The United States possesses an enormous nuclear arsenal, left over from the days of the cold war. We have about 6,000 operationally deployed nuclear weapons,1 of which roughly 2,000 are on intercontinental ballistic missiles, 3,500 on submarine-launched ballistic missiles, and a few hundred carried by bomber aircraft. These are thermonuclear weapons, considerably more powerful than the fission bombs that devastated Hiroshima and Nagasaki. Looking over these figures, one can hardly help asking, what are all these nuclear weapons for?

There was a rationale for maintaining a very large nuclear arsenal during the cold war: we had to be sure that the Soviets would be deterred from a surprise attack on the US by their certainty that enough of our arsenal would survive any such attack to allow us to deliver a devastating response. I don’t say that US strategic requirements were actually calculated in this way, but the need for such a deterrent at least provided a rational argument for a large arsenal.

This rationale for a large nuclear arsenal is now obsolete. No country in the world could threaten our submarine-based deterrent, and even with an implausibly rapid development of nuclear weapons and missiles, for decades to come no country except Russia will be able to threaten more than a tiny fraction of our land-based deterrent.

Russia maintains a nuclear arsenal of a size similar to ours, though with a different mix of delivery vehicles. On May 24, 2002, Presidents Bush and Putin signed a treaty calling for a reduction in operationally deployed nuclear weapons on both sides to about 3,800 in 2007 and to about 1,700 to 2,200 in 2012. This treaty will almost certainly be ratified by the Senate; Democrats will generally be glad of any reduction in nuclear arms, and few Republicans will want to oppose President Bush on a matter of foreign relations. President Bush has said, “This treaty will liquidate the legacy of the cold war.” But any celebration would be premature, for there is far less to this treaty than meets the eye.

For one thing, the rate of reduction is painfully slow. The START III agreement that was announced (though not signed or ratified) by Presidents Clinton and Yeltsin called for a reduction to about 2,000 to 2,500 “strategically deployed” nuclear weapons by 2007, not by the 2012 deadline of the Bush– Putin treaty. (The term “strategically deployed” differs from “operationally deployed” in including all weapons that are associated with delivery systems, whether or not they are actually ready to fire. Thus for instance the nuclear warheads of missiles on a submarine in dry-dock would be included on the list of strategically deployed weapons but not of operationally deployed weapons. When this difference is taken into account, the limit of 2,000 to 2,500 strategically deployed weapons in 2007 set by the START III agreement is the same as the limit of 1,700 to 2,200 operationally deployed nuclear missiles in 2012 set by the Bush–Putin treaty.) The treaty is highly reversible; either party can withdraw with forty-five days’ notice, and unless renewed the treaty will expire in 2012. Also, unlike former arms control agreements, the Bush–Putin treaty would not call for the destruction of missiles or bombers, only for the removal of their nuclear warheads or bombs.

Most important, the Bush administration has turned back Russian efforts to require in this treaty that the nuclear weapons withdrawn from deployed missiles and bombers should be destroyed. The Defense Department’s plans for nuclear weapons have been laid out in a classified Nuclear Posture Review,2 dated January 9, 2002, of which large sections were leaked a few months later to the press.3 The plans laid out in this review call for the retention of about 7,000 intact warheads that are not operationally deployed, not to mention a large number of plutonium “pits” (the fission bomb that triggers a thermonuclear explosion) and other weapon components. Of course, the treaty does not call for the destruction of demobilized Russian nuclear weapons either, and it does not constrain Russian nuclear tactical weapons, so it actually increases the danger that some Russian weapons could fall into the hands of rogue states or terrorists. This treaty is not designed to “liquidate the legacy of the cold war,” as President Bush claims, but to hold on to as much of that legacy as possible. Even taking into account the reductions called for by the Bush–Putin treaty, we are left still wondering what all these nuclear weapons are for.

There is one possible use of a large American nuclear arsenal: to launch a preemptive attack against Russian strategic weapons. I say “Russian,” because the large size of our arsenal would be irrelevant for a preemptive attack against any other power. If we ever found that a hostile “rogue” state were about to deploy a few dozen nuclear-armed ICBMs, and if we could locate them, then they could be destroyed by only a tiny fraction of our nuclear arsenal, and in fact even by conventionally armed cruise missiles. On the other hand, even though we were unable to neutralize the Soviet deterrent during the cold war, now as Russian nuclear forces become increasingly immobile, with their missile-launching submarines tied up at dockside and their land-based mobile ICBMs kept in fixed garrisons, our large nuclear arsenal may put Russian nuclear forces at risk of being destroyed by a preemptive US strike. In the letter of transmittal of the Nuclear Posture Review to Congress, Secretary Rumsfeld said that “the US will no longer plan, size, or sustain its forces as though Russia presented merely a smaller version of the threat posed by the former Soviet Union.” But that appears to be just what we are doing.


