GCHQ

by Richard Aldrich

  Harold Macmillan was clearly rattled. On 12 July he told Eisenhower that the whole question of using UK bases for reconnaissance flights would have to be looked at, and suggested that ‘Your and our intelligence authorities should conduct a joint review about the system and conditions under which these flights are conducted.’ Macmillan felt that this was particularly important, as the Soviets had announced that they were going to adopt ‘a policy of shooting down any aircraft, wherever it may be, on the grounds that it is conducting reconnaissance or is in an area from which they prefer to exclude foreign aircraft’. Macmillan wanted this review to take place as soon as possible. Although he told Eisenhower that the RB-47 incident had increased the ‘sense of solidarity between us’, in reality it had also heightened the sense of risk regarding the whole business of sigint surveillance from any platform.14

  Andrei Gromyko took the opportunity to highlight the issue of the intense surveillance of Soviet vessels in international waters. He complained to Western diplomats, including Patrick Reilly, about the ‘provocative buzzing of Soviet ships’ by American aircraft on the high seas, and asserted that this was happening on all the oceans of the world, with even civilian research ships being subjected to the practice. ‘The buzzing of Soviet ships by American planes is done, as a rule, at mast-top altitude in the dangerous proximity of Soviet ships, often involving dives down on the ships, simulated bomb and torpedo release movements, the dropping of various moving objects and incendiary devices…as well as other impermissible actions.’ Gromyko claimed that there had been 250 such incidents in the first five months of 1960.15 The truth of these assertions remains unknown, but there were clearly some cowboy incidents, and even NSA’s own official historian of the Cold War has admitted that, broadly speaking, reconnaissance ‘exacerbated an already touchy situation’.16

  In Britain, clearance procedures for sigint flights were already stringent, having been tightened up after the Buster Crabb incident in 1956. David Ormsby Gore, the Minister of State for Foreign Affairs, asked his officials, ‘Surely the RAF undertake no flights which entail some risk of incident unless the operation has been cleared by a Foreign Office Minister of State or by the Permanent Under-Secretary?’ Peter Wilkinson, head of the Permanent Under-Secretary’s Department, confirmed that this was correct. What happened, he explained, was that the RAF submitted a programme of ‘special flights’ for the Prime Minister’s approval six months in advance. Closer to the time, each day flight had to receive additional clearance from the Foreign Secretary, while night flights, which were thought less risky, were cleared by the Permanent Under-Secretary. This procedure included American reconnaissance missions from British bases.17

  Harold Macmillan absolutely hated the public fuss caused by spy scandals and espionage incidents. Accordingly, he took a deep personal interest in these matters.18 On 1 August 1960 he asked his Foreign Secretary to conduct an inquiry into ‘ “buzzing” in all its forms’. He was not referring to operations like the U-2 mission, which he considered to be clearly illegal. Instead he meant ‘things which are within the law but no doubt disagreeable to the victims…It seems to me these things may become very dangerous.’19 Unusually, Macmillan insisted on a series of special meetings for key Cabinet Ministers with the Chairman of the JIC, Sir Hugh Stephenson, in attendance. The rules for British sigint flights were firmed up, but the problem was what to do about the Americans. Against his instincts, Macmillan admitted that there was no benefit in raising the matter in the dying days of the Eisenhower administration.20

  In Britain, the twin shootdowns of 1960 had a similar impact to the Buster Crabb affair four years before. This time they served to crush a plan for increased airborne surveillance of the Soviet fleet that had been agreed between the First Sea Lord and the US Navy’s Chief of Naval Operations.21 By March 1961 these plans had been postponed, and they were later ‘put into cold storage indefinitely’.22 Other long-established British programmes were also brought to a close. Like Eden before him, Macmillan now required the JIC to prepare a review of all aerial and submarine surveillance tasks, covering both imagery and sigint, so he could assess the value of the intelligence gained from these activities and measure it against the risks.23 Defence intelligence chiefs were miserable, since the flow of short-range intercepts from these missions had been one of the few precious areas of sigint success. Air Vice Marshal Bufton, head of Air Intelligence, noted: ‘The bulk of the intelligence on Soviet scientific research and development on missiles comes from G.C.H.Q. intercepts,’ and ‘it is “hot” intelligence’.24

