Gordon Welchman Read online

  Once documented, traffic was passed to the Netz section (or Party as it was quaintly called). Their task was to find the daily key on particular German communication networks using the various tools developed by John Jeffreys and his team. Research and development of new tools was an ongoing activity and a bombe machine group was set up to test the early prototype of that device. It consisted of Miss E. E. Dawson and one assistant. An order for the first bombe had gone to the British Tabulating Machine Company (BTM) in November 1939, The Bombe Section was officially established and strengthened upon the arrival of the first prototype (called Victory) in Hut 1 on 18 March 1940.

  Once the key was ‘out’ the Deciphering section staff tested results from the Netz and Bombe sections and decrypted all the available traffic as quickly as possible. It was then passed to the service sections in Hut 3 for translation and analysis. In early 1940, Hut 3 had only three staff, Malcolm Saunders, S. C. Edgar and F. L. (Peter) Lucas.

  The Poles had exploited a weakness in the Germans’ procedures for Enigma in the 1930s. Henryk Zygalski had devised a method that used this weakness to help work out some of the settings used in the key for any given day. It was based upon a catalogue of perforated sheets and it had the huge advantage that it was not compromised by the plug connections on the Enigma machine. As there were only six wheel orders in operational use at this time (the operator could mix up his three wheels in six ways), a complete catalogue contained 26 perforated sheets each roughly 20 inches square, one for each of the possible ring settings on the right-hand wheel. Thus 6 × 26 = 156 sheets had to be manufactured. In December 1938 the Germans introduced two additional wheels so that each day, the Enigma operators chose three from a total of five available. This increased the number of wheel orders from 6 to 60 and, at that point, the Poles had managed to produce by hand, only two of the original six sets that they were working on.

  One of the positive outcomes of the conference in the Pyry Forest in July 1939 had been that the British, with greater resources at their disposal, agreed to produce the Zygalski sheets. The task was assigned to a team led by John Jeffreys and a machine was built to punch out the holes. It was a monumental task and apparently a small party was held to celebrate the punching of the two millionth hole. The Poles were doing the same work by hand with razor blades! The Zygalski sheet method was called the Netz method (or just Netz) at BP and because Jeffreys led the work, some authors have subsequently confused the Netz with another perforated sheet method which he developed. The Jeffreys sheet method was actually a catalogue of the effect of two wheels and the reflector. The Netz method was the one that Welchman reinvented in his early days at BP.

  The Poles had developed two machines to help with their work and they demonstrated both of them at the Pyry conference. The cyclometer helped with the task of constructing a card-index system containing information about all possible Enigma start positions that could have been used. The bomba had three pairs of Enigma wheel systems driven by an electric motor and exploited the same weakness as the Zygalski sheets. The Poles built six machines, one for each possible wheel order. When the number of possible wheel orders increased to sixty, the Poles would have needed sixty bomba machines and they did not have the resources to manufacture them.

  A key part of Welchman’s plan was to install Jeffreys and his team along with the Netz in Hut 6 as soon as they were operational. This would form the basis of Hut 6’s attack on the encrypted Enigma messages. However, when Jeffreys had produced two sets of the sheets, BP could not get them to work, so Turing was dispatched to meet the Poles, who had now fled Poland and were working at the headquarters of the French Cipher Bureau, code-named Bruno and located outside Paris at the Château de Vignolles. The Poles demonstrated to Turing how the British-made sheets could be used along with their other methods to find the settings in a wartime Enigma key. At the end of his trip, a farewell dinner was held and Zygalski asked Turing why each side of the square perforations on the British sheets was such an odd measurement, eight and a half millimetres. Turing explained that the British worked in inches and in fact that each perforation was one third of an inch square, much to the amusement of all present.6 Turing returned to BP and on or around 23 January 1940, a second set of the sheets was successfully used for the first time at BP. Shortly after this, the principal Luftwaffe key was broken. This, according to Milner-Barry, was a great event because:

  The first bombe was not yet in action, nor had cribs as yet been thought of. (Except probably in the fertile imagination of Welchman, ranging as usual a long way ahead of the event.)

