The Emperor's Codes

by Michael Smith

  By now the codebreakers had not only obtained proof that the secret protocol existed; there was a good deal of evidence in the decrypts that the Japanese and the Germans were looking to extend the pact to include Italy. An alliance of Germany in northern Europe, Japan in the Far East and Italy, whose navy was capable of dominating the Mediterranean, was a major threat to world peace.

  When Sir Eric Drummond, the British Ambassador in Rome, asked his Japanese counterpart Sujimura Yotaro if the reports in the Italian press of an imminent agreement between the two countries were true, Sujimura denied it. But the Japanese telegrams deciphered in Broadway Buildings told a different story. Nevertheless, Sir Eric was kept blissfully ignorant of the real facts for fear of compromising the codebreakers’ activities.

  A subsequent visit by Ribbentrop to Rome was played down in Rome and Berlin as being inconsequential. But again the deciphered Japanese telegrams kept the Foreign Office informed of what was really going on. ‘We have more than a shrewd idea that the object of the visit is connected with the possibility of Italy adhering to the German–Japanese anti-Communist pact,’ one official noted.

  The Foreign Office had no need of codebreakers to interpret many of the Japanese and German moves towards a grand alliance with Mussolini's Italy. Sujimura was regarded as a moderate in Japanese terms, but in an interview with the Italian newspaper Corriere della Sera, he spoke of an ‘identity of ideas’ between Japan and Italy: ‘We consider ourselves to be in the same condition. Overpopulation creates for obvious reasons the right to occupy more territory and the rights of civilization demand that people install themselves in those areas where the inhabitants stand in need of human evolution.’

  One of the codebreakers helping to decipher the Japanese diplomatic telegrams was the eccentric Hugh Foss. ‘He was a great six-foot-five Scot,’ recalled one of the codebreakers’ secretaries. ‘Very bright, very shy and a great Scottish dancer with his long legs and kilt.’ Foss's cousin Elizabeth Browning would later work alongside him in GC&CS's Japanese naval sub-section. But at this stage she was one of the Foreign Office staff handling the diplomatic intercepts. They arrived in blue-jacketed files and as a result became known as BJs, Browning recalled.

  I first knew of the existence of GC&CS in 1934. I had begun work at the Foreign Office as a shorthand-typist in January and after a few weeks became one of the two girls assigned to the American Department. One of our jobs was to file what were called the Red and Blue papers. The Reds were SIS reports and the Blues were decrypts from GC&CS which at that time was located in Broadway. These Most Secret papers had to be housed in appropriately coloured folders with a typed précis of their contents on a sticky label outside while a corresponding card went into a sort of box-file which was kept in a locked cupboard. The locked cupboards were a bit bogus. When one was moved out from the wall, because the room was being decorated, it was found to have no back to it.

  Browning had no idea that her cousin was one of the codebreakers producing the BJs.

  A colleague had herself once previously worked in Broadway and when I said to her one day that I was meeting my cousin Hugh Foss for lunch she said: ‘Oh I know him. He works there too.’ This was the first I knew of Hugh's job. During the next two to three years I saw a lot of Hugh and his wife Alison. One day they suggested that I might join the Chelsea Reel Club. They had belonged to it for some time and they said they had a friend who wanted a female partner as his wife didn't care for country dancing. The club met in Cheyne Walk not far from my base in Pimlico and it offered a cheap and amusing form of exercise (FO typist pay at that time was forty-eight shillings and sixpence, £2.42, a week). Hugh's friend turned out to be a quiet little man called Commander Denniston and for a year or so we ‘reeled’ together once a week. When I later joined GC&CS, I encountered Commander Denniston and said, ‘Hello, have you been roped in here too?’ But so tight was security that I did not discover my faux pas for some time.

  5

  PREPARING FOR WAR

  The Japanese military now began their expansion into China in earnest, protected from the Soviet Union to the north by the Anti-Comintern Pact. What was to become known as the China Incident began in July 1937 with a routine exchange of fire between Chinese and Japanese troops across the Marco Polo Bridge near Peking. The Japanese claimed one of their soldiers was missing and attempted to force their way into a small town where they alleged the man was being held – in fact he had simply wandered off looking for a woman. The Japanese force was repelled and with neither side prepared to back down for fear of losing face the incident escalated into full-scale war.

  The Japanese were oblivious to the international, and in particular the British, reaction to their behaviour in China. This included the notorious Rape of Nanking, where as many as a quarter of a million Chinese, the majority of them civilians, died. They strafed the British Ambassador's car, bombarded British enclaves in Chinese cities and, in an attack specifically designed to draw Britain into the war, shelled HMS Ladybird and HMS Bee as they lay at anchor on the Yangtze River. But with one eye on Hitler and Mussolini in the West, the British Government declined to respond with force.

  Neville Chamberlain, the British Prime Minister, believed that the only possible military solution to the Japanese problem lay in joint Anglo–US action but President Roosevelt rejected the idea and the British were not prepared to go it alone. ‘In the present state of European affairs with two dictators in a thoroughly nasty temper, we simply cannot afford to quarrel with Japan,’ Chamberlain said. ‘I very much fear therefore that after a lot of ballyhoo the Americans will somehow fade out and leave us to carry all the blame and the odium.’

  If the situation in China filled Whitehall with gloom, the code-breakers were at least able to practise on the never-ending stream of Japanese intercepts it produced. Six weeks before the Japanese troops landed, Hong Kong intercepted and decoded a long message to the Japanese 3rd Fleet providing full details of which forces were to be used and when they were to arrive. John Tiltman recalled that the plethora of military messages that followed kept him fully busy throughout the second half of 1937.

  My time was almost entirely taken up with research on the intercepts from Hong Kong. The major part of the traffic was in a succession of military systems used for transmission of intelligence reports from China, especially detailed as to the characteristics of key personalities. The small basic code-charts used at the time necessitated the spelling of most of the plain text in kana syllables and each character of a Chinese name had to be precisely described by giving the native Japanese reading for the character or, in some desperate cases, an elaborate description of the way in which a rare character was put together.

  Tiltman travelled to Hong Kong in the hope of persuading the Far East Combined Bureau to take over the military codebreaking. But the naval codebreakers had enough on their plate with the Japanese naval codes and the mission was a failure, leaving him with little choice but to continue with the Imperial Japanese Army codes and ciphers at the expense of other research.

  During this period the cipher systems used seemed to be changed quite drastically every nine months and I had a hard time keeping up with the changes. I had no Japanese interpreter attached to me and the other members of my military section were otherwise engaged. There was an underlying thread of continuity in all the systems, this being irregular switches from one substitution to another within messages. It was here that I first came across ‘bisection’, which remained the Japanese practice from that time until the end of World War Two. This is the practice of dividing the plain text into two portions irregularly and placing the second portion first, with the intention of shifting the stereotyped preambles into undefined positions in the bodies of the messages.

  At the end of 1937, the Japanese military began using a different system, a combination of a code and a cipher, Tiltman recalled. The message was encoded into blocks of figures using a codebook which gave numerical equivalents for all the main Jap
anese characters, phrases and kana syllables. The resultant code was enciphered using an additive system with predetermined groups of figures, taken from a cipher table or book, added to the code groups using the ‘addition modulo 10’ or Fibonacci system of addition in which no figures are carried over. Seven plus eight therefore becomes five rather than fifteen.

  The first cipher which was intercepted in sufficient quantity for attack had a two-part, four-figure code book in which the cipher groups were limited to multiples of three and the additive recipher consisted of ten thousand four-figure groups arranged on a hundred pages, each containing ten lines and ten columns. Both the starting and ending points in the additive were indicated in each message and these were reciphered each by one of a hundred four-figure additive groups controlled by a particular dinome of the first cipher group for the starting point indicator and the last cipher group for the ending point.

  It was not until the late summer of 1938 that Tiltman managed to make the vital breakthrough into the new Japanese military cipher. But by now the attentions of GC&CS were being concentrated elsewhere. With war looming in Europe, they were making frantic efforts to unravel the German Enigma cipher machine and were preparing for the seemingly inevitable confrontation with Hitler.

  Admiral Sinclair had bought a mansion at Bletchley Park, fifty miles north of London, in the spring of 1938 as a ‘War Station’ for both MI6 and GC&CS. He was acting entirely on his own initiative. Having realized that, if it came to war, he would need to protect his staff from the inevitable air raids, Sinclair had asked the Foreign Office to pay for a ‘War Station’. Its response was that the War Office was responsible for war: the generals should pay. The generals told Sinclair that as a former Director of Naval Intelligence he should go to the admirals, who told him he was part of the Foreign Office and the mandarins should pay.

  Frustrated by his inability to get anyone to pay the £7,500 asking price for Bletchley Park, Sir Hugh dipped into his own pocket to buy it. ‘We know he paid for it,’ said one former intelligence officer. ‘We're not sure if he was ever repaid. He died soon afterwards so he probably wasn't.’ The Park itself was given the covername ‘Station X’, not as might be assumed a symbol of mystery but simply the tenth of a large number of sites eventually acquired by MI6 for its various wartime operations and designated using Roman numerals.

  Shortly before the Munich Crisis of September 1938, some of the codebreakers and a number of MI6 sections moved to Bletchley Park briefly for ‘a rehearsal’. It was while they were there that Tiltman made the breakthrough into the main Japanese military cipher, working out how the Japanese operators indicated which part of the additive book they were using. ‘By the end of the year it seemed to me that we were sufficiently advanced to shift the responsibility for Japanese cryptanalysis to the Far East. At the beginning of 1939 I went again to Hong Kong.’

  Tiltman took with him two of his military codebreakers, Captain Peter Marr-Johnson, a Royal Artillery officer who had studied in Japan for a number of years and was immersed in both the language and codebreaking, and Lieutenant Geoffrey Stevens, a straightforward codebreaker. They would lead the FECB's efforts against the Japanese army codes and ciphers. Now at last Tiltman was successful in persuading the bureau to take on military as well as naval codebreaking.

  At the same time, the RAF had decided to step up its own presence within the bureau. There was no separate Japanese air force: both the Imperial Japanese Navy and the Imperial Japanese Army operated their own aircraft. But although there were seven RAF wireless intercept operators at the Stonecutters Island site and an RAF officer in the bureau's intelligence section, there were no RAF codebreakers. Squadron Leader H. T. ‘Alf’ Bennett, who was an experienced Japan expert, having spent three years as British Air Attaché in Tokyo, was sent on a codebreaking course at GC&CS, arriving in Hong Kong in March 1939.

  ‘As far as Air sigint was concerned no Japanese aircraft normally operated within 600 miles of Hong Kong and therefore traffic was mainly inaudible,’ wrote Nigel de Grey, one of the leading code-breakers at GC&CS. ‘On the other hand, the Japanese Naval Air Arm figured to a considerable extent in the naval messages that could be decrypted.’

  Eric Nave had now returned to Hong Kong and under his leadership the naval codebreakers exercised a fair degree of control over the Japanese naval traffic, Shaw recalled.

  Book-building on the 4-kana General Code went on steadily and progress was made with the Flag Officers’ Code, which mainly concerned local politics in China. A new naval kana code came into use between Japanese naval bases and hence was known as the Dockyard Code. The book was alphabetical but the vocabulary, being largely technical, was slow to build up. In the early stages of the kana codes, when transposition was fairly simple, messages were decrypted by two selected and trained Y operators at Stonecutters.

  By the end of 1938 Japanese communications security had been stepped up and this was no longer possible. New codes and ciphers were constantly being introduced and they were so complicated that trained cryptographers were required to unravel them. Nevertheless, according to Denniston, the codebreakers in London and Hong Kong remained ‘reasonably fluent in their reading of all main Japanese naval ciphers and knew quite a lot about Japanese army ciphers used in China’.

  They had also set up a separate section in London, under G. L. N. Hope, one of the First World War codebreakers, to decipher Japanese commercial codes and ciphers which Naval Intelligence was hoping to use to track supply convoys, Denniston recalled.

  Some time in 1938, the Admiral [Sinclair] and the newly appointed Director of Naval Intelligence [Admiral John Godfrey] formed the opinion that, in the event of a troublous political situation in the Far East, the Japanese might take steps to render their diplomatic and service material illegible, and that the communications of the big Japanese firms, particularly as to shipping, might be the only available source of intelligence. Therefore, Hope started a very small section to investigate commercial traffic, more especially the telegrams of the big Japanese firms.

  The concerns of the intelligence chiefs soon proved justified. At the end of 1938 the codebreakers began to read messages enciphered on the Type A diplomatic cipher machine referring to a new device angoo-ki taipu B, the Type B machine. It made its first appearance on 20 March 1939 and while the Type A machine continued in service for most messages, it was clear that it was being gradually replaced for the secret messages sent between the Japanese embassies in the major world capitals and Tokyo.

  In November 1938 the Imperial Japanese Navy replaced its codes, introducing a new 4- kana system for its main General Operational Code. Then the codebreakers began to find messages referring to yet another system to be called kaigun angoo-sho D, or Navy Code D, which was introduced on 1 June 1939 as the new General Operational Code. The recently introduced system it replaced was now reserved only for the use of Flag Officers. The Japanese also introduced yet another Dockyard Code.

  Nave concentrated on trying to break the Dockyard Code while all intercepted messages in the new five-figure General Operational Code were sent back to GC&CS in London for investigation. Nobby Clarke called in Tiltman, who immediately recognized a number of similarities to the Japanese army codes he had been working on.

  Barbara Abernethy was one of those working in the naval section at the time. A fluent German-speaker, she had been plucked out of the Foreign Office and sent to Broadway Buildings.

  Nobby Clarke was the most delightful man, very quiet, but very bright. I know he spent a lot of time fighting with the Admiralty for more staff. But it was a most civilized office. I was posted over there for a week not knowing what I was doing and told that it was strict secrecy. I was there for a week and they apparently approved of me because I was kept on and I stayed there. Life was very civilized in those days, you know, we stopped for tea and it was brought in by messengers. We had our own cups. I was very impressed by this, first job I'd ever had and it seemed paradise to me. I thought: ‘Well
this is the life, isn't it? Thank God I'm not back in the Foreign Office.’ The head of the Japanese naval sub-section, Lieutenant-Commander Bruce Keith, and one of his sidekicks, Lieutenant Neil Barham, used to come into our office to have tea but we rarely saw John Tiltman. He was head of the military section, a very nice, good-looking guy who always wore tartan trews. Very bright.

  Since many of the messages sent in the General Operational Code were reporting ships’ positions, they used a lot of figures and this was one of the ways into the system. Tiltman noticed that the code groups representing figures were stepped in a similar way to the army codes. Each code group for a figure differed from the ones below and above it by 102. The most common figure in longitude and latitude positions is zero and the most common group in the encoded positions was also the lowest. Clearly it was zero. The next lowest, just 102 on, was found to be one and from then on it was a simple task to work out the other figures. Within the space of a few weeks Tiltman had broken the new code.

  Tiltman's breakthrough into the new General Operational Code, brilliant though it undoubtedly was, did not mean that the British could now read the Japanese messages. The code, which would later become known to the Allies as JN25, was typical of the combinations of codebook and cipher additive that were to become common throughout the Japanese armed forces. The codebook contained more than 30,000 words or phrases designed to cover every contingency faced by the Imperial Japanese Navy, with the most frequently used repeated a number of times to enhance security. Alongside each was a five-figure group which the operator used to encode his message.