The Voynich Ninja

The drift could be explained by a scribe changing or refining his invented script over time, it happens often that people forget scripts they invented themselves, or that they come up with improvements or detect shortcomings to the script.

Still, not only is it strange that someone or multiple people would choose to create a new script instead of using a standard alphabet, but that it would coincidentally reuse share glyphs seen in other cipher manuscripts, many of them used in typical abbreviations to be fair, sure † , and placing them in their usual position (y at wordbreaks). Shouldn't we expect to see more marginalia in some European language if there were no motivation to conceal the information to some extent? Some small footnote like chol = cho4 or some glossa for ambiguous words like 'shi'? I would argue that some desire to conceal information must exist at the very least (not that it excludes the possibility of a non-european language underlying an invented script)

Edit:
† A set of glyphs standing for whole syllables (like y = us) would be a good starting point for writing languages like these I guess, but why keep a reduced inventory of glyphs and why not use them more freely within vords?

(13-06-2025, 03:52 PM)Letieum Wrote: You are not allowed to view links. Register or Login to view.

(13-06-2025, 03:28 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.These are more consistent with a poorly done imitation/parody/fake than an honest transliteration of a language.

"consistent with a fake" maybe, but not "poorly done" ...
If the VM is the result of some sort of "gibberish-generating process", it is a a surprisingly complex process for its time, with a clever mix of rules and randomness !

Clever yes, but naïve, without control of even the most basic statistics like frequencies of glyphs and their bi-grams, tri-grams, let alone more advanced statistics like MATTR, word pair correlation, etc. I'm sure a Chi² test would be interesting on these basic statistics: I don't remember if a study has done exactly that and compared with real historical texts, to show just how weird these statistics are in the VM.

It is not Chinese for the same reason that it is not any transliteration or simple substitution cipher of any language: Voynichese is not flat.

Nick Pelling Wrote:Even back in 1962, Elizebeth Friedman – having been a top US Government code-breaker for several decades – was able to note that all attempts to decrypt the Voynich Manuscript as if it were a simple language or single-substitution alphabet were “doomed to utter frustration”. That is, if you wind the clock back half a century from the present day, it was already clear then that Voynichese’s curious lack of flatness was strongly incompatible with:
* natural languages
* exotic languages
* lost languages
* monoalphabetic (simple) substitution ciphers, and even
* straightforward hoaxes

You are not allowed to view links. Register or Login to view.

(12-06-2025, 08:34 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.What I find strange in this theory:
1) Total lack of notes in Latin or another native language of the Author. If the Author's goal was not the concealment of information, but spreading/keeping the knowledge, I can imaging each page of the original transcription would be heavily annotated.

First, in that proposal, the Author wrote down what some local Reader was dictating.  At that point, there would be no time to add notes to the transcript.

But, more importantly, the Author had only a limited knowledge of the language.   As a foreigner who got there as an adult, he probably knew enough of the (spoken) language to order food, ask the way to the bathroom, negotiate purchases, shipments, bribes, ... So he probably could not understand most of the contents of those books, and would have no useful notes to add.  

And, even if he did, he would have written those on paper, and they probably would not be worth transposing to vellum. 

Again, while he was not trying to conceal information, he was not overly worried about making the transcription accessible to others.  If not for himself, he may have intended to give the book to one other scholar.  In this case, besides telling this scholar the sounds of the script, and providing a dictionary of the terms which he knew, he would leave to this scholar the problem of understanding the text.

Quote: 2) Lack of any images or inscriptions that would unambiguously link it to the Orient. Even if the Author couldn't draw well, sketching a few important symbols or objects seemed reasonable.

What images would there be in the images in the books that the author presumably would choose to transcribe?  Unidentifiable plants and plant parts -- check.  Incomprehensible  astronomical diagrams -- check.  Organs of the human body -- check.  What else?

When copying the illustrations, the Author must have copied only those details that (he thought) were significant, leaving out any ornamentation or obvious details.  As I proposed, the drawings in the VMS were mainly created by the Scribe -- based on sketches by the Author, but will all ornamentation and non-essential details (like the battlements on castle walls) made up.

That said, the Chinese solar calendar consisted of 360 (not 365, not 365.25) divisions, each covering one degree of the Ecliptic, grouped as 24 You are not allowed to view links. Register or Login to view. of 15 degrees each.  The cycle started around 4-5 February.  Coincidentally, the VMS zodiac starts with Pisces, has 360 nymphs/stars, and the first few diagrams have 15 each.

