lying by it containing nothing but hieroglyphicks..." In other words, if this is the Voynich (and based on all the Dee material we've discussed, it seems to be the only candidate), it was brought to Dee by Kelly. It is hard to interpret the speculations about the Duke of Northumberland as anything more than wild wishful thinking by Voynich and Co. trying to tie the Mss. to Bacon when the quote by Browne is so clear. In looking through Whitby's _John Dee's Actions With Spirits..._. On page 147 it says, "In his preface to his Enochian dictionary Donald Laycock argues that there are two versions of the angelic language: the first consists of the two series of 49 calls that are in Sloane MS 3188 and the second is the series of calls with interlineated translation that are in _48 Claues angelicae_. He reaches this conclusion on the grounds that the first series of calls are largely pronounceable and that some of the texts run fluently with much phonetic patterning such as repetition rhyme and alliteration, while the second version of the language is taken letter by letter from the tables of the _Book of Enoch_ and 'is less pronounceable than the old [language], and it has awkward sequences of letters, such as long strings of vowels (_ooaona_, _mooah_) and difficult consonant clusters (_paombd_, _smnad_, _noncf_)' (p. 40). It is altogether 'the type of text produced if one generates a string of letters on some random pattern' (p. 40)." While the Voynich is clearly not in Enochian, it is reasonable to suppose that if Kelly invented two languages, he would tend to conform to a pattern in his constructions. I think an examination of the Enochian vocabulary for Voynichesque properties might be fruitful. I have four of the Calls on line, if anyone has the whole corpus let me know (I couldn't find the Enochian material on the ftp site mentioned in alt.magick in their recent discussion). Jacques -- could you elaborate on Laycock's comments about the structure of Enochian? From rcarter@nyx.cs.du.edu Sat Nov 07 03:31:56 1992 Date: Fri, 6 Nov 92 11:31:56 MST From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9211061831.AA15792@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: Karl.Kluge@G.GP.CS.CMU.EDU, voynich@rand.org Subject: Re: Dee, Kelly, the Browne quote, Laycock, and Enochian Status: OR Good ideas on comparing the VM structure and Enochian... Be aware, that among other things, thr root structure of Enochian is discernable, and the time scale would say the VM predates the Enoch work... Not saying the comparison shouldn't be made, but I maintain the links `twixt the VM and Dee (brought to? written by?) via Kelly is tenuous at best, and further stretching it to say that the VM was brought to Rudolph's court by Kelly. I would propose that Kelly had faked a manuscript that showed how to transmutate base metals into gold. This manuscript would have been pure hieroglyphic nonsense, that only Kelly could read. This same fake was brought to Rudolph by Kelly, and got Kelly locked up when he couldn't produce gold by it. Kelly strikes me as being on hundred percent con-man, and not even particularly smart at that; the Enoch work is amateur at best. Why isn't the VM the `hieroglyphic' work of Kelly? The VM doesn't appear to deal with transmutation at all; it appears to be a very unusual herbal/biological/astrological manual of sorts. Granted, it might be fake; but for what purpose; religous/treasure/magic manuscripts/artifacts are much easier to fake -and- sell. If the VM -had- been the `hieroglyphic' work of Kelly, I think that either Dee would have written in his catalogs `one straynge and myfteriouse manuscrft wifh drawinges offe Angeles, herbes, and starres aboundeing' or the court of Rudolph would have it mentioned as such as well, as being brought as a gift from Dee. Granted, it appears Rudolph paid Kelly for a manuscript, most likely the hieroglyphic ms, but I think that is was coincidence that the VM shows up in the hands of Rudolph's physician later on. After all, all sorts of strange things landed in that court. Regards, Ron | rcarter@nyx.cs.du.edu | 06 NOV 92 From Karl.Kluge@G.GP.CS.CMU.EDU Sat Nov 07 04:32:00 1992 Date: Sat, 7 Nov 1992 04:32-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Whoops -- the Book of Dunstan not the Voynich Message-Id: <721128768/kck@G.GP.CS.CMU.EDU> Status: OR Thanks to Kelly having a little accident with an oil lamp, we have description of the book that Kelly "found" with the powder -- it was 40 leaves in quarto (Diary entry for 12 Dec 1597). From nelson@reed.edu Sun Nov 08 10:07:00 1992 Message-Id: Date: Sat, 7 Nov 92 17:07 PST From: nelson@reed.edu (Nelson Minar) To: voynich@rand.org Subject: more GIFs? Status: OR (Hello, just joined the list. My interest in the Voynich manuscript, so far, is purely curiousity.) There are a couple of GIFs of pages of the Voynich manuscripts available on the ftp site. They're very beautiful. I want more! Is anyone with a high quality scanner willing to scan a few more pages and make them available? I'm willing to scan a few, but the only scanner I have available to me is a 16 grey poor quality one. I'm going to give it a shot anyway (a friend has a few photocopies of pages), but I suspect that someone else could do better. From Karl.Kluge@G.GP.CS.CMU.EDU Mon Nov 09 02:08:00 1992 Date: Mon, 9 Nov 1992 02:08-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Enochian entropy tidbit Message-Id: <721292917/kck@G.GP.CS.CMU.EDU> Status: OR I typed in the names of the 30 Aethyrs and their 91 Govenors, as well as the names of the 49 Angels of the Bonorum. The second set of names is particularly interesting, as the names all start with "B", remininiscent of the tendancy of Voynich labels to start with "O". I had the first three of the Enochian Calls online, so I tested them as well. Here are some per-character entropies: Voynich text: 3.856 Voynich labels: 3.821 Aeth,Gov names: 4.026 Bonorum names: 3.830 Enoch call 1-3: 3.986 I haven't had a chance to look at the digraph structure yet for the sort of initial-medial-final structure Voynich has, but it's interesting that Kelly's method for producing bogus angel names and the Enochian call vocabulary produces strings with just about the same per-character entropy as the Voynich labels and text. The table for generating the names of the angels of the Bonorum comes from _De heptarchia mystica_, which according to Whitby "is essentially a collation of the information given in the Actions between 29 April and 21 November 1582". Karl From Karl.Kluge@G.GP.CS.CMU.EDU Tue Nov 10 02:52:00 1992 Date: Tue, 10 Nov 1992 02:52-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Enochian transcriptions Message-Id: <721381975/kck@G.GP.CS.CMU.EDU> Status: OR Neither the Enochian calls or the words used in them show the sort of distinctive digraph features that Voynich "words" do. Oh well. Below is the raw data I transcribed. 000xxA: Sloane 3188, f74a 00YxxA: Enochian call Y, as transcribed by Robert Turner in _Elizabethan Magic_. Line divisions are not as per the version in Turner. 00001A /GESCO/A/TAFFOM/GES/NAT/GAM/PAMPHE/ORDAQUAF/CESTO/KIDMAP/MISCHNA/IAIAG/ 00002A /IAIALPAZUDPH/A/DAMSET/VNBAN/CAF/RANSEMBLOH/DAFNA/VP/ASCHEM/GRAS/ 00003A /CHRAMSA/ASCO/DAH/VIANA/GEN/ALDE/OS/PAPEAM/OCH/LAUAN/VNAD/OH/DROSAD/ 00004A /UDRIOS/NAGEL/PANZO/AB/SESCU/VORGE/AFCAL/VALAFFDA/MORMAB/GAF/HAM/DE/ 00005A /PELEH/ASCA/ 00101A /OL/SONF/VORSG/GOHO/IAD/BALT/LANSH/CALZ/VONPHO/SOBRA/ZOL/ROR/I/TA/ 00102A /NAZPSAD/GRAA/TA/MALPRAG/DS/HOLQ/QAA/NOTHOA/ZIMZ/OD/COMMAH/TA/NOBLOH/ 00103A /ZIEN/SOBA/THIL/GNONP/PRGE/ALDI/OD/VRBS/OBOLEH/GRSAM/CASARM/OHORELA/ 00104A /CABA/PIR/OD/ZONSRENSG/CAB/ERM/IADNAH/PILAH/FARZM/ZURZA/ADNA/DS/GONO/ 00105A /IADPIL/DS/HOM/OD/TOH/SOBA/IPAM/LU/IPAMIS/DS/LOHOLO/VEP/ZOMD/POAMAL/OD/ 00106A /BOGPA/AAI/TA/PIAP/PIAMOI/OD/VAOAN/ZACARE/CA/OD/ZAMRAN/ODO/CICLE/QAA/ 00107A /ZORGE/LAP/ZIRDO/NOCO/MAD/HOATH/IAIDA/ 00201A /ADGT/VPAAH/ZONGOM/FAAIP/SALD/VIIV/L/SOBAM/IALPRG/IZAZAZ/ 00202A /PIADPH/CASARMA/ABRAMG/TA/TALHO/PARACLEDA/QTA/LORSLQ/TURBS/OOGE/ 00203A /BALTCH/GIUI/CHIS/LUSD/ORRI/OD/MICALP/CHIS/BIA/OZONGON/LAP/NOAN/TROF/ 00204A /CORS/TAGE/OQ/MANIN/IAIDON/TORZU/GOHEL/ZACAR/CA/CNOQOD/ZAMRAN/ 00205A /MICALZO/OD/OZAZM/VRELP/LAP/ZIR/IOIAD/ 00301A /MICMA/GOHO/PIAD/ZIR/COMSELH/A/ZIEN/BLAB/OS/LONDOH/NORZ/CHIS/OTHIL/ 00302A /GIGIPAH/VNDL/CHIS/TAPUIM/Q/MOSPLEH/TELOCH/QUIIN/TOLTORG/CHIS/I/ 00303A /CHIS/GE/M/OZIEN/DST/BRGDA/OD/TORZUL/I/LI/F/OL/BALZARG/OD/AALA/THILN/ 00304A /OS/NE/TA/AB/DLUGA/VOMSARG/LONSA/CAPMIALI/VORS/CLA/HOMIL/COCASB/ 00305A /FAFEN/IZIZOP/OD/MI/I/NOAG/DE/GNETAAB/VAUN/NANAEEL/PANPIR/MALPIRGI/ 00306A /CAOSG/PILD/NOAN/VNALAH/BALT/OD/VOOAN/DO/OIAP/MAD/GOHOLOR/GOHUS/ 00307A /AMIRAN/MICMA/IEHUSOZ/CACACOM/OD/DOOAIN/NOAR/MICAOLZ/AAIOM/ 00308A /CASARMG/GOHIA/ZACAR/VNIGLAG/OD/IMUAMAR/PUGO/PLAPLI/ANANAEL/Q/A/AN/ 00401A /OTHIL/LASDI/BABAGE/OD/DORPHA/GOHOL/G/CHIS/GE/AUAUAGO/CORMP/PD/ 00402A /DSONF/VI/VDIV/CASARMI/OALI/MAP/M/SOBAM/AG/CORMPO/CRPL/CASARMG/ 00403A /CRO/OD/ZI/CHIS/OD/VGEG/DST/CA/PI/MALI/CHIS/CA/PI/MA/ON/IONSHIN/ 00404A /CHIS/TA/LO/CLA/TORGU/NOR/QUASAHI/OD/F/CAOSGA/BAGLE/ZI/RE/NAI/AD/ 00405A /DSI/OD/APILA/DO/O/A/IP/QAAL/ZACAR/OD/ZAMRAN/OBELISONG/RESTEL/AAF/ 00406A /NORMOLAP/ 00501A /SA/PAH/ZIMII/DUIV/OD/NOAS/TAQUANIS/ADROCH/DORPHAL/CA/OSG/OD/FAONTS/ 00502A /PERIPSOL/TABLIOR/CASARM/AMIPZI/NA/ZARTH/AF/OD/DLUGAR/ZIZOP/ZLIDA/ 00503A /CAOSGI/TOL/TORGI/OD/ZCHIS/E/SI/ASCH/L/TA/VI/U/OD/IAOD/THILD/DS/ 00504A /PERAL/HUBAR/PE/O/AL/SOBA/CORMFA/CHIS/TA/LA/VIS/OD/QCOCASB/CA/NILS/OD/ 00505A /DARBS/Q/A/AS/FETHARZI/OD/BLIORA/IAIAL/ED/NAS/CICLES/BAGLE/GE/IAD/ 00506A /I/L/ 00601A /GAH/S/DI/U/CHIS/EM/MICALZO/PIL/ZIN/SOBAM/EL/HARG/MIR/BABALON/OD/ 00602A /OBLOC/SAMVELG/DLUGAR/MALPRG/ARCAOSGI/OD/ACAM/CANAL/SO/BOL/ZAR/ 00603A /FBLIARD/CAOSGI/OD/CHIS/A/NE/TAB/OD/MIAM/TA/VI/V/OD/D/DARSAR/SOL/PETH/ 00604A /BI/EN/B/RI/TA/OD/ZACAM/G/MI/CALZO/SOB/HA/HATH/TRIAN/LU/IA/HE/ODECRIN/ 00605A /MAD/Q/A/A/ON/ 00701A /R/A/AS/ISALMAN/PARA/DI/ZOD/OE/CRI/NI/AAO/IAL/PURGAH/QUI/IN/ENAY/ 00702A /BUTMON/OD/IN/OAS/NI/PARA/DIAL/CASARMG/VGEAR/CHIRLAN/OD/ZONAC/LU/CIF/ 00703A /TIAN/CORS/TO/VAUL/ZIRN/TOL/HA/MI/SOBO/LONDOH/OD/MIAM/CHIS/TAD/O/DES/ 00704A /VMADEA/OD/PIBLIAR/OTHIL/RIT/OD/MIAM/C/NO/QUOL/RIT/ZACAR/ZAMRAN/ 00705A /OECRIMI/Q/A/DAH/OD/O/MI/CA/OLZ/AAIOM/BAGLE/PAP/NOR/ID/LUGAM/LONSHI/ 00706A /OD/VMPLIF/VGEGI/BIGLIAD/ 00000A - 00012A: names of the Thirty Aires (3 letter names) and their 91 Governors. 00013A - 00019A: Names of the Angels of the Bonorum. 