Gordon Welchman Page 10
The machines themselves were spectacularly successful and the performance records recorded in the Official History of Hut 6 show availability figures across all machines throughout the war at just under 98 per cent. Given that there was regularly scheduled down time for maintenance and that they ran twenty-four hours a day, seven days a week, their performance was extraordinary and a testament to the skills of Doc Keen and his team at BTM. Other variants of the bombe would follow including the Jumbo which did more automatic testing than the ‘spider’. It also printed out results which eliminated much of the hand-based work which had previously been necessary. As not many of these machines were made, they were reserved for important messages which caused difficulties for the spider. A ‘Giant’ bombe was built in June 1944 just after D-Day and consisted of four bombes linked together. However, it never left BTM’s Letchworth factory as it proved to be unreliable. Once the Americans became involved in BP’s work, they began producing their own version of the bombe.
Joan Clarke had been recruited by Welchman in 1940, but rather than working in Hut 6 she had joined Turing in Hut 8 on 17 June of that year to help with work on the German Naval Enigma problem. The German Navy was much more security-conscious than its Army and Air Force counterparts and in the early years of the war, its Enigma-encrypted messages had proved to be unbreakable by BP. An excellent cryptanalyst in her own right, Clarke had been present when Turing had fully realized that Welchman’s invention of the diagonal board would significantly improve the performance of the bombe. She had also been engaged to Turing for a time and while it had not worked out (Turing had warned her that this might happen because of his ‘homosexual tendencies’), they did remain colleagues and good friends even after the war. Sometime in the late 1970s, as Joan Murray, she produced a document for the NSA about her personal memories of the bombe development.5 The following extract provides some insight into Welchman and Turing’s working relationship:
The heads of both Hut 6 and Hut 8 were involved in the vital developments in the logical design of bombes, which took place near the time of my arrival. The first was Welchman’s idea of the diagonal board, which made use of the reciprocal property of the stecker. I understood later from Turing that Welchman’s objective in specifying this was simply to provide entry to a secondary chain of constatations [one pair of matching plain/cipher text]. With the original form of test on the bombe, this secondary chain would need to include a closure in order to be of any value in reducing the number of bombe answers. Meanwhile, both Welchman and Turing were looking for a general method of achieving simultaneous scanning, i.e. testing all stecker assumptions for the input letter at the same time. I remember Turing jumping up with the remark that ‘the diagonal board will give us simultaneous scanning,’ and rushing across to Hut 6 to tell Welchman. Turing’s contribution was the realization that a wrong stecker assumption for the input letter would imply all wrong steckers, if one allowed an unlimited number of re-entries into the chain. In the electrical implementation which provided simultaneous scanning, 25 relays represented the other steckers for the input letter, and the new test was whether any of these relays was not activated. When I mentioned the subject to Turing after the war, when he was visiting GCHQ at Eastcote as a consultant, he minimized his own contribution compared with Welchman’s idea of the diagonal board, saying that Welchman or someone else was bound to have realized it before long – but I doubt whether anyone else would rate Turing’s contribution to bombe theory so lightly.
On 15 July 1941, Welchman, Turing and Alexander were invited to the Foreign Office in London by Lord Cadogan.6 Cadogan was Permanent Under-Secretary for Foreign Affairs, 1938–46. In an unpublished extract from his diaries he had mentioned a visit to Bletchley on 11 January 1941: ‘Very interesting – I should like to spend a week there so as to try to understand it. A charming young Cambridge Professor of Geometry – Welchman – did his best. A good show, I think.’ According to Cadogan the three of them were referred to simply as Menzies’s ‘Brains Trust’ and all presented by Menzies and given £200 each.’7 It was Welchman’s impression, supported by Joan Clarke, that they were recommended for a decoration but had to be given money instead.
Chapter 5
Expansion and Consolidation
By the end of May 1942 Welchman had already presided over three key periods in the development of Hut 6 and, through its efforts, the success of BP. The first period had begun with Dilly Knox’s pioneering prewar work in breaking the early version of the Enigma machine without a plugboard. Knox had devised a number of ingenious tools and methods to help with this work with arcane names such as rodding, buttoning-up, lobsters, crabs and boils. He had also been the recipient (although not initially a grateful one) of the results of the ground-breaking work by the Polish cryptanalysts in the 1930s. This had led to the construction of the Zygalski sheets and the successful adoption of the Polish methods at BP in late January 1940. A further development called the Jeffreys sheets was also put to good use. As this initial period ended, Welchman had assembled the nucleus of the original members of Hut 6.
The second period of activity from January to July 1940 had seen the Germans occupy Norway and France as well as the disaster of Dunkirk. Hut 6 was a limited operation initially with the Netz Room engaged in moving the sheets around to allow the holes punched in them magically to reveal part of the daily key; the Machine Room arranging the data for them and completing the key; the Registration Room logging all incoming traffic and finally the Decoding Room turning the incoming messages from gibberish to German and, it was hoped, vital intelligence. By March they had managed to break fifty daily keys across three German communication networks (called Red, Blue and Green). However, the Norwegian campaign saw the real start of Hut 6’s continuous operational breaking of German keys.
Disaster had struck on 29 and 30 April when the new indicator system appeared in some of the traffic coming into Hut 6. As was often the case, the Germans had provided Hut 6 with ample warning of this change. Remarkably, they used the very system that Hut 6 was breaking on a daily basis to inform Enigma operators about new security measures being introduced to the system. In fact, throughout the whole history of Hut 6, there appear to be very few examples of any innovations in the German cryptographic system of which Hut 6 did not have adequate warning. However, the new indicator system rendered the Polish method, including the Netz, obsolete.
The 10th of May 1940 was a turning point, not only for Hut 6, but for Britain and Europe as a whole. As Hitler launched the invasion of Belgium, the Netherlands, Luxembourg and northern France, the volume of wireless traffic coming into Hut 6 rose to levels previously unimagined by most at BP, except perhaps Welchman. On the same day, the man who would become BP’s biggest supporter stepped onto the world stage. Winston Churchill became Prime Minister at the head of a coalition government. With limited resources at his disposal, Welchman had a difficult decision to make. It was obvious from the volume of traffic on the main Air Force network (using the Red key) that it was of paramount importance. While there was a considerable amount of Army traffic coming in which could have contained high-grade intelligence, resources were not yet available to break it. Therefore, Welchman took the decision, in consultation with other senior figures at BP, to concentrate Hut 6’s attack on Red.
With the Netz Room out of action and the machines which would ultimately lead the attack not yet available, Hut 6 turned to hand methods. Breaks at this stage were based on cillies, which helped to reduce significantly the number of possible key settings used by the operator. A method based around cillies had originated in The Cottage under Knox and new classes of cillies were regularly exploited by Hut 6. The German Army and Air Force operators were allowed to choose ‘at random’ the three letters which would make up the message setting. This was in effect a final scrambling of the wheels before encrypting the message. It proved to be a serious mistake as human beings are not random and their choice of letters often proved to be anything but (s
ee Appendix 3). The Herivel Tip was another example of how analysing the behaviour of Enigma operators could reveal errors in procedure which could be exploited by the skilful cryptanalyst.
The third period ran from August 1940 to May 1941, when Britain stood alone against the threat posed by Nazi Germany. It also saw the arrival of the bombes and the evolution of the crib into an art form in its own right, known at BP as cribbery. With the first bombes incorporating the diagonal board now operational, Welchman turned his attention to integrating cribbery into Hut 6’s production-line approach. Likely looking messages were usually identifiable by a combination of length, frequency, time of origin, time of intercept, call sign, and similar standard features. Different sections of the message would be written down on strips of paper. Based on their experience, the cribsters would make an educated guess as to a phrase which was likely to be hidden somewhere in the encrypted message. They would also use their experience to determine which section of the message was likely to contain the phrase. The guessed phrase would be written down on a separate slip of paper and placed on a surface above or below the appropriate section of the encrypted message. By moving the guessed phrase left and right, the cribsters hoped to identify the encrypted version of the phrase. The technique was based on the fact that the Enigma machine could not encrypt a letter as itself. If they were successful the cribs were handed on to the Machine Room where they were turned into their graphical representations known as menus. The Crib Room became the ‘chief breaking agency’ and under Welchman’s direction, took charge of general bombe policy and relations with both Hut 3 and the naval sections.
The whole course of the Second World War changed when Adolf Hitler, realizing that Britain would not be bombed into submission, instead looked east and on 22 June 1941, attacked the Soviet Union. At the same time, BP entered its fourth period of development, one of expansion and consolidation. The peak of its success in terms of the quantity and quality of intelligence was probably achieved in the months following D-Day. However, from June 1941 to the end of 1943, BP, and particularly Hut 6, got almost completely on top of the Enigma system. As access to resources such as bombes, radio sets and staff increased, the Luftwaffe’s communication system was at the mercy of Hut 6. By the end of December 1942 the Hut 6 team was breaking almost every key that was generating a large amount of traffic with a reasonable frequency. Even more impressive was the fact that a satisfactory solution had been found for every technical problem presented to them.
As Germany initiated new campaigns, prepared to meet expected attacks by the Allies or split up existing keys because of alterations in their system of key distribution and allocation, the number of keys increased dramatically.
By June 1941 cillies had become very rare and the opportunities for the mathematicians to show their prowess in hand-breaking were scarce. The Machine Room spent its time in testing endless bombe ‘stops’1 and morale began to suffer. In the end the testing of the bombe stops was put out to the bombe outstations. Hut 6 would only be sent the output from tested stops which offered possible key settings.
As long as there were enough different coloured pencils, the Army and Air Force keys were just given corresponding colour names. With the invasion of Russia, Hut 6 began to run out of coloured pencils and keys named after birds, such as Kestrel and Vulture, sprang up. In 1942 the German Air Force started to use a large number of subsidiary keys and John Colman set about giving them names on some kind of logical basis and from then onwards, he was responsible for key name allocation. The Army keys were birds, Air Force keys fell into a number of groups including insects, vegetables, flowers and mammals.
