Discussion of "A possible generating algorithm of the Voynich manuscript"

[MarcoP]

I have also noticed that the amounts of the most frequent words in the Voynich manuscript are varying more when comparing their partial amounts in the different sections, if I do equal comparisons in texts of known languages. But then the text samples I have analysed is mostly fiction-type and other stories or novels. I have not compared to more technical texts, such as manuals or encyclopedias, so I don't know if this would be significant with regards to determining if the text is meaningless or not.

Hi Jonas,
I agree about the importance of focussing on technical/scientific texts! Anton recently started a thread about the analysis of a specific type of medieval texts: You are not allowed to view links. Register or Login to view.. We have not made much progress yet, but exploring this "genre" will certainly result in interesting information, if we are persistent enough to collect and compare a sufficient number of documents.


The variability among Voynich sections does not only affect the most frequent words, but goes down to something as basic as digraphs. For instance, the HerbalA and BioQ13 sections have similar length.
The EVA sequence 'ed' occurs 12 times in HerbalA and 1928 in BioQ13.
The EVA sequence 'dy' occurs 510 times in HerbalA and 2089 in BioQ13.
The combined sequence 'edy': 5 vs 1771

The fact that 'chedy' is one of the 3 most frequent words in Currier B sections like Q20 and Q13 and not in HerbalA appears to be related with the most basic differences in digraph distribution. These appear to be a progressive drift in the writing/encoding system and it is extremely unlikely to depend on the contents.

The digraph drift is nicely illustrated in the graphs at the bottom of You are not allowed to view links. Register or Login to view. by Rene Zandbergen (they are based on the CUVA transliteration system).