Gordon Welchman Page 11
After faithfully doing this for several months, Parker finally discovered that this was exactly what was happening. The normal practice was to make up a key out of the wheel order and ring settings of an existing key and the discriminants and plug connections from another. John Monroe remembered one month during which Red had the plug connections of the previous month’s Mustard. As this latter key had been broken two days out of three, it could be used to help break Red quickly. It became even more interesting when it was discovered that the wheel order and ring settings for Mustard had also been used the next month by a fairly unimportant key named Cockroach. It was thus possible to break Red by first breaking Cockroach, then using the resultant wheel order and ring settings on the previous month’s Mustard to work out its plug connections which had then been used in Red. This method became known as Parkerismus in Hut 6.
The whole catalogue of German mistakes seems to have been founded on their belief that the Enigma machine and their procedures for its use could not be broken. The ‘Official History of Hut 6’ concluded that:
It must be emphasised that, theoretically, if the enemy is not breaking your keys then you can use repeats, partial or otherwise, as much as you like, although there is a chance that a capture might give you away. If, however, you feel you must use repeats then use them in a completely patternless manner so that even if the enemy does establish that one key is constructed from another he can only ascertain which day is used for which day by breaking both without the aid of the repeat.
From all this it seems that one of the most desirable attributes of a key compiler is full confidence in the ability of the opposing cryptographers.2
It could well be advice such as this that persuaded GCHQ to wait until June 2006 to declassify the three volumes of the ‘Official History of Hut 6’.
In June 1942, 1,170 messages per day were being decrypted and by August, 500 breaks had been made and at least 50 separate German communication networks recognized. Increasingly, a definite attempt was also made to use known and forthcoming Allied intentions to guide Hut 6’s activities. As the amount of decrypted material flowing to Hut 3 increased, Welchman wanted a broad overview both of the cryptographic situation and of the present and future work of Hut 3. In the hope of achieving a much closer and more effective interrelationship between the two huts, he persuaded the new head of Hut 3, Eric Jones, to appoint a spokesperson for Hut 3. This individual would acquire an overview of the ever-changing needs of both huts and establish an effective intelligence liaison between them.
Jones appointed Oscar Oeser to the role as head of a new group called 3L. Throughout his professional career, before, during and after the war, Welchman demonstrated the admirable quality of being able to separate personal issues from professional ones. His relationship with Dilly Knox was a good example of this and another was his relationship with Oeser. They became good friends and Oeser occasionally visited the Welchman family at their home during the war. While he even enjoyed working with him, he told colleagues that Oeser was a bad choice to head 3L. According to Welchman, Oeser never understood the co-ordination of many activities that was essential to the success of Hut 6, so he could not explain what really mattered either to his own 3L staff or to the key people in Hut 3.
More systematic consultations were arranged with Hut 8, Hut 6’s sister section for naval decryption and Hut 4, the sister section for Hut 3 which had moved into Block A adjacent to Hut 8. Access to the bombes was also hotly contested so a rota of bombe directors from Huts 6 and 8 was set up and given the power, during their turn of duty, to decide on the distribution of the bombes. This group was eventually disbanded and replaced by weekly meetings at which intelligence and cryptography staff from both huts were represented.
Hut 6’s final period of activity would run from the beginning of 1944 until the end of the war in Europe in May 1945. The work would be led by Milner-Barry as Welchman would be promoted and moved to a new role. For almost three years he had been the driving force behind Hut 6’s many successes. He also carried the burden of being one of the few people at BP who understood the reality of their position. In a note to Travis dated 15 December 1941, he warned that:
The enigma code as now used by the German Air Force and German Army is, as far as is known at present, theoretically unbreakable. Our methods of breaking now are purely opportunist and are based on taking advantage of the enemy’s mistakes. We have been extraordinarily fortunate in that he has continued to make mistakes in just sufficient number to enable us to break some of his codes fairly regularly.
Our success has always hung by a thread, and must continue to do so, unless indeed we can devise some new methods or invent some new and powerful machinery; and nobody has yet any idea how this could be done. What is required and is essential in order to break, is a certain minimum of positive data, and if the Germans were to use their machine properly it is admirably adapted to withholding such data from us altogether.
