In military and technical terms, we can envisage agreements to limit strategic arms which would be sufficiently verifiable to be enforceable and which would enhance both the security of the United States and the security of the Soviet Union. On-site inspection is no longer the immovable roadblock that it has been in the past. Unilateral means of verification, available to both sides, provide forms of inspection as effective for some purposes as on-the-ground surveys.

It remains true that weapons have certain important characteristics that cannot be limited effectively by agreement without disclosures which go beyond what can be obtained unilaterally. And, without such inspection, there would be a greater possibility that we would have to withdraw from an agreement because of uncertainty about weapons improvements by the other side. But these reservations need not block the negotiation of a useful treaty that would increase our national security.

Strategic arms limitations could produce extra bonuses in addition to improved military security-among them the possibility of increasing the resources available for other purposes, of increasing East-West coöperation and of adopting additional arms-control measures. For the present I will leave to others the discussion of these latter hopes, prospects and advantages, as well as the problems of negotiating arms-limitation agreements, and concentrate on what I see as the most critical objective: increased security in a nuclear world.

Exactly how can we judge whether a particular agreement to limit strategic weapons would contribute to the security of the United States? In general, there are three principal goals to be taken into account in considering any strategic policy:

To deter a deliberate nuclear attack by maintaining military forces that could survive such an attack and still inflict an unacceptable level of damage on the attacker, providing thereby a high degree of assurance that no country would decide to attack.

To have the means to limit casualties to some degree in the event of an "irrational" war, or a limited accidental launching of weapons, or a so- called "third-country" attack-that is, an attack by a nuclear power other than one of the two nuclear superpowers.

To have the power to initiate strategic war while limiting one's own casualties to an "acceptable" level-a potential overwhelming first strike.

At the present time both the United States and the Soviet Union have the capability to deter strategic war, and both also have some small capability to limit casualties in an irrational or accidental war. To say that both can deter, however, is tantamount to saying that neither has the capacity to make an overwhelming first strike.

A carefully designed agreement to limit strategic weapons could help to preserve deterrence on both sides. It could also provide both sides with some ability to limit damage in case of irrational or accidental war. But in order to be acceptable an agreement could not permit one side to achieve the capacity for an overwhelming first strike, since this would be incompatible with the other side's requirement to be able to deter such an attack. Hence the third objective above is inconsistent with an effort to reach agreement on strategic arms control. An agreement should be such as to help ensure that if one side strikes first it will not gain, and preferably will lose, in relative military strength.

The maintenance of such a relationship depends on the weapon characteristics of the force structure on each side. Let us take as an example the use of missiles, each with a single warhead, to attack an equal number of hardened missile sites on the other side. Since missile reliability is less than 100 percent, not every attacking missile would destroy a missile. Such an attack would therefore reduce the number of missiles available to the attacker by more than the number of missiles lost by the defender.

Some insist that deterrence is not enough, and that the United States must have both the capability to deter Soviet attack and, if necessary, to launch an overwhelming first strike against the Soviet Union-that is, a strike which would effectively disarm it. In my judgment, this is not feasible, even if the policy were one we wished to adopt. An overwhelming first strike cannot be attained against a nation which has the technical and economic resources of the Soviet Union and plans its forces intelligently. The Soviets can put forth the necessary effort to prevent our achieving a first-strike capability-and there is no doubt that they would do so.

If we reject the pursuit of a first-strike capability, then an agreement to limit strategic weapons is compatible with our security. Of course, this works both ways. If there is to be an agreement, the Soviets must also be willing to give up the first-strike goal. Our clear intention to prevent its achievement should persuade them.

It follows, then, that our principal criterion for judging an agreement to limit strategic arms should be whether the resulting balance of forces assures us of possessing as much or more certainty of deterrence as we would have in its absence. We should compare the more likely, as well as the worst, possibilities with and without an agreement. The Soviets might make clandestine improvements under the protection of an agreement. But it seems at least as likely that, in the absence of an agreement, they might make a sudden massive effort to tilt the balance in their favor. If the agreement gives the United States a higher deterrent capability as compared with the likely evolution of forces in the absence of agreement, it is a positive gain for American security even in strictly military terms.

An agreement which meets this basic criterion could still provide some limitation of damage in case of accidental or irrational war, and a substantial limitation of damage from attacks by third countries. There will be no real threat from third countries such as Communist China until the mid-1970s, and negotiated limitations could still permit measures-e.g. a modest level of anti-ballistic missile defense-to counter such a threat. In addition, we should weigh any agreement in terms of its effect on factors involved in actually fighting a war if deterrence should fail-the ability of forces to survive, accuracy of delivery, damage assessment, command and control. These factors include the capability and flexibility of surviving forces.


Having concluded that the principal objective of a strategic arms- limitation agreement should be the maintenance of deterrence, we must examine in detail the meaning of deterrence in order to judge the effect of relative weapon capabilities. The success or failure of the U.S. deterrent ultimately depends on the judgment of Soviet leaders-that is to say, on their view of what the world would be like for them and for us after any attack which they might conceivably contemplate. To be realistic, we must assume a serious confrontation, such as a European crisis in which each side considered its vital interests involved. In that case, Soviet leaders might be willing to accept heavy casualties and industrial damage at home, if they thought the United States would cease to exist while the U.S.S.R. could successfully recover and ultimately use its surviving nuclear forces to dominate the European industrial complex.

Deterrence requires the maintenance of retaliatory forces which can destroy any attacker, together with defensive measures to complicate the enemy's attack problem and to protect our missiles from destruction before they can be launched. The would-be attacker must be convinced that his own country will suffer unacceptable damage, but his conception of what is unacceptable will vary with the nature of what may be gained in terms of conquest and power. The Soviets are likely to be thoroughly deterred if they are convinced that, after a Soviet first strike and U.S. retaliation, the military balance would be against them and ratios of surviving population and industry would not be adverse to the United States.

The relation between defensive and offensive systems is of critical importance in determining the probable outcome of strategic war. An attacker aiming at a first strike would coördinate the use of all his weapons to increase the probability of avoiding retaliation. He would try to penetrate our defenses and destroy enough of our weapons so that his own defenses could handle the disorganized remainder of our retaliatory capability.

Unfortunately, a side which engaged in a first strike could man?uvre so as to have the advantage of surprise, and would try to bring offensive and defensive forces to peak effectiveness simultaneously. We therefore cannot afford to rely on a single retaliatory system to preserve deterrence; that would make the first-strike problem too simple for the attacker, since he could concentrate both his offensive and defensive efforts against one type of retaliatory weapon and would have a much better chance of achieving increased effectiveness. To avoid this danger, we maintain a mix of land- based missiles, sea-based missiles and bombers. The Soviets would have to attack each in a different way and would require two completely different defensive systems to deal with our surviving missiles and bombers.

At the present time, the United States has an excellent capability to deter. Our retaliatory forces are so well protected that anyone seeking a first strike would need many more weapons than we maintain. Our measures for survival include warning radars, interceptor aircraft, hardened missile silos, dispersed Polaris submarines and the alert status of our bombers. If the Soviets should attack today, they would expend their weapons without disarming ours. We would be left with relatively stronger forces and our society would suffer less damage than we could inflict in return. The Soviet missile force is not accurate enough nor has it enough re-entry vehicles to attack our land-based missiles effectively; Soviet anti- submarine forces are not good enough to destroy many of our Polaris submarines; and our alert bombers have enough warning to escape destruction on the ground. Soviet air defenses are large, but not effective enough to deal with our alert bombers, and the Soviet's anti-missile system has as yet so few interceptors that even if each one were to kill a U.S. missile the overall effect would be small.

