3. Letter from Fisk to McCloy, March 21

[Facsimile Page 1]

Dear Mr. McCloy:

We submit herewith the final report of the Ad Hoc Panel on the Technical Capabilities and Implications of the Geneva System. All members of the Panel except General Loper have concurred in this report in full. While General Loper concurs in the first six sections of the report, he has not concurred with Section VII and will submit his comments to you separately. Dr. York concurred in the report in substance on the basis of an earlier draft, but he has not as yet had an opportunity to review the final report. The various members of the panel participated as individuals and not as representatives of their respective organizations.

Dr. J.B. Fisk, Chairman

Dr. Nano A. Bothe

General Austin Botts

Dr. Harold Brown

Mr. Spurgeon N. Keeny, Jr.

Dr. Richard Latter

General Herbert Loper

Dr. J. Carson Mark

Mr. Doyle Northrup

Dr. Wolfgang K.H. Panofsky

Dr. Frank Press

Dr. Herbert Scoville

General Alfred D. Starbird

Dr. Herbert York

[Typeset Page 8]



This report attempts to compile the technical material which has bearing on the broader questions of policy formulation in connection with the Geneva Conference on Cessation of Nuclear Tests.

We are submitting this report with the earnest desire that policy decisions on our future course be made with full understanding of the technical and related military considerations but with the realization that this is not a problem where positions should be controlled by the technical issues.

Policy on an arms limitation measure such as the test ban question is dependent on many technical, military, and political factors. Among these factors are:

1. The importance of the military considerations to all sides. These considerations must be analyzed for each of a variety of assumptions as to the response to an arms limitation agreement.

2. The technical means of verifying violations. Analysis of this question will naturally involve a mixture of well-established as well as speculative scientific information. In particular, we note that technical means of detection and of evasion are subject to change.

3. The unilateral technical and non-technical intelligence means of discovering violations.

4. Cost of the control operations.

5. Cost of evasion of a control system.

6. The judgment as to the degree of control which would “deter” a violator from evasion or make evasion relatively unproductive.

7. The degree of access provided by the control measures.

8. Growth potential of a control system to future arms limitation functions.

9. The relation of the negotiations to domestic and world opinion.

These and further purely political factors must be balanced with the risks or gain involved and thus form the basis of progress in this field.

[Facsimile Page 2]

This report deals with questions 1–5 in the above tabulation in relation to the proposed ban. Even under the restriction to these technical-military issues we meet questions where the results depend on over-all military policy. In such cases, we have attempted to give answers under alternate military assumptions. The report is divided into the following sections:

I. Capabilities of the Unilateral U.S. Long-Range Nuclear Test Detection System

1. Present Capabilities

2. Programmed Improvements

[Typeset Page 9]

3. Contributions of Other Intelligence Sources

II. Capabilities of the International Control System to Monitor a Nuclear Test Ban Agreement

1. U.S. Proposed International Control System

a. Atmospheric tests

b. Underwater tests

c. Underground tests

d. High altitude tests

e. Anticipated contributions from other intelligence sources

2. Cost of the U.S. Proposed Control System

3. Possible Control System Modifications

III. Estimated Changes in the Control System Capability from Future Research

1. Changes Resulting from the Seismic Research Program

2. Changes Resulting from the High Altitude Research Program

IV. Present and Projected U.S. and Soviet Nuclear Weapons Developments

The weapons developments are analyzed under various conditions as to test cessation and possibilities of evasion.

[Facsimile Page 3]

V. “Nth” Country Nuclear Weapon Developments

VI. The Cost of Evasion

A final section evaluating the impact of nuclear weapons development on U.S. and Soviet military weapons systems under a nuclear test ban as compared with continued unlimited testing is still under consideration by the Panel and is not included in this report.

[Facsimile Page 4]

I. Capabilities of the Unilateral U.S. Long-Range Nuclear Detection System

1. Present Capabilities

The U.S. Long-range detection system consisting of acoustic, seismic, electromagnetic, and radioactivity detection components is presently deployed around the USSR and its satellites. This system has a 60 to 90 per cent probability for detecting and identifying nuclear explosions of 5 KT or greater conducted within the USSR and China on the surface or in the atmosphere up to 10 km altitude. For explosions occurring between 10 to 30 km it may not be possible to collect samples of radioactivity required for identification.

Underground nuclear explosions greater than 5 to 10 KT (Rainier coupling) in the USSR and 10 to 20 KT (Rainier coupling) in China can be detected, but not identified, with 60 to 90 per cent, or greater, probability. The system cannot identify underground explosions.

[Typeset Page 10]

Underwater explosions (depths of about 500 feet) of ½ to 1 KT near the USSR and 1 to 2 KT near China can be detected seismically with 60 to 90 per cent probability. The probability of identification is uncertain.

The present system has essentially no capability for detecting high-altitude nuclear explosions (above 30–50 km).

Since the present system was designed to detect and identify atmospheric explosions carried out in the USSR, its capabilities for explosions outside the USSR and in other environments are limited. Atmospheric nuclear explosions as large as a few hundred kilotons or more might be missed if conducted in areas remote from the present detection system. Megaton explosions at extreme altitudes would not be detected.

2. Programmed Improvements

The U.S. nuclear test detection system is currently being expanded to improve its capability to detect explosions in China, at high altitude and in space, and in the atmosphere of the Southern Hemisphere. By 1965, it is programmed to have the following improved capability.

a. 60 to 90 per cent probability of detecting and identifying atmospheric nuclear explosions (below 10 km) of about 5 KT in the USSR and China and of about 20 KT outside these areas.

b. 60 to 90 per cent probability of detecting underground nuclear explosions of 1 to 2 KT (Rainier coupling) in the USSR and China and a small probability of identifying large earthquakes.

[Facsimile Page 5]

c. 60 to 90 per cent probability of detecting underwater explosions near the USSR and China of 0.2 to 0.4 KT.

d. Substantial capability of detecting high-altitude unshielded nuclear explosions of 5 KT at altitudes between 30 km and about 105 km above the earth.

The capabilities in b. and d. above are based upon the assumption that the present estimated research results actually materialize in the next three years for the VELA program of the DOD.

3. Contributions of Other Intelligence Sources

During the past twelve years other intelligence sources have made a significant contribution to the detection and acquisition of information on Soviet nuclear weapons test activities. This contribution has included:

(a) Alerting the US unilateral nuclear test detection system (NTDS) to the general location and timing of test preparations.

(b) Obtaining information from other sources on tests detected by the NTDS.

(c) [text not declassified]

During this period the intelligence community has no evidence that the Soviets have endeavored to conceal completely the geophysical [Typeset Page 11] signals produced by Soviet tests. However, the Soviets have made an obvious attempt to maintain a high degree of security in test operations.

Although there has been no specific experience in detecting very high-altitude or space tests, a variety of intelligence sources can detect major Soviet missile and space launchings and high vertical firings with a high degree of probability from present test ranges, and a good probability from any new ranges in the USSR.

Intelligence sources were able to detect test-related naval maneuvers and logistical preparations for Soviet underwater tests and furnished collaborative information subsequent to these events. It is quite probable that future tests of this nature would be similarly detected and reported, particularly if the test site were located distant from the USSR.

[text not declassified]

[Facsimile Page 6]

II. Capabilities of the International Control System to Monitor a Test Ban Agreement

1. U.S. Proposed International Control System

The U.S. proposed international control system is based on the report of the 1958 Conference of Experts. It consists of 170 control stations (21 in USSR, 14 in U.S. and 1 in U.K.) located about 1700 km apart in aseismic, and 1000 km apart in seismic regions. In addition, some control stations and 10 ships are equipped to detect underwater explosions. For the detection of high-altitude explosions, Technical Working Group I of the Conference on the Discontinuance of Nuclear Weapons Tests recommended the inclusion at control stations of specific ground-based instrumented earth satellites. Solar satellites were to be included if deemed necessary.

The U.S. proposed phasing for installation of the system requires half of the control stations installed in the USSR, U.S. and U.K. in 2 years, the remainder of the stations installed in the USSR, U.S. and U.K. in 4 years, and the worldwide system completed in 6 years. The recommendation of TWGI on the high-altitude detection system have as yet not been made a specific U.S. proposal. However, according to U.S. estimates, the Argus satellite system can be put up in 2 years, the far earth satellite system and the solar satellite system in 4 years.

After 6 years the capability of the system will depend upon the countries which participate and the agreed phasing for the high-altitude component.

Based on quite limited and uncertain information, the capability of the system has been estimated for the time periods, 2, 4 and 6 years (assuming all countries participate) after initiation of the system to be as follows:

[Typeset Page 12]

a. Atmospheric Explosions (Surface to 10 km Altitude)

Table 1 summarizes the probability of detecting and identifying explosions of 1 and 5 KT set off in the atmosphere below 10 km altitude in USSR.

Yield 2 years 4 years 6 years
1 KT 5–50% 15%–90% same
5 KT 10–80% 60%–90%
[Facsimile Page 7]

It should be noted that this average capability was estimated from predicted acoustic noise levels at control post locations selected in the USSR primarily on the basis of their being favorable seismic sites. The resulting degradation of performance from colocating the acoustic technique with the seismic results in an overall system capability in the USSR which is only about the same as that of the present unilateral U.S. detection system.

The range of probabilities in Table 1 corresponds to the limits in the capabilities of the system in going from the worst to the best area for detection in the USSR.

For explosions above 10 km the system will have an uncertain chance of obtaining a sample of radioactive debris required for identification. Its capability will depend upon whether sampling aircraft can reach the radioactive cloud.

At the end of six years it is expected that the worldwide system will have a capability similar to that in the USSR in other countries. There will, however, be a lower capability over certain large ocean areas. In these areas there are cases where it will only be possible to detect and identify 20 kiloton nuclear explosions with a probability of 60 to 90 per cent. The capability for detecting explosions occuring between 10 and 30 kilometers in these ocean areas may even be considerably less and identification as nuclear may not be possible.

b. Underwater Explosions

After the 6-year period the control system will be able to detect with high probability, underwater nuclear explosions of 1 KT and greater when set off in the deep ocean. A similar capability (not including certain large ocean areas remote from Phase I installations) will exist after 4 years. The system will have little probability for detecting explosions less than 20 to 100 tons when set off in some regions of the ocean.

Identification of an underwater explosion will require inspection of the site of the event. The explosion site can be located within an area less than 75 square kilometers if three stations observe the direct [Typeset Page 13] hydro-acoustic signals, but the area may be as large as 7500 square kilometers if only reflected signals are detected at only three stations. For explosions conducted so that radioactive debris reaches the surface, inspection will have a high probability of success if undertaken within a few days after the event. Inspection can be made quite difficult and possibly ineffective by conducting the explosion at such a depth that radioactivity does not rise to, or near, the surface or by conducting the explosion under the polar icecap or in remote areas where adverse weather conditions extend the time for initiating inspection.

[Facsimile Page 8]

c. Underground Explosions

There is no known way to identify an underground nuclear explosion by its seismic signals alone. Seismic control stations can only identify some earthquakes. A seismic event which is not identified as an earthquake by control stations could be suspected of being a nuclear explosion. Inspection of the site of the suspected event will be necessary to determine, if possible, the cause.

Table 2 summarizes the capability of the control system to detect and identify underground seismic events in the USSR. The identification capability of the Geneva system in phase IA, (10 stations in 2 years) exceeds that of the present U.S. unilateral system. However, the corresponding detection capability for events in certain unfavorable locations is comparable in phase 1. While identification requires new close-in stations, detection can use data from distant stations. The restrictions imposed by spacing criteria on the locations of the initial stations in the USSR may not permit optimal location in order to obtain minimum background noise. Therefore, the estimates in Table 2 conservatively assumed station noise levels in the USSR about 5 times as great as those at the quietest stations in the U.S. unilateral net. If optimal location can in fact be obtained within the agreed grid, there will be substantial improvement. If the stations are relocated to optimize identification and detection, the improved results in Table 4 are obtained even though noise levels are still assumed to be about 5 times as great as those at the quietest stations in the U.S. unilateral net. If, under these circumstances, sites with background noise comparable to the best stations in the unilateral U.S. net can be found, there will again be a substantial improvement.

If China is not in the control system at the end of 6 years, the number of unidentified events in the USSR will be increased by 10 to 30 per cent.

If the Soviet proposal of 15 stations in the USSR is accepted, the number of unidentified events in the USSR is increased by less than 10 per cent.

After 6 years, if China is in the system, there will be between 26 and 68 unidentified events per year in China above magnitude 4.75. [Typeset Page 14] In the U.S. there will be between 120 and 165 events above magnitude 4.75. However, the numbers for the U.S. may be too high by a factor of 2 if more recent seismicity estimates prove correct.

It should be noted that these estimates are based on the average number of earthquakes per year. The number of earthquakes per year may fluctuate by a factor of two or so from year to year. The [Facsimile Page 9] number of unidentified events can be expected to fluctuate by the same factor.

Identification of an event which is not identified as an earthquake will require inspection of the site of the event. Inspection can only be undertaken if the event is located (detected). Location accuracy may be within an area of about 200 to 500 square kilometers for an interior continental event and within an area greater than 500 square kilometers for a coastal event.

It is presently impossible to determine quantitatively the capability of on-site inspection to identify a clandestine underground test. However, there is a possibility, with imagination and thorough inspection, that such tests can be identified. This possibility of identification depends to a large extent on the persistence of the inspection team. If the team has some reason—possibly evidence either from the seismic network or from intelligence—that will indicate that a test had taken place in the area, it would then have the motivation required to perform a thorough, persistent, and exhaustive investigation, thus substantially increasing chances of identification.

Although through conventional intelligence there will be some chance of being aware of a possible violation, unless the event is detected by the system, inspection of the site and obtaining a radioactive sample as physical proof of a violation will not be possible.

d. High Altitude Explosions (above 30–50 km)

The capability of a control system to detect high-altitude nuclear explosion depends on the competition of the signals from nuclear explosions in space with the background noise signals. The direct radiation signals from a nuclear explosion in space are well understood, but knowledge of background noise signals is based on incomplete experimental information and theoretical considerations. The capability of a high-altitude detection system is limited by the statistical fluctuations of the background noise as well as by possible unexpected short-time noise signals. The occurrence of a detected signal (even when detected on more than one instrument) can only be a strong presumption of a nuclear explosion, rather than positive identification.

The development time for a reliable detection system can be shortened at the expense of system capability.

Table 3 presents recent estimates of the capability of a system for detecting high-yield explosions which might be achieved for various [Typeset Page 15] time periods. The estimates apply to a system whose capability has been degraded over potential capabilities in the interest of system simplification and reliability. Two years have been added to the time estimated for the U.S. unilaterally to achieve the system as an estimate of the effect of internationalization of the system.

[Facsimile Page 10]
2 years 4 years 6 years
Number of Unidentified Seismic Events Above (3)
(a) Magnitude 4.75 (20 KT Rainier Coupling) 70–75 53–70 30–60
(b) Magnitude 4.35 (5 KT Rainier Coupling) 190–195 170–190 130–180
Detection Limit (4)
(a) Equivalent Yield—Rainier Coupling 4–16KT(5) 0.9–3.3KT 6–1.8KT
(b) Equivalent Yield—Decoupled (6) 1200–4800KT 270–1000KT 180–560KT

(1) This table assumes station background noise levels about 5 times as great as those at the best stations in the US unilateral net.

(2) The numbers in this table are from a Rand study based on somewhat different assumptions on relative seismicity of the USSR and on station locations than those used in a similar [text not declassified] study. As a result the numbers are lower by about 25 percent in most cases than the [text not declassified] numbers but still within the fundamental uncertainty of “at least a factor of 2 up or down” which has been agreed between Rand and [text not declassified] to apply to all such numbers.

(3) The range of numbers in the tables expresses the uncertainty in the assumed noise conditions at the control stations and may vary by a factor of at least 2 up or down depending on the year selected.

(4) The detection limit refers to the minimum yield which can be located with 50 per cent probability when placed at the most unfavorable positions relative to the control posts.

(5) This detection limit depends on where stations are installed. In particular, this limit could be considerably improved, perhaps to 1–5KT by locating stations for detection.

(6) The full decoupling factor of 300 was assumed without regard to feasibility and cost of constructing the required cavity.

[Facsimile Page 11] [Typeset Page 16]
2 years 4 years 6 years
Yield 100 KT 1 MT 100 KT 1 KT 100 KT 1 MT
Limit of detectable range—in kilometers Capability dependent results of VELA research program.
a. Unshielded 106 3 × 106 6 × 107 2 × 108
b. Shielded (100–200 lb shield) 105 – 106 106 3 × 106
Capability for Detecting Explosions behind the Moon None Uncertain Possible with solar satellites
Other Limitations Regions between control stations extending up to several hundred kilometers not covered. The 46 Phase I control posts will not provide coverage of high altitude explosions over considerable portions of the earth where no Phase I control posts are scheduled. (1) Regions between control stations extending up to several hundred kilometers covered only by satellites. If solar satellites are included, the required shielding weights are increased to several thousand pounds.
(2) No solar satellites.
[Facsimile Page 12] [Typeset Page 17]

The capabilities expected after two and four years are shown separately in the Table. It is assumed that after two years only the Argus satellite and ground-based instrumentation in the U.S., USSR, and U.K. could be operative in the control system. It is assumed that at the end of four years, the ground stations could be increased in number and the remaining five to six earth satellites put in orbit. Improvement in the system capabilities beyond that predicted for the four-year period depends upon the use of solar satellites and on background measurements which at present have not been carried out. It is not possible to say whether the system will improve or worsen as a result of future information.

e. Anticipated Contributions from Other Intelligence Sources

Intelligence cannot be expected to provide advance warning of the time and precise location of a clandestine Soviet nuclear test. However, it has a strong capability to delineate areas within which such tests may be held and a poor-to-good capability to detect test preparations and to collect information about the event subsequent to its occurrence. Detection of preparations and collection of data on the occurrence of Soviet clandestine nuclear tests will be more difficult than in the past when no legal restraints against testing were in effect. Soviet attempts to evade the treaty would undoubtedly require tests in environments less susceptible to detection, i.e., underground and high-altitude/space.

[text not declassified]

Many intelligence sources have a good capability to detect Soviet space and high vertical missile firings which might serve to alert the treaty system for possible nuclear tests.

Present intelligence capabilities to detect tests in the atmosphere and underwater, particularly with respect to tests distant from the USSR are quite good. The latter situation would deny to the Soviets the advantages of their extremely effective internal security system.

Soviet officials planning an illegal, clandestine test will not know the precise extent and nature of intelligence detection [Facsimile Page 13] capabilities. In view of a wide scope of intelligence detection possibilities and the uncertainty which they interject in Soviet evasion planning, the violation will undoubtedly assume a greater capability of intelligence than it does in fact have.

[text not declassified]

2. Cost of U.S. Proposed Control System

a. Control Post System

Table 4 shows numbers derived by an approximate prorating of the costs of central post. Headquarters and Regional Offices, Airborne Operations and Communication costs into the phasing pattern agreed [Typeset Page 18] at Geneva. While the total installation and operating costs are based on extensive cost studies, the cost of each phase has not been examined in detail and the costs shown should be used only for a general idea of rate of expenditure during installation.

($ in millions)
Hq. & Reg. Off. 16 8 32 24 80 33
Control Posts 260 130 520 390 1299 229
Ships 29 29 37
Air Sampling 36 142 178 35
Communications* 45 23 91 66 225 63
386 161 785 480 1811 402

* Assumes use can be made of existing national communications networks.

b. Satellites Systems

Costs of satellite-based detection systems, derived from TWG I considerations and subsequent studies, are listed in Table 5.

[Facsimile Page 14]
($ in millions)
Initial Costs Annual Operations
ARGUS 14 10
Launch Facilities 100 34
Tracking & Data Acquisition 20 8
234 120

Thus, the total cost of ground-based and satellite-based control equipment would be about 2 billion and the total annual operating cost would be $500 million.

3. Possible Control System Modifications

a. It appears possible by slight, and possible politically acceptable, changes in the control system proposed by the U.S. to improve substan [Typeset Page 19] tially its capability for identifying earthquakes. The changes are, first, the dropping of the requirement on spacing of the control stations (1700 km in aseismic and 1000 km in seismic areas) imposed by the 1958 Conference of Experts; and second, the possible addition of 3 or 4 more stations into the USSR. If the 21 control stations in the USSR are relocated so that most of them are concentrated within the highly seismic areas of Kamchatka, Karafuto and the Pamirs and only a few stations are spread on a wide grid over the USSR for the purpose of locating small seismic events wherever they might occur within the USSR, then it has been estimated that the improved capability shown in Table 6 might be achievable.

Since the distribution of seismicity in the USSR is quite uncertain, further research may reveal a few other highly seismic areas which will also require some stations if the capabilities in Table 6 are to be achieved. This contingency can be met in part by maintaining the right to relocate stations as new data are required. It might also be necessary to increase the number of stations—perhaps by adding another 3 or 4.

[Facsimile Page 15]
Capabilities Relocated Geneva Control System (1) (2)
2 years 4 years 6 years
Number of Unidentified Seismic Events Above (3)
(a) Magnitude 4.75 (20 KT, Rainier coupling) 56–62 14–20 14–18
(b) Magnitude 4.35 (5 KT, Rainier coupling) 160–170 46–60 46–60
Detection Limit
(a) Equivalent Yield—Rainier coupling— 1.4–5.5 KT 1.3–4.8 KT .5–1.4 KT
(b) Equivalent Yield—Decoupled (4) 420–1650 KT 390–1440 KT 150–420 KT

(1) This table assumes station background noise levels about 5 times as great as those at the best stations in the U.S. unilateral net.

(2) See footnote (1) to table II.

(3) The range of numbers in the tables expressed the uncertainty in the noise conditions at the control stations.

(4) The full decoupling factor of 300 was assumed without regard to feasibility and cost of constructing the required cavity.

[Facsimile Page 16]

b. The system capability for detecting high-altitude nuclear explosions would be enhanced by requiring pre-launch inspection of all space vehicles. There are many obvious political difficulties and some [Typeset Page 20] technical difficulties which would have to be overcome. Pre-launch inspection of the space vehicle might require disassembly of the vehicle in order to detect the presence of fissionable material. In addition, in order to assure that all space vehicles launchings were actually announced for inspection, this system would presumably require a system to detect rocket launching unless a unilateral intelligence capacity was adequate and acceptable for this purpose.

III. Estimated Changes in the Control System Capability from Future Research

Expected results from the moratorium research program (Project VELA) are heavily dependent on the form and degree of support at the top levels of Government. The present forecast of results is based on the assumptions:

a. The budget for seismic research under Project VELA is restored to its original level and for high-altitude research the budget is increased to the original level recommended by ARPA.

b. Approval is given to fire the planned underground nuclear explosions presently scheduled under Project VELA.

If either of these assumptions is not met, improvements will be much less likely and, for some aspects of the control problem, impossible.

1. Changes Resulting from the Seismic Research Program

The capability of the system to detect and identify nuclear explosions depends on the following: (1) the number of unidentified earthquakes of a given magnitude; (2) the yield of nuclear explosions which corresponds in amplitude of signal to a particular magnitude; and (3) the probability of success and number of on-site inspections. In this context, the following estimates of the change of the performance of the control system have been made:

a. The seismic research program will remove many of the uncertainties which are presently involved in determining system capability. With reference to the capability described in Table 6 for a Geneva system with station relocation, it is reasonably possible that new seismic information obtained in two to three years will reduce the equivalent yield above which a given number of events will remain unidentified by a factor of about three. Some techniques are already forseen wherein this might be accomplished.

