GCHQ

by Richard Aldrich

  The German newspaper Der Spiegel has alleged that later, a secret meeting of the BND had discussed how to change the ownership of Crypto AG, perhaps through a merger. Reportedly, the BND also considered how the Swedish communications company Ericsson might be persuaded, through the medium of its German counterpart Siemens, to terminate its expanding cryptographic business. One former employee of Crypto AG stated that he had to coordinate his developments with shadowy figures ‘from Bad Godesberg’ – the location of the Central Office for Encryption Affairs of the BND. Employees also recalled the arrival of NSA officials who advised on the use of certain encryption methods.53 All this is perhaps not as surprising as it appears, since Switzerland and Sweden had been secret de facto allies of NATO since the early Cold War.54 A recently declassified British document shows that NATO had already come to an unofficial understanding with Switzerland: in peace it would be neutral, but in wartime it would support NATO.55

  Looming over all these cypher-machine issues like a vast black cloud was the troublesome issue called ‘Tempest’, the term used for the tendency of all electronic devices to radiate an electromagnetic field over a short range, usually less than a hundred yards.56 There has been much debate about when this phenomenon of radiation or ‘emanation’ was first identified. The NSA official history states that Germans had discovered it during the Second World War, but that its real importance was only recognised by a CIA technician in 1951.57 Whoever discovered it, during the early 1950s it had become clear that this radiation could allow an enemy to listen in to cypher machines at short range in a way no one had previously envisaged. The British do not seem to have been aware of Tempest until 1952, when it was accidentally discovered that a cypher machine in the Washington Embassy was radiating out the plain text of the messages it was processing. Tempest was a huge problem, because many expensive cypher machines were suddenly found to be vulnerable.58

  Nevertheless, Tempest also allowed the British to attack the cypher machines of their enemies. In his memoirs the former MI5 officer Peter Wright claims to have been one of the first users of Tempest to eavesdrop on cypher machines in the French and Egyptian Embassies in London. There is no doubt that such operations took place. During the late 1950s and early 1960s Wright joined John Hawkes from SIS in planning these sophisticated activities.59 Ralph Benjamin, GCHQ’s Chief Scientist, remembers Wright as a superb self-taught technician.60 Wright recounts an operation in 1956 in which British intelligence successfully attacked the Hagelin cypher machines in the Egyptian Embassy in London by picking up emanations from the keyboard using a microphone in an operation code-named ‘Engulf’.61 In the late 1950s, in what appears to have been a coordinated effort with the Americans, British intelligence collected signals generated by French diplomatic cypher machines in London and Washington. These efforts ended in late 1964, when the French finally installed protective copper shielding in their cypher rooms to prevent radiation.62

  What none of these accounts captures is the exquisite dilemma of offence versus defence. With Tempest, the conflicting demands of offensive sigint and defensive communications security were so complex as to make the head spin. The optimum solution was for GCHQ and NSA to be able to read the traffic of minor allies and neutral countries themselves, but to provide them with enough defensive expertise to make their communications immune to similar code-breaking efforts by the sigint specialists of the KGB. Tempest allowed GCHQ and NSA to launch innovative attacks on all sorts of machines that had not yet been broken, but this was only an unalloyed virtue if the Soviets were lagging behind in its use. This, in turn, raised another awkward question: how much did the Soviets know?

  In April 1958 Fred Stannard visited Ottawa to brief the Canadian Cipher Policy Committee on the latest developments. His message was that Tempest was a terrible and complex security problem. While some cypher machines radiated out an electromagnetic field, others gave out an acoustic echo that could travel down a wire for hundreds of feet. Still others caused tell-tale fluctuations in the power supply. Even those machines that did not radiate of their own accord were susceptible to ‘interrogation attack’. This meant that the KGB could potentially direct a microwave or laser beam at the cypher room, reflecting back reverberations that might give away the clear text of a message as it was typed on the keyboard. The only saving grace was that the danger was confined to a few hundred yards at most, so the enemy had to get close to the cypher machine they wanted to listen to.63

  Stannard explained that the most awkward issue presented by Tempest was ‘how we may best distribute our responsibilities for advising our allies, particularly in NATO’. Tempest was a nuisance in terms of security, but of course it was beneficial for offensive sigint operations by GCHQ. The British did not want information about Tempest to spread to ‘countries or organisations from which signal intelligence is required’. Where the balance of advantage fell depended on the relative importance attached to either sigint or security. Understandably, perhaps, there were some arguments between GCHQ and LCSA over this matter.64

  By the early 1960s the awkward NATO Tempest question was gradually being resolved. A handbook was provided to the European allies that explained how to install cypher equipment so as to minimise radiation risks from Tempest.65 However, LCSA emphasised that, within the inner circle constituted by the British, Americans and Canadians, the standard NATO briefing had always to be accompanied by ‘advice on certain aspects of the problem which it is undesirable to disseminate to NATO at large’. This circumlocutory language suggests that Britain was offering some of its NATO partners incomplete advice, leaving open certain avenues for exploitation. At the same time, GCHQ hoped fervently that the KGB was not using the same techniques.66