We have striven to avoid any gap in our defenses that would permit even a single element of our forces to be largely destroyed before it could be launched. Even as the Soviets developed a Fractional Orbital Bombardment System (FOBS) that could bypass our older warning systems because it has a low trajectory, we quickly deployed over-the-horizon radars, previously developed, that could detect an FOBS launch and provide adequate warning. As the Soviets have increased the number and effectiveness of their missiles, we have begun development of "harder" silos that can withstand much closer nuclear explosions and an anti-missile system that can be used for local defense of our missile fields, if necessary.

In addition to protecting against destruction of our weapons before launch, we have improved our offensive systems so that our surviving warheads would not be neutralized by improved Soviet defenses. As the Soviets have installed their Anti-Ballistic Missile system (ABM), we have added penetration aids and are developing multiple warheads. As the Soviets have worked on more effective air defenses, we have begun development of the SRAM air-to-surface missile, and have continued preliminary work on the aircraft technology needed for a new heavy bomber and on penetration aids suitable for both current and future bombers.

The United States can continue to be confident of maintaining deterrence in the face of Soviet weapon improvements, but some risk always exists that we may let our deterrent capability decline as a result of not acting quickly enough or wisely. For their part, the Soviets would also have to take continuing measures to protect their forces, for although we deploy no purely first-strike weapons, some of our second-strike improvements would also have possible first-strike capabilities. In such a race, each weapon in our complex mix of weapons would have to be improved to stay ahead of the potential threat.

In particular, the deployment of extensive anti-ballistic missile systems for the defense of cities and of highly accurate multiple warhead systems would at the very least introduce uncertainties resulting from differing estimates of their effectiveness. The decrease in the stability of deterrence that would result is a matter of considerable concern and provides a strong argument for limiting such deployments before they occur.


As is apparent from this analysis of deterrence, we need to take a new road to reach a stable situation. We need an effective agreement to limit strategic weapons. If we assume that the Soviets are, or can be, convinced of the futility of seeking a first-strike capability, then it is worth examining the possibility of strategic arms limitations. An agreement to that effect need not foreclose all improvements. Both effective limitations and certain allowable force improvements can proceed with a common aim of maintaining stability as well as preserving the survivability and effectiveness of deterrent forces.

At the present time, the strategic forces of both sides are roughly comparable in the sense that both can deter attack and neither side can launch an overwhelming first strike. Both sides have an adequate second- strike capability. That is to say, the second strike has superiority over the first strike.

In number of ICBMs, the Soviets are approaching equality with our own forces of 1,054 and, since they have more large missiles, they can deliver a greater total payload. This greater payload could be important if used for multiple warheads, though we believe that we are well ahead in multiple- warhead development. They also have some 700 medium-range missiles capable of attacking our NATO allies and our forces overseas. But we retain the advantage in submarine ballistic missiles, with 656 to the Soviet total of around 100. In heavy bombers we are ahead by some 600 to 150, but they also have some 600 medium bombers that could be used on one-way overflights. However, even with Soviet medium bombers included, we retain the advantage in terms of bomber payload. Neither side is yet deploying an extensive ballistic missile defense and neither can defend its cities against an all- out bomber attack with nuclear weapons.

An agreement to limit strategic forces would have to continue to preserve mutual deterrence. It would have to provide for force structures that would make it impractical for either side to approach a first-strike capability, either by altering forces covered by the agreement or by increasing forces not so covered. Ideally, each improvement in forces which might increase the first-strike threat would be prevented by some sort of arms-control agreement. Of course, some limitations will be more important than others, and some will be more feasible because they can be verified without arranging intensive inspection.

Perhaps the most destabilizing threat to mutual deterrence would be the deployment of a large Soviet ABM system. It now seems likely that an ABM system could prove quite effective in the absence of extensive countermeasures by the offensive forces of the other side. ABM weapons have been designed with long range and a large kill radius. Depending on radar coverage, a single ABM site might be able to defend a circular area 800 miles in diameter against small numbers of the present generation of missiles. An ABM system would have even greater effectiveness in support of a first strike, since the retaliatory missiles, already reduced in number, might not achieve the degree of coordination needed to saturate the system.

