GCHQ

Relations with the Australians were somewhat awkward. Typically, London had reluctantly agreed that Sir Frederick Shedden, the new Australian Defence Minister, could be indoctrinated into the secrets of sigint, but only so he could use his power to prevent a reduction of Australian spending on intelligence. There had also been alleged leaks about intelligence in Canberra, and in September 1945 there were momentary doubts as to whether any cooperation with Australia on sigint would be authorised.11 Indeed, by December 1945 a ‘dangerous position’ had developed, with the Australians seeming to want to go it alone with their own system; what was worse, there were rival elements within the Australian armed services. British liaison officers warned, ‘If we are not prompt to give a lead there may even be 3 or 4 rival shows in Australia with no hope of proper security.’ During the war, material collected in Australia and the Far East had often been sent back to Britain for analysis. However, there was now a possibility that the Australians might end up ‘insisting on full exploitation in Australia’. This was a situation that British code-breakers wanted to avoid at all costs, since final exploitation was power, and they wished to keep their Commonwealth associates in a subordinate position.12

The crucial moment in the creation of the global sigint alliance occurred on 22 February 1946, when Britain opened a two-week Commonwealth conference for ‘Signals Intelligence Authorities’. This gave them critical mass prior to concluding a deal with the Americans the following month. The attendance of Australia and Canada was a foregone conclusion, and given the significant contribution that New Zealand had made during the war to naval sigint, there were hopes that she would also join in.13 The conference was also attended by senior officers from GCHQ’s regional centres. Bruce Keith, the commander of HMS Anderson, the massive sigint collection station in Ceylon, was there, accompanied by his deputy, Teddy Poulden.14 At this conference Australia offered sixty-five operating teams, amounting to 417 personnel, from the three armed services as its contribution to a new global sigint network.15 Australia was persuaded to set up a British-style Joint Intelligence Committee and, most importantly, a unitary Signals Intelligence Centre along the lines of GCHQ, which was given the cover name Defence Signals Branch.16

The big issue was the choice of the director of Australian sigint. The Australians fielded four candidates, all experienced wartime intelligence officers. However, Travis told them bluntly that it would be a British officer. Some Australians were affronted, but on balance Travis’s decision was probably correct, since it ensured that Australia would have good access to British sigint. Travis’s choice was Teddy Poulden, who had spent the last two years of the war as deputy to Bruce Keith, the commander of HMS Anderson, on Ceylon.17 Poulden took over in April 1947, commanding a staff of around two hundred, about twenty of whom were GCHQ personnel on secondment. Although senior Australian sigint officers resented the fact that Poulden had his own private cypher for communicating with Travis, he was broadly considered to have done a good job. In the early 1950s he was succeeded by an Australian, Ralph Thompson, who remained in the position until 1978, making him easily the longest-serving Western sigint chief.18 In January 1947 a further Commonwealth sigint conference was held in London, and the Chifley government gave final approval for the integration of Australian sigint into UKUSA at the end of the year.19 However, this was delayed by the security concerns raised by Venona, so there was little sigint contact between Australia and the United States until 1949.20

Canada’s sigint organisation under the long-serving Lieutenant Colonel Edward Drake suffered similar ‘colonial’ treatment. Although Drake was a Canadian, his deputy was the stalwart British code-breaker and expert on Russian systems Geoffrey Stevens, who arrived to take up his post in Ottawa in March 1946. A few weeks later, on 13 April, the Canadian Prime Minister Mackenzie King authorised the consolidation of a number of wartime organisations into a small post-war unit of about a hundred staff known as the Communications Branch of the National Research Council (CBNRC). A number of senior posts were filled by staff seconded by GCHQ, prompting locals to observe that CBNRC stood for ‘Communications Branch – No Room for Canadians’. By the late 1940s Drake had resolved to offset this British oligopoly by developing better relations with the US Army code-breakers.21

