An LGM-25C Titan II missile is launched at Vandenberg Air Force Base, California (USA), in 1975.
U.S. Department of Defense, National Museum of the U.S. Air Force

IN THE past, a nation's security was guaranteed essentially by its military power. Though proposals for disarmament were made from time to time, particularly after World War I, there was an air of unreality about them. In the era of what we now call conventional weapons, the force-in-being was not nearly so significant as the industrial potential and the mobilization base. A country planning to attack had to engage in extended preparations very difficult to hide. Technology was relatively stable. Travel was unrestricted. Since surprise was not so crucial, and since victory could generally be achieved only through a prolonged mobilization of resources after a war had started, the contribution which arms control might make to stability seemed marginal. Armaments, it was then correctly said, were the symptom and not the cause of tension. The best method of achieving a stable peace was to remove the causes of political conflict.

Although this remains true today, physical conditions have basically altered. Technology is volatile. The advantage of surprise can be overwhelming. The forces-in-being are almost surely decisive--at least in all-out war. A major cause of instability is the very rate of technological change. Every country lives with the nightmare that even if it puts forth its best efforts its survival may be jeopardized by a technological breakthrough on the part of its opponent. It knows also that every invention opens up the prospect of many others. No country can protect itself against all the technological possibilities increasingly open to its opponents. Conversely, an advantage once achieved will produce a powerful incentive to exploit it, for the scientific revolution which made it possible also insures that it will be transitory.

The fear of a momentary weakness is compounded by the dangers of being surprised. As long as the retaliatory forces are composed primarily of liquid-fuel missiles and airplanes--as they will be until the middle sixties--the side which strikes first will have a perhaps decisive advantage unless the defender's retaliatory force is in a state of high readiness. But when two retaliatory forces so constituted confront each other, their very structure may contribute to instability regardless of the intentions of the two sides. Even if those intentions are peaceful, each side must seek to protect itself against catastrophe by increasing the readiness and security of its retaliatory force. Yet such measures, taken in what is conceived to be self-defense, may be indistinguishable to the other side from a decision to launch a surprise attack. For a force which requires no extended preparation before retaliating is also capable of striking without warning. There will consequently always exist a powerful incentive to anticipate this eventuality by launching a preëmptive attack, or at least to obtain the greatest degree of warning possible by any means available. Our aerial reconnaissance of the Soviet Union is the direct consequence of the nature of the retaliatory forces. Since the penalty for being surprised can be national catastrophe, the need to protect the retaliatory force may over-ride all other considerations.

After mobile, solid-fuel missiles become operational in the 1960s, vulnerability will be substantially reduced and the need for extreme readiness correspondingly lessened. Nevertheless, invulnerability is a relative term which depends on numbers, accuracy and defensive capabilities. In a volatile technology it will always be precarious. Both sides should therefore have an interest in stabilizing the arms race as much as possible. All countries should be concerned with preventing a war which might break out simply because of the automatism of the retaliatory forces. At the very least, they should strive to make certain that war, if it does start, is the result of a deliberate decision and is not produced because the opponents, in taking measures which they deem to be defensive, push the other side into an attack in self-defense. If the two sides cannot give expression to their desire to achieve this situation by proposing serious concrete programs, the future of negotiations is dim indeed.


The most obvious solution to the problem of surprise attack is to eliminate retaliatory forces and nuclear weapons altogether. But the seeming simplicity of such proposals is deceptive. To be effective, an arms control scheme must have a built-in incentive to observe it. The participants must not be able to achieve a decisive advantage through evasion or feel that they will put themselves at the mercy of the opponent if they observe the agreement but the other side cheats. Complete elimination of stockpiles of weapons and of retaliatory forces would have a built-in incentive for evasion. In case of complete nuclear disarmament even 50 hidden weapons would confer a perhaps overwhelming superiority. Since no conceivable inspection system could insure against such a possibility, both sides would be almost forced from the very beginning to hold back a part of their stockpile for fear that the opponent might do so. To observe the agreement might magnify a country's sense of insecurity. To evade it would be demoralizing and wreck all prospects for further progress in the field of arms control.

After a decade and a half of the growth of nuclear stockpiles and nearly a decade of the development of missiles, simple remedies can no longer work. The goal of responsible arms-control measures must be to determine, free of sentimentality, not how to eliminate retaliatory forces but how to maintain an equilibrium between them. It is more worthwhile--at least for the immediate future--to seek to reduce the incentive to attack than the capability for it.

