True to the words in the Room 40 ditty, AGD did indeed ‘go on’, and at the conclusion of hostilities, he was sent to Scapa Flow to interpret for Admiral Sir David Beatty, Commander-in-Chief of the Grand Fleet, at the surrender of the German fleet. He would later document his time at Scapa Flow1 and the problems that he faced:
I was a landsman on board a battleship, and a lot of my time was taken up in trying to conform to the life therein, no easy matter if you remember that it is really a foreign land where the inhabitants have a distinctive mode of life, even a distinctive language and very distinctive habits which to learn in a few days is not an easy matter for a visitor to this foreign land. My impressions of this particular foreign land have nothing at all to do with the matter in hand, namely the surrender of the German Fleet and the end of the motive power which has driven the British Navy for the last twenty years and made it such a wonderful machine. First of all I should like to confess that for the last four years I considered myself, and the department in which I worked, a very important cog in the machine; now for the first time I ran across the ‘business end’ of the weapon and I realised most strongly what a little cog we were. Practically no one I met had any idea of the existence of such a cog, which was satisfactory to know, as we had tried to conceal our identity. I had to keep a straight face, and lie right well to many an old friend from Osborne days whom I met up there, who wanted to know what my job was. On the whole I fancy I gave myself and my department a highly sensational appearance, such as would rejoice the readers of William le Quex.2
On arriving at Rosyth, AGD went directly to HMS Queen Elizabeth, Beatty’s flagship, to find out what duties were assigned to him. He met an old friend by the name of Spickerwell, who was secretary to the admiral. He actually had no interpreting to do but instead was to act as an intermediary between Beatty, who was concentrating on general policy and his opposite number, the commander-in-chief of the German High Sea Fleet. This dialogue was conducted by means of wireless. AGD was not impressed with Beatty, whom he regarded as ‘a very wilful man, and has no mercy on a man or nation he despises’.
They set sail around 2.00 am on 21 November and when Queen Elizabeth approached the location of the German fleet, action stations was sounded. AGD reported to Beatty on the signalling bridge and the German ships were spotted at 9.30 am. Beatty informed the Admiralty that he had taken them over and told AGD that he would rather have been able to report that he had sunk them instead. The British fleet then escorted the German fleet into the Firth of Forth.
After completing his duties with Beatty, AGD was transferred to HMS Lion, the flagship of the Battle Cruiser Fleet, to escort the German battlecruisers to the Orkneys. Once in Scapa Flow, he boarded the German flagship, Seydlitz, to oversee the voyage. AGD felt sorry for the senior German officers:
They were keen efficient men, who had learnt their work, and made the German Navy their career, and this was the end of it. We knew that many of them had fought a gallant action at Jutland, in fact, the Commodore, a fine looking old norseman with now a very sad expression had been captain of the Seydlitz at Jutland where she had been very badly hit. Only fine seamanship in his part had got her home, and now he had to save his ship for this end.
The rest of AGD’s stay at Scapa was, in his words, ‘dull beyond words’. On returning home, AGD would soon be asked to apply for a job which would shape the rest of his career.
***
In November 1918, along with his scheme for an amalgamated Secret Service, the DMI, William Thwaites, proposed to his naval counterpart, Admiral Hall, that their signals intelligence sections should be united into a single ‘School’. Calling the new unit a School would provide cover by stressing the organisation’s positive side, for example by studying ways to achieve secure communications.3 Thwaites had succeeded General Macdonagh, DMI at the War Office from 1916 to 1918, who had been promoted to adjutant-general. Colonel C.N. French, a senior figure in the Military Intelligence Directorate in London, had left the War Office as well and been replaced by an officer from GHQ, Malcolm Hay. Hay believed that Sir Henry Wilson4 was responsible for breaking up the staff of Military Intelligence at the War Office and that it was a great blunder. The reorganised Intelligence directorate was now composed almost entirely of officers from GHQ and Hay felt that:
Thwaites had no previous experience of Intelligence work, and no obvious qualifications for the position, or for the difficult task which it now involved of acting as military adviser at the Peace Conference. Sir George Macdonagh had acquired some useful knowledge about the political situation of Europe, his successor had none. The loss of Colonel Charles French was irreparable; he was perhaps the one man in the Directorate of Military Intelligence during the war who was indispensable.5
Hall agreed with Hay’s assessment and offered rooms in the Admiralty for the military side to merge ‘brain power’. French opposed a rapid amalgamation and argued that during the peace negotiations, the information produced by MI1(b) would be as, or perhaps more important than it had ever been during hostilities. Furthermore, the temperamental nature of the cryptographers meant that their move from Cork Street to the Admiralty could cause problems. Hay was unhappy with the proposed changes as well as the management of the Honours List:
The distribution of foreign decorations seems to have been successfully controlled at GHQ. An order was issued in 1915 that no officer would be allowed to wear any foreign decoration unless it had been personally presented to him for some specific service. A growing disgust with the traffic in medals was noticeable among soldiers during the latter period of the war. People in the War Office and at GHQ who stood, as the saying was, ‘nearest to the bag’, always seemed to secure the lion’s share of the spoil.6
On 10 December 1918, French proposed that a combined cryptographic unit be set up as part of a joint intelligence organisation.7 Hall was opposed to this but by mid-January 1919 he had been replaced by Commander Hugh Sinclair,8 who was more cooperative. A new unit would need to be responsible for the construction of British governmental codes and ciphers, work against foreign codes and ciphers and be able to secure a supply of raw material to replace the wartime censorship regime, which would expire with the Government’s emergency powers as soon as the peace treaty was ratified.
In January 1919, Lord Curzon (acting Foreign Secretary while Balfour was at the Paris Peace Conference), stated that the Foreign Office was ‘the proper place for the new school to be housed’. Sinclair disagreed and argued that it should be located in the Admiralty, as the Service Ministries had the required expertise and ‘all the arrangements as regards deciphering messages were already in existence in the Admiralty building’. He went on to say: ‘Without wishing to disparage the Foreign Office in the least, it is considered that the atmosphere of calm deliberation which characterizes that department is not suited to an organisation such as the proposed Code and Cypher School, which, above all things, must be a “live” undertaking, especially in connection with the “breaking” of codes and cyphers.’
By 27 February the Admiralty, War Office, Air Ministry and Foreign Office were agreed that there should be a new ‘Code and Cypher Department’ comprising a code-making ‘Code and Cypher School’ of about twenty-five people with a budget of £5,000–£6,000 per year and a codebreaking ‘secret sub-section’ with a budget of £10,000 per year. The War Office drafted a clause for the new Official Secrets Bill which required cable companies to provide all telegrams to the new organisation. In March, the War Office proposed that the new organisation should be headed by Hay with twenty-six staff and a budget of £8,250–£13,400 per year.
In a memo to Lord Drogheda of the Foreign Office dated 28 March 1919, Churchill, now serving as Secretary of State for War and Secretary of State for Air, gave the Admiralty’s view of the proposed Code & Cypher School:
Private and confidential
March 28, 1919,
Dear Lord Drogheda,
I had a few minutes conversation with Lord Curzon yesterday on the subject of the new Cypher Department which it is proposed to establish, and concerning which a memorandum is now in his hands, containing the views of the Admiralty and the War Office.
Lord Curzon told me that he hoped to summon a conference at the Foreign Office one day next week to consider the matter, when I should have an opportunity of stating my opinions, and he asked me to send you in advance a memorandum of the points I wished to raise. I therefore send you herewith the following notes of matters which the Admiralty consider essential in any scheme that may be adopted.
We have in the Section of the Naval Intelligence Department which has dealt with enemy wireless during the war, a great deal of material, some of which is worked out and filed for reference or historical purposes and some of which will require further study. We also have a small remaining nucleus of the expert staff which has done this work during the war. If the Admiralty is to join the new Department, we regard it as essential that this material and staff should be kept together.
Wherever the new Department may be located in peace time, we should have to stipulate that on the outbreak of war the naval portion of its staff should immediately be mobilised and take up their work in the Admiralty. Our experience has proved that in war the deciphering staff must be in the closest possible proximity to the War Staff. We have had to work day and night all the year round, and as immediate action has often had to be taken in consequence of the information which we have supplied, no avoidable delay in transmitting the information to the Operations Division can be allowed.
We should only consent to pool our staff with that of the War Office on condition that Commander A. G. Denniston is placed in charge of the new Department.9 I do not say this on account of any jealousy of the War Office, or any reluctance to accept a War Office man, but because no one who has not been trained in the conditions under which we have to work could meet the requirements of the Admiralty in time of war. Our work has been done in the face of the enemy and always against time. The messages we have had to decipher were from ships at sea, engaged in actual operations, or from airships also operating. We have had to master a new key every morning before we could begin to read the messages, and sometimes we have had to grapple with two keys in one day!
This has of necessity developed a particular kind of aptitude for the work, which depends for its success more on a study of the psychology of the persons sending out the messages and a sort of instinctive ‘flair’ for the kind of things they are saying, than upon careful study and analysis for which there is no time.
In the War Office they have dealt with cables which are far more accurate than wireless, and have never had to work against time, and the aptitude they have developed is different from – I do not for a moment suggest it is inferior to that of which the conditions of our work have produced. Denniston is not only the best man we have had, but he is the only one we have left with special genius for this work. We shall not be able to retain him in a subordinate capacity, and no advantages of concentration and cooperation with the War Office would compensate us for the loss of his services. If the War Office people are not willing to accept this condition, we should prefer to retain our staff in the Admiralty, but should of course cooperate with them in every other way that is possible.
A conference was held at the Foreign Office on 28 April 1919 to consider the question of the proposed new Code & Cypher School. Present were The Right Hon. Earl Curzon of Kedleston, KG (Chairman), The First Lord of the Admiralty (Walter Long); The Secretary of State for War and Air (Winston Churchill); The DNI, Admiralty; The Deputy Director, Military Intelligence (DDMI), War Office; Captain W.M. James, Deputy Director, Naval Intelligence (DDNI); Captain R.L. Nicholson, Director of Signals Division; Major H.E. Franklin DSO, MC, Secretary; The Earl of Drogheda, Foreign Office. It noted that:
The Chairman summarized briefly the recommendations of the Inter-departmental conference which recently met to consider the matter, and said that the main question now before the meeting was the housing of the new department, with the establishment of which everyone in principle agreed. In his opinion the arguments in favour of housing the new department in the Admiralty in time of war were unanswerable, but we were providing for its establishment under peace conditions, and in time of peace he thought that the fact that the interest of the intercepted telegrams was practically entirely political indicated that the new department should be housed in the Foreign Office.
It was decided that Curzon, as acting Foreign Secretary, should receive all intercepted telegrams and be responsible for passing them on to the Prime Minister or other Cabinet Ministers concerned when they were of sufficient importance.10
A further meeting was held at the Admiralty on 8 May 1919 to further consider the formation of a Code & Cypher School. Present were the DNI, Commodore H.P. Sinclair CB, RN; the Earl of Drogheda, Foreign Office; Captain R.L. Nicholson DSO, RN, Director of Signals Division; Captain W.M. James, DDNI; Lieutenant Colonel W.E. Wynn OBE, Air Intelligence; Commander B. Buxton DSO, Admiralty; Mr. A.P. Waterfield, Treasury; Major M.V. Hay, Reserve of Officers; and Major H.E. Franklyn, DSO, MM. It agreed that the new unit would be staffed by a head (salary of £1,200), senior assistants (salary of £600–£800 plus war bonus), junior assistants (salary of £200–£500), translators (salary of £200–£300) and clerks (salary in accordance with ordinary gradings).
The question of who would head the new organisation was yet to be decided, with AGD being the Admiralty candidate and Hay that of the War Office. A meeting was held at the Admiralty on 5 August 1919 to resolve the issue.11 The meeting was chaired by Sinclair, with James, Drogheda, Major-General Bartholomew (DDMI), Colonel Dick (Assistant Director of Military Intelligence, ADMI) and Franklyn in attendance. Sinclair and Bartholomew had interviewed the two candidates. Hay had made it very clear that he would only serve as Head and while he would accept AGD as part of his team, he didn’t rate him or want him on it. AGD on the other hand, would work as head or for Hay ‘at all events for a time’. Bartholomew thought it was ‘intolerable’ that Hay ‘should attempt to dictate his terms of service in such a manner’. Franklyn noted of Hay and AGD that ‘when they were together they could not agree’ and was of the view that MI1(b) workers at Cork Street didn’t like AGD. James replied that everyone at the Admiralty ‘had the highest respect amounting to affection’ for AGD. Drogheda thought that Hay was probably cleverer but AGD was a better administrator.
After due consideration, it was agreed that AGD would head up the new organisation.12 While no official statement of the reason for this decision exists, AGD was probably regarded as a safer pair of hands who looked upon Hall as a father-figure and the Royal Navy as his home. He could be trusted to know his place in the hierarchy as a subordinate to DNI and provide leadership to the cryptanalysts. The words ‘Denniston will never desert his solitary post’ would prove to be prophetic. Hay, on the other hand, clearly had strong views and pressed for independence from DNI as long as he ‘produced the goods’. His notes13 show him to be a prickly individual who ‘for many years after the war refused even to speak to a General’. He refused the OBE offered to him on 12 December 1919.14 His attitude is perhaps best summed up after he departed quickly on 21 August 1919 by his one-sided view of WW1 codebreaking successes:
Before decoding the messages, [we] had to reconstruct the code books … All these difficulties were overcome. Cork Street was never defeated … Some publicity has been given to the fact that German Naval messages and German Diplomatic wireless messages between Berlin and Washington were intercepted by the Admiralty and read by a section of Naval intelligence housed in Room 40 O.B. All or nearly all of these German intercepts were in code. Various stories are current about the way copies of the German code books were obtained. I do not know which of these stories is the true one. But it is certain that these encoded messages were not read by reconstructing the code books without some outside assistance.
On 24 October 1919, R.R. Scott wrote to the Secretary, War Office; The Secretary, Air Ministry; The Under Secretary of State for Foreign Affairs, Foreign Office; The Under Secretary of State for India, India Office; The Secretary, Ministry of Munitions; The Secretary, Ministry of Food; The Secretary, Ministry of Transport; The Secretary, General Post Office; and the Under-Secretary of State for the Colonies, Colonial Office, making the following points:
That the War Cabinet has now given approval for the formation of a Government Code & Cypher School under the control of the Director of Naval Intelligence, and that it is proposed that it should commence its duties on the 1st November, 1919.
It has been decided to appoint Commander A.G. Denniston, O.B.E., R.N.V.R. as Head of the Government Code & Cypher School, which will be accommodated in Watergate House, Adelphi, W.C.I. The duties of the Code & Cypher School will be as follows:-
To compile and be responsible for printing all codes and cyphers used by the British Government Departments with the sole exception of those mentioned in paragraph 5 below (this refers to Signal Books and purely Departmental Codes of the three fighting services. However, GC&CS was to advise on the general principles of their construction and the limitation of their ‘life’. GC& CS would decide on what is classified as Departmental Codes after consulting relevant Department).
To examine all the British Government Codes and cyphers now in force and the purpose for which they are used, mainly with a view to ascertain and, where necessary, increasing their degree of security; but also so as to ensure that messages shall be free from ambiguity and undue delay ensuing from mutilation in transit, and that they shall be coded in the most economical manner possible.
To maintain the closest liaison with all British Government Departments using codes and cyphers, and to advise them generally in matters relating thereto.
To instruct as large a proportion of Officers as possible who may be employed at any time in coding or cyphering.
To assist in the preparation of any hand-books or instructions relating to coding or cyphering, or of those concerning the handling of code and cypher messages in general.
It was suggested that the following departments appoint a Liaison Officer with GC&CS: War Office, Air Ministry, Foreign Office, India Office, Colonial Office, Ministry of Munitions, Ministry of Food, Ministry of Transport, General Post Office.
AGD’s new organisation was formally called the Government Code and Cypher School (GC&CS). The name itself was invented by Courtney Forbes, a member of the Communications Department of the Foreign Office. Publicly it was ‘to advise as to the security of codes and cyphers used by all Government departments and to assist in their provision’. However, its secret directive was ‘to study the methods of cypher communications used by foreign powers’. Pressure from DNI and others forced the inclusion in the new Official Secrets Act of a clause instructing all cable companies operating in the UK to hand over for scrutiny copies of all cable traffic passing over their systems within ten days of despatch or receipt. The discussions in 1919 made it clear that the Foreign Office and Lord Curzon recognised the potential of the diplomatic decrypts from GC&CS.15 Interestingly, the three services were expressly excluded from needing GC&CS’s advice.
GC&CS was up and running very quickly and its first decrypts were issued two days after its formation. The new organisation was housed as agreed in Watergate House in the Adelphi in London. Lieutenant-Commander Edward Travis16 was appointed to run the Construction Section and act as Deputy Head. Travis had experience of naval code book construction, so he took responsibility for cipher security while AGD supervised cipher-breaking. Travis also advised the Admiralty on communications security. However, GC&CS had no authority to advise on good security practice, so his role was limited. This may have led to his losing interest in security, hence the two Admiralty security staff assigned to GC&CS ended up working as cryptanalysts. In the end, the advice from GC&CS to the Admiralty about code and cipher security was very poor.
By December 1919, the GC&CS staff included five seniors (three from MI1(b), two from Room 40), seventeen juniors (ten from MI1(b) and six from Room 40, one from the Foreign Office), three female translators from MI1(b) and thirty female clerks for the Construction Section. Almost all of the traffic being dealt with was of a diplomatic or commercial nature, and it is estimated that between 1 November 1919 and 21 January 1920, 900 decrypts were distributed as follows: Argentina, Denmark, France, Italy, Japan, Norway, Persia, Romania, Sweden, United States, Uruguay: 54.3 per cent; Austria, Chile, Germany, Russia: 14.6 per cent; Greece, Spain: 26.8 per cent; Kingdom of the Hejaz, Poland, Syria: 4.3 per cent.17
In mid-1921, Curzon had succeeded Balfour as Foreign Secretary and described the work of GC&CS as ‘by far the most important branch of our confidential work. The decrypted telegrams of foreign Govts., are without doubt the most valuable source of our secret information respecting their policy and actions. They provide the most accurate and, withal, intrinsically the cheapest, means of obtaining secret political information that exists.’18 In February 1921, Walter Hulme Long, who was interested in intelligence matters, was replaced as First Lord at the Admiralty by Arthur Hamilton Lee. On 25 April 1921, Lord Curzon, who had previously argued for GC&CS to be in the Foreign Office, wrote to Lee:
I think I mentioned to you a little while ago that I proposed with your assent that the Code and Cypher School, which during the war was for very good reason placed under the Admiralty [Curzon seems to have forgotten about MI1(b)], should now be taken over by the Foreign Office, to which nine-tenths of its work appertains.
I was the Chairman of a Conference just two years ago … at which it was decided to continue the Code & Cypher School in existence [sic], and to house it, at any rate for the present, in the Admiralty. The reasons for this decision were that it would be undesirable to deprive the School of the shelter which the Admiralty buildings afforded, that the vote would be more likely to slip through the House of Commons if it came under the Admiralty disguise, and that the Foreign Office might feel more compunction if through the interviews of the Secretary of State with foreign Ambassadors and Ministers he were to profit by information which his own Department had secretly acquired [this rather obscure point does not appear in the minutes].
These reasons struck me at the time as very flimsy – nor did I entertain any of the qualms described in the last paragraph. I deferred however, to the representations of my colleagues … I now realise that [the arrangement] is both illogical and indefensible.
It is wrong in principle … the sphere of the activity of the School is now purely political, and the intelligence procured has, except in very rare cases, no relation to, and can be of no value to, the Admiralty.
In some cases, as you may know, our possession of the ciphers has been detected by foreign Powers, and in the consequent change of ciphers by them we have lost almost immeasurably.
I cannot doubt that there will be an increase in efficiency if the School is brought under the Foreign Office, since the greater part of its work ought to be done in the closest cooperation with us, and our experience of the matter is now very considerable.
Even the argument of the Admiralty ‘cover’ has ceased to apply; for whereas I was led to believe, two years ago, the Department was likely to be housed in the Admiralty building, it is now domiciled in separate quarters, as detached from one Office as from the other.
What has terrified me most has been the too generous and careless distribution of the material under the existing system. I have already been obliged to resume the powers conferred upon me by the Cabinet Committee of deciding to whom the intercepted telegrams should go, and the danger of a profuse or undiscriminating distribution, I hope, no longer exists.
In May, Lee, who had little interest in intelligence, agreed with Curzon’s views and Sinclair did not object. Another factor might well have been the proposed cuts in public expenditure which were aimed mainly at the armed services. These were introduced in 1922 and became known as the ‘Geddes Axe’.19
On 23 July 1921, the Foreign Office told the Treasury it was taking over GC&CS from the Admiralty along with its staff of eighty-seven and operating costs of £31,464 per year. The change was effected on 1 April 1922 for Treasury budgetary reasons. Sinclair returned to Intelligence as Head of SIS following the death in June of Mansfield Cumming, the Service’s first head.20 By September, he had arranged for GC&CS to come under his control. To appease the Service Ministries, who had complained vigorously in April 1923 that GC&CS had lost its interdepartmental character since it had come under the Foreign Office, the Foreign Office agreed to return five named individuals to the Admiralty in the event of war.21 The Permanent Under-Secretary at the Foreign Office, Sir Eyre Crowe, devised the compromise of AGD reporting to Sinclair as both Head of SIS and Director of GC&CS but in turn, Sinclair would report to the Foreign Office.22
Sinclair and his successor would remain both Head of SIS and Director of GC&CS for more than twenty years. While the armed services kept some intelligence-gathering expertise in the field,23 SIS in effect acquired monopoly control over British Sigint. This proved to be an effective strategy, and GC&CS provided Whitehall with a constant stream of intercepted and decrypted foreign governments’ telegrams. It read the communications of France, Italy, the United States and Japan, and that of many smaller countries. The historian John Ferris later noted that ‘the GC&CS was one of the world’s largest code-breaking agencies, perhaps the biggest; as effective as any other, better than most, possibly the best on earth between 1919 and 1935’.
