by Alastair Denniston- Code-breaking From Room 40 to Berkeley Street
In the early 1900s, the ‘Special Section’ was set up within the Department of Military Intelligence. This organisation eventually helped to give rise to the Security Service and the Secret Intelligence Service, better known to most as MI5 and MI6. It housed any cryptographic and cryptanalytic activity within the War Office. Although activity was minimal, it did cover cryptography, cryptanalytical training, planning for war and foreign relations. Colonel George Macdonagh, head of the ‘Special Section’, attended the International Wireless Conference in London in 1912 and established a working relationship with Commandant Francois François Cartier, chief of the French War Ministry’s cryptanalytical bureau. By the start of WW1, the French understood from their peacetime work the German Army’s wireless cipher system.15
Most of the modern techniques for gathering and exploiting Sigint had to be developed during WW1. At the beginning of the war, the British Army had a radio service of mediocre efficiency and primitive signals security procedures. However, it did have an elementary grasp of cryptanalysis techniques along with some expertise in solving the codes and ciphers of other countries. As cipher systems were cumbersome, most countries were sending radio signals in unencrypted form (known as ‘clear’) and this would have a significant impact on the early campaigns of WW1. According to the historian John Ferris:
During September-November 1914 French and British forces intercepted at least some 50 radio messages in plain language from German divisions, corps, armies and army groups. These provided otherwise unavailable insights into the collapse of enemy command and the yawning gap in its line during mid-September 1914. Victory on the Marne was no miracle. Over the next two months similar en clair transmissions (combined with solutions of encrypted German traffic) warned the British Expeditionary Force of the precise time, location and strength of six full scale attacks on its front, each involving four or more German corps. Without this material, the BEF might well have lost the race to the sea, or even have been destroyed. At no time in this century has signals intelligence affected campaigns more significantly than at the very hours of its birth, in 1914.16
Another problem facing the British Army, as well as its enemies, was that secure cipher systems were slow to implement. As late as 1918, experts using two different versions of the British Army’s ‘field cipher’ required four and thirteen minutes respectively to encrypt, transmit and decrypt a message of fourteen words.17 Therefore, most armies opted for usability over security.
While the Admiralty ran the naval war from London, the Army made operational decisions in the field by commanders there with their own staff. Military intelligence, consisting of both interception and cryptanalysis, was therefore a widely distributed function, with practitioners both in London and at GHQ. The War Office played a strategic role, providing training, a central repository of information and longer-term backup to units in the field. Ewing had visited the War Office in early August 1914 to see what steps they were taking, if any, to deal with military wireless traffic.18 Intelligence was part of the Directorate of Military Operations and he found that a new War Office group, MO5(e), had been set up in early 1914 shortly after the creation of Room 40. It had been done with the cooperation of Colonel. F.J. Anderson of the Royal Engineers, who had been involved in reading ciphers during the Boer War19, and Major G.R.M. Church who was in charge of cryptographic duties at Simla. In 1912, Macdonagh had established very close relations with Commandant Cartier, head of the Bureau des Chiffres of the French Ministry of War. Anderson offered his services to the War Office in August 1914. He was ordered to set up a subsection of MO5 to deal with and decipher intercepted German wireless messages, and initially he had four civilian assistants. MO5(e) became MO6(b) in April 1915 and MI1(b) in January 1916.
However, there was very little exploitation of wireless traffic for military purposes, or even a tradition of military interception and cryptanalysis in the War Office. The German Army had expanded its use of wireless telegraphy for command and control during its rapid advance through Belgium and Northern France in the opening months of the war. The War Office unit was also helped by the more experienced French War Ministry Sigint team, which cracked the main German military code used on the Western Front early in October 1914 and passed the solution to the British. The French, following early American work, also broke the new German cipher in June 1918.20
In April 1915, MO5 section became a sub-directorate entitled ‘The Directorate of Special Intelligence’. From January 1916 onwards, a completely separate Directorate of Military Intelligence was established. Initially MO5 handled censorship and cryptanalysis and was headed by Macdonagh, who at the beginning of the war departed to GHQ along with most of the staff.21 MO5(d), under Colonel A.G. Churchill, the Chief Cable Censor handled censorship, and MO5(e), under Anderson, investigated enemy ciphers.
Censorship was introduced in August 1914 and all enemy telegrams were stopped and private codes banned. This included cable traffic in transit between Europe and the Americas but the coded telegrams of Allied governments and neutral ones were allowed to pass. MO5(d) became MO8 in April 1915 and MI8 in January 1916. Postal censorship (later MO/MI9) had by March 1916 extended to the scrutiny of mail in transit on the high seas on neutral ships intercepted by RN ships. This would later prove useful to Room 40’s work on diplomatic Sigint.
The group collaborated with the American Military Intelligence Division in France and the British, French and Americans exchanged technical information and results with a limited division of tasks. When MO5b was formed, Ewing had sent AGD, Norton and Herschell as watchkeepers and there was initial collaboration. This ended in October 1914 when Churchill showed Kitchener, his opposite number at the War Office, the contents of a military intercept before his own section had managed to get the information through to him. Liaison was not resumed until the spring of 1917, when limited cooperation in the form of exchanges of decrypts was resumed.
MI1(b)22 consisted of only four staff by the end of 1915, with Major Malcolm Vivian Hay23 in charge. Thirty-four years of age and the grandson of the second son of the seventh Marquess of Tweeddale, Hay was educated at Beaumont College and joined the Gordon Highlanders as a captain at the outbreak of war. He suffered a head wound and was captured by the Germans. He was left partly paralyzed and eventually repatriated, being unfit for military service. After learning to walk with the aid of a stick, Hay was promoted to major and put in command of MI1(b). Hay seems to have taken over from Anderson sometime between March 1916 and early 1917. His recollection was that it was 1916:
In the Spring of 1916, I began to realise the potential value of the enormous mass of encoded messages from all over the world which was accumulating in certain War Office cupboards. And about this time I had an interview with the [Deputy] Cable Censor, Lord Arthur Browne, who arranged for copies of all diplomatic cables to be sent to my office. In 1916 information about what was going on in Greece was badly wanted and I decided to make a start with the Greek code. No one in this country had hitherto succeeded in breaking a diplomatic code book without what we called a ‘crib’.24 The problem was undertaken and solved by Mr. John Fraser, who is now Professor of Celtic at Oxford. … In June Fraser sent me a telegram: ‘Pillars of Hercules have fallen’.25
In July 1916, a new section, MI1(e), was split from MI1(b), initially for work in the UK against air raids. By autumn 1916, MI1(b) had ten staff but MI1(e) shared the same offices and some MI1(b) staff had to move temporarily to 2 Whitehall Court, which also housed SIS. The original members of MI1(b) were: J. St. Vincent Pletts, a radio engineer from Marconi’s Wireless Telegraph Company; J.D. Crocker, a Cambridge scholar; and Oliver Strachey of the Indian Civil Service. Hay started recruiting new members from the universities, with a preference for language scholars. They included John Fraser, 32, later professor of Celtic as a fellow of Jesus College, Oxford (and Hay’s chief assistant); Arthur Surridge Hunt, 45, then and later Professor of Papyrology at Oxford and a world expert on ancient writing; David Sa
muel Margoliouth, 58, professor of Arabic at Oxford and later president of the Royal Asiatic Society and author of many books on Arabic literature and history; Zachary Nugent Brooke, 34, then lecturer in history at Cambridge, later professor of medieval history there and an editor of the Cambridge Medieval History; Edward Thurloe Leeds, 39, then assistant keeper of the department of antiquities of the Ashmolean Museum; Ellis H. Minns, 42, then and later lecturer in palaeography at Cambridge; Norman Brooke Jopson, 26, from Cambridge and later professor of Comparative philology there; George Baily Sansom of the embassy in Tokyo; and Henry E.G. Tyndale, 28, later a housemaster at Winchester College.
