Thomas Nagel - War and Massacre

Philosophy and Public Affairs, Vol. 1, No. 2 (Winter, 1972), 123-144.

{123} From the apathetic reaction to atrocities committed in Vietnam by the United States and its allies, one may conclude that moral restrictions, on the conduct of war command almost as little sympathy among the general public as they do among those charged with the formation of U.S. military policy. Even when restrictions on the conduct of warfare are defended, it is usually on legal grounds alone: their moral basis is often poorly understood. I wish to argue that certain restrictions are neither arbitrary nor merely conventional, and that their validity does not depend simply on their usefulness. There is, in other words, a moral basis for the rules of war, even though the conventions now officially in force are far from giving it perfect expression.


No elaborate moral theory is required to account for what is wrong in cases like the Mylai massacre, since it did not serve, and was not intended to serve, any strategic purpose. Moreover, if the participation of the United States in the Indo-Chinese war is entirely wrong to begin with, then that engagement is incapable of providing a justification for any measures ,taken in its pursuit-not only for the measures which are atrocities in every war, however just its aims.

But this war has revealed attitudes of a more general kind, that influenced the conduct of earlier wars as well. After it has ended, we

{124} shall still be faced with the problem of how warfare may be conducted, and the attitudes that have resulted in the specific conduct of this war will not have disappeared. Moreover, similar problems can arise in wars or rebellions fought for very different reasons, and against very different opponents. It is not easy to keep a firm grip on the idea of what is not permissible in warfare, because while some military actions are obvious atrocities, other cases are more difficult to assess, and the general principles underlying these judgments remain obscure. Such obscurity can lead to the abandonment of sound intuitions in favor of criteria whose rationale may be more obvious. If such a tendency is to be resisted, it will require a better understanding of the restrictions than we now have.

I propose to discuss the most general moral problem raised by the conduct of warfare: the problem of means and ends. In one view, there are limits on what may be done even in the service of an end worth pursuing-and even when adherence to the restriction may be very costly. A person who acknowledges the force of such restrictions can find himself in acute moral dilemmas. He may believe, for example, that by torcuring a prisoner he can obtain information necessary to prevent a disaster, or that by obliterating one village with bombs he can halt a campaign of terrorism. If he believes that the gains from a certain measure will clearly outweigh its costs, yet still suspects that he ought not to adopt it, then he is in a dilemma produced by the conflict between two disparate categories of moral reason: categories that may be called utilitarian and absolutist.

Utilitarianism gives primacy to a concern with what will happen. Absolutism gives primacy to a concern with what one is doing. The conflict between them arises because the alternatives we face are rarely just choices between total outcomes: they are also choices between alternative pathways or measures to be taken. When one of the choices is to do terrible things to another person, the problem is altered fundamentally; it is no longer merely a question of which outcome would be worse.

Few of us are completely immune to either of these types of moral intuition, though in some people, either naturally or for doctrinal reasons, one type will be dominant and the other suppressed or weak. But it is perfectly possible to feel the force of both types of reason

{125} very strongly; in that case the moral dilemma in certain situations of crisis will be acute, and it may appear that every possible course of action or inaction is unacceptable for one reason or another.


Although it is this dilemma that I propose to explore, most of the discussion will be devoted to its absolutist component. The utilitarian component is straightforward by comparison, and has a natural appeal to anyone who is not a complete skeptic about ethics. Utilitarianism says that one should try, either individually or through institutions, to maximize good and minimize evil (the definition of these categories need not enter into the schematic formulation of the view), and that if faced with the possibility of preventing a great evil by producing a lesser, one should choose the lesser evil. There are certainly problems about the formulation of utilitarianism, and much has been written about it, but its intent is morally transparent. Nevertheless, despite the addition of various refinements, it continues to leave large portions of ethics unaccounted for. I do not suggest that some form of absolutism can account for them all, only that an examination of absolutism will lead us to see the complexity, and perhaps the incoherence, of our moral ideas.

Utilitarianism certainly justifies some restrictions on the conduct of warfare. There are strong utilitarian reasons for adhering to any limitation which seems natural to most people-particularly if the limitation is widely accepted already. An exceptional measure which seems to be justified by its results in a particular conflict may create a precedent with disastrous long-term effect2. It may even be argued that war involves violence on such a scale that it is never justified on utilitarian grounds-the consequences of refusing to go to war will never be as bad as the war itself would be, even if atrocities were not committed. Or in a more sophisticated vein it might be claimed that a uniform policy of never resorting to military force would do less harm in the long run, if followed consistently, than a policy of deciding each case on utilitarian grounds (even though on occasion

{126} particular applications of the pacifist policy might have worse results than a specific utilitarian decision). But I shall not consider these arguments, for my concern is with reasons of a different kind, which may remain when reasons of utility and interest fail.3