00000A /LIL/OCCODON/PASCOMB/VALGARS/ARN/DOAGNIS/PACASNA/DIALIVA/ZOM/SAMAPHA/ 00001A /VIROCHI/ANDISPI/PAZ/THOTANF/AXZIARG/POTHNIR/LIT/LAZDIXI/NOCAMAL/ 00002A /TIARPAZ/MAZ/SAXTOMP/VAVAAMP/ZIRZIRD/DEO/OBMACAS/GENADOL/ASPIAON/ZID/ 00003A /ZAMFRES/TODNAON/PRISTAC/ZIP/ODDIORG/CRALPIR/DOANZIN/ZAX/LEXARPH/ 00004A /COMANAN/TABITOM/ICH/MOLPAND/VSNARDA/PONODOL/LOE/TAPAMAL/GEDOONS/ 00005A /AMBRIOL/ZIM/GECAOND/LAPARIN/DOCEPAX/VTA/TEDOAND/VIVIPOS/OOANAMB/OXO/ 00006A /TAHANDO/NOCIABI/TASTOXO/LEA/CUCARPT/LAUACON/SOCHIAL/TAN/SIGMORF/ 00007A /AYDROPT/TOCARZI/ZEN/NABAOMI/ZAFASAI/YALPAMB/POP/TORZOXI/ABAIOND/ 00008A /OMAGRAP/KHR/ZILDRON/PARZIBA/TOTCAN/ASP/CHIRSPA/TOANTOM/VIXPALG/LIN/ 00009A /OZIDAIA/PARAOAN/CALZIRG/TOR/RONOAMB/ONIZIMP/ZAXANIN/NIA/ORCAMIR/ 00010A /CHIALPS/SOAGEEL/VTI/MIRZIND/OBUAORS/RANGLAM/PES/POPHAND/NIGRANA/ 00011A /BAZCHIM/ZAA/SAZIAMI/MATHULA/ORPANIB/BAG/LABNIXP/FOCISNI/OXLOPAR/RII/ 00012A /VASTRIM/ODRAXTI/GOMZIAM/TEX/TAOAGLA/GEMNIMB/ADVORPT/DOZINAL/ 00013B /BALIGON/BORNOGO/BAPNIDO/BESGEME/BLUMAPO/BMAMGAL/BASLEDF/BOBOGEL/ 00014B /BEFAFES/BASMELO/BERNOLE/BRANGLO/BRISFLI/BNAGLOE/BABALEL/BUTMONOM/ 00015B /BAZPAMA/BLINTOM/BRAGIOP/BERMALE/BONEFON/BYNEPOR/BLISDON/BALCEOR/ 00016B /BELMARA/BENPAGI/BARNAFA/BMILGES/BNASPOL/BRORGES/BASPALO/BINODAB/ 00017B /BARIGES/BINOFON/BALDAGO/BNAPSEN/BRAGLES/BORMILA/BUSCNAB/BMNINPOL/ 00018B /BARTINO/BLIIGAN/BLUMAZA/BAGENOL/BABLIBO/BUSDANA/BLINGEF/BARFORT/ 00019A /BAMNODE/ From reeds@gauss.att.com Tue Nov 10 13:08:48 1992 Message-Id: <199211100408.AA17935@rand.org> From: reeds@gauss.att.com Date: Mon, 9 Nov 92 23:08:48 EST To: voynich@rand.org Subject: Voynich images Status: OR Nelson Minar recently asked: > There are a couple of GIFs of pages of the Voynich manuscripts > available on the ftp site. They're very beautiful. I want more! > > Is anyone with a high quality scanner willing to scan a few more pages > and make them available? I'm willing to scan a few, but the only > scanner I have available to me is a 16 grey poor quality one. I'm > going to give it a shot anyway (a friend has a few photocopies of > pages), but I suspect that someone else could do better. This, after a year of effort, is still one of our big problems. We just DON'T have access to high quality copies of the Voynich MS. Here is a summary of Voynich images. (Many details can be found in my 'finder.Z' document on rand.org:/pub/jim.) 1. W. Voynich, the modern discoverer of the manuscript, had several sets of good photographs made. One set ended up at NSA, at least one set was in the possession of W. F. Friedman. Maybe there were others. I don't know how many of these sets still exist. (I am making inquiries.) The plates in Newbold's book came from these photographs. 2. The next owner of the VMS, H.P.Krauss, was camera shy. He published several pages in several of his catalogs (I have seen his "35 Manuscripts", it is probably the best source of clear V images) but made no sets of photos of the whole MS available to scholars. 3. Yale, the current owner of the VMS has (1) made a perfunctory microfilm and sells execrable "Copyflow" prints therefrom, and (2) has allowed better quality copies of a few pages to appear in books. The best of this lot are in Blunt and Raphael; Brumbaugh's book has quite a few of worse quality printed on rough paper ; Bennett has one or two, and Poundstone (source of f79v.gif) has one. When you buy a copy of a Yale photo you sign an agreement to the effect that you will not copy the photo. (It is possible, but unethical, to scan copies of the Yale microfilm. One old timer left his copy to Lou Kruh in his will, Kruh lent me that copy, and I did not feel bound by any contract the decedent might have signed. So I plopped it in a film scanner -- as opposed to a flat- bed scanner -- and (1) had to take extraordinary measures to fit the 200 exposure microfilm onto the spindles of a contraption designed for 24 or 36 exposures, and (2) bust a disk storing the scan of f6v/f7r.) 4. The British Museum made a microfilm from (I believe) Brigadier Tiltman's set of Voynich prints made by Voynich. (Or maybe Tiltman borrowed the NSA set; I don't know.) Jim Gillogly has a copy of that microfilm, and has made prints from it, and has scanned one of them. That's f3v.gif. Not very bad, given the number of intemediate copies. 5. And Father Petersen made a hand written copy of the whole VMS, based on Voynich photographs and on inspection of the MS itself. Alas, it does not xerox terribly well. Conclusion: if you want to get an overall impression of what the VMS looks like, there are enough good pictures and lots of bad pictures. But if you want to transcribe from very clear prints showing every detail worth seeing, forget it. You will need aspirin. In an ideal world, Yale would see as its number one duty to the world of scholarship the publication of a 10,000 dots per inch multispectral scan (soft X rays down through S- and P-waves, in addition to the usual ultra- vibgyor and rumble-red), all on a 10 disk CD set. Also would fund a ``Voynich Studies Institute'' with 50 resident scholars with tenure for life, and with a branch in Victoria. (Seminars alternate between New Haven and Melbourne, with free shuttle service provided.) Jim Reeds From Michael.Roe@cl.cam.ac.uk Wed Nov 11 04:39:02 1992 To: reeds@gausS.att.com cc: voynich@rand.org Subject: Re: Voynich images In-reply-to: Your message of Mon, 09 Nov 92 23:08:48 -0500. <199211100408.AA17935@rand.org> Date: Tue, 10 Nov 92 19:39:02 +0000 From: Mike Roe Message-ID: <"nene.cl.ca.234:10.11.92.19.39.07"@cl.cam.ac.uk> Status: OR > 4. The British Museum made a microfilm from (I believe) Brigadier Tiltman's > set of Voynich prints made by Voynich. (Or maybe Tiltman borrowed the NSA > set; I don't know.) Jim Gillogly has a copy of that microfilm, and has > made prints from it, and has scanned one of them. That's f3v.gif. > Not very bad, given the number of intemediate copies. I believe that it was John Manly's set of prints of the VMS. I have a copy of the BL microfilm, and the quality is quite good. (Although the BL wrote a very nice letter saying that they regretted the poor quality, but it was the best they could do with the copies they had). I've tried making prints from the microfilm (for my own personal use, and other such standard disclaimers) but the prints I've made have been much worse than the film. It may be possible to do better by using a film scanner and doing some computer enhancement of the image afterwards. (Of course, I would need permission from the BL before giving the result to a third party). The main reason I haven't tried scanning the film is that I don't have a film scanner! If any members of this group in the UK have access to a film scanner, I'd be interested in hearing from them. I'll ask around Cambridge and see if some University department has one. If it turns out that we can produce reasonable scans this way, then we can try asking the British Library for permission to make them available to others (subject to whatever conditions of use the BL might impose). Michael Roe Cambridge University Computer Laboratory From jim@rand.org Wed Nov 11 04:57:41 1992 Message-Id: <9211101957.AA00698@mycroft.rand.org> To: voynich@rand.org Subject: Re: Voynich images In-Reply-To: Your message of Tue, 10 Nov 92 19:39:02 +0000. <"nene.cl.ca.234:10.11.92.19.39.07"@cl.cam.ac.uk> From: Jim Gillogly Reply-To: jim@rand.org Date: Tue, 10 Nov 92 11:57:41 PST Sender: jim@mycroft.rand.org Status: OR I'll echo Mike Roe's offer on this side of the water -- if any US researcher has access to a film scanner, I'd like to work out some arrangement with my copy of the BL film. Note that the size of the images is pretty huge, and I can't make it all ftp-able from RAND. We'll need to find a site that has more room available. The current small page (f3v.gif) would give acceptable resolution if done directly from the film (that was via a microfilm printer before the scanning) but the 96K f79v.gif (a full page) is not acceptable -- needs to be considerably more resolution. I have a 1.4M gif of it which looks better. Hurm. That's for black-and-white (gray scale). The film is not in color. Obviously a color copy would be much preferable, but would increase the file size further. Jim Gillogly U.S. National Debt: $4,128,230,894,995 Your Share: $16,142 (U.S. only -- overseas users are free) From reeds@gauss.att.com Wed Nov 11 21:00:07 1992 Message-Id: <199211111201.AA29564@rand.org> From: reeds@gauss.att.com Date: Wed, 11 Nov 92 07:00:07 EST To: voynich@rand.org Subject: Voynich prints in Chicago? Status: OR I was wrong: the British Museum filmed J. Manley's set of Voynich prints, not Tiltman's. I wonder where Manley's prints are now? Last I looked, Manley's papers were in the Regenstein Library at U. Chicago; maybe his Voynich prints are among his other papers? Is anyone reading this within visiting distance of Chicago? Jim Reeds From j.guy@trl.oz.au Wed Nov 11 21:36:17 1992 Date: Wed, 11 Nov 92 07:36:17 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9211102036.AA05154@medici.trl.OZ.AU> To: voynich@rand.org Subject: Re: Back to "Kelly in the Drawing Room with Enochian tablets..." Status: OR Karl Kluge: "Now those of you who have studied Chinese (I am sure there are some) will have recognized there something very similar to the fanqie of the traditional analysis of Chinese words...." That's precisely what made dream up the "Marco Polo hypothesis", in which I don't believe... but do not disbelieve too firmly either. Except for that label-initial , which rules out Chinese. To come back to the method of production of Enochian, Donald Laycock, who had been through all Dee's notes and diaries with a fine comb, could not figure out how the text was generated. Dee's notes tell how "angels" instructed them to draw up mammoth tables (some 49x49) and go through a complicated rigmarole of picking letter (in the seventeenth column...etc). The resulting language, at any rate, has nothing of the strange properties of Voynechese. On the other hand, Voynechese might well have been generate using the tabular method described by Karl, it seems to me. But then, so could Chinese, a language which we *know* is for real! The fact that, under that hypothesis, there were at least two generating tables (one for language A, and one for B), does not help at all deciding whether the language is real or not: you can generate Mandarin words with such a table, and Shanghaiese ones with another table, with different "roots" and "suffixes" e.g. Mandarin zh-an l-an h-ui (ch-an l-an h-wei if you prefer Giles-Wade), Shanghai z-e l-e w-e. The three 'words' are etymologically the same, only pronounced very differently. So it Voynichese could still be the result of a random-generating process as well as of a real, natural evolutionary one. From j.guy@trl.oz.au Wed Nov 11 22:10:31 1992 Date: Wed, 11 Nov 92 08:10:31 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9211102110.AA05214@medici.trl.OZ.AU> To: voynich@rand.org Subject: Re: Dee, Kelly, the Browne quote, Laycock, and Enochian Status: OR Karl Kluge asks me: Jacques -- could you elaborate on Laycock's comments about the structure of Enochian? Oh, yes! You won't find this in Laycock's book, because you just don't write these things to an audience of people most of whom believe Enochian is for real. Don used to say that the phonology of Enochian was straight Elizabethan