A reorganization of Bletchley Park, designed to make the operation more efficient, had seen Commander Edward ‘Jumbo’ Travis put in charge and Denniston moved sideways. He retained control of the diplomatic section, which was moved back to London to make room for the new codebreakers needed for the service sections. The decision infuriated Malcolm Kennedy, who complained in his diary that he had only just bought a new house in nearby Woburn Sands.
Orders definitely issued for our branch of the FO to return to London next week. Feel very sore about it as the move has been manoeuvred by certain ‘interested parties’ and by gross misrepresentation of facts while AGD [Denniston] is so utterly spineless that, on his own admission, he has made no attempt to point out all the serious snags and difficulties involved. No small portion of the personnel who, for one reason or another, are unable to go have had to resign and for many of us the move will entail 4 to 5 hours’ travel daily to and from work. Yet AGD has the brass to contend that efficiency will be increased ‘in the tenser atmosphere of London’!
The British naval codebreakers in Colombo had solved some of their manpower difficulties by recruiting an extra fifty Temporary Women Assistants from among the local military and expatriate communities. They had also acquired the services of one of Kamer 14's codebreaking experts, Shaw recalled. ‘When Batavia fell, the Dutch naval C-in-C with a few key personnel, including Lieutenant-Commander Leo Brouwer RNN, flew across the Indian Ocean and, failing to find Ceylon, landed in India with their last pint of fuel. They refuelled and flew back to Colombo. Lieutenant-Commander Brouwer, a Japanese linguist, joined our JN25 team where his Special Intelligence experience and knowledge of the area around the Dutch East Indies were of great use.’
But there remained an acute shortage of Japanese linguists both at Bletchley and in Colombo. John Tiltman, the head of Bletchley Park's military section, consulted the School of Oriental and African Studies in London, which had already been training a small number of RAF Japanese interpreters. The SOAS experts told Tiltman that it took five years to train someone properly and two years was the absolute minimum required for any kind of decent standard. This was clearly far too long so Tiltman decided to set up an ‘experimental course’ of only six months using young men who had already shown aptitude for learning unusual languages as potential codebreakers. ‘I was advised to recruit classical scholars of from eighteen to twenty years of age from Oxford and Cambridge,’ Tiltman said. ‘For a short but glorious period I achieved considerable personal popularity at both universities because I was the only person who wanted classical scholars because of their attainments in Latin and Greek.’
Jon Cohen was reading Greats at Balliol when the Japanese launched the attacks on Malaya and Pearl Harbor. Shortly afterwards, he was asked if he would like to learn Japanese.
A lot of the British people who knew Japanese in the Diplomatic, Consular, and Colonial Service and suchlike had been rounded up by the Japanese in places like Hong Kong and Shanghai. So there was a shortage of people who could translate Japanese in the Government Code and Cypher School. I was approached by the master of my college, Sandy Lindsay, who had no doubt been asked by Colonel Tiltman to recruit some people who might be capable of learning Japanese in a very short space of time. He didn't tell me anything about what I was to do and I don't think he knew.
The School of Oriental and African Studies said it would take two years to teach anybody Japanese. But Tiltman had taught himself in a few months and thought it would be possible with students like myself. So we were recruited and went first of all to Bedford to the Gas Showrooms at Ardour House where about two dozen of us, some from Cambridge and some from Oxford, were taught Japanese by an elderly captain in the navy. Oswald Tuck taught us, I think brilliantly, for a period of about four months.
Tuck, who had served in naval intelligence during the First World War and was now sixty-five, was brought out of retirement to run the course. Tiltman asked him if he felt he could manage to teach young students enough Japanese to break codes in six months. ‘The idea sounded impossible but was worth trying,’ he noted in his diary. Tuck had left school at the age of fifteen to go to sea. He had learned to speak fluent Japanese, of his own volition and very much against the mood of the time, during a spell on the China Station, and subsequently served as Assistant Naval Attaché in Tokyo.