[Facsimile Page 17]

b. Unmanned stations would result in a great deal of improvement in detection and identification of explosions. With a grid spacing of 200 km detection capability is about 10 tons. Identification capability for such a network is probably limited by natural events which are similar to explosions. Assuming 95% limit to identification capability, [Typeset Page 21] this system would leave 60 unidentified events with equivalent yield greater than 150 tons. Location of epicenters would probably be better than 100 square km.

c. Many suggestions have been made for improvement which will be tested during the research program. This includes use of long waves, improved first motion criteria, depth of focus determination and other techniques which are difficult to evaluate at this early date but offer the possibility of major improvements. Experience gained in the operation of the Geneva network can also lead to improvements.

d. Progress in on-site inspection now offers the possibility of using aftershocks, precision epicenters and focal depth determination to localize the suspicious area to better than 50–100 km2 and identify earthquake aftershocks by their greater depth. This improvement depends on the rapidity in reaching the suspicious area.

e. Degradation by improvement of decoupling beyond the factor of 300 is possible. Factors of 2–10 have been estimated for further decoupling. Confusion of the explosion source to simulate earthquakes is a remote possibility.

Table 7 gives rough estimates of improvements (with respect to Table 6) following a 2–3 year research program. Realization of the capability would require subsequent installation of appropriate hardware.

2. Changes Resulting from the High Altitude Research Program

The primary outcome of the high altitude research program will be more complete information on background noise signals which interfere with detection. This information will allow more certain evaluation of the control system. It is not known whether the system capability will be better or worse, but in view of the conservatism of Table 3 it is more likely to be better. For example, ground-based techniques will probably cover completely the low altitude regions. A second outcome of the research program will be some new detection techniques. Two new ones are already promising, namely, the VLF phase shift method and the detection of the radio signal from x-ray shields. Others are still somewhat speculative. Finally, more effective x-ray shields, and possibly other concealment measures, might be found.

[Facsimile Page 18]
[Typeset Page 22]
Reasonably Possible
About 20 control posts, relocated
½ KT
About 20 control posts, relocated Plus ocean bottom seismographs (1) (Yield above which 60 unidentified events)
Augmentation by unattended stations 200 KM grid
10 Tons
Identification with unattended stations (Yield above which 60 unidentified events) 150 Tons (2)
Degradation beyond decoupling factor of 300 2–10
Confusion of source, earthquake triggering, etc. as to misidentify explosion ?

(1) Based upon only one ocean bottom measurement. 1½ KT identification will only be achieved if all installed stations have capabilities comparable to this measurement.

(2) Although theoretically this net would permit identification in excess of 99 per cent, a limitation of 95 per cent identification has been assumed to cover conservatively the case of seismic events which may have characteristics of explosions.

[Facsimile Page 19]


The present and future capabilities of the U.S. and the USSR in the nuclear weapons fields have been summarized as a function of the yields obtainable for different [illegible in the original] weight classes in Tables 8 and 9. In addition the respective capabilities in various specialized types of weapons, i.e., clean or radiation types, are also included. Extrapolations into the future have been made on several different assumptions which are compatible with various situations resulting from a nuclear test agreement, e.g., underground testing below a threshold, space tests, and several types of Soviet evasion.

The past tests of the Soviet Union and U.S. appear in Column I of the U.S. and Soviet tables respectively. Items presently scheduled for stockpile in the U.S. are listed in Column II. In Column III of each table are given the expected improvements possible in the U.S. and in the Soviet Union if testing is prohibited but laboratory experiments (hydronuclear tests) are permitted up to a nuclear energy release of 1 ton. In Column IV are listed the expected advances by each country with unlimited testing and in Column V with testing everywhere [Facsimile Page 20] except in the atmosphere. It is expected that under both these latter situations the potential future weapons development in the two countries would tend to converge to the same final achievements. It should be pointed [Typeset Page 23] out, however, that there could be considerable differences between the U.S. and the USSR in the rate of achievement of those capabilities, with specific times depending on their relative state now (and in some areas of development we have no specific information on the relevant situation for the Soviets), on the effort available, and on the experience with particular kinds of tests involved.

The remaining columns do not correspond for the two countries. For the U.S., Column VI represents achievements which could be made using legal tests which would be below the underground threshold. This is a seismic magnitude of 4.75, which corresponds to 20 KT with Rainier coupling, but represents considerably higher yield for decoupled experiments. It has been assumed that partial decoupling would then allow experiments up to 150 KT to be carried out. In Column VII are listed the developments which could be achieved under a high altitude threshold arrangement, i.e., about 1 KT.

Columns VIII, IX, and X represent what the Soviets would be able to accomplish by clandestine testing, that is, [Facsimile Page 21] with evasion to an extent which would not be detected by the Geneva system. These three columns represent different levels of evasion involving different parts of the Geneva system and different efforts at evasion. Column VIII and IX assumed tests underground up to 1 KT and 5 KT. Column X allows experiments up to 50 KT, which would require decoupling, either partial or full, for the experiments in the range of 10’s of KT and Column XI adds to this the possibility of evasion in outer space with experiments up to 200 KT.

In a situation involving no testing, there is an additional effect (not allowed for in the tables of this summary) deriving from espionage. As time passes and particularly as U.S. items go into production and deployment, some U.S. developments may become known to the USSR and in these cases the Soviet capability would converge beyond [illegible in the original] U.S.

Each column contains the yield, or other features which characterize that class of weapon which one might be able to achieve in a particular weight. Where relevant it also lists the materials (oralloy or reactor product) required. There is also a sub-column having to do with the number of tests required and their yields in order to achieve that capability. A time is given in some cases which represents a compounding (not necessarily addition) of the design time for experiments [Facsimile Page 22] and the analysis of the data before the next experiments, with the time necessary to carry out the experiments, involving the construction of decoupling holes, launch of heavy payloads, etc. The times given correspond to the time at which the weapon design and testing could be completed. Additional time to reach initial stockpile will depend on development procedures and may be anything from two [Typeset Page 24] years down to zero depending on how much the country involved is willing to commit the weaponization procedures before completing development.

The present and currently planned U.S. stockpile weapons can, of course, be established with a high degree of confidence, and in many cases longer range extrapolations of U.S. capabilities are reasonably reliable since these developments are based on already tested principles. Development of weapons in the various weight classes depends in varying degrees on testing, as shown in Table 8 and discussed later in this section. While no tests of [text not declassified] the [text not declassified] research is sufficiently advanced to give confidence that this class of weapon can be developed. [text not declassified] This [Facsimile Page 23] can probably be done by calculation and hydronuclear experiments (maximum yield 1 ton). [text not declassified] The details on the number of tests required and time to achieve these or other advanced weapons are subject to large uncertainties. In this connection, it should be remembered that major advances often come from surprises and therefore do not permit sound forecasting.

In the case of Soviet weapons, the uncertainties are much greater, but at least in the larger yield categories the long-range detection system has permitted reasonably reliable activities of the minimum weights for the yields obtained in Soviet tests. [text not declassified] There is no evidence from any source [text not declassified]. [Facsimile Page 24] Extrapolations into the future are subject to major uncertainties and are perforce based largely on U.S. weapon design principles. For instance, whether the Soviets could in fact reach the high performance indicated in Column X of the Table for [text not declassified] the figures given are based on the assumption that this experience is about the same as that of the U.S. It is known that the Soviets are continuing a vigorous weapons development program; but, although the Soviets could have conducted clandestine tests, intelligence does not support the thesis that the Soviets have in fact violated the moratorium.


[text not declassified]

[Facsimile Page 25] [Facsimile Page 26] [Facsimile Page 27] [Facsimile Page 28] [Facsimile Page 29] [Facsimile Page 30] [Facsimile Page 31] [Facsimile Page 32] [Facsimile Page 33] [Facsimile Page 34] [Facsimile Page 35] [Facsimile Page 36] [Facsimile Page 37]

Surprises: It is by now trite to say that the greatest advances in the course of further nuclear weapons testing and design may come from surprises which by their nature, cannot be predicted. That testing can produce new ideas or invite attention to the significance of certain effects or new techniques is obvious from a history of such occurrences over the past fifteen years. [text not declassified] [Facsimile Page 38] would probably not have been recognized or accepted without full scale testing. Limiting the scale or type testing may slow down or preclude the discovery and exploitation of new ideas. No matter how small the scale of testing, [Typeset Page 25] some developments of usefulness not now foreseen will probably be found and exploited by the tester.

[Facsimile Page 39]

[text not declassified]

[Facsimile Page 40] [Facsimile Page 41] [Facsimile Page 42]



In considering possible “Nth” Country nuclear weapon developments, one must take into account technical and scientific capabilities, motivation, availability of test sites, test environments, and possible delivery vehicles for operational employment of nuclear weapons. These factors are considered in the discussion below for France, Communist China, and “other countries”.

