- Options
- Login
- Desktop Style
Options
Voynichese is a numeric cipher?
- Voynichese is a numeric cipher?
-
RE: Voynichese is a numeric cipher?
JoJo_Jost > 12-07-2026, 09:34 AM
In my thread, you presented your theory as if it required fewer assumptions or rules to explain all the statistical peculiarities that VBM can explain.
You are not allowed to view links. Register or Login to view.
in reply to:
You are not allowed to view links. Register or Login to view.
I’m replying in your thread in accordance with the forum rules so as not to confuse the thread topics; I hope that’s okay.
I took the time to check this out—though I’m not entirely sure what exactly is part of your theory. So please feel free to correct me if I’ve made a mistake.
Your system using Roman numerals, as you describe it here, requires:
1. Plain text to which a Roman numeral is assigned for each letter.
2. Sorting in descending order within a word.
3. A second artificial alphabet that maps these numerals to Voynich glyphs.
4. bigrams that encode letters OR abbreviations OR morphemes as needed (actually, that’s already several rules, but never mind.
5. freely insertable zero characters from an open list
6. arbitrary word separation and merging (without any rules?))
7. two cipher tables—one with letters and the other with abbreviations. (which, however, makes decryption nearly impossible).
8. Symbol values that are “invented” to force individual glyphs into the system — for example, EVA x = 150 is justified by the astrological quincunx angle. But what about all the other numbers???
9. Adjustments at the end of words “for aesthetic reasons.”
9 rules versus 4 to 6. So Ockham’s razor works against you, not for you.
And even with these 6-9 rules—some of which allow for a high degree of flexibility—your model still lacks a mechanism for:
a. Long-range dependencies that span three glyphs AND cross a word boundary. Your per-word sorting makes cross-word dependencies impossible by design.
b. The steep Zipf curve of the VMS bigram distribution. The descending sorting per word smooths out precisely the structure that the language generates.
c. The fact that a small set of positional rules predicts over 90% of placements at word boundaries, because free zero insertions and arbitrary word breaks have too many degrees of freedom to produce such strict constraints.
d. Repeated sequences of four “chol” or five “okedy.” According to your scheme, the plaintext would have to contain the same content four to five times in a row—which is syntactically implausible in a herbal.
e. Your system does not explain the distribution of word lengths among the VMS words, unless you used arbitrary word breaks, which in turn allows for a high degree of freedom and makes the cipher less plausible, once again making it very difficult to decrypt.
f. You have no idea why these LAAFU effects should occur; it makes no sense in your cipher.
g. The system does not explain the sequence of similar words (words that differ by only one letter), unless it involves consistently inserting zeros in the same sequence—which would result in the system having completely nonsensical degrees of freedom. After all, the VMS could then consist of just three sentences, with the rest being all zeros. Such an assignment is meaningless.
h. The many hapaxes that don’t fit natural language—and they’d have that problem anyway, since the words remain roughly the same regardless.
i. The problem that descending sort order causes many (!) words to map to identical words, which in turn cannot be back-translated.
I'm sorry, but aside from the fact that your system suffers massively from too many lines of freedom, it makes too many assumptions and still can't explain all these statistical peculiarities clearly enough without making further additional assumptions. But that's exactly what I don't do with the VBM
No hard feelings—I don’t mean any harm! I normally wouldn’t have chimed in here at all, because I found the cipher idea fundamentally clever and interesting. But as it stands, it’s still too underdeveloped to fit the VMS (I read the thread). And I started out with a lot of “nonsense” myself—that’s normal and part of the process; it might never stop until, perhaps by chance or not, we can finally translate the entire VMS.
But you threw the ball into my court, so I had to pick it up. I hope you understand. 
-
RE: Voynichese is a numeric cipher?
ololololo > 12-07-2026, 10:27 AM
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.I’m replying in your thread in accordance with the forum rules so as not to confuse the thread topics; I hope that’s okay.
No problems
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.arbitrary word separation and merging (without any rules?))
There is no problem with combining conditional, for example, sheor, because you can immediately see that it is XX XIX. If we go by the example, there are no rules, and you can combine the words.
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.two cipher tables—one with letters and the other with abbreviations. (which, however, makes decryption nearly impossible).
In fact, I'm not so sure that abbreviations can be used here. I don't see a difference in encrypting -us as two characters instead of one, as it will still be in its correct position (according to logic. If it were to move to a different location, it would be more difficult to decipher).
