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

Re: Voynich letters

Hi Brian,

First off, I'm not so sure about the key thing, too many repeated words
it seems, but maybe.  Perhaps the 'ain' words and their brothers and
sisters are outside the key.  Anyway, if you think it's a key, you may
not need to know what the key means to check your assumption.  What
about writing a program that separated every 'line', a line being
defined as a gallows character and everything after that up until the
next gallows character.  Then make files to analyze composed of only
lines that corresponded to a particular gallows.

I've already done this, and posted my results to the list six-odd weeks ago. To my eyes, these seem to indicate that each of the 8 or 9 gallows 'states' has its own frequency distribution: but these are difficult results to interpret definitively. :-|

I really ought to tidy up my JavaScript state analysis program and make it openly available. *sigh*

Then in a second program look for 'keys' to play with, abcb in one
gallows code compared to other gallows code words that show patterns
like szxz, gvhv... in Perl /^\w(\w)\w\1$/, also try abbc and any other
dual or triple letter combos where the positions of the reduplicated
values is consistent.  You could even allow for optional unencoded 'ain'
family endings:
/^\w(\w)\w\1[ain aiin]?$/.

Again, I've thought of this: but I'm not 100% convinced that <space> isn't actually an encoded character - perhaps indicating "repeat last character"... the lack of repeated characters in the VMS is a bit of a mystery otherwise... but then again, how many languages become unreadable if you discard every space and then discard every duplicated character in the resulting stream?

Note that <space> has a significantly different frequency distribution in the picnic table state(s) as compared to the other gallows states... I don't know what that indicates. :-/

FYI, my theory du jour looks like:-
(1) underlying language is mutated Greek, with 8am == "kai"
(2) o = omicron = the (male), y = eta = the (female), both elided into the following word
(3) it's a state-based language, flipping on gallows and picnic tables (in some way)
(4) <space> = new word OR "use same letter as last letter"? (Don't know)
(5) text is left-to-right, but labels can go in either direction (just to be difficult)
(6) the various picnic table variants are punctuation marks
(7) lines can end with nonsense characters (just to be difficult)
(8) some parts are just nonsensical (just to be difficult)

I'm still figuring out what to do next to move this along. Looking for astrology label cribs is one traditional way to bust VMS theories... perhaps I should stay true to tradition. :-)

Also, compare word frequencies and letter frequencies.  It would be
interesting if the characters inside the gallows showed up much more
often or not at all.  For instance, if the key meant, "Hey, I'm leaving
these characters out of the text, fill them in as needed", or "Hey,
these are the wild cards, make an educated guess what they are supposed
to represent."

Unfortunately, I've only ever seen the first part of the VMS at the British Library - the Beinecke Library haven't returned my emails about buying a copyflow set. And the EVA transcription discards this information. :-( Hence, I've only been able to find the long gallows in the first 60-odd folio pages (two sides per image), otherwise I'd have posted a list of them long ago :-(

Perhaps they are clues as to missing vowels in a vowel
less system, given as hints on lines where the author felt the code
might be too ambiguous, or maybe they are the word delimiters.

Some pages have several long gallows: others have none. Some groups of pages have none as well - this destabilises a lot of possible theories.

Best regards, .....Nick Pelling.....