The Bletchley Park Codebreakers Page 6

  Here, in a paper written in September 1949, Foss describes the early British work on the commercial ‘C Model’ Enigma, and the beginning of the contacts with the French and Poles that were to become so important to the subsequent British breaks. The mention of the QWERTZU, or diagonal, in this paper is a reference to the order of the wiring between the keyboard and the first set of contacts inside the machine. The British had been unable to work out the order in which the keys were connected to the ‘entry plate’ in the Wehrmacht’s plugboard machine. Given the enormous number of different permutations available to the Germans, Foss and Knox had not imagined that it might simply be in alphabetical order. This was probably the most important single piece of information that the Poles provided.

  MS

  My earliest recollections of the Enigma date back to 1926. We then knew of two models: large typing [B model] and small index [C model], I never saw the large machine and don’t know if it was ever widely used. It was the small index model that was later developed and used by German services and others.

  It will be best to dispose of the large Enigma in a few paragraphs to avoid confusion. Its UK patent specification was 231,502, application date 25 March 1925, convention date (Germany) 25 March 1924. A letter from the Aeronautical Committee of Guarantee, Berlin, to the Air Ministry, dated 19 June 1924, said: ‘The invention, in an incomplete state, was examined in about 1921 by Lieutenant Hume, Office of the Military Attaché, British Embassy, Berlin, and it is believed, an expert was sent out from the War Office.’

  [A letter from] Hume to [Edward] Travis, 29 July 1926, says that the company have informed him that they have sold out all the big machines and no more of this pattern will be manufactured. An improved model may be ready in 10 months’ time. Perhaps the Admiralty would meanwhile like to buy a model of the small machine. The improved model of the large machine was on view in 1928.

  There is a brochure in French ‘La Machine à Chiffrer, Enigma’ issued by Chiffriermaschinen Aktiengesellschaft, Berlin W, 35 (undated, but presumed about 1924). This deals with the machine from the user’s point of view and gives no details (apart from the usual astronomical number of key variations) of the ciphering. There is also a brochure in German.

  An undated report describes a demonstration of the large machine at the Foreign Office on 27 March (possibly 1926). It was a typewriting model (called the ‘Typing Machine’ as opposed to the small ‘Index’ machine) and worked from the main current (DC). It had previously been demonstrated in Stockholm. It was a one-way machine (i.e. it had no reflector wheel), with four drums which were moved by four ‘gap-tooth’ cog wheels with different numbers of teeth on each, some of the teeth being operative and others not.

  Sometime in 1927 or so Travis gave me a small machine to examine. I was not told where it came from, but presumed it had been bought as a sample. This was the Enigma referred to [in the GC&CS history] German Abwehr Cryptographic Systems and their Solution vol. 1 (The Unsteckered Enigma).

  A worker [Keith Batey] on the SD (Sicherheitsdienst) Enigma having recovered the upright, the next most natural step to take was to see whether any known machine had a wheel defined by this upright. This led to a most surprising discovery: that the wheel recovered was identical with wheel I of a certain commercial machine said to have been purchased by Mr [Dilly] Knox in Vienna in 1925. It had in fact been lying in a cupboard behind the person who made the original break.

  The only difference between the two machines was that the turnover notch had been transferred to the tyre [ring] [in the SD machine]. I don’t know when this model [the small (Index) machine] was first made. It had movable tyres but the turnover notch was on the wheel and not on the tyre. Incidentally, the Air Ministry used this model as an inspiration for Typex* which also had turnover notches on the wheel and not on the tyres.

  I wrote a paper entitled ‘The Reciprocal Enigma’ (the large Enigma was not reciprocal) in which I showed how, if the wiring was known, a crib of fifteen letters would give away the identity and setting of the right-hand wheel and how, if the wiring was unknown, a crib of 180 letters would give away the wiring of the right-hand and middle wheels. The methods I used were rather clumsy as they were geometrical rather than algebraical and, when Dilly Knox came to study the subject ten years later, he invented the ‘rods’ and the process known as ‘buttoning up’, which used the same properties as I had done, but did so in a more effective way.

