Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site randvax.UUCP
Path: utzoo!linus!decvax!ittvax!dcdwest!sdcsvax!sdcrdcf!randvax!jim
From: jim@randvax.UUCP
Newsgroups: net.crypt,net.misc
Subject: Re: The Beale Ciphers (long)
Message-ID: <1836@randvax.UUCP>
Date: Sun, 8-Jul-84 21:41:18 EDT
Article-I.D.: randvax.1836
Posted: Sun Jul  8 21:41:18 1984
Date-Received: Tue, 10-Jul-84 07:28:30 EDT
References: <148@oakhill.UUCP>
Organization: Rand Corp., Santa Monica
Lines: 76
Xref: 69 1151

--------
As the discoverer of the so-called "Gillogly strings" referred to in
the article to which this is a followup, I ought to enlarge on the
description of my discovery and give my opinion on the Beale Ciphers
in general.

My article was published in _Cryptologia_, vol 4 # 2 (Apr 1980),
titled "The Beale Cypher: A Dissenting Opinion."

To recap, the cipher algorithm of Paper #2 consists of numbering the
words in the Declaration of Independence (DOI) and replacing each plaintext
letter in the message with the number of a word in the DOI which begins
with that number.  I applied the DOI to Paper #1 and discovered garbage at
the beginning of the cipher, but significant patches of non-garbage a ways
into it.  The most striking sequence is "ABFDEFGHIIJKLMMNOHPP".  The
first F is encrypted as 195, and letter 194 of the DOI is a C.  The last
H is 301, and letter 302 of the DOI is an O.  Carl Hammer, a strong
defender of the legitimacy of the Beale Cyphers, did a thorough analysis
of Paper #2 and discovered 23 examples where the encryptor made errors
of this type, or about one every 33 letters.  There are several other
sequences that look non-random, such as AABBCCACDD.

The point of all this is that if one of the papers is decrypted with the
wrong document, the result should be a random assortment of initial letters
from the document being tried, and all sequences occurring in it must be
the result of random processes.  In the paper I computed the odds against
that sequence occurring randomly, but that was a bit silly since anyone
can see that it's totally out of the question.

So any explanation of the Beale as a legitimate cipher must explain how
that sequence could legitimately come about.  Hammer suggests that the
first paper could be encrypted as a "pun", where this sequence is there
intentionally, but if one used the right document a real message would
appear again.  He attempted to construct an example using two documents,
but could not achieve such a long sequence, nor use such small numbers
(one needs to go farther in the documents to find an initial letter that
fits both your intended message and the "pun").  I think this is
unlikely, if only because the writer was unlikely to have the kind of
time necessary to find two documents which would allow this extended
pun, nor any particular reason to do it.  (I discount Hammer's theory
that it was put there as a signal that the decryptor is on the right
track, because the original story assumed that the legitimate recipient
would have the keys and method for all three papers.)

My opinion is that a hoaxer made up the second message, then produced
two sets of garbage numbers.  While making up the first set of garbage
numbers, he picked numbers from the DOI that he had numbered in front
of him that ran in patterns (Hammer calls this my "doodle" hypothesis).

This opinion is supported by a number of internal and external clues.
One is alluded to by another writer in the followup:  why give a synopsis
of papers #1 and #3, since the expected legitimate recipient will know
exactly what is in them as soon as he deciphers them!  The numerous
errors made in the encryption testify that it was a painful process for
the writer, so why add extra verbiage to it?  Second, Paper #3 is supposed
to contain the names and addresses of all 30 members of the party (or
their heirs).  Paper #3 has only 600 or 700 numbers, so that each name
and address can have only about 20 characters.  A little sparse.

Greg Mellen discusses at some length the fact that the originals are
missing and that there are odd discrepancies in the different versions
of the ciphers that are still intact (BCA Newsletter, March 1980).
He also gives further examples that another net.crypt writer noticed about
Paper #2:
	The "vault" is "an excavation or vault".  It is not someplace,
	but has an "exact locality."  It is not "six feet down," but
	"six feet below the surface of the ground."  It is not "lined",
	but "roughly lined."  Things are not "packed", but "securely
	packed."  Vessels rest not on "stone," but on "solid stone".

A number of other points have been made by other authors, but this should
be plenty to convince those who haven't yet decided one way or the other...
(Was that a little snotty?  I apologize for any offense taken...)

	Jim Gillogly
	{vortex, decvax}!randvax!jim