Absolute Deontology and the Problem of Risk



Absolute Deontology and the Problem of Risk

ABSTRACT: Absolutists hold that there are some ways of harming others that are absolutely impermissible, that is, cannot be justified no matter what the benefits. Absolutists may then take one of three attitudes towards actions that impose a risk of harming others in these ways: either (a) all such actions are absolutely impermissible, (b) such actions are absolutely impermissible only when the probability of harm is 1, or (c) such actions become absolutely impermissible when the probability of harm exceeds some threshold strictly between 0 and 1. I show that each of these alternatives results in an implausible ethical theory.

1. Absolute Deontology

It is a widely held tenet of deontological ethical theories that there are certain ways of harming individuals that cannot be justified merely by the production of a greater benefit, or the avoidance of a greater harm, for others.[1] Moderate deontologists qualify this by saying that if the good to be produced is vastly greater or of a qualitatively different and more important kind than the harm in question, then sometimes the harm can be justified. Absolute deontologists, however, hold that certain kinds of harms imposed on others can never be justified, no matter what the benefits.[2] I shall call the relevant way of harming individuals, “harming in the rights violating way,” or simply “rights-violation,” and I shall call the deontological principle just mentioned the “absolute rights principle”:

Absolute Rights Principle: It is always wrong to violate a person’s rights, regardless of the benefit that might be produced by doing so.

My use of “rights violation” is a stipulative one, not meant to capture the ordinary sense of that expression;[3] in my sense, a rights-violation is simply a harm of the kind, assuming there is such a kind, that cannot be justified by the production of a greater benefit for others. The need for this technical notion is shown by examples such as the following:

Trolley Car: A runaway trolley is heading for a fork in the track. If it takes the left fork, it will collide with and kill five people; if it takes the right fork, it will collide with and kill one person. None of the people can be moved out of the way in time. There is a switch that determines which fork the trolley takes. It is presently set to send the trolley to the left. You can flip the switch, sending the trolley to the right instead. Should you flip the switch?[4]

Organ Harvesting: A doctor in a hospital has five patients who need organ transplants; otherwise, they will die. They all need different organs. He also has one healthy patient, in for a routine checkup, who happens to be compatible with the five. Should the doctor kill the healthy patient and distribute his organs to the other five?[5]

Most people say that you should kill the one person to save the five in the first case but not in the second. A proponent of the absolute rights principle can accept this, maintaining that there is some relevant difference between the killing in the Trolley Car example and the killing in the Organ Harvesting example. For instance, perhaps the difference is that the harm in the latter case is intended as a means to saving the five patients, whereas in the former case the harm is only a foreseen but unintended side effect of the action that saves the five other people.[6] Or perhaps the difference is that in the Trolley Car case, an existing threat is diverted from one group of victims to another, whereas in the Organ Harvesting case a distinct threat is created.[7] Or perhaps there is some other difference between the two cases. Regardless, if the killing of the one person is justified in the Trolley Car case but not in the Organ Harvesting case, then the killing is “of the rights violating kind” in the latter case but not the former.

2. The Problem of Risk

A difficult problem for absolute deontologists is that of how to deal with the imposition of risk—specifically, with cases in which one imposes a risk of harm on an individual for the sake of producing greater benefits, where one’s action would count as a rights-violation if the situation were otherwise the same except that the harm was certain to occur. I shall call risks of this kind “potential rights violations.” For example, consider the following sort of scenario:

Partial Organ Harvesting: A doctor in a hospital has five patients who need parts of their organs replaced; otherwise, they will die. They all need parts of different organs. He also has one healthy patient who happens to be compatible with the five. If parts are taken from the healthy patient’s organs, the first five patients can be saved; however, there is a probability of x that the healthy patient will then die. The healthy patient declines to donate any parts of his organs. Should the doctor forcibly extract them?

If we let x = 1, then Partial Organ Harvesting is morally equivalent to Organ Harvesting, and deontologists will pronounce the taking of the organ-parts a clear rights violation. Therefore, when x < 1, the taking of the organ-parts is what I call a potential rights violation. How should an absolutist regard mere potential rights violations committed for the purpose of producing a greater benefit for others?[8]

There are three natural ways of answering this question. One way is to say that the potential rights violation is absolutely impermissible, as long as any nonzero risk of harm exists. Another answer is that the potential rights violation becomes absolutely impermissible only when the probability of harm is 100%; for any lower level of risk, the potential rights violation can be justified by sufficiently good consequences.[9] A third answer is that there is some threshold level of risk, strictly between zero and one, at which potential rights violations become absolutely impermissible; when the probability of harm is below this threshold, the potential rights violation can be justified by sufficiently good consequences, but when the probability of harm is above the threshold, it cannot.

