120. Memorandum From the President’s Deputy Special Assistant for National Security Affairs (Kaysen) to President Kennedy0
SUBJECT
- Atmospheric Test Ban Treaty
This paper examines the merits of a U.S. initiative in proposing a treaty to the USSR and others banning nuclear tests in the atmosphere, as the immediate step in our “supreme effort to break the log jam on disarmament and nuclear tests” promised in the State of the Union message.1
- 1.
- If we propose the treaty, and the Soviets reject it, we will be in a better public position when we begin our proposed Spring test series. This alone is not a sufficient reason for proposing it; although it is much more likely than not that the Soviets will refuse our offer.
- 2.
-
If the Soviets do sign such a treaty, we will have achieved a politically significant pause in the arms race. In the first instance we can expect [Page 298] the UK to join us, and France and Communist China not to. While nuclear tests are not the most critical element in our strategic posture, they are the most highly visible and emotionally charged aspect of the arms race. Thus, the value of any first step involving them cannot be judged merely by its technical military significance, but must take into account its symbolic value. The pause might naturally last three or four years; after which increasing attention to development of kinds of new weapons not dependent on atmospheric testing and problems posed by nations outside the treaty—France and Communist China—will erode the usefulness of the treaty as an arms control device. The prospect that it can be extended to cover these nations is poor. Nonetheless, a pause of several years’ duration is of high importance: a first step is necessary, and this might be an achievable one. The period of the pause would create both the opportunity and the incentive to take a second step. For these reasons the proposed treaty should run for a definite term, say, four years, with provision for review. Review could take into account whether other countries with nuclear capability have acceded to the treaty, and the achievement of other further steps in arms control.
The present moment offers what is almost certainly the last opportunity to achieve such a pause in this Presidential term. Our Spring tests in the atmosphere will certainly provoke Soviet response. Further, they clearly will not, in themselves, provide satisfactory answers to many of the most interesting questions to which they are directed, and thus must be viewed not as a self-contained series of tests but as the first step in what will be a continuing and expanding program of testing in the atmosphere, especially taking into account Soviet responses. The next opportunity for a pause will not arise till we and the Soviets have each completed at least two test series, a period of three or four years.
- 3.
- Are there any military risks in signing such a treaty? This depends on our assumptions about Soviet behavior during and after the term of the treaty, and the goals of our own military policy. It is clear that we can continue to keep a secure deterrent, capable of surviving the largest Soviet strike that is possible in this period, and thereafter inflicting such damage on the Soviets that they would not knowingly initiate general nuclear war as a rational policy choice. It will also be possible to maintain some war-fighting capability beyond this; to what extent cannot be so simply stated and will be discussed in (6) below. In (4) the chief points bearing on the security of our deterrent are examined.
- 4.
-
Our confidence in the deterrent power of our strategic striking force rests on our estimate of its capacity to achieve what we confidently consider an unacceptable level of damage to Soviet society, in the most unfavorable circumstances which we should take into account in our planning. Now, and for at least the next five years, our present and proposed deployments of aircraft and missiles promise to achieve the [Page 299] required capability, unless the Soviets (1) deploy a much larger missile force than our present intelligence estimates lead us to expect and (2) are capable of a coordinated attack on a large number of targets simultaneously, with no strategic warning and less than 15 minutes tactical warning, and (3) achieve and deploy an AICBM capability far superior to any we now foresee. It is only the conjunction of these three capabilities which would permit the Soviet Union to feel secure from retaliation by the alert-bomber force, which can be launched on 15 minutes warning, the Minuteman force, and submarine-launched Polaris missiles. The argument is further strengthened if account is taken of our bombers on airborne alert, naval strike forces and tactical forces in Europe.
The growth of the Soviet missile force is not directly related to the presence or absence of an atmospheric test ban; neither is its capacity to achieve a highly-compressed, zero-warning attack. (To be sure, there may be an interaction between Soviet force goals and Soviet technical plans, such that their acceptance of an atmospheric test-ban treaty might lead them to revise their missile deployment program, but this possibility does not affect the argument.) The possibility of achieving an effective and reliable AICBM capability based on present concepts is associated with testing in the atmosphere, and it is this which must be examined first. In the worst case, let us assume that the Soviets sign the treaty, and prepare for atmospheric testing, so that they can begin a large series of tests at the end of two or three years—having indeed signed the treaty with this end in view. We on our side will be in the state of maintaining such readiness for atmospheric testing as we can in circumstances in which it is not clear that we will ever conduct such tests again. In these circumstances we can maintain some reserve capacity, consisting of test facilities and support facilities on Christmas Island, and a specially outfitted group of test ships, and such scientific readiness as can result from the existence of laboratories and development teams engaged in underground and possibly space testing. The support facilities would involve costs on the order of an initial $30 million in capital investment and then $15 million per year of operating costs.