English, i.e. the English of Dee's times. Only the spelling of Enochian makes it look un-English. For instance, the word "prge", he reconstructed as pronounced like modern "purge", "qting" as "kwutting" (I'm using English-like spelling here, his transcription was more IPA-like: pe"rj, kwe"ting). The words themselves were often reminiscent of English, or Latin words as Kelley may have known, or Greek and Hebrew ones as he may have picked up from Dee, or from a Bible. "Prge" for instance, meaning "fire" is reminiscent both of Greek "pur" and of English "purge", fire being a purging agent. The meaning is often different or only vaguely related, e.g. paradiz = virgin. The word order, as the interlinear translation in Dee's notes show, is also straight English, e.g. beginning of 17th call: O thou third flame, whose wings are thorns to-stir-up vexation Ils d ialprt, soba upaah chis nanba zixlay dodsih and hast 7336 living lamps going before thee, whose God is wrath in anger od brint taxs hubaro tustax ylsi, soba Iad i vonpo unph Don opined that Enochian had all the marks of glossolalia, and was a fabrication by Kelly. From bgrant@ais.org Thu Nov 12 11:20:21 1992 Message-Id: From: bgrant@ais.org (Bruce Grant) Subject: A pattern in the V. alphabet? To: voynich@rand.org Date: Wed, 11 Nov 1992 21:20:21 -0500 (EST) X-Mailer: ELM [version 2.4 PL8] Content-Type: text Content-Length: 2073 Status: OR Though I've been interested in the Voynich manuscript since Jim Reeds introduced me to it about 25 years ago, I'm new to the mailing list, so I don't know if an idea like this has already been brought up here. It appears to me that many characters of the Voynich alphabet could consist of the two characters Currier calls 'C' and 'I' with a variety of flourishes. In fact, I think a case could be made for a parallel sequence of characters (in increasing order of implausibility): Based Based on C on I ------- ------- C I (no flourish) 8 J/6 (with an ampersand-like flourish from the top of the character) 2 R (with an upward flourish from the top of the character) 9 E (maybe) (with a downward circular flourish) --- D (with an upward flourish from the bottom of the character) P F (elevated, with a downward tail and a trailing loop) B V (elevated, with a downward tail and a trailing loop that crosses back over the tail) O A (i.e. C + (reversed) C, C + I) I don't know where characters like S, Z, and S with something superimposed would fit in this scheme, or why there is no C-based character corresponding to 'D'. Other observations which make me think there may be something to this: - the fact that characters such as '8' and '2', which could easily be written with a single stroke, often appear to consist of a 'C' and a second stroke; - the large number of unadorned 'C' and 'I' characters (i.e. freestanding 'C' characters and the many 'I'-like strokes in the characters T, U, K, L etc. etc.) - the speculation that the manuscript may be a bilateral cipher. Bruce Grant (bgrant@umcc.ais.org)  From Karl.Kluge@G.GP.CS.CMU.EDU Thu Nov 12 02:56:00 1992 Date: Thu, 12 Nov 1992 02:56-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Question on vowel recog algorithms Message-Id: <721555001/kck@G.GP.CS.CMU.EDU> Status: OR I'm reading D'Imperio again (I've found I get something new out of it every time), and was thinking about the section on cryptanalytic hypotheses. Several questions come to mind: How robust are the vowel recognition algorithms that have been described on the group (Sukhotin and the SVD method) in the presence of nulls in the text? D'Imperio quotes Miss Nill as saying that Manly told Voynich in a letter of March 20, 1920 (and Newbold in a letter of simlar date) that freq. counts on 8 pages showed "a comparatively simple cipher disguised by extensive use of nulls". Manly's correspondence with Voynich appears to be listed in the Krauss catalog, but isn't obvious in the Yale catalog. It would be interesting to track down these letters. The reason I ask it that *IF* the vowel recog algorithms are robust in the presense of many nulls, then that may provide leverage to locate the pattern of null insertion. (Sometimes I can almost pursuade myself that the A/B difference is an artifact of B being lazy and using nulls in a much more stereotyped way.) A misc thought...I wonder if Yale is so blase about studies of the Voynich because they don't want another Vinland Map situation on their hands. Forgeries (even donated ones) are embarassing. Karl From SL500000@BROWNVM.brown.edu Thu Nov 12 21:05:40 1992 Message-Id: <199211121233.AA18141@rand.org> Date: Thu, 12 Nov 92 07:05:40 EST From: Robert Mathiesen Subject: Re: Question on vowel recog algorithms To: voynich@rand.org Status: OR Having a mild interest in the VM (and being a specialist in Medieval manu- scripts in general), I have been reading this list for some time now, but nothing has inspired me to write Karl's recent posts. -- Surely the fore- most reason why the Beinecke is not focussing much attention on the VM is simply because (1) there is no reason to expect that the solution of its mystery, if ever achieved, will prove either useful or interesting to scholars in the major academic professions and (2) they have literally thousands of manuscripts and early printed books which we already know how to read, and which are certain to be of some interest once read, but which nobody has had time to study yet. The VM may well be the most mysterious manuscript in the world, but the odds are very much against its being one of the more important manuscripts in the world. I, by the way, have seen the original and leafed through it, and even been in the Beinecke stacks and seen the many, many boxes of papers connected with it, from the letters of the first tentative discoverers and would-be decipherers down to the most recent published "solutions," all carefully saved and available for use. The Beinecke is definitely interested in the VM as in all their other holdings, but they have an academic sense of proportion. For what it may be worth as a casual opinion from someone who has experience with Medieval manuscripts, the VM seems really to be late Medieval, not a recent fake of any kind; and, what is more interesting, seems to have had really heavy use at one time by someone who *could* read it. Also, the hand which numbered the pages is believed to be John Dee's by French, who wrote a book on Dee; so it is more than likely that Dee had it once, despite any reference in his catalogue. -- This is a casual opinion only, not the result of careful study (which would be needed to deal with the possibility of a truly great forgery, where the forger had taken competent pains to re- produce traces of heavy use long ago). On the subject of a VM conference, it all comes down in the last analysis to the question of who will pay for all the support staff and time needed, and what guarantees can be given that payment will really be forthcoming when the bills are due. Many of the Ivy League universities, including Yale and my own Brown, are so strapped for cash, and have been for so long, that their very survival over the next 50-100 years is a genuinely open question, and even more central academic pursuits are being cut back se- verely. So there's no money to spend on less central matters like a VM conference. Anyone on the list who has genuinely deep pockets and an in- dustrial-strength credit rating, and who wanted to organize a VM conference, would get a respectful hearing from the Beinecke, I am sure. (To judge by the extent of postings on relatively minor costs for microfilms and such, though, it sounds like all of you are as poor as I am.) Failing that, any- one who has a realistic chance of writing a successful grant proposal to fund such a conference would be listened to respectfully, I'm sure. But no one's *ever* going to do it for you ...; you're on your own here. Robert Mathiesen Slavic Languages / Medieval Studies Brown University Providence, RI 02912 From reeds@gauss.att.com Thu Nov 12 23:25:48 1992 Message-Id: <199211121426.AA21157@rand.org> From: reeds@gauss.att.com Date: Thu, 12 Nov 92 09:25:48 EST To: voynich@rand.org Subject: Voynich vowel spotting Status: OR Karl asks > How robust are the vowel recognition algorithms that have been > described on the group (Sukhotin and the SVD method) in the presence of > nulls in the text? I think these algorithms are fairly robust against random insertion of nulls, but not against malicious insertion. If, for example, after each meaningful text letter you added a null if you got a 6 on a die roll, I would expect all reasonable vowel spotting procedures to work, but might require more text. (Because the digraphic frequency table would be a mixture of 5/6 of the un-nulled table, exhibiting its V/C patterning, and 1/6 of a relatively unpatterned table.) But if you were malicious you could make your null letter look like a vowel if you only inserted it between adjacent consonants, or you could make it look like a consonant if you only inserted it between adjacent vowels (if your language allows them) or next to isolated consonants, etc. To be trickier yet (to make a consonant seem like a vowel, say) would be harder, but might be done as follows. Suppose the consonant in question is English F. Suppose the null is G. We will try to make G look like a consonant and we will put it next to F so often that F will seem like a vowel. So we put G between adjacent vowels and next to isolated consonants and next to F. This just assumes that some individual letters are designated as nulls, and are to be simply elided on reading. But a rule like "elide any word containing letter X" might be in effect, and then you can skew the statistics drastically. (Suppose one word in 3 is nulled this way. Then essentially all frequency counts will be 2/3 of what they ought to be, plus 1/3 adulturation.) Come to think of it, the "null word" method seems much more likely to have been used than anything approaching the die model described above. (Of course, one can quickly check for usage of this particular scheme, by tabulating digraphs in words which contain letter X and digraphs in words which do not contain letter X, and so on.) Jim Reeds From rcarter@nyx.cs.du.edu Fri Nov 13 04:28:35 1992 Date: Thu, 12 Nov 92 12:28:35 MST From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9211121928.AA13394@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: Karl.Kluge@G.GP.CS.CMU.EDU, voynich@rand.org Subject: Re: Question on vowel recog algorithms Status: OR RE: why the BL is so blase... I would imagine the BL couldn't care less about the VM for alot of reasons... 