Delta Cipher

[Fontanellean]

Delta ciphers haven't been excluded or investigated AFAIK. How would you investigate something so elusive?

I did some statistics on oshfdk's cipher (without special rule to ignore text until the first p/f) on paragraphs of the VMS to see how many differences exist within the cipher's specification (max. 1 char added, max. 1 char removed): they could store between 5000 and 15000 characters of plaintext depending on how lines are reparsed (or not).

Everything Voynich is elusive. I suppose you'd do it the natural way: Create sequences of word differences according to various methods and then examine their statistical properties.