What measures can reduce the incentive to make a surprise attack? An all-out nuclear attack is likely to result from one of two motives. An aggressor may feel sufficiently confident that a sudden attack could reduce the counterblow to acceptable proportions. Or a threatened country may feel so vulnerable that it seeks to lessen the danger through a preëmptive attack. Thus a control system will add to stability if it complicates the calculations of the attacker and facilitates those of the defender. Or, put another way, the objective should be to increase the uncertainty about the possibility of success in the mind of the aggressor and to diminish the vulnerability of the defender. To the degree that the aggressor feels uncertain either about the chances of success or, what amounts to the same thing, the likelihood of escaping unacceptable damage, deterrence will be increased. Stability will be enhanced if the defender does not feel so vulnerable that he feels obliged to launch his retaliatory force at the first ambiguous warning.

It follows that both sides can take unilateral measures designed to reduce vulnerability and thereby increase stability. It is essential to recognize, however, that unilateral actions, whatever their intent, may be interpreted as threatening by the opponent. Thus whenever a choice exists, it is desirable to opt for a measure whose defensive characteristics are easily identified. For example, the invulnerability of a retaliatory force is undoubtedly increased by multiplying the number of missiles and dispersing them. Even if each individual weapon is highly vulnerable, an aggressor now requires a more substantial force for success. This, coupled with the difficulty of coordinating a larger attack, complicates his calculations and thus adds to deterrence. At the same time, as the defending retaliatory force grows, it not only becomes more invulnerable but more threatening as well. The response of the other side may be to launch a preëmptive blow. Or more likely, the result would be a spiraling arms race. In either case, the result is to increase instability.

On the other hand, if we seek to achieve invulnerability not through numbers but through mobility or hardening of our retaliatory force, deterrence will be improved without adding to our offensive threat. Such a step can be considered a unilateral contribution to arms control. It enhances our ability to strike back; it does not add to our capacity for surprise attack. The Soviet Union loses only in the ability to destroy our retaliatory force; no additional threat is posed to its security. Here, then, is a unilateral measure which meets the two requirements for arms control outlined earlier: it complicates the calculations of the aggressor and it eases those of the defender. It magnifies the uncertainty of success of the attacker while increasing the security of the side which concedes the first blow.

While a secure retaliatory force is basic to any arms control scheme, invulnerability is a relative term. It depends on numbers, dispersal, mobility, hardening and similar factors. These relationships are so complicated that to seek to protect the retaliatory force solely through unilateral measures is almost certain to produce an arms race. If the goal is stability, negotiated arms-control schemes must therefore accompany unilateral measures. Their objective should be to define a stable equilibrium between the opposing retaliatory forces and then to devise a control system which protects both sides against violations. This raises the question of the significance of numerical limitation.

The foregoing discussion about nuclear disarmament has revealed the paradoxical fact that there is a certain safety in numbers. And this is true even if both sides religiously observe an agreement to limit nuclear weapons. Instability is greater if each side possesses ten missiles than if the equilibrium is stabilized at, say, 500. For an attack which is 90 percent successful when the opponent has ten missiles leaves him one--or a number hardly likely to inflict unacceptable damage. An attack of similar effectiveness when the opponent possesses 500 missiles leaves 50--perhaps sufficient to pose an unacceptable risk. And of course it is technically more complicated to destroy such a large number. Reduction of numbers is thus not an infallible remedy. A very small retaliatory force may increase the danger of war.

This leads to the conclusion that stability is greatest when numbers are sufficiently large to complicate the calculations of the aggressor and to provide a minimum incentive for evasion but not so substantial that they defeat control. The efficacy of a control scheme thus depends on the answer to two queries: What advantage will the side violating the agreement gain through its first violation? How difficult is the system to inspect for violations? If the number of permissible long-range missiles is set at zero--if, in other words, both sides agreed to destroy all I.C.B.M.'s and nuclear weapons--even a small evasion, say ten hidden missiles, will confer a decisive advantage. And such an evasion is almost impossible to discover. If the number is set very low, say at ten, an additional 15 may make a surprise attack possible. In such circumstances, there would be a dual incentive for evasion: fear of the opponent's evasion and the temptation to deal with the security problem once and for all by launching a surprise attack.

