Urgent messages began with one of several 'priority words' of which KRKR, BINE, MUKE, and WESP were the commonest and Fillibuster took some 80 messages at a time and tried them on all 4 cribs simultaneously. Lists of machine positions where the messages could be made to say one of these 4 things were sent back from the bombe hut and each one was tested by hand to see if the rest of the message decoded. Testing the stops, of which there were hundreds, was a great labour and results were less good than we had hoped because the Germans conceived of the intrinsically futile, but by a fluke successful, security device of burying these words in the middle of the messages though fortunately this did not become a general practice. There can be no doubt that Fillibuster saved the situation and when some 75 messages had been broken the clip possibilities had been reduced to 9 from 1352. Greatly to our relief the correct clips had not been 'boiled away' and fairly soon they were definitely established.
With the clips definitely established it became possible to discover possible Grunds and soon 2 messages produced the same Grund I E W E. Examination of these messages revealed a phenomenon of this sort if the underlined trigrams from the indicator groups after bigramming were looked up in the K book and the numbers belonging to them noted:
The encircled number giving the position of the trigram in the column of the reciprocal part of the K book was clearly the key to the system and it was soon found that other messages with coordinates 1 : 1 could be decoded with Grund I E W E. The assumption that there was a Grund table 26 x 26 = 676 proved incorrect; in fact it was 26 x 12 = 288, columns 1 and 2 being identical for no very apparent reason.
It was soon obvious that the Grund table was likely to last at least a month and we set to work to discover the rest of the 288 Grunds. This lengthy and laborious process was made more difficult by the fact that the traffic had not been prepared suitably for this unexpected phenomenon and a great deal of work had to be done in the form of looking up the Grund coordinates and sorting the traffic. In this we received very welcome assistance from Hut 6. By half way through the month about ¾ of the traffic could be decoded at once and by the end of the month most of the table had been built up.
There was some evidence to support the view that the Grund table would not change at the end of the month and would be the same on