The elimination of first strike threats and of large-scale nuclear war by accident or miscalculation are some of the most urgent priorities for de-alerting. However, de-alerting should be carried out in such a way as to represent the clearest and most significant progress towards complete nuclear disarmament, in fulfillment of Article VI of the Nuclear Non-Proliferation Treaty (NPT), as interpreted by the World Court in its advisory opinion. In other words, partial de-alerting measures cannot be seen as ends in themselves, any more than dismantling some nuclear weapons can be a substitute for complete nuclear disarmament.

In the immediate term, de-alerting measures can proceed even without a prior commitment to complete nuclear disarmament since nuclear weapons can be returned to alert status. However, de-alerting all nuclear weapons will essentially eliminate the risk of large-scale accidental nuclear war, and greatly lower the risk of war by miscalculation.¹ Therefore, de-alerting can allow for a nuclear weapons stand-down that will allow the political room and the time to achieve complete nuclear disarmament in a safe and verifiable way.

It can also allow for a process in which the five nuclear weapons states parties to the NPT can bring the other three nuclear weapons states into a process that neither denies the existence of their arsenals, nor legitimizes them. This is important, as stable maintenance of a state of complete de-alerting will require participation in verification by all eight nuclear weapons states.² Specifically, a verifiable halt to production of new nuclear weapons will be required in order to prevent clandestine deployment.

However, there is ample room for unilateral actions. For instance, partial de-alerting does not require prior agreement on verification, and can be carried out in order to test verification procedures and build confidence. Partial or even complete de-alerting can also be carried out unilaterally by any nuclear weapon state that subscribes to a second-strike deterrence policy, regardless of the differences in political and strategic situations of different nuclear weapon states.

Further, de-alerting measures are complementary to existing arms reduction processes, such as those which are occurring under START I and are scheduled to occur under START II. Most of the world's countries and many other leaders and NGOs have been insistently calling for nuclear disarmament, and de-alerting is widely seen as a crucial first step. For instance, the Canberra Commission endorsed de-alerting, as have retired US Admiral Stansfield Turner (former CIA director), and General Lee Butler (former chief of the US Strategic Air Command). The recent initiative by Brazil, Egypt, Ireland, Mexico, New Zealand, Slovenia, South Africa, and Sweden calls on the nuclear weapons states

Short-term De-alerting Measures

There are three main approaches to de-alerting: a) removing warheads from delivery systems; b) prolonging missile firing time; and c) reducing risks of first-strike and accidental launch. Generally speaking, explicit abandonment of first-use and first-strike options or launch-on-warning postures broadens the range of possible de-alerting measures, enhances their verifiability, and improves their connection to the process of complete nuclear disarmament.

a) Removing warheads from delivery systems

The surest way of preventing accidental large-scale nuclear war is through sequestration - that is, removing all warheads from their delivery vehicles and storing them at remote locations.⁴ The time it would take to put such warheads on re-alert would depend on how far the warheads were stored from the delivery system, what other means of disabling warheads had been implemented, and whether any multilateral monitoring and verification of de-alerted warheads had been implemented. The surest and most stable de-alerting arrangement would be tagging, disabling, and storing all warheads at considerable distances from their delivery systems, under physical monitoring of all eight nuclear weapons states and some non-nuclear weapons states. Disabling of delivery vehicles and its verification would complement these steps.

There are varying degrees of technical difficulty in achieving sequestration of nuclear weapons. Bombs can be separated from bombers easily and stored apart from them. They routinely are. For example, this was done for all nuclear bombs included in the unilateral de-alerting ordered by President Bush in September 1991 in the wake of the coup attempt in the Soviet Union, which was followed by a similar step by President Gorbachev. De-alerting bombs can be accomplished within a few hours or days, depending on the total number of warheads and locations involved. By the same token, this de-alerting measure can be simply and quickly reversed, depending on the distance bombs are stored from the bombers capable of carrying them.

Tactical nuclear weapons were also withdrawn from deployment on a large scale in 1991, as part of the same US and Soviet actions mentioned above. These withdrawals from surface ships, attack submarines, bombers, and land-based delivery systems (such as artillery guns) were accomplished relatively easily and quickly (within days, weeks, or months, depending on the specific circumstances).⁵

Land-based missiles pose greater complications. While their warheads can be separated from their missiles, the large numbers of warheads involved may require that new storage facilities be built. Moreover, given the dangers of diversion, it may be safer in some cases to store disabled warheads within missiles than to remove warheads from the missiles, until appropriate storage, monitoring and verification arrangements can be made.