The two-week sigint conference that GCHQ had convened with the Commonwealth partners in February 1946 was a vital prelude to business with the Americans the following month.22 Indeed, the Australians and Canadians had given GCHQ permission to negotiate on their behalf.23 On 6 March, William Friedman, one of the US Army’s most senior code-breakers, arrived in London to complete a revised version of the previous wartime agreements between Britain and the United States. The main section of the agreement which followed this policy conference between the principals was only four pages long; however, a UK-USA Technical Conference followed in June 1946 which added many annexes and appendices. Much of this new material was about attempting to agree on security procedures for handling sigint.24 The terms of the 1946 agreement are still highly secret. Both parties agreed to ‘pool their knowledge of foreign comint organizations’, and that in any future negotiations with other parties ‘every effort should be made to avoid disclosure of US/UK collaboration in the COMINT field’.25 Joseph Wenger, the head of America’s naval code-breakers, accepted that the 1946 conference had only dealt with generalities, and this had generated ‘some criticism’. Nevertheless, his priority was to ‘set up the framework and establish the will to make it work’, so in his view it was a great success, and ‘laid the foundation of a very fruitful and important partnership’.26

For GCHQ, liaison with allies was all-important to its new status as a proper intelligence agency in its own right. By the spring of 1946, Edward Travis was operating with two deputies at Eastcote. Nigel de Grey was the senior deputy, and had responsibility for operational coordination between the five main groups at GCHQ, together with recruitment, training and security. Following the important allied sigint conferences of February and March 1946, Travis added a second deputy, a naval officer called Captain Edward Hastings who had much wartime experience of working with Canada. His responsibilities included liaison with the US, the Commonwealth and India, together with managing GCHQ’s overseas collection stations.27

GCHQ’s strategy for cooperation with the Americans was to rapidly reorientate its collection towards Russia. Typically, the vast Forest Moor wireless station near Harrogate in Yorkshire, with an aerial farm of some ten square miles, was switched from collecting German traffic from the Eastern Front to Russian traffic as soon as the war drew to an end. British field units in Germany, Austria and Italy joined the suborned Italians in collecting Russian military traffic. The re-established sigint stations in Singapore and Hong Kong also focused on Russian traffic, with the latter specialising in KGB messages. All this made Britain an attractive partner for the United States.28 The core of Anglo-American cooperation was a ‘relentless attack’ on the wartime generation of Russian cyphers. Figures like John Tiltman and Hugh Alexander provided the code-breaking expertise, while the Americans provided most of the processing capability.29

All the three American armed services were routinely circulated with GCHQ finished product on Russia. A key instrument was the ‘Comintsum’, a digest of the latest ‘hot’ material which made its way around comint-cleared centres. London would send twenty copies of this sort of document to Washington on a regular basis, with two copies going to US Air Force intelligence, two to US Army intelligence and so forth.30 On Russian military targets at least, the British and Americans operated smoothly as one machine. A very high priority was given to joint planning for the use of nuclear weapons in any future war. As early as 28 April 1948, General Charles Cabell, head of US Air Force intelligence, reviewed the intelligence arrangements in support of the current emergency atomic strike plan ‘Operation Halfmoon’. ‘At the present time,’ he noted with satisfaction, ‘there is complete interchange of communications intelligence information between the cognizant United States and British agencies. It is not believed that the present arrangements…could be improved.’31 This was cemented by a further Anglo– American agreement on communications intelligence signed in June 1948.

However, the sharing of material on other parts of the world remained selective, reflecting the political tensions of the moment. In 1948, even while the UKUSA alliance was gradually being drawn together, Britain and America were at loggerheads over Palestine and the emerging state of Israel. There was anxiety in London about sharing intelligence on the Middle East with the Americans. On 15 February 1948, Britain’s Joint Intelligence Committee discussed the problem of circulating its own reports, which included material from SIS and GCHQ, to the newly formed CIA. Although British intelligence representatives in Washington were not aware of a specific ‘pro-Zionist bloc in the Central Intelligence Agency’, nevertheless they asserted that ‘Jewish sympathisers were no doubt included in its establishment’, and complained that there had been leaks. William Hayter, the Chair of the JIC, insisted that its material on Palestine should be shown in the first instance only to the Director of Central Intelligence in person. He added that ‘It should be explained to him that if he could not guarantee that they would not fall into pro-Zionist hands, then he could not be left with them.’ Even so, it was decided to withhold more sensitive recent reports on Palestine from the Americans.32 Meanwhile, ‘Operation Gold’, run by US Navy intelligence, was intercepting the cable traffic of Jewish arms smugglers, but this was not being shared with Britain, or indeed acted upon.33

