The Emperor's Codes

by Michael Smith

  Nave, whose Australian operators had also been monitoring the Japanese preparations for war, said the Japanese Navy had mounted a massive deception exercise to prevent anyone realizing that Pearl Harbor was a target. ‘One step which undoubtedly deceived the US Navy was the transfer of the wireless operators from aircraft carriers taking part in the Pearl Harbor attack to other ships in the Inland Sea,’ Nave said. Since the Allied intercept operators were using the ‘fists’ of their Japanese counterparts to identify the various enemy radio stations, this move completely threw them, leading them to place the carrier force that was to attack Pearl Harbor in Japanese home waters. The JN25 additive change on 4 December meant, ‘there was no help from the codebreakers,’ Nave said. However, wireless traffic to the Philippines and South-east Asia not only continued but increased with the intention of focusing interest on that area.’

  The entry of America into the war was of little use to the British in Malaya. The Japanese, with far better aerial power, destroyed all but ten of the RAF aircraft within two days. Without air support, the retreating British troops had no hope of holding up the advancing Japanese. They were forced into a long, ignominious retreat down the Malay Peninsula.

  Defeat followed upon defeat. One of the earliest and most demoralizing was the demise of Force Z on Wednesday 10 December, only a few days after the Japanese onslaught began. The British codebreakers monitored the Japanese aircraft messages reporting the sighting of HMS Prince of Wales and HMS Repulse but it was too late. Caught out in the open in the Gulf of Siam with no air cover, they were sitting ducks for the 33 high-altitude bombers and 53 torpedo-bombers of the Japanese 22nd Air Flotilla which attacked and sank them.

  At a stroke, the Royal Navy's only two capital ships in Eastern waters had been destroyed. ‘Something like a gasp went round the dining room at BP when the 1 p.m. news opened with the announcement that the Prince of Wales and the Repulse had been sunk off Singapore by J air attack,’ Malcolm Kennedy wrote in his diary that day. ‘A terrible blow just at this time when with the US Pacific Fleet so severely hit at Pearl Harbor we are so badly in need of every ship we can get.’

  Three days later, one of the Far East Combined Bureau's outstations, a direction-finding unit at Penang on the west coast of Malaya, was forced to pull back. On Christmas Eve another team of direction-finders at Kuching, in northern Borneo, had to destroy their equipment and flee by sea. The detachment left at Stonecutters Island was not so lucky. Despite having been warned by the code-breakers that a Japanese army division was preparing to take Hong Kong, the military authorities preferred to believe the Japanese would not bother with the British colony. It was forced to surrender on Christmas Day. The four intercept operators destroyed their direction-finding and wireless equipment, but they and Alf Bennett were captured by the Japanese and spent the next four years in Japanese prisoner-of-war camps.

  This was a fate that could not be risked for the members of the Far East Combined Bureau itself. They knew far too much about the breaks into the Japanese ciphers to be allowed to fall into enemy hands. Some of the family members had already been sent by ship to Australia and it was initially suggested that the unit should reform there, but eventually it was decided to evacuate the naval section to Colombo, the new headquarters for the Royal Navy's Eastern Fleet, where it was thought reception would be better.

  The army codebreakers were to go to Burma and mainland India while the army and RAF wireless operators stayed behind to monitor the Japanese air attacks. Arthur Cooper, a Foreign Office code-breaker who had lived in Japan for some time and whose brother Josh was head of Bletchley Park's air intelligence section, volunteered to stay behind. The Cooper brothers were both known for their brilliance: Arthur was the younger of the two and was rumoured to be able to complete the Times crossword in his head. When working at Bletchley Park, later in the war, he was said to have watched colleagues playing tennis each day for several weeks, analysing their technique. Then, never having played before, he borrowed a racket and performed like a veteran. Cooper was joined by a Royal Navy interpreter, Lieutenant-Commander E. H. M. Colegrave, four Royal Army Service Corps clerks and Lieutenant Norman Webb, who had worked for a Japanese subsidiary of Shell in Tokyo before the war. Webb had been commissioned into the army on the outbreak of the war in Europe.

