The Beginning of the Operational Period During the Autumn of 1941, the outlook became steadily brighter. In August there were 6 bombes in action and we had the bigram tables without which Banburismus was impossible. The results speak for themselves. All August was broken except 14, 2425, all September, and all October except 34 and 1213. From October 14th, 1941, Dolphin was broken consistently until March 7th, 1945. During these months, methods of Banburismus and cribbing improved considerably. The situation gradually developed that stability which can only arise from a long period of regular breaking.

Early in October, on the 3rd, 4th, or 5th, we were mystiWed by the failure of the U-boat traYc to decode when Dolphin for the day had been broken. The situation was saved by a Werft crib, one of Werfts earliest triumphs, and it transpired that the key was the same as for the rest of the traYc except for the Grundstellung, which was entirely diVerent. This was a further development of the innovation of April when the U-boats started to use a Grundstellung which was the reverse of that used by surface craft. This new development made very little diVerence except that it delayed slightly the reading of the U-boat traYc. November 29th brought the Wrst real crisis in the form of a change of bigram tables and K book. 6 months earlier this would have beaten us, but we now knew enough about cribs to be able to break the traYc without Banburismus if suYcient bombes were available, and there were now 12 bombes in action. Rather curiously, we thought it worth while to indulge in the luxury of a F

The policy was, however, to some extent justiWed, as on November 30th we got a right dummy depth crib and Commander Travis, then the arbiter of bombe policy, decreed that we might have all 12 bombes to run it.

This crib and many others duly came out in the course of December, and the process of e i n s ing, twiddling, and bigram table building, described in an earlier [section], proceeded merrily enough and considerably more smoothly than before. By the end of December, the tables were near enough to completion to begin to think of restarting Banburismus and Turing was just starting to 312 | Patrick Mahon reconstruct the K bookalmost the last theoretical problem he tackled in the Section, although he remained with us for some time to come. On December 30th a pinch of keys, bigram tables, and K book made further work unnecessary and we were able to restart Banburismus at once. C H A P T E R 6 Bombe and Spider (1940) Alan Turing