Alastair Denniston
Page 26
Sandwith proceeded to give a fairly detailed description of Britain’s Sigint organisation:
The Admiralty is responsible for reading every available intercepted and decrypted naval or naval air signal generated by Britain’s enemies or potential enemies.
The War Office is responsible for the intercepted signals of enemy military organisations.
The Air Ministry is responsible for reading the intercepted signals of enemy air organisations, except those signals generated by aircraft over water.
The Foreign Office is responsible for reading intercepted and decrypted commercial and diplomatic messages.
The Radio Security Service (reporting to MI6) is responsible for illicit wireless transmissions in the UK, although it has recently expanded this responsibility to include illicit transmissions emanating from countries of the ‘Empire’ and neutral countries.
Each of the five organisations cited above has its own wireless intercept or Y service and these are administered by a Y Committee composed of the heads of the 5 Y services plus certain senior intelligence officials. The head of this committee is an admiral [probably DNI].
The cryptographic centre is ‘Station X’ [GC&CS at BP] and consists of about 2000 people drawn from the 2 services and the Foreign Office. It does both cryptanalysis and TA. The results are pooled.
The wireless interception programme of the 5 Y services is determined half by the traffic needs of the cryptographers and half by the needs of the TA.
Because the cryptographic centre is in England, the main intercepting stations, Scarborough and Flowerdown, are there also. For traffic that cannot be received in Britain, there are intercept stations at Freetown, Pretoria, Durban, Colombo, Melbourne, Bermuda, Alexandria, and several in Canada.
The 65 receivers at Scarborough are all on German traffic whereas those at Flowerdown cover Italy, Spain, France, and any other countries heard in Britain.6
Sandwith also provided details of eighteen direction finding stations in Britain plus six in Australia, four in New Zealand, four in South Africa and fourteen in Canada. He provided details of successful operations such as the sinking of the Bismarck and the battle against the U-boats.
In late April, Lieutenant C.H. Little of the Canadian Navy was sent to London to confer with intelligence authorities there. His main task was to persuade the British to supply Canada with decrypts from non-naval Japanese traffic being sent to Britain from Canadian listening stations on the West Coast.7 In mid-May 1942, Little visited Section V of MI6 in St Albans and met with Felix Cowgill, MI6’s counter-espionage chief. He wanted to know why Canada was receiving so little South American espionage traffic. Cowgill in turn questioned whether any agreement to provide it was in place
However, AGD got on well with Little. He gave him a tour of the London offices and even invited him home for the weekend. AGD offered Canada the decrypts of high-grade Japanese diplomatic traffic. This included the ‘Purple’ encrypted messages from the Japanese ambassador in Berlin. In return, AGD asked that Canada continue to monitor Japanese traffic, concentrating specifically on the Japanese diplomatic and commercial messages being received by the Point Grey station near Vancouver. 8 AGD also supplied Little with a list of countries and call signs being monitored as of 3 June 1942. They included: Germany, Japan, Afghanistan, Argentina, Brazil, Bulgaria, Chile, Eire, France, Germany, Hungary, Iran, Italy, Japan, Portugal, Romania, Spain, Sweden, Switzerland, Thailand, Turkey, Russia and the Vatican.9
AGD made repeated requests that Canada concentrate on Japanese commercial traffic but Tommy Stone, the head of the Canadian Examination Unit, said that their work was for Canada, not as a service for Britain. On 6 June AGD wrote to Little, suggesting they try to work together so as to avoid duplication of effort:
You will remember at our conference with the Director that he stressed the importance of your Japanese work and asked that steps to expedite this material should be taken. I therefore asked that the Admiralty (whilst you were at sea) arrange for the Japanese Government material obtained by Point Grey [Vancouver] to be wired home, and it is now possible to make use of the Ferry Command to transmit raw material, preferably unsorted frequently. Where the material is known to be either diplomatic or commercial it would save delay if you could address the package to DD(C) London. I believe it arrives via the Air Ministry.
