Bletchley Park’s Sturgeon—The Fish That Laid No Eggs

This chapter describes the Siemens & Halske T52 cipher machines and explains how Bletchley Park broke them. (See photograph 51.) Many authors have confused the T52 with the Tunny machine, and have erroneously linked the T52 to Colossus. The German armed forces employed three different types of teleprinter cipher machines during the Second World War: the Lorenz SZ40/42a/42b (Tunny), the Siemens & Halske Schlüs-selfernschreibmaschine (SFM—Cipher Teleprinter Machine) T52, and the one-time-tape machine T43, also manufactured by Siemens. The Siemens T52 existed in four functionally distinct models: T52a/b, T52c, T52d, and T52e (there was also the T52ca, a modified version of the T52c). At Bletchley Park all T52 models went under the code name ‘Sturgeon’. The Siemens T43 was probably the unbreakable machine that BP called ‘Thrasher’. (This came into use relatively late in the war, and appears to have been used only on a few selected links.) In 1964 Erik Boheman, the Swedish Under-Secretary of State, first revealed that Sweden had broken the T52 during the Second World War. The Swedish successes against the T52 are the topic of Chapter 26. It was only in 1984 that the British officially acknowledged that Bletchley Park had also enjoyed some success against the T52. Not only did BP intercept traffic enciphered on the T52; it also broke all the different models that it discovered. It was clear from the beginning that the T52 was a very difficult machine to break. Probably it would have remained unbroken had it not been for German security blunders in using the machines. The blame should not be put entirely on the German teleprinter operators, however: the designers of the machines at Siemens, who failed to listen to the advice of the German cryptographic experts, were also responsible. The Siemens engineers seem to have focused more on the engineering problems than on the cryptographic security of the machine. The T52a/b and the original T52c were machines with quite limited security. The T52c is an extraordinary example of how not to go about designing a cryptographic machine. The wheel-combining logic, which was meant to strengthen the machine, had exactly the opposite effect—it eased the task of breaking the machine.

How Colossus was Built and Operated—One of its Engineers Reveals its Secrets

Flowers’ team, which included me, became involved with the design of the logic units of Newman’s proposed machine after Morrell’s Telegraph Group, which had been assigned the job, got into difficulties. For modulo-2 addition (‘exclusive-or’, or XOR) Morrell was proposing to use a type of frequency-modulator employed for voice-frequency telegraph signals. This might have been all right for adding only two signals, but it was useless for adding many signals, because the device was analogue in nature (i.e. not digital or discrete, but using continuously variable voltages). The small variations added up, with the result that the device often produced a wrong answer. After some clever work by Gil Hayward, it just about worked for the number of additions that were required. The Heath Robinson’s ‘bedstead’, containing the tape drive and the photoelectric tape-reader, was designed and built at Dollis Hill. Our people Eric Speight and Arnold Lynch had very recently used photoelectric cells to do what was required. Fighter Command had asked Dollis Hill for a fast means of recording the telegraphic signals from their aircraft observers. Speight and Lynch, working together with Morrell’s group, had designed some photoelectric equipment that would record these signals directly from the telegraphic punched tape. The device they built, called the ‘Auto-Teller’, was never in fact used, but this photoelectric technology formed the basis for the bedstead. When we finished our part of Newman’s machine at Dollis Hill I moved to Bletchley Park, and Alan Bruce from TRE accompanied their part of the machine, the counter and display rack. The Heath Robinson was installed in the wooden Hut 11—the Newmanry. I was privileged to be one of those present at the Heath’s inauguration before the VIPs—and I can confirm that smoke did rise from it at switch-on. I was able to deal with this. A large resistor had overloaded, which I bypassed, and we carried on. (The machine never did catch fire, on this or any other occasion, but as mentioned in Chapter 13, we had a benzene fire in our workshop, at a much later date, and this may have contributed to the erroneous stories of Heath Robinson catching fire.)

Dollis Hill at War

The Dollis Hill building was erected in 1933 as the headquarters of the Post Office Engineering Department Research Station. Here T. H. Flowers pioneered digital electronics. The imposing brick building looks out from its hilltop site over the suburbs of North London (see photograph 41). It housed what was probably the most active telecommunications research centre in Europe. The building still stands today. Now converted into condominiums, it flanks a road named Flowers Close. Dollis Hill (DH) supplied much of the cryptanalytical machinery for Bletchley Park. Another of its roles was to provide an emergency alternative to the underground Cabinet War Rooms in Whitehall. Early in the war a secret underground citadel was excavated at DH. A massive structure of reinforced concrete, the citadel extended three floors into the ground. It is said that Churchill took against the new bunker, and the War Cabinet met at DH only once. Gil Hayward joined the Post Office Research Station in 1934. He describes the ethos of the new research laboratory: I went to DH at the age of 16, straight from school. The Research Station had existed in permanent form for less than two years, having previously been accommodated in a series of wooden huts. ‘Research is the Door to Tomorrow’ was inscribed in stone above the main entrance to the new building. The atmosphere at DH was unique. Original thinking was encouraged and there was a substantial amount of freedom. Norman Thurlow entered the Engineering Department of the Post Office as a recruit some three years before the war. In 1942, he joined the Dollis Hill group and participated in Flowers’ engineering revolution. The Post Office included the post and telephone businesses. The Engineering Department served both operations for all engineering work, including R&D. The Research Branch at Dollis Hill consisted of several different groups. Among them were the telegraph, switching, and physics groups, headed by Frank Morrell, Tom Flowers, and Eric Speight, respectively. These three groups all became involved in some way with the Bletchley Park operation. The state of the art was defined by the telephone and telegraph systems.

Mr Newman’s Section

When Turing arrived at Bletchley Park, the day after Chamberlain’s announcement of war with Germany, he joined Dilly Knox’s Research Section. The job of the Research Section was to study enemy ciphers and operating procedures, and to devise methods of attack. The techniques invented by this high-powered think-tank were then handed over to other sections where they were used operationally against the enemy traffic. In 1939, Enigma was the focus of research. By the time Tutte joined the Research Section, in mid-1941, Captain Gerry Morgan headed it. Tunny soon became the leading problem. Thanks to Tutte, the Research Section broke the Tunny machine in January 1942, and in July read up-to-date traffic for the first time. Tunny could now be tackled operationally, and a Tunny-breaking section was immediately set up under Major Ralph Tester. Several members of the Research Section moved over to the ‘Testery’. Armed with Turingery and other hand methods, the Testery read nearly every message from July to October 1942—thanks to the insecure 12-letter indicator system (see Chapter 3). In October, however, the 12-letter indicators were replaced by QEP numbers and the Testery, now completely reliant on depths, fell on leaner times (see Chapter 5). The Research Section renewed its efforts against Tunny, looking for a means of wheel setting that did not depend on depths. With the invention of Tutte’s method, Newman was given the job of developing the necessary machinery, and when the Heath Robinson was delivered in June 1943 the ‘Newmanry’ became a separate section. From December 1943 the Newmanry would be responsible for breaking and setting the chi-wheels, and the Testery for breaking and setting the remaining wheels manually. The two sections worked hand-in-glove. Initially Newman’s staff consisted of one cryptographer (Michie), two engineers, and 16 Wrens. Soon a second cryptographer arrived (Good), and after three months of experimentation, two or three messages were being set each week. ‘Cryptographers’, ‘engineers’, and ‘Wrens’ remained the principal staffing categories of the Newmanry throughout the war. By May 1945, there were 26 cryptographers, 28 engineers, and 273 Wrens. During the period June 1943 to July 1944, two Americans joined the cryptographic staff of the Newmanry.

The Colossus Rebuild

In 1991, some colleagues and I started the campaign to save Bletchley Park from demolition by property developers. At this time I was working at the Science Museum in London restoring some early British computers. I believed it would be possible to rebuild Colossus, but nobody else believed me. In 1993, I gathered together all the information available. This amounted to no more than eight 1945 wartime photographs of Colossus (some of which are printed in this book), plus brief descriptions by Flowers, Coombs, and Chandler, and—crucially—circuit diagrams which some engineers had kept, quite illegally, as engineers always do! I spent nine months poring over the wartime photographs, using a sophisticated modern CAD system on my PC to recreate machine drawings of the racks. I found that, fortunately, sufficient wartime valves were still available, as were various pieces of Post Office equipment used in the original construction. In July 1994, His Royal Highness the Duke of Kent opened the Bletchley Park Museum and inaugurated the Colossus rebuild project. At that time I had not managed to obtain any sponsorship for the project, so my wife Margaret and I decided to put our own money into it, to get it started. We both felt that if the effort was not made immediately there would be nobody still alive to help us with memories of Colossus. Over the next few years various private sponsors came to our aid and some current and retired Post Office and radio engineers formed the team that helped me in the rebuild. In 1995, the American National Security Agency was forced by application of the Freedom of Information Act to release about 5000 Second World War documents into the US National Archive. A list of these documents was put onto the Internet. When I read it I was amazed to see titles like ‘The Cryptographic Attack on FISH’. I obtained copies of these documents and found that they were invaluable reports written by American servicemen seconded to Bletchley Park when America entered the war. I was also fortunate enough to be given access to the then still classified General Report on Tunny (parts of which are published for the first time in this book).

Colossus, Codebreaking, and the Digital Age

The paths that took men and women from their ordinary lives and deposited them on the doorstep of the odd profession of cryptanalysis were always tortuous, accidental, and unpredictable. The full story of the Colossus, the pioneering electronic device developed by the Government Code and Cypher School (GC & CS) to break German teleprinter ciphers in the Second World War, is fundamentally a story of several of these accidental paths converging at a remarkable moment in the history of electronics—and of the wartime urgency that set these men and women on these odd paths. Were it not for the wartime necessity of codebreaking, and were it not for particular statistical and logical properties of the teleprinter ciphers that were so eminently suited to electronic analysis, the history of computing might have taken a very different course. The fact that Britain’s codebreakers cracked the high-level teleprinter ciphers of the German Army and Luftwaffe high command during the Second World War has been public knowledge since the 1970s. But the recent declassification of new documents about Colossus and the teleprinter ciphers, and the willingness of key participants to discuss their roles more fully, has laid bare as never before the technical challenges they faced—not to mention the intense pressures, the false steps, and the extraordinary risks and leaps of faith along the way. It has also clarified the true role that the Colossus machines played in the advent of the digital age. Though they were neither general-purpose nor stored-program computers themselves, the Colossi sparked the imaginations of many scientists, among them Alan Turing and Max Newman, who would go on to help launch the post-war revolution that ushered in the age of the digital, general-purpose, stored-program electronic computer. Yet the story of Colossus really begins not with electronics at all, but with codebreaking; and to understand how and why the Colossi were developed and to properly place their capabilities in historical context, it is necessary to understand the problem they were built to solve, and the people who were given the job of solving it.

Living with Fish: Breaking Tunny in the Newmanry and the Testery

I should begin by explaining how I happened to find myself, on 12 January 1942, at the age of 18, awaiting permission to enter the gates of Bletchley Park, to undertake work, on behalf of the British Foreign Office, of whose nature I had essentially no knowledge. In October 1941, four very distinguished members of the Bletchley Park team (Alan Turing, Hugh Alexander, Stuart Milner-Barry, and Gordon Welchman) had written a letter to Churchill, drawing his attention to the importance of the work being done at BP on deciphering Enigma, and, therefore, the urgency of recruiting appropriately trained people (principally mathematicians), and of making funds available for building more of the high-speed Bombes needed to expedite the decoding process. Churchill, to his great credit, did not react like a bureaucrat appalled that the writers of the letter had not gone through the proper channels; he immediately saw the importance and good sense of the letter and minuted it ‘Action this day’ to his chief of staff, General Ismay. The result was the empanelling of an interviewing team which toured the universities looking for mathematicians with a knowledge of modern European languages. Such people, however, were not easy to find, since the British higher education system of the time, based as it was on the principle of premature specialisation, virtually guaranteed that no such people would exist. (Ironically there were many German Jewish refugee mathematicians in Britain at that time, but they were ‘enemy aliens’ and so not to be trusted!) The team came to Oxford and I was, I believe, the only person presenting himself for interview. I was not a mathematician, merely a second-year undergraduate specialising in mathematics, and my knowledge of German was very rudimentary, acquired by self-study. But the interviewing team snapped me up, and offered me a position in the Foreign Office, to carry out certain entirely unspecified duties, provided I was willing to start in January 1942. I suspect the interviewing team did not themselves know the nature of the work I would be doing.

