Foreign Relations of the United States, 1961–1963, Volume VII, Arms Control and Disarmament

109. Letter From the Director of Defense Research and Engineering (Brown) to President Kennedy⁰

Washington, December 12, 1961.

Dear Mr. President: In response to your request, I am submitting my personal views and comments on the nuclear weapons test shots proposed for inclusion in atmospheric test program during the Spring of 1962. I have taken as my point of departure the lists included in the letter to you of November 29, 1961, written by the Chairman of the Atomic Energy Commission.¹ I participated in the generation of that statement as a member of the National Security Council Committee on Atmospheric Testing Policy, and I endorse in general the tests proposed in that letter. However, some of the tests are more important than others. I believe that test series plans should be laid out on a time scale determined by only those experiments which meet the following criteria: (a) they can be fully justified on the basis of real (though not necessarily immediate) military importance, and (b) there would be very great difficulty in performing them in other environments. The decision on whether or not to proceed should be made on the basis of these tests only.

If an affirmative decision is made, it is likely that there will be additional useful tests arising either before or during the test series which should be included providing that (a) they do not lengthen or delay the schedule, and (b) they do not substantially increase the total fission yield of the series. What follows is my own evaluation with respect to these criteria of the weapons effects tests proposed by the Department of Defense, my estimate of the military importance of the experiments which are suggested by the Atomic Energy Commission for the purpose of weapons development, and my comments on the utility of atmospheric nuclear explosions as part of weapons systems tests.

1. The nuclear weapons effects program which the Department of Defense has developed involves high altitude, surface, and underwater tests. Not all of the important high altitude effects tests can be done in the proposed series, nor can the best possible surface (electromagnetic pulse and ground-shock) effects tests. However, I consider the high-altitude effects and surface effects experiments which can be done in the proposed series to be of such importance that I believe that resumed U.S. atmospheric testing is justified on the basis of these tests considered alone.

2. The effects of nuclear weapons exploded at very high altitudes has been the subject of much calculation, and a few experiments, but little knowledge. An understanding of the nature and extent of these effects is equally applicable to the problems of defending against Soviet ballistic missiles and of U.S. penetration through Soviet AICBM defense. Whatever the relative difficulty of these two jobs (and I have said several times that I regard the penetration problem as the more feasible one) the information about such effects is of great importance to the design of weapons systems for either purpose.

The relevant effects at high altitudes are of two kinds. One is the effect of a nuclear explosion on an incoming ICBM. To understand this problem completely one would have to know the detailed design of the incoming re-entry vehicle, its warhead, and the yield and efficiency of the AICBM warhead. Since the construction of the Soviet warheads in re-entry vehicles will not be known to us (and the details of construction in ours may not be very well known to them) even a precise knowledge of the effects will not permit the defender to be sure of the lethal range of his AICBM warhead against an incoming warhead. Furthermore, the U.S. at least will be unlikely to be sure of the Soviet AICBM yield or warhead characteristics and so even with effects tests will not be able to design its penetration program with an exact knowledge of the distance at which a Soviet AICBM explosion can “kill” our incoming ICBMs.

However, there are defensive systems in which it makes a great deal of difference whether the kill radius is about one kilometer or about ten kilometers. For this reason it may well be necessary for us to “harden” our re-entry vehicles to insure that their high vulnerability does not make the Soviet AICBM problem much easier. Certain general characteristics of vulnerability can be explored and major causes of vulnerability eliminated understanding of the lethal radii is obtainable readily and completely only from nuclear weapons tests performed at high altitudes in the atmosphere, particularly when one considers that there may be some phenomena not predicted on a theoretical basis which could produce a very large radius of vulnerability. Some of this data could be obtained by underground tests, but there are many phenomena (such as the motion of the nuclear fireball and its interaction with the earth’s magnetic field) of which account can be taken only by tests in the atmosphere.

The other important kind of effect of nuclear weapons explosions at high altitude is the “blackout” effect in which radio communications and (of more interest for the AICBM and ICBM penetration problems) radars are prevented from functioning properly. In particular it is almost certainly possible within some limits to prevent an acquisition radar from seeing an incoming re-entry vehicle for some period of time by the detonation of a nuclear weapon at an altitude of about 400 kilometers. It is also possible, within limits, to prevent the discrimination radar (which distinguishes between re-entry vehicles and decoys) from being able to locate correctly these objects by exploding a nuclear weapon at lower altitudes (50-80 kilometers). Something is already known about these phenomena, their dependence on radar frequency, altitude of burst, etc. from the tests which were carried out in 1958. However, they depend in great detail on these things and on the characteristics of the warhead. There are various causes of the ionization which produces this blackout effect, including the fireball itself, and X-rays, gamma rays and delayed fission products from the explosion. The blackout due to the fission products and the fireball itself, which may be most important from the point of view of influencing the radar for long periods of time or most completely for a short time, are matters about which we are in almost complete ignorance. We can fully understand them only with high-altitude tests. The 1.45 MT explosion at 400 kilometers and 165 KT explosion at 50 kilometers have been selected as the two which will give the most information within the limited time available for preparation, out of a longer list of such tests which are needed. The cut-off date of July 1 makes it infeasible to carry out more than two such experiments.