Though we have already developed penetration aids for our missiles, the creation of an extensive Soviet ABM system would require us to deploy larger numbers of re-entry vehicles than presently programmed in order to make certain that enough of our missiles would get through. We could accomplish this by building more land-based or sea-based missiles, putting larger missiles in present silos, subdividing payloads into even smaller re- entry vehicles, or by various combinations of these measures.

In view of the explosive sort of interaction between a large ABM system and the missile payload required for deterrence, an agreement limiting the ABM defense of cities would make a major contribution to preserving deterrence. Offensive capabilities are so far ahead of the defense that a limit of 100 or even 1000 ABMs would help preserve this advantage. Yet a limited ABM system would still be of value in dealing with any accidental or third- country launch of a small number of missiles.

A related problem is the possibility of numerous Multiple Individually Targeted Re-entry Vehicles (MIRVs). Accurate MIRVs would tend to lessen the chance of land-based missiles surviving an attack, since one first-strike missile with accurate MIRVs could destroy several deterrent missiles in their silos. Defensive measures that could be taken to preserve deterrence in the face of MIRVs would be further hardening of silos or local ABM defense of silos. Harder missile silos would require the attacker to expend several warheads to destroy one missile in its silo, thus helping to counterbalance the effect of MIRV. And local ABM defense of silos would also increase the number of warheads needed by the attacker. Missile defense could be designed so as to protect only silos or population centers as well (though this clearly creates a problem in connection with the extent of the ABM limitation). Such a distinction could be worked out in principle; negotiating it would be another matter.

Clearly an agreement to ban the deployment of MIRVs would be desirable in order to forestall erosion of the capability to deter. But the agreement should allow multiple re-entry vehicles that are not capable of being individually guided. Since these lack the required accuracy to destroy the opponent's missile silos, they would not threaten his deterrent-but they would be essential for penetrating an expanded ABM system and would thus contribute to deterrence. Unfortunately, a check on deployment of different types of payload clearly requires on-site inspection of a very obtrusive kind.

Nevertheless, it is possible that even without on-site inspection we can tell enough about each other's missiles to obtain reasonable assurance. This is so because the probable number of warheads per missile is proportionate to the payload of that missile, and payload, in turn, is directly related to the gross volume, which we may be able to determine unilaterally. Thus, a ceiling on numbers and sizes of missiles could also limit MIRVs to a number less than that needed for an effective first strike, and yet permit enough re-entry vehicles to penetrate missile defenses-as required for deterrence. The size of the missile force and its general characteristics can probably be monitored satisfactorily without on- site inspection.

It is clear, then, that a limit on missile payload would require a carefully related agreement on ABM systems, if deterrence is to be preserved. Similarly, if defenses against the bomber could be frozen or slowed up by agreement, bomber forces could also be limited. However, the latter agreement is likely to be very complex.


The foregoing considerations suggest a number of possible limitations which would enhance our security. In my view, the essence of a useful agreement would be a fixed limit on both ABMs and the number and size of offensive missiles.

The exact number of ABMs that could be permitted would depend on the importance given to limiting damage in situations other than an all-out nuclear exchange. An ABM limit of 100 would protect against small accidental strikes, but might not be adequate against, for instance, the ICBM forces which Communist China is expected to have available in the late 1970s. An ABM limit of some 1000 weapons would be much more satisfactory from the viewpoint of defense against third-country attacks, but that size might begin to threaten the stability of U.S.-Soviet deterrence, since it might give the attacker the capacity to eliminate most of the retaliatory weapons that would survive a first strike employing improved MIRV technology. Thus, perhaps a figure between 100 and 1000 weapons should be considered.

An agreement setting an ABM limit for city defense (as opposed to missile- site defense) in the neighborhood of several hundred launchers, and a limit on numbers of offensive missiles at levels currently deployed or under construction, would ensure that, at least until the middle 1970s, each side would have enough surviving missiles to penetrate the opponent's ABM system in a second strike. It would also be necessary to ban completely any mobile land-based missiles, since they would make it easier to conceal illicit increases in a missile force. A subsequent agreement should then be worked out to limit the size of missiles and thus prevent the unlimited growth of the MIRV threat through development of larger and larger missiles.

