[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: VMs: Further investigatio of folio f1r
> right you are. My only excuse is that while "letter" frequency of the VM
> followed Latin (but the text was still hidden even after the conversion) the
> transposition cipher was the only one which, additional used, would keep the
> frequency intact.
Have a look at my paper
Under "Verbose Ciphers", I discuss a verbose cipher I
devised called "Cat Latin". Cat Latin takes a Latin
plaintext and produces a ciphertext with about the same
single-letter frequencies as Latin but with digraph
frequencies that are very different. The first- and
second-order entropies of Cat Latin match those of
Voynichese closely! However, Cat Latin looks nothing
like Voynichese and has far longer word lengths.
> It is also true that transposition ciphers were not too popular in middle
> ages, when the art of substitution cipher was blossoming, they already
> developed the grill. I guess our problem is complicated by the fact we take our
> rules indiscriminately: Zipf's laws and second entropy rules are based on certain
> assumptions only - which, I have to admit, are not quite clear to me.
I think you will agree that the combinations EVA
<[i][i][i]n>, <[i][i][i]r>, <ch>, <sh>, <ee>, <qo>,
<dy>, <ol>, <or>, <os> and <al> are very common in the
VMs. If these are present in the source text, a
transposition cipher that *operates on the single
characters* would break them up. I find it hard to
imagine a transposition cipher that would *produce*
them in the ciphertext.
Of course, if these combinations are in the plaintxt
and the transposition cipher *rearranges them as
units*, it would be conceivable.
Finally, transposition ciphers typically do not leave
word divisions intact.
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying: