20 August 1987


  The following remarks intend to highlight the differences between the control of nuclear weapons in NATO Europe as compared with the intercontinental systems of the United States, which have been much more thoroughly investigated in recent literature. I do not purport to provide either a comprehensive account or a definitive assessment, but I would like to offer an outline of the main issues involved. I will first single out the main characteristics of NATO's nuclear arsenal, from the perspective of their control. I will then show how these result in a set of unsolvable problems that the alliance must live with. And I will finally address the question of what could be done to at least alleviate these problems.

1. The Nuclear Situation in NATO Europe

The control of NATO nuclear systems in Europe is made both difficult and highly sensitive by eight geographical, technical, and political factors. The nuclear control environment in Europe, unlike that of the U.S. intercontinental systems, is made highly complex by what I will call the "triple duality" of the nuclear arsenals; this triple duality encompasses the first three of those factors. First, many NATO nuclear weapons in Europe are "dual purpose" meaning they have both a strategic and a tactical capability, depending on their range or basing location; yet, from the point of view of someone who lives in Europe, it is hardly possible to think of any war, conventional or nuclear, that would be anything but strategic. This complicates the operational planning for systems that are supposed to accomplish both types of missions according to the prevailing circumstances.

The second duality is constituted by the "dual capability" of many systems that can deliver both conventional and nuclear ordnance. Most of NATO's delivery systems, artillery, aircraft, and missiles, fall into this category, and control arrangements are complicated by the obviously different safety and security requirements of either type of ordnance.

Third, many systems are "dual key" meaning they are jointly controlled by two allied countries, the U.S. providing the warhead while another ally operates the delivery system. Moreover, according to NATO policy, also the countries which host allied weapons on their territory, or on whose territory NATO targets might be located, would have a say in the decision to use these weapons, even though the precise arrangements covering consultation in such circumstances are ambiguous.

A related problem is posed by the control of U.S. manned systems which are hosted on allied soil. Strictly speaking, these are governed by a singlekey arrangement, but the U.S. is still committed to take into consideration the views of the host countries as well as of the other allies who would be involved in their operations. The extent to which it would actually do so, however, is highly uncertain, and the allies themselves provide inconsistent evidence of their stipulations with the U.S. in this regard. Recently, for example, the West Germans have played down the degree of their involvement in the control of the Pershing II and GLCMs installed on their territory, while the British and Italian governments have gone to great lengths trying to reassure their respective electorates that the U.S. would not and perhaps could not launch these systems without their prior approval.

Fourth, control for European-based systems must by necessity be much more decentralized than for central systems, as in war many would be dispersed among a high number of operational units around the continent. Some believe that this is indeed the core of deterrence, that it is in fact the inherent uncontrollability that makes the risk of aggression incalculable for the Soviets. Others counter that decentralized control increases the dangers of accidental war. The two positions are not necessarily incompatible, and both in fact contain a kernel of truth: decentralization complicates control requirements and presents a difficult tradeoff between deterrence and crisis stability.

Fifth, control is complicated by the technical complexity of the NATO arsenal, which is composed of dozens of different systems, and is regulated by different national procedures, only partially standardized at the alliance level.

Sixth, the geography of the continent sharpens the difficulty of responsible control by shortening reaction times and thus putting pressure on decision-makers and increasing the danger of ill-considered responses and overreaction to events in the field. This difficulty might be compounded by the likely divergence among the assessments of the situation which the various countries involved might arrive at through their national intelligence means.

Seventh, differences in the political and military goals of the allies may arise during the course of a crisis or while NATO might be negotiating peace with the Warsaw Pact; in a situation where thealliance might be trying to pursue a combination of fighting and negotiating political and military interests of 16 so diverse countries such as the NATO members might easily diverge. The 1985 incident between the U.S. and Italy over the handling of the terrorists of the "Achille Lauro" highlighted how even minor issues might degenerate into intraallied crises when political perceptions diverge and time for decisions and consultation is short.

Eighth, most systems are physically vulnerable. Vulnerability is due both to geography and to the nature of the basing of the systems. Most NATO nuclear weapons are vulnerable to conventional as well as to nuclear attack, and so are their C3I networks. The latter are also susceptible to disruption caused by indirect effects of nuclear explosions, such as electromagnetic pulses (EMP) and fallout. While most nuclear weapons are mobile, their mobility is limited if their performance in terms of accuracy and quick reaction is to remain at the highest standards.

2. Unsolvable Nuclear Dilemmas in Europe

The eight factors just described all contribute to four broad categories of problems for nuclear control in Europe. These problems are essentially unsolvable, and will remain with the alliance for the foreseeable future, and the best NATO can do is improve them at the margin. Let us briefly analyze each in turn. First, the military and political complexities of nuclear issues, and particularly of nuclear control procedures, are such that the responsible chief political leaders of NATO can not possibly be expected to understand them well. Presidents and Prime Ministers change, nuclear procedures change, and there are just too many things on the agenda of top leaders for them to devote as much time to nuclear issues as it would be necessary to understand them sufficiently well as to be able to make an informed judgment in a crisis.

Second, given NATO's doctrine of flexible response, which is intended to leave the maximum room for maneuver to NATO leaders, it is impossible to preplan all possible contingencies which might arise in Europe and require a response through NATO's nuclear arsenal. Even assuming that top leaders did devote much more time to understand nuclear issues, there would still remain a considerable margin of uncertainty and therefore nuclear control in Europe would always depend on a degree of extemporizing, with its inherent unpredictability and dangers. The military and technical complexity of the situation in Europe make the problem worse than it is with the central systems of the U.S.

Third, NATO remains locked into the dilemmas of collective control. Ever since the late fifties, the Europeans have tried to increase their say in the control of U.S. nuclear weapons dedicated to their defense, while the U.S. has sought, successfully, to preserve tight central control. Clearly, it has been impossible to accommodate both requirements. The Europeans have been concerned with acquiring both some form of a trigger on NATO weapons so as to make the deterrent more credible in Soviet eyes in light of possible American hesitancy and some form of safety catch over those same weapons so as to avoid the danger of the Americans fighting a limited nuclear war in Europe. Clearly, however, if any one country has a unilateral trigger to launch the arsenal, then none can have a safety catch, while if anyone has a unilateral safety catch, then none will have a trigger. This unsolvable dilemma has been with NATO for at least thirty years and, even if its political salience has considerably decreased in recent years if compared with what it was a couple of decades ago, will remain for the future the only exception being the case of France, which decided to pursue her own independent nuclear capability so as to have her own trigger.

The fourth unsolvable problem is what to do about the enemy's C3I network. From a military point of view, it would be desirable to target it so as to diminish the military effectiveness of enemy forces. On the other hand, from the point of view of crisis management and war termination, the enemy command must be spared so as to have someone to negotiate with. The Provisional Political Guidelines approved by NATO in 1969, and elaborated into the General Political Guidelines of 1986, stipulate that the top Soviet leadership will not be attacked. Yet, it is hard to believe that this same principle would also apply to low level Soviet and Warsaw Pact command. While the details of NATO targeting are obviously highly classified, it is reasonable to assume that NATO plans to attack low-level enemy command, possibly up to the division level, while sparing the higher commands, who would thus be able to conduct negotiations with NATO. At what point the damage inherent in the loss of negotiating partners becomes greater than the gain inherent in the enemy's loss of military effectiveness because of crippled command is highly uncertain. In the early stages of a war in Europe, it might intuitively make more sense to preserve enemy command so as to limit escalation, but it may well be argued that minimizing the chances of escalation through discriminate targeting will weaken deterrence and thus be undesirable. Again, this problem is more difficult in Europe than for central systems because the command of theater system is more articulated and intermingled with that for conventional forces.

