¹ Taylor, Richard W., in his introductory essay to the collection, Life, Language, Law: Essays in Honor of Arthur F. Bentley (1957), p. 16Google Scholar.

² The Political System (1953), p. 224Google Scholar. It is doubtful that we would really want to say that a locution expressed a proposition if all that it did was “express a sentiment.”

³ The Governmental Process (1951), p. 296Google Scholar.

⁴ In Chapter VII of Scientific Explanation (1955), R. B. Braithwaite expounds a theory according to which the choice between statistical hypotheses depends upon “the gains or losses occasioned by choosing one or the other of these hypotheses when it is true or when it is false.” Following Wald, Braithwaite holds that on the basis of these expected gains or losses we can develop “prudential strategies” for choosing between hypotheses; and, drawing, like Wald, upon the theory of games, he argues that these strategies when fully developed will be randomized ones. He concedes that “a strategy which is the prudential one for the assignment of values to the ‘utilities’ by one person will not be the prudential strategy for another such assignment by another person.” This raises the possibility, which Braithwaite does not sufficiently investigate, that people may differ in the statistical hypotheses which they accept or reject without being able to have recourse to common standards of scientific acceptability. I can only point out here that it is a long and complicated road from Braithwaite's theory to contesting the position I have adopted regarding the independence of facts and values. In the course of developing their strategies, people would, on Braithwaite's theory, have to consider not only the gains that they could expect from believing that a hypothesis was true when it was true, but also the losses that they would suffer from believing that it was true when it was false. Thus it is not a simple matter of believing something just because one very much wants it to be true. Furthermore, Braithwaite's theory is not the only one in the field; and whether it can be accepted will depend, I would say, on whether it can be reconciled with the position which I have defended in the text above. This is a problem which has hardly been ventilated, much less explored.

⁵ Kant's categorical imperative belongs here, too, in spite of his protestations that attending to consequences was “heteronomous” and unworthy. See the fine article by Singer, Marcus G., “The Categorical Imperative,” Philosophical Review, Vol. 63 (October, 1954), pp. 577–591CrossRefGoogle Scholar.

⁶ New York (1958). See “Creeping Inflation Resulting from Wage Increases in Excess of Productivity,” pp. 137–146.

⁷ Loc. cit., p. 146.

⁸ The Politics of Democracy (1940), p. 158Google Scholar.

⁹ Schattschneider, E. E., Party Government (1942), p. 208Google Scholar.

¹⁰ See Merton, R. K., Social Theory and Social Structure (1949)Google Scholar; and Levy, Marion, The Structure of Society (1952)Google Scholar; also various works of Talcott Parsons. Dorothy Emmet, in her recent book, Function, Purpose and Powers (1958), gives a good extended discussion of the matters that I treat in this section; social scientists would find it, I think, both interesting and sensible.

¹¹ See Merton, op. cit.; and Levy, op. cit. I am also drawing upon a lecture on “Functional Analysis in Social Science,” given at Yale, May 9, 1958, by Carl G. Hempel. During the discussion following that lecture Professor Omar K. Moore maintained that the functional approach had been useful earlier, for example in the days of Malinowski's field work in anthropology; but that it was not helpful today, since the so-called “functionalists” are not trying to do anything so ambitious as formulate the concept of a social system from the ground up, and are, in fact, obstructing the growth of stricter theories and explanations. If the functional approach is something of an anachronism in sociology, however, it may not be so in political science; and a number of points will emerge in this and the subsequent sections of this paper that could be interpreted as reasons for thinking it is not an anachronism.

¹² Among the philosophers chargeable with this error are Cohen, L. Jonathan, in “Teleological Explanation,” Proceedings of the Aristotelian Society, 1950–1951, pp. 255–293Google Scholar (cited by Miss Emmet, op. cit. p. 48); and Nagel, Ernest, in “Teleological Explanation and Teleological Systems,” reprinted in Feigl, Herbert and Brodbeck, May, Readings in the Philosophy of Science (1953), pp. 537–558Google Scholar. In the passage which I have in mind from the latter article, p. 541, Nagel allows, as a secondary possibility, “O is a necessary and sufficient condition of F”; this allowance does not go far enough, or even depart in the right direction.

¹³ So for the general case, when “function” is used descriptively. In some particular cases (not biological ones), it is also implied (c) that human beings have planned that O should do F. On the other hand, when “function” is used prescriptively, (a) may drop out, as we shall see below.

¹⁴ Herring, op. cit., ch. 16.

¹⁵ Schattschneider, op. cit., pp. 157–158.

¹⁶ American State Politics, 1956, p. 12Google Scholar.

¹⁷ These may be compared with, and taken to summarize, the purposes avowed in the formation of the Consitution of the United States, as set forth in the Preamble.

¹⁸ This does not, of course, exhaust the ordinary meaning of “justice,” a concept that would in fact be invoked in discussing rights and procedures under each of the three chief heads. Nevertheless, “justice” seems to be the most appropriate term for what I have in mind for the first chief head: the pattern of present sharing, considered as a moral topic.

¹⁹ The theory mentioned is what many people have in mind when they speak of the “emotive” theory of ethics, attributing it to Ayer, A. J., Language, Truth, and Logic (1936; 2d ed., 1946)Google Scholar, and Stevenson, C. L., Ethics and Language (1945)Google Scholar. It is not their theory, however; nor do I know of anyone who champions it. Both Ayer and Stevenson recognize that most of the judgments seriously considered in ethical discourse state facts as well as express emotions or commands; and they neither deny that there are reasons generally held to have precedence over personal inclinations as grounds for accepting or rejecting moral judgments, nor do they commit the absurdity of making such a denial as an incident of making moral judgments of their own. It is true that both of them grossly underrate the importance of these other reasons, and of such moves in ordinary ethical discourse as to discredit a moral judgment by showing that it does not conform to the received necessary conditions for valid moral judgments (such as consistency and disinterestedness), and merely expresses a personal emotion. Nor is their theory adequate to explaining our attachment to such criteria. For a more adequate explanation, see Toulmin, Stephen, The Place of Reason in Ethics (1950)Google Scholar.

²⁰ It is easy to exaggerate the extent of disagreements relating to ethics. If one starts with certain questions—questions about foundations, for example, or hard cases in which received principles notoriously conflict—a disturbing picture of hopeless confusion emerges very promptly. Things are bad; but not so bad as that. The trick, both in theorizing about ethics and in doing it, is to raise questions in the order that will maximize agreement, and (I would guess) no philosopher has yet brought the trick off completely. I think David Hume came nearest.

²¹ It could not, I think, be proved that all the logically distinct theorems that could be derived from the given axioms had in fact been derived and written down. Even a very simple logical system allows us to state infinitely many well-formed formulas, and which of these could be interpreted as causal laws could not be determined in advance. We would have to formulate them one by one and see.

²² Especially upon his monograph, The Formal Analysis of Normative Systems (1956, Yale Interaction Laboratory)Google Scholar. See also Anderson, A. R. and Moore, O. K., “The Formal Analysis of Normative Concepts,” American Sociological Review, Vol. 22 (February 1957), pp. 9–17CrossRefGoogle Scholar.

²³ This expression is given, following Anderson, in the so-called Polish notation. It could be rendered (again adding operators to the basic stock) equally well in the more familiar, but typographically more demanding, notation of Principia Mathematica.

²⁴ Anderson's illustration is that we should not borrow money when it is impossible for us to repay it.

²⁵ By coincidence, a quantified logical calculus is called a “functional calculus”; this usage springs from the mathematical concept of function, not the social and political one.