  These developments contained an element of irony. In the early 1950s, Britain and the United States had placed growing emphasis on ‘technical means’ of espionage, such as sigint-gath-ering aircraft and overhead photography, because the use of human agents inside Russia had proved increasingly hazardous and, with a few exceptions, rather unproductive. But now, forward surveillance with aircraft and submarines was also proving risky. Safer alternatives certainly existed. As early as 1956 the British had begun cultivating an alternative form of sea-borne surveillance: gathering intelligence on the Soviet fleet from the relative safety of British trawlers operating in northern waters. This was a direct copy of the Soviet sigint trawlers that became a ubiquitous plague by the 1960s. However, the real future of risk-free intelligence-gathering lay with satellites. The Americans were already experimenting with the collection of both sigint and overhead photography from space.25

  On 22 June 1960, less than two months after the shoot-down of the Gary Powers U-2, the United States launched its first sigint satellite. For many years it was believed that the first operational intelligence-gathering space project was a satellite called ‘Corona’, which collected imagery. In fact the first intelligence satellite was ‘Grab’, which stood for ‘Galactic Radiation and Background’. It was so secret that no details were released until the late 1990s. Publicly announced as a US Navy satellite designed to monitor the effects of solar radiation on radio communications, in reality it had a more secret second mission, namely gathering elint from Russia. Naval engineers simply took the electronic intelligence antennae that were being used on submarines penetrating Soviet harbours and bolted them onto the outside of the satellite. In the highest traditions of space engineering, the first designs were drawn on the back of a paper placemat when scientists were stranded in a Howard Johnson’s restaurant during a snowstorm. Although crude, the Grab satellite was highly effective, and soon torrents of elint from its listening missions high over the Soviet Union were pouring down from space. In the 1960s such satellites remained small, since it was more than a decade before their potential for listening in to communications as well as elint from radars became apparent. However, this was clearly the long-term future of sigint collection, and eventually they would be followed into space by intelligence-gathering giants.26

  The British and Americans were also enjoying new scientific breakthroughs in the highly secret world of cypher machines. The target of these operations was not the Soviet Union, but rather their West European friends and allies. These countries, whether allied like West Germany or neutral like Sweden, had a curious dual status as both lowly intelligence collaborators and also sigint targets. The management of relations with such allies and neutrals in Europe, some of which eventually became ‘third party’ members of the UKUSA alliance, was often controversial. As we have seen, immediately after the Second World War, more than half the traffic being analysed by American code-breakers was French.27 During the late 1950s and early 1960s, one of the big issues for the British and the Americans had been supplying cypher machines to fellow NATO countries. Most of those that had been supplied to European countries after the war were now regarded as insecure.28 This was because in the 1950s it had been discovered that many cypher machines that produced secure codes nevertheless radiated out a clear signal of the message they were sending over a distance of about a hundred yards – a problem known as ‘Tempest’.29 To deal with this, a new Ameri
can cypher machine called the KL-7 was rushed into service within NATO, and was also given to other partners, including Rhodesia and South Africa.30

  Cypher machines are extremely expensive, and the United States bore much of the cost of re-equipping Western Europe’s communicators. This was partly to improve security in order to thwart Soviet code-breakers. However, the more secret explanation, which no one wished to admit to, was that NSA and GCHQ wanted to suppress the efforts of European countries to make their own cypher machines.31 The British and Americans did not want the Continental Europeans to develop their own commercial cypher-machine industry, exporting machines around the world which would then generate messages that would prove difficult for GCHQ or NSA to break. In short, supplying American machines like the KL-7 virtually for free was intended to undercut the market and drive European manufacturers out of business.

