chapter three

---

THE VIEW FROM ABOVE

THE SOVIET TEST of Sakharov’s Layer Cake bomb would be followed, within thirty days, by four fission bomb tests: an airdrop on August 23 of the RDS-4 Tatyana, which would eventually be issued to bomber regiments, and three tower tests between September 3 and September 10. The airdrop produced a yield of 28 kilotons, while its successors produced smaller blasts, each below 6 kilotons.¹

During one of the 1953 Soviet tests, the Air Weather Service unveiled a new aerial sampling capability. Maj. James T. Corn and Lt. William H. Wright flew a modified B-57A to an altitude of fifty-four thousand feet into the debris cloud—a flight that marked the start of the special B-57s being used for intelligence missions. Under a classified program designated Black Knight, the air force directed the Air Material Command to procure six Martin Model-294 aircraft, which became the RB-57D Intruder.²

SOMETIME AFTER those four tests, the CIA’s nuclear intelligence analysts began preparing another assessment of the Soviet nuclear program. The information available to those analysts included the data gathered during and after Soviet tests, the results of krypton-85 monitoring, communications intelligence, and human intelligence. A national intelligence estimate provided a brief review of Soviet atomic weapons activities since 1945, and noted that while there was no doubt that espionage, German assistance, and Western scientific and technical literature made significant contributions to the Soviet program, the independent research required to adapt the information obtained to Soviet needs “was apparently carried out with a high degree of competence.” It also revealed that the Soviets had departed from U.S., British, and Canadian atomic energy practices on a number of occasions.³

Beyond history, the report offered some admissions of uncertainty and information gaps, some estimates and facts, and some speculation about alternative stockpiles. The exact extent of the Soviet capability to produce nuclear weapons remained uncertain, and the reason why U-235 production lagged behind plutonium production was unknown. In the latter case, there was “only meager evidence . . . available . . . relevant to the isotope separation program.” The lack of intelligence on that program was, the analysts wrote, “one of the most serious gaps in intelligence information on the Soviet atomic energy program.” That gap was one reason why there was “no clear evidence” to guide analysts trying to estimate the specific types and numbers of weapons that would make up the Soviet stockpile.⁴

The report did estimate that the total East German production of uranium metal available to the Soviet Union up to the end of 1953 was between ten and fifteen thousand tons, and noted that an equal amount could have been produced from domestic and other Soviet satellite sources. The CIA’s analysts also noted that the Soviets were depending on very low-grade uranium ore, the type of ore that Leslie Groves and others had believed could not be effectively exploited to produce bomb-grade material. The analysts did try, despite their limited knowledge, to calculate the number of warheads the Soviets might have available between the end of 1953 and mid-1957, under different combinations of weapons type and yield. But the numbers produced—somewhere between 12 and 550 at the end of 1953 and from 80 to 2,400 in mid-1957—primarily demonstrated how much the United States did not know about the Soviet nuclear program.⁵

During the remaining months of 1954, the Soviet atomic energy program continued to be a moving target for U.S. intelligence agencies and to provide more data through ongoing testing. On March 10, 1954, construction began on a gaseous diffusion plant, Combine 820, in Angarsk—about thirty miles north of the western tip of Lake Baikal. Then, on July 10, a government decree established the Naval Scientific Research Test Range, code-named Installation 700. Novaya Zemlya, a convex-shaped island over five hundred miles in length located between the Soviet mainland and the Arctic Ocean, whose southern tip is at the same latitude as Alaska, was chosen as the home for the new test site. In addition to substantial snowfall, the island also experiences winds of up to one hundred miles per hour, winter months of total darkness, and life-threatening cold weather. While the site’s stated purpose was to test the effects of nuclear weapons, primarily torpedoes, on naval equipment, its remote location made it a viable alternative site for high-yield tests that could cause significant environmental problems for the small towns and villages near Semipalatinsk.⁶

The initial test of 1954 was also the first Soviet one outside of Semipalatinsk. On September 14, as part of a field exercise, a modified version of the 40-kiloton RDS-3 bomb was detonated over the Totsk range, about one thousand miles to the northwest of the usual test site. Four further tests, in late September and early October, validated designs for tactical weapons, with yields ranging between 0.03 and 4 kilotons. Before the testing activity for October, and the year, was over, there would be another four tests. The first, of a device developed for the T-5 torpedo, failed. The most notable of the remaining three was the October 23 test of the RDS-3I bomb—an RDS-3 equipped with a neutron initiator. The modification was intended to allow about a 50 percent increase in yield. The bomb designers’ expectations were fulfilled when the device produced a blast equivalent to 62,000 tons of TNT.⁷

THE FIRST PART of 1955 was marked by the publication, within the intelligence community, of two studies that provided some indication of what American nuclear intelligence analysts knew or believed about the Soviet atomic energy program. One is the still classified national intelligence estimate on the Soviet program, prepared by the Joint Atomic Energy Intelligence Committee and released in April. That estimate, as the ones before it, was primarily a scientific and technical analysis of the Soviet ability to produce fissionable material, as well as the types of warhead sizes and yields the Soviets could turn out in the near future, and informed speculation on alternative stockpiles they could possess.⁸

The previous month, the Nuclear Energy Division in the Office of Scientific Intelligence had completed a revision of the atomic energy section of the National Intelligence Survey (NIS) on the Soviet Union. The NIS was a lengthy top-secret compendium of what the CIA knew about countries of interest. The seventy-seven-page treatment of Soviet nuclear activities included a discussion of the program’s history and growth, its current management structure, the personalities involved, research facilities, uranium mining, and the production of fissionable materials.⁹

The report reflected a combination of knowledge and ignorance, at least at the top-secret level. It provided a detailed account of the roles of the First and Second Chief Directorates in supervising the Soviet program—information that appeared to have come from knowledgeable sources. At the same time, the authors of the survey admitted that they had no current knowledge of the activities of the two directorates, or even if they still existed. There was also a description of the assorted laboratories and institutes, mostly in the Moscow area, as well as those created for German scientists.¹⁰

The list and commentary on ninety-six scientists filled up nineteen pages. Kurchatov, Khariton, Artsimovich, Flerov, Sakharov, and others who had played major roles in the Soviet development of atomic weapons were all included. Kurchatov was noted to be the head of Laboratory No. 2, and described as “probably the key scientist directing research on reactors and weapons for the atomic energy program.” That “probably” was one possible example of the limits of U.S. knowledge at the time. So was the lack of information on Sakharov’s role in designing the Layer Cake bomb. There were references to sites in the Urals and Siberia involved in uranium enrichment and plutonium production, but no specific locations were given. As the authors noted about the sites, “Important details remain undetermined.”¹¹

One explanation for some missing data was not a lack of knowledge but the level of classification of the survey. While the uninitiated might assume that there is nothing more sensitive than “top secret,” in World War II the United States began creating classifications that were, in effect, above top secret. Several of those, each designated by a code word, which was itself classified, covered intelligence obtained by communications intercepts. The authors of the survey relied on a number of different sources: returned German and Japanese prisoners of war, defectors, intercepted letters from German scientists, open literature, and espionage directed principally against uranium-mining activities. Communications intelligence was not one of them. But, by 1955, the United States had apparently discovered the existence of Sarov by detecting unusually heavy telephone traffic between the secret city and Moscow. Thus, the CIA knew more than it was telling the readers of the survey.¹²

But it is clear that some of the omissions were not due to restricted communications intelligence knowledge, but rather no knowledge at all. In April German scientists began returning from Sukhumi to East Germany. In 1951 the CIA, in anticipation of their return, initiated Operation Dragon. Largely relying on the analysis of mail intercepted by the Army Security Agency, the CIA began to prepare defection pitches targeted at the returnees. CIA attempts to induce Hertz, von Ardennes, Vollmer, and Max Steenbeck to defect failed, but some of their subordinates were more receptive. In addition, Nikholaus Riehl, upon learning that he could no more keep the money from his Stalin Prize than a runner-up on Jeopardy could keep his winnings, decided to take the CIA up on its offer.¹³

The agency knew from reading their mail that the defector returnees would not be able to offer fresh information, as they had not been allowed to work in the Soviet nuclear program for three years. While their Soviet masters had intended the cooling-off period to reduce any damage from defections or loose talk, “skillful and exhaustive interrogation . . . revealed technical details, individual names . . . in a richness unbelievable to one who has never witnessed this procedure,” according to Henry Lowenhaupt.¹⁴

The returnees were able to confirm the CIA’s supposition that the Germans working in the vicinity of Sukhumi had focused on isotope separation. Naturally, they were able to provide more detail on the von Ardenne group’s research into isotope separation, and the Hertz group’s investigation of a modified form of gaseous diffusion. Beyond that, the interrogations helped the CIA pin down the precise location of one of the secret cities in the Urals. The agency had concluded that there was a gaseous diffusion plant in the northern Urals, as indicated in the national intelligence survey. And it knew the existence of some sort of atomic facilities at Nizhnyaya Tura and Verkh Nevyinsk, but not which one was the diffusion plant. Several returning POWs had actually been to the plant but knew it only as “Kefirstadt,” a nickname derived from the name of the area’s favorite nonalcoholic beverage—kefir. The best guess was Nizhnaya Tura. But that guess turned out to be wrong. The new batch of defectors told the CIA that Kefirstadt was actually Verkh Neyvinsk. The agency had found Sverdlovsk-44.¹⁵

