Portal de Periódicos da CAPES

The Poles Reveal their Secrets: Alastair Denniston's Account of the July 1939 Meeting at Pyry

2006; Taylor & Francis; Volume: 30; Issue: 4 Linguagem: Inglês

10.1080/01611190600920944

1558-1586

Ralph Erskine,

Historical Geopolitical and Social Dynamics

Abstract Sets out, with comments, an account by A. G. Denniston (the operational head of the British Government Code and Cypher School in 1939) of the meeting near Pyry, Poland, in July 1939, when Polish cryptanalysts revealed how they broke Enigma, and a letter by Dillwyn Knox on the meeting's outcome. Keywords: Alastair DennistonCaptain Gustave BertrandColonel Gwido LangerCommander Humphrey SandwithDillwyn KnoxEnigmaGCCSHenryk Zygalski Jerzy RóżyckiMarian RejewskiPyry Notes 1Other first-hand accounts state that the meetings were on 25 and 26 July, e.g., Bertrand [8, p. 59]; Rejewski [40, p. 269]. However, both Bertrand and Rejewski were writing over 30 years later, and although Denniston did not write these recollections until May 1948, he derived the dates in his account from his pocket diary. In addition, a report written by Rejewski in 1940 [38 p. 31] gives the date of the conference as 26 July 1939, although it mentions only that date on its own, and not 25–26 or 26–27 July. While that is far from conclusive, it would be slightly odd to date a conference by reference to its second day, and not its first. 2On Denniston, see Hinsley [Citation22]; Denniston [Citation16]. 3Knox had been with GCCS since its formation in 1919. See Batey [Citation5]; Fitzgerald [Citation19]. 4Beesly [Citation7, p. 16]. 5See further [Citation2]. 6Stengers [Citation47, p. 133]. 7Fitzgerald [Citation19, p. 234]. 8Bertrand [Citation8, p. 18]; Paillole [Citation37, p. 27]. 9For details of the documents, including keylists, see Bloch [Citation10, Annexe 2-B, C]. 10Kahn [Citation26, p. 76]. 11For a biography of , see Kapera [Citation29]. 12The message setting, or message key (Spruchschlüssel), gave the starting position of Enigma's rotors for a specific message. 13Captain (later Brigadier) John Tiltman was the head of GCCS's military section. His principal task in the late 1930s was attacking Japanese military systems, but he also did some work on Wehrmacht Enigma: Erskine and Freeman [Citation17, p. 289]. 14Foss [Citation20]. 15Knox [Citation32]. 16The perforated sheets were a method for solving an Enigma key's wheel-order and Ringstellung from the coincidence of holes in superimposed sheets: Rejewski [Citation39, p. 267]. On the Polish bombe, see Rejewski [Citation41, p. 289]. 17On doubly enciphered message settings and 1 May 1940, see Bloch and Erskine [Citation9, p. 134]. For the effect on the perforated sheets when those settings were dropped then, see Hinsley [Citation23, p. 953]. 18Rejewski [40, p. 242]; Rejewski [42, p. 237]. The Polish bombes depended substantially on the plugboard not changing too many letters. When 20 out of 26 letters were altered after 1 January 1939, the Polish bombes would often have been defeated, even if 60 had been operational instead of the six actually in service. 19Denniston [Citation14]. 20Captain Gustave Bertrand: Kahn [Citation26, p. 77]. 21Vice-Admiral Sir Hugh Sinclair, the head of the British SIS from 1923 until his death in 1939. He was also the titular head of GCCS. See further Andrew [Citation3]. 22Wilfred ("Biffy") Dunderdale, the SIS representative in Paris. Dunderdale had been posted to Paris in 1926. He was a good linguist who had "an element of the pirate at times:" Lockhart [Citation34]. He was also a shrewd operator, as Jones [Citation24, p. 216] shows in a story about his use of a Ju 88 technical manual. 23Colonel Louis Rivet, the head of the French Army's 2ième Bureau (the Services de Renseignements et de ContreEspionnage militaire). 24The French Foreign Office. 25Foreign Office, or Foreign Ministry. 26Russian diplomatic messages used one-time pads (OTPs); German diplomatic messages employed OTPs and a long additive system codenamed Floradora by GCCS: Kahn [Citation25, p. 44]; Alvarez [Citation1, p. 88]. 27The third and fourth floors of Broadway Buildings (Denniston [Citation15]), 54 Broadway Road, London, opposite St. James's Park underground station. GCCS shared this building with the SIS before the war. 28The Spanish Civil War. 29British intercept facilities were thin on the ground in 1939. In addition, German army and air force traffic was mostly on medium frequencies, which were difficult to intercept in Britain. Naval traffic more often used high frequencies, which were easier to intercept there. 