The Emperor's Codes

by Michael Smith

  The first two retired officers taken on were J. W. Marsden, a former assistant military attaché at the British Embassy in Tokyo, and N. K. Roscoe, another former army officer who had served in Japan. They were followed by Captain Malcolm Kennedy, who during his military service had been seconded to the Imperial Japanese Army. After a spell in military intelligence, Kennedy had returned to Japan as the Tokyo correspondent of the Reuters news agency. He was now a leading member of an influential clique of ex-officers who, while working in Japan during the 1920s and early 1930s, had become confirmed Nipponophiles.

  None of this group was a natural codebreaker; they normally worked alongside the more experienced members of GC&CS translating and offering tips on the language structure. All of them were dismayed by what they saw as the anti-Japanese attitude prevalent in Britain, none more so than Kennedy who continually railed against ‘the mistake we made in scrapping the Anglo-Japanese Alliance in order to please America, thereby leaving Japan with a feeling of grievance’.

  When Denniston first offered him the post, Kennedy was attracted to the idea of ‘getting to grips with the Far East question’ and of working with Marsden and Roscoe, both of whom he knew well. But he openly admitted his distaste for spying on his Japanese friends, saying that while the work was ‘more in my line’ than anything else on offer, ‘there are aspects of it which I would dislike intensely’. Nevertheless, he accepted the post, noting in his diary on 1 October 1935: ‘Started work in my new job at the FO where working with a number of old friends – Marsden, Hobart-Hampden, Nave, Roscoe, etc.’

  The diplomatic section received its raw material from two separate sources. The Royal Navy codebreakers based at the Far East Combined Bureau continued to obtain drop copies of Japanese telegrams from the Cable and Wireless offices in Singapore and Hong Kong, and, as Shaw recalled, also had surreptitious assistance from MI6 in obtaining others. ‘Copies of Japanese diplomatic and consular codes in use in the Far East were supplied from GC&CS. Very little diplomatic material was taken at Stonecutters. But arrangements were made semi-officially with Cable and Wireless to supply the bureau with copies of Japanese Government telegrams passing over their lines in Hong Kong and with SIS [the Secret Intelligence Service, MI6] in Shanghai to supply copies of those which were handled by the Chinese Post Office.’

  Much of the Japanese diplomatic material had to be sent back to London where it supplemented the codebreakers’ main source, the diplomatic telegrams obtained by Henry Maine. As international tension rose throughout the mid-1930s, he persuaded the international cable companies to be even more co-operative, Denniston recalled.

  Maine's excellent liaison proved of the greatest help. When the state of unrest in the world became intense, from 1935 onwards, it was found that the ten days’ delay granted by the warrant became intolerable. Maine was able to cut it down to twenty-four to forty-eight hours in the case of foreign companies, and to instant service, where necessary, in the case of the Central Telegraph Office and Cable and Wireless. The Japanese traffic to France and Germany always went via Malta. All Italian cable traffic passed there. Thus we were, throughout, enabled to watch the growth of the Axis combination.

  Increasingly, diplomats began to use wireless transmissions. The three services had their own intercept sites but were reluctant to use them for diplomatic traffic. So Admiral Sinclair co-opted a small Metropolitan Police wireless intercept unit which had been operating from the attic of Scotland Yard, tracking the operations of Soviet spies. Harold Kenworthy, the unit's chief engineer, later recalled how Sinclair agreed to fund the unit which was moved from Scotland Yard to the grounds of the Metropolitan Police Nursing Home at Denmark Hill in south London. ‘In the middle 1930s, secret government stations were set up by various foreign powers. These various groups were mainly covered by our own intercept set up in Denmark Hill. The services were disinclined to intercept diplomatic wireless to any extent as it would lead to curtailment of the examination of their particular service channels. The Commissioner came to an agreement and SIS paid a lump sum a year for the service and a number of police operators.’

  The increased use of wireless transmitters was not the only new development the codebreakers had to face. Their success during the First World War had led to the development of a number of machines designed to encipher messages more securely. The most famous of these was the Enigma machine used by the Germans to encipher their most secret communications. But by the early 1930s there were a number of others in operation and there were clear signs that the Japanese were using one for some of the messages that had previously produced the most important intelligence.

  While most of the diplomatic codes and ciphers employed by the Japanese were relatively simple to unravel, the codebreakers had found difficulty making an entry into a five-letter system introduced in the early 1930s. The traffic, which like other Japanese diplomatic messages was sent using the romaji characters, was being passed on two separate sets of links. The first was the Far East network including the Singapore, Hong Kong and Shanghai cables received in the Far East Combined Bureau. The other was the main diplomatic network connecting the gaimushoo in Tokyo to the embassies in the major international capitals such as London, Paris, Berlin, Rome and Washington.

  It was immediately obvious to the codebreakers that the Japanese diplomatic links had gone over to a machine cipher for all of the most important ‘State Secret’ communications. If the British were to have any chance of predicting and countering the Japanese machinations behind the scenes in the Far East, the five-letter cipher had to be broken.

  The machine was known to the Japanese as the angoo-ki taipu A, the Type A cipher machine, or as the 91-shiki oo-bun inji-ki, Alphabetical Typewriter 91. The 91 was derived from the year of its development, 1931 being 2591 in the Japanese calendar. The Type A machine was similar to the Japanese naval attaché machine broken by Foss and Strachey, although it used romaji letters instead of the kana syllables. It had two typewriters, one to input the plain text, the other to type out the enciphered message; a standard telephone exchange plugboard; and the encipherment mechanism. This last comprised two so-called ‘half-rotors’ on a fixed shaft, each of which had twenty-six electrical contacts wired around the circumference of one of its sides, and a forty-seven-pin gearwheel which moved the half-rotors either one or two positions at every stroke of the typewriter keys.

  The operator connected the two typewriters to the machine through the plugboard using a daily changing setting. Each message was preceded by a five-figure indicator group which told the receiving station the starting positions for both the rotor and the gearwheel, as well as the pins which were to be removed from the gearwheel.

  He then typed the plain text into the input typewriter. The depression of each key sent an electrical impulse through the socket on the plugboard to which it was wired, then on through an endplate and into the circuitry of the two half-rotors. This reflected it back out via the endplate to the output typewriter which typed the enciphered letter. The decipherment was thus determined largely by the plugboard settings and the fixed wiring of the half-rotors, and only varied by the movement of the gearwheel. This normally moved the rotor forward by one contact for every letter inputted, but where a pin had been removed it jumped a contact.

  Basic cryptographic analysis of the messages enciphered on the machine, almost certainly carried out by Foss and Strachey, found that the frequency patterns of the letters seemed to change every ten days and that six of the letters in use stood out as occurring either more or less often than the other twenty.

  The first piece of evidence obviously indicated that the cipher machine's predetermined positions or ‘keys’ were changed every ten days. The six letters that stood out were more interesting and seemed to offer the means of making an entry into the system. They were all the vowels plus V. If, as seemed likely, they represented Japanese vowels, the vowel–consonant patterns in the enciphered text would exactly reflect those in the original plain text
.

  Using the common associations between the Japanese vowels, particularly the yoo and yuu sequences, the British codebreakers soon found that the cipher was vulnerable to attack, and by November 1934 they had not only found a way in but had discovered that the ten-day key periods operated on a predictable cycle.

  The codebreakers realized that recovering the messages was likely to prove a tedious, time-consuming task if it were done by hand. So they turned to Kenworthy and his Metropolitan Police wireless staff who were intercepting diplomatic and commercial traffic for GC&CS and had become its advisers on virtually any technical matter. Kenworthy recalled being asked to produce a machine that would allow the codebreakers easy access to the Japanese diplomatic cipher.

  As codes and ciphers became more advanced the cryptographers required mechanical aids. In November 1934 our branch was asked to help as we possessed a small workshop. The particular project was a Japanese decoding machine on a mechanical hand-operated system where a number of geared wheels were made to revolve in a certain order. This process was quite slow as the answer had to be written down for every movement of the wheels against the keys. It occurred to us that this could be done electrically with relays and coupled to a keyboard such as used on a teleprinter. The first machine was working by August 1935. It was very successful and was known as the J machine.

  This was spectacular progress. The American codebreakers of the US Army's Signal Intelligence Service led by William Friedman are always fêted as the pacesetters in the race to break the Japanese codes and ciphers. But it was not until late 1936 that they attacked and broke the Type A machine, which they designated the Red machine, and a further two years before they built a similar device to Kenworthy's J machine.

  The mid-1930s had seen the appearance of a new threat to the British Empire that more than rivalled that from Japan. The rise to power of Adolf Hitler and the establishment of the Third Reich had created another state that was not only, like Japan, keen to expand its territory at the expense of its neighbours but was uncomfortably closer to home. Even more worryingly, both the diplomatic telegrams deciphered using Kenworthy's J machine and the military attaché messages broken by John Tiltman provided clear evidence that the Germans and the Japanese were preparing to form an alliance which would threaten peace not just in the Far East but throughout the world.

  The British codebreakers had virtually ignored German radio traffic after the end of the First World War. Germany was thought to have been ‘crushed never again to rise’, recalled Clarke. Added to this, its diplomatic ciphers used the theoretically unbreakable onetime pad system while its armed forces employed the Enigma cipher machine, which the British initially assumed was too difficult to break.

  ‘Considering what Room 40 had achieved in 1914-18, it seems extraordinary that anyone should believe this,’ recalled Josh Cooper, another of the interwar codebreakers. ‘But it was generally assumed that no civilized nation that had once been through the traumatic experience of having its ciphers read would ever allow it to happen again and that, after the wide publicity given to Room 40's results, it would be a waste of time to work on German high-grade systems.’

  But the codebreakers did not need to read the German diplomatic traffic to build up a good picture of the new closeness between Germany and Japan. The enciphered telegrams of the Japanese Ambassador to Germany and, in particular, the messages that Oshima Hiroshi, the Japanese Military Attaché in Berlin, was sending back to Tokyo told their own story. Oshima was a key member of the group within the Imperial Army that was leading the drive to expand Japanese territory in the Far East. He was also an ardent advocate of closer ties with Germany. He had been born in Japan's Gifu prefecture in 1886, the son of a former Japanese War Minister. Oshima's father Kenichi was a keen Anglophile who had stayed at Windsor Castle as a guest of Queen Victoria. His period as War Minister had come during the First World War when Japan was allied with Britain against Germany. The younger Oshima had no such affection for the British. After graduating from military academy in 1905, he served with distinction as an army officer and was first posted to Germany in 1921. He returned to Berlin in 1934, speaking perfect German and determined to build up closer relations with the new regime. Within a year, he had gained a private audience with Hitler and was a friend and confidant of Joachim von Ribbentrop, then Hitler's Ambassador-at-Large. He was also close to a number of leading members of the German general staff, as one British military attaché noted in a report to London.

  Gen Oshima is a typically courteous and polite Japanese officer of undoubted intelligence and considerable personality. The German War Office have a high opinion of his attainments. Unlike many of his countrymen, he is a gregarious creature and enjoys society. This is probably mainly due to the fact that he drinks like a fish. He is very rarely sober in the evening and the more he drinks the more he talks. On the other hand, his discretion seems to increase under the influence of alcohol and I have often seen him surrounded by a circle of his colleagues, all hoping that he will produce some pearl of information, and all invariably disappointed at getting absolutely no return for their outlay of Kirsch, which is his favourite tipple. During the recent manoeuvres in East Prussia, he finished off a bottle of brandy in his car every day and after General Beck's dinner on the last night at Koenigsberg, he drank himself into a state of extremely noisy intoxication on a mixture of Cosacken-Café and Pillkaller.

