Testimony by Project Director Joseph Cirincione for the House Committee on Government Reform Subcommittee on National Security, Veterans Affairs, and International Security, September 8, 2000

Thank you, Chairman Shays and Members of the Subcommittee, for the opportunity to submit written testimony on the critical issue of the technical feasibility of national missile defense. As a former professional staff member of the Subcommittee, I have great respect for the important role Subcommittee hearings play in informing the deliberations of the House of Representatives and the understanding of the public on national security issues.

From the beginning of 1985 to the end of 1993, I was privileged to serve as a member of the professional staff of the House Armed Services Committee and as a professional staff member of the Government Operations Subcommittee on Legislation and National Security. One of my chief responsibilities was the review and analysis of the technical performance of the Strategic Defense Initiative (SDI) programs. I have closely followed missile defense programs since then, first, as a Senior Associate at the Henry L. Stimson Center, and, since 1998, from my current position as Director of the Non-Proliferation Project at the Carnegie Endowment for International Peace. Much of the history I cite in my testimony can be found on our Project's web site at: www.ceip.org/npp. I have also appended a recent article for Inside Missile Defense, "Lost at Sea" on the problems of sea-based missile defense systems.

As I received the request for testimony only this week, my testimony will be brief. That is just as well, as the issues of technical feasibility are actually very simple: None of the dozens of national missile defense systems proposed over the past 20 years has ever proven to be technically feasible. This includes the wide-range of systems researched and developed under the SDI program and the current candidates proposed by the Ballistic Missile Defense Organization and missile defense advocates.

It is highly unlikely that any candidate system can be shown to be militarily effective during the next eight years.
That is, during the next two presidential terms neither the technology nor our testing methods will provide an assured capability to defeat long-range ballistic missiles. It is possible that the next president may decide to proceed with deployment of a national missile defense system during that time, but that decision will be based on political considerations or the perception that the threat justifies early deployment, not on demonstrated ability to defeat the likely threats.

Given the overwhelming advantage enjoyed by offensive nuclear forces, and the enormous technical difficulties inherent in any missile defense, this should not be surprising. It may be possible to someday construct a system that could provide at least some defense against intercontinental ballistic missiles. However, we are years away from conducting the kinds of realistic tests that could provide our military and political leaders with the minimum confidence they must have before risking the lives of millions of citizens.

The investigations into the performance, cost and schedule of missile defense program conducted by the Government Operations Committee during the 1980s and early 1990s offer a rich trove of material critical to understanding the national missile defense issues before the Congress today. I have mined some of that material for my testimony.

False Claims

The past two decades of efforts to invent a viable national missile defense have been characterized by exaggerated claims of success and promises of performance that later proved false.
It is difficult to recall a missile defense proponent who understated the actual performance of a system. The problems began with the false claims of proponents of the X-ray laser that helped launch the SDI program1 and continue through claims today that Aegis destroyers and cruisers can quickly and inexpensively provide a highly-effective defense against both intermediate- and intercontinental-range ballistic missiles.

In 1992, the Chairman of the Government Operations Committee requested that the General Accounting Office review the accuracy of the Strategic Defense Initiative Organization's claims about the results of all flight tests of kinetic kill interceptors. This remains the only independent review of missile defense test claims. The GAO found that SDIO officials claimed five of the seven flight tests were successes and the other two were failures (one interceptor blew up on launch and one "Brilliant Pebbles" space-based interceptor shut down seconds after launch). However, GAO found a pattern of misleading claims, forcing them to conclude that SDIO officials had inaccurately portrayed four of the five tests as successes, when they were not.

