The Voynich Ninja

A while back I counted the letters in vords and also tried to identify the "rhytm" in the words (specific letterpatterns)
needless to say that I failed in finding anything worthwhile.

But, as a mindgame I regularly go over my past theories and think about improvements.
Today I was asking myself this question: Is it possible if you count the letters in every vord, to find a pattern?

This is useful, because when the text contains a cipher, it is very probable that it has a fixed (block) length.
I am talking about the text, not specifically about the labels.

For example:
* every vord has an even amount of letters.  like 64246228
* every vord has an odd amount of letters 
* every significant vord -length is repeated once  like 66 44 22 32648 66 44 21684 22 44
* ...

In order to do this you will have to make decisions on how to count several letters. For example you could count a gallow as 1 or 2 or count Sh as 3 or 1 or 2 etc.
A quick exercise showed me this is not so simple as it looks and should not be waved immediately.

Perhaps there are other possible patterns. 
If so, we can try to find a way to rule everything out by writing a method that check every possible method.

A related idea is to look at the physical length of each vord as written on the page, rather than attempting to count glyphs. So one could make a probability distribution with length in millimeters on the X axis. Or look at the sequence of length measurements.

Quote:A related idea is to look at the physical length of each vord as written on the page, rather than attempting to count glyphs. So one could make a probability distribution with length in millimeters on the X axis. Or look at the sequence of length measurements.

That's what I also have been thinking of. But, on the other hand, the free way in which vords are flowing through images (plants) does not support the idea.

(28-12-2016, 09:21 PM)Davidsch Wrote: You are not allowed to view links. Register or Login to view......

In order to do this you will have to make decisions on how to count several letters. For example you could count a gallow as 1 or 2 or count Sh as 3 or 1 or 2 etc.
A quick exercise showed me this is not so simple as it looks and should not be waved immediately.
.....

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^  THIS!!!!!!!!!!!!!!!!!!!!!!


Yes, yes, yes, yes, yes, yes, yes!!!!!! THIS is EXACTLY what has been be-deviling me for years (I've mentioned it in some of my responses to Thomas Coon's work). I noticed it years ago but whenever I think I have it ALMOST figured out, I discover that the last 10% is very difficult to reconcile.

I don't want to speak for Thomas, he can comment on his own, but I'm sure he has discovered the same thing—his charts indicate that he has stepped into this murky pool as well.


For those who wonder why all the exclamation points, it's because this is a key aspect of the text. There is a pattern in the way the text is laid down and whether something is a monoglyph, biglyph, or ligature, affects everything... entropy, interpretation of the letters, identification of possible nulls, significance of the gallows, significance of the spaces and more. In other words, the "count" matters—it determines how one breaks down the text.

(28-12-2016, 10:07 PM)-JKP- Wrote: You are not allowed to view links. Register or Login to view.

(28-12-2016, 09:21 PM)Davidsch Wrote: You are not allowed to view links. Register or Login to view......

In order to do this you will have to make decisions on how to count several letters. For example you could count a gallow as 1 or 2 or count Sh as 3 or 1 or 2 etc.
A quick exercise showed me this is not so simple as it looks and should not be waved immediately.
.....

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^  THIS!!!!!!!!!!!!!!!!!!!!!!

Yes, yes, yes, yes, yes, yes, yes!!!!!! THIS is EXACTLY what has been be-deviling me for years (I've mentioned it in some of my responses to Thomas Coon's work). I noticed it years ago but whenever I think I have it ALMOST figured out, I discover that the last 10% is very difficult to reconcile.

I don't want to speak for Thomas, he can comment on his own, but I'm sure he has discovered the same thing—his charts indicate that he has stepped into this murky pool as well.

For those who wonder why all the exclamation points, it's because this is a key aspect of the text. There is a pattern in the way the text is laid down and whether something is a monoglyph, biglyph, or ligature, affects everything... entropy, interpretation of the letters, identification of possible nulls, significance of the gallows, significance of the spaces and more. In other words, the "count" matters—it determines how one breaks down the text.

