Alastair Denniston
Page 28
The cryptanalytical process at Berkeley Street started with the head of a country section (head of a subsection for larger country sections) deciding if a decrypt was sufficiently complete and reliable to be passed on. He would then decide if it was of sufficient interest to be submitted for issue and if the translation was accurate. Translation was not always necessary, as in the case of the Foreign Office which preferred to see French decrypts in their original text. The D&R section was crucial to this process as it ultimately decided which material should be circulated and supervised the mechanics of doing so. They in effect acted as a reference bureau for the cryptanalysts. The section was headed by Eric Earnshaw-Smith, who had been a member of MI1(b) before joining the fledgling GC&CS in 1919. He also acted as AGD’s deputy. His assistant was Ore Jenkins, a Professor of Medieval History and Modern Greek at Cambridge. D&R kept indexes of all circulated material in longhand on cards. It had mainly a reporting function and never more than five staff. Its working model was to use the Heads of sections as outposts of a Central Distribution Office and, in effect, Intelligence Officers. The Head of D&R kept the Heads briefed so that they could deliver on this intelligence role. Information continually flowed to them and included directions received from ‘user’ departments for specific intelligence and valuations of recently circulated product, collateral material such as Foreign Office telegrams and dispatches, foreign newspapers, and press cuttings. While all collateral material was returned to D&R for filing, larger sections kept their own card indexes and other intelligence records. This was all coordinated by a Central Intelligence Officer.
D&R’s first task was an editorial one, as it had to ensure that the judgment of the section heads was sound. In some cases, intelligence deemed worthy of distribution might already be in the hands of the user department from another source or known to the Central Distribution Officer but not yet to the Section. However, this never exceeded five per cent of submitted translations. Secondly, checks were required to judge the clarity of the translation, and the Section maintained a very high standard in this regard. The nuances which exist in Diplomatic material can be lost without precise translations in place. An example of this is a telegram from the Italian Ambassador saying: ‘I believe that the Foreign Secretary was at some pains during our conversation to leave me with the impression that his government was uneasy about the course of events.’ It was translated as: ‘I think that at the end of our conversation what the Foreign Secretary was laboriously trying to say was that his Government was afraid of what might happen’ – which misses the finer points of the message and would not be trusted by a Foreign Office reader. Continuity from pre-war experience enabled the Central Distribution Officer to press for more clarity in translation, even for languages such as Arabic. Finally, a check was needed for consistency with known facts. Subject matter of a day’s traffic could be extremely varied and even, on occasion, quite technical.
The cryptanalyst/translator required an encyclopaedic knowledge of current affairs and to support them, a central library of reference books, card indices and records of unpublished information was built up under the Head of D&R. He was tasked with ensuring that nothing was issued which would not make sense to a user department. He was also able to refer failed decryption back to a Country section and his analysis usually proved to be correct. In one instance, a pilot from the Italian LATI airline reported details of an Allied convoy to Rome (this was standard and done twice a week). The Italian encoder mistakenly located the position by longitude and latitude in the middle of Spain. This was not picked up by the translator but by his section head. It would almost certainly have been picked up by D&R.
One D&R index was based on names, and over 19,000 were on file by mid-1943. One interesting name was that of the Japanese Ambassador to Berlin, Hiroshi Ōshima, who frequently sent messages in code from Berlin to Tokyo, containing information of great interest to the Allies. The largest index dealt with treaties, catalogued by country, and included details about treaties and negotiations for treaties among various governments. A ‘Cabinet Book’ recorded important officials in various governments, subdivided by country. It contained newspaper clippings and notes giving names of cabinet officials and elected officers. A file of ships recorded all ships which were named in intercepted messages. The ‘Diary’ was a large bound book which was assigned two opposite pages for each day. All events were recorded which might be of interest to anyone working on material and might involve such events. Each item was referenced to its source and thus fuller information if required. A ‘Who’s Who’ file illustrated the thoroughness of the Section. It consisted of Foreign Office printed bulletins with details of foreign diplomats, supplied by British diplomats abroad. One diplomat was described as having a young wife who was very short-sighted and therefore wore very thick glasses!