Most importantly for the Subcommittee's purposes, these inaccurate reports included claims that a "Brilliant Pebbles" test was "a "90-percent success," and was ready to proceed to more advanced testing. GAO found that the "90-percent success" claim was based on a substantially downward revision of the original goals for the test to correspond with what the interceptor was able to achieve, not what was originally planned. Of the original four goals, none was fully met, including its complete inability to detect, acquire and track a target. Since the program's "accomplishments were significantly less than planned," GAO concluded, the first phase of the program's testing "was completed only in the sense that SDIO had decided to proceed into Phase II."2

Equally relevant, is GAO refutation of the claim by SDIO officials that the ERIS ground-based interceptor had discriminated between the target and two decoys on its only successful intercept. SDIO officials claimed in a May 16, 1991 Subcommittee hearing that the ERIS "did its own thing in…determining which of the targets to go after, whether the decoy or the target vehicle…The principal algorithms we have to prove can work in doing the discrimination task I think were effectively proven as part of that test."3 Investigators uncovered a different story. The ERIS sensors actually detect three objects in the test: the reentry vehicle target and two balloon decoys tethered to the either side of the target. The ERIS interceptor was pre-programmed to intercept the target in the middle. If the middle object had been a balloon decoy, the ERIS would have hit the balloon; it could not tell the difference. The SDIO director, in a letter to the Chairman, then clarified that the ERIS role in target selection "did not constitute discrimination."4 However, the original test claims are still cited "proving" the discrimination problem has been solved.5

This history is part of the reason why it is common to hear advocates of missile defense claim that their proposed system is ready to go, inexpensive to build and highly effective. But as former BMDO Director Air Force General Lester Lyles said, in directly rebutting before the Senate last year the claims of the Heritage Foundation that effective sea-based missile defenses could be rapidly deployed, "When it comes to missile defense, there is nothing quick, cheap and easy."

President Clinton rediscovered this truth when, despite large funding increases by both Congress and the Administration over the past four years, the technical problems with the proposed National Missile Defense System proved overwhelming. He said on September 1, "I simply cannot conclude with the information I have today that we have enough confidence in the technology, and the operational effectiveness of the entire NMD system, to move forward to deployment."

Misplaced Faith

Some argue that when President Clinton assumed office in 1993, he sabotaged plans that, if allowed to continue, would have by now produced a working and affordable missile defense. The Global Protection Against Limited Strikes (GPALS) plan introduced by President George Bush and Defense Secretary Richard Cheney in January 1991 scaled-back the original SDI program. It proposed instead a space-, sea- and land-based system to destroy from 10 to 200 warheads "delivered by ballistic missiles launched from anywhere in the world to attack areas anywhere else in the world." The system relied heavily on the so-called "Brilliant Pebbles" satellite weapons.

Some advocate a return to such a system today. But their confidence in the concept is based on faith, not fact.

The Congressional Budge Office at the time estimated that the 12-year cost of the GPALS plan would be at least $85 billion (in 1992 dollars).6 Annual expenditures would have averaged $8 billion. Since the system consisted largely of view graphs and concept studies, CBO pointed out that "the complexity of the Grand Forks and GPALS defenses suggests that total costs could exceed planned levels." 7

The General Accounting Office, in a report to the Chairman of the Legislation and National Security Subcommittee in 1992, warned that the plan would have to "overcome tremendous technical challenges."8 This report was the last independent evaluation conducted of the GPALS program. It was not optimistic about the technical feasibility of the weapons proposed. "Such a system will push the cutting edge of technology," GAO warned, "SDIO must rely on some technologies that are as yet unproven and learn how to integrate them into a reliable system" For the system to work, the GAO advised, "significant advances must be made over the next several years in critical areas . . . if these advances are not achieved, schedule delays, escalating costs, and performance problems could occur."

Even if the technologies became available, the analysts said, there was still "the enormous challenge of integrating them into a cohesive system."

In short, space- and sea-based systems proposed in the waning days of the previous administration were hardly the ready-to-go weapons that advocates now fondly remember. Outside of those with a direct financial or career interest in the programs, few experts or military officers thought any of these programs could deliver real, near-term military benefit.

