The Bletchley Park Codebreakers Read online

  The Mask operation led by Tiltman continued until the middle of 1937. Its significance is confirmed not just by Curry but by Josh Cooper who remarked that while the work carried out by Kenworthy’s police unit was of no particular interest to the police it was ‘of great importance to the future GCHQ’.

  Exploitation of Soviet armed forces traffic during the interwar period appears to have been less systematic and therefore less successful than that of the diplomatic and Comintern networks. It is not clear how much military material was deciphered in India. But both Simla and Sarafand had limited success with Russian military ciphers. There was also some early work on the ciphers of the KGB. But during the 1920s and early 1930s, the main focus inside GC&CS with regard to Soviet armed forces was on the Russian Navy.

  William ‘Nobby’ Clarke, a former member of Room 40, was one of a number of GC&CS members unhappy at the way in which the codebreakers and the intercept operators were increasingly being asked to work on diplomatic material at the expense of service traffic. He managed to persuade the Admiralty that there should be a naval section within GC&CS and then made the rounds of Royal Navy establishments and ships, coaxing a number of officers into agreeing to intercept Russian, French and Japanese wireless communications in their spare time while at sea. The Royal Navy intercept station at Flowerdown also began taking Russian naval material and the Army station in Sarafand covered the Black Sea Fleet. But the use of one-time pads and a lack of depth ensured that few messages were deciphered.

  Clarke’s report on Naval Section work for 1927 admitted that there had been very little naval traffic intercepted, all of it between shore-based establishments. There had, however, been some success in solving a super-enciphered system in which encoded messages were reciphered before transmission, a practice designed to make them more difficult for an eavesdropper to read. The messages were first encoded using a codebook, which provided groups of randomly selected figures for common words or phrases. This produced a series of groups of figures, normally uniformly four-figure or five-figure groups. The operator then took a stream of predetermined but randomly selected figures, placed them underneath his encoded message, and added the two together figure by figure, using non-carrying arithmetic, to produce the reciphered message.

  In an attempt to find out whether it was worthwhile continuing to work on the Russian Navy ciphers, Clarke decided to see the problems faced by his volunteer intercept operators for himself: ‘In the hope of clearing up the problem of Russian naval ciphers, I joined HMS Curacao for the Baltic cruise of the 2nd Light Cruiser Squadron … Unfortunately the amount of intercepts available does not permit of much work being undertaken on this kind of traffic nor is it considered that it would be profitable.’ The situation improved slightly in 1929. Although no messages had been broken, enough intercepts were coming in to allow some limited traffic analysis.

  During late 1929 and 1930, the Naval Section launched a concerted effort to try to solve the Russian naval codes and ciphers. Josh Cooper was sent to Sarafand to carry out a fifteen-month investigation of Black Sea Fleet communications while Lieutenant-Commander G. A. ‘Titts’ Titterton, a former member of the Naval Section, returned from a Russian interpreter’s course to work on Russian Navy material in London.

  Cooper’s report shows that there were a number of different codes and ciphers in use, ranging from relatively simple systems to high-grade ciphers. But while the lower grade systems proved vulnerable he was not able to break any of the high-grade systems and Titterton seems to have had even less success. He left GC&CS briefly in December 1932 and, as one senior codebreaker noted, was replaced by another officer whose ‘skirmish with Russian was also short and unsuccessful’. The lack of success and the need to divert resources to Italian systems as a result of the 1935 Abyssinia Crisis eventually led to Russian Navy coverage being dropped.

