Spaces

[Koen G]

Part of the difference may be what you're counting as a glyph. Is aiiin one glyph or five?

Ah yes, it's counted as five. So if tests against other texts continuously give a higher percentage of spaces than the VM, it may be as you say an indication that we're splitting certain glyphs into strokes.

[Koen G]

So I replaced "iiin" by "x", "iin" by "m" and "in" by "n". In other words, the frequent word endings we transcribe as strokes were turned into single glyphs. 
Then I divided the total number of words by the total number of characters, including spaces. The number of words normally equals the number of spaces including at the end of the line, only it's easier to count.

The result is 17,5 %. 
I also tested Orwell's 1984, but this time I normalized it by removing all punctuation. Here the number of words/spaces compared to the total number of signs is 18,7%. 



This means that if we transcribe "in"-clusters differently, Voynichese comes much closer to modern language, though it's not quite there yet.

Edit: hmm, for the original text of Beowulf without punctuation I only get 16,9%.

   

[ReneZ]

This statistic is of course depending directly on the average word length, which is known to vary for different languages, and also for different texts in the same language.

(If one does not add spaces at the end of each line, if also becomes dependent on the average number of words per line, which is primarily a layout issue, and hardly related with the language).