IN HIS MARCH 1993 address in Pretoria, F. W. de Klerk told the world that his nation was no longer interested in being a member of the nuclear club. But during the first half of the decade, at test sites in different parts of the world, other nations would demonstrate their continued desire to maintain and improve their nuclear arsenals. Those activities would be followed closely by U.S. nuclear intelligence analysts, who sought to determine exactly what those nations were doing and why.
FRANCE WAS AMONG those countries. During the 1980s the French had used their Mururoa and Fangataufa testing grounds on ninety-two separate occasions. Included were eight tests in 1984 used to validate an improved warhead for the multiple-warhead M4 submarine-launched ballistic missile that would enter service in 1985 on the L’Inflexible, to test an improved warhead for the ASMP air-to-surface missile that would be first deployed on Mirage IVP aircraft in 1986, and to conduct research for a warhead intended for the never-to-be-deployed Hades short-range tactical missile.1
The French testing campaign in the 1980s was also notable for the two nonnuclear explosive devices that tore through the Rainbow Warrior, a ship operated by the Greenpeace environmental group, on July 10, 1985. The environmental group was planning to use it to lead a flotilla of vessels to the vicinity of Mururoa to protest the French testing activities. It would soon be discovered that the bombing, which destroyed the ship and killed Greenpeace photographer Fernando Pereira, was the work of the General Directorate for External Security (DGSE), the French secret service.2
During 1990 and 1991, while South Africa was dismantling its arsenal and signing the nonproliferation treaty, France conducted another twelve tests, nine at Mururoa and three at Fangataufa. While the yields of the tests varied widely, from 10 to 130 kilotons, they did have a common purpose. Each was for evaluating the performance of the miniaturized, hardened TN 75 warhead that was to rest atop the M45 submarine-launched missile, which was to be carried aboard the Triomphant-class submarines beginning in the mid-1990s. The Commissariat l’Energie Atomique had promised that the warhead would be “virtually invisible to radar.”3
A three-year moratorium followed, imposed by French president François Mitterand. But that hiatus ended in 1995, when his successor, Jacques Chirac, assumed power. The CEA as well as nuclear specialists in the Ministry of Defense convinced the new chief of state that a new testing campaign would serve two purposes. It would allow French weapons designers to perfect a new warhead for the M5 missile, scheduled to enter service with the third Triomphant-class submarine around the turn of the century, and to ensure the reliability of the French arsenal once the country signed the comprehensive test ban treaty. The treaty, whose origins went back to the 1950s, would go beyond the prohibition of the 1974 threshold test ban treaty and prohibit all nuclear tests of any yield.4
The moratorium ended with an underground test on September 5, 1995, which Chirac had postponed from August in response to President Bill Clinton’s plea that France not conduct a test while he was in the region celebrating the fiftieth anniversary of the end of World War II in Hawaii. The test took place at 12:30 p.m. at Mururoa, when the nuclear device inside the sixty-five-foot-long canister at the bottom of a shaft about a half mile deep detonated with a force of 8 kilotons, an explosion that was detected in Australia within minutes. French television pictures of the event showed the turquoise lagoon rise several feet, turn a milky white for several seconds, and then settle back into its original state. French nuclear scientists watched the test from deck chairs on the atoll and politely applauded the explosion. French defense chief Charles Millon issued a statement declaring that the testing program was “indispensable to enable us to guarantee the reliability and safety of our nuclear arsenal in the long term.”5
The new series of tests was met with strong protests from governments and activists. While Russia, Great Britain, and the United States had ceased testing between 1990 and 1992, France and China had continued. Japan, Australia, and New Zealand were among the nations that lodged complaints. New Zealand, whose prime minister in 1984 had suggested the French do their testing “somewhere near Strasbourg,” recalled its ambassador from Paris. Australian foreign minister Gareth Evans objected that the test “is not the action of a good international citizen; it is not the action of a good neighbor.” Throughout the region there were consumer boycotts of French wine, cheese, and fashion products. In Papeete the protests were violent. A flotilla of twenty-five protest ships, including the Rainbow Warrior II and Greenpeace, ringed Mururoa. Once again the French interfered, seizing both vessels as they moved toward the exclusion zone, although without killing anyone.6
A second test, at Fangataufa, followed on October 1. The fifth, and last, test of 1995 took place in late December. The protesters would have what they wanted—an end to French testing—on January 27, 1996 when France conducted its final nuclear test, number 210. Approximately eight months later, on September 24, it signed the comprehensive test ban treaty along with the other acknowledged nuclear powers.7
THE FRENCH TESTS in the 1990s did draw the attention of both real and fictional spies. To protest the 1995 round of tests, actor Pierce Brosnan refused to attend the French premiere of the film Goldeneye, his debut appearance as James Bond, leading to the premiere’s cancellation. Meanwhile, on the other side of the world, very real spies were monitoring those nuclear tests.8
When Mitterand announced his moratorium in April 1992, analysts with the New Zealand External Assessments Bureau who followed French testing were shifted to other duties. Once testing resumed in 1995, New Zealand’s Government Communications Security Bureau (GCSB) resumed intercepting French military communications to and from Mururoa. Analysts at the assessments bureau also went back to monitoring French testing activity. They used the signals intercepted by GCSB concerning French military aircraft movements to and from Mururoa, along with other information, to evaluate and predict the times and yields of French tests. Their reports were of interest not only to New Zealand officials, but also to those in the U.S. intelligence community who monitored French testing activity—and the Americans received the product of the New Zealand analysts’ work, as they had in the past.9
The reports from New Zealand about French nuclear activities helped the United States target its more substantial intelligence capabilities on nuclear developments in the South Pacific. In the mid-1990s the National Reconnaissance Office was operating both later versions (launched in late 1987 and 1988) of the original KH-11 satellites, as well as one advanced electro-optical satellite—the Improved Crystal, launched in November 1992, which was equipped with infrared sensors for nighttime imaging. Another Improved Crystal would be in orbit at the time of the final two French tests. All would pass over the French test grounds and could return their high-resolution imagery in real time.10
Were any of the Vela satellites still operational, they would, of course, have been useless in detecting French tests. Other satellite systems—including the Defense Support Program, Defense Meteorological Satellite Program, Global Positioning System, and Satellite Data System satellites—able to detect atmospheric tests were also of no use against underground tests.* Aerial monitoring would be of no value unless nuclear material escaped from the underground test shaft, and it does not appear any attempt was made to collect such debris. But the Air Force Technical Applications Center was still operating seismic stations and underwater arrays that could detect the earth-shaking signals produced by a nuclear weapons test.