The Historic Record

Based on current schedules and all available evidence it is reasonable to assume that if proposed high-altitude, ballistic missile defense systems are used in combat they will fall far short of predicted effectiveness.
It is unlikely that the systems will completely fail, but the evidence indicates that they will perform significantly below either tested or predicted kill rates. Military commanders, therefore, would be wise not to base troop deployments or enemy engagement strategies on unrealistic expectations of the protection these defenses will offer. Officials should consider reallocating the excessive funds devoted to overlapping and duplicative new systems and planning more realistic development schedules for missile defense efforts.

The evidence available includes:

· the performance of the Patriot missile system in the Gulf War
· the performance of high-altitude missile defense systems in tests to-date
· current test plans for proposed systems prior to production and deployment

The Patriot Experience

In the United States, confusion over the Patriot's performance in the Gulf War still fuels overly optimistic estimates of the effectiveness of new, proposed defensive systems. During the war, many believed that the Patriot had achieved a near-perfect intercept rate, as was reported initially from the battlefield and Washington. Claims were revised downwards from 96 percent in testimony to Congress after the war, to 80 percent, 70 percent, and-after a investigation by the Government Operations Subcommittee on Legislation and National Security in 1992-to 52 percent, though the Army report notes that destruction of only 25 percent of the Scud warheads is supported by evidence with high confidence levels.9

Independent evaluations are more pessimistic, concluding that the Patriot hit few if any Scuds during the war. These include assessments conducted by the Israeli Defense Force, the Congressional Research Service, the General Accounting Office, and the Massachusetts Institute of Technology and staff of the Government Operations Committee.

The General Accounting Office review of the evidence in support of the Army claims revealed that, using the Army's own methodology and evidence, a strong case can be made that Patriots hit only 9 percent of the Scud warheads engaged, and there are serious questions about these few hits. The speed of the Scuds, the limitations of the Patriot missile system, and the confusion and targeting difficulties caused by the break-up of the Scud missile as it re-entered the atmosphere seem to have contributed to the high failure rate.

The Patriot missile, equipped with a new multi-mode seeker, failed in two out of three intercept tests conducted after the war. The Army declared it "operationally unacceptable." The new replacement interceptor missile for the PAC-3 configuration, the ERINT, will not initially deployed until 2001. Until then, US forces cannot reliably intercept even the short-range Scuds encountered in the Gulf War.

Whatever the kill ratio attributed to Patriot, the few unclassified hard figures released by the Army should serve as a sobering reminder of how combat conditions can wreck havoc even on systems that perform well on the test ranges, as the Patriot did.

A total of 158 Patriot missiles were fired at fewer than 47 Scuds during the war:
· 86 Patriots were launched at real Scud targets, but
· 30 per cent of the Patriots were launched as Scud debris mistaken for targets
· 15 per cent of the Patriots were launched against false targets caused by radar backlobe and sidelobe interference (including one
launched by accident in Turkey.)

The fragmentation and EMC problems were known at the time (the Scud fragmentation had been observed during the Iran-Iraq war) but were not included in deployment and operational planning for the Patriot nor were they included in any tests of the system.

It is my personal evaluation, after extensive investigation for the Subcommittee and review of all available evidence, that the Patriot hit few, if any, Scud warheads. The Patriot was simple overmatched. It was never designed to hit a target as complex as that presented by the Scud. As Raytheon executive Robert Stein explained after the war:

"Upon reentry, the resulting forces caused the missile to break apart into several pieces. These extra pieces looked to the Patriot software like targets that were diving at high speed and were going to impact in the areas that the defense design was laid out to defend. In effect, they became 'decoys' that were indistinguishable from TBMs to the Patriot radar, since no discrimination features had been implemented in anticipation of these types of targets.

"The anomalous behavior that the operators were seeing was created by the aerodynamic instability of the warhead section after the missile started to break up. It was spiraling, rather than travelling on an expected ballistic trajectory, because of changes in its center of gravity and center of aerodynamic pressure after breakup. In addition, its radar reflectivity had dropped significantly because of its smaller size. In effect, what Iraqi engineers had created, purely unintentionally and by poor workmanship and design, was a high-speed, low radar-cross-section, maneuvering reentry vehicle (RV0 accompanied by decoys…"10

This is the type of target that TMD and NMD systems should expect in combat and should be used extensively in all test programs now.

The Historic Test Record and Current Plans

All the proposed new missile defense systems except for the Navy Area-Wide system, employ hit-to-kill interceptors. That is, unlike the Patriot interceptors, which used a proximity fuse and an explosive warhead to scatter pellet-size fragment in the path of the intended target, the new interceptors will attempt to hit the target head-on using the kinetic energy of the encounter to destroy the target.

The track record for test of exo-atmospheric hit-to-kill interceptors should indicate caution in projections of future capabilities. There have only been 15 intercept attempts outside the atmosphere conducted by the Department of Defense since 1982. Of these, only 4, or 26 per cent, actually hit their targets.11 The low number of past tests and the weak success rate warrant deep skepticism for much success in the near future with the proposed systems.

Lower-Tier Systems

The most promising new system, the improved Patriot system, or PAC-3, is designed to intercept Scud-type missiles of the type now deployed by potential Third World adversaries. These 300- to 1000-kilometer-range missiles will represent a challenge, but one which the PAC-3 should be capable of intercepting. The new ERINT missile for the system successfully intercepted two targets (although at relatively short ranges) in a shoot-off with the Patriot multi-mode missile in 1993, but its has since undergone some design changes. It has enjoyed five successful intercepts over the past two years; three against short-range Hera targets and two against cruise missile targets The Navy Area-Wide (Lower Tier) system (an upgrade to the AEGIS radar system and Standard missile) and the multi-national MEADS program are also aimed at these lower-range threats, but have yet to have any intercept tests.

Some experts still voice concern, however. David Eshel, a retired career officer in the Israeli Defense Force, writes in the September Janes' Intelligence Review, "Although this system [the PAC-3] has an increased range and an onboard terminal radar guidance system it is doubtful that this could overcome the unique corkscrewing effect of the Iraqi Al-Hussayin Scud missile." Without realistic tests it is impossible to predict performance, but these lower-tier systems appear to hold out the best possibility of successfully intercepting the existing Third World missile threats armed with single warheads. (Missiles armed with submunitions released after the boost phase would defeat any known kinetic energy missile defense system.) They rely on previously developed radar and hardware systems and, because they intercept their targets within the atmosphere after any decoys deployed would have been stripped away, they do not encounter the difficult discrimination problems facing higher, outside the atmosphere interceptors. Countermeasures remain one of the major unsolved technical barriers to effective missile defense despite decades of effort.

Higher-Tier Systems

Potentially more threatening than Scuds are medium-range missiles that travel from 1000 to 3,500 kilometers. No nation hostile to the United States currently fields such missiles, except for several Nodong missiles deployed by North Korea with a range of 1000 km. But this is the threat represented by systems reportedly under development in North Korea and Iran. Both the Administration and Congress favor developing systems to intercept these missiles, with Congress urging a faster development and deployment schedule. To-date, tests of the most promising candidates, the Army's Theater High-Altitude Area Defense system (THAAD) and the Navy Theater-Wide (Upper Tier) system, have been disappointing. While both systems may be technically feasible, THAAD has failed in six of its eight test intercept attempts, and the Navy has gone zero for four in tests of the LEAP kill vehicle (Lightweight Exo-Atmospheric Projectile).

These were tests against specially designed targets, with known trajectories and characteristics, well within the expected performance range of the systems. The THAAD tests were against Storm and Hera targets, which have a maximum range of about 750 and 1,100 kilometers, respectively. A suitable long-range target of 2,000 kilometers or more, does not yet exist. The Navy plans to use surplus Terrier missiles as targets for the Theater-Wide tests.

National Missile Defense System

Noting that the NMD schedule is shorter than most other major system acquisition programs, the General Accounting Office warned in 1997 of the high risks inherent in the program:

"Because of the compressed development schedule, only a limited amount of flight test data will be available for the system deployment decision in fiscal year 2000. By that time, BMDO will have conducted only one system-level flight test, and that test may not include all system elements or involve stressing conditions such as targets that employ sophisticated countermeasure or multiple warheads. As a result, not all technical issues, such as discrimination, will be resolved by the time of the deployment review. Also the current schedule will permit only a single test of the integrated ground-based interceptor before production of the interceptor's booster element must begin. If subsequent tests reveal problems, costly redesign or modification of already produced hardware may be required."12

By comparison, the only other U.S.-based ballistic missile defense system, the Safeguard, had an acquisition schedule twice as long as planned for the NMD program. Safeguard also had 111 flight tests, compared to only three intercept tests and one system-level flight test before a fiscal year 2000 deployment decision. The GAO noted that even this system-level test will not be comprehensive because it will not include all system elements, and:

"…the single integrated system test…will not assess the NMD system's capabilities against stressing threats such as those that use sophisticated countermeasures or multiple warheads. The test is to be conducted against a single target with only simple countermeasures such as decoys. No test against multiple warheads is planned."

In June 1998, the GAO reaffirmed its findings, concluding that even with increased funding technical and schedule risks are high.

The Bottom Line

There are no current plans to test the THAAD, the Navy Theater-Wide or the NMD system against realistic threats such as multiple warhead missiles that deploy warheads with realistic decoys or jammers.
Department of Defense Director of Operation Test and Evaluation Phil Coyle concluded in a August 11 memorandum (reported by Bloomberg News on August 23) that "test results so far do not support a recommendation at this time to deploy in 20005." President Clinton apparently agreed.

Director Coyle also warned:

"Deployment means the fielding of an operational system with some military utility which is effective under realist combat conditions, against realist threats and countermeasures when operated by military personnel at all times of day or night and in all weather. Such a capability is yet to be shown to be practicable for NMD." (emphasis added)

The same, of course, is true of the higher altitude TMD systems. This should give military commanders and policy-makers low confidence in the ability of these systems, if deployed, to provide their troops, the nation or US allies any appreciable degree of protection against longer-range ballistic missile threats. Defense planner should consider whether more realistic schedules and elimination of duplicative programs could reduce the approximately $20 billion planned for missile defense efforts over the next five years and the savings allocated to more pressing defense needs.

Countering and Negating Missile Defenses

The 1999 National Intelligence Estimate provides the most elaborate unclassified intelligence description to-date on the steps nations are likely to take in response to deployment of U.S. theater and national missile defenses.

First, it notes:

"We assess that countries developing ballistic missiles would also develop various responses to US theater and national defenses. Russia and China each have developed numerous countermeasures and probably are willing to sell the requisite technologies."13

This possibility should not be lightly dismissed. Over the decades the United States, Russia, the United Kingdom, France and China have all developed and deployed sophisticated countermeasures to overcome the defensive systems erected by their adversaries.

The inability to discriminate among decoys and overcome other likely counter-measures remains the Achilles' heel of all currently envisioned ballistic missile defense systems. This is not a hypothetical contest. This is the experience of the existing nuclear arsenals when confronted by defensive systems.

For example, in March 1987 Lawrence Woodruff, then deputy undersecretary of defense for strategic and theater nuclear forces, described the contest between the offense and the defense to the House Armed Services Committee this way:

"The Soviets have been developing their Moscow [ABM] defenses for over ten years at a cost of billions of dollars. For much less expense we believe we can still penetrate these defenses with a small number of Minuteman missiles equipped with highly effective chaff and decoys. And if the Soviet should deploy more advanced or proliferated defenses, we have new penetration aids as counters under development…We are developing a new maneuvering re-entry vehicle that could evade interceptor missiles." 14

For these reasons, the Joints Chiefs of Staff were always supremely confident of our ability to overwhelm and penetrate the Moscow anti-ballistic missile systems. (For similar reasons, the Joint Chiefs have never been enthusiastic about U.S. national missile defense plans. In recent requests for a reported $30 billion increase in the FY 2001 budget, the Chiefs did not include any national missile defense programs. They cited other, more pressing needs.)