AGD’s organisation had moved to Queens Gate in 1921 and when Sinclair took over in 1923, he brought SIS and GC&CS together. By 1925, they occupied the third and fourth floors of the Broadway Buildings. This move was opposed by Lord Curzon as it was in Kensington, two miles from Whitehall,24 but according to AGD, in Kensington ‘we were more comfortable rather remote from other departments’.25 At the end of 1923, GC&CS’s staff numbered ninety-one and Sinclair told the Foreign Office that he wanted to reorganise the Service to meet the demands of war.26 The burden of keeping GC&CS afloat fell on the shoulders of AGD and he regarded his organisation as ‘an adopted child of the Foreign Office with no family rights, and the poor relation of the SIS, whose peacetime activities left no cash to spare’. In reality, GC&CS’s funding was comparable to other parts of Sinclair’s Service. However, he rarely complained since his dogged fidelity as a public servant was equalled by his dislike of publicity. AGD’s problem was that because of the secrecy, few in Whitehall knew of the war-winning work of Room 40 and AGD’s organisation had few supporters. He was not in a strong position to fight for funds and staff in the lean interwar years. He and his colleagues worked on a shoestring and in a kind of Civil Service limbo. Most Whitehall insiders thought that Room 40 had been wound up at the end of WW1 and were unaware of AGD’s new organisation.
Sinclair met with AGD in January 1924 and confirmed that he wanted to integrate the work of GC&CS with that of SIS.27 GC&CS would be responsible for cryptography and SIS for the distribution of intelligence derived from them. SIS would also supply intelligence and criticism to GC&CS to assist cryptography. Section I of SIS was to ‘supply GC&CS with a list of general subjects on which to concentrate…Armed Forces Sections of SIS to collaborate’. GC&CS was to ‘have full access to SIS records’. By June 1924, because of the large volume of ‘intelligence product’ available, AGD was told to distribute it directly to GC&CS’s customer departments and send copies to Sinclair. However, SIS kept overall control of the distribution of ‘Sigint product’.
In 1924, the Cabinet appointed a committee, with General Romer as chairman to advise on the anti-aircraft defence of the UK. One of its recommendations was that the control of the ‘Wireless Interception Service’ should be taken over by AGD ‘at the request of the Fighting Services and with the consent of the Foreign Office’.28 To oversee formal coordination within the intelligence services, the ‘Cryptography and Interception Committee’ (later the ‘Co-ordination of W/T Interception Committee’) was set up.29 Sinclair chaired the committee, with membership drawn from the three Service Ministries and GC&CS. The Secretary was an SIS officer, Colonel Arthur Peel, who had previously been Assistant DNI to Sinclair during his tenure as DNI. Peel helped develop a relationship between GC&CS and the Metropolitan Police during the 1920s. In the late 1930s, he helped coordinate the development of wireless intercept service in the Dominions.30 The Committee’s remit went beyond interception and covered cryptanalytic training and war planning, although it never had a controlling role, only one of coordination. While the police interception work was initially intended for suspected illicit wireless transmitters in Britain, from 1927 they began to intercept diplomatic wireless traffic for GC&CS.31 A subcommittee was set up initially under Peel and then AGD from the mid-1930s. As it only met every two years, it was agreed in July 1928 to form a standing sub-committee which ‘should concentrate on formulating definitive recommendations, and should in future prepare agenda for the main committee’. This new ‘Y Sub-Committee’ as it became known, met fortnightly under AGD and included representatives from the three Services, Scotland Yard, the GPO and the Head of the W/T Board.
GC&CS had to handle its own telegram collection service with an SIS car collecting material from the General Post Office and commercial communications companies daily and delivering them to Broadway, where they were copied and returned within twenty-four hours. The Cable Intercept Section, under Henry Maine, controlled the collection effort. Similar collections took place in India, Tehran, Haifa and Jerusalem. Between 1920 and 1927, GC&CS was producing on average 3,500 reports annually. These were known as ‘flimsies’ or ‘BJs’ (the file covers used to circulate GC&CS reports in the Foreign Office were blue, hence ‘blue jackets’. Confusingly, in the early 1920s, army reports were referred to as ‘black jumbos’.)
The GC&CS reports were verbatim transcriptions of decrypted messages, translated unless the original was in English or French, in which case the customer would be able to read it for himself. GC&CS did not attempt to provide context or explanation apart from crossreferencing them with related reports. AGD told colleagues within GC&CS that their role at this point was simply decryption. This was openly opposed by some in GC&CS such as William Clarke of the Naval Section, a veteran of Room 40. He firmly believed that GC&CS should be producing intelligence reports, as its predecessor Room 40 had done. However, Sinclair and AGD knew that customer departments regarded intelligence assessments as their domain and would have resisted any attempt by GC&CS to overstep agreed boundaries.
The distribution of ‘BJs’ changed from a restricted circulation of Admiralty, Foreign Office and War Office to include the Prime Minister, Lord Privy Seal, Colonial Office, India Office, Air Ministry and the Home Office ‘Directorate of Intelligence’. GC&CS’s Naval Section was the first in GC&CS to be dedicated to military matters, and worked against specific foreign navies.
It was the only GC&CS military section funded by GC&CS and the only one to use non-cryptanalytic techniques such as traffic analysis. When GC&CS had moved to Foreign Office control, Clarke became one of GC&CS’s liaison officers to the Admiralty and eventually head of the new Naval Section in GC&CS. Sinclair, as a former DNI, took a personal interest in their work and Clarke frequently bypassed AGD. However, from 1925, Sinclair, AGD and the Admiralty put constraints on the overzealous Clarke.32
***
With his heavy workload, AGD found it difficult to find time for friends and family. His wife Dorothy worked regularly at the Chelsea Day Nursery and at the Infant Welfare Centres in Chelsea and Shadwell and would continue to do so until 1939. The couple were very much in love and he talked freely to her about his work. When she became pregnant in 1924, they called the child ‘X’, assuming it was male. When their daughter Margaret was born instead on 21 May 1925, they called her ‘Y’. The much anticipated son Robin was born 19 months later on Christmas Day 1926. AGD was a good father, even though he was so busy, as Robin remembered:
Through all this my father was perhaps the most important part. When I fell downstairs during a party (at age four) I was brought down, screaming with pain and fear, to my father sitting alone in the drawing room, on plumped up cushions; and he read me Winnie the Pooh to our mutual delight until the panic was over. He went with me to fetch the car from the garage in King’s Street, and was interested to note that I noted that the steering did not work well, thanks to there being a puncture in one of the front tyres. He drove a snub-nosed Morris with great care, particularly down to Bartonon-Sea for the summer holidays, where we had a small bungalow and rented a bath hut. Modest family picnics, sea bathing, some agreeable adult company including several from the Office; the beginnings of golf and tennis.
Robin had a nanny called Sheila. She was young and the daughter of a sergeant-major at Camberley. She was followed by a number of governesses, who he despised, until Dorothy decided to look after her own son, unusual in those days for those of their social standing. He attended nursery classes in Tedworth Square and then a day school called Mrs Spencers (an early form of elementary school, usually taught by women in their own homes) in South Kensington. ‘Y’ also went there and AGD would walk them to school every morning before catching the Underground from Gloucester Road to St James’ Park and his office at Broadway. AGD worked long hours, six days a week but took a few weeks off in the summer. As war came nearer, his holidays were frequently interrupted. He was a benign presence who obviously enjoyed family life. He often went to play golf or tennis but preferred if the family came with him. The family attended King George V’s funeral in January 1936, and Robin was the bearer of a bouquet for Mrs Stanley Baldwin when she visited Shadwell Infant Welfare Centre where Dorothy did volunteer work.
***
In April 1927, information from a Chinese raid on the Soviet Embassy in Peking reached London. It revealed that Arcos, the official Soviet trading organisation in London, was a centre of Soviet espionage in Britain. On 12 May 1927, several hundred policemen raided the Arcos offices in ‘Soviet House’ at 49 Moorgate in London. Over several days, police and intelligence officials removed numerous documents. Considerable intelligence was obtained of interest to both Britain and the US. However, to justify the raid, the Government revealed to Parliament that the intelligence obtained demonstrated Russia’s hostile intentions with regard to diplomatic and trade relations. Remarkably, the Prime Minister revealed that Britain could read the most secret Soviet cipher traffic. This was to provide proof to the Government’s parliamentary opponents that the intelligence was accurate. However, it was done despite lobbying from AGD and Stewart Menzies,33 Assistant Director for Special Intelligence in SIS. As AGD later noted: ‘The only real operational intelligence came from our work on Soviet traffic. We were able to attack their systems step by step with success from the days of Litinov’s first visit to Copenhagen, of Kamenev as their first representative in London followed by Krassin.’34 The Soviet reaction to the Prime Minister’s statement was immediate: ‘Until the famous Arcos Raid in 1927 when HMG found it necessary to compromise our work beyond any question. From that time the Soviet Government introduced OTPs [one-time pads]35 for their diplomatic and commercial traffic to all capitals where they had diplomatic representatives.’ However, Soviet diplomatic traffic continued to be read by GC&CS for some time.
By the mid-1920s, GC&CS staff numbered twenty-five officers (one head, six senior assistants, eighteen junior assistants) and about twentyeight clerical staff (six typists, twelve clerks for code construction and ten traffic sorters and slip readers36). AGD began recruiting staff from the universities as early as 1925. Initially, those brought in were classicists, linguists and papyrologists, and the establishment increased to ten seniors and twenty juniors. As there was virtually no difference between the work of good juniors and seniors, in the early 1930s the balance was changed to fifteen seniors and fifteen juniors. The services also contributed staff and from 1923, the Admiralty’s interest in Japanese diplomatic and naval attaché traffic led to the permanent placing of a Japanese interpreter officer in GC&CS.
The War Office established a station at Sarafand in Palestine, with an intercepting and cryptographic unit which had close links to GC&CS. In November 1925 AGD visited Sarafand to research the setting-up of wireless stations there. He arrived on 7 November and wrote to Dorothy every few days. ‘Y’ was almost six months old and AGD was clearly missing his family, ending one letter ‘Good night dearest D.G. & kiss wee Y on the brow for me’. He carefully restricted himself to general descriptions of life in the areas he visited. While much of his time was spent travelling to potential wireless station sites in Egypt and Palestine, he still managed to fit in the odd game of tennis and round of golf. He boarded a ship, the Maloja, on Sunday, 22 November, at Port Said and arrived in Marseilles on Friday, 27 November. He then took the train to Paris and the boat train from Calais, arriving home the following day.
The War Office also sent officers to GC&CS for training before being posted abroad. Pre-eminent amongst these was John Tiltman,37 who would remain a close friend of AGD for the rest of his life. Older specialists were drawn from the original Room 40/MI1(b) staff, such as Ernest Hobart-Hampden, former consul in China and Japan, and Ernst Fetterlein,38 a former Tsarist cryptanalyst.
On 29 July 1926, AGD’s deputy, Edward Travis, in his role of protecting British codes and cyphers, received a note from R. Hume, probably from the embassy in Berlin. The Admiralty wished to purchase two Enigma encryption machines which were commercially available in Germany. Hume informed Travis that a new machine was under development which was cheaper, simpler and more fool-proof, although not available for ten months. The existing large machines cost 2,000 marks (less than £100) while the smaller machine was 600 marks (less than £30). Travis replied by telegram in September, saying ‘Am proceeding Prague to inspect machine for War Office. Admiralty wish me to break journey at Berlin and ascertain particulars of new Enigma.’ Travis duly travelled to Berlin and purchased one machine for the British Government and brought it back to London.39 It survives to this day and is on display in alternate years at Bletchley Park and GCHQ.
The Armistice which marked the cessation of hostilities on the Western Front had seen the end of military wireless traffic. This had enabled Lord Curzon to argue successfully for the move of GC&CS to the Foreign Office. Over the following years, the focus of AGD’s organisation was very much on diplomatic traffic. Some elementary work had been carried out from 1914–18 by both Room 40 and MI1(b). AGD had taken a small Admiralty party to Paris in April 1919 to work with the French on German material during the Peace Conference and stayed until the signing of the Peace of Versailles. He regarded the trip as ‘useless though pleasant’, the latter because the party included Dorothy.40 The Germans came to the conference with new code books and methods, not surprisingly following the Zimmermann Telegram episode. In 1919, only a small number of staff had any real cryptography expertise and most were linguists. The reconstruction of code books being used by various governments and the translation and editing of the resulting text was GC&CS’s primary function in the early years of its existence. Fetterlein, Strachey and Knox were the key men initially, along with Turner as master-linguist and Hobart-Hampden in charge of Japanese work.
Code books, known as ‘hat books’, were used by the Germans and within one year of GC&CS’s inception, they could be solved by one good linguist. The work was based on a method used by Ernst Fetterlein for many years in this type of work in Russia. John Tiltman joined Fetterlein’s team in the summer of 1920 and later recalled their work:
I worked as one of a group of from 5 to 7 persons on Russian diplomatic ciphers under the direction of Ernst Fetterlein. Fetterlein had been Chief Cryptanalyst of the Russian Czarist Government and held the rank of both admiral and general; he had practiced cryptanalysis since 1898 or earlier. At the Revolution he walked out of Russia across the Finnish frontier and was specially naturalized on arrival in England.
At the time of my arrival, Fetterlein’s small section was entirely occupied with the solution of the current Moscow-London and London-Moscow diplomatic traffic intercepted in the cable office. All messages were enciphered by simple columnar transposition of Russian plain text conventionally transliterated out of Cyrillic characters, As each message was transposed on a different key, all messages had to be individually solved. The average delay, was I believe, 1 or 2 days. 41
The traffic of two British allies was also read, as AGD recalled:
The Americans celebrated the advent of peace by introducing a new hatted diplomatic code recyphered with tables changing quarterly. The solution of the first of these tables was a year’s work and thereafter the American Section had to be expanded for the increased task of breaking the tables and reconstructing the code. Good progress was made and the section was able to be of some assistance during the Washington Naval Conference of 1922.
The second really big task was to make a concentrated attack on French Diplomatic cyphers, which had received no attention during the war.
A large number of hatted books of 10,000 groups were used and with the constant practice of reconstruction of such books they never presented any difficulty. Given sufficient traffic, legibility appeared with a month of birth. Many recycled books also appeared and after the initial struggle to obtain the general system the constant change of tables presented little difficulty.
The reading of this traffic during the years of peace and intrigue did from time to time produce very interesting if not invaluable intelligence. But the proximity of the two capitals did mean that a great deal passed by bag.42
Only the Soviet traffic yielded operational intelligence. According to AGD:
The Revolutionary Government in 1919 had no codes and did not risk using the Czarist codes which they must have inherited. They began with simple transposition of plain Russian and gradually developed systems of increasing difficulty. The presence of Fetterlein as a senior member of the staff and two very competent girls, refugees from Russia, with a perfect knowledge of the language, who subsequently became permanent members of the staff, enabled us to succeed in this work. We were also able to borrow certain British Consuls who could not return to Russia.43
A major effort was made on Japanese diplomatic traffic and was largely productive. It was led by Hobart-Hampden who had thirty years of service in Asia. While no more than 20 per cent of the intercepted traffic was read and circulated, the Section was able to provide the views of the Japanese Government in advance of major conferences. Hobart-Hamden was joined by another former member of the British service in Japan, Sir Harold Parlett, in 1926.
A watch was kept on all former enemy countries, and it was known that Germany was using OTPs and a second method nicknamed ‘Floradora’, which was eventually broken during WW2. Austrian traffic was read in 1918–19, thanks to work by Fetterlein. Knox successfully read some Hungarian traffic and, building on work in MI1(b) and Room 40, traffic was read from Greece, Spain, Italy, Scandinavia and Persia. Targets were driven by the politics of the day so new sections looked at various South American republics, Portugal, Brazil, the Balkans and Near East.
AGD summarised GC&CS’s effort on diplomatic traffic over its first twenty years as follows:
To sum up the cryptographic effort of twenty years on diplomatic traffic: we started in 1919 at the period of bow-and-arrow methods, i.e., alphabetic books; we followed the various developments of security measures adopted in every country; we reached 1939 with a full knowledge of all of the methods evolved, and with the ability to read all diplomatic communications of all powers except those which had been forced, like Germany and Russia, to adopt OTP.
The authority who sanctioned our Establishment in 1919 clearly never envisaged a complete reading, translation and issue of every telegram received by us.
Such was a physical impossibility for the thirty specialists who composed the main body of the staff employed on the work.
Hence from the outset sections did exercise their own discretion as to what they translated and submitted for circulation. They got guidance from the D and R who in turn received intelligence directives from the Foreign Office, the circulating sections of SIS and the officers who used our material in the Service and other large departments.
During the thirties we did supplement our daily issue by a daily ‘Summary of telegrams decoded but not circulated’, for the benefit of SIS, Admiralty and War Office (occasionally the Foreign Office) and it is noteworthy that it was only a very small percentage that were ever asked for in complete form.
With personal satisfaction I maintain that GC and CS did during those twenty years fulfil its allotted function with success, with exiguous numbers and with an absence of publicity which greatly enhanced the value of its work.44
Apart from the diplomatic traffic, unusual transmissions were picked up around 1930 which turned out to be a worldwide network of clandestine stations controlled by a station near Moscow (the Comintern network). The police station at Denmark Hill45 in South London obtained German diplomatic traffic broadcasts in 1937–8 from an unlisted station in Germany to unknown recipient call signs as well as obvious replies from unknown stations. Interception, traffic analysis and direction finding helped in these early days to map the traffic between German embassies, legations and consulates.
GC&CS had no W/T intercept facilities of its own and was totally dependent on the Admiralty and War Office for material to work on. Sinclair had persuaded the Admiralty to retain its intercept stations at Scarborough and Pembroke at the end of WW1. The Military Directorate in the War Office had also agreed to retain its station at Chatham. While there was little German naval traffic in the early 1920s due to the fact that Germany had no real navy, GC&CS’s Naval Section was probably set up in 1921 when Clarke joined after completing a naval history of WW1. What German naval traffic there was could not be read and, by the mid-1920s, this could well have been early Enigma traffic. Italy did have a navy, however, and its traffic was read by reconstructing their main naval code book. This work was helped by the Italians’ habit of enciphering long political leaders from the daily press. Around 1934, Italian naval traffic increased, albeit in a more secure form, so GC&CS’s naval Italian section continued to grow to keep pace with it. Their reports kept DNI well informed about Italian naval plans. During the Spanish Civil War between 1936 and 1939, the Italians introduced the commercial Enigma machine for all of their secret naval communications. Hitler had issued an order for ten Enigma machines to be sent to Franco in October 1936, so it is likely that some were also provided to Mussolini at the same time. This gave GC&CS and a team led by Knox and including William Bodsworth, its first opportunity to research machine encryption. This pioneering work would prove invaluable in the years to come and it was Bodsworth who broke Enigma traffic between Franco’s navy and the Italian Navy, the Regia Marina, in late April 1937.46 By 1936, a large amount of German naval traffic started to appear in the Mediterranean. Knox made some progress but by 1937 the security of the German naval Enigma machine had been significantly enhanced with an attachment known as the steckerboard.47 The Germans also added code books as part of the daily Enigma setup procedure and Knox made no further progress. The only source of intelligence came from an early form of traffic analysis which looked at the origins of traffic and plotted the routes that it followed.
A start was also made on Japanese naval traffic, and while there was no interception in Britain, a steady flow of material was delivered by bag to London. A small bureau for interception and cryptography was established, initially in Hong Kong and then Singapore in 1939. AGD summarised the situation in 1939 as follows: ‘To sum up the situation of the Naval Section in 1939, including the Japanese branch in Hong Kong: they exercised a very fair measure of control of all Italian and Japanese naval cyphers; they had only seen German signals by the Enigma machine and this they could not read; they had started an intensive professional study of raw German traffic with a view to extracting any available intelligence.’
Many of the foundations of a united Sigint Service were laid in the post-WW1 arrangements, but the Service Ministries seemed to regard them as peacetime arrangements only. War Office policy was summarised in 1925 as follows:
On the outbreak of war the War Office will be responsible for intercepting the enemy’s field wireless sets, and for collecting all information obtainable from this source. For this purpose it will provide, from officers on the active list and on the reserve, the necessary personnel for wireless intelligence and cryptography.
