¹ Cambridge: Harvard University Press, 1971.

² Ibid., p. 303.

³ Indeed, the parallel to Fourier's “guarantee of the minimum” is very striking. Cf. Design for Utopia, edited by Manuel, Frank E. (New York: Schocker Books, 1971), pp. 19–92Google Scholar.

⁴ Rawls, p. 302.

⁵ See in particular Book V.

⁶ The phrase is Schaar's, John in “Some Ways of Thinking about Equality,” Journal of Politics, 26 (1964), 867–95CrossRefGoogle Scholar.

⁷ See Bell, Daniel, “Meritocracy and Equality,” The Public Interest (Fall, 1972), 29–68Google Scholar.

⁸ See Young, Michael, The Rise of the Meritocracy (Harmondsworth: Penguin Books, 1958)Google Scholar; or Schaar. If, as these authors suggest, meritocracy means that the privileged really are superior by going standards, then the position of the disadvantaged is in a sense worse than it would be under an arbitrary system of assignment.

⁹ For an elegant, if eventually unconvincing argument for a necessary link between laws or other social norms and social inequality, see Dahrendorf, Ralf, “On the Origin of Inequality Among Men,” in his Essays in the Theory of Society (Stanford: University Press, 1968)Google Scholar.

¹⁰ Cited in Bell.

¹¹ Translated by Ann S. Schwier (New York: Augustus M. Kelley, 1971); see also Meade, J. E., Efficiency, Equality and the Ownership of Property (London: George Allen and Unwin Ltd., 1964)Google Scholar.

¹² See, for example, Sen, Amartya K., Collective Choice and Social Welfare (San Francisco: Holden Day, 1970), p. 21 ff, 56 ffGoogle Scholar.

¹³ Rawls, especially pp. 90–95.

¹⁴ This is Rawls's name for the gathering at which a social contract is struck. I discuss its peculiarities in Part 2 below.

¹⁵ The inference that unanimity vouchsafes shared benefit presupposes both a static environment and a fair point of no agreement. See my article, “The Limits of Consensual Decision,” American Political Science Review (09, 1975)Google Scholar, forthcoming.

¹⁶ A convenient summary of these conditions will be found in Rawls, pp. 146–47.

¹⁷ Ibid., p. 19.

¹⁸ Ibid., pp. 18–19

¹⁹ Ibid., p. 148.

²⁰ The gist of the case for utilitarianism is as follows: (1) I assume that I am equally apt to occupy each social position, and (2) I want to make the resulting lottery as attractive as possible, so (3) I try to maximize the average or total without attention to distribution. This is of course complicated by attitudes toward risk and the non-linearity of personal utility functions. See in particular Harsanyi, John, “Cardinal Welfare, Individualistic Ethics, and Interpersonal Comparisons of Utility,” Journal of Political Economy, 63 (08, 1955), 309–321CrossRefGoogle Scholar.

²¹ Rawls, p. 154. The other two features are explained in the same passage.

²² If, however, inferences (c) and (d) are put aside, the connection is far less direct. For if the threshold is above the maximum point, then the rule must be rejected, yet if it falls below the maximin outcome then we have Rawls's principle as one of many possible principles, and the argument takes on a logic resembling:

the king of birds is in this forest;

a robin is in this forest;

therefore a robin is the king of birds.

See Douglas Rae and James Fishkin, “A Weak Theory of Justice,” unpublished paper, Yale University; and see Barry, Brian, The Liberal Theory of Justice (Oxford: Oxford University Press, 1973)Google Scholar.

²³ Rawls, p. 303.

²⁴ Fellner, William, Probability and Profit (Home-wood, Ill.: Richard D. Irwin Inc., 1965). p. 140 ffGoogle Scholar. Fellner in fact regards maximin as a very peculiar way to do business (which is after all his subject) and cautions that it ought never be adopted as a “mechanical decision-rule.” This is actually what Rawls proposed for the parties to his original position.

²⁵ Cf. Barry, p. 103.

²⁶ Rawls, p. 151.

²⁷ “Distributive Justice,” in Philosophy, Politics and Society, Series III, Lasket, Peter and Runciman, W. G., eds. (Oxford: Basil Blackwell, 1964), especially pp 80–81Google Scholar. See also Rawls, , Theory of Justice, p. 317 ffGoogle Scholar.

²⁸ Rawls, p. 83.

²⁹ Notice that this is much less general than the full utilitarian case for inequality, for we are not permitted to sacrifice the interests of the least advantaged. And if envy and calculations from lost opportunity are put aside, the poor must favor such inequalities.

³⁰ By “minimum test,” it will perhaps be recalled, I mean the rule that no acceptable principle should at once, in a single judgment, urge us to violate the principles of equality and utilitarian maximization.