In the final analysis, I believe that the dilemma cannot always be resolved. While not every conflict between absolutism and utilitarianism creates an insoluble dilemma, and while it is certainly right to adhere to absolutist restrictions unless the utilitarian considerations favoring violation are overpoweringly weighty and extremely certain—nevertheless, when that special condition is met, it may become impossible to adhere to an absolutist position. What I shall offer, therefore, is a somewhat qualified defense of absolutism. I believe it underlies a valid and fundamental type of moral judgment—which cannot be reduced to or overridden by other principles. And while there may be other principles just as fundamental, it is particularly important not to lose confidence in our absolutist intuitions, for they are often the only barrier before the abyss of utilitarian apologetics for large-scale murder.


One absolutist position that creates no problems of interpretation is pacifism: the view that one may not kill another person under any circumstances, no matter what good would be achieved or evil averted thereby. The type of absolutist position that I am going to discuss is different. Pacifism draws the conflict with utilitarian considerations very starkly. But there are other views according to which violence may be undertaken, even on a large scale, in a clearly just cause, so long as certain absolute restrictions on the character and direction of that violence are observed. The line is drawn somewhat closer to the bone, but it exists.

The philosopher who has done most to advance contemporary philosophical discussion of such a view, and to explain it to those

{127} unfamiliar with its extensive treatment in Roman Catholic moral theology, is G.E.M. Anscombe. In 1958 Miss Anscombe published a pamphlet entitled Mr. Truman's Degree,4 on the occasion of the award by Oxford University of an honorary doctorate to Harry Truman. The pamphlet explained why she had opposed the decision to award that degree, recounted the story of her unsuccessful opposition, and offered some reflections on the history of Truman's decision to drop atom bombs on Hiroshima and Nagasaki, and on the difference between murder and allowable killing in warfare. She pointed out that the policy of deliberately killing large numbers of civilians either as a means or as an end in itself did not originate with Truman, and was common practice among all parties during World War II for some time before Hiroshima. The Allied area bombings of German cities by conventional explosives included raids which killed more civilians than did the atomic attacks; the same is true of certain fire-bomb raids on Japan.

The policy of attacking the civilian population in order to induce an enemy to surrender, or to damage his morale, seems to have been widely accepted in the civilized world, and seems to be accepted still, at least if the stakes are high enough. It gives evidence of a moral conviction that the deliberate killing of noncombatants—women, children, old people—is permissible if enough can be gained by it. This follows from the more general position that any means can in principle be justified if it leads to a sufficiently worthy end. Such an attitude is evident not only in the more spectacular current weapons systems but also in the day-to-day conduct of the nonglobal war in Indochina: the indiscriminate destructiveness of antipersonnel weapons, napalm, and aerial bombardment; cruelty to prisoners; massive relocation of civilians; destruction of crops; and so forth. An abso-

{128}lutist position opposes to this the view that certain acts cannot be justified no matter what the consequences. Among those acts is murder—the deliberate killing of the harmless: civilians, prisoners of war, and medical personnel.

In the present war such measures are sometimes said to be regrettable, but they are generally defended by reference to military necessity and the importance of the long-term consequences of success or failure in the war. I shall pass over the inadequacy of this consequentialist defense in its own terms. (That is the dominant form of moral criticism of the war, for it is part of what people mean when they ask, "Is it worth it?") I am concerned rather to account for the inappropriateness of offering any defense of that kind for such actions.

Many people feel, without being able to say much more about it, that something has gone seriously wrong when certain measures are admitted into consideration in the first place. The fundamental mistake is made there, rather than at the point where the overall benefit of some monstrous measure is judged to outweigh its disadvantages, and it is adopted. An account of absolutism might help us to understand this. If it is not allowable to do certain things, such as killing unarmed prisoners or civilians, then no argument about what will happen if one doesn't do them can show that doing them would be all right.

Absolutism does not, of course, require one to ignore the consequences of one's acts. It operates as a limitation on utilitarian reasoning, not as a substitute for it. An absolutist can be expected to try to maximize good and minimize evil, so long as this does not require him to transgress an absolute prohibition like that against murder. But when such a conflict occurs, the prohibition takes complete precedence over any consideration of consequences. Some of the results of this view are clear enough. It requires us to forgo certain potentially useful military measures, such as the slaughter of hostages and prisoners or indiscriminate attempts to reduce the enemy civilian population by starvation, epidemic infectious diseases like anthrax and bubonic plague, or mass incineration. It means that we cannot deliberate on whether such measures are justified by the fact that they will avert still greater evils, for as intentional measures they cannot be justified in terms of any consequences whatever.