It might seem that the ability to launch a preemptive strike against Russian strategic nuclear forces is a good one to have, but in fact it poses enormous dangers, and to us as well as to Russia. The Russians can count missiles as well as we can, and as “prudent” defense planners they are likely to rate our chances of a successful preemptive attack more highly than we would. Even though they may understand that the US now has no plans for such a preemptive attack, they are bound to consider the possibility that this could change if relations between Russia and the US sour in future. This possibility is likely to seem more probable if the US proceeds with a national missile defense, which might be perceived to have some effectiveness against a ragged Russian second strike, or if we follow the recommendation of the Nuclear Posture Review that the US should develop real-time intelligence capabilities of a sort that would allow us to target even mobile Russian missiles on the road.

The danger is not that the Russians will get angry with us, or plan to attack us. The danger is that they will quietly adopt a cheap and easy defense against a preemptive American attack, by keeping their forces on a hair-trigger alert. This presents the US with the threat of a large-scale Russian attack by mistake during some future crisis; for instance, the Russians may receive misleading warnings of an imminent American attack and launch their own nuclear weapons before they can be destroyed on the ground. (According to Russian sources, it now takes fifteen seconds for the Russians to target their ICBMs, and then two to three minutes to carry out the launch.) This danger is exacerbated by the gradual decay of Russia’s capabilities for surveillance of possible attacks and control of their own forces, a decay that has already led them on one occasion to mistake a Norwegian research rocket for an offensive missile launched from an American submarine in the Norwegian sea.4

For those who may think that this is a paranoid worry, perhaps left over from cold war movies like Fail Safe or Dr. Strangelove, it is instructive to look back at mistakes made by American strategic forces during the Cuban missile crisis, the most dangerous crisis of the cold war5:

(1) On August 23, 1962, a navigational error led a B-52 bomber on airborne alert—i.e., ready to retaliate if the US were attacked—which was supposed to be on a nonprovocative course heading over the Arctic Ocean toward Alaska, to head instead directly toward the Soviet Union. Its error was noticed when the bomber was only three hundred miles away from Soviet airspace. Despite this incident, and the well-known difficulties of navigation above the Arctic Circle, the routes of US bombers on airborne alert were not changed for months, not until after the October missile crisis. Luckily no similar navigational errors were made by our bombers during the missile crisis.

(2) On October 26, 1962, when US and Soviet forces were already at a heightened state of alert, an intercontinental ballistic missile was launched from Vandenberg Air Force Base, as part of a test program that no one had thought to cancel. We do not know if the Soviets detected this launch, but they might have.

(3) The Cuban missile crisis happened to come at a time when new Minuteman I missiles were being installed at Malmstrom Air Force Base in Montana. In order to get these missiles ready for possible launch, Air Force and contractor personnel apparently bypassed safeguards that had been designed to prevent a launch by a single officer. Fortunately no officer decided to launch the missiles under his control.

We don’t know what mistakes may have been made on the Soviet side. Whatever mistakes were made on either side did not lead to war, but this was evidently not because national leaders are able to completely control their forces under crisis conditions. As President Kennedy said during the Cuban missile crisis, “There is always some son-of-a-bitch who doesn’t get the word.”


Even though the threat of a large Russian mistaken attack is not acute, it is chronic. It is also the only threat we face that could destroy our country beyond our ability to recover. Compared with this threat, all other concerns about terrorism or rogue countries shrink into insignificance.

This brings me to the one real value of our large nuclear arsenal: we can trade away most of our arsenal for corresponding cuts in Russian forces. I don’t mean cuts to about two thousand deployed weapons, but to not more than a few hundred deployed weapons on each side, and with each side having not more than a thousand nuclear weapons of all sorts, including those in various reserves, as called for by a 1997 report of the Committee on International Security and Arms Control of the National Academy of Sciences.6 In that way, although the danger of a mistaken Russian launch would not be eliminated, the stakes would be millions or tens of millions of casualties, not hundreds of millions.

Such cuts would also reduce the danger that Russian nuclear weapons or weapons material could be diverted to criminals or terrorists. Instead of seeking the maximum future flexibility for both sides in strategic agreements with the Russians, we should be seeking the greatest possible irreversibility on both sides, based on binding ratified treaties. We ought also to be spending more on the program, originally sponsored by former Senator Sam Nunn and Senator Richard Lugar, that assists the Russians in controlling or destroying their excess nuclear materials. At this moment, when the Russians are eager to improve relations with the West, when considerations of economics provide them with a powerful incentive to reduce their nuclear forces, and when for the first time they have a president powerful enough to push such reductions through their military and political establishments, we have an unprecedented opportunity to begin to escape from the risk of nuclear annihilation. It is tragic that we are letting this opportunity slip away from us.