3. Another experiment in the effects category is a surface explosion of one KT to tell us whether the electromagnetic signal induced by the explosion can burn out command links in varied ICBM sites, and to give us an experimental point in calculating whether our hardened sites will survive nearly multi-megaton explosions. Because of the many billions of dollars we will have spent on hardened missile sites by the mid-1960’s, experimental verification of their survivability (and consequent usability as a second-strike capability) becomes extremely important. This is so despite our confidence in calculations which indicate that they can survive the earth shock from a large nuclear explosion and that the control circuits can survive the electromagnetic signal from such an explosion.

The electromagnetic signal simply cannot be tested except with a surface explosion. An experiment at one or two kilotons can answer unresolved questions about the source and strength of the signal and assure that the arming circuits will not be burned out by an enemy attack. The situation with respect to earth shock is more complicated. One cannot obtain complete confidence except from a full-scale test. This is infeasible in the proposed series since it would require a multi-megaton surface explosion in a geological conformation similar to that present in the areas of location of our proposed missile bases (a coral atoll will not do for this purpose). Furthermore, one can make a very small start on the problem by carrying out experiments completely underground. However, even a small (one or two KT) surface explosion will allow a normalization of calculations. As a result we would depend on a very much smaller extrapolation than we now must in judging the effect of a surface nuclear explosion on hardened structures.

All this argues quite strongly, in my view, for the importance of the proposed Nevada surface test. Additionally, on the same test experiments could be carried out on the effects on a nearby nuclear warhead more conveniently than those particular measurements could be carried out on some of the high altitude tests.

4. The anti-submarine weapons effects test, under water, would give useful information on whether the sound detection gear would operate after use of nuclear weapons for ASW purposes. Though less important than the high altitude and surface shots, it should be included if the proposed test series is carried out.

5. The list contains a large number (up to 7) of experiments designed to produce [1-1/2 lines of source text not declassified]. Some such a number of experiments will be required because the area is one that has not been explored in previous developments and will undoubtedly prove difficult, interesting, and rewarding. One would expect as a result of these tests to be able to increase the yield available at these weights by a factor of [less than 1 line of source text not declassified] or alternatively to reduce the weights presently required to give these yields by a factor of [less than 1 line of source text not declassified]. This would allow us to [less than 1 line of source text not declassified] over the presently planned cluster of three to be carried by the Polaris A-3. Instead half the present weight of that re-entry vehicle could be allotted to penetration aids. The same arrangements could be made with each to increase the penetrability. A [less than 1 line of source text not declassified] would be required for the mobile MRBM now under consideration.

One could also use these warheads (two or three times lighter than what would otherwise be available) to produce correspondingly lighter missiles for air launching—[less than 1 line of source text not declassified]. This could allow such missiles to be launched from lighter aircraft. Alternatively, the lower warhead weight could reduce the weight of a [less than 1 line of source text not declassified] allowing them to be launched from aircraft or small trucks.

I believe there will also be applications in which bombers carrying rather small payloads might by use of these warheads be able to deliver twice or three times as many bombs as would otherwise be possible.

Not all of these are desirable weapons systems. They are mentioned to show what additional flexibility in systems design in the mid-1960’s is allowed by warhead development in this weight class.

[less than 1 line of source text not declassified] can be carried out underground, but it would cost considerably more in time (several years) and money to develop such warheads without atmospheric tests.

6. In the [less than 1 line of source text not declassified] category the list contains a number of development-proof tests. This is the weight of the current Minuteman and Polaris A-1 and A-2 warheads. These tests would allow the weapons design laboratories to establish the correctness or otherwise of their current design calculations on weapons which have been certified for stockpile without exact knowledge of their yield. I believe that still further gains can be made in this weight class, but not without the normalizations to experiment which these tests would provide. I believe that these tests are justified by the future progress to which they will lead. To carry out the normalization of calculations on the burning of thermonuclear fuel, they must be done at full yield, which precludes them from being done underground in the next few years.