{129} Someone unfamiliar with the events of this century might imagine that utilitarian arguments, or arguments of national interest, would suffice to deter measures of this sort. But it has become evident that such considerations are insufficient to prevent the adoption and employment of enormous antipopulation weapons once their use is considered a serious moral possibility. The same is true of the piecemeal wiping out of rural civilian populations in airborne antiguerrilla warfare. Once the door is opened to calculations of utility and national interest, the usual speculations about the future of freedom, peace, and economic prosperity can be brought to bear to ease the consciences of those responsible for a certain number of charred babies.

For this reason alone it is important to decide what is wrong with the frame of mind which allows such arguments to begin. But it is also important to understand absolutism in the cases where it genuinely conflicts with utility. Despite its appeal, it is a paradoxical position, for it can require that one refrain from choosing the lesser of two evils when that is the only choice one has. And it is additionally paradoxical because, unlike pacifism, it permits one to do horrible things to people in some circumstances but not in others.


Before going on to say what, if anything, lies behind the position, there remain a few relatively technical matters which are best discussed at this point.

First, it is important to specify as clearly as possible the kind of thing to which absolutist prohibitions can apply. We must take seriously the proviso that they concern what we deliberately do to people. There could not, for example, without incoherence, be an absolute prohibition against bringing about the death of an innocent person. For one may find oneself in a situation in which, no matter what one does, some innocent people will die as a result. I do not mean just that there are cases in which someone will die no matter what one does, because one is not in a position to affect the outcome one way or the other. That, it is to be hoped, is one's relation to the deaths of most innocent people. I have in mind, rather, a case in which someone is bound to die, but who it is will depend on what one does. Sometimes these situations have natural causes, as when too few resources (medicine, lifeboats) are available to rescue everyone threat-

{130}ened with a certain catastrophe. Sometimes the situations are man-made, as when the only way to control a campaign of terrorism is to employ terrorist tactics against the community from which it has arisen. Whatever one does in cases such as these, some innocent people will die as a result. If the absolutist prohibition forbade doing what would result in the deaths of innocent people, it would have the consequence that in such cases nothing one could do would be morally permissible.

This problem is avoided, however, because what absolutism forbids is doing certain things to people, rather than bringing about certain results. Not everything that happens to others as a result of what one does is something that one has done to them. Catholic moral theology seeks to make this distinction precise in a doctrine known as the law of double effect, which asserts that there is a morally relevant distinction between bringing about the death of an innocent person deliberately, either as an end in itself or as a means, and bringing it about as a side effect of something else one does deliberately. In the latter case, even if the outcome is foreseen, it is not murder, and does not fall under the absolute prohibition, though of course it may still be wrong for other reasons (reasons of utility, for example). Briefly, the principle states that one is sometimes permitted knowingly to bring about as a side effect of one's actions something which it would be absolutely impermissible to bring about deliberately as an end or as a means. In application to war or revolution, the law of double effect permits a certain amount of civilian carnage as a side effect of bombing munitions plants or attacking enemy soldiers. And even this is permissible only if the cost is not too great to be justified by one's objectives.

However, despite its importance and its usefulness in accounting for certain plausible moral judgments, I do not believe that the law of double effect is a generally applicable test for the consequences of an absolutist position. Its own application is not always clear, so that it introduces uncertainty where there need not be uncertainty.

In Indochina, for example, there is a great deal of aerial bombardment, strafing, spraying of napalm, and employment of pellet- or needle-spraying antipersonnel weapons against rural villages in which guerrillas are suspected to be hiding, or from which small-arms fire

{131} has been received. The majority of those killed and wounded in these aerial attacks are reported to be women and children, even when some combatants are caught as well. However, the government regards these civilian casualties as a regrettable side effect of what is a legitimate attack against an armed enemy.

It might be thought easy to dismiss this as sophistry: if one bombs, burns, or strafes a village containing a hundred people, twenty of whom one believes to be guerrillas, so that by killing most of them one will be statistically likely to kill most of the guerrillas, then isn't one's attack on the group of one hundred a means of destroying the guerrillas, pure and simple? If one makes no attempt to discriminate between guerrillas and civilians, as is impossible in a aerial attack on a small village, then one cannot regard as a mere side effect the deaths of those in the group that one would not have bothered to kill if more selective means had been available.

The difficulty is that this argument depends on one particular description of the act, and the reply might be that the means used against the guerrillas is not: killing everybody in the village—but rather: obliteration bombing of the area in which the twenty guerrillas are known to be located. If there are civilians in the area as well, they will be killed as a side effect of such action.5

Because of casuistical problems like this, I prefer to stay with the original, unanalyzed distinction between what one does to people and what merely happens to them as a result of what one does. The law of double effect provides an approximation to that distinction in many cases, and perhaps it can be sharpened to the point where it does better than that. Certainly the original distinction itself needs clarification, particularly since some of the things we do to people involve things happening to them as a result of other things we do. In a case like the one discussed, however, it is clear that by bombing the village one slaughters and maims the civilians in it. Whereas by giving the only available medicine to one of two sufferers from a disease, one does not kill the other, even if he dies as a result.