The absolutist’s answer might vary according to the type of harm in question and/or other circumstances of the case. Perhaps some kinds of potential rights violations would be absolutely impermissible regardless of the level of risk (greater than zero) that they impose, while others only become absolutely impermissible at some threshold level of risk. Furthermore, on the risk-threshold view, the threshold level of risk at which potential rights violations become absolutely impermissible might vary from case to case. Still, if we fix all the morally relevant factors other than the probability of harm, one of the three answers given in the previous paragraph should apply in any given type of case. In any case of potential rights violation, if the situation were the same except that the probability of harm were 1, then the action, according to the absolute deontologist, could not be justified by the production of a greater good. Either this remains true no matter how much the probability of harm is lowered (as long as it remains greater than zero), or it ceases to be true as soon as the probability of harm is less than 1, or it ceases to be true when the probability of harm falls below some threshold strictly between 0 and 1. As I shall argue presently, however, none of these three answers is satisfactory; none of them leads to a plausible form of absolutism.

2.1. Zero risk tolerance

Suppose we adopt the first version of absolutism, on which potential rights violations are always held impermissible. We may call this the Zero Risk Tolerance view. Absolutism is widely known as an exacting moral theory, but the Zero Risk Tolerance view is absurdly so. Consider:

Driving: A surgeon needs to drive to work to save some number of people’s lives. If he does not do so, some patients will die; however, if he drives to work, there is a probability of x that he will kill at least one person in a traffic accident. May he drive to work?

If we let x = 1, most absolutists will judge it wrong to drive to work. Suppose, for example, that the surgeon knows for certain that there is a child on the road whom he will have to run over, and that for whatever reason, there is no alternative to running over and killing the child, other than to refrain from going to work. A typical deontological intuition is that the surgeon may not run over the child in this case.[10] However, the difference between killing the child in this case, and killing a person in a typical traffic accident, appears to be simply a matter of probabilities: in the former case, one knows that a person will be killed by the impact of one’s car if one drives a certain route; in the latter case, one knows only that there is a chance that a person will be killed by the impact of one’s car if one drives a certain route. Therefore, if the Zero Risk Tolerance view is correct, then the surgeon may not drive to work no matter how low the probability of killing someone in a traffic accident. Since in reality, there is a nonzero probability of killing someone whenever one drives a car, one may not drive to work to save other people’s lives. A fortiori, one may not drive for any less important purpose.

Similar arguments could be constructed to show that a great many common practices would be impermissible on the Zero Risk Tolerance view. For example, consider:

Pollution: A factory is manufacturing life-saving pharmaceuticals. Unfortunately, the factory inevitably produces a small amount of pollution, which has a probability of x of killing at least one innocent bystander per year. This risk cannot be eliminated, nor are there any viable substitutes for the drugs produced. May the factory continue to operate?

When x = 1, the absolute deontologist will presumably answer “no.” Therefore, on the Zero Risk Tolerance view, the factory may not operate regardless of how small the risk is. Since in reality, there is always a nonzero probability that any pollution will cause at least one death (just as there is a nonzero, even if minuscule, probability that virtually anything will cause a death), no factory may produce pollution to create life-saving products. A fortiori, one may not produce any pollution for any less important purpose.

Not all absolutists would be affected by this line of reasoning. Some, appealing to the Doctrine of Double Effect, would say that only actions that intend harm to others, either as an end or as a means, count as rights-violations; actions for which harm to others is merely a potential side effect do not thereby violate rights. This is not, of course, to say that actions that harm others as mere side effects fail thereby to be prima facie wrong. Rather, such actions are prima facie wrong, but they can nevertheless be justified by the production of a sufficiently greater good. In contrast, actions that intend harm to others cannot be justified by the production of a greater good. In the Driving case, if the surgeon kills a person in a traffic accident, the victim’s death will only be an unintended side effect; such a death is no part either of the surgeon’s end or of the surgeon’s means to that end, as is shown by the fact that if a traffic accident fails to occur, the surgeon’s plans are in no way frustrated. We thus cannot parlay the Driving case into a general argument that all driving is morally impermissible. A similar point applies to the Pollution case.

This line of thinking avoids the counter-intuitive consequences in Driving and related examples. My objection to it is that it betrays some of the central intuitions that motivate a deontological approach to ethics to begin with. It avoids the counter-intuitive result that driving is impermissible in normal circumstances, only by embracing the counter-intuitive claim that knowingly running over the innocent child on the way to the hospital (that is, when x = 1 in the Driving case) is permissible. It can also be made to deliver a counter-intuitive result in the following sort of example:

Frame-Up: A crime arousing great public outrage has recently occurred. The sheriff of the town believes that, unless someone is arrested and tried for this crime, there will be riots in which several people will be unjustly killed. Unfortunately, he cannot identify the guilty party. Should he frame an innocent person to appease the mob?[11]

This is a paradigm of the sort of case used to attack consequentialism and motivate a deontological approach to ethics. Deontologists generally take it as clear that the sheriff may not frame the innocent party. Now add the following stipulations to the case: the mob will be appeased as soon as the framed individual has been arrested and put on trial, although the defendant’s sentence would not be carried out until much later. However, once the trial begins, the sheriff will be unable to prevent the innocent defendant from being convicted and executed, due to the prejudice of the judge and the jury. In this situation, the doctrine of double effect leads to the conclusion that it is permissible to frame the innocent person. The innocent party’s conviction and execution are not intended, either as ultimate ends or as means, by the sheriff. They are not means to appeasing the mob, since they will occur after the mob has been appeased. Intuitively, however, this is irrelevant—it is morally irrelevant whether the mob is appeased as soon as the frame-up victim is put on trial, or whether they are not appeased until the sentence has been carried out.

One might claim that arresting an innocent individual and putting him on trial is itself a rights-violation, over and above the rights-violation of actually punishing the defendant. If so, the doctrine of double effect yields the desired result that the sheriff may not arrest the innocent person and cause him to be put on trial, since the harm of being arrested and put on trial would be intended by the sheriff rather than being a mere side effect. However, when this view is combined with the Zero Risk Tolerance form of absolutism, it yields the result that the sheriff may not arrest and put on trial any individual who is less than 100% certain to be guilty, since to do so would constitute a potential rights-violation. Since in reality it is never absolutely certain that any individual is guilty, this version of absolutism would demand the complete abandonment of the criminal justice system.

Consider a final, related example:

Indiscriminate Punishment: Two suspects have recently been arrested for a crime, in a society with a rather harsher criminal justice system than ours. It is known that exactly one of the suspects is guilty, but the state has no idea which is the guilty party. In accordance with the laws of the society in which the crime occurred, the state decides to punish both parties, to ensure that the guilty party gets punished.

Again, I take it that this is a paradigm of the sort of case in which one has deontological intuitions—in this case, to the effect that it is unjust and therefore wrong for the state to punish both parties, even if a greater good is produced by doing so. However, any actual criminal justice system differs from the case of Indiscriminate Punishment only in the numbers of guilty and innocent parties involved. By establishing a criminal justice system in which a standard of proof less than absolute certainty is used for deciding when suspects are to be punished, we for all practical purposes guarantee that both some guilty parties and some innocent parties will be punished. We accept the punishment of a relatively small number of innocent individuals as the cost of punishing a larger number of guilty parties. But if we were to adopt an absolute deontological position, we would be committed to rejecting any such tradeoff, and hence rejecting any practicable criminal justice system. How the doctrine of double effect bears on this issue is unclear—it is intuitively unclear whether, in Indiscriminate Punishment, the innocent party is punished as a means to ensuring the punishment of the guilty party. But the advocate of the doctrine of double effect faces a dilemma: if the state’s action in Indiscriminate Punishment counts as one of harming the innocent intentionally, then, if we also accept absolutism, we must denounce any practicable criminal justice system, since presumably implementing any such system would also involve harming the innocent intentionally. If, on the other hand, the state’s action in Indiscriminate Punishment counts as one of harming the innocent merely as an unintended side effect, then the doctrine of double effect yields the counter-intuitive result that the state’s action in Indiscriminate Punishment is permissible, provided that it produces a greater benefit.[12]

