1 I use the word "practice" throughout as a sort of technical term meaning any form of activity specified by a system of rules which defines offices, roles, moves, penalties, defenses, and so on, and which gives the activity its structure. As examples one may think of games and rituals, trials and parliaments.

2 The distinction is central to Hume's discussion of justice in A Treatise of Human Nature, bk. III, pt. ii, esp. secs. 2-4. It is clearly stated by John Austin in the second lecture of Lectures on Jurisprudence (4th ed.; London, 1873), I, 116ff. (1st ed., 1832). Also it may be argued that J. S. Mill took it for granted in Utilitarianism; on this point cf. J. 0. Urmson, "The Interpretation of the Moral Philosophy of J. S. Mill," Philosophical Quarterly, vol. III (1953). In addition to the arguments given by Urmson there are several clear statements of the distinction in A System of Logic (8th ed.; London, 1872), bk. VI, ch. xii pars. 2, 3, 7. The distinction is fundamental to J. D. Mabbott's important paper, "Punishment," Mind, n.s., vol. XLVIII (April, 1939). More recently the distinction has been stated with particular emphasis by S. E. Toulmin in The Place of Reason in Ethics (Cambridge, 1950), see esp. ch. xi, where it plays a major part in his account of moral reasoning. Toulmin doesn't explain the basis of the distinction, nor how one might overlook its importance, as I try to in this paper, and in my review of this book (Philosophical Review, vol. LX [October, 1951]), as some of my criticisms show, I failed to understand the force of it. See also H. D. Aiken, "The Levels of Moral Discourse," Ethics, vol. LXII (1952), A. M. Quinton, "Punishment," Analysis,vol. XIV (June, 1954), and P. H. Nowell-Smith, Ethics (London, 1954), pp. 236-239, 271-273.

3 On the concept of explication see the author's paper Philosophical Review, vol. LX (April, 1951).

4 While this paper was being revised, Quinton's appeared; footnote 2 supra. There are several respects in which my remarks are similar to his. Yet as I consider some further questions and rely on somewhat different arguments, I have retained the discussion of punishment and promises together as two test cases for utilitarianism.

5 Note the fact that different sorts of arguments are suited to different offices. One way of taking the differences between ethical theories is to regard them as accounts of the reasons expected in different offices.

6 In this connection see Mabbott, op. cit., pp. 163-164.

7 On this point see Sir David Ross, The Right and the Good (Oxford, 1930), pp. 57-60.

8 See Hobbes's definition of punishment in Leviathan, ch. xxviii; and Bentham's definition in The Principle of Morals and Legislation, ch. xii, par. 36, ch. xv, par. 28, and in The Rationale of Punishment, (London, 1830), bk. I, ch. i. They could agree with Bradley that: "Punishment is punishment only when it is deserved. We pay the penalty, because we owe it, and for no other reason; and if punishment is inflicted for any other reason whatever than because it is merited by wrong, it is a gross immorality, a crying injustice, an abominable crime, and not what it pretends to be." Ethical Studies (2nd ed.; Oxford, 1927), pp. 26-27. Certainly by definition it isn't what it pretends to be. The innocent can only be punished by mistake; deliberate "punishment" of the innocent necessarily involves fraud.

9 Cf. Leon Radzinowicz, A History of English Criminal Law: The Movement for Reform 1750-1833 (London, 1948), esp. ch. xi on Bentham.

10 Bentham discusses how corresponding to a punitory provision of a criminal law there is another provision which stands to it as an antagonist and which needs a name as much as the punitory. He calls it, as one might expect, the anaetiosostic, and of it he says: "The punishment of guilt is the object of the former one: the preservation of innocence that of the latter." In the same connection he asserts that it is never thought fit to give the judge the option of deciding whether a thief (that is, a person whom he believes to be a thief, for the judge's belief is what the question must always turn upon) should hang or not, and so the law writes the provision: "The judge shall not cause a thief to be hanged unless he have been duly convicted and sentenced in course of law" (The Limits of Jurisprudence Defined, ed. C. W. Everett [New York, 1945], pp. 238-239).

11 By the classical utilitarians I understand Hobbes, Hume, Bentham, J. S. Mill, and Sidgwick.

12 All these features of punishment are mentioned by Hobbes; cf. Leviathan, ch. xxviii.

13 Ethical and Political Thinking (Oxford, 1947), p. 65.

14 The analogy with the price system suggests an answer to the question how utilitarian considerations insure that punishment is proportional to the offense. It is interesting to note that Sir David Ross, after making the distinction between justifying a penal law and justifying a particular application of it, and after stating that utilitarian considerations have a large place in determining the former, still holds back from accepting the utilitarian justification of punishment on the grounds that justice requires that punishment be proportional to the offense, and that utilitarianism is unable to account for this. Cf. The Right and the Good, pp. 61-62. I do not claim that utilitarianism can account for this requirement as Sir David might wish, but it happens, nevertheless, that if utilitarian considerations are followed penalties will be proportional to offenses in this sense: the order of offenses according to seriousness can be paired off with the order of penalties according to severity. Also the absolute level of penalties will be as low as possible. This follows from the assumption that people are rational (i.e., that they are able to take into account the "prices" the state puts on actions), the utilitarian rule that a penal system should provide a motive for preferring the less serious offense, and the principle that punishment as such is an evil. All this was carefully worked out by Bentham in The Principles of Morals and Legislation, chs. xiii-xv.

15 Ross, The Right and the Good, pp. 37-39, and Foundations of Ethics (Oxford, 1939), pp. 92-94. I know of no utilitarian who has used this argument except W. A. Pickard-Cambridge in "Two Problems about Duty," Mind, ii.s., XLI (April, 1932), 153-157, although the argument goes with G. E. Moore's version of utilitarianism in Principia Ethica (Cambridge, 1903). To my knowledge it does not appear in the classical utilitarians; and if one interprets their view correctly this is no accident.

16 Ross, The Right and the Good, pp. 38-39.

17 Ross, ibid., p. 39. The case of the nonpublic promise is discussed again in Foundations of Ethics, pp. 95-96, 104-105. It occurs also in Mabbott, "Punishment," op. cit., pp. 155-157, and in A. I. Melden "Two Comments on Utilitarianism," Philosophical Review, LX (October, 1951), 519-523, which discusses Carritt's example in Ethical and Political Thinking, p. 64.

18 Ross's example is described simply as thar of two men dying alone where one makes a promise to the other. Carritt's example (cf n. 17 supra) is that of two men at the North Pole. The example in the text is more realistic and is similar to Mabbott's. Another example is that of being told something in confidence by one who subsequently dies. Such cases need not be "desert-island arguments" as Nowell-Smith seems to believe (cf. his Ethics, pp. 239-244).

19 What I have said in this paragraph seems to me to coincide with Hume's important discussion in the Treatise of Human Nature, bk. III, pt. ii, sec. 5; and also sec. 6, par. 8.

20 For a discussion of these, see H. Sidgwick, The Methods of Ethics (6th ed.; London, 1901), bk. III, ch. vi.

21 So far as I can see it is not until Moore that the doctrine is expressly stated in this way. See, for example, Principia Ethica, p. 147, where it is said that the statement "I am morally bound to perform this action" is identical with the statement "This action will produce the greatest possible amount of good in the Universe" (my italics). It is important to remember that those whom I have called the classical utilitarians were largely interested in social institutions. They were among the leading economists and political theorists of their day, and they were not infrequently reformers interested in practical affairs. Utilitarianism historically goes together with a coherent view of society, and is not simply an ethical theory, much less an attempt at philosophical analysis in the modern sense. The utilitarian principle was quite naturally thought of, and used, as a criterion for judging social institutions (practices) and as a basis for urging reforms. It is not clear, therefore, how far it is necessary to amend utilitarianism in its classical form. For a discussion of utilitarianism as an integral part of a theory of society, see L. Robbins, The Theory of Economic Policy in English Classical Political Economy (London, 1952).

22 This footnote should be read after sec. 3 and presupposes what I have said there. It provides a few references to statements by leading utilitarians of the summary conception. In general it appears that when they discussed the logical features of rules the summary conception prevailed and that it was typical of the way they talked about moral rules. I cite a rather lengthy group of passages from Austin as a full illustration.