  In January 1939, the French cryptanalysts showed Denniston, Knox and myself their methods, which were even clumsier than mine, and ended with a flourish and a dramatic ‘Voici la méthode Française’. They asked Knox if he had understood and he replied in a very bored way ‘’Pas du tout’, meaning (I think) ‘Pas du tout à fait’ [sic]. Denniston and I rushed in with conciliatory remarks. The French were, however, delighted with the rods when Knox explained them and by the next interview had made a set of ‘réglettes’ of their own.

  At these two interviews, the Poles were mainly silent but one of them gave a lengthy description in German of the recovery of throw-on indicators when the operators used pronounceable settings. During this exposition Knox kept muttering to Denniston, ‘But this is what Tiltman did’, while Denniston hushed him and told him to listen politely. Knox went and looked out of the window.

  Some time in 1938 or 1939, I can’t remember when – Josh Cooper places the time as the autumn of 1938 and that suits me – we were given by the Poles or French cribs of four long steckered Enigma messages and, I believe, the Stecker-pairings. I think that at the time we did know how the Stecker worked, but I can’t remember who told us. There is an undated translation of a secret German document published in 1930 which describes the method of plugging of the Stecker, but does not give the cryptographic effect. This may have been the document given us by the French in 1931. Knox, [Oliver] Strachey, R. R. Jackson and I all worked on it in an effort to reconstruct the wiring, including the basic ‘diagonal’, which on the steckered Enigma was ABCD … Z and not QWERTZU … I don’t know why the others failed, but the reason I failed was because I assumed the turnover notch was on the wheel and not on the tyre. I believe Knox and Strachey were allowing for the turnover notch to be on the tyre. Later on, at Warsaw, the Poles, who must have considered us all very stupid, gave us the complete answer.

  * The high-grade cipher machine used by the British armed forces during the Second World War

  4

  BREAKING AIR FORCE AND ARMY ENIGMA

  RALPH ERSKINE

  Introduction

  GC&CS’s wartime solutions of plugboard Enigma owe much to Polish cryptanalysts, who were the first to solve it - in 1932. As is now well known, the Poles gave GC&CS a clone of plugboard Enigma in August 1939, following a meeting in Warsaw at the end of July. But it was not until January 1940 that GC&CS was able to solve any Enigma using new rotors that had come into service in December 1938. Hut 6, as it became known, did so using Polish methods until May 1940, when the Heer (German Army) and Luftwaffe (German Air Force) changed the indicating system used with Enigma. But even before then, Dilly Knox had improved significantly on the Polish methods. Alan Turing had also invented the British ‘bombe’, which bore little resemblance to the Polish bomba and was the basis of virtually all GC&CS’s successes against Enigma during the war after 1940.

  Chapter 4 illustrates the fundamental importance of good intercept facilities, which were in short supply for much of the war. Indeed, Hut 6’s successes ultimately depended on the skills of the humble intercept operator and good, easily operated intercept sets. Although the contribution of the US Navy bombes to Hut 6’s work is now better known, it is not generally appreciated that the British intercept stations had enough high-quality sets only because of superior American production facilities and know-how. A British history of the important Army intercept station at Beaumanor concluded that the intercept war ‘was won, in a very large measure, as a result of the availability of American communication receivers’.

 
; Breaking Enigma is sometimes thought to have been a task carried out by a few cryptanalysts, working on their own in back rooms. In fact, most of the work was done by relatively unskilled staff, and the cryptanalysts were very much in a minority - and seldom eccentric. There was so much traffic that Hut 6 had to adopt production-line methods to deal with it. Running Hut 6 as a well-organized and flexible unit, while maintaining staff morale at a high level, required considerable leadership and management skills, especially since many staff were doing boring and monotonous work under difficult conditions. Fortunately, Gordon Welchman, the first head of Hut 6, and his successor, Stuart Milner-Barry, rose to the challenge well. Not every unit at Bletchley was so well managed.

  It is sometimes claimed that Hut 6 was completely on top of Enigma by the end of 1944, and that it was often quickly and easily broken then. Chapter 4 shows that nothing could be further from the truth. In November 1944, when Hut 6 was attacking about sixty-five Heer and Luftwaffe Enigma ciphers, Hut 6’s demands for bombe time became prodigious, and it required considerable help from US Navy bombes in Washington. The US Navy had generously agreed in November 1943 to handle Hut 6 ciphers on its bombes on exactly the same basis as Kriegsmarine (German Navy) ciphers, except for the main naval cipher. Shark: it kept to that decision for the rest of the war, despite being worried about impinging on the functions of the US Army codebreakers, which was always a highly sensitive issue.