7. [less than 1 line of source text not declassified] three experiments are proposed which would about double the available yield at these weights. Alternatively they can be viewed as about [less than 1 line of source text not declassified] presently required for those yields. They would allow us to double the present yields of our Atlas and Titan I re-entry vehicles, or to devote half of that weight to penetration aids, or to use multiple warheads (two of the present yield, or four of half the present yield). Changes of the kind indicated are unlikely to make the difference between successful and unsuccessful deterrence unless the deterrence is marginal to begin with. At the same time the factor of two in yields or weights is not negligible in military calculations, and further tests would produce larger factors.

8. A number of tests of a more experimental or exploratory nature are also included in the proposed list, to be carried out in various weight classes. For example, there is one experiment which is aimed at reducing the weight required for a [less than 1 line of source text not declassified]. This would mean a reduction in weight at that yield of about [less than 1 line of source text not declassified] below what is presently planned for the stockpile. Following the inclusion of such an atmospheric test in the proposed series, it might be possible to complete the development with underground tests.

Another of the proposed ideas, which is in a considerably earlier stage of development, would involve in the actual experiment a very large and heavy box to examine the effects of a different kind of implosion. This could ultimately lead to a yield of [less than 1 line of source text not declassified]. Here again, additional underground and possibly atmospheric tests might be required to complete the development. If such development is successful, one could replace the currently planned cluster warhead of the Polaris A-3 by two warheads of [1 line of source text not declassified]. This would be a very much more effective system, considerably surer to be able to penetrate possible defenses than one using [less than 1 line of source text not declassified] (as presently planned) for the following reasons. If the warheads of an entering ICBM are of low yield, and known beforehand to be of low yield (the difficulty of knowing this would in any event, complicate the defender’s problem) intercept can be delayed until they have reached a rather low altitude. This allows considerably more sorting in the atmosphere by measurement of the different rates of slowing down, and makes the AICBM problem substantially easier. By increasing the yield of such warheads as we have in the Polaris A-3 we would have a noticeable effect in increasing the penetrability.

A third experiment of an exploratory kind is one involving the use of fissile material in the [1-1/2 lines of source text not declassified]. There have already been a number of not very successful experiments along this line. If success could be achieved it would allow a general increase in efficiency and correspondingly of the yield per unit weight throughout the spectrum of warhead weights from a [less than 1 line of source text not declassified].

Experiments of this kind, although they may not produce an immediate change in the characteristics of the nuclear stockpile, are precisely the kind of work which in the long run can lead to large factors of improvement. I believe therefore that they should be included in the proposed series; in fact I believe that they are the most important of all the weapons development experiments suggested, with the possible exception of the [less than 1 line of source text not declassified].

9. A number of tests have been suggested which have yields of less than [less than 1 line of source text not declassified]. These are the [less than 1 line of source text not declassified]. Though they are interesting and useful, I believe that they can be done adequately underground. I would therefore not recommend their inclusion in the basic atmospheric series, although if they could be sandwiched in after the series is laid out they would be worth doing. Since additional experiments in the series inevitably will make it more complicated and difficult to do, I advocate the exclusion of these experiments. I also advocate the exclusion of the suggested experiment [less than 1 line of source text not declassified] which would save about [less than 1 line of source text not declassified]. In my view a 10 percent change in weight does not justify inclusion in the series of this kind.

10. To summarize the effect of the weapons development tests (sections 5-9) on our military posture, one can expect from this test series a decrease in weight at a given yield ranging from [less than 1 line of source text not declassified] the larger numbers corresponding to smaller yields. Over a longer period of time, perhaps as a result of subsequent underground tests and perhaps requiring further atmospheric tests, one could get factors of [less than 1 line of source text not declassified]. This would result in greater penetrability of missiles through the use of larger yields and the allocation of more weight for penetration aids, and better survivability of missiles through the greater mobility of their launching systems. It would also allow bombers of limited payload to have a greater effect. For example, if strike-reconnaissance bombers can be developed their effectiveness will depend to some degree on the number of bombs of a given yield which they can carry on a single sortie.

How important these advantages are is a matter of individual judgment. The most likely military situation, in the period at which most of the weapons resulting from these tests would become available, is that the U.S. deterrent posture will be maintained by the variety and numbers of our delivery systems. At the same time a first-strike capability which would prevent unacceptable retaliatory damage to the United States will be very difficult no matter what we are willing to spend or do (including atmospheric nuclear testing). This argues that failing to reduce warhead weights by factors of two or even five is not likely to make deterrence infeasible instead of feasible, nor will systems made available by these factors of improvement by themselves make pre-emptive attack feasible instead of infeasible.

