Gordon Welchman Read online

  Eventually, Sixta (short for Hut 6 Traffic Analysis) was created in November 1943 and incorporated the Central Party and SLP. It was independent of Hut 6 and Hut 3 and had overall responsibility for traffic analysis of German Army and Air Force networks. The development of W/T I and subsequently Sixta, according to Milner-Barry, ‘was down to Welchman-Colman, Welchman-Blair-Conyngham [sic], Welchman-Lewis.’ Many of Welchman’s colleagues had thought that traffic analysis was a fad but Milner-Barry, effusive as ever in his admiration for Welchman’s contribution in this area, went on to say:

  Again it was Welchman with his strange and uncanny knack of grasping the ultimate significance of things who fought throughout for the recognition of the importance of W.T.I.

  Welchman’s prescience was to be brilliantly justified after his departure, when the knocking down of the various props which had made identification almost a rule of thumb matter – first discriminants, then changing frequencies and finally encoding of call signs – made the cryptographer largely dependent on the complete and accurate knowledge of the German W/T organization which Sixta had steadily been building up. Then what had been from the cryptographer’s standpoint a luxury became a basic necessity of life and, if Welchman had not fought for its development in the early years, we should have no hope of weathering the storms which nearly overwhelmed us in the last eighteen months.

  Battles were successfully fought to provide us with enough tools to do the job at the breaking as well as at the intercepting end. That again under the Director, was almost Welchman’s achievement, not only when he was Head of Hut 6 but when he was translated to a higher sphere.

  Chapter 6

  The Americans

  In late April 1943 a car drove through the gates of BP with three American visitors inside. One of the three, a dapper man with a tiny moustache, was William Friedman, the doyen of American cryptography. Friedman’s visit to BP was intended to cement further the relationship between his SIS operation and BP.1 Its Operational Director, Alastair Denniston had visited Friedman in 1941 and they had become good friends.2

  The US Army had created a cryptographic service during the First World War, designated MI-8. At the end of the war, it was felt to be un-American to snoop on other countries in peacetime so a successor organization to MI-8 was set up in New York and known by those involved as the American Black Chamber. It attacked the codes of many countries throughout the 1920s under the leadership of Herbert Yardley.3 The US Navy also had established a Communications Intelligence Organization in 1924. The American cryptographic groups were under threat from the American Secretary of State, Henry L. Stimson who, after being appointed in 1929, famously said ‘Gentlemen do not read each other’s mail.’ Nonetheless, on 19 July 1929, three Army officers and one civilian met to plan a new cryptographic service, the Signals Intelligence Service (SIS). The civilian present was William Friedman whose official role had been to create codes and ciphers for the US Army.4 Once the Army agreed that he should take up the position of Director of the SIS, he had set about recruiting young men who combined mathematical expertise with linguistic skills. What followed was remarkably similar to GC&CS’s recruitment drive for ‘men of the professor type’. Friedman trawled the listings of American Civil Service examination results looking for mathematicians who had top scores and a good knowledge of a foreign language. By 1939 the SIS had received a considerable increase in funding and by early December 1941 its numbers had swelled to 181 officers, enlisted men and civilians in Washington, and 150 in the field.

  In early February 1941, four passengers landed from HMS King George V after it had anchored in Scapa Flow, the Home Fleet’s remote base in the Orkneys. The ship was returning from its maiden voyage to the USA. The four passengers were greeted by John Tiltman, who was to be their host during their stay in Britain. The four had with them a package containing gifts and were on their way to Bletchley Park, the first Americans to enter BP. The American party were Abraham Sinkov and Leo Rosen from the US Army’s Signals Intelligence Service and two young US Navy communications officers, Robert W. Weeks and Prescott H. Currier.5 The gifts they had brought with them were replicas of the Japanese encryption machine that the Americans had named Purple. Japan had developed the machine to encrypt diplomatic traffic and, like the Enigma equipment used by the Germans, the Purple machine was also used to decrypt messages. The development of the American Purple replica was arguably an equivalent feat to the early Polish work in reconstructing the Enigma machine. It could even be compared to the work of the brilliant mathematician Bill Tutte in his later reconstruction of the Lorenz SZ 42 machine at BP (see Chapter 7).