WHILE THE CIA was debriefing the Germans who had been sent home, Soviet scientists and technicians continued their work. In an eight-day span, beginning on July 29 and ending on August 5, the Semipalatinsk area experienced three more atomic tests. All were low-yield tests of the RDS-9 developed for the T-5 torpedo. Then, on September 21, another RDS-9 device was detonated. But this time, the test occurred about one hundred feet underwater, in Chernaya Bay, located near the southern end of Novaya Zemlya. In contrast to the July and August tests, the purpose of this one was not weapons development, but to determine the damage that could be done by a T-5 when its nuclear warhead exploded.¹⁶

Testing was not the only way in which the Soviet nuclear program had expanded by the end of September 1955. In April, concerned that a too heavy workload would result in a decline in the quality of the research and development being done at Arzamas-16, Soviet leaders established a second weapons design bureau. The Scientific Research Institute was built about fifty miles south of Sverdlovsk and twelve miles north of Chelyabinsk-40, in the city of Snezhinsk. Its first scientific director was K. I. Shechelkin, whose name and specialty (the combustion and detonation of gases), but not his role as Khariton’s first deputy, had been noted in the national intelligence survey. The secret city’s new designation was Chelyabinsk-70. Before the year was out, there would be further developments. In November the first plutonium production reactor at Tomsk came online, while a second heavy-water reactor would began operating at Chelyabinsk-40 in late December.¹⁷

The most important development of the year occurred sometime after Andrei Sakharov boarded a train at Moscow’s Yaroslavl station. He was accompanied by bodyguards from a special KGB detachment who had been assigned to him since the summer of 1954. They served as both guardians and watchers. No counterpart of Moe Berg was going to assassinate one of the Soviet Union’s key weapons designers, and that designer was not going to associate with “undersirables.” Sakharov apparently reached the test site in time to view a test on November 6. Although the blast equaled the force of 250,000 tons of TNT, it “made no special impression on me,” he recalled years later. Two weeks later, Sakharov was anticipating something special when a plane carrying a nuclear device in its bomb bay took off from an airfield near Semipalatinsk. The device was the product of what Sakharov would refer to as the “Third Idea,” and promised to give the Soviet Union a full-fledged thermonuclear capability, one that would pass scrutiny even before the Bethe Panel.¹⁸

The first proposal for a two-stage thermonuclear device had been made in 1953. Then, on January 14, 1954, Zeldovich and Sakahrov sent Khariton a memo titled “Concerning Utilization of the Gadget for Implosion of the RDS-6s Supergadget” that included a schematic of a different two-stage device and an estimate of its performance. The device would contain both an atomic bomb and a thermonuclear core, and the gases from the detonation of the atomic bomb would compress the thermonuclear core. There was no suggestion of creating a thermonuclear detonation through radiation compression, the Teller-Ulam idea being pursued in the United States.¹⁹

That concept of radiation compression came to Sakharov and others a few months later, in March and April. It followed the U.S. Bravo test on March 1, whose yield of 14 megatons made it very clear that U.S. weapons designers had developed a high-yield thermonuclear device. “Intensive analytical dissection and interpretation of all the available evidence,” presumably including Soviet analysis of the American debris, followed. The subsequent Soviet-accelerated research and design effort resulted in the technical specifications for an experimental two-stage, radiation compression, thermonuclear device being completed on February 3, 1955, with the device being designated RDS-37. By the time the theoretical and engineering work had been done, about thirty physicists, including Sakharov and Zeldovich, had made contributions, along with a team of mathematicians.²⁰

But Sakharov and the others at Semipalatinsk were to be disappointed that November day, not because the bomb did not live up to expectations, but because it was not dropped at all. The weather had changed after the plane lifted off. Low clouds made it impossible to use visual sighting to deliver the bomb or optical systems to monitor the detonation, so the test was aborted and rescheduled for a few days later. On November 22, Sakharov recalls an hour of tedious waiting, and then the announcement that the plane was over its target. That was followed by notification that “the bomb has dropped!” and “the parachute has opened!” The countdown finally reached the “five, four, three, two, one, zero stage.”²¹

Sakharov saw “a blinding yellow-white sphere expand, turn orange in a fraction of a second, then turn bright red and touch the horizon, flattening out at its base. Soon everything was obscured by rising dust which formed an enormous, swirling, gray-blue cloud.” It was not long before he “felt heat like that from an open furnace on my face . . . in freezing weather, tens of miles from ground zero.” Several minutes later he was jumping from the platform, as “the sudden shock wave was coming at us, approaching swiftly, flattening the grass.” The test, he would write, “crowned years of effort. It opened the way for a whole range of devices with remarkable capabilities.”²²

WHILE THE SOVIETS were busy upgrading their nuclear infrastructure and conducting further weapons tests, the United States was enhancing its detection capabilities on land and in the air. In March 1954, Col. Jack Gibbs, the deputy chief of AFOAT-1, was seeking guidance from the State Department’s Gordon Arneson concerning the possible establishment of a seismic station in Australia. Arneson suggested relying on the U.S. ambassador in Australia to raise the issue, notifying the United Kingdom of U.S. plans, and stressing that “the seismic installation in question is directed solely toward the Soviet Union and that the equipment involved would not be suitable for monitoring of any future British tests that might take place on or near the continent of Australia (emphasis added).”²³

In December, Colonel Gibbs again consulted the State Department, inquiring about the feasibility of conducting seismic surveys in southern Germany, Spain, Norway, and Sweden. The seismic stations that the United States was operating at Camp King, Germany (Team 313A), and Thule Air Base, Greenland, were not producing the expected results, and AFOAT-1 was seeking two replacement stations—one in Norway or Sweden to replace Thule, and one in Spain or Germany to replace Camp King. The following June, a letter from a State Department official to L. Corrin Strong, the American ambassador in Oslo, requested his views about how the Norwegian government would react to a request to permit a U.S. team to survey Norwegian territory in search of a suitable site. The letter explained that such a station would consist of four small huts to house the instruments and a central recording station, with the huts being spaced along a four-mile line. Two officers and twelve airmen, “who could be in civilian clothes,” would be needed to man the central recording station around the clock.²⁴

A more important issue for the ambassador was undoubtedly the U.S. desire to tell the Norwegians as little as possible about what the Americans would be doing, at least in part because of legal restrictions on the sharing of nuclear intelligence. Norwegians would not be involved in the operation of the station, nor would the government receive finished intelligence derived from the signals detected. The raw data obtained could be shared, but those signals would be “relatively meaningless until correlated with similar data from several stations.” The issue would come before the Norwegian Cabinet Defense and Security Committee that summer. After discussion, particularly of whether they should demand that Norway play a greater role in operating the station and have greater access to the data it produced, it was agreed to leave negotiations in the hands of the National Defense Research Establishment and the Intelligence Staff. As a result, on September 5, an American site survey team began its trek across southern and northern Norway, conducting a “granite reconnaissance” in search of a suitable site.²⁵

The approach to Australia was also successful. Sometime in 1955, early enough for the station to be operational before the year was out, the United States and Australia entered into a secret agreement to permit AFOAT-1 to establish a seismic station in the vicinity of Alice Springs, a small remote city in the Australian outback. The unit was designated Team 421 and the station code-named Oak Tree. By the end of 1955, with the Thule site having been closed in June, the seismic network consisted of nine fixed stations, a mobile team, one research station, and one standby station. Foreign sites included those in Australia (Alice Springs), South Korea (mobile), the Philippines (Clark Air Force Base), Germany (Camp King), and Turkey (Ankara). Within U.S. borders there were ground stations in Wyoming (at Pole Mountain, Douglas, and Encampment) and Washington (Larson Air Force Base).²⁶

AFOAT-1, also at the end of 1955, continued to have the benefit of reporting from its three electromagnetic pulse stations, eight acoustic stations operated by the Army Signal Corps as well as five stations operated by the Coast and Geodetic Survey. America’s nuclear test monitors were also getting a little help from British and Canadian friends. One British acoustic station, Tag Day, had operated throughout the year, while a seismic station, Beaver, had been activated during the year at American request. Another acoustic station had also been operating during 1955. Both Tag Day and Beaver, one of which was located in Pakistan, “lay in close proximity to . . . Semipalatinsk.” Canada furnished data from its own ground and aerial filter operations and conducted radiochemical analyses of debris and air samples collected by both Canadians and Americans.²⁷

Airborne debris collection remained a key element of the monitoring activity, and was performed by the 56th Weather Reconnaissance Squadron (WRS) at Yokota Air Base in Japan, and the 58th WRS at Eielson Air Force Base in Alaska—along trajectories now labeled Buzzard Delta and Loon Charlie. Special flights were also flown by the 53rd WRS at Burtonwood, England, and the 55th WRS at McClellan Air Force Base in California. To improve their collection capabilities, that August the Air Weather Service began replacing WB-29 aircraft with the more modern WB-50, which could collect debris samples at thirty thousand feet, five thousand feet higher than the WB-29. Concern that debris from some atomic bombs might be carried in fast-moving airstreams at very high altitudes, with little fallout to the altitudes patrolled by the WB-50, led to the use of T-33 jet trainers for sampling operations, starting with a July 12 mission out of Yokota Air Base in Japan. The T-33 would not be the only aerial supplement to the WB-29s and WB-50s.²⁸

