Martin Gardner: Mathematical Games : A Cipher I broke that Defeated Poe

Peter Bakke : I solved this cipher in 1987 while working on my master’s degree in systems science at SUNY Binghamton (University of Binghamton, now).

“Ge Jeasgdxv,
Zij gl mw, laam. xzy zmlwhfzek
ejlvdxw kwke tx lbr atgh lbmx aanu
bai Vsmukkss pwn vlwk agh gnumk
wdlnzweg jnbxw oaeg enwb zwmgy
mo mlw wnbx mw al pnfdcfpkh wzkex
hssf xkiyahul. Mk num yexdm wbxy
sbc hv wyx Phwkgnamcuk?”

In 1839, in a regular column Edgar Allan Poe contributed to a Philadelphia periodical, Alexander’s Weekly Messenger, Poe challenged readers to send him {cryptograms (monoalphabetic substitution ciphers), asserting that he would solve them all “forthwith.” One G. W. Kulp submitted a ciphertext in longhand. It was printed as shown above in the issue of February 26, 1840. Poe “proved ” in a subsequent column that the cipher as a hoax—”a jargon of random characters having no meaning whatsoever.” In 1975 Brian J.Winkel, a mathematician at Albion College, and Mark Lyster, a chemistry major in Winkel’s cryptology class, cracked Kulp’s cipher. It is not a simple substitution — Poe was right — but neither is it nonsense. Poe can hardly be blamed for his opinion. In addition to a major error by Kulp there are 15 minor errors, probably printer’s mistakes in reading the longhand. Winkel is an editor of a new quarterly, Cryptologia, available from Albion College, Albion, Mich. 49224, at $16 per year. The magazine stresses the mathematical and computational aspects of cryptology. The first issue (January, 1977) tells the story of Kulp’s cipher and gives it as a challenge to readers. So far only three readers have broken it.