Atomic audit :the costs and consequences of U.S. nuclear weapons since 1940

Chapter One Building the Bomb Kevin O'Neill     Between 1945 and 1990 the United States produced more than 70,000 nuclear weapons. The cost--for the acquisition and production of fissionable materials and for weapons research, development, testing, and manufacturing--was approximately $409 billion and was divided among several agencies (see figure 1-1). Before World War II and up to the end of fiscal 1947, expenditures totaled $27.3 billion, of which nearly $26 billion went to the Manhattan Project and other, smaller government efforts during the war. After World War II, the Department of Defense invested more than $37 billion in nuclear weapons effects and related research. But by far the largest amount was spent by civilian agencies. Together, the Atomic Energy Commission (1947-75), the Energy Research and Development Administration (ERDA, 1975-77), and the Department of Energy (DOE, 1977 to the present) have spent $345 billion since 1947 to produce nuclear materials, to conduct research, and to develop, test, and manufacture nuclear weapons. This chapter covers the costs incurred during World War II, the Manhattan Project, and the cold war. Other costs, such as the historical and future costs of weapons dismantlement, environmental remediation, and waste management, are taken up in chapters 5 and 6. Methodology and Sources The inherent secrecy of the U.S. nuclear weapons program, the almost total lack of detailed budgetary data from the 1940s through at least the early 1960s, and the multipurpose nature of many weapons facilities make it virtually impossible to audit the U.S. nuclear weapons development program on a site-by-site basis. In addition, a calculation of actual year-by-year expenses by program is complicated by standard budgeting practices, which generally record budget authority and obligations.     Still, budget authority and obligations make it possible to estimate expenditures over time. Budget authority is the right granted to an agency by Congress to allocate a given amount of money for certain specified purposes, while obligations represent an agency's commitment to allocate this amount as specified. These obligations may be discharged within a single fiscal year or over a period of several years, as in the case of construction projects and capital equipment purchases. Therefore, budget authority and obligations provide an accurate record of historical expenditures, but not of actual spending in a given year.     Most of the data presented in this chapter can be found in government documents presented to the Congress and to the public. Beginning in the early 1960s, annual AEC, ERDA, and DOE budget requests were presented to Congress and printed as part of the official records of hearings of the Joint Committee of Atomic Energy. After this committee was abolished in 1977, separate House and Senate appropriations subcommittees continued the practice. Before the early 1960s, these records did not typically present budget data in a useful fashion, although the Bureau of the Budget (later, the Office of Management and Budget, or OMB) annually published such data in the Budget of the United States Government . Some of the data have only recently been released to the public or made available by the DOD and DOE at the request of this project's participants.     Several deflators have been used. OMB total defense outlays were used as a deflator for Manhattan Project expenditures, which are expressed as actual expenditures by the sources used. Otherwise, DOD deflators are used for total obligational authority and for budget authority, where applicable. The Nuclear Weapons Production Complex Historically, the United States has expanded or reduced its weapons production complex whenever policy decisions, perceived military requirements, and costs deemed it necessary. At present, the complex consists of thirteen major facilities around the country (three of which are inactive). These facilities occupy nearly 3,300 square miles (8,547 square kilometers), a land area greater than that of Delaware, Rhode Island, and the District of Columbia combined (see figure 1-2 and appendix C for a complete list of active and inactive facilities). Producing the Nuclear Materials The most difficult step in manufacturing nuclear weapons lies in producing highly enriched uranium (HEU) and plutonium, the fissionable materials used in nuclear weapons. Only certain isotopes of uranium and plutonium are suitable for making weapons. Isotopes are forms of an element that have nearly identical chemical properties but whose nuclei contain different numbers of neutrons. Naturally occurring uranium is composed of almost 99.3 percent uranium-238, which is not suitable for nuclear explosives, and only 0.7 percent uranium-235, the fissionable isotope used in weapons. The level of uranium-235 must be increased or enriched before it can be used in nuclear weapons or even in most common nuclear power reactors. Plutonium is also problematic: it occurs naturally only in minute quantities and must be produced in a nuclear reactor. Uranium Enrichment. Most nuclear power reactors use uranium fuel enriched to 2-5 percent, but the U.S. nuclear program used "highly enriched uranium," defined as greater than 20 percent uranium-235, to manufacture warhead pits and secondaries for thermonuclear weapons, for naval reactor fuel, and as fuel for some production reactors at its Savannah River Site (SRS).     The Manhattan Project--officially known as the Manhattan Engineer District (MED)--used a combination of three enrichment methods to produce HEU. The first method, called electromagnetic isotopic separation, achieved enrichment with the aid of large electromagnets in a device called a calutron. The second method relied on gaseous diffusion. Gaseous diffusion enrichment operates on the principle that heavier gaseous molecules move more slowly than lighter gaseous molecules. In this method of enrichment, gaseous uranium tetrachloride molecules are circulated into a series of cylindrical containers, each divided into two sections by a porous barrier. The molecules containing the lighter uranium-235 hit the barrier more frequently than the heavier molecules and therefore collect on one side of the barrier. This was the primary method used by the United States to enrich uranium during the cold war. Large plants housing the calutrons (Y-12) and gaseous diffusion "cascades" (K-25) were built at what is now the Oak Ridge Reservation in Tennessee. A third method, thermal diffusion, was initially considered and then abandoned after electromagnetic and gaseous diffusion were deemed more promising for large-scale production. However, the S-50 Thermal Diffusion Plant at Oak Ridge--built for an estimated $179 million--was used to provide enriched feed material to the Y-12 and K-25 plants.     After World War II, in response to the first Soviet detonation of an atomic bomb and the outbreak of the Korean War, the AEC expanded the Oak Ridge gaseous diffusion plant and built large new facilities in Portsmouth, Ohio, and Paducah, Kentucky. These facilities have also produced low enriched uranium for commercial nuclear power plants. In April 1964 the United States unilaterally ceased production of HEU for weapons, although two of the plants continued to produce HEU for naval nuclear reactors, plutonium production reactors, and civilian research reactors. As a result of this decision, HEU production declined 40 percent (at the same time, plutonium production fell 20 percent). Plutonium and Tritium Production. Unlike uranium-235, which can be isolated from other naturally occurring uranium isotopes, all of the plutonium used by the United States to build nuclear weapons has been produced in nuclear reactors. Plutonium is created when uranium-238 absorbs neutrons during a chain reaction. The uranium-238 forms the isotope uranium-239, which through a process called "beta decay" quickly transforms into another element, neptunium-239. A second beta decay then results in plutonium-239, the principal isotope used in weapons.     Once the plutonium is produced, it is chemically separated from the spent reactor fuel in a procedure known as "reprocessing." Because spent reactor fuel is highly radioactive, reprocessing must be conducted in remotely operated or automated facilities. Plutonium and uranium can be separated from the highly radioactive waste by dissolving the spent fuel in acid. Subsequently, the plutonium and uranium are separated from each other using nitric acid in a process known as solvent extraction. The plutonium is then used to manufacture weapons, and the uranium is recycled back into the fuel production pipeline of the weapons complex (see figure 1-3).     During the cold war, the United States produced plutonium at two sites: Hanford Reservation, near Richland, Washington; and the Savannah River Site, near Aiken, South Carolina (known as the Savannah River Plant until April 1, 1989). Hanford produced the plutonium for the Manhattan Project. Of the nine reactors built at Hanford, eight were in operation by the mid-1950s but were all shut down by 1971. A ninth reactor was constructed in the early 1960s and produced both weapon-grade and fuel-grade plutonium until 1987. The irradiated fuel containing the plutonium was stored on site, and the plutonium was extracted in as many as five reprocessing plants at Hanford, nicknamed "Queen Marys" by plant workers because of their great size (which significantly exceeded that of their maritime namesake).     Between 1951 and 1955, five reactors were constructed at the SRS. Unlike the light-water, graphite-moderated reactors at Hanford, the Savannah River reactors were moderated and cooled by heavy water and used HEU fuel. HEU and plutonium were recovered from spent fuel at the Savannah Site's H and F chemical separation plants.     The United States also used the Savannah River reactors to produce the materials needed to build boosted and thermonuclear weapons. These weapons employ tritium and deuterium to increase the explosive yield or decrease the amount of fissile materials required to achieve the same yield. In boosted weapons, a mix of tritium and deuterium gas is released into the hollow, fissioning mass of plutonium, or HEU, called a "pit" or "core," shortly before detonation. Energy generated by the fission explosion causes tritium atoms to join, or "fuse," with deuterium and release neutrons. The release of neutrons by the tritium causes more of the plutonium or HEU to fission and thereby increases, or "boosts," the yield.     In thermonuclear weapons, tritium is formed in situ from lithium deuteride components and fuses with deuterium in the thermonuclear "secondary." The energy released through fusion in thermonuclear weapons is sufficient to increase the yield of an atomic weapon from tons or kilotons to the megaton range. If a uranium tamper (composed of uranium-238) is placed around the secondary, high-energy neutrons released by the fusion reaction will fission the tamper and increase the yield even more.     The United States needed to isolate sufficient deuterium from heavy water and produce tritium to manufacture boosted and thermonuclear weapons. Heavy water occurs in nature in a concentration of about 0.015 percent and may be separated from raw water through a combination of distillation and chemical exchange between hydrogen sulfide and water or through electrolysis. To isolate heavy water, the Manhattan Project and the AEC constructed several plants, including one at Savannah River that operated from 1952 to 1982. Tritium was produced in the heavy water-moderated reactors by bombarding lithium-6 targets with neutrons. At the Tritium Recovery Facility at SRS, engineers then recovered the tritium from the irradiated targets in a fashion analogous to reprocessing spent uranium fuel.     The United States relied on various sources to obtain the natural uranium needed to produce HEU and plutonium and to fuel the production reactors. Much of the Manhattan Project uranium--4,810 tons in all--came from the Belgian Congo in colonial Africa (today the Democratic Republic of Congo). Belgium sent this uranium to the United States at the beginning of the war to prevent it from falling into the hands of Nazi Germany. In the spring of 1943 the Manhattan Project initiated a highly secretive effort, known as the Murray Hill Area Project (named after a neighborhood in Manhattan that was the location of MED headquarters), to purchase the rights to as many of the high-grade uranium ore deposits as possible around the world, in a deliberate effort to monopolize existing supplies and prevent other nations from acquiring the raw materials necessary to build atomic weapons. To keep the program secret and avoid having to obtain Congressional approval, the government transferred funds directly from the MED into the personal bank account of Brigadier General Leslie Groves, head of the MED. Initial funding in 1943 was in excess of $37 million (more than $400 million in 1996 dollars).     After the war, the AEC purchased uranium from a variety of foreign sources, including Canada, Portugal, Australia, and South Africa. By the end of 1955 some 925 uranium mines were in operation in the United States. That figure held relatively steady through the early 1960s (as total production per mine increased) and then declined until mining operations for the weapons program ceased in 1970. Accounting records do not allow for an estimate of uranium purchase costs in the immediate postwar years, but by 1954 the AEC was spending $1 billion or more annually on uranium (see appendix table A-1). Other Materials. Other important materials besides HEU, plutonium, tritium, and deuterium are needed to make nuclear weapons. Very pure fluorine gas is required to make uranium-hexafluoride (UF-6), the corrosive gas used in the enrichment process. The United States needed high-quality graphite to moderate Chicago Piles 1 and 2 (CP-1 and CP-2, the first experimental reactors at the University of Chicago), the X-10 reactor at Oak Ridge (the first plutonium production reactor and the model for the Hanford reactors), and weapons reactors at Hanford. Beryllium, a strong but lightweight metal, was used to fabricate shields around the core to reflect and amplify neutrons back into the critical mass (early weapon designs also incorporated beryllium in neutron initiators). Lithium-6 was needed to produce tritium, and mercury was needed to produce lithium-6. Significant quantities of precious metals, including gold, silver, and platinum, were also used in the manufacture of nuclear weapons (see chapter 5). Many of these materials--most notably plutonium, beryllium, and mercury--are highly toxic; environmental contamination by these and other materials continues to pose a significant concern at weapons complex sites (see chapter 6). From Materials to Warheads and Bombs More than two dozen facilities in the United States have conducted nuclear weapons research and helped to design, test, and manufacture nuclear warheads and bombs. The process can be illustrated by examining the primary functions of several of these facilities. Weapons Research, Design, and Testing. The primary mission of nuclear weapons research and development has been to create new nuclear explosive devices. The scope of the research and the direction and growth of the nuclear arsenal have often been dictated by laboratory breakthroughs rather than military requirements, especially during the early years of the cold war. From the beginning, the services were not greatly concerned about specialization: "As long as their specific requirements could be met, they felt that interchangeability of nuclear materials among several systems was a good thing.... As soon as the nuclear designers learned and pointed out that higher efficiencies would be obtained if weapons were tailored to specific systems, the Services accepted the idea gladly." Even as late as 1973, a senior DOD official told Congress, "the laboratories have often led. They have bright people. They know the military problems. In many cases, they travel, they interact and quite often, they come up with solutions to military problems before the military people know that a solution is possible."     