  At last, after two troublesome decades, Britain’s code-breakers and code-makers were finally getting to grips with Tempest. The issue, together with independent cryptography by European NATO allies and neutrals, had presented sensitive and costly problems. However, there had also been enormous achievements, many of which are still shrouded in deep secrecy, which provided continuous access to many streams of diplomatic traffic around the world. The main beneficiaries were K Division at GCHQ, whose task it was to read non-Soviet systems. However, these triumphs also owed much to the work of the unsung heroes of communications security, hiding out in one of Britain’s least-known secret service headquarters in Palmer Street in central London. In 1969, their last year of independent operation before they were merged with GCHQ, Fred Stannard’s colleagues declared that their secretive influence over cypher machines was the surest route to good intelligence: There is no better way to successful Sigint than to influence selected target countries by Comsec advice to use a source of equipment desired by Sigint.’ This, they added with quiet satisfaction, ‘can sometimes be done’.67

  Although the problems of Tempest were now being addressed, they were frighteningly expensive to resolve, and in the 1960s money was a major problem for all of Britain’s secret services. Indeed, if a single adjective had to be chosen to describe British intelligence during the Cold War, it might well be ‘impecunious’. The real world of British secret service at this time was a shabby one, and the persona of the average intelligence officer was less that of Ian Fleming’s glamorous James Bond than of Len Deighton’s down-at-heel Harry Palmer. In Cheltenham, Berlin, Cyprus or Hong Kong, thousands of intelligence operatives endured dingy offices notable only for their peeling paint and rotting linoleum. Moreover, there was remorseless pressure to economise.

  One obvious area for economies was the three sprawling intelligence empires owned by the Royal Navy, the Army and the RAF. As we have seen, each of the armed services had its own intelligence activities. Their largest collection task was sigint,which was mostly picked up by dedicated military units of listeners equipped with headphones sitting in Germany, Cyprus and Hong Kong and working in close collaboration with GCHQ. In December 1960 an inquiry by General Sir Gerald Templer, the ‘Tiger of Malaya’, argued that the three inte
lligence service directorates should be merged to form a single Defence Intelligence Staff, and that most of the secret military listeners who did sigint collection should be replaced by civilians and put under the direction of GCHQ. Templer was right, since National Service, which had provided an almost unlimited source of personnel for these military sigint units, was coming to an end.68

  GCHQ’s money problems were made worse by trying to keep up with the Americans. Back in 1955, President Dwight D. Eisenhower had appointed a major inquiry into intelligence, headed by the leading scientist James R. Killian. Like Templer, Killian had concluded that it was now ‘exceedingly difficult’ to run human agents inside Russia, where the security police were ‘brutally effective’. In future the USA would have to depend on science and technology for its intelligence. This would mean spending huge sums of money on spy satellites and code-breaking. American intelligence, he noted, was verging on being a billion-dollar-a-year business. Another intelligence inquiry, the Hoover Commission, came to much the same conclusion. It also proposed an all-out attack on Soviet cyphers using the finest minds and the best computers that money could buy – the equivalent of the Manhattan Project which had produced the first atomic bomb.69

  By 1962 the cost of British sigint was rising fast, while overall government expenditure was being cut. Something had to give. The issue came to a head in a secretive Whitehall committee called the Permanent Secretaries Committee on the Intelligence Services, or PSIS. Chaired by the Cabinet Secretary, Sir Norman Brook, it allowed the most senior denizens of Whitehall to debate the sharing out of the money allocated to the secret services.70 In January 1962, PSIS expressed undisguised horror at the rising cost of sigint. While careful to praise GCHQ’s achievements, it concluded that it was time for a comprehensive review of the organisation. It was moved in part by the success of Templer’s recent inquiry into service intelligence, and hoped fervently for more of the same. Norman Brook called for a wide-ranging investigation into how GCHQ should develop in the face of ‘increasing technical difficulties’ which were pushing up costs, and whether the expenditure involved was likely to be ‘commensurate with the intelligence obtained’. Over the past decade GCHQ had enjoyed two generous allocations of additional funds to boost its work on Soviet high-grade cyphers. These five-year plans for improvement were entitled ‘Methods to Improve’, or MTI. The first was launched in 1952–53 and the second in 1958–59. Although much knowledge about the Soviet military had been gained from special operations with aircraft and submarines, the golden prize of breaking high-grade cypher traffic and returning to the heady days of ‘Ultra’ had eluded GCHQ.71 The review would also look at the future availability of ‘overseas sites for interception’, since decolonisation was constantly whittling away at GCHQ’s precious overseas bases.72 GCHQ had been instrumental in the development of some of the best British computers in the 1950s, but was now having to buy from the American company IBM to obtain state-of-theart machines. These were expensive, but GCHQ knew that without constant investment in better computing it was dead in the water. In 1963–64 it hoped to double its spending on equipment, mostly on new computers, taking its budget from £7.5 million to £11 million. Set against a background of recent Whitehall economies that had imposed a 10 per cent cut on all overseas departments, this was an extravagant request.73 Moreover, the cost of GCHQ was only half the picture. Some of the costs of British sigint were hidden in the activities of the armed services, and were of a similar size and scale. In fact the total sigint spending was now £20.5 million, and it would soon overtake the cost of the whole of the Foreign Office, with all its embassies and diplomats, of £21.8 million. In other words, the servant would soon be paid more than its master. It was hardly surprising that a serious review had now been triggered.74

