The Structure of the VMS
Just a quick note for anyone who wants to visualize the beautiful structure of the VMS. This is a Sankey diagram. It’s structured using reverse Eva transcription (read backward line by line). From the first vowel bridge (i.e., the two glyphs to the right and left of the space) to Slot 3 (counting forward according to the old slot numbering) to Slot 2 to the next vowel bridge.
The symmetry arises in part because the path isn’t actually traced; instead, the calculation simply goes to the next slot and starts over from there. The diagram shows summarized partial transitions, but not necessarily the actual complete individual paths.
Still, it’s a nice visual, isn’t it? 
Notes: The image was generated using the AI after extensive calculations and structural analyses. However, I can’t guarantee that it’s entirely accurate—I did cross-check it, but in this case, I’m not sure if such a test is sufficient. The original is an HTML file with exact numbers, but I can’t post that here.
It is intended more as a nice illustration than as a scientific study...
Okay, I've been thinking about how to make it clearer what I'm doing here, and I came up with the idea of doing it from the cipherer's perspective,
How the Author of the VMS developed the VBM (Vowel Bridge Model)
To explain how the VMS was encoded, I’ll start with a text in English—though this is merely for illustrative purposes, since, of course, the vast majority of readers do not speak “German.” In fact, however, Bairish is in many respects a better fit for the VMS than some other 15th-century European languages.
The cryptographer knew that in Germanic (like english) languages, “e,” are by far the most common letters, and that, given this knowledge, any substitution cipher would no longer be sufficient in the 15th century.
So he had a brilliant idea: He writes every vowel in the text as a bigram— for example consisting of a “y” and another letter. A typical vowel bigram would look like this: "e" = "yo".
But of course, even that wouldn't have been enough, so he came up with the idea of deleting all the word boundaries in the text and placing the new word boundaries exactly between these bigrams:
"e" = "y [space] o"
And then it immediately becomes clear why he chose the “y”—one of the most well-known ligatures of the time; in Latin, the symbol that looks like a 9 is used for the ending “us.” Many endings used this ligature, so back then everyone must have assumed it was a Latin text.
To take this even further, he chose two vowel bridges for the vowel “e.” As mentioned earlier, these were “y[space]o” and “y[space]qo.” Many words begin with “qo,” and “qo” is reminiscent of the Latin “quod,” among other words.
The other vowels were for example encoded using different letters following the “y.” ="yd"
Let’s see what happens when you encrypt a normal English text using only this simple method. Original Text:
"To make an eye salve for a stye: take equal amounts of leek (or onion) and garlic and crush them well together. Take equal amounts of wine and ox bile and mix them with the alliums. Put the mixture into a brass vessel and leave it to stand for nine nights. Strain it through a cloth and clear it well, then store it in a horn. Around nighttime, apply it to the eye with a feather. This is the best remedy.For a white spot or film on the eye: smear hare's bile onto the eye; it is said to clear within two days. For blurred or dim eyesight: smear the juice of centaury (the herb the old text calls earthgall) onto the eyes to sharpen the sight. Adding honey helps as well. For a headache: take the lower part of a houseleek, crush it, soak it in cold water, and grind it well until it foams; then bathe the head with it."