  The breaking of the Japanese diplomatic traffic had been the crowning achievement of Friedman’s SIS and in particular the brainchild of one of his first recruits, Frank Rowlett. In the mid-1930s, the SIS team had been ordered to make decryption of Japanese diplomatic messages its top priority. The messages were being encrypted on a machine based on code wheels which the Japanese called Cipher Machine Type A. Again, paralleling Welchman’s use of colours to identify German communication network keys, the Type A became the Red machine. This was broken by Rowlett and his colleague Solomon Kullback, paralleling Rejewski’s early work on Enigma. It was a hugely significant development because the system would eventually be used to send diplomatic traffic between Japan and Germany. Hundreds of diplomatic messages were decrypted. In this case the US Army and Navy co-operated and pooled resources to deal with the huge volume of traffic. In late 1938, much like Enigma decrypts had warned BP of changes being made by the German cryptographers, a Red intercept revealed that the Red machine would be replaced in major embassies by the Type B cipher machine which was given the codename Purple. The new system was more complex than Red and the Navy cryptanalysts soon gave up trying to break it, Rowlett and Leo Rosen discovered that the new system was based on telephone switches rather than wheels. It took the SIS team about one year to design and build its own Purple replica. They decided to demonstrate it to the Director of the Navy Department’s Communications Intelligence Unit, Laurence Safford. They wanted approval to share their work with their naval intelligence counterparts. Following the demonstration, Safford assumed that they had stolen the machine and was astonished to learn that they had actually built it. General Mauborgne, the Chief Signals Officer, was also impressed and took to referring to his cryptanalysts as ‘magicians’. This led to the Americans referring to intelligence produced by cryptanalysis as ‘Magic’, much in the same way as BP’s output was known as ‘Ultra’.

  The Americans never achieved the complete inter-service co-operation, collaboration and communication which Welchman believed made BP unique. Evidence of the American failure in this regard is provided by events leading up to the infamous attack on Pearl Harbor by the Japanese Navy on 7 December 1941.6 Some historians have claimed that Churchill knew well in advance about the devastating air raid on Coventry on 14 November 1940 and that warnings were not issued because it might reveal the Ultra secret to the Germans. This has been categorically disproven by wartime records.7 Some authors have made the same claim about the Americans and Pearl Harbor. However, the American situation was different and more a result of inter-service rivalry.

  In the early hours of 7 December 1941, the Navy intercepted the fourteenth part of a long Japanese message to their Washington Embassy which said:

  The Japanese Government regrets to have to notify hereby the American Government that in view of the attitude of the American Government it cannot but consider that it is impossible to reach an agreement through further negotiations.

  Two more messages were subsequently decrypted. The first instructed the Japanese Embassy to destroy its cipher equipment, all machine codes and all secret documents. The second was very unusual and said:

  Will the ambassador please submit to the United States Government (if possible to the Secretary of State) our reply to the United States at 1:00 p.m. on the 7th, your time.

  This material was shown t
o General George Marshall, the US Army’s Chief of Staff. It was agreed that US outposts should receive fresh warnings about a possible attack as soon as possible. Unfortunately, he was not told that due to atmospheric problems, the Army’s communication facility was having trouble contacting Hawaii by radio that morning. The Navy had a more powerful transmitter but red tape and inter-service rivalry meant that the notion of the Army giving the Navy a message for transmission was out of the question. Instead, it was sent via Western Union with no greater priority attached to it than any other telegram being sent from one person to another. A messenger was attempting to deliver the telegram to Army headquarters at Fort Shafter in Hawaii as Japanese bombs and bullets were destroying much of the American naval fleet.