16th C. (or even 15th C.) botanicals aren't that rare at all, no one even remotely famous (no one knows who Dee was anymore) is connected to it, even some of the best crypto people can't crack it (yet! :-) ie looks like some old paper to them... Now, if we could show a connection twixt the VM and the Aztecs, or Roger Bacon (sure...) or better yet, Shakespeare (hee...), that would perk them up, and take us out of the ballgame; it is probably best (for us anyway) that it is of little mainstream interest. All -I- know, is that I since I have been on Team Voynich (and being the one to suggest that name!) I have learned more about crypto than I ever knew, found out about Dee (incredible man!), and found a project that continually holds my interest, no matter how frustrating it gets. Yeah, and I wouldn't be surprised that Yale -is- afraid of another forgery scandal as well... :-) Regards to all, Ron | rcarter@nyx.cs.du.edu | 12 NOV 92 From gehenna@deeptht.armory.com Fri Nov 13 09:54:41 1992 From: Trepan X-Mailer: SCO System V Mail (version 3.2) To: voynich@rand.org Subject: ATTN: JIM GILLOGY Date: Thu, 12 Nov 92 16:54:41 PST Message-Id: <9211121654.aa29414@deeptht.armory.com> Status: OR Dear Mr. Gillogy, I I had a partial loss of text in receipt of your transmission of Jacques Guy's message, dated Nov 13, 1992, titled: vowel recognition in the presence of nulls. I received up "He would have loved" in the sixth paragraph. Could you please re-transmit from that point. Thank You. Col-lea Lane gehenna@deeptht.armory.com (192.122.209) From reeds@gauss.att.com Fri Nov 13 12:19:40 1992 Message-Id: <199211130319.AA18562@rand.org> From: reeds@gauss.att.com Date: Thu, 12 Nov 92 22:19:40 EST To: voynich@rand.org Subject: Voynich photostats Status: OR I got a reply from the Marshall Libe, as follows: Dear Mr. Reeds Mr. Cook has recently moved back to Wisconsin and we are between archivists, so I will endeavor to answer your questions regareding the Voynich manuscript. We have Col. Friedman's photostat copy (one of 6 made in 1945). I enclose a photocopy of one pair of pages made at 100% so that you can see the size and the quality that we can get from photocopying. I also include a page copied at 150%. You will have to decide whether that are adequate for the purpose you intend and/or whether you can scan the originals digitally, but there is a great deal of low-contrast and grey in the old photostats; they are certainly not crisp and clear photgraphs. And of course all the copies are in black and white; most sheets have Friedman's handwritten descriptions of the colors on the back; if you desired to have this descriptive material, it would increase the photocopying considerably. There are 235 numbered pages in the version Friedman has. In addition, of course, we have a large collection of material that Friedman's group worked up on the manuscript, including his early use of the computer to analyze the character frequency. If you think that the quality we are capable of making is woth the effort, a complete copy at about 160% on 11x17 paper would cost about $60 (235 x .25). Sincerely, Larry I. Bland Director Marshal Papers Project The enclosed copies appear to have been made by putting the photostats on an ordinary flatbed xerox machine & hitting the copy button; the result is much more legibile than the corresponding Yale xeroxes. The 150% page is f1r, the two 100% percent pages are f1v and f2r. The photstats seem to be about 9.25 by 12.75 inches, the V pages are about 6 by 8.75 inches. All are clear enough to transcribe from. The general appearance is black ink on gray in contrast to the gray on light gray in the Yale copy. Maybe I should send the two copies to some other member of the list, say Karl, for a second opinion about legibility? Comments? Jim Reeds From j.guy@trl.oz.au Sat Nov 14 00:05:26 1992 Date: Fri, 13 Nov 92 10:05:26 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9211122305.AA10514@medici.trl.OZ.AU> To: voynich@rand.org Subject: vowel recognition in the presence of nulls Status: OR The last issue of Cryptologia -- that is, the last issue to hit our library shelves here at Telecom Australia -- had an article by [forgot his first name] Sassoon, examining the performance of Sukhotin's algorithm on a variety of languages, including Irish Gaelic and ... unpointed Hebrew! It certainly could not be right on unpointed Hebrew. Nevertheless, I remember it did identify such consonants as often serve as supports for vowels (he, yod...) as vowels. Very predictably, it identified the ubiquitous "h" of Irish as a vowel too. >From the Irish test case, we can infer that if nulls are inserted according to strict rules (a bit self-defeating is it not?) some may be identified as vowels, some as consonants, and that may somewhat mess up the identification of non-nulls. I have a seat-of-the-pants feeling that randomly inserted nulls would not affect the identification of non-nulls. Thinking on it further, I get this feeling that if nulls were inserted randomly enough, then Sukhotin's algorithm might provide a way of identifying them: as random insertions they should lack the property of non-nulls, i.e. vowels occur next to consonants. Nulls, then, should be found to occur randomly, regardless of their immediate environment. Yes, it seems that Karl Kluge has the right idea. Hypotheses that can be tested fairly easily by the Monte-Carlo method on a text in a known language. Perhaps I should lay aside my infuriating fiddly Voynich screen editor and spend a week-end on that ("What? Going to see your girl-friend *again*?" -- Girl-friend, that's what my wife calls my computer). On a different tack now. Why, there was a medievalist watching us! Robert Mathiesen, please, would you tell since when you have been reading our exchanges? What do you think of my hypothesis of a Beneventan influence in the Voynich script? We don't have sabbaticals here at Telecom Research, but in 3 years' time I will qualify for long-service leave -- 3 months on full pay or 6 months on half-pay. Do you think the Beinecke would let me go through their boxes of correspondance and photocopy what I find of interest? With a view to writing a book on it, a` la Don Laycock's "Enochian Dictionary". Pity Don died. He would have loved doing that and he would be dur for sabbatical just about now... From Karl.Kluge@G.GP.CS.CMU.EDU Fri Nov 13 20:43:00 1992 Date: Fri, 13 Nov 1992 20:43-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Photostats, vowels, some odd Herbal B folios, and Yale. Message-Id: <721705408/kck@G.GP.CS.CMU.EDU> Status: OR Jim -- I'd be happy to give a second opinion on legibility. The cost is disappointing -- more than Yale charges -- but I've wasn't overwhelmed by either the sample pages Yale sent or the comments people who ordered the print from the Yale film have made. On vowels -- looking for more diversion, I implemented what I believe to be Sukhotin's algorithm as described in Jacques's CRYPTOLOGIA article. I then ran the program on the A corpus, the B corpus, the entire D'Imperio transcript, and my label corpus. The results (vowels listed in order of identification) are: A corpus: O A 9 C 0 6 B corpus: C O A 9 L 0 A & B : O A C 9 0 labels : A O 9 S Z I 0 4 The labels are short. If Manly is right, and this is just a simple system concealed with nulls, and making the assumption that labels are words associated with the proximate drawings, etc., then the labels ought not contain many nulls (not long enough to be very padded). While C's are only 4% of the label characters vs. 11% in the D'Imperio, it may be significant that C (which often occurs in these -C89, -CCC9, etc. words) drops from the vowel list for the labels. Folios 33, 39, 40, 50, and 55 really stick out in the clusterings I've done on the B folios (A 13.5% vs 6-8% in other clusters, C 7.8% vs. 13-18%, M 4.0% vs 1.3-1.9%). I don't know how this fits in with the pattern of B folios within quires. Folios 26, 34, 43 and 46 also clump together, but not quite as distinctively, in their letter frequencies. This is in a 4 cluster clustering, and in the other pair of clusters recto and verso sides split between different clusters in many cases. I.e., the two small groups I listed appear to be "real" clusters, with the other two being artifacts of the number of clusters specified. Regarding Yale -- hello, Robert, great to have a specialist lurking in the group. I certainly recognize that limited resources and a vast body of material for potential study account for much of the lack of interest. On the other hand, their attempting to make a better archival microfilm print would be nice. Comparisons with earlier photos suggest that this should be possible. Karl From kibo@world.std.com Mon Nov 16 14:58:34 1992 Date: Mon, 16 Nov 1992 00:58:34 -0500 From: kibo@world.std.com (James "Kibo" Parry) Message-Id: <199211160558.AA08926@world.std.com> To: voynich@rand.org Subject: Fry's Pantographia Status: OR on something tangential to the Voynich... Does anyone have access to a copy of _Pantographia; containing accurate copies of all the known Alphabets in the World, together with an English explanation...[etc., very very long title]_ by Edmund Fry, 1799? This is a book in which are collected hundreds of 'hieroglyphic', foreign, archaic, and hermetic scripts... I would dearly love to borrow a copy or to get Xeroxes or stats of it. Cost is no object. :-) If nobody here knows of a copy that I can get reproductions from, I'll probably try to purchase it through the rare-book-dealer network. It's a long shot that one or more of the alphabets in it may be related to the VM script, but given that there are 200+ in there (many of which have not been published elsewhere that I know of; the samples I've seen are really bizarre), it's possible. And besides, I love alphabets anyway. -- K. From j.guy@trl.oz.au Wed Nov 18 01:16:21 1992 Date: Tue, 17 Nov 92 11:16:21 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9211170016.AA15537@medici.trl.OZ.AU> To: voynich@rand.org Subject: Entropy of Voynichese and Chinese Status: OR This morning, j'ai pris mon courage a` deux mains, and I started typing in the first fable in Zhong Qin's "Everyday Chinese -- Brighter Readings in Classical Chinese". I got sick of it soon, and stopped. But I still had 1167 letters, not counting spaces and tones. Now, Bennett gives, for the first 10 pages of the VMS: h0 = 3.66 h1 = 2.22 h2 = 1.86 and for Hawaiian (glottal stop and vowel length unrecorded): h0 = 3.20 h1 = 2.45 h2 = 1.98 I found for my bit of Classical Chinese, with Voynich in rightmost column for easy comparison: No tones, No tones, Tones, Tones no spaces spaces no spaces and spaces Voynich h0 4.64 4.70 4.86 4.91 3.66 h1 4.03 3.63 4.33 4.21 2.22 h2 2.77 2.14 2.76 2.40 1.86 Not what I expected (Voynich-like low h2 for Chinese). So I retransliterated the text into the Giles-Wade system. Painful, and I shouldn't have bothered. Just consider: No tones, No tones, Tones, Tones no spaces spaces no spaces and spaces Voynich h0 4.52 4.48 4.75 4.70 3.66 h1 4.01 3.86 4.32 4.25 2.22 h2 2.66 2.18 2.58 2.24 1.86 Doesn't look good for my Marco Polo theory. Voyniches definitely looks more Hawaiian than Chinese. On