In March also shift work was started and Hut 8 was manned 24 hours a day for the rest of the war. In April teleprinting of traffic from Scarborough was begun and the Registration Room was started, all traffic being registered currently. Banburismus was started on some March days and there was a rather depressing period of inexplicable failure before the first break and then the system began to get under way: sooner or later a large part of April and May were broken. There can be no doubt that at this stage the battle was won and the problem was simply one of perfecting methods, of gaining experience, and of obtaining and above all of training staff. These last stages were made much simpler by the pinch of June and July keys.
These last two pinches were a great stroke of good fortune for the bigram tables changed on June 15th and had once again to be reconstructed. The methods of reconstruction were, however, by now efficient and between June 15th and the end of July this task was easily accomplished. Had we not had the keys, all days subsequent to June would have had to be broken on all wheel order cribs and the messages subsequently broken by EINSing: with the lack of bombes and comparatively crude knowledge of cribs which existed at that stage this would have been a slow process and the beginning of what I have called the operational period of Hut 8 would have been delayed by perhaps 2 or 3 months.
Throughout these early chapters I have avoided as far as possible all mention of bombes. Bombes are a complicated subject and their workings are to a large extent incomprehensible to the layman but without them Hut 8 and Hut 6 could not have existed and it is essential to attempt to describe briefly the part they played.
The bombe was so called because of the ticking noise it made, supposedly similar to that made by an infernal machine regulated by a clock. From one side a bombe appears to consist of 9 rows of revolving drums, from the other of coils of coloured wire, reminiscent of a Fairisle sweater.
Put briefly the function of a bombe was to take one wheel order at a time and discover which of the 17,000 possible positions of the machine combined with which of the half billion possible Stecker combinations would satisfy the conditions of the problem presented to it. This problem was