VMs: fake letters

[Knox Mix <knoxmix@xxxxxxxxxxxxx>]

On the topic "Lookalike encipherment"
Nick Pelling wrote:
> . . . the issue of whether, 
> if EVA <o> is a fake letter solely used to construct verbose digraphs 
> (like qo, or, ol, etc), two adjacent fake <o>s would elide into a single 
> one - ie, should <qo.ol> contract to <qol>? Fortunately, <qol> only 
> seems to occur in pages where free-standing and word-initial <l>s occur, 
> which suggests that whatever <qo> and <ol> represent, <qo> cannot be 
> followed by <ol>.

Hello Nick,

From what followed, I think you are referring to the issue I brought 
up. If so I did not say <o> is necessarily fake. I considered the 
possibility that it is occasionally inferential whether fake elsewhere 
or not. Not just <o>. Any letter that would be one of a double if 
written. Writing it or not would not change the meaning. <qol> for a 
demo was not the best choice but to race a slow horse we do not know a 
letter (<l> for example) is free-standing in all positions or in any 
position adjacent another letter or across a space from another letter.
> Simialrly, there are related issues with <cho> and <chol> in Voynichese 
> (particularly in Currier A, IIRC): I strongly suspect that <ch> and 
> <cho> are tokens that code for quite separate things, and that <chol> 
> can only decompose into <ch.ol>, never <cho[o]l>: 

I think so, too. At the time I thought of inferred letters. I did not 
consider <cho> or <qok> or even <cth>. It was after a first look at a 
transcript and seeing (part of) what everyone sees. I, then attempted to 
sort out the bigrams. Isolating those I thought to be most obvious left 
quite a few others in between (the most prominent of which I took to be 
"real" bigrams), many singletons and some longer strings. If I remember 
right this left (sorry about that) only a few trigrams, four or five 
4-grams and maybe one longer string -- and most of the singletons -- and 
some dubious digrams. Well, they all were dubious but these more so. 
Anyway, I had about the right number of different elements to fill the 
used portion of a grid with an alphabet running across the top and at 
the side. I also made a frequency table for sequences. I knew some of 
these could be equivalent to, say <er> and <re> in English and that any 
18th Century cryptanalyst would have resolved the method if it were 
used. I knew I would not recognize a non-English script if it jumped 
out. No, not even a second very simple cipher of English behind the 
outer cipher. I knew I had gone in without sufficient familiarity with 
the script (<qok>, etc.). I never had any illusions about "solving" the 
script. I only wanted to get a better look at it. The repetition of 
n-grams and the lack of repetition of words and phrases, is a feature 
that interests me now.

The circumstantial evidence for implied/inferred letters is somewhat 
less than for the existence of a crenel or a knothole which we can 
determine by surrounding structure. The possibility that there is one or 
two or more does not have to lead anywhere; it only needs sit in the 
wings in the event it might serve a purpose in one scheme or another. If 
you have proven to your satisfaction they do not exist (tried to think 
of a substitution for "exist") I doubt there will be any loss.


>ie, that <cho> can
> never be followed by <ol> in the underlying language.

We have not read all the possible literature but it would seem so if 
there is an underlying language.

 Knox

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