[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Re: VMs: Further investigatio of folio f1r
From: "Jan" <hurychj@xxxxxxxxx>
Sent: 06 April 2004 13:56
> Hello Dennis,
> ======= At 2004-04-05, 22:33:00 you wrote: =======
> > Don't forget, EVA was never intended to represent one grapheme with one
> transcribed character, and it doesn't. EVA /iin/ is a case in point.
> Of course, but would another system show graph so similar to single
> curve of many languages (not only with Latin) ? Maybe, I do not know.
Yes I saw this too. You need to run a lot of different tests though to see
the structure fully. It took me months to put together the necessary tools.
> > It's interesting that Latin comes close but not >English and the others.
> Well, it is not perfect closeness, but the differences may be easily
> to the "crudeness" of the method, which letter frequency certainly is.
> what counts most is how the Latin curve follows the VM curve - the English
> instance crosses the line several times, having apparently altogether
I found Italian a very good match but never tried properly against Latin.
English, French and German failed.
> > As D'Imperio noted, others saw long ago that the single-character
> distributions of Voynichese are close to those of many European languages
> (though your study is the most complete one).
> Yes, this "hyperbolic characte" of the curve is obvious for all natural
> and they even do not differ too much - percentwise - but what differs
> is the order of letters (for different languages). Of course I also made
> corresponding table and subsequent "conversion" of the VM into "Latin"
> Needless to say, I did not get too far :-).
You are currently looking at it in the wrong way. Try group analysis!
> > Those early researchers quickly found that they could not solve
> >monoalphabetic substitution.
> Correction: "as single monoalphabetic substitution" - the "shortness"
> "words" shows there must be at least another encoding present. All that of
> course only if we take the "sign-for-letter" conversion, which as you
> said may not be the case at all.
> >The digraph distributions >are the ones that are really unusual, and
> >reflected in the second-order entropy.
> But bi-letter substitution cipher (say Vigenere table, where we replace
> letter by its two coordinates) may have something similar and
> cipher has letters so mixed up that original "digraphs" simply dissappear
. . .
Nope I tried exactly that and it don't fly. Feel free to try yourself I may
have missed something but the structure of word endings in the VMS makes
this difficult in the extreme.
> Best regards.
> http://www.angelfire.com/electronic2/ohlas/VM/ Voynich Manuscript
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying: