The Bletchley Park Codebreakers

by Michael Smith

  On the other hand, our relationship with the bookbuilding team, led by Michael Loewe, was close. There was to-and-fro consultation at all levels without inhibition or formality, depending only on the problem that had to be discussed: it could be Michael Loewe studying with the newest Wren the depth she was working on, to see what insight they could together find. By contrast we compared notes with no-one outside our JN-25 work area, not even with those on JN-11 who were our close neighbours both in task and on the corridor. Commentators now refer with a certain awe to Bletchley Park’s success in keeping secrets: it seemed only natural to us at the time.

  It is clear in Appendix VI how heavily we relied on the services of the ‘Freebornery’, particularly in providing successive editions of our Indexes of Good Groups and of Differences. Almost nothing has been published about Block C, but an account by Mr Freeborn’s Deputy, Ronald Whelan, has been released. It contains some vivid recollections (for example, about Winston Churchill’s visit) as well as technical information about the machines; and it strengthens the view that the Freebornery’s contribution to Bletchley Park’s overall success is still inadequately recognised.

  Referring to an earlier era, when the Freebornery was in Hut 7, Gordon Welchman commented:

  …some of the cryptanalytical departments had a tendency to try to tell Freeborn how to use his machines in support of each problem….. the cryptanalytical sections would have had a better service if they had simply discussed their needs with Freeborn instead of dictating to him. This would have given him the chance to program the overall use of his equipment and staff in a way that would have been advantageous not only for each individual problem solution, but for the overall service he was providing ….Of course not all the cryptanalytic sections made the mistake of dictating to Freeborn.

  By 1943, in my experience at least, this issue had been resolved in Mr Freeborn’s favour, but a certain tension remained between him and his clients. In several respects he stood apart from the generality of the Park’s professional staff. At my level he was always known as Mr Freeborn: no Christian name, no nickname. He was a business man (borrowed from the British Tabulating Machine Company at Letchworth) and always in a business suit. Though he had middle managers as well as Ronald Whelan to keep Block C running continuously through three shifts, I never met them. Every discussion I had about the work to be done for us was with Mr Freeborn himself and he was in total command of the detail. I always went to his office in Block C, never he to mine.

  The majority of the very numerous Block C staff were civilian women in low machine-operator grades and Wrens. They were with us in the canteen and in the buses (it looked like a regiment when they all walked together under our Block B windows at a shift-change) but I doubt that they mingled or joined much in the leisure activities. Mr Freeborn was fiercely protective of them. Suggestions from would-be customers that an extra bit of work could be wangled in on top of what was already scheduled were ill-advised and resulted only in a loss of his respect. It paid to ensure that the material submitted to Block C was immaculately prepared, for if it was shoddy and took his staff’s time to correct, Mr Freeborn would be harder to persuade to give priority next time. In return, the work delivered back from Block C was always on time and accurate.

  In his ‘A History of British Signals Intelligence 1914–1945’ Frank Birch has commented rather critically on the arrangements for determining priorities for the Freebornery’s services and on the failure to resolve ‘the issue between the Tabulating Section and its customers’. At our level we were more concerned with the outcome of the priorities than with their derivation. There was a general belief that, since Mr Freeborn was obviously working within guidelines from the higher directorate, you could judge the importance that higher authority attached to your work from his willingness to accept your material and the priority he gave it. By this test our JN-25 work increasingly mattered at the top.

  I left the history of our team at November 1943 when we had found our feet, had some success and were beginning to entertain the ambition to take over the smaller B Channel for ourselves. But the Japanese had other ideas. We did not know that we were in for eleven months of turmoil and disappointment before we could feel successful again in December 1944. I will not chronicle here all the changes that were made: they can be found in my 1945 history. Rather I will focus on three events, on 1 December 1943 and on 25 July and 23 October 1944, that had profound consequences for the role of our party.

  On 1 December 1943 the Japanese brought in new additive tables 36 and 38, still with codebook J, and a significantly different indicator system. On 10 January they changed the additive tables again to 42 and 44 and replaced codebook J by G. Washington, helped by the capture of two charred pages of Table 42, was quickly able to establish that the new tables had 501 pages each of 15 x 12 additives in place of the familiar 10 x 10, and that the position of each additive on the page was identified by a line and a column bigram, pairs of randomly chosen digits. They also solved the encipherment of the page numbers. But they could not break the system of two-digit ordinates that determined the starting position on the page.

  New ways were therefore needed to establish depth. Here are three examples of what we did. First, given two messages starting on the same page, we used scoring by Hall’s Weights (see Appendix VII) to try to place them correctly in relation to each other. When the first few additives had been recovered from a depth established in this way, we could then try to place more messages on that page in true depth by ‘dragging’: moving the message along past the additives one group at a time, subtracting at each position and checking the resulting codegroups for scanning and good groups. Both scoring and dragging were laborious and their results uncertain. A third approach, called ‘synthetics’, used a routine message and a long message known to be on the same page. A cliché of two or three codegroups forecast for the routine was run first against the enciphered version, then against the long message, by the Freebornery and the resulting speculative codegroups were tabulated for us to test by hand. This was less laborious but called for particularly accurate forecasting of the routines by GEN.