John Austin in his Lectures on Jurisprudence meets the objection that deciding in accordance with the utilitarian principle case by case is impractical by saying that this is a misinterpretation of utilitarianism. According to the utilitarian view ". . . our conduct would conform to rules inferred from the tendencies of actions, but would not be determined by a direct resort to the principle of general utility. Utility would be the test of our conduct, ultimately, but not immediately: the immediate test of the rules to which our conduct would conform, but not the immediate test of specific or individual actions. Our rules would be fashioned on utility; our conduct, on our rules" (vol. I, p. 116). As to how one decides on the tendency of an action he says: "If we would try the tendency of a specific or individual act, we must not contemplate the act as if it were single and insulated, but must look at the class of acts to which it belongs. We must suppose that acts of the class were generally done or omitted, and consider the probable effect upon the general happiness or good. We must guess the consequences which would follow, if the class of acts were general; and also the consequences which would follow, if they were generally omitted. We must then compare the consequences on the positive and negative sides, and determine on which of the two the balance of advantage lies. . . . If we truly try the tendency of a specific or individual act, we try the tendency of the class to which that act belongs. The particular conclusion which we draw, with regard to the single act, implies a general conclusion embracing all similar acts. . . . To the rules thus inferred, and lodged in the memory, our conduct would conform immediately if it were truly adjusted to utility" (ibid., p. 117). One might think that Austin meets the objection by stating the practice conception of rules; and perhaps he did intend to. But it is not clear that he has stated this conception. Is the generality he refers to of the statistical sort? This is suggested by the notion of tendency. Or does he refer to the utility of setting up a practice? I don't know; but what suggests the summary view is his subsequent remarks. He says: "To consider the specific consequences of single or individual acts, would seldom [my italics] consist with that ultimate principle" (ibid., p. 117). But would one ever do this? He continues: ". . . this being admitted, the necessity of pausing and calculating, which the objection in question supposes, is an imagined necessity. To preface each act or forebearance by a conjecture and comparison of consequences, were clearly superfluous [my italics] and mischievous. It were clearly superfluous, inasmuch as the result of that process [my italics] would be embodied in a known rule. It were clearly mischievous, inasmuch as the true result would be expressed by that rule, whilst the process would probably be faulty, if it were done on the spur of the occasion" (ibid., pp. 117-118). He goes on: "If our experience and observation of particulars were not generalized, our experience and observation of particulars would seldom avail us in practice. . . . The inferences suggested to our minds by repeated experience and observation are, therefore, drawn into principles, or compressed into maxims. These we carry about us ready for use, and apply to individual cases promptly . . . without reverting to the process by which they were obtained; or without recalling, and arraying before our minds, the numerous and intricate considerations of which they are handy abridgments [my italics]. . . . True theory is a compendium of particular truths. . . . Speaking then, generally, human conduct is inevitably guided [my italics) by rules, or by principles or maxims" (ibid., pp. 117-118). I need not trouble to show all these remarks incline to the summary view. Further, when Austin comes to deal with cases "of comparatively rare occurrence" he holds that specific considerations may outweigh the general. "Looking at the reasons from which we had inferred the rule, it were absurd to think it inflexible. We should therefore dismiss the rule; resort directly to the principle upon which our rules were fashioned; and calculate specific consequences to the best of our knowledge and ability" (ibid., pp. 120-121). Austin's view is interesting because it shows how one may come close to the practice conception and then slide away from it.

