There were three main codebreaking sections at Kilindini, one working on JN40, which consisted of three codebreakers, a pensioner clerk and two locally employed female clerks; a second working on the Fleet Auxillary System, JN11, a derivative of the merchant-shipping code, JN40, which had taken over the most productive traffic; and a third and very much larger section working on JN25 and led by the Dutch naval officer Lieutenant-Commander Leo Brouwer and the veteran Royal Navy codebreaker Lieutenant-Commander George Curnock. This had five codebreakers assisted by three pensioner clerks and seventeen female clerks. All deciphered material was channelled through an intelligence section run by Lieutenant-Commander Neil Barham.
Cohen recalled that he and Denham were set to work on Kilindini's revived JN25 operation under Brouwer and Curnock, rebuilding the codebook.
We did this in different ways. The best, of course, the nicest way is if somebody has been put in the way of stealing or somehow picking up this thing. The most direct way they might have done that was by a copy of the code being left on an island that the Japanese had evacuated. Then you might have received details of the individual code groups from the Americans. But you'd be lucky if you got them all. So what did we do? Well, we might know that a certain type of message at a certain time of day was being sent out from the Commander-in-Chief, or it might fit in with weather, or with some minor battle or engagement that had been going on. So there were these various things that we might expect the Japanese to be talking about and that would be the kind of thing that gave us a clue.
Then there might be patterns in the text of the message. You might find a message that echoed another message that had been sent a long time before and so then you might test if it was right to decode this new message in that way. There were some messages that were sent by a particular station at a certain time of day that were always the same. However strong the warnings sent by the Japanese intelligence people to their juniors that this was insecure, they still broke the rules and did things like send the same message at the same time every day.
The ability of the two new young codebreakers proved the wisdom of Tiltman's decision to implement the six-month Bedford course and impressed the older hands like John MacInnes. ‘These two very soon found themselves well in the hurly-burly of translating a mass of messages with an inadequate staff, and rose to the occasion,’ he recalled. ‘Their experience from April to July 1943, combined with their great ability, made them into translators and book-builders of a standard scarcely reached by later arrivals who did not have to face such a press of work.’
Both at Bletchley and in Kilindini, the new young graduates coming in found it difficult to relate to many of the older pre-war codebreakers, Cohen said.
There were, of course, all sorts in the navy who were recruited as intelligence officers or interpreters or whatever who might not have seemed to be the brightest ones around and, if you can imagine the difference between us in our very early twenties and these other people who were in their forties or fifties, they were senior to us but were not respected by us in the way they should have been. So there was some tension there. But we felt we still had a long way to go and the best people at the game were the ones who taught us how to get on with it. George Curnock in particular was just brilliant at the game, very intelligent. He was a three-ringed commander who had been in the FECB and would ‘gin himself up’ before work. But he had a very powerful mind and could talk, and sing, irrespective of how many gins he had.
Outside of work, Kilindini was a very relaxed place to be, Cohen recalled.
Mombasa was lovely, a very nice climate. The junior officers’ mess was right on the sea front with the roar of the sea hitting the rocks. We were members of the yacht club which was allowed to sail into the harbour. We made an effort to climb Kilimanjaro. Hugh did it and I dropped out 1,000 feet from the top. Then there was the beach. Nowadays, you see Mombasa as a tourist resort but in those days it was unspoilt. Once a week there was a classical music concert, records, of course, up on the ramparts of Fort Jesus, the castle that the Portuguese had built. We all sat out on the tropical night listening to classical records.
The spring and early summer of 1943 saw a gradual change in OP-20-G's attitude towards co-operation on Japanese naval codes and ciphers. Whether this was a result of the strong views being expressed by the Admiralty or the marked increase in the number of useful messages coming out of Kilindini which were not available to the Americans elsewhere is not clear. But at the end of March OP-20-G agreed to institute a weekly US courier service of recoveries and other findings between Washington and Kilindini. Melbourne was still holding back ‘crucial material’ as late as March but, as signs of a US rethink began to emerge, Fabian was recalled to Washington for discussions.