American links with the Commonwealth parties were also hesitant. The Americans were slow to do business with the Canadians. They worried about how much GCHQ had told the Canadians about Anglo-American agreements, and suspected that GCHQ was secretly giving Ottawa some American sigint.34 During the 1948 discussions of possible CAN-USA sigint agreements, it became clear that the US Communications Intelligence Board was anxious to prevent an information free-for-all. It preferred to hand material to the Canadians on a ‘need to know’ basis, and was anxious to prevent a proliferation of sigint liaison officers.35 Meanwhile, somewhat foolishly, Canada resisted the all-important standardisation of security procedures that was a foundation stone of the BRUSA agreement, so negotiations were ‘very difficult’, dragging on until 1953.36 The Americans were even more wary of sigint cooperation with the Australians due to the KGB spy cases uncovered by Venona in the late 1940s. This, in turn, retarded the joint sigint effort against the newly formed People’s Republic of China from 1949. In late 1953, the advent of a Liberal (i.e. conservative) government in Australia triggered a full resumption of cooperation, formalised at a tripartite sigint conference between the Americans, British and Australians. New Zealand also came in as fifth partner. It was only at this point that the name ‘UKUSA’ was adopted at GCHQ’s request.37

The most prickly area of alliance relations was the business of cypher security, which protected the secrecy of diplomatic communications. Foolishly perhaps, at a meeting in London in May 1947, the British launched an audacious bid to persuade the Americans to share the innermost secrets on matters of their code-making. The discussion revolved around the replacement of the Combined Cypher Machine, which had been developed for inter-allied communications during the war, but was now thought obsolete and vulnerable. The British were also keen to replace their own national machine, the Typex, but were desperately short of money as a result of post-war austerity, and argued that for reasons of economy any new cypher machines should be capable of inter-allied use, and proposed joint research and development with the Americans.38 The Americans were startled: it had been a cardinal principle never to share the secrets of their unique and highly prized Sigaba machine. Hoping to overcome this psychological barrier, the British revealed that they were in fact already knowledgeable about Sigaba. They not only described its inner workings ‘quite accurately’, but confessed that they had ‘incorporated its principles in a radioteletype machine for their own use’. Hoping that they had pushed the Sigaba obstacle aside, the British then made their pitch. They claimed that they had developed an approach to cypher machines that was ‘new and revolutionary’, and ‘superior to the Sigaba principle’. They were happy to share this with the Americans, and perhaps make use of it in joint machines that might be developed for both national and allied use.

Far from being reassured, the Americans were horrified. Discussion had to be ‘temporarily discontinued’ while they withdrew to confer amongst themselves. The US Army could see no objection to releasing the Sigaba principle for use in a combined allied machine, since the British had clearly unravelled it. However, the US Navy offered ‘serious objections’, and used their veto. Thus the British were told that Sigaba had to be completely eliminated from the discussions. At this point they revealed their ‘new and revolutionary idea’ for future cypher machines, only to find that their American colleagues sneered at it and dismissed it as ‘impractical’ on engineering grounds. The two sides parted without agreement.39