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.Symbol values that are “invented” to force individual glyphs into the system — for example, EVA x = 150 is justified by the astrological quincunx angle. But what about all the other numbers???
About x, it was just a guess. It may have some other meaning (but it is most likely related to astronomy or something similar, as it appears in the You are not allowed to view links. Register or Login to view. chart with other unidentified symbols).
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.9 rules versus 4 to 6. So Ockham’s razor works against you, not for you.
Are these the exact rules? If I wanted to combine a word to make it more "beautiful," that's not a rule; it's just my whim. Just because I can do something doesn't mean I should do it. "Rules" refers to the latter.
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.And even with these 6-9 rules—some of which allow for a high degree of flexibility—your model still lacks a mechanism for:
I know. I'm not particularly good at statistics, and I can't create something that will reproduce the Voynichese text as statistically similar as possible. But at the same time, I didn't pursue this goal. In that example, I showed that it's quite easy to transform a Latin text into a Voynich-like one.
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.Repeated sequences of four “chol” or five “okedy.” According to your scheme, the plaintext would have to contain the same content four to five times in a row—which is syntactically implausible in a herbal.
What I described in my post and Voynichese are different things! When I said that it was easy to turn Latin text into something similar to VMS, I didn't say that I had solved the encryption scheme and knew the contents of the manuscript. I didn't say that I had reconstructed the encryption algorithm of the manuscript. The purpose of this example is to show that Roman numerals produce a similar result. Similar, but not exactly the same!
Moreover, at the moment, I can't describe exactly how Voynichese might work. I can only provide a rough description of what it might contain. It could be features of the cipher alphabet that minimize the number of repetitions or something similar.
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.You have no idea why these LAAFU effects should occur; it makes no sense in your cipher.
I didn't think about LAAFU/BAAFU when I was doing this. It wasn't part of my plan
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.The system does not explain the sequence of similar words (words that differ by only one letter), unless it involves consistently inserting zeros in the same sequence—which would result in the system having completely nonsensical degrees of freedom. After all, the VMS could then consist of just three sentences, with the rest being all zeros. Such an assignment is meaningless
In my example, the similar words are just numbers that are aligned in order. It is only natural that if we have XVII XVI, we will have two words that start with the same letter (if XV is considered to be one letter).
I would also like to point out that my example is not to be confused with VMS. You do not claim that VMS is encrypted using the Naibbe cipher, do you?
Similarly, I can turn a Latin text into a Voynich-like text, but I can't translate or accurately reproduce the manuscript text.
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.it makes too many assumptions and still can't explain all these statistical peculiarities clearly enough without making further additional assumptions.
There's still a lot of work to be done on the manuscript's cipher... So far, that's all I have to say.
I think that the current drawback of my theory is that I don't have a methodology. I can only speculate about how the VMS text is formed, but I won't be able to provide any details. It might be helpful to create similar ciphers and compare them with the text, potentially leading to the development of even more similar models based on similarities and differences. This could potentially lead us to the actual cipher...
Thanks a lot for the reply!
P.S. I think the main problem is that a very inexperienced person took over the theory of the number cipher... That's first.
The second thing I want to point out is that Roman numerals are not You are not allowed to view links. Register or Login to view.. @Bluetoes101 did a very good job...
I think the fact that the numbers somehow parody Voynichese very well (not as well as I would like, but nevertheless) deserves special attention.
I also ask you to separate the degrees of freedom from the assumptions. The "freedom" of action will not affect the cipher in any way. I can change the order of the letters or combine the words, but the text itself will not change. An assumption would be if my text was not very similar, and I decided: "Let's randomly add daiin and other words to make it look similar." Possible examples of nulls and "numeric" repetitions can be found on You are not allowed to view links. Register or Login to view. and on f15v. For some reason, my example of encryption reminded me of the text on the first pages of botany, specifically f4r. The differences are in the "frequent" elements (VMS qo is more common). Maybe it's just my imagination. -
RE: Voynichese is a numeric cipher?
ololololo > 12-07-2026, 10:57 AM
(12-07-2026, 09:34 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.freely insertable zero characters from an open list
And also about the nulls. The thing is, there's nothing wrong with them, they're not mystical or politically incorrect, and they don't affect the cipher in any way. I think if I write the word AoPoPoLoE, you'll understand what kind of fruit I'm referring to. And I don't think the nulls are a serious rule. It's more of a trick or a cleverness. -
RE: Voynichese is a numeric cipher?