Strategic nuclear missiles on submarines present the most difficult case for complete removal from delivery systems. This is because strategic submarines represent the most "survivable" part of nuclear arsenals and are essentially invulnerable once deployed at their launch sites.⁶ According to present nuclear doctrine, nuclear weapons states rely on them for assured retaliatory capability in case land-based systems and aircraft are destroyed in a surprise nuclear first-strike. Thus, submarine launched ballistic missiles (SLBMs), hidden at sea and relatively invulnerable to detection and surprise attack, are considered the best deterrent to a surprise nuclear first strike. By contrast, land-based multiple warhead missiles are regarded as the most attractive target for a first strike, thus making them a candidate for earlier de-alerting by removal of the warheads from their delivery systems.

Another problem with removal of warheads from SLBMs is that the submarines must be brought to port, where they are more vulnerable to surprise attack. Hence, removal of SLBM warheads must be done in proper sequence or in conjunction with other de-alerting and verification methods (see below).

b) Prolonging missile firing time

Missile firing can be made more difficult by various means. They include:

• pinning open the motor ignition switches of missiles, making it impossible to fire them until the pins have been manually removed;⁷

• removing the pneumatic systems that enable the missile covers of land-based missiles to be opened automatically. Re-alerting would require the opening of the missile cover by a crane or by reinstallation of the pneumatic system, introducing delays of hours;

• putting mobile land-based systems in garrisons and putting appropriate barriers on their roofs. This does not introduce much delay but provides greater verifiability than if the systems are actually being moved around (though the measure also makes such systems more vulnerable to a first strike);

• covering missile silos with large mounds of earth: Richard Garwin, a long-time consultant to the US government on nuclear weapons issues, has suggested 20-meter mounds to introduce "some hours" of delay;⁸

• ordering crews not to prepare SLBMs for rapid launch by foregoing procedures such as removal of flood plates from launch tubes and inspection of weapons systems to ensure they are ready for rapid launches. This could buy eighteen hours of delay.⁹

c) Reducing first-strike and accident risks

First-strike dangers have sometimes been addressed in arms control debates by appeal to adoption of a "no-first-use" policy. In this context, we use the phrase "first-strike" to mean a nuclear attack on an adversary's nuclear arsenal with a view to destroying it. ("No-first-use" covers no-first-strike as well as all other possible first-use situations.) For instance, China has stated that it has a no-first-use policy and has called on other states to adopt the same. However, the policy consists essentially of a declaration that is not verifiable and is subject to quick reversal. There is some experience with such a reversal. The Soviet Union had a no-first-use policy, but in 1993 Russia reversed it though it had been in place for over a decade. Thus, while it is a useful confidence building measure, the durability and utility of no-first-use declarations have often been questioned.

Complete de-alerting with multilateral monitoring can be viewed as a verifiable no-first-use policy. Since all weapons will be under monitoring, the threat of a first strike would be eliminated. Such a policy would be robust, because even if weapons states cheated by hiding a few warheads or delivery systems, they could not achieve the objectives of a first strike. The purpose of a first strike is to disable and destroy essentially all of an opponent's nuclear weapons, but this would require more than "a few warheads."

A complete zero-alert requires changing SLBM warhead capabilities to ensure that a first strike is not possible. US SLBM warheads such as the W88 are very accurate and can be used for first strikes. To reduce this first-strike threat, highly accurate warheads could be replaced with types whose yields and accuracies are relatively low.

i) Removing tritium bottles

Removing the tritium bottles from all nuclear-boosted fission and thermonuclear warheads and bombs is an option to eliminate first-strike threats that require warheads with large yields. These kinds of first-strike threats are against weapons stored in hardened silos or other hardened locations. The tritium could be mixed with helium and stored under multilateral monitoring. Such a measure would still allow the first fission stage of the fission explosive to function, though not to its design explosive power.¹⁰ Since the booster would be eliminated, the secondary would not function. Hence the warhead could not be used for a first strike, as defined above. However, it could still be used for nuclear retaliation, since the weapons still would deliver a huge explosion (from hundreds to thousands of tons of TNT equivalent). For instance, a few hundred tons of TNT equivalent is roughly a hundred times larger than the bomb that destroyed the Alfred P. Murrah federal building in Oklahoma City.

Under these circumstances, nuclear arsenals, with the exception of their use regionally by India, Pakistan or Israel, would be limited to retaliatory deterrence, and the explosive power of nuclear weapons would be sharply diminished from thousands of megatons (global total) to roughly a hundred megatons total, perhaps less. This would greatly reduce the consequences of accidental nuclear war. Further, the replacement of tritium reservoirs would become politically and militarily impossible, since a demand for withdrawing tritium from storage would be tantamount to announcing the intention of a first strike. Moreover, mixing the tritium with helium would introduce a considerable delay in the recovery of tritium, making it even more unlikely that it would be withdrawn.