While GCHQ was overawed by the scale of American sigint resources, matters looked quite different from Washington. With the Second World War now over, and an economising Republican Congress controlling the federal purse-strings, resources for American co mint interception activities were remarkably tight. This contributed to American under-preparedness prior to the Korean War. It also enforced a division of labour between GCHQ and the Americans, and prevented American sigint from expanding its activities in Europe in the way it had hoped. In 1949, US Army Security Agency interception units in Europe were still passing much of their product to GCHQ for analysis, rather than back to Washington. Moreover, GCHQ retained primary responsibility for areas such as Eastern Europe, the Near East and Africa.40 Because of this division of labour, the late 1940s saw the gradual development of American and British spheres of influence. In Scandinavia, for example, relations with Norway were an American responsibility, while those with the Swedes belonged to GCHQ, although this demarcation was not always strictly adhered to.41 GCHQ enjoyed the additional benefits of the panoply of bases provided by Britain’s imperial and post-imperial presence. Although the Empire was shrinking, the very process of retreat and the euphoria of independence often rendered the new successor states willing to grant limited base facilities to the departing British. These ‘communications relay facilities’ may have seemed innocuous, but in fact many countries were unwitting hosts to important GCHQ collection sites.42

The outbreak of the Korean War early on the morning of Sunday, 25 June 1950 took Britain and the United States by complete surprise. Although they had comint units in locations such as Hong Kong and Japan, their main focus was Russian traffic, and their sigint capabilities against North Korea were non-existent. The NSA official history notes that there was ‘no person or group of persons working on the North Korean problem’, and even had they done so, they had ‘no Korean linguists, no Korean dictionaries and no typewriters’. Although the CIA had picked up what might be called ‘rumours of war’ from human agents, there was no high-profile attack warning delivered to policy-makers. During the first few weeks of the war, the Americans and their South Korean allies suffered serious reverses and were almost overrun. Sigint helped the Americans to beat back the attacks on their rapidly shrinking perimeter by providing excellent tactical intelligence, but they blundered again by missing the entry of the Chinese into the war in October 1950.43

The Korean War resulted in a headlong expansion of American sigint. More than two thousand additional staff were recruited, and more than $5 million of additional spending on comint and comsec was authorised within weeks of the war commencing.44 The outbreak of the war also meant crash expansion in Asia. The Americans informed London of their ‘urgent need’ for a US Air Force sigint unit to be deployed to Hong Kong, and other sites were quickly developed on Taiwan in an attempt to remedy the yawning intelligence gap in East Asia.45

Korea had another important impact. With new employees flooding into the training wing at Nebraska Avenue in central Washington DC, the Americans soon had a vast backlog of people requiring security clearance before they could begin work. By the end of 1950, more than a third of sigint employees were ‘uncleared’. It was then discovered that since 1948 the CIA had been using the polygraph, or lie detector test, initially only to screen people who had access to sigint, although its use was soon extended to all CIA employees. By May 1951 it had been adopted for all American code-breakers, and polygraph examiners were testing ‘from seven in the morning till eleven at night’ to clear the backlog. Polygraphs soon became an embedded part of American sigint culture, but were not introduced at Britain’s GCHQ.46

The Korean War was of enormous importance for GCHQ because it fundamentally reshaped the American sigint community. There had been several failed attempts to create a single unified American sigint organisation along the lines of GCHQ.47 The war broke the logjam. In 1952, President Truman suddenly insisted on the creation of a strong central body called the National Security Agency, or ‘NSA’, under General Ralph Canine. The armed services fought a desperate rearguard action: in August 1952, General Samford of US Air Force intelligence denounced Truman’s desire for ‘strong central control’ as nothing short of a ‘major error’. However, Truman’s mind was made up, and in November he signed the order for the reshaping of American comint.48 NSA was given unambiguous control over comint in a historic document called NSCID-9.49 This brought about a reduction in, but not the elimination of, what the leading historian of NSA has called ‘the fractious and seemingly never-ending internecine warfare’ between the American service comint organisations. The British were immensely relieved. In the background, figures like Edward Travis had been quietly urging inter-service unity on their American collaborators since the summer of 1945.50

The creation of NSA also had physical consequences. Up until this point there had been a plan to relocate the headquarters of American sigint to Fort Knox, near Louisville in Kentucky. However, it was now realised that the need for high-grade communications circuits and for civilian workers made this impossible. The policy-makers, who were the consumers of their intelligence ‘product’ in Washington, also protested about the move, rightly anticipating that it would mean a worse sigint service. They insisted that the new headquarters be within a twenty-five-mile radius of the Washington area.51 On 3 November 1952, Fort Meade on the northern edge of Washington’s Beltway was designated the likely new headquarters for NSA under a secret programme entitled ‘Project K’. Over the next five years this location would become the headquarters of the world’s largest, most expensive and most secretive intelligence agency52