JoJo_Jost > 12-07-2026, 11:38 AM
(12-07-2026, 10:27 AM)ololololo Wrote: You are not allowed to view links. Register or Login to view.There's still a lot of work to be done on the manuscript's cipher... So far, that's all I have to say.
I think that the current drawback of my theory is that I don't have a methodology. I can only speculate about how the VMS text is formed, but I won't be able to provide any details. It might be helpful to create similar ciphers and compare them with the text, potentially leading to the development of even more similar models based on similarities and differences. This could potentially lead us to the actual cipher...
I can only encourage you to do exactly that. In the beginning, it’s always just an idea that then develops, goes in the wrong direction, takes a different turn, gets discarded—only to give rise to a new theory, which then develops, goes in the wrong direction, takes a... ...just like all of us
I guess that only stops when you give up, die—or when you’ve really found the right one.... But maybe someone will figure it out someday—no one knows who it will ultimately be. -
RE: Voynichese is a numeric cipher?
ololololo > 12-07-2026, 03:39 PM
(12-07-2026, 11:38 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.
\(12-07-2026, 10:27 AM)ololololo Wrote: You are not allowed to view links. Register or Login to view.There's still a lot of work to be done on the manuscript's cipher... So far, that's all I have to say.
I think that the current drawback of my theory is that I don't have a methodology. I can only speculate about how the VMS text is formed, but I won't be able to provide any details. It might be helpful to create similar ciphers and compare them with the text, potentially leading to the development of even more similar models based on similarities and differences. This could potentially lead us to the actual cipher...
I can only encourage you to do exactly that. In the beginning, it’s always just an idea that then develops, goes in the wrong direction, takes a different turn, gets discarded—only to give rise to a new theory, which then develops, goes in the wrong direction, takes a... ...just like all of us
I guess that only stops when you give up, die—or when you’ve really found the right one.... But maybe someone will figure it out someday—no one knows who it will ultimately be.
I can never give up! In general, I can still do something.
-
RE: Voynichese is a numeric cipher?
ololololo > 12-07-2026, 05:44 PM
As for the mini-cipher, it is currently worth finding out whether it can create a plausible Voynichese text and how plausible it will be. I think I can do it.
By the way, I would like to clarify a few things about my thesis. When I said it was easier, I meant that it could be done manually, and that my cipher didn't require any phonetic schemes, special rules, etc. You simply replace the numbers with letters, and that's it. -
RE: Voynichese is a numeric cipher?
ololololo > 13-07-2026, 05:34 PM
I don't think this is seriously likely, but I have a theory that the scribes may have worked independently of each other in terms of encryption...
If we assume that the scribes are the authors of the manuscript, and they knew its content and knew everything about the encryption algorithm, and they made this book for themselves, then it doesn't seem so absurd.
This could be effective as a security measure against hacking. If we decrypted Botany, we wouldn't decipher Balneology, and vice versa.
It is possible that the numerical alphabet changed slightly, but the meaning of the Voynichese symbols remained the same. It is also possible that other methods were used (in Currier B, the simplification of the cipher seems to result in an abundance of chedy words and detached morphemes such as aiin and edy. However, this may not be the only explanation).
I can also assume that there may be some kind of "base number" underlying the alphabet. This could be, for example, 10 added to the ordinal number of the letters, or in some other cases.
This could explain some of the quirks of that mini-cipher and the peculiarities of Voynichese. For example, a (ei) could be a trace of this "base number" that was added to one (which somewhat explains why there is no combination of ie). -
RE: Voynichese is a numeric cipher?
ololololo > 14-07-2026, 08:01 PM
Here are a couple more sentences from Lorem ipsum that I decided to encrypt.
Donec quam felis, ultricies nec, pellentesque eu, pretium quis, sem. Nulla consequat massa quis enim. Donec pede justo, fringilla vel, aliquet nec, vulputate eget, arcu. In enim justo, rhoncus ut, imperdiet a, venenatis vitae, justo. Nullam dictum felis eu pede mollis pretium. Integer tincidunt. Cras dapibus. Vivamus elementum semper nisi.
