There were other less important rules. 1 or 8 on the left were unlikely on the first of the month when no restriction from the preceding day was available and Op-20-G also used a system of preferences on the middle wheel. There is no doubt that here also they had a true theory but we only made very restricted use of it. Their system of making up wheel orders lists was very complicated and we should have had difficulty in finding the labour, or preparing anything similar.
In March 1944 wheel orders behaved badly and caused many delays in breaking; for a time we thought that the preference rule for left hand wheels had broken down completely but the end of the month was more normal and the rule of threes still held. There were, however, signs of the beginning of a new era. Plaice in April used wheels 1 to 5 on the left 3 times each, wheels 6, 7, and 8 not at all; with one exception the same rule can be seen in Dolphin keys for June which were captured near the end of April. These keys were even more insecure than their predecessors as no wheel is used twice until each of the 5 has been used once and none 3 times until each has been used twice.
The lesson is clear for all key-makers to read. The only secure method of choosing wheel orders is to take them one by one out of a hat, throwing them back into the hat as soon as they have been written down. Human inability to make a random selection has frequently been proved so that the human element should be reduced to a minimum while any system is likely eventually to be discovered and to defeat its own ends.
Before digressing on the subject of wheel orders we had dealt with the developments of April 1944 - Bonito and Trumpeter. May 1st passed without increasing our commitments and throughout May we continued to break keys without much difficulty. The Second Front was now very much in the air and there was considerable nervousness about possible cryptographic developments. Apart from the probable change of bigram tables the most popular theory was that a special key would be used in the area of the Allied landings or alternatively that the whole Channel and Biscay area of Dolphin would split off. Much work was done with a view to making the crib situation as secure as possible if something of this sort should happen; in fact the Second Front was to bring no immediate cryptographic development.
On June 1st Porpoise and Grampus ceased to throw-on. There was great confusion in the Porpoise world when the instructions were issued to this effect at the end of May; this was, however, only due to the bad phrasing of the instructions and it was soon made clear that the tetragram was to be encyphered only once, the last 2 letters of the second indicator group being dummy. This did not make very much difference to Porpoise except that cribs had now to be run as normal 4 wheel jobs and that the reserve method of breaking by means of alphabets was no longer