Britain’s GCHQ had made precisely the opposite decision. In 1952 it moved away from its suburban site at Eastcote on the perimeter of London to a comparatively distant location at Cheltenham in Gloucestershire. Although GCHQ had only finished the move to Eastcote in June 1946, by April 1947 it was already looking for a new home. This was partly because of the physical limitations of the Eastcote site, and also because Travis realised that in any future war there would not be time to relocate to a safe place like Bletchley, away from Soviet bombing. Several possibilities were considered, and in October 1947 GCHQ scouts had found promising twin sites at Oakley Farm and Benhall Farm, near Cheltenham, which were occupied by the Ministry of Pensions. These single-storey temporary office complexes had been built in 1940, initially for the possible evacuation of government from London during the Blitz. After 1942 they were used for the logistical organisation of the US Army in Europe. The wartime presence of the Americans was the key, since it had left a helpful legacy of improved trunk cable communications.53

An alternative explanation for the choice of Cheltenham is offered by Professor R.V. Jones, one of the famous architects of the ‘Wizard War’ which deployed British science against Nazi Germany. Jones served as scientific adviser to GCHQ in the early 1950s, and recalls that the scout who initially found the Cheltenham site was Claude Daubney, one of the senior GCHQ staff who liaised with the Y Units of the armed services, and who spent much of his time in Whitehall. Daubney was a typical RAF officer of the thirties, handsome and ‘heavily moustached’. His main relaxation was betting on horseraces, and he argued endlessly with Jones about the theoretical possibility of beating the bookmakers. He ‘deliberately chose Cheltenham, as he told me, so that he could combine visits from his London office to GCHQ with attendance at Cheltenham races’.54 Others support the idea that GCHQ was attracted to Cheltenham by its proximity to the racecourse.55 Whatever the truth of the matter, the course would serve as an occasional helicopter landing pad for future visits by British Foreign Secretaries.56 The move from Eastcote took place between 1952 and 1954, but the rapid growth of GCHQ during the Korean War meant space was tight even at the new location. This contributed to the decision to constitute comsec as a separate organisation called the London Communications Security Agency. Its staff either moved to central London offices in Palmer Street, or stayed at Eastcote.57

In its move to Cheltenham, GCHQ created a unique intelligence town. It brought with it Bletchley Park’s formidable reputation for secrecy. Within a decade almost everyone in Cheltenham had a family member or friends who worked at GCHQ, but nobody talked about what they did. The town welcomed the ‘Foreign Office types’ with open arms, and GCHQ was soon contributing an enormous amount to its intellectual and artistic life, quite apart from being its biggest employer. The GCHQ staff were also sporty, providing most of the players in the Foreign Office football team that won the Civil Service Football Cup in 1952. This could present some peculiar problems. When local reporters covered matches in Cheltenham, they were told they could name the goal-scorers of the visitors, but not of the local team. Reporting these games tested their copywriting skills to the very limit.58

The previous ten years had been a formative decade for GCHQ. By 1944, the code-breakers of Bletchley Park had made themselves Britain’s premier intelligence service. Out of the chaotic brilliance housed in a few wooden huts in the Buckinghamshire countryside came one of Britain’s most forward-looking and innovative organisations. Another crucial legacy of the late 1940s was the agreements with the United States and the Commonwealth that laid the foundations of UKUSA, a worldwide sigint alliance, agreements that are still in force today. Anxiety about Moscow had been a driving force behind these agreements, and even before the Second World War was properly over, the Western allies had been paying increased attention to Soviet cyphers. Despite the triumphs of Venona, and the uncovering of key KGB agents like Klaus Fuchs and Donald Maclean, Moscow’s higher-level communications remained mostly unbreakable after 1948. Accordingly, GCHQ and its partners were already searching for new kinds of intelligence-gathering to use against the Soviet Union, opening up a whole new vista in the electronic war.59