304. Memorandum From Lay to the NSC 1

[Facsimile Page 1]


  • Technical Feasibility of Cessation of Nuclear Testing


  • NSC Action No. 1840–c

The Report on the subject,2 called for by NSC Action No. 1840-c-(1), has been prepared by the NSC Ad Hoc Panel established by the reference NSC Action (consisting of representatives of the President’s Science Advisory Committee, the Department of Defense, the Atomic Energy Commission, and the Central Intelligence Agency), and will be presented orally at the National Security Council meeting on Thursday, April 3, 1958.

[Typeset Page 1257]

Because of the sensitivity of this Report, copies have been circulated only to those agencies represented on the NSC Ad Hoc Panel and to the Department of State. A copy of the Report is available, in the office of the Executive Secretary, NSC, for reference by other regular participant members of the Council.

James S. Lay, Jr.
Executive Secretary

cc: The Chairman, Atomic Energy Commission

The Chairman, Joint Chiefs of Staff

The Director of Central Intelligence

The Special Assistant to the President for Science and Technology


Memorandum From Killian to Gray

[Facsimile Page 2]


  • Transmittal of Report

In accordance with Action No. 1840 of the National Security Council, as approved by the President on January 9, 1958, I submit herewith a report of the Ad Hoc Working Group devoted to the following three studies in the area of nuclear testing:

“(a) A study of the losses to the United States consequent on a total suspension of nuclear tests at specific future dates.

“(b) A symmetrical study of the losses to the USSR that would accrue from cessation of nuclear testing, using the same hypothetical dates.

“(c) A study of the technical feasibility of monitoring a test suspension, including the outlines of a surveillance and inspection system.”

The Ad Hoc Working Group submitting this report is made up of representatives nominated by the President’s Science Advisory Committee, the Department of Defense, the Atomic Energy Commission and the Central Intelligence Agency.

The Ad Hoc Working Group, in preparing this report, limited itself to the technical feasibility of monitoring nuclear tests and to the technical losses that would result to the U.S. and the U. S. S. R. from a cessation of tests. Although the Group considered some of the military implications of these technical losses to the U.S. and the U.S.S.R., a complete evaluation of these military implications would have required extensive studies by the Department of Defense and these are not yet [Typeset Page 1258] available. It excluded from its consideration any question of policy with respect to whether there should be a suspension of nuclear tests.

J.R. Killian, Jr.


Report of the NSC Ad Hoc Working Group

[Facsimile Page 3]


[Facsimile Page 4]


  • Letter of Transmittal
  • Summary
  • A. Capabilities of the Present U.S. Long Range Detection System
  • B. The Technical Feasibility of Monitoring A Test Suspension
    • 1. At the Earth’s Surface and at Low Altitudes
    • 2. At Very High Altitudes Over the USSR and China
    • 3. Below the Earth’s Surface
    • 4. Tests Conducted Outside the USSR
    • 5. Detection Net
    • 6. Risk of Detection
    • 7. Weapons Development Implications
  • C. The Losses to the U.S. and to the USSR from Cessation of Nuclear Testing
    • 1. U.S. and USSR Nuclear Warhead Capabilities
    • 2. Asymmetries in U.S. and USSR Nuclear Warhead Needs
    • 3. AICBM Warheads
    • 4. Very Small Warheads
    • 5. Boosted Warheads
    • 6. Summary of Relative Position
    • 7. Clean Weapons
    • 8. Military Effects of A Test Cessation
    • 9. Effects on Weapons Laboratories
    • 10. Soviet Gains Through Espionage
[Typeset Page 1259]

Table 1. U.S. and USSR Technical Capabilities in the Field of Nuclear Weapons

Appendix A. Report on the Detection of Nuclear Tests—Present and Potential Capabilities and Limitations for the Detection of Nuclear Tests, report by AFOAT–1, U.S. Air Force

Appendix B. Detection of High Altitude Nuclear Tests, report by Hans A. Bethe

[Facsimile Page 5]

Appendix C. Concealment and Detection of Nuclear Tests Underground, report by Harold Brown and Hans A. Bethe

Appendix D. Chart of Present and Future U.S. Nuclear Warhead Developments, report by the U.S. Atomic Energy Commission

Appendix E. Impact of a September 1958 Nuclear Test Moratorium on Soviet Nuclear Weapons Capabilities, report by the Central Intelligence Agency


[Facsimile Page 6]

Dear Dr. Killian:

We submit herewith for transmittal to the National Security Council the report of the Ad Hoc Working Group on the Technical Feasibility of a Cessation of Nuclear Testing established in accordance with NSC Action 1840 c. The report is concurred in by all members of the Working Group which included representation from the President’s Science Advisory Committee, Department of Defense, Atomic Energy Commission, and Central Intelligence Agency.

Hans Bethe, Cornell University


Harold Brown, University of California Radiation Laboratory

Maj. Gen. Richard Coiner, USAF

Herbert Loper, Office of the Secretary of Defense

Carson Mark, Los Alamos Scientific Laboratory

Doyle Northrup, AFOAT–1, USAF

Herbert Scoville, Jr., Central Intelligence Agency

Roderick Spence, Los Alamos Scientific Laboratory

Brig. Gen. Alfred Starbird, Atomic Energy Commission

Col. Lester Woodward, USAF

Herbert York, University of California Radiation Laboratory and Advanced Research Projects Agency, Department of Defense

[Typeset Page 1260]

Appendix A

[Facsimile Page 7]


Prepared for Inclusion in the Report of the


for the



[Facsimile Page 8]
    • Acoustic Net
    • Seismic Net
    • Electromagnetic Net
    • Nuclear Sampling
    • Surface to 50,000 feet
    • Subsurface
    • High Altitude
    • Tests in Remote Geographical Locations
    • Summary
    • Seismic Net
    • Acoustic Net
    • Electromagnetic Net
    • Nuclear Sampling
    • Inspection Teams (CIA)
    • Other Intelligence Sources (CIA)
    • Security
    • Cost
    • Acoustic Net
    • Seismic Net
    • Electromagnetic Net
    • Air Sampling
    • Capability
[Typeset Page 1261] [Facsimile Page 9]



A long range detection system consisting of seismic, acoustic, electromagnetic and air sampling components is presently deployed around the USSR. This system can detect and identify nuclear tests of 10 KT or larger conducted within the USSR and China as shallow sub-surface, surface or air bursts up to 50,000 ft with an estimated reliability of 90–100 per cent. Nuclear tests as small as 3 KT in the same environments can be detected and identified with a reliability of 30 per cent. Underground explosions of 10 KT or larger can be detected with a certainty of 90–100 per cent but cannot be identified as nuclear explosions. Underwater explosions of 20 KT or larger conducted in deep ocean areas of the Northern Hemisphere and some parts of the Southern Hemisphere can be detected with 90–100 per cent certainty and probably identified as an explosion rather than an earthquake. Since the present system was designed to detect tests conducted in the USSR, its capabilities for tests outside the USSR are limited. Nuclear tests as large as a few hundred kilotons and possibly even one megaton might be missed if conducted in areas remote from the present detection network.

A system of improved capability for the detection of nuclear tests possibly conducted on a clandestine basis in the USSR or China is described. This system would consist of about 70 geophysical stations in the USSR and China plus an aerial sampling network involving overflight of critical areas as required to intercept radioactive clouds for the purpose of proving the nuclear nature of the explosion. It is estimated that this system could detect and identify with 90–100 per cent certainty nuclear tests of 1 KT or larger conducted as shallow subsurface, surface or air bursts up to 50,000 ft within the USSR and China. Underground explosions of 1 KT or larger in the USSR and China could be detected with a certainty of 90–100 per cent but not identified as nuclear in nature. Identification of such underground disturbances may be possible through the use of on-the-spot inspection teams investigating about 300 unidentified sub-surface disturbances of 1 KT or larger per year in the USSR and China. If inspection only of those disturbances of 5 KT or larger is contemplated, only 35 events per year would require investigation.

[Facsimile Page 10]

A system for the detection and identification of nuclear tests in areas of the world remote from the present detection system is described. This system would consist of about 30 geophysical stations located principally in the Southern Hemisphere plus an aerial sampling network necessary to intercept radioactive clouds to prove the nuclear [Typeset Page 1262] nature of the explosion. It is estimated that this system could detect and identify with 90–100 per cent certainty nuclear tests of 20 KT or larger conducted as shallow sub-surface, surface or air bursts up to 50,000 ft in remote areas of the world. Underground explosions of 20 KT or larger in remote parts of the world could be detected with a certainty of 90–100 per cent but not identified as nuclear in nature. Identification of such underground disturbances may be possible through the use of inspection teams investigating about 100 suspected test areas per year in remote regions of the world. Underwater explosions of 20 KT or larger anywhere in the world could be reliably detected and probably identified as explosions rather than earthquakes.

[Facsimile Page 11]


INTRODUCTION: This report describes the present system for long range detection and identification of nuclear tests in the USSR including an evaluation of its capabilities and limitations. The results of a study of the technical feasibility of monitoring a test suspension is presented, including the outline of a surveillance and inspection system for detecting and identifying, if possible, nuclear tests conducted in the USSR, China and in remote areas of the world.


a. General. The present Long Range Detection System is deployed around the USSR and consists of four major components, i.e., acoustic, seismic, electromagnetic and nuclear. The purpose of this system is to determine the fact of an explosion, time, location, height of burst and yield, and finally from the analysis of the nuclear debris to reconstruct, insofar as possible, the detailed characteristics of the nuclear device tested.

Scientific observation posts for each of the four techniques are located as close as possible to the USSR and are operated by units of the Army, Navy and Air Force as appropriate. Preliminary analysis of the radioactive debris is accomplished in field radiochemical laboratories and more comprehensive and detailed analysis of the samples is carried out in a central laboratory in California, all operated by AFOAT–1. In addition, highly specialized measurements are made by laboratories of the Atomic Energy Commission, civilian contractors and university laboratories in the U.S. Collection and preliminary evaluation of data from all these laboratories are effected by AFOAT–1 analysts assisted by special studies at RAND Corporation. Finally, all data and evaluations are reviewed by the Foreign Weapons Evaluation Committee, Professor Hans A. Bethe, Chairman, a committee which is responsible jointly to the Division of Military Application, AEC, and to AFOAT–1. [Typeset Page 1263] The final conclusions are reported as intelligence to the Joint Atomic Energy Intelligence Committee, where the information is collated with intelligence from all other sources, and National Intelligence Estimates of Soviet nuclear capability are prepared.

[Facsimile Page 12]

A limited cooperation between AFOAT–1 and the United Kingdom has been in progress for a number of years. Geophysical stations are operated by the British Ministry of Defense. Measurements from these stations are reported by cable directly to headquarters, AFOAT–1. Radiochemical data obtained on samples of Soviet weapon debris are exchanged with the Atomic Energy Research Establishment, Harwell under Sir John Cockroft and the Atomic Weapons Research Establishment, Aldermaston under Sir William Penney.

b. The Acoustic Component. The acoustic element of the Long Range Detection System consists of 10 acoustic stations surrounding the USSR as shown in Figure 1. One of these stations is omitted from the figure for security reasons. Eight of these acoustic stations are operated by the U.S. Army Signal Corps from a headquarters at Ft. Monmouth, New Jersey. These stations are located in Japan (2), Philippine Islands, Turkey, Eritrea, Germany, Greenland and Alaska. Two acoustic stations are operated by the United Kingdom. In addition to the permanent net stations, data are frequently obtained from experimental stations in San Diego, Washington, D. C. and Ft. Monmouth, N. J.

Each acoustic station consists of four condenser microphones, one located at each corner of a 6- to 10-mile square. Although only three microphones are necessary to establish the azimuth and apparent velocity of the incoming acoustic wave, four are used to provide a factor of safety for instrument malfunction and to improve the precision of measurement. Time sequence of arrival of the acoustic wave at individual microphones provides a measure of the azimuth and apparent velocity. The microphones are coupled to the atmosphere through a pipe array 1000 ft in length with openings every five feet throughout the length. This coupling device greatly reduces the background noise from wind turbulence (about a factor of 10) and permits identification of acoustic waves having pressure amplitudes as low as 0.1 dyne/cm2 under favorable conditions. Each microphone is connected by wire lines to a central recording station where variations in atmospheric pressure in the subsonic frequency range from approximately 1.0 to 0.01 cps are amplified and recorded on Esterline-Angus pen recorders. Service personnel, trained especially for the purpose, continuously scan the records on a 24-hour basis. When signal characteristics meet the criteria determined by AFOAT–1 to be indicative of a large explosion, the station personnel itemize the principal signal characteristics in a coded message to the [Facsimile Page 13] net headquarters at Ft. Monmouth, N. J. There the data are checked, correlated with other acoustic data, and transmitted to [Typeset Page 1264] AFOAT–1 for further evaluation and correlation with data from other elements of the Atomic Energy Detection System.

Equipment of the same type has been supplied to the British for their operations. In a similar way they identify significant signals and send the signal characteristics by message through headquarters, AFOAT–1, to the net headquarters at Ft. Monmouth for correlation with U.S. data.

The acoustic data are used to determine the time of burst to about ± 5 to 10 minutes, the location of burst within a radius of about 100 miles, and the yield to about ± a factor of 2 for small shots and to about ± 15 to 20 per cent on large yield tests.

c. The Seismic Component. The seismic component is made up of eight surveillance stations under the operational control of AFOAT–1. These stations are located in Spain, Turkey, Korea, Alaska, Australia and three in the U.S. (See Figure 2.) In addition, one seismic station with similar equipment is operated by the U.K.

Each seismic station is composed of an array of four concrete piers poured on solid rock and deployed when possible at equal spacing along a two-mile line. Each pier accommodates a vertical sesimometer and one of the piers has, in addition, two horizontal instruments oriented at 90° with each other to permit recording of the three components of earth’s motion on that pier. In those cases where a linear array is possible, the axis of the array is oriented to favor reception of a signal from the USSR and to discriminate against known sources of microseisms. The individual seismometers are connected through wire lines several miles in length to a central recording station. Seismic waves from a distant source are refracted upward from the mantle to the seismometers at an angle which is nearly vertical. In-phase signals are thus produced on all four seismometers which result in a gain in amplitude proportional to the square root of the number of seismometers. Since microseismic disturbances, in general, travel horizontally from local sources to the station, the response of the seismometers to noise is uncorrelated. Earth motion of about 1–2 millimicrons can be detected under favorable background conditions.

[Facsimile Page 14]

At the central recording station the signals from individual seismometers are amplified and recorded on 35-mm film, together with standard time signals from WWV or WWVH as appropriate. The response characteristics of the seismographs favor the reception of seismic disturbances in the period range of 0.5 to 1.0 seconds where the maximum energy in the earth from atomic explosions is found. Background noise at other frequencies is excluded and as a result the signal-to-noise level at the station is increased.

The film at each of the seismic stations is developed three times during each 24-hour period and scanned by field personnel for evidence [Typeset Page 1265] of signal characteristics meeting certain criteria of significance. When, in the opinion of the team personnel, these criteria have been met, the principal characteristics of the signal of interest are transmitted by TWX to AFOAT–1 in Washington, D. C. A central analysis station at Laramie, Wyoming but programmed to be located at headquarters, AFOAT–1, presently receives all of the incoming messages and makes a careful study of the significance of these reports.

From the seismic data AFOAT–1 can ascertain the time of explosion to the nearest 1/2 second, and estimate to the nearest 1/10th second; can determine the location to the nearest 2–5 miles; and can obtain a rough estimate of yield (to an order of magnitude). Yield determinations by seismic methods are very uncertain since the coupling between the explosion and the earth is affected both by height of burst and the geological formations at ground zero. Neither of these factors is known in the case of an explosion in the USSR.

d. The Electromagnetic Component. The Long Range Detection System presently contains eight electromagnetic stations located in Minnesota, Washington State, Alaska (2), Japan, Pakistan, Turkey and Germany. Installation of two additional stations is in progress. These stations, located as shown in Figure 3, are under the operational control of AFOAT–1.

The equipment at each station is energized by two antennas: one a vertical whip and the other a pair of crossed loops. The signal from the crossed loops passes through amplifiers and records on a cathode ray oscilloscope. This record permits determination of the azimuth of the incoming signal. The energy from the vertical whip is utilized to remove the 180° ambiguity in determination of azimuth from the crossed loop circuits described above. The vertical whip also energizes the equipment for recording waveform and energy content of the pulse.

[Facsimile Page 15]

The direction finding circuits operate on a frequency of 10 kc with a 2 kc pass band. The equipment for recording waveform and energy content utilizes four channels. The waveform itself is recorded in a wide band from 3 kc to 300 kc and, in addition, three other narrow bands (2 kc) centered on 20 kc, 75 kc and 5 mc are recorded for sampling the spectrum of energy radiated by the explosion. The waveform is recorded on a 500 microsecond sweep which is triggered by the initial signal from the vertical whip and provided with a suitable delay circuit to permit recording of the entire waveform from start to finish.

The channels at 20 kc, 75 kc and 5 mc have been added in the hope that the distribution of energy from the bomb will be different from lightning flashes. The five channels described, together with a WWV channel for timing purposes, are all recorded on 35-mm film on a [Typeset Page 1266] 24-hour basis. Signal strengths as low as 30 millivolts/meter are detectable under favorable conditions.

The electromagnetic system suffers from a defect of recording millions of lightning flashes which look very similar to the pulse transmitted from a nuclear explosion. Therefore, the individual station cannot, on its own, determine whether the recordings have significance with respect to a suspected nuclear explosion. For that reason the film is transmitted to the analysis center at AFOAT–1 where correlation studies are made to determine the existence of multi-station time and azimuth coincidences. The usual procedure is to search that part of the electromagnetic film covering a time period considered to have been significant from reports of the other geophysical systems or from the date and time established by measurements on fresh debris from the nuclear system.

Signal significance is determined by three methods: coincidence in times of arrival, consistency of intersections of azimuth, and compatibility of recorded signal strength at widely separated stations, assuming a single source and applying known attenuation factors. These methods, however, are applicable only after some method of sorting has been accomplished. At present, this sorting depends upon establishing a time of the explosion by acoustic, seismic or nuclear means.

The electromagnetic system has produced time and azimuth, multi-station coincidences for all of the large Soviet tests as well as several smaller tests, and, of course, it has been checked for accuracy on U.S. tests. The time obtained by electromagnetic data can be determined to 50 milliseconds and estimated to 20 milliseconds. The azimuth can be determined to ± 3 degrees.

[Facsimile Page 16]

Recent experience, on one small Soviet test, indicates the possibility of detection of relatively small tests in the USSR which, if conducted on the surface, produce large electromagnetic signals. If this experience on one Soviet test can be duplicated on all small surface tests, an important detection capability for small tests may be realized. At present, however, the electromagnetic component of the Long Range Detection System does not report independently the detection of a nuclear test in the USSR. It is capable only of response to query.

An important contribution of the electromagnetic time, however, is that when combined with seismic time it is used to determine the height of burst of a nuclear test. The electromagnetic time, of course, corresponds to the exact detonation time of the device and the seismic time gives a reasonably accurate time at which the shock wave from the device strikes the earth. Since there is available a “time-distance curve” on the shock wave transmission through the air from the burst to the ground, it is possible to utilize this curve to determine the height of burst.

[Typeset Page 1267]

e. The Nuclear Component. A system for collecting radioactive debris from a test utilizes airborne filters for removing the particulate debris from the air along the flight tracks of the aircraft. Specially designed air filters have been mounted on WB/50 aircraft operated by the Air Weather Service out of bases in Japan, Alaska and Burtonwood, England. WB/50 aircraft fly tracks northeast and southwest out of Japan, north and southwest out of Alaska and to any predesignated point in Europe or the Middle East out of Burtonwood, England. (See Figure 4.) The flight tracks from Japanese and Alaskan bases are planned to cover on a once-each-24-hour basis all air mass trajectories coming out of the USSR between 30° N and 85° N. The aircraft usually fly at 10,000 feet outbound and about from 20,000 to 30,000 feet on the return track so that two altitudes are covered during each flight.

Instantaneous radiation detectors mounted behind the filter papers indicate to the pilot when the filter paper is collecting unusual amounts of radioactive debris. This information is radioed back to the base and used for vectoring special missions to intercept “hot” parts of the atomic cloud and is used as a guide to enable the aircraft to orbit within the cloud while sampling.

The airborne filters are analyzed in a field laboratory (Japan and Alaska) by physical and chemical methods to determine whether [Facsimile Page 17] or not fresh radioactive debris has been encountered on the flight. Although in most instances early warning by geophysical techniques alerts the air crews to the possibility of intercepting debris, in five cases the initial and only detection of Soviet tests has been by the radiochemical analysis of filter papers. The field laboratories at operational bases in Japan and Alaska also conduct comprehensive radio-chemical analyses of the filters, concentrating mostly on short-lived isotopes for dating and other purposes, and thus provide information which would not be obtainable by the time the filters had been sent back to the central laboratory in Sacramento, California.

Reports of unusual amounts of radioactive debris, or the detection of fresh debris together with the preliminary reports on chemical analysis of short-lived isotopes, are submitted by the field laboratories through priority dispatch to the headquarters, AFOAT–1, Washington, D.C. All significant filter papers are transmitted to the central laboratory at McClellan Air Force Base, where a decision is made with respect to the distribution of samples among the laboratories in the U.S. which contribute to various phases of the analysis program, as well as to the Atomic Weapons Research Establishment at Aldermaston, England. Laboratories participating are the McClellan Central Laboratory, Argonne National Laboratory, Knolls Atomic Power Laboratory, and Tracerlab, Inc. In certain cases, the Los Alamos Scientific [Typeset Page 1268] Laboratory, University of California Radiation Laboratory at Livermore and the Naval Radiological Defense Laboratory perform special analyses.

These laboratories make quantitative mass spectrometric studies, alpha pulse analyses, activation studies and chemical identifications of induced activities and isotopes of the transuranium elements and fission products over a period of one to three months after the collection of the debris. All of these data are forwarded by letter report to AFOAT–1 where the final evaluation is carried out. Data received from all of these laboratories are also transmitted to the Foreign Weapons Evaluation Committee and are reviewed at periodic meetings. AFOAT–1 and the Bethe Committee review and discuss conclusions concerning the type of nuclear reactions, materials and geometry utilized by the Soviets in each specific test.

[Facsimile Page 18]


The Long Range Detection System of AFOAT–1 was primarily designed to detect and identify surface bursts or air bursts and sub-surface bursts detonated within the boundaries of the USSR. In considering its capabilities and limitations as a system for monitoring an international agreement on nuclear test limitations, the following four environments for possible test detonations within the USSR have been considered: surface or air bursts below 50,000 feet; sub-surface tests; high altitude tests; and surface or air bursts in remote geographical locations.

a. Surface or Air Bursts below 50 Kilofeet within the USSR. To date 44 nuclear tests have been identified within the USSR which are believed to have been between the surface and 50 kft and two which may have been higher than 50 kft. The remaining two Soviet tests were known to be sub-surface, probably underwater. Table I shows the part played by each component of the Long Range Detection System in identifying the 48 Soviet tests. It is noteworthy that five of these tests were picked up by nuclear techniques only. No geophysical [Facsimile Page 19] response was obtained. Of interest to the study of detection capability, four of the tests did not produce nuclear debris which was detectable. All four of the components of the system seemed to be relatively successful in tests of 500 KT or higher. The absence of electromagnetic detection of some of the tests does not indicate necessarily a low capability but rather that it consisted of only one station overseas and three stations in the U.S. prior to 1957 and no stations prior to 1954. Finally, the table shows a variety of combinations of the techniques which produced successful detection on different Soviet tests, thus demonstrating the value of including all four components in the system.

[Typeset Page 1269]
Table I
AFOAT–1 Identification of USSR Tests
Component Yield
3.5–8 KT 10–45 KT 60–500 KT 750–4300 KT
N <20, <20, <5, 8, 8
A 30, 45
AS 20, 20
AN 3.5, 4, 4, 7, 8 15, <20, 20, 25, 25, 25, 25, 30, 30 70, 100
EN <5
AEN 7 15 70
ASN 25 90, 300, 500 ~1000
ASEN 30 60, 90, 100, 200 750, ~1000, 1300, 1700, 2200, 2700, ~3000, 3200, 4300

The electromagnetic technique does not contribute, at present, to detection without the assistance of one or more of the other techniques. It does show considerable promise for the detection of surface bursts of relatively low yield, but this promise requires further exploration before any definite statements can be made. Reference to Table I shows a detection capability for tests in the USSR as low as 5 KT. However, in general, it is believed that a lower limit of reliable detection for surface bursts is more like 25 KT at ranges of 4000–5000 kilometers. A reduction in detectability is observed at burst heights of 8000–10,000 feet. Pending completion of developments of machine sorting techniques and a discriminator between bomb pulses and lightning pulses, the electromagnetic system may not contribute greatly to detection of very low yield tests within the USSR.

Except for subsurface tests, the seismic technique is applicable principally to surface or low air bursts of 100–150 KT or larger. One surface test of 25 KT in the USSR was detected, but this is the exception and [Typeset Page 1270] not the rule. In general, we record about 300 seismic disturbances per year in the USSR above 100 KT, of which about 75% are identified as earthquakes, leaving about 75 disturbances which cannot be identified either as man-made or natural.

The problem of detecting and identifying clandestine tests in the environment (surface to 50 kilofeet in the USSR) under discussion will depend principally upon the success of the combined nuclear and acoustic components of the Long Range Detection System. For the purpose of studying the effect of these two techniques, AFOAT–1 detection data on 25 Soviet tests and 66 U.S. tests in the range of 3.5 to 50 KT were selected. U.S. data are fairly well spread out through all of the seasons. Soviet data does not include many tests in the summer and spring.

[Facsimile Page 20]

Acoustic detection at individual stations has a marked diurnal variation due to the fact that noise levels at night are normally lower by a factor of 2 or more than they are in the daytime. A marked seasonal variation is noted in the detection of acoustic waves, since stratosphere wind patterns in the winter time are most favorable for transmission toward the east and, conversely, stratosphere wind patterns in the summer favor transmission toward the west. Stations to the north or south in general show relatively smaller seasonal effects than do east-west stations. Normally, the stratosphere winds permit longer ranges of detection in summer and winter than in the spring and fall.

Further complicating the acoustic picture is the fact that the amplitude of pressure waves from an explosion varies greatly due to fluctuations in winds and temperatures encountered in the atmosphere at altitudes above the known meteorology. Pressure amplitudes are only very qualitatively related to yield.

In estimating the capabilities of the acoustic net, it has been necessary to rely mainly on actual results obtained from U.S. and Russian nuclear tests. These estimates cannot be precise but they are believed to be as realistic as it is possible to make them at present. Four ranges of detectability were established, as follows: excellent—90 to 100%; good—60 to 90%; fair—30 to 60%; poor—0 to 30%. The result of the study is shown in Table II.

[Typeset Page 1271]
Table II
Acoustic Detection* Capability for Low Yield Air or Tower Nuclear Tests
Detection Coefficient Based on USSR Test Results Based on U.S. Test Results Overall (KT)
Winter–Summer (KT) Spring–Fall (KT) Winter–Summer (KT) Spring–Fall (KT)
Excellent (90 – 100%) ≥15 ≥20 ≥10 ≥15 ≥15
Good (60 – 90%) 5–14 10–19 3–9 10–14 10–14
Fair (30 – 60%) 1–4 5–9 1–2 3–9 5–9
Poor (0 – 30%) <1 <5 <1 <3 <5
[Facsimile Page 21]

The last column of Table II shows a conservative overall acoustic detection capability obtained by averaging the winter/summer and spring/fall results obtained from USSR and U.S. tests.

Nuclear debris from relatively small yield low altitude shots within the USSR seems to be readily detectable. Shots of the order of 3–10 KT can be expected to produce substantial clouds of debris if detonated at or near the surface and this debris will not rise much above the 10,000 to 20,000 ft levels routinely patrolled by aircraft out of Japan and Alaska. However, when the yield increases to substantial fractions of a megaton or higher, so much of the debris is deposited in the stratosphere that collections on routine patrols at 10,000 to 20,000 feet are extremely unsatisfactory. From the standpoint of detection of the fact of a nuclear burst, however, it is believed that most very large tests will leave enough debris in the troposphere to permit the fact of fresh debris to be ascertained for the high yield tests.

Clandestine tests of low yield devices conducted at altitudes of 20,000 or 30,000 feet will probably produce debris which will pass over the top of surveillance flights at 20,000 feet. This actually occurred on the U.S. “HA” shot (3 KT) where debris was picked up over Washington at 40,000 feet while no debris was obtained from that same shot on flights between Bermuda and Washington at 10,000 and 20,000 feet. Since altitudes above 20,000 feet are not routinely patrolled but are searched by special high altitude aircraft only upon warning from the geophysical system, it is quite possible that low yield shots fired at [Typeset Page 1272] altitudes above 20,000 feet will not be detected by nuclear techniques. Figure 5 is a graphic portrayal of the situation that will occur in the case of low yield tests in the environment under discussion.