a lighter note of strange coincidences, I found Poundstone's "Labyrinths of Reason" (with a whole chapter on the VMS) and "Enochian Physics -- The Structure of the Magical Universe" on the remaindered books shelves at the university bookshop. The latter is Enochian in name only, with bugger-all real Enochian in it, not even the alphabet, or a proper call. Mentions John Dee of course. Nuts on the bandwagon. From Karl.Kluge@G.GP.CS.CMU.EDU Tue Nov 17 18:52:00 1992 Date: Tue, 17 Nov 1992 18:52-EST From: Karl.Kluge@G.GP.CS.CMU.EDU To: voynich@rand.org Subject: Massive Voynich coincidence of the week. Message-Id: <722044350/kck@G.GP.CS.CMU.EDU> Status: OR I typed in the 30 stations of the Moon from Agrippa and the left column of 15 "Hermetic" names of fixed stars from D'Imperio. I ran my program on the labels from f70v1&2 and f71r&v (Pisces, the two Aries, and one Taurus). In the Voynich label set, OP constituted 11.000% of the digraphs, and started 58.1081% of the labels. In the Moon stations/star names set, AL constituted 10.3125% of the digraphs, and started 58.1395% of the names. Spooky, ain't it? By the way, based on the Voynich label set mentioned above, the vowels are A, O, 9, S, and Z according to the Sukhotin algorithm. The vowels identified for the D'Imperio are O, A, C, 9, and 0. If you delete the C's from the D'Imperio transcript, Sukhotin's algorithm identifies O, A, 9, S, Z, and 0 as vowels (in that order). Given that C shows up in strings of more than 2 (i.e., Tiltman's -CCC9 and -CCC89 suffixes), C is starting to look like a candidate for null-hood. From j.guy@trl.oz.au Fri Nov 20 00:48:50 1992 Date: Thu, 19 Nov 92 10:48:50 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9211182348.AA18603@medici.trl.OZ.AU> To: voynich@rand.org Subject: Reduplicated words (long: 750 lines) Status: OR Reduplicated strings, e.g. 8AM8AM, are very common in the VMS (which makes me think that, if it is a language, it is non-Indoeuropean, welcome back to the Marco Polo theory!). I just did a brute strength list of reduplicated strings. Brute strength because I took only five minutes to write the program, so that it looks only for duplication within a line. Would have been better to ignore end-of-lines and stop at paragraphs, but I was, again, lazy. Here is the output, sorted alphabetically, and with the reference folio and line for each string (as per D'Imperio). There is at least one whole *sentence* duplicated: 8AMO8AK4OFAM9PCAEOFAEOJ. No, I don't know what to make of it. I've only had a cursory look. This list ought to help find word or morpheme boundaries, assuming of course that Voynichese is a real language. No ideas as to go about yet, though. I only noticed that languages A and B do not duplicate the same strings at all! Here it is, 748 lines long: <00310A> 2AM <15101B> 4OBSC89 <14814B> 4OE <15418B> 4OE <15804B> 4OE <15835B> 4OE <03403A> 4OF9 <15234B> 4OFAE <15703B> 4OFAE <15126B> 4OFAM <15230B> 4OFAN <15732B> 4OFAN <07903B> 4OFC89 <14713B> 4OFC89 <14713B> 4OFC89 <14713B> 4OFC89 <14738B> 4OFC89 <14808B> 4OFC89 <14820B> 4OFC89 <15219B> 4OFC89 <15305B> 4OFC89 <15305B> 4OFC89 <15619B> 4OFC89 <15619B> 4OFC89 <14713B> 4OFC894OFC89 <14739B> 4OFCC89 <15111B> 4OFCC89 <15121B> 4OFCC89 <15134B> 4OFCC89 <15604B> 4OFCC89 <16126B> 4OFCC89 <16416B> 4OFCC89 <16510B> 4OFCC89 <14819B> 4OFCC9 <14914B> 4OFCC9 <09705B> 4OFS89 <15808B> 4OFS89 <03402A> 4OFS9 <16128B> 4OP9 <05005B> 4OPC89 <15123B> 4OPC89 <15201B> 4OPC89 <16203B> 4OPC89 <15109B> 4OPCC89 <16307B> 4OPCC89 <03605A> 4OPS9 <05910B> 894OFC <07903B> 894OFC <07910B> 894OFC <14713B> 894OFC <14713B> 894OFC <14713B> 894OFC <14738B> 894OFC <14738B> 894OFC <14742B> 894OFC <14821B> 894OFC <15113B> 894OFC <15223B> 894OFC <15305B> 894OFC <15413B> 894OFC <15619B> 894OFC <16114B> 894OFC <16503B> 894OFC <16510B> 894OFC <14713B> 894OFC894OFC <14737B> 894OFCC <14739B> 894OFCC <15134B> 894OFCC <15134B> 894OFCC <16126B> 894OFCC <16416B> 894OFCC <16505B> 894OFCC <16510B> 894OFCC <05005B> 894OPC <15201B> 894OPC <16203B> 894OPC <16307B> 894OPCC <08310B> 8989 <03307A> 898OE <04905B> 89CC <15334B> 89ESC <16612B> 89OFC <16620B> 89OFC <06001B> 89OPC <06601B> 89S <08308B> 89SC <14917B> 89SC <06505B> 89ZC <14926B> 89ZC <16327B> 89ZXC <04216A> 8AE <09909A> 8AE <15213B> 8AE <15402B> 8AE <16427B> 8AE <01205A> 8AJ <10507A> 8AJ <01003A> 8AM <01511A> 8AM <01912A> 8AM <03512A> 8AM <03604A> 8AM <04004A> 8AM <04006A> 8AM <04705A> 8AM <04805A> 8AM <06812A> 8AM <07304A> 8AM <08114A> 8AM <09111A> 8AM <09515A> 8AM <15333B> 8AM <100r.2> 8AM <09807B> 8AMO <10608A> 8AMO8AK4OFAM9PCAEOFAEOJ <07406A> 8AN <14829B> 8AN <03306A> 8AR <14735B> 8AR <14736B> 8AR <00207A> 8OE <09615A> 8SO <09616A> 8SO <09503A> 8SOR <15101B> 94OBSC8 <15524B> 94OEFCC <03403A> 94OF <08702A> 94OF <16618B> 94OF <14903B> 94OFC <15833B> 94OFC <16204B> 94OFC <16628B> 94OFC <16631B> 94OFC <07903B> 94OFC8 <07910B> 94OFC8 <14713B> 94OFC8 <14713B> 94OFC8 <14713B> 94OFC8 <14738B> 94OFC8 <14808B> 94OFC8 <14820B> 94OFC8 <15223B> 94OFC8 <15305B> 94OFC8 <15619B> 94OFC8 <15619B> 94OFC8 <14713B> 94OFC894OFC8 <14914B> 94OFCC <14739B> 94OFCC8 <15111B> 94OFCC8 <15134B> 94OFCC8 <16126B> 94OFCC8 <16416B> 94OFCC8 <16510B> 94OFCC8 <03402A> 94OFS <09705B> 94OFS8 <02006A> 94OP <15824B> 94OP <16128B> 94OP <16513B> 94OP <05005B> 94OPC8 <15123B> 94OPC8 <15201B> 94OPC8 <16203B> 94OPC8 <14742B> 94OPCC <16307B> 94OPCC8 <15224B> 98AE <08410B> 9BSC <04905B> 9CC8 <16131B> 9ESC <15334B> 9ESC8 <16021B> 9EZC8 <08704A> 9FCOE <00608A> 9FS <02303A> 9FS <02513A> 9FS <09811B> 9FS <14833B> 9OEFC8 <15502B> 9OEZC <08402B> 9OF <08502A> 9OF <16612B> 9OFC8 <04301A> 9OP <08304B> 9OP <10811B> 9OR <08303B> 9P9 <08304B> 9PC <15005B> 9PC <15022B> 9PC <09304B> 9PC8 <16520B> 9PC8 <16013B> 9PC89 <15633B> 9PCC <00612A> 9PS <01707A> 9PS <15009B> 9RAESCC <14917B> 9SC8 <15019B> 9SCX <05712A> 9XC <14926B> 9ZC <06505B> 9ZC8 <08904B> 9ZCF <16327B> 9ZXC8 <02309A> AE8 <04516A> AE8 <06711A> AE8 <08706A> AE8 <15213B> AE8 <15402B> AE8 <16001B> AE8 <16126B> AE8 <03401A> AE8S <16422B> AEZ <00504A> AJS <01511A> AM8 <02103A> AM8 <06812A> AM8 <06812A> AM8 <100r.2> AM8 <06711A> AMES <07608B> AMOEF <07709B> AMOF <09809B> AMOF <10810B> AMOF <09707B> AMOP <16322B> AMOP <15829B> ANOEF <15829B> ANOEF <15518B> ANOP <15518B> ANOP <16231B> ANOP <03306A> AR8 <14736B> AR8 <14736B> AR8 <10710B> ARO <07702B> AROF <07705B> AROF <06302B> AROP <14727B> AROP <15610B> BSC89 <16412B> C894OES <05910B> C894OF <07910B> C894OF <14713B> C894OF <14713B> C894OF <14713B> C894OF <14713B> C894OF <14738B> C894OF <14738B> C894OF <14742B> C894OF <14808B> C894OF <14821B> C894OF <14824B> C894OF <15113B> C894OF <15211B> C894OF <15215B> C894OF <15223B> C894OF <15233B> C894OF <15303B> C894OF <15305B> C894OF <15413B> C894OF <15619B> C894OF <16114B> C894OF <16312B> C894OF <16503B> C894OF <16510B> C894OF <16519B> C894OF <04905B> C894OFC <14737B> C894OFC <14738B> C894OFC <14739B> C894OFC <14943B> C894OFC <15113B> C894OFC <15134B> C894OFC <15134B> C894OFC <15407B> C894OFC <16126B> C894OFC <16201B> C894OFC <16416B> C894OFC <16503B> C894OFC <16505B> C894OFC <16510B> C894OFC <14713B> C894OFC894OF <14713B> C894OFC894OF <15123B> C894OP <15201B> C894OP <15801B> C894OP <16203B> C894OP <16307B> C894OPC <05001B> C899BS <04905B> C89C <15334B> C89ES <16612B> C89OF <06001B> C89OP <15308B> C89OP <15533B> C89OPCC <14807B> C89P <07909B> C89S <08308B> C89S <14917B> C89S <07911B> C89SCFC <06505B> C89Z <14926B> C89Z <15931B> C89Z <16323B> C89ZCQ <16327B> C89ZX <14903B> C94OF <14934B> C94OF <15037B> C94OF <14903B> C94OFC <14914B> C94OFC <16325B> C94OFC <16131B> C9ES <15009B> C9RAESC <05712A> C9X <04603A> CAR <04703A> CC2 <04905B> CC89 <14736B> CC894OF <14738B> CC894OF <15111B> CC894OF <15118B> CC894OF <15134B> CC894OF <15615B> CC894OF <15631B> CC894OF <16330B> CC894OF <16416B> CC894OF <16417B> CC894OF <16422B> CC894OF <07911B> CC89SCF <15237B> CC94OF <16011B> CC9OF <10903A> CC9SOF <11103B> CO89S <11103B> CO89SC <10408A> CO89Z <03907A> CORS <14814B> E4O <15703B> E4OFA <03307A> E898O <04516A> E8A <06711A> E8A <08706A> E8A <15213B> E8A <16001B> E8A <00207A> E8O <01407A> E8O <15901B> EBSC89O <15829B> EFANO <09308B> EOPCO <15601B> EORO <15809B> EPA <15022B> ESC89 <15035B> ESC89 <15334B> ESC89 <16205B> ESC89 <14818B> ESC894O <15209B> ESC9 <16637B> ESC9 <00115A> ESO <01005A> ESO <01608A> ESO <01908A> ESO <02809A> ESO <02809A> ESO <03802A> ESO <03802A> ESO <04007A> ESO <08121A> ESO <08611A> ESO <09102A> ESO <09107A> ESO <09107A> ESO <09515A> ESO <10909A> ESO <11014A> ESO <11015A> ESO <11016A> ESO <16021B> EZC89 <15421B> EZC89O <08510A> F9SO <07709B> FAMO <15709B> FANO <15722B> FANO <16623B> FANO <15829B> FANOE <15829B> FANOE <07702B> FARO <07705B> FARO <10702B> FARO <10706B> FC89 <15311B> FC89 <05910B> FC894O <14713B> FC894O <14713B> FC894O <14713B> FC894O <14738B> FC894O <14808B> FC894O <14824B> FC894O <15406B> FC894O <15619B> FC894O <16312B> FC894O <16519B> FC894O <14713B> FC894OFC894O <14943B> FC89E <07909B> FC89SC <14736B> FCC894O <15111B> FCC894O <15121B> FCC894O <15134B> FCC894O <15604B> FCC894O <15631B> FCC894O <16330B> FCC894O <16416B> FCC894O <15008B> FCC89O <07911B> FCC89SC <16011B> FCC9O <10903A> FCC9SO <10904A> FORS <15808B> FS894O <02303A> FS9 <02513A> FS9 <04112A> FS9 <16229B> FSC9 <09109A> FSO <00112A> FZO <02501A> M8A <06812A> M8A <06812A> M8A <06711A> MESA <07709B> MOFA <07302A> MOPA <09707B> MOPA <00712A> MSA <15829B> NOEFA <15518B> NOPA <15518B> NOPA <16231B> NOPA <06503B> O89 <08305B> O894OP <11103B> O89SC <11103B> O89SCC <10408A> O89ZC <09518A> O8S <00511A> OE2 <10210A> OE2 <00207A> OE8 <00602A> OE8 <04305A> OE8 <03307A> OE898 <15901B> OEBSC89 <00208A> OEF <14824B> OEF <15829B> OEFAN <14833B> OEFC89 <06007B> OEOF <10510A> OEOF <09308B> OEOPC <15601B> OEOR <15809B> OEP <16505B> OEPC <00115A> OES <00703A> OES <01515A> OES <01608A> OES <01810A> OES <01908A> OES <02809A> OES <02809A> OES <03802A> OES <03802A> OES <04007A> OES <04007A> OES <05103A> OES <08121A> OES <08611A> OES <09102A> OES <09107A> OES <09107A> OES <09515A> OES <10909A> OES <11014A> OES <11015A> OES <11016A> OES <15505B> OES <16340B> OESC89 <14818B> OESC894 <14806B> OESC9 <15012B> OEZC <14815B> OEZC9 <15537B> OEZCC9 <16011B> OF9 <08510A> OF9S <03009A> OFAM <06306B> OFAM <07709B> OFAM <07709B> OFAM <16408B> OFAM <15709B> OFAN <15722B> OFAN <16623B> OFAN <07702B> OFAR <07705B> OFAR <10702B> OFAR <08206A> OFC <05910B> OFC89 <15313B> OFC89 <16534B> OFC89 <16606B> OFC89 <16612B> OFC89 <05910B> OFC894 <14713B> OFC894 <14713B> OFC894 <14713B> OFC894 <14738B> OFC894 <14808B> OFC894 <15406B> OFC894 <15619B> OFC894 <16312B> OFC894 <14713B> OFC894OFC894 <08406B> OFC8AEZC89 <15008B> OFCC89 <16606B> OFCC89 <15111B> OFCC894 <15121B> OFCC894 <15134B> OFCC894 <15604B> OFCC894 <16330B> OFCC894 <14912B> OFCC9 <16011B> OFCC9 <10903A> OFCC9S <09109A> OFS <15808B> OFS894 <00112A> OFZ <05502A> OP9S <10711B> OPAE <07302A> OPAM <15518B> OPAN <06302B> OPAR <06302B> OPAR <14734B> OPAR <14745B> OPC89 <15123B> OPC894 <15201B> OPC894 <15109B> OPCC894 <09308B> OPCOE <01910A> OPOR <09518A> OPS <11009A> OPS <11009A> OPS <11009A> OPS <08506A> OPS9 <10502A> OPSC9 <11009A> OPSOPS <02306A> OR8 <15601B> OROE <03210A> OROF <02802A> OROR <10204A> ORQC <00508A> ORS <00807A> ORS <01504A> ORS <01909A> ORS <02805A> ORS <02905A> ORS <03303A> ORS <03713A> ORS <04702A> ORS <05609A> ORS <10203A> ORS <10904A> ORS <01616A> ORSC <05711A> ORSC <03105A> ORW <09510A> ORZ <05201A> OVS <11102B> OX9S <15809B> PAE <06902A> PAM4O <15518B> PANO <06302B> PARO <14734B> PARO <07513B> PC89 <09304B> PC89 <14807B> PC89 <16520B> PC89 <15123B> PC894O <15201B> PC894O <16505B> PC894O <14745B> PC89O <15104B> PCC894O <15109B> PCC894O <15633B> PCC9 <03505A> PS9 <03712A> PS9 <06903A> PS9 <03501A> PS94O <06807A> PSO <10915A> PSO <11009A> PSO <11009A> PSO <11009A> PSO <11009A> PSOPSO <00102A> QAR <04103A> QOE <05303A> QOE <03306A> R8A <14736B> R8A <16210B> ROE <15930B> ROEFA <03210A> ROFO <06302B> ROPA <10204A> RQCO <00508A> RSO <00807A> RSO <01504A> RSO <02805A> RSO <06801A> RSO <10203A> RSO <10904A> RSO <06406B> S89 <07505B> S89 <14904B> S894OF <15417B> S898 <08509A> S94OF <04806A> S94OP <05501A> S94OP <02513A> S9F <03003A> S9F <01707A> S9P <00502A> SAJ <00504A> SAJ <00712A> SAM <08306B> SC89 <08308B> SC89 <09007B> SC89 <14702B> SC89 <14917B> SC89 <14937B> SC89 <16204B> SC89 <16411B> SC89 <16623B> SC89 <14818B> SC894OE <15236B> SC89E <15334B> SC89E <03903A> SC9 <09517A> SC9 <14929B> SC9 <05711A> SCORSC9 <15019B> SCX9 <00115A> SOE <00703A> SOE <01605A> SOE <01608A> SOE <01608A> SOE <01807A> SOE <01810A> SOE <01905A> SOE <01908A> SOE <02302A> SOE <02809A> SOE <03205A> SOE <03209A> SOE <03213A> SOE <03216A> SOE <04007A> SOE <04007A> SOE <04609A> SOE <05602A> SOE <05603A> SOE <08121A> SOE <08203A> SOE <08611A> SOE <08702A> SOE <09102A> SOE <09102A> SOE <09107A> SOE <09107A> SOE <09209A> SOE <09515A> SOE <10510A> SOE <10902A> SOE <10909A> SOE <10915A> SOE <11003A> SOE <11014A> SOE <11015A> SOE <11015A> SOE <11016A> SOE <11016A> SOE <100r.9> SOE <08510A> SOF9 <01511A> SOP <11009A> SOP <11009A> SOP <11009A> SOP <05502A> SOP9 <11009A> SOPSOP <01102A> SOR <01208A> SOR <01504A> SOR <01908A> SOR <02805A> SOR <02805A> SOR <02808A> SOR <02905A> SOR <03303A> SOR <03510A> SOR <03512A> SOR <03713A> SOR <04207A> SOR <05201A> SOR <05706A> SOR <05802A> SOR <05809A> SOR <06103A> SOR <06821A> SOR <10511A> SOR <10904A> SOR <09503A> SOR8 <05201A> SOV <09801B> VAR <05201A> VSO <05712A> XC9 <08909B> Z89 <09003B> Z89 <06505B> ZC89 <14914B> ZC89 <14922B> ZC89 <14926B> ZC89 <15010B> ZC89 <15023B> ZC89 <16117B> ZC89 <15025B> ZC894OF <08904B> ZCF9 <10408A> ZCO89 <01406A> ZOE <01604A> ZOE <08113A> ZOE <09106A> ZOE <01705A> ZOE8 <05201A> ZOE9 <03807A> ZOQ9 <02105A> ZOR <05506A> ZOR <08604A> ZOR <09510A> ZOR <16327B> ZXC89 From rcarter@nyx.cs.du.edu Fri Nov 20 03:18:57 1992 Date: Thu, 19 Nov 92 11:18:57 MST From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9211191818.AA12489@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: j.guy@trl.oz.au, voynich@rand.org Subject: Re: Reduplicated words (long: 750 lines) Status: OR Yikes! A couple of thought re: duplicate words. If the the VM is a code of some kind, would the writers be dumb enough to build in nulls that are `just' duplicated words? I doubt it. What if some of the dupes are numerical structures such as `three threes' (or 3 times 3 or 9) or four fours, etc.? Some of these dupes could have been produced (if we use the dictation theory) by the speaker (or transcriber) getting lost in the original text, and rereading the same section or phrase or word, twice... Since the VM doesn't have any crossed out areas, I think it is safe to assume that the writers of the VM made mistakes, but just chose to ignore them and continue on. Or perhaps one of the speakers had a bad stutter :-). I don't feel real comfortable that this would apply to all the dupes, however, because of the volume of them. Ron | rcarter@nyx.cs.du.edu | 19 NOV 92 From bgrant@ais.org Fri Nov 20 14:12:24 1992 Message-Id: From: bgrant@ais.org (Bruce Grant) Subject: Reduplicated words To: voynich@rand.org Date: Fri, 20 Nov 1992 00:12:24 -0500 (EST) X-Mailer: ELM [version 2.4 PL11] Content-Type: text Content-Length: 902 Status: OR Another explanation of reduplicated words might be that they are encoding some simpler underlying cipher, where many Voynich "words" represent the same underlying group of characters. With many equivalent "words" to choose from, the scribe might tend to use the same ones repeatedly. This could also explain why a different scribe would produce text with different frequencies, even if the underlying text is essentially the similar. Incidentally, since Jim Reeds' page list indicates that both sides of the same page are always in the hand of the same scribe, and the pages from scribes A and B are intermixed, it seems more likely to me that the use of two scribes represents simply a division of labor, rather than the use of two languages or vocabularies by two "specialists", and that the language encoded by scripts A and B _should_ be similar even if they don't appear to be on the surface. From rcarter@nyx.cs.du.edu Fri Nov 20 14:57:47 1992 Date: Thu, 19 Nov 92 22:57:47 MST From: rcarter@nyx.cs.du.edu (Ron Carter) Message-Id: <9211200557.AA20804@nyx.cs.du.edu> X-Disclaimer: Nyx is a public access Unix system run by the University of Denver. The University has neither control over nor responsibility for the opinions or correct identity of users. To: bgrant@ais.org, voynich@rand.org Subject: Re: Reduplicated words Status: OR Interesting points by Bruce Grant... repeatedly. This could also explain why a different scribe would produce text with different frequencies, even if the underlying text is essentially the similar. text with different frequencies, even if the underlying text is essentially the similar. If this holds true, then maybe (a big maybe) the most repeated A dupe phrase/sequence is directly replaced by the most repeated B dupe phrase/sequence ie we are seeing two different ways of writing the same thing; a sort of dialect of writing (or transcription) styles, like the old example of spelling fish with alternative phonetics (ghoti or whatever it is...) in place. Ron | rcarter@nyx.cs.du.edu | 19 NOV 92 From j.guy@trl.oz.au Fri Nov 20 22:48:16 1992 Date: Fri, 20 Nov 92 08:48:16 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9211192148.AA20121@medici.trl.OZ.AU> To: voynich@rand.org Subject: More on duplicated words (shortish) Status: OR I did a frequency count, and the winner is SOE (41 times duplicated). Second is C894OFC894OF. To find such a long string so often duplicated is unexpected. I have never before tried that angle on any real language and, consequently, do not know what to make of it. Certainly, one would find such duplicates in English, e.g. ING as in: SINGING, BRINGING, etc. And even trips: BRINGING IN G... ^^^^^^^^^^^ Next, I bundled together those dupes that shared common substrings. It makes it much easier if you write the whole doubled "word", e.g. SOESOE. Here are the partial results of my efforts: 26 C894OFC894OF These occur *only* in B 18 894OFC894OFC 12 94OFC894OFC8 12 4OFC894OFC89 9 OFC894OFC894 11 FC894OFC894O 8 FCC894OFCC894O Only in B 11 CC894OFCC894OF 15 C894OFCC894OFC 8 894OFCC894OFCC 6 94OFCC894OFCC8 8 4OFCC894OFCC89 5 OFCC894OFCC894 41 SOESOE Only in A 24 OESOES 19 ESOESO 21 SORSOR Only in A 12 ORSORS 7 RSORSO 16 8AM8AM Only in A 5 AM8AM8 3 M8AM8A 8 AE8AE8 In both A and B 5 E8AE8A 5 8AE8AE 5 PSOPSO Only in A 4 SOPSOP 4 OPSOPS 9 SC89SC89 Only in B 7 ZC89ZC89 Only in B 4 PC89PC89 Only in B 4 ZORZOR Only in A 4 ZOEZOE Only in A Once again, we are faced with how very differently A and B behave. As I just said, I do not really know what to make of it. Since I am a cryptological ignoramus, I would like some of you to comment on what kind of a cipher those patterns may betray. For my part, I would say that we have one word that is used exclusively by scribe B, and is often reduplicated, and which is (4)OFC(C)89. The duplicated in language A are too short to be much informative, and I'd say that we have in those accidental duplications, as in English ...INGING IN G..., which would produce patterns similar to that seen for SOESOE, SORSOR, 8AM8AM etc., viz: INGING NGINGI GINGIN From j.guy@trl.oz.au Sat Nov 21 00:01:25 1992 Date: Fri, 20 Nov 92 10:01:25 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9211192301.AA20233@medici.trl.OZ.AU> To: voynich@rand.org Subject: vowel identification and nulls Status: OR Randomly inserted nulls seem to make no difference at all. I took the whole text of Genesis, ran Sukhotin's algorithm on it, then inserted randomly 3 nulls. No difference. You'll notice that "t" is given as a vowel. It always is in English. The reason appears to be the very high frequency of "th". But "d" is also given as a vowel. I suspect that that is due to the very high frequency of "and" in the King James Version. Actually, this points to a possible improvement: one ought to run Sukhotin's algorithm on the vocabulary of a text, not on the text itself. Here are my results, with the null-inserting program in appendix; File 01GEN.KJV (Genesis, King James Version) 151854 letters. 8 vowels, 18 consonants Absolute frequency Relative Frequency (per 1000) V# Total Init Med Fin Isol L/I L/F Total Init Med Fin Isol L/I L/F 1 e 19278 850 10796 7632 0 81 813 e 127 23 142 204 0 20 197 2 a 15476 5412 9674 49 341 1459 42 a 102 145 127 1 314 354 10 5 t 13585 6423 4036 3126 0 558 322 t 89 172 53 84 0 135 78 h 13261 3050 8385 1826 0 206 234 h 87 81 110 49 0 50 57 n 11230 720 7811 2699 0 76 316 n 74 19 103 72 0 18 77 3 o 10254 1967 6577 1701 9 103 122 o 68 53 87 45 8 25 30 6 d 9204 994 922 7288 0 86 633 d 61 27 12 195 0 21 153 s 8701 2932 2409 3109 251 261 434 s 57 78 32 83 231 63 105 4 i 8334 1554 6247 49 484 139 46 i 55 42 82 1 446 34 11 r 7696 386 5222 2088 0 34 276 r 51 10 69 56 0 8 67 l 5468 1061 3212 1195 0 72 104 l 36 28 42 32 0 17 25 m 3998 1508 1213 1277 0 116 192 m 26 40 16 34 0 28 47 f 3618 1424 674 1520 0 133 138 f 24 38 9 41 0 32 33 7 u 3564 1015 2101 448 0 48 60 u 23 27 28 12 0 12 15 w 3090 2304 488 298 0 213 19 w 20 62 6 8 0 52 5 8 y 2692 482 343 1867 0 40 217 y 18 13 5 50 0 10 53 b 2663 1839 617 207 0 197 28 b 18 49 8 6 0 48 7 c 2605 1147 1377 81 0 108 12 c 17 31 18 2 0 26 3 g 2315 849 976 490 0 71 60 g 15 23 13 13 0 17 15 p 1849 699 951 199 0 63 17 p 12 19 13 5 0 15 4 v 1355 87 1268 0 0 7 0 v 9 2 17 0 0 2 0 k 937 221 465 251 0 25 36 k 6 6 6 7 0 6 9 j 488 453 35 0 0 27 0 j 3 12 0 0 0 7 0 z 108 47 40 21 0 3 3 z 1 1 1 1 0 1 1 x 68 0 60 8 0 0 2 x 0 0 1 0 0 0 0 q 17 5 12 0 0 0 0 q 0 0 0 0 0 0 0 File 01GEN.NUL 01GEN.KJV with three nulls (* # |) randomly inserted. Sample: 1:1 In t*he beginning| God* creat|ed the he|aven* and# th*e| earth. 1:2 And the earth# wa#s withou*t fo#rm, and# void; a*nd d*arknes*s [was#] upon the face of the d|eep. And t|h|e Spirit of Go|d move*d upon the fac*e of the wate*r|s. 1:3 A*nd God sai#d, Le#t the*re be lig*ht: an*d ther*e wa*s li*ght|. 