Countries attempting to develop medium-or long-range missiles would not, however, have to rely on the purchase or transfer of counter-measure technology. The NIE lists eight distinct currently available technologies that such countries could employ:

"Many countries, such as North Korea, Iran and Iraq probably would rely initially on readily available technology-including separating RVs, spin-stabilized RVs, RV reorientation, radar absorbing material, booster fragmentation, low-power jammers, chaff, and simple (balloon) decoys-to develop penetration aids and countermeasures."15

The NIE further concludes that these countries could develop these countermeasures "by the time they flight test their missiles." Moreover, foreign espionage and other collection efforts are likely to increase, says the NIE, increasing the likelihood that adversary nations could use critical information about U.S. defenses to improve their ability to overcome such defenses.

These "readily available technologies" could present severe problems for any missile interceptor. Again, these are not new technologies. An analysis prepared by the Office of Technology Assessment in 1988 confirmed that:

· "There are plausible decoy designs that would be very difficult to counter merely with passive infrared sensors in conjunction with radar."
· "It appears possible that chaff, if properly deployed with decoys, could be used to deny RV [re-entry vehicle] detection and more easily, deny RF [radio frequency] discrimination to the radar elements of a defense."
· "Whereas chaff would deny information to radar, aerosols would mask RVs and decoys from infrared sensors."16

In a review of sensor systems under consideration in 1987, including the ground-launched Probe system and the satellite-based Space Surveillance and Tracking System (SSTS), (the predecessor of the Space-Based Infrared System now planned), the Defense Science Board also noted:

"Serious questions remain unanswered about the ability of the passive IR [infrared] sensors on Probe and SSTS to carry out discrimination against anything but the most primitive decoys and debris. In addition, the presence of cooled RVs would greatly reduce the range of proposed sensors."17

These serious questions remain today. Some may believe that the United States has recently solved the discrimination problem. The first intercept test of a proposed national missile defense interceptor on October 2, 1999 contained a test element where the interceptor was to distinguish between the target and a decoy object. The interceptor vehicle, using "hit to kill" technology successfully collided with and destroyed the target. In briefings before the test, however, Ballistic Missile Defense officials provided important qualifying details of the test. In particular, there were four critical test enhancements that made the test conditions not entirely realistic:

1.) The target followed a pre-programmed flight path to a designated position.
2.) The interceptor missile also flew to a pre-programmed position.
3.) A Global Positioning Satellite (GPS) receiver was placed on the target to send its position to ground control, and the necessary target location information was uploaded to a computer in the kill vehicle.
4.) The decoy released had a significantly different thermal signature than the target, making it easier for the sensors on the kill vehicle to distinguish between the objects.

Subsequent reports have made clear other problematic aspects of the test:

5.) Incorrect star maps loaded into the kill-vehicle's computer prevented the vehicle from ascertaining its position once it had separated from the booster.
6.) Back-up inertial guidance systems led to inaccuracies in pointing the sensors used to locate the target.
7.) The sensors finally saw the large, bright balloon decoy, re-oriented, continued searching, and only by virtue of the proximity of the decoy to the target did they locate the cooler warhead that the kill vehicle had been programmed to recognize as the correct target.

The interceptor failed to hit its target in the second intercept test, on January 18, 2000. Initial reports blamed the failure on faulty sensors. The test again had to rely on the GPS transponder for tracking information. Officials believe that a leak in the gas lines used to cool the sensors may have caused the failure. This raises the obvious question: If a hand-built, meticulously prepared interceptor fails from leaky tubing, how well are assembly-line production models likely to perform after sitting for years in the frozen Alaskan tundra or aboard ships at sea?