GIVEN THE NUMBER and diversity of Soviet tests in 1955, the ground and aerial collection systems operated by the United States and Britain received a good workout, confirming the utility of the multiple means of detection. On July 29, the date of the first Soviet test of 1955 (Joe-15), the Ankara station as well as two additional acoustic sites detected signs of a possible low-yield explosion, in the area of 5 kilotons. Analysis of aerial samples confirmed that despite initial British skepticism due to the failure of its Pakistan station to detect any sign of a test, an atomic blast had been detected. There was no disagreement between the United States and Britain over the reality of Joe-16, the 12-kiloton test of August 2. That morning two British and two American stations detected what appeared to be an explosion of some sort, first picked up by acoustic sensors and seismic signals. That the event was an atomic test was subsequently confirmed by aerial debris, although the debris was apparently not suitable for extensive analysis. In both cases the initial estimates of yield overstated the power of the blasts. The yield of the second test was estimated at between 15 and 60 kilotons, with 30 kilotons being most likely, a figure 250 percent greater than the actual yield.²⁹

The shift in test sites from Semipalatinsk to Novaya Zemlya did not prevent either U.S. or British acoustic and seismic stations from detecting Joe-17, an underwater test on September 21. Conclusions as to the time and day of the test were based on seismic data, while the judgment that the explosion probably occurred underwater was the result of examining both seismic and acoustic data. U.S. analysts concluded that since there was a large discrepancy between the strength of the seismic signal and that of the acoustic signal, the explosion was probably “well tamped” as the result of having taken place underwater. They noted that at the apparent location of the test the depth of the water was between one hundred and two hundred feet. Once again, they significantly overestimated the yield of the 3.5-kiloton blast, concluding that the “true yield [is] on the order of 20 kilotons.”³⁰

The detection system missed one test entirely, a low-yield detonation that occurred on August 5, though it did not miss the final two Soviet tests of 1955. The first of those tests did not come as a surprise to Britain, nor presumably the United States. There had been evidence for two months, minister of defense Selwyn Lloyd informed British prime minister Anthony Eden on November 7, that “the Russians were preparing to carry out a large scale nuclear explosion.” At dawn of the previous day British stations in the United Kingdom and Pakistan detected the large explosion. Varied types of U.S. stations also detected the blast and contributed in different ways to characterizing it. Electromagnetic data provided the time; seismic data, the location (near Semipalatinsk); and acoustic data, the best estimate of yield—200 kilotons, an estimate that was reasonably close to the actual 250-kiloton yield.³¹

Analysis of the acoustic data obtained from the thermonuclear test of November 22, Joe-19, also helped produce an estimate—1.7 megatons—that was very close to the actual yield of 1.6 megatons. Other detection techniques allowed the U.S. analysts to go beyond simply announcing that a test had taken place. Radiochemical analysis of the debris confirmed the thermonuclear nature of the test. Electromagnetic data established the time, while seismic data pinpointed the location. The height at which the bomb detonated was estimated through analysis of seismic signals and the debris.³²

Debris from one or both of the November tests, most probably the November 22 blast, was also collected by modified B-36 aircraft, which sampled some of the gases released during a nuclear explosion, at altitudes up to forty thousand feet. Missions using the B-36 were dubbed Sea Fish. The first was flown from Guam by the 6th Bomb Wing, while the second took off from Fairchild Air Force Base in Spokane. Both missions succeeded in bringing back debris for analysis.³³

THE PROGRAM to determine the amount of krypton-85 being emitted by Soviet plutonium production reactors into the atmosphere, designated Music, continued in 1955, but was also a source of contention with America’s closest nuclear intelligence ally, Great Britain, for the United States was more content with the utility of the program than the British were. In a 1953 estimate on Soviet capabilities for attacking the United States, the CIA had claimed that while estimates of Soviet U-235 production were subject to large uncertainties, the estimates for plutonium production were “reasonably firm.” Even so, the authors suggested that the numbers they provided for the Soviet stock-pile for the three years beginning in mid-1953 (120, 200, 300) could be as much as one-third less than the actual numbers, or twice as high, giving the “uncertainty in the evidence concerning the production of fissionable material” as their prime reason.³⁴

But the British were also skeptical of the reliability of the Music effort, and annoyed at American postponements of meetings to discuss the issue. A conference had originally been scheduled for October 1954, and then put off until January 1955. When that meeting was also postponed, and the Americans suggested rescheduling for June, a senior British official wrote in a confidential memo that “there are many difficulties in making the assessment, for example how much gas is produced, and what happens to it, from U.S. and U.K. production, how much is produced, and what happens to it, from U.S. trials, and how the gas is distributed in the atmosphere.”³⁵

He continued that “this is a sorry tale. We ourselves need to take decisions urgently to continue or close a number of our overseas stations engaged in this programme. We have considerable doubts whether the programme continues to be worthwhile in view of the possible margins of error in the assessments resulting from it, but the assessment of the Russian plutonium stock is fundamental to intelligence on their total weapons supply.”³⁶ Eventually, the differences between the allies with regard to estimating plutonium production would be resolved, but it would take several more years.³⁷

OVER THE COURSE of 1956, there were eight tests at the Semipalatinsk test site, including two thermonuclear ones with yields of 900 kilotons. Meanwhile, the United States would seek to extend its network of ground stations that could provide data on the occurrence, location, and yield of such tests. The first overseas electromagnetic pulse station, apparently located in Pakistan, began operations in August 1956. Reestablishing a nuclear detection capability in Iran, involving either seismic or electromagnetic detection, was also discussed. While the 1955 site survey in Norway led to the conclusion that none of the sites were suitable for a seismic station, in 1956 the possibility of establishing an electromagnetic pulse facility in the vicinity of the Rygge Air Base southeast of Oslo became a subject of discussion between the two governments.³⁸

But the most important development that year in the U.S. quest to monitor Soviet nuclear activities involved an aircraft whose nuclear intelligence mission was not its most important. On July 4, Hervey Stockman took off from Wiesbaden, Germany, in a plane designated the U-2, which he guided over Poznan, Poland, before heading for Belorussia. The next leg of his mission included a turn north to take him over Leningrad, where Soviet submarines were being built in the city’s shipyards. He also overflew several major military airfields before heading back to Germany.³⁹

The concept for the aircraft Stockman was flying originated with Kelly Johnson of Lockheed Aircraft, and was strongly supported by key presidential advisers Edwin Land and James Killian. Their support led to a program code-named Aquatone and managed by the CIA’s Richard Bissell, a special assistant to agency director Allen Dulles. The plane produced by Johnson and his staff at the secretive Lockheed “Skunk Works” in Burbank could fly at 70,000 feet, at a speed of 500 knots, to a range of 3,000 nautical miles, with a pilot as the lone crew member. The special focal-length camera it carried could photograph objects as small as a man within a strip 200 miles wide by 2,500 miles long.⁴⁰

Stockman’s mission, the first for a U-2 over Soviet territory, was followed by another four overflights within a week. Among the targets were Moscow, a bomber test facility at Ramenskoye airfield outside of the capital, bomber bases, the Kaliningrad missile plant, the Fili airframe plant, and the Khimki rocket-engine plant. Those flights were followed by a hiatus, decreed by President Dwight Eisenhower in response to unexpected Soviet protests of the overflights, which the CIA had not expected to be detected. It would be November before the next U-2 lifted off on a mission that would take it over Soviet territory, and even then it was under presidential orders to stay as close to the border as possible, a flight path that precluded photographing possible atomic installations located in the center of the country or in Siberia.⁴¹

U-2 activity in the summer of 1957 would be much more extensive, with targets well within the interior of the Soviet Union and China. Flying from Incirlik Air Base in Turkey, Lahore Air Base in Pakistan, and Atsugi Air Base in Japan during a twenty-three-day period in August, the spy planes conducted Operation Soft Touch—seven overflights of the Soviet Union and two of the People’s Republic of China. Those overflights would substantially add to the U.S. intelligence community’s knowledge of not only Soviet military forces and industrial capability but also its nuclear facilities.⁴²

Soft Touch targets included the Soviet space launch facility (later known as Tyuratam), the antiballistic missile radar test site at Sary Shagan, and the aircraft construction facilities in Omsk and Novosibirsk, as well as nuclear installations selected by the Ad Hoc Requirements Committee, responsible for selecting targets for U-2 missions, in late May. The most important central Asian and Siberian atomic targets were near Krasnoyarsk, Semipalatinsk, and Tomsk. In July, Henry Lowenhaupt was instructed “to work up target briefs for all atomic targets in the enormous geographical area of central Asia and Siberia.” Because of the secrecy surrounding the U-2 program, Lowenhaupt did his work in the “Blue Room,” a small centrally located secure area that was actually painted light green.⁴³

Lowenhaupt’s work involved selecting a flight path that would allow photography of the highest-priority targets as well as coverage of as many of the lower-priority targets as possible. Those lower-priority targets included the uranium concentration plants in the Fergana Valley, including one near Taboshar, and another near Andizhan—facilities that had been located thanks to a 1947 refugee who had driven a bread truck to each of them. The job required familiarity with the defector, refugee, espionage, and technical intelligence reports.⁴⁴

In putting together the target brief for Krasnoyarsk, Lowenhaupt had, in addition to the the reports from Icarus, the testimony of a German prisoner of war who had spent several years there as a construction worker. Imagery was also recovered, earlier in the year, from the short-lived Genetrix balloon program, which commenced and ended in 1956 and had involved camera-carrying balloons sailing over Soviet territory. The few pieces of hard intelligence actually produced by the program included photographs of the Krasnoyarsk area, which showed an enormous construction effort—new apartment houses, laboratories, warehouses, machine shops, and a vast mining enterprise. Higher-resolution U-2 images, it was hoped, would reveal more information about what was actually going on at the site.⁴⁵

Although the Semipalatinsk test site was also on the target list, there was considerable uncertainty about its precise location. To come up with specific coordinates for the U-2 pilot to overfly, Lowenhaupt asked Donald Rock, Northrup’s deputy at AFOAT-1, to compute the average of the seismic epicenters of the five highest-yield detonations in the target area. The result was “a spot in the featureless desert some seventy miles due west of Semipalatinsk,” Lowenhaupt recalled.⁴⁶

But the primary nuclear target was the Siberian Chemical Combine, Tomsk-7, located nine miles northwest of Tomsk. While Kurchatov knew exactly where it was and what went on there, Lowenhaupt did not. He had only a minimal amount of information about its location and purpose, owing to the city’s closure in 1952 and the inability of attachés to photograph it from the Trans-Siberian Railroad. While there had been a number of comments in intelligence reports suggesting the presence of something related to atomic energy in the vicinity of Tomsk, in 1957 “the resulting evidence . . . was all contained in just three reports, two of which did not inspire much confidence, and the analysis of a fur hat,” Lowenhaupt would recall a decade later.⁴⁷

The latest report conveyed the comments of a German claiming to have been employed as a blacksmith in Tomsk in 1955. He had told his army interrogator that the local residents had suggested, tongue in cheek, that “Atomsk” would be a more appropriate name for the city, and that there was an underground secret plant to the northeast of the Tomsk railroad station.⁴⁸

Earlier, another returned German informed a British interrogator that he had heard of an industrial unit that “manufactured fillings for atomic weapons locally known as the Post Box,” and had seen a large building in Tomsk with all its windows barred and a large sign identifying it as “Information Office, Personnel Department, Post Box.” Subsequently, he mentioned that while riding a bus, he had seen railway trains carrying coal, wood, and building materials entering the the closed area. He also reported that from a distance of about four to five miles north of Tomsk, he could see three large chimneys emitting smoke. His interrogator observed that the source was plagued by a very poor memory, appeared to be suffering from some kind of mental problem, and was preoccupied with emigrating to Canada.⁴⁹

A more persuasive report came from another German, a returned prisoner of war. In 1949, he had been employed as a tailor in a small factory northwest of Beloborodova, about seven miles north of Tomsk. During his interrogation by air force intelligence officers, he claimed that in April or May 1949, about twelve thousand penal laborers were put to work in a fenced-off area between his factory and the village of Iglakovo, along the Tom River. Among his customers were military officers of the construction staff that managed the project who came to get their uniforms properly fitted. The tailor’s Soviet supervisor told him that the officers would be supervising the construction of an atomic energy plant.⁵⁰

A CIA officer, John R. Craig, also obtained physical evidence in the form of a fur hat that had been worn by one of the Germans who had recently lived in Tomsk. The hat was turned over to AFOAT-1 and the Atomic Energy Commission. Analysts concluded that its exterior contained fifty parts per billion of uranium that was, without a doubt, slightly enriched in U-235. They were able to eliminate the possibility that the uranium came from fallout or from a reactor. Tests for plutonium, radioiodine, and separated lithium all came back with negative results. The hat was consistent with the hypothesis that somewhere in the Tomsk area, there was a uranium enrichment plant. While the U-235 may also be residue from fabrication of nuclear warhead components, the size of the atomic facility seemed, to Lowenhaupt, too large for that to be the case. It was also unlikely to be a reactor with an associated chemical plant or a lithium isotope separator. Lowenhaupt thus classified the target as a uranium separation plant and centered it on the spot where the German tailor had seen twelve thousand prisoners go to work.⁵¹

The Soft Touch missions began on August 4, and the missions of August 21 and 22 brought back key imagery of the Soviet nuclear targets. On August 21, pilot Sammy Snyder lifted off from Lahore Air Base and piloted his spyplane over part of the Semipalatinsk test facility and eventually on to Tomsk. On August 22, James Cherbonneaux also left Lahore and flew directly over the testing ground, and discovered why he had been given those coordinates. On several occasions he had overflown the U.S. test site in Nevada and was able to recognize a test site when he saw one beneath him—large circular areas that had been cleared and graded, with paved support roads connecting to distant block houses. A slight adjustment in his course took Cherbonneaux over the center of a cleared area, where he noticed a large isolated shot tower, with a nuclear weapon “cab” in the center. At one of the block houses, close enough to witness a blast but not suffer from it, he detected a number of parked vehicles. It occurred to Cherbonneaux that the tower might be holding a nuclear bomb about to detonate.⁵²

Upon returning to Lahore, the relieved pilot told his debriefers what he had seen, but found his tale greeted with skepticism. While they were reluctant to credit Cherbonneaux’s observation skills, they did believe the images he brought back, as well as the report they received within twenty-four hours that less than four hours after his overflight the eighth Soviet test of 1957, Joe-36, had occurred. The photographs showed the bomb and the aircraft that was used to drop it. They also showed evidence of a recent low-yield test.⁵³

Two of the Soft Touch nuclear targets—Kyshtym and Krasnoyarsk—hid under cloud cover while U-2s were overhead. But Sammy Snyder brought back images of sufficient value to impress Allen Dulles and be shown to Eisenhower. The tailor’s information about location was as accurate for Tomsk as Donald Rock’s had been for Semipalatinsk. Allen Dulles is reported to have said, “You mean you really did know that something atomic was going on ’way out there in the wilds of Siberia!” Of course, knowing that something is going on is quite different from knowing exactly what is going on. The clear, vertical photographs of Tomsk provided plenty of surprises to the photointerpreters and nuclear intelligence analysts at CIA. The photographs revealed, in addition to the expected U-235 separation plant, a plutonium-producing reactor area and a plutonium chemical separation facility in midconstruction.⁵⁴

The extent of the atomic enterprise at Tomsk-7 was clear from the mission report, which noted that the installation “covers an irregular shaped area of about 40 square miles on the right bank of the Tom River. No single energy complex in the western world includes the range of processes taking place here.” It went on to specify that “on the west edge of the area, a large thermal power plant with an estimated capacity of 400 megawatts is undergoing further expansion. . . . East of this plant is . . . the feed and production section and gaseous diffusion plants. One gaseous diffusion building is uncompleted. On the east edge is . . . the reactor area. One of the two reactors appears to be in the final stages of construction. . . . On the northeast edge a plutonium chemical separation area is uncompleted.”⁵⁵

Full “exploitation” of the Tomsk imagery was not an overnight process and involved far more than simply identifying the key facilities of the complex. While obtaining the imagery was the dangerous part, turning it into finished intelligence was the hard part. It took Richard Kroeck, from the agency’s Photographic Intelligence Center, a full five months to complete his interpretation of the photographs that were snapped in a matter of minutes. Working along with Kroeck was the OSI’s William F. Howard, who directed the complementary engineering analysis. The combined efforts provided the agency with an evaluation of not only what facilities could be found in those forty square miles, but also what was inside them, their interrelationships, and how it all worked.⁵⁶

THE NUCLEAR WEAPON CAB that Cherbonneaux had seen during his overflight was for a low-yield device that was tested on September 13, the third test since his mission. The Soviet test program for 1957 concluded in late December after sixteen tests, the most powerful being one of the four conducted on Novaya Zemlya—a 2.9-megaton blast on October 6. That Arctic island was the site of five more tests during the first three months of the new year, with yields ranging from 40 kilotons to 1.5 megatons. Although Semipalatinsk could not compete with its northern neighbor in terms of the size of its blasts, it doubled the frequency of testing, with ten between the first of January and March 22.⁵⁷

The fifteen tests in early 1958 provided the usual variety of signals and debris associated with such events. A significant contribution to U.S. understanding of the Arctic tests was made by a detachment from the Strategic Air Command (SAC). SAC had tried to usurp the CIA’s control of the U-2 program and came away with the responsibility for noncovert overflights. The detachment flew specially modified U-2s, designated U-2A-1s, from Eielson Air Force Base in Alaska as part of Operation Toy Soldier. The modifications included a new hatch for the equipment bay, which contained a device to gather gaseous samples and store them in six spherical, shatterproof bottles. The hatch also carried an air scoop connected to a filter paper system, which allowed four filter papers to be rotated, placing a fresh one in front of the airduct at appropriate times. The U-2 itself had an advantage over other planes in that it could operate at substantially higher altitudes, where the debris from thermonuclear, megaton explosions rose and where the winds were minimal and the airflow more stable.⁵⁸

The deployment to Eielson began on January 30, with the first three of the ten modified U-2s assigned to SAC. Almost everyday over the next two months its pilots flew “long and boring” missions that frequently lasted eight hours or more, taking their planes far to the north, over Point Barrow, where the fallout from the Arctic tests usually appeared within a day. The SAC U-2 flights from Eielson were supplemented by CIA flights, also involving modified U-2s, along a track north from Atsugi.* Initially, the samples brought back to Eielson were sent to the AFOAT-1 laboratory at McClellan Air Force Base, but eventually Eielson had a lab of its own, since the very short half-lives of some of the radioactive samples demanded quick analysis.⁵⁹

Within a month after the March 22 test, another station joined the growing network of U.S. and allied ground stations dedicated to monitoring Soviet testing. Near the end of 1957, the United States had made another approach to Norway, this time proposing establishment of an electromagnetic station, in the form of a ten-ton trailer requiring between six and ten operators, to help monitor the Arctic tests. Within a few months the allies were able to agree on establishing a Norwegian-manned, American-financed station at Hoybuktmoen. The two nations agreed to pretend that the station, code-named Crock-Pot, was a “weather research station.” On April 18, it became an around-the-clock operation.⁶⁰ But it would be a while before that monitoring station or any other detected a Soviet test.

On the last day of March, Soviet foreign minister Andrei Gromyko appeared before the Supreme Soviet and proposed that in order “to contribute in every way to the great goal of mankind’s deliverance from the threat of atomic war . . . the Soviet Union should unilaterally cease the testing of all kinds of atomic and hydrogen weapons.” In addition, he challenged the United States and Britain to do the same. Igor Kurchatov also addressed the legislative body, complaining that during World War II, America and Britain had conducted their work on the atomic bomb “under conditions of most strict secrecy and did not help us,” resisting any temptation he might have had to add “at least not intentionally.” He asserted that the casualties at Hiroshima and Nagasaki were needed by “American military politicians . . . to begin a campaign of unparalleled atomic blackmail and cold war against the USSR.” He also offered reassurance that the Soviet Union possessed the means to detect “distant explosions of atomic and hydrogen bombs.” Not surprisingly, the rubber-stamp legislative body approved a resolution declaring a moratorium.⁶¹

Soviet tests would not resume until September, but through the summer several U.S. analysts continued trying to unravel the Soviet atomic energy infrastructure, relying, rather significantly, on photographs provided by the secretive Soviets themselves. One of those analysts was OSI’s Charlie Reeves, an MIT graduate who had worked his way through school as a heavyweight boxer and had been recommended to the CIA by the president of the New England Electric Power Company because of his professional experience and linguistic talent. In August 1958 Reeves was faced with the task of assembling data on suspected Soviet atomic energy facilities in the Urals, as a means of estimating the consumption of electric power at those sites. Such estimates could be used to gauge the production of fissionable materials, since a plant’s production of such materials was directly proportional to the amount of power consumed.⁶²

Reeves started with a single picture of the Sverdlovsk Central Dispatching Station of the Urals Electric Power System that appeared in the July issue of Ogonek, the Soviet equivalent of Look. The picture of the inside of the dispatching station showed what looked to Reeves like a schematic diagram of major power plants, with their transmission lines and user substations—all the information required for control of the entire Urals electric system. The CIA analyst then examined at least 103 articles in Soviet newspapers and technical journals, four reports of visits by delegations, eleven POW returnee reports, approximately twenty-five local photographs, as well as some of the photographs that had been obtained by the Genetrix program. Among the items consulted were the December 1948 issue of Elektricheskiye Stantsii, which contained a short report of a Moscow conference on a planned expansion of power in the Urals, and a 1958 book celebrating the fortieth anniversary of electric power in the Urals.⁶³

As a result, by April 1959 Reeves was able to map out the power distribution network in the Urals and determine the approximate power supplied to three of the Soviet Union’s most important facilities for producing fissile material. The U-235 production plant at Verkh Neyvinsk (Sverdlovsk-44) received 850 to 1,000 megawatts; the plutonium reactor at Kyshtym (Chelyabinsk-40), 105 to 195 megawatts; and the unidentified complex near Nizhnyaya Tura (Sverdlovsk-45), up to 100 megawatts.⁶⁴

Another Soviet photo proved vital in enhancing U.S. understanding of the Tomsk-7 reactor, although this one was somewhat more difficult to obtain than the one that appeared in Ogonek. The same month that the Soviets resumed testing, they also participated in the two-week-long Second Conference on the Peaceful Uses of Atomic Energy in Geneva, Switzerland. Among the featured topics was controlled fusion research; among its themes was collaboration, with U.S. and Soviet scientists calling for world cooperation “to unravel the mysteries of fusion power for peace.”⁶⁵

Hoping to derive some useful intelligence about Soviet atomic energy developments, Charles Reichardt, the AEC’s director of intelligence, went to the conference. From his temporary office at American delegation headquarters, he was prepared to serve as a liaison between intelligence officers and scientists attending the conference and to support intelligence collection activities, whether overt or clandestine. Henry Lowenhaupt was also there to assist, with files on the Soviet personalities attending, a list of what the Soviets had published on nuclear energy, and his memories of what U-2 imagery had shown about the nuclear facilities in Siberia.⁶⁶

Opportunity arrived about halfway through the conference when the Soviets announced that an atomic power station “somewhere in Siberia” had just begun operations. The chief Soviet delegate, V. S. Emelyanov, told his audience that one of the six natural-uranium reactors planned for the station was already operating, but declined to pinpoint the location. When a reporter asked where to send a letter to plant workers, he offered to mail the letter himself. The announcement was followed by visual evidence, in the form of a seventeen-minute color film and related exhibit in the conference exhibition hall. The description of the facility as shown in the movie led Lowenhaupt to believe it could well be the one north of Tomsk that had shown up on U-2 photographs from August 1957.⁶⁷

But the intelligence that could be derived from having a copy of the film was far greater than what could be obtained simply by viewing it. The Soviet delegation, however, would not let a copy out of its possession, particularly since I. I. Rabi, the head of the American delegation, had already promised the Soviets copies of all the U.S. movies shown at the conference. To get the most out of viewing the film, OSI’s John R. Craig recruited a group of reactor design engineers, whose firm employed them to evaluate Soviet reactor engineering practices for the OSI, to watch the film, with each engineer looking and listening for specific items. They also proposed taking still photographs of the movie, using two Leica cameras and the very high-speed film Lowenhaupt and Craig had brought along.⁶⁸

The results included detailed notes on the content of the film, along with photographs that turned out to be vital to subsequent analysis. Information was also gathered by non-AEC American delegates, some of whom had been asked by the CIA’s Domestic Contact Service to inquire about specific subjects. Others were assigned specific “situational gambits” devised by an air force intelligence officer who was in contact with the CIA representatives at Geneva.⁶⁹

Back in Washington, OSI’s Frank McKeon began the analysis by comparing the photographs from Geneva with U-2 images of the Tomsk reactor. The objective was to understand the internal workings of the reactor in order to estimate how much plutonium it could produce—the plutonium that would go into Soviet nuclear warheads. Attaining that understanding required obtaining data on a multitude of items, including the number of fuel rods, reactor dimensions, turbine performance, and the flow of cooling water through the reactors.⁷⁰

Producing such data demanded the application of an array of interpretation and research skills. In late October the agency’s Photographic Intelligence Center contributed a one-page brief interpreting the content of the motion picture photography showing the outside of the reactor building. The center, working with photographs of the building’s interior, was also able to provide measurements of the size of the reactor as well as the size of the blocks containing the fuel rods. The latter information, when combined with McKeon’s conclusion that there were 2,100 fuel rods, led to the assessment that the reactor was graphite-moderated (since the average space between the 2,100 rods when placed in the twenty-six blocks that contained them averaged 8.5 inches—close enough, given measurement limitations, to the 8 inches expected for a graphite-moderated natural-uranium reactor).⁷¹

Charlie Reeves relied on his ability to read technical Russian fluently, as well as his research skills and his five-shelf library, to provide answers needed about the facility’s turbines. Days were spent looking at engineering drawings of Russian turbines for a match to the ones shown in the Geneva photography. Neither the Leningrad VK-100-2 nor the Leningrad SVK-150 MW, a picture of which he found in his personal copy of Energetecheskoe Stroitel’stvo SSSR Za 40 Let, were it. But Elektricheskiye Stantssi, again, had the answer. This time it was the November 1957 issue. On page 46, he found the matching turbine and its specifications, which permitted an estimate of the power level of the reactor.⁷²

All the intermediary findings led to the conclusion that the reactor would produce a small amount of electric power and a large quantity of plutonium. The analysis of the Geneva photography not only allowed a more detailed assessment of Tomsk, but also became, Lowenhaupt recalled, “the key to understanding Russian facilities for the production of plutonium for nuclear weapons.”⁷³

DURING THE SUMMER of 1959 the CIA continued its attempts, on the ground and in the air, to acquire more information on the Soviet nuclear program. No information was available from collection of debris, since the Soviets had followed President Dwight D. Eisenhower’s declaration of a testing moratorium, which commenced after a U.S. test on October 31, 1958, with another of their own. Eisenhower hoped that the United States, the Soviet Union, and Britain, which also ceased testing, could negotiate a test ban treaty. The moratorium would continue until early September 1961, when the Soviets resumed testing, followed later that month by the United States.⁷⁴

But there were other means of uncovering Soviet nuclear activities. On July 8, Allen Dulles and Richard Bissell met with President Eisenhower. When they left the White House that day, they had his approval for Operation Touchdown. The U-2 mission would depart from Pakistan and, after completing its mission, land in Iran. Its targets included the test ranges at Tyuratam and Sary Shagan, the Semipalatinsk proving ground, Dolon airfield and its Bear bombers, and the never-before-photographed Kyshtym.⁷⁵

At six o’clock the following morning, pilot Marty Knutson took off from Peshawar Air Base, and while his U-2 was periodically detected by Soviet air defense radars, he was able to complete his mission and land in Iran. The imagery from the mission provided analysts of Soviet missile programs with valuable information, as it revealed the existence of a second launch complex under construction at Tyuratam. But the agency’s nuclear intelligence analysts in OSI were disappointed. The key areas of interest at Kyshtym were almost entirely obscured by clouds.⁷⁶

That summer Raymond Garthoff was working in the agency’s Directorate of Intelligence. He had joined the agency in late 1957 and was assigned to the Office of National Estimates. His graduation from Princeton in 1948, shortly after his nineteenth birthday, was followed by graduate studies at Yale, where he learned to speak and read Russian. While at Yale he also served with the 469th Strategic Intelligence Research and Analysis Team, a military reserve unit that engaged in library research on aspects of Soviet war potential. Before the end of 1949, with his master’s degree in hand, and his doctorate just a couple of years away, he was looking for a job. His search yielded three alternatives: teaching at Yale, working for the CIA, or joining the RAND Corporation. RAND was the winner, but by 1957 he was ready to join the CIA.⁷⁷

On July 23, 1959, Garthoff arrived in the Soviet Union, part of the contingent touring the country with Vice President Richard Nixon. Garthoff’s job was to serve as an interpreter for Vice Admiral Hyman Rickover, the head of the navy’s nuclear propulsion program. But Garthoff came equipped with more than language skills. He spotted and photographed a large, and previously unidentified, munitions storage center near Novosibirsk. Near Sverdlovsk he noticed and covertly photographed from his airplane seat two new SA-2 antiaircraft missile sites. He also scooped up samples of soil and water from Novosibirsk, Beloyarsk, Sverdlovsk, and Pervouralsk, samples that he recalls as “contributing to our understanding of the pattern of Soviet nuclear activities in those key and inaccessible regions.”⁷⁸

Shortly after returning, he was asked by Sherman Kent, the head of the Office of National Estimates, if he would do it again. This time he would be interpreting for John McCone, the chairman of the AEC, who would shortly be leading a high-level delegation to visit a number of Soviet nuclear facilities. Garthoff agreed, unaware at the time that Soviet intelligence had discovered his position with the CIA, thanks to a Swedish officer working for the GRU.⁷⁹

Part of his trip with McCone, which began on October 9, involved personal encounters with some of the key members of the Soviet atomic weapons program, including Igor Kurchatov at the Institute of Atomic Energy and A. I. Alikhanov at the Institute of Theoretical and Experimental Physics. The delegation also visited a uranium mine and uranium concentration and processing plant in the Ukraine. McCone, Garthoff, and other members of the delegation were first flown to Dnepropetrovsk, which was followed by a 120-mile drive to a uranium mine near Pervomaisk, and then to the processing plant at Zheltye Vody. To prevent the Americans from tracing their trip and determining exactly where they had been taken, the drive had been a long and circuitous one. It succeeded in confusing the visitors but failed in its objective. After arriving in the village near the uranium mill, they encountered some of the village’s inhabitants, including some boys, about twelve years old, whom the KGB had apparently failed to brief. When Garthoff asked where they were, the boys promptly told him “Zheltye Vody.” Their official hosts did provide detailed information on ore concentration, the separation process, and the purity of the product. But they had nothing to say about the current or cumulative quantity of production or about the mine’s production as a percentage of total Soviet production.⁸⁰

Despite the extra attention from the delegation’s Soviet security escorts, Garthoff still managed to bring back some photographs. He came equipped with several cameras, including a CIA camera with fast black-and-white film as well as his personal camera with color film, both of which he used “liberally.” He later recalled that his photographs of the mine tailings taken from ground level “complemented very well overhead [that is, U-2] photography.” The two sets of images permitted a more accurate determination of the height of the tailings, and were valuable in allowing analysts to estimate the mine’s cumulative production. Appreciation for his work would extend not only to the head of the Office of Central Reference, who wrote a glowing memorandum to Garthoff’s boss at the national estimates office, but also to agency director Allen Dulles, who invited him to a luncheon at his Georgetown home.⁸¹

THE SUMMER OF 1959 also marked a change in name for the Washington headquarters and field units of the air force’s nuclear monitoring organization. On July 7, AFOAT-1 became the Air Force Technical Applications Center (AFTAC) and the 1009th Special Weapons Squadron became the 1035th Field Activities Group (1035th FAG). AFTAC also had an unclassified nickname—Project Clear Sky. However, its mission, as was AFOAT-1’s, was officially classified.⁸²

The name change was prompted by a presidential award Doyle Northrup received in January 1959 for his work at AFOAT-1. Northrup would recall in an oral history interview years later that the “citation was so directly exactly what I had been doing—and it had always been classified secret—that the Air Force practically went into a tizzy.” Among those most upset was Maj. Gen. Jermain Rodenhauser, who had assumed command of the detection organization the previous August. According to Northrup, Rodenhauser went to the air force’s chief of staff and said, “This is a terrible security breach, and what we had better do is change the name.” So AFOAT-1 became AFTAC, a change Northrup described as “really comical,” explaining that he had “a list of all the [unclassified and easily available] Department of Defense telephone directory listings for all those years and it shows D.L. Northrup, Technical Director, Don Rock, Assistant Technical Director . . . and all of a sudden those people changed from AFSMW-1 to AFOAT-1 to AFTAC and that’s supposed to fool somebody.”⁸³

At the time of the name change, AFTAC was concerned, even more than usual, with increasing its ability to detect nuclear detonations through means other than air sampling—because of the prospect of a prolonged atmospheric testing halt. Late the previous August, in 1958, President Eisenhower had proposed negotiations among the United States, Britain, and Soviet Union to permanently end nuclear testing. He also announced that the United States would halt testing for a year from the time negotiations began. The Soviets soon agreed to negotiations beginning on October 31. With that date looming, all parties began a round of tests. The Soviet test series, which started on September 30 and ended on November 3, included twenty-one tests. Of the nineteen at Novaya Zemlya, six had yields of over 1 megaton.⁸⁴

During, and shortly after those tests, AFTAC had improved its ability to detect and gather intelligence on them. By the end of 1958, the B-36s that had been used to conduct the Sea Fish high-altitude gas-sampling missions had been replaced by jet-powered B-52s. In addition, the T-33 sampling aircraft operated by the Alaskan Air Command and Pacific Air Forces was replaced with a RB-57, providing an additional sampling capability above fifty thousand feet. December also marked the commencement of seismic detection operations at Pinedale, Wyoming, and May 1959 saw the opening of a seismic station at Flin Flon, Canada—although the Crock Pot station in Norway had been closed the previous month due to its failure to produce the information expected.⁸⁵

Throughout 1959, AFOAT-1/AFTAC was laying the groundwork for increasing its capabilities, both by expanding the number of stations using already developed techniques and by forging ahead with experimental work to validate new techniques. AFTAC would inform the State Department’s special assistant for atomic energy of its interest in conducting a site survey in a particular country, which might be followed by a letter from the special assistant to the American ambassador asking his opinion on the wisdom of making such a request to his hosts.

In March 1959 the special assistant was Philip J. Farley, who notified John J. Muccio, the ambassador to Iceland, of “Air Force” interest in possibly establishing an electromagnetic station in the Langanes area of Northwest Iceland to improve coverage of “atomic events in the northwest quadrant of the USSR.” Later in the month, an AFOAT-1 inquiry into the feasibility of establishing a detection station in East Pakistan, complementing the one in Lahore, brought back the response that there were no U.S. forces in the area to provide cover. Also rejected in the region were Cambodia, Laos, Malaya, Burma, and Singapore. South Vietnam was considered unlikely owing to the existing ceiling on military personnel. That left Thailand. At the end of the year the special assistant’s office reviewed the possibility of seismic, acoustic, or electromagnetic stations in Thailand, Ceylon, and Ecuador and noted potential problems: the ambassador wanted to limit the number of American military personnel in Thailand, the government of Ceylon was “hanging by a thread,” and the Ecuadorians might demand “an excessive quid pro quo.”⁸⁶

Despite such objections, Ceylon, Ecuador, Thailand, and a large number of other nations had been listed as part of AFTAC’s expansion program, which had been approved by the Air Staff in the early fall. The expansion was expected to improve the ability to detect tests in the Southern Hemisphere or at high altitudes. The five-year program called for new electromagnetic, seismic, and acoustic stations. Stations employing a new technique—backscatter radar, which would detect the ionospheric disturbances caused by nuclear detonations at high altitudes—were also part of the plan. All together, it envisioned sixty-five stations being added to the existing network of thirty-five stations—stations located on every continent, in major countries, and on the most obscure islands.⁸⁷

WHILE AFTAC endeavored to expand and improve its ground and aerial nuclear intelligence capabilities, the CIA tried to develop a revolutionary new intelligence capability—one that would significantly improve the spy agency’s ability to monitor a large slice of Soviet military activities, including missile and bomb deployments, military exercises and troop movements, and nuclear activities ranging from uranium mining to reactor construction to test preparations.

The man who was responsible for implementing President Eisenhower’s February 1958 decision to assign the CIA responsibility for developing a photographic reconnaissance satellite was Richard Bissell. As with the U-2, Bissell headed a CIA–air force–contractor program, code-named Corona, to develop the reconnaissance system. On February 28, 1959, a little over a year after the president’s go-ahead, test launches began from Vandenberg Air Force Base in California. But it would not be until August 18 and 19, 1960, that a camera-equipped Corona satellite would be successfully placed in orbit, photograph targets in the Soviet Union and other denied areas, and return its images back to earth the next day via a film capsule that was ejected from the satellite and recovered in the air in the vicinity of Hawaii.⁸⁸

The satellite’s orbit allowed it to overfly the entire Soviet Union, at times passing only 116 miles above its target. The camera carried on that August flight was designated Keyhole-1 (KH-1), its forty-foot resolution being far inferior to that of the U-2. But the spyplane would not be flying over Soviet territory any more, because Soviet air defenses had shot down a U-2 piloted by Francis Gary Powers on May 1 in the vicinity of Sverdlovsk, preventing Kyshtym and a number of other important targets from being photographed that day. Fortunately, successful satellite missions were conducted on December 7–10, 1960, and June 16—19, 1961, using the KH-2 camera, with a resolution of about twenty-five feet. In August the first mission using the KH-3 camera, with a resolution that varied between twelve and twenty-five feet, flew, and of the final four successful missions of 1961, three would be KH-3 missions.⁸⁹

Top priority for the early Corona missions were areas where intercontinental ballistic missiles might be deployed, for the U-2 missions had not been able to cover enough Soviet territory to determine whether there was a “missile gap,” as Democrats had charged during the 1960 election year. Finding airfields with heavy bombers, which could reach the United States carrying atomic bombs, was the second priority. Nuclear energy targets were third. For the August 18 mission they included Kyshtym and Nizhnyaya Tura, the name by which Sverdlovsk-45 was known in the U.S. intelligence community.⁹⁰

The Corona missions gave the army, navy, and CIA interpreters at the CIA-managed National Photographic Interpretation Center (NPIC)—which had been established in January 1961 as one of President Eisenhower’s last acts—a reason for updating reports on assorted Soviet atomic energy complexes. A February report focused on the uranium metal plant and related facilities at Novosibirsk, comparing the more recent satellite images to those obtained by the U-2 photos from Soft Touch, a comparison that revealed a number of additions and changes.⁹¹

It sometimes took a while before a target showed itself to the Corona cameras. A forty-four-page report issued in June, concerning the uranium-mining and -milling complex at Mayli-Say in the Fergana Valley, apparently relied more on the older U-2 imagery and a study by the U.S. Geological Survey (of the area’s geology) than on Corona images, possibly because the target was obscured by cloud cover. Throughout 1960 and a substantial part of 1961, clouds interfered with attempts to photograph Kyshtym. But during a September 1961 KH-3 mission, the clouds finally parted. The photographs showed that canals had been constructed to route the Techa River around Lake Kyzyltash and that two large cascaded basins with a combined area of approximately thirty miles had been created for retention and evaporation of drainage from the lake.⁹²

Much earlier that year, in February, the CIA’s espionage branch, the Directorate of Plans or Clandestine Service, had received a report of a possible accident at Kyshtym. The source had both relevant firsthand knowledge and hearsay evidence. The latter included reports from “several people that large areas north of Chelyabinsk were contaminated by radioactive waste from a nuclear plant operating at an unknown site near Kyshtym.” The source also revealed that “it was general knowledge that the Chelyabinsk area had an abnormally high incident of cancers.” Twenty years later a 1981 CIA report noted that the creation of the retention basins and construction of the bypass canals may have become necessary owing to the repeated release of “significant fission and activation products” from reactor operations and from site runoffs.⁹³

THE IMAGES OBTAINED by U-2 overflights prior to May 1960 and acquired by Corona satellites since then, the data and debris gathered by the Atomic Energy Detection System, the analysis of open-source material (including official Soviet statements as well as newspapers and magazines), communications intelligence, and human intelligence, both overt and clandestine, all went into producing a 1961 national intelligence estimate that was approved by the U.S. Intelligence Board on October 5, 1961: NIE 11-2-61, The Soviet Atomic Energy Program. Estimates with identical titles had been published since at least 1956, with each year’s edition providing additional certain knowledge of the Soviet program as well as revised estimates of key parameters such as the quantity of fissile material produced.⁹⁴

The 1961 report, which consisted of forty-five pages of text and twenty-five pages of photographs and maps, covered all aspects of the Soviet program: organization, nuclear reactors, nuclear materials production, nuclear weapons, possible allocations of fissionable material to weapons stockpiles, and research laboratories. Neither human nor communications intelligence provided information on the Soviet designations for the country’s secret cities—for example, there is no reference to “Arzamas-16” or “Chelyabinsk-40.” But more importantly, the estimate reported on the existence of atomic activities at such sites and specified the type of activity.⁹⁵

The report noted the presence of the gaseous diffusion plants at Tomsk and Verkh Neyvinsk, and suggested there was probably a third one at Angarsk, about which “considerable information” had become available in the preceding year. Angarsk had begun operations in 1954, but the authors reported that “we have been unable to confirm U-235 production in this area.” It estimated that the Soviets had produced about 167,000 pounds of U-235 by mid-1961, although the navy’s intelligence chief dissented, arguing that it was based on assumptions not supported by the available evidence.⁹⁶

The two sites that the estimate unequivocally identified as being associated with plutonium production, and provided a bit of detail about, were Kyshtym and Tomsk. The authors also noted that the site near Krasnoyarsk (Krasnoyarsk-26, where production had begun in August 1958) and “perhaps that at Angarsk” (where it had never taken place) might also include some plutonium production facilities. Outside of those locations it was doubtful there were any other large plutonium production facilities, since it “was unlikely that any sites large enough to have significant plutonium production capacity would have remained wholly unassociated by intelligence with the Soviet atomic energy program.” Indeed, there were no other plutonium production sites.⁹⁷

The secret city of Sarov was, as noted, apparently first detected by communications intelligence, and then identified as the principal site for nuclear weapon research, design, and development. The estimate reported that the U-2 images taken in February 1960 of the area revealed “a large and elaborate nuclear weapon research and development complex comparable in size to the combined facilities of the Los Alamos Scientific Laboratory and the Sandia Corporation in Albuquerque.” It was also noted that recent analysis of July 1959 U-2 photography indicated the existence of a facility near Kasli that was “probably concerned with nuclear weapon research and development.” What the analysts had found was Chelyabinsk-70, the Soviet Union’s second weapons design bureau.⁹⁸

Those with access to the national intelligence estimate could also read about Semipalatinsk, view a photograph, and examine drawings of the site—one of which showed the main shot area and the associated facilities, and another which showed the layout of the apparent ground zero. The text reported that three facilities had been constructed outside the fenced shot area since 1957. They consisted of a new research facility (most probably associated with laboratory experiments related to nuclear weapons development and located to the northwest of the main shot area), a rectangular grid pattern about three miles by five miles in size, and an apparent ground zero also located to the north-northwest of the main shot area.⁹⁹

PRODUCING A NATIONAL ESTIMATE is usually a long and involved effort. And during the last part of that process, the test ban moratorium came to an end, with a Soviet weapons test at Semipalatinsk on September 1. During the halt in testing, while the United States, Britain, and the Soviet Union had been negotiating a possible test ban, some had been concerned that the Soviet Union was using the negotiations as means of halting U.S. and British testing while secretly conducting its own tests. During a December 1960 interview with U.S. News & World Report, AEC chairman John McCone stated that he believed the Soviets were probably conducting clandestine tests, possibly underground, in order to gain on the United States.¹⁰⁰

The possibility of cheating and the need to detect it had, of course, occurred to the U.S. intelligence community, even before the moratorium had been declared. In December 1957, the community had produced a special national estimate on the topic: Feasibility and Likelihood of Soviet Evasion of a Nuclear Test Moratorium. The analysis focused on the ability of the Atomic Energy Detection System to sense an explosion and confirm that it was a nuclear blast. The estimators concluded that the existing system had an excellent capability (90 to 100 percent) for detecting airbursts of 10 kilotons or greater, a good capability (60 to 90 percent) with respect to 5- to 10-kiloton airbursts, and a fair one (30 to 60 percent) for 3- to 5-kiloton bursts. There was a poor capability (0 to 30 percent) for airbursts less than 3 kilotons.¹⁰¹

The report also observed that while it was conceivable that tests could be staged in remote areas such as Antarctica or southern waters, “such possibilities would probably be excluded by the Soviets, since various intelligence collection efforts could be almost certain to spot the activities which would be associated with test preparations, if not the test itself.” With respect to establishing a detonation’s nuclear origin, the estimate noted that it would be necessary to obtain radioactive debris, which was “generally possible” for tests involving fission conducted between the surface and thirty-five to forty-five thousand feet.¹⁰²

Herbert Scoville, the head of the CIA’s OSI, reported McCone’s remarks to CIA chief Allen Dulles and noted that the intelligence community had “reached the conclusion that balancing the potential gains versus the risk of detection, it appears unlikely that the Soviets have been conducting clandestine nuclear tests.” He also mentioned that he had recently heard an air force briefing purporting to provide evidence of such clandestine tests, and that while the agency was making a detailed analysis of the material presented, he believed that “the Air Force has provided no significantly new information and has primarily twisted the raw data to prove a prejudgment.”¹⁰³

In late April, Dulles, and then the U.S. Intelligence Board, approved a national intelligence estimate titled The Possibility of Soviet Nuclear Testing During the Moratorium. The three-page analysis examined the technical motivations and political considerations affecting Soviet actions, the techniques that could be employed to minimize the risk of detection, and the evidence of possible testing. With regard to the latter, it noted that the United States had collected no nuclear debris or other conclusive evidence of Soviet nuclear tests since November 3, 1958. Other than nuclear debris, conclusive evidence might be hard to come by since “other indicators of testing activity are susceptible to alternative explanations.” There had been seismic events in the Soviet Union that could have been the result of tests, and there had been indications from other intelligence sources that raised the possibility of Soviet evasion by underground testing, but those indications “are also susceptible to alternative explanations.”¹⁰⁴

The most suspicious evidence concerned the area around Osh, and Semipalatinsk. The data concerning Osh, undoubtedly measurements gathered by AFTAC seismic stations, was considered most consistent with the Soviets’ having conducted one or more detonations using high explosives in the winter of 1959–1960, as part of their seismic improvement program or to study methods of clandestine testing. U-2 photography during April 1960 showed evidence of testing after the U-2 mission in August 1957. But those who prepared the estimate considered the most likely explanation to be that the tests had occurred between August 1957 and November 1958. The assessment that the Soviets had conducted tests during the moratorium “cannot be drawn from the available evidence,” they concluded—a judgment disputed by the intelligence chiefs of the armed services and Joint Staff.¹⁰⁵

On August 30, 1961, a U.S. listening post on Cyprus, probably one of the stations that monitored radio and television broadcasts for the Foreign Broadcast Information Service, picked up a Tass transmission of items for the provincial papers that included a statement scheduled for release at seven that evening. The statement announced that the Soviet Union was going to resume nuclear testing. An alert monitor had “fished this item out of the reams of copy being teletyped from Moscow to the Caucasus in Cyrillic characters,” Glenn Seaborg recalled. About two hours later, the information had reached the State Department, and not long after that the president knew.¹⁰⁶

U.S. detection stations were probably notified within twenty-four hours. In any event, as Herbert Scoville, in his capacity as chairman of the JAEIC, reported on September 1, an explosion had been detected earlier that day by three acoustic stations. The estimated location was in the vicinity of Semipalatinsk while the estimated yield was in the 50- to 500-kiloton range, with a best estimate of 150 kilotons. The Soviets had indeed tested in the vicinity of Semipalatinsk, but the yield of 16 kilotons was far lower than the initial estimates relying solely on acoustic measurements.¹⁰⁷

BEFORE THE YEAR was out, another fifty-eight tests would follow, almost all conducted at either Novaya Zemlya or Semipalatinsk—the exceptions were a small number of missile-related tests at Kapustin Yar. Herbert Scoville was busy issuing statements like those he issued on September 1 and November 4, the latter reporting the detection of two explosions. The November 3 detonation, apparently with a yield of less than 10 kilotons, was an atmospheric blast in the vicinity of Semipalatinsk, first reported by two acoustic and ten electromagnetic stations. The second explosion, on November 4, took place on the east coast of Novaya Zemlya, the JAEIC chairman reported. The estimated yield was between 1 and 6 megatons, most probably 3 megatons. Its power was certainly sufficient to light up much of the U.S. detection network, with signals being reported from eight acoustic, nine seismic, and ten electromagnetic stations.¹⁰⁸

But that test was little more than a stick of dynamite compared to the Soviet test of October 30, also at Novaya Zemlya. Andrei Sakharov had appealed to Nikita Khrushchev and the rest of the Soviet leadership to cancel the test of the bomb he had helped design. The device, designated RDS-220, was about 6.5 feet in diameter and 26 feet long, and weighed about twenty-five tons. To ensure that the bomber dropping it did not become one of its victims, the bomb relied on an enormous set of four parachutes. In its most powerful form it could explode with the force of 100,000,000 tons of TNT—100 megatons. Concerns over the possible effects of such a blast led Khrushchev and his colleagues to order its yield reduced to a still impressive 50 megatons.¹⁰⁹

Decades later, Sakharov recalled being in Moscow at the beginning of October 1961 to discuss calculations concerning the “Big Bomb,” and that it was assembled in a special workshop on top of a platform car. “A few days later, when everything was ready, the superstructure would be dismantled and, under the cover of darkness, the platform would be coupled to a special train that would transport the device along an open track all the way to the airfield where it would be loaded into the bomb bay of a waiting plane.”¹¹⁰

On October 30, a specially modified Tu-95N bomber, flown by Maj. Andrei Durnovstev, took off with its payload, headed for a drop point over Novaya Zemlya. Sakharov sat by a telephone, waiting for news. A call that morning informed him that the plane had taken off. He recalled that he and his fellow scientists at Sarov “just couldn’t keep our minds on our work. My colleagues were hanging around in the corridor, continually dropping in and out of my office.” At noon he received another call and was told “there’s been no communication with the test site or the plane for over an hour. Congratulations on your victory!”* Ninety minutes later, Sakharov received another call, learning that the mushroom cloud had reached a height of over forty miles.¹¹¹

Due to the advance planning, the Soviet Tu-95 escaped without damage. The same could not be said for the specially equipped American plane monitoring the test. There appears to have been significant prior warning of the planned test date from intelligence sources—human intelligence, communications intelligence, or both—and from Khrushchev himself, who had bragged openly that the Soviets could detonate a 100-megaton bomb. After discussions between Scoville, acting in his capacity as JAEIC chairman, and Gerald Johnson, the assistant to the secretary of defense for atomic energy, a crash program had been initiated to modify a KC-135 to carry broadband electromagnetic and special optical equipment to monitor the test. The modification effort, code-named Speedlight, was carried out under the supervision of Doyle Northrup by an air force unit known as Big Safari, with headquarters at Wright-Patterson Air Force Base in Ohio, and detachments at the headquarters of key aircraft manufacturers. By October 27 the plane was ready to depart for its overseas staging base.¹¹²

Given Novaya Zemlya’s location in the Arctic, it was possible for the Speedlight aircraft to fly relatively close to the test area. Given the force of the explosion, it proved almost too close. The plane returned not only with the data it had been sent to gather but also with a scorched fuselage. Had the Soviets decided on a 100-megaton blast, the plane and its pilots would not have made it back.¹¹³

The Bethe Panel scrutinized the data and concluded, correctly, that the blast exceeded the 50-megaton objective by 7 megatons, large enough to ensure that “the whole earth’s atmosphere just vibrated for days after,” Northrup recalled. The panel also determined that the yield had been constrained by encasing the weapon in lead rather than uranium.¹¹⁴

In the aftermath of the huge blast there were repercussions in the Soviet Union and the United States. Khrushchev asked his scientists about the targets against which such a bomb could be used. West Germany could not be on the target list, they told him, because the westerly winds would blow the fallout over East Germany, inflicting damage not only on the East Germans but, more importantly, on the Soviet forces stationed there. On the other hand, Britain, Spain, France, and the United States were viable targets.¹¹⁵

In the United States, in addition to leading to the production of similarly modified planes, the success of the initial Speedlight mission produced a letter of appreciation from President Kennedy to Robert McNamara, asking his defense chief to “extend his sincere appreciation to the personnel” who participated in the operation. “The expeditious preparation of the complex technical equipment and the bold execution of this operation are excellent examples of the effective use of our resources. I am proud of those who took part in this operation,” Kennedy continued.¹¹⁶

CIA and other intelligence analysts evaluated the data collected from the Soviet tests of September and October and used it in a national estimate on Soviet strategic military posture that was completed in November. The estimate broke down the tests by their differing purposes, which included evaluating complete weapon systems, researching and developing future offensive systems, as well as obtaining weapons effects information needed to develop an antiballistic missile system. The new national estimate concluded that the new tests had given the Soviets increased confidence in weapons design, and had augmented their understanding of thermonuclear weapon technology and weapons effects. In addition, the “1961 test series will permit the Soviets to fabricate and stockpile, during the next year or so, new weapons of higher yields in the weight classes presently available,” which had significant implications for the Soviet ability to deliver their warheads to U.S. targets via missile rather than via slower and more vulnerable aircraft.* The analysts correctly concluded that the October 31 detonation was a test of a 100-megaton device at reduced yield so as to limit fallout.¹¹⁷

WHEN PRESIDENT John F. Kennedy wrote to Robert McNamara, praising those involved in Speedlight, the Soviet Union was no longer the only Communist state whose nuclear aspirations were of concern. There was no doubt that the People’s Republic of China was seeking to join the nuclear club, despite the substantial entrance fee—in terms of men, money, and resources. By late 1961 the club included not only its charter members, the United States and the Soviet Union, but also Great Britain and France. Britain had joined in October 1952, with a test off the northwest coast of Australia, and France in February 1960, when it detonated an atomic bomb in the Sahara Desert.¹¹⁸

But while French acquisition of the bomb was certainly not viewed with pleasure by Eisenhower or his successor, China’s pursuit of atomic weapons was far more distressing. Walt Rostow, who served as head of the State Department’s policy-planning council during part of Kennedy’s brief presidency, and then as national security adviser under Lyndon Johnson, recalled that Kennedy considered a possible Chinese nuclear test as “likely to be historically the most significant and worst event of the 1960s.”¹¹⁹ That event had its origins midway through the previous decade.

___________

* The CIA received the first five modified U-2s, and Detachment C began flying missions from Eielson in June 1957, before it began flying them from its home base at Atsugi. During a flight headed northeast along the Kurile Islands, CIA pilot John Shinn made the first interception of nuclear debris in a U-2. See Chris Pocock, The U-2 Spyplane: Toward the Unknown (Atglen, Pa.: Schiffer, 2000), p. 120.

* The lack of communication was a sign of success because ionized particles released by a powerful explosion interfere with radio transmissions. The more powerful the explosion, the longer the communications gap. Andrei Sakharov, Memoirs (New York: Knopf, 1990), p. 219.

* By November 1961 those with access to the latest national intelligence estimate on Soviet strategic forces knew that the feared missile gap did not exist, knowledge that was the direct result of the Corona missions. But, of course, Soviet intercontinental ballistic missiles (ICBMs) were inevitable, and Soviet tests that revealed the relationship between warhead size and yield also revealed Soviet ICBM options with regard to missile range, yield, and size.