The Voynich Ninja

<f75r.P.13;V>      pchedy.keedy.qokedy-
{figure}qokedy.qokedy.qokedy

Pchedy keed… end end end end?

@grove: You are not allowed to view links. Register or Login to view.

Ha! Forgot I read that before. I was thinking in English of course and thought four consecutive ends would have been interesting.

Continued from: You are not allowed to view links. Register or Login to view.

The big problem is figuring out the exact coding of the matrix. Since we don’t know the writing conventions, more complex frequency analysis of such a cipher is impossible.

Here’s an example. Let’s stick with “nd” and “nde.”

Let’s examine the various German texts I’m using as a basis for their occurrences:

[attachment=16341]

Here you can see the full extent of the problem: Calculated per word: Breslauer 7.65 percent, Ortloff 0.26 percent. And this is mainly due to a difference in the spelling of the word “and” as “unde” in Breslauer and as “vnd” and “vnnd” in Ortloff.

Added to this are spelling conventions: The “cooking recipes” contain an extremely high number of “und,” while the Bavarian text has relatively few. But the lack of consistent spelling rules—b/p, t/d, a/o, etc.—also makes it nearly impossible to analyze the text based on frequency if you don’t know which spelling convention it follows.

And the final problem is this: The other families will also encode clusters, and as long as we don’t know exactly which ones, it’s impossible anyway.

That’s why I’m still not sure if it can be solved at all. The most likely approach is probably via the “Crips”—that is, the peculiarities of word repetitions in the VMS—but even that is extremely time-consuming. I don’t know if I can manage it..

Your method is ingenious and surely interesting (and I think there have been few studies which considered spaces not to be true spaces). I always assumed spaces to be real spaces, if not because they are the most frequent character in the VMS, just as in every other language. But equating spaces to vowels works too, because after eliminating the spaces from a text the remaining most abundant characters are vowels.


A problem I see is that, if I understood correctly, the cipher will always encrypt the same plaintext string to the same encrypted string, which should create many cases where the same Voynichese words repeat in the text, which is not what is observed.

(05-07-2026, 05:22 PM)Mauro Wrote: You are not allowed to view links. Register or Login to view.Your method is ingenious and surely interesting (and I think there have been few studies which considered spaces not to be true spaces). I always assumed spaces to be real spaces, if not because they are the most frequent character in the VMS, just as in every other language. But equating spaces to vowels works too, because after eliminating the spaces from a text the remaining most abundant characters are vowels.

Thanks

(05-07-2026, 05:22 PM)Mauro Wrote: You are not allowed to view links. Register or Login to view.A problem I see is that, if I understood correctly, the cipher will always encrypt the same plaintext string to the same encrypted string, which should create many cases where the same Voynichese words repeat in the text, which is not what is observed.

That's a very interesting question, and it's essentially correct. 

If you only transcribe the vowels, then it would work:

Let’s define
e/ea = y[space]qo
i/ie = y[space]ch
o/oi = y[space]sh
a = y[space]o

Different variations of “Home”
y othy shmy qo... = at Home
y chnhy shmy qo... = in Home
ty shy shmy qo... = to Home
fry shmhy shmy qo... = from Home

(btw: here u see, why Line Start Markers are needed, like in VMS)

The consonant-vowel cluster “hy shmy qo” would therefore theoretically be the same and would have to repeat. I think that’s what you mean.

However, everything after “qo” would depend on what comes next;

But it’s not just the vowels that are encoded—the encoding is somewhat more complex:

As already mentioned, internal consonant clusters are encoded differently.

The “hy” would therefore depend on which word precedes it.

othy
chnhy
shy
shmhy

These consonant clusters are replaced partly by the E-family and partly by other families (aiin) or others. This, in turn, can also influence the vowel part on the right.

So “shmy qo” or only "my qo" would remain a possible repetition.

What I haven’t written yet, because I haven’t quite deciphered it myself (i have some ideas):

It must be the case that when a word ends on a vowel (like "home" here) and a new one begins with a vowel, a specific cipher is applied again—precisely because two vowels cannot be encrypted consecutively. In that case, the “y qo” would also be omitted in these instances in this example.

What would remain is the "shm" or “m” as the core of “Home,” which is repeated—and that would no longer be recognizable as a repetition.

In VMS, bigrams and trigrams do, in fact, recur quite frequently.

But i'm working from the Bavarian dialect, and it is an almost monosyllabic language, there aren't that many longer words. Many Words are: = Consonants Vokal Consonants = "CVC". If "CVC" meets "CVC" then you have CVCCVCCVCCVCC And the CC Cluster are part of the E-Family and others. So there is no repeat, too.

Addendum:

Yes, I think that so far no one has come up with the idea of viewing the spaces as part of the vowels, as apparent as that may be.

But to decipher the VMS, you have to think extremely outside the box—because everything else has already been explored a thousand times over....

And the beauty of it is that this is an extremely simple form that fits the 15th century.

(06-07-2026, 08:57 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.Yes, I think that so far no one has come up with the idea of viewing the spaces as part of the vowels, as apparent as that may be.

I'm absolutely sure many people have considered this before. It's been over 100 years of active research. However, this hypothesis runs into many problems, that's probably why it never got traction. For example, how does this approach explain the labels? Labels are overwhelmingly consist of a single word, labels with spaces are a minority. Are you suggesting there are no vowels in labels?

Why I'm sure many people have considered this before is because this hypothesis a simple subset of a more generic hypothesis that several Voynichese characters encode one plaintext character. There are many theories like this. When people run computational attacks on these generic theories this particular sub theory is also included.

(06-07-2026, 11:15 AM)oshfdk Wrote: You are not allowed to view links. Register or Login to view.

(06-07-2026, 08:57 AM)JoJo_Jost Wrote: You are not allowed to view links. Register or Login to view.Yes, I think that so far no one has come up with the idea of viewing the spaces as part of the vowels, as apparent as that may be.

I'm absolutely sure many people have considered this before. It's been over 100 years of active research. However, this hypothesis runs into many problems, that's probably why it never got traction. For example, how does this approach explain the labels? Labels are overwhelmingly consist of a single word, labels with spaces are a minority. Are you suggesting there are no vowels in labels?

Labelese could lack vowels if they are primarily numerical in nature.