As weapons technology evolved, however, the services began requesting not only more and different kinds of bombs and warheads, but also extra features or capabilities that significantly increased the per unit cost. Because the military services have never had to pay for their own nuclear bombs and warheads--all of the costs were borne by the AEC, ERDA, and the DOE--there was never any fiscal incentive not to "require" large quantities of the most sophisticated kinds that could be produced.     Between 1945 and 1991, the DOE and its predecessors designed and built 65 types of bombs and warheads for 116 kinds of delivery systems, ranging from battlefield, or "tactical," nuclear weapons to the strategic bombs and warheads based in the United States or on board ballistic missile submarines (see box 1-1). Land mines, antisubmarine depth charges, artillery shells, gravity bombs, and cruise missiles are among the various weapons systems that have employed nuclear warheads (a complete chronological list of U.S. nuclear bombs and warheads appears in table 1-3).     Three dedicated nuclear laboratories--at Los Alamos, New Mexico; Livermore, California; and Albuquerque, New Mexico--have designed, tested, and sometimes manufactured nuclear weapons. The AEC inherited the Manhattan Project's main research laboratory at Los Alamos in 1947. Scientists there continued to pursue new weapons concepts and to manufacture atomic bombs. As the weapons complex grew, manufacturing activities were performed elsewhere, while the Los Alamos National Laboratory (LANL), as it is known today, continued to be a primary focus of nuclear research and weapons development. Over time, its numerous assets have grown to include facilities for processing plutonium, handling other radioactive materials, manufacturing test devices, and testing new design concepts short of a full-scale test.     The Lawrence Livermore National Laboratory (LLNL), in Livermore, California, was established in 1952, primarily to expand research into thermonuclear weapons. In the late 1940s, some scientists, AEC commissioners, and influential members of Congress began to argue that insufficient resources at Los Alamos were being dedicated to thermonuclear work and that a new laboratory was needed. Manhattan Project physicists Edward Teller and Ernest O. Lawrence were particularly insistent that a new laboratory be established (Teller even resigned from Los Alamos to lobby the administration on this point). Pressure also came from Senator Brien McMahon (Democrat of Connecticut), chairman of the Joint Committee on Atomic Energy and an ardent advocate of increased production of nuclear weapons, who tried to interest the Department of Defense in establishing a second laboratory. Teller personally lobbied air force officials on this point, forcing AEC chairman Gordon Dean to choose between creating a laboratory at Livermore or ceding the hydrogen bomb program to the DOD. Under this pressure, the AEC established the Livermore Radiation Laboratory in 1952. Ironically, the first thermonuclear device tested by the United States in the November 1952 "Ivy-Mike" test and most of the more advanced weapons detonated during the "Castle" test series in 1954 were designed at Los Alamos. However, the AEC argued that two research laboratories were needed to create competition, fostered by "independent peer review" among weapons designers, which would lead to faster and cheaper breakthroughs. In turn, this competition contributed to the development of new weapons designs that exceeded military expectations and, by some accounts, spurred nuclear weapons developments as much as the competition with the Soviet Union.     Both the Los Alamos and Lawrence Livermore laboratories concentrated on developing sophisticated "physics packages" for nuclear weapons. A laboratory for nonnuclear research, known as the Sandia National Laboratories (SNL), was established at Albuquerque, New Mexico, in 1945 and also Livermore, California (in 1956). Sandia worked on other important bomb components, such as electronics equipment; safing, fuzing, and firing devices; and drogue parachutes. Consequently, while LANL and LLNL scientists independently produced new warhead designs, Sandia engineers contributed to the efforts of both laboratories.     Once new nuclear explosives were designed, the AEC and DOD collaborated to test them. The United States conducted "Operation Crossroads," its first post-World War II nuclear tests, at Bikini Atoll (part of the Marshall Islands) in the Pacific in July 1946, before the Manhattan Project was disbanded (the operation is discussed later in the chapter). In 1948, the United States began testing its nuclear weapons on Enewetak Atoll, also in the Marshalls and the site of the first thermonuclear test (Ivy-Mike) on November 1, 1952. Together, these atolls became the prime facilities of the AEC's Pacific Proving Ground (PPG), which was the site of sixty-six atmospheric and underwater tests until 1958.     However, only a limited number of tests could be conducted at this site, owing to the complex logistics of maintaining such a facility so far from U.S. shores and adverse weather conditions in the fall and winter. Defending the atolls against Soviet spies and possible outright attack was also difficult. Following the outbreak of the Korean War in June 1950, the army and navy--eager to break the air force's monopoly on atomic weapons and develop devices small enough to be used on the battlefield--requested small, low-yield weapons deliverable by carrier aircraft and artillery pieces. After studying several locations, the AEC in December 1950 designated the areas in and around the Las Vegas Bombing and Gunnery Range as the dedicated continental nuclear test site (on the selection of the test site, see box 1-2). The first test occurred a month later. This would be the primary test site for the remainder of the cold war. Although the United States continued to test high-yield thermonuclear weapons in the Pacific until the signing of the Partial Test Ban Treaty in 1963, 904 (88 percent) of the total 1,030 tests it conducted occurred at the Nevada Test Site (NTS) (see figure 1-5). Weapons Assembly, Production, and Storage. The manufacture and storage of nuclear weapons is a complicated process. Nuclear and nonnuclear components must be manufactured to stringent specifications. Assembly, production, and storage facilities must impose the strictest of security, health, and safety regulations to ensure that dangerous materials are not stolen and to minimize the health risk to workers and the offsite population. For example, metallic plutonium, which is susceptible to spontaneous combustion, must be handled in an enclosed and controlled environment. To shield workers from exposure and to reduce the risk of fire through accidental releases to the environment, this material is handled remotely in airtight "glove boxes."     Beginning in the early 1950s, components made of plutonium and HEU metal were assembled at the Rocky Flats Plant near Denver, Colorado, and the Y-12 Plant at Oak Ridge, Tennessee. Specially equipped machine shops at Rocky Flats produced pits and other nuclear weapons components from plutonium, HEU, depleted uranium, beryllium, and other materials. Scrap plutonium from the assembly line, plutonium metal not deemed waste, and plutonium metal from retired warheads was purified and recycled into new components at the site. Originally, HEU pit components were manufactured and shaped at Oak Ridge's Y-12 Plant before shipment to Rocky Flats, but Y-12 ultimately Assumed responsibility for the full assembly of these components while Rocky Flats consolidated its work on plutonium. Thermonuclear secondary components were also assembled at Y-12.     Of the many facilities that contributed nonnuclear components, only the Kansas City Plant, near Kansas City, Missouri, remains in operation. It supplies various electrical, electronic, and plastic components, including arming, fuzing, and firing systems, radars, and coded safety locks known as permissive action links (PALs).     Since 1945 nuclear and nonnuclear components have been assembled into completed nuclear explosive devices at seven locations: Los Alamos; Sandia National Laboratories in Albuquerque, New Mexico; Buffalo Works in Buffalo, New York; the Burlington AEC Plant in Burlington, Iowa; the Clarksville Modification Center in Clarksville, Tennessee; the Medina Modification Center near San Antonio, Texas; and the Pantex Plant, outside Amarillo, Texas. Only Pantex performs these functions today (see appendix C). The most sensitive and dangerous procedure takes place in containment cells known as "Gravel Gerties," where high-explosive chemicals are layered around the spherical nuclear explosive pits. Each Gravel Gertie (there are thirteen in all at Pantex) holds some 17 feet (5.2 meters) of gravel on its roof and is designed to collapse should the chemical explosive detonate during the assembly process, thus containing the explosion and any radioactive debris it products.     At first, high explosives for the Manhattan Engineer District were manufactured at the Salt Wells Pilot Plant in Inyokern, California (at what is now China Lake Naval Air Warfare Center), under the code name Project Camel. In 1955 the bulk of production moved to the Holston Army Ammunition Plant at Kingsport, Tennessee (which has been the sole supplier since 1961). Newly manufactured warheads (or those returned to Pantex for routine surveillance or modification) are stored at Pantex and remain in DOE custody until the military takes control of them at an operational base or military storage depot.     Pantex also served as one of four plants that routinely disassembled retired warheads, a function it still performs (see chapter 5). After disassembly, nuclear components are transferred from Pantex back to their point of origin--either Rocky Flats or Y-12--for recycling and reuse. Since the cessation of production activities at the Rocky Flats Plant in 1989, Pantex has stored all plutonium components on site, although HEU secondaries are returned to the Y-12 Plant at Oak Ridge. Nonnuclear components were returned to the Pinellas Plant near Clearwater, Florida; the Mound Laboratory in Miamisburg, Ohio; and the Kansas City Plant. Since 1995, most work on nonnuclear components has been consolidated at the Kansas City Plant. Assessing the Costs Of course, many laboratories besides Los Alamos, Livermore, and Sandia conducted research and developed processes for handling nuclear materials, and dozens of facilities produced nonnuclear components and handled radioactive materials (see appendix C and chapters 5 and 6). The very size of the nuclear weapons complex, not to mention the secrecy surrounding many of its activities, the lack of rigorous accounting procedures, and poor historical records, make it extremely difficult to determine the costs incurred at each facility. It is much easier to assess costs by program among the lead agencies that contributed to the development of the U.S. nuclear stockpile. The ones considered here are the Manhattan Project; the AEC, ERDA, the DOE, and DOD. The Manhattan Project The U.S. nuclear weapons program began shortly before the country entered into World War II. At first it was a modest program of basic research conducted under the supervision of President Franklin Delano Roosevelt and his closest advisers. Under wartime conditions it quickly expanded into a $25 billion project. Three offices--the National Defense Research Committee (NDRC), the Office of Scientific Research and Development (OSRD), and the Manhattan Engineer District--guided this work under what is more commonly known as the Manhattan Project. National Defense Research Committee. The NDRC was created in June 1940, when atomic physics was still in its infancy. Indeed, that uranium was fissionable had only been demonstrated two years earlier, and it was not yet known if building an atomic bomb was practical. In the fall of 1939, in response to a letter from Albert Einstein, President Roosevelt established the Uranium Committee of the National Research Council under the leadership of Lyman J. Briggs, Director of the National Bureau of Standards. The committee coordinated research into the properties of uranium, but very little money was spent on this effort. Less than a year later, information that Germany had begun its own research into uranium spurred Roosevelt to expand U.S. activities, and in June 1940 the Uranium Committee was reconstituted under the newly created NDRC.     Led by Vannevar Bush, a former vice president of the Massachusetts Institute of Technology, the NDRC investigated the fissile properties of uranium-235 and uranium-238 and began to evaluate methods for separating isotopes of uranium. The NDRC also coordinated research into the fissile properties and potential weapons applications of the newly discovered transuranic element number 94, later to be called plutonium.     Work coordinated by the NDRC was absorbed by the OSRD beginning in the summer of 1941. According to historical financial reports released by the AEC, the NDRC had spent $468,000, or $6.4 million in 1996 dollars, up to this time. Office of Scientific Research and Development. Although the creation of the NDRC had demonstrated the importance of atomic research, organizational problems led the government to change its support in the following year. As a scientific and research organization, the NDRC was not equipped to "fill the gap between research and procurement orders that engineers called `development."' Furthermore, it had to compete for resources with the laboratories operated by the army and navy and thus found access to funding restricted. To remedy these shortcomings, Bush persuaded President Roosevelt to elevate the priority assigned to the military application of atomic energy.     At Bush's recommendation, Roosevelt created the Office of Scientific Research and Development (OSRD) in June 1941. The OSRD was located within the Office of Emergency Management in the Executive Office of the President and was placed under Bush's direction. James Bryant Conant, the president of Harvard University, replaced Bush at the NDRC, which continued to exist under the OSRD. Under this arrangement, the NDRC made recommendations for further research, while the OSRD guaranteed access to the president and the promise of more resources. Research directed by Bush and Conant continued to explore different isotope separation methods, such as electromagnetic separation in calutrons and centrifuges and by means of gaseous diffusion. Work also continued on producing plutonium in a uranium-graphite "pile," although real breakthroughs were about a year away.     The OSRD was succeeded by the Manhattan Engineer District in 1942. It remained in existence until 1947, however, and its budget was intimately tied to the hidden accounts of the MED as well as other military research efforts. Official AEC sources report that the OSRD's independent contribution to the bomb program was $14.6 million, or $170.8 million in 1996 dollars. Manhattan Engineer District. By far the largest and best-known World War II organization involved in the building of the bomb was the Army Corps of Engineers' Manhattan Engineer District. Although the OSRD had made progress in all its endeavors to separate uranium isotopes and produce plutonium, the design and construction of production-scale plants were beyond the scope of the office's expertise and mandate. It had long been recognized that the help of the army would be needed to construct facilities large enough for full-scale atomic bomb production, and the military became involved while the program was still under the direction of the OSRD. By September 1942 the project was in the hands of the MED.     Under the leadership of Brigadier General Leslie Groves, who had recently supervised construction of the Pentagon, the MED built the nucleus of the U.S. nuclear weapons production complex. Far from pilot facilities, however, these full-scale plants were designed from the start to generate enough materials for a sizable stockpile of atomic bombs.     In September 1942 Groves selected Oak Ridge, a site near Clinton, Tennessee, for the production of enriched uranium. Over the next two years the MED constructed several facilities there, including the Y-12 Plant, which housed the calutrons; the K-25 Plant for gaseous diffusion; and the S-50 Plant, where uranium was enriched through thermal diffusion. In the process, about one thousand families were displaced.     The Oak Ridge calutrons and gaseous diffusion separators were developed by physicists and engineers at several university laboratories, primarily the University of California, the University of Chicago, and Columbia University, while the MED contracted out construction, labor, and management tasks to firms such as the Tennessee Eastman Corporation (a subsidiary of Eastman Kodak), Union Carbide, and Stone and Webster, establishing a pattern that survives to the present. Shortages of essential materials caused by the war sometimes led to innovative substitutions in the design and production of these technologies.     A notable and expensive example of such ersatz craftsmanship involved the use of silver, rather than copper, to manufacture the electromagnetic coils for the calutrons. Because copper was scarce, silver--a good electrical conductor and not a critical war material--became a substitute. On August 29, 1942, Secretary of War Henry L. Stimson requested the transfer of 175 million fine troy ounces (about 6,000 tons) of silver to the War Department for a "highly secret" project. By 1944 the amount had increased to 14,700 tons, worth nearly $3.3 billion in 1996 dollars.     On December 2, 1942, Enrico Fermi achieved the first chain reaction in natural uranium in CP-1, a graphite-moderated "pile" under the University of Chicago's Stagg Field. Surveying for the proper location for full-scale plutonium production reactors took place during the last two weeks of 1942 even before Fermi's experiment was complete. General Groves, in consultation with scientists in Chicago and Du Pont engineers, drew up a list of eight criteria for the site, encompassing available water and power supplies, land area, distance to existing communities, highways and railroads, and a suitable climate. From a list of twenty potential sites, Groves indicated that he preferred the ones in the Pacific Northwest. A three-man team--two engineers with E. I. Du Pont de Nemours and a lieutenant colonel from the Manhattan Engineer District--inspected the five most promising: Coulee and Hanford in Washington State; and Pit River, Needles, and Blythe in California. Their report, completed by January 2, 1943, recommended Hanford, finding that the other sites were deficient in one or more of the requirements (Needles, for example, was in an earthquake zone and would necessitate relocating a highway, and Blythe, which was 50 miles (80.5 kilometers) from the Mexican border, was excluded on security grounds). Hanford's only disadvantage was a lack of natural camouflage, owing to the flat, desertlike land. Formal acquisition of the land around Hanford began on February 9, 1943. Meanwhile, the MED would construct the X-l0 "pile" at Oak Ridge, to validate Fermi's research on controlled chain reactions.     In all, three reactors and three separation facilities were constructed at the Hanford Reservation during World War II. In September 1944, less than two years after the start of the unprecedented project, the first reactor went "critical," and soon separated plutonium was available for research and weapons use.     As at Oak Ridge, the MED relied on inexpensive construction labor and contracted out with private corporations to carry out most of the design and operations. E. I. Du Pont de Nemours and Company was selected to produce the uranium slugs and construct the reactors at the site. With the influx of tens of thousands of workers and technicians, new communities sprouted up and disrupted the lives of about five hundred residents in the nearby farming communities of Richland, Hanford, and White Bluffs, along with a handful of Wanaapum Indians who fished and foraged along the Columbia River. The purchase of the land itself involved complicated negotiations with the two thousand owners of the more than three thousand tracts that were required. Not only the living were inconvenienced; the area's cemeteries were all closed and the graves exhumed and relocated.     Weapons design and production took place at Los Alamos, run by the University of California under contract with the MED. The property was acquired in part from a boys' school at the site, although possession was disputed by Pueblo Indian and Hispano farmers, ranchers, and sheepherders living in the area. Until Los Alamos was founded in 1942, scientists had conducted weapons research at many different facilities across the country. By putting them all in one place, General Groves hoped to minimize security risks and foster collaboration. Isolated atop a high mesa in New Mexico, Los Alamos was ideal for both purposes. What Did the Manhattan Project Cost? On July 16, 1945, an experimental plutonium-fueled device, nicknamed the "Gadget," was detonated from a tower 100 feet (30.5 meters) above the New Mexico desert near Alamogordo. The test was code-named "Trinity." Less than three weeks later, on August 6, a uranium-fueled atomic bomb, nicknamed "Little Boy," was detonated over Hiroshima (this "gun-type" device was considered so reliable that it was never tested). On August 9 a third bomb, named "Fat Man" and a virtual copy of the Trinity device, was dropped on Nagasaki. Two days later, Japan surrendered and World War II ended.     The cost of building the facilities to produce these bombs far exceeded the government's expectations (see footnote 39). According to the AEC, actual Manhattan Project (NDRC, OSRD, and MED) expenditures through the end of 1945 totaled $1.9 billion in then-year dollars ($21.6 billion in 1996 dollars). Of that amount, uranium enrichment accounted for $1.2 billion (about $13.6 billion in 1996 dollars); plutonium production for $390.1 million (about $4.5 billion in 1996 dollars); and weapons research, design, testing, and production (including work at Los Alamos) for $143.7 million (about $1.6 billion). In addition, $103.4 million ($1.2 billion) was spent on raw materials, primarily uranium, including ore from Canada and Belgium, along with extremely pure graphite, fluorine, and other materials needed to produce separated plutonium, enriched uranium, and nuclear weapons. The costs of individual programs encompassed by the Manhattan Project are presented in table 1-1.     Only a handful of people knew the enormous sums of money involved. General Groves and other officials secured what was essentially a blank check for the MED from the congressional leadership. To facilitate the appropriations process without revealing too many details, senior War Department and military officials (not including Groves) briefed House leaders in February 1944 (seventeen months after the project got under way) about finances, construction, procurement, and schedules, as well as the project's overall connection to the war. "The Congressmen indicated their approval without reservation" and said further explanations "would not be necessary." The Senate leadership concurred. Thus most members of Congress remained "completely in the dark" about the MED's work. Atomic Energy Commission After World War II ended in August 1945, the MED, under the direction of General Groves, continued to manage the weapons complex. At least one more bomb was in the pipeline in August 1945, and Oak Ridge and Hanford remained operational under their wartime contracts. Gradually, the production of uranium-235 at Oak Ridge and plutonium at Hanford declined. Work at Los Alamos also wound down, but preparations for "Operation Crossroads" kept about one-eighth of the scientists busy. There was no question, however, that the program would continue after the war. At a meeting of the Interim Committee on May 31, 1945 (formed by Secretary of War Stimson to consider postwar policy options for the atomic bomb and including Stimson, Groves, Army Chief of Staff George C. Marshall, Oppenheimer, Lawrence, Bush, MIT president Karl T. Compton, Undersecretary of the Navy Ralph A. Bard, Assistant Secretary of State William L. Clayton, and Secretary of State-designate James F. Byrnes), Lawrence spoke forcefully in favor of continued production, recommending "that a program of plant expansion be vigorously pursued and at the same time a sizable stock pile of bombs and material should be built up" to ensure that the nation would "stay out in front." Later in the meeting, Byrnes "expressed the view, which was generally agreed to by all present , that the most desirable program would be to push ahead as fast as possible in production and research to make certain that we stay ahead and at the same time make every effort to better our political relations with Russia."     The only issue was under whose authority it would do so. Because the entire endeavor was conducted in strict secrecy, there was never any public or congressional discussion as to the profound military, economic, political, or environmental implications of this decision.     Meanwhile, Congress decided to replace the Manhattan Project with a civilian peacetime agency. In 1946, through the Atomic Energy Act (sponsored by freshman senator Brien McMahon), it established the Atomic Energy Commission, with broad powers to conduct, control, and regulate nuclear research. All materials, facilities, equipment, items, and property related to atomic energy research, and all property in the custody or control of the Manhattan Engineer District--including nine completed "Fat Man"-type bombs--were to be turned over to the AEC by January 1, 1947. The military aspects of the MED were incorporated in the new Armed Forces Special Weapons Project (which in turn became the Defense Atomic Support Agency, DASA, on May 6, 1959; the Defense Nuclear Agency, DNA, on July 1, 1971; and the Defense Special Weapons Agency on June 26, 1996).     Before this transfer was complete, the MED spent an additional $3.1 billion, most notably on the manufacture of nine new bombs, the development of more advanced plutonium weapons, continued research into thermonuclear weapons, and support for the Army-Navy Joint Task Force in conducting Operation Crossroads. Except for Crossroads, specific costs for these activities are unknown. On January 1, 1947, the AEC formally assumed control of the Manhattan Project. The transfer list included thirty-seven installations in nineteen states and Canada; and authority over 254 military officers, 1,688 enlisted men, and about 38,000 contractor employees.     As table 1-2 indicates, the AEC spent $1.6 billion during the last six months of fiscal 1947. The accuracy and completeness of this figure is uncertain. When the newly formed AEC went to Capitol Hill in 1947 to present its budget request for fiscal 1948, its chairman, David E. Lilienthal, wrote: The committee itself refused to look at the secret tables we had prepared, but I insisted on it; then we withdrew them. But the secret tables contained large lump sum amounts--140 million for this, 200 for that. The committee members were baffled, and we admitted frankly we didn't know how to present a budget when (a) we didn't have a set of books showing costs, since the Army's Manhattan District didn't have or keep any; and (b) to disclose details would be to breach security. As the AEC's controller subsequently explained: "The weapon--not the expense--was properly the primary consideration during the war. For this reason, financial controls as a tool of management were largely lacking.... In the main, Manhattan District financial management was aimed merely at justifying the reimbursement of expenditures made by cost-type contractors, in conformity with law and Government regulations." As a result, the commission had "an exceptionally poor basic program on which to work." AEC, ERDA, and DOE Weapons-Related Activities Between 1947 and 1996, the AEC and its successor agencies--ERDA and DOE--spent about $345 billion on nuclear weapons, roughly divided between materials production, on one hand, and weapons research, development, testing, and manufacturing, on the other. Construction at the weapons complex during this period cost an estimated $79 billion, and operations an estimated $227 billion. Nuclear Materials Production Costs. Official budget information suggests that the AEC, ERDA, and DOE have spent more than $165 billion to produce fissile and special materials for nuclear weapons, including an estimated 725 metric tons of HEU (estimated average enrichment of 93 percent), 103.5 metric tons of fuel and weapon-grade plutonium, and an estimated 225 kilograms of tritium since fiscal 1948. Of this amount, $76.7 billion was spent on operating costs, $35.1 billion on source materials procurement, and $51.1 billion on construction and capital equipment purchases (see appendix table A-1). An additional $1.6 billion was spent between 1968 and 1995 to safeguard these materials against theft or diversion while in DOE custody and (to a lesser extent) to implement production-related safety measures. (This amount is for program direction only; each facility has its own safeguards and security budget, which historically was not disaggregated from other spending.) Data on the unit cost of fissile and special nuclear materials, though tracked by the DOE, are classified. Whatever data exist on the total costs of producing each material (or even the total costs at each facility) also appear to be classified. It is therefore not possible to ascertain with any degree of specificity how much was expended just to produce, say, plutonium, highly enriched uranium or tritium (see figure 1-6).     When examined against the background of the ebb and flow of the AEC's materials production activities, the available time-consistent budget data fall into four distinct periods: 1948 to 1964, which was when fissile materials stockpiles rapidly increased; 1964 to 1980, a period of retrenchment in which the United States ceased producing HEU for weapons and sharply curtailed plutonium production activities; 1981 to 1988, which saw a rapid resurgence in materials production under the Reagan administration's military buildup; and 1989 to the present, which has witnessed the collapse of the materials production complex because of age and neglect and the reorientation of DOE's nuclear weapons missions. Expansion: 1948-64 . Not long after it was established, the AEC aggressively expanded U.S. capacity to produce fissile materials in response to the breakup of the wartime "Grand Alliance" and the deterioration, between 1946 and 1948, of U.S.-Soviet relations. Three important factors behind these developments were the domestic discord over the agreement negotiated at Yalta in February 1945 by President Roosevelt, British Prime Minister Winston Churchill, and Soviet leader Josef Stalin, which granted control of postwar Eastern Europe to the Soviet Union (in part to maintain the military alliance and end WWII); the resulting "German question"; and the Marshall Plan. Although the United States had a massive industrial mobilization base and a booming economy (both thanks to the war), not to mention the atomic bomb, the perceived superiority of the Red Army and the thought of it occupying Eastern Europe raised alarms in Washington about Soviet intentions and Western disadvantages should conflict break out.     To prevent war and maintain global influence, the Truman administration and its successors decided, among other things, to further embrace nuclear weapons, especially thermonuclear weapons. The first Soviet detonation of an atomic bomb in 1949 (which came as a shock to many observers), the fall of China to the Communists that same year, and the outbreak of the Korean War on June 25, 1950, seemed to confirm policymakers' worst fears about Soviet expansionism (embodied most strongly in NSC 68, issued in April 1950) and spurred the United States to accelerate its nuclear weapons program. Apprehension about possible Soviet thermonuclear weapons also played a significant role in the expansion of the U.S. effort.     One day after North Korean forces attacked South Korea, Senator Brien McMahon (Democrat of Connecticut), now chairman of the Joint Committee on Atomic Energy (JCAE), asked AEC commissioner Sumner Pike to estimate the cost of increasing planned nuclear weapon production rates by 50 percent over the next few years. Before the AEC could respond, Truman submitted to Congress on July 7 a supplemental appropriation of $260 million ($1.8 billion in 1996 dollars) for the AEC. McMahon supported the request and took the relatively small figure (3 percent of the overall 1951 defense budget) to mean that thermonuclear weapons were not very expensive and therefore were well within the reach of the Soviet Union. McMahon then asked William L. Borden, the JCAE's executive director, to assess the requirements for further expanding weapons production. Borden prepared a three-page memorandum outlining his belief that the Soviet Union was not only pursuing the hydrogen bomb but was likely well ahead of the United States. To keep the United States in the lead, Borden called for the establishment of a second Hanford, with three to five reactors of the existing graphite and the newer heavy-water design (according to intelligence reports at this time, the United States had only a one-pile advantage over the Soviet Union, and superiority in gaseous diffusion technology was also threatened; such dire analyses anticipated the bomber and missile "gaps" that were to materialize--and just as quickly fade away--later in the decade). "If we act to increase our supply of atomic weapons and they turn out to be unnecessary, we may lose a few hundred million dollars," Borden explained. "If we fail to produce these weapons and they do turn out to be necessary, we may lose our country."     Convinced of Soviet capabilities, the need for nuclear weapons in the U.S. defense posture, and the importance of research on these weapons, the Truman administration expanded the weapons complex during 1949-52. All the while the pressure for more nuclear weapons, especially large numbers of tactical weapons, and hence for fissionable materials, was building: it came from the JCAE, primarily Senators McMahon and Henry M. Jackson (Democrat of Washington); Robert LeBaron, assistant to the secretary of defense for atomic energy and chairman of the Military Liaison Committee (MLC), the DOD's link to the AEC: and the Joint Chiefs of Staff (largely following LeBaron's recommendations). The AEC, which preferred to base its production goals on a clear number of weapons rather than an arbitrary percentage increase in production, was troubled both by the huge costs associated with the increases and by the DOD's position that the AEC had no role in setting nuclear weapons policy but rather served essentially as a contractor to DOD, providing nuclear weapons and materials as required. (However, the AEC, as authorized by the Atomic Energy Act, retained custody of the nuclear stockpile, releasing weapons to the military only upon the authorization of the president. The JCS chafed at this arrangement, eventually gaining control of the arsenal from the AEC in stages from 1951 to 1967).     At a meeting with the MLC on October 5, 1951, Commissioner Henry Smyth, speaking for the AEC, stated that the military was moving far too quickly in deciding to increase production so dramatically. Given the cost of the proposals and their impact on the economy, Smyth argued, the decision to proceed lay not with the JCS or the AEC but with the president, the National Security Council, and Congress. The AEC, the MLC, and others were fully aware that the sizable outlays for the expansion program proposed in 1951 (and approved by Truman on January 16, 1952) would not have any meaningful impact on the size of the nuclear stockpile until 1956 and thus were of no use in addressing the threats raised by the Korean War (and could potentially divert resources necessary to prosecute the war). Indeed, Charles E. Wilson, with the Office of Defense Mobilization (and secretary of defense from 1953 to 1957), had written a memorandum criticizing the JCS for failing to justify the expansion program in light of these facts and during a meeting with Truman at the White House on January 16, 1952, argued that he saw no alternative, given the stated need for large numbers of weapons, although "the estimated requirements for critical materials and equipment had appalled him."     Between 1951 and 1955, the AEC obligated nearly $35 billion for construction at fissile material production sites, primarily to build new reactors and to expand the capacity of the gaseous diffusion plants. Carleton Shugg, the AEC's acting general manager, assessing the expansion proposed in 1950 (as a direct result of the outbreak of the Korean War) for 1951, considered the proposed expenditure of $883 million ($6.2 billion in 1996 dollars) in that one year to be a "fantastic sum," exceeding the peak annual expenditures of the Manhattan Project just seven years earlier. Completing the program, he realized, would clearly exceed the costs incurred during World War II. In 1953, at the beginning of the so-called 50-150 expansion program, the AEC obligated more than $19 billion for construction activities at materials production sites, or slightly less than the amount spent on the entire Manhattan Project during all of World War II.     As a result of these expenditures, both uranium enrichment and plutonium production capabilities were greatly increased. By 1954 the AEC had added four new buildings to the Oak Ridge gaseous diffusion plant and constructed a new five-building gaseous diffusion plant near Paducah, Kentucky. Two years later, a three-building plant was finished at Portsmouth, Ohio. Five new reactors--H, DR, C, KW, and KE--were constructed at Hanford and were producing plutonium by 1955. In addition, five new heavy-water reactors--R, L, P, K and C--were built at the new Savannah River Plant especially for producing plutonium, tritium, and other materials needed to build thermonuclear weapons. Large new reprocessing plants were built at both Hanford and Savannah River as well.     Operating costs also increased, although not as steeply as those for construction. Between 1950 and 1954, operating expenses increased threefold, from $710 million to more than $2.1 billion. By 1957, with all thirteen reactors and all three gaseous diffusion plants operating, the AEC had obligated more than $3.8 billion to produce fissile materials. These operations would make possible the rapid increases in the weapons stockpile to come (see box 1-1).     Obligations for procuring source materials, especially uranium, increased with the AEC's uranium enrichment and plutonium production capacities. Initially, the availability of uranium ore severely constrained Defense Department planning for nuclear weapons and AEC capabilities to produce uranium and plutonium. According to AEC documents, "the shortage of raw materials dictated the magnitude of military requirements. Not only were existing sources small in comparison with needs [in 1947] ... no known uranium source could supply the projected military requirements, and the extractive metallurgy of these sources had only begun to be studied."     During the closing days of World War II, the United States and Great Britain jointly sought rights to the high-grade uranium ore deposits in the Belgian Congo and moved to purchase lesser ores from gold mine tailings in South Africa and other foreign sources. When it was discovered that even lower grade ores could be processed to extract uranium concentrates, these foreign efforts were largely abandoned and significant resources were devoted to domestic production.     In 1948 the AEC stimulated domestic production by guaranteeing minimum prices for ten years for high-grade uranium concentrates and offered a $10,000 (then-year dollars, about $99,000 in 1996 dollars) bonus for the discovery of new high-grade deposits. Over the next twelve years, domestic raw materials production increased rapidly, as did procurement costs. Hence spending on raw materials increased throughout the 1950s. Eventually, actual uranium purchases by the AEC surpassed planned or even optimal processing requirements. Whereas the 1953 expansion assumed an optimal receiving rate of 9,150 tons of ore a year through the end of the decade, by 1958 domestic sources alone were producing nearly 18,000 tons of concentrate, while significant quantities of foreign uranium were also being purchased. Domestic production was then curtailed, but receipts from all sources did not peak until 1960, when the AEC purchased 34,600 tons.     By 1955 obligations for source materials were more than $1.5 billion, up from $334 million in 1950. By 1959 spending was in excess of $4.3 billion, and twenty-three ore-processing plants were operating in the United States. Curtailment: Mid-1960s through 1980 . The rapid growth of plutonium and HEU production in the 1950s led the United States to consider curtailing its production activities in the early 1960s. Satisfied that fresh material stocks, together with materials recycled from retired weapons, were sufficient to accommodate planned weapons production, the Johnson administration determined that the need to produce additional fresh plutonium and HEU had dramatically decreased.     These considerations led the United States to propose a fissile material production cutoff to the Soviet Union in 1963. The United States hoped to capitalize on the momentum generated by the recent signing of the Partial Test Ban Treaty (PTBT) and the Washington-Moscow Hotline Agreement, and it sought to further reduce tensions between the superpowers while retaining a clear nuclear advantage.     Citing the substantial U.S. advantage, the Soviet Union refused to curtail fissile material production. In fact, some aspects of the Soviet program, in particular the gas centrifuge program to produce enriched uranium, were only then reaching maturity and full-scale operations. Had the Soviet Union cut production, it would have been left with far less fissile material than the United States, and more than a decade's worth of research and development would have been wasted.     Despite the Soviet refusal, the Johnson administration took several steps to reduce fissile material production. First, it unilaterally stopped producing HEU for weapons in 1964. While the AEC would continue to produce enriched material, primarily low-enriched uranium (LEU) for commercial nuclear power plants and HEU for naval propulsion reactors, the gaseous diffusion plants ceased to play a central role in materials production for nuclear weapons.     Plutonium production also was scaled back dramatically in the mid-1960s. Between 1964 and 1971 the administration closed the original eight Hanford reactors and converted the N reactor, in operation between 1964 and 1966, to maximize electricity production (it would be reconfigured to produce weapon-grade material in the early 1980s). At the Savannah River Plant, two reactors were shut down between 1964 and 1968, while the remaining three continued to produce plutonium and tritium until mid-1988.     Reprocessing was also scaled back. At Hanford, operations ceased in 1972 (although they resumed in the early 1980s). Spent fuel from the N reactor was stored in large concrete pools on site beginning in 1972. The fuel-grade plutonium in the spent fuel was to be recovered in the future for use in the never-realized U.S. breeder reactor program. However, the F and H canyons at Savannah River continued to operate.     Production budgets soon felt the impact of these trends. In 1964 annual obligations for operating expenses were more than $2.5 billion, not including the additional billions dedicated to procuring uranium and other materials. Ten years later, spending on source materials had virtually disappeared, and annual operating expenses were nearly one quarter of what they had been. By 1979, total budget authority for materials production, including both operating expenses and capital equipment and construction purchases, had fallen below $1 billion for the first time ever. Resurgence: 1981-88 . With the start of the Reagan administration, trends in materials production spending again reversed direction. Several Soviet policies---including continued adventurism in Central America, Africa, and Asia; the decision to deploy mobile SS-20 intermediate-range ballistic missiles capable of targeting bases and cities in Western Europe; and the 1979 invasion of Afghanistan--soured U.S. decisionmakers on detente. The Reagan administration responded in part by expanding significantly fissile materials production to fuel a planned nuclear weapons modernization program calling for as many as 14,000 to 17,000 new warheads, according to some unofficial calculations. Between 1981 and 1988 the DOE obligated or was authorized to spend $14 billion, including more than $10 billion for operating expenses and $3.5 billion for construction and capital equipment (appendix table A-1).     Several programs were initiated or resumed under the Reagan administration to increase the U.S. fissile materials stockpile. The Hanford N reactor was reconfigured to produce weapon-grade plutonium beginning in 1982, Hanford was reprocessing again in 1983, and the Savannah River L reactor was restarted in 1985. These programs incurred significant costs: $112 million was spent in 1983-84 on restarting Hanford reprocessing and an initial $41.3 million on restarting the L reactor.     With the demise of the breeder reactor program in late 1983, the DOE also sought to increase weapon-grade plutonium stocks by recovering the fuel-grade material produced in the N reactor since 1966. This effort took two main tracks. First, in 1985, the DOE configured the Savannah P, K, and C reactors to produce "supergrade plutonium." This material was then blended with some of the N-reactor material to produce weapon-grade plutonium.     Second, scientists explored new methods of isotope separation. After the fuel-grade plutonium was recovered, the DOE proposed that the material should be enriched by using lasers at a new facility called the Special Isotope Separation (SIS) Plant, to be located at the Idaho National Engineering Laboratory (INEL; now known as the Idaho National Engineering and Environmental Laboratory, or INEEL). The SIS would outlast the second Reagan administration, with work continuing under President George Bush. Estimated research and development obligations for the SIS Plant, including a pilot-scale facility built at Lawrence Livermore National Laboratory, totaled $774 million before the program was canceled in 1990 owing to a lack of need and serious proliferation concerns. Collapse: 1988-93 . During the mid-1980s, citizens' groups living near Hanford, Savannah River, the Fernald Plant in southwestern Ohio, and other such sites became increasingly concerned that the restarted facilities posed unacceptable risks to the environment and to the health and safety of both workers and the surrounding populations. Early in the decade, a few groups managed to keep facilities closed or to postpone their reopening. The Natural Resources Defense Council, for one, challenged DOE efforts to restart the L reactor and succeeded in delaying restart by two years. However, it was not until the Chernobyl accident in April 1986 that these grass roots groups began to achieve real success. Like Chernobyl, the Hanford N reactor was graphite-moderated. Unlike Chernobyl, it did not even have a system for monitoring or controlling hydrogen. It also lacked a containment vessel. Indeed, only the five reactors at the Savannah River Site had containment vessels and none of the fourteen military production reactors built and operated in the United States had the pressurized steel and reinforced concrete containment building required by law for all civilian power reactors.     Under mounting public and congressional pressure, the DOE eventually acknowledged that environmental, safety, and health protection measures at production reactor sites and at other materials complex facilities were inadequate or nonexistent. Secretary of Energy John Herrington tried to persuade Congress to obligate billions of dollars to rebuild the complex and continue weapons production, but severe environmental and safety problems eclipsed stockpile concerns and the strategy backfired (see chapter 9). Ultimately, the DOE's failure to comply with environmental standards or adequately invest in maintaining its increasingly brittle infrastructure--despite reports of serious deficiencies as early as 1981--brought production and reprocessing activities to an abrupt halt in 1988.     Spending at materials production sites did end, however. Under George Bush and Bill Clinton, the DOE committed almost $12 billion to the materials production facilities--nearly as much as President Reagan spent during his two terms in office--even though the United States has not produced any new plutonium or tritium since mid-1988. Much of the DOE's investment since 1988 has gone toward upgrading facilities to satisfy environmental, health, and safety concerns. Between 1988 and 1993, the DOE spent more than $2 billion to restore the Savannah River K, L, and P reactors alone, although efforts to restart them were ultimately abandoned, and they never resumed operation.     In addition, both the Bush and Clinton administrations sought to resume the production of tritium, which had stopped when the K reactor shut down on April 10, 1988. Although plutonium stocks eventually were deemed more than sufficient to permit the reactor shutdowns, some officials within the Reagan and Bush administrations argued that because tritium had only a 12.3-year half-life such shutdowns would render the U.S. nuclear arsenal impotent, and its weapons would have to be dismantled and cannibalized. The Bush administration initially proposed building a New Production Reactor (NPR) at SRS to produce tritium but deferred construction in September 1992, citing a lack of immediate need for new tritium. The Clinton administration terminated the reactor program in 1995, after $2.2 billion in expenditures, but has not given up efforts to produce tritium entirely. It spent $182 million on basic research on nonreactor options such as particle accelerators in 1993, $75 million in 1996, and $150 million in 1997. Weapons Production Costs. Between 1948 and 1996 the DOE and its predecessors spent an estimated $178 billion on weapons research, development, testing, and production. This estimate, based on data in appendix table A-2, includes $150 billion in operating costs and $28 billion in construction and capital purchases (see figure 1-7). From 1948 through 1956 these costs kept pace with the steady increase in weapons production as new facilities were built and new weapons designs were introduced. They reached their first peak in 1964, the period of the most rapid buildup in the stockpile. Since 1965 they have had a mixed pattern. Steady Growth: 1948-56 . The AEC inherited many wartime facilities, notably the Manhattan Project's Los Alamos laboratory. As in the case of materials production, weapons production expenditures increased as the tension between the United States and the Soviet Union mounted. Advances in the Soviet nuclear arsenal and breakthroughs by weapons designers in the United States also helped push up expenditures.     The only constraint on the growth of the weapons production complex in the 1950s was the supply of nuclear materials. Scientists at Los Alamos were already designing new atomic weapons at the end of World War II, trying to optimize the yield of the weapons without increasing the amount of fissile materials and thus maximize the ratio of yield to weight. Such concepts as levitated cores, composite pits, and "boosted" weapons were considered during the Manhattan Project but not acted upon until the war had ended.     The AEC also spent increasingly large sums of money on thermonuclear research after World War II. Even before President Truman ordered the crash development of the "so-called hydrogen or super-bomb" on January 31, 1950, the AEC had investigated the question of building thermonuclear weapons. Significantly, with the exception of certain air force planners and members of the AEC's MLC, the military--particularly the Joint Chiefs of Staff--was not active in the early debates. In fact, the JCS concluded in January 1948 that "the majority of targets do not require a more powerful bomb because of area limitations [that is, hydrogen bombs were deemed too big for all but a limited number of targets]."     In October 1949, shortly after the Soviet Union tested its first atomic bomb, the AEC's General Advisory Committee, chaired by J. Robert Oppenheimer, was asked to make recommendations on pursuing a highly accelerated program to build the Super (as the thermonuclear program was then known). In its October 30, 1949, report, the six-member GAC unanimously opposed such a program, arguing that the extreme dangers to mankind inherent in the proposal wholly outweigh any military advantage that could come from this development. Let it be clearly realized that this is a super weapon; it is in a totally different category from an atomic bomb.... We are alarmed as to the possible global effects of the radioactivity generated by the explosion of a few super bombs of conceivable magnitude. If super bombs will work at all, there is no inherent limit in the destructive power that may be attained with them. Therefore, a super bomb might become a weapon of genocide.... In determining not to proceed to develop the superbomb, we see a unique opportunity of providing by example some limitations on the totality of war and thus of limiting the fear and arousing the hopes of mankind.     The AEC split on the recommendation of the GAC, with the majority agreeing not to proceed with a crash program. But physicists Edward Teller, Ernest O. Lawrence, and Luis Alvarez challenged the AEC's decision and aggressively promoted the idea that scientists had a fundamental duty to pursue new knowledge, whatever the costs. AEC chairman Lilienthal, in a meeting with Truman on November 9, 1949, expressed deep-seated reservations about the hydrogen bomb, arguing that it would "intensify in a new way" the arms race and signal that the United States had "abandoned our program for peace and are resigned to war." Staking national security on the Super, he asserted, would result in a "costly cycle of misconception and illusion."     The JCS responded to the GAC's recommendation in a January 13, 1950, memorandum to Secretary of Defense Louis A. Johnson siding with Teller and his colleagues: "Such a weapon would improve our defense in its broadest sense, as a potential offensive weapon, a possible deterrent to war, a potential retaliatory weapon, as well as a defensive weapon against enemy forces.... [T]he United States would be in an intolerable position if a possible enemy possessed the bomb and the United States did not." Through the end of fiscal 1952, just before the Ivy-Mike test, total Manhattan Project and AEC operating costs on thermonuclear weapons research exceeded $586 million (in then-year dollars).     As the capacity to produce fissile materials increased, weapons production costs also increased. Until 1957 obligations for operating expenses remained below $2 billion a year (appendix table A-2). The growth of the stockpile and low annual production rates through 1957 suggest that the total obligations for operating expenses from 1948 to 1957 ($11 billion), were principally for research, development, and testing.     In comparison with the large obligations for building additional nuclear reactors, reprocessing plants, and gaseous diffusion plants, the cost of adding to or building new weapons production facilities was small. At this point, there was less need to build new facilities. Many weapons production complex sites were left over from the Manhattan Project, while industrial plants that would contribute nonnuclear components only needed to be converted from other defense contracting work. Of the new facilities, most were built or converted in the late 1940s and early 1950s. Construction at the Mound Laboratory began in 1947, and at the Kansas City Plant in 1949. Operations at NTS and Rocky Flats began in 1951 and 1952, respectively. The Pantex Plant (originally used to load TNT into conventional shells during WWII) was renovated in 1950 and began assembling nuclear weapons in 1952. The Lawrence Radiation Laboratory was founded in 1952. In 1956, a branch of Sandia Laboratories opened at Livermore. Peak of Buildup: 1957-65 . By 1960 the output of AEC's plutonium production and uranium enrichment plants was sufficient to produce nuclear weapons at an unprecedented rate: more than 7,000 warheads annually, compared with only about 140 in 1950-51. Between 1959 and 1961, the United States produced 14,884 warheads, bringing its total stockpile to 24,173 warheads and bombs, or three times the total number in 1957.     The growth in R&D and production obligations at the end of the decade is consistent with the development and manufacturing of large quantities of new and more sophisticated weapons. Research costs between 1957 and 1961 increased by $202 million, while production obligations nearly doubled, from $1 billion in 1957 to $1.9 billion during the same period. Obligations for both R&D and production would remain above $1 billion a year for the rest of the decade.     By contrast, obligations for nuclear testing at the end of the 1950s steadily declined. This reflected the Eisenhower administration's decision, effective October 31, 1958, to observe a one-year nuclear testing moratorium (joined by the Soviet Union and the United Kingdom and subsequently extended through the end of 1959) and to attempt to negotiate a comprehensive test ban with the Soviet Union. During the course of these negotiations, in 1960 and 1961, both countries refrained from testing despite the expiration of the joint moratorium. When General Secretary Nikita Khrushchev announced on August 30, 1961, that the Soviet Union intended to resume testing--and then proceeded, between September 1 and November 4, to detonate fifty-nine nuclear devices, including a massive 50-megaton blast (a reduced yield version of a 100-megaton bomb)--U.S. officials were taken by surprise, and the AEC scrambled to resume testing. (Within two weeks the first U.S. test was conducted underground at NTS; eight more would follow by the end of the year. It was not until the following spring that the United States resumed atmospheric testing, reflecting the difficulty and expense of testing at the Pacific Proving Ground.) The test ban negotiations eventually resumed during the spring of 1962. Following the Cuban Missile Crisis in October 1962, which sobered both Khrushchev and President John F. Kennedy on the potential for nuclear war, the two sides continued to talk. Concerns about verification and the opposition of the JCS to a total ban (see footnote 112) eventually led to the Partial Test Ban Treaty, which prohibited all nuclear testing in the atmosphere, underwater, and in outer space.     As a result of the moratorium, testing expenditures declined from $473 million in 1958 to $124 million in 1961. Testing aboveground resumed briefly in 1961, before the PTBT forced all nuclear testing underground. The decision to conduct underground tests freed the U.S. testing program from the constraints imposed by seasonal weather changes and also limited (but did not eliminate) the incidence of nuclear fallout and its impact on the environment, an issue that had made aboveground testing controversial from the beginning. Now that the AEC was able to conduct more tests per year, costs again rose. In addition, the cost per test increased because preparing for an underground test necessitated the drilling of test and monitoring shafts. These factors pushed annual testing obligations beyond $1 billion in 1964, where they stayed for the next several years.     With the resumption of nuclear testing, the AEC ran into serious public relations problems. Attempting to cast nuclear weapons as a constructive rather than destructive force, it proposed Project Plowshare, which would demonstrate the use of atomic explosions to dig harbors, cut through mountain passes, and perhaps even dig a new Panama Canal. One group, which included Edward Teller, proposed detonating nuclear weapons in natural gas fields to stimulate the production of that resource. Studies of that concept began in 1958, with twenty-seven nuclear tests conducted in Nevada, Colorado, and New Mexico from 1961 to 1973. Although the AEC claimed that fallout from such projects would be minimal, experience proved otherwise. The PTBT also severely hampered the ability to test the basic principles of the program. That, coupled with fierce public opposition and a decrease in available funds due to the escalating costs of the war in Vietnam, led to Plowshare's ultimate demise in 1977 (by which time the program was known as Peaceful Nuclear Explosions). Costs for the program were more than $770 million. Mixed Pattern: 1965 to the Present . Between 1965 and 1975 the total budget authority, including operating expenses and capital purchases, at first rose (to $4.8 billion in 1969) and then dropped below $3 billion by the mid-1970s. Then, during the late 1970s and continuing into the early 1990s, annual spending increased to levels well above the peak production years of the late 1950s. Research and Development. Between 1966 and 1996 research and development budget authority totaled an estimated $34 billion, or approximately 27 percent of the overall weapons program operating expenses. As appendix table A-2 shows, R&D expenditures divided among the three nuclear weapons laboratories and other DOE laboratories surpassed $1 billion a year for the better part of the 1960s before declining during the next decade. Indeed, estimated annual operating costs for R&D in 1979 were $278.3 million lower than they had been in 1966.     These trends reversed course toward the end of the Carter administration with the commencement of the Stockpile Improvement Program (SIP). In 1979 the DOD and DOE initiated the SIP to identify both strategic and tactical warheads and weapon systems that needed to be "modernized, retired, or replaced." Embracing this effort, the Reagan administration rapidly increased nuclear weapons R&D to well over $1 billion a year by the middle of the decade.     Today research and development spending often is committed to ongoing, "big-ticket" programs, such as lasers, particle accelerators, or other sophisticated test and research facilities that are designed to simulate the conditions of a nuclear explosion in the absence of actual full-scale tests. The inertial confinement fusion (ICF) program is such an attempt, with research efforts divided between the three weapons labs. In the ICF program, high-powered lasers are focused on a small target of tritium and deuterium, causing the target materials to fuse. Since 1974 this program has cost approximately $4 billion, which includes operating and capital expenses. The DOE is currently building a new, multibillion ICF facility called the National Ignition Facility (NIF) at Livermore. NIF has a projected $4.5 billion price tag (not factoring in inflation), which includes annual operating costs for its expected thirty-year life span. Proponents of NIF say it will help maintain confidence in the nuclear stockpile in the absence of nuclear testing and keep weapons scientists from leaving the weapons laboratories by allowing them to conduct cutting-edge physics experiments. Critics counter that crucial precursor experiments to demonstrate NIF's validity will not be completed until well after construction has begun and that the program is actually a payoff to the laboratories in exchange for their support for the CTBT.     A second major program to receive significant R&D funding during the 1980s and early 1990s was the effort to develop a space-based X-ray laser to shoot down Soviet ballistic missiles. Developed in conjunction with President Reagan's Strategic Defense Initiative (SDI), this program proposed using the detonation of a nuclear device to generate and direct beams of focused X-ray energy at Soviet ballistic missiles, destroying them in their first few minutes of flight.     Press accounts in October 1987 revealed that two years earlier Roy D. Woodruff, a former associate director for defense systems at Livermore, had charged that Edward Teller and others at the laboratory had provided many senior policymakers, including President Reagan's science adviser George Keyworth and CIA director William Casey, with "overly optimistic, technically incorrect" information about the program (they suggested, for example, that the laser could be made as small as a desk and could fire up to 100,000 separate laser beams to destroy ballistic missiles as they were boosted into orbit). According to Woodruff, laboratory director Roger E. Batzel was "fully aware" of the misleading information but refused to correct it and forbade Woodruff to do so. Unable to reign in the erroneous statements and unwilling to make the classified matter public, Woodruff resigned as associate director on October 29, 1985 (his charges were later leaked to the press). A 1988 GAO report and a subsequent congressional investigation confirmed Woodruff's accusations.     This misinformation altered, perhaps profoundly, the course of and expenditures for SDI, as well as strategic arms control negotiations with the Soviet Union. At the October 1986 summit in Reykjavik, Iceland, President Reagan adamantly refused to limit SDI research and testing to the laboratory (thus dashing any hope that all nuclear weapons would be eliminated within ten years, as agreed in earlier talks at the summit between Reagan and Soviet President Mikhail Gorbachev), at least in part because he believed that the X-ray laser--then the centerpiece of SDI--was a viable weapon system. Yet, had SDI been limited and this sweeping and historic agreement enacted, the threat SDI was designed to address might by now have disappeared. Between 1984 and 1992 (when the X-ray laser was effectively canceled), the DOE labs obligated an estimated $2.2 billion of R&D funds to this and other weapons concepts making up the directed-energy weapons program. This cost is covered in chapter 4 and does not count toward the overall total in this chapter.     After leveling off during the Bush administration, expenditures once again began to rise under President Clinton. Much of this funding has been directed to basic research. In recent years, the DOE has assigned high priority to "the fundamental scientific, engineering and facility capabilities" that underlie specific nuclear weapons research, even though there are currently no plans to develop new weapons. Proponents of technology-based research argue that even under a comprehensive test ban weapons labs need to maintain a certain level of expertise in order to ensure that the nuclear stockpile remains safe and reliable in the absence of nuclear tests. Much of the funding under the Bush and Clinton administrations has gone to such activities. In fiscal 1996, the DOE moved in this direction under the stockpile stewardship and management program. Projected average annual costs over the next decade for stockpile stewardship (formerly R&D, testing, and weapons production)--$4.5 billion (in 1998 dollars)--are now higher than the average annual costs during the cold war ($3.6 billion). Nuclear Testing. The United States formally discontinued nuclear tests aboveground, underwater, and in outer space following the enactment of the Partial Test Ban Treaty on October 10, 1963. However, 825 underground tests were conducted at NTS beginning in 1957, the last in September 1992. Although the PTBT outlawed all but underground tests, a little-noticed provision, known as Safeguard C, allowed the United States to maintain the capability to resume atmospheric testing promptly, should national security imperatives require it to do so. This safeguard was secured by the Joint Chiefs of Staff in exchange for the military's support for the treaty's ratification. For thirty years, the DOE (and its predecessor agencies), working with the Defense Nuclear Agency and its predecessors, maintained Safeguard C facilities in Nevada, Hawaii, and Johnston Island (the site of a command post shielded against the effects of nuclear blasts), about 800 miles (1,287 kilometers) southwest of Honolulu, Hawaii. If a president decided to resume testing, scientists and engineers at the post would remove Thor ballistic missiles and nuclear warheads from storage and would mate and launch them. The program was equipped with ships, aircraft, laboratories, launch pads, and a dedicated staff to keep everything ready for use on short notice. Up to 1993--when Congress became aware of the program and canceled it--the DOD and the DOE and its predecessors had spent at least $1.6 billion, and perhaps as much as $3 billion, on Safeguard C.     Testing costs between 1966 and 1996 were approximately $20 billion, or about 20 percent of the total weapons budget since 1966. Annual costs are listed in appendix table A-2. The expenditures on nuclear testing, like those for R&D, rose steadily through the 1960s, declined during the 1970s, and increased rapidly in the 1980s. After peaking in 1969 at nearly $1.4 billion, spending began to decline. By 1974, as superpower relations warmed under detente, spending on nuclear testing dropped to $515 million in new budget authority. New test ban initiatives, such as the Threshold Test Ban Treaty (TTBT) in 1974 and the Peaceful Nuclear Explosions Treaty (PNET) in 1976 were initialed by the United States and the Soviet Union, and in 1977 new talks on a comprehensive test ban were begun. Testing expenditures during the 1970s were regularly one-half of expenditures during the previous decade.     Under Presidents Reagan and Bush, this trend was reversed with estimated annual operating expenses for nuclear testing reaching as high as $700 million or more in the mid-1980s. Between 1981 and 1992, when the United States conducted its most recent nuclear test, the DOE spent some $7.2 billion on testing. Production and Surveillance. Producing new warheads and maintaining the operational readiness of existing warheads is by far the most expensive aspect of the nuclear weapons program. Between 1966 and 1996 the DOE and its predecessor agencies spent $53 billion on weapons production activities, or 49 percent of the total operating budget (appendix table A-2). In every year of this period except 1976 and 1977, new budget authority for weapons production and surveillance was in excess of $1 billion. Production spending fell during 1966-69, rose to more than $1.4 billion per year in 1971-72, fell again (to $953 million) in 1977, then resumed its climb. Annual spending exceeded $2 billion in 1983--the first time in nearly twenty years--and remained in excess of $2 billion a year for the next ten years.     Paradoxically, spending increased while the number of stockpiled weapons declined. According to reliable, unclassified estimates, the stockpile peaked at about 32,200 warheads in 1966 and declined to approximately 10,400 in 1996. In fact, this overall decline was irregular. After initially declining in the late 1960s, the stockpile increased to about 28,500 warheads in 1973. Much of this increase was due to the production of multiple independently targeted reentry vehicle (MIRV) warhead systems on intercontinental and submarine-launched missiles. The strategic component of the U.S. stockpile increased by 5,300 bombs and warheads between 1966 and 1974, even as the total stockpile declined from 32,200 warheads to 28,300 warheads over the same period.     The next period of significant budgetary increase--during the Reagan administration--also corresponds to increases in the strategic arsenal. Calling for substantial modernization, the administration forged ahead with new bombs and warheads for several new weapons systems. Many systems, notably the W87 (MX), W88 (Trident II/D-5), the W-80 (Air-Launched Cruise Missile), and the B-83 gravity bomb, replaced existing weapons, while others, such as the W79 artillery shell, W84 (Ground-Launched Cruise Missile), and W85 (Pershing II Missile) were deployed along with existing forces (see table 1-3).     Stockpile maintenance also pushed up costs even as the arsenal dwindled in size. The AEC, ERDA, and DOE periodically disassembled and inspected stockpiled warheads to certify that they would explode as designed. Since the AEC began its Stockpile Evaluation Program (SEP) in 1958, more than 13,800 weapons of forty-five types have been disassembled, inspected, and tested, and only 400 findings have been deemed "actionable," requiring a modification to either the warhead or the production process that created it or a lowering of the warhead's presumed yield or reliability. Most of these findings occurred within the first few years of a warhead's entry into the stockpile; as weapons age, fewer fixes have been required. What age-related problems have occurred are almost entirely within the realm of nonnuclear components and thus can be fully assessed and corrected without resorting to nuclear tests. Between 1980 and 1992, 1,414 weapons were disassembled, inspected, and ultimately disposed of while another 2,449 were disassembled, inspected, and then reassembled using original or replacement parts. The weapons in this latter category were then redeployed. This work continues today (see chapter 5). Warheads were first disassembled at Pantex (or one of the other assembly/disassembly plants) before the component parts were returned to their points of origin. Rocky Flats (and later, Los Alamos) inspected plutonium pits, while the HEU secondaries were sent to the Y-12 Plant at Oak Ridge. The Savannah River Site examined tritium reservoirs, and other nonnuclear components were evaluated at the Kansas City Plant, Mound Laboratory, and the Pinellas Plant.     The cost of warhead inspections and dismantlements is not disaggregated from other production-related costs (estimated costs since fiscal 1991 are included in chapter 5). As the United States retired large numbers of warheads in the 1970s, they were dismantled at the Burlington AEC Plant and at Pantex, and their components and materials were recycled into new warheads or discarded as waste. In the late 1980s and early 1990s, as the United States greatly reduced its active stockpile of tactical nuclear bombs and warheads deployed in Europe and elsewhere, retirement and dismantlement activities increased, keeping so-called production costs high. All warhead production ceased in 1990, but the DOE is at present upgrading a facility at Los Alamos National Laboratory known as TA-55 to give it the capability to produce up to twenty pits a year. This capability was demonstrated in March 1998, and full production is planned to be available by 2007, although problems encountered in modifying the necessary facilities and the resulting cost overruns threaten to delay this.     