+- The Voynich Ninja (https://www.voynich.ninja)
+-- Forum: Voynich Research (https://www.voynich.ninja/forum-27.html)
+--- Forum: Theories & Solutions (https://www.voynich.ninja/forum-58.html)
+--- Thread: A Numeric Solution (/thread-5204.html)
RE: A Numeric Solution - Jorge_Stolfi - 04-01-2026
(04-01-2026, 03:43 AM)Legit Wrote: You are not allowed to view links. Register or Login to view.All translation attempts so far that use Latin number glyphs as abbreviations have been invalidated. This is also why I am suggesting these number glyphs could be intended to be used as numbers.
One problem with this theory is that the frequencies of glyphs are quite non-uniform. If the text of the VMS is a bunch of numbers, then presumably each "word" would refer a word in a code book. But then the digits should have roughly equal frequencies. (Maybe someone has a sample of actual codebook-encrypted text that we could use to check this claim?)
Worse, the frequencies of glyph pairs are extremely different from those expected in a codebook-encrypted text. Like, in the file I have at hand, freq(qX) = 5405/85614 = 0.063 (where X means "any letter"), freq(Xo) = 16308/85614 = 0.19, but freq(qo) = 5297/85614 = 0.061 -- not 0.063 x 0.19 = 0.012. That would be like finding that, in some codebook-encrypted text, almost every digit "7" is followed by a "3".
All the best, --stolfi
RE: A Numeric Solution - Legit - 04-01-2026
(03-01-2026, 09:28 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.Thank you for this refreshing (and LLM-free!) first post!
To be fair, EVA was constructed with the express purpose to be able to talk in a more human-friendly and typing-friendly way about Voynichese. otaiin is just easier to work with for a person typing and reading than 0#a111. In a way it became too successful, because as you point out, it makes the thing seem more alphabet-like than it actually is.
The resemblance to numerical and non-alphabetical characters has been remarked before, but I'm not sure if many people are aware of it to this extent. You present it in a pretty clear and structured way.
Now EVA-v is extremely rare, and I wouldn't take it into account for something like this. So you're looking at potential allusions to 0, 1, 2, 4, 8, 9.
Then there's the fact that Voynichese glyphs sort themselves in a word in a way similar to Roman numerals. This gives you EVA e, ch... as potential C (100).
The question would be if any of that is relevant, and where to go from there...
(04-01-2026, 12:03 PM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.(04-01-2026, 03:43 AM)Legit Wrote: You are not allowed to view links. Register or Login to view.I was wondering about someone like Baresch. Keeping in mind that Baresch was looking at the manuscript 2 centuries after it was written and the common number system was closer to what we use today, I'm not surprised he was confused. The amount of information he would have had available was miniscule. Drinking mercury and leach treatments didn't help either.
I can't agree with your assessment of Baresch's knowledge on the matter.
While it's not impossible that there is a numeric solution to the MS and it has been suggested many times before (You are not allowed to view links. Register or Login to view.), but visually I think the script is designed to look like text and I think medieval readers who don't know what this is would perceive it as some kind of text and not as strings of numbers.
Edit: by the way, the post I linked above mentions letter assignments for different digit-like shapes:
"In Latin, all of the VMS “numeral” shapes can double as letters:
- The “7” that looks like a caret sometimes stood for “a”.
- The “8” often stood for “s” or “d”.
- The “9” symbol could be a number or a very common Latin abbreviation used at the beginnings and ends of words.
- In early medieval Latin texts, EVA-l (numeral “4”) was used as an abbreviation symbol, a convention that had mostly disappeared by the 15th century."
I mean the amount of information Baresch had versus what we have at our fingertips today. Yesterday I scrubbed over a 200 page manuscript to examine the few pages containing illustrations in a few minutes. He would have had to travel and know what he was looking for to begin to cross reference such information.
I'm aware of the Latin VMS numerals doubling as letters in Latin scripts. Reasoning against this is my main point. I would add that if these Latin number glyphs were intended as doubling for letters, there is already an "a" symbol EVA-a. If EVA-v was meant as an "a" and not a 7, why still have EVA-a? While possible, to have some numeric symbols represent abbreviations and others not represent abbreviations seems more unlikely in any cypher or substitution/abbreviation language scheme.
Stolfi: until it's solved, there's no evidence of a language or a cypher either right? Would you agree that claiming these glyphs as abbreviations is a claim to the source being Latin?
If we continue to profile our scribe. Their spoken language is either medieval German or Italian. So their initial draft would be in one of these languages. Then their method of obfuscation would not need another step before writing it as voynichese. Is there some argument to be made that a medieval German person would translate their ideas in Latin and then translate again to voynichese? If Latin is not part of this process, there is no reason to expect these glyphs used presently in the 15th century as numbers (and not commonly as abbreviations) to be considered in their alphabetical form.
I'm afraid treating these symbols as Latin and not 15th century German numbers is begging the question, and implies a Latin solution that we don't have.
RE: A Numeric Solution - Jorge_Stolfi - 04-01-2026
(04-01-2026, 01:46 PM)Legit Wrote: You are not allowed to view links. Register or Login to view.If we continue to profile our scribe. Their spoken language is either medieval German or Italian.
The Scribe was probably from North Italy or Central Europe, judging by the style of dresses, castles, etc.
Quote:So their initial draft would be in one of these languages
The Author -- the person(s) who devised the alphabet, chose the contents, composed the text, specified the figures -- was probably not the same as the Scribe -- the person who actually put pen to vellum. There is some evidence and some arguments indicating that they were different persons; and AFAIK there is no evidence or argument for the opposite.
But even if the Scribe was the same as the Author, the underlying language of the VMS was not necessarily his native language. I bet that most vellum books written in Europe in the 1400s were in Latin, and many were in Greek or Hebrew; even though practically no one spoke those languages (outside of the Church and Greece, of course).
All the best, --stolfi
RE: A Numeric Solution - oshfdk - 04-01-2026
(04-01-2026, 01:46 PM)Legit Wrote: You are not allowed to view links. Register or Login to view.I'm aware of the Latin VMS numerals doubling as letters in Latin scripts. Reasoning against this is my main point. I would add that if these Latin number glyphs were intended as doubling for letters, there is already an "a" symbol EVA-a. If EVA-v was meant as an "a" and not a 7, why still have EVA-a? While possible, to have some numeric symbols represent abbreviations and others not represent abbreviations seems more unlikely in any cypher or substitution/abbreviation language scheme.
I don't think one can build any argument about the script using EVA-v. I think it appears only once in the whole MS if we don't count f57v. And I don't think we can treat f57v rings as a generic sample of the writing, since they show a large number of unique glyphs that look quite unlike any other Voynichese glyphs.
Also, design-wise, EVA-a and EVA-v could be just the same symbol. a is absent from the 4x17 ring, and the first appearance of a d-like symbol in this ring is of rare j form, so it's not very far fetched to assume that v is just a rare form of a.
(04-01-2026, 01:46 PM)Legit Wrote: You are not allowed to view links. Register or Login to view.I mean the amount of information Baresch had versus what we have at our fingertips today. Yesterday I scrubbed over a 200 page manuscript to examine the few pages containing illustrations in a few minutes. He would have had to travel and know what he was looking for to begin to cross reference such information.
This is not obvious to me. Baresch could have had access to a large number of 100-200 hundred year old manuscripts contemporary to the Voynich manuscript ('s vellum), those that have been completely lost since then. I would say that if anything Baresch likely had more information than we do. In any case, this is just about hunches and intuition, either way.
RE: A Numeric Solution - pfeaster - 04-01-2026
(04-01-2026, 12:47 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.One problem with this theory is that the frequencies of glyphs are quite non-uniform. If the text of the VMS is a bunch of numbers, then presumably each "word" would refer a word in a code book.
There are numerical scenarios that don't involve sequentially numbered word lists -- for example, simply substituting ascending numbers for letters of the alphabet (A = 1, B = 2, L = 10, M = 11, etc.) or encipherment via Polybius Square (A = 1-1, B = 1-2, F = 2-1, etc.), either of which would or could involve uneven digit frequencies.
Not that I'm suggesting either of those particular examples would be a much better statistical fit overall. But the hypothesis that Voynichese is partly or wholly numerical doesn't necessarily go hand in hand with any one specific type of encoding mechanism. It just implies some kind of reliance on numerical ordering or counting.
Or number forms could also be borrowed, even consciously, w1th0ut r3f3r3nce 2 the1r num3r1cal valu3s.
RE: A Numeric Solution - Rafal - 04-01-2026
In many real cases numbers aren't equally frequent in sets of data:
You are not allowed to view links. Register or Login to view.
RE: A Numeric Solution - Jorge_Stolfi - 04-01-2026
(04-01-2026, 03:03 PM)pfeaster Wrote: You are not allowed to view links. Register or Login to view.simply substituting ascending numbers for letters of the alphabet (A = 1, B = 2, L = 10, M = 11, etc.)
Well, if "a" means 1, and 1 means "k", then "a" means "k". That would be a simple alphabetic substitution cipher. And those were already excluded...
(... if the underlying language is assumed to be European ...)
(04-01-2026, 03:03 PM)pfeaster Wrote: You are not allowed to view links. Register or Login to view.or encipherment via Polybius Square (A = 1-1, B = 1-2, F = 2-1, etc.)
That would leave the words twice as long as those of the original language. But one vexing feature of Voynichese is that its words are shorter than normal...
All the best, --stolfi
RE: A Numeric Solution - Jorge_Stolfi - 04-01-2026
(04-01-2026, 03:09 PM)Rafal Wrote: You are not allowed to view links. Register or Login to view.In many real cases numbers aren't equally frequent in sets of data:
You are not allowed to view links. Register or Login to view.
Indeed, but that law applies to the leading digit only. It is a consequence of the fact that smaller numbers are more likely to occur than larger ones. The frequencies of digits in general (not just the leading ones) should be much more uniform.
All the best, --stolfi
RE: A Numeric Solution - Aga Tentakulus - 04-01-2026
If you take all the available information and look for the common denominator, the target area looks something like this.
For anything written before 2010, regardless of who wrote it, I don't give a penny for their answer unless they confirmed it again after 2010.
RE: A Numeric Solution - pfeaster - 04-01-2026
(04-01-2026, 03:27 PM)Jorge_Stolfi Wrote: You are not allowed to view links. Register or Login to view.Well, if "a" means 1, and 1 means "k", then "a" means "k". That would be a simple alphabetic substitution cipher. And those were already excluded...
In that particular scenario, the digit 1 could also form part of a compound such as K (10) or X (21).
11o9o10o5 19o8o9o18 6o14o17 5o21o1o12o15o11o5
1417 o51118o5 12o122o2o5 11o910o5 19o8o918
(Note twenty-five tokens of "1" versus no tokens at all of "3", plus reduced entropy....)
But again, I wasn't intending to promote any specific numerical hypothesis, and certainly not that one. I'm just pointing out that an argument against one particular numerical system -- a sequentially numbered word list -- doesn't serve as an argument against a numerical system in general.