In A System of Logic, bk. VI, ch. xii, par. 2, Mill distinguishes clearly between the position of judge and legislator and in doing so suggests the distinction between the two concepts of rules. However, he distinguishes the two positions to illustrate the difference between cases where one is to apply a rule already established and cases where one must formulate a rule to govern subsequent conduct. It's the latter case that interests him and he takes the "maxim of policy" of a legislator as typical of rules. In par. 3 the summary conception is very clearly stated. For example, he says of rules of conduct that they should be taken provisionally, as they are made for the most numerous cases. He says that they "point out" the manner in which it is least perilous to act; they serve as an "admonition" that a certain mode of conduct has been found suited to the most common occurrences. In Utilitarianism, ch. ii, par. 24, the summary conception appears in Mill's answer to the same objection Austin considered. Here he speaks of rules as "corollaries" from the principle of utility; these "secondary" rules are compared to "landmarks" and "direction-posts." They are based on long experience and so make it unnecessary to apply the utilitarian principle to each case. In par. 25 Mill refers to the task of the utilitarian principle in adjudicating between competing moral rules. He talks here as if one then applies the utilitarian principle directly to the particular case. On the practice view one would rather use the principle to decide which of the ways that make the practice consistent is the best. it should be noted that while in par. 10 Mill's definition of utilitarianism makes the utilitarian principle apply to morality, i.e., to the rules and precepts of human conduct, the definitioc in par. 2 uses the phrase "actions are right in proportion as they tend to promote happiness" [my italics] and this inclines towards the summary view. In the last paragraph of the essay "On the Definition of Political Economy," Westminster Review (October, 1836), Mill says that it is only in art, as distinguished from science, that one can properly speak of exceptions. In a question of practice, if something is fit to be done "in the majority of cases" then it is made the rule. "We may . . . in talking of art unobjectionably speak of the rule and the exception, meaning by the rule the cases in which there exists a preponderance . . . of inducements for acting in a particular way; and by the exception, the cases in which the preponderance is on the contrary side." These remarks, too, suggest the summary view.

In Moores Principia Ethica, ch. v, there is a complicated and difficult discussion of moral rules. I will not examine it here except to express my suspicion that the summary conception prevails. To be sure, Moore speaks frequently of the utility of rules as generally followed, and of actions as generally practiced, but it is possible that these passages fit the statistical notion of generality which the summary conception allows. This conception is suggested by Moore's taking the utilitarian principle as applying directly to particular actions (pp. 147-148) and by his notion of a rule as something indicating which of the few alternatives likely to occur to anyone will generally produce a greater total good in the immediate future (p. 154). He talks of an "ethical law" as a prediction, and as a generalization (pp. 146, 155). The summary conception is also suggested by his discussion of exceptions (pp. 162-163) and of the force of examples of breaching a rule (pp. 163-164).

23 One might feel that it is a mistake to say that a practice is logically prior to the forms of action it specifies on the grounds that if there were never any instances of actions falling under a practice then we should be strongly inclined to say that there wasn't the practice either. Blue-prints for a practice do not make a practice. That there is a practice entails that there are instances of people having been engaged and now being engaged in it (with suitable qualifications). This is correct, but it doesn't hurt the claim that any given particular instance of a form of action specified by a practice presupposes the practice. This isn't so on the summary picture, as each instance must be "there" prior to the rules, so to speak, as something from which one gets the rule by applying the utilitarian principle to it directly.

24 A philosophical joke (in the mouth of Jeremy Bentham): "When I run to the other wicket after my partner has struck a good ball I do so because it is best on the whole."

25 How do these remarks apply to the case of the promise known only to father and son? Well, at first sight the son certainIy holds the office of promisor, and so he isn't allowed by the practice to weigh the particular case on general utilitarian grounds. Suppose instead that he wishes to consider himself in the office of one empowered to criticize and change the practice, leaving aside the question as to his right to move from his previously assumed office to another. Then he may consider utilitarian arguments as applied to the practice; but once he does this he will see that there are such arguments for not allowing a general utilitarian defense in the practice for this sort of case. For to do so would make it impossible to ask for and to give a kind of promise which one often wants to be able to ask for and to give. Tlierefore he will not want to change the practice, and so as a promisor he has no option but to keep his promise.

26 Philosophical Investigations (Oxford, 1953), 1, pars. 65-71, for example.

27 As I have already stated, it is not always easy to say where the conception is appropriate. Nor do I care to discuss at this point the general sorts of cases to which it does apply except to say that one should not take it for granted that it applies to many so-called "moral rules." It is my feeling that relatively few actions of the moral life are defined by practices and that the practice conception is more relevant to understanding legal and legal-like arguments than it is to the more complex sort of moral arguments. Utilitarianism must be fitted to different conceptions of rules depending on the case, and no doubt the failure to do this has been one source of difficulty in interpreting it correctly.