AYEAR and a half ago the Western powers began negotiations with the Soviet Union for a treaty to end the testing of nuclear weapons. The negotiations were warmly welcomed by the public, and by scientists in particular. It appeared to many people that this was one area of disarmament in which agreement should not be too difficult to reach. The generally optimistic feeling about the negotiations had its origin in three widely held beliefs. It was believed that the development of nuclear weapons had reached a point of technical stagnation; that the military consequences of new inventions in this field would be small; and that a political agreement to cease further development could be adequately controlled by a system of long-range explosion-detectors. These views have been repeatedly expressed by scientific experts, and many citizens and politicians have come to accept them without serious question.

It is my purpose here to argue the contrary views. I believe that radically new kinds of nuclear weapons are technically possible, that the military and political effects of such weapons would be important, and that the development of such weapons can hardly be arrested by any means less drastic than international control of all nuclear operations.

This by no means implies that negotiations to end weapon-testing should be abandoned. If my views are correct, then to end weapons development becomes vastly more difficult to achieve than had been thought, but it also becomes vastly more rewarding. Any agreement to cease weapon-testing upon a sound and verifiable basis will require an opening of communications and a breaking down of barriers among the nuclear laboratories of the world. The aim of our negotiations will thus be not merely the prohibition of weapon testing, but the establishment of an open world. To many thoughtful people, since the early days of nuclear energy, the great hope has been that nuclear weapons would compel mankind to coöperate not only in nuclear affairs but in other areas too. This dream could well be realized if the negotiations which were begun in order to control weapon tests should end in a system adequate to control more substantial disarmament.


Until now we have had two essentially different types of nuclear weapons, the fission bomb and the hydrogen bomb. It is not necessary to discuss in detail how these things work. The essential facts about their performance are the following. A fission bomb cannot explode at all unless it contains a certain quantity (the critical mass) of extremely expensive metal. Thus every fission bomb, regardless of its power, costs a certain fraction of a million dollars and eats up a certain fraction of the available supply of fuel. Every fission bomb using its fuel with reasonable efficiency has an explosive power of the order of kilotons of T.N.T. A hydrogen bomb is able to extract its energy from a much cheaper and more abundant fuel (heavy hydrogen), but it requires at least a moderately efficient fission bomb to ignite it. Thus every hydrogen bomb costs at least as much as a fission bomb. Its cheapness becomes impressive only when its explosive power moves toward the megaton range.

So the basic built-in characteristic of all existing weapons is that it is relatively much cheaper to make a big bang than a small one. Below a certain explosive yield of the order of a kiloton, nuclear weapons are grossly inefficient and extravagant. However, for military purposes other than wholesale annihilation, a kiloton is already an unreasonably big bang. There is a clear and acute military need for an explosive which would fill the gap between a ton and a kiloton of T.N.T. with a cost which is proportional to the yield instead of being independent of it.

There is theoretically a simple way to escape from the tyranny of the critical mass. This is to burn heavy hydrogen without a fission bomb to ignite it. A fission-free bomb, containing a small quantity of heavy hydrogen and no fissionable metal, is logically the third major step in weapon development after the existing fission and hydrogen bombs. Such a bomb has been occasionally mentioned in newspapers and magazines and described as a "100-percent clean bomb." It would not be 100-percent clean. It would contaminate the countryside enormously less than existing fission or hydrogen bombs, but this is not its main advantage. The decisive advantage of a fission-free bomb is that it could be built economically in small sizes. It would have no critical mass. It would provide without gross inefficiency an explosive power adapted to the needs of small-scale and local warfare.

There seems to be no law of nature forbidding the construction of fission-free bombs. The question remains whether this theoretical possibility is likely soon to be realized. In this connection some sentences from a report by L. S. Artsimovitch entitled "Research on Controlled Thermonuclear Reactions in the U.S.S.R." are interesting. The report was published in Uspekhi Fizicheskikh Nauk in December 1958 and is now available in translation. The quotation is condensed and some technical phrases are omitted.

It may also be possible to realize a pulsed thermonuclear reaction under conditions in which the high temperature is produced by a charge of conventional explosive (such as T.N.T. or something more powerful) which surrounds a capsule containing heavy hydrogen. Without dwelling on the experimental details, we may note that conditions have been found under which the generation of neutrons in hydrogen reactions has been established reliably and reproducibly. In experiments carried out in 1952, there is no doubt that we have observed neutrons which are formed as a result of the heating of matter to extremely high temperatures. This process takes place under conditions in which the density of matter is very high (significantly exceeding the normal density of solids).

Artsimovitch is a well-known and first-rate physicist who played a prominent part in the Soviet weapons program (at the appropriate times he received Stalin and Lenin Prizes). These remarks which he published are an indiscretion, and presumably a calculated indiscretion. It is at least highly interesting for us to know that the Russians were experimenting with fission-free explosive systems in 1952. No doubt somebody on the Russian side hopes by such a minor indiscretion to stimulate a more serious indiscretion on our side. This is the kind of poker game we are playing.