In the following it is assumed that there is no wilfull communication of weapons know-how, or gift of fissile material for weapons purposes, by the nuclear powers to the non-nuclear ones.

In the first place, it should be noted that the principal difficulty in starting a weapons program is in acquisition of fissile material. This might be done either through gifts, its own production, or diversion from its peaceful nuclear power program. [text not declassified]

So far as developing advanced fission weapons is concerned, one can expect that a limitation to underground tests (space testing would presumably not be available to nth countries for a very long time) would not be very severe. Such a limitation might slow down development and make it more expensive. [text not declassified]

[Facsimile Page 43]

It is important to point out that in contrast to the situation at the start of the U.S. effort, there now exists in unclassified form a great deal of information important to the design of nuclear weapons. For example, the size, weight and yield of several U.S. designs have been released. Experimental techniques in hydrodynamics and neutronics have been described [text not declassified]. Neutron cross sections for all relevant materials are available. Finally, the calculational procedures in hydrodynamics, neutronics and radiation transport have been published, and computers to perform these calculations can be purchased.

[text not declassified] It is questionable, but not necessarily impossible, for them to succeed.

Thus, limitation to non-atmospheric testing would [text not declassified].

[Facsimile Page 44]


[text not declassified]

Underground test sites are available in Metropolitan France, and in the Sahara (pending the outcome of the Algerian problem). From the standpoint of the technical capabilities of the various detection [Typeset Page 26] systems, French ability to evade detection would range from poor in Metropolitan France to fair in the Sahara. [text not declassified] In view of the high risk of detection, it is believed unlikely that the French Government would attempt evasion. However, it would probably continue overt testing as long as it is politically feasible.


China has the scientific and technical capability to develop nuclear weapons, light and medium bombers for delivery, and many possible test areas. China also is presumed to have a strong desire for such development.

[text not declassified]

[Facsimile Page 45]

China’s capability for evasion is high, both from a technical and intelligence standpoint [text not declassified]. It is doubtful that China would consider herself bound by a treaty which she did not sign.


There are, of course, many other countries who have an actual or planned reactor capability which could produce plutonium for weapons. The most likely candidates are Israel, the UAR, Sweden, West Germany, and India. None have existing or planned delivery capabilities for large weapons. With the exception of West Germany’s active gas-centrifuge development program, none are planning U–235 facilities and therefore are not in a position to stockpile gun-type weapons for many years.

Israel and the UAR have the strongest motivation for nuclear weapon development. Both are developing reactor facilities—Israel with considerable French help, the UAR with some Soviet and perhaps West German aid. Sweden is waiting the outcome of the Geneva negotiations before deciding to embark on a weapons program, [text not declassified]. India officially deplores nuclear weapons, but a demonstrated Chinese Communist capability may substantially change this position.

All of these countries except India have a poor to fair capability to evade detection from the technical standpoint, while for India it may be fair to good. [text not declassified]

The expense of decoupling would be a major factor in the consideration of all of these countries, and would be particularly important in the case of India and the UAR.

All of these countries, with the possible exception of the UAR, have the technical knowhow to design simple, implosion-type fission weapons which could probably be stockpiled without nuclear testing. These non-tested devices would be very large and heavy and none of these countries presently have a delivery capability for such a device. [Typeset Page 27] While these countries would prefer an actual proof test, they would probably be willing to forego this since the principles are fairly well established by US, USSR, and UK successes. The development of advanced types of fission weapons or TN devices without tests is unlikely by these countries. This may not be a major factor in the security posture of these countries since the availability of a few nominal yield weapons would be the most critical factor.

[Facsimile Page 46]

VI. The Cost of Evasion

Any control systems, such as the ones described in Sections I–III of this Report will have natural limits of detection and identification. In addition, a determined violator can take specific measures which will broaden the range over which tests can be carried out and still escape detection by technical means. Such methods of evasion are:

a) Small underground shots below the detection threshold followed by cover-up operation.

b) Decoupled or partially decoupled underground events using the “large hole”.

c) Tests in space beyond the capabilities of a detection system.

d) Tests in space requiring specially deployed shields to escape detection.

Each of the evasion tactics imposes a penalty on the violator in the terms of one or more of the following: financial cost, stretched out time scale and/or reduced test effectiveness. In addition, there exists the risk of suspicion arising out of conventional intelligence.

Evasion costs have been analyzed only in a preliminary way since by necessity they are based only on studies.

1) Test effectiveness

Underground clandestine operations including decoupled shots would not differ materially from our previous underground experience in the range of diagnostic methods which could be applied.

Space tests restrict the diagnostic tools which can be used but are still adequate for weapons development by measurements of yield and limited diagnostics.

2) Cost—Underground

Table 10 gives the size of cavity required to achieve full decoupling and partial decoupling to the extent indicated. The specific case of a test for 50 KT in a overdriven sphere 400′ in diameter at a depth of 2000′ has been analyzed in some detail under a variety of site conditions. These conditions would give a decoupling factor of about 75.

Holes sufficient to decouple a 3 KT explosion exist. Although larger holes have not been excavated, it is considered possible to construct a [Facsimile Page 47] [Typeset Page 28] 750 feet diameter hole in salt (sufficient to decouple 90 KT). The feasibility of larger cavities is uncertain.

In the case of a hole of sufficient size for complete decoupling, holes are almost certain to be re-usable to some extent. Contamination by the debris prevents re-use in less than six weeks. If radio-chemistry is considered desirable, the total number of shots in a given cavity may be limited. In the case of an overdriven cavity re-usability is less certain and subject to the same constraints as in the fully decoupled case.

In round numbers the cost of a 50 KT test program might be roughly $7,000,000 per shot assuming three shots in a given hole. We might estimate that in the 50 KT neighborhood the decoupling operation increase costs by about a factor of 3–5.

If the yield is sufficiently small that the event would not be inspectable even without decoupling, the need for clandestine operation will increase costs much less, possibly by a factor of 2.

3) Cost-Space tests

Tests in space are in themselves a means of evasion; their difficulty and cost is directly dependent on the type of control they are to evade. With a ground-based control system, the evader, if he is to avoid all possibility of detection, is forced to distances of roughly 106 kilometers, which is a distance not involving large flight times; with the far-earth satellite system, an evader is forced to distances near 108 kilometers where flight times and vehicle requirements are much more substantial. Evasion with the use of shields, which again complicates the problem of evasion, might reduce these distances to 106 kilometers.

Using U.S. cost estimates, the cost to prepare for a space test in the 106 to the 107 kilometer range is roughly $100 million, with a cost of $10–12 million per additional launching. Actual test costs would depend on the reliability obtained, i.e., the number of launches required per successful test and the launch sites required. Costs to conduct one megaton test beyond 108 kilometers depend on detailed test requirements but may not be much larger than the shorter range costs unless waiting times are objectionable.

4) Time delays

Table 10 in the Appendix gives the construction and cover-up time scale for a 50-kiloton decoupled, pre-shot construction program; times [Facsimile Page 48] under the various conditions vary from 2 to 5 years, after the initiation of preparation, assuming U.S. conditions. The delays in small clandestine tests not requiring decoupling are small.

The delays involved in space tests depend partially on vehicle availability and reliability; present USSR vehicles are adequate for space tests under the conditions considered. It has been estimated that

[Typeset Page 29]
Table of yield in kilotons of test shots which can be placed in a cavity of a given size.
Sphere Depth of Burial
Volume Diameter 1000 Ft 2000 Ft 3000 Ft Decoupling Factor
1.9 × 104 cu yds 100’ [text not declassified] [text not declassified] [text not declassified] 300
1.6 × 105 cu yds 200’ [text not declassified] [text not declassified] [text not declassified]
1.3 × 106 cu yds 400’ [text not declassified] [text not declassified] [text not declassified]
4.2 × 106 cu yds 600’ [text not declassified] [text not declassified] [text not declassified]
1.9 × 104 cu yds 100’ [text not declassified] [text not declassified] [text not declassified] 40
1.6 × 105 cu yds 200’ [text not declassified] [text not declassified] [text not declassified]
1.3 × 106 cu yds 400’ [text not declassified] [text not declassified] [text not declassified]
4.2 × 106 cu yds 600’ [text not declassified] [text not declassified] [text not declassified]
[Typeset Page 30]

space tests would stretch a test program out in time by a factor of two after the initial development program which might last 2–4 years, assuming long waiting times for vehicle travel and depending on the complexity of the evasion tactic required.

5) Summary

a. Underground evasion tactics will not restrict test diagnostics. Space tests will restrict diagnostics somewhat but are adequate for weapons development.

b. Cost in yield ranges requiring large-hole decoupling space tests are increased roughly a factor of 3–5. Small clandestine underground tests can be carried out without substantial cost penalty. Space tests increase costs by a factor of 3–10 depending primarily on reliability.

c. Time delays in clandestine tests requiring hole decoupling are 2–5 years. After construction holes might be reusable. Time delay in the space tests stretch a program out by possibly a factor of two after an initial development program of 2–4 years.



A nuclear test ban will place limits on both U.S. and Soviet nuclear weapons systems. An analysis of the significance of these limitations on the relative military positions of the two countries involves assumptions as to the extent of the limitations actually imposed by a ban and the nature of the military problem.

Among the large variety of possible responses to the outcome of the treaty negotiations, we have focussed on the following three:

Case I No test ban Unlimited testing Unlimited testing
Case II Total test ban No testing No testing
Case III Total test ban No testing Maximum evasion possible technically under Geneva System

Note: (The Nth country problem is not considered in this section)

From the strictly military point of view, the most conservative approach to the test ban problem rests on the comparison of the relative positions of the U.S. and the USSR assuming, on the one hand, unlimited testing (Case I) and, on the other hand, a test ban with no further testing by the U.S. but with maximum Soviet evasion (Case III) technically feasible under the Geneva System. Attention has therefore been [Typeset Page 31] focussed on comparing Case I and Case III, realizing that Case III is the maximum possible risk rather than a certain development. It must, of course, be [Facsimile Page 51] recognized that the conduct of tests in a clandestine manner would generally involve political risks and be more expensive and would retard progress both as a result of the physical circumstances required (discussed in Section 6) and the extreme security precautions involved to avoid detection by conventional intelligence. Such factors are difficult to evaluate in quantitative terms.

[Facsimile Page 49] [Facsimile Page 50]

Although most weapons now in U.S. stockpile have not actually been proof tested, they are straightforward extrapolations of tested devices. In a weapons test ban, additional extrapolations will have to be made in the physics to correspond to engineering changes dictated by altering weapon environments and military requirements. Many of the changes of this kind now foreseen can be made with full assurance that the weapons performance will be as predicted. However, if over a period of ten or more years many of the weapons design developments were carried out in the absence of experiments involving nuclear explosions, substantial doubts might arise about weapons performance. With changes in personnel and loss of experience, some of these doubts might also apply to the “older” weapons designed in a period of testing or immediately thereafter, but not tested in exactly their stockpiled configuration. The relative effect of this factor depends on which alternatives are being considered.

The discussion which follows concentrates on the characteristics of specific weapons systems. There also has been an attempt to indicate where some of the unknown possibilities may lie. It must be pointed out, however, that in all weapons technology one of the most important considerations in further developments is a possibility of the appearance of actual surprises. In general, we believe that as far as yield to weight improvement is concerned, surprises are very unlikely in the strategic weapons beyond the developments predicted in Section IV. There could be considerable surprises in weapons effects of various kinds. Finally, in the area of tactical nuclear weapons, where the room for invention is large, the possibility of important surprises is correspondingly great.

The following military areas have been considered: (1) Strategic systems, (2) AICBM, and (3) Tactical Systems.

[Facsimile Page 52]


1. General

The effectiveness of the strategic weapon systems depends not only on weapons design factors such as yield-to-weight ratio, and materials requirements, but also on the delivery system design factors such as accuracy (CEP), reliability and vulnerability to enemy attack as well [Typeset Page 32] as level of intelligence on enemy targets and enemy defence capabilities affecting penetration.

In addition to these technical questions, the problem is complicated by the differences in the impact of a test ban on a deterrence strategy as compared with a counterforce strategy.

It is difficult to evaluate the U.S. requirements for a counterforce strategy for nuclear weapons (even if it is conducted in a preemptive manner) since its effectiveness depends on firm and precise knowledge of the location of a very large proportion of Soviet strategic delivery vehicles. We do not now have this knowledge on Soviet missiles; and, even if increased intelligence capabilities through space or other means improves our knowledge of Soviet targets, the mobility of the Soviet strategic force presumably will also have increased by that time. Also, the U.S. will develop [text not declassified] thus, future Soviet counterforce strategy, if possible at all, must include extremely difficult new methods for determining the location of mobile U.S. systems, or very much larger force levels.

A counterforce strategy would emphasize attack on hard and mobile targets in addition to soft targets such as airfields. For attacks on hard and mobile targets an increase in yield is equivalent to a reduction in CEP or an equivalent increase in the number of weapons delivered on target as given by the following table: (For an area attack on mobile targets whose exact location is not known, such as Polaris, CEP is not important).

[Facsimile Page 53]
Yield increased by a factor Equivalent reduction of CEP Equivalent increase in number of weapons delivered on target
2 20% 1.6
3 30% 2.0
4 40% 2.5
10–20 50%–60% 5–7

It must be emphasized that, unless one is certain to take preemptive action and has accurate intelligence, increases in yield would not permit nearly as great a decrease in over-all force size as implied by the figures since such a reduction in force level would substantially reduce the ability of the system to survive an initial enemy attack and thereby decrease its second strike capability. In fact, if there is a question as to whether a sufficient force will survive an initial enemy attack to provide a deterrent, increases in yield would permit no reduction in force level and an improvement could only be found by increased force levels.

[Typeset Page 33]

In a deterrent strategy, there is a requirement for a minimum number of delivery systems (missiles or aircraft) to survive any enemy attack and penetrate energy defenses. Survival depends on such factors as hardness, readiness, reliability, mobility and secrecy. Decrease in warhead weight at a given yield has contributed to the mobility of deterrent systems. Increased mobility [Facsimile Page 54] is at present of greater value for a deterrent strategy to the U.S. than to the USSR since secrecy of the USSR deterrent force has a similar effect as mobility on the U.S. force. This may decrease in value, however, depending on improved intelligence measures, including the Samos system.

In a deterrent strategy, an increase in the yield of a warhead at a given weight would in principle increase the effects from both blast and fallout against urban areas and industrial complexes more than against hard or mobile targets. However, these relations have significance only in the case of relatively small yields and levels of attack. In the case of blast damage, warheads of present yields delivered with the CEP’s of existing systems would so completely over-kill the population and overdestroy the floor space of urban area targets by blast and fire, that further increases in yields would produce little additional damage. Similarly, probable attack levels during the period in question would result in such extremely high casualty levels from fallout with existing yields that further increases in fallout would produce only small increases in casualties in the surviving population.

The problem of accidental detonation or unauthorized use of nuclear weapons remains a matter of continuing concern. Prevention of such events involves both technical and non-technical problems. Some, but not all, of the technical devices that can substantially improve the degree of control without sacrifice in readiness can be incorporated into weapons without testing.

2. Unlimited Testing (Case I)

In the event that the U.S. and USSR both undertake unlimited testing, both are likely to achieve eventually (though not necessarily at the same time) comparable yield-to-weight ratios in the weight classes in which they are interested. The nuclear technology of both countries is sufficiently advanced that in the 1965–1970 time period, weapon yield [text not declassified].

[Facsimile Page 55]

In the case of a counterforce strategy, the increases in yield possible with unlimited testing would probably have the effect of reducing the counterforce capabilities of both the U.S. and the USSR. While the increase in yield for very large weapons [text not declassified] through testing would tend to improve counterforce capability somewhat against a static enemy force, this would be more than compensated by the increased mobility resulting from the substantially improved yield-to-weight ratio for the USSR in [text not declassified] and from the proba [Typeset Page 34] ble development of more mobile ICBM systems than they now have. However, the Soviets already have a larger degree of invulnerability as a result of their general secrecy. Therefore, the Soviet counterforce problem might be increased in difficulty by an even larger factor by the U.S. development of warheads in [text not declassified]. In addition, some members of the Panel seriously question whether such systems are desirable from the point of view of safety.