  Following the attack on Pearl Harbor, the scale of the whole SIS operation was expanded, much as BP had moved from the days of The Cottage to a hut-based production-line approach. In June 1942, the SIS was given total responsibility for handling Japanese diplomatic traffic. The Navy’s cryptographic section was now free to concentrate on Japanese naval codes and ciphers. At the same time, the Army decided to set up a signal intelligence school and the SIS was tasked with taking charge of it. Larger accommodation was needed and, again, with echoes of BP, a posh girl’s school called Arlington Hall, three miles from Arlington, Virginia, was purchased and became Arlington Hall Station. The Navy was also faced with a shortage of accommodation and soon secured another girl’s school in north-west Washington, DC, called Mount Vernon Academy. The facility became known as Naval Communications Annex and would house the main US Navy cryptographic section, known as OP-20-G. By December 1942 this section had grown to 1,188 staff.

  Denniston had visited Washington in 1941 after the first American visit to BP and established what would become a lifetime friendship with Friedman. He also laid the groundwork for possible future technological collaboration between the two countries.8 After he was replaced as Director of GC&CS, Denniston’s new role was as head of a 200-strong operation based in London and responsible for diplomatic and commercial signals intelligence. In 1943 Friedman was able to spend some time in London with Denniston before travelling on to BP. In a future role at BP, Welchman would be responsible for the ongoing liaison between BP and the American cryptographic stations Arlington Hall and Mount Vernon. He would become friends with one of OP-20-G’s top technical experts and his American counterpart, Howard Engstrom, and their paths would cross again in peacetime.

  The urgent need, and doubts about the British engineering workload, had already prompted the Americans to start investigating designs for a bombe, based on the full blueprints and wiring diagrams received by US Navy Lieutenants Robert Ely and Joseph Eachus at BP in July 1942. Throughout the summer of 1942, engineers of OP-20-G hoped for breakthroughs in technology which would allow them to develop a high-speed electronic bombe. For this work to succeed, they needed the advice of the more experienced British engineers. In August, OP-20-G informed GC&CS that it intended to build its own machine. Joseph Wenger, OP-20-G’s operational head, had been given all of the resources that he needed for such a project. Due to its urgency, the project was also free from the usual bureaucracy of the US Navy’s procurement arm, the Bureau of Ships. The electrical-research laboratory of the National Cash Register Corporation (NCR) based in Dayton, Ohio, was taken over by OP-20-G as a research centre and a possible production site for the machine. The project would be led by the laboratory’s head, Joseph (Joe) Desch.

  Funding for a full $2 million Navy development effort was requested on 3 September 1942 and approved the following day. Edward Travis and Frank Birch, head of BP’s Naval Section, travelled to Washington later that month. On 2 October 1942 they signed an agreement with Carl Frederick Holden, Director of US Naval Communications, one of the first deals to establish the special signals intelligence (sigint) relationship between the two countries. The agreement went some way to clarifying the respective roles of Britain and the USA in the European war and in naval codebreaking work.

  At the end of 1942, Alan Turing was sent to Dayton to check on the American bombe project and duly arrived on 21 December. The advice that Turing was able to offer was limited, given Desch’s apparent desire to design his own machine. However, he was able to pass on GC&CS’s experience in a number of areas which would prove invaluable to the American project. For example, the British had learned that laboratory testing alone could not anticipate all of the problems with key components of the bombe. Turing wrote a short report on his visit to Dayton9 and was impressed enough with some of Desch’s design ideas to say that ‘Starting from scratch on the design of a Bombe, this method is about as good as our own.’

  Turing was also able to show that it was not necessary to build 336 bombes, one for each possible wheel order available to the German Navy Enigma operators (who chose three wheels from eight). In 1940, Turing had developed a technique called Banburismus which significantly reduced the number of possible wheel orders that needed to be tested so the initial order was scaled down to ninety-six machines. In the end, the US Navy bombes contained sixteen four-rotor Enigma-equivalents and were much faster than their British counterparts.