On the other hand, if the number is set relatively high, say at 500, even a fairly substantial violation would not confer a decisive advantage. In that case, 50 additional long-range missiles would not enable the violator to launch a surprise attack. A decisive advantage can be obtained only by so many weapons that the risk of detection is likely to appear excessive.

These considerations can be summed up as follows:

(a) The primary goal of any arms control scheme must be to increase stability. A precondition is that both sides should strive to develop invulnerable retaliatory forces.

(b) Since invulnerability is a relative term, the purpose of arms control measures must be to strengthen so far as possible the relative position of the defender, either by enhancing the security of his force or by complicating the calculations of the aggressor.

(c) The equilibrium should be established at a level at which a strategic advantage can be gained only by such a substantial violation that likelihood of its going undetected is very small. The goal should be a stable balance of forces such that deterrence will not fail even if the agreement is upset. A violation, in short, should start an arms race and not a war.

(d) In addition to negotiated schemes, or even in their absence, both sides can take unilateral steps which promote stability. Wherever possible they should seek to assure the invulnerability of their retaliatory force by measures which are manifestly defensive.


As originally conceived, inspection against surprise attack was intended to provide tactical warning. It was believed that a sudden blow required such extensive preparations that they could not be hidden. Planes would have to be serviced and perhaps moved to forward bases. Warning radar would be placed on increased alert. A control system on the ground coupled with periodic aerial inspection could not fail to detect these efforts. We would then be able to protect our retaliatory force by making it airborne or increasing its readiness by other measures. Once the advantage of surprise was removed and the possibility of destroying our retaliatory forces eliminated there would remain no incentive for aggression.

Whatever might have been the significance of obtaining tactical warning when airplanes were the backbone of the retaliatory force, the quest for it is extremely dubious in the missile age. The greater the readiness of the retaliatory force the more difficult it is to determine whether there is an intention to use it. And instant readiness is a special characteristic of missile forces. Even the present generation of liquid-fuel missiles is reported to require a countdown of less than an hour. When solid-fuel missiles become operational, the interval between the command and the actual firing will be a matter of minutes. Then, perfect inspection perfectly communicated could add about ten minutes to the warning time.

If the security of a retaliatory force depended on a warning of ten minutes the situation would be unstable regardless of the effectiveness of the inspection system. The narrow margin of survival would make it prey to all kinds of ambiguous alarms and increase the risk of its becoming the catalyst of accidental war. Indeed, a properly designed missile force should not have to rely on tactical warning at all. It is safe to launch airplanes on the basis of an unconfirmed warning because a relatively long time-interval is available to determine the accuracy of the information and in that period the planes can be recalled. Missiles, on the other hand, are in transit such a short time and the decision to launch them is so irrevocable that they must be designed to ride out an attack.

Inspection in order to obtain tactical warning may also prove inconsistent with the security of the retaliatory forces. Because of the special characteristics of missiles, the only sure method for obtaining tactical warning against a missile attack is to establish continuous surveillance of the launching sites. But continuous surveillance of the retaliatory forces may help a potential aggressor more than the defender, thus violating one of the cardinal principles of arms control. The defender learns only what he already knows: the instant readiness of the aggressor's force. At best he gains an additional warning time which is so short that his retaliatory force cannot possibly be designed to make use of it.

The aggressor, on the other hand, gains vital strategic information. He will learn the exact location of every missile at every moment--thus nullifying to a considerable extent whatever advantage his opponent may have achieved through mobility. He will know precisely the pattern of operation of the retaliatory force he is planning to destroy. The conclusion is inescapable that inspection to obtain tactical warning may detract from stability rather than add to it.

The Soviet Union's adamant refusal to permit inspection of its retaliatory forces is understandable. The Kremlin undoubtedly has a reasonably accurate notion of where our retaliatory forces are based. We, on the other hand, probably do not know the precise location of many Soviet bases. In these circumstances, a system of constant surveillance may appear to the Soviet leaders as a form of unilateral disarmament, which would improve our capacity for surprise attack without increasing the security of the Soviet Union.

But even from our point of view there is a serious question whether constant surveillance of the retaliatory forces is desirable. The practical difficulties of inspection are complicated by the sheer mass of information that must be reported and the speed with which it must be transmitted and analyzed. Moreover, the moment when surveillance is most needed is also the time when the opponent has most incentive to interrupt it.