Not only are we not moving fast or far enough in the right direction—in some respects the Bush administration seems to be moving in just the wrong directions. One example is the abrogation of the 1972 treaty limiting anti-ballistic missile systems.7 Another example is the revival of the idea of developing nuclear weapons for use, rather than solely for deterrence.

For instance, the Nuclear Posture Review calls for the development of low-yield, earth-penetrating nuclear weapons for attacks on underground facilities, such as biological warfare laboratories in countries like Iraq. There are great technical difficulties here, which might prevent our using such a weapon even if we had it. When dropped from a bomber, our present earth-penetrating weapon, the B61-11, has penetrated only about ten feet into frozen tundra. The depth of penetration can be increased by accelerating the weapon down to the surface with a rocket; but increasing the velocity of impact beyond a certain point just causes the weapon to crumple, so that instead of the depth of penetration increasing, it decreases. Recent calculations show that an earth-penetrating weapon cannot be driven down to a depth greater than about four times its length in concrete.8 This sets an upper limit on the depth of penetration of about eighty feet for a weapon that is twice the length of our B61-11. The actual depth that may be reached in practice may be considerably less, because the velocity of impact must be kept low enough to preserve the weapon’s electrical circuits.

It is true that an eighty-foot depth is sufficient to put most of the energy of the explosion into a destructive underground blast wave, which can destroy facilities below the actual explosion, but even so, a one-kiloton explosion would only destroy tunnels that are at depths considerably less than three hundred feet, and not much more than that in a horizontal direction; the precise ranges are sensitive to geological details that we are not likely to know. An earth-penetrating nuclear weapon would be effective only against an underground target that is not too deep, and whose location is accurately known. To have confidence that the underground target had been destroyed we would have to have troops on the ground anyway, so that a missile attack might not even be necessary.

Even if an earth-penetrating nuclear weapon could destroy its target, we would be unlikely to use it because of radiation effects. In the 1950s a project known as Plowshare exploded a number of nuclear devices at various depths underground, with the hope of developing peaceful uses for nuclear explosions, like digging canals. Experience in these tests showed that to keep a nuclear explosion from breaking through the surface and spreading radioactive dirt into the atmosphere, a one-kiloton explosion would have to be kept below three hundred feet, with the depth required decreasing only slowly as the yield is decreased. (The penetration of a weapon through the earth would create a shaft to the surface, something that did not exist in the Plowshare tests, so the depth required to avoid fallout is bound to be even larger than indicated by these tests.) To avoid fallout from a nuclear explosion at a depth of only eighty feet it would be necessary to reduce the yield to nineteen tons of TNT, not much more than could be delivered using conventional explosives. I don’t believe that there is any way for a nuclear weapon with a yield greater than a few tenths of a kiloton to penetrate to depths sufficient to avoid producing a great deal of radioactive fallout, without someone carrying it down in an elevator.

The fallout produced by a one-kiloton explosion at a depth of eighty feet would kill everyone on the surface within a radius of about half a mile. This estimate is for fallout under conditions of still air; wind could carry the fallout for tens of miles. We could be killing not only the local population, which (as in Afghanistan) we might be trying to enlist on our side, but also whatever forces we or our allies have on the ground.

There was another sign of increased interest in developing nuclear weapons for actual use in a recent statement by William Schneider, the chairman of the Defense Science Board. He announced a renewed study of nuclear-armed interceptor missiles as part of a system of missile defense. Nuclear-armed missile defense interceptors would have technical and political problems of their own, problems that have led to the abandonment of nuclear-armed interceptors as components in missile defense since the administration of Ronald Reagan.

For the dubious advantages of developing new nuclear weapons, we would pay a high price, including pressure for resumed testing of nuclear weapons. As I mentioned, calculations indicate that any nuclear weapon that would be effective against underground targets would release large quantities of radioactivity. Even if the depth of penetration of a nuclear weapon were somehow increased and the yield decreased enough so that no fallout was expected, how, without testing these weapons in action, could anyone ever have confidence that fallout would not escape, especially after the US weapon has created its own shaft to the surface? And how could anyone have confidence in a missile defense system based on nuclear-armed interceptors without tests that involve nuclear explosions in or above the atmosphere? We have not carried out even underground tests since the previous Bush administration. And, as is very much in our interest, neither has Russia or China.

The development of new nuclear weapons for war-fighting would in itself violate our commitment under the 1970 Nuclear Nonproliferation Treaty to deemphasize the role of nuclear weapons and to work toward their total elimination. The resumption of nuclear testing for this purpose would make this violation concrete and dramatic, and would thereby gravely undermine the effectiveness of the Nonproliferation Treaty in discouraging nuclear weapons programs throughout the world.