3. Categories of Nuclear Control in NATO Europe

Three types of devices and procedures are used to control NATO nuclear weapons in Europe: administrative, physical, and informational. This section will briefly analyze each in turn. Administrative controls are procedural arrangements through which the central authorities strengthen their control over the forces. Examples of administrative controls in NATO include measures such as the so-called "two-man rule", meaning that no one man is allowed to have access to a weapon without the presence of a companion with equivalent expertise, i.e. an electrician must be accompanied by an electrician, etc. For example, aircraft on Quick Reaction Alert (QRA)10 to 15-minute reaction time are parked in so-called "no lone zones", where an armed U.S. soldier watches that no single individual gains access to the aircraft the only exception being, of course, when a pilot has to take off for a mission.

Additional administrative control is obtained through a detailed accounting of the location and the movements of the weapons and of their components, with redundant checks throughout the process; or through periodic as well as random tests and certifications that the weapons are not tampered with. There have been times in the recent past, such as during the GreekTurkish crisis of the mid-seventies, when the U.S. became concerned that allies involved in conflicts other than those where the mutual defense pledge of the alliance would apply might have tried to gain access to NATO weapons without authority.

Administrative controls are intended to reduce the probability of unauthorized use. They can not by themselves offer an absolute guarantee against it, because they do not hamper the capability of those with access to the weapons to use it without authorization. They do however require that two or more people cooperate to breach the orders of the authorities, thus making such a breach more unlikely.

Second, there are physical controls. These are measures that are intended to provide to the forces the physical capability to launch at a time of the authorities' choosing and, conversely, to preclude accidental launches at all other times. Controls of this type in NATO Europe include storing warheads separately from their delivery systems, in appropriate storage sites configured according to the specific systems and warhead characteristics. Alternatively, in the past, critical non-nuclear components were separated from the rest of the warhead.

Another example of physical control devices are the so-called environmental sensing devices (ESD). These are devices that are sensitive to variations in the weapon's position and velocity, in the air density around it, etc. and can disable the weapon if those data do not correspond to those for which the use of the weapon is envisaged.

Physical control devices can however be bypassed. In particular, they are threatened by "front door" and "backdoor" circumvention. One would have a front door circumvention if an unauthorized user succeeded in activating the weapon by acquiring and then using the physical means e.g. keys that were supposed to prevent unauthorized use. Backdoor circumvention will take place when an unauthorized user will bypass the control device by neutralizing its function in the weapon e.g. by shorting the electrical circuits connected to a lock. Once more the problem in Europe is more serious than for central systems in light of the inescapable higher exposure that systems there have to tampering and unauthorized access.

The third type of use control devices and procedures are informational, that is arrangements whereby indispensable data for the execution of a launch is transmitted by the authorities to the forces only at the appropriate time and place for the forces to execute authorized use decisions. No hardware needs to change hands for informational controls to release their weapons. Such controls will include electronic or combination locks that are opened by, for example, turning knobs or keying the proper digital codes.

Physical and informational controls share several similarities. Both are intended to deprive the units that man the forces of the capability to use them without authorization, while administrative controls can only make it more difficult for unauthorized personnel to use weapons that are however in themselves ready for use at all times. Both face frontend as well as backend circumvention problems: hence, both have been equipped with similar features to face those two sets of problems. For example, the probability of frontend circumvention has been reduced by designing the devices with mechanisms that will render the weapons temporarily or permanently inoperable if some conditions obtain. Such conditions include, among others, the succession within a given period of time of a limited number of activation attempts with wrong keys or codes; or attempts to handle or move the weapon without scrupulously following a pre-established sequence of actions.

Backend circumvention, on the other hand, is prevented only in the latest devices which are designed in the weapons, and not attached to themor, still worse, attached to the delivery systemso that it would take unauthorized personnel a great deal of time and expertise to possibly bypass them. Like in the case of frontend problems, the devices are so designed that attempts to bypass them would render the weapon inoperable.

Finally, one can have devices that combine two or more of the above three categories of use controls. Since the early sixties, electronic locks called Permissive Action Links (PALs) have been first attached to and later designed into the weapon so that the latter could not be launched or detonated without both an alphanumerical code informational control and a decoder to insert that code. Physical control and custody of both of which has been appropriately assigned to U.S. personnel other than that which mans the weapon system.