The second technical point to take up concerns a possible misinterpretation of this feature of the position. The absolutist focus on actions rather than outcomes does not merely introduce a new, out-

{132}standing item into the catalogue of evils. That is, it does not say that the worst thing in the world is the deliberate murder of an innocent person. For if that were all, then one could presumably justify one such murder on the ground that it would prevent several others, or ten thousand on the ground that they would prevent a hundred thousand more. That is a familiar argument. But if this is allowable, then there is no absolute prohibition against murder after all. Absolutism requires that we avoid murder at all costs, not that we prevent it at all costs.6

Finally, let me remark on a frequent criticism of absolutism that depends on a misunderstanding. It is sometimes suggested that such prohibitions depend on a kind of moral self-interest, a primary obligation to preserve one's own moral purity, to keep one's hands clean no matter what happens to the rest of the world. If this were the position, it might be exposed to the charge of self-indulgence. After all, what gives one man a right to put the purity of his soul or the cleanness of his hands above the lives or welfare of large numbers of other people? It might be argued that a public servant like Truman has no right to put himself first in that way; therefore if he is convinced that the alternatives would be worse, he must give the order to drop the bombs, and take the burden of those deaths on himself, as he must do other distasteful things for the general good.

But there are two confusions behind the view that moral self-interest underlies moral absolutism. First, it is a confusion to suggest that the need to preserve one's moral purity might be the source of an obligation. For if by committing murder one sacrifices one's moral purity or integrity, that can only be because there is already something wrong with murder. The general reason against committing murder cannot therefore be merely that it makes one an immoral person. Secondly, the notion that one might sacrifice one's moral integrity justifiably, in the service of a sufficiently worthy end, is an incoherent notion. For if one were justified in making such a sacrifice (or even morally required to make it), then one would not be

{133} sacrificing one's moral integrity by adopting that course: one would be preserving it.

Moral absolutism is not unique among moral theories in requiring each person to do what will preserve his own moral purity in all circumstances. This is equally true of utilitarianism, or of any other theory which distinguishes between right and wrong. Any theory which defines the right course of action in various circumstances and asserts that one should adopt that course, ipso facto asserts that one should do what will preserve one's moral purity, simply because the right course of action is what will preserve one's moral purity in those circumstances. Of course utilitarianism does not assert that this is why one should adopt that course, but we have seen that the same is true of absolutism.


It is easier to dispose of false explanations of absolutism than to produce a true one. A positive account of the matter must begin with the observation that war, conflict, and aggression are relations between persons. The view that it can be wrong to consider merely the overall effect of one's actions on the general welfare comes into prominence when those actions involve relations with others. A man's acts usually affect more people than he deals with directly, and those effects must naturally be considered in his decisions. But if there are special principles governing the manner in which he should treat people, that will require special attention to the particular persons toward whom the act is directed, rather than just to its total effect.

Absolutist restrictions in warfare appear to be of two types: restrictions on the class of persons at whom aggression or violence may be directed and restrictions on the manner of attack, given that the object falls within that class. These can be combined, however, under the principle that hostile treatment of any person must be justified in terms of something about that person which makes the treatment appropriate. Hostility is a personal relation, and it must be suited to its target. One consequence of this condition will be that certain persons may not be subjected to hostile treatment in war at all, since nothing about them justifies such treatment. Others will be proper objects of hostility only in certain circumstances, or when they are

{134} engaged in certain pursuits. And the appropriate manner and extent of hostile treatment will depend on what is justified by the particular case.

A coherent view of this type will hold that extremely hostile behavior toward another is compatible with treating him as a person—even perhaps as an end in himself. This is possible only if one has not automatically stopped treating him as a person as soon as one starts to fight with him. If hostile, aggressive, or combative treatment of others always violated the condition that they be treated as human beings, it would be difficult to make further distinctions on that score within the class of hostile actions. That point of view, on the level of international relations, leads to the position that if complete pacifism is not accepted, no holds need be barred at all, and we may slaughter and massacre to our hearts' content,'if it seems advisable. Such a position is often expressed in discussions of war crimes.

But the fact is that ordinary people do not believe this about conflicts, physical or otherwise, between individuals, and there is no more reason why it should be true of conflicts between nations. There seems to be a perfectly natural conception of the distinction between fighting clean and fighting dirty. To fight dirty is to direct one's hostility or aggression not at its proper object, but at a peripheral target which may be more vulnerable, and through which the proper object can be attacked indirectly. This applies in a fist fight, an election campaign, a duel, or a philosophical argument. If the concept is general enough to apply to all these matters, it should apply to war—both to the conduct of individual soldiers and to the conduct of nations.

Suppose that you are a candidate for public office, convinced that the election of your opponent would be a disaster, that he is an unscrupulous demagogue who will serve a narrow range of interests and seriously infringe the rights of those who disagree with him; and suppose you are convinced that you cannot defeat him by conventional means. Now imagine that various unconventional means present themselves as possibilities: you possess information about his sex life which would scandalize the electorate if made public; or you learn that his wife is an alcoholic or that in his youth he was associated for a brief period with a proscribed political party, and you believe that this information could be used to blackmail him into with-

{135}drawing his candidacy; or you can have a team of your supporters flatten the tires of a crucial subset of his supporters on election day; or you are in a position to stuff the ballot boxes; or, more simply, you can have him assassinated. What is wrong with these methods, given that they will achieve an overwhelmingly desirable result?

There are, of course, many things wrong with them: some are against the law; some infringe the procedures of an electoral process to which you are presumably committed by taking part in it; very importantly, some may backfire, and it is in the interest of all political candidates to adhere to an unspoken agreement not to allow certain personal matters to intrude into a campaign. But that is not all. We have in addition the feeling that these measures, these methods of attack are irrelevant to the issue between you and your opponent, that in taking them up you would not be directing yourself to that which makes him an object of your opposition. You would be directing your attack not at the true target of your hostility, but at peripheral targets that happen to be vulnerable.

The same is true of a fight or argument outside the framework of any system of regulations or law. In an altercation with a taxi driver over an excessive fare, it is inappropriate to taunt him about his accent, flatten one of his tires, or smear chewing gum on his windshield; and it remains inappropriate even if he casts aspersions on your race, politics, or religion, or dumps the contents of your suitcase into the street.7

The importance of such restrictions may vary with the seriousness of the case; and what is unjustifiable in one case may be justified in a more extreme one. But they all derive from a single principle: that hostility or aggression should be directed at its true object. This means both that it should be directed at the person or persons who provoke it and that it should aim more specifically at what is provocative about them. The second condition will determine what form the hostility may appropriately take.

{136} It is evident that some idea of the relation in which one should stand to other people underlies this principle, but the idea is difficult to state. I believe it is roughly this: whatever one does to another person intentionally must be aimed at him as a subject, with the intention that he receive it as a subject. It should manifest an attitude to him rather than just to the situation, and he should be able to recognize it and identify himself as its object. The procedures by which such an attitude is manifested need not be addressed to the person directly. Surgery, for example, is not a form of personal confrontation but part of a medical treatment that can be offered to a patient face to face and received by him as a response to his needs and the natural outcome of an attitude toward him.

Hostile treatment, unlike surgery, is already addressed to a person, and does not take its interpersonal meaning from a wider context. But hostile acts can serve as the expression or implementation of only a limited range of attitudes to the person who is attacked. Those attitudes in turn have as objects certain real or presumed characteristics or activities of the person which are thought to justify them. When this background is absent, hostile or aggressive behavior can no longer be intended for the reception of the victim as a subject. Instead it takes on the character of a purely bureaucratic operation. This occurs when one attacks someone who is not the true object of one's hostility—the true object may be someone else, who can be attacked through the victim; or one may not be manifesting a hostile attitude toward anyone, but merely using the easiest available path to some desired goal. One finds oneself not facing or addressing the victim at all, but operating on him-without the larger context of personal interaction that surrounds a surgical operation.

If absolutism is to defend its claim to priority over considerations of utility, it must hold that the maintenance of a direct interpersonal response to the people one deals with is a requirement which no advantages can justify one in abandoning. The requirement is absolute only if it rules out any calculation of what would justify its violation. I have said earlier that there may be circumstances so extreme that they render an absolutist position untenable. One may find then that one has no choice but to do something terrible. Neverthe-

{137}less, even in such cases absolutism retains its force in that one cannot claim justification for the violation. It does not become all right.

As a tentative effort to explain this, let me try to connect absolutist limitations with the possibility of justifying to the victim what is being done to him. If one abandons a person in the course of rescuing several others from a fire or a sinking ship, one could say to him, "You understand, I have to leave you to save the others." Similarly, if one subjects an unwilling child to a painful surgical procedure, one can say to him, "If you could understand, you would realize that I am doing this to help you." One could even say, as one bayonets an enemy soldier, "It's either you or me." But one cannot really say while torturing a prisoner, "You understand, I have to pull out your fingernails because it is absolutely essential that we have the names of your confederates"; nor can one say to the victims of Hiroshima, "You understand, we have to incinerate you to provide the Japanese government with an incentive to surrender."