³¹ In the proof given at p. 22 above, the inferences given at (4) and (5) collapse if a middle group is introduced.

³² The numerical calculations are as follows:

The example is not peculiar, and a little tinkering will suggest that innumerable analogous ones are lo be found.

³³ Rawls, p. 80.

³⁴ Namely, those with exactly two positions and those in which one available alternative is absolute equality.

³⁵ The continuum would be measured out against Def. 1, p. 10, (1.3) above. It should be noted that the measurement of inequality is far from simple, and indeed our intuitive sense may lead only to a partial ordering. See Sen, Amartya K., Economic Inequality (New York: W. W. Norton, 1973)CrossRefGoogle Scholar.

³⁶ The assumption is not of course a novel one. It appears in the Nicomachean Ethics, Book V, and a great many other places. See, for instance, William K. Frankena, “The Concept of Social Justice,” and Vlastos, Gregory, “Justice and Equality,” both in Social Justice, ed. Brandt, Richard (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1962)Google Scholar.

³⁷ The inclusion of information about social strata may appear arbitrary, but it is not. For, if we are to agree upon principles of allocation, relative benefits from social allocation appear to be referents of argument by analytic necessity and this holds, so far as I can discover, for no other facts.

³⁸ Suppose, for example, that you represent one stratum and I another. We make a log-rolling bargain under which my stratum suffers so that yours can gain in Case A, while yours suffers in favor of mine in Case B. I cannot explain my agency to those affected by A and you cannot explain your agency to those affected by B, for the particular individuals (or positions) may very well turn out to be different. Neither of us can say we have represented Smith by sacrificing his interests in one case in favor of some Jones who happens to hold the same rank in another case and perhaps even another historical period.

³⁹ The argument from unanimity here depends on two quite unworldly aspects of the process: (1) there is an enforceable point of no-agreement which gives absolute equality, and (2) the environment is so static that “nothing happens” except by collective decision. This is obviously a most artificial construct, and bears little resemblance to experience. Cf. Rae. I am also using unanimity in the slightly unconventional sense that indifference is ignored: it counts as unanimity if there is at least one yea, no nays, and an unlimited number of abstentions.

⁴⁰ This intuition, of course, arises almost everywhere from the Golden Rule to the Categorical Imperative. Mechanisms for its implementation by means of forced empathy have been with us from ancient times. See in particular Baumol, William J., “Economics of Athenian Drama,” Quarterly Journal of Economics, 85, no. 3 (1971), 365–76CrossRefGoogle Scholar; or Luce, R. D. and Raiffa, Howard, Games and Decisions (New York: John Wiley and Sons, 1957), p. 363 ffGoogle Scholar. or Rawls, p. 85.

⁴¹ This rule is indeed a cardinal (versus ordinal) analog for what the welfare economists would call a “compensation test”: a gain for one person is legitimate if and only if he can compensate all losers yet still retain a surplus for himself.

By using a cardinal framework, I have put aside much of the complexity which surrounds these tests in the work of Kaldor, Skitovsky, Hicks, and Samuelson. As will be seen below, however, my principle does not escape the transitivity problems which plague tests of compensation in welfare economies. For a convenient introduction to this literature, see Mishan, E. J., Welfare Economics: An Appraisal (The Hague: North Holland, 1970)Google Scholar; or Little, J. M. D., A Critique of Welfare Economics, 2nd edition (Oxford: Oxford University Press, 1957)Google Scholar.

⁴² Rawls, p. 303.

⁴³ Social Choice and Individual Values, 2nd edition (New Haven: Yale University Press, 1963)Google Scholar. For more on the present point see Barry, Brian and Rae, Douglas, “Political Evaluation,” Handbook of Political Science, Polsby, Nelson and Greenstein, Fred, eds. (Reading, Mass.: Addison-Wesley, forthcoming, 1975)Google Scholar.

⁴⁴ Suppose we have some X, Y, and Z such that:

(i) X is chosen from (X, Y) by general advantage,

(ii) Y is chosen from (Y, Z) by equality and the absence of general advantage,

(iii) Z is chosen from (X, Z) by equality and the absence of general advantage.

We will, in such a case, be led to an intransitivity of the kind suggested in the text. If a numerical example is required, let X provide 100 to the first stratum and twenty to all others; let Y provide twenty to all strata; and let Z provide fifty to all strata save the lowest, which receives 10. Rawls escapes this by violating another of Arrow's conditions, “non-dictatorship,” as is shown by Charles Plott's “Rawls's Theory of Justice: An Impossibility Result,” unpublished paper, Public Choice Society, New Haven, 1974.