1:4 And G*od saw th#e light, th*at [i#t was] good: and Go|d divi*ded th*e light fr|om* the dark#ness*. 182256 letters. 8 vowels, 21 consonants Absolute frequency Relative Frequency (per 1000) V# Total Init Med Fin Isol L/I L/F Total Init Med Fin Isol L/I L/F 1 e 19278 850 12359 6069 0 81 650 e 106 23 117 161 0 20 158 2 a 15476 5493 9685 38 260 1459 32 a 85 146 91 1 304 354 8 3 t 13585 6423 4668 2494 0 558 256 t 75 171 44 66 0 135 62 h 13261 3050 8772 1439 0 206 176 h 73 81 83 38 0 50 43 * 12184 0 9092 3092 0 0 347 * 67 0 86 82 0 0 84 n 11230 720 8358 2152 0 76 245 n 62 19 79 57 0 18 59 # 10583 0 7905 2678 0 0 285 # 58 0 75 71 0 0 69 4 o 10254 1968 6927 1351 8 103 96 o 56 52 65 36 9 25 23 6 d 9203 994 2344 5865 0 86 518 d 50 26 22 156 0 21 126 s 8700 2982 3024 2493 201 261 345 s 48 79 29 66 235 63 84 5 i 8334 1652 6260 36 386 139 35 i 46 44 59 1 451 34 8 r 7696 386 5646 1664 0 34 216 r 42 10 53 44 0 8 52 | 7638 0 5704 1934 0 0 204 | 42 0 54 51 0 0 49 l 5468 1061 3432 975 0 72 84 l 30 28 32 26 0 17 20 m 3998 1508 1457 1033 0 116 154 m 22 40 14 27 0 28 37 f 3617 1424 984 1209 0 133 107 f 20 38 9 32 0 32 26 7 u 3564 1015 2186 363 0 48 53 u 20 27 21 10 0 12 13 w 3090 2304 543 243 0 213 16 w 17 61 5 6 0 52 4 8 y 2692 482 703 1507 0 40 176 y 15 13 7 40 0 10 43 b 2663 1839 664 160 0 197 23 b 15 49 6 4 0 48 6 c 2605 1147 1392 66 0 108 9 c 14 30 13 2 0 26 2 g 2315 849 1051 415 0 71 54 g 13 23 10 11 0 17 13 p 1849 699 992 158 0 63 14 p 10 19 9 4 0 15 3 v 1355 87 1268 0 0 7 0 v 7 2 12 0 0 2 0 k 937 221 513 203 0 25 28 k 5 6 5 5 0 6 7 j 488 453 35 0 0 27 0 j 3 12 0 0 0 7 0 z 108 47 43 18 0 3 2 z 1 1 0 0 0 1 0 x 68 0 64 4 0 0 1 x 0 0 1 0 0 0 0 q 17 5 12 0 0 0 0 q 0 0 0 0 0 0 0 Program used to generate 01GEN.NUL from 01GEN.KJV var inf,outf: text; procedure OpenFiles(infn,outfn: string); begin assign(inf,infn); reset(inf); assign(outf,outfn); rewrite(outf); end; procedure CloseFiles; begin close(inf); close(outf); end; procedure Process; var BufferIn,BufferOut: string; i: byte; ch: char; begin while not eof(inf) do begin readln(inf,BufferIn); BufferOut:=''; for i:=1 to length(BufferIn) do begin ch:=BufferIn[i]; BufferOut:=BufferOut+ch; case Upcase(ch) of 'A'..'Z': case trunc(Random*100) of 0..79: ; (* do nothing *) 80..84: BufferOut:=BufferOut+'|'; 85..92: BufferOut:=BufferOut+'*'; 93..99: BufferOut:=BufferOut+'#'; end; end; end; writeln(outf,BufferOut); end end; begin OpenFiles('01GEN.KJV','01GEN.NUL'); Randomize; Process; CloseFiles; end. From bgrant@ais.org Fri Nov 27 19:07:15 1992 Message-Id: From: bgrant@ais.org (Bruce Grant) Subject: POSSIBLE NULLS To: voynich@rand.org Date: Fri, 27 Nov 1992 05:07:15 -0500 (EST) Reply-To: bgrant@umcc.ais.org X-Mailer: ELM [version 2.4 PL11] Content-Type: text Content-Length: 5890 Status: OR If the Voynich text consists of some characters which are meaningful and others which are nulls, then perhaps the statistical differences between the A and B hands are due to different selection of nulls by two different scribes. If so, the characters whose frequencies vary widely between the two hands would be likely candidates to be nulls, and removing them should leave the A and B texts more like one another. Using the D'Imperio transcription file I calculated the following letter frequencies and word length distributions for the two hands, expressed in percentages: ---Null Test--- File-A: vms_a File-B: vms_b Nulls: CHARACTER FREQUENCIES Chr Pct-A Pct-B Chr Pct-A Pct-B Chr Pct-A Pct-B Chr Pct-A Pct-B 0 0.02 0.00 1 0.00 0.00 2 1.82 1.30 3 0.13 0.12 4 2.39 4.81 5 0.00 0.00 6 0.13 0.05 7 0.03 0.01 8 7.61 9.93 9 10.89 13.22 A 7.75 7.63 B 0.76 0.75 C 5.67 14.16 D 0.29 0.05 E 5.76 7.07 F 4.90 6.96 G 0.02 0.02 H 0.02 0.01 I 0.18 0.02 J 0.75 0.38 K 0.04 0.01 L 0.02 0.01 M 3.82 1.76 N 0.73 1.38 O 19.49 13.15 P 4.41 3.06 Q 1.63 0.43 R 4.99 3.40 S 10.79 5.49 T 0.21 0.19 U 0.08 0.08 V 0.23 0.24 W 0.34 0.08 X 0.62 0.75 Y 0.08 0.03 Z 3.42 3.40 WORD LENGTH DISTRIBUTION Len Pct-A Pct-B Len Pct-A Pct-B Len Pct-A Pct-B Len Pct-A Pct-B 0 0.06 0.03 1 3.30 1.01 2 12.00 8.27 3 34.95 20.29 4 22.29 24.75 5 16.87 23.65 6 7.26 13.39 7 2.50 6.42 8 0.62 1.43 9 0.13 0.47 10 0.03 0.13 11 0.00 0.10 12 0.00 0.04 13 0.00 0.01 14 0.00 0.00 15 0.00 0.00 16 0.00 0.00 17 0.00 0.00 18 0.00 0.00 19 0.00 0.00 STATISTICS Txt Avg len Pages Paras Lines Words Chars Known Unknown A 3.71 86 177 1263 7124 26578 26437 141 B 4.39 45 44 1165 9142 40199 40162 37 Rat 1.18 I then removed the two letters whose frequencies were most widely divergent (C and S) which generated the following distributions: ---Null Test--- File-A: vms_a File-B: vms_b Nulls: CS CHARACTER FREQUENCIES Chr Pct-A Pct-B Chr Pct-A Pct-B Chr Pct-A Pct-B Chr Pct-A Pct-B 0 0.02 0.01 1 0.00 0.00 2 2.18 1.62 3 0.15 0.15 4 2.86 5.99 5 0.00 0.01 6 0.15 0.06 7 0.04 0.02 8 9.11 12.37 9 13.03 16.45 A 9.28 9.50 B 0.91 0.94 C 0.00 0.00 D 0.34 0.07 E 6.90 8.80 F 5.87 8.66 G 0.02 0.03 H 0.02 0.01 I 0.21 0.02 J 0.90 0.47 K 0.05 0.02 L 0.02 0.01 M 4.57 2.19 N 0.87 1.72 O 23.33 16.37 P 5.28 3.81 Q 1.96 0.53 R 5.98 4.23 S 0.00 0.00 T 0.25 0.24 U 0.10 0.11 V 0.27 0.30 W 0.40 0.10 X 0.74 0.93 Y 0.09 0.04 Z 4.09 4.24 WORD LENGTH DISTRIBUTION Len Pct-A Pct-B Len Pct-A Pct-B Len Pct-A Pct-B Len Pct-A Pct-B 0 0.07 0.05 1 7.06 3.33 2 24.26 19.44 3 35.42 29.52 4 20.28 22.85 5 9.52 19.81 6 2.53 3.56 7 0.70 0.96 8 0.14 0.31 9 0.03 0.13 10 0.00 0.03 11 0.00 0.01 12 0.00 0.00 13 0.00 0.00 14 0.00 0.00 15 0.00 0.00 16 0.00 0.00 17 0.00 0.00 18 0.00 0.00 19 0.00 0.00 STATISTICS Txt Avg len Pages Paras Lines Words Chars Known Unknown A 3.10 86 177 1263 7124 22226 22085 141 B 3.53 45 44 1165 9142 32305 32268 37 Rat 1.14 The resulting word length distributions still vary quite a bit between the two hands, however. Finally, (by trial and error) I removed several other divergent characters (894R) which left the character frequencies much closer than in the original and also made the word length distribution and average word length much more similar: ---Null Test--- File-A: vms_a File-B: vms_b Nulls: CS894R CHARACTER FREQUENCIES Chr Pct-A Pct-B Chr Pct-A Pct-B Chr Pct-A Pct-B Chr Pct-A Pct-B 0 0.03 0.01 1 0.00 0.00 2 3.16 2.66 3 0.22 0.24 4 0.00 0.00 5 0.00 0.01 6 0.22 0.10 7 0.05 0.03 8 0.00 0.00 9 0.00 0.00 A 13.45 15.58 B 1.31 1.54 C 0.00 0.00 D 0.50 0.11 E 9.99 14.43 F 8.50 14.20 G 0.03 0.05 H 0.03 0.02 I 0.31 0.04 J 1.30 0.77 K 0.07 0.03 L 0.03 0.02 M 6.62 3.60 N 1.27 2.83 O 33.79 26.86 P 7.65 6.24 Q 2.83 0.87 R 0.00 0.00 S 0.00 0.00 T 0.37 0.40 U 0.14 0.17 V 0.39 0.49 W 0.58 0.17 X 1.07 1.53 Y 0.13 0.07 Z 5.92 6.95 WORD LENGTH DISTRIBUTION Len Pct-A Pct-B Len Pct-A Pct-B Len Pct-A Pct-B Len Pct-A Pct-B 0 6.50 7.71 1 21.67 22.49 2 37.49 36.45 3 20.97 17.09 4 11.05 13.33 5 1.78 2.18 6 0.46 0.61 7 0.04 0.11 8 0.03 0.02 9 0.00 0.01 10 0.00 0.00 11 0.00 0.00 12 0.00 0.00 13 0.00 0.00 14 0.00 0.00 15 0.00 0.00 16 0.00 0.00 17 0.00 0.00 18 0.00 0.00 19 0.00 0.00 STATISTICS Txt Avg len Pages Paras Lines Words Chars Known Unknown A 2.14 86 177 1263 7124 15386 15245 141 B 2.15 45 44 1165 9142 19708 19671 37 Rat 1.01 Of course, this doesn't prove that these characters _are_ nulls, but I thought it was kind of interesting. Bruce Grant From bgrant@ais.org Sat Nov 28 10:52:05 1992 Message-Id: From: bgrant@ais.org (Bruce Grant) Subject: POSSIBLE NULLS To: voynich@rand.org Date: Fri, 27 Nov 1992 20:52:05 -0500 (EST) Reply-To: bgrant@umcc.ais.org X-Mailer: ELM [version 2.4 PL11] Content-Type: text Content-Length: 4084 Status: OR In another experiment I made the following substitutions to test the hypothesis that letters like T, Q, and S are composites of simpler letters. The idea again was to see how the statistics for the A and B parts of the manuscript compared under this transformation. The specific substitutions I used were these: N->ID M->IID 3->IIID T->IR U->IIR 0->IIIR K->IJ L->IIJ 5->IIIJ G->IE H->IIE 1->IIIE Q->CPC X->CFC Y->CVC W->CBC S->CC Z->2C I made these substitutions to the A and B parts of the D'Imperio transcription file. This produced the following distribution of letter and word-length frequencies expressed as percentages: ---Null Test--- File-A: vms_a1 File-B: vms_b1 Nulls: CHARACTER FREQUENCIES Chr Pct-A Pct-B Chr Pct-A Pct-B Chr Pct-A Pct-B Chr Pct-A Pct-B 0 0.00 0.00 1 0.00 0.00 2 4.07 4.02 3 0.00 0.00 4 1.85 4.11 5 0.00 0.00 6 0.10 0.04 7 0.02 0.01 8 5.90 8.48 9 8.45 11.28 A 6.02 6.51 B 0.85 0.71 C 27.94 26.56 D 3.85 2.83 E 4.50 6.06 F 4.28 6.57 G 0.00 0.00 H 0.00 0.00 I 7.36 4.90 J 0.63 0.34 K 0.00 0.00 L 0.00 0.00 M 0.00 0.00 N 0.00 0.00 O 15.13 11.22 P 4.69 2.97 Q 0.00 0.00 R 4.12 3.14 S 0.00 0.00 T 0.00 0.00 U 0.00 0.00 V 0.23 0.23 W 0.00 0.00 X 0.00 0.00 Y 0.00 0.00 Z 0.00 0.00 WORD LENGTH DISTRIBUTION Len Pct-A Pct-B Len Pct-A Pct-B Len Pct-A Pct-B Len Pct-A Pct-B 0 0.06 0.03 1 3.17 0.97 2 5.31 6.18 3 12.30 9.77 4 23.10 15.26 5 24.96 26.15 6 15.23 22.11 7 9.49 12.47 8 4.06 4.39 9 1.70 1.66 10 0.44 0.58 11 0.11 0.23 12 0.04 0.11 13 0.04 0.07 14 0.00 0.00 15 0.00 0.00 16 0.00 0.02 17 0.00 0.00 18 0.00 0.00 19 0.00 0.00 STATISTICS Txt Avg len Pages Paras Lines Words Chars Known Unknown A 4.78 86 177 1263 7124 34200 34059 141 B 5.15 45 44 1165 9142 47106 47069 37 Ratio 1.08 I then removed the letters 4/8/9/O as possible nulls, due to the wide difference in letter frequencies between the A and B parts. That resulted in the following distribution in which the letter frequencies appear fairly close and the word length distribution very close: ---Null Test--- File-A: vms_a1 File-B: vms_b1 Nulls: 489O CHARACTER FREQUENCIES Chr Pct-A Pct-B Chr Pct-A Pct-B Chr Pct-A Pct-B Chr Pct-A Pct-B 0 0.00 0.00 1 0.00 0.00 2 5.92 6.19 3 0.00 0.00 4 0.00 0.00 5 0.00 0.00 6 0.15 0.06 7 0.03 0.02 8 0.00 0.00 9 0.00 0.00 A 8.77 10.03 B 1.24 1.10 C 40.68 40.91 D 5.61 4.37 E 6.55 9.33 F 6.24 10.13 G 0.00 0.00 H 0.00 0.00 I 10.72 7.55 J 0.91 0.53 K 0.00 0.00 L 0.00 0.00 M 0.00 0.00 N 0.00 0.00 O 0.00 0.00 P 6.83 4.58 Q 0.00 0.00 R 6.00 4.84 S 0.00 0.00 T 0.00 0.00 U 0.00 0.00 V 0.34 0.36 W 0.00 0.00 X 0.00 0.00 Y 0.00 0.00 Z 0.00 0.00 WORD LENGTH DISTRIBUTION Len Pct-A Pct-B Len Pct-A Pct-B Len Pct-A Pct-B Len Pct-A Pct-B 0 3.54 1.66 1 10.08 9.31 2 15.02 17.10 3 28.19 29.19 4 23.25 22.45 5 11.54 12.26 6 4.88 5.27 7 2.79 1.78 8 0.46 0.58 9 0.20 0.27 10 0.04 0.10 11 0.01 0.01 12 0.00 0.01 13 0.00 0.00 14 0.00 0.00 15 0.00 0.00 16 0.00 0.00 17 0.00 0.00 18 0.00 0.00 19 0.00 0.00 STATISTICS Txt Avg len Pages Paras Lines Words Chars Known Unknown A 3.28 86 177 1263 7124 23528 23387 141 B 3.34 45 44 1165 9142 30592 30555 37 Ratio 1.02 Bruce Grant From j.guy@trl.oz.au Wed Dec 02 07:37:55 1992 Date: Tue, 1 Dec 92 17:37:55 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9212010637.AA01907@medici.trl.OZ.AU> To: voynich@rand.org Subject: state of the Voynich editor Status: OR I am, at long last, getting there. The interface is finished, and is indifferently menu, keyboard, or mouse-driven. I have been hacking deep into the source code