At this stage the help of the GC&CS will only be required in the event of the enemy using a cypher which cannot be broken by the cryptographers in the field. Should this occur the GC&CS will be provided with the necessary material and asked to break the cypher. When this has been done, the results will be handed over to the cryptographers in the field who will thenceforth decipher the messages.48
The Navy also intended to run its own ‘show’, and as the DNI wrote in November 1927: ‘On the outbreak of war, the entire naval section of the Government Code and Cypher School will be transferred to the Admiralty, who may require it to go abroad. This transfer may be called for in an emergency other than war, and the Admiralty will always decide when the transfer is necessary. The naval section will then come entirely under the orders of the Admiralty.’49
The RAF was still happy, in the event of offensive warfare abroad, to ‘obtain its intelligence either from the Army or Navy, therefore GC&CS needs take no steps in this matter’. It assumed that it was responsible for erecting W/T and direction finding (D/F) stations to locate enemy aircraft and that: ‘The best solution [to the cryptanalytic problem] would be for the Air Force to possess a small nucleus of officers who had received training in simple cryptography. These trained Air Force officers would be attached on the outbreak of a European war to Air Defence headquarters, and they could be reinforced, if necessary, either by one or two members of the GC&CS or by the recruits called up by the GC&CS.’50 The higher authorities in the RAF had no WW1 Sigint experience and as AGD wrote in 1932 ‘the higher authorities were frankly very sceptical about the value of wireless interception and intelligence obtained therefrom’.51 In 1927, following recommendations by the Romer Committee in 1924, an Air Ministry Y station was erected near Waddington. Initially dealing with diplomatic traffic, by 1932 it was processing enough Russian air material to appoint a cryptanalyst and when he died in 1934, two others were appointed and stationed with GC&CS. In 1936, with the threat of war looming, the Air Ministry intelligence authorities felt that they needed their own experts, so a cryptanalytic Air Section was attached to GC&CS under J.E.S. (Josh) Cooper. He had been a member of GC&CS since 1925 and was transferred from the Foreign Office to the Air Ministry Civil Establishment. By 1938, the section had been expanded and Waddington transferred to Chatham to provided foreign air traffic to GC&CS. The RAF also recruited a former Royal Signals NCO who was promised a commission. A station was set up at Mere Branston and then moved to Cheadle in 1937.
The Army opened a Y station at Fort Bridgewoods, Chatham in 1926 and made progress thanks to Lieutenant-Commander (later Lieutenant-Colonel) M.J.W. Ellingworth. The War Office had been sending serving Army officers to GC&CS for training ever since GC&CS was formed, but this resulted in the Military Wing in GC&CS not being as cohesive as its Naval and RAF counterparts.
A separate GC&CS Military Section was not set up until after 1930. It was established under Tiltman, although he remained on the payroll of the War Office. F.A. Jacobs, a recently retired Army captain, joined him as deputy along with eight staff, including at least two civilian cryptanalysts. Experience had modified War Office policy and all Sigint stations overseas were not regarded as being under the control of AGD and the Committee. Furthermore, ‘the GC&CS should be responsible for the control of the interception of traffic by permanent stations, and the War Office for expeditionary force traffic’. Also ‘the War Office would certainly require a section of GC&CS to continue foreign military intelligence as well as a cryptographic staff with the expeditionary force’.52
Military Intelligence had maintained an interest in interception and cryptography, which is why they sent officers to GC&CS for training. The Admiralty would lend officers to GC&CS to assist in producing results. The War Office had maintained posts abroad and set up a permanent intercept station in the Middle East in the early 1920s and in 1923 at Sarafand, with three officers attached to No. 2 W/T Company there. GC&CS’s Military Section worked closely with the intercept station at Chatham, which produced the first army and air force material and German police transmissions in 1937. Knox led the attack on the German military Enigma, having failed to make progress on their naval Enigma traffic.
Commercial work was not in the original GC&CS mandate but in 1938, Sinclair and DNI agreed that in the event of political turmoil in the Far East, the Japanese might enhance the security of their diplomatic and service traffic. Therefore, the communications of large Japanese companies might provide useful intelligence. A small section was set up under Hope in 1938 to investigate commercial traffic, telegrams of large Japanese firms, less likely to be enciphered than Japanese diplomatic and service material.
While initially set up as a ‘School’ to study world-wide cryptographic methods and practices, events led GC&CS to develop a dual role as a Sigint ‘Centre’, tasked with extracting as much intelligence as possible, as quickly as possible from communications which were of interest to the British Government and relevant departments. The original sanctioned staff consisted of twenty-five cryptanalysts and around thirty support staff. As many as ten Service officers were loaned to GC&CS until the Military and Air Sections were established. Clearly, the size of GC&CS meant that it could not cover all of the diplomatic and commercial traffic circulating around the world. AGD, as Head was responsible for allocating tasks, directing all branches of work and line managing all subordinate Section Heads. This remained the model for diplomatic work both between the wars and during WW2. He had a registry for correspondence and a small number of administration staff. Travis, as his deputy, was responsible for advising on the security of British codes and ciphers and assisting on their provision. He was eventually assigned a junior assistant and clerical staff which formed the nucleus of the Cipher Security Department. On 2 January 1933, AGD received another honour in his capacity as the head of a department of the Foreign Office. He was appointed a Companion of the Order of the British Empire (CBE).
Diplomatic ‘product’ did not require much comment or interpretation. It was processed by a Distribution and Reference Section which was led by a senior officer with cryptanalytic expertise. The section became a reference library which was indexed for research and editorial ease. It never exceeded five staff, including support staff. Unlike the rest of GC&CS, which allocated tasks to a team of cryptanalysts, the Diplomatic Section assigned tasks to one senior cryptanalyst who possessed wide linguistic expertise, thus bypassing the ‘country section’ organisation which was in place. This suited the ‘country section’ model of flexibility and variability of size. For example, during the Italian-Ethiopian War in 1935–6, Italian diplomatic traffic could be exploited so the Italian Section was increased from five to twenty. This was done by passing the work to experienced cryptanalysts and their support staff in other sections such as the French, Romanian and Scandinavian.
In the early days of GC&CS, several experienced cryptanalysts with mathematical expertise were used on more difficult problems such as the German OTP or when a country section was unable to make progress. A senior linguist was also employed in this way and these men reported directly to AGD. At that time there were a number of small country sections, including French, Belgian, Japanese, Italian, Spanish, Near Eastern, Balkan States, Scandinavia and Greek. Not all countries could be covered due to a shortage of staff. The Heads of all of these country sections reported to AGD. The Head of each section selected which deciphered material should be translated and submitted for circulation. The Distribution and Reference Section, by its contacts with the user departments, became the conveyer to the country sections of current priorities; and its Head, by continuity of experience and knowledge of current needs, had become a guide to the exploiting sections in this matter. No more formal ‘priorities’ machinery existed. During the Ethiopian and Spanish Civil wars, the Diplomatic Sections gained valuable experience about the value of its product in supporting military, naval and air operations. It also realised that standard commercial coded traffic might be of considerable value to ‘Economic Warfare’. In 1938, a special Commercial Section’ was set up. Therefore, the Diplomatic and Commercial Sections were already operational and experienced in their work at the beginning of WW2. While the organisation would have to expand to provide full cover as required by their ‘customers’, they would never face the expansion, evolution and exploitation problems faced by the Service Sections. It was able to retain its pre-war structure throughout the war.
GC&CS continued to deal with Japan’s diplomatic and naval attaché communications which could be intercepted at home. Consular, military and naval traffic was handled by outstations in the Far East. Copies of translations were sent to DNI and intercepts to GC&CS, which used them as training material or for cryptanalysis on unbroken systems. There was an agreement between DNI and AGD that in the event of war, GC&CS’s Naval Section would transfer to the Far East theatre.53 A Far Eastern Bureau was considered necessary and it would be interservice in nature. In the Spring of 1934, the bureau was set up on Stonecutters Island in Hong Kong. By June 1935, it was completely staffed and while intelligence was handled on an inter-service basis, apart from five RAF staff the cryptanalysts and interception staff were from the Navy.54 From 1935 to 1939 there was a shift in Sigint concern from the Far East to the Middle East and from the Mediterranean to Western Europe. There was still no Army or RAF presence at the Hong Kong Bureau (Hong Kong Combined Bureau or Far East Combined Bureau [FECB]). The Army was waiting for preliminary cryptanalytical research in the Military Section of GC&CS to facilitate local exploitation. By 1939, progress at GC&CS was sufficient for two Army officers and one RAF officer to be sent to FECB. Up until 1938, GC&CS’s Naval Section and the FECB were successful in reading Japan’s diplomatic and naval attaché machine cipher. FECB reported being able to be ‘in the happy position of being able to read all Japanese Naval cypher messages’.55At the end of 1938, Japan rapidly changed its cryptographic systems. Its fleet introduced a five-figure subtractor cipher56, while diplomatic and consular traffic used a new electrical cipher machine, called the ‘Purple Machine’ by the Americans. Little progress was made against the Japanese fleet system, called JN25, by FECB staff, but it was broken by Tiltman (he had made the earlier break into the main military cipher) and the first decrypted messages were dispatched to FECB in September 1939. By then plans were in place to move FECB to Singapore. GC&CS had provided FECB with JN25 codebreaking material in September 1939, which was used on 75 per cent of Japanese naval traffic. However, GC&CS made virtually no contribution to anti-Japanese Service Sigint from 3 September 1939 until the outbreak of war with Japan in December 1941. The Japanese Diplomatic Subsection translated and did minor research on Japanese diplomatic ciphers.
In early 1934, Italian naval Sigint became of interest as Italy seemed to be on a war footing and communications contained operational intelligence. By the end of 1937, GC&CS’s Naval Section (eighteen strong excluding Japanese staff)) was engaged almost entirely on Italian Sigint. Encrypted traffic was provided by intercept stations at Flowerdown, Malta and Gibraltar. The Military Section was also concentrating on Italian work by 1936, with seven officers involved, supported by six female clerks. By 1938, with six current Italian code books and much of the traffic being read, Tiltman complained about the lack of resources. Encrypted traffic was provided by the intercept station at Chatham and abroad from Sarafand in Palestine and detachments from there. Tiltman’s section also tackled Italian air ciphers.
The Air Ministry was finally stirred into action with the resurgence of the German Air Force (GAF) in 1935, and a section, A.I.1e, was formed to scan, translate and summarise plain text intercepts. In July 1938, it was suggested to AGD that coordination could best be achieved in Malta by a small inter-service Y Intelligence Centre in which intelligence officers from the three Services would produce intelligence from intercepts, direction finding bearings and ciphers broken in GC&CS. AGD proposed the creation of a Combined Bureau in the Middle East, similar to FECB, to perform all Sigint functions, including cryptanalysis. Each Service favoured its own main station: Malta for the Navy, Sarafand for the Army and Cairo for the Air Force. In the end a Middle East Intelligence Centre was set up in Cairo in the summer of 1939, receiving decrypts direct from Sarafand and via the War Office, from GC&CS. NID got Board approval for an Operations Intelligence Centre in Malta and AGD reluctantly agreed in July 1939 to supply, for local exploitation, ‘any broken cyphers used unrecyphered by the Italians since such books can be used without the assistance of a skilled cryptographic staff’.57
In May 1938, a German subsection of the Naval Section at GC&CS was set up to analyse traffic, consisting of one officer and one lady clerk. However, German military W/T activity was suspended at Chatham in 1935. The study of German Army systems did not even rate a mention in the Military Section report of July 1938.58 Even though the Air Ministry was aware of the growth of GAF activity in 1935, it was a low priority of the Air Section in 1936 59 and Italian traffic remained its main concern.
