The website http://www.voynichese.com/ is a fascinating addition to our tools for analysing the manuscript. Here I will just point out one or two things it has helped me to notice which I think are interesting:
1. Look at the word EVA:qokeedy as analysed in voynichese.com: http://www.voynichese.com/#/exa:qokeedy/517
What is curious about this is how the word is relatively rare in the first 73 folios, but then suddenly explodes in the Balneological section from f75r onwards until f84v, the last page of the Balneological section, when it suddenly stops, to resume again at the first page of the Recipes section (f103r) and continue quite frequently to the end of the manuscript. In other words it is very frequent in two sections only, but in those sections it is strikingly common.
2. The same seems true of another word beginning with the same sequence: EVA:qokain, as you can see here: http://www.voynichese.com/#/exa:qokain/917 .(Note a particular concentration f111 and 116r)
We should not draw too many conclusions from this without a fuller statistical analysis, simply because there is more text on those pages than on the pages with more or larger illustrations. But it does seem to be true that earlier pages of the manuscript have relatively fewer occurrences of these sequences.
Why is this interesting? It seems to me that the reasons for such imbalances in word distribution, if they prove to be true, could be due to a number of factors.
One of these might be to do with theme – these words might be particularly related to the ‘balneological’ topic and the ‘recipe ‘topic (whatever they might be).
Another might relate to scribal choices – it could be that we have different scribes working on different parts of the manuscript and they chose different spellings or ways of writing words from those in other parts – so we might be able from such analyses to identify who wrote which part. (See this discussion of scribal variation)
Another value of such analysis might be to unlock the recipe pages at the end. It has been suggested that each of the ‘paragraphs’ in those pages might relate to different parts of the manuscript.
If we can find clusters of words on particular pages, and then see if they are repeated in particular paragraphs of the recipe pages, maybe we be able top see if indeed those recipe paragraphs link up with particular pages in the earlier parts of the manuscript?
Note that with some smaller clusters there seem to be a similar phenomenon. For example, the same pattern appears to be true with EVA: l at the start of a word: http://www.voynichese.com/#/exa:l-/0.
This could again be a scribal choice. Whatever the reason, it is odd that it is so frequent on some later pages ( e.g. http://www.voynichese.com/#/f111r/exa:l-/317) while on earlier pages with a lot of text it does not occur at all (e.g http://www.voynichese.com/#/f58r/exa:l-/300).