This does not take us very far, of course, since a utilitarian would presumably be willing to offer justifications of the latter sort to his victims, in cases where he thought they were sufficient. They are really justifications to the world at large, which the victim, as a reasonable man, would be expected to appreciate. However, there seems to me something wrong with this view, for it ignores the possibility that to treat someone else horribly puts you in a special relation to him, which may have to be defended in terms of other features of your relation to him. The suggestion needs much more development; but it may help us to understand how there may be requirements which are absolute in the sense that there can be no justification for violating them. If the justification for what one did to another person had to be such that it could be offered to him specifically, rather than just to the world at large, that would be a significant source of restraint.

If the account is to be deepened, I would hope for some results along the following lines. Absolutism is associated with a view of oneself as a small being interacting with others in a large world. The justifications it requires are primarily interpersonal. Utilitarianism is associated with a view of oneself as a benevolent bureaucrat dis-

{138}tributing such benefits as one can control to countless other beings, with whom one may have various relations or none. The justifications it requires are primarily administrative. The argument between the two moral attitudes may depend on the relative priority of these two conceptions.8


Some of the restrictions on methods of warfare which have been adhered to from time to time are to be explained by the mutual interests of the involved parties: restrictions on weaponry, treatment of prisoners, etc. But that is not all there is to it. The conditions of directness and relevance which I have argued apply to relations of conflict and aggression apply to war as well. I have said that there are two types of absolutist restrictions on the conduct of war: those that limit the legitimate targets of hostility and those that limit its character, even when the target is acceptable. I shall say something about each of these. As will become clear, the principle I have sketched does not yield an unambiguous answer in every case.

First let us see how it implies that attacks on some people are allowed, but not attacks on others. It may seem paradoxical to assert that to fire a machine gun at someone who is throwing hand grenades at your emplacement is to treat him as a human being. Yet the relation with him is direct and straightforward.9 The attack is aimed specifically against the threat presented by a dangerous adversary, and not against a peripheral target through which he happens to be vulnerable but which has nothing to do with that threat. For example, you might stop him by machine-gunning his wife and children, who are standing nearby, thus distracting him from his aim of blowing you up and enabling you to capture him. But if his wife and children are not threatening your life, that would be to treat them as means with a vengeance.

{139} This, however, is just Hiroshima on a smaller scale. One objection to weapons of mass annihilation—nuclear, thermonuclear, biological, or chemical—is that their indiscriminateness disqualifies them as direct instruments for the expression of hostile relations. In attacking the civilian population, one treats neither the military enemy nor the civilians with that minimal respect which is owed to them as human beings. This is clearly true of the direct attack on people who present no threat at all. But it is also true of the character of the attack on those who are threatening you, viz., the government and military forces of the enemy. Your aggression is directed against an area of vulnerability quite distinct from any threat presented by them which you may be justified in meeting. You are taking aim at them through the mundane life and survival of their countrymen, instead of aiming at the destruction of their military capacity. And of course it does not require hydrogen bombs to commit such crimes.

This way of looking at the matter also helps us to understand the importance of the distinction between combatants and noncombatants, and the irrelevance of much of the criticism offered against its intelligibility and moral significance. According to an absolutist position, deliberate killing of the innocent is murder, and in warfare the role of the innocent is filled by noncombatants. This has been thought to raise two sorts of problems: first, the widely imagined difficulty of making a division, in modern warfare, between combatants and noncombatants; second, problems deriving from the connotation of the word "innocence."

Let me take up the latter question first.10 In the absolutist position, the operative notion of innocence is not moral innocence, and it is not opposed to moral guilt. If it were, then we would be justified in killing a wicked but noncombatant hairdresser in an enemy city who supported the evil policies of his government, and unjustified in killing a morally pure conscript who was driving a tank toward us with the profoundest regrets and nothing but love in his heart. But moral innocence has very little to do with it, for in the definition of murder "innocent" means "currently harmless," and it is opposed not to "guilty" but to "doing harm." It should be noted that such an analysis has the consequence that in war we may often be justified in kill-

{140}ing people who do not deserve to die, and unjustified in killing people who do deserve to die, if anyone does. So we must distinguish combatants from noncombatants on the basis of their immediate threat or harmfulness. I do not claim that the line is a sharp one, but it is not so difficult as is often supposed to place individuals on one side of it or the other. Children are not combatants even though they may join the armed forces if they are allowed to grow up. Women are not combatants just because they bear children or offer comfort to the soldiers. More problematic are the supporting personnel, whether in or out of uniform, from drivers of munitions trucks and army cooks to civilian munitions workers and farmers. I believe they can be plausibly classified by applying the condition that the prosecution of conflict must direct itself to the cause of danger, and not to what is peripheral. The threat presented by an army and its members does not consist merely in the fact that they are men, but in the fact that they are armed and are using their arms in the pursuit of certain objectives. Contributions to their arms and logistics are contributions to this threat; contributions to their mere existence as men are not. It is therefore wrong to direct an attack against those who merely serve the combatants' needs as human beings, such as farmers and food suppliers, even though survival as a human being is a necessary condition of efficient functioning as a soldier.