  Cypher machines remained a hot NATO issue throughout the 1960s. Partly because of their massive resources, the Americans were in the driving seat, and were advancing radical ideas. NSA argued that NATO countries in Europe should be allowed to manufacture British- and American-designed machines under a scheme called ‘free licensing’. This meant making these designs available without cost, and ensured that NATO communications would enjoy the greatest degree of commonality and inter-operability. More importantly, the Americans also wished to continue to discourage smaller NATO countries from developing their own cypher-machine industries. Although GCHQ liked the idea, other parts of British government did not. Many in Whitehall felt they had no power to compel British manufacturers of cypher equipment to grant free licences to the Europeans, and would have to pay them expensive compensation. Moreover, London had almost completed negotiations for the Canadians to manufacture the impressive new British Alvis on-line cypher system under licence, and feared that this deal would be undermined.32

  In June 1962 there was a serious clash between Clive Loehnis, the new Director of GCHQ, and the Ministry of Defence.33 GCHQ supported the American proposal, explaining that ‘SIGINT wanted to avoid the spreading of cryptographic techniques and knowledge,’ and that NSA’s proposal would remove ‘the financial incentive from the development of cryptographic equipment in NATO’. Loehnis’s colleague Fred Stannard, who looked after the communications security organisation known as LCSA, heartily agreed.34 However, commercial interests won the day. To the utter dismay of GCHQ and LCSA, Whitehall offered a stern rejection of the American ‘free-licensing’ idea. Stannard had the unpleasant duty of telling the Americans that their ideas had been vetoed.35 He later reported that the news had ‘shocked’ NSA, and GCHQ’s liaison officers in America feared that ‘our attitude will in fact impair the relations between NSA and GCHQ and LCSA’.36

  Loehnis and Stannard did not give up, and now argued that Britain’s refusal to cooperate might lead to NSA hardening its approach to Britain, especially in the area of the Mutual Weapons Development Program, a secret route by which the Americans subsidised equipment for both GCHQ and LCSA. NSA also saved Britain costs by assisting with the testing of the security of its new cypher equipment. Loehnis argued that agreeing with NSA on this matter would not only ‘prevent damage to our most important and unique SIGINT and COMSEC relations with the United States’, but would also save money in the long run.37 It was certainly true that over the previous decade large amounts of expensive equipment used by GCHQ’s sigint programme had been supplied through this scheme, mostly for elint and electronic warfare.38

  The matter culminated in a remarkable showdown at the Treasury on 10 July 1962. Loehnis and Stannard set out what they insisted was the irreplaceable value of close sigint cooperation between GCHQ and NSA, emphasising that the ‘principal feature of the partnership was the full interchange of the “product” ‘.39 LCSA’s relationship with the communications security division of NSA was, they declared, no less close. With perhaps a little hyperbole, they argued that Britain might ‘endanger this partnership’ by appearing to act unsympatheti-cally. They warned that:

  N.S.A. would see United Kingdom reluctance to meet them on free licensing as positively damaging, because they regarded their own proposal as the only way of removing from other N.A.T.O. countries the incentive to do their own research and development in COMSEC. And the continuation of this research and development would, N.S.A. considered, damage the United Kingdom/U.S.A. SIGINT interests.

  Fundamentals were at stake. Meanwhile, they asserted, the cost of supporting the NSA plan was a relatively small amount of compensation to British companies such as Plessey for the free licensing of its new Alvis machine.40 In the event GCHQ won the day, and equipment was supplied to NATO nations at a highly subsidised cost.41

  As an additional backstop, GCHQ and NSA quietly encouraged their NATO allies to introduce legislation that regulated the export of cypher machines in the same way that arms exports were tightly controlled. However, there was an alarming loop-hole. The restrictions did not apply to neutral countries such as Sweden or Switzerland. Switzerland was one of the world’s most advanced manufacturers of a range of technical defence equipment, from optical instruments to radar. It had a long-established record as a manufacturer of excellent cypher machines, the reputation of which was enhanced by public confidence in Swiss neutrality. This ‘neutrals problem’ was being energetically addressed by GCHQ and NSA. The main issue was a well-known Swiss company called Crypto AG, which was owned by a Swede, Boris Hagelin Senior, one of the world’s top experts on cypher machines and a significant innovator in the field of cryptography. Crypto AG had become an increasingly important supplier of cypher equipment after the Second World War, trading on the Swiss reputation for excellent technology and neutrality.