  The radio security of the Japanese naval bomber aircraft attacking Singapore was not high, Cooper recalled. The codebreakers were soon able to work out that specific units, recognizable by the nicknames they used, specialized in certain targets. ‘Thus when the unit airborne was recognized its probable target could be foretold, while from counting the number of individual calls and acknowledgements the strength of the forces airborne could be assessed. Results were telephoned to the local RAF defence and several successful actions were fought. The RAF became enthusiastically appreciative.’

  Cooper, Colegrave and Webb were assisted by the capture, from one of the Japanese aircraft downed as a result of their work, of a codebook for the kuuchi renraku kanji-hyoo 1-goo three-figure air–ground liaison code used for air–ground communications. They stayed in Singapore until almost the last moment, getting out on a ship to Java on 11 February 1942, just four days before Singapore finally fell to the Japanese.

  The fall of Singapore was a disaster of immense proportions for the British. The island was supposed to be an ‘impregnable fortress’, defending the Empire against all attackers – ‘the Gibraltar of the East’. Its capture by the Japanese was described bitterly by Churchill as ‘the worst disaster and largest capitulation in British history’. He was not to know that within weeks, the codebreakers would have sown the seeds of revenge, turning the tide of Japanese victory and putting them continuously on the defensive for the rest of the war.

  9

  THE AMERICANS TAKE THE LEAD

  The news that the naval codebreakers were to move to Colombo sparked a series of weddings and a rush to crate up the unit's records. On 2 January, the day after Rene Skipp's wedding, they began loading the crates and boxes of personal belongings on to lorries and taking them to Keppel Harbour, where the troopship HMS Devonshire was waiting. The Temporary Women Assistants were given the option of going to Colombo with the codebreakers but most decided to stay with their families. Some of the staff took their cars with them to the docks in the hope of shipping them to Ceylon on board the Devonshire. The captain initially refused but finally agreed to take twelve of the codebreakers’ cars as deck cargo.

  ‘We left Kranji on 5 January 1942, on board the Devonshire, accompanied by HMT Lancashire, two destroyers and two cruisers,’ Joan Sprinks recalled. ‘As the Japanese had by this time driven a long way down the Malayan peninsula, we were unable to take the usual route through the Straits of Malacca and headed south through the Banka and Sunda straits. The food was pretty grim, the bread and porridge being full of weevils, but who were we to complain about such a minor detail?’

  The voyage caused something of a problem for Lillie Feeney's attempts to keep her marriage secret and ensure the navy did not split her up from her new husband. ‘When we got on to the ship to Colombo you had to say if you were married or single and so I had to tell my senior officer. She said, “I know.” The best man had asked for leave and had to explain why. But we were very lucky, I think; because of the shortage of kana Morse operators, they were very good, they let us live out together.’

  The codebreakers arrived in Ceylon on 14 January and moved into Pembroke College, an Indian boys’ school about two miles inland from Colombo which had been requisitioned as a combined codebreaking and wireless interception centre.

  The officers stayed in hotels while the clerks, wireless operators and Wrens were in hostels and boarding houses. All but the Wrens were free to find their own alternative accommodation. ‘We were quartered in Galle Road, Colpetty, near Colombo, very near to the sea,’ said Joan Sprinks. ‘The kitchens were most unhygienic with piles of raw meat covered with flies lying on the floor.’

  Ae
rials were draped over palm trees, wireless sets installed, and within five days the intercept station was up and running. The code-breakers started to try to rebuild the JN25b codebook using messages collected before departure from Singapore and what messages were being intercepted currently. They re-established communication with the codebreakers at Corregidor and began exchanging results. ‘Working conditions were quite good,’ recalled Joan Sprinks. ‘One of us would leave half an hour before the change of watch to “shake” the relieving operators and this involved walking past long lines of rickshawmen sleeping near their vehicles, with only their coconut oil lamps to light our way.’

  But while the personnel themselves had arrived safe and sound, some of the equipment had not. The bulk of the Hollerith tabulating machine was eventually traced to the dockside but a key part was missing and an officer had to be sent to Bombay to borrow one from the Indian State Railways.

  More worrying was the non-arrival of a Purple machine promised to the codebreakers several months earlier by Bletchley Park. Shaw sent an urgent signal to London enquiring as to its whereabouts.