It became necessary to have more than a single person on each shift as there were so many keys to tackle. In June 1941 there was often only one person in the Watch room. In later years the number increased, often to five. In preparation for D-Day, two separate watches were put in place, one for Army and one for Air Force keys. There were three main tasks on the watches. The first of these, essentially supervised by the head of the Air Force shift, was to know what the bombes were doing and to have a ready supply of jobs waiting as bombes became free if a key ‘came out’ or ‘went down’. The head of shift had to be in a position to brief his successor as head of shift. New jobs were passed through a hatch to the Machine Room where its head of shift was in direct telephone contact with the bombe outstations and could advise them on job priorities. The head of the Army Watch usually left instructions with the head of the Air Force Watch shift about their priorities.
The second task under the head of shift was known as e.p. (after the chess term en passant, meaning ‘in passing’) and involved receiving decrypted messages from the Decoding Room (through another hatch) and noting anything of cryptographic interest before putting them on a conveyor belt to Hut 3. Each key had someone responsible for it (at least one parent) taken from members of the Watch, whose job included keeping records about the key, The final check by the head of Watch had to be done fairly quickly to avoid unnecessary delays.
The third main task was to deal with incidents of re-encrypted messages. With the proliferation of keys many messages were encrypted on at least two different keys. This was manna from heaven for the cryptanalysts in Hut 6 and it was essential to record them carefully. The Registration Room made out a slip of paper with brief particulars of a re-encrypted message and added a cross in the corner in the colour associated with the key on which it had been sent. These slips were consequently called kisses. The person responsible for this in the Registration Room sorted the kisses and checked for other re-encrypted messages on a key, regardless of whether it was ‘out’ or not. If a promising re-encrypted message was discovered and neither the original or new key was ‘out’, he would wait until one key was ‘out’ and then ensure that the message on the other key was promptly decrypted and record it. If one key was already ‘out’, this person had to obtain a copy of the decrypt either by having a duplicate done in the Decoding Room or by finding it in Hut 3.
Other things needed attending to on shifts, including assisting the Machine Room with difficult settings and preparing menus, which were in effect, instructions for the bombe operators. In the latter stages of the war two female operators were employed to prepare the menus on the instructions of one member of the Watch. These then had to be checked by another member of the Watch. There were also ‘jobs’ which, when they became available, the US-based bombes were better suited to deal with. While the American bombes gave priority to naval jobs, they had a great deal of spare time, which was used by Hut 6 for research projects under the direction of John Manisty.
By 6 August 1941, the various inter-relationships in Hut 6 had settled down pretty well, but Welchman was still finding that, perhaps because he was the originator of the various information-handling procedures, he still had a more complete overview than anyone else. This was partly why he started regular management meetings and insisted on weekly reports from all sections being posted on all bulletin boards. The big expansion of Hut 6 and Hut 3 and their supporting activities had started in early 1941. The people who arrived after that were fitted into an existing organization. While their jobs were well defined, they were under pressure as soon as they started working in the huts.
As the workload increased, Welchman started to look at streamlining the organization to make it even more efficient. While he prided himself on having created a very friendly and collegial working environment, problems did occur. Space was becoming increasingly limited and when the Machine, Crib and Registration Rooms established research sections, they had to be housed in the Mansion. This problem was relieved in February 1943 when they were able to move into the spacious accommodation of the newly opened Block D. This building had state-of-the-art pneumatic tubing and pulley systems installed to allow documents to be moved from section to section. Block D was also large enough to house Huts 3 and 8 as well as several other related BP sections.
Setting up of the Watch in February 194
3 was Welchman’s last major reform in Hut 6 and produced an immediate and permanent improvement in the atmosphere. The changes allowed one man to have complete responsibility for a key and also raised the status of research in the Hut 6 operation. Up to the spring of 1943, research had always been something of a poor relation, but not in Welchman’s eyes. Harold Fletcher remembered Welchman as someone ‘who always saw the long term importance of things more clearly than anyone else’. For a time, at least two bombes were even allocated to research but this proved hard to maintain.
As the number of keys started to multiply during 1942, many of them proved to be a mixture of other keys. This was detected because Reg Parker had been keeping meticulous records of broken German keys. Parker had joined the Hut 6 Control team in May 1940 and had taken it upon himself to keep logs of all broken keys. His reasoning followed the classic approach to cryptography, that is to get inside the head of your enemy. Parker reckoned that someone in Germany had to produce a multitude of setting sheets every month and this must be a very boring job. To produce a monthly setting sheet, he would have to create for each day of the month a wheel order (three numbers from one to five), the ring settings for each wheel (a three-letter group) and ten pairs of letters to be plugged together on the plugboard. There were also the discriminants and call signs which had to be included with each day’s key settings. The chap was bound to be tempted to re-use some of the settings that he had previously worked out, following the various rules that had been imposed upon him for this task.