Page 166: There is a more complete technical description of the Enigma available on-line from Tony Sale's Codes and Ciphers in the Second World War. This includes the exact specification of the rotor wirings.
Page 185 and note 4.10: Turing was indeed 'the emissary' to the Polish cryptanalysts near Paris in January 1940. The main reason why I cast doubt on the statement in The Secret War was that Harry Hinsley, when I interviewed him, had thought it unlikely. In fact, the evidence was already available in 1979 but only in a Polish-language source. This book, Enigma by W. Kozaczuk, appeared in an English translation by C. Kasparek in 1984.It was a great pity that I missed this source, but I was able to include it in the 1992 Vintage edition. Kasparek drew on Rejewski's 1975 recollections and gave a very striking passage describing Turing's table talk.
"Why, that's a powerful poison!" said Turing in a raised voice.This passage is quoted on this Sources page, together with further comment on the codebreaking methods used at this period before Turing's first Bombe was ready.
Page 197: In 1983 I could only guess at the uses of Turing's weights of evidence and what the 'important method called Banburismus' was. My guesses were not bad: in fact Banburismus was a method of first (1) identifying where parts of different messages had probably been enciphered on the same rotor positions (2) using these 'depths' to deduce the identity of the outermost rotor and if possible the middle rotor too. The effect of this would be to reduce the Bombe search time by a considerable factor. This was revealed by Jack Good in his article in Codebreakers, F. H. Hinsley and Alan Stripp (eds.), Oxford University Press (1993), paperback edition with corrections (1994).
Page 224: With the addition of the fourth rotor, I said, the cryptanalytic problem had become 'only 26 times worse'. But this 26-fold expansion of the space of rotor start positions had the effect that there would be not be enough 'depths' ̵ parts of messages enciphered on the same rotor positions — thus making the Banburismus method impossible. So the situation was in fact much worse than that factor of 26 might indicate.
Page 230 The transmission error that gave away the Fish cipher system was in fact made on 30 August 1941. The error was not quite as simple as that of advancing the key by one place, and the work of teasing out the key was not trivial. It is attributed to J. H. Tiltman (see page 204).
There is further material in the Alan Turing Internet Scrapbook: Critical Cryptanalysis