  It was learned from Admiralty that the case containing the machine had left England in a warship and was trans-shipped at Durban to the freighter Sussex, although ordered to be conveyed only in a warship or military transport. At the time of the enquiry the Sussex was at an Australian port and in reply to a wireless message the master said that he had landed the case in question at Singapore at the end of December 1941 and obtained a receipt from the Naval Stores Officer.

  A frantic signal to Singapore failed to find any trace of the crate containing the Purple machine. The Naval Stores Officer denied all knowledge of it. The Master of the Sussex insisted he had left it in Singapore. ‘By this time the situation in Singapore was chaotic and it could only be hoped that if the case were there it had been destroyed or dumped in the sea by a demolition party.’

  The alternative was too awful to contemplate. Had the Japanese found the Purple machine, the Allies’ insights into the intentions of both Tojo and, through Oshima, Hitler would have been lost and the fledgling exchange of signals intelligence between Britain and America stifled at birth. Sadly, it would not be the last time that the precious secret of the extent to which the Allies were reading the Japanese codes and ciphers was to be put at risk.

  By now Singapore had fallen and the three codebreakers left behind had gone to join the army operators who were at Kamer 14's base in Bandung on Java, Geoff Day recalled:

  We arrived on Monday 2 February to a very heavy storm. The streets of Batavia were flooded and we took a three-and-a-half-hour train ride to Bandung. Fourteen of the seventeen-man rear party arrived on Tuesday 17 February. These boys were given quarter of an hour to get to a boat, the Wang Chow, which was loaded up with high explosives and which they themselves had to stoke. It took them four days to reach Batavia. Six of the rear party had been sent to work as a mobile DF section at Serangoon village on the north-east shore of Singapore island and three of them were left behind. Another man went missing in Batavia, but turned up later, having been ‘jugged’ for alleged desertion and failure to destroy secret papers.

  Two days after the attack on Pearl Harbor, the US Navy's Hypo codebreaking unit at Hawaii was ordered to forget the Flag Officers’ code and join the attack on JN25b. At the same time, both Corregidor, in tandem with the British codebreakers at Colombo, and OP-20-G, the US Navy's codebreaking headquarters in Washington, had increased their own efforts to reconstruct the code-book, leaving the US Army codebreakers to provide the main effort on Purple and other diplomatic ciphers. By the end of January there was a marked increase in the additive recoveries.

  As competition built up between the various Allied intercept sites to produce operational messages, Colombo found itself left behind. Their very closeness to the Japanese theatre of operations alone meant that the US Navy codebreakers on Corregidor led the way in breaking messages, Shaw recalled. ‘Corregidor's output increased at an amazing rate. The telegrams from Corregidor were coming in such numbers that, even with augmented staff, we had difficulty in keeping pace with decoding them and applying the recoveries.’

  The move from Singapore crippled the British attempts to break the Japanese naval codes. The number of messages they were able to intercept dropped dramatically. The loss of the Temporary Women Assistants caused problems; so too did the initial British reluctance to embrace machine technology. Breaking JN25 by hand was extremely labour-intensive. Although contact with their US counterparts at Corregidor had made the British codebreakers realize the potential of the Hollerith tabulating machines, the Admiralty did not, and continued to ignore their repeated urgent signals requesting a replacement for the missing part. The final straw was an outbreak of dengue fever that laid low the whole of the JN25b codebreaking section shortly after the arrival in Colombo.

  The British codebreakers had managed to penetrate the new additive cipher before they left Singapore. But the move had broken the all-important continuity and, despite a number of crucial successes, it would be another two years before they would fully catch up with the breaking of JN25, said John MacInnes, one of the GC&CS codebreakers based in Colombo.

  The original work on this had all been British but from the start of cooperation with Corregidor in 1941, the burden was carried more and more on the broad shoulders of the US stations. Work on the British side was badly dislocated by the move to Colombo. The loss of depth on the cipher table caused by the break in interception during the move, and the subsequent reduction in volume, greatly hampered stripping in bulk.