There were about twenty-three students on the first course, most of them classicists and including only one woman. They studied from 9.30 to 5 on weekdays plus Saturday morning. Tuck was a natural teacher, able to instil his own enthusiasm for the Japanese language in his students and willing to employ a number of innovative teaching methods. The youngest student on the first course was Maurice Wiles, who had been recruited by his brother's tutor at Cambridge.
My elder brother Christopher had been up for two years at Christ's before the war and had won the Porson Prize for Greek verse two years running. When his tutor Sidney Grose recommended him to Colonel Tiltman, he recommended me too. So we started at this course in Bedford, above the noisiest crossroads in the town, and in walked Captain Tuck, who was a very small, very dignified ex-naval man with a neat white beard and this old-world civility which seemed old-world even in those days.
Tuck's opening gambit was to tell his new students that they had not greeted him properly. ‘When I come into the room,’ he said, ‘you are all to stand up. I shall then say “shokun ohayo,” which means: all you princes are honourably early. You will then reply, “ohayo gozaimasu,” which means: honourably early it honourably is. I shall now leave the room and come in again and we shall do this.’ Tuck's teaching methods were innovative. Every day the students moved back one row, with those in the back row moving to the front.
Prior to taking on the Bedford course, Tuck had worked at the Ministry of Information, censoring dispatches by Japanese journalists based in London, and these provided the students’ texts. ‘We did a lot of learning and practice on our own,’ said Wiles, ‘but Tuck was very effective in getting us into the language. The pressure of the war and the knowledge that many of our peer group were in circumstances far less comfortable than ours rather concentrated the mind…’
The students’ spare time was spent in the pubs of Bedford and in privileged access to performances by leading classical musicians, recalled Jon Cohen. ‘The BBC moved its classical music section to Bedford and they gave the services free tickets, so we had endless wonderful first-hand music.’
After about four months, Tiltman decided to send two of the students to the Japanese diplomatic section which had now moved back to London and was based with the rest of the GC&CS diplomatic and commercial sections in Berkeley Street, off Piccadilly. ‘The question was whether the course was good enough,’ Cohen said. ‘So two of us, myself and another, were sent to work in this office in London for a short while to be tested by the people who were long-term interpreters into Japanese. We seemed to satisfy them.’
The success of the course sparked a mixture of disbelief and embarrassment at SOAS, which had suggested it could never be done, said Wiles. ‘They had always said that the war would be over before anyone was anywhere near good enough at Japanese to be useful. When their experts tested two of the students at the end of the course, they were amazed.’
Bletchley Park's air intelligence section under Josh Cooper ran its own special short-term Japanese course since it required a different kind of linguist capable of listening to the clear speech passing between pilots and their ground controllers. Cooper was one of the more eccentric codebreakers at Bletchley Park, a fact reflected in the course he selected for his students of Japanese. It was described by Tiltman, with some understatement, as ‘a rather more tricky experiment’ than the Bedford codebreakers’ course.
What the Royal Air Force needed were interpreters who could read air-to-ground and air-to-air conversations. For this purpose my counterpart in the air section, J. E. S. Cooper, started an intensive eleven-week course at which the students were bombarded inc
essantly with Japanese phonograph records, ringing the changes on a very limited vocabulary. The course was directed, not by a Japanese linguist, but by a phonetics expert. I remember taking a US Army Japanese interpreter, Colonel Svensson, round the course. Stunned by the volume of sound in every room, Svensson mildly asked the Director whether all the students made the grade and the reply he received was: ‘After the fifth week, they're either carried away screaming or they're nipponified.’
The American codebreakers were also desperately trying to recruit new personnel to be sent to the various outstations. Japanese-speakers were the obvious priority but they also needed more wireless operators to intercept and DF the Japanese Navy's radio stations. So urgent was the need for reinforcements, Rochefort recalled having to recruit navy bandsmen to operate his punch-card machines behind the dividing wall in the Hypo basement.
Our whole operation depended on IBM machines. We'd just assemble the cards and we would be able at any moment to pick out any group or two or three groups, or half-a-dozen groups, in a message. Suppose the message said, A, B, C, D, E. If we wanted to use this for some reason or another, all we'd tell the officer in charge of this operation, this IBM thing, ‘Give me all the A, B, C, D, Es you have,’ and he'd run this whole thing through the collators, through the tabulators, and he would then come up with printed forms where the use of this thing had ever been made together with the messages from the groups ahead of it and, of course, after it.