The problem of adequate sanctions is more difficult still. If the penalty for suspected evasion is very rapid and drastic, the inspection system may contribute to the tenseness of relationships; its instabilities will in effect be added to those inherent in the existence of retaliatory forces. On the other hand, if the reaction is slow or weak, no benefit has been derived from the additional warning time.

A country which becomes aware of what it believes to be an imminent attack has four broad choices:

(a) It can await the blow and gear its retaliation to the scale of attack. This can be done only if its retaliatory force is invulnerable.

(b) It can increase the readiness of its own retaliatory force in the hope that, having lost the element of surprise, the aggressor will desist from his course.

(c) It can give the opponent an ultimatum demanding an end to his preparations and a return to a more peaceful posture--perhaps feeding information into the inspection system on its side to reinforce the threat.

(d) It can launch a preëmptive blow.

Of these options the safest course is the one which is least dependent on tactical warning: that of developing a retaliatory force so well protected that it can await the blow. All other measures are likely to make the situation more tense. Increasing one's readiness--whatever the significance of this phrase may be in the missile age--may deter; but it may also cause the opponent to accelerate his preparations and even to transform what may have been merely a threat of surprise attack into a preëmptive blow. In the missile age, an ultimatum is almost bound to bring on what it seeks to avoid. The recipient--unless his forces are highly invulnerable--cannot possibly run the risk of waiting to see whether the threat is serious. And he may be sufficiently doubtful about his ability to demonstrate compliance with the demand to consider a preëmptive attack the only safe course. This raises the question whether a country would not have a powerful incentive not to recognize hostile preparations. In certain circumstances, indeed, the best contribution to stability may be to seem not to notice the information supplied by the inspection system.

Even if war is avoided, a pattern of diplomacy may well develop which uses the inspection system for purposes of blackmail. When missile forces are in a high state of readiness it is difficult to support a verbal threat with concrete actions that are convincing. But with an inspection system, overt preparations could be demonstrated to the opponent. Inspection can then become a device for blackmail. The blackmailer can take steps he is certain will be detected in order to bring pressure on his opponents, particularly on those which are known to be most reluctant to risk all-out war.

Even if the threatened country does not succumb to the menace, the potential aggressor can use the inspection system to obtain vital strategic information. Whether the threatened country ignores his blackmail or makes counter-preparations, the aggressor will learn the likely pattern of its response. In short, inspection in order to obtain tactical warning is either illusory or dangerous.


Though inspection for the purpose of providing tactical warning does not seem promising and may be dangerous, surveillance could be of great significance in stabilizing the opposing retaliatory forces. This raises two questions: How can inspection reveal the size of a weapons system which is mobile? How can it be reconciled with the unilateral measures to promote invulnerability described earlier?

The answer is that inspection to determine numbers can and should be different from inspection designed to obtain tactical warning. The difficulty with constant surveillance is that it reveals not only the number of missiles but their locations. It contributes to stability by supplying information about the size of the retaliatory force. But it increases vulnerability by revealing the location of the most sensitive targets.

It therefore becomes important to distinguish two kinds of inspection. Most negotiations heretofore have addressed themselves to the problem of inspection in situations in which the opponent is assumed to be interested in hiding something. But a control scheme designed to produce what may be called "negative evidence"--the absence of violation--suffers from a built-in factor of uncertainty. In the nature of things it can never be clear whether the absence of a violation is due to the integrity of the signatories or their skill in evasion. The greater the distrust which has produced the demand for inspection in the first place, the less reassured the parties may be by a control scheme. Lack of evidence of evasion may not seem proof of compliance so much as a symptom of the inadequacy of the inspection system.

It is different, however, with inspection which seeks to produce "positive evidence"--that is, not what each side may want to hide but what it is eager for its opponent to know. In certain circumstances both sides may desperately wish to make certain that the opponent understands they are not preparing an all-out blow. If accidental war is to be avoided, there must be means by which the nuclear powers are able to inform each other rapidly and convincingly that an ambiguous action was not intended to be the prelude to a surprise attack. In the extremenly unlikely event that one of our bombers crashed on a training mission and its hydrogen bomb exploded, it would be vital to have some means to convince the Soviet leaders rapidly that a genuine accident had occurred. The same would be true if a nuclear-powered rocket designed to place a satellite into orbit should malfunction and land on the opponent's territory. Similarly, limited wars--even those not sought by any of the major powers--may threaten suddenly to boil over into a final showdown. In these situations, everything may depend on the ability to reassure the opponent clearly and convincingly.[i]

In the contemporary turmoil, many situations are conceivable which could flare up into hostilities not sought by any of the major powers and yet threatening to embroil them--as, for example the revolution in Iraq. It becomes important to provide safeguards to prevent a spread of local incidents into a general conflagration. All-out war must not occur simply because the nuclear powers have not considered how to back away from the precipice.