  Thoughout the following months the Japanese multiplied their cryptographic changes: the life of their additive tables tumbled from six weeks to 20 days or less. In June 1944 they introduced grilles which chose the additives to be used from different places on the page instead of consecutively. During this period of uncertainty as to what was going to happen next, we developed a close collaboration with Lieutenant Peter Laslett RNVR in Naval Intelligence NS V. He had developed a specialism in crypto-intelligence: reading and analysing those Japanese messages which were about their own cryptographic arrangements; and from this he could often forewarn of impending changes. We remained in close touch after he moved to join Bletchley Park’s Liaison Unit in Washington. Later he was Fellow of Trinity, Cambridge, social historian and originator of The Dawn University which led into The Open University.

  Fortunately the flood of changes made life at least as difficult for the Japanese as for the cryptanalysts. Their overseas stations were so numerous and so distantly distributed that the task of getting the new codebooks and tables to everyone in time was beyond them, so some stations had to continue with the old for the time being. Having some messages transmitted in both the old and the new systems was a gift to the cryptanalysts, which Washington with its huge resources could exploit to keep abreast most of the time. But we were struggling and failing to keep up. We could not attempt any current system after G-59. Even when we dropped back to superseded systems, the clerical overheads left so little time for stripping that our recovery of additives dropped to some 200 weekly with just one peak of 320. Our crippling problems were the shortages of traffic and of staff.

  The new more complex systems required a bigger quantum of messages to be available before a break could be achieved. With the shorter life of tables, and the same time-lag in receiving traffic from Washington as when we started in Septem
ber, we were sinking perilously to the point where the B Channel would be unreadable, not just for the time being but for ever. Early in March 1944 we asked for all Washington’s intercepts to be sent to us by cable. Our request was declined; not unreasonably from Washington’s point of view but dismayingly from ours.

  Throughout 1944, up to November, our team numbered only 25: little bigger than our original 20 in September 1943. Of these, seven were men of TJAO or commissioned rank. Tony Phelps and David Foxon had gone to Colombo at the end of May in response to their appeal for skilled cryptanalytical help. Brennan and Vernon Pilling had joined us. Original members who left the Big Room were now replaced by very able Wrens. But we rarely if ever had 25 on the job at once. Ian Cassels and Jimmy Whitworth took the Japanese Language Course in November 1943, then three of the most promising girls in January 1944, another four in March and this rate continued until almost all had some knowledge of the language. For nine weeks in May–June, we lent six girls including two very capable strippers to help out the NS IIJ Registry who were losing ground as traffic volume increased. Higher authority recognised that even the B Channel could not be worked successfully by only 25 people and authorised an increase to 80. Only this promise and a steady growth in what we could get the Freebornery to do for us kept our dismay and disappointment within bounds.

  On 25 July 1944 the Japanese, unusually for them, played the winning trick of introducing simultaneously a new codebook (N), new additive tables (62 and 63) with 20 lines of 10 additives per page and a new operating system, still with grilles; and made no serious error to compromise the new system. For the first time in over two years of currently reading JN-25, Washington had no break and no prospect of a break. With no sign of progress by mid-August, our cryptanalysts began to tackle N-62 too, while the Big Room continued to work G-57. Towards the end of September, with the problem completely unsolved and Washington recognising ‘the greatest crisis’, Brigadier John Tiltman and Major Jerry Morgan joined in our cryptanalytic efforts. At Tiltman’s request our entire team began preparing all the enciphered groups page by page, for indexing by the Freebornery. But this came to nothing and in October, with very little progress on the problem anywhere in the world, we gave it up.

  My 1945 history is frank in recognising with hindsight that it was a mistake to have persevered so long. It cites three reasons why I and – more importantly – those above me did so. First, we supposed that the turmoil of cryptographic change was a temporary phase and that stability would return. We were wrong about that. Secondly, it was feared that Washington would abandon work on the B Channel whose intelligence value they judged to be less, whereas Colombo rated it as 20 times as valuable as the A. Thirdly, we saw ourselves as holding on, as keeping our hand in, ready to move into high gear when the promised extra staff arrived. My 1945 history put it:

  there is no doubt that the ease with which promises of staff increases could be obtained was responsible for much over-ambitious planning which proved a liability when the promised staff failed to appear.

  Alongside the disappointments, and thanks to them, this was a period when the team grew up. The technical difficulties drove us to work hard to find solutions and we learned a lot in the process. We became much better at the job than we would have been if it had remained as straightforward as in the autumn of 1943. Moreover, once we were obliged to surrender the imperative of working on current traffic, we could afford to devote some time to systematic training. We coached the Big Room members, and especially the senior ones, in all the techniques they used – indicators, stripping, grilles, in case we had to work them, a smattering of probability theory – so that with a fuller understanding of the processes they could do more on their own initiative. As a result we developed a power and a confidence which served us well later.

