by Alastair Denniston- Code-breaking From Room 40 to Berkeley Street
Friedman and McCormack visited AGD at Berkeley Street on 6 May and got a revised draft of a proposal for division of work on ‘Floradora’. On 22 May, Friedman had a chat with AGD about a paper drafted by AGD which outlined the proposed basis of a talk with Taylor, McCormack and himself about the future relations in neutral and Allied fields. AGD had written that after McCormack had seen all of AGD’s sections and knew ‘all our methods’, he wanted to draft a detailed plan for continued liaison, but that it would need to be submitted ‘to our respective chiefs’.20 There was already substantial cooperation on enemy diplomatic and commercial traffic which was just as well, since the BRUSA Agreement had avoided the issue of diplomatic traffic. In fact, it only called for full cooperation and complete exchange of cryptanalytical results and intercepts from the traffic of the German Army and Air Force, as well as the German intelligence service, the Abwehr. Even with this cooperation, the British were reluctant to share with Washington any diplomatic messages from neutral countries which were sent by cables controlled by Britain. The Americans in turn went to great lengths to conceal from the British their work at Arlington Hall on Russian diplomatic codes.
On 17 May 1943, the US Army agreed formally that Arlington Hall would concentrate on Japanese military traffic and leave the breaking of German and Italian military ciphers primarily to the British. Each side would be responsible for ensuring that the resulting intelligence got to Allied commanders in the field. AGD’s 21 May paper attempted to address the existing problems:
It has occurred to me and others here that your visit provides a good opportunity to define the scope and limits of the liaison which we are trying to build up between G.C. & C.S. (Civil side) and Arlington Hall and G2 (Diplomatic and Commercial).
For my own personal part in this matter I have urged during and since my visit to Washington in August 1941 that Arlington’s greatest contribution to the war effort is the effective and operational reading of Japanese military cyphers and that G.C. & C.S. was and is prepared to fill any intelligence gap in diplomatic work which may result from a supreme effort on Japanese Military by Arlington. I wish to repeat this and I know I shall have the support of my superior officers.
Colonel McCormack’s letter to me on his arrival gave me great hope that mutual misunderstandings were going to be cleared up and that we should straighten out the line of liaison. I have arranged that you should see every section and every detail in order that you may be familiar with ‘our methods’ and appreciate our aims, which are, in short, to provide our several customers with all possible intelligence derived by cryptography from telegrams from enemy and neutral sources. When you know all our departments who are in any way affected by our liaison, e.g. (in England) Foreign Office, M.E.W. and the Service Departments, whose efforts may be influenced by the knowledge of our cooperation. The bases of the liaison between A and B are:
Cryptographic documents i.e. code books and key tables obtained either by cryptography or by S.S. methods.
Raw material, i.e. telegrams obtained by W/T interception, by cable censorship or by S.S. methods.
The translated versions of the raw material.
Note: In London it is not the duty of G.C. & C.S. to extract intelligence from these translations: that is left to the Intelligence Sections in the receiving Ministries with whom G.C. & C.S. is in close touch.
So far as enemy countries are concerned (Germany, Japan, Italy) it should be our aim to make the liaison absolutely complete and I believe we are already achieving this. If either A or B requires a telegram in cypher or en clair, it is passed without delay. It might be noted that Arlington helped us into the Japanese purple.
The immediate problem is the prosecution of the war and I consider it would be to our mutual advantage if G2 had their representative in London (Lt. Col. Taylor) as a liaison officer to G.C. & C.S. (civil side). He should have the entry to our D & R and all sections … As to the purely cryptographic part of the liaison, this should continue to be direct between Arlington Hall and the sections but Colonel Taylor would be available here to clear up questions hard to solve by letter or telegram.
A.G.D. 21/5/4321
McCormack summed up their discussions as follows: ‘Denniston says that if Arlington wants to divert any talent from present Japanese operations to turn them to JAC he is prepared to take up the full slack and to transmit finished translations of all material here. Second, he expresses a desire to give Arlington traffic and information of every kind that has to do with winning the war.’22 Meanwhile, AGD and Friedman had become good friends and they subsequently spent a golfing weekend together at AGD’s home in Surrey.
AGD was keen to give the Americans the benefits of British experience and of traffic obtained over many years. He was however, limited by a Foreign Office dictate forbidding the sharing of cable traffic into and out of London. He was also reluctant to share traffic in areas which he considered to be of prime British concern such as from various Near East areas. However, he did accept that given the American commitment in those areas, he may have to share information with them. He was prepared to share everything his team knew about methods of solution but he did not want to share so-called crypt documents such as code books, key tables, obtained either through cryptography or secret service methods. AGD thought it was more important to concentrate on liaison on the intelligence rather than the cryptography side of the work and he was annoyed by the interface role played by the BSC under Stephenson.
McCormack summed up AGD’s views as follows:
Denniston has said on several occasions that he did not like being unable to deal directly with Arlington on traffic exchange problems or on how exchanges should be handled. He also is very anxious to obtain good liaison with G-2 and has been going on the assumption that Taylor is to be quartered with him to function in that capacity. While there would be no point in trying to work out any revised traffic deal now, certainly not for us to try it, it appears to me that there is some value in pursuing discussions with Denniston along the above lines, so as to test his general ideas by specific cases, in order to carry back a fairly good idea of what sort of deals his authority permits him to make and what the general viewpoint here is on these various problems. His attitude, in my opinion, will permit all important intelligence problems along his alley to be solved satisfactorily, one way or another. Note also that Denniston, more than anybody else here, has turned his people over to us for questioning and given us a free run of his place.
According to McCormack, AGD was willing to let American staff in London read anything of interest and forward it on to Washington. However, in some cases, the Americans insisted on the material in its raw form and the appropriate keys to read it. He went to say:
It [i.e. Washington] has just thrown Denniston into a state of bafflement by asking for Iraqi keys. This request would be roughly equivalent – if we still had the Philippines – to the British asking Arlington for Philippine keys, since Iraq is not only in reality part of the British Empire but the cornerstone, because of its oil, on which the whole Middle East and Eastern Mediterranean situation rests. Denniston however, wanted to put Arlington in a position to read whatever Iraq stuff it might intercept, insofar as he could do it without going to the Foreign Office for specific authority. Hence he authorized his people to explain the Iraq system to Taylor, and Taylor duly passed the information along, and Denniston felt that he had met Arlington’s demands without having to create an issue here. If questioned by the Foreign Office, he could always say that Iraq uses simple substitution which any cryptanalyst who had an Arabic linguist at hand could solve in half an hour, and so he really only told Arlington what it could have found out for itself in a very short time.
Now however, Arlington turns this matter into a major issue by a formal request for keys, forcing Denniston either to put himself on record as refusing something that Arlington specifically asked for or to create an issue by asking the Foreign office for permission. If Denniston is right about th
e difficulty of solving the Iraq system, what sense does all this make?
AGD had offered, through McCormack, to take on Japanese work and, according to McCormack, had expressed to him ‘a desire to give Arlington traffic and information of every kind that has to do with winning the war, by which he means complete exchanges of all enemy traffic and crypt information plus anything that Arlington wants to get out of the non-enemy field where, as in the case of Turkey, we have asked for it and stated reasons connected with the war effort’. AGD also agreed that ‘in the crypt field each country wants to establish for itself a position of independence so that it can get and turn its efforts toward any class of traffic that may interest it’.
On 21 May 1943, McCormack sent his first dispatch to Washington, describing AGD’s organisation:
Denniston’s show, commonly called Berkeley Street, has none of the hectic atmosphere of Park but rather gives [the] impression of a well established operation that goes along through wars and peace. General impression is typified by the two ladies who receive and sort incoming traffic. [This was corrected in a later dispatch.] (They started as telegraph clerks in Post Office in reign of Queen Victoria and were fully familiar with general field when they joined present organisation in 1919.) These little birdlike old ladies receive and register incoming material and they have acquired such great familiarity with it that [they] can do everything except actually decipher it. Whole organization is very simple and they seem to accomplish a great deal with quite limited personnel. Whole outfit consists of two hundred.
The Americans assigned Taylor to work with the British on diplomatic traffic as the US liaison officer at GC&CS with full access to BP and Berkeley Street and their decrypts. However, he was ordered to confine himself exclusively to the diplomatic decrypts being produced by AGD’s team.23 His eventual replacement, Major Bancroft Littlefield became a close friend of AGD. He corresponded with his wife during his stay in Britain for the last few years of the war and his letters24 provide a snapshot of the life enjoyed by American officers stationed in Britain.
The following extract is from a letter written on 26 March 1944:
I had luncheon Friday with the British Major (ex school teacher) whom I went on the automobile trip with (I can’t remember what I then told you his name was). He called me Thursday and (as one does over here) rather formally invited me to luncheon the following day – to meet him at a restaurant on Deane [sic – Dean] Street in Soho (Gennaro’s Restaurant). When I got there I found he had brought along his brother and half brother – the former a private in the Medical Corps – the latter a clerk in the War Office. We had a most entertaining time – and I thoroughly enjoyed myself. After lunch I came back to work, en route seeing a February Atlantic Monthly in a bookshop window (quite unusual I think; at least the first time I’ve seen one) and buying it.
Saturday (yesterday) was another social day. I worked all morning until about 2 pm – then had soup and a sandwich with Ed Kellog at a funny English Snack Bar near where he lives (Queen Street). Then we went together over to a performance by the French players here in town of the ‘Paquebot Tenacity’ – a very inconsequential 3 act creation whose chief merit was that it attracted a wholly French audience at the Institut Francais du Royamme-Uni on Queensway Place, Kensington, and that we soaked up a good deal of French during the course of the afternoon.
McCormack attempted to report on the amount of traffic being handled by Berkeley Street in a dispatch to Washington dated 21 May 1943:
Records maintained at Berkeley Street do not include figures for total traffic and I did not ask them to go to geographical sections and add up all these figures, but they receive ‘thousands’ of messages per day. Their serial numbers of circulated diplomatic items crossed one hundred thousand early in 1942. Do not have present range of number.