By 1995 AFTAC’s seismic network included stations on four continents, although the number had declined with the end of the Cold War. There were stations in North America (Cambridge Bay, Canada; Flin Flon, Canada; Eielson Air Force Base, Alaska; Lajitas, Texas), Asia (Chiang Mai, Thailand; Wonju, Korea), Australia (Alice Springs), and Europe (Belbasi, Turkey). In addition, as part of AFTAC’s response to the September 1979 Vela incident, it engaged the U.S. Geological Survey to establish a set of nine stations distributed across Latin America, Antarctica, and Africa. The first of those stations to become operational was located in South Africa (in February 1993) while the final station to come online was the Argentina station (in November 1994), so all were in operation when the French began their new round of tests.11
The end of the Cold War had also taken its toll on the Navy’s Sound Surveillance System and AFTAC’s collocated stations. But at least some of the arrays that resided in portions of the Pacific Ocean and their associated ground stations remained in operation, and the hydroacoustic signals generated by the French tests at Mururoa and Fangataufa would have been detected by the arrays and been transmitted to the naval facility that recorded their data.12
WHILE THE U.S. intelligence community continued to take an interest in French nuclear testing activities in the 1990s, it did not take nearly as great an interest as it did in earlier years. In the 1960s it had produced a number of national intelligence estimates, the community’s most prestigious product. But in the 1990s, it would not publish a single national, or special national, intelligence estimate on French nuclear weapons programs.13*
In contrast, there was still great interest in China’s continuing efforts to upgrade its nuclear capabilities. China, like France, would sign the comprehensive test ban treaty on September 24, 1996, and cease testing. But, like France, China had conducted nuclear tests throughout the first half of the decade and into 1996, stopping with its forty-fifth test on July 29, 1996, by which time, according to CIA estimates, the Chinese had an arsenal of between two hundred and three hundred warheads.14
From the late 1960s and up to the final Chinese test, the Americans had continued to gather intelligence on China’s nuclear activities. Each new generation of imagery satellite photographed Lop Nur, Baotou, Jiuquan, and the other elements of China’s nuclear weapons establishment, while the National Security Agency continued to monitor Chinese communications for relevant data, and the CIA’s operations directorate tried, and occasionally succeeded in, recruiting human sources—as illustrated by the information obtained in the 1980s on the links between the Pakistani and Chinese programs.15
Through China’s twenty-seventh test, on October 16, 1980—the six-teenth anniversary of its first test—significant intelligence was also obtained from the satellites equipped with nuclear detonation detection sensors as well as aerial collection of the nuclear debris that resulted from the explosions. In 1976, for example, Strategic Air Command U-2Rs conducted Olympic Race debris collection operations following China’s tests of January 23, 1976, and November 17, 1976.16
Thanks to advance warning from AFTAC that a test was imminent—which the organization had probably received as a result of the combined work of the Gambit (KH-8) and Hexagon (KH-9) satellites orbiting the earth during the final months of 1975 and into 1976, and America’s eavesdropping operations—SAC had a U-2R ready at Osan Air Base in Korea for China’s first test of 1976. The plane flew its first sampling mission on January 24, over the Sea of Japan, and gathered debris from the test. Four more missions followed but failed to collect additional debris.17
While the test of January 23 was relatively small, characterized by the CIA as being of low yield, the November 17 test was estimated at 4 megatons, the largest Chinese test since joining the world’s nuclear fraternity. In order to sample the resulting debris from the large nuclear cloud as it slowly moved eastward through the stratosphere, U-2Rs flew from four locations—from Osan as well as air force bases in Alaska (Eielson), California (Beale), and New Hampshire (Pease). The planes logged 143 hours of flying time conducting seventeen sampling missions, six of which yielded debris.18
But China’s October 1980 test would be its last atmospheric test, as the nation joined the other declared nuclear powers in observing the terms of the 1963 partial test ban treaty. During the remainder of the decade China would conduct another seven tests.19 Thus, by the time the 1990s began, the U.S. nuclear intelligence community would have had a decade of experience in monitoring Chinese nuclear testing activity without the benefit of satellites equipped with nuclear detonation gear or the guarantee that there would be atmospheric debris.
While over the years France and Russia had moved the locations of their main test sites, Lop Nur had remained China’s one and only site for nuclear weapons tests. It was also the world’s largest test area, occupying over 62,000 square miles, with over 1,200 miles of highways. About one-fifth of the site was used for testing purposes. The entire facility consisted of three districts. The northwest district included the scientific city of Malan, where test site head-quarters and the residences for scientists, engineers, and technicians were, and still are, located. The southeast district was where China had set off atmospheric tests, while the central district was where underground tests were conducted. The Qinggir region, in the eastern part of the district, was used for vertical shaft tests. Tests were conducted in horizontal tunnels in the Beishan region, to the southwest of Qinggir, and the Nansahn region, to Qinggir’s northwest.20
Gathering the details about the test site—where the shafts were, whether they were horizontal or vertical, and how deep they were—was part of the nuclear test intelligence effort. Information was also needed on the geology of the test site. That some of the vertical shafts had been dug in Carboniferous granite and Upper Paleozoic metasandstone and conglomerate was important. The magnitude of the seismic signals from an underground test is a product not only of the true yield of the device but also the type of rock that the resulting seismic signals pass through on their way to a monitoring station. Therefore, turning the magnitude into an estimated yield required consideration of the site’s geology. Some of the needed information could be found in, or teased out of, the open Chinese literature, such as articles that appeared in 1980 and 1983 in the journal Shuiwendizhi Gongchengdizhi with the titles “Physical Geological Reactions to Underground Nuclear Explosions” and “A Preliminary Study of Abnormal Movement of Groundwater Influenced by an Underground Nuclear Explosion.” But there was also a need to produce classified studies such as the Joint Atomic Energy Intelligence Committee’s 1992 Geology of the Qinggir Underground Nuclear Test Site, China, based partly on secret sources.21
Of course, those in the U.S. intelligence community—whether at the CIA in northern Virginia or at Z Division in northern California—who monitored Chinese nuclear testing activities in the 1990s wanted to know about more than the rocks at Lop Nur. They also wanted information about the people involved in the test program—about their responsibilities, skills, and backgrounds, information that helped in understanding China’s testing activities.
One means of gathering such information involved Chinese nuclear scientists visiting facilities such as Livermore and Los Alamos, an activity that began in the late 1970s. Those scientists, according to a Livermore counterintelligence briefing, “sought close personal relationships with individual Lab employees.” The Chinese delegation that visited Livermore in February 1994 included Hu Renyu, director of the Chinese Academy of Engineering Physics, which ran the nuclear weapons program, along with academy deputy director Hu Side. Also along for the visit was Ye Lirun, the chief engineer at Lop Nur.22
Even more valuable were the U.S. scientists who were permitted to travel in China and visit China’s nuclear weapons facilities, including the test site. The first was Harold Agnew, the veteran of Los Alamos, who had become the institution’s director in 1970. In January 1979, a few months before he was due to retire, Agnew received an invitation to a reception for Deng Xiaoping at Washington’s Ritz-Carlton Hotel. Also attending the gathering was Chen Ning Yang, a prominent Chinese physicist who had studied, along with Agnew, under Enrico Fermi. Yang introduced Agnew to the deputy director of the Second Ministry of Machine Building, K. C. Wang, and suggested that Agnew eat dinner with the official in a back room. During dinner, Wang asked some questions about Agnew’s scientific interests. Two weeks later Agnew received another invitation, which he also accepted—to visit China.23
