Gordon Welchman Read online

  Remarkably, the Royal Air Force experienced a very similar and equally embarrassing incident. On the evening of 6 May 1944, a Lancaster bomber, was shot down over southern France and the crew bailed out.15 Amongst them was Air Commodore Ronald Ivelaw-Chapman, who had commanded the Bomber Command base at Elshaw Wolds in Lincolnshire. Like Vanaman, he wanted first-hand experience of action even though he was ‘Ultra cleared’. He was captured by the Gestapo on 8 June, the most senior Bomber Command officer to have been captured by the Germans. Fortunately, the Germans did not realize his importance and he was treated as an ordinary prisoner of war. Unlike Vanaman, Ivelaw-Chapman was promoted to air vice marshal after the war.

  GW as a young boy in Bristol.

  GW while a student at Marlborough College.

  GW’s father William Welchman, Archdeacon of Bristol.

  GW’s brother Eric, one of the first British officers to be killed in WWI.

  GW playing hockey for Cambridge Wanderers around 1930.

  GW led a Cambridge University expedition to Spitzbergen, Norway, in 1932.

  GW with his father and sister Enid at his mother’s funeral in 1938.

  GW’s marriage to his first wife, Katharine Hodgson, in 1937. They had three children.

  Bletchley Park, home to the Leon family from 1882 to 1937.

  Alastair Denniston, first Operational Director of GC&CS. He made several trips to the USA during the war. GW felt he never received the credit he was due.

  Edward Travis, who took over from Denniston in early 1942. GW admired his organizational skills. The photo was taken in Pat Bayly’s New York apartment in 1944.

  The standard three-wheel Enigma machine, as used by the German Army and Air Force.

  Inside the Enigma machine. It is believed that only the German Navy used letters A-Z on the wheels rather than numbers 1–26. Note the reflector on the left marked as ‘B’. Reflectors with other wirings were available.

  Dispatch riders who delivered Enigma messages to BP from the intercept stations.

  General Heinz Guderian in his command vehicle in 1940 during the battle for France. Note his Enigma machine and three operators.

  A rare wartime photo of the original wooden Hut 6 taken from the Mansion. Note the bomb blast walls around the huts.

  Hut 6 Machine Room in Block D.

  Hut 6 structure chart, circa 1942.

  Hut 6 Control Room in Block D. It provided the interface between Hut 6 and the intercept stations.

  Boxes holding some of the 2 million Hollerith punch cards processed weekly in Block C.

  ‘Cobra’ high speed four-wheel bombe attachments later described by GW as ‘the monstrosity designed by Flowers’.

  Bombes under construction at BTM’s Letchworth factory.

  The British Typex cipher machine, modified to replicate an Enigma machine. They were produced in large numbers for Hut 6’s Decoding Room.

  Rare photograph of a prototype of the RM-26, designed by Welchman as a replacement for the Typex.

  One of 59 operational four-wheel bombes built by BTM.

  Leaders of Allied signal intelligence work meet at the secret J.A.C. Conference on 13 March 1944 at Arlington Hall, Virginia, headquarters of the US Army’s Signal Intelligence Service (SIS) during the Second World War. Notable attendees include John Tiltman (1), Edward Travis (2), William Friedman (3), Pat Bayly (4), Philip Lewis (5), Solomon Kullback (6), Joe Hooper (7), Joseph Wenger (8), Abe Sinkov (9) and Tony Kendrick (10).

  GW while manager of Ferranti’s New York subsidiary, Ferranti Electric Inc. circa 1957.

  GW with his second wife, Fannie Hillsmith, a prominent New York artist, in 1969.

  GW while visiting his daughter Ros in Paris in 1972.

  GW with his third wife Teeny and her son Tom in 1972.

  Bletchley Park in the early 1970s.

  Diana Lucy, a former intercept operator at Chicksands revisits the room in Welchman’s former Newburyport house where she first learned of his wartime exploits, and he of hers.

  GW with MITRE colleagues Bobbie Statkus and Bob Coltman following publication of The Hut Six Story.

  GW with former BP colleague William Bundy at the 1982 convention of the American Cryptogram Association.

  GW on holiday in the Virgin Islands at the end of March 1985. He learned of his terminal illness shortly after returning home.

  GW’s son Nick at lunch with the author in September 2011 and delighted to find a ‘bomba’ on the dessert menu.

  Chapter 7

  Bletchley Park: The Last Two Years

  When Edward Travis replaced Alastair Denniston as the Operational Director of Bletchley Park in February 1942, he increasingly turned to Welchman to assist him in a range of activities not directly related to the work of Hut 6. In September 1943 Welchman was appointed head of the Machine Co-ordination and Development Section and his role in Hut 6 came to an end. At his recommendation, Milner-Barry took over as the Head of Hut 6 and Manisty took over the Watch. It was a tough act for Milner-Barry to follow.

  By the end of 1943, BP had moved from an operation whose infrastructure included the movement of documents between Huts 6 and 3 in a wooden tray on wheels pulled by string, to the pneumatic tube and conveyor belt production line of Block D. Staff numbers in Hut 6 had grown to 401 and they occupied 29,230 of the 40,000 sq ft of the available office space in their new building.

  Welchman was now based in Room 10 in the newly opened Block F and getting to grips with his new role. He wrote to Travis on 2 December and categorized the areas he believed he would now be working in: assistance to cryptographers, security of machine ciphers, development of cipher machines, communications, interception, after the war, policy and plans.1

  In March 1944 Travis restructured his senior management team.2 Three Deputy Directors would be responsible for the three main functions of the establishment, designated DD1, DD2, and DD3. Nigel de Grey (DD1) became Travis’s chief deputy and took on responsibility for security, and the production and distribution of intelligence. Alan Bradshaw (DD2) took on responsibility for planning and administration matters, while Edward Hastings (DD3) took on liaison with Allies and the research sections, comprising Intelligence Service Oliver Strachey (ISOS) and Intelligence Service Knox (ISK). The different services at BP were also under the command of a Deputy Director: Tiltman – Military, Birch – Naval, Cooper – Air and Wilson – Cipher Security. Finally, four Assistant Directors were appointed: Welchman – Machines and Mechanical Devices, Jones – Hut 3, Elsdale – Communications and Interception and Page – ISOS & ISK.

  One of Welchman’s functions in his new role was to keep an eye on the security aspects of new technological developments in British communications. He quickly found himself concerned about trends in military thinking on communications which were beginning to appear. He continued to work closely with Keen and BTM,3 Flowers and the Post Office Research Station at Dollis Hill and Wynn-Williams at the Telecommunications Research Establishment in Malvern. Welchman was also responsible for ensuring that BP had a sufficient stock of replica Enigma machines for the decoding rooms in Huts 6 and 8.

  In early 1944, Katharine was pregnant with their third child, Rosamond, and left the ATS. The Welchmans were at last able to buy the home that they had sought for a number of years. Rippington Manor was a fine Elizabethan house in the village of Great Gransden in Bedfordshire. The original house was a cell of the monastery at Repton, which was then called Rippington. The house the Welchmans acquired was built towards the end of the sixteenth century, during the reign of Queen Elizabeth I. They had arranged to purchase it some years before and, once they took possession, they remained there until the end of the war. Great Gransden was a remote place, not easily visited by people who lacked a motor car and the petrol to make it go. Two country buses ran from the village, one to Cambridge, the other to St Neots in Huntingdonshire, a small county that no longer exists. The village was also fairly close to the Cambridge/Bletchley rail line and Welchman had fond memories in la
ter life of cycling to and from the village of Gamlingay, the location of the nearest station. Despite its remoteness, the Welchmans were able to entertain friends and colleagues from BP and Cambridge. On several occasions, Welchman was able to take his children to the rooms he had lived in at Sidney Sussex College and for a punt on the river.