By the end of the period the attention of the best people in the laboratories would certainly have turned away from questions which can be answered only by atmospheric testing. To the extent that there were problems of this category which were not at all related to the questions involved in underground testing, some desirable people would have drifted away from the laboratories. On the other hand, the problems we are proposing to examine in the Spring series of tests can be further studied, and instrumentation for these problems can be brought to a higher state of development. All this can, so to speak, be put on the shelf as a reserve against the resumption of testing in the atmosphere. The extent to which still further theoretical work on problems and instrumentation [Page 300] could be maintained is not clear. Presumably the best people would find it less interesting than working on ongoing underground testing. On the other hand, the underground tests and laboratory work would continue to suggest new questions to be answered by atmospheric tests.
We would be prepared to perform atmospheric tests on some 60 to 90 days’ notice. This notice might be provided by Russian testing or, in a less likely case, by some other event which we would interpret as a violation of the treaty. The first series of tests we would be prepared to perform would be directed in part toward warming up interest in the problems, and recruiting personnel, but it would also produce immediately useful knowledge. This series would probably be more fruitful than the presently proposed Spring series both because there had been more time for thought on the problems and preparation of instrumentation and because the continuing underground testing and laboratory work would provide more ideas to be tested. However, it is equally clear that these first tests would not operate at maximum efficiency and it might take as much as a year after atmospheric testing was resumed to achieve a really good posture by the process of rebuilding interest and recruiting additional personnel not then engaged in underground testing and associated weapons development.
How significant is this gap? Assuming that both sides have used the interval to do whatever can be done without nuclear testing in the atmosphere in developing guidance systems, radars, and missiles, as well as in exploring quite different concepts of AICBM systems and that neither has made any significant breakthrough in this way (which is what we expect, and the only situation relevant to our discussion) the gap is not militarily threatening. The most likely results of the atmospheric testing efforts will be to reveal further the weaknesses of AICBM systems. But if this is not the case, the time from a test revealing significant new information about AICBM to the achievement of a large-scale deployment of weapons will be of the order of five years. Many of the steps in this sequence will be such as to give us notice, from the first nuclear tests themselves to the systems-testing of weapons and the initial phases of deployment. In this period we would have sufficient time to ready our own responses to the new developments. What these responses might be would, of course, depend in part on what we learned about Soviet developments and from our own resumed tests. In the interim, however, we could continue to explore the possibilities of weapons which would not be vulnerable to AICBM’s, such as low-trajectory aerodynamic re-entry vehicles for Minuteman and Polaris, and the proposed SLAM supersonic, low-level ramjet, and similar devices. Further, and with more certainly predictable success, we can now intensify our programs to modify existing missiles to make them less vulnerable to AICBM systems of the general Nike-Zeus type. This can be done on the basis of our present [Page 301] knowledge and of new knowledge from underground testing and laboratory experiments, and does not depend on atmospheric tests. Finally, if we become aware of what appear to be substantial deployments of Soviet AICBM’s, we can build up our own missile force at a faster rate. The economic exchange ratio in such a case would favor us by a wide margin; it would be quicker and cheaper for us to add to our offensive strength than for them to add to their defensive strength. The same point can be made in relation to an increased build-up of Soviet missile forces over that which we now anticipate. In the first case it might be mainly further hardened Minutemen and advanced Titans that we would deploy; in the second, mainly further Polaris submarines. The essential point in this argument is that significant changes in Soviet weaponry and deployment take time, reveal themselves to us in the course of their occurrence, and allow us time to react: the margin of deterrence cannot disappear in six months or a year, much less overnight.
The discussion above started from the assumption that the Soviets do what is most damaging to our capacity to learn from atmospheric testing, namely, wait essentially the whole of the two or three years and then test. A shorter life for the treaty would not change the argument significantly. A much shorter life would probably result in a smaller degradation in the capacity of our weapons establishment to produce significant tests in the first new series.
- 5.
- This discussion has not attempted to explore what the next step in arms control might be. The proposed treaty would give us the space to explore this step, and it would be quite important to begin preparations for that exploration now. As a first step, an atmospheric test-ban has the advantages and disadvantages of being chiefly political. It avoids the difficult problems of inspection and control. A natural next step might be the present Geneva test-ban treaty or some revised version of it. However, it is possible that the very justification put forward for accepting the proposed ban on atmospheric testing would argue against an extension of the ban to other environments. It might be that a more acceptable next step should be a much bigger effort, going either in the direction of an attempt to limit development of new nuclear weapons altogether or tackling the problem of limiting fissionable-material production, or controlling inventories of delivery vehicles. The difficulty of predicting the next step is no reason to refuse to take the first one now, unless we never intend to take it, since we will always face this difficulty.