  In mid-May 1962 the Chair of the JIC, Hugh Stephenson, went to the Treasury for an informal brainstorming session with Burke Trend to decide who should chair the review.75 Trend was a rising star in the Treasury who had a special fascination with the intelligence services, and presided over their budgets. Their eventual choice was Sir Stuart Hampshire, an Oxford philosophy don. Hampshire seemed an improbable figure to review the future of Britain’s most technical intelligence agency. With a First in Greats, he may have commanded the respect of civil servants, but he had no background in maths or science. Having been elected a Fellow of All Souls in 1936, he joined the Army in 1940. His otherworldly nature ensured that his superiors were reluctant to let him loose with a compass and a pistol, so he was moved into intelligence. By late 1942 he was in SIS, working alongside Hugh Trevor-Roper and the Radio Security Service. Some of his time was spent at Bletchley Park, and in July 1944 he found himself deliberating over the July Bomb Plot against Hitler with several other young SIS officers.76 Like Trend, Hampshire had a romantic fascination with intelligence, and he later confessed that he had greatly enjoyed ‘the spectacle of duplicity and deceit in secret intelligence during the war’.77

  GCHQ prepared for the review by producing a report on the ‘State of Sigint’ and material on ‘Interception Deployment in the Sixties and Seventies’, and forecasting the costs of research and development.78 In July it offered a historical summary of its post-war activities running to almost forty pages. The first versions were regarded as evasive by Whitehall, offering ‘no assessment of the extent to which their policies had succeeded’. After much coaxing, later drafts identified precisely what had been achieved by the two five-year MTI programmes.79 Bill Millward, GCHQ’s Principal Establishments Officer, offered an account of the core activities against Soviet systems. This was painful for several reasons. It not only put an unsuccessful programme under the spotlight, it also required GCHQ to let the Treasury have ‘specially sensitive information’. Millward found this excruciating, and asked that it be seen only by ‘the smallest number of people’. The most interesting part of the report detailed GCHQ’s expensive failed attack on the main high-grade Soviet diplomatic cyphers. The Treasury was asking whether this should ever have been attempted.80

  The star performer during the review was Joe Hooper, the Deputy Director of GCHQ. Hooper knew that cooperation with the Americans was Cheltenham’s trump card, and went to great lengths to explain how fortunate Britain was to reap the benefits of ‘the much more expansive and extravagant effort of the Americans’. He added that senior figures at Cheltenham were genuinely worried about whether they might ‘not be making an adequate contribution to the UKUSA partnership’. In the late 1940s sigint had often been a story of British cryptanalytical skill combined with American computer power, but now ‘the Americans are becoming less dependent upon us because they are getting better themselves’. The Treasury pressed GCHQ to revisit the UKUSA agreement in order to ‘reach a new understanding’. It wanted GCHQ to admit frankly that it could no longer afford to partner NSA’s more ambitious programmes, and needed to make a more selective contribution. This made economic sense, but GCHQ regarded such ideas as utterly unthinkable.81 The Hampshire review also raised the awkward question of ‘overlap’. This was ultimately about the degree of mutual trust between GCHQ and NSA. GCHQ had often boasted that there was complete exchange between the British and the Americans, but when pressed hard it admitted that there was duplication of effort – even on the Soviet target. More importantly, the head of K Division, which looked after non-Soviet targets, acknowledged that there was ‘considerable and inevitable overlap’ in his province, and the British and the Americans were duplicating much of their work on countries in the Middle East, Africa and Asia. ‘The reason is we cannot trust each other on sensitive matters,’ he admitted. This was because British and American foreign policy diverged in these regions. Inevitably, on intelligence relating to economic matters there was almost no exchange at all. The Treasury demanded ‘more of a carve-up and less overlap’.82

  Hampshire’s probe of sigint lasted several months, and included a six-week visit to the sprawling NSA headquarters at Fort Meade that lasted into January 1963.83 He
re, Hampshire was briefed on the super-secret plans of NSA to expand sigint collection using satellites like the highly successful ‘Grab’ launched in June 1960. He soon concluded that this was one road GCHQ could not travel, since the costs of satellite operations were prohibitive.84 This also reflected his worries about poor scientific recruitment at GCHQ.85 His work gave him insights into some of the most secret activities of MI5 and SIS. He met the MI5 officer Peter Wright to discuss GCHQ support for scientific efforts to locate the radios of KGB agents in Britain. Hampshire was broadly supportive of this, but wished to wind up ‘Airborne Rafter’, the element of the operation conducted with the help of the RAF’s 51 Squadron, which was proving very expensive. Wright, a zealous counter-espionage enthusiast, resisted, but eventually agreed with Hampshire that the flights were not cost-effective.86