key (1) Cribbing (2) Banburismus. Cribbing - given some knowledge of the traffic - can be done on 3 wheel or 4 wheel keys and provided there are some routine messages does not depend on volume of traffic. Banburismus does not require knowledge of the traffic but needs a fairly large volume of traffic (300 messages or more) and can only be done on a 3 wheel key. It has the advantage of indicating which wheels are in the machine and is thus economical in bombe time. It is however a large scale job needing very much larger staff than cribbing and also slower since work on a day cannot start until all the traffic has been received whereas a crib can be used as soon as it comes in. When a throw-on system is used the indicators alone, on quite small traffic volumes, provide material for a break.
48. All these methods have to be rounded off by use of the bombes. The most economical procedure in bombe time, where possible, is to find "the alphabets" either by Banburismus or (where that system is used) from the weaknesses of the throw-on system; the next most economical method is to run cribs and the most extravagant method is to run (on throw-on systems) menus derived directly from the indicators without finding the alphabets. We shall see that those considerations - in particular the balancing of the economy of bombe time obtained from Banburismus against the quicker break made by cribs - dictated to a great degree the policy and staffing of the section.
49. "Paired days" give no trouble to break since cribs have only to be run on one wheel order; they are always broken on a crib, usually very quickly.
E. Recovery of Bigram Tables.
50. The methods of breaking daily keys when the bigram tables were known has been outlined above. When the bigram tables were changed our procedure was as follows
(1) A day would be broken on a crib (this was unaffected by absence of the bigram tables).
(2) As many messages as possible would be got out by one of the various methods known for dealing with messages when stecker and W.O. have been found.
(3) It was then possible to find the Grund and decoding the