Voynich Paper Suggestion #4: Unpicking the k/t gallows

13 thoughts on “Voynich Paper Suggestion #4: Unpicking the k/t gallows”

1. D.N.O'Donovan on April 8, 2022 at 9:50 am said:

   NIck – not exactly on topic, but I thought you might be interested to hear that there’s a Greek-to-Latin translation of Euclid which, though certainly dated to the 13thC includes what one scholar describes as “an almost artificial humanist script” but which I should call proto-humanist because it too derives from Latins encountering the simplicity of Greek and Byzantine script and layout.

   If you’d like to see the manuscript in question, its digitised.
   Paris, Bibliotheque Nationale, Fonds latin 7373, 18 x 27 cm., vellum, 13thC.

   The article which directed me to that MS and the script which begins from f.140 is

   John E. Murdoch, ‘Euclides Graeco-Latinus: a Hitherto Unknown Medieval Latin Translation of the Elements Made Directly from the Greek’, Harvard Studies in Classical Philology, Vol. 71 (1967), pp. 249-302. [JSTOR]
   He makes a few other interesting comments.
2. Sharon J Lindimore on April 9, 2022 at 4:25 pm said:

   Hi Nick: I got my answer, please disregard prior message – thank you so much.
3. Darius on April 9, 2022 at 8:01 pm said:

   Nick, a lot of statistics and observations about “clasic” EVA-Voynichese. So, from here you try to build a bridge to the plaintext? May I take the liberty to make some points…

   1. EVA doesn’t contain all glyphs! E. g. f1r, 3rd passage, 8 line, 9th glyph from the right, just following an ‘8’ isn’t the same glyph like 15th glyph from the right likewise following an ‘8’. For EVA these glyphs are identical – an ‘a’, but the first one has an open belly – for me not the same.
   2. EVA suggests every glyph represents only one letter – not the case.
   3. For a few letters there are two representations (say like in modern alphabets capital and small letters – 2 representations but the same letter) – no adequate representation in transcriptions.
   4. No adequate representation for the double c’s + a glyph above. These are combinations of 3 letters, a double letter + one additional (at least on Currier-B pages)
   5. Vords are not representations of plaintext words! Else Voynichese would high probably satisfy Zipf’s law. These are concatenations of 1 till 4 words.
   6. Plaintext words are very short, because they are vowelless ketiv words in a language which knows a plenty of short words anyway. The avarage string length of vords is 5.6. For plaintext words I estimate 2 – 3 without vocalisation.
   7. You don’t know, how to read the text – left to right or right to left! So, you don’t know if special cumulative occurrences of glyphs at the beginning of vords or lines aren’t possibly rhymes if you read other way round. A strong indication for right to left direction are spirals (if you assume the bottom line of the text on the inner side of a counter clockwise spiral), because so you have some flexibility where your passage ends. Other way round you need a precision landing in the centre.

   You don’t need to believe even only one of these points. It’s up to you! Solely, their sheer potential validity, and there isn’t any rational reason to rule them out a priori, should make you sceptical. So, with a fuzzy heuristic and statistics of a particular dogmatic sting interpretation (e. g. new glyphs or vord de-concatenation would change these statistics dramatically) in the bag we should hunt out the plaintext.
   I guess, this doesn’t work that way – it didn’t as for the past…

   Take your EVA t/k examples – as for my dogma EVA k(K) stand for both a small and a capital letter and t(T) only for a capital letter (the corresponding small letter for t and T is EVA o). Here the relevant ratio would be k(K) vs t(T) and o and hence completely different.
4. D.N.O'Donovan on April 10, 2022 at 5:42 am said:

   Darius – Just two minutes ago I said the same as your point 2 in a post I’m preparing.

   I’m guessing neither you nor I is the first to make the observation, but can any reader help me credit the first person recorded as doing so?

   Sounds like the sort of thing Glen Claston might have dared say.