  But as the attack on Russian Navy systems waned, the Army Section of GC&CS, set up in 1930 under John Tiltman, was beginning to get to grips with an upsurge of Russian Army traffic. This had become available in 1933 as a result of an arrangement with Estonia’s codebreakers, who offered all the material they were intercepting in return for radio equipment. With Tiltman busy, first on the Comintern traffic and later on super-enciphered Japanese army systems, P. K. Fetterlein, the brother of Ernst, was recruited to work on the Estonian material. The creation in 1936 of an Air Section of GC&CS, led by Cooper, gave fresh impetus to this work. Cooper’s experience of Russian material and Tiltman’s preoccupation with Japanese systems appear to have been the main reasons for the Air Section being asked to deal with the Estonian material, since little if any of it was from Russian Air Force units. ‘The Russian traffic was a mess,’ Cooper said. ‘What we received was a mixed bag from a wide range of stations (some of them in the Leningrad Military District, just over the border, some from at least as far away as the Ukraine), with little or no continuity. Some of the material was in very low-grade systems, which had usually been broken by the Estonians before we got it; the content was of no face value. There was also a variety of higher-grade systems but never enough of any one line to make a crypt attack possible. Controlled interception of selected lines of traffic with good T/A backup might at this time have produced very interesting results, but we could not control the Estonians.’

  By 1938, with war against Germany beginning to look inevitable, the main priority had become the breaking of the German Enigma machine cipher, although Russia still remained high on the GC&CS list of target countries. Lieutenant-Commander Titterton had now returned to Broadway and the Home Fleet had resumed interception of Russian Navy traffic. As late as mid-1938, two new junior assistants were recruited solely for their knowledge of Russian, one of them being Alexis Vlasto, whose A Linguistic History of Russia to the End of the Eighteenth Century remains a standard textbook. Russia also remained a high priority in India, where the codebreakers had broken the Soviet secret service super-enciphered code.

  The responsibility for the Russian material was subsequently handed back to the military section, together with the services of P. K. Fetterlein. The 1939 co-operation with the Polish and French codebreakers, that was to be of immense importance in the breaking of the German Enigma machine cipher, had the added bonus of providing a batch of material from Lithuanian and Latvian codebreakers, Cooper recalled. ‘The raw material was similar in quality to the Estonian and work on this untidy mass of miscellaneous Soviet services was co-ordinated in the Military Section with Vlasto as Air Section’s contribution. There was only one consignment of Lithuanian and Latvian. Estonian went on, I think, for a while with deliveries by diplomatic bag.’

  By now, the codebreakers had been moved out of London to the MI6 ‘War Station’ at Bletchley Park, designated Station X. This was not, as is sometimes supposed, a mark of mystery, but simply because it was the tenth in a number of properties acquired by MI6, all of which were designated using Roman numerals. The number of staff remained limited – only 137 of those who went to Bletchley in August 1939 were members of the GC&CS codebreaking sections. But the alliance between Stalin and Hitler continued to ensure that Russian material retained a high priority. Co-operation with the French was expanded. A dedicated Russian section was set up under Tiltman’s tutelage at Wavendon, a country house close to Bletchley Park, and another Russian section was set up at Sarafand, in Palestine.

  The crucial breakthrough on Russian armed forces material came during the Russo-Finnish War in late 1939 and early 1940. The large amount of traffic created by the Russian Army’s invasion of Finland, and the Finns’ determined defence of their territory, gave the codebreakers enough depth to solve two high-grade Russian systems: the Soviet Army’s GKK super-enciphered code and the OKF super-enciphered naval code.

  It is not clear who made the break into the two Russian high-grade systems, but it seems likely that Tiltman was yet again involved. He was in charge of the section, had worked on a number of similar Japanese Army systems and, the previous yea
r, had broken the Japanese Navy’s main super-enciphered code, known to the Allies as JN-25, within weeks of its introduction. It was also Tiltman who set up an arrangement with the Finnish Army’s codebreaking unit in order to obtain as much of the Russian traffic as possible to allow further recovery of the system. ‘He had the foresight to note the extreme importance of Finnish collaboration in our Russian work,’ Denniston said. ‘He spent a fortnight in Finland and established a close and friendly liaison with their cryptographic unit and his persistent drive … may well seal an alliance which should prove of the greatest value to the intelligence departments of all three services.’