This brings us to the second group of restrictions: those that limit what may be done even to combatants. These limits are harder to explain clearly. Some of them may be arbitrary or conventional, and some may have to be derived from other sources; but I believe that the condition of directness and relevance in hostile relations accounts for them to a considerable extent.

Consider first a case which involves both a protected class of non-combatants and a restriction on the measures that may be used against combatants. One provision of the rules of war which is universally recognized, though it seems to be turning into a dead letter in Vietnam, is the special status of medical personnel and the wounded in warfare. It might be more efficient to shoot medical officers on sight and to let the enemy wounded die rather than be patched up to fight another day. But someone with medical insignia

{141} is supposed to be left alone and permitted to tend and retrieve the wounded. I believe this is because medical attention is a species of attention to completely general human needs, not specifically the needs of a combat soldier, and our conflict with the soldier is not with his existence as a human being.

By extending the application of this idea, one can justify prohibitions against certain particularly cruel weapons: starvation, poisoning, infectious diseases (supposing they could be inflicted on combatants only), weapons designed to maim or disfigure or torture the opponent rather than merely to stop him. It is not, I think, mere casuistry to claim that such weapons attack the men, not the soldiers. The effect of dum-dum bullets, for example, is much more extended than necessary to cope with the combat situation in which they are used. They abandon any attempt to discriminate in their effects between the combatant and the human being. For this reason the use of flamethrowers and napalm is an atrocity in all circumstances that I can imagine, whoever the target may be. Burns are both extremely painful and extremely disfiguring—far more than any other category of wound. That this well-known fact plays no (inhibiting) part in the determination of U.S. weapons policy suggests that moral sensitivity among public officials has not increased markedly since the Spanish Inquisition.11

{142} Finally, the same condition of appropriateness to the true object of hostility should limit the scope of attacks on an enemy country: its economy, agriculture, transportation system, and so forth. Even if the parties to a military conflict are considered to be not armies or governments but entire nations (which is usually a grave error), that does not justify one nation in warring against every aspect or element of another nation. That is not justified in a conflict between individuals, and nations are even more complex than individuals, so the same reasons apply. Like a human being, a nation is engaged in countless other pursuits while waging war, and it is not in those respects that it is an enemy.

The burden of the argument has been that absolutism about murder has a foundation in principles governing all one's relations to other persons, whether aggressive or amiable, and that these principles, and that absolutism, apply to warfare as well, with the result that certain measures are impermissible no matter what the consequences.12 I do not mean to romanticize war. It is sufficiently utopian to suggest that when nations conflict they might rise to the level of limited barbarity that typically characterizes violent conflict between individuals, rather than wallowing in the moral pit where they appear to have settled, surrounded by enormous arsenals.


Having described the elements of the absolutist position, we must now return to the conflict between it and utilitarianism. Even if certain types of dirty tactics become acceptable when the stakes are high enough, the most serious of the prohibited acts, like murder and torture, are not just supposed to require unusually strong justifica-

{143}tion. They are supposed never to be done, because no quantity of resulting benefit is thought capable of justifying such treatment of a person.

The fact remains that when an absolutist knows or believes that the utilitarian cost of refusing to adopt a prohibited course will be very high, he may hold to his refusal to adopt it, but he will find it difficult to feel that a moral dilemma has been satisfactorily resolved. The same may be true of someone who rejects an absolutist requirement and adopts instead the course yielding the most acceptable consequences. In either case, it is possible to feel that one has acted for reasons insufficient to justify violation of the opposing principle. In situations of deadly conflict, particularly where a weaker party is threatened with annihilation or enslavement by a stronger one, the argument for resorting to atrocities can be powerful, and the dilemma acute.

There may exist principles, not yet codified, which would enable us to resolve such dilemmas. But then again there may not. We must face the pessimistic alternative that these two forms of moral intuition are not capable of being brought together into a single, coherent moral system, and that the world can present us with situations in which there is no honorable or moral course for a man to take, no course free of guilt and responsibility for evil.

