VMs: Reducing the VMS to a stream of grouped glyphs...?

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi everyone,

Here's one approach we might try out to help us move towards a deeper 
understanding of the properties of the VMS' text.

The distributional statistics (entropy etc) for the various EVA versions of 
the text point to EVA's level of representation as being located (as per 
its original intention) below the level of the actual text - ie, if there 
is an underlying text, it is probably formed of groups of EVA letters/strokes.

The obvious examples of this are EVA <ch>, <sh>, <qo> and <ee>, and GC 
points to others like <ckh> (etc) and <ccc>: but (to my eyes) the actual 
level of grouping might well be higher still. For example, <dy> occurs 
extremely frequently, as do the members of the <dain> family etc.

My suggestion is this: that we try to suggest groupings for EVA letters 
that have the effect of remapping the VMS' statistics closer to that of 
natural languages, and to compare those statistics with several European 
languages' statistics (from circa 1500).

As a general rule, I think that a good set of groupings should contain 
somewhere between 20 and 60 elements - more than that would be fairly 
impractical? (Your comments, please!)

Note that there may well be instances of nulls, misspellings, coding 
errors, and copying errors in the actual text, so we do have some margin 
for additional interpretation.

To start, here is my initial suggested set of 51 groupings, based on 
examining <f78r> (which is where Strong also started from, because of its 
clarity), and guessing at an underlying structure.

I've put them in sequential order of replacement, so that (for example) 
<qo> comes before <q>. I'd also suggest removing spaces (and EVA null 
characters like <!>) before you start, as spaces can be inserted mid-pair 
to misdirect the cryptologer:-

	dy
	ol	al	or	ar	am
	ee	cc
	ain	aiin	aiiin	air	aiir	aiiir
	oin	oiin	oiiin	oir	oiir	oiiir
	qo	q
	ofe	of	fe	f
	oke	ok	ke	k
	ope	op	pe	p
	ote	ot	te	t
	ocfhe	ocfh	cfhe	cfh
	ockhe	ockh	ckhe	ckh
	ocphe	ocph	cphe	cph
	octhe	octh	cthe	cth
	she	sh
	che	ch
	s	d	y	r

FWIW, this came about from trying to explain how the behaviour of the "o" 
and "e" arose, while looking at the various cipherbets from 1440-1460 which 
contain "4o". Here, I'm suggesting that both "o" and "e" are used in glyph 
combinations (both before and after key trigger letters): which would be 
the logical extension of the ideas in those ciphers.

If (as I believe) the <ain> family is steganographically concealing Roman 
numerals of some sort, that would have the effect of inflating the 
effective grouping count towards the top end of the practical range I 
suggested.

I haven't tried this out yet (except by hand), but would be very interested 
to hear your comments (especially on the resulting statistics!) and your 
suggestions both for improvement and for other groupings to try out. :-)

Cheers, .....Nick Pelling.....

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