17. On November 29th the bigram tables and K book changed. This would have been very serious a few months earlier, but now the number of bombes had increased to 15 and consequently we were able to take the change more or less in our stride. All wheel order cribs were successfully run and the tables built up by EINSING: towards the end of September we were ready to re-start Banburismus and Turing was beginning to tackle the problem of building up the new K book - a difficult and laborious process - when a pinch was made of both K book and bigram tables.
18. Two more cryptographers, Perkins and Golombek, joined in December. Looking back on this, it was a mistake to engage any more senior people at this time. We overestimated the amount of work that could be profitably put into a day's Banburismus; at this time we rather thought given sufficient effort the alphabets could ultimately be got on any day and also that the extra skilled staff would speed up work. Neither of these views was right - some days were impossible on Banburismus and the result of having three people instead of two working on a day was more that one worked slightly less hard than anything else. We were always, rightly I think, of the opinion that it was better to be a little understaffed rather than overstaffed and this was an unsuccessful departure from this principle.
19. On December 28th there was another step in the break away of Shark from Dolphin, the introduction of a Stichwort (see Chap. I, Para. 19). This was decidedly more of a nuisance than the Grundstellung changes and it took 48 hours to recover the Stichwort; fortunately the same Stichwort stayed in force until Shark broke away and became an independent key, otherwise we should probably have had serious delays in Shark throughout January.