This suggests that the West and the Communist countries may have a common interest in setting up a control system which will enable them to exchange and verify information, particularly in periods of crisis. A minimum requirement is for a Joint Soviet-Western technical study, to examine the types of accident and miscalculation that can now be imagined. This study should seek to devise means for avoiding them--and, if they should occur nevertheless, to keep them from spreading into a cataclysm.

Schemes that merit attention are the establishment of a communications system to enable the leaders of both countries to communicate instantaneously. Joint Western-Soviet offices might be established in Moscow and Washington with their own communications equipment. In addition, special surveillance teams could be set up either under United Nations or joint Western-Soviet control. These teams should be trained to move rapidly to trouble spots to verify information which one side wishes to convey rapidly to its opponents. Both the communications and the special surveillance teams should run frequent exercises designed to guard against the dangers of accidental war and miscalculation revealed by the technical study.

To be sure, in their present state of mind the Soviet leaders are likely to dismiss such a scheme out of hand. And probably it will be ridiculed in the West by those persons who are so horrified by the prospect of nuclear war that they refuse even to consider the possibility that deterrence may fail despite our best efforts. But it is necessary to provide some means for preventing the automatic spread of every conflict into a holocaust. "Positive evidence" inspection can reduce the danger of a war brought about because one side is afraid that its opponent is afraid.

The notion of establishing a control system especially designed for critical periods admittedly sounds strange. But its strangeness is due to the fact that we still have not yet comprehended the revolutionary nature of our present world. The new technology can be mastered only by political innovations as dramatic as those in the field of science. One test of the sincerity of the Soviet leaders with respect to arms control would be their willingness to participate at least in the technical discussions proposed here.

The problem of designing arms-control measures to deal with deliberate surprise attack is more complicated. "Positive evidence" inspection--to verify information voluntarily given--will not by itself provide adequate protection, while "negative evidence" inspection involves all the liabilities of continuous surveillance. Yet, if arms control is to have any meaning, a method of coping with the problem of surprise attack must be devised. This is important not only as a means to stabilize the arms race; it is even more significant for the future of other arms-control negotiations. If the two sides cannot act constructively on the basis of this manifestly common interest, the danger is real that future arms-control negotiations will be reduced to a largely symbolic effort, devoted to ritualistic incantations of general goals and producing agreement only on measures that are meaningless. Too many proposals in the Western plan submitted to the Ten-power Disarmament Conference have precisely this characteristic.

This is a most unpropitious moment for the detailed and dispassionate negotiations which are necessary for a serious effort to deal with surprise attack. But we cannot gear our position to Mr. Khrushchev's erratic moods. We must be clear in our own minds about the elements of a responsible program. When negotiations do prove feasible we can then enter them with confidence instead of preparing our position under the pressure of a deadline.

There are two prerequisites to an effective system to control surprise attacks:

(1) The opposing retaliatory forces should be stabilized at a level which reduces the incentive to attack to a minimum.

(2) The inspection system should be sufficiently reliable to prevent evasions which can upset the strategic balance, yet not so pervasive as to destroy the security of the retaliatory force. How can these objectives be reconciled?

In a subject of such technical complexity it would be rash to offer any proposal as a final solution. But if a scheme does no more than indicate the complexity of the problem, it will have served its purpose. In arms control, simple answers are almost surely wrong and the quest for them is an obstacle to real progress. If we really want to deal with the problem of surprise attack we must be prepared to face complicated situations.

Thus it may be useful to consider a combination of "negative evidence" inspection with what may be termed "inventory" inspection, the purpose of which is to determine the actual strength of the opposing forces.

How would this work in stabilizing retaliatory forces at a level which minimizes the incentive to attack? Under this system, each country possessing nuclear weapons would designate regions where it would agree to station no retaliatory weapons, and another group of areas in which retaliatory weapons up to a certain number would be permitted. The areas stripped of retaliatory weapons would be open to unlimited inspection. Here would be installed "negative evidence" inspection designed to find out the absence of retaliatory weapons.