The removal of tritium bottles from warheads could be done at any stage of de-alerting and is compatible with all other methods of de-alerting. While complete verifiability would be difficult, since some uncertainties regarding materials accounting will remain, storage under multilateral monitoring of almost all tritium would ensure that nuclear weapon systems would not be used for a first strike. To accomplish the same goal for India, Pakistan, and Israel would require sequestration because in these regions a first strike may be contemplated with relatively low-yield warheads.

Removing tritium bottles and disabling the boosting and thermonuclear portions of warheads would obviate any need for those parts of stockpile stewardship programs that deal with thermonuclear reactions. Thus, inertial confinement fusion programs would become unnecessary for stockpile stewardship, though they may or may not be pursued for other reasons. Plans for tritium production for use in nuclear warheads, which in themselves exacerbate nonproliferation efforts, would also become superfluous.

ii) Pit Stuffing

After removing tritium bottles it would also be possible to disable nuclear warheads quickly without dismantling them. The method, developed at Los Alamos National Laboratory to deal with weapons deemed to be unsafe, is called "pit stuffing." ¹¹ In this method, the warhead is disabled by stuffing a wire into the tube through which the tritium is "fed" into the primary. The wire fills the hollow portion of the pit, and is stuffed so that it is tangled inside. Once the end is stuffed into the pit, the warhead cannot be re-activated except by completely re-manufacturing the pit. Pit stuffing can be done relatively quickly and does not require the construction of expensive storage facilities for warheads or for pits prior to the complete and permanent disablement of large numbers of nuclear warheads. As with other de-alerting and disarmament measures, verification issues would need to be addressed.

iii) Reducing risks from the Y2K Problem

The need for de-alerting should also be considered in the context of the potential problems associated with computer hardware and software expected at the turn of the century (called the Year 2000 or Y2K problem). In view of the considerable uncertainties associated with the operation of command, communication, and control systems within and associated with nuclear weapons, it would be prudent to implement at least one physical de-alerting measure for all nuclear warheads as much before the end of 1999 as technically possible. Y2K dangers include possible blacking-out of radar screens and malfunctioning of command and control systems. Such occurrences might provide no information or wrong information to those responsible for making decisions to launch nuclear weapons. As a result, the dangers associated with "use-it-or-lose-it" hair trigger postures could increase considerably. One de-alerting measure to address the Year 2000 problem would be to disable the warheads so that the explosives in them cannot be ignited by any malfunction of the electronic system. For instance, this could be achieved by putting a wire into the pit in a manner analagous to the pit stuffing described above, but with the end of the wire left in a position that allows it to be removed and the warhead restored to operability.

De-alerting and Nuclear Disarmament

Most de-alerting can be carried out by nuclear weapons states unilaterally. In fact, it will likely be necessary for the US to undertake some unilateral de-alerting before it can engage Russia in a bilateral program. For instance, Bruce Blair of the Brookings Institution and others have presented a detailed plan under which the United States could unilaterally reduce its nuclear warheads to about 600 and immobilize (for instance by pit stuffing) or dismantle the rest.¹² The specific mix of measures taken to de-alert nuclear weapons will depend on their design, on verification measures, and on the delays that are to be introduced by the de-alerting. For instance, a retaliatory deterrence posture does not require as many as eight or even four warheads per SLBM. The number of warheads per SLBM could be reduced to one, and the nature of the warheads changed from large yield (in the hundreds of kilotons) to relatively low-yield. The number of submarines on patrol at any one time could also be greatly reduced. These measures, if undertaken unilaterally by the United States, would greatly increase confidence on the part of Russia, whose submarine fleet cannot be maintained at sea at anywhere near full strength due to lack of funds. Specifically, it would increase the likelihood of Russia's participation in a global de-alerting process.

All tactical nuclear weapons as well as nuclear bombs can be de-alerted by storing the warheads apart from the delivery systems, and establishing physical as well as remote technical means of verification. This can be done rapidly, unilaterally, and in the near term, without any coupling to any other measure. Withdrawal of all weapons from foreign bases is another de-alerting measure for nuclear weapons that would enhance the prospects of complete disarmament. At present, only the United States has nuclear weapons based abroad. An estimated 150 US warheads are based in Europe. The unilateral measures most likely to reduce Russian concerns would be de-alerting of weapons based near Russia, such as those in Europe and in the Arctic region.