   To quote Darius (above)

   [quote] 2. EVA suggests every glyph represents only one letter – not the case.[unquote]

   I said ‘not necessarily the case’ but it’s much the same thing.

   If no-one can recall who should be credited, I’ll try hunting posts to Reeds’ mailing list.
5. Darius on April 10, 2022 at 1:50 pm said:

   Diane – as for Aramaic the multiple occupancy of some glyphs can very probably even be explained etymological. I will give an example in next days.
   I am sure, some of these points were already raised by other people.
6. Karl on April 11, 2022 at 10:46 am said:

   Darius and Diane on Darius’ Point 2: EVA (or Currier or v101 or FSG or Bennett or Frogguy or etc.) “suggests” that a particular group of adjacent pen strokes in one place on the vellum is sufficiently similar in shape to a bunch of other groups of adjacent pen strokes elsewhere on the vellum to record all of them the same way in the transcription. That’s all. The question of how those sets of sufficiently similarly shaped groups of adjacent strokes map onto letters in some other script (if at all) is outside the scope of what they’re trying to accomplish.

   That’s not to deny or bury the issue that “sufficiently similar” in any particular case is a judgement call on which reasonable people can differ pending a widely agreed upon solution. Glen tried very hard with the v101 transcription to capture any shape variant that occurred more than a handful of times (he had 6 by my count for Currier ‘R’, at least two for Currier ‘2’, and at least 7 for Currier ‘Z’), but he missed the distinction Darius points out — not just in the specific line, where both those glyphs are transcribed as ‘a’:

   (ko.3oe.3c.ho82c9.Jcae9.8ay.an.8an.H98s

   but at all in his transcription alphabet as far as I can see. He also missed the clear difference in how far back the right-most stroke hooks in things he transcribes under the umbrella of ‘n’ as shown in Figure 7 of Lisa Fagin Davis’ paper (https://repository.upenn.edu/mss_sims/vol5/iss1/6/). Which I suppose is as good an object lesson as any in the nature, difficulty, and limitations of transcription as a process.

   Nor is that to deny or bury the issue of deciding what groups of adjacent strokes correspond to individual glyphs in the script (i.e., are the things that look like ligatured gallows individual glyphs per Currier or combinations of multiple glyphs per Bennett). Disappointingly, after raising the question “How many letterforms are there?” in her abstract Lisa doesn’t weigh in on that issue in the rest of the paper [:-(]^googleplex

   Darius’ point 2 doesn’t sound like the sort of thing Glen would have said because Glen held to a fundamentally different type of theory regarding the text than what Darius is proposing — or if he did say it, he would have clarified it to mean that glyphs corresponded to different letters in the sense that individual glyphs didn’t have any fixed correspondence to a specific letter or set of letters at all, which from having looked at his blog is not what I understand Darius to be proposing.
7. R. Sale on April 17, 2022 at 6:10 pm said:

   2. EVA suggests every glyph represents only one letter – not the case.

   Multiple interpretations??? Really?
   Examples: See the ‘so-called’ 4 x 17 symbol sequence of VMs f57v. In the first five symbols (EVA: o, l, d, r, v), the fifth symbol has three interpretations.

   EVA v has the shape of an inverted “v”. As such, it has the appearance of the Greek letter lambda. Appearance is only appearance, as we all know, however this interpretation is backed up by structural verification. The distance to the first symbol, EVA o, is the same as for the Greek letter omicron. Greek reads either way.

   The second interpretation is based on medieval numerals as found on Typus Arithmetica. In the VMs sequence, the spacing between numerals “4” and “7” is correct. This is structural verification.

   The third interpretation is the Roman numeral “V”, which represents five “5”. In the VMs it is inverted. In the sequence it is “five” in the fifth place.

   Three different cultural interpretations of one VMs symbol. Each interpretation from a significant cultural tradition and based on appearance. Each interpretation supported by a structural verification from within the relevant frame of cultural interpretation built into the structure of the VMs sequence.