The German Tunny Machine

The Enigma cipher machine was slow and cumbersome to use. Sending a message was a complicated procedure requiring the participation of several operators (see photograph 24). The process started with the German plain-language, known as the ‘clear’ or the ‘plaintext’. Encrypting this produced the ‘ciphertext’. Typically, the plaintext or clear consisted of ordinary German words mixed with military abbreviations and jargon (such as WEWA for Wetter Warte, meaning ‘weather station’, and BINE, literally ‘bee’, meaning ‘very very urgent’). A cipher clerk typed the plaintext at the keyboard of an Enigma machine (see the diagram on page 17). Each time the clerk pressed a key, a letter on the lampboard would light. For example, typing HITLER might produce the letters FLKPIM. As the letters of the ciphertext appeared one by one at the lampboard, they were painstakingly noted down by an assistant. Various items of information were then added to the ciphertext, including the intended recipient’s radio call-sign, and a radio operator transmitted the complete message in Morse code. At the receiving end, the process had to be carried out in reverse. The radio operator turned the dit-dit-dahs of the Morse transmission back into letters of ciphertext and handed the result to the cipher clerk. The clerk typed the ciphertext at the keyboard of an Enigma, which had been set up identically to the sender’s machine. The letters of the plaintext lit up at the lampboard one by one and were recorded by the assistant. The Tunny system was much more sophisticated. The process of sending and receiving a message was largely automated. Encryption and decryption were entirely automatic. The transmitted ciphertext was never even seen by the German operators. At the sending end, a single operator typed plaintext at the keyboard of a teleprinter. At the receiving end, the plaintext was printed out automatically by another teleprinter. (A teleprinter is called a teletypewriter in the US.) The sender could switch his teleprinter equipment from ‘hand mode’ to ‘auto mode’. In auto mode, a pre-punched paper tape was fed into the equipment. The plaintext punched on the tape was encrypted and transmitted at high speed.

How It Began: Bletchley Park Goes to War

The breaking of the German teleprinter cipher that led to the construction of the Colossus computer was the culmination of a series of triumphs for British codebreakers. British interception of other countries’ radio communications had begun in earnest during the First World War. The War Office ‘censored’ diplomatic communications passing through the hands of the international telegraph companies, setting up a codebreaking operation to decipher the secret messages. The British Army intercepted German military wireless communications with a great deal of success. E. W. B. Gill, one of the army officers involved in decoding the messages, recalled that ‘the orderly Teutonic mind was especially suited for devising schemes which any child could unravel’. One of the most notable successes for the British cryptanalysts came in December 1916 when the commander of the German Middle-East signals operation sent a drunken message to all his operators wishing them a Merry Christmas. With little other activity taking place over the Christmas period, the same isolated and clearly identical message was sent out in six different codes, only one of which, until this point, the British had managed to break. The army codebreaking operation became known as MI1b and was commanded by Major Malcolm Hay, a noted historian and eminent academic. It enjoyed a somewhat fractious relationship with its junior counterpart in the Admiralty, formally the Naval Intelligence Department 25 (NID25) but much better known as Room 40, after the office in the Old Admiralty Buildings in Whitehall that it occupied. The navy codebreaking organisation had an even more successful war than MI1b, recruiting a number of the future employees of Britain’s Second World War codebreaking centre at Bletchley Park, including Dillwyn ‘Dilly’ Knox, Frank Birch, Nigel de Grey, and Alastair Denniston, who by the end of the war was head of Room 40. Among the many successes of the Royal Navy codebreakers was the breaking of the Zimmermann telegram, which showed that Germany had asked Mexico to join an alliance against the United States, offering Mexico’s ‘lost territory’ in Texas, New Mexico, and Arizona in return, and brought the United States into the war.

The British Tunny Machine

Early in 1944 I returned to the UK from top-secret work in the Middle East. Two days after my arrival I received instructions to report to Tommy Flowers at Dollis Hill. I had joined DH in 1934 at the age of 16, straight from school, and had left in 1940 to carry out intelligence work overseas. Flowers introduced me to my new colleagues, Doc Coombs, Bill Chandler, and Sid Broadhurst, the last of whom I had met in 1938, during a course of training for the rank of probationary inspector—I had enjoyed his lectures on automatic telephony. The introductions over, an awkward silence fell. Here was an army captain in the intelligence corps who knew nothing about their project and who was still being vetted by the security services. This would preclude their discussing anything of a secret nature in my presence, probably for another two weeks, until my security clearance came through. On the third day of this ridiculous state of affairs, Broadhurst could stand it no longer. After lunch he said, to no one in particular, ‘Let’s tell him.’ The others agreed, and in less than an hour I had a fairly detailed outline of what our project was. By the end of the afternoon I was deeply immersed in the design of the wiring and layout of the rotary switches that would simulate the 12 wheels of the German Tunny machine. Broadhurst saved two precious weeks by taking the bull by the horns as he did. As it was, it was a near-run thing to get the equipment in operation by D-day. Our Tunny would be deciphering the encrypted teleprinter traffic after the cryptanalysts had determined the wheel patterns and wheel settings. The tedious hand-work required to produce the decrypts, once the settings were known, had not been able to keep pace once Colossus went into operation. This situation called for a copy of the Lorenz machine to produce decrypts. The Lorenz, one of which I was able to examine after the end of hostilities, was a beautifully made piece of mechanism, but it lacked the flexibility that our electromechanical copy possessed.