Now there is conclusive evidence that the German security officers are very much alive to the dangers of this practice even though they may not be aware of the exact means by which we take advantage of it. We often decode messages in which operators are hauled over the coals for doing just this kind of thing. This source of success is, therefore, now of relatively less importance.
Cribs are now our main line of attack against the major keys. Provided we can guess at the exact (letter for letter) German text of any single message or part of a message up to a length of say, 20–40 letters, this will give us sufficient data (if everything else goes right) to enable the bombe to give us the answer.
In sum, it would, I think, be much easier for the German security officers to prevent their operators from giving us the benefit of cribs than stop them from using favourite indicators when they are in a hurry. But, secondly – and this is the main point of these notes – even on the hypothesis that the Germans never guess, we can never be confident of continued success. The good luck which has so far attended us in the matter of cribs is little short of miraculous.3
*
1943 was something of an anti-climax after the rapid pace of the previous two years. It was also a transitional period as key repeats like those exploited by Parkerismus ceased and thus the pace slackened. It also saw the move to Block D in the spring as well as setting up of the Watch and the larger Research Section under Babbage. With a lull as the Germans were forced back into fortress Europe, the main attack switched to the research keys, especially the Army ones. Derek Taunt and Bob Roseveare led on the Air Force research while David Gaunt and Doug Nicoll led on Army research. Babbage and Aitken were known as the ‘senior partners’ and oversaw their work.
In September 1943, the Germans dropped discriminants from Army traffic and in Milner-Barry’s opinion this was the most dangerous period BP ever went through. The Air Force duly followed suit by dropping them at the beginning of November. Hut 6 had been forewarned of this from a decrypted message in March. The Germans had been using discriminants for years which had made sorting and identification comparatively straightforward. Hut 6 was desperately afraid that sorting might become next to impossible. A scheme was devised and duly put into action on 1 April 1944. Welchman regarded it as overly complicated, but it did produce a solution. A new section was set up called the Duddery and run by David Gaunt and subsequently saved the day by successfully identifying the keys as they came into the Watch.
Hut 6 managed to defeat most of the technical innovations thrown at it by the German cryptographic sections. In the last year of the war, new modifications to the German Enigma machine presented Hut 6 with fresh challenges. Each machine had a component called the reflector (Umkehrwalze in German) which acted like a stationary wheel with letters wired in pairs. The current running through the Enigma machine passed out of the left-hand wheel and back into it through the reflector before making its way to the lampboard. The first modification was the introduction of a pluggable reflector, Umkehrwalze D. Up to that point, th
e Air Force and Army had used Umkehrwalze B and the Navy, Umkehrwalze B followed by C. The wirings of both of these reflectors were known, so there was quite a stir when a decrypted message revealed the existence of an unknown reflector. Preparations were made to deal with it and when the Red key was ‘out’ on 1 January 1944, it was found that some signals on certain networks could not be read (known as duds at BP). It was fairly quickly established that this new reflector was pluggable and could be changed daily along with the rest of the key settings. While the D reflector proved to be a major problem when it was used exclusively on a network, again German mistakes provided the solution. On most keys, the B reflector was used as well as the D reflector, so once the key was broken with the B reflector, the connections on the D reflector could be found,
Just after D-Day, Hut 6 started finding a number of messages that started, when decrypted, with a number spelt out in German and then ended with meaningless characters. Once again, another decrypted message revealed the existence of the Enigma Uhr. This turned out to be an attachment to the Enigma machine which made it possible to change the plug connections in a number of additional ways. Remarkably, this new problem was solved on a regular basis by Hut 6. The recovery of the connections on the D reflector and the Uhr settings in connection with any broken key was normally an additional function of the Watch, with the task being assigned by the head of the shift to a member competent enough to take it on.
Annually, in the spring, the Germans rearranged all the frequencies of their networks, so they all had to be rediscovered. Intercept operators were, however, able to identify the idiosyncrasies of the wireless operators. In February 1945 the Germans started to change the frequencies daily and encrypt the call signs, which proved to be unbreakable. John Monroe remembered working two twelve-hour shifts with Nigel Forward in an attempt to identify frequencies. Fortunately, the Germans reverted to their normal frequency-setting regime fairly quickly.