AGD worked hard to build on the success of the visit to BP by representatives of SIS and OP-20-G in February 1941, known as the Sinkov Mission. The material that the Americans brought to BP included a copy of the German diplomatic code book Deutsches Satzbuch Nr. 3 (called ‘DESAB 3’ by the SIS). His efforts led to full cooperation between SIS and GC&CS and ultimately to success against two very difficult German diplomatic cipher systems. The German Foreign Office used three different methods to encipher DESAB 3: OTPs (a system codenamed GEE by the SIS), a system using double encipherment (called GEC by SIS and ‘Floradora’ by GC&CS) 10 and reciprocal bigram tables, known as Spalierverfahren. Germany was confident about the security of OTPs and double encipherment systems, and lazy practices enabled SIS to recover part of the code book using cryptanalytic methods. A complete copy was obtained from the Army and FBI, taken from a German courier in July 1940, as he passed through the Panama Canal on a Japanese steamship. He was also carrying 3,500 OTP sheets. GC&CS had already reconstructed part of DESAB 3 and had made more progress than SIS on the method of encipherment being used.11 It had also received a copy of the book from the east coast of Africa.
De Grey had been overseeing work on German Diplomatic systems in March 1941, along with Patricia Bartley12 and several others including Filby. They had possession of material captured in Iceland in May 1940 from the German Consulate which was part of the ‘Floradora’ system. However, the absence of daily indicators had made breaks impossible, and work on ‘Floradora’ had almost been abandoned March 1941. As AGD would later note, ‘Floradora’ was broken because of three factors: ‘The basic book fell into our hands; close cooperation with USA; and SS [Secret Service] work by an able ally who obtained first-hand information and one page of figures from a German cypher officer.’
Progress was slow but, by July 1941, considerable progress had been made with messages from 1940 and early 1941 being frequently readable. 13 In August, AGD was able to report that ‘liaison with America has been conspicuously successful’ and that ‘‘Floradora’ could not have been broken at all without an initial pooling of our resources’.14
***
By the end of 1942, AGD had a staff of around 200 and his organisation’s primary responsibility was to break diplomatic messages for the Foreign Office. This could yield useful intelligence which would then be passed to Britain’s Joint Intelligence Committee (JIC). It also processed plain-language and encoded commercial traffic for Britain’s MEW, and AGD had considerable autonomy in this work. Information from both enemy and neutral nations, combined with that from Censorship, was used to plot the economic progress of the war and to set strategic priorities. While the German Section had been considerably enhanced, it was still not producing a consistent amount of decrypted material. The ‘Floradora’ cipher was not broken until August 1943, so work was concentrated on registration and key recovery.
As well as the ‘Purple’ machine, the Japanese used a transposition cipher on a code, of which 75 per cent had been recovered, along with 90 per cent of the code itself. This had resulted in some material of considerable intelligence value. The ten linguists involved needed to be well qualified to deal with the sometimes imperfectly recovered telegrams to and from the Japanese Ambassadors. The cryptanalysts working on the machine encrypted material collaborated with the linguists on some of the keys being used with the hand systems. While the main Italian systems had been changed in July and the new systems could have been difficult to master if used correctly, Italian misuse meant that messages continued to be read.
After the fall of France, the French Section targeted the traffic of the Vichy Government. Up to N
ovember 1942, they continued to use existing ciphers, which allowed the section to exploit its previous work. The Free French government was established in Algiers and also started to use old systems, but eventually both ‘Governments’ introduced some new systems towards the end of the year. Eastern European government traffic was also targeted, whether ‘free’, friendly, or hostile. Polish traffic was read when the code and keys used by the Polish General Staff in London and its military attachés in Berne, Washington and Stockholm were obtained. Work was carried out on Yugoslav, Bulgarian and Romanian traffic. As Germany advanced to Stalingrad and El Alamein, Near Eastern traffic became of interest, such as Persian, Turkish and Arabic, with sixteen staff covering the traffic of six governments. Sections in India and Sarafand had been reinforced, and an officer was sent to Baghdad and later Teheran to work on Persian Diplomatic traffic on the spot. Success was achieved with Saudi-Arabian, Iraqi, Afghan and Egyptian.