30Hugh Foss had joined GCCS in 1924. He wrote a paper in 1927 showing how the C version of commercial Enigma could be solved, and was heavily involved in GCCS's attempts to crack the Wehrmacht's machine: Foss [Citation20]; see also Smith [Citation44]. 31Knox had solved a rewired version of commercial Enigma (the K Model) in 1937, while it was being used by the Italian Navy during the Spanish Civil War: Batey [Citation4, p. 94]. 32Wilfred Bodsworth, then a member of GCCS's Naval Section, who had joined GCCS in 1927 [Citation46]. He solved a rewired version of commercial Enigma used by the Spanish naval attachés in Berlin and Rome during World War II: [Citation45]. In July 1944 he was appointed a liaison officer in Washington with Op-20-G, and wrote numerous reports on Op-20-G's work: see e.g., [Citation11]. 33Tiltman, Knox and Foss did in fact attend the January meeting in Paris: Foss [Citation20]. 34Major Maksymilian . 35This is very unlikely for German, since he had been brought up in the part of Poland that was occupied by Germany until 1918, attended German-speaking schools and been forced to serve in the German army in 1917 and 1918. 36This was not quite correct for January 1939. Between 15 September 1938 and 1 May 1940, after the time and number of letters in the signal (which appeared in clear, e.g., 1755 145), the Grundstellung was given once in clear, followed by the five letters of the Kenngruppe (to indicate the specific cipher involved), then the first five letters of the doubly enciphered message setting, followed by the sixth letter, and the first four letters of the enciphered text: [Citation43], as amended by [Citation12]; Bloch and Erskine [ 9, p. 134]; Erskine and Smith [18, Appendix II]. 37This refers to the complex Kenngruppenbuch indicating system employed by the principal naval Enigma cipher, Heimisch. On this system, see Kahn [27, p. 285]; Erskine and Smith [18, Appendix III]. However, the Kenngruppenbuch system was introduced on 1 May 1937—not in May 1939, as suggested by Denniston: Turing [48, p. 135]. It is unlikely that Knox "diagnosed" the system in any meaningful sense: all the evidence indicates that Turing solved it, in late 1939: Mahon [35, p. 16]. For an explanation of how Turing did so, see Turing [48, p. 136]. 38Cf. the text to note 13. 39Knox had believed at the time that the Poles' knowledge of Wehrmacht Enigma was "nil:" Knox [Citation30]. 40After the war, Rejewski stated that the Poles "read military messages [in 1939] when the original drums [rotors] happened to be … in the machine, as occurred on the average of one time in ten:" Rejewski [39, p. 269]. However, he did not explain how that was achieved when the usefulness of the bombes had been greatly decreased following the increase to ten Stecker—see note 18 and the text to it. Moreover, on 15 December 1938, the Germans had introduced rotors IV and V, increasing the number of rotor orders from 6 to 60. The Poles, therefore, needed 60 sets of perforated sheets and 60 bombes, but had only two sets of sheets, and six bombes. His figure of one time in ten therefore appears to be much too high. 41Denniston was correct in this surmise. The Poles had been instructed not to disclose the Polish achievements against Enigma unless the British and French could show that they had made progress in solving Enigma and were prepared to share their knowledge: Mayer [Citation36]. 42For the dates of some of these visits, see Bloch [9, Annexe 2-B, C]. 43The Director of Naval Intelligence, Rear Admiral John Godfrey. See further, Beesly [Citation6]. 44Poznán. 45This was the Polish codebreaking centre (codenamed Wicher (Gale) by the Poles) near Pyry, outside Warsaw. 46It is not clear which officer chaired the meeting. Since Denniston remembered Colonel Langer, the head of the Cipher Bureau, it may have been Colonel Mayer. 47Colonel Gwido Langer (codenamed Luc). For a biography, see Kapera [Citation28]. 48The wheel order was the order in which the rotors were placed in the machine (e.g., III, I, V), but Denniston is probably referring to the wiring of Enigma's rotors (wheels, in GCCS's terminology). 