[Cryptography] GHCQ Penetration of Belgacom

Henry Baker hbaker1 at pipeline.com
Wed Dec 24 15:18:16 EST 2014

At 07:37 AM 12/24/2014, Bill Frantz wrote:
>On 12/23/14 at 4:12 PM, leichter at lrw.com (Jerry Leichter) wrote:
>>Ah, yes, the IBM 1620 - sometimes affectionately known as CADET, for Can't Add Doesn't Even Try.
>And the IBM 1401 did right to left decimal add/subtract as well.  At least it didn't use table look up for the results like the 1620, saving a number of memory cycles.

The 1620 model II got rid of the addition table, and exchanged the older model IBM electric typewriter for the newer IBM Selectric (on both machines the console typewriter was the most unreliable part of the whole machine).



I think that there were "stripped down" versions of the 1401 that couldn't multiply/divide.  The first 1401 I worked on had only 4,000 characters of memory; newer models went up to 16,000 characters using a bizarre 3-character addressing scheme.

The 1401 was often used as a "front end" for bigger scientific machines such as the 7040 or 7090.  I was tired of waiting for all the card punching for the "intermediate" file of the 1401 assembler, so I used the 7040's main core memory as /tmp for the 1401.  This also saved a lot of trees in the form of discarded cards.


The 1401 printer (called the 1410, I recall) used a spinning chain with 5(?) sets of characters in a particular sequence; these characters were imprinted using electromechanical hammers fired at the precise time that the chain character appeared in the position that that character was supposed to appear on the paper.

The sequence of characters on the chain was chosen to minimize the number of hammers that were likely to fire at exactly the same time for words in the English language.  Needless to say, a visual examination of the chain sequence would tell you how to "print" a line that would fire all 132 of the hammers simultaneously, usually breaking the chain in a spectacular manner.

More information about the cryptography mailing list