1 Elster, Jon The Cement of Society: A Study of Social Order (New York: Cambridge University Press 1989), 215CrossRefGoogle Scholar. Henceforth, CS.

2 CS, ch. 3; see also Elster, Jon ‘Norms of Revenge,’ Ethics 100 (1990) 862–85CrossRefGoogle Scholar.

3 CS, 98; italics in original.

4 Elster, Jon'Rationality, Emotions, and Social Norms,’ Synthese 98 (1994) 21–49Google Scholar, at 28

5 He discusses three varieties of reductionism: ‘The first, (A), says that norms are nothing but ex post rationalizations of self-interest; the second, (B), that people follow norms because of a(n outcome-oriented) fear of being punished when they do not; and the third, (C), that non-consequentialist behavior can be explained by the fact that it often has good consequences’ (ibid.). I find the arguments against reductionism quite convincing.

6 Elster, Jon Sour Grapes (New York: Cambridge University Press 1983)CrossRefGoogle Scholar, ch. 1, esp. 5f.

7 This type of norm, which includes the norm of fairness ('cooperate if and only if everybody else, or at least a substantial number of others, cooperate’) is important for explaining collective action; see CS ch. 5.

8 Elster, Jon Nuts and Bolts for the Social Sciences (New York: Cambridge University Press 1989), 113Google Scholar

9 A reader drew my attention to the fact that regret, which is backward-looking, can still lead to future improvement. If one regrets having carelessly miscalculated, one may resolve to do better next time. The statement has to be qualified for other reasons, too. I will later argue that motives like envy and honor do not exclude calculation.

10 It is significant that coordination norms, such as traffic rules, which are one of the three prime types of norms distinguished by Ullmann-Margalit, EdnaThe Emergence of Norms (Oxford: Clarendon 1977)Google Scholar, are not even counted as social norms by Elster; he calls them ‘convention equilibria,’ which he treats as a distinct category: see his CS, 101-2.

11 This way of defining social norms is common among sociologists and cultural anthropologists. Philosophers influenced by the theory of games prefer to define social norms as solutions to problems of social interaction; whether the solution is group-specific is irrelevant. In that case social norms include moral norms.

12 That such a defence is at all possible has been questioned, for instance, by Macintyre, Alasdair After Virtue, 2nd ed. (Notre Dame, IN: Notre Dame University Press 1984)Google Scholar.

13 An alternative explanation was suggested to me, in terms of semi-conscious reasons such as ‘Avoid seeming threatening.’ This might be plausible in other cases, but it presupposes a shared norm. I do not think there was one in this case: the limit that the accordion player overstepped was important to me, but held no significance for him. That is what made the situation comical.

14 It should be stressed that in Elster's version of rational choice theory, unselfish motives (altruism on the one hand, envy and spite on the other) have a place: compare the typology of individual motivations in CS, 35f. Note, too, that Elster did not always place social norms outside the realm of rational choice. In an earlier article, he distinguished four types of motivation that may enter into rational choice explanations of collective behavior. The fourth was motivation by ‘unselfish, process- oriented benefits,’ and he suggested that ‘the elusive category of social norms' might have a place there Elster, Jon ‘Rationality, Morality, and Collective Action,' Ethics 96 (1985) 136–55CrossRefGoogle Scholar, at 145).

15 Strawson, P.F. ‘Freedom and Resentment,’ Proceedings of the British Academy 48 (1962) 1–25Google Scholar, esp. section III

16 CS, 137; the reference is to Coleman, James ‘Systems of Trust,’ Angewandte Sozialforschung 10 (1982) 277–300Google Scholar.

17 Hence, the conduct of the merchant bankers is rational in the sense to be explained later: it can be justified by an appeal to reasons their fellow bankers cannot reject. By contrast, saying that someone pursues revenge ‘relentlessly’ or that a trade union leader ‘would rather fight and lose than not fight at all’ is to indicate that this person will have difficulty in persuading even his allies that his course is the right one.

18 It has proved difficult to pinpoint the reasons for treating some norms as inflexible. I am grateful for discussions on this issue with Gijs van Donselaar and Pieter Pekelharing.

