Why and how the text could be Bavarian

[JoJo_Jost]

The entropy numbers

You can't apply h2 to 19 characters; it only produces nonsense, you need much more text for that. Your own control shows it: the genuine German sentence ("ist heiz unde trucken", Breslauer Arzneibuch) has the lowest h2 of the three. It's true that this fragment is extremely heavy on e and n, but you can't draw any conclusions about the entropy of the whole text from that.

With my generators I get fairly easily close to the VMS. That is not the problem of the VBM.

The repetitions

More problematic, and this is indeed one of my two fundamental problems: the hapax rates and the repetition rates of the VMS.

For the hapax rates I found a solution that was very simple: they depend strongly on the underlying text. By choosing a different underlying text I get close to the hapax rates of the VMS.

The repetitions are harder. The VBM does produce such repetition sequences quite naturally, but across the whole corpus not to the extent the VMS does. However, there are clear indications that the repetition rates also depend on the text. See the following table:

   

You can see that a highly repetitive text like a cookbook can get almost into the range of the VMS - but the gap is actually still somewhat larger than it looks here. How far the repetition rate could be produced by "strange phrasing" is the question that is still open. Open mainly because I haven't yet been able to pin down an assignment for the core groups of the VMS. Only with an assignment do I know which clusters repeat, and can I search specifically for phrasings that do this in German. - There really are very strange forms of writing in medieval texts - for example things like litanies or blessing formulas, or actual magical repetitions of single words as reinforcement.

But the repetitions are also where my generators have failed so far, otherwise I would have published them already. Because the rest fits quite well, especially with German as the language. If you look at the repetitions in Bavarian dialects in a phonetic spelling, a solution might lie there, but that too is still speculative. (Other European languages do considerably worse - although of course I haven't tested all of them with the VBM. Mainly the European standard languages.)

But there are other questions as well: Is there another "reasonable" explanation for the repetitions?

1. LAAFU: If you look at the LAAFU effects, the question remains, for example, whether the line-start markers build up an additional layer of encryption that could be connected with the repetition rate. Because what else are they supposed to be? My tests so far, however, show only small effects of the start marker on the repetitions within the line.
   
2. Null texts: Or are there really null texts? Only, null texts are a "standard excuse" for solvers. You have to be very careful with that, even though it was a very common method in the ciphers of the time.
   
3. A further layer of encryption: I'm currently testing, for example, page-wise encryption approaches.
   

What is so strange: the VBM is theoretically able to turn the VMS into readable text in a very simple way. The strong dependence between the final and the initial letter in the VMS speaks for not reading the spaces as real spaces.

The structures of the VMS, with the different cores and table-like structures, also fit a vowel/consonant encryption fairly well.

Only the repetition rates of the VMS don't fit yet - that's where things stand.