Isn't that the same as Bazeries' Cylinder Cipher, where you have a bunch of disks with mixed alphabets around the edges, you put them on a rod in some order, set the plaintext along a row, then read the ciphertext off any other row? If so, a short paper of mine was just printed in "The Cryptogram" (JF95 issue) about cryptanalyzing it. A US military field cipher was based on the Jefferson version; I handled a copy (with metal disks) in the National Cryptologic Museum just outside Ft. Meade last summer. There may even be an Aegean Park Press publication on it. The servicemen hated it -- fiddling with all the disks in the heat of battle made it cumbersome for tactical use. If it's as I describe, I'd call it a multiple-key polyalphabetic: keyed alphabets, keyed order, and (potentially ambiguous) offset key for deciding how far around the cylinder to go. The period is how many disks you have, since they stay in the same order (but with different relative offset) for the whole message. It's not really an autokey under the Meaning of the Act, since there's no feed-forward from one letter to the next. I'm sure Kahn's "The Codebreakers" will have more info. Jim Gillogly Highday, 18 Solmath S.R. 1995, 21:59