[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

VMs: Re: VMS numbering systems hypotheses...

Hi all,

On Saturday 08 Jun 2002 10:26 pm, Nick Pelling wrote:
> (1) I don't believe that every instance of a gallows character indicates a
> number

I agree.

> (2) The first few characters on each page (& under some long gallows) still
> look anomalous

I agree here too, and to me the character <o> has a special meaning, in the 
labels, which is not numerical.

Some time ago I had a brief look at the sequences of characters that 
could behave as Roman numerals *as they appear* (i.e. in eva, separated by 
spaces) based on the word patterns, but I could not find any  obvious 
candidates (it is in the mail archives,  if anybody is interested).
Doing it with the word patterns allows to search the possible 
character=numeral  candidates from the patterns that already exist and then 
search for which characters in those patterns, rather than guessing which 
they may be and later finding out that they are not there.

So I am still a bit cautions about recovering a number system from the vms 
transcriptions, because I have the feeling that if there are "words" and 
"numbers", the words has to be much more common than the numbers (unless the 
document is coded like the Beale cipher) . 

You may ask why? Because it would be quite unusual for a distribution of 
numbers to follow Zipf's law, unless the vms uses a numerical system to index 
a dictionary (and so each number is just a substitution for a particular 

If we are sure that the all vms words are numbers, shouldn't we forget about 
cracking it, until the code book is found? :-/

I guess that if there is no "number flag" to indicate what is a number and 
what is a word, then there is no way of testing whether the system is working 
or not, and whether "numbers" are so or not.

Despite this, Nick's number encoding, if correct, could explain why the 
character repetition (and low entropy) of the vms text.
But at the same time I  was surprised that about half the numbers (17 out of 
31) decoded in the decoded document finish in "1" . 
(So again, my question: Is this anomaly because the system is incorrect or 
because most of the words are *not* numbers? And if so, which are the numbers 
correctly identified?)

You mention that <e> is the only character that appears in sets of 3 (and also 
4), but this combination does not appear on its own.  In <f39r.7> there is a 
<eee,s> though, so let's give it a chance.  Even if <eee> is 3 and <e> is 1 
(it appears 5 times), <ee> (2) and. <eo> (4) never appear on their own.

This is no reason to dismiss the idea, but it is a bit curious that there are 
no 2, 3 and 4s.  Of course if the numbers have a "flag" indicating a numeral, 
then this comment is irrelevant.

All this is worth to further investigation!
It would be ideal if the vms had some tables... :-(