In May 1943 Bletchley Park reached a groundbreaking cooperation agreement with the US Army. The BRUSA Agreement is widely seen as the predecessor of the post-war UKUSA Accord which continues even now to ensure that British and US signals intelligence operations are very closely co-ordinated. It had not been reached without a great deal of acrimony over the vexed issue of the extent of US Army involvement in the breaking of the German Enigma machine ciphers. But with respect to the Japanese it did little other than to rubber-stamp the willingness of both sides to co-operate fully.
On the naval side, however, despite the apparent thaw in relations between OP-20-G and Bletchley Park, considerable difficulties remained. Joe Redman and Wenger were invited to a series of conferences at Bletchley and the Admiralty during July and August at which the British pushed for better co-operation on Japanese naval matters. The Americans agreed to pass urgent material and all recoveries to the British. The move from Kilindini to Colombo was now only weeks away and Redman promised that the new station could have an American cipher link direct to Melbourne, controlled by a US naval liaison officer. The problems appeared to be over.
Not the least important result of the apparent end to a damaging dispute that had lasted for more than a year was the arrival for the first time at the Admiralty, South-east Asia Command and HQ Eastern Fleet of substantial numbers of US Navy signals intelligence reports of interest to the British.
Another interesting skirmish in the final days of the rows over cooperation involved claims made by a number of British officers that a small Royal New Zealand Navy intercept operation which ostensibly came under Fabian's control was not being given enough ‘guidance’. The section was run by a Lieutenant Philpott, assisted by Professor Campbell, Professor of Mathematics at Wellington University, a part-time civilian Japanese interpreter and two women assistants, ‘both of whom were above average ability and one of whom knew quite a bit of Japanese’. Interception, DF and radio-fingerprinting were carried out by New Zealand Wrens at a number of sites in the South Pacific, extending as far as Suva in Fiji, and despite its small size and apparent lack of assistance or direction from Melbourne the section was remarkably productive.
New Zealand was not the only Commonwealth country to do what it could to intercept Japanese naval communications. The Canadian signals intelligence organization, the Examination Unit, whose main target was diplomatic and commercial traffic, also intercepted Japanese naval communications, including JN25 material picked up at its intercept site in Esquimalt. The Examination Unit received assistance with decoding from Bletchley Park, which stationed a senior codebreaker in Ottawa on a permanent basis to control operations, but its naval intercepts were co-ordinated by OP-20-G, albeit with a similar lack of control and guidance to that exercised by FRUMEL over the New Zealand operation.
There was also an ‘unofficial’ intercept operation in Mauritius, which was set up and run entirely of his own volition by Edward Twining, later Baron Twining and a member of the tea family, who was the island's chief censor and information officer. His civilian staff worked under the cover of the colony's censorship department, and although the extraordinary size of his organization led to local criticism that they were a waste of public money and wer
e engaged in everything from censoring letters to ‘fixing the prices of cabbages’, their true purpose appears never to have been suspected.
The main codebreakers were a local Mauritian civil servant, a botanist from the colony's agricultural department who had studied Japanese, a former headmaster's secretary who had ‘a crossword-puzzle mind’ and a trained chemist who came in on a part-time basis. There were a number of clerical assistants, described by Twining as being recruited from ‘good Mauritian families who, if their standard of intelligence is not high, are keen, hardworking, sound and adequate’, and the intelligence passed to London, which was deemed by the Admiralty to be ‘of considerable value’, was enciphered before transmission by wives of local British officials.
That the operation, codenamed CHESOR, was never compromised is all the more remarkable given that, in Twining's words, the eighty intercept operators were ‘mainly drawn from the colonial community and are an undisciplined lot of ragamuffins. We possess no disciplinary powers except dismissal, which in most cases is undesirable.’ Order was kept by force of Twining's personality and a daily bonus of between three and five rupees for turning up to work and satisfactory conduct while there.