Debris collection seems to be at its greatest disadvantage in the summer time. From fall through winter and spring collection activities are at their best. It is, therefore, believed that the air sampling technique will substantially add to the overall system capability but it is difficult to make a quantitative assessment. The overall detection capability for the system under the environment being discussed must, therefore, rely to a certain extent on the judgment as to what size nuclear cloud could escape routine sampling flights from 30° N to 85° N during short periods when interruptions of flight schedules occur and to what extent clouds from small shots would overpass routine flight lines.

Taking everything into consideration, it is believed that for tests in the USSR between the surface and 50,000 feet there will be a substantial contribution to the acoustic component by the nuclear [Facsimile Page 22] detection component. A combined acoustic and nuclear detection capability, which is believed to be the determining factor in the low yield range, is shown in Table III.

Table III
Overall Detection and Identification Capability for Airbursts between the Surface and 50,000 feet
Overall Acoustic (KT) Acoustic plus Nuclear (KT)
Excellent (90–100%) ≥15 ≥10
Good (60–90%) 10–14 5–10
Fair (30–60%) 5–9 3–5
Poor (0–30%) <5 <3

The limited capability for seismic and electromagnetic techniques to contribute in the low yield air burst range does not permit any quantitative evaluation of the small contributions which they may make. Both techniques are limited, however, in any case to small tests conducted on or near the surface as indicated above and to some large yield tests at moderate heights of burst. There is the possibility mentioned in paragraph 1c of some assistance from the electromagnetic component on surface or low air bursts.

b. Subsurface Tests. It is assumed that a clandestine subsurface test by the USSR will be conducted in such a way that no radioactive debris will be cast into the atmosphere, that the shot will be sufficiently tamped to prevent any energy appearing in acoustic waves and, of course, that there will be no electromagnetic radiation. The seismic component of the Long Range Detection System will then be the sole component responsible for the detection of such a test.

[Typeset Page 1273] [Facsimile Page 23]

The seismic system is limited to the determination that a large subsurface disturbance has occurred. There is no unique way of identifying with certainty the signature of an underground nuclear explosion. At long range detection distances the individual characteristics of the source of disturbance are very difficult to detect, since the character of the medium through which the wave is propagated plays a predominant part in determining the character of the recorded signal. The problem is considerably simplified, however, by the existence of several useful, though not unique indicators of earthquakes or blasts.

For example, many large earthquakes persist for many minutes or hours whereas the record from a small underground blast has a maximum duration of a few seconds to a few minutes. Another useful indicator is the fact that all subsurface blasts produce a compressional first wave, while many earthquakes produce alternate compressional and dilational first waves as one proceeds around the source with a seismic detector. Thus, a dilational first wave is an almost certain indicator of an earthquake. The usefulness of this indicator, however, is limited to the large shots where the first wave can be clearly distinguished from the background fluctuations due to microseismic disturbance. In the yield range below 5 KT, it is doubtful if this criterion would be of much use at long ranges. A few earthquakes occur so far below the surface that the seismograms clearly indicate earthquake origin. However, a very large percentage of earthquakes occur at shallow depths (20 kilometers) and do not give any indication that would be useful in separating them from blasts. Most earthquakes produce large clearly identifiable shear waves, whereas most explosions do not produce shear waves. However, there are earthquakes on record where no detectable shear wave was recorded at long range detection distances and there are blasts on record, for example, Rainier, Wigwam and many of the small Nevada shots that have produced noticeable shear waves. Finally, repeated blasts at the same underground location will produce almost identical wave shapes at the same long range detection station. This might be useful in the event that repeated use were made of the same location for clandestine tests. It is concluded, therefore, that the seismic surveillance for underground tests might detect the presence of an underground explosion but would not uniquely identify the source as nuclear in origin.

The ability to detect underground explosions by seismic means has been carefully studied. The limitation in pushing to smaller and smaller yields appears to reside principally in noise generated by storms and man-made disturbances in the vicinity of the stations. A general deterioration of capability for the entire net in the wintertime [Facsimile Page 24] is observed because of the storm activity in the Northern Hemisphere. There are also short periods of relatively poor capability during summertime windstorms. Natural earthquakes and, occasionally, volcanic activity produce additional background noise which reduces capability for the detection of small yield [Typeset Page 1274] underground tests. Large earthquakes produce disturbances which persist for many hours and are often accompanied by aftershocks on several succeeding days which individually persist for many minutes. Natural seismicity of the earth varies considerably from one point to another within the USSR. (See Figure 9.) Natural seismicity is, in general, higher to the south near the Himalayas and along the eastern coast.

The transmission characteristics of the earth for seismic waves varies considerably with the distance between the source and the seismic observatory. (See Figure 6.) It will be noted that a distinct loss of sensitivity occurs at distances between 500 and 1100 miles from the source. A second band of loss in sensitivity occurs between 2000 and 2700 miles from the source. Since all the stations of the existing network are beyond 1000 miles from known Soviet test sites, the seismic detection capability for the existing system is not affected by insensitivity in this range. However, the region of loss in sensitivity at 2000–2700 miles does affect the seismic capability for detecting low yield subsurface shots with the existing network of stations. The reason for the loss in sensitivity in the range of 2000 to 2700 miles is believed to result from a rapid change in the physical characteristics of the mantle at approximately 1200 kilometers depth which results in a dispersion of seismic energy at the surface for this particular range from the source.

The effect of the transmission characteristics of seismic waves as a function of distance on detection by existing stations is shown in a map of the USSR in Figure 7. The areas of poor detection for surface tests resulting from this effect are cross hatched. Subsurface bursts in these areas will also be the most difficult to detect with the present network. A reasonable estimate of yield for subsurface shots which would be detected in these areas is obtained by dividing the yield for surface tests by a factor of 25 for summertime conditions and by 12-1/2 for wintertime. Thus, general statements about seismic detection capability cannot be applied uniformly to the entire Soviet-dominated area. However, for tests within approximately 90–95% of the area some general estimate can be made. Taking all the above factors into consideration, it is believed that the present seismic detection capability is as indicated in Table IV. The capability indicated in Table IV applies to the clear areas in Figure 7. In the cross hatched areas, however, subsurface tests of 10–15 KT might escape detection by the present network.

[Typeset Page 1275] [Facsimile Page 25]
Table IV
Seismic Detection* Capability for Low Yield Subsurface Nuclear Tests
Detection Coefficient YIELD (KT)
Summer Winter
Excellent 90–100% 4 8
Good 60–90% 3 6
Fair 30–60% 2 4
Poor 0–30% 1 2
*Detection but not identification as nuclear in nature. [Footnote is in the original.]

c. High Altitude Tests. The Long Range Detection System has no evidence to date that the Soviets have tested a nuclear weapon at extremely high altitudes. U.S. tests conducted so far have been limited to altitudes below 50,000 feet (HA, 37,000 ft). Theoretical investigations of the effects of very high altitude tests (100,000 to 250,000 ft) have been made by Professor Hans A. Bethe and others in recent months. These studies indicate many very interesting and unusual phenomena which will be discussed in detail by Professor Bethe in another paper.

Of the four existing detection techniques, only the acoustic and electromagnetic techniques appear to offer promise for the detection of very high altitude tests and these are debatable in the light of present knowledge. It is believed that adequate detection of high altitude tests will involve development of new or improved detection techniques following experimental high altitude tests at HARDTACK involving new influence fields, i.e., optical, magnetic, etc., as well as possible modification of existing acoustic, seismic and electromagnetic surveillance methods.

d. Nuclear Tests Conducted in Remote Geographical Locations. The AFOAT–1 Long Range Detection System is deployed around the USSR as the primary target. It therefore has very great [Facsimile Page 26] limitations for shot points in such remote geographical locations as the Antarctic, South Pacific and Indian Ocean areas. It is believed possible that shots of several hundred kilotons and possibly even a megaton could be detonated in these areas far from the present acoustic stations without fear of detection by acoustic means. British tests in Australia at points only 20°–30° S produced no nuclear, no seismic, no electromagnetic and, in six shots out of twelve tests, no acoustic data although advance notification of time and place of the test was provided by the U.K. Six tests did excite one and in two cases two acoustic stations but detection even in these cases was certainly greatly assisted by advance knowledge of U.K. intentions.

The seismic capability suffers very seriously from core shadow for shots detonated in these remote areas. Over 60% of the Southern [Typeset Page 1276] Hemisphere would produce very poor capability for shots in the low megaton range. Electromagnetic capability for detection of tests in the Southern Hemisphere is unusually poor because of the high frequency of thunderstorms in the inter-tropical front along the Equator. The ability to collect samples of debris in the Southern Hemisphere is very limited.

e. Summary of Existing Capabilities and Limitations.

(1) Surface or Air Burst 50 kilofeet within the USSR.

(a) The electromagnetic technique may possibly detect shots within the USSR of 25 KT or greater but requires a determination of time independently by some other geophysical technique to take care of the sorting problem.

(b) The seismic technique is capable of detecting surface or low air bursts of 100 KT or larger.

(c) The acoustic system is capable of detecting tests of 15 KT or larger.

(d) The nuclear technique is capable of detecting tests as small as 3 KT as long as the debris does not rise much above 20,000 feet.

(e) The overall detection capability for the Atomic Energy Detection System is excellent (90–100%) for surface or air burst tests of 10 KT and greater between the surface and 50 kilofeet.

[Facsimile Page 27]

(2) Subsurface Tests. The seismic system presently has an excellent capability of detecting subsurface shots of about 10 KT or larger over 90–95% of the area of the USSR. Subsurface tests as large as 10 to 20 KT in the remaining 5–10% of the USSR might escape detection by the present AEDS. It is estimated from Figure 8 that a total of about 140 earthquakes per year of 10 KT or larger occur in the USSR. Identification of about half this number is possible, leaving about 70 events which may be either explosions or earthquakes.