*
In the summer of 1941 BP had its first warning that German submarines were being issued with new four-wheel Enigma machines and new code books. The German Navy as a whole was much more security conscious than its Air Force and Army counterparts. It did not use the stereotyped language which had provided many of the cribs being used by Hut 6. It also did not leave the choice of message setting to the operators but instead issued them with elaborate code books for this purpose. The message setting gave the starting position of the wheels in the Enigma machine and had to be worked out along with the daily key before the decryption of messages could begin.
The Navy Enigma was generally considered to be unbreakable in 1939 and when Alan Turing arrived at BP in early September little work was being done on it. He became interested in it ‘because no one else was doing anything about it and I could have it to myself’. Through a brilliant piece of analysis, he fairly quickly worked out how this code book system worked. However, even Turing could not reproduce the tables in the code books and it was felt at BP that only ‘a capture’ would allow them to make further progress. The situation got so desperate that an extraordinary scheme to capture the secret code books was considered. Lieutenant Commander Ian Fleming was personal assistant to Rear Admiral John Godfrey, Director of Naval Intelligence. He visited BP twice a month and was well aware of the problem. He wrote to Godfrey on 12 September 1940 with the following plan:
I suggest we obtain the loot by the following means:
Obtain from Air Ministry an air-worthy German bomber.
Pick a tough crew of five, including a pilot, W/T operator and word-perfect German speaker.
Dress them in German Air Force uniform, add blood and bandages to suit.
Crash plane in the Channel after making S.O.S. to rescue service in P/L [plain language].
Once aboard rescue boat, shoot German crew, dump overboard, bring rescue boat back to English port.
In order to increase the chances of capturing an R. or M. with its richer booty, the crash might be staged in mid-Channel. The Germans would presumably employ one of this type for the longer and more hazardous journey.
While similar scenarios by Fleming would eventually yield him fame and fortune through his James Bond books, in this case common sense prevailed. The cancellation of Operation Ruthless, the name given to Fleming’s plan, was a major disappointment to Turing and Twinn. As Frank Birch, the head of the Navy Section at BP, wrote in a letter on 20 October 1940:
Turing and Twinn came to me like undertakers cheated of a nice corpse two days ago, all in a stew about the cancellation of Operation Ruthless. The burden of their song was the importance of a pinch.
Eventually captures (or pinches as they were called at BP) were made and Hut 8 began to decrypt German Navy Enigma messages.
In February 1942 the warnings about a four-wheel Enigma become a reality and Hut 8 could not read the Enigma traffic coming from the U-boats. They were attacking Allied convoys crossing the North Atlantic and almost 700,000 tons of vital Allied ships and cargoes were sunk around the world that month. Only eleven U-boats were sunk in the first six months of 1942. Salvation came in October and the story behind it was worthy of a Hollywood film. Regrettably, one was made years later in which the heroes of the hour were American rather than British.
On 30 October 1942, the submarine U-559 under Lieutenant Hans Heidtmann was operating in the eastern Mediterranean. It was detected by British radar and four destroyers were sent to search for it. One of these was HMS Petard, a P-class (for Pakenham) fleet destroyer. Its captain, Lieutenant Commander Mark Thornton, and his first lieutenant, Antony Fasson, had discussed in detail how they would board a captured U-boat and had drilled a boarding team. The British ships attacked U-559 that night and badly damaged it. Once it was clear that it was stopped and being abandoned, Fasson rang the cease-fire bells. Rather than wait for a whaler to be launched, he stripped off and swam over to the submarine. He was joined by a young seaman, Colin Grazier, and a fifteen-year-old civilian canteen assistant, Tommy Brown, who had lied about his age to get aboard ship.
Once on board the badly listing submarine, Fasson and Grazier managed to pass documents up to Brown who in turned handed them to shipmates who had pulled up alongside in the whaler. After three bundles of documents were passed up to him Brown shouted at his colleagues to come up. They had just started on their way when the submarine sank with Fasson and Grazier on board. Brown managed to jump off and was picked up by the whaler. Three weeks later, the documents rescued by Fasson and Grazier reached BP, including the current edition of the Short Signal Book and the second edition of the Short Weather Cipher. Their heroic efforts had helped to turn the tide in the Battle of the Atlantic in favour of the Allies.