  Tiltman agreed a similar deal to that with Estonia, promising to provide radio equipment in return for the traffic. Unfortunately, there were immediate complications. No sooner had he returned to England than the Russo-Finnish War came to an end. Nevertheless, the Finns insisted that they were still determined to expand their Sigint operations against the Russians. There was then difficulty persuading Stewart Menzies, who had succeeded Sinclair as Chief of MI6, to pay for the Finnish equipment. Even once he had agreed, some of the wireless receivers went missing. But eventually the arrangement began to work and the Finns provided an increased flow of Russian military and KGB traffic as well as five captured codebooks: three Russian Army codebooks – two of which had already been partially worked out by the Russian section – and two Soviet Navy codebooks. These were the KKF2 code used by the Black Sea Fleet and the KKF3 code used in the Baltic, which the British had broken a few months earlier but had not totally reconstructed.

  This provided fresh impetus to the work against the Russian Navy. Russian armed forces traffic now became the subject of a major effort by GC&CS and its outstations. Radio receivers were installed in the British Legation in Stockholm. Scarborough joined Flowerdown in monitoring Russian Navy frequencies. Vlasto was sent out to bolster the Russian section at Sarafand which, along with India, weighed in with a full range of interception, traffic analysis and codebreaking, sending high-grade cipher back to England via the diplomatic bag. The RAF set up an experimental intercept site in Baghdad to target Russian army and air circuits in the Caucasus. It also monitored Russian Navy intercepts from Cairo. The Royal Navy intercept sites at Dingli in Malta, Alexandria in northern Egypt, and Ismailia on the Suez Canal took Russian Navy material and fixed locations of the transmitters using direction finding. The Far East Combined Bureau (FECB), the main British Sigint outpost in the Far East, began taking Russian traffic from the Vladivostok area and kept watch on Baltic Fleet frequencies. It even asked the Australian and New Zealand Navy intercept organizations to provide any Russian naval material they could. At GC&CS itself, French naval code-breakers, who had been transferred to Britain, began working on high-grade Baltic and Black Sea Fleet ciphers.

  The flow of material from Estonia dried up in June 1940 when Russia occupied the Baltic States. But the fall of France brought willing replacements in the form of Polish wireless operators and codebreakers who, having escaped Poland to work with Gustave Bertrand, had now been forced to flee the German Blitzkrieg for a second time. Based at Stanmore in west London, they found they were able to monitor Russian material from the Ukraine and were co-opted by Denniston to provide more material for the Russian section.

  The increased effort brought yet another break, this time at Sarafand where the KKF4 Black Sea Fleet super-enciphered code was broken in November 1940. A few weeks later, the British and Americans exchanged Sigint material primarily relating to the Axis countries. The exchange took place at Bletchley Park in February 1941 and as part of their contribution, the British handed over four Russian codebooks: the OKF high-grade Baltic Fleet super-enciphered code broken at Wavendon and Bletchley; the EPRON codebook used by the Russian Navy Salvage Corps, which had also been worked out by GC&CS; the KKF4 Black Sea Fleet codebook broken at Sarafand; and the KKF3 Black Sea Fleet codebook provided by the Finns.

  The British operation against the Russian armed forces traffic was subsequently disrupted by a major change in codes and ciphers. But Denniston told Menzies that there was no doubt that it had ‘benefited largely by the increased interception from the Poles and that our Finnish liaison is becoming really attractive. This liaison, owing to recent changes in all Russian codes, is of first importance.’ By the beginning of June, there was serious concern that the increasing collaboration between the Finnish and German General Staffs might compromise the arrangement with Finland, leading the Germans to question how vulnerable their own ciphers were. The Poles were asked to reinforce their Stanmore operation and Tiltman even sought the advice of the MI6 Head of Station in Helsinki as to whether the Finns could be trusted to keep the secret. However this debate was overtaken by events, in the shape of Operation Barbarossa – the German invasion of Russia – which began on 22 June.