  While all the American services had technical liaison people at BP, as the war developed this was proving to be inadequate. Following protracted negotiations, it was agreed that an American detachment would be integrated into BP. A strong team from the US Army arrived on 30 August 1943, led by William (Bill) Bundy10 and consisting of nine officers and ten enlisted men with the designation 6813th Signal Service Detachment. There was also the 6812th group11 that would work on the bombes and the 6811th which would be involved in radio interception. The cover story of the cryptanalysts among them was that they were messenger pigeon experts in the Signal Corps. This aroused the suspicions of an officer who was checking their identities. He was concerned that there was no record of any of them taking the Army General Classification Test. They agreed to take the test so that the paperwork would be in order. One of their number, Art Levinson, remembered the officer coming back after grading their test and saying: ‘What scores! You guys ought to be in intelligence.’12

  Bundy’s group consisted of seven cryptanalysts and two translators and, after settling into their billets, they made their way to Block D at BP. The two translators were taken to the offices occupied by Hut 3 and indoctrinated by its head, Eric Jones. Bundy and his team of cryptanalysts reported to Welchman for an introduction to the work of Hut 6. Welchman admitted years later to being slightly nervous until one of the Americans, Bill Bijur, asked if he could smoke and proceeded to produce an enormous cigar which he lit. With the ice broken, Welchman went on to amaze them with the exploits of his team. As he recalled in The Hut Six Story:

  Among my most pleasant memories of the whole war is the way those Americans came to Hut 6. By then we had moved to our new brick building, so I was able to receive them in respectable though by no means luxurious surroundings. There must have been eight or ten of them. They were ushered into my office to be briefed on our activities. I felt somewhat ill at ease as I started to tell them the Hut 6 story. But their attitude was simply that they wanted to be told what to do so that they could be helpful as soon as possible. There were no fanfares. No arguments,. No difficulties. They simply melted into Hut 6 and were liked and welcomed by everyone. Very soon each of them had found their niche and was contributing. Their leader joined our management group and before long he became a major contributor to the key-breaking activities of the Hut 6 watch.

  The success of the Americans in Hut 6 would owe much to the leadership of Bundy and the close working relationship that Welchman established with him. They remained friends for the rest of Welchman’s life. Years later Bundy wrote a paper which concluded with a point that would prove poignant, given his long association with Welchman:

  The only gap it leaves, which at least one of my wartime colleagues wishes to fill but has so far been thwarted by British Security concerns, is a truly professi
onal account of the breaking techniques. An extraordinary combination of mathematical theory and brilliant opportunism in catching constant German errors produced the first Enigma breaks in 1940; the machine techniques then developed were in a very real sense the foundations of modern electronic computer technology. This is a book that remains to be written.13

  *

  Frederick Winterbotham’s Special Liaison Units (SLU) were tasked with briefing Allied commanders in the field as Ultra intelligence was sent to them from BP. When the USA entered the war, it was agreed that American commanders would also have to be briefed, and mobile SLU units would be assigned to them. There were concerns expressed in British intelligence circles about sharing the Ultra secret with the Americans and these fears were almost justified in 1944. At the beginning of the year, the legendary James Doolittle, who had led the carrier-borne air strike on Tokyo in April 1942, had taken command of the US Eighth Air Force. In due course, Doolittle had been briefed personally about Ultra by Winterbotham. In May, a vacancy on Doolittle’s staff was filled by Arthur Vanaman, who was in turn briefed by Doolittle about Ultra. As Vanaman’s job was to select targets and ask men to risk their lives to attack them from the air, he felt he should experience action for himself. He presented his case to Doolittle who gave him the green light to go on a real mission. On 27 June 1944, Brigadier General Vanaman boarded an American bomber for a raid over eastern France. German flak soon put an end to the mission and, after successfully parachuting to safety, Vanaman was picked up by German troops. He was taken to a hospital in Frankfurt for treatment to flak wounds and then on to Berlin. By August, he was in the largest German prisoner of war camp, Stalag Luft III. Conscious of the fact that he talked in his sleep, Vanaman would put tape over his mouth before falling asleep and developed techniques for self-hypnosis in an attempt to exorcise any memory of Ultra from his mind. In the end, Vanaman was released, apparently so that he could carry special radio codes to Supreme Allied Headquarters for possible negotiations over prisoners of war. Vanaman eventually returned to the US in 1945 without having given away the Ultra secret to the enemy. Nonetheless, he was demoted to colonel, evidently for embarrassing the US Air Force in the eyes of the British and other US services. Perhaps not surprisingly, his superior Doolittle, who had approved Vanaman’s mission, was never reprimanded for the event.14