At Bletchley Park, Hut 7 was expanding rapidly as more and more staff from the Italian naval section moved across to work on Japanese material and new recruits arrived, many of them Wrens like Rosemary Calder, who recalled being put to work in the Japanese Navy traffic analysis section which was run by the Cambridge historian Sir John Plumb.
I was interviewed by Jack Plumb, who told me, ‘we analyse traffic’. I had no idea what this meant. I had this picture in my mind of people sat on camp stools by the side of the road counting lorries and gun carriages. I spent most of my time at BP attached to the room of which Angus Wilson the famous novelist was head. We considered ourselves to be a small, exclusive group who were all given scope for initiative and intelligence despite the bulk of the work being of a repetitive clerical nature. Any of us could do any of the jobs in the office. It was a very democratic place. Wrens mixed up with civilians. We might as well have not been in uniform. We were having a marvellous time. It was like being back at college.
Angus was a great darling who spoilt us all and we spoilt him in return. He called us all Ducky and he had this special friend called Bentley Bridgwater who took over the traffic analysis section from Jack Plumb and later went on to become Secretary of the British Museum. Angus was known to be very brilliant but crazy. He had at least one nervous breakdown before I got there and was still going to Oxford to see a psychiatrist, writing all his dreams down. But he was very good-natured most of the time and if he started getting agitated, we would just give him a copy of Vogue or Tatler and he could go off and sit down by the lake flicking through it and come back as happy as a sandboy.
Anne Petrides, another of the young Wrens in Hut 7, worked on an index of merchant-shipping movements.
I joined naval section at BP the day after my eighteenth birthday ‘celebrated’ at the WRNS training centre at Mill Hill on 31 May 1944 and was flung into the work of cataloguing ships, entering brief notes on their cards about the decoded signals as they came to us from the translations. Most of their warships had been sunk by then and we were dealing with maru – merchant ships.
The naval officers in my office included Gorley Putt and ‘Shrimp’ Hordern, brother of Michael Hordern, the actor. As a very young girl I was petrified to be left all alone at lunchtime, in four interconnecting rooms – and in fact justifiably so, as a senior officer from the Admiralty phoned and said, ‘Can we go over…?’, followed by a burble of words. He came back in clear language and was outraged to find that not only had no-one seen fit to tell me which button to press for the scrambler but that no ‘duty officer’ was present.
A regular visitor from ‘down the passage’, usually on quieter night watches, was Angus Wilson. His first book of stories was said to have originated in a series of sessions he had with a psychiatrist. I believe the men cracked more easily under the strain, whereas girls found it easier to have a crying breakdown. We Wrens were extremely spoilt in our accommodation, nothing but the best country houses in the area, including Woburn Abbey, while the ATS lived in barracks at the back of the park. I started out at Wavendon House. Then I lived at Stockgrove Park. It had been rather knocked about by the 51st Highland Division which had been there before us. I remember dances attended by locally billeted GIs and drinking draught cider. Very heady.
The authorities at Bletchley Park also continued to rely on the Oxbridge old-boy network to talent-spot new recruits. Norman Scott was reading Mathematics at Brasenose College, Oxford, when his tutor Theodore Chaundy, who had himself spent some time at Bletchley Park early in the war, recommended him for work on Japanese codes. Scott was assigned to Hut 7 to work on the relatively new JN11:
There were two familiar faces in the group: Patrick Taylor had been at Brasenose with me and I knew Harry Field from New College. I also recall meeting Hugh Foss, the head of Hut 7. Apparently, he had been in the code and cipher-breaking business for many years. As a ‘break-in’ guru, his talents were not those needed for the exploitation of partially cracked ciphers. It was not long before all of us working on JN11 moved to the Japanese naval block to become part of Commander McIntyre's group which already dealt with JN25 and other Japanese naval codes and ciphers.