Result:
"ty lmy dky qoy dny ty qosy dlvy qofy lry dstyy qoty dky qoy qoqy shy dly dmy sntsy lfly rky lry lny chy lny dndgy drly chcy dndcry shshthy qomwy qollty lgy othy orty dky qoy oqy shy dly dmy sntsy lfwy chny qoy dndy lxby chly oy dndmy chxthy qomwy chththy qoy dlly chy shmspy shtthy omy chxty shry qoy chnty ly dbry dssvy ossy qoly dndly kvy oy chtty lsty dndfy lrny chny qony chghtsstry yny chtthry sghy dcly lthy dndcly kry chtwy qollthy qonsty lry qoy chty chny dhy lrny dry sndny chghtty chmy oy dpplyy chtty lthy oy ty owy chthy dfy kthy qorthy chsy chsthy qoby qostry omy odyfy lry dwhy chty ospy lty lrfy chlmy lnthy qoy ty osmy krhy dry osby chly qoy lnty lthy qoy ty oy chty chssy ydty lcly krwy chthy chntwy ldy ysfy lrbly shrry qody lrdy chmy ty qosy chghtsmy krthy ojy pcy oy lfcy qonty arythy ohy qorbthy qoy lldty oxtcy dllsy krthgy dlly lnty lthy oy ty osty lshy drpy qonthy osy chghty dddy chnghy lny thy qolpsy dswy ollfy lry dhy kdy dchy oty dky othy qoly sy qorpy drty lfy dhy ssy oly rkcry shshy chtsy cthky chty chncy lldwy dty qory dndgry chndy chtwy olly shnty chly chtfy cthmsthy qonby dthy othy qohy kdwy chthy cht"
(Note: If you're viewing this translation using a browser translation tool, it may not display in its entirety.)
I coded the english vowels using the VMS vowel bridges based on their frequency:
e = y qo / e = y o
i = y ch
a = y d
o = y l
u = y sh
ea = y k
ou = y s
ow = y s
ey = y t
ai / ay = y y
ee = y r
oa = y cth
ui = y p
au = y a
What’s immediately obvious: You can’t make out anything anymore. That’s why I used an English text at the beginning here—to make that point clear. It looks like pure nonsense.
This achieves the following:
- The actual word boundaries are made completely invisible.
- The consonant clusters of the original language are completely broken up and regrouped.
- Artificial “prefixes” and “suffixes” are created, which arise purely statistically because certain vowels in English often follow certain consonants.
I fed this text into Gemini’s Pro version to try to decrypt it. It didn’t succeed—but it was on the right track. The many “y”s at the end are striking. Now, almost all the vowels are missing except for “o.” And there are consonants that repeat very often.
A skilled cryptologist would probably be able to decrypt the text quickly.
But what’s already striking about the similarities to the VMS is:
- The uniform, almost static word distribution
- The limited word lengths
- The almost repetitive occurrences of similar words: cthky, chty, chncy
- Words that differ by only one letter, such as qoqy, qohy, qody, qoby, etc. This is already very reminiscent of our qokeedy words.
- Repetitive phrases are also already appearing: “lry dhy kdy dchy”—4x, “shy dly dmy sntsy”—3x, etc.
- And the text’s seemingly highly structured nature.
- And the many hapax, which contradict any normal linguistic structure!
-----------------------
Of course, the cipherer was also aware of these weaknesses, so he had to do three things:
1. Conceal the “y” endings
2. Reintroduce vowels into this cipher
3. Disguise the consonants. In German, these would be “n” and “d”
What am I trying to show with this?
This single, very simple step alone transforms a text into something resembling Voynich structures. Of course, there are no Voynich-like words yet—a few more cipher steps are needed for that—but it points in the right direction. That was the reason why I began to explore this idea more deeply, and why I continue to do so.
It would be an easy and sufficient explanation for the possibility that VMS is indeed based on a language. But I am aware that more needs to be done to prove this. And I will continue to approach this further here.
I think I have the same message for several threads today
Looking similar to Voynichese and being statistically similar to Voynichese are two different things. It is not hard to invent a scheme that makes a text took somewhat like Voynichese. Your simplified scheme here does that. However, this has never been the problem, as far as I know. The problem is to reproduce the whole collection of paradoxical features, like word/token length distribution, proportion of single token words (hapax), lack of large scale repetitions, line position preferences, etc.
It's not hard to invent a scheme that would explain any single statistic, it's when trying to combine these schemes to reproduce the whole everything usually breaks down.
(26-06-2026, 07:48 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.The problem is reproducing the entire set of paradoxical features, such as word/token length distribution, the proportion of single-token words (hapax), the lack of large-scale repetitions, line position preferences, and so on.
Yes, oshfdk, I agree with you 100 percent! Precisely because I see it exactly the same way.
