chapter eleven


POKHRAN SURPRISE

IT WAS 8:00 A.M. on May 11, 1998, when an aide handed John Lauder, who had become head of the DCI’s Nonproliferation Center in November 1997, what appeared to be a wire-service report. It stated that India had detonated one or more nuclear devices at its Pokhran test site. Lauder’s immediate response was to ask his aide, “Is this some sort of joke?”1 But his aide was not joking. Almost twenty-four years after its initial nuclear detonation, India had conducted another. This time there was no pretense that a peaceful nuclear explosion was involved.

DURING THOSE TWENTY-FOUR years, India had continued to develop its civilian and military nuclear capabilities, and on three occasions had come close to conducting further nuclear tests. Soon after Morarji Desai was sworn in as prime minister in March 1977, he convened a meeting of his cabinet’s political affairs committee to discuss Indian nuclear strategy. Although no test was approved, Desai, according to Homi Sethna, gave him the “green signal to refine the design [of the explosive device],” which involved reducing the weight and diameter of the device through miniaturization.2

Intelligence about Desai’s instructions to Sethna apparently reached U.S. officials, since in May 1977 President Carter hurriedly appointed Robert Goheen as U.S. ambassador and requested he meet with Desai immediately and ask him to restrain India’s nuclear weapons program. When the two met, Desai pledged, “I will never develop a bomb.”3

By January 1981 Desai had been displaced as prime minister by Indira Gandhi. That month she met with Raja Ramanna, director of the Bhabha Atomic Research Center, who reported that Indian scientists had asked permission to test two devices: a streamlined version of the 1974 device that could be delivered by aircraft, and a more advanced boosted-fission device that would produce four times the force with the same amount of plutonium. He also noted intelligence reports indicating that Pakistan’s bomb program was moving ahead. Gandhi gave her approval.4

Specifically, she approved the digging of two shafts for the devices and instructed that the bomb team get the devices ready. In February 1981 the 113 Engineer Regiment began work at Pokhran, where it had to battle both oppressive heat and an assortment of vipers, cobras, and scorpions. Team members were not told the purpose of the shafts, and to prevent U.S. and Soviet spy satellites from detecting their work, digging was done under camouflage netting. Trespassers were subject to being shot on sight. In May 1982 Gandhi would give approval for the tests, but then called it off only hours later.5

Gandhi’s reversal was the direct consequence of the images obtained by the National Reconnaissance Office’s spy satellites of the era—the KH-8, KH-9, and KH-11—and whatever other intelligence the CIA and National Security Agency had managed to collect about activities at Pokhran. That May, India’s foreign secretary, Maharaja Krishna Rasgotra, traveled to the United States. During a private meeting, U.S. undersecretary of state for political affairs Lawrence Eagleburger asked him, “What are you doing in Pokhran?” After Rasgotra denied that anything was taking place, Eagleburger took out maps and satellite pictures of the area. He pointed to them and told his guest that “there is a lot of activity going on there and it looks like you are sinking shafts. Are you going to conduct a test again?”6

Ragostra, who had been kept in the dark about India’s nuclear plans, told Eagleburger that to his knowledge there was no such plan but he would investigate when he returned. It was hours after Gandhi approved the test that he briefed her on his meeting with the undersecretary and told her that there would be major unfavorable consequences should India test. The prime minister listened but had no comment. Hours later she canceled the planned tests.7

The CIA and other intelligence agencies continued, of course, to monitor Indian nuclear weapons developments. In July 1982 a CIA study of the Indian program reported that scientists at the Bhabha Atomic Research Center had been conducting research on laser isotope separation. In October NSA produced a report on India’s shortage of heavy water, classified Top Secret Umbra, indicating that at least some of the information in the report came from high-level communications intercepts. NSA’s analysts noted that India’s inability to produce sufficient quantities of heavy water, along with its aversion to international safeguards, was a major factor constraining its nuclear power program. The report also inventoried the country’s heavy-water plants, from its first at Nangal in Uttar Pradesh to its fifth and largest at Kota in Rajasthan, and facilities to upgrade the quality of heavy water, including the operational ones at the Madras Atomic Power Station and the Rajasthan Atomic Power Station along with the one under construction at the Narora Atomic Power Station.8

Both the CIA and NSA probably also reported, later in the 1980s, on India’s covert acquisition of heavy water from foreign sources, including Norway. The CIA detected an illegal shipment of beryllium from West Germany to India late in the decade. In May 1989 director of central intelligence William Webster told the Senate Committee on Governmental Affairs that one of the indicators to the CIA of a country’s interest in developing thermonuclear weapons was the acquisition of beryllium, which he explained was “usually used in enhancing fission reaction.” In addition, the CIA had noted a number of other indicators of Indian interest in developing a hydrogen bomb, including purification of lithium, which is needed to produce the tritium used in thermonuclear explosions, and the separation of lithium isotopes.9

THE WORK on nuclear weapons by India and Pakistan was far from the only source of tension between them in the 1980s and beyond. In addition to the other nation’s nuclear progress, several incidents undoubtedly encouraged continued work on advanced weapons. In 1986–1987 India conducted Operation Brasstacks, a triservice military exercise of unprecedented size involving nine divisions and three brigades, held in training areas near the border with Pakistan. In response, Pakistan’s army extended and expanded its normal field training. Before the resulting crisis was over, the two countries had massed more than a quarter of a million troops on their border, threatening to turn normal winter exercises into a major confrontation.10