The general view within the British Service and intelligence communities in the mid- to late 1930s was that the likelihood of solving Enigma was ‘not yet sufficient to justify real confidence in eventual success’.60 German diplomatic traffic was unreadable, and even lowlevel German Army transposition and stencil systems had not been broken and no effort was being made on naval Enigma until military Enigma was solved. The Air Ministry believed that Enigma was not being used by the GAF. AGD himself seemed to be pessimistic ‘as to the possible value of cryptography in another war’61 and Sinclair was concerned after the Munich crisis ‘that as soon as matters become serious, wireless silence is enforced, and that therefore this organisation of ours is useless for the purpose for which it is intended’.62 The Naval Section of GC&CS wrote to AGD in February 1938 saying that ‘D/F will certainly be our only real source of information as to enemy movements by sea, land and air in the early and therefore probably the most important days of a modern war’.63 However, Service Ministry thinking was starting to change and, increasingly, their view was that in the event of war, the Sigint effort shouldn’t be split between them. The Main Committee recommended the interconnection of all intercept and D/F stations in the UK to GC&CS by telephone and teleprinter and the formation of a ‘joint Inter-Service Operational Intelligence Section at the GC&CS’.64 But the Ministries preferred to keep ‘operational intelligence’ under their control and the section was never formed. However, it did speed up the expansion and improvement of D/F facilities and the communications needed to combine the stations into a network. All Y stations were linked to GC&CS and a Defence Services Lines Telecommunications Board (DSLTB) created by the Director-General of the Post Office, with representatives of all Services and AGD. The teleprinter and telephone lines converged on a special room at the Central Telegraph Office (CTO) in London, which was connected by a private cable to GC&CS in the Broadway Buildings, to operate the system from another centre such as GC&CS’s war station at BP. It was only necessary to switch the system over from the CTO room to the new site.
The RAF needed Sigint to be delivered swiftly and a test was carried out in March 1938 to demonstrate the advantages of decentralised cryptanalysis. Italian and Spanish Nationalist Air Force traffic was sent to Waddington where Cooper and two other cryptanalysts, supported by a colleague at GC&CS, spent three days decrypting and reporting on the traffic. This convinced the Air Ministry that a cryptanalysis section should be located at the central Y station in peace or war. The role of the GC&CS Air Section would be restricted to cryptanalysis and training new personnel. Meanwhile, the Admiralty set up an Operational Intelligence Centre in 1937, but Operational Intelligence was another name for traffic analysis which was really carried out in the Naval Section of GC&CS. The new centre collated intelligence from non-Sigint sources or decrypts from GC&CS.
By 1938, it was increasingly clear to AGD that he was going to have to expand his organisation. With Sinclair’s backing, he asked two former Cambridge dons, Frank Birch and Frank Adcock, both veterans of WW1 codebreaking, to trawl through the staff and student lists at both Oxford and Cambridge and other universities. They were looking for men who were deemed suitable for secret intelligence work within the Foreign Office. While the WW1 cryptanalysts did not have much time for mathematicians, GC&CS had recruited one before 1935 and was already putting a Cambridge mathematician through preliminary training in London. A second was recruited from Oxford in February 1939. Through the Chief Clerk’s Department, AGD got the Treasury to sanction fifty-six senior assistants at £600 per year and thirty women with a graduate’s knowledge of at least two of the languages required at £3 per week.
According to AGD: ‘It was naturally at that time impossible to give details of the work, nor was it always advisable to insist too much in these circles on the imminence of war. At certain universities, however, there were men now in senior positions who had worked in our ranks during 1914–18. These men knew the type required.’65 One such recruit was E.R.P. Vincent, a Fellow of Corpus Christi and a professor of Italian. He had learned German during his internment in Germany during WW1. He later recalled dinner with Adcock in the spring of 1937:
We dined very well, for he was something of an epicure, and the meal was very suitably concluded with a bottle of 1920 port. It was then that he did something which seemed to me most extraordinary; he went quickly to the door, looked outside and then came back to his seat. As a reader of spy fiction I recognised the procedure, but I never expected to witness it. He then told me that he was authorised to offer me a post in an organisation working under the Foreign Office, but which was so secret he couldn’t tell me anything about it. I thought that if that was the case he need not have been so cautious about eavesdropping, but I didn’t say so. He told me that war with Germany was inevitable and that it would be an advantage for one of my qualifications to prepare to have something useful to do.66
Vincent was summoned by telephone to Broadway Buildings in London shortly afterwards and returned periodically until war was declared. He was able to learn something about cryptographic problems and he ‘picked up the jargon and got to know some of the people’.
Another recruit in 1938 would certainly make his mark during the impending war. Alan Turing was identified by either Adcock or Birch as having the right skills for GC&CS and he was invited to a training course at GC&CS’s offices in London in the summer of 1938. He attended another course at Christmas and then visited every two or three weeks to help with the work. He was attached to a team led by Dilly Knox and worked alongside Peter Twinn,67 the young mathematician who had been recruited from Oxford in February 1939.
Throughout the 1930s, Sinclair had become concerned that his intelligence organisations were based in Central London, which would make them vulnerable in the event of war. He started to look for somewhere outside of London to establish a war station for intelligence activities. In early 1937, it was brought to his attention that the remains of a large Victorian estate, located around 50 miles from London, was on the market.68 Its 55 acres would provide ample space for a growing intelligence organisation. Furthermore, it was only five minutes’ walk to a railway station, which sat on the main line from London to the North. Also close by was the main north/south arterial road. Sinclair initially leased the property and, on 9 June 1938, he purchased the property, called Bletchley Park, for £6,000. This was done using his own initiative and many authors believe that he was following a Service tradition by paying for it out of his own pocket. The Official History of MI6 is less certain:
The relevant property transaction documents show him personally as the sole owner, and after he died in November 1939, apart from legacies of £3,500 to each of his two sons, his sister Evelyn inherited the remainder of his property, with a total value of £21,391. In April 1940 Evelyn (as personal representative of ‘Sir Hugh Sinclair deceased’) transferred BP to William Ridley and Percy Stanley Sykes [the Service’s Finance Officer] for ten shillings [40p]. In their turn, on 3 March 1947, Ridley and Sykes transferred the property to the Ministry of Works, again for ten shillings, all of which strongly suggests that the original purchase money had come from public, if not also SIS, funds.69
Engineers from the GPO began installing telecommunications equipment and by the autumn, enough was in place for Sinclair to order a dress rehearsal at his new intelligence war station. As the Head of GC&CS’s Air Section, Josh Cooper, remembered:
In Autumn 1938 GC&CS had no administrative staff and the Admiral (who appears to have taken the decision to move unilaterally) put Captain Ridley RN of SIS in charge. All personnel of every grade were accommodated in hotels in Bletchley and surrounding towns and villages. The Admiral sent out an excellent chef from London and we all sat down to lunch together at one long table in the House. All this was simply paid for out of SIS funds; Captain Ridley was not concerned with Civil Service regulations. A large room on the ground floor of the House had been set aside for Air Section. Tables and chairs had been provided but there were no cupboards and I remember coming into a scene of chaos with a great mound of books and papers piled on the floor. After Munich we all trooped back to London.70
The rehearsal at BP revealed a number of shortcomings in accommodation (work and billeting), in staff and in efficiency of communication. So while GC&CS staff returned to London after the signing of the Munich Agreement71 on 30 September 1938, work continued on preparing BP for war and improving arrangements for accommodation and catering.
By the end of 1938, the Air Ministry not only recognised its responsibility for the air defence of Britain, but also that it couldn’t live off the work of other Services. A Section at GC&CS would have to be engaged in cryptanalytic research and training and would work with a W/T and direction finding complex centred at Cheadle. For the War Office, MI8 would take over Sigint responsibility in September 1939 and ‘the collection, correlation and dissemination of all Military Intelligence obtained from the study of foreign military intercepted communications, for the plotting and identification of foreign enemy wireless stations, the breaking of foreign code call systems and the measure of the enemy’s wireless activity’.72 Chatham was concerned with purely strategic interception work and the Military section of GC&CS, like its Air Force counterpart, with higher-grade cryptanalysis and training. At this stage, the Air Ministry was concerned about the defence of Britain and had made no provision for strategic interception, on the basis that there was no such traffic. The War Office was concerned with Sigint in the field and their Chatham Y station alone intercepted GAF Enigma traffic. The Army and Air Ministries controlled their respective sections at GC&CS, while the Admiralty did not, following the removal of its GC &CS’s sections operational intelligence work.
GC&CS was changing from two groups working on diplomatic cryptographic systems as a sideline to their Service work, to one which developed Service sections as an adjunct to its diplomatic work. Traffic Section was responsible for the intake and sorting of intercepts. Distribution and Reference Section edited and distributed decrypts translated by cryptanalysts and kept in contact with users to assess their needs. The Civil side in 1939 outnumbered all of the Service sections combined and consisted of a large number of cryptanalytic units, ranging from one to fifteen staff. The Head of each continued to report directly to AGD. Some of the units carried out cryptanalytic research and others operated as a centre for the production of decrypt intelligence and unit boundaries were flexible. A Commercial Section with six staff was set up in 1938 to scan and select from a mass of intercepted correspondence, mainly in plain language or public commercial codes. This was passed to a centre in London which eventually became the Ministry of Economic Welfare. However, the civil side of GC&CS was functionally far more limited than the military side. Its sole functions were cryptanalysis and translation. It had little involvement with traffic analysis or intelligence and even interception presented few problems, as the most reliable source of supply were the copies of cable traffic that had been handed in for transmission and held by the Post Offices and Cable Companies after they had been sent.