  Chapter 4 describes the methods used by Hut 6 to penetrate Heer and Luftwaffe Enigma, and shows the many difficulties that they faced, especially with Heer ciphers. It was a close-run thing, especially in late 1944 and 1945, when Milner-Barry thought that Hut 6 was on the verge of losing its hold on Enigma.

  RE

  The Polish Cipher Bureau, Bureau Szyfrow (BS), had attacked German codes successfully in the 1920s until 1926, when the German Navy adopted two simple versions of Enigma. In 1928, the Bureau became blind against much of the German Army traffic, which had also started to use Enigma. The Cipher Bureau realized that a machine cipher required special talents to solve it, and in 1929 gave selected mathematics students a course in cryptology. The only three to complete the course, Jerzy Różycki, Henryk Zygalski and Marian Rejewski, were recruited by the Bureau, although initially they worked only on a part-time basis.

  The German Army added a plugboard to Enigma in 1930, but it was not until September 1932 that Rejewski, who was the star among the young cryptanalysts, was given the plugboard Enigma to attack on his own – the older cryptanalysts in the Bureau had been trying to solve it since its introduction, but had completely failed. By the end of that year, Rejewski had reconstructed the wiring of Enigma’s rotors mathematically, using permutation theory, in an outstanding feat of cryptanalysis. The fatal flaw in Enigma had been its indicating system, which used doubly enciphered message keys (see Appendix II). Rejewski described the system as ‘the third secret’ of military Enigma, although he also received invaluable help from Enigma key-lists which had been received from Gustave Bertrand, in French military intelligence: Bertrand had bought them from Hans-Thilo Schmidt, who was working in the German Defence Ministry’s Chiffrierstelle (Cipher Centre). The indicating system was also exploited by the Poles to solve Enigma traffic in 1938 and 1939 using electro-mechanical machinery known as bombas (bombes), and a system of perforated sheets invented by Zygalski. Rejewski also reconstructed the wiring of rotors IV and V when they were introduced in December 1938.

  Plugboard Enigma, which was the only type used by the Wehrmacht during the war, measured about 28×35×15 cm (11×14×6 in) and weighed about 12 kg (26 lb). Its main components were a keyboard, lampboard, rotors, reflector (Umkehrwalze) and plugboard. A battery provided the electric current for the internal circuitry (see Figure 4.1). The standard machine used three rotors, which during the war were chosen from a set of five for the army, or eight for the navy. Rotor turnover notches were in a different place on rotors I to V – a weakness that was later to be exploited by Huts 6 and 8. Double notches in the special naval rotors, VI to VII, were in the same positions in all three rotors, which made it impossible for Hut 8 to differentiate between them in a procedure known as Banburismus. Unlike the rotors, the reflector in three-rotor Wehrmacht Enigma could not move. A later version of naval Enigma, M4, used a rotor as part of a settable reflector, but it did not rotate during use.

  The plugboard contained twenty-six dual sockets, into which cables with jacks were inserted to connect pairs of letters – A to P, B to K, and so on. It was reciprocal: in the previous examples, P swapped for A, and K for B. Ten pairs of letters were generally connected during the war, although eleven pairs would have produced the maximum number of combinations. Without a plugboard, Enigma was relatively easy to solve, but the plugboard version was a very impressive machine.

  Dilly Knox, GC&CS’s chief cryptanalyst, and other members of GC&CS tried to reconstruct Wehrmacht Enigma’s wiring in the 1930s, but were thwarted by one factor - the wiring to Enigma’s entry disc or rotor (the Eintrittwalze). Knox called the wiring the ‘QWERTZU’, on account of its link to Enigma’s keyboard. It had also stalled Rejewski, although only for a relatively short time. In January 1939, three representatives from GC&CS, including Knox, met two senior members of the Polish Cipher Bureau and Bertrand in Paris, but the Poles were under orders to disclose nothing substantive about their successes. They did well: one of the GC&CS party wrote about the Poles: ‘Practical knowledge of [Wehrmacht] enigma nil.’ However, Knox and Alastair Denniston, the operational head of GC&CS, together with Bertrand and a French cryptanalyst, were invited by the Poles to Warsaw at the end of July. To their great surprise, a Polish clone of Enigma was revealed to them on 26 July at the Cipher Bureau’s centre at Pyry, outside Warsaw, and they were told that the wiring to the entry rotor was the identity permutation (A wired to A, B to B, and so on). Knox was far from pleased that the Poles had beaten him to it, maintaining ‘a stony silence’ during the conference. The Polish methods to solve Enigma signals were explained at some length; the visitors were shown the bombas during the second part of the conference. Fortunately, when they met the Poles the next day, ‘Knox was his own bright self & won the hearts & admiration of the young men with whom he was in touch.’