Then in 1990 the two adversaries squared off over Kashmir, the territory on India’s northern border that the two had claimed for over forty years, ever since the 1947 collapse of the British Empire in India. Over the course of several months India assembled two hundred thousand troops, including paramilitary forces and five brigades of the Indian Army Strike Corps, its most sophisticated attack force. Pakistan deployed its primary armored tank units along the Indian border and, according to one account, placed its nuclear weapons arsenal on alert. The United States, either through NSA eavesdroppers or the CIA-NSA Special Collection Service unit in Islamabad, intercepted a message to the Pakistani Atomic Energy Commission directing it to assemble at least one nuclear bomb. How close the two nations actually came to war, particularly nuclear war, is a matter of dispute. But the situation was serious enough for President Bush to send deputy national security adviser Robert Gates to the subcontinent to caution Pakistani and Indian leaders against the use of force and urge them to adopt assorted confidence-building measures.11

During the first half of the 1990s India and Pakistan continued their work on nuclear weapons as well as the means to deliver them, including ballistic missiles. In January 1992 Robert Gates, who had become director of central intelligence in 1991, told a congressional committee that while there was no evidence that India maintained, assembled, or deployed nuclear devices, “such weapons could be assembled quickly.” The next month, Shahryar Khan, Pakistan’s foreign minister, told the Washington Post that his nation possessed the components to assemble at least one nuclear bomb. By the end of that year India’s inventory of weapons-grade plutonium was estimated to have reached seven hundred pounds, and the expectation was that the inventory would reach over nine hundred pounds by the end of 1995. By 1995 Pakistan was still in the early stages of work on a ballistic missile that could cover northwestern India, including New Delhi. But it had acquired M-11 missiles from China, which had a range of 186 miles. India was more advanced, having successfully tested the two-stage Agni in 1994, whose expected range was 1,240 miles.12

In April 1995 prime minister Narasihma Rao simultaneously approved development of the Agni II and instructed the army to prepare the shafts at Pokhran needed to conduct nuclear tests. Rao was responding to the requests of two of India’s most important military scientists. One was A. P. J. Abdul Kalam, head of the Defence Research and Development Organization since 1992, who had achieved fame as the mastermind of India’s missile and space launch vehicle programs, including the Agni. The other was Rajagopala Chidambaram, who had become chairman of the Indian Atomic Energy Commission in 1993. Their scientists had been advocating testing for three reasons, including the need to perfect and demonstrate the technological innovations they had made. They also felt that their work could be validated only by full-scale tests and that recruiting top-flight scientists and engineers for the program required testing.13

Appointed as mission director that August was K. Santhanam, the defense research organization’s chief technical adviser. Santhanam had joined the organization in 1986 after a career that included heading the health physics division of the atomic research center as well as stints in the Research and Analysis Wing, India’s CIA, where he read intelligence reports on the Pakistani and Chinese nuclear programs and briefed the government, and with the foreign ministry.14

Following Rao’s instructions, the 8 Engineer Regiment at Pokhran began to refurbish the two shafts that had been built in 1982, and began digging a third. The deeper of the three shafts, code-named White House, was discovered to be filled three-quarters of the way to the top with water, the result of its having been capped after the previous regiment at Pokhran got tired of maintaining it. It took until November to complete the process of pumping out the accumulated water and finish refurbishing the shaft.15

While the engineer regiment was busy on the ground in Pokhran, U.S. imagery interpreters in the United States were taking an interest in its activities. Located at Fort Belvoir, Virginia, just south of the nation’s capital, was the main ground station for the KH-11 and advanced KH-11 satellites—with the cover name Defense Communications Electronics Evaluation and Testing Activity (Defense CEETA) and the designation Area 58. Also located there was the Priority Exploitation Group, a contingent from the CIA’s National Photographic Interpretation Center, whose job was to examine incoming satellite imagery to determine if any of it merited immediate attention. Further analysis of imagery of strategic importance was done at NPIC headquarters, a windowless building in the Washington Navy Yard, located in a rundown section of Washington.16

At the time, the interpreters could have had access to images from as many as six satellites: the two KH-11s launched in 1987 and 1988; the first advanced KH-11, orbited in 1992; the Misty stealth satellite, which had been operating since the spring of 1990; and the two Onyx radar imagery satellites launched in December 1988 and March 1991. In addition to obtaining images during daylight hours, it was also possible to acquire images at night, using the advanced KH-11’s infrared sensor or the radar-imaging capability of the Onyx satellites.17

Starting in November 1995 images from at least some of those satellites showed increased scientific and technical activity at the Pokhran test site. When the U.S. ambassador to India, Frank Wisner Jr., was in Washington at the beginning of the month, he paid a courtesy call on secretary of state Warren Christopher. Minutes after that meeting he learned that imagery from a satellite passing over Pokhran had caught sight of suspicious activity, including cables running through L-shaped tunnels, apparently to transmit diagnostic data from an underground test.18

It was not completely clear whether that activity was related to preparations for a nuclear test or some nonnuclear experiment intended to increase India’s expertise in making nuclear weapons. One U.S. official told the New York Times in December, “We’re not sure what they’re up to.” The official added, “If their motive is to get scientific knowledge, it might be months or years before they do the test. If it’s purely for political reasons, it could be this weekend. We don’t know the answer to those questions.” Intelligence officials also told the Washington Post that the images, which showed the clearing out of a deep underground shaft and possible preparations for instrumentation, depicted “activities [at the test site] going beyond what we’ve seen in the past” and that Indian scientists were trying to develop boosted atomic bombs as well as a hydrogen bomb. The Indian government first denied plans for a nuclear test and then characterized the Times report as “highly-speculative.”19

That politics might be a motivating force, and a test could come sooner than later, was a serious possibility. With an election coming up, Rao’s party was facing a serious challenge from the right-wing Hindu nationalist Bharatiya Janata Party (BJP), which called for India to come out of the nuclear closet. A secret cable drafted by State’s intelligence and research bureau noted that “Rao’s effort to recover his political reputation and to refute BJP charges that he has compromised the defense of India could soon result in the testing of a nuclear device in the Rajasthan desert.” A State Department official observed that if India exploded a nuclear bomb, it “would be a matter of great concern and a serious setback to nonproliferation efforts.”20

In an attempt to prevent a test, Wisner met with A. N. Varma, Rao’s private secretary, on December 15, only hours after returning from Washington. He was equipped, as other American representatives before him had been when challenging foreign nations about military activities, with satellite imagery—in this case a single image, which he showed Varma and then “put in his back pocket,” according to Strobe Talbott who was undersecretary of state at the time. Wisner warned that a test would backfire and bring sanctions. President Bill Clinton followed with a call to Rao, urging him not to proceed with any tests. The effort proved successful, or at least not unsuccessful, when foreign minister Pranab Mukherjee denied that India was preparing to conduct a nuclear test.21

Of course, Indian officials had denied that there was any substance in the first place to the public reports of India preparing to test nuclear weapons. The press minister for the Indian embassy in Washington claimed that the movements seen by U.S. satellites “have been absurdly misinterpreted.” Close surveillance of the test site continued, with U.S. satellites producing four images a day in January 1996—an activity certainly not discouraged by accounts early that month in the Indian press, monitored by the CIA’s Foreign Broadcast Information Service, of more army troops and increased activity at Pokhran, including fencing off the old test site and laying down cables.22

Then in mid-May, the results of India’s parliamentary elections made Mukherjee’s pledge irrelevant. While the BJP finished with less than a majority of seats, it did have the largest plurality and received the first opportunity to put together a government. On May 16 its leader, Atal Behari Vajpayee, took the oath as prime minister and received fifteen days to win a parliamentary vote of confidence for his government—which would require it to add 75 votes to the 186 it could count on from the party’s representatives.23

During the campaign the BJP had promised to reevaluate India’s nuclear policy. Vajpayee himself had asked, in December 1964 after China’s first test, “What is the answer to the atom bomb?” He also provided his answer: “the answer to an atom bomb is an atom bomb, nothing else.” When the party’s general secretary claimed, in early April, that if the BJP came to power it would test a nuclear weapon, another party spokesman claimed that while it was the party’s position that India should possess a nuclear deterrent, “the issue of testing has not been discussed.” Maybe no discussion was needed, or it was discussed in the six weeks between early April and Vajpayee’s appointment, but almost immediately after he assumed the prime ministership Vajpayee told Kalam and Chidambaram to proceed with the tests, which were to involve an improved version of the 1974 device as well as a boosted-fission bomb.24

Once again NRO’s constellation of imagery satellites provided indications that a test might be upcoming, picking up signs of renewed activity that spring, although they failed to detect that the team at Pokhran had placed at least one nuclear device in a test shaft. While senior U.S. officials did not expect an imminent test, the Clinton administration did follow up on the new intelligence and urged Indian officials to refrain from testing.25

But it was not Bill Clinton, spurred on by the information gathered from space, that prevented an Indian test. Instead, it was Vajpayee and developments on the ground. With a vote of confidence pending, the prime minister decided to wait for the outcome before giving the final go-ahead. On May 28, 1996, failing to attract the additional 75 supporters, the BJP lost the vote of confidence and was replaced by a United Front government consisting of thirteen parties. But in March 1998 the BJP would return to power when it won 250 seats in the parliamentary elections, 22 short of a majority but sufficient to give Vajpayee the opportunity to form a government. And this time the coalition he assembled prevailed, albeit by a slim margin, in the confidence vote that followed before the month was over.26

DURING THE CAMPAIGN the BJP issued an election manifesto with the title “Our Vision, Our Will, Our Way,” which covered topics ranging from cow protection (a total ban on the slaughter of cows was called for) to policy on weapons of mass destruction. It echoed 1996 campaign rhetoric with respect to nuclear weapons, noting that “the BJP rejects the notion of nuclear apartheid” and would “re-evaluate the country’s nuclear policy and exercise the option to induct nuclear weapons.” The BJP saw a nuclear-armed India as a way not only of deterring Pakistan, but of deterring another nation that presented an even greater threat—China.27

On March 20, the day after Vajpayee and the BJP assumed control of the Indian government for the second time, the new prime minister was visited by Chidambaram. “It was not,” one of the Indian leader’s aides recalled, “a pure courtesy call.” The atomic energy chief was there to make the case for conducting nuclear tests. Early the next month, on April 6, Pakistan tested its Ghauri missile, which could hit targets 930 miles away and carry a payload of 1,540 pounds. Two days later, Chidambaram and Kalam received the answer they were hoping for when they were summoned to see the prime minister and told to go ahead with the tests, with Vajpayee telling them that the Ghauri test was the last straw. A contingent of one hundred scientists and engineers soon packed up and headed for the test site.28

Sometime during the night of May 1 an Indian air force Antonov-32 plane took off from Santa Cruz airport in Mumbai, carrying the plutonium cores to be used in the tests. The cores weighed between eleven and twenty-two pounds and had been produced at the BARC in Trombay and stored in underground vaults in Mumbai. Two hours later a convoy of trucks lined up at the Jodhpur airport and the crates were loaded onto one of them. To make the whole activity appear routine, no extra security was provided, and the convoy set out for the test site under the cover of darkness. Once it arrived at Pokhran, the crates were moved to temporary labs, code-named Prayer Hall, where the bomb team began mating the cores with the conventional explosives, detonators, and triggers, which had been flown in separately.29

On the morning of May 7 a team of scientists from the Kalam’s defense research organization and Chidambaram’s Department of Atomic Energy arrived in Jodhpur. That night they left for Pokhran, arriving early the next morning. On May 10 preparations for lowering the devices into the shafts began. The device with an expected subkiloton yield was the first lowered down its shaft, which was sealed by 8:30 in the evening. A hydrogen bomb was lowered into the shaft code-named White House, about 655 feet deep. It was sealed with concrete and sand by 4:00 a.m. on May 11, while a fission bomb was lowered into a third shaft, designated Taj Mahal, which was sealed by 7:30 a.m. Back in New Delhi senior officials had only just learned of the impending tests. Defense minister Georges Fernandes was told on May 9, while the military service chiefs and foreign minister were let in on the secret the following day.30