- 6.
-
The extension of the foregoing argument to the problem of assuring that under an atmospheric test-ban treaty we can maintain that something extra, more than survivable deterrence, which our currently planned strategic force is supposed to provide presents no new issues of principle, but it does raise a difficult quantitative problem. The purpose of maintaining such a margin in the event of nuclear war is to be able to [Page 302] prevent Soviet follow-on attacks after a first strike, and still maintain a reserve capacity for deterring yet further attacks. The highest figure that could be set for this margin is an ability to strike all identifiable targets in the relevant category. However, neither the proportion which is ideally desirable nor a rationale that would determine it has as yet been defined.
Our previous arguments apply; the possible Soviet developments in AICBM which could diminish the capacity of our force to strike at the Soviets could be countered in the same ways. What is not clear is what would be the magnitude of fluctuations in the extra margin above deterrence which we might have to accept, since the exact times of Soviet developments and subsequent deployments, and of our reactions thereto, are unpredictable. We must recognize that the extra margin might fluctuate somewhat in the period in question under the proposed treaty. However, the same thing can be said in its absence. It is in the nature of this margin that it is never known to even a fair degree of precision at any time. One way to define it is by the proportion of the total of Soviet offensive targets—missile sites, long-range bomber fields, and the like—we can attack after absorbing a first strike and still maintain a defined reserve capacity to attack cities or other targets. Our estimate of this magnitude depends, of course, on our knowledge of the Soviet striking force. It seems that inevitably this knowledge is subject to a substantial margin of uncertainty, and further, that there are elements of the force, such as missile-firing submarines or air-alert aircraft which it is difficult for us to account for and impossible for us to target. Even the maximum possible figures could be maintained over time by a sufficient increase in our own forces, although of course the scale and expense of so doing might rapidly climb to astronomical heights. But it is clear that it cannot be asserted that this margin can be maintained at its maximum level continuously, whether or not there is a test-ban treaty. What can be asserted is that, under the adverse circumstances discussed above—namely, a treaty which expired after four years with no further arms control measures—the period of our uncertainty as to how near the maximum margin we were might be greater. Whether we should tolerate this additional uncertainty depends on the relative value we place on the political benefits of an atmospheric test-ban treaty on the one hand, including its effect on international tensions, and the unknown possible loss of counterforce capability in the light of whatever chance there may be that we in fact may be forced to general nuclear war during this period on the other.
- 7.
- A “full first strike” counterforce capacity can never be any more than our ability to hit all the offensive targets we know. In what sense this is “full” is open to debate since we never know just how ignorant we are. It is clear that the argument in paragraph (6) above applies to this case.
- 8.
- On the other hand, there are advantages to us if the Russians sign and abide by such a treaty. Indeed, it is precisely because of these advantages that the Soviets might not sign. The U.S. is far ahead of the Soviet Union in the capacity to test underground. The Russians have had little experience here, and it would be both costly and time-consuming for them to develop suitable facilities. This might well slow down their testing progress for the equivalent of a year in relation to our own efforts. This is no more or no less significant on their side than a corresponding lag would be on ours.
- 9.
- In sum, on the one side are the opportunities to move on arms control in a striking and dramatic fashion. The probability of success is not high, but the value of success would be substantial. On the other, the military effects of a signed treaty are such as appear risky only in terms of the proposition that anything short of the maximum effort of which we are capable in building up our strategic forces, is dangerous. This view makes any step toward arms control or disarmament impossible; and, logically, it would require that we do much more than we are now doing. The great difficulty that has attended past discussions of arms control measures within our own government arises from the desire to secure assurance of an exact balance between the impact of the proposed measures on ourselves and our opponents. The whole thrust of the preceding argument has been that this is an unnecessary goal. In the framework of a strategy of deterrence, quite large fluctuations in specific elements of relative strength need not alter the fundamental military equilibrium. Even in the framework of a strategy of deterrence plus controlled counterforce, moderate fluctuations in effective strength are tolerable unless we in effect assume that general nuclear war is both inevitable and imminent. If we consider general war an unlikely event and, further, think that its probability is affected by the level of international tension, on which arms control measures, among others, may have an effect, we need not seek an unattainable precision of balance in the effect of any arms control measure.