At the end of 1914 MO6(b) started to look at ‘the great volume of diplomatic code messages sent by cable routes which, owing to the censorship, were now for the first time accessible.’26 According to Hay:
In 1914 and 1915, before my appointment to M.O.6(b) some progress had already been made by Messrs. Strachey and Pletts with the American diplomatic code. I am not able to say definitely how this code was first broken. I was told that some clear texts were obtained which facilitated solution. Until the beginning of 1916 the work of the War Office cryptographic section was limited to reconstruction of the American Diplomatic code books.27
America was an obvious target, given the amount of material available. Three code books were solved and super-enciphering tables recovered, so some convalescent officers were temporarily attached to the subsection to help in decoding although they did not tackle ‘the more technical work’.28 During 1916, MI1(b) broadened its work and
the section successfully attacked and solved the current Greek, Swiss and Spanish code books. The Greek book was of the greatest interest; very long messages which were passing in great numbers between King Constantine of Greece and Berlin had, owing to the presence of Allied forces at Salonika, to be sent by wireless between Athens and Sofia and were duly intercepted. The solving of the code in which these messages were being sent proved of the very highest importance.29
It was not until 1917 that there was any exchange between Room 40 and MI1(b), and this was restricted to results. Ironically, MI1(b) exchanged technical information as well as results with the French military bureau and the American Military Intelligence Division in Paris. Room 40 received Mediterranean intercepts from the French Ministre de la Marine, collaborated with the Italians in the Adriatic and established liaison with and shared information with the Russians before the revolution. However, much could be learnt from the primitive but pioneering work of WW1 Service Sigint as it comprised the four main elements of Sigint – Interception, Traffic Analysis, Cryptanalysis and Intelligence (operational and research).30
According to AGD: ‘Room 40’s work was principally German naval, latterly a certain amount of diplomatic enemy and neutral, only using material obtained by interception of W/T by stations under Admiralty control.’ MI1(b) worked on ‘principally German military and some neutral and later even Allied diplomatic, the material for the latter being obtained from Cable Censorship under War Office control’ and that ‘only from 1917 was there any exchange’ between Room 40 and MI1(b) and even then ‘principally of results’. 31
The diplomatic work of MI1(b) and Room 40 was quite complimentary, given the different initial sources of material. Room 40 concentrated on Germany while MI1(b) monitored every other country. MI1(b) relied on censorship while Room 40 relied on radio and physical interception of diplomatic material from enemy states, mainly Germany. In September 1916, Room 40 discovered how Germany was evading interception of their diplomatic telegrams. Those to their missions in the Americas were going through Sweden’s Foreign Office and sent as Swedish telegrams. Also, the US State Department sometimes laundered German telegrams to and from Washington and MI1(b) had not spotted this. De Grey wrote to Hall on 21 September, noting:
It is now abundantly clear that telegrams are passing to Washington not intercepted by us [as postal mail] and not transmitted via Buenos Aires [as Swedish]. Neither can the telegrams omitted from our series be fitted into the wireless messages from Sayville or Tuckerton [US radio stations] – their number is by no means large enough. I consider it likely that they are sent via the State Dept. and the USA Embassy and might consequently be interceptable there.32
This kindled Room 40’s interest in cable censorship, which had been the preserve of MI1(b) and perhaps established collaboration between the two departments. The earliest communication is a note from AGD dated 15 October 1916 and by late December the bureaux were exchanging technical details of Greek and Spanish code book recoveries and usage.33
Room 40’s interest in cable censorship continued to grow and, from the end of September 1916, it received ‘American’/German telegrams from censorship. This inevitably led to cooperation between Room 40 and MI1(b). The earliest known contact is a note from AGD dated 15 October 1916. On 26 December, AGD mentioned, in a note about Greek keys, that ‘I am very glad this opportunity to avoid duplication of work has arisen and hope it may be carried through successfully.’
MI1(b) was brought together again in offices at 5 Cork Street Piccadilly, in the latter half of 1917. They had two outside telephone numbers and twenty-one extensions off the main War Office switchboard.