The idea of a moral blind alley is a perfectly intelligible one. It is possible to get into such a situation by one's own fault, and people do it all the time. If, for example, one makes two incompatible promises or commitments—becomes engaged to two people, for example—then there is no course one can take which is not wrong, for one must break one's promise to at least one of them. Making a clean breast of the whole thing will not be enough to remove one's reprehensibility. The existence of such cases is not morally disturbing, however, because we feel that the situation was not unavoidable: one had to do something wrong in the first place to get into it. But what if the world itself, or someone else's actions, could face a previously innocent person with a choice between morally abominable courses of action, and leave him no way to escape with his honor? Our intuitions rebel at the idea, for we feel that the constructibility of such a case must show a contradiction in our moral views. But it is not in

{144} itself a contradiction to say that someone can do X or not do X, and that for him to take either course would be wrong. It merely contradicts the supposition that ought implies can—since presumably one ought to refrain from what is wrong, and in such a case it is impossible to do so.13 Given the limitations on human action, it is naive to suppose that there is a solution to every moral problem with which the world can face us. We have always known that the world is a bad place. It appears that it may be an evil place as well.

1. This paper grew out of discussions at the Society for Ethical and Legal Philosophy, and I am indebted to my fellow members for their help.

2. Straightforward considerations of national interest often tend in the same direction: the inadvisability of using nuclear weapons seems to be overdetermined in this way.

3. These reasons, moreover, have special importance in that they are available even to one who denies the appropriateness of utilitarian considerations in international matters. He may acknowledge limitations on what may be done to the soldiers and civilians of other countries in pursuit of his nation's military objectives, while denying that one country should in general consider the interests of nationals of other countries in determining its policies.

4. (Privately printed.) See also her essay "War and Murder," in Nuclear Weapons and Christian Conscience, ed. Walter Stein (London, 1963). The present paper is much indebted to these two essays throughout. These and related subjects are extensively treated by Paul Ramsey in The Just War (New York, 1968). Among recent writings that bear on the moral problem are Jonathan Bennett, "Whatever the Consequences," Analysis 26, no. 3 (1966): 83-102; and Philippa Foot, "The Problem of Abortion and the Doctrine of the Double Effect," The Oxford Review 5 (1967): 5-15, Miss Anscombe's replies are "A Note on Mr. Bennett," Analysis 26, no. 3 (1966): 208, and "Who is Wronged?" The Oxford Review 5 (1967): 16-17.

5. This counterargument was suggested by Rogers Albritton.

6. Someone might of course acknowledge the moral relevance of the distinction between deliberate and nondeliberate killing, without being an absolutist. That is, he might believe simply that it was worse to bring about a death deliberately than as a secondary effect. But that would be merely a special assignment of value, and not an absolute prohibition.

7. Why, on the other hand, does it seem appropriate, rather than irrelevant, to punch someone in the mouth if he insults you? The answer is that in our culture it is an insult to punch someone in the mouth, and not just an injury. This reveals, by the way, a perfectly unobjectionable sense in which convention may play a part in determining exactly what falls under an absolutist restriction and what does not. I am indebted to Robert Fogelin for this point.

8. Finally, I should mention a different possibility, suggested by Robert Nozick: that there is a strong general presumption against benefiting from the calamity of another, whether or not it has been deliberately inflicted for that or any other reason. This broader principle may well lend its force to the absolutist position.

9. It has been remarked that according to my view, shooting at someone establishes an I-thou relationship.

10. What I say on this subject derives from Anscombe.

11. Beyond this I feel uncertain. Ordinary bullets, after all, can cause death, and nothing is more permanent than that. I am not at all sure why we are justified in trying to kill those who are trying to kill us (rather than merely in trying to stop them with force which may also result in their deaths). It is often argued that incapacitating gases are a relatively humane weapon (when not used, as in Vietnam, merely to make people easier to shoot). Perhaps the legitimacy of restrictions against them must depend on the dangers of escalation, and the great utility of maintaining any conventional category of restriction so long as nations are willing to adhere to it. Let me make clear that I do not regard my argument as a defense of the moral immutability of the Hague and Geneva Conventions. Rather, I believe that they rest partly on a moral foundation, and that modifications of them should also be assessed on moral grounds. But even this connection with the actual laws of war is not essential to my claims about what is permissible and what is not. Since completing this paper I have read an essay by Richard Wasserstrom entitled "The Laws of War" (forthcoming in The Monist), which argues that the existing laws and conventions do not even attempt to embody a decent moral position: that their provisions have been determined by other interests, that they are in fact immoral in substance, and that it is a grave mistake to refer to them as standards in forming moral judgments about warfare. This possibility deserves serious consideration, and I am not sure what to say about it, but it does not affect my view of the moral issues.

12. It is possible to draw a more radical conclusion, which I shall not pursue here. Perhaps the technology and organization of modern war are such as to make it impossible to wage as an acceptable form of interpersonal or even international hostility. Perhaps it is too impersonal and large-scale for that. If so, then absolutism would in practice imply pacifism, given the present state of things. On the other hand, I am skeptical about the unstated assumption that a technology dictates its own use.

13. This was first pointed out to me by Christopher Boorse.