  It was at this time that the US Navy first took the lead in cryptanalysis. The unit at Corregidor had, in return for its small size and its physical discomfort in the bowels of the fortress, the advantage of a very large intercept of clean messages, as it was at the geographical centre of the main Japanese operations. The unit made full use of this and the volume of cipher recoveries received from it in Colombo was very large.

  As Corregidor began to evacuate its staff in the face of the advancing Japanese, the Hypo site at Hawaii took over its mantle as the lead station in breaking JN25. Hypo had just moved into the cellar of the Naval Headquarters at Pearl Harbor, dug deep into the volcanic rock. It was a completely open-plan office apart from one dividing wall to protect the codebreakers from the noise of the tabulating machines. The officer in charge was the ‘tall, thin and humorously caustic’ Commander Joe Rochefort, the leading Japanese expert in US naval intelligence for more than a decade. Born in Dayton, Ohio, he had served as an ordinary enlisted man during the First World War. He was commissioned in 1919 and six years later put in charge of the US Navy's cryptologic section.

  Rochefort took over the Pearl Harbor Hypo site in June 1941, inheriting around fifty intercept or DF operators and an intelligence staff of twenty codebreakers and analysts. He transformed the station, turning it from a sedate peacetime organization into a bustling twenty-four-hour operation. But the decision from on high that it concentrate on the Flag Officers’ code and traffic analysis, leaving JN25 to Corregidor and the British, seriously hampered its ability to produce intelligence. Let loose on the new JN25b, the Hypo codebreakers soon showed their true worth. For the next three months, Rochefort rarely left the office. He slept there and ate there, kept going by US Navy-issue amphetamines, and even when working was invariably to be seen wearing a silk smoking jacket and slippers. Rochefort insisted that there was a logical explanation for this eccentric attire.

  I'd spend all of the time in the office, putting in around twenty or twenty-two hours a day. I would only leave when I had to leave and that was just about all. This went on for about forty-eight hours at a stretch. So it was a very fantastic operation down there. I started to wear a smoking jacket over the uniform. It was a sort of reddish smoking jacket somebody had given me. The main reason for wearing it besides keeping me warm was that it had pockets where I could keep my pouch and pipe. Then my feet got sore from the concrete floor we ha
d down there. So I started wearing slippers because the shoes hurt my feet.

  Long-term research was dominated by Washington with its larger resources both in terms of personnel and access to tabulating machinery. But while OP-20-G led the way in reconstructing successive JN25 codebooks, and in building up the necessary database on which the outstations would come to rely, it was Rochefort and his Pearl Harbor codebreakers who dominated the race to break messages in real time.

  By mid-February Malaya, including Singapore, Hong Kong, much of the Philippines, Guam, Wake and Sumatra were in Japanese hands. On 19 February, the 1st Air Fleet, the force that had attacked Pearl Harbor, struck a crippling and demoralizing blow against the northern Australian town of Darwin. A few days later the Allied codebreakers reported a build-up of Japanese naval forces in the Java Sea, apparently preparing to land troops on Java itself. On 27 February a joint US, British, Dutch and Australian naval force under the Dutch Rear-Admiral Karel Doorman intercepted the Japanese ships. Doorman made an audacious but ultimately unsuccessful attempt to outsmart the Japanese, going down with his flagship RNNS De Ruyter. The Allied force lost two cruisers and several destroyers. Japanese control of the sea and air around Java was now total; within days their troops were pouring on to the beaches with little or no effective opposition.

  Two weeks later, General Douglas MacArthur, the commander of the US forces in the Philippines, pulled his men out, famously vowing to return. But some of the Cast codebreakers remained in the tunnel at Monkey Point until the very last moment, when they were taken off by submarine and ferried to Melbourne, where Lieutenant Rudi Fabian, the Cast unit's commander, had already set up a new intercept operation. Fabian had been born in Butte, Montana, in 1908. After a year spent at the Montana School of Mines he had entered the US Naval Academy. Known to his men as an often unreasonable ‘hard charger’, Fabian was a purely administrative head, in charge of a number of codebreakers with more senior rank. He was to create a number of ripples within the as yet undeveloped Australian sigint community and the greater field of Anglo-American collaboration.