The tests were scheduled for 9:00 a.m. on May 11. Chidambaram and Kalam waited in a tiny bunker, code-named Deer Park, containing computers and control panels with an assortment of colored switches. They wore battle fatigues in accord with their cover identities as Major-General Natraj and Major-General Prithviraj, identities adopted to prevent knowledge of their repeated visits to the test site from leaking. But at eight that morning the test was put on hold, because of a west wind that would have blown any radioactive debris accidentally released not only toward nearby villages but also into Pakistan.31

While the wind was still a problem at noon, it would eventually subside, and at 3:45 p.m. the Pokhran test site would be rocked by India’s first nuclear test since 1974. Ten minutes later a phone rang in a room in the prime minister’s residence, where Vajpayee and five other senior Indian officials, including Fernandes and the defense, home, and finance ministers, were gathered. When the prime minister’s principal secretary picked up the phone, he heard a voice cry, “Done!,” telling him that the tests, which would subsequently be designated Operation Shatki, had been a success.32

At 5:00 p.m. Indian time, 7:30 a.m. in Washington, Vajpayee walked over to reporters and cameras waiting outside his office and announced that India had tested three nuclear weapons. Later, his government issued an official statement providing specifics: the weapons included a fission device, a low-yield device, and a thermonuclear device and the measured yields were “in line with expected values.” No radioactivity was released into the atmosphere, according to the statement. Two days later, on May 13, the government issued another statement, announcing that two subkiloton tests were carried out at Pokhran earlier that day.33

The Indian claims came as a surprise to the United States once again. The CIA and other intelligence agencies had provided no warning of a second round of tests. At a classified briefing on Capitol Hill on May 12, Nonproliferation Center chief John Lauder gave no indication that further tests were anticipated. And on May 13 deputy assistant secretary of state for nonproliferation Robert J. Einhorn told a Senate committee, “I personally woke up this morning and I did not know about it.”34

THE SURPRISE INDIAN TESTS left the U.S. intelligence community with three clear tasks. The most pressing was gathering intelligence on how Pakistan planned to respond and when. Then there was the need to gather and evaluate intelligence on the Indian tests—to verify, if possible, Indian claims about what had taken place and to obtain information about the tests that India did not wish to volunteer. It would also be necessary to consider why the almost $30 billion that the United States spent each year on its large array of intelligence agencies, analysts, and collection systems had failed to buy advanced warning of India’s tests.

That India’s nuclear tests would make a Pakistani test a certainty—and sooner rather than later—seemed clear to U.S. intelligence analysts. The State Department’s intelligence and research bureau commented on May 13 that “though some Pakistani officials will counsel patience to allow the weight of the international opprobrium to fall exclusively on India, mounting domestic political pressure makes a Pakistani nuclear test virtually inevitable.” The following day State informed national security adviser Sandy Berger that various sources, including the Pakistani press and (apparently) communications intercepts, indicated that Pakistani president Nawaz Sharif had given the “green light” for tests.35

Also on May 14 director of central intelligence George Tenet told the House and Senate intelligence committees that U.S. satellites had detected Pakistani preparations for an underground test. In closed session, the CIA chief presented satellite imagery showing an increase in equipment, technicians, and security activities at the Ras Koh test site in the Chagai Hills. The satellite images also indicated heavy traffic on the roads, with some being used for the first time, and heavier traffic on others.36

What the satellites were seeing were preparations for a nuclear test, but a nuclear test that Pakistan was well prepared for. Six months earlier, a multidisciplinary team of Energy Department nuclear experts—from Z Division, Oak Ridge Laboratory, and Savannah River Laboratory—noted that Pakistan was making significant progress in developing nuclear weaponry. The key intelligence that led to this conclusion concerned Pakistan’s progress in producing yield-enhancing tritium. While the information had been acquired by the CIA and initially kept away from the Energy Department’s intelligence personnel, a CIA detailee to Energy was able to get his agency to share the crucial intelligence.37

By May 27 most of the U.S. intelligence community believed that Pakistan was close to conducting its first test—although an INR analysis, consistent with other analyses it had produced since mid-May, suggested that Sharif “probably still hopes to avoid having to order a test.” But “it could happen any time,” a U.S. intelligence official told the Washington Times. The test site activities continued to be the target of both U.S. imagery satellites as well as those capable of intercepting communications from 22,300 miles above the earth. The sum of U.S. intelligence indicated that a nuclear device had been placed in an underground shaft and that the sensors, cables, and other equipment needed to monitor a detonation were in place.38

Sometime on May 27 President Clinton himself collected a definitive bit of human intelligence from Pakistan’s president, intelligence indicating that his efforts to prevent a Pakistani test were going to fail. According to Strobe Talbott, during a phone conversation “a timorous sounding Sharif apologized to Clinton for ‘disappointing’ him, but [he] simply had no choice but to go ahead with the test.”39

At 3:30 p.m. on May 28 seismic stations recorded the signals commonly associated with nuclear testing, and Sharif issued a statement claiming that “Pakistan today successfully conducted five nuclear tests” and congratulating “all Pakistani scientists, engineers and technicians for their dedicated team work.” In a somber early-evening television address, Sharif announced, “Today, we have evened the score with India.” Two days later Pakistan one-upped its adversary by announcing a single test, at a site about sixty-two miles from the May 28 tests, bringing the number of claimed Pakistani tests to six.40

THE PAKISTANI CLAIMS meant that now two nations’ assertions about their nuclear accomplishments needed attention from U.S. nuclear intelligence experts. The nuclear intelligence community and its customers would want to know if the claims about the number of tests conducted, their yields, and the nature of the devices—for example, whether the claimed hydrogen bombs were indeed thermonuclear—were accurate.

