[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Concensus transcription?
> This is probably from Stolfi: http://www.dcc.unicamp.br/~stolfi/voynich/
Thanks, he also sent a pointer to this.
> > BTW, the initial results are interesting...in particular, I don't think
> > it can be what we've been calling a "verbose cipher" (single plaintext
> > letter maps to Voynich letter combinations of different lengths) unless
> > Currier C, S, and Z are nulls (at least in some/many cases). When I hit
> > a sufficiently frustrating wall, I'll write up what I've been doing and
> > post it so folks can see what the evidence and reasoning is.
> This sounds very interesting, pretty close to what I'm plannig. I'll be
> looking forward to what you get
With respect to the issue of nulls, the reasoning is somewhat as follows:
If you have an induced DFSA derived from some word list (in this case,
words which occur 3+ times in Biological B), then if you enumerate possible
substrings that can occur between final states in the DFSA that should give
you character combos and fragments of character combos from the verbose cipher
(verified using what I called "Tiltman-encoded Genesis"). What you in fact
get with the Biol B vocab list is a list of substrings longer than the input
word list (!), and it is clear that Currier C, S, and Z are to blame.
Unfortunately, if this is a verbose cipher then it looks as if the construction
of the letter combinations was not terribly orthogonal (for want of a better
word). My current best guess involves [OA][EMRNJ] (i.e., OE, AE, OM, AM, etc.)
+ A[TD6] + A, B, E, F, M, O, R, U, 2, 4, 8 and 9. (The assumes V = B and P = F
as per D'Imperio's speculation based on the one repeated marginal sequence and
that Q, W, X, and Y map to P, B, F, and V when the surrounding S is removed.)
Unfortunately, F, 2, and 9 are too common with this mapping, and the entropy is
still "too low."