There is, however, a broad spectrum of intermediate situations between pure minimal deterrence and a full first-strike capability, and almost inevitably that is the situation in which the U.S. will find itself. If a war ensues under these conditions, the limitation of U.S. civilian damage resulting from Soviet follow-on attacks or as a result of spill-over from Soviet counter-force attacks will depend to a considerable extent on the details of the survivability, penetrability and deliverability of our own counter-force attack. This statement is not intended to gloss over the fact that in a nuclear war civilian casualties would be enormous in any event and that the societies of the countries involved would be catastrophically and perhaps irreparably damaged. There remains the additional fact that damage will depend in detail on how much explosive power is delivered, and at what stage of the conflict, on the enemy’s nuclear delivery capability. These factors depend upon the quality of the nuclear weapons as well as of the delivery systems on each side. This is turn cannot help but affect the calculations made by each side in determining its own estimate of its military strength and its resulting political behavior.²

One could probably not justify resumption of U.S. atmospheric tests on the basis of any one or two of the development experiments proposed for this series. However, the entire set of development tests, and the disadvantages which would probably result from a growing lag in U.S. nuclear weapons technology in the middle and late 60’s if we do not carry out atmospheric tests, argue strongly for, and I believe justify, resumption.

11. The argument in favor of the systems tests (Polaris and Atlas) is as follows:

There is no real doubt that, if a missile is tested and works, with experiments being done on a dummy warhead in the missile (perhaps including a chemical but not a nuclear explosion), and if the warhead is separately tested and works, the system will function as a whole.

However, the confidence obtained in operational forces by virtue of having carried out tests with actual nuclear weapons in actual systems is of very great importance. It is notable that the Soviets in their recent test series have included a large number of air drops of thermonuclear weapons involving several aircraft, apparently designed to provide operational training for at least a fraction of their force, and have also carried out several explosions of large nuclear weapons in missiles fired operationally. Though by themselves such a set of tests would not justify the resumption of atmospheric nuclear testing, I would strongly urge that if the proposed test series is carried out these two tests be included. They will neither lengthen nor delay the proposed test series, since they would be carried out by quite different groups from those carrying out the other experiments.

12. I have omitted in the body of this paper discussion of political considerations for and against resumption of atmospheric nuclear testing as being outside my area of responsibility. In order to make a recommendation, however, I must include my personal and undoubtedly oversimplified estimate of these factors. The long-term political issue of greatest significance is the arms race, and the need to limit it so as to improve the security of all nations. To the extent that this aim is advanced by refraining from atmospheric tests, resumption is undesirable. To what extent is the aim of arms limitation so advanced?

Atmospheric testing is not open to “cheating.” However, the Soviets have not offered to sign a treaty banning atmospheric tests, and even if they did it is not clear what would keep them from breaking it at their convenience. Thus we are faced with a unilateral and unreciprocated exclusion of the U.S. from atmospheric nuclear testing.

The unilateral limitation imposed by not resuming U.S. atmospheric tests is unlikely to prove fatal. But it will allow a clear shift in the military balance in favor of the Soviet Union, and there is in high-altitude nuclear effects a small chance of decisive military advantage. To close off or seriously hamper one line of military development unilaterally while other areas of military technology go forward seriously limits U.S. flexibility in design of our military systems. Seeing no counterbalancing political gains, and on the basis of the military-technical considerations given in earlier sections, I recommend that the series comprising items 2-3 and 5-8 above be carried out. The tests mentioned in paragraphs 4 and 11 should be included only if they do not lengthen the series.

Respectfully yours,

Harold Brown

0. Source: Kennedy Library, National Security Files, Subjects Series, Nuclear Weapons Tests, 12/7-18/61. Secret; Restricted Data. Another copy is in Washington National Records Center, RG 330, OSD Files: FRC 65 A 3464, Atomic 400.112 (8 Apr 61).
1. Document 102.
2. Kaysen commented on the preceding two paragraphs in a December 27 memorandum, arguing that what the United States would give up by postponing testing in the atmosphere was “a reduction of an unknown but not very large amount in the size of the margin which our strategic striking force will offer over the minimum survivable force which provides deterrence. Since we are in some doubt as to how big this margin ought to be, the military significance of a reduction in it is small.” (Kennedy Library, National Security Files, Subjects Series, Nuclear Weapons Tests, 12/21/61-1/8/62) In a revised version, dated January 5, 1962, Kaysen added that a second consequence of refraining from atmospheric tests would be “a slowdown in the growth of our knowledge on the problems of designing AICBM systems, which we expect can only reach a moderate degree of effectiveness in the most favorable circumstances.” (Ibid.) See the Supplement.