In the regions where retaliatory weapons are stationed only "inventory inspection" would be permitted. Inventory inspection would mean that at some agreed interval, say twice a year, inspectors would have free access to determine the strength of available forces. But they would be barred at other times. During the period of inventory, the retaliatory force would have to be stationary, a fact which could probably be monitored through an adequate combination of ground and aerial inspection. But it could be mobile at other times.

It may be objected that such a system would create an enormous vulnerability during the period of inventory inspection when the precise location of the opposing force would be known. This danger can be overcome by placing the land-based retaliatory force in several different areas separated by territory in which uncontrolled inspection is permitted. The inventory in each area could then be taken at different times. Thus the retaliatory force would continue to be mobile in some regions even while it is being counted in others. Uncontrolled inspection in the territory separating these areas could monitor the shifting of weapons from one region to another.

Such a system should be effective in preserving the equilibrium or at least in keeping evasions within limits that prevent either side from obtaining a decisive advantage. Uncontrolled inspection in the "disarmed" regions would provide also a measure of control over the production of retaliatory weapons particularly if the "armed" areas were located so that they contained a minimum of industry--as they should be. The system could be further strengthened by stationing inspectors at all access points to armed areas. In these circumstances it would seem unlikely that either side would be able to increase its retaliatory force beyond agreed levels on a scale sufficient to be worth the risk of detection, either by means of uncontrolled surveillance or through the system of inventory control. Inventory inspection would be adequate to determine the strength of the retaliatory force within acceptable limits of error. It would not be sufficient to determine its location and thereby compromise its security.

The scheme here described may be a means also to give effect to another common interest of the two sides: to protect as much of the civilian population as possible should deterrence fail by accident or miscalculation. The first objective of a surprise attack must be to destroy the opposing retaliatory force. Even the maddest aggressor would know that an attack on cities which failed to eliminate the defender's capacity for retaliation would merely guarantee a devastating counterblow. Moreover, once the retaliatory force of the opponent is destroyed, the aggressor may have a positive interest in sparing the civilian population. For blackmail purposes, the opponent's civilian population is more useful alive than dead. Thus surprise attack against centers of population is likely only if they are thought to shelter retaliatory weapons. If major population centers are in the "disarmed" areas, the consequences of a failure of deterrence would be much mitigated.

This raises the problem that placing all major population centers in the disarmed areas may defeat the possibility of land-based mobility. Cities are usually also the hub of the road and rail network. Another objection may be that a system which opened all cities to uncontrolled inspection--implicit in disarming them--would be bound to be unacceptable to the Soviet Union. One solution is of course to design land-based retaliatory systems so that they can be moved cross-country without using roads or railways. If this is deemed impractical, each signatory of the agreement suggested here could be given two choices. It could "disarm" its centers of population even if they were located in armed areas. In that case, it would have to admit permanent inspection teams whose task would be to verify that the city sheltered retaliatory weapons only when in transit--and this term should be rigidly defined to exclude any pause of more than an hour or so. Or else a country could refuse to admit inspectors into the population centers in "armed areas" except for inventory, in which case it would have to take its chances in case of war. This would place the population of the cities concerned in the position of inhabitants of fortified towns in conventional warfare.

As in any arms-control scheme, a number of loopholes suggest themselves. One concerns that part of the retaliatory force based at sea. Here the system of inventory control would obviously be impossible and constant surveillance, even if it could be designed, might defeat one of the chief advantages of a sea-based retaliatory force: the uncertainty about its location. One solution that occurs to the author--though more intensive study would undoubtedly reveal others--is to include all port cities and harbors in the regions of unlimited inspection. This would provide a check on new construction--a useful form of surveillance if the system came into being before sea-based forces have become too large. And since ships, whatever their endurance, have to return to port from time to time, an inventory of the total force could probably be obtained in this manner.

Another lesser difficulty is posed by airplanes. An inventory by region is not possible here because airplanes can be shifted around so easily and are not subject to control while in transit. Thus if each area were inspected in turn, the count could be falsified by shifting planes out of the region being inspected. The solution is to begin the inventory with a count of all airplanes in all armed regions. At the beginning of the inventory all planes would be grounded and inspection teams would move into the airfields. After the airplanes are counted, the inventory of missiles would proceed region by region. To prevent the airlifting of missiles out of areas where inventory is about to take place, the inspection teams would remain at the airfields until the inventory in all regions is completed.

It would be idle to pretend that such a system--if it could be negotiated--would guarantee stability for the indefinite future. The problem of a technological breakthrough will always remain with us. Stability in numbers of offensive weapons--which is essentially the purpose of the scheme outlined here--could be made irrelevant by a major advance in defensive weapons and systems. If one side developed a defense against ballistic missiles which it considered very highly effective, it could use even its controlled retaliatory force for blackmail since it might feel safe from a counterblow.

The difficulty of devising a method to obtain a count of weapons already produced suggests the complexity of controlling scientific research. An effort of this nature would have to wait until the two sides gain competence through the operation of simpler control schemes. In the meantime, no alternative exists except to keep up in the technological race and to seek to maintain stability, in addition to arms-control schemes, by an unremitting effort in research and development.

It may be objected that the Soviet Union will never accept so comprehensive a scheme. But surely we cannot fail to make a responsible proposal simply because it may be rejected. If neither negative evidence nor positive evidence nor inventory inspection prove negotiable, Mr. Khrushchev's speech at the United Nations in favor of complete disarmament will stand revealed as a cynical propaganda hoax. If no meaningful scheme involving inspection is acceptable, the melancholy conclusion may be that the requirements of the Soviet domestic system are incompatible with serious arms-control measures.


It is essential to understand that a price must be paid for each step towards arms control. Refusal to recognize or to admit this fact would have the consequence that arms control instead of increasing security would create new weaknesses. To the degree that the risk of surprise attack is reduced the threat of local aggression may be increased. To the extent that the "positive evidence" concept of inspection makes accidental war less likely, it may encourage probing action, and new forms of blackmail.

One of the key sanctions against local aggression is the fear that, whatever the intentions of the side that initiates it, neither side may know how to limit its scope. The Communist bloc has been deterred to some extent from exploiting its local preponderance in certain areas because it is not sure that the resulting war could be limited. The control measures outlined above would remove much of this constraint. One of the chief purposes of the positive evidence type of inspection is, after all, to enable the nuclear powers to control the scope of conflict and to reassure each other in tense situations. Similarly, if inspection of the retaliatory forces proves reliable the result must be the total incredibility of the threat of all-out war. For the purpose of such a control scheme is to deprive all-out war of any rational goal.

Effective control against surprise attack may make limited wars not only more likely, it may cause them to increase in intensity. If communication between the protagonists is assured and if a mechanism exists for verifying that a given act is not a prelude to surprise attack, an aggressor may become confident that regardless of the scale of aggression he will always be able to keep matters under control. When both sides have accepted the notion that accidents must not lead to all-out war, and when they have examined means of containing a conflict even in the face of very substantial provocation, an aggressor may be tempted to step up pressure. It would be ironic indeed if the frequency and perhaps the violence of conflicts increased because the powers are too much in control of events.

Nevertheless, the benefits of a scheme to prevent surprise attack outweigh the risks. The perils can be met if we are willing to admit them and if we do not treat an agreement on surprise attack as an excuse to relax our efforts. As arms control reduces the possibility of surprise attack, greater stress must be placed on local defense. The possibility of Communist pressure will not be ended but it will be channeled into an arena of conflict where the alternatives are less stark and the risks less dreadful.

Moreover, it would be a mistake to look at the problem only strategically. A control system against surprise attack could not fail to have profound political and psychological consequences. If both sides can coöperate to the extent envisaged here, the solution of many other issues will become much easier. For example, the regions of unlimited inspection could be so arranged as to provide buffers against local aggression. Inventory inspection might be a device to determine reduction to agreed force levels. Positive inspection could be applied to the reduction of nuclear stockpiles.

But we must not delude ourselves. The prerequisite is a Soviet decision to allow meaningful inspection. If Mr. Khrushchev is serious about his disarmament proposal, the best first step is to agree to the kind of inspection described here.


No aspect of American policy has received less systematic attention than arms control. Substantial intellectual as well as material resources have been devoted to the study of strategy. Yet arms control, which is its reverse side, has lacked a focus of attention. As a result, it has been difficult to achieve agreement about desirable goals and, even more, to develop a dynamic program. Before there can be a successful negotiation on arms control we must get our intellectual house in order.

The need is all the more urgent because the rate of technological change has continually outstripped the pace of negotiations. Whatever may have been the possibilities of controlling nuclear weapons when the Baruch-Lilienthal plan was first formulated, so many weapons have been produced in the interval, and the stockpiles have grown so large, that complete abolition would be impossible to inspect. Similarly, in the early stages of the missile age it might have been possible to arrest the proliferation of rockets, either by a ban on testing or through a severe limitation on production, or both. But neither side sufficiently understood the implications of missile technology for this proposal ever to be seriously advanced. Today, measures to control the missile race are infinitely more complicated.

Two conclusions follow: First, arms control schemes should be developed with a clear understanding that their rate of obsolescence is as rapid as the rate of technological change. They must therefore have a built-in mechanism for adaptation. Second, there is a critical point in the development of any weapon after which arms control becomes impossible or at least extremely intricate. After this point is passed, it will seem to both sides--and probably correctly--that they will be more secure through unilateral efforts than through any control system that they are able to devise, agree on and inspect. This suggests the crucial importance of controlling new weapons in the very early stages of development. From this point of view, the adamant Soviet refusal even to discuss control of earth satellites armed with nuclear weapons is not encouraging.

Understanding of the purpose of arms control is made all the more difficult by the divisions within our military establishment about the nature of security. For without an agreed strategic doctrine we will possess no criteria to determine whether a given scheme would add to stability or detract from it. The view about the significance of local defense determines the conclusions about the feasibility of demilitarized zones. The interpretations of vulnerability influence the schemes for the prevention of surprise attack. Since we have failed to resolve these issues in our own military establishment or within the Western Alliance, it is not surprising that the proposals in the field of arms control have been so tentative. And the situation has not been helped by the attitudes of some of the most passionate and vocal defenders of arms control who advance the deceptively simple proposition that only the elimination of all weapons of mass destruction can maintain the peace and who resist efforts for limited measures as an attempt to defeat their cherished objective.

The failure to achieve an arms-control agreement is not, of course, entirely or even largely the fault of the United States. Soviet proposals so far have been consistent only with one of two interpretations: either the Soviet leaders have as much difficulty in understanding the problem as we do, or else they are using the negotiations in a deliberate effort to demoralize the free world and to induce it to disarm unilaterally. It is discouraging that almost every Soviet proposal seems more designed to exploit the weariness of the West than to slow down the arms race. Our recognition of the importance of the problem and our eagerness to find a solution must not blind us to the cynicism of much of the Soviet performance.

But our judgment about the Soviet approach does not excuse the uncertainty of our response. Perhaps no serious negotiation is possible at all. But we will be able to determine this only by becoming clear in our own minds about the purpose of arms control and by devising responsible schemes for attaining it. If the Soviet Union rejects proposals which are designed to increase its security together with ours--and this is the essence of any responsible program--it will have given clear proof that there is no alternative to the arms race.

Before we can advance these proposals, we have to end the confusion in our own minds as to our purposes. It is said that we must engage in arms control to free resources for the real competition which is in the field of economics. It is argued that arms control may reduce the burden of taxation. Arms control is advocated as a means of speeding up the evolution of the Soviet system.

Almost all these arguments are essentially irrelevant. The Soviet leaders can hardly be attracted by schemes whose primary purpose is announced to be the transformation of their system. The argument about freeing resources for economic competition begs the principal question which is, after all, precisely whether it is possible to develop arms-control measures which achieve this end. It is not at all certain that arms control will in fact free resources, particularly in its early phases. Inspection is expensive. Additional funds for research are essential. A recasting of our military establishment may have to accompany arms control. To justify arms control as a device to save money may cause us to be attracted to the wrong schemes for the wrong reasons.

If we are to make progress in the field of arms control, the military establishment must come to understand that in the present state of technology an arms race is the most unstable of all forms of security, and that properly conceived arms control may increase the security of all countries. And many enthusiasts for arms control must realize that ardor is no substitute for precision. A great deal depends on the ability to be concrete. In the next few years we may have perhaps our last opportunity to stabilize the arms race. Perhaps Soviet obduracy will foil our most earnest efforts. But it would be unforgivable if we failed because we have refused to face either the importance or the complexity of the challenge.

[i] Thomas C. Schelling has discussed this problem brilliantly in "The Strategy of Conflict." Cambridge: Havard University Press, 1960, p. 246-254.

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
  • HENRY A. KISSINGER, Associate Director of the Harvard Center for International Affairs; author of "Nuclear Weapons and Foreign Policy"
  • More By Henry A. Kissinger