(3) High Altitude Tests. It is not possible to evaluate the present capability of the AEDS to detect high altitude tests.

(4) Nuclear Tests Conducted in Remote Geographical Locations. Tests in remote areas in the Southern Hemisphere of from several hundred kilotons to a megaton or two might escape detection by the present Atomic Energy Detection System.

Summarizing, the present Long Range Detection System will detect and identify surface or air bursts below 50 kilofeet in the USSR having a yield of 10 KT or larger. It will detect about 140 subsurface events per year (by seismic means) of 10 KT yield or larger of which only 70 will be positively identified as earthquakes leaving 70 subsurface events of unknown origin. The detection capability for very high altitude bursts over the USSR is unknown and for tests in remote geographical locations is limited to the high kiloton or low megaton range.

[Typeset Page 1277]


This section will discuss a Long Range Detection System within the USSR and China designed to meet the following requirements: (1) The detection and identification of tests as small as 1 KT between the surface and 50 kilofeet, and (2) Detection of underground explosions, and possibly their identification, in a yield range of 1 KT and above.

a. Seismic Net. In considering the deployment of seismic stations within the USSR to give 90% capability of detecting tests of 1 KT or larger, three problems of major magnitude appear:

The band of low signal level between 500 and 1100 miles from the source.
The high microseismic noise level expected at stations in average localities.
The frequency of earthquakes of yield equivalent to 1 to 5 KT which because of small signal amplitudes are the most difficult to identify.

As a result of the loss in signal level between 500 and 1100 miles, it is expected that signals from a one kiloton explosion in this distance range will be smaller than the prevailing microseismic noise. It is to be expected that the noise level at a station removed from sources of cultural noise would be as high as .005 microns. An inspection of Figure 6 will show that such a station would be unlikely to detect an explosion at a distance greater than 400 miles. This loss in detectability makes it necessary to use more elaborate techniques than are currently employed at AFOAT–1 stations in order to detect 1 KT without covering the USSR and China with an inordinately large number of stations. It is believed possible to obtain an improvement in detectability of about a factor of four at individual seismic stations by employing arrays of about 20 seismometers at each station. A spacing between the seismometers can be chosen which should result in considerable cancellation of microseismic noise without appreciable degradation of signals of interest. The seismometers would be distributed approximately uniformly over an area of about six square kilometers. With the signal-to-noise improvement of about four to one, it may be seen from Figure 6 that signals of the order of three millimicrons would be detectable between 0 and 500 miles and again between 1100 and 2000 miles.

In addition, it is believed that ultra-long period seismometers (par. 2c(1)) under study may add considerably to the detection of long period surface waves from small subsurface tests. To date, long period surface waves were observed on Wigwam but not on Rainier. However, these instruments would be included in addition to the improved arrays of the present instruments to supplement detection at ranges possibly unfavorable to the present equipment mentioned above if future tests confirm the existence of long period surface waves from small subsurface tests.

[Typeset Page 1278]

Relative to identification of the “detected” events, the annual number of earthquakes within the USSR-dominated area which are larger than a given yield is shown in Figure 8. This relationship is derived from statistical data on world-wide distribution of earthquakes from Gutenberg and Richter, “Seismicity of the Earth,” and AFOAT–1 measurements of seismic wave amplitudes from subsurface explosions. From this curve the number of earthquakes in the USSR and China which produce energy in the earth equal to subsurface explosions of 1–5 KT is found to be 2100. The areas of most frequent occurrence of these earthquakes are shown in Figure 9. The indicators mentioned in par. 2b [Facsimile Page 29] permit identification of about 85% this number as natural earthquakes provided there are at least three stations within 400 miles of the burst. There remain about 300 earthquakes per year in this yield range which cannot be distinguished from explosions by scientific means.

Considering all of the above factors, it is believed that it will be necessary to install about 43 seismic stations with improved arrays and ultra low frequency seismometers within the USSR and China. Possible locations of such stations are shown in Figure 10. The stations shown are concentrated near seismic regions to obtain the maximum amount of information leading to the diagnosis of natural earthquakes. Any shock originating in the vast aseismic areas of the USSR and China would be strongly suspected of being man-made. It is believed that this network of stations would permit a 3-station fix on a completely tamped subsurface shot as small [text not declassified] in about 90–95% of the area of the USSR and China.

It is estimated that about 300 earth shocks in the range of 1–5 KT would occur each year which could not be distinguished from subsurface nuclear tests. It would, therefore, be necessary for inspection teams to investigate most of these sites for the purpose of determining whether or not the seismic indications were from an explosion or from an actual earthquake.

b. Acoustic Net. It is estimated that about 30 acoustic stations within the USSR and China (see Figure 11) placed roughly on a 700 nautical mile square grid will provide an excellent (90–100%) capability to detect shallow subsurface, surface, and air bursts with a yield as low as [text not declassified] any place in the USSR and China and immediate surrounding water and island areas. This estimate is based on a study of the actual detection results obtained on all U.S. shots (29) in the yield range from .1 to 10 KT. These results indicate a reliable detection range of at least 700 nautical miles for a yield as [text not declassified]. This detection range, coupled with the requirement for detection by a least two acoustic stations in order to determine the location and time of explosion, established the grid pattern stated above.

c. Electromagnetic Net. Data available on electromagnetic detection of small U.S. surface or tower bursts indicate that a 1000-mile range [Typeset Page 1279] would give good assurance of obtaining electromagnetic data on shots as [text not declassified]. Data on U.S. high air bursts is, of necessity, very limited. It does indicate that the electromagnetic signal strengths from high altitude tests are considerably smaller than from surface tests. However, the electromagnetic signals [text not declassified] were [Facsimile Page 30] obtained at ranges of 1350 miles and 2400 miles, respectively.

It is, therefore, recommended that an installation of about 30 electromagnetic stations co-located either with seismic or acoustic stations (see Figures 10 and 11) would permit taking maximum advantage of the electromagnetic technique. Since the technique does not provide independent detection at the present time, its purpose would be to strengthen acoustic and seismic evidence of a clandestine test. In about one year improved sorting techniques may permit independent detection by the electromagnetic component and correspondingly increase its value.

It should be noted that relatively simple gamma ray shields of lead or water surrounding the bomb will prevent detection at useful ranges by the electromagnetic system. For example, there was a calculated and measured attenuation of the electromagnetic signal by a factor of [text not declassified] for the U.S. test which was shielded by water. [text not declassified] The signal strength, at long range detection distances, would thus be reduced well below the detectable limit.

d. Nuclear Sampling. The nuclear sampling network required to detect radioactive clouds from tests of 1 kiloton or larger in the USSR and China introduces a requirement to intercept air masses emanating from tests in this area at all altitudes from the surface to 50,000 ft. The difficulty of this problem arises principally from the fact that clouds from tests in the very low kiloton range have been observed to be no more than a few thousand feet thick. Setting up a picket line to insure interception of all such clouds by sampling aircraft is obviously impractical. It is estimated that a daily air filtering effort along a meridian at 135°E, at altitudes of 10,000, 20,000, 30,000 and 40,000 feet would provide ample coverage of air masses emanating from tests in both the USSR and China. Tests conducted near the east coast of these countries would produce clouds which would still be relatively small in extent as they pass the flight lines and would have the greatest chance of escaping detection. Clouds from tests further inland in Asia would be stretched out to considerably greater lengths and permit a greater chance of detection.

Several months after a moratorium on testing nuclear devices has been in effect the analysis of nuclear debris samples would become considerably more sensitive than they now are for the detection of fresh nuclear debris, since background levels from old debris would be considerably reduced. As a result, smaller samples would still provide [Facsimile Page 31] significant radiochemical analyses. The number of disintegrations per minute of 2.7-day Mo99 at various times after detonation is given in the following table for a sample size of 107 fissions, a rough [Typeset Page 1280] average of the intensity of a single particle of nuclear debris. From this table it can be seen that even ten days after zero time the Mo99 is still detectable even in a very small sample. In general, samples of debris are at least an order of magnitude larger than 107 fissions. Of course, other short-lived activities could be investigated; e.g., 17-hour Zr97, 33-hour Pr143, 7.5-day Ag111, 2.3-day Cd115, and 2.3-day Np239, to detect the presence of fresh debris.

Days after Zero Time dpm Mo 99 in 10 7 Fissions
2 66
4 40
6 24
8 14.5
10 8.7
12 5.3
14 3.2

In general, it is believed that daily aerial filtering at the altitudes and over the flight tracks defined above would provide 90–100 per cent certainty for detection and identification for yields of 3 KT or larger detonated between the surface and about 25,000 feet. Debris from tests as small as 3 KT detonated at altitudes between 25,000 and 50,000 feet will probably be found at altitudes in excess of those recommended for routine patrols. In this case, the air sampling technique will require an assist from the early warning geophysical network in order that high altitude aircraft can be dispatched to altitudes above those routinely filtered for interception of radioactive clouds from such tests.

Radiochemical laboratories should be located at or near flight terminals; in this case, probably Tokyo, Japan and Fairbanks, Alaska. These laboratories would be equipped with the latest techniques for determining the presence of fresh radioactive debris.

It is expected that for small tests deep in Asia, it may be necessary to overfly certain areas of the USSR and China in order to intercept debris which might not come out across the Japanese flight lines. These overflights might require permission to land and refuel at Soviet bases.

[Typeset Page 1281] [Facsimile Page 32]

e. Detection of Soviet Concealed Underground Nuclear Tests.3

“Under a test moratorium, it is possible that the Soviets might endeavor, because of overriding technical considerations, to conduct a limited series of low yield, underground tests on a surreptitious basis. Therefore, the seismic component of the inspection system within the USSR must be capable of detecting signals generated by an underground test, and the moratorium agreement should permit access for on-the-spot ground inspection of suspect areas upon presentation of seismic records that cannot be properly explained as a natural phenomenon.