Shortly after his arrival at Shanghai airport a group of six individuals, who Agnew concluded were weapons scientists, showed up to take charge of him. Their first stop was Beijing, where the scientists wanted Agnew to give an address in the Great Hall of the People. As a result, Agnew found himself behind a podium, with thirty or more Chinese weapons scientists in front of him. Over the following weeks, Agnew’s tour continued, with visits to Beijing’s Forbidden City and a massive bomb shelter underneath the city, with the weapons scientists as his companions.24
Once he returned to New Mexico, Agnew drafted a report that intrigued U.S. intelligence officials, who were hungry for more information about the Chinese bombmakers. In addition to memories, Agnew brought back photographs and a tiny spiral notebook in which his Chinese colleagues had written their names in Mandarin as well as English. The chief of Los Alamos’s International Technology Division (ITD), the lab’s intelligence unit, had Agnew’s report typed up, stamped “Secret,” and sent back to Washington, where it left analysts at the CIA and Defense Intelligence Agency hungry for more.25
Daniel B. Stillman, the head of ITD, hoped to provide more. He encouraged George A. “Jay” Keyworth II, head of the lab’s physics division, to follow in Agnew’s footsteps. Although reluctant, his respect for Agnew, who also urged him to go, led him to agree. With intelligence collection a clear objective, Keyworth was given some pointers by the new CIA liaison to Los Alamos, Robert S. Vrooman. He also received a visit from CIA representatives stationed in Denver, probably from the agency’s Domestic Collection Division, which collected intelligence from U.S. residents who traveled abroad. Key-worth recalled that “they gave me tips, taught me methods—‘Here are some ways you can do things.’ It was just like the spy books, writing things that can’t be seen.”26
In 1980 Keyworth’s plane deposited him in Shanghai, and he soon discovered that his host was Gen. Zhang Aiping, first commander of the Lop Nur test site who had become chairman of the National Defense Science and Technology Commission in 1975. In 1977 he had also become a deputy chief of the General Staff, as well as member of the Chinese Communist Party Central Committee. He was also believed to be a close associate of Deng Xiaoping.27
Keyworth’s quarters were in a presidential guest house in Beijing. To neutralize the bugs he assumed were liberally distributed throughout his rooms, he would turn on the shower to maximum intensity and record each day’s events on microcassettes as he stood in a cloud of steam, using the lone CIA technical device he had taken along. His reports covered what he learned, what his hosts seemed to know, and what they wanted to know. When he slept, the cassettes were in his pillowcase, and when he was awake, he carried them with him. Return trips followed in 1980 and 1981, trips that lasted for weeks and brought back more secretly recorded reports.28
In May 1981 Keyworth’s role as “scientist-spy” ended just before he was about to leave for a visit to Lop Nur—he was informed that he was about to be nominated as President Ronald Reagan’s science adviser. The next year, Agnew volunteered to take Keyworth’s place if the CIA would provide him with a map of China and airfare for his wife. A flight to western China was followed by a drive to test site headquarters at Malan. At night Agnew ate in the home of Gen. Zhang Zhishang, the test site’s commander, and they traded stories of life at the Nevada site and Lop Nur. Agnew’s days as the first American visitor to the test site included watching a movie and examining color photographs of China’s atmospheric tests that no American had seen, as well as a look at the tunnels that had been dug for testing.29
Eight years later, the man who sent Agnew’s first report to Washington, and asked Keyworth to go to China, made his first trip. By 1990 Daniel Stillman had worked at Los Alamos for twenty-five years. He had come to the lab as a specialist in devices used to simulate and measure nuclear detonations. In 1978 he became head of ITD, Los Alamos’s version of Livermore’s Z Division. In June 1988 lunch with five Chinese weapons scientists, including Professor Yang Fujia, the director of the Shanghai Institute of Nuclear Research, at a conference in Albuquerque led to an invitation to visit a number of Chinese facilities. China’s suppression of the Tiananmen Square uprising in June 1989 would delay the trip until spring 1990, when Stillman and his deputy, Terry Hawkins, arrived for their tour. One stop was the Southwest Institute of the Chinese Academy of Engineering Physics at Mianyang, which was scattered in several valleys, often under cloud cover, making satellite photography difficult. But photographs of the facilities could be found in a brochure Hawkins brought back and which he felt lucky to get his hands on—“When I was in DIA doing real intelligence work, I would have given ten million dollars for this book,” he observed.30
Stillman would make nine visits through the summer of 1999, continuing to tour Chinese nuclear facilities and equipped with video recorders and cameras—first as representative of Los Alamos and then, after his October 1993 retirement, as a private citizen, sometimes escorting Los Alamos officials. In a talk at MIT in 2001, he recalled, “I visited virtually all of China’s nuclear weapons laboratories.” In Shanghai, he visited Fudan University and the Shanghai Institute of Nuclear Research, where scientists worked on neutron initiators and sources. In Mianyang, near Chengdu, he toured the headquarters of the Chinese Academy of Engineering Physics, “China’s equivalent to our Los Alamos, Sandia, and Lawrence Livermore nuclear laboratories.”31
Stillman also visited the Northwest Institute of Nuclear Technology, which designed and produced diagnostic equipment to monitor nuclear weapons tests. It also assembled the instrumentation trailers used for each test as well as conducted radiochemical analysis after the test to determine the yield of the detonation. Lop Nur was also on his itinerary. There he toured several vertical-shaft test sites and was able to walk in a tunnel that had been used for a horizontal test. He was also told that China’s first seven tests had employed uranium consisting of 93.5 percent U-235.32
The visits were followed by reports detailing where he went, who he saw, and what they said. The written record produced by Stillman’s visits included the names of more than two thousand Chinese scientists working at the nuclear weapons facilities, detailed histories of the Chinese program from senior scientists, descriptions of his inspections of nuclear weapons labs and Lop Nur, and reports of interviews with Chinese weapons designers. There were also photographs of the nuclear facilities. All that was passed on to debriefers from the U.S. intelligence community who came to visit him after each trip.33
Vrooman observed that “Danny’s approach was disarmingly simple: You just go to China, find the guys who designed the bombs and ask them questions.” According to Robert Daniel, a former CIA officer and congressman who headed the Energy Department’s intelligence office in 1991 and traveled to China with Stillman in October 1990 and in 1991, “We saw things no outsider had ever seen before . . . we went to the test site . . . and saw them getting ready to place a device down a 600-meter hole.” George Keyworth concluded that “the whole activity that he was involved in was extraordinarily successful for the United States.” Stillman’s assessment was that “the information the Chinese scientists willingly gave to me and my fellow travelers would have cost the government several millions of dollars to collect by traditional methods.” As for his trip to Lop Nur, he observed that “more Americans have walked on the Moon than on China’s nuclear weapons test site.”34
NOT SURPRISINGLY there were critics of the entire exchange and contact effort, which included not only the high profile visits of Agnew, Keyworth, and Stillman, but groups of Chinese and American scientists visiting each others’ facilities and attending conferences—after which many of the U.S. scientists would be debriefed by CIA representatives. The skeptics believed that whatever increased understanding the United States obtained about the state of the Chinese nuclear weapons program was outweighed by what Chinese scientists would learn about ways to improve China’s nuclear capabilities.35
Whether the benefits outweighed the costs may have been a matter of debate, but it was certainly clear that the CIA and other U.S. intelligence organizations watching China’s nuclear activities benefited from the information obtained. That information, however, did not eliminate the need to spy on China’s nuclear program using imagery, signals intelligence, seismic detection, and spies as China proceeded with its nuclear test program.
Testing activity at the site in the 1990s would begin on July 26, 1990, and August 16, 1990, with yields of at least 15 and 50 kilotons, respectively. The CIA’s Office of Scientific and Weapons Research located the July test at the Qinggir Underground Test Site East and noted that “the current test may be related to development of a warhead for a Chinese short-range ballistic missile.” The same analysis also speculated that the upcoming August test was being conducted for the same reason.36
It would be almost two years before China would test again. In early January 1992 an article in the CIA’s Science and Weapons Review estimated that China would conduct two nuclear tests that year, with at least one taking place in May. One possible source for the information in the report, as well as other information on the Chinese nuclear program, was Hua Tianqiang, a senior engineer and director of the intelligence unit at the nuclear research institute of the Shanghai Academy of Sciences who had access to a variety of secrets about that program. Hua disappeared in mid-October while touring Emei Mountain in Sichuan Province, an area full of crags, caves, and forests. In the following weeks groups searched the region and came up empty. After three months Chinese authorities began to suspect that either he had been killed in an accident or by a wild animal, or he had been smuggled out of the country by a foreign intelligence service.37
In any case, U.S. imagery satellites detected preparations for a test weeks ahead of time, and on May 21 China detonated a nuclear device, with a yield estimated to be between 700 kilotons and 1.8 megatons. The second and last test of the year followed on September 25, with a far smaller yield of 15 kilotons. The approximately two hundred recipients of the elite, top-secret National Intelligence Daily read about the test in the May 22 edition. The following day the unclassified Washington Times revealed some of what U.S. officials told reporter Bill Gertz. The “huge underground blast,” the paper reported, was a test of a warhead for a new intercontinental ballistic missile under development. U.S. intelligence agencies were also expecting that radioactive gases from the explosion would be vented into the atmosphere and spread to areas outside of Chinese territory, despite the test having taken place at a depth of more than three thousand feet.38
Intelligence coverage of the event and its aftermath included detection at AFTAC seismic stations and satellite images obtained by KH-11 and advanced KH-11 satellites as they passed overhead, images that showed a cave-in around the center of the blast. And, as expected, despite the blast having occurred far underground, a radioactive cloud passed over the Sea of Japan. A WC-135 from a base in Japan flew through the cloud to monitor the level of radioactivity and gather whatever debris it could.39
In August 1992 the May 21 test would be the subject of an article in the CIA’s Science and Weapons Review, which examined the reported claim of Chinese deputy premier Deng Xiaoping, carried in a Hong Kong newspaper, that the test was actually a multiple-device test. The CIA article noted that if it was a multiple-event test, it would have been the first known one of that type in the Chinese program, that obtaining high-quality diagnostic data from multiple devices is the major problem in conducting multiple tests in a single test shaft, and it was unclear why the Chinese would wish to test two devices with identical yield given the cost and complexity of testing more than one device in a single shaft.40
As had been predicted, a second test followed that year, on September 25. Another seven would follow that one. The sole test of 1993, on October 5, would be followed by two tests in 1994 (on June 10 and October 7), two in 1995 (on May 15 and August 17), and the final two in 1996 (on June 8 and July 29). Yields of the tests were estimated, in open sources, as ranging from 1 to 5 kilotons for the July 29, 1996, test to 40 to 150 kilotons for the October 1994 and May 1995 tests.41
Intelligence gathering directed at the Chinese program would continue to provide notice of impending tests. By mid-September 1993, if not earlier, the U.S. intelligence community had concluded that another Chinese test would occur in the near future, based on satellite imagery of the Lop Nur site, which indicated that Chinese engineers had recently lowered a device down a deep shaft, among the final preparations for a test. And in May 1996 undersecretary of defense Walter Slocombe told reporters that preparations for a test were underway at the Lop Nur test site, and that “what we see them preparing to do is to conduct a nuclear test.” Slocombe’s remarks were based on the same intelligence that allowed the authors of an article in the May 31 issue of the CIA’s Proliferation Digest to note that two of the four test sites at Lop Nur were in late-stage preparations, one was in early-stage preparations, and one was in the site-preparation stage.42
U.S. intelligence gathering and analysis allowed not only short-term alerts based on observed activities, but also longer-term projections of Chinese testing and explanations of the objectives associated with the tests. The February 19, 1993, issue of the National Intelligence Daily noted that “China has accelerated its nuclear test schedule and plans to conduct seven tests by 1996” devoted to strategic and, possibly, tactical systems development, and speculated that the speedup in its schedule might be the result of “growing international pressure for a comprehensive test ban in 1996.” The October 1994 edition of Proliferation Digest explained that the planned tests were part of China’s nuclear weapons modernization program, which included the development of warheads for new intercontinental and submarine-launched ballistic missiles as well as technologies to enhance confidence in the reliability of warheads that would be part of China’s stockpile after it signed the comprehensive test ban.43
The same classified publications that contained long-term projections of Chinese activity also provided updates on China’s test plans. The August 29, 1994, edition of the National Intelligence Daily reported that China had planned a test early in the month but postponed it for six weeks. That September, readers of the NID learned that the director of Chinese nuclear weapons research had indicated that China would continue testing through 1996, with a minimum number of tests, unless some failed. The next month, Proliferation Digest reported that the Chinese testing program was facing delays that could set back China’s adherence to the comprehensive test ban treaty. In March 1995 the Digest and NID contained items noting that China appeared to have taken steps to put its nuclear testing program back on schedule. In April 1996, an NID informed its readers that “China is accelerating its nuclear testing preparations at Lop Nur.”44*
Of course, the Chinese program was far from transparent and the information available to CIA analysts was not sufficient to leave them confident that they always understood what the Chinese were doing in the remote western part of China. A September 1994 joint CIA-DIA memorandum observed that although their analysis indicated that China had several modernization goals, “the specific purpose of each Chinese test is unknown.” A November 1995 article in Proliferation Digest illustrated that point. It noted that “an underground event—probably a nuclear test—occurred between 4 and 6 September at China’s nuclear test site.” While three probable nuclear testing scenarios had been evaluated—test of a nuclear artillery shell, a safety test, or a hydronuclear experiment—none was considered “completely consistent with all of the evidence.”45
WHILE THE CHINESE ceased testing in July 1996, a controversy over the significance of one test emerged in 1995 and would continue until the end of the decade and beyond. The controversy was not contained within the highly classified world of the nuclear intelligence establishment, but would result in a very public trial for a Los Alamos scientist, Department of Justice and intelligence community investigations, congressional inquiries, and, of course, books and articles.46
It began with China’s nuclear test of September 25, 1992. Like the test that May, the United States collected a variety of data on preparations for the test and the test itself, using imagery satellites, the seismic stations spread across the planet, and a WC-135 to intercept the radioactive gases that were vented into the atmosphere and passed over the Sea of Japan. The test, which took place in a tunnel at the Qinggir site, was estimated to have generated the equivalent of about 10,000 tons of TNT, a small fraction of the energy released by the test in May.47
But it was not until almost three years later that the intelligence gathered concerning the test began to have a dramatic impact. During the summer of 1995 Dan Bruno, the Energy Department’s chief counterintelligence investigator, was called into the office of Notra Trulock III, the head of the department’s Office of Energy Intelligence, established in 1990 as the Office of Intelligence by merging the department’s foreign intelligence, threat assessment, and counterintelligence offices.48
Trulock had started his intelligence career as an eavesdropper. Graduation from Indiana University, with a degree in political science, was followed by a stint with the Army Security Agency beginning in 1971. After attending the Defense Language Institute in Monterey, California, to learn Russian, he headed for West Germany to listen to the radio traffic transmitted by Soviet troops across the border. When he returned to the United States, he put in several years at NSA, starting in 1975, where he became a specialist in the analysis of command and control systems. That job was followed by positions in the contract research world and at the National Defense University and, starting in 1990, at the Los Alamos Center for National Security Studies. In October 1993 he returned to Washington to join that what was then the Energy Department’s Office of Intelligence. In May 1994 he became director of the renamed office.49
Trulock’s message to Bruno was a simple one: “The Chinese have stolen the design of the W88,” the country’s most “highly optimized” warhead. Each W88 warhead, eight of which reside on top of each of the twenty-four Trident II (D5) submarine-launched ballistic missiles carried by Ohio-class submarines, would detonate with a force of 455 kilotons.50 Chinese acquisition of W88 technology would certainly increase their ability to produce smaller, but more deadly missiles.