Medium and Long-Term Steps

De-alerting as a part of an approach to enduring nuclear disarmament should be carried out in phases. We have laid out some steps of the first phase above. The next phase, described below, would be very close to disarmament, but may be better described as "deep de-alerting." The basic technical approaches are similar to short-term steps, but are more complete. As they would involve greater multilateral verification and control, they would be politically and technically more difficult to reverse. Both phases would enhance efforts to reach a much longer term third stage, a state of enduring nuclear disarmament that is far more resistant to attempts to reverse it in time of crisis.

Medium-term de-alerting measures include:

• removal and remote storage of all warheads separately from delivery systems under multilateral monitoring;

• storage of all guidance systems at locations remote from delivery systems under multilateral monitoring;

• multilateral verification of all materials accounts for weapons-usable materials to ensure compatibility of warhead declarations, numbers of de-alerted warheads, and stored weapons-usable materials. The technical aspects of these measures could take many months to a few years to complete, as time will be needed to design and implement verification systems, and design and construct storage facilities.

Interim-alerting measures would be enhanced if all eight states possessing nuclear weapons would put all their commercial and military weapons-usable fissile materials that are not actually in warheads or in classified shapes under the safeguards system of the International Atomic Energy Agency (IAEA). In any case these materials should be put into some kind of bilateral or multilateral accounting, monitoring and verification scheme (such schemes often go under the rubric of "transparency measures").

The longest-term de-alerting approaches slide into disarmament measures. They include dismantlement of warheads and storage of all weapons-usable fissile materials under IAEA safeguards or in non-weapons usable forms. It would take one or more decades to accomplish the technical requirements of these measures, depending on the technologies chosen to implement them. They would be considered part of a de-alerting process if the facilities to reconstitute nuclear arsenals are maintained. They would be part of nuclear disarmament if the warhead and associated materials production and processing facilities are also dismantled.

1. An early article advocating de-alerting was: Bruce G. Blair and Henry W. Kendall, "Accidental Nuclear War," Scientific American, Vol. 253, December 1990, pp. 53-58.

2. According to Bruce Blair, "only Russia and the U.S. today maintain [their nuclear] forces on high combat alert under normal conditions. The others maintain a de facto policy of de-alert." Personal e-mail communication, 30 August 1998. That might change if Pakistan carries out its threat to arm its delivery vehicles and if India follows suit.

3. Statement of Brazil, Egypt, Ireland, Mexico, New Zealand, Slovenia, South Africa, and Sweden, 9 June 1998.

4. Arjun Makhijani and Katherine Yih, "What to Do at Doomsday's End," The Washington Post, 29 March 1992.

5. Bruce G. Blair, Harold A. Feiveson. and Frank N. von Hippel, "Taking Nuclear Weapons Off Alert," Scientific American, November 1997, pp. 74-81.

6. Bruce G. Blair, Global Zero Alert for Nuclear Forces, (Washington: Brookings Institution, 1995), pp. 90-107.

7. Ibid., p. 87.

8. Richard L. Garwin, "De-alerting of Nuclear Retaliatory Forces," presented to the Amaldi Conference, Paris, France, November 20-22, 1997.

9. Blair, 1995, op cit., pp. 88-89.

10. The idea of removing tritium bottles from nuclear weapons has been proposed as a qualitative de-alerting measure by Martin Kalinowski in "Qualitative Disarmament by Tritium Control," INESAP Information Bulletin, Issue No. 15, April 1998, p. 48. In an earlier paper, Kalinowski and Colschen have calculated that the yields of various US warheads would be reduced from a typical level of several hundred kilotons to yields in the range of a few hundred tons to a few kilotons of TNT equivalent in all but one case. In the case of the W89, the removal of the tritium bottle would cause the warhead not to operate. The overall effect of removing tritium from all warheads is estimated to be a reduction of yield by two orders of magnitude or more. See Martin B. Kalinowski and Lars C. Colschen, "International Control of Tritium to Prevent Horizontal Proliferation and to Foster Nuclear Disarmament," Science and Global Security, Vol. 5, 1995, pp. 131-203.

11. Matthew Bunn, "'Pit-Stuffing:' How to Disable Thousands of Warheads and Easily Verify Their Dismantlement," and Richard L. Garwin, "Comment on Matt Bunn's 'Pit-Stuffing' Proposal," in F.A.S. Public Interest Report, Journal of the Federation of American Scientists, Vol. 51, No. 2, March/April 1998. Available at: "www.fas.org/faspir/pir0498.htm."

12. Blair, Feiveson, and von Hippel, 1997.