Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site alice.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!alice!reeds
From: reeds@alice.UUCP (Jim Reeds)
Newsgroups: net.crypt
Subject: Re: crypt and enigma
Message-ID: <3703@alice.UUCP>
Date: Wed, 8-May-85 10:42:00 EDT
Article-I.D.: alice.3703
Posted: Wed May  8 10:42:00 1985
Date-Received: Thu, 9-May-85 02:51:38 EDT
References: <4051@allegra.UUCP>
Organization: Bell Labs, Murray Hill
Lines: 19

>From allegra!don Wed Dec 31 19:00:00 1969
>
>Actually, the German "enigma" code was not broken in the complete sense
>in WWII.  The Germans put the key into the message in a stupid way.
>The method Reeds used to break crypt is a relatively new idea; probably
>discovered around 1960 by the NSA.  The government still used rotor
>machines up until then.
>
Flattering but false.  I used old ideas, well explained (for example) in 
H. F. Gaines's book.  Although the paper does not say it, the crypt(1)
command is really not an "enigma" cipher, but something simpler.  A simple
transformation of the cipher text turns the output of crypt(1) into  a
"sliderule" type cipher with (unknown) mixed ciphertext component and 
unmixed pt component with known key sequence.  Recovery of the mixed 
component is straightforward.  The alphabet size makes it harder, the self 
inverse nature of the CT component makes it easier.  These ideas occurred 
to several people at the same time.  I just rushed into C code first.

Jim Reeds.