  Oshima's discussions with the Nazi leaders frequently centred on the possibility of a pact between the two countries. By August 1935 the codebreakers’ reports of traffic passing on the diplomatic circuits between Berlin and Tokyo had convinced the author of a British military intelligence report that an alliance was inevitable.

  The Japanese and German nations have many traits in common. Both have strong leanings towards nationalism. In both the military caste have traditionally been – in Japan they still are – the real rulers of the country. Recent events have served to accentuate this resemblance. Both countries have left the League of Nations, for not dissimilar reasons. While the Japanese, whose thirst for a safe source of raw materials has led to their annexation of Manchukuo and forthcoming annexation of North China, view with sympathy Germany's aspirations for the return of her colonies and for expansion into the Ukraine.

  The diplomatic telegrams and the messages of both Oshima and his Imperial Japanese Navy counterpart revealed that, as part of the rapprochement, Berlin had said it would not be seeking the return of its Far Eastern colonies seized by Japan during the First World War. They also showed that the two countries’ armed forces had begun exchanging technical know-how.

  By early 1936 indications of an imminent German–Japanese pact, engineered primarily by the Imperial Japanese Army, were increasing. Oshima was one of its chief architects, leading Malcolm Kennedy to note that the evidence suggested it would be ‘more in the nature of an understanding between the naval and military staffs of the two nations than a binding agreement between the two governments’.

  The Anti-Comintern Pact was eventually signed in November 1936. Taken at face value, it was little more than an agreement to exchange information on the spread of communism. But a secret protocol to the pact went much further, effectively aligning the two countries against the Soviet Union.

  The Foreign Office appears to have had no definite proof that the secret protocol existed, but Kennedy noted that, according to the Soviet Ambassador to Britain, ‘Moscow has obtained a complete copy of the J–German agreement and, despite official denials to the contrary, it includes certain secret protocols. One wonders just how much the Soviet does know and how much is insidious gossip on its part.’

  Meanwhile, the staff of the Government Code and Cipher School, like the rest of Britain, were in thrall to a different kind of gossip, concerning King Edward VIII's love affair with Wallis Simpson, an American divorcee. The very future of the British royal family seemed threatened by the scandal, as Kennedy recorded in his diary entry for Thursday, 3 December
1936.

  The King's affair with Mrs Simpson, about which the US press has been talking so freely for some time past, has now been brought to light in our own press, which hitherto has shown a laudable readiness to keep silent on the subject. The US press, of course, has been simply gloating over the whole thing and publishing columns and columns of the most sordid and sensational details. A most ghastly business and likely to do untold harm to British prestige in general and to the prestige of the Throne in particular. I heard of it first from Roscoe some three or four weeks ago and, of course, knowledge of it was spreading gradually in this country; but to the general public it has come like a bolt from the blue and most people are naturally horrified.

  A week later the newspapers reported that the King was expected that afternoon to announce that he was abdicating. The staff of the Government Code and Cipher School were as interested in the events as everyone else in Britain, Kennedy noted.

  Knowing that the final decision was to be broadcast at 4 p.m., I went along with Marsden to a room which had imported a wireless set for the occasion. Found most of the office congregated there – a curious, almost uncanny gathering, with everyone doing his or her best to appear calm and collected, though all present were clearly in a state of nervous tension. A low, subdued chatter. Then complete silence as the announcer began to speak, the tense silence continuing until the announcement was finished, when everyone dispersed quietly and returned to work. A most tragic ending to a reign which had seemed to hold out such great prospects. But in view of all that has come out, the King's decision to abdicate may perhaps prove all for the best. And even though his successor may not, at first sight, appear so well-fitted as his brother and may be lacking his magnetic personality, he seems likely to prove a far more conscientious, steady-going monarch, and his wife has all the qualities of a most popular Queen.