The Voynich Ninja

I thought of two simpler tests.

In cases where there are multiple paragraphs that seem to be "related", and the paragraphs are of different sizes, the musical hypothesis predicts that there should be statistical differences between them. In the simplest case, probably some glyphs are more common in large or small paragraphs. My thinking is that if the paragraphs are intended to have the same duration, but some contain more written text, then:

• If the larger paragraphs contain more notes, the notes must be shorter, which probably means different rhythmic indications.
  
• If the larger paragraphs contain more lyrics, then the proportion of lyrics to non-lyrics will be larger.
  
• If the larger paragraphs contain notes which are more complicated somehow, then the proportion of the text which encodes those complications will be larger.
  

Browsing through the pages at the beginning of the MS, I don't see a lot of good test cases, though. If, to my eye, the paragraphs appear to be "related", then they also usually have similar sizes.

The other prediction is that the last words of paragraphs should have something in common: a longer rhythmic value. Across the whole MS, shorter durations shouldn't appear at the end of paragraphs. In sets of "related" paragraphs, the last words of each should be even more likely to have the same rhythmic value.
Also, the last words of "related" paragraphs might often have the same pitch, but this is a weaker prediction. Across the whole MS, the final words of paragraphs should not be as clearly aligned in pitch as they are in rhythm. The ideal result for this hypothesis would be that final words of "related" paragraphs have two things in common, one of which is shared across unrelated paragraphs and one of which isn't.

I did a quick manual examination, starting at the beginning. I skipped You are not allowed to view links. Register or Login to view. and You are not allowed to view links. Register or Login to view. because they had more than two paragraphs, and You are not allowed to view links. Register or Login to view. f6r You are not allowed to view links. Register or Login to view. because they had only one paragraph, and stopped there. I'm including the first and last word of each of the paragraphs in each pair, since it seems reasonable that pieces will begin on a consonance, if both parts begin at the same time. (By the way, that's an explanation for paragraphs of different lengths that I overlooked: some of the parts might begin with long rests.)

f1v:
kchsy… chokody
potoy… chodar
f2r:
kydainy… chol dan
kydain… chcthy
f2v:
kooiin… chodaiin
kchor... dolody
f3v:
koaiin… chom
tchor… yteam
f4r:
kodalchy… cthy
pydaiin… cthey
f4v:
pchooiin... cphody
torchy… chtody
f5r:
kshody… chocthy
tshy… shody


Some observations:

• All of the paragraphs here begin with a gallows.
  
• There are several strong resemblances: beginning of You are not allowed to view links. Register or Login to view. (kydainy/kydain), end of You are not allowed to view links. Register or Login to view. (cthy/cthey), end of You are not allowed to view links. Register or Login to view. (cphody/chtody).
  
• There are some weaker resemblances: end of You are not allowed to view links. Register or Login to view. (cho-), end of You are not allowed to view links. Register or Login to view. (-m), and maybe the use of "s" in f5r.
  
• The final words tend to end in -y, but I'm aware that's not unusual.
  

I wonder if this could be explained by someone generating a nonsense text and partially copying from the paragraph above?

Hm, the beginnings of f2v  f3v  You are not allowed to view links. Register or Login to view. all have -iin in the first paragraph (more specifically, -ooiin or -oaiin) and in the second paragraph they all have a permutation of "ch", "or", and a gallows with "both feet down" ("t" or "k").

"I wonder if this could be explained by someone generating a nonsense text and partially copying from the paragraph above?"

You would be interested in Torsten Timm's work

I additionally looked at f3r, which has four paragraphs with the following first and last words:
tsheos... cthom
pcheol... qokeey
tsheoarom... sam
pcheoldom... am

Again, all the paragraphs begin with a gallows. It's tempting to go on a flight of fancy and imagine that the last notes, with -m on three "parts", have three notes of the same pitch class, maybe even with the tenor and bass lines in unison (-am), and the soprano an octave above (-om), and the alto line in between, maybe a fourth or fifth above the bass. This would be, I believe, historically appropriate.

But alas. I did the simplest statistical test, comparing frequency distribution of letters in paragraphs of different sizes, for three sets of "related" paragraphs - You are not allowed to view links. Register or Login to view. f3r and You are not allowed to view links. Register or Login to view. - with a negative result. In some folios, some letters are proportionally more common in longer or shorter paragraphs, but the patterns do not hold across the three folios.

For differences caused by rhythm in particular, I suppose you could say that maybe the pattern would reverse depending on the meter of the piece. For example, in modern notation, quarter notes are more common in the faster-moving part of a piece in 3/2, but more common in the slower moving part in a piece in 3/8. We could also say that maybe the rhythm is not specified for every note, only when it changes, or something. But that's definitely adding complications.

Perhaps the patterns we do see are from pitches, if different parts have different smallish ranges, or simply tend to hang out on different pitches. The letters which vary in frequency the most between paragraphs on the same folio are: c and h, strongly, and e i l, less strongly. In f3r, m has this characteristic too somewhat.

If lyrics are present, I think they must be present on every note. Is there any small set of segments, exactly one of which is in just about every word?

Overall, this is evidence against the music hypothesis. It's not impossible, but there are fewer ways for it to be possible.