I have appealed above to what I take to be paradigmatic deontological intuitions to argue against the doctrine of double effect. My argument is not that because these deontological intuitions are correct, the doctrine of double effect is mistaken—in fact, I am suspicious of these intuitions. Rather, my argument is that because the doctrine of double effect clashes with the sort of intuitions that motivate deontology to begin with, we are not justified in accepting a version of deontology that employs the doctrine of double effect to demarcate rights-violations. Though the doctrine of double effect delivers the intuitively right results in some cases, it clashes with our intuitions in other, equally clear cases. If any deontological theory is correct, it seems most likely that it is a theory that accommodates our deontological intuitions in all the central, intuitively clear cases.

2.2. Maximal risk tolerance

Next, suppose we adopt an absolute deontological theory in which merely potential rights violations are always held to be justifiable by the production of greater benefits, as long as the risk of harm is less than 100%. While such a theory might, for all I have to say, be true, it is uninteresting. No action performed in any realistic circumstances is ever such that, at the time of acting, the agent can be 100% certain of causing a harm. There is always at least some possibility, however remote, that any would-be victims will be miraculously saved, or even benefitted by an action that would normally be harmful. Consequently, the absolute deontological prohibition would never be relevant to any actual choices. I therefore think it unnecessary to discuss this form of absolute deontology.

2.3. The risk threshold view

Finally, suppose we adopt an absolute deontological theory with a risk threshold somewhere strictly between zero and one, where potential rights violations are held to be justifiable by the production of greater benefits when the probability of harm falls short of the risk threshold, but to be absolutely impermissible when the probability of harm exceeds this threshold.

This view yields paradoxical results. Suppose two actions, A and B, each pose a risk slightly below risk threshold r of causing a rights-violating harm, and that each action is justified by the fact that it will produce a greater overall good. Now consider the compound action (A+B), the “action” of doing both A and B. (A+B) poses a risk greater than r of causing a rights-violating harm, so (A+B) cannot be justified by the production of a greater good.[13] Thus, it is permissible to do A, and it is permissible to do B, but it is impermissible to do A and B. To illustrate, consider:

Repeated Driving: A surgeon needs to drive to work on Tuesday to save some number of people’s lives. If he does not do so, some patients will die; however, if he drives to work, there is a probability of x that he will kill at least one person in a traffic accident. Again on Thursday, the surgeon has the option of driving to work to save more patients, with a risk of x that he will kill at least one person in a traffic accident. May he drive to work on Tuesday? May he do so again on Thursday?

Suppose that x is just slightly below r. Then it seems that the absolute deontologist is committed to holding that the surgeon may drive to work on Tuesday, and he may drive to work on Thursday, but he may not drive to work on Tuesday and Thursday.

One might deny that this result is paradoxical. Consider the following case:

Bomb Squad: Tim works for the bomb squad. He has just been called in to defuse a bomb. He knows that if he cuts only the red wire, the bomb will be defused. Also, if he cuts only the blue wire, the bomb will be defused. But if he cuts both wires, the bomb will explode, killing several innocent people.

Here, it is permissible to cut the red wire, and it is permissible to cut the blue wire, but it is impermissible to cut the blue wire and the red wire. This example refutes the principle that if it is permissible to do A, and it is permissible to do B, then it is permissible to do both A and B. However, the example fails to refute the following, more sophisticated principle:

Two Rights Don’t Make a Wrong: If it is permissible to do A, and it is permissible to do B given that one does A, then it is permissible to do both A and B.

In the Bomb Squad case, it is permissible to cut the red wire, but given that one cuts the red wire it is not permissible to cut the blue wire. This is because given that Tim is cutting the red wire, if he cuts the blue wire, the bomb will explode. So the Two Rights Don’t Make a Wrong principle does not entail that it is permissible to cut both wires, and the principle is not open to this sort of counter-example.

However, in Repeated Driving, there is no analogous relationship between the two actions in question. The effects of the surgeon’s driving to work on Thursday are independent of whether the surgeon has at that point driven to work on Tuesday; what happens on Tuesday thus seems irrelevant to what the surgeon may do on Thursday. Therefore, we should say that not only is it permissible to drive to work on Tuesday, but it is also permissible, given that he has driven to work on Tuesday, for the surgeon to drive to work on Thursday.