Also, Chinese books were bound on the right side.  Now turn page You are not allowed to view links. Register or Login to view. upside down, so that the binding is on the right side.  What do you see?

Quote:3) Generally, I think this would have been partially solved a long time ago, if this was just a faithful phonetic representation of a natural language, regardless of the language. Start with the labels, try identifying what they mean using images and repeated labels, find these labels in the text, ..., get a solution

IIRC, the labels are mostly all distinct; and when a label occurs in the text, it is usually near its occurrence in the figures.  Isn't that so?

Early missionaries who tried to write primers for Chinese claimed in frustration that the language "had no grammar".  AFAIK it does have a grammar alright, only that it lacks the most salient features of "Classical" languages (Indo-European, Semitic, Finnish, Hungarian, Turkish, etc.) .  No articles, no genders, no plurals, no cases, no verb inflections, and even the verb "to be" is rarely used. 

Quote:There are attempts like this a few times a year, but they quickly break down, because there is no consistency in labels and how the same words are used in the text.

Yes, there have been many such attempts.  But most of them failed because their strategy was 

  1. use those image clues to identify the meaning of a few words, 
  2. guess the language, 
  3. using 1 and 2, determine the encoding, 
  4. using 3, decode the rest of the text 

The attempt usually failed at 2.  Because of the style of the drawings, people
typically guessed a language that could have been known in Europe at the time
(initially the 1500s, then early 1400s after the carbon dating).

It is an easy guess that the label doaro dcholday oalcheol on f68r3 (or part thereof)
is the name of the Pleiades.  But in which language? Could be Vietnamese, Burmese, Tibetan, Thai, Khmer, or one of the 20+ "dialects" of Chinese ... as they were 600 years ago. 

And we know that most of those languages changed a lot in those 600 years.  While the pronunciation of languages with alphabetic or syllabic script changes mostly by general phonetic shifts, a language with a logographic script can change the pronunciation of any single ideogram from "aaba" to "zzyzx", independently of all the others, in a few decades.

(13-06-2025, 04:29 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.It is not Chinese for the same reason that it is not any transliteration or simple substitution cipher of any language: Voynichese is not flat.

Some of the "bumpiness" features listed by Nick seem to affect a relatively small fraction of the text, and (like transcription errors, or missing fragments on clay tablets) should not be a big obstacle to decipherment.  

It is almost certain now that one-leg gallows are variants of the two-leg gallows, possibly combined with e or other letters, that are used mostly on the first line of a paragraph -- a convention that the Author picked up from the typical European manuscript style.   When single-leg gallows occur inside a parag, I would guess that the Scribe failed to see a parag break in the Author's draft and thus started the first line of the second parag as continuation of the last line of the previous one.   I don't see reason to ascribe any other meaning to single-leg gallows, just as I don't think that split two-leg gallows or fancy decoration on gallows have any linguistic or semantic value.

Some, if not all, of the free-standing columns of letters are likely to be the work of later owners, and thus should be ignored.

The peculiar features at the "margins" of the pages can have banal explanations too.  For one thing, on many languages the final letters or words of a sentence may be strongly affected by the topic.  In a narrative of past events, sentences are more likely to end with "-ed" in English, with "-ta" in Japanese; whereas in a herbal the sentences should be mostly in the present tense, hence  more likely to end with "-desu" or "-masu" in Japanese.  

Others have pointed out the increased use of abbreviations at end of lines in European manuscripts.  And, based on the few "classical" texts I have looked at, narratives in Hebrew and Ge'ez (and presumably Arabic) will often start a sentence with "Ve/We/Wa" (= "And").

Quote:Either you learn the original script, or you use the script you know to represent the sounds of the foreign language. [...] there is no way that Voynichese would have been the spontaneous result of Latins representing a foreign language in a more accessible way.

That is not an absolute fact; it depends on the goal of the script's inventor.

I would say that, in general, when someone creates a script for a language A that is based on the script of a language B, the goal is either to bring the speakers of A into the cultural sphere of B, or teach the spoken language A to many speakers of B.

An example of the former goal is the creation of the modern Turkish alphabet by Atatürk, based on the Latin alphabet with a few sound changes and some diacritics (and treating the dot on the "i" as a diacritic).  It was part of his goal of bringing Turkey out of the "Arab" sphere and into the "European" one.  Other examples are the alphabets created by the Soviets for many non-IE languages, based on cyrilic.   Another example is the pre-Atatürk Turkish script and the Farsi script, both based on the Arabic script, which were part of the inclusion of Turkey and Persia into the Islamic sphere.