‘At Cork Street a private line had been installed from Major Hay’s office to the Admiralty and Sir Reginald Hall. Malcolm greatly admired the efficiency of this man, though he was of the opinion that he could be almost entirely ruthless, especially in what he considered to be the execution of his duty.’34
Staff numbers grew throughout 1917 as they solved the diplomatic codes of Argentina, Brazil, Denmark, Italy, Japan, the Netherlands, Norway, Persia, Sweden, Uruguay and the Vatican. By the end of the year, forty-three out of the fifty staff were working on diplomatic codes according to the official history. This compares to below ten in Room 40. Correspondence between AGD and Crocker continued through the year, with the last available example dated 30 October. During 1918, MI1(b) solved the code books of France, Peru and Romania.
By the end of the war, MI1(b) had eighty-four staff including thirty women and they were mainly based in a private house requisitioned by the War Office several streets away from its main building in Whitehall. Hay was very security-conscious and instigated an elaborate entry procedure which prevented visitors from wandering around the building. MI1(b)’s work was helped by keys and techniques for German military ciphers provided by the French. Intelligence from MI1(b) to army command soon started to flow.35 The unit broke the German Army political section’s cipher which was used to communicate to agents. Through this source and resumed liaison with its Admiralty counterpart, the navy sunk a German submarine in the Mediterranean carrying arms for Arab nationalists in North Africa.
An Admiralty memorandum in May 1918 says that the Director of Naval Intelligence (DNI) and the Director of Military Intelligence (DMI) each received all the available information from their own Service’s work and then communicated it ‘to whom they think fit’. MI1(b) is reported to have solved fifty-two diplomatic codes during WW1 and seems to have done most of the work on the American diplomatic telegrams for the period 1916–18, from which the British authorities learned about American attitudes to the war in Europe and the state of their relations with Germany. Again, AGD later noted that the diplomatic codes of the period were ‘most elementary’, only the German Foreign Office was using ‘hatted’ books and reciphering methods.36 All others were arranged alphabetically. Codes were being read from Japan, Greece, Spain and Scandinavia.
Hall had set up a diplomatic annex of Room 40 in 1915, reporting directly to him and headed by Sir George Young, a former diplomat. By 1917, Room 40 was receiving copies of German diplomatic telegrams from the War Office censorship as well as from the Admiralty’s intercept stations. Their work was helped by the acquisition of a German diplomatic code book retrieved from the baggage left behind in Persia by a German consul, Wassmuss, and acquired by Hall from the India Office in 1915. By 1916, Room 40 was reading Ge
rman telegrams between Berlin and the German Embassy in Washington. From these, Hall learned of plans for German assistance to the Irish nationalists in an armed uprising. He then arranged for the navy to intercept a shipment of arms on 21 April 1916. This enabled the authorities to capture Roger Casement, an Irish nationalist, activist and diplomat who was plotting an uprising with the help of Germany, within hours of him landing from a U-boat on the Irish coast at Tralee Bay. He was subsequently tried for treason and executed on 3 August 1916. Hall’s source of intelligence on the Irish problem dried up when the German Embassy in Washington closed in February 1917, the US having broken off diplomatic relations over Germany’s declaration of unrestricted submarine warfare.
MI1(b) never developed into an integrated military Sigint centre like the Admiralty section would become but it did have all of the hallmarks of an integrated diplomatic Sigint centre. It had little control over intercepts and poor interaction between cryptanalysts and linguists. However, Sigint activities were carried out in the field in France in support of the BEF. It achieved results of some value from decryption, traffic analysis and direction finding. Sections on each front could break the codes being used by its opponents. Results were sent to the Intelligence Branch of the General Staff. William Friedman, at the time a lieutenant with the American Sigint unit, G2A6, in France, was of the view that in 1917, very little cipher work was done at British GHQ ‘as they depended more or less on their offices in London for this’. During the static trench warfare on the Western Front, both sides obtained Sigint through the interception of line telegraphy and telephone conversations. Poor British security allowed the Germans to make a transcript of an entire British operational plan read out over the telephone in 1916 and inflict significant casualties. In early 1917, the Germans introduced a number of special code books as they extended wireless telegraphy in more forward positions.37