Our capability to deter would be more certain if there were such an agreement; without one, the uncertainties multiply. The larger the Soviet missile force, the larger will be the probable error in our estimate of the Soviet MIRV threat. Similarly, the larger the Soviet ABM system, the more uncertainties we will have about its effectiveness. And the combination of an uncontrolled MIRV threat and a massive ABM system would work together to produce a geometric increase in our uncertainty. Thus, an agreement to limit these weapons would increase our confidence in deterrence and would therefore be in our interest in strictly military terms-even apart from conserving resources.

It is imperative, however, that we avoid looking on a strategic arms- limitation agreement as a panacea for our national security problem. It may help to maintain deterrence, but many dangers would remain. War could still occur by escalation from lower levels of conflict or as a result of accidental or irrational actions. New technology and presently inconceivable weapon systems may be devised to threaten such security as is provided by deterrence. Further, many serious problems present themselves in implementing an agreement effectively.

In considering the type of agreement which I judge to be preferable, two specific problems of this sort will have to be explored in considerable technical detail. Whether they can be solved in the negotiating process is an open question. One is control of the MIRV threat through inspection or through limiting missile payload (already discussed briefly). A second is that of a possible upgrading of air-defense systems which have or could be given some anti-ballistic missile capability.

An example of the latter problem concerns the Soviet "Tallinn" air-defense system. If the extensive Tallinn system were given the missile and radar capability to perform as an ABM defense, we would be seriously concerned, regardless of whether our forces were limited by an agreement. Our intelligence should provide warning of major missile or radar improvements, and we would have to take action to ensure that our missiles could penetrate the new Soviet defenses. This would probably mean the end of any agreement limiting our offensive forces. Therefore, in order to attain a lasting treaty we might need an understanding on exactly what sort of improvements would or would not be allowed. We might also require some form of relatively simple on-site inspection to ensure compliance.

In the area of other possible weapon improvements, we must be sure that the agreement does not prohibit activities necessary to maintain deterrence. An agreement will reduce the danger of significant improvements by the other side, but will not eliminate it entirely. All deterrence systems tend to obsolesce. Thus, the increasing accuracy and miniaturization of missile warheads reduce the survivability of missiles in silos; improved anti- submarine warfare techniques have the same effect on Polaris; improved FOBS or submarine-launched missiles lessen the effectiveness of the bomber; and new ABM or anti-aircraft techniques might seriously affect weapons penetration-all without an increase in the quantity of weapons.

Treaty limitations should be designed to control the first-strike threat, but they should not inhibit the countermeasures which must be taken against gradual improvements in the other side's forces or against illicit improvements contrary to the agreement. The problem is complicated by the fine line between first-strike and second-strike capabilities. MIRVs provide both, while multiple warheads that are not individually guided add chiefly to second-strike capacity. Extensive ABM weapons for defense of populations contribute to the first, while ABM weapons to defend silos contribute only to the second; increased missile pay-loads provide both, while the increased hardening of silos is useful chiefly to protect a second-strike capability. Provisions for point defense of missiles and increased hardening of silos would not contribute to first-strike capabilities, but would allow us to preserve deterrence in the face of qualitative improvements in MIRV and ABM systems.

A properly designed agreement to limit strategic forces can better insure the security of both the United States and the Soviet Union. Our task in negotiating such an agreement should be eased by the growing realization on both sides that a first-strike capability will not be permitted to develop regardless of the effort and treasure expended for that purpose.

The peaceful needs of society are too great to waste the world's wealth on a useless competition in strategic weapons. But the outlook for ending this competition depends on the choices, attitudes and actions of both the Soviet Union and ourselves-not on us alone.

You are reading a free article.

Subscribe to Foreign Affairs to get unlimited access.

  • Paywall-free reading of new articles and a century of archives
  • Unlock access to iOS/Android apps to save editions for offline reading
  • Six issues a year in print, online, and audio editions
Subscribe Now