49A pinch really referred to capture or theft, which was unlikely, since it would have been noticed. Knox was really thinking of espionage or treason (Verrat). 50See note 48. 51Denniston—or —was much too optimistic in this assessment (see note 18). 52The Poles had not broken any naval traffic since 8 May 1937, shortly after the Kenngruppenbuch system was introduced, since they could not solve that indicator system: Mahon [35, p. 18]. 53Bertrand gave the machine to Menzies at Victoria Station, London, on 16 August 1939: Bertrand [8, p. 61]; Kahn [27, p. 81]. 54Some naval decrypts did reach GCCS from the Poles, who had found the keys for 2 to 5 and 8 May 1937, and read about 100 messages by breaking them individually. Indicators and other data for 5 May 1937 received from the Poles helped Turing to solve the Kenngruppenbuch indicator system: Turing [48, p. 136]. 55GCCS may have wanted to concentrate "on current air/army traffic" from August 1939 onwards, but it was unable to break any new Enigma traffic until around 25 January 1940, after a full set of the perforated sheets became available. During 1939, its naval section could therefore only make a detailed study of the 1937 naval back traffic held by it: [Citation21]. 56Rejewski, Różycki, and Zygalski. 57Knox [Citation31]. 58Alastair Denniston. 59After the military ciphers introduced a new indicating system on 15 September 1938 (see note 36), card catalogues and a grille method formerly used to solve Enigma keys became useless against those ciphers, but the Poles quickly developed the bombes to attack them. 60This refers to the wiring to the entry disc (see above). 61Rotors IV and V. Rejewski solved them by means of the Sicherheitsdienst (SD—the intelligence service of the Nazi party) traffic, which continued to use the old indicating system, with its common daily Grundstellung, until 1 July 1939: Rejewski [39, p. 268]. 62"machine" probably refers to the Polish bombe, and "SSD" to the Sicherheitsdienst. "O.S." and "n.s." refer to the "old" and "new" indicating systems in force before and after 15 September 1938 (see note 36). 63It is not known why Knox referred to 1 April, but see the text to note 51, referring to "up to the previous May." However, no change to Enigma or its indicating system is known to have occurred then. His reference to this date may have been due to a misunderstanding at the Pyry meeting. 64The Poles failed mainly because they did not have the manpower to produce 60 sets of perforated sheets (1560 sheets in all): Rejewski [39, p. 268]. However, the sheets would have been very slow and difficult to use without some method, such as the Cillies (see Erskine and Smith [18, Appendix IV]—note that Gordon Welchman got them wrong in [49, p. 99]), for reducing the number of wheel-orders to be tested considerably, but Cillies had not been discovered in 1939. Making the sheets was a very laborious task for the Poles, since they had to use razor blades to cut about 1,000 precisely located holes in each sheet (for an example of a sheet, see Rejewski [41, p. 288]). GCCS employed machinery for the purpose but even so, preparing and making two copies of the sheets took it over three months, until 7 January 1940. The sheets were known to GCCS as "Netz," from Netzverfahren—net method). Welchman [49, p. 90] greatly confused matters—and upset some Poles — by describing them as "Jeffreys sheets," which were completely different. Turing [48, p. 95], makes it clear that Jeffreys sheets were a catalogue of the effect of any two Enigma rotors and the reflector (Umkehrwalze). The confusion probably arose because John Jeffreys was in charge of the small section making both the Jeffreys sheets and the perforated sheets. 65It is very difficult to know what Knox meant here. 66The meaning of "statistics" is unknown. It may have referred to data (perhaps coordinates) for the perforated sheets. Denniston [Citation13] distinguished between "Netz," "statistiques," and "filets Jeffreys" (Jeffreys sheets). Sheets for rotor V were sent to the French and Poles on 28 December 1939: Denniston [Citation13]. The sheets for rotors I to IV followed, at some date between 7 and 17 January 1940. Once the Poles and GCCS had full sets of sheets, breaking Enigma was resumed, with the Poles making the first break of a wartime key, Green, the German army's home administration cipher, on 17 January. 67Major . Knox's misspelling arose because he transcribed the name phonetically.