“Should such evidence of a possible test be obtained from seismic data or other intelligence sources, it must be promptly evaluated in terms of all available information, including the pattern of normal seismic disturbances within the USSR. If the data cannot be discounted as a natural phenomenon, a small mobile inspection team should be dispatched to the area pin-pointed by the information at hand. In the event of underground tests, a mobile inspection team would provide the only capability for confirming the fact of a nuclear explosion, other than that derived inferentially from repeated suspect activity at a given site.

“Such teams must be able to move rapidly into the suspect area before instrumentation and other physical evidences of testing have been removed; conduct low altitude aerial reconnaissance; follow roads and car tracks, and inspect on the ground any unusual activity; and, if justified, interrogate residents in the vicinity and obtain, under extreme conditions, the right to drill for core samples. Such drilling operations are time-consuming and difficult, but provide the only proof that a very low yield deep underground test has been conducted.

“This elaborate inspection system and the concurrent use of all intelligence sources will not guarantee detection of such underground tests, particularly since seismic signals from a low yield underground test cannot be consistently distinguished from normal earthquake signals. However, these elaborate precautions should raise serious doubts in the minds of the Soviets as to whether they should risk such an attempted evasion. We believe it more likely under these circumstances that they would abrogate a test moratorium under some false pretense rather than by surreptitious testing.

[Facsimile Page 33]

“The expansion of the nuclear test detection system to include geophysical stations and mobile inspection teams within the USSR will greatly increase our overt and covert intelligence collection capabilities against a wide range of other Soviet activities. While this does not of itself justify implementing the proposed test moratorium and inspection system, it is a bonus therefrom that would have direct bearing upon military order of battle, early warning and economic intelligence.”

[Typeset Page 1282]

f. Contribution of Other Intelligence Sources to the Detection of Clandestine Nuclear Test 4

“In addition to the aforementioned capabilities of the U.S. Atomic Energy Detection System as augmented, it is anticipated that a concurrent expansion of the clandestine collection effort will be attempted. This covert effort will be designed to help detect attempted evasion of a test moratorium via the development of selected informants and the surveillance of activities which may be indicative of such an attempt.

“The routine flow of information from all other intelligence sources should provide assistance in alerting us to possible attempts at evasion within the USSR, but they will not provide any direct evidence that a test in violation of the moratorium has been conducted. The major contribution of such sources would be the detection of preparations for such an evasion and the targeting of the general area involved.

“Conceivably, Soviet clandestine nuclear tests could be staged in remote areas outside the Soviet Bloc such as Antarctica or southern waters. However, all intelligence agencies have agreed that such possibilities would probably be excluded by the Soviets, since various conventional intelligence collection efforts would be almost certain to spot the activities which would be associated with test preparations, if not the test itself. Difficulty would probably be encountered in proving such a test had actually been conducted unless fresh radioactive debris was obtained, and this could be associated in some way with Soviet operations.”

g. Security and Classification Problems. It would probably not be advisable to reveal the existing operational network of the AFOAT–1 Long Range Detection System to other nations for the following reasons: [Facsimile Page 34]

Termination of the moratorium would leave the U.S. with the requirement to use a compromised world-wide network of stations.
Many small countries presently giving hospitality to AFOAT–1 stations would be compromised by release of the information to the Soviet Union.

The operation of an International Test Monitoring System will presumably be for the purpose of producing scientific proof of a violation by the U.S., the USSR, the U.K., or any other nation. The type of scientific proof will be as follows:

Seismic, acoustic or electromagnetic records of nuclear tests which permit determinations of time, place, height of burst, and yield.
Radiochemical data on samples of debris which establish its date of origin, nature of the device (fission or fusion), etc.

[Typeset Page 1283]

The present Atomic Energy Act of 1954 classifies data in both of these categories for both U.S. and U.K. weapons as well as the samples of debris themselves as Restricted Data because they may reveal “important information on the design and fabrication of nuclear weapons.”

Two problems arise, therefore, in the operation of a monitoring system. First, the data in the above categories would have to be cleared for distribution to foreign nationals, and second, some procedure would have to be developed to limit analysis of debris to those measurements necessary to establish the freshness of the debris but to exclude more esoteric measurements which may reveal important information on nuclear devices exploded by either the U.S. or the U.K.

h. Installation Costs of a Long Range Detection System for Clandestine Tests in the USSR . The Long Range Detection System within the USSR discussed above involves 30 acoustic stations of the present type, 43 seismic stations utilizing improved instrumentation at each station, 30 electromagnetic stations utilizing present equipment plus complete aerial filtering coverage of air masses emanating from the USSR and China. The seismic and acoustic stations, because of their mutually incompatible technical site requirements, cannot [Facsimile Page 35] generally be co-located. The electromagnetic stations, however, are usually found, from a technical standpoint, to be compatible with either seismic or acoustic locations and they therefore may be co-located with one or the other. Therefore, a total of 70 detection stations within the USSR and China would be required.

The time required to install such a system will depend on a number of factors. First, negotiation with the Russians for station locations could turn out to be quite protracted. Technical surveys of possible site locations will have to be made for each of the acoustic, seismic and electromagnetic components installed. Low, flat land remote from electrical disturbances both man-made and natural, as well as from large metal buildings, fences, etc., are the principal requirements for a good electromagnetic station. Level or slightly rolling terrain in areas of relatively low wind velocity provide good acoustic locations. Residential areas of cities are practical since man-made noise does not affect the ultra low frequency acoustic detectors. The geology of the region is critical to satisfactory seismic locations. Of importance also to a good seismic site are areas of low storm activity and areas remote from man-made or natural noise.

The above general statements concerning the requirements for adequate site locations are greatly simplified and presented only to indicate that the survey for the location of all the proposed stations will be a relatively involved and time-consuming operation. Usually it is desirable for the survey team to remain in the area taking background measurements for a period of approximately 1 to 3 months, depending on the technique which is to be used at the site. If several survey teams could operate simultaneously, the time could be reduced. [Typeset Page 1284] A conservative estimate indicates that the survey would take of the order of at least six months for all the stations suggested.

The procurement of equipment, which could be initiated concurrently with the surveys, is estimated to take from 6 to 18 months, depending upon the type of equipment. The installation of the equipment, which must take place in series with the surveys, will take from 12 to 18 months, depending on the type of installation. Training of operators, which can start concurrently with the start of the survey, will take from 6 to 12 months, depending on the type of training and upon the procurement of personnel to be trained. This could become relatively involved if utilization of personnel of several nations is planned.

[Facsimile Page 36]

Based upon the above time factors, it is conservatively estimated that two years would be required for the installation. Assuming a minimum of ten months to determine U.S. policy and to negotiate an agreement with the USSR, the system would not be in operation prior to 1 January 1961.

A rough estimate of cost of the installation has been made based upon the assumption that each installation be self-sufficient and independent of the local economy. It was further assumed that all materials and labor would be provided from the U.S. Individual station costs include water supply and filtration, electrical power supply, sewage disposal, access roads, pole lines and vaults, equipment shelters, barracks and latrines, mess hall, vehicle storage and maintenance, fuel storage, technical and support equipment and, most costly of all, a complete communications center. Obviously, depending upon local conditions at each site, the extent of local support in manpower, materials and communications, considerable reduction in cost figures could be effected. These rough estimates indicate that the equipment and installation would cost about $100,000,000 and operation might run $30,000,000 per year.


In Section 3 of this report, a comprehensive detection system has been described which would detect clandestine tests of relatively low yield within the USSR. While this system is by no means fool-proof, principally because of the possibility of evasion by conducting tests underground where detection and identification of all tests is extremely difficult, it would certainly force the Soviets to consider other environments in which clandestine tests might be conducted either with less risk of detection or with greater facility.

It has been pointed out previously in this report that the existing AFOAT–1 Long Range Detection System, as well as the system within the USSR and China described in Section 3, will have a very poor detection and identification capability for tests in areas of the earth remote from those detection systems. The Soviets certainly are aware of this fact and therefore might be strongly tempted to conduct tests in these remote areas using a submarine task force. Tests under these conditions would probably permit limited but adequate diagnostics for the purpose.

[Typeset Page 1285] [Facsimile Page 37]

One rather disconcerting possibility in this connection would be for a Soviet submarine to deliver a nuclear device to the Marshall Islands and detonate it, fully aware that it would be detected. Such a maneuver would probably be followed up by Soviet propaganda that the U.S. had conducted another test in violation of a moratorium agreement. Proving the national origin of such a test by any scientific means would be extremely difficult.

With possibilities such as these in mind, a system is now described which would permit a greatly increased detection capability in the most remote areas of the world. This system includes acoustic, seismic, electromagnetic and air sampling components designed on the basis of availability of land masses in the Southern Hemisphere and in the large ocean areas of the Northern Hemisphere rather than being designed for a specific detection capability. The system has then been evaluated for its probable detection capability.

a. Acoustic Net. It is estimated that about 22 acoustic stations (see Figure 12) would be required to supplement the present long range detection net in order to achieve a capability of 90–100 per cent to detect shallow sub-surface, surface and air bursts with a yield as low as 40 KT anywhere in the world outside the USSR and China. This estimate is based on experience in detecting U.S., U.K and USSR nuclear tests which indicates that explosions as low as 40 KT can be detected with excellent reliability at a range of 2500 nautical miles. The requirement that at least two acoustic stations detect an explosion in order to establish the location and time of the event also influences the choice of stations.

It should be noted that two factors make the capability estimates less reliable for places outside the USSR than for those within the USSR. One is the fact that there have been only 15 nuclear explosions conducted in the Southern Hemisphere, e.g., U.K. tests in Australia, and acoustic data on these are sparse. The other is the fact that a number of island locations for stations had to be chosen in order to cover the Antarctic and remote ocean areas. Noise levels at these locations will probably be considerably higher than at present sites, making detection more difficult.

It is estimated that such a net could detect shallow sub-surface, surface or air bursts with a capability of 90–100 per cent for 40 KT or larger, 60–90 per cent for 10–40 KT, 30–60 per cent for 5–10 KT and 0–30 per cent for less than 5 KT.

[Facsimile Page 38]

b. Seismic Net. About 10 seismic stations (see Figure 12) of the type presently used by AFOAT–1 would be required to supplement the present long range detection net in order to achieve a capability of 90–100 per cent to detect underground bursts with a yield as low as 20 KT anywhere in the world. This estimate is based on experience in detecting the Rainier shot and a number of underground high explosive charges. The limited data available indicate that seismic signal amplitudes expected from a 20 KT sub-surface burst would be about twenty times larger than shown in Figure 6. Assuming good station locations could be found with [Typeset Page 1286] noise levels of the order of 5 millimicrons, signals should be detectable at all distances out to 6000 nautical miles. A minimum of four detecting stations is required to determine the location, time of occurrence and depth of focus of each shock. Suitable station sites south of the equator are rather limited in number, since the Southern Hemisphere is predominantly oceanic, and island locations are generally found to be extremely noisy. With the addition of stations shown in Figure 12, it is believed that signals could be received at four or more stations from large subsurface shots in all parts of the world. The region of least detection capability is an area of about 1000 miles radius centered near 10° S, 165° W in the South Pacific Ocean. It is estimated that such a worldwide net would have a capability of 90–100 per cent to detect subsurface bursts of 20 KT or larger. Below 10 KT the detection capability deteriorates very rapidly, e.g., at 5 KT there is virtually no possibility of three-station detection.

The identification of those natural earthquakes producing signals of the same size as a 20 KT sub-surface burst is discussed in paragraph 2b. It is estimated that 400 shallow earthquakes per year will be in this class. Of these, about 300 can be identified as earthquakes with reasonable certainty, leaving 100 events per year requiring detailed investigation by inspection teams.

c. Electromagnetic Net. Electromagnetic detection of tests in the Southern Hemisphere is favored by the fact that most of the propagation paths will be over water, which will give somewhat less attenuation than over land. However, an unfavorable characteristic of the Southern Hemisphere with respect to detection is that South America, Africa and the East Indies are areas of high thunderstorm activity. Furthermore, the inter-tropical front which encircles the globe approximately at the equator with seasonal shifts from the north to south, constitutes a band of additional thunderstorm activity.

[Facsimile Page 39]

A network of 23 electromagnetic stations has been selected, with the stations made parasitic on the acoustic or seismic locations shown in Figure 12. The detection capability for the network as a whole is estimated to be 90–100 per cent for tests of 100 KT or greater throughout these remote areas of the world.

d. Air Sampling Net. In designing a network of air sampling flights for nuclear detection in the remote ocean areas of the world, consideration has been given to the complexity of the wind patterns on a global basis. It is estimated that daily aerial filtering along three meridians with approximately pole-to-pole coverage at each of two altitudes, e.g., 20,000 and 40,000 feet, will provide a detection capability of 90 per cent or greater for tests of about 5 KT detonated between the surface and 25,000 feet. Tests detonated at altitudes higher than 25,000 feet will require special-vectored high altitude sampling flights based on early warning from other components of the detection system.

As is shown in Figure 13, the pole-to-pole coverage is along meridians at approximately 140° E longitude, 50° W longitude and 20° E [Typeset Page 1287] longitude. Radiochemical laboratories would be located at Tokyo, Melbourne, Montreal, Buenos Aires, Rome and Capetown. These laboratories would be equipped with the latest devices for the detection of fresh nuclear debris.

e. Overall Detection Capability. It is estimated that this system of about 30 geophysical stations plus an air sampling network could detect and identify with 90–100 per cent certainty nuclear tests of 20 KT or larger conducted as shallow sub-surface, surface or air bursts up to 50,000 feet in remote areas of the world. Underground shots of 20 KT or larger in remote parts of the world would be detected with a certainty of 90–100 per cent but not identified as nuclear explosions. Identification of such underground disturbances may be possible through the use of inspection teams investigating about 100 disturbances per year in these remote regions of the world.

The U.S. underwater test (WIGWAM), conducted at a depth of about 2000 feet, produced a unique spectrum of seismic waves at distances of 3000 miles. Ultra-long period seismometers would probably extend this range at least by a factor of two. The existence of these waves suggests that any similar occurrence in remote parts of the world would attract immediate notice on the network [Facsimile Page 40] of seismic stations described above. It is therefore believed that an underwater explosion of 20 KT or larger anywhere in the world would be reliably detected and probably identified as an explosion rather than an earthquake, because of the marked difference between such signatures and normal earthquake signatures.

D.L. Northrup
Technical Director, AFOAT–1
Office for Atomic Energy, DCS/O
[Facsimile Page 41]

List of Figures

1. AFOAT–1 Acoustic Net

2. AFOAT–1 Seismic Net

3. AFOAT–1 Electromagnetic Net

4. AFOAT–1 Flight Tracks

5. Air Sampling

6. Seismic Amplitude vs Distance

7. Seismic Detection Capability

8. Earthquake Frequency vs Yield

9. Seismicity of Parts of Europe and Asia

10. Proposed Seismic Stations

11. Proposed Acoustic Stations

12. Proposed Geophysical Stations (Southern Hemisphere)

13. Proposed Daily Surveillance Flight Tracks

[Typeset Page 1288] [Facsimile Page 42]
Figure 1
[Typeset Page 1289] [Facsimile Page 43]
Figure 2
[Typeset Page 1290] [Facsimile Page 44]
Figure 3
[Typeset Page 1291] [Facsimile Page 45]
Figure 4
[Typeset Page 1292] [Facsimile Page 46]
Figure 5
[Typeset Page 1293] [Facsimile Page 47]
Figure 6
[Typeset Page 1294] [Facsimile Page 48]
Figure 7
[Typeset Page 1295] [Facsimile Page 49]
Figure 8
[Typeset Page 1296] [Facsimile Page 50]
Figure 9
[Typeset Page 1297] [Facsimile Page 51]
Figure 10
[Typeset Page 1298] [Facsimile Page 52]
Figure 11
[Typeset Page 1299] [Facsimile Page 53]
Figure 12
[Typeset Page 1300] [Facsimile Page 54]
Figure 13
[Typeset Page 1301]

Appendix B

[Facsimile Page 55]

[Appendix B not declassified (4 pages of source text).]

Appendix C

[Facsimile Page 56]


Harold Brown and Hans A. Bethe

I. General

Complete containment of a test explosion underground, though devised as a method for making testing easier by eliminating fallout, may also serve as the most effective method of concealing the existence of tests, and may make it very difficult to gather effective proof that such tests have been carried out in violation of a suspension agreement.

Such an explosion provides no electromagnetic signal, and the acoustic signal if it exists at all will be so muffled and distorted that it will not be characteristic even at a distance of a few hundred miles. No activity is released into the atmosphere, so that the only detection method is the seismic. To provide proof by scientific means the residual activity from the explosion must be located underground and sampled.

II. Results from Rainier Test

The only such shot carried out in this country which casts any light on such procedures is the Rainier shot of Operation Plumbbob. This was 1.7 KT in yield, and was buried 800 feet from the nearest ground surface, in volcanic tuff. About 1% of the energy appears to have gone into the seismic wave, producing a magnitude 4.2 earthquake indication on seismographs a few hundred miles distant from the shot. Accelerometers indicate that the top of the mesa under which the device was located moved up about a foot and then fell back again. Rocks on top of the mesa were displaced somewhat, and some were rolled down the side, but the appearance of the surroundings after the shot was not inconsistent with the results of a small earthquake; rocks moved by the shot could not be distinguished from those moved by past earth motions, and fissures were present both before and after. Most observers at a distance of 2-1/2 miles felt no earth shock. This is due principally to the absence of hard rock between the source and the observation point. (It is possible that an underground shot will create less disturbance above ground than an earthquake of the same seismic magnitude, [Facsimile Page 57] and that this might be ascertained by examination of the ground around the event and by questioning of the local population, if [Typeset Page 1302] any, but extensive experience in the local effects of underground shots would be needed before any such difference could be established.)

One could expect to contain the shot completely without venting by using as little as 500 feet distance to the surface if more disturbance of the surface were allowed; such disturbance would still not be characteristic of an explosion rather than an earthquake. Burial depth required will vary with somewhere between the 1/3 and 1/4 power of the yield; a reasonable formula is 400 W0.3 feet (W in KT).

No activity above background was discernable either above the ground or in the tunnel leading to the explosion chamber, which was blocked off only 200 feet from the shot site. Thus the absence of activity is not merely absence of a radioactive cloud at several hundred miles, but of any radioactivity above ground or in any other region accessible without drilling. The horizontal access tunnel, 1700 feet long from the portal, showed some slabbing and cavein for several hundred feet beyond the point where it was blocked.

Exploration of the region around the zero point by drilling in from the tunnel at a distance of 210 feet has revealed that the solid fission products are contained in a shell a few feet thick at a radius of 55 feet. After four months the peak activity measured along a line at the level of the zero point was 800 mr/hour, while along a line aimed at a point 50 feet below zero from a point 210 feet away horizontally the peak was 40 r/hour. Outside of the shell the activity as measured by a counter was indistinguishable from background. Peak temperature along the horizontal line was 45°C, along the other line it was 65°. Diffusion appears to have carried elevated temperature into the zero point, and some rise above ambient is also noted out to about 70 feet.

Thus a 55-foot radius hole appears to have been established momentarily but then to have collapsed, and the falling in appears to have continued up to a point 400 feet above zero, where a hole 25 feet in radius and 25 feet high was discovered in drilling. This hole contained gaseous fission products at the same concentration as they appear inside the 55-foot radius region around zero, so the entire volume in between appears to be simply connected. This accounts for only a few percent of the gaseous fission products, and it is thought that the remainder were trapped in the resolidification of the [Facsimile Page 58] molten rock. The region from 55 feet to 130 feet is still impervious but was apparently crushed since the drilling shows water return but no core return. It has not yet been feasible to detect this crushed region by sonic measurements even from inside the tunnel, so that detection by sonic means from above the surface is at least very difficult.