OP-20-G grew dramatically in size, with large numbers of young sailors straight out of college and Waves, the US Navy's female contingent. There was already a large pool of trained Japanese-speakers to draw on from among America's business community. They were sent on refresher courses set up at Columbia and Yale while a Japanese-language school was set up at the University of California at Berkeley, moving later to Boulder, Colorado, with new students being put through an intensive fourteen-month course.
The US operators were trained to take Japanese kana Morse on top of the Navy Building and as a result became known as the ‘On the Roof Gang’. But after the attack on Pearl Harbor Hawaii was in urgent need of operators. Phillip H. Jacobsen was one of a number of standard US Navy radiomen in Hawaii who found themselves moved on to interception duties.
Near the end of March 1942, I and about seventeen others of those who were at the Transmitting Station were transferred to the Radio Receiving Station at Wahiawa, Oahu, to commence training in Japanese kana code and Japanese naval communications. This first class at Wahiawa was augmented by eight or so radio operators from the submarine base and a musician off the USS California who knew the Morse code.
As an attempt at security, we were all quartered in houses formerly used for married men. It was a bit crowded at first, but it was much nicer than living in the barracks. We had a reserve first-class radioman in our house to supervise us. Six men in double-decker bunks slept in the larger bedroom while I and three others slept in two double-decker bunks in the small bedroom. Having a bathtub was a bit of luxury as the navy only provided showers. Those of use who were still seamen second class were given the third-class radiomen's test and all eighteen of us passed. We bought out all the beer at the navy exchange and held a big party. Apparently, it got a bit out of hand because the Exchange Officer wouldn't let us do it again.
Tiltman travelled to Washington in March to discuss arrangements for joint coverage of Japanese Army and Navy communications with the Americans. His brief was ‘to urge upon them the necessity of a division of labour, and that they should concentrate on Japanese, leaving the German and Italian to us’. Given that the British were way ahead on the breaking of Enigma and rapidly losing touch with the Americans on Japanese naval codes and ciphers, this division of responsibility made sense. A good deal of suspicion remained at higher levels over the burgeoning exchange of signals intelligence. But among those actually working on the Axis codes and ciphers there was little if any antipathy.
Tiltman agreed that OP-20-G should take the lead on JN25, with Fabian's Melbourne unit, the Hypo station at Pearl Harbor and Colombo sending all their messages and findings to Washington, which was to be codenamed ‘Susan’. The British Chief Cryptographer, who was used to working by hand on his own, standing up at a custom-made desk, was impressed by OP-20-G's JN25 section but staggered by the number of tabulating machines in use. ‘It is controlled by a psychologist, Dr Ford, and appears to be extremely well run, if somewhat over-mechanized. This section divides its work with Melbourne and Colombo and the general result is I believe quite satisfactory.’ But the section dealing with all other naval codes and ciphers, which was run by Preston Currier, was not so efficient, Tiltman said. ‘He is a very good man with experience of Japanese naval ciphers extending over at least ten years. But he is the only man in the section who knows Japanese and much of his time is occupied in trying to cope with unreciphered codebooks to the possible detriment of research on the Flag Officer and Submarine ciphers.’
The Americans asked that Bletchley Park attempt to crack the Flag Officers’ codes, JN16 and JN49, and to assist on the breaking of the Japanese naval attaché machine cipher, which had changed in September 1939, and was now codenamed ‘Coral’ by the Allies. In return, they would provide a complete six-month study of JN25 recoveries in order to help the British codebreakers to catch up. Both sides agreed to a full exchange of data.
Liaison with the Signals Intelligence Service was much less productive, largely because the US Army codebreakers had been concentrating on Purple at the expense of the army codes and ciphers. But they also agreed to exchange all their research.
