I had been stuck with my Mark I version of the stroke theory for quite some time and thought of it as a nice, but wrong idea.
Enter Robert Firth:
But then I came across Robert Firth’s Voynich notes, especially his #24. (Which was, BTW, the last one he posted… apparently, he had come too close to the truth and The Powers That Be(TM) silenced him.)
In the noteworthy note #24 Robert had attempted to break up the Voynich words into grammatical constituents. But as opposed to most others, he didn’t employ a “prefix — stem — suffix” system, but only two classes of letter groups which he termed “odd” and “even” and which supposedly alternated. (Ie, the odd groups made up word starts, while the even groups made up the word endings.)
By using roughly 20 odd and even groups each, Robert was able to reproduce the majority of Currier A words in the VM. (He didn’t achieve a complete match; nobody does. But he apparently could reproduce virtually all of the more frequent words.) Some of the ciphertext letters were unique to odd or even groups, while some occured in both. The groups varied in length between one and five EVA letters.
In EVA, the groups looked like this: