The Emperor's Codes Read online

  It was unlikely, in the circumstances, that Japanese espionage in the Far East was limited to Malaya, he added. ‘There is probably some interesting matter coming from the Japanese Consul-General at Hong Kong and it is proposed that the former's cables should be intercepted and sent to Shaw.’

  Dickens was determined to ensure that there was a substantial increase in British intelligence operations against the Japanese and persuaded the Admiralty to send an extra cryptographer to Hong Kong together with a dozen trained wireless operators to intercept any Japanese radio messages they could find. ‘It will be necessary to go on increasing the number of Japanese cryptographers, as it is essential now to create a strong centre on the China Station as well as in London,’ he said. ‘Unless we can have a good grip of Japanese wireless, our intelligence will be quite inadequate and we may find ourselves critically at fault.’

  In the meantime, Dickens said, he was asking MI5 to be discreet in its investigation of the Singapore spy scandal. ‘Unless great care is taken,’ he added, ‘there is, of course, a danger that the Japanese will discover that we are intercepting their Consul-General's cables.’

  The Japanese were already showing increased awareness of the need for more security, upgrading the naval attaché system by introducing a machine cipher. The success of the British codebreakers during the First World War had made foreign governments aware of the ease with which relatively complex ciphers could be solved using basic characteristics of the language, such as letter frequency. Machine ciphers were developed to try to protect against this, changing the system of encryption with each letter to ensure that the codebreakers could never build up sufficient depth to break the keys.

  The new naval attaché system, which enciphered the kana syllables rather than romaji letters, was broken by Hugh Foss, a tall, gangly, red-haired eccentric who was the main Japanese systems-breaker, and Oliver Strachey, a former First World War military codebreaker. This was the codebreakers’ first experience of machine ciphers, Nave said.

  There were two wheels, one containing twenty consonants and the other the six vowels, V being a vowel. The first trial was made in the office using a brown foolscap file cover with a collar stud retrieved from a returning laundry parcel, a piece of string and slots cut in the cover for the letters. This worked, so we asked the Signal School at Portsmouth to help and received some expertly finished models in Bakelite. To find the starting point or key, the Japanese language provided good assistance. Japanese is not constructed like most European languages on a consonant–vowel basis, but is monosyllabic: YO KO HA MA is four syllables. The word oyobi [and] was a great help, giving a sequence of three vowels and at times four or even five on the smaller wheel. Then when one of these vowels represented itself we could have a zero position necessary to place the enciphered text underneath each other in sections of sixty symbols to establish the order of letters.

  This first naval attaché machine was not solved by the Americans, who called it Orange, until February 1936, and then apparently only with the aid of a ‘pinch’, or theft, of information, possibly even a machine, from the Washington apartment of the Japanese Naval Attaché. By this time the Japanese had introduced a number of additional security measures which hampered the British code-breaking effort. ‘Firstly, the machine could be set to jump certain positions,’ said Nave. ‘But this proved a minor nuisance. More tiresome was the subsequent monthly change of the order of letters on the wheel and introduction of one hundred keys based on the message serial numbers.’

  Shortly after the British broke the naval attaché system, in November 1934, Admiral Sinclair chaired the annual meeting of an obscure Whitehall committee. The Committee of Co-ordination, or the Y Committee, as it was known for security reasons, was in charge of the interception of foreign communications. It met on the fourth floor of Broadway Buildings. Those who attended included Denniston and the directors of the three service intelligence departments. Since the threat from Japan was at the top of the agenda, Nave was also asked to take part.

  The committee considered Captain Tait's report of his visit to the Far East to investigate Japan's ‘provocative’ activities in the region. Admiral Dickens insisted it showed that the British had no choice but to counter the massive increase in Japanese espionage activities in the Far East with their own substantial intelligence-gathering operation.

  Since the navy was responsible for the collection of ‘special intelligence’ in the Far East, it had strengthened the Japanese naval section at GC&CS, providing Nave with two other Japanese-speaking officers, while Shaw in Shanghai had been given another officer to assist him. Dickens had also sent a further naval code-breaker to Hong Kong with around a dozen wireless operators to set up a small intercept and codebreaking unit. But despite these moves to strengthen the British intelligence effort, more was needed.

  There was general consensus around the table that Admiral Dickens was right. Japan constituted a major threat to British interests in the region. They would be shirking their responsibilities if they did not take the necessary steps to ensure there was early warning of any fresh Japanese moves. They needed their own spy base in South-east Asia and Hong Kong was ideally placed.

  ‘The present trouble between Russia and Japan will provide a first-class rehearsal for interception and cryptography in the Far East,’ Sinclair said. The committee agreed that in order to disguise its real role, the new organization should have a ‘fairly bland’ title, the Far East Combined Bureau (FECB). But its activities would be far from bland. It would collect intelligence from every possible source: spies, informants and, of course, the secret Japanese messages, ‘the decipherment of which affords the most valuable source of information – especially in time of crisis, emergency or war – it is possible to conceive’.

  The expansion of intelligence operations against the Japanese was hampered only by a lack of anyone with the necessary linguistic ability to staff the new bureau, most particularly in the army and RAF. All three services approached Denniston for assistance. Initially, he insisted that he was unable to help. ‘It is not possible to deal with such traffic in the GC&CS in London owing to the distance and the delay in sending such material home. Nor are there any Japanese cryptographers at present available here to send out to the Far East as when the numerical peacetime establishment of the Code and Cipher School was arrived at the question of the provision of cryptographers for duty overseas was not contemplated.’

  As a result of the difficulty finding the necessary language specialists, the Far East Combined Bureau was not set up until March 1935. It was housed in part of a block of offices on the Hong Kong naval dockyard, separated from other parts of the building by iron grilles which were manned permanently by armed guards carrying loaded revolvers. The guards excited a great deal of local curiosity, rendering useless any attempts to keep the bureau's presence secret.

  The move to a more comprehensive intelligence-collection role led to the new organization having ‘a somewhat shaky start’, one former officer recalled. It now consisted of three main components. The first, and the part that was to cause the problems, was an intelligence-reporting section, made up of naval and army staff officers. It was headed by the Chief of Intelligence Staff (COIS), who was also the bureau commander. The department's task was to evaluate material from a variety of sources, ranging from the deciphered messages and MI6 agent reports to ‘local gossip’, in order to keep the authorities both in London and in South-east Asia informed of what the Japanese were doing.

  The second component was the Special Intelligence, or code-breaking, section, also known as the Y Section, which consisted of a number of Japanese, Chinese and Russian interpreters, initially at least all naval or former naval officers. They mainly worked on Japanese messages but also on Chinese and Russian naval communications, deciphering any messages they could, and passing translations of anything interesting to the intelligence section. When they were unable to decode the messages, they sent them back to London, along wit
h all enciphered army and air material, in the diplomatic bag.

  The final component was a team of a dozen wireless intercept, or Y Service, operators, who lived and worked on Stonecutters Island, about four miles across the harbour from the naval dockyard. The operators, initially a mixture of Royal Navy ratings and RAF airmen, faced a number of problems. Their watchroom was ramshackle, with some of the wireless receivers located in a former washroom. They were located alongside the main Royal Navy radio transmission station, which interfered with their signals. The intercepted messages had to be carried to the bureau offices by courier on a steamboat ferry which took around an hour at a time. The only telephone link went through the local public exchange and was therefore insecure. During the summer months the area was buffeted by a series of typhoons which not only led to the suspension of the steamboat service, and therefore the courier operation, but forced the interceptors to take down their aerials until the storm had passed. To compound the difficulties of working there, the island was plagued by ‘a plethora of winged and other insects’.