All these administrative, physical and informational controls are in fact jointly used in NATO nuclear weapons in Europe, though not uniformly. The introduction of administrative controls in NATO forces has been relatively uncontroversial. All concerned agree that they are necessary. The same can not be said for physical and informational controls, which, while operational on all weapons on land, have been successfully resisted by the U.S. Navy and have not been installed in naval forces, including those assigned to operations within the NATO command and in the European theater.

The U.S. Navy has put forward a variety of arguments against the need for them. I will briefly analyze and critique each in turn. My conclusion is that the U.S. Navy has failed to provide a logical explanation for its continuing refusal to accept physical or informational controls on its weapons. First, sailors have traditionally detested "rudder orders from the beach", and PALs would be just that. This is admittedly a psychological argument with no logic whatsoever to it.

Second, physical controls are unnecessary because the highly selected navy personnel would in no case act without proper authority moreover, naval personnel with nuclear command responsibility are more senior than Army or Air Force personnel with similar functions. Yet, it is difficult to understand why it is acceptable for Navy commanders to have to wait for authorization messages to arrive before launching their force, while it should be unreasonable to wait for enabling codes. Admittedly, the EAM would be lengthened somewhat if enabling codes were added to authorization codes; this would mean a somewhat longer processing time for VLF or ELF transmission to the SSBNs, but the amount of extra time needed would be measured in very few minutes at most: most likely an insignificant loss considering the fact that the SSBN would be highly invulnerable and not under tight time pressure to act the Navy is of course well aware of this invulnerability, as can be seen in the ninth argument below. In any case, this argument does not apply to nonsubmarine-based weapons at all carrier based nuclear weapons, for example, do not rely on VLF/ELF communication.

Third, naval procedural controls ensure that no one in a naval vessel could execute an unauthorized launch. The problem with this argument is that while no individual has the capability to execute an unauthorized launch, a small number of officers could.

Fourth, there have been no accidents at sea to indicate that a danger of accidental launch exists. That there have been no accidents does not mean we should wait for one to happen before thinking about remedies.

Fifth, naval budgets should be spent on other more important priorities. It is indeed hard to imagine of more important missions than the avoidance of unauthorized or accidental nuclear use.

Sixth, physical or informational controls would constitute yet another complex mechanism that would have to function in an emergency, and could therefore impair the reliability of naval weapons: electronic equipment will fail more often than people. Yet, modern naval weapon systems already rely on a panoply of gadgetry that is much more complex than electronic locks: the added "complexity" which the latter would add at the margin would be negligible. In any case, the malfunctions of a few of the locks would hardly compromise the missions of a force several thousand of weapons strong: physical control would constitute, at worst, a slight risk of very partial and graceful degradation of the naval deterrent, never of catastrophic degradation.

Seventh, the navy's capability to launch nuclear weapons without authorization from the NCA strengthens deterrence because the Soviets know that even a successful preemptive attack aiming at decapitation would not disarm the U.S. retaliatory force. It is not clear why the Soviets should be deterred from a decapitating attack by the fear of retaliation by a force that allegedly would under no circumstances be launched without authorization from the NCA. To the extent that the Soviets are to be deterred by the lack of control devices on navy weapons, the navy's commitment never to launch without authorization and hence the second argument cited above loses credibility.

Eighth, unauthorized use of at least some naval forces, while extremely unlikely, would, in any case, have less disastrous consequences than that of landbased forces, since some naval weapons, e.g. those targeted against naval targets would likely not cause an unwanted escalation in a conflict, because the area involved would be very limited and there would be no collateral damage. While it is true that in some cases nuclear use at sea might remain well-circumscribed, this is bound to be highly certain; moreover, most naval weapons have a capability to hit land targets and the most highly valued by an enemy among them. Hence this argument could apply only to a few naval systems e.g. nuclear anti-ship missiles. If accepted, this argument would call for leaving these systems free from use control devices, which would however be installed on SLBMs, carrier-based aircraft, land-attack cruise missiles, etc. However, neither the Navy, nor anyone else, has called for differentiation in the use of control arrangements among different naval nuclear systems.