I shall not be so indiscreet as to say anything about the more recent history of fission-free bombs. What I wish to make clear is that these bombs are a theoretical possibility, and that their importance is well understood by at least some individuals on both sides of the iron curtain. Any political arrangement which fails to take these facts into account is doomed to failure.

The future growth of nuclear explosion technology will have military importance in many ways which are hard to foresee now. There are many areas of weaponry in which important problems remain unsolved. The unsolved problems are not, as is often claimed, to make slight or marginal improvements in existing designs. There are several directions in which qualitatively new methods of design may lead to qualitatively superior weapons, or to weapons performing qualitatively new functions. To give only one example, it is possible that a nuclear explosive system designed for the efficient propulsion of a space ship could be the key to long-range and economical space travel. Any country possessing a markedly superior propulsion system in space would reap important advantages, both military and non-military.

To illustrate as forcibly as possible the military importance of new weapons, I return to the example of fission-free systems. Imagine a hypothetical situation in which the United States is armed with its existing weapons, while some adversary (not necessarily the Soviet Union) has a comparable supply of nuclear fuel and has learned how to ignite it fission-free. The adversary's bombs would then outnumber ours ten or a hundred to one, and theirs could be used with far greater versatility in infantry warfare. Suppose that in this situation a local war of the Korean type should begin. God help the American infantryman who is sent in to fight against these odds. Practically speaking, our army would have only two alternatives, either to retreat precipitously or to strike back with our much more limited number of heavier nuclear weapons and thoroughly destroy the whole country. This is not a pleasant situation to contemplate, and yet it is necessary that our people understand that it is a possibility. Any country which renounces for itself the development of nuclear weapons, without certain knowledge that its adversaries have done the same, is likely to find itself in the position of the Polish Army in 1939, fighting tanks with horses.


Next let us turn to the stubborn subject of long-range explosion-detectors and their effectiveness for policing an international agreement to halt weapon testing.

A conference was held at Geneva in the summer of 1958 at which scientific experts from East and West met and discussed the problems of long-range detection of explosions. The records of the existing long-range detectors and the responses of these detectors to past weapon tests were examined in detail. After some strenuous argument, the conference agreed upon a statement of conclusions. The statement describes a particular system of detectors which is suggested as a basis for a future international control of weapon-testing. This system, now generally known as the "Geneva system," consists principally of 180 control posts equipped with a variety of scientific instruments and distributed over all parts of the earth. The statement concludes "that it is technically feasible to establish, with the capabilities and limitations indicated below, a workable and effective control system to detect violations of an agreement on the world-wide suspension of nuclear weapons tests."

The agreement of the Geneva "conference of experts" was widely hailed as a shining example of the triumph of scientific objectivity over national and political differences. In many ways it was indeed a triumph. For the first time since the advent of nuclear weapons, men from East and West were sitting together to discuss the nuclear facts of life with some degree of openness. And the speed with which this technical conference could reach an agreement was in startling contrast to the interminable sterility of earlier negotiations with the Soviet Union.

The experts at Geneva in 1958 did not know to what extent artificial concealment of nuclear explosions was technically possible. Probably for this reason they decided to say nothing about it. Scientific objectivity would, however, have required them to report that the concealment problem had not been explored.

Since the summer of 1958, political negotiations between the Soviet Union and the Western powers have been in progress, aiming at an agreement to cease weapon tests. Throughout these negotiations, the Soviet delegates have adamantly held to the line that the Geneva system could effectively police such an agreement. They have refused even to enter into any realistic discussion of the concealment problem. Meanwhile, a small number of Americans have started belatedly to think about concealment, and some small non-nuclear experiments have been carried out. It has gradually become clear to the American experts that a high degree of concealment is theoretically possible. In a report issued in June 1959 by a panel appointed by the Science Advisory Committee of the United States Government to study the possible improvement of the Geneva detection system, it was stated:

In considering the possibility that the capabilities, now or in the future, of the Geneva System might be reduced by the intentional concealment of underground tests, the panel concluded that decoupling techniques existed which could reduce the seismic signal by a factor of ten or more. Moreover, preliminary theoretical studies have shown that it is possible in principle to reduce the seismic signal from an explosion by a much greater factor than this.

The technical facts concerning concealment were held secret until December 1959. It will still be some time before the facts are published in detail and exposed to the scrutiny of the international community of scientists. Only after such public scrutiny, and a great deal more experimental work, will it become possible to make an objective judgment of the effectiveness of concealment. Meanwhile, the limits of uncertainty remain very wide.

My personal opinion is that nuclear explosions in the kiloton range could be concealed very completely. That is to say, I believe it will be feasible to build a building, looking externally like a normal industrial structure, within which kiloton explosions can be contained. The cost of such a facility may not be exorbitant, and the earth tremors which it produces may be no larger than those produced by ordinary industrial operations. Giving free rein to the imagination, one may envisage a weapon testing facility bearing on the outside the inscription "Kazakhstan Consolidated Steel Mills" and carrying on a legitimate business of steel fabrication as a side line. Concealment of this sort could be detected only if the international control authority were empowered to travel everywhere and to open all doors.