  The codebreaking section was placed under the command of Harry Shaw, who had now retired with the rank of captain. He was assisted by two other Royal Navy Japanese cryptographers, Paymaster Lieutenant-Commander Dick Thatcher, who had been working with Nave in London, and Lieutenant Neil Barham. They concentrated on three Japanese naval codes and ciphers, Shaw recalled:

  The main commitment was the Japanese Naval General Cipher which was being worked up from an embryonic stage. This was a basic book of about 90,000 4-kana groups, with a simple transposition which was changed at intervals. Fortunately the book was alphabetical and was not changed for three years. This, and continuity of work, enabled the changes in transposition to be followed with comparative ease. Later, when Japanese naval codes and ciphers were upgraded, this earlier experience afforded a useful background. Second priority was the Flag Officer system used by Japanese naval staff officers stationed in China. The book was rather smaller than that of the general system and the transposition was of a higher grade. Limitation of personnel restricted work on this system until staff was increased in 1937.

  They also decoded the small number of messages sent in the tasogare reporting code originally broken by Nave. Any enciphered army messages were sent back to London, Shaw said. ‘Special Intelligence translations were passed to the Chief of Intelligence Staff. If he and his colleagues thought the contents warranted it, a précis was drafted, usually in consultation with the Head of Special Intelligence. A comment was added and it was signalled to C-in-C China, then repeated to DNI, Admiralty, who was asked to pass it to the War Office, etc.’

  But the need to pass messages through the Intelligence Department led to a great deal of tension between Shaw and the COIS and head of bureau, Captain John Waller. Shaw arrived expecting to be able to act much as he had in Shanghai, dealing direct with GC&CS with regard to the codes, and to the Commander-in-Chief Far East and the Admiralty on matters of intelligence. But despite being in charge of a much bigger operation, he found his hands tied by the creation of the Intelligence Department. Waller insisted everything went through him and appeared to pay scant regard to the need to protect the source of their material from becoming public knowledge, Shaw recalled.

  Dissemination of Special Intelligence to local authorities was for long a bone of contention. The fact that some of the intelligence staff officers had no Far Eastern background militated against a proper appreciation by them of items of intelligence. None of them knew the implications of cryptography and security mindedness was inclined to be sacrificed to the desire to make definite use of Special Intelligence, particularly in dealing with civilian authorities. There were instances of leakage of information thus passed, and after one case of leakage the Admiralty ordered that Special Intelligence was not to be communicated to the head of government.

  With the bureau's two most senior naval officers barely on speaking terms, morale suffered and divisions between the codebreakers and the intelligence-reporting section, and between the three services, became accentuated. The absence of any army or RAF codebreakers and any army intercept operators also caused problems, recalled Colonel Valentine Burkhardt, who arrived in the spring of 1936 to find the bureau embroiled in a series of petty turf wars. ‘No great intimacy existed between the heads of the naval and military sectors. Each service ran its own work on its own, though all were under the same roof. Visits to the Y room were distinctly discouraged and very little information was handed out to the army on the grounds that it did nothing to contribute to production in this line. To overcome this monopoly of Y, four army signallers were sent by the War Office to assist with reception on Stonecutters.’

  In an attempt to break down the ‘atmosphere of mutual suspicion’, Burkhardt obtained a grant of £1,500 a year from the Army Entertainment Fund. He used the money to set up a series of social gatherings, including a weekly ‘Chinese dinner’ at his own home for all the bureau staff, which ‘did much to break down inter-service rivalries’.

  Gradually, Shaw said, the codebreakers began to accept, and live with, a situation where they had no control over what intelligence reports were issued.

  No intelligence records were kept in the Special Intelligence (code-breaking) department. The translators used their register and vocabulary indexes to refer to previous messages and private notebooks for matters of current interest. The translators generally had better local background than the Intelligence Department and they were expected to insert explanations of matters which called for local knowledge. Usually translations with notes appended were taken personally to the Chief of Intelligence Staff by the head of Special Intelligence who gave any further explanation asked for. The Chief of Intelligence Staff then distributed the sigint with comments of his own.