D&R also kept a reference library for use by all Berkeley Street staff. It included a ‘Geographical Handbook Series’ published by the ‘Naval Intelligence Section’ of the Admiralty which contained considerable information about various countries. An annual publication of the Empire Parliamentary Association called ‘Report on Foreign Affairs’ gave a chronology of all important events during the year for each country. Foreign language press was also monitored with relevant articles sent to interested sections. All geographical sections kept their own reference material. McCormack summed up this part of the operations as follows: ‘Whole key to this British operation lies in the infinite pains which they take with the files while never losing sight of their very practical objectives.’
D&R also graded and annotated diplomatic translation, and from 1942, user departments asked that all translations include both the grading believed to be attached by the sender to the cipher used by him for that message and their own rating of the security of the cipher used.25 Accurate and consistent gradings proved to be difficult to achieve but user departments felt that they gave some indication of the secrecy desired by the sender and the likelihood of them being read by other governments. Ratings/gradings were supplied by Sections under the guidance of the D&R.
Pre-war experience had shown that for the diplomatic country sections, linguists could be trained as cryptanalysts and be responsible for translation. In some cases, they would even be tasked with the selection of decrypts for submission. The exception to this was the Japanese ‘Purple’ traffic. This differed from the model adopted for the Service sections because of the sheer volume of traffic faced by them and machine rather than hand ciphers in general use. The merger of cryptanalytic and linguistic skills was made possible by recruiting people of a very high intellectual standard with a good knowledge of international affairs.
The head of a country Section could, on his own discretion, translate and issue only part of a message. In some cases, this was driven by the subject matter changing and the remainder not being of interest. Summaries or digests of message subjects were not kept as it was found to be too labour intensive with little accrued value. More reliance was placed on the judgment and working relationship between the section Heads and Central Diplomatic Intelligence Officer. The diplomatic sections were part of a stable organisation at the beginning of the war with considerable experience. Once it moved to Berkeley Street, AGD was able to maintain and even enhance its efficient operation.
Diplomatic country Section work concentrated primarily on the systems of Germany, Italy and Japan. Secondary targets were China and France. In addition, some forty-five countries in Europe, South America and the Near and Middle East were targeted. Of the individual country sections, the Japanese and German were the biggest and most fruitful in terms of intelligence. The Japanese Section consisted of fourteen experienced and able staff who received all messages after they were logged by their clerks. One member of the Section handled liaison with the US and Australian Japanese intelligence sections dealing with diplomatic and commercial traffic. Head of the Section was Oswald White, who at the beginning of the war was Consul General in Tientsin and had l
ong service experience in Japan and China. He, or a deputy in his absence, prioritised all messages before sending them for translation. The staff who under,took this work included W.B. Cunningham, formerly Consul to the Japanese Imperial Government (JIG26), Norman Roscoe who was Consul to JIG during WW1 and E. Hobart-Hamden, probably the world’s leading JIG scholar and co-editor of the Standard Dictionary. They described themselves as the Translation Bureau and the leading lights included, Captain Rayment, an officer of Naval Intelligence and JIG expert and translator for the Navy, and Captain Harold Shaw, a civilian and ex-consul.
McCormack illustrated the effectiveness of this Section’s work as follows:
Lord Farrar, who sits as Economic Warfare Minister for purpose of handling most secret information in continental economic field and in fact for whole world, told us yesterday that 85 per cent of all important information about South Central economic picture has been built up from Berkeley Street production, the contributions by censorship, foreign press, radio broadcasts and agents being trivial in comparison with our material.