But that is precisely what is so amazing about this method—that it achieves exactly that with just a few simple tools. Above all, the small number of repetitions, because this is precisely how an extremely large number of hapaxes are formed—since they are, so to speak, consonant runs.
In the next steps, I will show that the word-length distribution matches almost exactly The hapax rate is an approximation (and I’ll also explain why it can only be an approximation), and it reveals the word repetitions (qokedy qokedy qokedy qokedy)—I’ve already covered that in the Crib, etc.
Believe me, I’ve come to understand a great deal about the structures of the VMS by now.
And I’m not adapting the system to the VMS; rather, I’m developing a cipher in which that happens almost automatically… that’s a huge difference.
(26-06-2026, 06:30 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view."ty lmy dky qoy dny ... qonby dthy othy qohy kdwy chthy cht"
Your encypherment will be difficult to read. Suppose the text was somewhere in a page packed with words and you wanted to reread the bit about hare's bile. How would you search for it? In the English text you just scan the words and find "hare's bile". Even if you could not locate the words immediately the associated words "eye", "remedy", "smear" would guide you to the right place. But would you recognise "krhy dry osby chly qoy" in the encyphered text?
This sort of problem is going to be awkward for any of the cypher hypotheses. I mentioned it before, the gunpowder scenario,
You are not allowed to view links.
Register or
Login to view.
(26-06-2026, 08:03 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.And I’m not adapting the system to the VMS; rather, I’m developing a cipher in which that happens almost automatically… that’s a huge difference.
So far I have no idea how this could possibly work the way you propose, and I think I understand your method. However, this is an unnecessary discussion, if you can actually pull this off and achieve a good statistical match, the numbers will speak for themselves.
As a side note,
(26-06-2026, 08:03 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.and it reveals the word repetitions (qokedy qokedy qokedy qokedy)
I was talking about the lack of large scale repetitions, not about the repeated words.
(26-06-2026, 08:33 AM)dashstofsk Wrote: You are not allowed to view links. Register or Login to view. (26-06-2026, 06:30 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view."ty lmy dky qoy dny ... qonby dthy othy qohy kdwy chthy cht"
Your encypherment will be difficult to read...
First, you have to decode the text along with the ciphers—that much is certain.
I’ve come to believe that anatomical topics were being discussed, and that was extremely dangerous in the Catholic regions of Bavaria. And so it had to be encoded in such a complex way.
(26-06-2026, 08:56 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.So far I have no idea how this could possibly work the way you propose, and I think I understand your method. However, this is an unnecessary discussion, if you can actually pull this off and achieve a good statistical match, the numbers will speak for themselves.
Yes i agree again.
Quote:As a side note,
I was talking about the lack of large scale repetitions, not about the repeated words.
Yes, I realized that; as I said, this type of cipher produces a Hapax rate similar to that of the vms. I am aware that only a small number of words are repeated, while the vast majority are not. This clearly argues against a “normal” language in a word-for-word translation or substitution.
And that’s exactly what makes this theory so interesting. Because that’s exactly how it is. Precisely because consonant bridges—those from the previous word (without a space) and those from the following word—are encoded. And that’s exactly what leads to this.
(26-06-2026, 09:40 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.I’ve come to believe that anatomical topics were being discussed, and that was extremely dangerous in the Catholic regions of Bavaria. And so it had to be encoded in such a complex way.
In your opinion, was our author not only a physician knowledgeable in anatomy, a pharmacist, and a herbalist, but also an exceptional cryptographer, ahead of his time?
(26-06-2026, 10:18 AM)Ruby Novacna Wrote: You are not allowed to view links. Register or Login to view.In your opinion, was our author not only a physician knowledgeable in anatomy, a pharmacist, and a herbalist, but also an exceptional cryptographer, ahead of his time?
No, not at all; all of this fits perfectly with the cryptographic practices of the early 15th century. In fact, I tend to think it's not really a very good cipher.
And anatomical examinations of corpses were being conducted at the same time (in other Countries) - and even earlier.