Re: VMs: Three frequency tables

["Jeff" <jeff@xxxxxxxxxxxxxxxxxxxxxx>]

Rene Zandbergen" r_zandbergen@xxxxxxxxx wrote


>
> --- Jeff <jeff@xxxxxxxxxxxxxxxxxxxxxx> wrote:
>
> > The polymorphic cipher will not give a letter for
> > letter substitution. Like
> > a polyalphabetic. However single substitution is not
> > used. An example.
> >
> > Take a line from the VMS.
> >
> >
> shoy.ckhey.kodaiin.cphy.cphodaiils.cthey.she.oldain.d
>
>  (description deleted)
>
> > Again all comments welcome.
>
> Two comments:
>
> 1) Thanks, this is a very clear description!
>
> 2) It describes the decryption process. I am pretty
> sure, however that the opposite process (encryption)
> cannot exist. Bear with me:
> The scribe would always be in the
> following situation: a new plaintext character
> must be encoded, and there are already two code
> characters on the papper (Vellum) to make a new
> triplet. That means that for any given (code) pair
> on the vellum there should be many different
> triplets in his tables in order to accommodate the
> many different plaintext characers to be encoded.
> However, in Voynichese the possible set of triplets
> is very restricted.
>
> I know that this is a qualitative analysis, but
> before pursuing this path you should be concerned
> that the encryption matching your decryption
> actually exists.
>
> A suggestion: it might be worth trying the opposite:
> the VMs was encrypted using a method that you are
> using for decryption...
>
> All the best,
>            Rene
>

I know what you are saying and yes it could get very complicated for a
scribe after a while.

I have tried the encryption approach already and the table develops as you
go. Some method of easing the encipherment would be needed. This is where a
feature of the VMS comes in. I noticed a while ago that the tests I was
running would only make sense if the words were written back to front. So
that the key letter was the last in a line or paragraph. I assume paragraph
for obvious reasons.

The repetative word endings seem to be the key. These I believe were
determined first. On an individual page this would be fairly
straightforward. The reason this seems apparent is that mid word positions
tend to have triplets with lower occurance counts than those at end of word.
I have not as yet run any analysis on word starts for average occurance.
This will be one of my next tasks.

Lets take an example. Take the phrase "I am going out shopping and I will be
leaving at noon."

This text has three word endings that are the same. All ending 'ing.' If we
assume word endings are the key then the pattern for these could be put in
place first and the links added to fit. This would aid with the encipherment
of later pages. It could even have been done on well constructed test text
before the actually VMS encryption.

With the phrase "I am going out shopping and I will be leaving at noon" all
ing combinations would be replaced by iin or any other repeating pattern.
Also other dissimilar endings could have the same pattern to totally foul
the statistics. Also some occurances of ing could be made to look different
depending upon the text that leads up to the ing pattern. As you can see
from this description it would be an enormous task to plot the plaintext
letters back into position.

Jaqcues Guy now seems to have a new idea which might be interesting to
follow and it would be simpler to impliment than my idea.

Jeff


______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list