  Part of the original land belonging to their house had been requisitioned by the government and used as a camp for prisoners of war. Initially, Italian prisoners were housed there and Welchman’s son Nick befriended some of them. Eventually the Italians were relocated and replaced by German prisoners, who were segregated from the locals. The Canadian Air Force had a base near Great Gransden, and the village children were invited to a splendid Christmas party in 1944. Like many families living in the countryside, the Welchmans took in a boy who had been evacuated from London.

  In 1943, another major activity was rapidly growing in complexity at BP. The German high command had commissioned a new encryption system which was more suited to very long strategic-level communications. This system did not use Morse code to transmit messages but instead was based on the international teleprinter system. This had a thirty-two character alphabet and was an online system. The operator would type his message on a teleprinter attachment to the encryption machine which would add a randomly generated character to each character of the message using the properties of the teleprinter alphabet and its addition table. So, for example, if the first character of the message was ‘A’ and the encryption machine generated a ‘B’, the two would be ‘added’ together producing ‘G’. A radio transmitter was integrated into the system to send the messages off to another operator. Using the example above, if the receiving encryption machine was set up exactly the same as that of the sender, the first random character it generated would be ‘B’. It would then automatically add it to the first incoming character, ‘G’. Following the rules of the teleprinter addition table, this would produce ‘A’, the first character of the message. Each character in the message would be processed in the same way and printed on the teleprinter attached to it. A mistake by a German operator using this system on 30 August 1941 had allowed Bill Tutte, following some crucial work by John Tiltman, to reconstruct the encryption machine producing the random characters and subsequently break the system. After developing machines called Robinson to speed up the decryption process, the world’s first electronic computer, Colossus, was designed by Tommy Flowers and his colleagues at the Post Office’s research laboratory at Dollis Hill in London. Colossus speeded up the decryption process to an acceptable level for intelligence purposes. While not directly involved in the day to day operation of this work, Welchman did have overall responsibility for it as the work of BP’s Newmanry and Testery sections that did take on this system was part of his remit.4

  His assistant, Houston Wallace, remembered accompanying Welchman to one of the intercept facilities to address operational problems:

  You and I went to Denmark Hill together to see what could be done to reduce the error rate in the tapes that reached Newman’s people at BP. In those days, we had no equipment for comparing two tapes. I believe we decided that, to get error-free operation, it was necessary to punch three separate tapes from each oscillograph trace (undulator tape), and to compare them visually by laying them on top of each other on a long table.5

  In his new role, Welchman established a committee to discuss desirable improvements to BP’s various technological capabilities. Its members were Alexander, Turing, Freeborn, Gerry Morgan, Head of Research in the Military Section, and Max Newman, head of the section responsible for the Colossus machines.

  On 17 February 1944 Welchman embarked on a month-long visit to the USA in his role as Assistant Director for Mechanization. His first task was to work closely with a Canadian colonel, Benjamin Deforest (Pat) Bayly.6 Bayly, was a communications expert who had developed a new encryption machine based on teleprinter traffic which was christened the Rockex. Travis wanted Welchman to work with Bayly on improving the security of this machine as well as the whole problem of government communications. He also wanted the two of them to carry on the British–American technological liaison on Enigma and other problems that had already been initiated.

  One project which emerged from this liaison was the development of a replacement for the Typex machine. The new machine needed to be highly secure, easy to operate and deployable in forward areas. Welchman, now a cryptanalyst-turned-cryptographer, worked with Bayly, Turing and Alexander to design the RM-26. It was to house 24 × 26-point rotors, each containing the same scrambled wiring. The 24 rotors were to be in three banks of eight, with complicated interdependent motion. Unfortunately, it never progressed beyond the prototype stage due to the increasing necessity for a machine compatible with the new generation of American cipher devices.