   Two numerical systems of interpretation and one that is alpha-numeric. The question of existence should be a question of function.
8. Darius on April 19, 2022 at 4:25 pm said:

   R. Sale, if you have 3 different interpretations for one glyph so you need 3 representations to show which one applies in a concrete context. It’s exactly what EVA doesn’t provide.

   Concerning f57v: ‘EVA-v’ is a relatively rare used glyph and of all things as for ‘EVA-v’ I have only one interpretation. The circular text you mentioned seems to be a simple list of glyphs in whatever order, then the whole list is repeated 4x. For me more interesting is the next inner circle text. Here some of the rare glyphs stand separated and some “usual” vords follow. The first one, just after the vertical stroke, is ‘EVA-v’ (I read them right to left). Then 4 multi-glyph vords follow. When I transliterate and translate this small passage with my method/alphabet I get:

   ‘EVA-v’ אוֹת ‘owth {oth}/קָרָא qara’ {kaw-raw’}/בְּעָא b@’a’ {beh-aw’}/עָבָה abah {aw-baw’}/כֹּה koh {ko}/הֶאָח heach {heh-awkh’}/כֵּהֶה keheh {kay-heh’}/אַחַר ‘achar {akh-ar’}

   ‘EVA-v’ sign appoints/be read aloud/utters a loud sound requesting/desiring to be fat/thick/gross, in this manner “hah” (aleph/alap) dimming/dulling the following part/following/further

   For me ‘EVA-v’ is a long vowel (‘fat aleph/alap’) (/oː/, /uː/) following by a dimmed /r/ like in אוּר ‘uwr {oor}. So far, the few occurrences in my transliterations correspond with this explanation.

   Karl, Diane, in my view the scribes undoubtful used deliberate similar ‘adjacent pen strokes’ for two different instances to obscure even more. That can easily be taken for a sloppy writing but I believe that people who wrote and copied probably on a daily basis were skilful enough to make use of the very subtle differences. Regarding 2. I see indeed some glyphs which can be assigned univocal to only one letter or letter sequence like the example above ‘EVA-v’ -> אוּר (in Hebrew transcription the Waw after Aleph indicates here a long vowel /oː/, /uː/) and other glyphs which can be assigned to different letters like ‘EVA-o’
   -> ב /b/or ו /v/,וֹ /w/. But, if they are (or were) actually vocal in each case or change only the adjacent Auslaut etc. is a different story.
9. John Sanders on April 19, 2022 at 10:13 pm said:

   Voynicheros been chomping on old EVA’s door, bin at it two score years and more. When they gits through, sure gonna be sore…for there ain’t a durn thing in there.
10. Darius on April 27, 2022 at 5:38 pm said:

    Goddess EVA doesn’t let anybody walk in through this door into the temple…

    As posted above, I see more different glyphs in the VMS than EVA specifies. One another example: EVA-l, I see here two different glyphs, one with equal legs (for me a shin ש) and one where the left leg is significantly longer than the right (a semkath ס). In contrast, I don’t see here different glyphs: EVA-b and EVA-n, but probably it makes a difference how long the upper stroke is to indicate where is the last of the following EVA-i’s which belong to the same word. No wonder that my statistics will be quite different to EVA statistics. Naturally, I don’t criticize the attempt to use keyboard letters for glyphs to make communication easier.

    I read @ Ninja that some people are convinced to see corrections in the text. Yes, I see corrections too. I have pointed to places in the VMS in my documents, which I think are corrections. A new one: f1r, 4th passage, line 5, 3rd word from the left, 2nd glyph from the left – the scribe made a big dot from a ‘c’ – one ‘c’ too much? Corrections naturally disprove hoax theories. One more indication for a meaningful text: a very rare glyph (not specified in EVA) in the middle rosette text direct left to the paragraph division mark (looks like an ‘8’ but the upper circle is open to the left), there is an identical counterpart glyph on the opposite side of the rosette, squeezed between 2 other glyphs. In my opinion 2 bracket signs. But why to take this passage into brackets with a short explanation (5 glyphs) outside of the bracket, if it were not meaningful? Ok, you need to assume it is a bracket.

    Regarding Aramaic – this lingua franca for a period of time had various regional characteristics and time specific characteristics (Chaldean, Palestinian, Nabatean, Imperial Aramaic of the Persian Empire, Babylonian Aramaic, Aramaic found in Qumran Scripts…).
    I would expect to find some Greek loanwords for my assumed period of recording: 1CE, but so far nothing. Otherwise, I read about Daniel (partly written in Aramaic, partly in Hebrew, last revisions assumed 2BCE), where the Greek influence were already drastic, that only 5 Greek loanwords had been found, all connected to music instruments like קַתְרוֹס qiytharoc {kee-thaw-roce’} lyre or zither -> gittare?. Can be the scribes were ‘picky’ in wording in scripts with religious content. On the other hand, I think to have identified the word אִגְּרָה ‘igg@ra’ {ig-er-aw’} missive/letter in VMS and the frequently used דָּת dath {dawth} law/rule which seem to be loanwords from Persian. The language of Daniel is considered to be an example for Biblical Aramaic. But there isn’t a clear definition what Biblical Aramaic in fact is and, as from my point of view, we are here a couple of centuries later than Daniel or Ezra. I see some differences in word-endings as well. But I assume it’s still BA. However, all the language considerations deserve a separate document… something for the future.
11. Darius on April 29, 2022 at 7:12 pm said:

    Distributions and entropies. I wanted to revert to math and to make a big thing out of it later on but now all historians have fallen in love with math… so two words about it from my side.

    Zipf’s law in its simplest interpretation says that the probability to find a particular word A in a longer text is inversely proportional to his rank n in the frequency list: P(X=A) ~ 1/n. For a finite set of values A0…AN (fortunately we have here a finite set) we additionally need a norm factor H, which makes the sum of all the involved probabilities sum(P(X=Ax))=1; x є (0,..,N) so we have P(X=A) ~ H/n when n is the rank of A.

    Now let’s code a text with prime number coding. We take one prime number for every letter and build a product for every word. We take high prime numbers so the factorization would be time consuming. Rather simple coding method, but we can retrieve at least the ‘letter set’ for every word (we let the valid permutations out of scope – all words you can build of the letter set using every letter in it) if we had the key table for the prime numbers involved. Our code would be like 11271 5865 77… (here the numbers are small, it’s only an example). Would we still see the Zipf’s law in the whole coded text? Yes, we would see Zipf’s law for permutation groups, every group had a probability which equals the sum of probabilities of the valid group members. The same would apply to an abjad/ketiv word, all possible combinations of given consonants + vowels insertions which would result into a valid plaintext word, would build a group. And for this group we could calculate a probability. We could build a derived Zipf distribution.

    No matter which coding method, a substitution or mathematical procedure, we could find this distribution pattern and we already had the solution for years!
    But what happened in Voynich to this Zipf or derived Zipf distribution?

    Imagine you build a further product of your coded numerical words as multiplicands, so your final numbers would be e.g. (**) (*)… etc. but you build a product only if it fits into a range of numbers, say v0 < (**…) < vk otherwise you rebuild your chain e.g. (*) (*) (*).