Most of the 1,000 JIG commercial messages which came into the section each day were in plain text so translation was the main task. Their sources included Point Grey and other west-coast Canadian stations and were sent over every week or ten days by bomber. Much of it was of little use, consisting of small financial and commercial transactions but about a dozen items per day were of interest. They might reference ships by name or routing and include the address of enemy military or naval units. The volume of this material was around 1,500 items per week. A second source was material in commercial code intercepted in Mauritius and a third came from UK stations picking up continental traffic. The output of the JIG Commercial Section went in longhand to Alford House where Berkeley Street’s Commercial Section was housed.
The German Section dealt primarily with ‘Floradora’ and was headed from October 1939 until 1943 by 25-year-old Patricia Bartley, who had been recruited from Oxford at the beginning of the war by John Tiltman. She was assisted by Percy Filby, formerly librarian of one of the libraries attached to Trinity College, Cambridge. Staff numbers under Hartley reached thirty-seven and key members were Adcock and Fetterlein, S.A. Trantor, Pallinger, a former school master and Potter, a German Foreign Office expert on diplomatic English. Bartley left Berkeley Street to return to Oxford to complete her Ph.D. However, her health deteriorated and she never returned to London.27 By early 1944, they were passing English versions of every message of importance in ‘Floradora’ with a delay of perhaps 24 hours, sometimes less. Progress was even made against OTPs, which should be unbreakable as settings are only used for one message. However, messages were being sent by Berlin (so-called Multex messages) to all posts in OTP but to Dublin in ‘Floradora’. The messages were long and the same length in both systems, which could make tests productive. At the same time, several German pads were acquired from American sources. They could be assembled to work out the logical design of the machine that produced them. German diplomatic traffic of high importance was encrypted using OTPs. In collaboration with the Americans, the machine which produced these was reconstructed, but only one series of messages was read between Berlin and Madrid in March and April 1945.28
The Vichy Government traffic was read until the North African landings when they made belated cipher changes. They introduced three new systems which remained unsolved, but for Far East traffic they used improvised modifications of existing systems which GC&CS read.
The Near East Section was headed by Dr Thomas Thacker, Professor of Semitic Languages at Durham University. His team included Raymond Thornhill, a clergyman and former pupil of Thacker’s at Durham, Simpson, a geographer from the University of London, and Gungry from Cambridge University. They dealt with messages in Turkish, Persian and Arabic. The Persian sub-section was run by Frederick Humphreys, who was the Archivist for many years at the British Embassy in Persia. He was assisted by John Boyle who had learned Persian at the School of Oriental Studies. Thacker and Dr Bernard Lewis, who had lectured in the Islamic History School for Oriental Studies, ran the Arabic sub-section. Other code books were being read which were used by Turkey, Saudi Arabia, Iraq, Egypt, Iran, Afghanistan, Syria, Nepal and Ethiopia. The Turkish sub-section received about fifteen to thirty messages per day in 1943 and circulated on average six per day. Other traffic received by country per month was Saudi Arabia fifteen to twenty; Iraq four to five; Iran 1,000; and Afghanistan twenty to twenty-five.
The Italian Section was headed by Frederic Catty from the Spring of 1942 until September 1944. Two women handled Vatican traffic from April 1942. By 1 December 1942, the Section had read 550 out of 664 messages received but only sixty-four were considered worth translating and circulating as it was mostly of an ecclesiastic, charitable or personal nature. Italian diplomatic systems were strengthened at German instigation in July 1942, but misuse led to it being broken by the end of the year. Further innovations by the Italians in January 1943 were also overcome due to their carelessness in use.
The Portuguese and Brazilian Section was run by Arthur Exell, a botanist who had researched Portugal and its colonies. It had no permanent staff and the Brazilian side was handled by Exell’s wife, Mildred, and an assistant. The Portuguese used a combination of book and machine ciphers, all of which were read by this Section.
Of the smaller European countries, a Romanian Section was run by R. Greiffenhagen, a former diplomat. Most of the traffic was commercial in nature, dealing with, for example, purchases of Spanish lead and blankets and Spanish demands for oil in exchange. The Bulgarian Section was run by Gabriel Woods but processed very little traffic. He also handled Yugoslavian and other Balkan country traffic.