  I did not know in the summer of 1944, nor when writing the internal history a year later, how our struggling to keep up fitted in to the bigger picture of tensions between Bletchley Park and Washington. Reports released and books and articles written only in the last ten years have shown that these existed at various levels, and that they were sometimes acrimonious. At the highest level, for example, Britain feared that American concentration on the naval war in the Pacific might deprive our Commander-in-Chief Far East of intelligence that he might have had. And at the level of our party’s concerns, there were those in Washington who thought that keeping us involved was a needless expense from which they could expect no return.

  The resolution of our role came on 23 October 1944 with the signing at Director level in Washington of an Agreement on the disposition of forces on Japanese Naval cryptanalysis. So far as large-scale recovery was concerned, Bletchley Park would work on tables assigned to it by Washington, while Washington undertook the recovery of current systems. This was realistic, and right, and it came none too soon.

  Also in October, for quite other reasons, Washington emerged from the darkness which had shrouded JN-25 since 25 July. On 15 October N-66 and 67 were superseded by N-70 and 71, astonishingly without the use of grilles. Washington then made rapid progress, and vanished out of our sight and out of this chapter: we were no longer expected to keep up.

  At about the same time as the Washington Agreement, and with the battle against the Atlantic U-boats virtually over, Hugh Alexander left the naval Enigma section Hut 8 and moved to Japanese Naval. He took over the cryptanalytic teams and the Processing Party as NS II J. The bookbuilders and translators, under Commander McIntyre, became NS III J. Under the new arrangements our staffing at last began to be reinforced though never to the level once promised. By mid-December we had gained 15 Wrens, including two who had stripping experience on JN-11, and five more by the end of April 1945. Another experienced JN-11 stripper, Lieutenant Fred Ponting RNVR, joined us. A week after the German surrender in May we received reinforcements from Hut 8. Our eventual maximum number was about 45.

  The first assignment we were given under the Agreement was the totally unworked Table 53. This was neither in the A Channel nor in the B but in the Ancillary or RU Channel. This ran alongside A and B but usually at a more humdrum level, concerned with personnel and supplies. With one exception its systems continued familiar patterns. The RU’s additive tables stayed with pages of 10×10 when the A and B Channels went to 15×12 and then 20×10, and it did not use ordinates or grilles.

  Its indicator system, however, was radically different. After the text of a message had been enciphered with consecutive additives taken from a starting-point of the cipher clerk’s choosing, he used the first enciphered text group to identify a line, column and page position elsewhere in the table; the additive in that position was then used to encipher the 5-digit line, column and page starting-point of the message itself. As a check he used the immediately following additive to encipher the starting-point for a second time. Similarly the last enciphered text group was used to identify by its line, column and page a third additive elsewhere in the table, which then enciphered the line, column and page of the ending-point. Outwardly simple, this was in fact very hard to break: the indicator keys for start and end positions could be any two of the 50,100 additives in the table.

  Given this difficulty, the small quantity of the traffic at the time and the relatively lesser importance of the content, Washington had being doing little more than observe the RU Channel work its way through successive codebooks and additive tables. However, when General MacArthur’s forces landed at Biak in the last week of May 1944, the Japanese feared that all JN-25 systems might be compromised. They immediately switched a lot of what would have been A and B Channel traffic into the RU Channel, then L-49 with an emergency usage. On 1 June they replaced L-49 by L-53 and on 15 June they added a second emergency usage.

  The RU Channel had now to be tackled, and Washington chose to attack L-49 rather than L-53. A huge effort made little progress. On 17 July L-53 was superseded by L-64. In October Washington closed down work on L-49, but Honolulu made a break into L-64 which, with more
traffic available, was quickly exploited by the two stations. Thus at the time of the Agreement no attempt had been made to break into Table 53, and it was accordingly assigned to Bletchley Park. Helpfully, the American work on the adjoining tables had made cumulative progress with the L codebook which had continued throughout, so that part of our assignment was not starting from scratch.

  Under the Agreement Washington also agreed to Hugh Alexander’s suggestion that we tackle the recovery of columnar key additives on N-66, to which they attached some urgency: a smaller-scale but technically more complex problem. Ian Cassels began work on it in November with Hugh Stanton, recently returned from Colombo and with long experience on JN-25. At times Jimmy Whitworth and Vernon Pilling worked with them, and they drew on the Big Room as they needed. This work was outside our party and therefore outside this chapter. But I will record that when (to our dismay) Washington took the task back again at the end of February, along with the work done on it, they commented:

  The British keys proved to be of exceedingly good quality … a good test of the Washington method although they had made certain modifications including an elaborate use of Bayes’ Theorem.

  Additive recovery on L-53 made a slow start while we prepared the half-million enciphered groups for Hollerith punching. Commander Belfield and Brennan worked on the indicator system and must have made quick progress, for we made our first few additive recoveries in the last few days of November. Given the difficulty of this problem described earlier, it would be good to know how they did it. My 1945 history does not say: perhaps this is in another internal history which Ian Cassels and I wrote and which has not yet been released to the National Archives.