(31-08-2026, 09:03 PM)Mauro wrote: You are not allowed to view links. Register or Login to view.Some time ago I did something similar: I calculated the geometric distances of Voynich glyphs according to the distribution of their previous and following characters (the 'routes' in your diagram) and I made some observations you might find interesting. [...]
Thank you —
what you wrote matches very closely with what I now measured. Your distance_following(k,t) = 0.12 comes out almost identical for me (0.122), even though I use a different transcription.
In the letter-by-letter count, the immediate predecessor of k in “okaiin” and “qokaiin” is the same—an o. The q does not come immediately before the k and is invisible in this count. So “...ok...” and “...qok...” end up in the same category: “Predecessor = o.” Am I seeing this correctly?
Counted this way, an “o” precedes “k” about 61% of the time and precedes “t” about 65% of the time—almost the same, which explains your small distance of 0.12.
But if you separate the two categories, it looks like this:
only o before it qo before it
k 30% 31%
t 46% 19%
k is neutral toward both channels, whereas t favors the pure “o” and avoids “qo.”
In absolute numbers: ok:ot ≈ 3032:2798 (almost 1:1), but qok:qot ≈ 3121:1121 (almost 3:1).
That’s why I think the totals (61% and 65%) are similar more by chance. The apparent similarity between k and t seems to be an effect of the merging.
But of course, we don’t know exactly what “units” are. However, since q is bound to o 97.7% of the time, I treat the two glyphs in this combination—from q’s perspective—as a single unit—whether that’s correct, who knows
(P.S. Not only in Latin but also in German, “qu” is bound nearly 100 percent of the time—just as a side note)
![[+]](https://www.voynich.ninja/images/collapse_collapsed.png)