The Royal Navy codebreakers at Pembroke College just outside Colombo were still struggling, but somehow managing to keep pace with the Americans in terms of breaking JN25b. But with the British intercepting far fewer messages and working largely by hand, while the Americans made extensive use of the tabulating machines, it was a ‘long and laboured’ process, said Lieutenant-Commander Neil Barham, one of the leading Royal Navy codebreakers.
The first step was to strip off the cipher additive that had been added to the coded messages. This depended on being able to line up messages that had been sent using the same portion of the additive table or cipher book. Each message had a preamble which told the receiving operators the serial number and priority of the message, who originated it, who it was for, and how many groups it contained. The message itself began with a six-figure group giving the date and time of origin in clear. It was followed by a series of five-figure groups. The last five figures of the date–time group were repeated as the penultimate group and the last group was a repeat of the first five-figure group. Buried somewhere in the message was the five-figure indicator group, which told the receiving station the so-called starting point, the precise place in the additive table from which the originator had begun taking additive groups. In addition to the preamble notification of the number of groups in the message, the receiving station had a built-in garble check – the sum of every five-figure JN25 group was divisible by three. This was as invaluable to the Allied codebreakers as it was to the Japanese.
Finding repeated sequences could be very difficult, particularly with the limited number of messages being received at Colombo. But the insistence, common among most service radio operators, on sending routine messages at precisely the same time every day was invariably the way in. Once the format of these messages had been recovered, they could very often be predicted. A message sent at 0900 hours every morning by an isolated station and with the same low group count was almost inevitably to say that he had nothing to report. Such messages were often the easiest way into the code.
The Japanese mistake in changing codebooks and additive tables at different times meant that either the code groups or the additive in use could be recovered simply by subtracting whichever of these two values was known or could be guessed. Weather messages, which were invariably transmitted in a number of different codes and ciphers, some of which were already broken, were
also a useful way in, as were low-level intelligence reports of enemy shipping movements. Since these messages were always sent in a stereotyped format and the Allies knew where their ships were, such messages were relatively easy to break. From this lowly start, other messages containing some of those known code groups or different parts of the same sequence of additive groups could be recovered.
The codebreakers had to find the indicators and identify as many messages as possible that had been enciphered using additive sequences that were identical or overlapping, said Barham. ‘The lining-up of these messages, or parts of messages, in their correct relative positions on the reciphering table may be termed “forcing”.’
The messages were written out horizontally in rows on a large sheet of graph paper, known as a worksheet, and lined up under each other so that each vertical column provided a ‘depth’ of five-figure groups all enciphered using the same additive group, Barham said.
Once it has been possible to force two or more messages into their correct relative positions on the reciphering table, it is not impossible to discover the subtracting figure common to all groups for a certain position. This becomes a definite possibility if groups in the underlying codebook are known. It is possible to break into the reciphering table, to subtract the group found from the textual group in the message and thus produce the corresponding unreciphered group. This process is known as ‘stripping’.
There were a number of key code groups that the codebreakers would look for. The intrinsic difficulties of the Japanese language, in which one word can have a variety of different meanings depending on its context, meant that the Japanese operators had often to spell things out to ensure they were correctly understood. The only certain way of doing this was to use a commercial code known as the Chinese Telegraphic Code which provided a four-figure group for each ideograph. In order to warn the receiving station that the message was about to start using this system, all the main Japanese codes had special code groups that indicated its use was beginning or ending. This was also true of figures – which were frequently used in predictable places in the proforma sighting reports passed on the JN25 code – and a special code which had individual groups for each of the ships of the Japanese Merchant Navy, known by the code-breakers as the maru, from the suffix used after the name of each ship. So the codebreakers would look at the enciphered groups in their columns for ones that might be hiding the basic code groups indicating the start of the use of the Chinese Telegraphic Code, the maru code or figures. They subtracted this basic code group from each of the enciphered groups in the likely positions until they found one that fitted the pattern. The difference between the two must then represent the additive group. This could be subtracted from all the five-figure groups in that column, providing yet more basic code groups. Some of these would already have been recovered and would provide a clue as to further basic groups in their own horizontally written message.
The Emperor's Codes Page 15