  Bayly was originally from Moose Jaw, Saskatchewan, and had become a professor at the University of Toronto. At the height of the war, he had been hired by William Stephenson, the senior representative of British intelligence for the entire western hemisphere. Stephenson had been sent to the United States on 21 June 1940 covertly to open and run British Security Co-ordination (BSC) in New York City. The BSC office, headquartered in room 3603 in the Rockefeller Center, became an umbrella organization that by the end of the war, represented the British intelligence agencies MI 5, MI 6 (SIS or Secret Intelligence Service), SOE (Special Operations Executive) and PWE (Political Warfare Executive) throughout North America, South America and the Caribbean.

  Bayly had first visited BP near the end of Denniston’s tenure as its operational head. Sir Stewart Menzies had instructed Richard Gambier-Parry to escort him to BP and introduce him to Denniston. Gambier-Parry had been given the task at the beginning of the war of leading MI 6 Section VIII (SIS Communications), the section devoted to all forms of secret communication. He went on to orchestrate its complete overhaul and modernization. Menzies had forgotten to tell Denniston of Bayly’s visit, which was unfortunate, because, according to Bayly, it was clear that Denniston ‘loathed Gambier’s guts’. Bayly left them to their inevitable row, retiring to the guardhouse until it was all over. Denniston met him a year later in London and was most apologetic about their previous meeting.

  On arrival, Bayly’s first request was to be taken to the teleprinter room where he promptly sent a message to his office to ask his wife Bun, to be there in thirty minutes or so. Shortly after, Bayly and his wife were having an asynchronous conversation via the teleprinter link. While operators on security networks frequently used ‘chat’ as a way of testing their communications, the ultimate purpose was always to send a secure and/or receive a secure message. Here, Welchman witnessed Mr and Mrs Bayly, years ahead of their time, using the same technology for normal family conversation!

  Bayly recalled his first meeting with the legendary American cryptographer William Friedman, in a series of letters to Welchman in 1983. It gives a good insight into the subsequent work which would occupy not only much of Welchman’s and Bayly’s time for the remainder of the war, but also that of Alan Turing:

  My first meeting with Billy Friedman was interesting. Bill Stephenson was faced with the problem of being in New York and having to talk to Washington about rather delicate matters. This was just after Pearl Harbor. I was supposedly doing purchasing with side tones of setting up communications in the Western Hemisphere where they might become necessary, so Bill asked my opinion on phone scramblers. I told him that as far as I had been able to find out, that they were a joke. His answer was typical, ‘Find me something that isn’t.’ The Journal of the A.I.E.E. Feb 1926 pp 109–115 had an article by G. S. Vernam of what later was to become Bell Laboratories, on the subject of Cipher Printing Telegraphs. This came out at the time that I was installing a network of Teleprinters for the Canadian Pacific Railway, and had aroused my interest. I couldn’t find Vernam but it mentioned Capt W. F. Friedman, cryptanalyst of the U.S. Signals Corps. I persuaded ‘Wild Bill’ Donovan of O.S.S. to find him
for me and I visited him at his home in Washington. He was naturally not very forthcoming, but told me that Western Union had developed such a machine called a Telekrypton. We found that they had made two of these machines and had stored them as there had been no commercial acceptance. We bought them and rapidly found the troubles. A small cipher tape of about 200 characters, dependence on synchronization at both ends and a hopeless method of starting such ‘togetherness’. As their circuits had an average of a hit every ten minutes this meant a communications speed of zero and a security rating of slightly less! A hectic weekend with Hoover of Western Union led to a table model, continuous code tape, perforated tape at the receiving end so the operator could set the code tape at the proper point. The machine was an instant success on the grounds that it was all we had. The code tape was made by putting two girls on hand perforators and telling them to punch complete random nonsense. It soon turned out that once a girl had worked to produce nonsense for a couple of days she unwittingly started a cycle of up to 200 characters that occurred over and over. I then went to shaking steel and glass balls and then feeding them into five tubes with contactors. This gave me a very high grade randomness but unmerciful wear and tear. The final solution was a very high frequency oscillator of poor stability beating against a low frequency oscillator also of low stability. I judged randomness by the ratio of the tape holes to blanks. It worked very well.