  With the rest of his team in place, Welchman arranged for Jeffreys and his group to move into Hut 6. During the transition period, other breaks were occurring but not on an operational basis. Much of this traffic was days or weeks old and of little use for intelligence purposes. It did, however, prove useful for training. In his early work on call signs and discriminants, Welchman had adopted the technique of colour-coding different German communication networks by using coloured pencils. Thus the principal Luftwaffe key came to be known at BP as ‘Red’.

  Milner-Barry would later say that:

  I can still remember Welchman talking about dealing with the traffic on an operational basis at a time when the phoney war was still in progress and when the terms carried no significant meaning to me.

  By summer 1940 Welchman had put the hut on a twenty-four-hour rota to avoid the Registration Room having an accumulation of some fifteen hours of traffic. The first members of the night shift were David Gaunt, Michael Banister, Sheila Dunlop and June Canney. There were three shifts: 0000–0800, 0800–1600, 1600–2400 with three or four staff per shift. With keys changing every day at midnight, the main task was to break the current day’s Red key. On one very successful day, it was broken at 5.00 a.m. and a thousand messages which subsequently came in on it were read. Plans were also put in place for a mobile Hut 6 after Dunkirk. At this stage, the whole Hut 6 operation was still fairly informal with none of the rigid differentiation of function between various members of a shift that later proved necessary. The Machine Room had been set up and the staff there would start by examining traffic on the registers, which were a continuous list of messages by each station with essential preamble detail. These lists were known as blists, a shortening of Banister Lists after their originator, Michael Banister. Machine Room staff would arrange the data in a suitable form for the Netz Room to try to complete the key by using the Netz method and other techniques. The registrar would underline Red discriminants with a red pencil to distinguish them from traffic on another network.

  Many of the secrets of the Enigma were unlocked by Knox and his team, which on occasion included visitors from Hut 6 such as Rees and Babbage. Knox had pioneered the technique of looking for wheel order tips by using Enigma operator mistakes, which were called cillies. It was an esoteric mystery presided over by Knox in The Cottage, assisted by Hut 6 visitors. One example of a cilli occurred when a German Enigma operator, after encrypting a message, proceeded to encrypt the next message without moving his wheels.

  Another extraordinary mistake made by the German cryptographic departments was the procedure imposed on those producing the monthly key sheets for each communication network. BP called these the Rules of Keys7 and discovered them for Red in June 1940, and confirmed them during the following months. Many of the rules were discovered by members of Hut 6 whose names would be attached to them for the amusement of all involved as well as posterity. Nigel Forward, for example, arrived in Hut 6 in February 1941 and made the remarkable discovery that all German Air Force keys selected their wheel orders, not from the sixty possible, but from a list of only thirty. This dramatically reduced the amount of bombe time needed, particularly towards the end of each month. The discovery became known as the Nigelian Wheelorder Rule. Lionel Clarke arrived in July 1940 and about a year later discovered that in the columns on the monthly sheets which gave the wheel orders for each day, no wheel was followed by a conse
cutive wheel. In other words, if the wheel order on day one was 123, then on day two, the left-hand wheel would not be 2, the middle wheel 3 nor the right wheel 4; this rule was named the Clarkian Wheelorder Rule. John Monroe joined this illustrious band of rule discoverers when he realized that, on some keys, all five wheels would always be used on consecutive days. While this was not as absolute as the other rules, the Monrovian Wheelorder Rule was born. Some other rules were discovered including:

  • On any two consecutive days in the same month, the same wheel could not be used in the same place. Within a month the same wheel order could not be used twice.

  • In the first 26 days of the month, each ring setting was a unique letter, then arbitrary for days 27 to 31.

