have to do is examine these 336 x 26^{3} or 336 x 26^{4} positions [i.e. about 6,000,000 in one case and 150,000,000 in the other], find the right one and having found it work out the stecker.

37. This implies the possession of two things (1) a test to distinguish between right and wrong positions (2) a very rapid means of applying this test.

__Cribs.__

38. There are two kinds of test which we are able to produce under suitable circumstances (1) a CRIB i.e. knowledge of the exact decode of all or part of a message. We might be able to say, for instance, that a certain message - because of its length, time of origin, call sign etc. - must begin VORHERSAGEBEREICH SIEBEN (weather forecast from Area seven). To insist that a stretch of letters in the message text should decode in exactly that way is clearly an extremely severe test (there is less than one chance in 100 million million million million million of it happening fortuitously!) and it supplies adequate information, given a quick way of testing. Cribs are of two main kinds (a) routine messages which are sent out day after day at about the same time from the same place about the same length and starting off in exactly the same way. (b) RE-ENCODEMENTS, which are retransmissions in exactly (or almost exactly) the same form of messages that have already been sent out in another key; for instance, Northern U-boats were on a different key from other U-boats, so whenever a message was sent out to all U-boats one obtained a re-encodement from one key into the other. [See Chap. ].

__Banburismus.__

39. The second test is one obtained by discovering, through a mathematical technique known as "BANBURISMUS" the effect of encoding the indicator at the Grundstellung. Look back to the example in Pars. 27 - 29. Since we are assuming the bigram tables to be known we can work out that "the trigram" (i.e. the Verfahren kenngruppe) chosen was ARQ. "Banburismus" enables us to say that the actual position at which the message started was ?VN (we cannot determine the first letter). This tells us that at consecutive machine positions R encodes as V and Q as N; similar information about all other message settings gives us the encodes of all the other letters of the alphabets at these consecutive machine positions. This also is a sufficiently severe test to determine the correct machine set-up. The two alphabets are known as the "middle" and "right wheel" alphabets.

-10-

Back to Cryptographic History of Work on the German Naval Enigma contents