[split] Verbose cipher?

[kckluge]

Taking account of the entire resulting inventory of the remaining single characters, plus all of Koen's bigrams, trigrams, and n-grams above, in this way we have produced a character "alphabet" of about 25 values:

[e], [y]
[d], [k], [t]
[s], [l], [r]
[ch], [sh]
[ain], [aiin], [aiiin]
[air], [ar], [al], [am]
[or], [ol]
[ok], [ot], [od]
[qo], [qok], [qot]

Of course there are a handful of additional rare characters/letters as well. But the above inventory should account for the overwhelmingly vast majority of the Voynich ms text.

This general idea is a seductive one that's been floating around for 20ish plus years (I've been one of the people floating it off and on). There is a big problem with it which it took me far too long to recognize, and which I think calls the whole idea into question:

Consider the specific breakdown proposed above. With r, l, d, k, and t serving as the second glyph in a pair as well as standing alone, then you should see substantial numbers of doubled glyphs where the plaintext letter represented by [xy] is followed by the plaintext letter represented by [y] (i.e., if [ok] = 't' and [k] = 'h', then every time "th" appears in the plaintext you'd have "okk" in the ciphertext).

This does not happen much in the Voynich text. Doing a quick and dirty count on the pages in the D'Imperio transcript and translating to EVA, even ignoring spaces the only repeated glyph pairs (other than EVA 'i' in word-final ai*<x> combinations) with greater than single-digit counts are:

    Count    % digrams (neither is weirdo)
ii:   15     0.0283%
dd:   21     0.0397%
oo:   40     0.0756%
ll:   46     0.0869%
yy:  145     0.274%
cc: 1420     2.68%

Note that d and l are the only glyphs from the proposed breakdown that serve as both the second glyph in a pair and a standalone glyph. That's a big constraint on the contact statistics of the underlying language and the assignment of plaintext letters to elements of the cipher (which, by the way, would massively weaken the cipher), or else it requires a mechanism for keeping those repetitions from happening. Obviously other decompositions into standalone glyphs and glyph pairs are possible, but the problem is still there.