  The arrival of the army signallers and increasing tension in the region led to more interest from the War Office in the work of the Far East Combined Bureau, Burkhardt recalled.

  Japanese hatred of all foreigners made an outbreak of war, particularly against Great Britain, an imminent possibility. Responsibility for keeping track of political trends in Japan, of military and strategic developments, and of any secret preparations for a military blow to be delivered without warning, therefore devolved upon FECB.

  During the summer of 1936 a murder in Peking, involving accusations of complicity against a British soldier by the Japanese Secret Service, led to the discovery that the interception of Japanese consular and military messages had lapsed. Two of the high-speed army operators who were then functioning on Stonecutters were moved to Tientsin [in northern China] to intercept Japanese Army traffic which was out of range of Hong Kong. The army code was far more complicated than either the naval or the consular and just when it was becoming readable the reciphering table would be changed. The preliminary work proved, however, extremely valuable.

  The Tientsin listening post, staffed by around two dozen wireless intercept operators, was known as the North China Signals Section. At one stage it had its own codebreaker, but most of the intercepted material was sent back to London to Captain John Tiltman, the head of the GC&CS military section, which included an RAF Japanese interpreter.

  John H. Tiltman was arguably one of the best codebreakers, if not the best, working during this period. Born in London on 24 May 1894, he was so obviously brilliant as a child that, at the remarkably young age of thirteen, he was offered a place at Oxford. He served with the King's Own Scottish Borderers in France during the First World War, winning the Military Cross for his bravery, and was seconded to MI1b shortly before it merged with Room 40 to become GC&CS. Although Tiltman spoke very little Japanese, he later recalled that his main work as head of the military section during the mid-1930s was on the material sent back from Hong Kong and Tientsin.

  I learned what I know of the written form of Japanese the hard way! We had a British intercept station at Hong Kong which from about 1935 forwarded to us considerable quantities of Japanese mil
itary intercept. The major part of the traffic was in a succession of military systems used for transmission of intelligence reports from China. During this period, the cipher systems used seemed to be changed quite drastically about every nine months and I had a hard time keeping up with the changes.

  There were no Japanese interpreters working in the military section. Tiltman could occasionally call on the expertise of men like Hobart-Hampden and Parlett. But they were sometimes resentful of any forays Tiltman made into breaking Japanese codes and ciphers.

  In 1933, I solved the Japanese military attaché system which had been in use since 1927. There was a small basic code-chart of, I think, 240 units which meant that a large part of the plain text had to be spelled out in syllables. I don't remember the details of the system except that the code-chart had to be reconstructed and forty different sets of lines and column co-ordinates recovered. There were at the time two very distinguished Japanese scholars in the office who had each retired as British Consul-General in Japan. One of them partially reconstructed the code-chart for me from the material in one of the forty keys, and I set about recovering the other keys. Shortly afterwards the other expert came and said he had heard I had some interesting work in progress and that he would like to help. So I gave him the material in one of the keys in which I had then recovered most of the syllabic values, numbers and so forth. He returned it to me three weeks later unchanged, saying my solution was ‘plausible’. This should have warned me that I was treading on hallowed ground.

  4

  DIPLOMATIC SECRETS

  With the Japanese threat increasing, Denniston applied to the Treasury for more funding to take on additional Japanese-language experts for the diplomatic section.

  Normally, the Code and Cipher School recruit their cryptographers from men of nineteen to twenty-four years of age from the universities. They are untrained and have no special language qualifications. This entails considerable time spent on the study of oriental languages followed by a cryptographic course at the Code and Cypher School. Obviously, such a method of recruitment would not provide the necessary cryptographers for some years to come. Such delay cannot be accepted. It is suggested therefore that such individuals should be sought for amongst retired service or consular officers who have the necessary language qualifications for four additional appointments as Junior Assistants in the Code and Cipher School at say a fixed salary of £500 per annum.