Most of my working time was spent in a long, narrow upper-floor room with one door to the corridor. This accommodated all of us except our sub-section chiefs Brian Augarde and John English. Both were lieutenants in the Army Intelligence Corps. Immediately opposite our door was an equally large room for the JN11 translators headed by an army captain. His chief assistant was a Bedford-trained RAF sergeant named Harris-Jones. He would walk over with his needs for missing portions of a partly deciphered message and leave us with helpful suggestions on how we might tackle the blank stretches on our worksheets.
JN11 was similar to the JN25 code used by the Japanese Fleet proper. However, it was much simpler to use. (Merchant Navy wireless operators could not be trained as well as their counterparts in the Imperial Navy.) When I joined the team many of the JN11 code groups and stretches of the first book of additive tables had already been identified. Looking at our worksheets, one could see the human factor in play. It was easier for the sender to use the right-hand page of the random-number books to copy out the additive groups below those in his encoded message. Very few messages exceeded 100 groups. To save turning the page, the encipherer would start at or near the top-left of the page. Positions 90, 01, 10 and 11 were very popular. Finally, if there are three books to choose from, it can be expected that the first book will be used more often.
As messages came in and had their starting positions determined, they would be copied out by hand on to wide sheets of fifty columns with three or more vertical spaces between each message. At the top of each column would be the additive, if known. If it were known, it would be added to each enciphered group in the column to yield the underlying code group. Beneath that code group would be its meaning in Japanese if this meaning were known. The number of messages under each column heading was known as ‘depth’.
The most common first code group of a message was 7879, meaning hatsu, ‘from’. After that, 4276 meaning maru, ‘merchant ship’ could be hypothesized in the next few columns. Then there were pairs like 2269 for ‘open bracket’ which would lead to a search of following columns for 0328, meaning ‘close bracket’. If hokui, ‘North’, showed up, one could search for tookei, ‘East’, to follow. Then there were the Japanese grammatical suffixes ha (nominative case), wo (accusative case), no (genitive). The group 5566 for shuushifu, meaning ‘full-stop’, often followed the ‘from so & so’ opening.
Like JN25, the JN11 code had a built-in error check. The sum of the four figures would be divisible by three with one over, and this helped the codebreakers to check that their calculations were correct, Scott said.
For instance, on a de
pth of three, hypothesizing 4276 as one of the underlying code groups would yield acceptable code groups by chance in only one in nine tries. As more of the code group meanings were established, even arithmetically acceptable code groups whose meaning was still undetermined could be doubted as a valid solution for the additive.
Very much a part of the scene were the young women who performed, manually, a great deal of the work. As I recall, all but two Wrens were civilians from the London area. They copied out new messages on to the single set of 600 wide sheets, each with fifty columns. They wrote in additives received from the US Navy team in Washington and applied them to the enciphered groups in the column. They kept our codebooks up-to-date as new meanings came from the translators across the hall and from Washington. If and when these tasks were all completed, they would ‘drag’ commonly used code groups through wide stretches of unbroken messages where some depth existed. The process of discovering the 30,000 additive groups was akin to stripping away the disguise which hid all the code groups in any one column. Consequently the process itself was known as ‘stripping’. This led, of course, to the term ‘stripper’ for a person who did it. Our hardworking helpers declared it a great pity that they could not tell their families and friends that they worked at Bletchley Park as strippers.
Olive Humble, one of the Temporary Women Assistants drafted in to work as ‘strippers’, was put into another sub-section of Hut 7. She had been called up in early 1943 for the WRNS but there were no places and she found herself sent off instead to the Foreign Office as a civil servant.
So in February 1943 I arrived at BP and was escorted to the Billeting Office by an armed soldier, to my great consternation. I had never left home before, having worked in an insurance office in the City when I left school. I was suitably impressed with my new surroundings, until I saw the Mansion, which no-one can say is beautiful to the eye.
The Emperor's Codes Page 24