In 1994 the nuclear weapons laboratories formed the Nuclear Weapons Information Group (NWIG) to document and archive their nuclear weapons engineering knowledge to allow the stockpile to be maintained into the future and--if so desired--provide the data necessary to resume production. This multiyear effort was spurred by the Defense Nuclear Facilities Safety Board, which in December 1993 urged the DOE to develop a means of retaining the collective knowledge of fifty years of nuclear weapons production so that it is not lost as facilities shut down and personnel retire. John D. Immele, director of nuclear weapons technology at Los Alamos, told the Los Angeles Times in 1995: "You have to have the intellectual capability to respond to future developments. This lets our adversaries know that we still know what to do." With an annual budget of about $14 million, researchers are scouring old safes and file cabinets for useful materials, which are then recorded on optical disks, magnetic tapes, and printouts. For the time being, each lab is archiving its own materials, but future plans call for a secure computer network that will allow the labs and other nuclear weapons facilities to share information. By August 1995 researchers at Los Alamos had collected about 800,000 drawings, 1 million photographs, 2,000 videotapes, and 7 million pages of documentation. As part of the project, current and former weapons designers are being interviewed and videotaped to preserve knowledge that would otherwise be lost. New technology has been developed to permit indexed keyword searches of these materials to allow users to locate information about specific warheads or production processes. At Los Alamos, this effort is called the Nuclear Weapons Archiving Project; at Sandia, the Knowledge Preservation Project; and at Livermore, the Nuclear Weapons Information Project. What Do Nuclear Bombs Cost? The government has never disclosed the cost of individual nuclear warheads and bombs. However, over the past half century the occasional lapses of government censors have revealed selected warhead cost data for a handful of systems. An otherwise heavily censored transcript from a 1973 congressional hearing on nuclear weapons and technology, for example, contains this question: "Whereas the conventional shell for the 155-millimeter weapon [the W74] costs $191 each, the nuclear shell in production costs $452,000 each; and, whereas the 8-inch shell [the W75] in production will eventually cost $56 apiece, the nuclear shell in production cost over $400,000 apiece?" Within a month of this hearing, these weapons were canceled by the JCAE and never produced because they were considered to be "more of the same." The committee stated it was important to develop weapons that "can utilize the latest glide bomb technology, radar guidance and other remote controlled delivery means." In constant 1996 dollars, the unit cost of these weapons would have been $1.5 million and more than $1.3 million each, respectively. However, the army and the nuclear weapons laboratories continued to pursue the nuclear artillery shell and later in the decade proposed building the W79 8-inch shell and the W82 155-millimeter shell, both of which were to be "enhanced radiation warheads," commonly known as neutron bombs (the W79 was produced beginning in 1981, but the W82 was canceled in 1983; a non-ER version was later canceled in 1992). The W82 would have been even more expensive, estimated at more than $4 million a copy.     Congressional testimony in 1981 revealed that the W84 warhead for the Ground-Launched Cruise Missile was expected to cost $630 million for 530 warheads, or $1.1 million apiece ($1.9 million in 1996 dollars). The actual cost may have even been higher, owing to a shortened production run (400 were built but were retired in the late 1980s under the Intermediate Nuclear Forces [INF] Treaty) and production problems.     In August 1990 the GAO inadvertently revealed the W80-1 warhead costs for the Advanced Cruise Missile (ACM) to be about $720,000 apiece ($845,000 in 1996 dollars), on the basis of the eventual procurement of 1,461 ACMs. In the face of rising costs and operational deficiencies, only 460 were built and only 400 warheads were produced for these; as with the W84, these changes almost certainly pushed up the cost. The W80-1 is also used in the Air-Launched Cruise Missile (ALCM) and a variant, the W80-0, is in the Sea-Launched Cruise Missile (SLCM), so costs for these systems are likely similar (see footnote 130).     The preceding is all that has ever been released (albeit in error) from official sources on the cost of nuclear warheads since 1945. Since many of these weapons no longer exist and all production, not to mention the cold war, has ceased, we hoped that the DOE might be willing to release historical data on unit costs. To make matters easy, we requested a declassified copy of a 1985 presidential study conducted at the behest of Congress (which at the time was greatly concerned about the escalating cost of nuclear warheads). This study contains tables and other data on the costs of nuclear weapons going back to 1960 and thus includes most of the information we needed to conduct our analysis. Although large portions of the report and its supporting documentation were ultimately released to us after nearly two years, all of the data pertaining to warhead costs (with a single partial exception) were deleted. Despite repeated efforts to obtain any additional data from this report, the DOE (in conjunction with the DOD, which under the Atomic Energy Act must concur in any release of formerly restricted data) has steadfastly refused our requests.     Why such secrecy, even today? The information, it seems, remains classified because "it is counter to U.S. nonproliferation goals to tell proliferants precisely how much special nuclear material is used in a weapon, or to provide details on the mix of fuel materials the United States has found optimal for its weapons." Such reasoning makes good sense, which is why we scrupulously avoided asking for the cost of the nuclear materials themselves, requesting data on the total cost of the weapon, on a unit basis, and, where possible, broken down by its components (with the cost of the nuclear materials lumped together under the "physics package" category). We had no desire to thwart U.S. nonproliferation objectives; on the contrary, the release of such data could bolster those objectives by revealing the high cost of such weapons. Fundamentally, we wanted to report to taxpayers what they had received for their multibillion dollar investment in nuclear deterrence. To argue that decades-old data on the total unadjusted cost of a particular type of weapon would somehow reveal the precise quantities of materials contained within it is farfetched, akin to being given the price of an automobile and from that and that alone ascertaining the exact types and quantities of materials used to assemble it and how all those parts function together.     In fact the GAO reported in 1972 that for at least the first twenty-five years of its existence the AEC had no means of calculating accurately discrete warhead costs, and in any case its accounting system "was not designed to accumulate production costs by weapon system." Thus production costs for each weapon system delivered during the year, as computed by the AEC, represented the standard costs of the system adjusted for a share of the variance between total standard costs and total production costs incurred during the year for all weapon systems. The AEC assigned a share of the total variance to each weapon system on the basis of a uniform percentage, regardless of the variance experienced in producing each system. This system created significant distortions. For example, the unit costs incurred by one contractor during fiscal 1970 for its part of a warhead were more than the total unit costs that the AEC attributed to the production of the entire warhead for the year. According to the GAO, when the AEC provided estimates of the total costs of producing individual weapon systems, its manner of assigning cost variance to individual weapon systems wag not a satisfactory basis for identifying cost growth that may have occurred on such systems. To complicate matters, contractors were permitted to develop production data from their own cost accounting systems (subject to AEC approval).     In 1977, a new cost-estimating procedure was formally adopted under Phase 2A (Design Definition and Cost Study), and the Weapon Design and Cost Report (WDCR) was established. The 1985 presidential study--requested by members of Congress frustrated with the DOE's explanation of weapons costs--found that the DOE's reporting of costs to Congress was inadequate and that the department failed to use the production costs of individual weapons "as a basis for program management," and declared: "DOE's apparent reluctance to adopt customary, and thus more understandable, concepts of cost measurement and presentation requires serious attention."     AEC and DOE data on weapons costs appear to have consisted largely of estimates, and the way those estimates were derived, especially before 1977, makes it highly questionable that they accurately represent the total cost of each type of nuclear warhead. Of the sixty-five types of weapons actually produced since 1945, evidently only two--the W87 (MK/Peacekeeper) and W88 (Trident II/D-5)--were ever subjected to the improved and more rigorous accounting method (comprising a scant 3 percent of all warhead types ever manufactured). Therefore, any attempt to use these data to obtain any insight into the inner workings of any particular type of weapon would be fraught with tremendous methodological difficulties.     Because the DOE released figures in June 1994 documenting the total number of weapons built, dismantled, and in the stockpile from 1945 through 1961 (release of complete data after 1961 was overruled by DOD, on the grounds that it would reveal information about weapons currently in the stockpile), a gross calculation can be made from the data in appendix tables A-1 and A-2 of the average annual cost of warheads produced from 1948 through 1961. While this method is far from satisfactory, not least because larger weapons utilizing more nuclear materials would tend to be more expensive than smaller ones, it is unfortunately the only method available that provides any degree of accuracy for any given year. Another drawback of this approach is that the AEC and DOE historically entered the cost of a particular warhead on the books only at the time the first production units were delivered, even though significant costs may have been incurred in earlier years (another practice that tends to distort actual costs).     This method of calculation reveals that when the cost of weapons research, development, testing, production, and materials production are combined and divided by the total number of warheads produced in a given year, the average per unit cost of nuclear weapons declined from $92.5 million each in 1948 to $2.1 million each by 1961. When materials production costs are excluded--which might be considered reasonable given that in any one year the total spent on materials was applicable to more than one year, especially in light of the tremendous sums expended on the expansion programs in the early 1950s--the figures are $33.2 million and $633,000, respectively. However, these latter figures ignore all nuclear materials costs and thus dramatically understate the true cost of the weapons in any given year.     Another approach is to add the cost of RD&T and production ($151.4 billion), and materials production ($158.4 billion) from 1948 (the first available data, aside from the Manhattan Engineer District) through 1990 (the last year of actual production) and divide it by the estimated total number of warheads built (70,299, see table 1-3). The resulting figure is $4.4 million per warhead. Adding in the estimated and known costs attributable to DOD weapons research and testing ($37.4 billion) raises this figure to $4.8 million per warhead. As before, this is a less than satisfactory approach, masking as it does both those weapons that were very expensive along with those that were less so. Furthermore, not all of these costs are attributable to the actual production of new weapons; in 1985, for example, "nearly one-third of the [DOE production and surveillance] workload was in support of stockpile maintenance, stockpile reliability, process development, and other weapon programs." Such an estimate also provides no sense of how costs have fluctuated over time and whether costs per unit had gone down over time, as production increased and the technology improved (as would be expected in other large-scale industrial processes).     The 1985 presidential task group report found that the average unit cost of production (in constant dollars) increased at an average rate of 7 to 8 percent a year from 1962 to 1985. This trend reflected the increasing emphasis on more complex and technically sophisticated weapons designed to provide greater flexibility of employment, enhanced safety and control, and improved yield efficiency.     The W79 and W82 AFAPs and the W84 for the GLCM all "experienced design and production difficulties which have increased costs above early estimates." In addition, "the costs of weapons which rely to a greater extent on well-understood technology and processes, or existing components appear to be estimated with considerably greater accuracy than more challenging designs." Absent actual data, however, and forced to approximate costs via gross annual averages, these important trends remain invisible. Military Expenditures With the exception of the Manhattan Project, this chapter has thus far concentrated on expenditures by civilian agencies. However, the DOD--through the Defense Special Weapons Agency (DSWA) and its predecessor agencies, the Defense Nuclear Agency (1971-96), the Defense Atomic Support Agency (1959-71), and the Armed Forces Special Weapons Project (1947-59)--has allocated billions of dollars to nuclear weapons research, development, and testing. The individual services also supported their own research efforts to develop new weapons. The DOD appears to have spent more than $37 billion to support nuclear weapons research, development, and testing (see figure 1-8). A more accurate estimate is not possible, given the lack of complete expenditure data before 1962 and possible classified expenditures not yet publicly available.     Most of these expenditures went to support studies of the effects of nuclear explosions on different pieces of equipment, communications systems, weapons systems, and materials and to devise techniques to make these more survivable. Some nuclear weapons tests, such as those including military maneuvers at Camp Desert Rock (the Nevada Test Site) in the early- to mid-1950s, studied blast effects on tens of thousands of troops, many of whom were exposed to dangerously high levels of radiation. The tests were designed to simulate a nuclear battlefield and provide both combat training and data on the psychological effects of nuclear attack on troops (see chapter 7). However, all troops involved in the exercise "were thoroughly briefed in advance on the nature and characteristics of the weapon and were told that they were safe." Hence, as a secret 1952 Army study reported, "the results were highly indeterminate and unconvincing.... No well-controlled studies could be undertaken which could presume even superficial validity.... To attempt to probe into men's private fears and anxieties when all danger of death or injury ... has been excluded by the exercise seemed superfluous." Despite the enormous expense and risk of these exercises, the Army still did not know "how--or if--troops would endure close-in nuclear fires, unannounced detonations or multiple bursts." Weapons Research and Testing 1946-95. The military's involvement in post-World War II nuclear testing began with Operation Crossroads at Bikini Atoll in July 1946. Organized under the command of Admiral George Blandy, the stated objective was to determine the effects of atmospheric, shallow underwater, and deep underwater nuclear blasts on naval vessels. The entire exercise was also strongly motivated by a desire to counter claims by the U.S. Army Air Forces that atomic bombs had rendered naval vessels obsolete. Two atomic implosion-type bombs, similar to the bomb dropped on Nagasaki, were detonated. The first bomb, code-named "Able," was dropped by a modified B-29 and missed its target by several hundred yards. The second shot, "Baker," was moored 90 feet (27 meters) beneath a landing craft in Bikini's lagoon. The Baker explosion produced a massive surge of radioactive water, spray, and other debris, which severely contaminated many of the ships in the lagoon to an extent not foreseen by the military. In part because of this contamination, the third deep underwater blast, code-named "Charlie," was canceled.     Operation Crossroads presaged the enormity of the DOD's eventual involvement in nuclear testing. To prepare the atoll for the tests, military authorities persuaded the native population to relocate from their homes to other, less habitable islands in the Pacific, foreshadowing the forced removal of the people of Enewetak Atoll in the 1950s. The operation was carried out by 42,000 military and civilian personnel. Ninety-five ships, including personnel landing ships, submarines, destroyers, aircraft carriers, and battleships, were moored in Bikini's lagoon. The total estimated cost of Operation Crossroads was an astonishing $1.3 billion, far more than any of the subsequent thermonuclear test series conducted during the 1950s, most of which tested many more thermonuclear devices than the two atomic bombs detonated at Crossroads.     During the 1950s, the DOD sponsored numerous additional weapons effects tests both in the Pacific and in Nevada. Because testing in the Pacific involved formidable logistical considerations, many of these tests were conducted in conjunction with AEC-sponsored weapons development tests. The AEC and the military maintained two test sites in the Pacific (Enewetak and Bikini) for similar reasons. The DOD also conducted weapons effects tests in Nevada; between 1975 and 1992 twenty-five such tests were conducted at a cost of nearly $1.3 billion. Some of these were used to determine whether proposed civil defense measures (such as hardened structures) and military equipment could withstand the impact of atomic explosions. All of them were carefully documented by extensive still and motion-picture photography. A full cost accounting of this documentary effort is beyond the scope of this study (if complete records even exist), but an AEC report on Operation Castle (a four-test series in 1954 to verify the successful design of the hydrogen bomb) indicates that costs for "scientific photography" were more than $7.5 million. Data from official reports prepared by the 1352d motion picture squadron (known as Lookout Mountain Laboratory) for the air force historical office indicate that for fiscal 1958 and 1959 photographic costs for the testing program in Nevada and Enewetak were $10.3 million. During this period, 108 tests were conducted--70 in Nevada, 32 at Enewetak or Bikini, 3 in the South Atlantic, 2 over Johnston Island, and one 85 miles (137 kilometers) northwest of Enewetak. In 1997 the DOE began to declassify and release many of these films.     According to the best available evidence, the Defense Department spent at least $4.7 billion on nuclear weapons research and testing through the end of fiscal 1961 (table 1-4). This figure comes from a tabulation of known expenditures for test series conducted during the period together with partial expenditures from other test series (four major test series involving forty-eight detonations remain unaccounted for). Therefore, this estimate may undercount total costs by as much as several billion-dollars.     From 1962 to 1995 the DOD spent in the neighborhood of $32.8 billion to support the development of nuclear weapons and to explore the effects of atomic and thermonuclear explosions. (appendix table A-3) This estimate includes research and testing programs ($27.5 billion), nuclear weapons development programs ($1.6 billion), and overhead costs ($3.7 billion). Conclusion Since 1940 the United States has spent at least $409 billion to produce fissile materials and to amass a nuclear weapons stockpile of more than 70,000 bombs and warheads (although these weapons were not all deployed at the same time). As large as this total figure seems, it is only 7 percent of total estimated U.S. spending on nuclear weapons. Producing a small, yet potent nuclear arsenal is relatively inexpensive; one recent estimate placed the costs at $130 million to $326 million. Building delivery systems, creating the military infrastructure to deploy the arsenal, maintaining the weapons, and developing effective command and control systems entail far greater costs, as is discussed in the following chapters.     Despite our best efforts to compile a complete record of historical spending, poor record-keeping and continued classification of key documents ensure that there will always be more details to uncover. The Department of Defense has released substantial and unprecedented amounts of formerly classified budgetary data to the authors of this book, some of which is incorporated in this chapter. However, no complete records of spending in the 1950s have been kept by either the AEC or the DOD. According to one DOE official, any such records were stored in "shoebox-like" fashion. To what extent they can be retrieved or have even survived to the present day remains unknown. We encourage future researchers to build upon what we have presented here despite the enormity of the challenge.     Designing and building nuclear weapons was only the first step in creating a nuclear arsenal. For these weapons to be useful, they had to be delivered to their targets. Chapter 2 examines the development of these delivery systems, including strategic bombers, ballistic missiles, and submarines. Box 1-1 Growth and Evolution of the U.S. Nuclear Stockpile     Sixty-five warhead types have been produced and deployed, configured for approximately 116 weapons systems. The air force has used 42 types of nuclear weapons, the navy and Marine Corps 34 types, and the army 21 types. Another 25 warhead types were canceled before production, because another warhead type was chosen, or in some cases, because the delivery system itself was canceled. Some warhead types have had wide applicability, used in one configuration as a bomb, and in another as a warhead for one or perhaps several kinds of missiles. The last completely new warhead was a W88 assembled at the Pantex Plant on July 31, 1990, for the Trident II missile. Although production has not resumed, Pantex recently converted approximately fifty existing B61-7 bombs into B61-11 bombs, to allow them to penetrate 10 to 20 feet (3 to 6 meters) into the earth's surface and destroy hardened underground targets. The principal modification is the emplacement of the existing "physics package" into a new needle-nosed, hardened, depleted uranium casing. This has increased the bomb's weight by 449 pounds (204 kilograms) and its length by just over 3 inches (8 centimeters). In addition, the drogue parachute has been removed, to permit the bomb to fall freely upon release and thus achieve maximum velocity before impact. While this program does not appear to contravene U.S. government pledges to foreswear development and deployment of entirely new nuclear weapons, it has raised serious questions about when a modified bomb becomes a new weapon. The B61-11 was first deployed with the B-2A bomber at Whiteman AFB in Missouri in April 1997. Scientists at Sandia National Laboratories are already designing a new nuclear glide bomb (based on the B61-11 but utilizing a new guidance system) to be dropped from the B-2A bomber, even though the air force has no requirement for such a weapon. Other warhead modifications are under way: the W76 warhead for the Trident I missile is being revalidated (to ensure its continued conformity with military requirements in the absence of nuclear testing) and the pit for the W88 is being rebuilt (to address safety concerns pertaining to accidental, detonation and plutonium scattering accidents first raised in 1990). In addition, the navy and the weapons laboratories are engaged in a joint SLBM warhead protection program, the goal of which is to design and fabricate (but not actually produce) a new warhead for either the current or the next generation of SLBMs by 2004. The historic high for the stockpile was reached in 1966 when about 32,200 nuclear warheads were simultaneously active. As can be seen in figure 1-4, the U.S. buildup started in the late 1950s and consisted almost entirely of tactical, nuclear weapons, made possible by the rapid AEC production expansions in the early 1950s and the Eisenhower administration's "New Look" military program, which emphasized nuclear weapons over conventional forces. Annual production of nuclear weapons exceeded 7,000 a year in 1959-60, and more than 5,000 in 1961: this amounted to almost 19,500 new warheads in three years, or a rate of about 25 per workday. In 1954 hydrogen bombs--hundreds of times more powerful than their fission predecessors--began to enter the stockpile in great numbers, and the megatonnage increased sixtyfold in five years. It peaked in 1960 when it equaled almost 20.5 billion tons of TNT--equivalent to nearly 1.4 million Hiroshima-sized bombs--largely because the Strategic Air Command dominated the nuclear force of the day with a fleet of some 1,600 bombers, armed with thousands of high-yield bombs (the explosive power of the arsenal today equals some 120,000-130,000 Hiroshima-sized bombs). With the sudden retirement of about 940 warheads in 1961, the megatonnage was cut almost in half. The reason was that the retired bomb, the B36, had a yield of 10 megatons. Until recently, the largest warhead in the arsenal was the 9-megaton B53 bomb, though only about fifty remained and were replaced following the introduction of the B61-11 into the active stockpile in April 1997. As ballistic missiles were introduced and accuracy improved, high-yield weapons were further reduced. The rule of thumb is that making a weapon twice as accurate allows an eightfold reduction in yield to achieve the same level of destruction. Lower yields also permitted the use of substantially less plutonium and highly enriched uranium in warheads, lowering the cost of many weapons and contributing to the eventual surplus of fissile materials. --Robert S. Norris and Stephen I. Schwartz Box 1-2 Selecting a Continental Test Site     By 1948 the Joint Chiefs of Staff had initiated plans for a continental test site. But they had to contend with public concerns about the radiological hazards of such a move. To encourage the public to take a more "realistic" attitude and to reduce public relations and political difficulties, an aide to the army chief of staff thought it prudent to "reeducate" the public, in order to allay its "unhealthy, dangerous and unjustified fear of atomic detonations" and to lay to rest "the ghost of an all pervading lethal radioactive cloud which can only be evaded by people on ships, airplanes and sandpits in the Marshall Islands." Officials in charge of selecting a continental site understood that nearly the only parts of the country that would escape fallout were on the East Coast. As a U.S. Air Force meteorologist reported in 1948, that region was "predominantly under the influence of westerly winds" and might provide a suitable site. By 1950 the government had ruled out anything north of Cape Hatteras (because ocean currents could bring fallout on shore and because fallout could also affect important fishing grounds) and concentrated on five locations, three of which were owned by the Department of Defense, in order to avoid the costly and time-consuming process of acquiring the land (a site south of Cape Hatteras, which the aforementioned meteorologist indicated would be ideal for avoiding "all fallout," was rejected for this reason): Alamogordo-White Sands Guided Missile Range, New Mexico (where the Trinity test took place); Dugway Proving Ground-Wendover Bombing Range, Utah; Las Vegas-Tonopah Bombing and Gunnery Range, Nevada; an area in Nevada, about 50 miles wide, extending from Fallon to Eureka; and the Pamlico Sound-Camp Lejeune area of North Carolina. The last two sites were not under government control, and, faced with possible delays in their acquisition, they were removed from further consideration. In the end, convenience carried the day, despite the warnings of Colonel Stafford Warren (the chief of radiological safety at the Trinity test who later headed the AEC's Division of Biology and Medicine). After monitoring the fallout from Trinity, he urged that tests not be conducted within a 150-mile radius of human habitation, Gordon Dean, then chairman of the AEC, stated in 1950: "A geographical location as close as possible to the Los Alamos Laboratory, to enable accelerating the pace of the weapons development program, is obviously a characteristic of such desirability that it could outweigh partial deficiencies in other respects." On January 8, 1951, three days before the selection of the test site was officially announced, Dean was informed by Dr. Charles Dunham, chief of the medical branch of the AEC's Division of Biology and Medicine, that while no evaluation had actually been done, there were no serious problems with potential fallout in Nevada and that monitoring would be done mainly for "record purposes." Dunham said: "The only ones we are concerned with are the people who live in the first valley--a town called Alamo. It is not the first valley from the test site; it is about fifty miles as the crow flies." Dean continued to ask what would happen to the people in the surrounding areas. "If there was a pretty good sized burst," replied Dunham, "there might be trouble in the valley, if it poured right after, that is a remote possibility." But the people would not be affected "if they get out and ... they would have several hours to evacuate--it would not have to be done in a matter of a few minutes," and even if rain fell within two hours of a test, the worst thing that could possibly happen would be "minor skin burns." The sheep, on the other hand, "cannot be controlled." In reality, large amounts of fallout routinely drifted far from the test site, depositing highly radioactive debris across the United States, but especially in eastern Nevada, southern Utah, Idaho, and Montana (see figure 7-3). Radiation levels were monitored, but even when they became quite high no evacuations were ordered. Instead, reassuring press releases were issued and people were told to stay indoors and, if caught outdoors, to simply remove and wash their clothes and themselves to eliminate any marginal hazards (see chapters 7 and 9). Even before the very first test, the AEC dearly intended to announce that testing posed no risk whatsoever. --Stephen I. Schwartz Copyright (c) 1998 THE BROOKINGS INSTITUTION. All rights reserved.