The basis for Trulock’s statement came from work done at Los Alamos. It began with Bobby Henson, a former hydrogen bomb designer assigned to the laboratory’s intelligence division who had considerable experience in analyzing the results of Chinese atomic tests. He had begun in 1967 when he was asked to study China’s initial test of October 1964. From there he went on to examine the data on each of China’s nuclear tests. The data from September 25, 1992, and later tests would lead him to a disturbing conclusion.51
The seismic signals from those tests indicated smaller yields, which Henson believed meant China was learning to build smaller primaries—the atomic bombs used to set off the thermonuclear fuel. Smaller and therefore lighter primaries opened up a number of possibilities for China’s strategic force—more warheads per missile, missiles with more mobility, missiles with longer ranges—none of which were good for the United States. That could mean a Chinese missile force that was harder to monitor, more likely to survive a U.S. attack, and capable of delivering greater death and destruction to the United States.52
The intelligence about the September 25, 1992, test included the fact that Chinese weapons scientists, for the first time since 1980, used a horizontal tunnel and ran a far larger set of fiber-optic cables out of the entrance. The implication was that the test device’s performance had been monitored by an exceptionally large number of sensors, an indication that something unusual was taking place. To Henson that meant the Chinese were capable of producing smaller H-bombs, that, as he would remark, “they quit driving a Model T and started driving a Cadillac.”53
Henson then began to wonder how the Chinese had made such a dramatic advance, whether they had developed the Cadillac themselves or simply stolen it from someone’s driveway. Given that he was less than enamored with the contacts between U.S. and Chinese weapons scientists, believing them to be a security risk, and doubted the Chinese scientists’ ability to make such a rapid advance, he was open to the possibility that espionage was involved. Then, in the summer of 1994 he attended a lecture given by Chinese theoretical physicist Sun Chen Wei, an explosives expert who worked on nuclear primaries. During his lecture Sun mentioned how Chinese scientists had relied on ball-shaped primaries for decades, but that in the last few years “we’ve just been working with these watermelons,” using his hands to illustrate his point.54
Sun’s remarks startled Henson because he had just disclosed what was officially highly classified nuclear weapons design data in the United States—the watermelon configuration reduced the size and number (to two) of explosives used to trigger nuclear blasts. Although outside experts understood that some primaries of U.S. nuclear weapons were spherically shaped, the fact that China’s primaries were similarly shaped alarmed Henson and suggested that his conclusions from studying the Chinese test data had some substance. Then, in January 1995 he was informed that a classified cable was awaiting his attention. When he read it, he discovered that a Chinese nuclear expert, recruited by U.S. intelligence years earlier, had revealed that the device tested in September 1992 was miniaturized and used a hollow plutonium, watermelon-shaped core. The cable also explained that the primary was an oblong plutonium shell wrapped in high explosives and about as wide as a soccer ball. A follow-up test, which used a similar core surrounded by insensitive high explosives, also detonated as expected.55
It had taken America’s weapons scientists over a dozen tests to make really small primaries behave as desired. To also succeed with the insensitive high explosives on the first try, without help from a source in the United States, was not possible, in Henson’s view. He went to see another Los Alamos intelligence analyst, Larry Booth. The result was an April 25 coauthored Top Secret Codeword memo and an appointment to see Trulock, who listened to what they had to say and decided to ask for a second opinion.56
The person Trulock consulted was John Richter, a Los Alamos physicist, an expert on primaries, and the chief designer for forty nuclear test devices. In May 1995 he traveled to Washington and read the same cable that had excited Henson and also examined the data from the September 25, 1992, test. He concluded that a primary had been tested and its width (nine inches, similar to that of a soccer ball) was pretty close to that of the W88. Before the end of the month, Richter joined Henson and Booth in producing a new memo for Trulock. It told the Energy Department’s intelligence chief that the Chinese had probably tested the primary for the W88, and that they had likely acquired the necessary design information from a spy.57
That China might have tried to obtain data that would help improve its nuclear capabilities by any means possible—from assiduous collection of unclassified (open-source) material through espionage—was hardly a shock. In 1984 a DIA “estimative brief” noted that “qualitative improvements that the Chinese are developing for their nuclear warheads will depend on the benefits that [the] Chinese are now deriving from both overt contact with U.S. scientists and technology, and the covert acquisition of U.S. technology.”58
The Henson-Booth-Richter memo led Trulock to call in Bruno and inform officials in his department as well as the FBI of “potential espionage involving nuclear weapons data.” Bruno’s suggestion that a scientific working group be established to “assist in the development of a logical investigative effort” led to the creation of the Kindred Spirit Analysis Group, “Kindred Spirit” being the designation for the investigation of possible espionage. Appointed to head the group was Michael Henderson, a longtime primary designer at Los Alamos. The panel also included, in addition to Richter, Booth, and Henson, additional Los Alamos scientists, and representatives from Sandia, Livermore, the CIA (a nuclear physicist), and DIA (a senior nuclear intelligence analyst).59
By late July some of the weapons designers and CIA analysts were unconvinced that espionage was the only reasonable explanation, suggesting that indigenous development, possibly combined with some Russian help, might explain the advance. A majority, including some representatives of Los Alamos, were leaning toward dismissing the notion that espionage was the cause for China’s new capabilities. But before the issue could be settled, new data became available.60
That information had first arrived earlier in the year in the arms of a middle-aged Chinese man claiming to be a missile expert. He was what was known as a “walk-in,” someone who brought documents or volunteered to spy (or both) rather than being targeted and recruited by an intelligence service. Indeed, almost all the success that either the CIA or KGB had during the Cold War in penetrating the other’s national security organizations (other than through technical means) was due to walk-ins, including Oleg Penkovskiy and Aldrich Ames. The initial beneficiary of this walk-in was Taiwan’s internal security service.61
Among the documents he delivered was a twenty-page memo, dated 1988, prepared for the China’s First Ministry of Machine Building, which employed missile designers and builders. A five-year strategic plan for China’s future missile forces, it described in text, diagrams, and graphs the characteristics of both Chinese and American weapons. There were descriptions and hand-drawn sketches of a variety of U.S. strategic warheads for the Trident, the MX and Minuteman intercontinental ballistic missiles, and the cruise missile. The document noted five key attributes of the warhead and accurately described the shape of the primary as well as the width of the casing surrounding it to within a millimeter (four hundredths of an inch), a description Trulock characterized as “pretty damn accurate.” Taiwan passed both the defector and the documents to the CIA station in the American Institute in Taiwan, which had served as the unofficial U.S. embassy since President Jimmy Carter had recognized the People’s Republic of China (PRC) as China’s “legitimate” government.62
The walk-in made frequent trips back to China and returned with more documents—over seven hundred, totaling thirteen thousand pages, before he was through. A CIA translation team was flown to Taiwan to begin working on them. A CIA polygrapher worked on the walk-in, and found his answer to whether he was operating on behalf of a foreign intelligence agency deceptive. He was then flown back to the United States so the CIA and FBI could try to determine whether the entire operation had been managed by China’s intelligence services.63
Theories as to why Beijing might have sent the documents varied. The goal might have been to intimidate Taiwan, or discourage the United States from defending Taiwan should China attack, to misinform the CIA, or divert attention from a more valuable agent. It was also possible that somebody had just made a mistake. Richter suggested that China might have wanted to know if the information it had gathered on U.S. weapons was correct, and was hoping to judge so by the American reaction to the walk-in’s documents. What was certain was that the documents contained a great deal of classified information, including the description of the W88.64
By this time Henson’s conclusions were no longer driving the Kindred Spirit investigation. While seismic signals and satellite images generated initial suspicion, the words and gestures of a Chinese weapons designer and classified Chinese documents revealing secret details of the most advanced U.S. warheads provided the basis for continuing investigations. The alleged espionage became part of a broader congressional investigation into Chinese acquisition of U.S. technology led by California congressman Christopher Cox. The investigation concluded that “the PRC stole classified information on every currently developed U.S. intercontinental ballistic missile (ICBM) and submarine-launched ballistic missile (SLBM),” a conclusion that did not go unchallenged. A Los Alamos scientist, Wen Ho Lee, would come to be suspected of providing China with classified data on U.S. nuclear weapons systems. In the end he would experience a lengthy pretrial detention and lose his job, but regain his liberty. While it was established that he created a private collection of the computer programs used to design U.S. nuclear weapons and left them vulnerable to outside hackers, including those who might be in the pay of foreign intelligence services, and had put the codes on portable cassette tapes, which he claimed to have destroyed, there was no definitive proof of espionage. The case would end with almost everyone involved tarnished, whether justly or not, in one way or another.65