Ninth, naval commanders would be less prone to hasty action because, thanks to their low vulnerability, in an emergency they would not be under the same time pressure as their Army or Air Force peers would likely be, especially if located near the battle area. By the same token, however, they could afford to wait the few extra minutes which might be necessary for an EAM lengthened by enabling codes to be copied but, if so, argument two above is decisively weakened.

The final, and most powerful argument for the lack of use control devices in the Navy, is that the danger of host country takeover, which was the most important rationale for installing PALs in Europe, does not exist for most Navy weapons except for ASW warheads based on land in Europe, which are equipped with PALs. If indeed the main purpose of use control devices was to insure against the dangers of potentially unstable allied political leaders or overly entrepreneurial allied military commanders, then indeed there is no need for such devices on Navy weapons, which would be virtually impossible to seize for the allies sand even more, of course, for terrorists or psychotics.

4. Conclusions and Policy Implications

The main conclusion which emerges from this overview of the control of NATO nuclear weapons in Europe is that the problems and the dilemmas posed by the technical, military, and political characteristics of the arsenal are such that they defy any clear and definitive solution, at least under the current doctrine of flexible response, about which a few words are in order at this point.

The current doctrine of flexible response requires continuous adjustment to the evolving conditions on the part of the decisionmakers; because of differences between the European and U.S. interests over when and how to resort to nuclear weapons for the defense of Europe, uncertain and ambiguous control premises had to be adopted so as to provide an appearance of coherence among otherwise irreconcilable national interests. To put it very briefly, Europeans have tended to favor a doctrine of early use in the hope of stopping any war right at the outset without too much collateral damage to their homelands. The U.S., on the contrary, fearing an escalation of nuclear warfare to its own territory, has pushed a gradual move towards delaying the resort to nuclear weapons until all else has failed to stop the aggressor, including protracted conventional fighting. The U.S. has guaranteed its right to enforce this delay by keeping control of all NATO nuclear warheads while committing itself to consult the Europeans as feasible and necessary in any crisis.

Flexibility has therefore been interpreted by NATO to imply uncertainty for the adversary about what NATO might do in response to varying levels of threats. In turn, uncertainty for the adversary has been interpreted to require ambiguity of one's own control arrangements, so that the adversary could never be sure about what circumstances would trigger the arsenal, and would be thus deterred from initiating any offensive action. However, there is no logical reason why flexibility of operations for NATO should require uncertainty for its potential adversaries. Nor is there any why uncertainty for the adversaries should require ambiguity for NATO's own control arrangements. In fact, might be better to eliminate uncertainty in eyes of an adversary, and instead convince him that his aggression will surely be met with unacceptable damage because of nuclear retaliation.

But as long as the U.S. controls NATO's nuclear arsenal both uncertainty and ambiguity are likely to remain. It would not be rational for the U.S. to risk involving its territory in a nuclear war for the sake of Europe. The U.S. did intervene in the two past major European wars in this century, but then it was assured that its society was invulnerable to attack. This time Washington would be well aware that American society would be vulnerable to annihilation, and thus would certainly be more cautious. Uncertainty of strategy and ambiguity of control serves to ensure U.S. options to be cautious while presenting to the Europeans a politically acceptable perception of involvement in the nuclear defense of the alliance. 

The ultimate question, from a European point of view, is whether better alternatives exist which could be acceptable to the U.S. Such alternatives have been sought for decades, and have not been found. The answer to that question is no, at least as long as the Europeans continue to rely on the U.S. nuclear guarantee. Whether the Europeans should cease doing that, and provide for their own nuclear defense like the French and the British are already doing can be debated, but there are many and most compelling political reasons, which go beyond the scope of this paper, why they should not.

        In sum, the current arrangement for the control of NATO nuclear weapons is far from optimal: NATO's doctrine sets requirements which the control structure can not fulfill. But the current arrangement is likely to remain for the foreseeable future as no better alternative is acceptable to all the allies concerned.