Cryptographic History of Work on the German Naval Enigma

19.  In view of all this Turing decided to go back and work on the 1938 keys; more was known about this early traffic and as there were only 6 stecker hand methods of attacking cribs were possible. This decision was confirmed by an interesting discovery about the Forty Weepy cribs. These had inexplicably begun to "crash" in 1939 (see Chap. I, Para. 38) and could not therefore any longer be right. Interrogation of a P.O.W., Funkmaat Meyer, disclosed that the German Navy had ceased to use the top row of keyboard for numerals and were spelling them out instead - so that Forty Weepy would now be sent as FORT ZWO DREI DREI NUL. When Fort's were written in this new form under suitable messages it was found that the number of crashes was only 30% of what would have been expected if the cribs had been wrong thus indicating that the Fort procedure was still continuing and that 70% of the placements were correct. These messages could unfortunately only be traced up to the beginning of the war as owing to the dropping of addressee callsigns in September 1939 it was not possible to spot which messages were Fort's after this time. It was also known that wheels 4 and 5 had been introduced in Dec. 1938 - and possibly others later - so that on every count it was right to tackle the earlier period.

20. Turing's break of the indicating system had occurred at the end of 1939 and in 1940, joined by Twinn and two girls; he started an attack on Nov. 28th, 1938, using the Forty Weepy crib. After a fortnight's work this day was broken and four others came out on the same W.O. There proved to be still only 6 stecker and furthermore a letter was never steckered for two days in succession. This was a most encouraging discovery since it meant that having broken one day 12 unsteckered letters for the next day were known so that methods applicable to the easier unsteckered machine could be used.

21. This break signalizes the first use of the EINS catalogue later to be a standard method of getting out messages on a known key but with unknown bigram tables - and hence unknown message setting. It


< previous

next >