of the Turbo Editor Toolbox, and reduced it from 5000+ lines down to 3700 so far with no loss of real functionality (for instance, lines of up to 32000+ columns are no longer allowed, as they were in the original. You'll have to make do with an upper limit of 255 characters per line). That Toolbox is like the Voynich manuscript: written by at least four programmers, each with different styles and naming conventions. Some of it looks translated word-for-word from C, assembler, or worse. Horrible stuff, and the block move, copy, write, and delete commands are not fully functional. I am slaving at making them so. Once that is done, it seems that all that will be left to do is hook it to the interface. Back to the grinding stone... From j.guy@trl.oz.au Sat Dec 05 04:59:51 1992 Date: Fri, 4 Dec 92 14:59:51 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9212040359.AA07031@medici.trl.OZ.AU> To: voynich@rand.org Subject: Countdown to the Voynich editor for PCs Status: OR At long last, a bright light at the end of the tunnel. I have fixed the original Turbo Editor Toolbox block commands so that they operate properly and sensibly, whilst reducing the code from 5000+ lines down to 3100. The EXE file is down from 52K to 44K. Today, I am tackling the search-and-replace procedures which should shrink to 1/4 the original size when properly re-written. I had thought that, later, I might try to rewrite some parts in Assembler, but it is so surprisingly fast on my old 33MHz 386 that I will not bother. After I have fixed that search-and-replace procedure, there will only be a few touch-up jobs left to do (I abhorr messy code), and to link the lot to the interface. It might even be ready for Xmas! From firth@SEI.CMU.EDU Thu Dec 10 22:34:13 1992 Date: Thu, 10 Dec 92 08:34:13 -0500 From: firth@SEI.CMU.EDU Message-Id: <9212101334.AA18393@bp.sei.cmu.edu> To: voynich@rand.org Subject: Fourth Mail Try!!! Status: OR Notes on the Voynich Manuscript - Part 19 ----------------------------------------- Yes, I'm still here, and with the end of term approaching might even get my head above water soon! Is the Voynich manuscript writen in Arabic? I once looked at this conjecture, and alas rather think not. let me share those thoughts. My starting point was the star names, so many of which began with 'opq-'. It seemed an obvious guess that this stood for "al-". Now, that doesn't tell us anything about the language, since we too use the arabic star names, but it seemed worth a little speculation. Such a provenance is not impossible. If it's arabic, the place of origin is surely moslem Spain - al-Andalus - and very likely Catalonia or the Balearic Islands. Now, that civilization produced some major work in astrology and herbal medicine. It also produced a linguist called al-Zubaydi, who proposed a simplified form of arabic, with most of the grammar regularised or eliminated, that in many ways anticipated Peano's "latino sine flexione". Well now... consider a small, esoteric school, who carried through al-Zabaydi's plan, designed a synthetic script and grammar, and translated a few popular texts into it, as a proof of concept. Or even, took a copy of a real arabic book, after the illustrations had been drawn but before the text, and wrote the text instead in their new language. Is that why some folios have the pictures in the right margin? Anyway, one or more of these documents fell into the hands of James I, King of Aragon, when he conquered the territory in 1229. And one of his clerics took such a document to the first Council of Lyons, in 1245, where he gave it to that noted english scholar and arabist, Robert Grosseteste (1175-1253), who, of course, was the tutor of Roger Bacon. You see, you can connect anything to anything, if you try hard enough! So, what about the text? Well, if 'oqp-' is "al-", then we have an obvious guess at the infamous 4: it does indeed mean 'and', and so has the value "wa-" or "w'", and '4oqp-' is read "w'al-". However, the definite article isn't always written (or pronounced) "al-". Before some consonants, it mutates; the L turns into the following consonant. So we have "al-Kitab" ('the book'), but "ad-Din" ('the faith', ie Islam), "ar-Rahman" ('the Merciful'), "an-Nur" ('the light'), and so on. In all, there are 13 such consonants in the arabic alphabet. Well, let's suppose the other gallows letters represent such duplicated consonants, so that 'olp-' stands for "ar-R", which by frequency is plausible. Unfortunately, there aren't enough gallows letters, and their frequencies are a little too low. This, of course, is a problem with any Voynich speculation - there seem to be just too few consonants to make sense, both in terms of numbers of letters and their frequencies. The point at which the entire speculation collapses, however, is the text groups. These cannot be arabic words - they show no sign of triliteral roots, internal vowel mutations, or indeed any feature typical of arabic. They look much more like arabic with the prefixes and suffixes, but with the roots taken out of the middle - an eccentric and not very useful encryption system. Or, of course, an indo-european language with initial stems and end inflexions, which our analyses keep throwing up in our faces. And, finally, why would they write from left to right? So, unless I've missed something big, this too was a dead end. Robert From j.guy@trl.oz.au Thu Dec 17 04:04:03 1992 Date: Wed, 16 Dec 92 14:04:03 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9212160304.AA26885@medici.trl.OZ.AU> To: voynich@rand.org Subject: Voynich editor countdown Status: OR Still ticking. The block move/copy/delete functions now work as you'd expect them to. I am now struggling with the "reformat paragraph" command that did nothing sensible at all. Executable code is down to 43K and still shrinking. That means that, even after whacking on the interface, the whole thing should be well under 70K and able to edit a 500K file. Back to the grinding stone... From j.guy@trl.oz.au Fri Dec 18 04:32:26 1992 Date: Thu, 17 Dec 92 14:32:26 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9212170332.AA28781@medici.trl.OZ.AU> To: voynich@rand.org Subject: Voynich editor countdown Status: OR I know, I know, you'd rather have the thing up and going, and you don't care a fig, but it's my way of keeping my spirits up. After much swearing and pushing of the reset button, the paragraph reformatting routine now works as it ought to. There are three interesting things to note there -- if you are into writing programs: 1. Source code for the routine is down to 113 lines (from 220). 2. Once I decided, in despair, to junk the whole lot and rewrite ReformatParagraph from scratch, it took me about 30 minutes. 3. The compiled code has shrunk by 976 bytes! Doesn't say much for the quality of the original Turbo Editor Toolbox, does it? Now to add right-justifying, which the wretched thing did not even provide... From j.guy@trl.oz.au Mon Dec 21 23:51:15 1992 Date: Mon, 21 Dec 92 09:51:15 EST From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9212202251.AA03120@medici.trl.OZ.AU> To: voynich@rand.org Subject: Voynich editor countdown... Status: OR It seems that the last kinks are, at long last, ironed out of the editor. I have added right-justifying and a switchable keyboard map display. I got rid of tab-settings. The tab key now sends the cursor under the beginning of the next word of the previous line when in overwrite mode; when in insert mode it moves the rest of the line to align under that word. Still remains to hook the editor to the interface. But before that I ought to allow to reload fonts from within the editor. For instance, you should be able to switch from Roman-and-Voynich to Roman-and-Cyrillic, or, why not, Voynich-and-Cyrillic or Enochian-and-Voynich. Now about the interface. The editor program itself, adapted from FIRST-ED of the Turbo Editor Toolbox, uses Wordstar-like commands. The interface program uses menus, keyboard and mouse-driven, and is inspired from the built-in editor of Turbo Pascal (and Turbo C). So far it does nothing useful, except open and close windows. The second row on the screen is occupied by a menu bar, white on blue: File Block Replace/find Delete Go to Edit Toggle Quit The initial of each menu item is highlighted in cyan. There are three ways of accessing the commands. 1. Move the mouse to the menu bar and click. This highlights the command in white on red. For instance, moving the mouse anywhere inside Replace/find makes it light up in white on red. At first I had written it so that the menu item would be selected my merely moving the mouse there, and the menu window would open automatically. But I found that I was forever moving the mouse unintentionally -- or the cat (a real, furry cat) would jump onto my desk! -- and open menus I didn't care for. 2. Press F10. This highlight a menu item. Use the left and right arrows to move to other items, press [Enter] to select and open corresponding menu. 3. Press Alt and the highlighted (cyan) initial of the menu item you want to select. E.g. press Alt-F and the "File" menu opens. A menu opens with its top item highlighted, i.e. white on red. The other items are, again, white on blue, with one letter capitalized and highlighted in cyan. You can use the mouse, or the up-arrow and down-arrow keys to move the highlight. You select the command by clicking or pressing [Enter]. Or if you prefer you need only press one of the capital letters to execute the corresponding command; for instance press Alt-F then Alt-L or just L (upper or lower case) to load a file. I have allowed a number of shortcuts. Load file, for instance, is also executed by directly pressing F3. At this stage I am in two minds about whether or not I should let this interface accept Wordstar-like commands. That would be easy, since Wordstar commands all start with Ctrl-something and my menu items are activated by Alt-something. On the other hand, I was thinking of having Ctrl-something select a menu item *without* opening the corresponding window. That would disallow Wordstar-like commands. But again, I could implement a toggle: Open menus/Closed menus. So I am really in three minds about that. At any rate, here are the menu and shortcut commands as they stand now: File Load file F3 Save F2 save As Alt-F2 Block mark Start of block F7 mark End of block F8 Copy block at cursor Move block to cursor Delete block Hide block Alt-H Show block Alt-S Write block to file Read block from file Replace/find Find F9 Replace Alt-F9 Next Alt-N Delete Word Line to End of line Block All Go to Start of block End of block Line #... Mark #... Previous page PgUp Next page PgDn Top of file Bottom of file Edit Reformat paragraph Center this line set Left margin set Right margin Mark this line Toggle Overwrite/insert Ins Alphabet Alt-A autoIndent Alt-I Keyboard display Alt-K Right-justify Quit Not yet Yes, quit now