Among the sources of data were the written statements released by the Indian and Pakistani governments, and the oral remarks. There was also the question-and-answer session held on May 17 by AEC chairman, Chidambaram, Kalam, BARC director Anil Kakodkar, and DRDO chief adviser for technology Santhanam. Kalam repeated the claim that the tests on May 11 included a hydrogen bomb, a fission device, and a subkiloton device, while Chidambaram stated that the hydrogen and fission devices were six-tenths of a mile (one kilometer) apart and the yield of the hydrogen device was 45 kilotons. Chidambaram also reported, as he had previously, that the fission device was significantly lighter and more compact than the 1974 bomb and produced a yield of 12 kilotons, while the yield of the third device was 0.2 kiloton. Kalam also told his audience that the test had been approved thirty days prior to May 11.41

The scientists at the press conference released a videotape of the blasts that allowed viewers to see the countdown at the test site, hear a deep boom, witness the ground shake violently, and see a huge cloud of dust rise into the sky above the test site. Cheers could also be heard. Views from a helicopter showed a crater that appeared to be seventy feet deep and several hundred feet wide. One could also see what appeared to be a concrete walkway leading into the shaft. Its sides had been shattered and covered with netting, twisted steel, and broken sandbags.42

Analysts would pay close attention to the BARC Newsletter, whose May 1998 issue contained a note written by Kakodkar and S. K. Sikka, a longtime research associate of Chidambaram. They reported the yield of the hydrogen device as 45 kilotons and of the fission device as 15 kilotons. The three smaller devices were described as “experimental” and their yields given as 0.2, 0.5, and 0.3 kiloton.43

There were additional things U.S. intelligence analysts would have liked to know that the Indian scientists weren’t telling. During the press conference, AEC chairman Chidambaram would not reveal the ingredients used in the hydrogen bomb or the depth of the shafts. Kalam had no comment when asked about how many warheads India needed, when production would begin, and whether the “U.S. surveillance system was deliberately fooled by you or was it accidental?”44

To gather whatever information it could on the Indian and Pakistan tests, the U.S. intelligence community employed a variety of assets. Imagery satellites photographed the test sites in the aftermath of the detonations, while communications intelligence systems on the ground and in space continued to intercept and analyze whatever relevant communications they could. The imagery from advanced KH-11 satellites could show where radiation from the tests had hit the surface, rapidly deoxidizing the rock and turning the mountainside white.45

Satellite imagery can also be employed to identify “throw-out” craters and their ejecta blanket, created by explosions at shallow depths, as well as any domelike structure (referred to as a retarc — for “crater” backwards) created by detonations at greater depths. The surface signatures of an underground blast, along with knowledge of the geology of the site (which influences the surface features) and the depth at which a device was buried, could be used to produce estimates of its yield.46

In addition, a WC-135, stationed at Offutt Air Force Base in Nebraska, was deployed to South Asia in search of any debris that had been injected into the atmosphere. It was an opportunity that was almost missed. By May 1998 the nation’s aerial nuclear sampling fleet, whose mission was designated Constant Phoenix, was down to a single plane, the result of the decline in worldwide nuclear testing and budget constraints. Later, Terry Hawkins, the director of the nonproliferation and national security division at Los Alamos, would observe that “sampling capability is really important. If a nuclear explosion occurs somewhere, and if you want to attribute it to a country that conducted the test, the only credible way to do that is to get a piece of the debris and analyze it.” But by May 1998 there had not been a nuclear test for almost two years, and the plane used for debris collection was scheduled for six months of maintenance, with its 1,200-pound collection system headed for storage. The Indian tests resulted in a crash effort to reconstitute the aircraft in time to fly the sampling missions, manned by the last remaining personnel trained to use the equipment for detecting and gathering debris.47

In addition to imagery, communications intelligence, and debris collection, seismic signals were available to American nuclear intelligence analysts from both unclassified seismic detection systems as well as the stations operated by the Air Force Technical Applications Center or allies. In the immediate aftermath of the Indian tests, the AFTAC-collected seismic signals seemed to indicate a combined yield for the May 11 Indian tests of only 15 to 20 kilotons, about a third of that claimed by Indian scientists. Such a relatively low yield, even for a single test, raised questions as to whether the tests really included a full-fledged, first-generation hydrogen bomb, which could be expected to produce yields up to a half million tons of TNT or more. Nor was there proof in the seismic signals that more than one test had been conducted, since only one signature had been detected.48

There was even more skepticism concerning the Indian claims about the May 13 tests. No seismic signal could be detected that would serve as confirmation that India had conducted even one, much less two, tests that day—not by the International Data Center network, or the U.S. Geological Survey, or the seismic station at Nilore in Pakistan, which detected the May 11 explosions, and apparently not by any of the AFTAC stations.49

At their press conference Indian scientists tried to answer some of the skepticism that had emerged within days after the tests, largely owing to the seismic data obtained by unclassified stations. Chidambaram emphasized that India had detonated a true hydrogen bomb—a plausible claim since BARC reportedly began, in the mid-1980s, separating lithium-6, which could then be placed in reactor cores and transformed into tritium or mixed with deuterium to produce lithium-6 deuteride. “We used a fission trigger and a secondary fusion,” he said. Kalam and his colleagues explained the thermonuclear yield of 43 kilotons as the result of a decision to limit yield in order to minimize damage to nearby villages. It was also suggested that the simultaneous detonations caused “interference” in the seismic signals being used to evaluate yield. Failure to detect the May 13 tests were, the scientists explained, due to their having been conducted in a sand dune.50 Their explanations, of course, were available not only to the world at large but to the U.S. nuclear intelligence establishment.