***
On 7 December 1931, a representative of the Polish cipher department met a French intelligence officer at the Central Station in Warsaw. The officer, Captain Gustave Bertrand, carried with him operational instructions for the Enigma machine being used by the German military. The Gebrauchanleitung für die Chiffriermaschine Enigma also contained four drawings of the machine. A further document in his possession, the Schlüsselanleitung, provided vital information about the various settings of the machine. His source for these documents was a German by the name of Hans-Thilo Schmidt, who had first contacted the intelligence representative at the French Embassy in Berlin on 8 June 1931. Through his brother Rudolf, Schmidt had been given a job in the German Defence Ministry’s Cipher Centre, the Chiffrierstelle, known as the ChiStelle. Rudolf Schmidt had been the previous head of the ChiStelle and, ironically, had approved the Enigma machine for use by the German Army. This was the very machine which his brother had now betrayed for no other reason than money to fund his lavish lifestyle. The Deuxième Bureau’s Service de Renseignements had assigned one of its operatives, Rudolf Lemoine, to make contact with Hans-Thilo and they duly met on 1 November 1931 at the Grand Hotel in Verviers, a town in eastern Belgium around 15 miles from the German border. Bertrand attended a subsequent meeting on 8 November, along with a photographer. As well as the two Enigma documents, Schmidt produced an organisation chart of the ChiStelle, an army hand cipher and a memorandum on poison gas. Schmidt was duly paid 10,000 marks (about £41,000 today) and Bertrand returned to Paris. He handed them to Colonel Bassières, a top French cryptanalyst who, after analysing them for two weeks, told Bertrand that they were of little use to French intelligence. Bertrand now turned to British intelligence and on 23 November, handed copies to their Paris representative, Wilfred (Bill) Dunderdale. Three days later, Dunderdale also rejected the documents, as they would be of little use in decrypting Enigma messages. Thus it was that Bertrand got permission to offer the material to Polish intelligence and was subsequently greeted by them with open arms.73
Polish intelligence’s interest in the material from Hans-Thilo Schmidt was understandable, given the difficult historical relationship between Poland and Germany. At the end of WW1, the newly-reborn Polish state had taken part of Silesia, Pomerania and territory around Poznan, as it had been part of Poland before the partition by Frederick the Great. This was a cause of considerable anger in Germany, and an atmosphere of enmity and continual tension persisted, fuelled by the long history of numerous and bloody wars between Poland and Germany. The re-born Polish state felt seriously threatened by its neighbour to the west. On 15 July 1928, Polish radio-monitoring stations in Starogard, Gdanski, Poznan and Krakow-Krzeslawice had intercepted the first German messages which were in a machine-generated cipher. The following year a course in cryptology was started at the University of Poznan for twenty of its most advanced mathematics students who could also speak German. The course was set up at the initiative of the radio intelligence department and some of their specialist officers would give lectures to the students, the most gifted of whom would be asked to volunteer to continue their studies within Military Intelligence. Throughout 1931, a cipher bureau was created which was an amalgamation of the radio intelligence and cryptography sections. The new bureau was headed by Major (later Lieutenant-Colonel) Gwido Langer. As the new bureau was being organised in 1931, the course at Poznan was coming to an end, and three students had frequently managed to solve the German ciphers that had been set for them on the course. It was decided to set up a small section of the department for them in Poznan. Their names were Marian Rejewski, Hendryk Zygalski and Jerzy Różycki.
The German military had introduced a new version of an Enigma machine with a plugboard at the beginning of June 1930 and, the following year, the Poles purchased a commercial model on the open market in Germany. At the beginning of September 1931, Rejewski, Zygalski and Różycki were employed by Section BS 4 (German intelligence) of the Cipher Office of Department II of the General Staff in Warsaw. This section was under the command of Captain (later Major) Maksymilian Ciez’ki. Rejewski was the most advanced academically of the three young students, having just been awarded his degree in mathematics. He had also spent a further year studying the subject in depth at the University of Gottingen in Germany. He began work on his own in October 1932 with the commercial Enigma machine which had been purchased earlier. He was also receiving several dozen messages daily which had been encrypted on the military version of the Enigma machine. By December, he had a photograph of the military machine, user instructions for it and a schedule of daily keys for September and October 1932. Remarkably, by the end of December 1932, Rejewski had reconstructed the internal connections within the Enigma machine and identified the indicator system74 currently being used by the German Army. He had done it entirely using a mathematical technique called permutation theory. Along with significant contributions from Zygalski and Różycki, the first breakthrough in solving the German military Enigma machine had been achieved. The three young Polish cryptanalysts handed to their superior the first completely decrypted Reichswehr75 signal which had been encrypted using the military Enigma.
In early 1933, Rejewski, Zygalski and Różycki began working together and were successful for a number of years in reading encrypted German Army messages. After the GAF introduced Enigma at the end of 1934, they grouped radio operators who worked alongside Enigma operators into different radio networks. The number of these grew rapidly and the Poles had to monitor an increasing amount of traffic. By 1 February 1936, the Germans had introduced significant changes which improved the security of the Enigma machine. Each radio network had its own setting of the machine which changed every twenty-four hours. The following year, the Polish General Staff transferred their German cryptographic section BS4 to a camouflaged and high-security new headquarters in the Kabacki Woods near Pyry outside Warsaw. They developed technologies such as the Zygalski sheets, the cyclometer and the bomba to help work out the Enigma settings being used by operators on different networks. Up until December 1938, GAF and Army Enigma operators had three wheels for their Enigma machines. They would change the order of the wheels in the machine each day, and as each wheel was wired uniquely, this would in effect change the wiring in the machine. This meant that there were six possible wheel configurations and the Poles had built six bombas to speed up their work. On 15 December, the Enigma operators were given two additional wheels and could now choose three from a set of five uniquely wired. This increased the number of possible wheel configurations to sixty and the Poles would need a bomba for each additional configuration. They simply did not have the resources to build fifty-four more machines. To make matters worse, the Germans had introduced a plugboard at the front of the Enigma machine with twenty-six sockets and a number of letters were plugged together. When they also increased the number of plugboard connections to ten, working out the daily settings of the Enigma machine became a daunting task indeed!
The deployment of the Enigma machine across all of its services was becoming an increasing concern to AGD. From 1937 onward, it was obviously desirable that British naval, military and air intelligence should get in very close touch with their French colleagues for political and military reasons. Sinclair had always been keen to maintain a close liaison between SIS and GC&CS and had established links with the French intelligence agency the Deuxième Bureau de l’État-major général, directed by Colonel Maurice-Henri Gauché. Sinclair’s deputy, Menzies, had a close relationship with Colonel Louis Rivet, the head of the French Army’s ‘2ième Bureau’ (the Services de Renseignements et de Contre-Espionnage militaire). Bertrand worked under Rivet but AGD believed that it was Dunderdale, although he had little knowledge of cryptography, who urged the British to liaise with the French on a technical level.76
AGD was concerned about how far he could go in collaborating with the French. He wrote to Sinclair on 2 November 1938, asking for guidance.
Liaison with the French
I should like to have your guidance as to the limits to which this liaison should go.
We have received about 100 documents of varying types and varying value:-
Three photographs of codes of which one might well have had great value in the event of war with Germany last month.
Photographs of documents relating to the use of the Enigma machine which did increase our knowledge of the machine and have greatly aided our researches.
Studies of the German Military and Air services.
A full description of the German Y Service.
Reconstructed German and Italian codes (unrecyphered).
Considerable amount of German and Italian intercepted telegrams.
The results of French D/F work.
We have given them purely cryptographic assistance in the shape of :-
Our reconstruction to date of the Italian recyphered and unrecyphered books.
Copies of two Italian recyphered books on which they set great store.
We propose to give them results of our Y work in the shape of:-
German Naval Call Signs.
A complete study of the German Air Force.
A complete study of the German Military.
It appears to us that their cryptographic work is less ambitious than ours. They have worked on the German and Italian unrecyphered codes with success and on the German Military hand cypher (double transposition).
As to their Y Service it appears:-
That the Navy works separately. I was informed by a Naval Officer liaison officer with ? that they relied entirely on their station at Dunkerque for interception and D/F.
We have received no Naval intercepts but the French version of situation report of the German Navy from time to time through? We have reciprocated with our version through Section III.
That the army, Y and D/F service is considerable; M/F interception is more productive that ours.
The Military Section report that the organisation is not so complete as here, no night watch being kept and no work done on Sundays.
That the Y watch on German Air Force is very poor (vide Air Section report).
That they have an organisation for watching N. Italy, Libya and to a certain extent Eritrea (vide reports from Italian Diplomatic Section and Military Section).
It is very remarkable that we have received no example of any interception in Spain, either German, Italian or Spanish, work which occupies a large part of our staff. It occurs to me that this may have been a special study for the Navy who will no doubt watch the Mediterranean and Bay of Biscay.
This may also account for the entire absence of reference to Italian Naval Cyphers. On my own work on French Naval Cyphers during the last two years I have come across references to interception, etc. which lead me to suppose that it is the French Navy and the Bureau de Chiffre of the Marine who carry out this side of the work.
Our main reason for seeking this liaison in the first place was the desire to leave no stone unturned which might lead to a solution of the Enigma Machine as used by the various German services. This is of vital importance for us and the French have furnished us with documents which have assisted us but we are still doubtful if success can be obtained without further documents. During the coming meetings we hope to show Mr. X. the lines on which we are working and make clear to him what other evidence we need in the hope that his agents may produce it.
We also hope to obtain from him more material prior to September 15th, 1938, as there was a slight prospect of a break-in on the military machine before their mobilisation change.
We shall continue to work on the machine and will naturally give Mr. X. any results we obtain.
An interchange of current intercepted material may be necessary, unless the French prefer to put the onus of research on us.
We can also continue to cooperate on the reconstruction of the two Italian codes. As yet there has been no discussion on Italian Naval and Military cyphers and on this question I should like to receive orders in view of the fact (a) that I suspect that the Italian Naval is being worked in another branch in Paris and (b) that the Foreign Office considered that cooperation on the Italian Military telegrams between Spain and Rome was, at that time, undesirable.77
In early January, Bertrand asked AGD to come to Paris to meet some Polish experts in cryptography. In the hope that such a meeting would help GC&CS with work on German and Soviet ciphers, AGD attended meetings on 9 and 10 January along with Tiltman, Knox and Hugh Foss.78 In attendance on the Polish side were Langer and Ciez’ki; on the French side Bertrand and a French cryptanalyst, Henri Braquenie. The meeting was shrouded in secrecy and mystery and began with a presentation by Ciez’ki on the results of the Polish research into the Enigma machine as used by the German services. The British delegation was not impressed, as Ciez’ki gave a long and pedantic account of results which Tiltman had already achieved in an hour’s work. AGD and his colleagues felt that the Poles had little to contribute and Knox went further, claiming that the Poles’ knowledge of the Wehrmacht Enigma was ‘nil’.79 Writing of the meeting in May 1948 (see Appendix 5), AGD soon realised that in fact the Poles had been vetting the British, and had instructions not to disclose any of their real work on Enigma unless the British and French could demonstrate that they had made comparable progress and were prepared to share it with them.80
While nothing more was heard from the Poles for six months, tensions between Poland and Germany were close to breaking point. In the middle of July, Langer sent an invitation, through Bertrand, for AGD and his colleagues to come to Warsaw. The invitation specifically included Knox, as both the Poles and the French knew that he was working on the Enigma machine as used by the German military. AGD was very reluctant to take Knox with him, given his temperamental nature. However, Sinclair instructed him to do so and after discussions with the DNI, AGD was also told to include Commander Humphrey Sandwith, Head of the Admiralty’s interception service, in his group.81 He wanted to talk to the Poles about placing a site in Poland to help with British direction finding efforts. Bertrand proposed that the second tri-lateral meeting be held in Warsaw from 24–27 July. Bertrand and Sandwith travelled by air to Warsaw but AGD and Knox travelled by train as they wished to see Germany, possibly for the last time for quite some while. The latter arrived on the 24th and were met by Bertrand and the Poles. They were lodged at the Hotel Bristol while the French stayed at the Hotel Polonia.