Examples of the second goal are the scripts invented by Christian missionaries for Mandarin, Vietnamese, Quechua, Maya, Nahuatl, Guarani, etc..  Their goal was to train missionaries who could be able to preach and convert in those languages.  Thus staying as close as possible to the Latin script was an obvious necessity.  Ditto for the Esperanto script.

But when neither of these goals exist, new scripts are often invented from scratch, without any attempt to imitate any script that was familiar to the inventor or to other speakers  of the language.  Saint Mesrop was familiar with the Greek and Latin scripts, and probably also Hebrew and Arabic; but the alphabet he designed for Armenian has practically no similarity with any of these, except a couple of letters.    The You are not allowed to view links. Register or Login to view. syllabary resembles Chinese characters in the shapes of the strokes, but it is not a modification of the Chinese script.   Japanese hiragana and katakana, as well as Korean hangul, are similarly (not) based on the scripts known by the creators and their intended audience.  The You are not allowed to view links. Register or Login to view. are another example.  And the You are not allowed to view links. Register or Login to view. is another.

And, according to the "Chinese theory" as I outlined before, the Author of the VMS di not have any of those two goals.  So his choice of a totally different alphabet, rather than a Latin-based one, is not at all unreasonable.  As noted before, the Voynichese script seems to have been designed with writing speed as the main goal.

(13-06-2025, 07:01 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.First, in that proposal, the Author wrote down what some local Reader was dictating.  At that point, there would be no time to add notes to the transcript.

But, more importantly, the Author had only a limited knowledge of the language.   As a foreigner who got there as an adult, he probably knew enough of the (spoken) language to order food, ask the way to the bathroom, negotiate purchases, shipments, bribes, ... So he probably could not understand most of the contents of those books, and would have no useful notes to add.  

Any arrangement is possible, but overall it looks extremely weird that the Author concentrated on recording the sounds of incomprehensible speech instead of trying to figure out the meaning by discussing it with the Reader and translating or at least annotating it on the go. The Reader was probably an educated person, being able to read in the XV century.

One scenario that would look somewhat plausible to me, is if the initiator of the whole process was the Reader and not the Author. So, certain Oriental scientist wanted to spread their teachings to the world and being unfamiliar with the Latin script and understanding that the local logographic script would take too much time to teach, decided on inventing a simple phonetic alphabet that would be easy to the European pupil to write down with a quill. Then insisted that the whole book was written this way without any barbarian Latin scribbles, to be able to verify that the message is properly recorded.

(13-06-2025, 07:01 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.What images would there be in the images in the books that the author presumably would choose to transcribe?  Unidentifiable plants and plant parts -- check.  Incomprehensible  astronomical diagrams -- check.  Organs of the human body -- check.  What else?

For example, a single word written in the original Oriental script, a sample of Oriental characters, a few numbers in the original number system, an unambiguously Oriental object, fruit, utensil. I understand, that the Scribe had to recreate images from imperfect sketches, but generally the Scribe had no problem drawing things that he never witnessed (like the Scorpio or the Dragon - which looks like a European dragon to me), so why are there no images showing likely Oriental objects?

(13-06-2025, 07:01 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.Also, Chinese books were bound on the right side.  Now turn page You are not allowed to view links. Register or Login to view. upside down, so that the binding is on the right side.  What do you see?

Sorry, I don't like riddles. I know, it sounds ironic. I see a manuscript upside down. I remember from a while ago, that the red weirdos are supposed to look like some Chinese characters, but personally I don't see a lot of similarity between the red weirdos and properly written 兀 or 几, if that's the proposed reading. Also, if the folios are supposed to be read upside down, then we have to conclude that the nymphs and plants were likely imported from Australia. Sorry for being a bit snarky, but there is more than enough ambiguity and uncertainty in the Voynich Manuscript itself for me to want some more on top 

(13-06-2025, 07:01 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.IIRC, the labels are mostly all distinct; and when a label occurs in the text, it is usually near its occurrence in the figures.  Isn't that so?