One might argue that what happens on Tuesday is relevant to what the surgeon may do on Thursday, because if the surgeon drives to work on Tuesday, then his driving to work on Thursday pushes him above the acceptable risk threshold (when we consider the compound action that he will have performed, of driving to work on both days), whereas if he does not drive to work on Tuesday, his driving to work on Thursday does not push him over the risk threshold. On this view, the surgeon may drive to work on Tuesday, but then he may not drive again on Thursday. This seems implausible. Consider another case:

Two Surgeons: A hospital has two surgeons, Mary and Jon, on call. Each of them can drive to the hospital on Tuesday to save some patients, with a probability of x of killing someone in a traffic accident. The hospital calls Jon, and he comes in. The same situation obtains again on Thursday. Jon is a slightly better surgeon than Mary, however, so he will probably save more patients. Whom should the hospital call in on Thursday?

It is implausible to suppose that Mary may drive on Thursday but Jon may not, or that the hospital has good reason for calling Mary because Jon drove on Tuesday.

Another way of attempting to avoid paradox is to maintain that in the Repeated Driving case, “driving on both Tuesday and Thursday” does not name an action—driving on Tuesday is an action, and driving on Thursday is an action, but there is no conjunctive action formed from the two. Our deontological theory only pronounces verdicts on the permissibility of individual actions, so it delivers no verdict on the permissibility of “driving on both Tuesday and Thursday.” It allows the surgeon to drive on Tuesday, it allows him to drive again on Thursday given that he drove on Tuesday, and it does not prohibit his driving on both days.

This view also has implausible results. Consider:

Three Hospitals: A surgeon is deliberating between two options: (i) He can drive to hospital A to save five lives, creating a risk of x of killing someone in a traffic accident, and then drive to hospital B to save five more lives, again creating a risk of x of killing someone in an accident. (ii) He can drive to hospital C to save twenty lives. Hospital C is sufficiently far away that if he drives there, he will have the same probability of killing someone in a traffic accident as he will if he drives to both A and B.

Assume again that x is slightly below the risk threshold. It seems absurd to claim that the surgeon has good reason to prefer option (i), on the grounds that neither driving to hospital A nor driving to hospital B exceeds the risk threshold, whereas driving to hospital C does. It is absurd to suppose, that is, that option (i) is preferable merely by virtue of its spreading the total risk between two actions. One’s moral theory should not be so sensitive to one’s method of individuating actions.

3. Prospects for Absolute Deontology

The problem of risk is a serious one for absolutists. Is it insurmountable? Perhaps not. The most naive form of absolutism is one that takes all rights violations to be absolutely (that is, irrespective of consequences) impermissible and that identifies rights violations by means of our deontological intuitions (that is, it takes rights violations to occur in all those cases in which we intuitively feel some harmful action to be wrong despite its producing better overall consequences than the alternatives). It is this naive sort of absolutism that the preceding arguments refute, but more sophisticated versions of absolutism are possible.

I think the arguments of sections 2.2 and 2.3 show—as perhaps is clear intuitively in any case—that an absolutist ought to take a Zero Risk Tolerance attitude towards potential rights violations. The problem with such a view, as we saw, is that it threatens to condemn too much of ordinary life, since virtually any action creates at least some risk of harming others, and since many of these cases of risk-imposition would, intuitively, constitute rights violations if only the probability of harm were greater. The absolutist’s best move may be to retreat to the position that certain kinds of rights violations are absolutely impermissible, while others may be permissible if they produce sufficiently good results. One might, that is, take an absolute deontological view with respect to some rights violations and a moderate deontological view with respect to others.

One would then need a principle for distinguishing these two kinds of rights violations. One might, for example, employ the doctrine of double effect to identify the absolutely impermissible rights-violations, taking these to be those in which a harm is intended either as an end or a means. One could then take the moderate deontological view with respect to all other intuitive cases of rights violations or potential rights violations, including the Driving, Pollution, Frame-Up, and Indiscriminate Punishment cases. This would enable one to embrace the Zero Risk Tolerance view without forcing one to embrace such counter-intuitive results as that it is always wrong to drive, create pollution, or punish people convicted of crimes. In ordinary circumstances, driving, creating pollution, and (arguably) punishing convicts do not aim at harming innocent persons, even though they create a risk of doing so. This is shown by the fact that, if we should avoid ever having a traffic accident, if polluters should avoid ever causing illness to bystanders, and if our criminal justice system should miraculously avoid ever punishing the innocent, the goals of the relevant agents will be in no way frustrated.

This is not to say that such a hybrid absolute/moderate deontology would be without problems. Most notably, there would be the problem of supplying a rationale for the hierarchy of rights-violations—why should only some rights violations be absolutely impermissible? And the view would still generate some of the more obvious counter-intuitive results—for example, cases in which one’s deontological theory demands that one allow the rest of humanity to perish rather than violate one individual’s rights. But these problems are not the subject of this paper, and they are probably less serious than the problem of risk faced by the naive absolutist view.

The problem of risk, then, does not provide a complete refutation of absolutism. It does, however, put strong pressure on absolutists to make concessions in the direction of either moderate deontology or consequentialism.

References

Anscombe, G. E. M. 1981. Ethics, Religion, and Politics: The Collected Philosophical Papers of G. E. M. Anscombe, vol. 3. Minneapolis, Minn.: University of Minnesota Press.

Foot, Philippa. 1967. “The Problem of Abortion and the Doctrine of the Double Effect,” Oxford Review 5: 5-15.

Kamm, F. M. 1992. “Non-consequentialism, the Person as an End-in-itself, and the Significance of Status,” Philosophy and Public Affairs 21: 354-89.

Kant, Immanuel. 1959. Foundations of the Metaphysics of Morals, tr. Lewis White Beck. Indianapolis, Ind.: Bobbs-Merrill.

McCloskey, H. J. 1957. “An Examination of Restricted Utilitarianism,” Philosophical Review 66: 466-85.

Nozick, Robert. 1974. Anarchy, State, and Utopia. New York: Basic Books.

Simons, Kenneth. 1999. “Negligence,” Social Philosophy and Policy 16(2): 52-93.

Thomson, Judith Jarvis. 1986. Rights, Restitution, and Risk, ed. William Parent. Cambridge, Mass.: Harvard University Press.

Quinn, Warren. 1989. “Actions, Intentions, and Consequences: The Doctrine of Double Effect,” Philosophy and Public Affairs 18: 334-51.

-----------------------

[1]Hereafter, I shall count the avoidance of a harm as a kind of “benefit.”

[2]Kant 1959, p. 17; Anscombe 1981, pp. 39-40, 64-6. Nozick (1974, pp. 28-33) comes close to endorsing such a view, but stops just short (p. 30n).

[3]In particular, I am not suggesting that all rights-violations in the ordinary sense of the term constitute harms, nor must a deontological theorist formulate the principle in terms of rights in the ordinary sense.

[4]This example is from Foot (1967, p. 8), with a modification suggested by Thomson (1986, p. 96).

[5]This example derives originally from James Rachels (in informal conversations in the 1960’s).

[6]Foot (1967) defends this account based on the doctrine of double effect.

[7]Thomson 1986, pp. 83-4.

[8]I assume that in assessing what counts as a “greater” benefit, one compares expected harms and benefits, that is, the probability-weighted sums of the possible harms and benefits.

[9]This leaves open the possibility that the latter sort of potential rights violations are justified, e.g., only when they produce much greater benefits, and not merely whenever they produce greater benefits. One might, that is, take the moderate deontologist’s position with respect to potential rights violations in which the probability of harm is less than 1.

[10]Kamm 1992, pp. 367-8; Quinn 1989, p. 345.

[11]The example is from McCloskey (1957, pp. 468-9).

[12]Strictly speaking, the doctrine of double effect merely fails to provide a rationale for holding that the state’s action in Indiscriminate Punishment is wrong. An advocate of the doctrine of double effect could supplement his view with some further principle that allows one to condemn the state’s action in that case. However, such a view no longer provides a basis for objecting to the argument of this section; as long as one holds the actions in such cases as Driving, Pollution, Frame-up, and Indiscriminate Punishment to be wrong when the risk of harming the innocent is 100%, the Zero Risk Tolerance view implies that many ordinary actions that we take to be permissible are absolutely impermissible.

[13]A related point appears in Simons (1999, p. 63), though he does not draw the conclusion I advance here.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download