The History of Hut Eight

The next interesting development was the interrogation of Funkmaat Meyer who revealed valuable information about Short Signals and also the fact that the German Navy now spelt out numerals in full instead of using the top row of the keyboard. This encouraged Turing to look again at the FORTY WEEPY cribs which in 1937 had begun unexplicably to crash and he came to the conclusion that the cribs remained fundamentally correct provided that the numerals were spelt out.

In early 1940, now joined by Twinn and 2 girls, he started an attack on November 1938 by the FORTY WEEPY method and the new style crib. The reasons for choosing a period so long ago were various but were primarily based on the knowledge that modern keys were more complicated and would require more work. Two new wheels (4 and 5) had been introduced in December 1938 and from the beginning of the war they were unable to trace the FORTY WEEPY messages owing to call-signs being no longer used.

After about a fortnight's work they broke November 28th and 4 further days were broken on the same wheel order. Only the Spanish Waters came out, the rest of the traffic was on a different key. There were still only 6 Stecker and there was a powerful and extremely helpful rule by which a letter was never steckered 2 days running: if continuity was preserved, 12 self-stecker were known in advance. No Grundstellungs and no bigrams were broken, messages being broken individually or on the EINS catalogue which was invented at this time and was to play an important part in the exploitation of Enigma.

EINS was the commonest tetragram in German Naval traffic, something in the region of 90% of the genuine messages contained at least one EINS. An EINS catalogue consisted of the results of encyphering EINS at all the 17,000 positions of the machine on the keys of the day in question. These 17,000 tetragrams were then compared with the messages of the day for repeats. When a repeat was found, it meant that a certain position of the machine the messages could be made to say EINS and further letters were then decoded to see if the answer was a genuine one: if it was, the starting position of the message was known and it could be decoded. In fact about one answer in 4 was right so that messages were broken fairly rapidly. In later days the whole process of preparation and comparison was done rapidly and efficiently by Hollerith machinery, but at first slow and laborious hand methods were used.

The plan was to read as many messages as possible, to gain some knowledge of cribs, and then to make rapid progress with the help of the Stecker rule. "There seemed", says Turing in


< previous

next >