In March 1999 director of central intelligence George J. Tenet selected Robert Walpole, the national intelligence officer for strategic and nuclear programs, to lead an interagency group that would assess possible damage to U.S. national security from any disclosure of classified nuclear weapons information to China. Walpole, who joined the CIA in 1978 as an imagery analyst and spent five years as the deputy director of the Nonproliferation Center, would lead a team of intelligence analysts from across the intelligence community and the national labs. Tenet also appointed an outside review group, headed by Adm. David Jeremiah, former vice chairman of the Joint Chiefs of Staff, to head an outside panel that would review the Walpole group’s report and write one of its own.66
The next month Jeremiah’s panel—whose members also included Richard Kerr, a former deputy director of central intelligence, Gen. Brent Scowcroft, national security adviser in two administrations, and John Foster, the head of defense research and engineering in the Nixon administration—issued their report, which according to a statement from Tenet agreed with those produced by Walpole’s group. The outside panel concluded that China had obtained via espionage classified U.S. nuclear weapons information that probably accelerated its nuclear modernization program, had obtained at least basic design information on the W88, and had acquired information on a variety of U.S. weapons design concepts, including those of the neutron bomb. The group also concluded that “China’s technical advances have been made on the basis of classified and unclassified information derived from espionage, contact with US and other countries’ scientists, conferences and publications, unauthorized media disclosures, declassified US weapons information, and Chinese indigenous development. The relative contribution of each cannot be determined.”67
WHILE CHINA had been the last of the acknowledged nuclear powers to halt testing, the Soviet Union had been the first. But in 1997 some in the U.S. nuclear intelligence community suspected that Russia might be conducting low-yield nuclear tests. Although the comprehensive test ban treaty had not come into force, Russia had pledged to refrain from further testing. If it were conducting secret tests, in violation of that pledge, that was something the United States and its political leadership needed to know.
Not surprisingly, during the Cold War there had been suspicions that the Soviets might be cheating on one or both of the nuclear testing treaties as well as other arms control agreements. In late 1979 the DIA had raised the possibility that the Vela detection that September might have been due to a Soviet test, in violation of the 1963 partial test ban treaty. During Ronald Reagan’s presidency there were concerns about Soviet compliance with the antiballistic missile treaty, the strategic arms limitation treaty, and the agreement concerning biological weapons. And during the middle of the Reagan years the Soviet Union’s possible failure to comply with the other nuclear testing treaty—the 1974 threshold test ban treaty, which restricted weapons tests to those with a yield no greater than 150 kilotons—was a serious matter.68
Between 1976, when the Soviets pledged to abide by the still unratified treaty, to the end of 1984, the Soviets had conducted 128 underground weapons tests. The tests in the Semipalatinsk region took place in three distinct areas: The eastern and central testing areas were commonly known as the Shagan River and Degelen Mountain test sites. A third site, to the west, was referred to in the U.S. intelligence community as the Konystan testing area, due to its proximity to the nearby town. There had been no tests at Konystan site since 1980.69
By 1985 analysis of the seismic signals detected by AFTAC’s nineteen detachments—which included manned sites at Alice Springs, Australia; Lakenheath, United Kingdom; Torrejon, Spain; Clark Air Force Base, Philippines; Crete, Greece; Okinawa and Misawa, Japan—seemed to indicate that a number of tests had exceeded the 150-kiloton limit, in some cases significantly. Concern that the Soviets were not living up to their pledge of treaty compliance was noted in a secret national security decision directive signed by President Ronald Reagan on January 14, 1984. Soviet Noncompliance with Arms Control Agreements reported that while “the available evidence is ambiguous and we have been unable to reach a definite conclusion, this evidence indicates that Soviet nuclear testing activities for a number of tests constitutes a likely violation of legal obligations under the Threshold Test Ban Treaty of 1974 . . . which banned underground nuclear tests with yield exceeding 150 kilotons.”70
The same judgment conveyed in the 1984 directive was repeated in a publically released February 1, 1985, report on arms control compliance, as well as two identically titled directives, signed by Reagan in February and December 1985. The December directive, NSDD 202, also elaborated on the reason for concern: “If the yields of Soviet nuclear tests have been substantially above 150 kilotons, then Soviet testing would allow proportionately greater gains in nuclear weapons development than the U.S. could achieve.” The directive also observed that violations, even if of little apparent military significance, could become “precedents for future, more threatening violations.”71
That the nuclear intelligence analysts at the CIA and other intelligence organizations were forced to rely on ambiguous evidence indicated that the agency was not operating a human source who could provide a definitive answer, although the United States had received some human intelligence from a source with knowledge of activities at the Degelen Mountain test area in recent years. Nor, clearly, had NSA’s eavesdropping efforts settled the issue. And while the National Reconnaissance Office’s imagery satellites could monitor pretest activities on the surface and some of the surface effects of an underground test, the images could not, in the absence of other information, reveal the force with which Soviet bombs exploded beneath the earth. And while AFTAC had no problems collecting the seismic signals generated by the Soviet tests, estimating the yield of the tests from those signals was not a simple matter—and became a topic of contention within the intelligence community.72
The fundamental problem was the same one that had led the United States to gather information about the geological structure of Lop Nur. While the magnitude of the “body waves” generated by an underground nuclear explosion, which pass through the earth’s mantle and core, can be transformed into an estimate of the explosion’s yield through a simple mathematical formula, the result needs to be adjusted. The adjustment, a “discount factor,” takes account of the specific geological structure at the test site, because it influences the magnitude of the waves generated by an explosion. In contrast to some of the rocks beneath the Nevada site, which were believed to be partly molten, the Soviet test site at Semipalatinsk was geologically older and more stable. As a result, the magnitude of the body waves generated by a blast of any given yield would be stronger when the blast took place at Semipalatinsk.73
Because the treaty had not been ratified, neither the Soviet Union nor the United States had exchanged information on the geological coordinates of the boundaries of each test site, the geology of the testing areas, the geographic coordinates of underground tests, and the yield, date, time, depth, and coordinates for two tests from each geophysically distinct testing area where tests had or were going to be conducted—as called for in the treaty protocol. By the early 1980s some unclassified Soviet publications that concerned the geology of the Shagan River test area were available, but they did not provide the definitive information required by U.S. nuclear intelligence analysts.74
In 1985, in the absence of definitive data, CIA and Energy Department experts argued that the discount factor then being used, 30 percent, was too low, and contended that too much emphasis had been placed on the body waves and that more weight should be given to the “surface waves” that travel through the earth’s upper crust—an argument also made by outside seismic experts who followed the controversy such as Columbia University’s Lynn Sykes and the U.S. Geological Survey’s Jack Evernden.75
In October 1985 the Defense Advanced Research Projects Agency (DARPA) Seismic Review Panel completed an analysis which concluded that the method being used to estimate the yield of the Soviet tests was, as had been charged, based on erroneous assumptions. That information was then forwarded to the JAEIC. It was also supported by another report, one commissioned by AFTAC itself. In mid-December the committee echoed the reports’ recommendation, that the intelligence community adopt the defense research agency panel’s suggestion that the discount factor be increased to a level that would reduce the estimated yield of Soviet tests by 20 percent.76
The recommendation was not without opposition, including the Defense Intelligence Agency as well as assistant secretary of defense Richard Perle, who feared that a recalculation would become public and would be used by the Soviet Union to build and test more powerful weapons. The recommendation was pending before DCI William J. Casey when NSDD 202 was issued. On January 21, 1986, a little over a month after the directive was signed, Casey, despite the opposition, approved the recommendation.77
Casey’s approval did not automatically end the controversy. Some believed that a number of tests would still be evaluated as being over 150 kilotons. In February 1987 Reagan signed another directive on Soviet noncompliance, which noted an ongoing review of methodologies for estimating Soviet nuclear test yields. The new directive stated that until the review was completed, the conclusion of the December 1985 directive, that a number of Soviet tests constituted likely violations of the threshold treaty, “stands.”78
The issue of Soviet compliance with the threshold test ban treaty would fade away, as the U.S.-Soviet relations took a dramatic turn—first with the assumption of power by Mikhail Gorbachev and then, after the failed coup of August 1991, with the collapse of the Soviet Union and Boris Yeltsin’s rise to power as president of Russia. But in 1996 there was concern within the U.S. national security establishment about Russian compliance with another testing treaty that the Soviet successor state had pledged to abide by—the comprehensive test ban treaty.