Such explanations still left some outside experts unpersuaded, in part because India’s 1974 claim that its bomb had exploded with the force of fifteen thousand tons of TNT, when subsequent analysis produced estimates in the 6 to 8 kiloton range, indicated a willingness to exaggerate its accomplishments. “They definitely hyped it the first time around,” noted George Perkovich, an expert on the Indian nuclear weapons program.51

“The whole thing sounds odd,” Herbert A. York, a former nuclear bomb designer and Pentagon chief of research and engineering observed. “It’s not odd enough to make me say it’s not true, but it’s still a very strange story.” Experts also noted India’s claim that it had conducted a full thermonuclear test without having gone through the intermediate step of testing a boosted-fission device using thermonuclear material, as did other nations. In addition, experts speculated that the two tests of May 13, which had the equivalent of two hundred and six hundred tons of TNT according to India, might have been failed boosting tests. “Maybe they tried and failed,” observed former Livermore bomb designer Ray E. Kidder. Peter Zimmerman, a physicist and former arms control official, wrote that sand dunes were porous and carried the risk of letting radioactive gases escape, making them a strange choice for a medium within which to conduct a nuclear test.52

The public controversy over the yields of the May 11 and May 13 tests would continue for several years. Numerous foreign experts argued that the total yield was significantly lower than the claimed yield, and the yield of the alleged hydrogen bomb test was closer to 30 than 43 kilotons, while others, particularly Indian scientists, defended the Indian government’s estimates.53 But settling that issue would not settle the more important concern about India’s tests—did they demonstrate that India possessed a hydrogen bomb?

In late November the trade journal Nucleonics Week reported that analysts at Z Division had completed several months of analysis of seismic, human, and signals intelligence, which apparently included intelligence on contacts between the Department of Atomic Energy and senior Indian decisionmakers. They concluded that India had attempted to detonate a hydrogen bomb on May 11, not just a boosted-fission device. However, “the secondary didn’t work,” according to one of the journal’s sources. While the fission primary detonated as planned, the heat failed to ignite the second stage, which contained the thermonuclear material. As a result, “if India really wants a thermonuclear capability, they will have to test again and hope they get it right,” one U.S. official said.54

PAKISTAN’S CLAIMS were also subject to similar scrutiny and skepticism. While no one doubted that India’s adversary had detonated at least one nuclear device, there was no initial evidence of more than one. The unclassified global seismic monitoring network detected a signal from deep beneath Pakistani territory on May 28, a confirmation that Pakistan had conducted a nuclear test. The preliminary measurements of the signal’s magnitude indicated a yield of between 8 and 17 kilotons. As with India’s May 11 tests, there was only one signal, which could have been caused by a single detonation or the five simultaneous explosions claimed by Pakistan.55

U.S. intelligence analysts believed that it was likely that at least two bombs had been tested but were skeptical of the claim that five had exploded. They thought that while it was not implausible to detonate two bombs at identical moments, it would be technically challenging as well as highly unusual to fire off five detonations simultaneously. Analysts also speculated that the device or devices had not performed up to specifications. U.S. data indicated that the magnitude of the blast(s) was between 9 and 12 kilotons, with 6 kilotons being the most likely value, which was less than the estimated yield of even one of the principal bombs in the Pakistani arsenal. One obvious possibility was that Pakistan had exaggerated its achievements in order to match India’s claims. “We don’t believe either nation is really telling the truth about what they did,” one U.S. official commented.56

Unlike India’s claim of follow-on tests, Pakistan’s assertion that it tested again, shortly after its initial tests, was backed up by seismological data. But the faintness of the signal picked up indicated to many specialists that the test was either a failure or successful but small. One of those specialists, University of Arizona seismologist Terry C. Wallace, remarked at the time, “It’s a small event.”57

There was also skepticism, within at least some segments of the U.S. nuclear intelligence community, concerning A. Q. Khan’s claim that the Pakistani bombs used highly enriched uranium—a conclusion reached after analysis of debris collected by the WC-135 dispatched to South Asia in May 1998, specifically debris collected from the May 30 test. That December the CIA informed President Clinton, in a highly classified report based on a preliminary analysis done at Los Alamos, that the debris contained low levels of weapons-grade plutonium. The finding implied that Pakistan had been manufacturing or importing plutonium, which would allow the development of smaller warheads and thus either longer-range missiles or missiles with more deadly payloads, without detection by American intelligence agencies.58

The conclusion reached by the Los Alamos scientists came as a surprise to other elements of the U.S. nuclear intelligence community. The United States had been routinely monitoring work on a reactor at Khushab in the Punjab, where construction had begun in the late 1980s, but which Islamabad had first announced as going critical in the spring of 1998. There was also an awareness of increased Pakistani research and development activities with respect to plutonium production. But prior to the Pakistani tests, the U.S. officials claimed that Pakistan had no significant reprocessing capability other than the small pilot facility at the Pakistan Institute of Nuclear Science in Rawalpindi.59

The conclusion by the Los Alamos scientists was not shared by their counterparts at Lawrence Livermore, those who worked or consulted for Z Division. It was not the first disagreement. Former Energy Department intelligence chief Notra Trulock has noted that Los Alamos’s International Technology Division was fiercely competitive with Livermore’s Z Division and “IT’s harshest invective was reserved not for a foreign espionage operation, but for a Z-division report or Z-division personnel.” David Kay, the former UN weapons inspector and chief of the Iraq Survey Group, recalled that it was “interesting to watch them denigrate each other’s assessment.”60

In this case the Livermore scientists claimed that Los Alamos first contaminated the May 30 sample. Any opportunity to reevaluate disappeared when the sample was lost. A second sample was available at another laboratory, but whether it was an identical sample was another matter of dispute. One U.S. intelligence official acknowledged that “there is some disagreement here, and experts at the labs need to sort it out.”61

BUT THE MOST PRESSING question facing the U.S. intelligence community was why it had failed to provide advance warning of India’s intentions, possibly allowing the United States to exert diplomatic pressure as it did in 1995 and preventing India’s tests and Pakistan’s reaction.

The United States was not the only nation that had been surprised by the tests. Russian president Boris Yeltsin and his foreign minister, Yvgeny Primakov, characterized the tests as a “big surprise.” An official of the Russian Foreign Intelligence Service, the SVR, told one newspaper that the service “had no information” about India’s plans to carry out the tests, despite the reputation of India being “transparent” to Russian intelligence and the presence of a large number of India specialists in the SVR, including director Vyacheslav Trubnikov.62

But the Russians did little to mitigate the concern that the U.S. intelligence community, with significantly greater resources, particularly in space, had failed to provide notice of such a significant event. And it was a failure that stretched across the entire nuclear intelligence establishment, from the CIA and INR in the Washington area to Z Division in California. When Phyllis Oakley, the director of INR, was grilled at a closed Senate hearing on May 14, 1998, she conceded, “We were wrong. We were all wrong.” One former intelligence official recalled that the “guys at Livermore thought they owned the Indian program” and “blew off BJP claims that they would test.” The State Department, he said, accepted their view. Senator Richard Shelby, chairman of the Senate Select Committee on Intelligence, labeled the episode “a colossal failure of our intelligence-gathering system, perhaps the greatest failure in a decade.” He also bemoaned the lost opportunity for the president or secretary of state to attempt to intervene diplomatically. A senior administration official characterized the failure “as a very big deal, especially because nonproliferation is supposed to be the No. 1 priority of the intelligence community.”63

On May 13 former director of central intelligence James Woolsey suggested that the failure to detect blast preparations could be traced back to a decision earlier in the decade to cut the projected size of the reconnaissance satellite constellation.64 However, it would soon become clear that the problem was not in space but on the ground. The United States had the intelligence that would have provided last-minute warning of a test, but it went unanalyzed until after India’s first round of tests had been completed. That intelligence was in the form of imagery from an advanced KH-11 satellite that was in the hands of interpreters in the Priority Exploitation Group.

Beginning on October 1, 1996, the interpreters at Fort Belvoir reported to a new organization, the National Imagery and Mapping Agency (NIMA) within the Department of Defense. NIMA had been established on the initiative of director of central intelligence John Deutch to combine the national imagery interpretation and mapping efforts. The new entity absorbed the NPIC, the Defense Mapping Agency, the Central Imagery Office, the imagery interpretation elements of the CIA’s Directorate of Intelligence and Defense Intelligence Agency, along with other offices and programs. The imagery interpreters at Fort Belvoir would be given a new designation but their mission remained the same.65

On May 8 a satellite transmitted imagery showing signs of renewed activity at the test site, including the presence of bulldozers nearby. That imagery may also have shown activity at the wellheads on top of deep holes where the nuclear devices would be detonated—if not, other imagery taken about the same time did. But not until sometime in the early hours of May 11 did an Area 58 interpreter notice something of interest in newly received images of the Indian test site—fences being removed and some panels, possibly motion detectors, being laid on the ground. The analyst marked the imagery for further analysis by the more experienced day-shift analysts, who were not expecting a test, and were presumably asleep rather than on alert. By the beginning of their workday, the opportunity for warning had passed.66

The BJP’s platforms and the statements of Indian officials might also have been a source of warning. Days after the first Indian tests, Senator Daniel Patrick Moynahan, a former U.S. ambassador to India, posed the question, “Why don’t we learn to read? . . . The political leadership in India as much as said they were going to begin testing.” In early March AEC chairman Chidambaram stated publicly, “We are technologically prepared to go nuclear, but it is for the policy-makers to decide whether to go nuclear.” At about the same time, Lal Krishna Advani, president of the BJP, and soon to become home minister, described a nuclear-free world as “a distant dream.” But in classified reports analysts from the CIA, DIA, and State Department discounted the promise to go nuclear as mere campaign rhetoric, possibly believing it would not stand up to the threat of sanctions. It was a view communicated to higher levels and accepted. During his mid-April meeting with Pakistani leader Nawaz Sharif, U.S. ambassador to the UN Bill Richardson dismissed the BJP’s vow to develop nuclear weapons as “election rhetoric.”67

There might also have been an opportunity to uncover Indian plans to test if the CIA had had a human source in one of the villages in the Pokhran area, since the imminence of a test had been an open secret in those villages. Alternatively, attention to a Sikh community newsletter in Ontario, Canada, which speculated on an upcoming test four days before it happened, could have provided advance warning.68

THE U.S. INTELLIGENCE community’s lack of attention was augmented by India’s deception campaign. It employed the same technique that foreign officials had used for years in the face of a U.S. challenge concerning the nuclear intentions: they lied. State Department spokesman Jamie Rubin labeled it a “campaign of duplicity.” In late March 1998, senior foreign policy adviser N. N. Jha reassured officials at the American embassy that over the next three to six months his government would be reviewing Indian national security policy and had no plans for weapons tests. Possibly Jha offered his assurances in good faith, with no knowledge of what was being planned. But his message, and one that was repeated to American officials in March and April, was false.69

On April 14 UN ambassador Bill Richardson met with Vajpayee and others in New Delhi, stressed U.S. opposition to nuclear testing, and was persuaded that there was no plan for tests. Sandy Berger, the president’s national security adviser, did not raise the issue directly during a May 1 meeting in Washington with India’s foreign minister, but praised India’s restraint in not responding to Pakistan’s missile test and came away reassured. That the Indian officials were telling their Clinton administration contacts exactly what they wanted to hear—that restraint would prevail—made it easier to believe.70

Some of the Indian officials were not aware they were providing misinformation because they did not know that the prime minister had already given Kalam and Chidambaram the go-ahead for the tests. The small number of Indian officials let in on the planned tests made it easy for others, including the foreign minister, to tell plausible lies since they did not believe they were being deceptive. Of course, no matter how plausible or sincere the lies, hard intelligence would convince those in the CIA and other agencies who monitored Indian nuclear activities that a test was in the works.