  • Consecutive plugs were never used on the plugboard apart from A and Z. There were only 300 possible pairings but repeats in a month were avoided if possible.

  Many of the Rules of Keys provide compelling evidence that the Germans believed throughout the war that their Enigma systems could not be broken. The rules must have been introduced on the basis that the enemy would never know any of the settings from a previous day of the month!

  Hut 6 was now ready to take the lead in the attack on Enigma. Despite his outbursts and peculiar behaviour, Knox had been the Enigma pioneer in Britain and his energy and enthusiasm had inspired all in the early years. But now that enough of the Enigma secrets were exposed and Hut 6 was up and running under Welchman. Knox had to accept his new ‘Research’ role and turned his attention to other things such as the Enigma machine being used by the German military intelligence organization, the Abwehr. This device did not have a plugboard but was extremely complex with three moving wheels with multiple turnover positions and a reflector which moved during operation (in contrast to the standard service Enigma which had a fixed reflector). Yet Knox’s team achieved great success against it. Knox worked on until, tragically, he succumbed to the cancer which had plagued him throughout the war, on 27 February 1943.

  Throughout the whole history of Hut 6, there appear to have been extraordinarily few examples of innovations in German technique of which BP did not have adequate warning. Someone always seemed to be willing to take on the next problem and come up with a solution. By the beginning of May 1940, Hut 6 had broken four Army and five Air Force keys. Red, which was used by the Luftwaffe’s general operational network and first seen by Hut 6 in September 1939, was being broken on a daily basis. Then, around 10 May, Hut 6 traffic registers showed up a change in German operator procedures.

  The existing procedure had been for the sending operator to choose a three-letter message setting, say KAR, which would be the starting position of the Enigma wheels after the machine had been set up using the daily key settings for both encryption and decryption of the message. He would then choose another three letters, say GQX, and turn the wheels until these letters appeared in the window for each wheel. He would then type in KAR twice which might yield TLFQEP. The letters GQX would be sent in the preamble and TLFQEP would be the first six letters of the encrypted message. The receiving operator, having set his machine up with the daily key settings, would turn his wheels to GQX, type in TLFQEP to recover KARKAR. He would then set his wheels to KAR and begin decrypting the message by simply typing the encrypted message text on his machine. As the lamps lit up, they would reveal the plain text of the message. When staff in the Hut 6 Machine Room saw that there was an extra three-letter group in the preamble of the message, they quickly guessed that the Germans had stopped the double encryption of the message setting. For the example above, the preamble now contained the two three-letter groups, GQX TLF.

  The news must have been shattering to the staff in Hut 6 because it meant that the Netz method was useless to them and consigned to history. It had been based entirely on the double encryption of the message settings and patterns which resulted from this. However, it could not have been entirely surprising, particularly to Welchman, Turing and other colleagues who had recognized that the Netz method was fragile, based as it was on a German mistake which could be rectified at any time. In fact, a machine-aided method which was not dependant on the German indicator system had been under discussion at BP for some while. According to Peter Twinn, in 1984 correspondence with Welchman, the idea of Enigmas linked together and going through all possible positions was certainly in many people’s minds before Turing started thinking about their use after the change to the German indicator system. Thinking about a machine-based solution had yielded the bombe but the prototype was not fit for general use.

  Before the German procedure change, with his usual foresight, Welchman had asked Milner-Barry to make a study of decrypted messages. He hoped that this would yield an intimate knowledge of the Enigma operators and mistakes that they might introduce due to ‘human factors’. What it yielded was a catalogue of operator errors which helped Hut 6 break back into Red on 22 May. During the campaign in France, this resulted in over 1,000 decrypted messages a day and Welchman decided to concentrate most of his then limited resources on it.