XV XIV V IV III / XVI V V IV / XXI XX XIX XV X / XVIII XIV XII XII IX IX VII VI I / XXI XII V / XXI XX XVII XII IX V I / XIV V III / XXI XXI XXI XX XX XVI XII V I / XX VII V V / XXI XVIII III I
XIV IX / XIV XIII IX V / XXI XX XIX XV X / XXI XIX XVIII XV XIV VIII III / XXI XX / XX XVIII XVI XIII IX IX V V IV / I / XXI XX XIX XIV XIV IX V V I / XXI XX IX V I / XXI XX XIX XV X
XXI XIV XIII XII XII I / XXI XX XIII IX IV III / XIX XII IX VI V / XXI V / XVI V V IV / XIX XV XIII XII XII IX / XXI XX XVIII XVI XIII IX V
XX XVIII XIV IX VII V V / XXI XX XX XIV XIV IX IX IV III / XIX XVIII III I / XXI XIX XVI IX IV II I / XXI XXI XXI XIX XIII IX I / XXI XX XIV XIII XIII XII V V V / XIX XVIII XVI XIII V V / XIX XIV IX IX
o oke a ok iin / chor oin daiin dy / eor dan ol d a / chol sheol oiin dan or or aiin / ote aiin / chor sheor oin on ote dan dan a a a a / chor a / chor sh oiin on daiin or a / chor cheor oin ol / eol daiin dy (without nulls)
Inside an anagram, we have the right to rearrange the letters. We can also add numbers inside the anagram to get a different letter from the alphabet. In the case of Roman numerals, this won't result in any loss of information or make the decryption more difficult (for example, to get VI, you can simply add V and I. I believe this can be quickly achieved visually).
1). o oke a ok iin - a moves between ok and iin: okeo okaiin
2). chor oin daiin dy - The problematic oin can be replaced with... The gallows p/cph: chor cphdaiin dy (or chor cphdy daiin)
3). eor dan ol d a - The last a turns into ey. Add a garbage ch: cheor chedan chol chd chey (maybe cha)
4). chol sheol oiin dan or or aiin - A little trick: we split or or into oro, and r goes into the next word: chol sheol oiin dan oro raiin
5). ote aiin - oteaiin
6). chor sheor oin on ote dan dan a a a a - a series of the last a's is disturbing, but we can add them up to get XX (sh): chor sheor cphy qo otey dan dan sh (maybe dan shdan).
7). chor a - chor ey - chor chey
8). chor sh oiin on daiin or a - chor shoiin qodaiin or am
9). chor eor oin ol - chor cheor cphol
10). eol daiin y - cheol daiin chy
The nulls are highlighted in bold. As you can see, the text can be "edited" without using any rules.
Again, I would like to point out that this is a clear example, not an attempt to accurately reproduce the text of the manuscript. -
RE: Voynichese is a numeric cipher?
ololololo > Yesterday, 07:13 PM
By comparing the Voynichese texts with my cipher, I can identify areas for improvement.
There may have been more variations for numbers in a specific position in a number, next to other numbers, in a specific area of a line, etc.
For example, XV can be o, but XV in the number XVI is, let's say, cth. I in all positions is i(n), but in combination with XV it is or. Therefore, XVI will be written as cthor, not as on.
This can explain some of the LAAFU effects, as well as the placement of certain characters (for example, at the beginning of a new paragraph, a number turns into a gallows, a character at the beginning of a line turns into an s, etc.)
Again, this variability does not complicate the cipher in any way, but it does complicate its analysis for an unwanted decrypter. -
RE: Voynichese is a numeric cipher?
magnesium > Today, 12:45 AM
My general point of view on cipher models is that it's healthy to build them out and then test their properties against Voynichese. For example, every cipher model out there makes some kind of assumption as to what a Voynichese token represents on average. This should have immediate implications for how Voynichese tokens relate to each other. In principle, this sort of thing could be modeled and tested through statistics like mutual information and the co-occurrence of ciphertext word types, akin to what Montemurro and Zanette did for the Voynich Manuscript itself: You are not allowed to view links. Register or Login to view.
This would be a fun way of testing the behavior of various cipher proposals, such as my Naibbe cipher, Matlach et al.'s steganographic cipher, this numeric cipher, and JoJo_Jost's "vowel bridge" cipher (on specifically Bavarian).