***
By the end of 1943, AGD’s Diplomatic Section now numbered 250 with his own office which directed administration, traffic sorting, the D&R Section and the typing section numbering forty-five staff. The main enemy country sections were the Japanese (dealing with hand and machine systems) and German Sections, each with fifty staff. The French and smaller enemy sections had thirty staff, while neutral and friendly country sections had seventy. Small research sections were being built up to deal with new developments such as the use of the Hagelin machine. The total number of translations issued in 1943 was 14,050, of which the Foreign Office received all, the Admiralty 5,481, the War Office 5,697, the Air Ministry 4,162, the MEW 1,702 and MI5 9,850.
Space became short at Berkeley Street, so smaller sections were moved to Aldford House. From October onwards, night raids gradually increased and the staff (serving by rota as part of the Fire Watch of the area) manned fire-fighting equipment through some noisy nights. Staff also contributed to the 2nd Home Guard Battalion of the City of London Regiment. By the autumn of 1943, old German traffic had been cleared and current traffic was being exploited. This allowed some resource to be freed up to look at the OTP system, some of which had been exposed.
The Japanese had introduced three new systems of encrypted code, so cryptanalysts and linguists worked together to produce an increased volume of translated decrypts of good quality which more than compensated for the loss of Italian intelligence. The French Section was working on both Vichy and Free French systems. The Vichy Government had only introduced new systems in Europe and the Near East, which were never identified. In the Far East, a mix of former systems being used or keys originally developed for one system being used with another, meant that the section maintained control of the Vichy traffic to some extent. Ethiopian traffic was investigated at the end of the year and read by April 1944, and the Syrian main system was also broken.
By the end of 1943 no action had been taken on Redman’s proposal of the previous year. On 13 December, Frank Rowlett, now a Lieutenant Colonel in the Signal Security Agency (SSA, the new name of SIS), reported on a meeting between Army and Navy representatives to discuss the allocation of commercial traffic. It appeared that while the President had initialled an informal note to the Director of the Budget on 8 July 1942, no action had been taken by the Joint Chiefs of Staff. It appeared that there was confusion about whether or not the agreement was in force. One of the problems seemed to be the difficulty in separating Japanese commercial and diplomatic traffic, as Japan treated them as one and the same.
By the beginning of 1944, AGD’s organisation was in full production with established procedures in place. Apart from material obtained from the US, the principal source was through the censorship measures put in place in telegram ‘clearing houses’ such as those in London and Ottawa. Foreign Office intercept stations in the UK, Canada, India, Australia and South Africa also provided material, as did secret intelligence services in neutral capitals. When all non-commercial material came into Berkeley Street, clerks routed it to the appropriate country section where it was registered under a number of headings. It was then passed to cryptanalysts for processing and if successful, results were passed to the Head of Section who decided which material should go to D&R.
All internal processing was done in longhand but typed before being distributed externally as GC&CS reports, and continued to be known as BJs. Copies of all distributed material were also sent to the Foreign Office, Menzies, the War Office, Air Ministry and Admiralty. Some material was not sent to service ministries if it revealed that a British diplomat abroad had been compromised in some way. Teleprinter links were available to Menzies’s office, BP and the intercept stations at Denmark Hill and Sandridge. However, they were not used for circulating material, apart from some of exceptional importance to Menzies himself. Various government offices had liaison officers at Berkeley Street and most of them were senior figures. These included those of Menzies, the Prime Minister, the Foreign Office, the War Cabinet and SIS amongst others.
Distribution policy was simple; distribute everything that might have some intelligence value, however small. Berkeley Street w
as more a production organisation like Arlington Hall than an intelligence organisation with important operational functions like BP. McCormack noted in his report that ‘In case of Berkeley Street as in case of BP, what impressed us was that they are taken to see that all information gets out to those who can make use of it.’
Berkeley Street was able to analyse the messages that it processed through its imaginative record-keeping system. A ledger or book contained sheets, roughly thirteen by sixteen inches, for every city which originated messages. Each of these sheets in turn catalogued messages by the city to which it was destined. Each city was then subdivided into nationalities and for each of these was recorded, the call signs of the originating and destination station, intercepting station or stations, estimated number of messages sent, number of messages sent, number missed and additional information to aid analysis. McCormack was very impressed with this system and recommended that his colleagues at Arlington Hall consider implementing something similar. He noted that it was possible ‘to see what intercept stations produce important material and in what volume, to determine at a glance what sources of information there are in each principal city and how many of them are being tapped and of course to see how much of each type of traffic is being brought in’.
The CMY (Commercial Y) Section was moved from BP to Berkeley Street in 1944 and accommodated in the same building as the cryptanalytic sections. It controlled the intercept positions employed on international commercial W/T. Research sections were expanded to respond to monthly cryptanalytic reports from country sections on unsolved problems. A senior cryptanalyst coordinated these reports and kept AGD informed about developments. By the end of the war, the diplomatic sections had created a similar research-orientated structure to that which existed before the war began.