the Hoofden weather and once again Dolphin was set on its feet. The reencoding was somewhat sporadic and uncertain but it kept Dolphin going for the last 10 days of the war.
The second and far more serious development was the introduction of the Grundstellung table on Plaice, one of the rare cypher developments which took place without any advance warning from German sources; the significance of the one message which might have given a hint of what was to happen was completely missed both by ourselves and by Naval Section.
Plaice broke normally at the beginning of March but we found we were quite unable to discover a Grundstellung which would decode the traffic. After a few hours work the information we had collected for discovering the Grund was so extensive and so contradictory that it became apparent that there had been some change in procedure. We guessed that some form of multiple Grund was in force and soon disproved the theory that it was dependent on the Time of Origin of the message. We decided then to attempt to discover the Ringstellung on the assumption that the indicating system was fundamentally unchanged i.e. that the bigram tables were still used in the same way and the message set-up obtained by transposing the tetragram at a Grundstellung. The method adopted was as follows: if a message has the tetragram N Q D G and we break it by means of a crib we can decode the message although we know neither the Grundstellung nor the Ringstellung; the position of the Ringstellung is comparatively unimportant as we know the correct position of the core of the wheel. Assume then that we put clips at ZZZZ and the message decodes at position J P D B; from this we can get some information about clips:
|Tetragram||N Q D G|
|Set-up||J P D B|
We know that the true set-up - the set-up with the correct clips - was obtained by transposing N Q D G and therefore it cannot crash with that tetragram; therefore the third letter of the set-up cannot be D and the clip cannot be Z although it can be any of the other 25 letters; similarly the left hand clip cannot be D as J + D = N, the second clip A or the 4th E. Thus every message broken knocked out one possible clip for each wheel (unless as often happened it had been knocked out already) and eventually only the correct clips should remain. This process was known as "boiling out" the clips and took a long time; in actual fact the right hand clip could be obtained by other methods and the one next to it had only 2 possibilities so that the total number of clip possibilities to consider was 26 x 26 x 2 = 1,352.
The first and worst part of the problem was to break sufficient individual messages. On the 4 wheel machine the EINS catalogue method was impossible and we now made use of the Fillibuster which Squadron Leader Jones had evolved when presented with the problem of how to break messages when the bigram tables next changed. Fillibuster was based on the assumption that the Germans would continue to consider that large numbers of their messages were urgent.