In general, improvements in missile technology can be achieved by ordinary engineering measures, but are aided substantially by decreases in warhead weight. [text not declassified]

The effectiveness of a strategic system is also determined by its ability to penetrate enemy defenses. In the case of [Facsimile Page 56] strategic missiles, penetration through enemy AICBM’s is aided by decoys, radar camouflage, maneuvering targets and multiple warheads. The use of multiple warheads which make the AICBM problem even more difficult than it is at the present time is strongly dependent on warhead weights. In addition, the use of a smaller weight warhead in a given system would permit the inclusion of additional decoys. [text not declassified] In this case, the potential of a non-nuclear kill by AICBM would become dominant. In any event, miss distance is not at present the determining parameter in the AICBM problem for the U.S.

Further testing will probably reduce costs to maintain a given strategic posture. The amount of reduction depends greatly on the posture desired and the cost reduction achieved may vary from a negligible to a substantial proportion of system cost.

In summary, continued testing would make a counterforce strategy more difficult through increased mobility and survivability of the second strike force. Continued testing could increase deterrence by adding, through lowered weapon weights, a factor toward achieving survival and penetration. This would be an essential factor to the U.S. only if our deterrence becomes marginal and [text not declassified] becomes the factor which makes the difference between a marginal and non-marginal deterrence. There is disagreement among members of the panel as to the likelihood of such a situation occurring.

3. No Further Tests (Case II)

If nuclear weapons of importance to strategic systems were stockpiled without further nuclear tests but after extensive laboratory tests, it is believed that the U.S. would have some initial advantage over the USSR in the yield of warheads [text not declassified].

[Facsimile Page 57]

As time passes and particularly as U.S. items go into production and deployment, some U.S. developments may become known to the USSR and in these cases the Soviet capability would converge toward that of the U.S.

[Typeset Page 35]

With this spectrum of warheads, which could be stockpiled without test, the U.S. and the USSR would appear to have and to be able to maintain a very strong deterrent strategy by intelligent planning of delivery systems. However, such a deterrent position could in principle be unbalanced in favor of either party by the appropriate combination of difficult developments in ASW Air Defense, AICBM systems, or shortening to 15 minutes or less the time in which intelligence information can be available on the position of a large fraction of the mobile systems.

It is not possible to evaluate the adequacy of a counterforce strategy in terms of these weapons for the inherent reasons discussed above.

4. No Further U.S. Testing and Maximum Soviet Evasion (Case III)

If there is a treaty barring tests and U.S. activities were limited to hydro-nuclear tests while the USSR evaded the treaty to the maximum extent possible by testing up to 50 KT underground with big-hole decoupling and up to 200 KT in outer space, there would still not be any significant developments by either country in weight classes [text not declassified].

[Facsimile Page 58]

In summary, under a test ban obeyed by the U.S. but evaded by the USSR to the maximum extent technically possible under the Geneva system, the preemptive counterforce capability of the U.S. would eventually be about the same as under the condition of unlimited testing. Consequently, the deterrent capability of the USSR would also eventually be about the same under these two conditions. On the other hand, the counterforce capability of the USSR would eventually be improved and the deterrent capability of the U.S. correspondingly degraded as compared with the condition of unlimited testing. The extent and significance of this change depends on how marginal U.S. deterrance is considered to become and how important very small warheads [text not declassified] are considered to be to assure survival of mobile systems.

5. [text not declassified]

[Facsimile Page 59]


Developments of AICBM techniques can also be regarded either as aiding counter forces (in case one intends a first strike and wants to defend against a retaliatory force) or as a pure defense (which depends on having a better AICBM than anyone now conceives of). An AICBM system could perhaps be considered more seriously as a pure defense in a situation where one has an agreed upon and observed limit on the number of missiles (at a rather low number).

Further nuclear tests have a bearing on the AICBM problem in the following areas:

1) Decreased warhead and missile weight and cost.

[Typeset Page 36]

2) Possibilities of increased kill radii [text not declassified] and of using nuclear blast as a means of sorting out decoys.

3) The problem of “blackout” effects from a nuclear burst on AICBM radar and communications.

[text not declassified]

2. The possibility of using nuclear blast as a means of sorting out decoys.

Studies made on the use of nuclear blast in sorting out decoys are not encouraging. There exists the possibility that with additional theoretical [Facsimile Page 60] work and nuclear tests [text not declassified] can be developed, (presumably both by the U.S. and USSR) in a period of four years or more. Tests are such that concealed underground testing is possible in the development phases of the device proper. Speculations concerning the effect predict large kill radii. However, for this improvement to be useful the decoy problem must be solved. [text not declassified]

3. Special effects.

The understanding of how radars, communication system, etc., behave when nuclear explosions take place at high altitudes is an important part of the AICBM problem. In principle, radar difficulties can be gotten around by going to higher frequencies to overcome the effects of blackout. Furthermore, the U.S. presumably knows more about such effects from the experimental point of view than do the Soviets, who are not known to have had any high altitude shots over 50,000 feet.

Some experiments in this area could be carried out clandestinely although with some risk of detection, until a high altitude control system is established, but probably not afterwards.

Beyond the “blackout” problem the main interest in high altitude tests focusses on better understanding of re-entry phenomena, study of the state of ionization of debris in a vacuum, study of magnetic trapping phenomena, etc. Whether such studies will have significant bearing on the AICBM problem is not clear.

4. The USSR has tested a number of warheads suitable for AICBM application. [text not declassified] without tests. Further development could be carried out by clandestine outer-space tests.

5. In summary, the present problems critical to the solution of the AICBM problem are not in the nuclear explosion field. It is unlikely that any good solutions will be found. Improvements in nuclear weapons would become very significant only if new inventions are made which reduce drastically the cost of target acquisitions, target tracking and data handling.


1. Tactical nuclear weapons are defined as ammunition for defensive and offensive systems whose primary purposes are the conduct [Typeset Page 37] of operations, (ranging from very small use of force to large operations), short of a strategic exchange between the primary contestants. The weapons cannot be defined as to yield, size, methods of delivery or effects, but only as to purpose. Tactical nuclear systems can be considered in the role of a “deterrent” strategy to discourage enemy actions (either nuclear or non-nuclear) short of a strategic exchange. [Facsimile Page 61] Alternatively, tactical nuclear weapons can be considered in the role of a “counterforce” strategy for actual use in large or small quantities on either a broad battlefront or in isolated limited engagements. While there exist strong differences of opinion as to whether nuclear weapons can be employed in many cases without escalation into general war, stated basic national policy at present is that main but not sole reliance should be placed on nuclear weapons. No attempt has been made here to judge this issue; however, conclusions depend significantly on the policies adopted.

2. [text not declassified]

In view of the number of possible theaters and areas within a theater where nuclear weapons could be used tactically, there are situations where stockpile limitations could become significant, even aside from competition while other requirements (Air Defense, ASW, and advanced strategic systems) or other possible production limitations.

[text not declassified]

3. It should be noted that the possibilities for further development and inventing of nuclear warheads for tactical weapons systems are substantial. The possibility of important surprises is correspondingly great. Current and proposed warheads are [text not declassified].

4. Unlimited testing.

In the event that the US and the USSR both undertake unlimited testing, both would probably eventually, though not necessarily at the same time, achieve [text not declassified].

[Facsimile Page 62]

With unlimited testing, it should eventually be possible to reduce the requirement [text not declassified].

Such developments would provide the military commander with small, light-weight weapons systems whose warhead produces prompt incapacitation of troops within a well-defined radius of the explosion and without the attendant material damage and residual radiation associated with presently available nuclear warheads.

The rapid fall-off of radiation dose with distance might permit in certain tactical situations elimination of enemy troops with comparative safety to friendly ones, if the position of the latter is known, at separations where blast weapons would make this selectivity impossible. The penetrating nature of the radiation also would allow neutralization of such hard items as tanks or pillboxes. Such strong points could be [Typeset Page 38] attacked by conventional nuclear weapons only by exploding those quite near the target, with risk of fallout and certainty of blast damage over a wide area which could include friendly troops or existing structures.