The Chinese Section had grown to five linguists and four supporting staff, and traffic was regularly being read. Given the vulnerability of the Chinese systems, a cipher expert was sent from India to Chunking (now Chongqing in south-west China) to improve their diplomatic and military ciphers. As new staff became available, the traffic of more neutral countries was covered if it was felt that the enemy may seek to exploit their strategic position or natural resources. A South American Section was established to cover Argentina, Chile and Mexico, all of which were of particular interest to American diplomatic agencies. Traffic was easy to break but a shortage of linguists over a number of languages hampered progress. By the end of the year, the section numbered twelve and systems in use were sufficiently recovered to enable most of the traffic to be read.
Swiss traffic was targeted for its ‘economic warfare’ value, and the Section was set up in January 1942 with two expert cryptanalysts with other ongoing work and a few linguists. By the end of the year about 1,000 messages had been circulated with most being of ‘economic warfare’ value, and others dealt with the affairs of a number of governments whose interests were represented by Switzerland in enemy countries. As the Italians might use Vatican representatives to seek peace negotiations, the Papal ciphers were also targeted.
In the first half of 1943 the Japanese Section was enhanced, along with a growth in both volume and interest in the translated messages issued. In particular, the communications of the Japanese Ambassador in Moscow proved to be strategically valuable. The German Section was expanded to fifty and considerable progress was made, with current material on the ‘Floradora’ cipher being read from the beginning of the year. The Italians had made their new systems more secure by changing substitution keys daily. However, some posts continued to use old daily keys, thus neutralising the latest precautions. Mussolini was deposed in July, and in September the Royal Italian Government surrendered unconditionally. However, Mussolini was freed from captivity by the Germans and set up a Fascist government at Salo in northern Italy, supported by the Germans. Posts in various countries were reporting to and taking orders from Mussolini. As the Italian diplomatic service was always royalist by tradition, all major ambassadors and ministers in countries where they could safely do so, supported the King and handed all ciphering equipment over to the Allies. It was therefore assumed that these ciphers would be obviously compromised and never used again. However, when cipher traffic was intercepted it was clear that compromised ciphers were still in use. While the Italian Section had been dissolved when Italy surrendered, a small group remained to look at traffic. For seven months, most Fascist traffic was in plain text. Traffic was intercepted between Salo on Lake Garda and Berlin, Madrid, Tokyo and Budapest. Eight hundred messages were intercepted between September 1943 and January 1944, giving a picture of the Italian set-up at Salo and the attitude of the Germans. Given the vulnerability of Italian systems after their surrender, a joint Anglo-American Mission with Berkeley Street and ASA staff were attached to the Allied Control Commission in Italy from 1 October 1944.
On the American front, there continued to be problems over the ownership of diplomatic and commercial traffic. On 25 November 1942, John R. Redman had written to the Vice Chief of Naval Operations via the Director of Naval Communications. Redman was the Communications Officer on the staff of the Commander-in-Chief, US Pacific Fleet, Admiral Chester Nimitz, and in February 1942, had been put in charge of OP-20-G. His memo dealt with the current problem of cryptanalytical and decryption operations on diplomatic traffic.15 He recommended that the Army be permitted to take over all of the diplomatic work rather than persisting with the current arrangement, which was inefficient and not desired by either OP-20-G or SIS. His proposal was approved by the Standing Committee for Co-ordination of Cryptanalytic Work on 25 August.
