begin and end-letter analysis

[Torsten]

We know that 'y'-ending takes about 40% in the text of the VMs.

There is a systematic difference between the two Currier 'languages'.
In Currier A 3664 out of 11415 tokens (32%) end with EVA-[-y].
In Currier B 9845 out of 21582 tokens (46%) end with EVA-[-y].
The value for the whole text is 40.6 %.

Other typical ending glyphs are [n], [l], and [r].
6064 out of 37919 tokens (16.0 %) end with [-n]. (Currier A = 15.7 % | Currier B  = 16.9 %)
5909 out of 37919 tokens (15.6 %) end with [-l]. (Currier A = 18.4 % | Currier B = 14.4 %)
5689 out of 37919 tokens (15.0 %) end with [-r]. (Currier A = 16.8 % | Currier B = 13.2 %)

Rarely used ending glyphs are [s] and [o].
1345 out of 37919 tokens (3.6 %) end with [-s]. (Currier A = 4.6 % | Currier B  = 2.3 %)
1173 out of 37919 tokens (3.1 %) end with [-o]. (Currier A = 5.0 % | Currier B = 2.1 %)

The dominance of certain glyphs in fixed locations is in my eyes an interesting observation for the Voynich text (see Timm & Schinner 2020, p. 10).