5. No Further Tests.

If nuclear weapons of importance to tactical systems were stockpiled without further tests, the US would have a wide range of yields, [text not declassified].

[Facsimile Page 63]

By hydronuclear experiment involving energy releases up to about a ton of high explosive equivalent and calculations, the US (and the USSR) could make substantial improvements in [text not declassified].

In some cases it would not be possible to design warheads of optimum size and weight for specific future weapons systems. Planned weapons systems for tactical warfare have not made full use of advanced low-yield warheads which have been designed but have not yet reached stockpile.

It should be noted that changes in requirements and tactics take place only after the deployment or even the use of such weapons in the hands of troops.

6. No Further U.S. Testing and Maximum Soviet Evasion.

If there were a treaty barring tests and U.S. activities were limited to hydronuclear experiments, while [illegible in the original] the [illegible in the original] even to the extent of 1-kiloton, experiments, the USSR could achieve all foreseen nuclear weapon developments in the [illegible in the original] and medium yields (up to 20–50 kilotons) relevant to [illegible in the original] warfare, while U.S. nuclear weapon developments would be limited to those improvements possible with laboratory experiments and calculations. This would mean that the USSR could, over a period of time, achieve the limits in economy of conventional fission weapons and thereby more than compensate for any relative deficiency in the availability of fissile material. However, the US, by that time, could have made large inventory of fissionable material, such of which would be available for tactical weapons.

[Facsimile Page 64]

[text not declassified] In this situation, the US would be at a disadvantage relative to the USSR in the field of tactical nuclear weapons. This disadvantage would be of importance in tactical situations such as the following:

1) Possibility of extensive use of nuclear weapons in multiple engagements requiring the deployment of [text not declassified].

2) The enemy is able to force a situation in which friend and foe are closely “diffused” so that large nuclear warheads cannot be used, but where a large number of small nuclear warheads can be effective.

3) The lethal radius must be sharply defined and material damage is to be avoided.

[Typeset Page 39]

The US would, however, have a large inventory of fissionable material available in the 1965–70 time period. The Soviets would therefore still have to consider the possibility of U.S. response with those nuclear weapons to which Soviet tactics might be vulnerable, as well as the possibility of escalation of nuclear warfare outside the existing geographical boundaries or even into strategic war.

The panel members did not reach a consensus as to how important specific problems of this nature would be in the over-all picture of limited war tactics. In particular, no consensus was reached as to the degree of latitude available to the US to balance its strength and weakness by selection of alternative tactics.

[Facsimile Page 65]



The following changes and additions to Section VII are submitted as representing the views of the undersigned.

Page 3—End of first full paragraph:

Add the following:

For such an attack yield is important since there is a direct relationship between yield and the area effectively covered.

Page 3—End of paragraph following the tabulation:

Add the following:

However, the quantitative increase required would be determined by the effectiveness of the force expected to survive the attack. Thus, if the yield of the surviving weapons had been increased by a factor of four much smaller initial and surviving forces would be required.

Page 4—

Omit the first full paragraph and substitute the following:

A major effect expected of a deterrent strategy is the creation of casualties from fallout which may be measured in terms of total fission yield delivered by surface bursts. In this case the yield of the individual weapon makes little difference as long as the total fission yield is substantially the same and the distribution pattern is sufficient to cover the major population centers. As in the case of the counterforce [Facsimile Page 66] strategy, a certain minimum surviving force is essential. One of the several yardsticks by which the effectiveness (and thus the essential strength) of the surviving force may be measured is the total yield deliverable by that force. If the strike second capability is marginal, individual weapon yield assumes considerable importance whether the deterrent strategy is based primarily on floor space destruction from fire and blast, upon the effects of fallout or both. Unfortunately, without accu [Typeset Page 40] rate intelligence as to the enemy’s first strike capabilities, a measure of the actual margin of safety is impossible. A realistic approach to a greater margin of safety without increased numbers of delivery vehicles, and thus a significant increase in system costs, is an improvement of the effectiveness of each weapon.

Page 5—First full paragraph, 2nd sentence:

Delete the parenthetical phrase, [text not declassified].

Page 6—

The 2nd and 3rd full paragraphs should be combined in the Summary and should read as follows:

Further testing will certainly reduce costs to maintain a given strategic posture in the long term. The amount of reduction depends greatly on the posture desired and the cost reduction achieved may vary from a negligible to a substantial proportion of system costs. Continued testing would make a counterforce strategy more difficult in increased mobility and survivability of the strike second force. Continued testing could increase deterrence by adding, through lowered weapon weights, a factor toward achieving survival and penetration, or through higher weapon yields for a given payload by loading greater [Facsimile Page 67] effectiveness to the surviving force. These factors would become essential if our deterrence becomes marginal. Aside from their contribution to cost reduction for the maintenance of a given strategic posture, these factors assume importance in relation to our ability to accurately appraise the margin of safety of our deterrent system. Accordingly, continued improvement of yield/weight ratios adds a factor of safety to compensate for our lack of intelligence concerning the enemy’s capabilities. On the other hand, if our intelligence as to his capabilities is accurate, these factors assume importance in relation to a counterforce strategy.

Page 7—First full paragraph:

Revise as follows:

[text not declassified]

Page 7—8th line from bottom:

Rewrite the sentence beginning with “Specifically” to read as follows:

[text not declassified]

Page 8—First full paragraph:

Revise to read as follows:

In summary, under a test ban obeyed by the United States but evaded by the USSR to the maximum extent technically possible under [Facsimile Page 68] the Geneva System, the attainment of an effective, preemptive counter [Typeset Page 41] force capability, if at all feasible, would require a much larger U.S. force structure than would be needed under the condition of unlimited testing. The deterrent capability of the USSR would remain about the same under these two conditions since the Soviets could apply their improvements in the lower weight classes to increased mobility and survivability in either case. On the other hand . . . .

Page 9—Last full paragraph:

Revise to read as follows:

[text not declassified] It is particularly pertinent to observe that, although a nuclear stalemate seems to be approaching and is likely to remain for a considerable period, it must not be conceived as a static stalemate. It is essential that all promising avenues of research which might break the stalemate to our advantage, particularly in the AICBM area, should be vigorously pursued. The nation that can develop an effective anti-missile defense, even in the face of countermeasures, will be well on the way to achieving strategic superiority.

Page 11—Second full paragraph:

Add the following phrase:

However, it would not be to the advantage of the national economy to adopt this solution.

[Facsimile Page 69]

Page 13—Second full paragraph:

Omit the first three sentences and substitute the following:

[text not declassified]

Page 13—Final paragraph:

Add the following sentence:

Here, again, a large inventory of materials is not a desirable substitute for economy in their use.

Page 14—Omit the penultimate paragraph and substitute the following:

Assuming a continuation of production at approximately present levels, the United States will have a large inventory of special nuclear materials in the 1965–70 period. The adequacy of these prospective supplies to deal with the threat and potential capabilities of the Sino-Soviet Bloc in the area of tactical warfare, whether local or general, or indeed in any other area, cannot be judged independently of basic national policy with respect to the use of nuclear weapons nor the military force structure maintained as a consequence of that policy. In any case, as regards both strategies and tactical uses, the military value of the available materials may be appreciably enhanced by taking full advantage of weapon technology now available and greatly enhanced by improvements possible with future testing. If a cut-off of production [Typeset Page 42] is negotiated and adequately monitored it may be assumed that both sides will endeavor to make maximum use of the materials available. In this case the advantage will rest with the side employing the most advanced technology.

Analysis of the three general cases considered in this report indicate that under Cases I and II, equivalent technologies may be developed by both sides in due course [text not declassified]. From the strictly military standpoint, therefore, Case III is most advantageous to the USSR and most disadvantageous to the United States.

Herbert B. Loper
Assistant to the Secretary of Defense (Atomic Energy)
  1. Transmits final report of Ad Hoc Panel on Technical Capabilities and Implications of Geneva System. Top Secret. 71 pp. Kennedy Library, National Security Files, Subjects Series, Nuclear, Fisk.