The policy was to send the traffic to Freeborn in the hope of finding some fits which would prove something and possibly of cribbing them. Fits duly appeared and established amongst other things that the keys changed at mid-day and that there was only one key in the whole area. We noticed the curious phenomenon that most fits had letters 2 and 6 of their indicators the same and that in some 3 and 7 the same also, but still did not jump immediately to what had happened. It was actually a day or two later that the phenomenon was described to a member of Hut 6 who realized immediately what it was as he had been familiar with a similar system. From this point all was plain sailing; the rest of the theory was obvious and a first break was obtained shortly afterwards.
A start has to be made by listing the throw-ons for the left, middle, and right hand wheels, i.e., the information as to which letters in position 2 are followed by which letters in position 6. If A throws on to L and L to F and F to O etc. they are listed like this:
A L F O P A = (A L F O P)
As can be seen the 'box', as these sets of letters were called, has closed and it is a property of this type of indicator that each box implies a second box of identical size: we have here a five letter box, there is therefore another 5 letter box and, possibly, two 8 letter boxes to make up all 26 letters.
It is possible to break a key on throw-on information alone; the bombe is willing to tackle information in the form of discovering unknown letters which will encypher to known letters at fixed relative positions of the machine, but such "query menus" raise a variety of problems and are slow and expensive to run. Actually it is possible to proceed to find the Grundstellung alphabets from the throw-ons.
On our left hand wheel we have 2 five boxes:
( A L F O P ) (D Q K Z G)
This implies that the 10 letters of the alphabet involved may be paired together in 5 different ways: