Silencing the Bomb Page 14

  In one of Perle’s statements, he commented that he had additional classified information that led him to different conclusions about verifying Soviet nuclear testing. His responses to similar previous statements had led people with high-level clearances for all of the relevant documents to reexamine them. Perle was found wanting.

  In a later frightening interview on the Public Broadcasting Service’s program Frontline, Perle made the case for using a war with Iraq to remake the Middle East. He stressed the significance of the World Trade Center disaster of September 11, 2001, in shaping the Bush administration’s thinking about the links between terrorism and weapons of mass destruction.

  SENATE AND HOUSE ACTIONS ON NUCLEAR TESTING

  The U.S. House of Representatives and the Senate, which were controlled by Democrats, held hearings on Soviet compliance with the Threshold Treaty and verification of a full test ban in 1985 and 1986. I gave oral and written testimony on both topics, twice to House subcommittees and once to the Senate Foreign Relations Committee.

  Alewine sent a letter on December 16, 1985, to Representative Beverly Byron of the House Committee on Armed Services criticizing several statements in my testimony before her subcommittee on November 20, 1985. Alewine said, “Dr. Sykes represents one extreme in the assumptions he advocates (leading to lower yield estimates than most). There are others, just as responsible and knowledgeable, who advocate extreme assumptions on the opposite end (leading to higher yield estimates). The reviews conducted by the DoD have tended to balance these extremes.” He made no mention of the several scientists and officials at Livermore and Britain who disagreed with his views. Alewine also stated, “The flippant remarks of Dr. Sykes concerning the possibilities for cavity decoupling evasion gloss over a very difficult and threatening problem that must be addressed.”

  In the 1980s I debated Alewine several times about Soviet yields and the verifiability of a full test ban treaty. I had no doubt that Alewine, Bache, Perle, and others consistently played “hard ball.” Someone said that if you want to argue with those people, you have to be willing to jump into the pigpen. The debate went on for a very long time, and I am glad I kept at it. My testimony in 1985 and 1986 likely contributed to Congress’s taking an active role in the verification of a full test ban and ascertaining if the Soviet Union had cheated on the Threshold Treaty.

  INDEPENDENT REVIEW BY THE OFFICE OF TECHNOLOGY ASSESSMENT

  The Senate Select Committee on Intelligence, the House Committee on Foreign Affairs, and the House Permanent Select Committee on Intelligence requested that Congress’s Office of Technology Assessment (OTA) undertake an assessment of these test ban issues in 1986. OTA had previously performed studies for Congress on arms control and other scientific and technical issues.

  OTA conducted the first major independent review on seismic verification of nuclear testing from 1986 to 1988. For the first time, the departments of Defense and Energy were not the sole sources of information in the government.

  OTA formed an advisory panel of nineteen individuals from various government agencies, universities, the weapons labs, and consulting firms. Paul Richards of Lamont, who had been involved in test ban issues for some time, and I were members. Gregory Van der Vink of OTA was the project director. He received a PhD in geophysics from Princeton and became interested in science policy, especially arms control, through Frank von Hippel and his group at Princeton. Frank had long worked on national security issues and energy. Greg persuaded OTA to conduct the test ban study.

  The study itself and its 1988 publication Seismic Verification of Nuclear Testing Treaties were approved by the Technology Assessment Board of Congress, which consisted of twelve members chosen equally from Republicans and Democrats and from the House and Senate.

  The 1988 report stated, “Seismic monitoring is central to considerations of verification, test ban treaties, and national security.” It addressed two key questions:

  1. Down to what size explosion could underground testing be monitored seismically with high confidence?

  2. How accurately could the yields of underground explosions be measured seismically?

  The answers to these questions would provide the technical information that lay at the heart of the political debate over:

  1. how low a threshold test ban treaty with the Soviet Union we could verify;

  2. whether the 1976 Threshold Test Ban Treaty was verifiable; and

  3. whether the Soviet Union had complied with present testing restrictions.

  OTA held workshops of two days each on (1) Seismic Network Capabilities, (2) Identification, (3) Evasion, and (4) Yield Determination. About a dozen individuals participated on each panel. Richards and I were on the panels on Identification and Yield Determination; Richards was also on Network Capabilities, and I was also on Evasion. While some of the briefings were at the secret level, the 1988 report was cleared by relevant government agencies prior to its open publication. The report is a good primer on seismology as it relates to nuclear monitoring and on the role of verification in the context of national security.

  In its executive summary, the report concluded that unless differences in the transmission of seismic P waves beneath Eastern Kazakhstan and Nevada were taken into account, the sizes of Soviet explosions were greatly overestimated. Once these differences were appropriately considered, the report stated, “All of the estimates of Soviet and U.S. tests [since the TTBT became effective in 1976] are within the 90 percent confidence level that one would expect if the yields were 150 kt or less. Extensive statistical studies have examined the distribution of estimated yields of explosions at Soviet test sites. These studies have concluded that the Soviets are observing a yield limit consistent with compliance with the 150 kt limit of the Threshold Test Ban Treaty [original boldface].” I consider this an important victory for good science, sound science policy, and arms control in general.

  The 1988 report has a table on page 124 showing my calculations of the yields of the six largest explosions at Eastern Kazakhstan since 1976. All of those yields were close to the 150-kiloton limit and well within the uncertainty expected for its observance.

  ON-SITE MEASUREMENTS OF YIELD

  Two nonseismic methods are available for determining the yield of explosions near the 150-kiloton limit of the Threshold Treaty. One involves drilling back into the explosion point and obtaining samples of various radioactive materials produced by the explosion. This method, often called radiochemical, or rad-chem, has an uncertainty of about 10 percent. Although the United States often used rad-chem for its tests, applying them to explosions on-site could reveal information about the characteristics of weapons that the Soviets and Americans likely wanted to keep secret.

  Another method for yield determination, abbreviated CORRTEX, involves drilling a second hole very close to and equal in depth to that used for a large underground explosion. When the nuclear explosion occurs, a cable in the second hole is crushed. This method measures the speed of the shock wave, which travels faster than the speed of sound, close to a nuclear explosion as it crushes the cable. The satellite hole must be within about 33 feet (10 meters) of the hole containing an explosion with a yield of about 150 kilotons.

  These measurements permit yield to be determined with an uncertainty of about 30 percent. This method, of course, is very intrusive because it is so close to the explosion. Its advantage is that it does not reveal the radiochemical contents of the materials produced by the explosion. It is not applicable, however, to small nuclear explosions. The 1988 report contains an extensive appendix on this method. An idea was for the United States to use CORRTEX to monitor a Soviet test and for them to use similar equipment for a U.S. test in Nevada.

  By 1986 General Secretary Gorbachev of the Soviet Union had put President Reagan under pressure to push for a full test ban. Reagan proposed that if Gorbachev would agree to work on the verification of the TTBT and PNET first, the United States would then negotiate ways to implement a step-by-step program
limiting and ultimately ending nuclear testing. Reagan said, however, that he would not ratify the TTBT even if the Senate gave its advice and consent.

  During expert talks in July 1987, the Soviets proposed calibration of test sites to reduce uncertainty in yield estimation. They invited the United States to measure yields at one of their main sites using CORRTEX and regional seismic observations. The two countries then signed a bilateral agreement to conduct Joint Verification Experiments (JVEs) in 1988 at Eastern Kazakhstan and Nevada. Each country was to detonate one explosion with a yield between 115 and 150 kilotons.

  The United States did not release the yield of the Soviet JVE test of September 14, 1988, which it monitored using CORRTEX. Each country released to the other the yields of the JVE and five previous tests at each site. The 1988 agreement stated that yields were not to be released to additional people or countries without the consent of the other. As far as I know, these yields were kept secret, and still are. Nevertheless, a detailed report on the two Joint Verification Experiments in the New York Times by Michael Gordon states that American and Soviet on-site measurements were said to give yields of 115 and 122 kilotons, respectively, for the Soviet JVE, an average of 118.5 kt.

  In 1989 Göran Ekström, a geophysicist then at Harvard, and I published an average magnitude, mb, of 6.115 with a very small uncertainty of +/- 0.018 for the Soviet JVE using reports of P waves from sixty-eight seismic stations. We then used three mb-yield relationships derived from various test sites where yields were available. Each was corrected for mb bias with respect to Eastern Kazakhstan. We chose a bias of 0.35 for explosions in hard rock and those below the water table in Nevada. The three calibrations gave yields very similar to those reported in the New York Times for the Soviet JVE. We extrapolated the yield of 118.5 kilotons measured on-site for the Soviet JVE to 150 kilotons. It gave a magnitude mb of 6.20, very similar to the magnitudes of the six largest explosions at that test site published in the 1988 OTA report.

  YIELDS OF SOVIET NUCLEAR TESTS MEASURED USING LG SEISMIC WAVES

  A third method of determining yields from seismic waves came of age in 1992. In 1973 Otto Nuttli of St. Louis University started to develop a magnitude scale called mbLg that uses short-period seismic waves called Lg with periods near one second (one cycle per second frequency). In 1952 Press and Ewing of Lamont had described and named Lg and another slow surface wave called Rg, which propagate in continental areas. Lg is often the largest wave on a short-period seismic record from earthquakes to stations at regional distances within continents. P waves from explosions, however, are relatively large compared to Lg, making Lg a good method for identifying underground explosions and potentially for determining their yield.

  Nuttli developed mbLg magnitudes suitable for all parts of the United States. In regions of older crust, Lg is observed at distances of more than 3000 miles (5000 km). Examples of this are Lg recordings at Palisades, New York, of waves that cross the ancient rocks of Canada from earthquakes in the Yukon and northern Alaska.

  Lg radiates symmetrically from explosions and earthquakes, and it does not have the major difficulty of P waves, which need a correction for differences in wave propagation through the uppermost mantle of the Earth. Hence, it became very useful for yield determinations.

  In 1992 Frode Ringdal of Norway, Marshall of Britain, and Alewine of DARPA obtained more precise estimates of mbLg than Nuttli had for explosions in Eastern Kazakhstan using data from the NORSAR (Norway) and Grafenberg (Germany) seismic arrays. Instead of measuring the amplitude of a single wiggle of Lg, they obtained an average over several minutes. Lg, as shown in figure 10.3, is not a single pulse but a train of seismic waves that builds up and then decays slowly. Its records are more like those of waves that travel great distances in the oceans, called the T phase, and those made on the moon.

  FIGURE 10.3

  Upper seismogram is from an event at a regional seismic station that propagated along a continental path. Lg is its largest signal. Lower seismogram is from an event to a station at a large distance (a teleseismic signal). The P wave is its largest signal, and Lg is very small.

  Source: Office of Technology Assessment, 1988.

  Ringdal and colleagues analyzed seismic data for 101 nuclear explosions at the Shagan River portion of the Eastern Kazakhstan test site, which was the site of most large underground tests. Their measurements of mbLg and mb determined from P waves differed systematically by as much as 0.15 magnitude units between two subareas of the Shagan River site. They discovered that mb values for four explosions with yields published in 1989 by V. S. Bocharov and his Russian colleagues also varied in the same way with respect to values of mbLg. Those subareas are separated by a series of major faults oriented northwesterly. Hence, Ringdal and his colleagues concluded that mbLg is a more reliable measure of yield than either mb or surface waves.

  They found a magnitude mb bias with respect to Nevada of 0.45 for the southwestern part of that area and 0.30 for the northeastern part. These are close to the biases of 0.3 to 0.4 that several of us determined earlier, and clearly not zero. By subdividing that test site into two parts, Ringdal and colleagues provided better determinations of yields than any of us had obtained previously. They obtained an average yield for the Soviet JVE explosion of 1988 in the southwest part of the test site of 108 kilotons, which compares very well with the 118.5 kilotons reported in the New York Times for the two close-in CORRTEX measurements.

  A lingering question is whether the Soviet Union has cheated on the Threshold Treaty by conducting any tests above its 150-kiloton limit since its start date in 1976. It is important to understand that all yield determinations have uncertainties associated with them. For explosions of 150 kilotons, some estimates will be somewhat larger, others somewhat smaller, and some right on, as shown for the three following explosions with yields near 150 kilotons.