Indian officials and scientists, who certainly had not forgotten Frank Wisner’s use of satellite imagery to make his point in 1995, tried to limit what U.S. spy satellites photographed in 1998. “They knew from the 1995 experience we were watching them closely,” one U.S. official observed. Indian countermeasures included burying the cables and wires running into the shaft where they conducted the tests, placing camouflage netting over the test area, and conducting as many operations as possible at night or when satellites were not overhead. Working largely during the night the 58 Engineers dug two new shafts during the first half of April. On the night of May 5 the regiment laid the cables at the various shafts and then tried to cover their tracks, replacing the vegetation in the hopes that satellite images would show no sign that the area had been disturbed. The Indians also took advantage of the sandstorms that occur during May and block the view of KH-11–type satellites.71

The 58 Engineers also believed that its activity could be detected by U.S. imagery interpreters by tracking the movements of sand mounds required to seal the shafts. Mounds that occurred through the forces of nature would form in the direction of the blowing winds. Mounds created by bulldozers could stand out from those created naturally. To prevent discovery, the winds were monitored to ensure that any artificially created mounds were aligned with the wind.72

The Indians also tried, successfully, to prevent imagery from revealing vehicle movements that signaled increased and test-related activity. Each vehicle had an assigned parking space for the day. No matter how extensively they were used at night, they would be back in their designated spaces by morning, hoping to give the impression that nothing of consequence was happening at the test site.73

Also contributing to the failure to detect the tests well in advance was the limited targeting of the satellites on the test site. By mid-April, having heard the reassurances of Indian officials, the Imagery Requirements Subcommittee of NIMA’s Central Imagery Tasking Office—which decided which targets would be photographed by U.S. spy satellites, from what angle and height, and when—decided to concentrate on India’s missile sites, with the reasonable expectation that a response to Pakistan’s Ghauri test would be forthcoming. Coverage of the nuclear test site was scheduled at three-day intervals, in stark contrast to the four images a day being obtained in January 1996. And the less the satellites saw, the more administration fears decreased.74

India’s success in preventing U.S. spy satellites from seeing signs of the planned tests days to weeks in advance was matched by its success in preventing acquisition of other types of intelligence. India’s Intelligence Bureau ran an aggressive counterintelligence program, and the CIA, despite a large station in New Delhi, was unable to recruit a single Indian with information about the Vajpayee government’s nuclear plans. Instead, the deputy chief of the CIA station in New Delhi was expelled after a botched try at recruiting the chief of Indian counterintelligence operations. Former ambassador Frank Wisner recalled that “we didn’t have . . . the humans who would have given us an insight into their intentions.”75

Nor had NSA’s eavesdropping activities detected test preparations. “It’s a tough problem,” one nuclear intelligence expert told investigative journalist Seymour Hersh, because India’s nuclear weapons establishment would communicate via encrypted digital messages relayed via small dishes through satellites, using a system known as VSAT (very small-aperture terminal), a two-way version of the system used by satellite television companies.76

IT WAS SHORTLY AFTER the first Indian tests that director of central intelligence George Tenet appointed a panel, headed by former Joint Chiefs of Staff vice chairman Adm. David Jeremiah, to review intelligence community performance. While the full report remains highly classified, Jeremiah did hold a press conference when the report was finished, in early June 1998. The CIA also released a set of unclassified recommendations.

In his press conference Jeremiah noted that the identification of Indian nuclear test preparations represented a difficult collection problem as well as a difficult analytical problem. Since the program was not derived from a foreign one but was indigenous, some characteristics were difficult to observe. He mentioned the attention to limiting the number of Indian officials aware of the program as well as efforts to limit observable actions related to test preparations.77

He also noted the assumption, both in the intelligence and policy communities, that the BJP would behave as a Western political party, making promises in a political platform but not necessarily keeping those promises once they reached office and were confronted with the problems or costs of keeping their pledge. It was also assumed that the BJP would not be willing to suffer the economic sanctions that would follow a test.78

Unclassified recommendations discussed at the press conference ranged across all aspects of intelligence community activity. It would be necessary to add rigor to analysts’ thinking, bringing in outside experts in a more systematic fashion; to reexamine the formal warning process in anticipation of altering it significantly; and to find means to better integrate regional and technical analyses. There was also a need to devote more resources to the processing and interpretation of imagery, relative to the resources expended on collecting imagery. For years the collection of imagery far outpaced the ability of the machines and people on the ground to turn it into finished intelligence.79

It was also recommended that collection priorities be realigned so that high-priority issues within individual nations, such as Indo-Pakistani weapons of mass destruction programs, would be treated with similar urgency as rogue states. Another recommendation was to create a management structure to integrate collection systems so that collection is tasked as a “system of systems” rather than as individual activities.80

Asked to what degree the failure to predict the tests was an intelligence failure and to what degree it represented an Indian success in keeping its plans secret, the admiral responded that the two factors had equal weight. As to whether advance warning could have been used to avert tests, his personal opinion was, “No, I don’t think you were going to turn them around.”81

In a statement issued the day of Jeremiah’s press conference, Tenet accepted all of Jeremiah’s recommendations and stated that he was “making it my highest priority to implement them as quickly as possible.”82