  How NSA and GCHQ ‘neutralised the neutrals’ is an extraordinary tale. The story seems to have begun on 4 September 1956, when the Australian Embassy in Washington ‘had a ring from a character who wanted to sell us some cipher machines’.42 The ‘character’ was none other than Boris Hagelin Junior, who represented two of his father’s companies – Crypto AG in Switzerland and AB Cryptoteknik in Sweden – and ran a sales office in Maryland. He boasted that, together, the two companies had manufactured over 150,000 cypher machines which were in service with over thirty countries.43

  Hagelin Junior was busy moving around Washington selling cypher machines to the many foreign embassies there. He persisted with his pitch to the Australians, supplying copies of Crypto AG technical literature which the Embassy sent back to Canberra. There it was passed to Vic Rolf at Defence Signals Branch, Australia’s sigint agency. In turn, he forwarded it to Fred Stannard and LCSA in London.44 LCSA was very interested, since the literature revealed that Crypto AG was producing some new and alarmingly advanced machines. LCSA was soon relaying more and more detailed questions about the new machines through its Australian proxy, which now posed plausibly as a buyer. Australian officials in Canberra noted: ‘At no stage did we inform our Embassy of the use to which the papers were put.’ Later, the Australian code-breakers purchased some samples but did ‘not wish to become known to any overseas contact as the ultimate addressee’.45 Meanwhile, Hagelin remained exceedingly keen to sell its cypher machines and ‘random generators’ used for making one-time pads.46

  In August 1957 William Friedman, one of NSA’s most experienced code-breakers and a long-standing friend of John Tiltman, was sent on a special mission to Britain. Some have suggested that this was because GCHQ had become suspicious that the Americans had been reading British messages prior to the Suez invasion – something forbidden by the UKUSA agreement – and Friedman was being sent to offer reassurance. But that was not the main purpose of the visit. On 26 August Friedman arrived at GCHQ to be greeted by old friends such as Josh Cooper. For a week he shuttled between GCHQ’s new site at Cheltenham and the offices of LCSA in Palmer Street. The main focus of discussion was collaboration against the West Europeans, and a key issue was the increasingly advanced machines being produced by Crypto AG and AB Cryptoteknik. NSA later admi
tted that the purpose of Friedman’s visit was ensuring that it continued to have ‘the daily information enabling NSA to read NATO countries’ messages’. Friedman’s biographer recounts: ‘Ciphering machines incorporating ingenious variants and improvements were being produced in Europe by more than one manufacturer and were being bought and adapted by more than one NATO country.’ Unsurprisingly, Friedman’s subsequent destinations on his European grand tour were Sweden and Switzerland.47

  Having consulted closely with GCHQ and LCSA, it appears that Friedman came to a private arrangement with a number of companies in ‘neutral’ Sweden and Switzerland. The exact nature of the negotiations between him and Hagelin remains obscure, but this was a historic meeting, since they were two of the twentieth century’s most experienced cypher experts. Certainly by 1975 it appears that NSA was actively involved in subverting the next range of Crypto AG machines. Reportedly, NSA’s approach was to introduce a flaw so that the machines secretly broadcast the cypher key, undermining their own encryption. The alleged result was that NSA could read the encoded messages even faster than the intended recipient. This approach is known as a ‘trap door’, and has also been used more recently in computer encryptions.48

  The secret of NSA’s programme to penetrate much of the traffic carried by ‘neutral’ cypher machines did not begin to leak out until the 1980s.49 Publicly, NSA and GCHQ have never admitted the alleged operation against Crypto AG, but by the summer of 1986 many of the details of this sensitive secret had reached the ears of Congressional committees on Capitol Hill, and it seemed only a matter of time before they appeared on the pages of the New York Times and the Washington Post. General Bill Odom, then the Director of NSA, was furious, and held a personal meeting with Robert Gates, Deputy Director of the CIA, on 6 November 1986. There was only one question on the agenda: ‘Who told Cong[ress] about Cry AG?’50 NSA was also working with the West Germans on this project. In his daily log, Odom referred to a periodic review of the ‘Swiss connection’ in his meetings with George Wieck, the head of the Bundesnachrichtendienst (BND), the West German intelligence service that handled both sigint and human agents.51 Cooperation between NSA and the BND was central to the subversion of the neutral machines.52