So long as the concealment problem has not been thoroughly and openly investigated, the effectiveness of long-range detectors against concealed explosions will be known only to those governments which conduct their own clandestine experiments. Under these conditions the idea of international control of explosions by long-range detectors is a dangerous illusion. If, as I suspect, the total concealment of explosions is possible, then a long-range detection system can never guarantee the observance of a test-cessation agreement. No party to the agreement can be sure that his adversaries are not secretly and successfully carrying on the development of new weapons. Especially dangerous is the circumstance that many of the militarily important new weapons will require test explosions only of very low yield.

If we sign an agreement to cease weapon-testing on the basis of a long-range detection system, ignoring the possibilities of concealment, we are trusting to the good faith of all the signatory powers. Now it can be argued that an agreement with the Soviet Union based upon good faith is not worthless. One need not assert that the Soviet Union would feel inclined to violate such an agreement whenever it were technically possible to do so with impunity. It is enough that a world-wide agreement to stop testing, based on good faith alone, would be very unlikely to last forever. Somewhere, sometime, some government would yield to temptation and resume testing in secret.

Public opinion in the West has a terrifying readiness to take refuge in illusions. For some years after 1945 we took refuge in the illusion that our preponderance in nuclear weapons would be more or less permanent. Now many of us are taking refuge in the illusion that the development of nuclear weapons can be stopped all over the world by a painless international agreement and a few hundred seismographs. It is necessary that we face facts resolutely. One fact is that militarily important weapons can be developed in secret by any government keeping a large part of its territory closed to foreigners. Another fact is that, in the absence of an effective international control of armaments, the pressures driving other governments to surpass ours in the technology of nuclear weapons will be almost irresistible.

To ensure that new weapons are not developed, there must exist an international force of detectives with unrestricted rights of travel and inspection. Such an international detective force is at present unacceptable to the Soviet Union. For this reason it seems inevitable that the development of nuclear weapons will continue, either openly or in secret, for many years to come. For the peace of the world, it is far better that the new weapons should be developed openly than in secret. We must forever be thankful that the unconcealed testing of large hydrogen bombs has made the power of these weapons known to everyone.

The only wise policy for the United States at present is to continue the exploration of nuclear weapons technology, including the testing of weapons, until a reliable international control of testing is established. We should not be discouraged by the fact that the Soviet Union will not now accept such a control. In the last ten years the Soviet Union has come a very long way toward accepting the unpleasant facts of life in the nuclear age. It is quite possible that in another ten years the pressure of logic will carry the Soviet leaders a great deal further. In particular, it is possible that the Soviet authorities may in ten years recognize the fact that a rigid international control of all nuclear activities is the only guarantee of their safety as well as of ours.


To many people the continuation of any sort of nuclear weapons development is morally repugnant. To these people my arguments will carry no conviction. They will feel that these arguments merely reveal a moral blindness and an insensitivity to the overwhelming evil of nuclear warfare.

Everyone is agreed that war is evil and that nuclear war is surpassingly evil. And we live in a world where independent sovereign states have undisputed power over stockpiles of nuclear weapons. What then is to be done? I find no escape from the logic which impelled Lilienthal and his colleagues in 1946 when they drew up their plan for the international control of nuclear energy. If nuclear war is to be permanently avoided, it is necessary that somehow, sooner or later, sovereign states voluntarily hand over their nuclear armaments to an international authority. And the international authority must control everything, not only weapon tests but also production facilities, stockpiles and research.

The prospect of an adequate international control of armaments is remote, although possibly not as remote as it seems. Our whole endeavor must now be directed to stay alive somehow and to maintain the stability of the world through the interim period while international anarchy prevails.

During the period of international anarchy, we have a moral choice, either to strive to the utmost to remain in the forefront of nuclear weapons technology or to let the leadership pass into the hands of others. These are the only two alternatives before us. We do not have the power to change the laws of nature so that the possibilities for developing new weapons will disappear. Our moral choice is, either to possess the new weapons ourselves, or to leave it to chance to decide who shall possess them.

The moral dilemma facing the designers of weapons today is not essentially different from that which faced the builders of the fission bomb in 1943 and the builders of the hydrogen bomb in 1951. In each case the choice is the same. You discover that a new and horrible kind of weapon can be made. Either you make it yourself, or you leave it to chance to decide who makes it.

I believe the morally right decisions were made in 1943 and 1951. And I believe the morally right decisions now must be the same. Apart from all considerations of patriotism, it would be wrong for us to leave the future development of nuclear weapons to chance. Our development and possession of these weapons will help to maintain the stability of the world, until in the fullness of time we can hand over all such devilish inventions to an international authority powerful enough to prevent their abuse.

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
  • FREEMAN J. DYSON, Professor at the Institute for Advanced Study, Princeton; Professor of Physics, Cornell University, 1951-53
  • More By Freeman J. Dyson