A joint British, Canadian and US conference was held at Arlington Hall on 15 January 1943 to formalise the sharing of diplomatic and commercial traffic. MI6 and GC&CS were represented by Kenneth Maidment and Tiltman, Canada by Drake, Tony Kendrick16 and Murray, and the US by Colonel Frank W. Bullock, head of Signal Security Service (formally Signal Intelligence Service), Major Telford Taylor from Military Intelligence Service (G-2), Colonel Alfred McCormack, deputy chief of the Special Branch, and William Friedman. Remarkably, the US Navy’s only representative was an ensign.17 The US Army wanted total American ‘self-sufficiency’ in the worldwide monitoring of diplomatic traffic and wanted Canada and Britain to fill in gaps in their coverage. It was agreed that once a month they would exchange lists of what they were monitoring, and the exchange of data would be through Maidment’s office in New York. Maidment worked for GC&CS and had been sent along with a small team to work with Benjamin Deforest (Pat) Bayly at British Security Coordination (BSC). Before the war, Bayly was a professor at the University of Toronto, and he had been hired by William Stephenson, the senior representative of British intelligence for the entire western hemisphere. Stephenson himself had been sent to the United States on 21 June 1940 to covertly open and run BSC, over a year prior to the US entering the war. The BSC office, headquartered in Room 3603 in the Rockefeller Center, became an umbrella organisation that by the end of the war represented the British intelligence agencies MI5, MI6/SIS, SOE (Special Operations Executive) and PWE (Political Warfare Executive) throughout North America, South America and the Caribbean.
In April 1943, new negotiations began between GC&CS and the US Army’s Special Branch in London and Washington. William Friedman, Colonel Alfred McCormack and Lieutenant Colonel Telford Taylor arrived in London on 24 April. AGD paid a courtesy call on the American delegation at their hotel on 1 May. In the afternoon, Friedman, McCormack and Taylor made their first visit to Berkeley Street. AGD had already discussed the visit with Menzies, and a schedule of what the Americans wanted to see was agreed. They spent three days from 4 May going through the various sections at Berkeley Street.
On 2 May they met with Travis and other British officials and on 8 May Travis and Eddie Hastings18 travelled to the US. Britain wanted the US to concentrate on Japanese Army and Air Force traffic. While there, Travis completed, on 17 May, an agreement in Washington with General George Strong, the US Army’s Military Intelligence chief. The so-called ‘BRUSA Agreement’ was signed on 24 May and formed the basis of all subsequent Anglo-American Sigint cooperation. Travis returned to Britain on 11 June and met with Friedman and McCormack before they returned to the US the following day. McCormack would return to the UK and he and Taylor subsequently produced detailed reports for the Army Intelligence Group at Arlington Hall about the Berkeley Street operation (see Appendix 11). While Travis had been in the US, Friedman and McCormack had met with various senior military figures in Britain, and visited BP as well as Berkeley Street. Friedman was effusive in his praise for British code breaking activities, many of which had been set in motion by AGD. He commented that:
Their achievement is astounding not only because of the breadth of the concept upon which the operations are based and of the directness with which they are prosecute
d, but also because of the manner in which the British tackled and successfully solved a cryptographic system which apparently presents insurmountable and impenetrable bulwarks against attack by pure cryptanalysis. It must be understood that the solution was attained not by cryptanalysis at all but by exploiting to the fullest degree possible the weaknesses injected into the system by the methodicalness of the Germans in their formulation and operation of the system, by studying and making use of the well-known German addiction to fixed habits, and by taking advantage of the occasional carelessnesses and blunders on the part of German cipher clerks. The margin upon which the British had and still have to operate in their solution is indeed a very narrow one, so far as technical cryptanalytic weaknesses in the system are concerned, but dogged British persistence, extremely painstaking attention to minute details, and brilliance in coordinating and integrating into one vast picture the many small operations involved, have brought about a success beyond the wildest expectations of any cryptanalyst’s fancy.
Moreover, the success the British have attained and continue to attain in this field is also astounding in that they have been able to keep the whole operation utterly secret from the enemy for so long a time, despite the hundreds of people who participate in the operations and despite the various tenuous threads upon which these operations rest – threads which might be broken by a mere whisper in the proper place at any moment.19