JKP is right, I've been working on this frustrating thing too. And whenever it feels like I have pulled away the curtain and become the first person in 600 years to stare at the hidden inner mechanisms of the VMS, everything falls apart within a day. But for those few hours it's a wonderful feeling.

JKP is clearly farther along the trail than I am and can see much more of the overall pattern. But it's 100% clear that the way letters are arranged - which letters pair up, which letters follow each other - is often beyond statistical chance and is therefore significant. So Davidsch and JKP, I support both of your work

So is a pattern arising here where the VMS points to being a language.  Can someone here explain the count as a pattern and make it simple please?

Instead of focussing on direct results, I was wondering if we can create a small list of patterns or methods we would like to see.


For example the Roger Bacon cipher works with blocks of 5 letters. So I would have loved to find a way to make blocks of 4, 5 or 6 letters.
Helas, that is not possible.

Another method: take the first letter, then the second, then the third, then the third etc.

All those methods have one big flaw:  even when you do work like that, the letters must add up to an alphabet. That alphabet must contain al least 22 variants and the frequency must add up or the letters must have a variation of about 10. Ten because of the digits 1......10. With those digits we can recreate all the letters we want because 10+9=19 for example.

Let's consider that system, 1-to-10. What would be the function of wordlengths in such a system? Logically the length would give 1 letter. That is not really convenient, from the knowledge point that we have an average of 5 a 6 word length. There would be an enormous amount of Nulls.  Or is it still plausible.

It is more probable that a word contains more letters, for example the letters are always groups of 2. But then we have a problem with all the singles letters.
Also it is not logical that there are so little double letters. At least very few letters You are not allowed to view links. Register or Login to view.. (almosty only ee and ii exist)

Are there more possible systems? Yes, we can take a special character such as a gallow or the 4o or the c_c  as a "signal"-character.  Or the last and/or the first letter of a word.
Although the 4o does not occur on one page, wasn't it f2, that could signal us to take a particular transposition table. 
In general there are 12 letters as first letters and 9 last letters. The most striking and highest occurrences are there for o h and y. 
If we would use grouping for letters we could take  o a e c  or o c a y which You are not allowed to view links. Register or Login to view.. 

But then what? We can also use the gallows and assume all other words are nulls. But what transposition table must we now use. Any suggestions?

There is also a more complex grouping, similar to thomascoon his work, although I hesitate to apply such grouping because it seems it bit too artificial and over-constructed.
Using n-grams as elements does also make the text very small. And also there are the labels. A label of two or three characters then translates to one character or simply one digit.
Of course it is possible that the labels are not descriptive but read a line of text scatted throughout the labels (such as:  david was here) but then the labels would follow a particular line or a more flowing pattern, like an invisible banner.

(29-12-2016, 12:52 PM)Davidsch Wrote: You are not allowed to view links. Register or Login to view....
It is more probable that a word contains more letters, for example the letters are always groups of 2. But then we have a problem with all the singles letters.
...

David, THIS, is the sticking point. This is one of the essential problems/challenges with interpreting the VMS. This one seemingly simple point affects everything, not only interpretation of the glyphs, the glyph patterns, entropy, etc., but also the results of computational attacks.

This is the basic problem that has kept me up more nights than I care to count. It's possible to figure out some of them, I feel I have a reasonably good grasp of it, but some are difficult to interpret and without them, it's difficult to advance further.


Here's why I think it might be so difficult to sort this out even after one has noticed this dynamic in the text... I think it's POSSIBLE that there are nulls, but that nulls may be positional. In other words, in one position, a glyph may carry meaning, but in another position, it may carry a different meaning or none at all.

Ok..why is it not possible to advance further. What are you doing when you have groups? Assigning 1 letter to group, or what?
Enlighten me.  (Did you know, it's very simple to check whether the assignment are good, even though it does not yet form valid words?)

Yes, agreed, there are possible many Nulls, perhaps in a system. You can focus at the nulls, but also at the non-nulls.


I also have a system where almost every letter is a null. The page then was 95% empty. Take for example all the letters that represent a number. o,4,8,9 perhaps some others. We've talked about that beginning of the year. But then, again, we got stuck with less letters. The frequency is too low or ackward and the variety is too low to form 0...9 or alphabet letters.