  According to the official history of intelligence, all intelligence operations against the Soviet Union now came to an abrupt halt on Winston Churchill’s orders. The reality was nowhere near as clear-cut. Indeed, initially, coverage was increased. There was a long drawn-out debate over whether or not to drop Soviet traffic. The codebreakers in India were only too happy to dispose of their Russian Air Force and KGB tasks which had produced ‘nothing of intelligence value’. But there were concerns that the lack of continuity would hamper attempts to break the Russian codes and ciphers in the future. As a result, it was not until December 1941 that the Russian section was closed down. Even then the Poles were told to continue intercepting traffic and trying to break it, while the British kept two sets monitoring known Russian frequencies at the Scarborough Royal Navy site and the RAF station in Cheadle.

  Within weeks, the Metropolitan Police intercept site at Denmark Hill and the Radio Security Service, an organization set up to monitor clandestine radio stations which now came under the aegis of MI6, had begun to pick up messages between Moscow and its agents in Europe. Despite the Churchill edict, MI5 had continued to keep a watch on the Russians and their links to the CPGB. But it was unaware of the ‘bundles of Russian traffic’ that had been intercepted until February 1943 when it discovered through its own sources within the party that Jean Jefferson, a party member, had been asked to resign in order to take an ‘illegal’ post operating the radio link to Moscow. Sir David Petrie, the Director-General of MI5, had a fraught discussion with Menzies at which it was agreed that Soviet espionage links should be monitored. This was subsequently refined to interception and decryption of the links between Moscow and communist parties across Europe. A small Russian team was secretly set up at a GC&CS outpost overlooking London’s Park Lane to break the cipher. The keys were taken from an English edition of Shakespeare’s plays, in order to avoid the risk of the operator being caught with a codebook. By the late summer of 1943, less than two years after they had closed their Russian section, ostensibly for the duration of the war, the British code-breakers were again reading Soviet traffic. The end of the Second World War was almost two years away, but the preparations for the second Cold War had already begun.

  3

  REMINISCENCES ON THE ENIGMA

  HUGH FOSS

  Introduction

  Although Bolshevik codes and ciphers had been the main target of the interwar years, by the time the codebreakers moved to Bletchley Park in late August 1939, these had been replaced by the attempts to break Enigma. It has been suggested that it was not until shortly before the war that the GC&CS codebreakers began to make any effort to break an Enigma machine. Although it is certainly true that Enigma had not enjoyed the highest priority during the interwar years, it was far from ignored. As the late Hugh Foss, who joined GC&CS in 1924, explains in the following chapter, he first looked into the possibility of breaking the machine in 1927, paradoxically with a view to seeing whether the British might want to use it themselves.

  The workings of the machine will be explained in later chapters. But for those as yet unfamiliar with the subject, it looked essentially like a small typewriter in
a wooden box. On most models, there was a standard continental QWERTZU keyboard, as opposed to the British QWERTY, and above that a lampboard with a series of lights, one for each letter of the alphabet. Inside the machine were a series of three, or sometimes four, rotors, which were the main elements of the encipherment system. The operator typed in the letters of the plain-text message. The action of depressing the key sent an electrical impulse through the machine and the enciphered letters lit up on the lampboard. On the later Wehrmacht models, there was a plugboard, or Stecker system, which increased the variations of encipherment to around 159 million million million possible settings. Put like that, it seemed impossible to break. But the reality was that there were still only twenty-six letters in the German alphabet and that gave the codebreakers a chance. Examining the commercial C Model he was given, where the number of different possible settings was merely several million, Foss decided that it had a high degree of security but, given certain conditions, it could be broken. If you knew a piece of original plain-text, a ‘crib’, that was at least 180 letters long, the wiring of the first two wheels could be worked out; and if the wiring was known, a crib of just fifteen letters would be sufficient to break the machine settings.

  GC&CS did a very small amount of work on the machine during the early 1930s. But it was not until the Spanish Civil War in 1936 that there was any real attempt to break live Enigma traffic. After some initial work by Josh Cooper, an Enigma machine, given by the Germans to the Italians and Spanish (the K Model), was broken by Dilly Knox on 24 April 1937, using an improved version of the system recommended by Foss ten years earlier.