  The possibility of using operator errors had been pointed out by Dilly Knox as early as January 1940 and the name for them – cillies – could well have been originated by him or one of his team in The Cottage. In any event, it led to what Welchman referred to as a comedy of errors and a more detailed explanation about how they were exploited can be found in Appendix 3. The classic example of a cilli was an operator’s choice of the six letters for his indicator settings. As only the first three letters would appear as plain text in the preamble, operators were choosing memorable six-letter words like HITLER or BERLIN. A particularly amusing example was recalled by Art Levinson, one of the American contingent at BP.8 The first three letters sent by the operator in plain text were TOM, followed by the encryption of his second three letters, say XGH. Levinson and his colleagues immediately tried to think of six-letter words beginning with TOM such as TOMTOM or TOMMEY. Much to their surprise, it turned out that the six-letter word chosen by the operator was TOMMIX. Tom Mix was a Hollywood cowboy who, much to the astonishment of the Americans at BP, appeared to have a large fan club in Nazi Germany!

  John Herivel was a man on a mission and once he understood the working of the Enigma machine, he spent much of his time thinking about how it could be broken. His eureka moment occurred when he imagined himself as an Enigma operator setting up his machine for the day. He realized that when the operator set the position of the ring around each wheel, the physical arrangement of the machine meant that he might well end up with the ring setting at the top of the wheel. All three ring settings would then be viewable through the windows adjacent to them. A lazy operator might then use these as his indicator setting for his next message or perhaps just turn one or two wheels by a position or two. If this happened, Herivel came up with a method which, if successful, would produce the ring setting for the day. Welchman encouraged him to carry on with it, even though it did not yield useful results at first. Then, when he arrived at Hut 6 on 10 May 1940 to take up the evening shift, Welchman was waiting for him. In the corner of the Hut surrounded by a group of his colleagues, David Rees was in the process of completing the breaking of Red with the aid of Herivel’s method. Welchman took him to one side and said ‘Herivel, this will not be forgotten.’9 The Herivel Tip (or Herivelismus) as it became known, would prove to be a key tool for Hut 6 until the arrival of the bombes.

  Welchman was true to his word the following year when Prime Minister Churchill visited BP in early September. Travis accompanied Churchill on his tour around the site and, on arriving in Hut 6, informed Welchman, who had been asked to give a short presentation, that he had five minutes. Welchman said that he had three points and after completing the first two, Travis said ‘That will do Welchman.’ The Prime Minister winked at him and said ‘I believe there was a third point, Welchman?’ He then took Churchill over to where Herivel was standing and said ‘Sir, I would like to present John Herivel who was responsible for breaking the G
erman Enigma last year.’ Herivel was moved to end his own memoir published sixty-seven years later with the words ‘So Floreat Welchmani memoriam!’

  As his workload increased, Welchman was finding it more and more difficult to get back to Cambridge. After Katharine had given birth to their second child, Susanna, on 1 January 1941, the family decided to move closer to BP. After living in the villages of Fenny Stratford and Loughton, they eventually settled in the town of Stony Stratford. Their Queen Annestyle fronted house was on Watling Street, a Roman road which ran though the centre of the town. The town had been an important stopping-off point for mail and passenger coaches travelling between London and the North of England. This coaching history is the source of the supposed origin of the phrase ‘cock and bull story’. In the height of the coaching era – the eighteenth and early nineteenth centuries – the Cock and the Bull were two of the main coaching inns in the town and the banter and rivalry between groups of travellers is said to have resulted in exaggerated and fanciful stories, which became known as ‘cock and bull stories’. The Welchman house lay along the main road, a short walk from the two inns which were still thriving businesses. However, rather than horse-drawn coaches, the road in front of the house was now choked with military convoys. On one occasion, Welchman was involved in an accident on the main road when his car was sandwiched between two military vehicles. His black car and the heavily blinkered headlamps in use in wartime no doubt contributed to it. The Welchmans would often entertain friends at home and, in particular, musicians would arrive for impromptu musical evenings with Welchman providing beer from one of the nearby pubs. The household also included a nanny, Miss Ring, and occasionally a maid, which became essential when Katharine joined the ATS in 1942.