orders of priority in general terms but the detailed decisions depended on technical considerations such as how the crib would run on the bombes, what was likely to happen to things already running and so on and so forth. A body of five "bombe controllers", Milner-Barry, Manisty and Munro from Hut 6 and Wylie and myself from Hut 8, was therefore formed and a rota arranged so that one of us was always on duty and acting as "bombe controller". He was responsible for allocations of bombes between the huts basing his policy on the priorities allotted by the intelligence sections. This procedure worked extremely well and was one of the many things that showed the overwhelming advantage of an inter-Service organization; no arrangement whereby each Service "owned" or even had prior claims to a certain fixed number of machines could have compared with it in efficiency.
25. We now come to an episode in the history of the section over which even the least sensitive of us would gladly draw a veil of considerable opacity. Besides the Home Waters and U-boat traffic which was teleprinted to us currently there was some Mediterranean traffic sent down by bag and held by Naval Section. This was known not to be on the same bigram tables as Home Waters and it was thought probable by Naval Section that it was on three different keys, Mediterranean, Aegean and Adriatic. The volume of traffic was round about 100 messages a day during the latter part of 1941 and early 1942 so that even had the bigram tables been known Banburismus was impossible. For these reasons we had not examined the traffic at all.
26. During 1942 the traffic began to increase until days of 200 - 300 become quite common; also the messages were very long and it seemed therefore that if (1) all three areas had a common key (2) the bigram tables being used were old ones or current ones used in some different order, then we had a chance of success. Encouraged by this, and perhaps even more by being rather underemployed, we decided to examine a few days for fits (i.e. long textual repeats which would connect messages with each other) and see if we could get anywhere.