    You distribution function (for Zipf’s law a hyperbolic function) becomes discontinuous, with vast empty value ranges. In the prime number codification example, you would have e. g. |(v0…vk)| = x >> 1 , |(vl…vm)| = 0, |(vn…vp)| = y >> 1, |(vr…vs)| = 0 etc. (|X| is the cardinal number of the set X). The prime number example makes clear: it isn’t even important to use a concrete coding method, the goal is simple to make the distribution discontinuous. You can achieve this with simple substitution + merging of plaintext words into similar vords. That’s what we see in the VMS. You have the freedom to put the words together according to your own rules. E. g you can say all vords should begin with the same letter, if possible (you have to take a letter which frequently occurs as starting letter in your language like EVA-y י (yodh) in Aramaic – that fits perfectly!) and your vords shouldn’t be shorter than 4 and not longer than 12 chars. And you start to code:
    ‘we don’t know what to do when we encounter a weird code’ -> ‘wedontknow whattodo when weencountera weirdcode’

    You generate similar vords and reject concatenations which produce too much diversity, exceptions are still allowed and unavoidable but should be limited. Your distribution is not more continuous but erratic, Zipfs or other continuous distributions are not more recognizable (distributions of finite sets are discrete so they are not continuous in math sense but we don’t need to differentiate anything).

    And then some additional camouflage: for some letters (for 4-5) we introduce capital letters, for some at the beginning of vords for some others at a special position inside the vord (EVA-a as small mem/nun, EVA-m as their capital letter). And some hokus-pokus with the double chars we have sometimes genuine in our plaintext words but much more these double chars we have built with the concatenations:

    ‘wedontknow wha(tt)odo when w(ee)ncountera weirdcode’

    We don’t need to treat them all this way, we can take only the most frequent (in VMS transcribed ll/dd/bb/ww/vv – all these connected c’s with a glyph above), for the rest try even to avoid such concatenations, which would result in doubling of letters (for some letters, however, that wouldn’t be possible, like EVA-i ח heth – too many of these beasts around – too many words with this ending and important words with a starting heth: הֵם hem {haym} they, these, the same, who; הֵן hen {hane} behold, if, whether, though; הַר har {har} hill, mountain, hill country, mount, etc..). So, take them at least at the vord starting part so they heap up there. Make an illusion of one word where are mostly a couple of them and fool the knowledge based non-believers (if one dared to read).
12. Darius on May 1, 2022 at 11:44 am said:

    The font converted all indices to normal letters. In mathematics you can’t show anything without indices.

    However, I hope my intention was apparent: to show how a continuous distribution can be ‘destroyed’ and made erratic by means of building of vast numerical value ranges which are empty and others which are overpopulated in your code (prime number code example) or similar strings patterns which are overrepresented and others with none-occurrence (VMS) through an appropriate concatenation, which e. g. allows only a limited group of glyphs to be used as a starting glyph in a vord. In doing so rare exceptions are however allowed if no chance to concatenate over several words according to the general rule.
13. Darius on May 6, 2022 at 6:31 pm said:

    With my glyph interpretation dogma (in fact my glyph interpretation is different to EVA in at least 12 points: more glyphs, double occupancy, capital/small letter, letter merge over the plaintext word boundaries, …) I was able to decode and translate the entire folio 1r.

    Prof said one should start at the beginning and might be 1r was always the first folio of the script, thematical this folio belongs at the end. If it was the first, so the story is told from the end. It’s in 3 passages about the exodus of the Hebrew Christians from destructed Jerusalem 70 CE. Their faith will later change or will be founded anew in the Greco-Roman tradition, so for them it’s an endpoint time. Their gnostic exegesis will still remain a few centuries onward but will then be considered as a dangerous heresy by the orthodoxy.

    The words here are clear, fair and powerful. They speak for themselves so my comments are short. You can find the decryption documents (the last two are about f1r passage 3 and 4) on my website. Passage 3 describes the aftermath of the devastating war – the Great Jewish Rebellion – and passage 4 gives advices how to act decent in the diaspora, particularly in times of shortage and hunger, and gives concrete guidance for the way out of Israel. Passage 2 is short, the words of passage 1 are unbelievable, I need more time for an adequate comment.

    When you click on my name above you get directly to the salad bar.

This site uses Akismet to reduce spam. Learn how your comment data is processed.