  Figure 4.1 Path of current through Enigma

  Surprisingly, one cryptanalyst at GC&CS, a ‘Mrs B. B.’ (it has not been possible to identify her) ‘had seriously contemplated’ that the wiring was indeed an identity. But she had not been given a crib (known plain-text) supplied to GC&CS by Bertrand, either because organization was not Knox’s forte, or because he thought that she would be wasting her time in following it up: she had therefore been unable to make any progress. Even more surprisingly, although Knox understood her hypothesis about its wiring, he had not pursued it himself, probably because he could not believe that the Germans had been so stupid. Knox wrote that ‘had she worked on the crib we should be reaching them [the Poles]’. However, in this he was somewhat over-optimistic, since it is most unlikely that GC&CS could have recovered the wiring of rotors IV and V. Rejewski had been able to do so only because the Sicherheitsdienst (SD) continued to use a pre-15 September 1938 indicating system when the new rotors were introduced. It is almost certain that the British were unable to intercept the SD traffic at that stage, since it seems to have been sent at low-power, and British intercept facilities were very thin on the ground.

  The Poles disclosed the design of the Zygalski sheets (called ‘Netz verfahren’ - net method) at the Warsaw meeting, and sent Enigma clones to London and Paris in August. GC&CS soon began to prepare two sets of sheets, based on Zygalski’s design, although the actual punching of the sheets could not start until mid-November, since a special machine to print the positions of certain repeated indicator letters, known as ‘females’, had first to be made. The first set of Zygalski sheets was completed around late December 1939, in one third of the time predicted by Knox. Making the sheets so quickly was a major achievement, since sixty sets (on
e for each rotor order), each with twenty-six sheets, had to be prepared, with many of the sheets containing about 1,000 holes cut precisely to give the relevant co-ordinates. Knox apparently had to resort to subterfuge and contravene Denniston’s orders for preparing them, in order for them to be ready so soon.

  The British cryptanalysts must have been dismayed when they started to use the sheets, since they could not solve any Enigma traffic with them. Unknown to GC&CS, crucial data received from the Poles about rotors IV and V had been incorrect. GC&CS’s failure with the sheets may have been the reason why Denniston asked Brigadier Stewart Menzies (‘C’, who was also the director of GC&CS) to see whether Rejewski and his colleagues could visit GC&CS to help with Enigma. Menzies duly wrote to Colonel Louis Rivet, the head of the Cinquiéme Bureau de l’État Major de l’Armée (Fifth Section of the Army General Staff – the Services de Renseignements et de Contre-Espionnage militaire), whose functions included radio intelligence. But Rivet was unwilling to send the Polish cryptanalysts, who were working with the French Army at the Château de Vignolles, in Gretz- Armainvillers, near Paris, after escaping from Poland. He felt that since his Service was paying the Poles, it was entitled to keep them.

  Part of a second set of Zygalski sheets was sent to the Poles in France on 28 December 1939, together with a set of Jeffreys sheets (which were a punched sheet catalogue of the effect of two rotors and the reflector). Alan Turing took the balance of the Zygalski sheets soon after they were completed around 7 January 1940. According to Knox’s accounts, Denniston did not want the balance to be sent, perhaps for reasons of security. Knox had to threaten to resign unless they were sent immediately. It was fortunate that he took such a strong line, since GC&CS would not otherwise have learned the correct data for rotors IV and V so quickly, which might well have had disastrous consequences for its attack on Enigma. The Poles used the sheets to make the first break into wartime Enigma on 17 January, when they solved Green (the cipher used by the Heer in Germany’s military districts, the Wehrkreis) for 28 October. At around this time, the GC&CS Enigma section moved into Hut 6, pursuant to a decision taken in early December 1939.