Voynich Reconsidered

[kckluge]

I never quite sure precisely what people mean when they use the term "substitution cipher". Do they just mean a cipher when each letter is substituted with a letter or symbol? Or do they include a more complex cipher with homophones, nulls, substring substitutions, word substitutions etc.

Mark,

I *think* "substitution cipher" is a class that subsumes most (if not all) of those things. When the mapping is 1-1 (no nulls or homophones), it's a "simple substitution cipher", with an addition axis of whether the mapping is fixed (a monoalphabetic substitution cipher) or varies according to some key sequence and alphabet table (a polyalphabetic substitution cipher). Unless I'm mistaken, variable-length substitutions for letters (straddling checkboards, for instance) fall under this umbrella. If by "substring substitutions" you mean something like replacing digrams with other digrams, I don't believe those sorts of schemes are considered "substitution ciphers". But I'm saying that off the cuff without double checking standard nomenclature against something like the Riverbank manuals or Lanaki's "Classical Cryptography Course" (You are not allowed to view links. Register or Login to view. relevant document I came across recently that may help answer questions about what buckets different cipher schemes fall into is the following (1918) taxonomy by William Friedman: You are not allowed to view links. Register or Login to view.