Following the Nazi-Soviet pact in August 1939, GC&CS had broken the Soviet meteorological cipher and read a number of naval signals along with some 1,000 army and police methods. While useful as a source of tactical information, it yielded little of strategic importance. According to Hinsley: ‘All work on Russian codes and cyphers was stopped from 22 June 1941, the day on which Germany attacked Russia, except that, to meet the need for daily appreciations of the weather on the Eastern Front, the Russian meteorological cypher was read again for a period beginning in October 1942.’29 However, it does appear that Britain was working on Soviet traffic again by 1943. This had been decided at meeting that year between ‘C’ and the Director General of the Security Service in the US.30 A covert section was set up at Berkeley Street under Professor David Bernard Scott, a senior administration officer and a Cambridge mathematician. It was tasked with breaking a system which appeared in late 1943 when the RSS picked up Morse code signals which was presumed to be a reinstatement of the Communist International or Comintern network (officially abolished in May 1943) and had been inactive for some time. Scott reported directly to AGD and the section had three to four staff. Scott broke the cipher being used in about a month. None of this traffic was shared with the US.
As the Soviet Army advanced, traffic increased on this network and the section was increased to around twelve staff. According to John Croft who worked in the section, they included ‘the wife of a University of London professor, two 18-year-olds straight from school, one a classical scholar from Belfast, the other from Winchester College; the wife of one of the directors of J. Lyons and Co., the caterers; and a lady who asked for a transfer from MI5 in Curzon Street because she had had enough, so she said, of the society debs (who in wartime had been recruited to that outfit) gossiping about all the other society people under surveillance.’ Croft was one of four or five cryptanalysts and a Russian expert.31 He worked alongside one of the Fetterlein brothers who were brought out of retirement to translate the decrypts. These were then typed up, classified as top security with limited circulation such as the CSS and Churchill’s assistant on security matters, Major Desmond Morton. Only AGD visited the section, and telephone communications with RSS or Berkeley Street was through scramblers. In early 1945, the traffic could no longer be b
roken as it was now being enciphered using OTPs.
In late 1944, a special section was working on non-Morse traffic on military and civil circuits. The Strong/Travis agreement of 1943 did not cover traffic from non-service enemy or neutral sources. Between August and September 1943, Roger Randolph of G-2 Special Branch visited Berkeley Street. He was shown almost everything and in his report he covered Russian issues under ‘Miscellaneous matters’. His summary was: ‘Prior to 1941 Russian diplomatic traffic was studied. The conclusion was reached that it was one-time pad and accordingly the research was abandoned. At the present time Russian diplomatic traffic is not being analysed and none of it is being read.’32
However, Randolph did not say that GC&CS had stopped collecting Soviet diplomatic traffic. Clearly, the whole issue was very sensitive and it may be that AGD was under instructions not to share this with the Americans. According to David Alvarez: ‘Ironically, the Americans remained unaware that in late 1944 GC&CS established its own secret unit to study Russian internal traffic (civil and military) in Sloane Square, London.’33 He went on to say that:
By the end of the war, the United States had developed a large and productive signals intelligence organization that was reading the diplomatic traffic of almost every government in the world from Afghanistan to Yugoslavia. Much of this success was due to the diligence and skill of American cryptanalysts. No small part, however, was played by the British codebreakers at the Government Code and Cypher School. Throughout the war GC&CS provided timely advice and assistance that significantly advanced the American programme in diplomatic signals intelligence. The value of the British contribution was accurately summarized by the US Signal Security Agency (previously SIS) in a post-war review of operations: ‘It is doubtful whether success in solution of certain diplomatic systems could have been achieved in time to be useful had not the British supplied the necessary information. The debt of the SSA to GC&CS in shortening the period between the beginning of study and the production of translations was in the case of the diplomatic traffic of certain governments very great indeed.’ Without doubt, the British had been most helpful and cooperative.