Not to my knowledge. As far as I remember, there are quite a few repeated labels, including relatively long ones, and absolutely none of them make any obvious sense. If my greps are correct, these are the repeating labels of 7+ characters:

2x  otcheody
2x  otaraldy
2x  otaldar
2x  otalaiin
2x  olkeedal
2x  okolShy
2x  okeey.ary
2x  oeeodaiin
2x  cheoldy

If we include 6 character labels, there are some that repeat 5 times. If each label meant the same thing each time it appeared, there would have been very plausible explanations by now. I don't know, maybe the labels are not words but references ("fig. 1").

EDIT: I don't know if they preferably appear close in the text. I took the first label in the list above and tried to find it in the transcription:


It appears fairly close once in a circular inscription and once more in a completely different part of the manuscript.

(13-06-2025, 07:01 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.

Quote:There are attempts like this a few times a year, but they quickly break down, because there is no consistency in labels and how the same words are used in the text.

Yes, there have been many such attempts.  But most of them failed because their strategy was 

  1. use those image clues to identify the meaning of a few words, 
  2. guess the language, 
  3. using 1 and 2, determine the encoding, 
  4. using 3, decode the rest of the text 

As far as I know, there have been a few language agnostic manual and computational attempts, trying to ascribe meaning to various sequences of characters based on context. Some of them looked a bit naïve, some more sophisticated. I didn't study them in depth, but I know that so far they failed to produce anything convincing for the public.

(13-06-2025, 07:41 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.

(13-06-2025, 04:29 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.It is not Chinese for the same reason that it is not any transliteration or simple substitution cipher of any language: Voynichese is not flat.

Some of the "bumpiness" features listed by Nick seem to affect a relatively small fraction of the text, and (like transcription errors, or missing fragments on clay tablets) should not be a big obstacle to decipherment.

There are small oddities, perfectly explainable as mistakes, spelling variants, abbreviations, etc. and then there are massive inconsistencies, much less explainable: the big differences between Currier "languages" and all the many "dialects" on a smaller scale.

(13-06-2025, 09:29 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.There are small oddities, perfectly explainable as mistakes, spelling variants, abbreviations, etc. and then there are massive inconsistencies, much less explainable: the big differences between Currier "languages" and all the many "dialects" on a smaller scale.

I think most of these large scale changes can be explained by the evolution of the script. If it was a newly invented script, it was very likely to go through various reinterpretations and optimizations when used in practice for the first time. Also true if the intent was writing down an unfamiliar language, only imagine the consequences of finding out that <of> and <'ve> are two different things half way through the manuscript 

(13-06-2025, 09:37 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.

(13-06-2025, 09:29 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.There are small oddities, perfectly explainable as mistakes, spelling variants, abbreviations, etc. and then there are massive inconsistencies, much less explainable: the big differences between Currier "languages" and all the many "dialects" on a smaller scale.

I think most of these large scale changes can be explained by the evolution of the script. If it was a newly invented script, it was very likely to go through various reinterpretations and optimizations when used in practice for the first time. Also true if the intent was writing down an unfamiliar language, only imagine the consequences of finding out that <of> and <'ve> are two different things half way through the manuscript 

Perhaps, but it's a bit extreme. Why would they want to:

- Omit all e on two random pages: You are not allowed to view links. Register or Login to view. and f36r.

- Concentrate some patterns on some pages and omit them in an entire section, for example You are not allowed to view links. Register or Login to view. has a lot of [e:h]od, [e:h]os: 14 eod, 3 hod, 2 eos, 1 hos. No [e:h]or, [e:h]ol, -eo. It is the opposite of Q13, but both are Currier language B. There are many examples of extreme concentrations and total absence of common patterns on many pages.

(13-06-2025, 09:52 PM)nablator Wrote: You are not allowed to view links. Register or Login to view.Perhaps, but it's a bit extreme. Why would they want to:

- Omit all e on two random pages: You are not allowed to view links. Register or Login to view. and f36r.

- Concentrate some patterns on some pages and omit them in an entire section, for example You are not allowed to view links. Register or Login to view. has a lot of [e:h]od, [e:h]os: 14 eod, 3 hod, 2 eos, 1 hos. No [e:h]or, [e:h]ol, -eo. It is the opposite of Q13, but both are Currier language B. There are many examples of extreme concentrations and total absence of common patterns on many pages.

I have no in-universe answer for this (I mean the Reader-Author-Scribe plot), but out-of-universe these are simple. I'm a happy follower of "it's a cipher" cult, so different cipher keys cause different glyph sequences. There was nothing assigned to e on You are not allowed to view links. Register or Login to view. or f36r, so no e was recorded. And f57v is an oddity in itself.