Hugh Alexander had originally joined Hut 6 but was then moved to the newly formed Hut 8 to act as Turing’s deputy. He eventually became head of Hut 8 when Turing moved on to other projects and remained in the position for most of the war. Huts 8 and 4 copied much of the Hut 6/3 model, which was not surprising given that Alexander had started off in Hut 6 and was a close friend of Milner-Barry. With the vital code books in their hands, a start could be made on the messages emanating from the U-boats. In his post-war report on Hut 8 Alexander said that in November they finally started to read the U-boat traffic using weather reports and the captured Short Signal Book (from U-559).4 The German’s Navy’s obsession with weather forecasts would provide Hut 8 with cribs for the rest of the war.
The U-boats had the four-wheel machine, but for a time they used only three of the wheels for some of their messages and this enabled Hut 8 to work out the wiring of the new wheel and decrypt a small number of signals. However, four-wheel bombes were needed to read the U-boat messages on any significant scale.
Travis asked Welchman to oversee the development of a four-wheel bombe. Travis had been told, presumably by the head of MI 6, Stewart Menzies, not to give the job to BTM and Keen’s production team. The reasoning seems to have been that if they gave the job to Keen, they would be ‘putting all
of their eggs in one basket’. Travis was put in touch with Professor Patrick Blackett, a government advisor on military strategy and developing operational research, who suggested that they approach the Post Office’s research establishment at Dollis Hill in north-west London. In the late summer of 1942, one of Dollis Hill’s leading engineers, Tommy Flowers, was briefed on the problem and Charles Wynn-Williams, a brilliant physicist who was doing research at the Telecommunications Research Establishment (TRE) at Malvern was called in to help on the project.
The Dollis Hill team developed an attachment to the standard three-wheel bombe being manufactured by BTM. The device was called the Cobra because of the snake-like form of its multi-wire connections leading to the bombe (approximately 2,000 wires). It provided the extra-fast readout required for the four-wheel bombe, and used electronics for the first time to provide the twenty-six-fold increase in tests needed because of the fourth wheel. It used gas-filled thyratron valves as electrical switches as well as radio-type vacuum valves and had no moving mechanical parts. Its valves performed the same functions as relays but much faster. Harry Fensom, one of Flowers’s key assistants, helped install the first Cobra at Stanmore in the spring of 1943. Its cable apparatus was built by an engineering firm at Dursley near Stroud in Gloucestershire and the valve attachment at Dollis Hill. The Cobra proved to be noisy and unreliable and was not a success.
As the pressure grew on Hut 8, Welchman had a series of meetings with Frank Birch, starting on 7 December 1942 and continuing until 5 April 1943. Birch wanted assurances that everything possible was being done. As late as 29 March 1943 Welchman was at Dollis Hill talking to Flowers and the Director of the Dollis Hill Research Station, W. G. Radley, about the ‘four-wheel problem’. All concerned had been instructed not to tell Keen about this further development of his bombe. When he found out, he refused to have anything more to do with bombe-related developments. Faced with what he described as ‘the monstrosity designed by Flowers’, Welchman needed Keen back on board. He and Fletcher went to see him and eventually Welchman managed to talk him around.5 Keen added a row of super-slow drums, using available parts and then drove the whole assembly a lot faster using valves instead of relays for the sensing. This was an extremely simple use of valves compared with Flowers’s design. Neither Flowers nor Wynn-Williams would have known, as Keen did, that the fast drums of the three-wheelers could be driven a lot faster and still provide good electrical contacts. Keen’s relays operated in one millisecond and were copied from a Siemens design. BTM’s development of its four-wheel bombe did not have a significant effect on their existing three-wheel bombe production as the basic three-wheel design was retained. Four-wheel machines were installed at Stanmore and Eastcote from early 1943 and a total of fifty-nine were built. Fletcher estimated that the delay caused by Keen not being briefed initially had been from four to six months.