This time the troubling seismic signals were not coming from Semipalatinsk. In early 1990 Col. Gen. Vladimir Gerasimov informed the Supreme Soviet that nuclear testing would end at Semipalatinsk by 1993, and any further testing would be conducted at Novaya Zemlya, where 130 nuclear tests had been conducted between 1954 and 1989.79 Even without such a plan, the collapse of the Soviet Union would have necessitated a change in the main test site. On Christmas Day 1991, when the Soviet Union’s existence ended, Semipalatinsk became part of, not the Russian-successor state, but the new nation of Kazakhstan.
Far to the north, and within the Russian successor state, the Central Test Site, as Novaya Zemlya was designated, had been closed down following the collapse of the Soviet Union. Then, in 1992 it was reopened to allow Russian leaders to order a resumption in testing. In March 1996 it was reported that, according to Clinton administration officials, “U.S. intelligence agencies suspect Russia secretly set off an underground nuclear test this year.” The information that led to such suspicions included both seismic data and satellite imagery. According to one U.S. official, “There was some activity you would expect to see that is associated with a nuclear test,” possibly a reference to drilling of a large hole or laying of cables. The same official also noted, however, that the data was “inconclusive.”80
Another anonymous official reported that a number of Pentagon officials had few doubts and believed Russia had tested a small nuclear weapon. “There’s no question it was a nuclear test,” the official told the Washington Times, “the only question is the yield.” The official went on to tell the paper that “it was a low-yield test in mid-January,” and that intelligence reports located the test on the northern island of the test site.81
But despite the certainty of some officials, the case was less clear to others. Seismologists in Europe looking for evidence of a Russian test had not detected any underground blasts. Back in the United States, one senior official who was willing to be named, secretary of defense William J. Perry, told a congressional committee that “there is some evidence on the subject, there’s also some ambiguity in the evidence.” Sometime after Perry’s statement, it was concluded that the seismic disturbance had been caused by an earthquake rather than an explosion.82
But in the summer of 1997 there would be new concern and controversy over whether the Russians had been fully complying with the terms of the now signed, but still unratified comprehensive test ban treaty. The catalyst was a seismic signal coming from the vicinity of Novaya Zemlya on August 16, which registered at 3.2 on the Richter scale—consistent with a very small nuclear blast of between 0.1 and 1.0 kiloton, which might indicate scaled-down tests of a warhead primary. The signal was first detected, not by one of AFTAC’s sites, but by a station in Russia, at Norilsk, operated by the Russian Defense Ministry. That station had been designated as one of the 320 monitoring sites that would gather the data used to monitor compliance with the comprehensive test ban treaty.83
The signals from Norilsk were automatically relayed to the International Data Center in Arlington, Virginia, where signals from each of the monitoring network’s stations were analyzed. Within minutes of the Norilsk signal’s arrival, additional data was received from two monitoring sites in Norway, one in Finland, and another in Sweden. Together they indicated that the event that generated the signal occurred at 5:00 a.m. Novaya Zemlya time, a time consistent with past Russian nuclear tests.84
After the data center informed AFTAC of the event, on August 18 the U.S. nuclear detection organization passed on the information to the CIA. Some of the CIA’s analysts were alarmed, in part because some of the signals recorded by seismometers looked like those from previous nuclear tests at the site. Satellite images also appeared consistent with a test. The constellation of advanced KH-11 satellites passing overhead had returned imagery over the previous months showing Russian scientists unusually active at the test site. Helicopters were flying technicians around the site. On the ground, those technicians were lowering equipment, plugging test holes, and stringing cables for diagnostic equipment. Activities photographed on August 14 and 16 were, according to one intelligence analyst, “a dead ringer for [those in] test shots” conducted during the previous ten years and prompted AFTAC to order a WC-135 to fly downwind of the test site on the first of those dates. Several weeks earlier there had also been a visit from Viktor Mikhailov, Russia’s atomic energy minister.85
To further study the data, the CIA called a meeting of the Nuclear Test Intelligence Subcommittee, a component of the JAEIC, which included representatives from the key agencies involved in analyzing test-related intelligence. Signals from the stations in Russia and Finland were thrown away because their sensors were not properly calibrated. The subcommittee then concluded that while the seismic event might have originated at the test site, its origin was probably at sea.86
On August 16 the International Data Center reached a more definitive conclusion based on computer analysis of the signals from the sensors or arrays in Norway, Sweden, Finland, and Russia, pinpointing the location of the event as being more than 60 miles from the test site. Further analysis completed by August 18 confirmed that conclusion.87
The nuclear test intelligence group may have been more open to the explanation that a test had occurred for a number of reasons: their access to the highly classified images showing activity at the test site, concern about the validity of the data obtained by the sensors in Finland and Russia, as well as their responsibilities. But the message that the JAEIC subcommittee conveyed to policymakers was even more categorical than the conclusions of the subcommittee. An initial alert message, dated August 18, erroneously described a probable nuclear test as having taken place at the test site, without any qualification, and the coordinates from a previous test on Novaya Zemlya were given as those for the event instead of the most probable coordinates, which corresponded to a location at sea about forty miles from the test facility. “We were trying to be very, very careful,” one official told the Washington Post. A policymaker who read the alert message said it conveyed “very high confidence that it was explosive . . . and right at Novaya Zemlya.”88
Based on the alert, officials at the National Security Council called an interagency meeting on August 20 and ordered that an extensive effort be made to get an explanation from Moscow. The Russian ambassador was called to the State Department to hear of U.S. concerns, while the senior American diplomat in Russia issued a similar statement at the Foreign Ministry in Moscow. In response, Moscow would claim that the seismic signals had been caused by an underwater earthquake, and that activity at the test site was related to preparations for treaty-sanctioned “sub-critical” nuclear tests involving chemical explosions that blast apart fissile material without producing a chain reaction.89
The CIA, in Notra Trulock’s judgment, “jumped the gun” and acted as if it wanted the event to be a test. In particular, according to Trulock, CIA seismologist Larry Turnbull, one in a long line of intelligence and defense officials who had worried about “evasive” nuclear testing by foreign countries, and Nonproliferation Center chief John Lauder “drove the intelligence process harder than it should have been pushed.” There was a lot of talk of tunnels, branching off under the ocean bed, to permit covert testing. An underwater tunnel would have to have been well under the ocean, according to former Arms Control and Disarmament Agency deputy director Spurgeon Keeny, and would have required an “impossible engineering feat.” Keeny considered that the whole episode was a “shocking example of the rigidity of the intelligence community, particularly CIA.”90