  Only one of the explosions studied by Ekström and Richards, that on April 3, 1987, for which they calculated a yield of 176 kilotons, exceeded the 150-kiloton threshold. Ringdal and colleagues, however, obtained 140 kilotons for that explosion using mb and mbLg. For a second explosion, on August 4, 1979, the two sets of authors obtained 153 and 132 kilotons. For a third event, on October 27, 1984, they calculated 165, 104, and 140 kilotons. Those measurements taken together indicate that the three largest explosions at Eastern Kazakhstan testing area since 1976 were at or close to the 150-kiloton limit within the uncertainties of the measurements.

  When asked by a member of a subcommittee of the U.S. House of Representatives in 1985, I stated that the yields of Soviet tests in Eastern Kazakhstan near the threshold of the treaty could be determined [then] with an uncertainty of about 30 percent. Donald Kerr, the director of Los Alamos, said my estimate of the uncertainty in yield of 30 percent was small compared to extrapolations that could be made up to a factor of about two times. With the introduction of Lg measurements, the uncertainty in yield estimation for Eastern Kazakhstan was reduced further to about 25 percent.

  To appreciate what these numbers mean in terms of accuracy, a person driving 25 percent faster than a 55 mile per hour speed limit would be traveling at 69 miles per hour and might or might not receive a speeding ticket. If the person were traveling 25 percent slower, the speed would be 44 miles per hour, and he or she certainly would not get a ticket. If Russia had decided to test weapons at yields 25 percent higher than the 150-kiloton limit, or 187 kilotons, the yield determined by the United States likely would be between 150 and 234 kilotons but more likely close to 187 kilotons. The analogy is that they probably would be “given a speeding ticket”—that is, accused of cheating on the TTBT. More than one test at 187 kilotons would have substantially increased the chances of the United States’ determining that the Soviet Union was testing above the threshold of the TTBT.

  Alewine joined Ringdal and Marshall as third author of their 1992 paper. This probably was an unstated acknowledgment by Alewine, as well as by the U.S. Defense Department, that the det
ermination of yields at that test site had been resolved. It was clear that earlier U.S. charges of Soviet cheating, spearheaded by Alewine, Bache, Perle, and Romney, were false. The Soviets, in fact, had been in compliance with the 150-kiloton limit of the Threshold Test Ban Treaty. With that, the “yield wars” were finally over. Alewine told me in 2009, at a meeting in Vienna on nuclear test testing, “I guess you got about 50 percent of things right and we [DARPA] 50 percent.” I would give them no more than a 5 percent.

  In 1989 Bocharov, Zelentsov, and Mikhailov of the Soviet Union published an official list of ninety-six underground nuclear explosions at the Eastern Kazakhstan test site through 1972, before the Threshold Treaty became effective. Mikhailov became minister of atomic energy of the Russian Republic in 1992. While most yields are listed within a broad range, several are given exactly. They list two larger than 125 kilotons: 165 kilotons on November 2, 1972, and 140 kilotons on December 10, 1972. My determinations of those yields were 154 and 138 kilotons; Ringdal and colleagues calculated 169 and 158 kilotons. I estimated the yield of the largest underground explosion at that test site, on July 23, 1973, as 193 kilotons. All three of those explosions, of course, occurred before the Threshold Treaty became effective in 1976.

  From 1990 to 1992, physicist David Hafemeister worked for the Senate Foreign Relations Committee to examine arms control treaties at the end of the Cold War. In an article published in 2005, he stated, “This charge [of a probable violation of the Threshold Test Ban Treaty] was removed in 1990 after the 1988 CORRTEX measurements at Semipalitinsk [Eastern Kazakhstan] Test Site and after properly taking into account the geological differences between test sites…. The U.S. record on TTBT noncompliance charges was not entirely honorable.”

  By 1990 the United States and Russia completed a considerably revised protocol for the Threshold Test Ban Treaty. I doubt if more than a few people ever read the new protocol, which is exceedingly long. It called for CORRTEX-type measurements for U.S. and Russian weapons tests larger than 50 kilotons. Because the Russians stopped testing before the treaty entered into force in late 1990, the United States has not been able to make additional CORRTEX measurements of Russian nuclear explosions. Under the ratified TTBT, the Russians, however, were able to monitor two U.S. explosions in Nevada before President Clinton halted U.S. testing in 1992. The TTBT and its companion Peaceful Nuclear Explosions Treaty (PNET) are still in force.