How We Solved the $100,000 Decipher Puzzle
			 (16 Hours Too Late)

			  Robert W. Baldwin*
			Tandem Computers, Inc.
			 19333 Valco Parkway
		     Cupertino, California 95014
				   
			   Alan T. Sherman*
		     Computer Science Department
	       University of Maryland Baltimore County
		      Baltimore, Maryland 21228

			    July 25, 1989
		     (revised December 15, 1989)

[*This work was done while the authors were graduate students in the
Department of Electrical Engineering and Computer Science at the
Massachusetts Institute of Technology.]

			       ABSTRACT

On Saturday March 30, 1985, we solved a cryptography puzzle that had
remained unsolved for over two years.  Unfortunately for us, we missed
by one day the deadline for sharing what had become a $117,000 prize.
This paper describes the puzzle and how we solved it.

The $100,000 Decipher Puzzle consists of a simple two-sided jigsaw
puzzle that contains a sequence of 376 codenumbers, ranging from 1 to
1252, which were encrypted by a multiple-substitution cipher similar
to the Beale cipher.  Clues state that the key was derived from some
keytext in the public domain.  We solved the puzzle by building a
Zetalisp program that tested candidate keytexts by trying a variety of
ways to extract candidate keys from each keytext and by checking the
resulting candidate plaintexts for English.  Our program exploited a
novel ``windowing'' technique that detected when part of a candidate
key was correct and an effective test for English based on digraph
frequencies.

KEYWORDS.  Beale cipher, contests, cryptanalysis, cryptography,
cryptology, homophonic cipher, multiple-substitution cipher, $100,000
Decipher Puzzle, puzzles, Warren Holland.