Near the end of the month, U.S. officials confirmed their concern about a possible Russian test. On August 27 the NSC released a statement which read, “We do have information that a seismic event with explosive characteristics occurred in the vicinity of Novaya Zemlya.” Ralph Alewine, director of the Pentagon’s nuclear treaty office, added that “the information is still under review, and we are discussing this with other countries including Russia.” An anonymous Pentagon official noted that the seismic signal created “very sharp” waves on the detection equipment, waves not usually associated with an earthquake. From Martha’s Vineyard, where he was with a vacationing President Bill Clinton, White House spokesman Joe Lockhart reported, “We are currently in conversation and dialogue with [the Russian government].”91
Not long afterward, the test intelligence subcommittee formally began to retreat, with one official describing it as “the last to join the crowd.” The group issued a new classified report in early September, which included additional data and stated that there was no connection between the test site activities and the seismic event and that the event occurred at sea. Before the middle of the month British government scientists seconded the opinion, telling the Defense Department that the event had “a similar location and mechanism” as an earthquake that occurred in the Kara Sea eleven years earlier.92
But that was not the end of story. While the CIA had no evidence that the August 16 signal had been caused by a detonation, it was reluctant to conclude that it was the result of an earthquake. One intelligence official explained that “we like our judgment to be based on positive evidence.” An alternative explanation that some officials were interested in investigating in late August involved the possibility of a sudden compression of the hull of one of the obsolete nuclear submarines that Russia had dumped into the Kara Sea and the signal being the result of the shock waves generated by that compression.93
In an attempt to put the issue to rest, because questions remained both within and outside the government, DCI George J. Tenet appointed a four-member panel to examine the evidence. The members included Sidney Drell, a Stanford University physicist, member of the President’s Foreign Intelligence Advisory Board, and longtime adviser to the U.S. intelligence community; Richard Kerr, who had spent three decades with the CIA and served as deputy director of central intelligence (1989–1992); Eugene Herrin, a professor of geological sciences at Southern Methodist University who had also chaired the DARPA Seismic Review Panel for fifteen years; and Roger Hagengruber, who held a doctorate in experimental nuclear physics from the University of Wisconsin and was vice president of the Sandia National Laboratories.94
The panel evaluated the imagery, communications intelligence, seismic signals, and the results of the AFTAC-directed WC-135 flight. The data included not only that gathered in the immediate aftermath. There was also the negative evidence that accumulated during two months of intelligence community efforts to find potentially corroborating evidence. There were no signals indicating telltale underwater blast sounds, no signs of unusual radioactivity, and no evidence of underwater drilling or other special activities in the Kara Sea prior to, during, or after the event.95
In late October the members of the panel delivered a two-page secret report to Tenet. They reported that two seismic events had been detected on August 16, about four hours apart. The second one was smaller but had the same signature as the initial event, “indicating it to be of the same source.” The centroid of the region, according to the report, “was located in the Kara Sea some 130 km southeast of the test area.” Most importantly, they noted that “available data leads to a firm conclusion that the site of the seismic activity is offshore, and, therefore, almost certainly not associated with the activities at [Novaya Zemlya].”96
The panel also concluded that “the seismic event on 16 August triggered a process that worked in many respects as it should within the monitoring community,” adding that “given the brief reporting times and limited data, the association of the seismic event with a nuclear test at NZ was sound.” Furthermore, the panel reported that until the seismic event “became unambiguously centered in the Kara Sea,” the intelligence community “needed to act with the emphasis and dispatch associated with a subkiloton nuclear test at the NZ site.”97
As a result of the report, both the CIA and the White House formally dropped any claim that the August 16 event had resulted from a clandestine nuclear test. On October 28 Tenet transmitted the report to higher authority, with a covering memo summarizing the panel’s key findings. A White House spokesman told one newspaper, “We agree with the judgment of the DCI based on the findings of the outside panel that this event was not nuclear.”98
Many also believed that the event in the Kara Sea was an earthquake, rather than an underwater volcano. In October, Harold P. Smith, assistant to the secretary of defense for nuclear, chemical, and biological programs, told the Washington Post, “I personally think it was an earthquake,” and added that other scientists at the Pentagon agreed. It was a view also shared by at least one member of the DCI’s panel, Eugene Herrin, who remarked that “it was not an ambiguous event. . . . It’s an earthquake.”99
U.S. spy satellites would continue to keep watch on Novaya Zemlya for the rest of the decade. In September 1998 images returned by those satellites revealed, to the interpreters at the National Imagery and Mapping Agency, activity at the test site that was normally associated with an underground test, including vehicle activity near a deep hole at the site and trucks unloading filter material at the mouth of the shaft. A Pentagon spokesman acknowledged, “We have observed some activity at this Russian test range suggesting that some nuclear-related experiments are underway.”100
What the imagery interpreters had seen and warned of in September was the beginning of a series of five subcritical tests that would take place between September 14 and December 13.101 But earlier in the year, the interpreters had not warned about another nation beginning a series of full-scale nuclear tests.
___________
* The Global Positioning System (GPS) is a constellation of twenty-one primary satellites, and three spares, in near-circular orbits 11,000 miles above the earth at an inclination of 55 degrees to its equator. While the satellites are best known for their role in allowing the precise location of individuals and objects, by 1995 they had been performing an important secondary mission for over decade. Every GPS satellite since GPS-8, launched in 1983, had carried a Nuclear Detonation (NUDET) Detection System (NDS) package on board. The NDS includes X-ray and optical sensors, bhangmeters, electromagnetic pulse sensors, and a data-processing capability that can locate a nuclear explosion to within one hundred meters. Data is reported on a real-time basis directly to either AFTAC or ground stations at Diego Garcia, Kwajelein Atoll, Ascension Island, or Kaena Point, Hawaii. See Jeffrey T. Richelson, The U.S. Intelligence Community, 4th ed. (Boulder, Colo.: Westview, 1999), pp. 218–219.
* There was even less need after 1996 to gather information about the French program. The U.S.-French nuclear cooperation that had begun in the early 1970s expanded further, and included an exchange of each country’s data base of nuclear weapons tests. The United States also provided the results of computer simulations on the workings of the “primary” (fission bomb) component of thermonuclear bombs. See R. Jeffrey Smith, “France, U.S. Secretly Enter Pact to Share Nuclear Weapons Data,” Washington Post, June 17, 1996, p. A9; Kenneth Timmerman, The French Betrayal of America (New York: Crown, 2004), p. 267.
* An April issue of National Intelligence Daily demonstrated that the U.S. intelligence sources were able to provide information on the worries of military personnel at the Lop Nur site. An article reported that “some military personnel at . . . Lop Nur . . . are concerned because the three nuclear tests planned for this year may require them to remain at the test site for long periods of time. The site has been contaminated by fallout from 17 years of atmospheric nuclear tests and improper venting from most of the underground tests.” See Director of Central Intelligence, “China: Nuclear Test Plans Prompt Health Concerns,” National Intelligence Daily, April 19, 1994, p. 10.