The third test failure on July 7, 2000 confirmed that missile defense programs are still struggling with basic engineering problems and are years away from tests that would rigorously stress defense capabilities under real combat conditions.

For test purposes, there is nothing wrong with minimizing the number of variables in order to test key elements of the weapon system. It is vital, however, that test officials provide full disclosure of test limitations to policymakers at every stage of the process, lest test results be interpreted to have greater significance than, in fact, they do. The October 1999 test was much more a demonstration of two missiles intercepting each other than it was a test of intercepting an enemy missile under combat conditions. Proceeding with the program based on that test, unvalidated computer simulations and program manager guarantees would have been folly.

In sum, until interceptor tests are conducted under combat-like conditions in the presence of realistic decoys and countermeasures and independently assessed by objective evaluators, it will be impossible to ascertain the effectiveness of any proposed ballistic missile defense systems.

1 In February 1981, Aviation Week and Space Technology reported, based on briefings by Lawrence Livermore National Laboratory scientists Lowell Wood and Edward Teller, "X-ray lasers based on the successful Dauphin test…are so small that a single payload bay on the space shuttle could carry to orbit a number sufficient to stop a Soviet nuclear weapons attack." (cited by William Broad in Teller's War, p. 92, (New York: Simon & Schuster, 1992).

2 United States General Accounting Office, "Strategic Defense Initiative: Some Claims Overstated for Early Flight Tests of Interceptors," September 1992, GAO/NSIAD-92-282.

3 Ibid, p. 23.

4 See Hearings before the Legislation and National Security Subcommittee of the Committee on Government Operations, House of Representatives, 102nd Congress, First Session, "Strategic Defense Initiative: What Are the Costs, What are the Threats," May 16 and October 1, 1991, p. 368.

5 See, for example, K. Scott McMahon, Pursuit of the Shield: The US Quest for Limited Ballistic Missile Defense, (University Press of America, 1997), p. 94. ""…the SDIO and the Army Strategic Defense Command announced that an experimental ground-based interceptor built by the Lockheed Company had successfully intercepted a mock ICBM RV after discriminating it from balloon decoys." (emphasis in original).

6 Congressional Budget Office, "Costs of Alternative Approaches to SDI," May 1992, p. 20.

7 Ibid.

8 United States General Accounting Office, Report to the Chairman, "Strategic Defense Initiative: Changing Design and Technological Uncertainties Create Significant Risk," February 1992 (GAO/IMTEC-92-18).

9 See Hearings before the Legislation and National Security Subcommittee of the Committee on Government Operations, House of Representatives, 102nd Congress, Second Session, "Performance of the Patriot Missile in the Gulf War." April 7, 1992, summarized in Activities of the House Committee on Government Operations, 102nd Congress, First and Second Sessions, 1991-1992, December 31, 1992, pp. 179-185.

10 Robert M. Stein, "Correspondence: Patriot Experience in the Gulf War," International Security, Summer 1992, at p. 199.

11 The four hits were by the Homing Overlay Experiment (HOE) in 1984, the Exoatmospheric Reentry-Vehicle Interceptor Subsystem (ERIS) in 1991, the National Missile Defense system interceptor in 1999 and one of the two THAAD intercepts which can be considered outside the atmosphere in June 1999 (THAAD hit two out of eight targets flying at various altitudes).

12 United States General Accounting Office, "National Missile Defense: Schedule and Technical Risks Represent Significant Development Challenges," December 12, 1997 , GAO/NSIAD-98-28.

13 National Intelligence Council, "Foreign Missile Developments and the Ballistic Missile Threat to the United States Through 2015."

14 See, Staff Report on the Strategic Defense Initiative, Democratic Caucus of the U.S. House of Representatives, "Strategic Defense, Strategic Choices," May 1988, available at www.ceip.org/npp.

15 Ibid.

16 Ibid.

17 Ibid.