III. Diagnostic Experiments

Diagnostic experiments necessary for weapon development can be easily carried out underground. The yield can be measured by shock [Typeset Page 1303] arrival time measurements in the rock, analagous to the fireball measurements above ground. This was done on Rainier and appears to be accurate to 10% even without a calibration. The radius of the radioactive debris or the amount of material melted might also be used if the medium is calibrated by a shot of known yield. The prompt diagnostics such as neutron and gamma ray measurements to give time interval or propagation burning data, streak camera work, etc., can be done better below than above ground since the shielding is free and one need only drill holes as desired. Radiochemistry has not been demonstrated to be satisfactory, since fractionation does occur. However, the use of hollow pipes leading into reception chambers from the device may give a substantial fraction of debris unfractionated, and may lead to satisfactory radiochemical diagnostics.

Preliminary estimates indicate that a test operation can be carried out more cheaply underground than on towers and balloons. The diagnostic stations could also be underground for clandestine tests (in fact they probably will be even if the tests are not hidden). Keeping underground tests secret will increase the costs by preventing the use of a single diagnostic bunker for many shots on the basis that more than one in a given vicinity increases suspicion and the possibility of proving a violation. It may mean that each shot must be in a completely different area, but this conclusion may be modified to some extent, because also natural earthquakes have aftershocks. In any event, such extra costs are associated with clandestine tests generally rather than underground tests specifically and are not likely to be more than a few million per shot, which is not a large percentage increase.

IV. Dependence of Seismic Signal on Yield and Medium

On the basis of observations, it is believed that the amplitude of the earth motion from an underground explosion increases as the [Facsimile Page 59] 1.2-power of the energy released. This scaling law is obtained on the basis of explosions of conventional explosives underground (quarry blasts). The law is somewhat surprising; theoretically one would expect that the seismic amplitude would go as the square root of the energy release. The empirical law has been used in Appendix A to predict the frequency of earthquakes in the USSR which might be confused with subsurface shots of various yields. The empirical law clearly gives more larger results for the seismic signal to be expected from shots of larger yield than Rainier than the “theoretical expectation” would give.

The empirical law indicates that a larger fraction of the energy release goes into seismic waves at higher yield. This effect certainly must stop at some point; at about 100 kilotons the entire energy would be converted into seismic energy if the 1.2-power law held up to that yield. Experiments are urgently needed to establish the actual relation between yield and seismic signal. These should be carried out with [Typeset Page 1304] nuclear explosions since conventional explosives may not give the same effect due to the evolution of large amounts of gas.

The seismic signal will depend strongly on the medium in which the test is conducted. The volcanic tuff in which the Rainier test was conducted probably gives a relatively small seismic signal; it is only equivalent to an air shot of about 20 times greater yield. Hard rock would almost certainly give a stronger seismic signal while on the other hand it may contain the radioactive products in an even smaller volume. On the other hand, unconsolidated material which is found in many places near the surface of the earth may well reduce the seismic effects below those observed in the tuff because the signal should decrease with decreasing yield stress, and unconsolidated material may have a yield stress as low as one-tenth of that of tuff (which has about 10,000 psi).

It may also be possible, by excavating a large chamber to begin with, to reduce the energy found at large distances by a factor of 10. One possibility which may reduce the seismic energy is the excavation of large cavities in salt domes. Such cavities may be tens or even hundreds of millions of cubic feet in volume, and need not be spherical. For example, a cavity 150 feet in diameter and 3000 feet long may have nearly the same effect as a spherical one of the [Facsimile Page 60] same volume. The excavation of such a cavity would be fairly costly, and its use might be limited to a single occasion because it might cave in. To find out to what extent the seismic signal from an underground explosion could be reduced by suitable choice of medium, many tests would be required but most of these could be carried out at low yield.

It is likely that reduction of seismic signal is easier for low-yield shots than for high-yield ones. Unconsolidated material is found only in the top layers of the earth and the required burial depth increases with yield, so that it may be difficult to find such material deep enough to successfully contain a 50-kiloton test. The digging of underground caves large enough to give a substantial reduction of the seismic signal from a 50-kiloton explosion will be very costly and may in fact be impossible, especially since for mechanical stability a cave must be smaller at great depth than near the surface. Thus it may well be possible to reduce the signal from a 5-kiloton explosion so that it “looks like” 1/2 kiloton, but more difficult to make 50-kiloton explosion appear like 5 kilotons.

V. Identification

It is shown in Appendix A that the seismic wave from a 1-kiloton subsurface explosion in surroundings similar to those of the Rainier shot will be detected by the net of seismic stations proposed for the USSR in that appendix. However, there are about 2500 earthquakes per year in the USSR which give signals of similar strength. The most promising feature of seismic signals from underground explosions distinguishing them from earthquakes is that the first pulse from explosions always [Typeset Page 1305] corresponds to compression while the first pulse from an earthquake is compressive in two quadrants, while it corresponds to dilatation in the other two. It is estimated in Appendix A that there will be about 300 earthquakes of strength equivalent to 1 kiloton or over which will give signals in the proposed seismic detection net which cannot be distinguished from nuclear explosions and therefore will require further investigation on the spot. If the limit is set at 5 kilotons the number of unidentifiable earthquakes will be about 35.

It should be pointed out that 1 and 5 kilotons refer to the size of the seismic signal, not to the actual yield. By proper choice of the [Facsimile Page 61] medium as discussed in Section IV, tests of 10 kilotons might be made to look like a normal 1-kiloton explosion, and perhaps, with more difficulty, 50 kilotons to look like 5. According to seismologists, it is unlikely that a nuclear explosion could be so conducted (by proper shaping of the explosion chamber) that the signal is dilatational in some directions.

The seismic signals would locate the source within about 5 miles. Investigation on the spot will then be necessary to decide whether the signal could be due to a test, this is described in Section 3e of Appendix A.

One would presumably try to find the entrance to the tunnel which was used for the test. The experiment could be carried out in a remote area, where there would be no people to give away the game, but then such indicators as roads, unusual human activity, etc., might make the inspection team dispatched on receipt and study of the seismic signal suspicious. They would still have to find the entrance (say a 6 foot hole, since covered up), proceed to the correct part of the tunnel, and drill successfully to get proof. This is made difficult by the small radius of the shell in which the radioactivity is concentrated (55 feet for Rainier). Alternatively one might use an area of substantial human activity, thus producing less unusual change in what is going on, but perhaps requiring more local people to know about what was going on or become suspicious about it.

To summarize:

Detection of underground shots depends entirely on a seismic net. Identification depends on local investigation.
Adequate proof of violation probably depends on location of the debris, which is confined to a shell whose radius is of the order of 40 W1/3 feet where W is in KT, and whose depth is of the order of 400 W1/3 or larger as desired. Within the five mile radius circle of uncertainty identified by the seismic signals, the entrance to the tunnel or hole must be found as a beginning in finding the activity. Broad access is required to have a good chance of locating the debris and thus providing proof.
Adequate diagnostic information for weapon development can almost certainly be obtained at no great increase (and perhaps some decrease) in cost by testing underground. It has not been proven that radiochemical detectors can be used, but it appears possible that some [Typeset Page 1306] can by appropriate design of underground chambers. Some extra cost may be incurred if it is required to duplicate diagnostic bunkers, etc., in order to avoid testing several devices in one region so as to reduce suspicion; this is characteristic of clandestine rather than of underground shots.
Experimental data is lacking or insufficient and should be acquired either prior to or as part of an agreement on the following subjects:
Reducing the seismic energy by choice of medium and design of explosion chamber. Reducing the radius of the radioactive region by choice of medium.
The complete range of radiochemical detectors in diagnostics of underground shots.
Possible distinctive characteristics of underground explosions which will enable them to be told surely from natural earthquakes. This includes the seismograph records at a distance, and earth motions nearby. Possible special chamber design to remove such distinctions if any are found to exist must also be studied.
Use of acoustic sounding from above the surface to detect the disturbed region below the surface. This has not yet proven feasible even from inside the tunnel.

[Facsimile Page 63]

Appendix D

[Appendix D not declassified (5 pages of source text).]

[Facsimile Page 64]

Appendix E

[Appendix E not declassified (16 pages of source text).]

  1. Source: Appendices A–E to Report of the NSC Ad Hoc Working Group (Bethe Report) (print Document 147). Technical feasibility of cessation of nuclear testing; Appendices B, D, and E not declassified (25 pp.). Top Secret; Restricted Data. Appendix A is Secret. 64 pp. Eisenhower Library, White House Office Files, Additional Records of the Office of the Special Assistant for Science and Technology, Disarmament—Bethe Report.
  2. Report enclosed with this copy. Special security precautions should be observed in the handling of the enclosures, and access to them should be limited on a strict need-to-know basis. [Footnote is in the original.]
  3. A – Acoustic

    S – Seismic

    E – Electromagnetic

    N – Nuclear

  4. * Detection but not identification as nuclear in nature. [Footnote is in the original.]

  5. These opinions of the Central Intelligence Agency were prepared by the Assistant Director of CIA for Scientific Intelligence, Dr. Herbert Scoville, Jr. and are included at this point in the AFOAT–1 report at the request of the Chairman of the Ad Hoc Panel on Nuclear Test Limitation. [Footnote is in the original.]
  6. These opinions of the Central Intelligence Agency were prepared by the Assistant Director of CIA for Scientific Intelligence, Dr. Herbert Scoville, Jr. and are included at this point in the AFOAT–1 report at the request of the Chairman of the Ad Hoc Panel on Nuclear Test Limitation. [Footnote is in the original.]