The Poles entertained their visitors to lunch at the Hotel Bristol and ironically the fairly banal conversation was conducted in German as it was the only common language of all in attendance. On Wednesday, the 27th, the Poles called for AGD and Knox at 7.00 am and drove them to a clearing in the Kabacki woods about 20 kilometres from Warsaw. About a kilometre from the small town of Pyry, a clearing had been made and here lay the Polish Biuro Szyfrow’s secret headquarters which was partly underground. It was in the shape of a square with sides of about 200m surrounded by a high wall enclosing two brick buildings. The smaller of the two served as an air-raid shelter and radio station which was serviced by aerials which could be lowered to avoid detection.
The Poles had exploited a weakness in the Germans’ procedures for Enigma in the 1930s. Hendryk Zygalski had devised a method which exploited this weakness to help work out some of the settings used in the Enigma machine for any given day. It was based upon a catalogue of perforated sheets and it had the huge advantage that it was not compromised by the plug connections on the Enigma machine. As there were only six wheel orders in operational use at this time (the operator could mix up his three wheels in six ways), a complete catalogue contained twenty-six perforated sheets, one for each of the possible ring settings82 on the right-hand wheel. Thus, 6 x 26 = 156 sheets had to be manufactured. When the Germans introduced two additional wheels in December 1938, the number of wheel orders had increased from six to sixty and 1,560 sheets were required. The Poles had also developed two machines to help with their work. The cyclometer helped with the task of constructing a card index system containing information about all possible Enigma start positions that could have been used. The bomba had three pairs of Enigma wheel systems driven by an electric motor and exploited the same weakness as the Zygalski sheets. However, the increase in possible wheel orders and lack of resources meant that the bomba could no longer be used.
Ciez’ki proceeded to demonstrate, with the assistance of an Enigma machine which included a plugboard, how the Poles had gone about working out the daily setting of the machine. Once this was known, they could read the messages by simply typing the encrypted text into their Enigma machines. The British and the French attendees were then shown the Zygalski sheets, cyclometer and bombas. Knox, in his position as the British expert, stood close to the machines during the demonstration. During a break for tea after three hours, and then afterwards, AGD’s worst fears about Knox’s attendance at the meeting were realised. He maintained a stony silence throughout the meeting and was clearly upset by what he was hearing. Matters came to a head when they got into a car to leave, as AGD later recalled:
It was only when we got back in the car to drive away that he (Knox) suddenly let himself go and assuming that no one understood any English raged and raged that they were lying to us now as in Paris. The whole thing was a pinch he kept on repeating – they never worked it out – they pinched it years ago and have followed developments as anyone could but they must have bought it or pinched it.83
Knox probably knew more about the Enigma machine than anyone in Britain at the time of the Pyry meetings. While he had some success in breaking versions of the Enigma machine which were similar to the early commercial version, he had been unable to break the new military version with the plugboard. His problem had been the connections between the keyboard and the entry drum inside the machine. On the models of Enigma machines that he had successfully broken, the connection pattern followed the order of the keys left to right, row by row and alphabetically around the entry drum. So the Q key was connected to A, W key to B, E key to C, R key to D, etc. On the model of Enigma in mass use by the German Army and Air Force, the connection pattern had been changed and Knox’s team (which included Turing) could not work out the new pattern. This problem had also stumped Rejewski initially, and he described his solution in a paper written in 1980:
What, then, were the connections in the entry drum? It turned out later that they can be found by deduction, but in December 1932, or perhaps in the first days of 1933, I obtained those connections by guessing. I assumed that since the keyboard keys were not connected with the successive contacts in the entry drum in the order of the letters on the keyboard, then maybe they were connected up in alphabetical order; that is, that the permutation caused by the entry drum was an identity and need not be taken into account at all. This time luck smiled upon me. The hypothesis proved correct, and the very first trial yielded a positive result.84
When Knox met Rejewski he had quickly asked him: ‘What are the connections to the entry drum?’ Knox was furious when he heard the answer; ‘A, B, C, —-’. In other words, the Germans had wired it up in the simplest possible way, the Q key to Q, the W key to W, the E key to E, the R key to R, etc. He remained aloof and alone over dinner and gave the impression that he had a grudge against the Poles. AGD spoke with Bertrand and Sandwith and it was decided to return to Warsaw as soon as possible. The next day, Knox met with Langer, Ciez’ki, Rejewski, Zygalski and Różycki. According to AGD, Knox was ‘his own bright self & won the hearts & admiration of the young men with whom he was so much in touch’. Ever the loyal colleague, AGD explained Knox’s behaviour as follows: ‘If only that first day of formal disclosure could have been avoided & pompous declarations by senior officers had been omitted. Knox’s mind & personality in touch with men who really knew their job would have made that visit a very real success. They were all simple & straightforward.’
AGD and Sandwith left on the 28th to return to home and he was back in London by Sunday the 30th. The following day he wrote to Bertrand:
My dear Colonel,
Our party returned safely if with difficulty. Mr. Knox’s papers were not quite in order and he had to spend the night in Posen, but arrived safely 24 hours later.
I wish again to assure you our extreme gratitude for all that you did for us and our appreciation of your very fine achievements. I have reported the matter in brief to my Chief who also expressed freely his appreciation of your work as I described it. He empowers me to repeat our invitation for you and the Major to visit us.
I trust such a visit may be in the near future, but the next few weeks are unusual with us owing to manoeuvres, and I also think it might be a good thing to get our cooperation started on the lines we outlined before you came, in order that any weak points might be noted and eliminated.
In accordance with the desire of your staff I hope to send by an early courier our intercepts of the T.G.D. group and these I will continue to send by each courier via our mutual friend Captain B. I am also sending our interception of the Naval material during the month of May 1937, as our experts felt that the additional material at that period might prove of great value to them.
Mr. Knox is now organising his staff to continue investigations on the Military material on the lines suggested by your experts.
I shall be grateful, my dear Colonel, if you will convey to Colonel S. and Colonel M. our gratitude and in the hope that our close collaboration may achieve further useful results.85
AGD had agreed with the Poles that the British, with greater resources at their disposal, would produce the Zygalski sheets.86
Knox’s passport had been wrongly stamped for a return journey through Germany. Instead, he had to travel to Poznan to get a visa from the British Consul there. AGD’s patience with Knox must have been sorely tested when he received a letter from him, written on Hotel Bristol notepaper and dated 30 July:
My Dear Alastair
Let’s get this straight.
The Poles have got the machine to Sept 15th 38 out by luck. As I have said only Mrs B.B. had seriously contemplated the equation A = 1 B = 2. Had she worked on the crib we should be teaching them.
They must have done very well to determine the two new wheels. How they did it might be important. I have not discussed this.
Their machine for determining all ciphers (O.S.) and SSD (n.s.) may be good. If we are going to read them we should be given a detailed study: if not, we only want the broadest outline of its electrical principles.
Military (?) Sept 15 – ?? April 1 ??. Here they seem to have failed badly Mathematically the data are more not less than with the other cipher (if there be over 300 messages). If we are to attempt this we should examine their system and statistics (if any) with considerable scepticism*. Why have they failed on an astronomical but simple calculation.
April 1st (??) to now. The whole basis of diagnosis may have changed though doubtless the machine remains.
Before deciding anything we must settle
Whether we are going to give a miss to the S.S.D. stuff
Whether they have got the right system for the other.
I am fairly clear that Schessky [Knox is referring to Ciez’ki] knows very little about the machine & may try to conceal facts from us.
The young men seem very capable and honest.
A.D.K.
Finally (f) It cannot too strongly be emphasised that all successes have depended on a factor (the machine coding of indicators) which may at any moment be cancelled.
Even the principle of electrical selection must be viewed with distrust.87
It is typical of Knox that he ends his rant against the Poles with kind words for Rejewski, Zygalski and Różycki. He was also correct in pointing out the weakness in the Polish methods, i.e. their dependence on the machine coding of indicators. This was indeed cancelled in early May 1940, rendering the Polish methods useless.88
Despite being friends with Knox since their Room 40 days, AGD revealed his exasperation in a remarkably candid letter to Bertrand, dated 3 August 1939. It was written on the same Hotel Bristol note paper that Knox had used:
My dear Bertrand,
I have finally had a day off and I take this opportunity to write to you a very personal letter, ‘from the heart’, which seems necessary to me.
I have seen D;89 in his opinion I may have said something bad about you and that is why I wish to emphasise that we owe everything solely to you and I look forward to the cooperation of our trio and, that to reach our goal you must remain in the leading position. In Warsaw it was you who advised me to return and think about it – and you were right.
Maybe you understand my problem in the shape of Knox. He is a man of exceptional intelligence, but he does not know the word cooperation. You surely must have noticed that off duty, he is a pleasant chap loved by all. But in the office his behaviour is different.
In Warsaw I had some deplorable experiences with him. He wants to do everything himself. He does not know how to explain anything. He can’t stand it when someone knows more than him. Unfortunately, I cannot do without him, he knows more about the machine than anyone else in the country. He built a machine of the type used by the Spanish, and frequently by the Italians in Spain, which is not to be sneered at, even if not so much has been done as has been done by our friends in Z.
You must forgive me for being so keen to keep him, but I will tell you in all sincerity, that I will never take him to a conference again if I can only avoid it. From now on, we must establish the rules of our cooperation in order to avoid unnecessary effort.
The Pyry meeting came just in time because on 1 September, Germany invaded Poland and two days later, Britain declared war on Germany. By 10 September 1939, German forces were rolling relentlessly across Poland and their armoured columns broke through near Warsaw. The Polish Cipher Department was ordered to leave the capital and the Pyry Centre as soon as possible and move its essential equipment. However, the advance of the Red Army into Poland on 17 September made the Poles change their plans. The cryptanalysts were forced to destroy all traces of their equipment and documents, retaining several Enigmas. The military and civilian sections of the Cipher Department split after they crossed the border into Romania. On 20 September Bertrand delivered a replica Enigma to Menzies in London as a gift from the Poles. The Polish contribution to the early success of BP would prove to be significant. They had recognised in the late 1920s that the age of machine cryptography had begun and that mathematicians would be effective as cryptanalysts.
While it is likely that Knox came away from the Pyry conference with the missing link in his attempt to construct the Enigma machine itself, AGD clearly saw the bigger picture, and on returning to Britain began to recruit more mathematicians for GC&CS. The Poles had demonstrated to the British that encryption machines like Enigma could be broken, if the right mathematical minds were allowed to concentrate on the problem. With the prospect of many years of war ahead, AGD’s knowledge and experience of Sigint and its effectiveness in military conflict, would certainly be put to the test.