Cryptographic History of Work on the German Naval Enigma


between messages was made and our old figures revised, and sets of charts were prepared to make possible much quicker and more accurate assessment of the mathematical value of the various data used in Banburismus. The cribbing possibilities on the various groups of traffic were also carefully studied preparatory to re-starting current work in August.

9. An even greater advantage of the pinch was that it tided us over a change in bigram tables on June 15th. At this time there were still only five bombes which had to suffice for the Army and Air keys as well as for the Naval; to have had to run a series of all wheel order cribs while the new tables were being built up would have caused a serious hold-up and would probably have meant two months' delay before we began to break currently.

10. August saw us all fully primed for an attack on the new keys. The crib room was now for the first time an independent entity and with the Banburists armed with fresh charts, scoring tables and statistics all felt fully confident of breaking anything. The first immediate result was somewhat deflating as no success was obtained on any of the first six days. August 7th was less intractable however and it, together with the rest of August, yielded to treatment satisfactorily. Though we continued for some time to refine our methods there were no more major troubles with Banburismus - September 18th/19th was the only other pair of days we failed to get out on DOLPHIN for the rest of the war. This is not to say, I must hasten to add, that we were successful on all days - owing to the "paired day" phenomenon we only had to break every other day on Banburismus - so if the first day failed we could try the second. Also we sometimes had a partial success, some but not all of the letters of the alphabet, indicating that a certain wheel was in the right hand position in the machine and we could then run a crib on this. Finally - a disgrace we rarely suffered - an all wheel order crib could be run if Banburizing was completely unsuccessful. This latter process was not quite so expensive in bombe time as it sounds, owing to the assistance

-31-


< previous

next >