Voynich Reconsidered

[MarcoP]

I guess that, when considering anagrams, the idea is that each Voynich word-token corresponds to a plain-text word token? Since word lengths are more or less comparable, this scenario does not seem to allow for nulls. In addition to character frequencies, one could check character co-occurrence in the same word. E.g. if o corresponds to ‘a’, q must be a character that basically can only occur in words that include ‘a’. This can be extended to qok and similar checks can be made for the characters decoding edy, aiin etc.