A special danger of programs to develop nuclear weapons for use is that they may stand in the way of a really large-scale mutual reduction of nuclear arms. I’m not sure whether we are retaining a huge nuclear arsenal in order to facilitate such new weapons programs, or whether the weapons programs are being proposed in order to slow down cuts in our nuclear arsenal. Probably something of both. Back in the days when the first test ban treaty was being debated, one of the arguments against it was that it would stand in the way of Project Plowshare and also Project Orion, the development of a spacecraft propelled by nuclear explosions. (Both programs have long since been abandoned.) The development of nuclear weapons for attacking underground facilities or for missile defense may be today’s Orion and Plowshare.

But the current proposals for new nuclear weapons are much more dangerous than the Plowshare or Orion programs. As the world’s leader in conventional weaponry, we have a very strong interest in preserving the taboo against the use of nuclear weapons that has survived since 1945. Developing and testing new nuclear weapons for actual use rather than deterrence teaches the world a lesson that nuclear weapons are a good thing to have. This is not entirely a rational matter. I remember that once in the late 1960s I had lunch at MIT with the chief scientific adviser to the government of India. I asked about India’s plans for developing and testing nuclear weapons, and he said that it all depended on whether the US and USSR could reach an agreement banning all future nuclear testing. I said that that seemed irrational, because it was not the US or the USSR that presented a military threat to India, and even if such a threat did develop, American and Soviet nuclear forces would in any case be so much greater than India’s that it would not matter to India if the US or the USSR had stopped testing or gone on testing.

The Indian science adviser answered that politics is not always based on rational calculations, that there was great political dissension in Indian governing circles over whether to develop nuclear weapons, and that the spectacle of continued testing of nuclear weapons by the US or the USSR would strengthen the hands of those in India who favored developing nuclear weapons. Of course, the US and USSR did not stop testing at that time; India did develop nuclear weapons; and Pakistan followed suit. Is it likely that resumed US nuclear testing would have no effect on decisions about nuclear weapons in countries like Japan, or Egypt, or Germany? Is it likely that the Nonproliferation Treaty will survive when the US is developing and testing nuclear weapons for actual use?

After the signing of the Bush–Putin treaty, President Bush was asked why it was necessary for us to keep 2,000 nuclear weapons loaded on missiles. He answered that the future was uncertain. The same argument is often made to defend the development of new nuclear weapons. It is true that the future is uncertain, but where is it written that the way to reduce uncertainty is always to maximize our nuclear capabilities? We cannot tell what crisis may occur in US–Russian relations, a crisis that could put the US at risk from a mistaken launch on their part. We cannot tell what terrorists may take over or steal part of the Russian arsenal. We cannot tell what dangers we may face from a large Chinese arsenal, built to preserve their deterrent from the threat of an American first strike backed up by a missile defense system. We cannot tell what countries may be tipped toward a decision to develop nuclear weapons by new US weapons programs or resumed nuclear testing. There is no certainty whatever we do. We have to decide what are the most important dangers, and these dangers may be increased rather than decreased by other countries’ responses to our own weapons programs. The Nuclear Posture Review strikingly fails to consider what other countries might do in response to our plans for nuclear weapons.

At the beginning of the twentieth century, Britain was overwhelmingly the world’s greatest naval power, much as the US is today the world’s leader in conventional arms. Then in 1905 Admiral Sir John Fisher, the First Sea Lord, pushed forward the construction of a new design for a fast battleship armed solely with twelve-inch guns, the biggest guns then available. The prototype was named after one of Nelson’s ships, the Dreadnought. Dreadnoughts really were superior to all previous battleships, and suddenly what counted was not the size of a country’s fleet, in which Britain was supreme, but the number of its Dreadnoughts. Other countries could now compete with Britain by building Dreadnoughts, and a naval arms race began between Britain and Germany, in which Britain would stay ahead only with great expense and difficulty. Admiral of the Fleet Sir Frederick Richards complained in Parliament that “the whole British fleet was morally scrapped and labeled obsolete at the moment when it was at the zenith of its efficiency and equal not to two but practically to all the other navies of the world combined.”9 Like Dreadnoughts, nuclear weapons can act as an equalizer between strong nations like the US, with great economic and conventional military power, and weaker countries or even terrorist organizations. It should be clear by now that national security is not always best served by building the best weapons.

As a scientist, I can recognize a kind of technological restlessness at work, from the building of the Dreadnought to this year’s Nuclear Posture Review. Years before he pioneered the Dreadnought, as a newly appointed captain in charge of the Royal Navy’s torpedo school, Fisher explained that “if you are a gunnery man, you must believe and teach that the world is saved by gunnery, and will only be saved by gunnery. If you are a torpedo man, you must lecture and teach the same thing about torpedoes.” There is nothing corrupt or unpatriotic about such attitudes, but their consequences could be catastrophic.

—June 19, 2002

This Issue

July 18, 2002