19 Robert Edgerton's study of the ways in which societies treat exceptions to their own rules supports this pragmatic explanation of inflexible constraints. He identifies three conditions that make for strict rules: perception of danger, need for close cooperation (as in military endeavors), and great inequality prompting measures to protect existing advantages. In all three conditions, the features I mentioned are present: the importance of the value to be protected, and its vulnerability. See Edgerton, Robert B. Rules, Exceptions, and Social Order (Berkeley, CA: University of California Press 1985)Google Scholar, ch. 11.

20 This discussion of rigid constraints and inflexibility was partly inspired by Nozick, Robert's analysis of side-constraints in Anarchy, State, and Utopia (New York: Basic Books 1974), 28–33Google Scholar and 48-51. But he would probably not approve of the pragmatic twist I have given to the concept.

21 On the difference between social practices as they are officially presented, and as they are exploited for strategic purposes, see the account of how marriages are arranged among the Kabyle in Bourdieu, P. Outline of a Theory of Practice, trans. Nice, Richard (New York: Cambridge University Press 1977), 58–71CrossRefGoogle Scholar.

22 See Hamlin, Alan P. ‘Rational Revenge,’ Ethics 101 (1991) 374–81CrossRefGoogle Scholar; this is a reply to Elster's ‘Norms of Revenge.'

23 Sen, A.K. ‘Rational Fools,’ Philosophy and Public Affairs 6 (1977) 317–44Google Scholar

24 The argument about the shortcomings of rational choice theory and game theory in this respect owes much to Hollis, Martin The Philosophy of Social Science (New York: Cambridge University Press 1994)CrossRefGoogle Scholar, ch. 9.

25 Hume, David A Treatise of Human Nature, Selby-Bigge, L.A. and Nidditch, P.H. eds. (Oxford: Clarendon 1978)Google Scholar III, II, section 1; the miser example is at 482.

26 Scanlon, T.M. ‘Contractualism and Utilitarianism,’ in Sen, Amartya and Williams, Bernard eds., Utilitarianism and Beyond (New York: Cambridge University Press 1982)103–28Google Scholar,at116

27 In Rawls, John A Theory of Justice (Cambridge, MA: Harvard University Press 1971),Google Scholar the notion of impartial grounds is connected with that of choice behind a veil of ignorance. Scanlon, on the other hand, appeals to ‘the idea of what it would be unreasonable for people to reject given that they are seeking a basis for general agreement’ (122). This implies that people may be fully aware of their social position, but using this knowledge for purposes of special pleading is simply counterproductive. Obviously, Scanlon's approach is a more promising starting point for an account of how social norms can be binding than the Rawlsian model.

28 Barry, Brian Theories of Justice (Berkeley, CA: University of California Press 1989)CrossRefGoogle Scholar, section 35. Scanlon is quoted on 284. See also section 19.

29 Coleman, James Foundations of Social Theory (Cambridge, MA: Harvard University Press 1990)Google Scholar, chs. 10 and 11, esp. 246f. Disjoint norms correspond to what UllmannMargalit calls ‘norms of partiality'; see her ch. 4.

30 Notice that the rationality of compliance with a norm does not depend on what people actually think or say on the moment of action. A man refusing payment for mowing his neighbor's lawn may just indignantly point out that it is not done. But further reasons are available within the practice, and might be employed when he reports the incident to his family: ‘Who does he think he is? My employer?’ Because Elster is looking for actual thoughts rather than available reasons, he is too dismissive about purposes of norms. See the discussion of the lawn-mowing example in his ‘Rationality, Morality, and Collective Action,’ 153; CS, 111; Nuts and Bolts, 115.

31 Hart, H.L.A. Tile Concept of Law (Oxford: Clarendon 1961), 84–5Google Scholar. Hart stresses that such rules ‘are thought important because they are believed to be necessary to the maintenance of social life or some highly prized feature of it.’ Many misgivings about social norms turn on the question whether this belief in the necessity of the rules is justified in particular cases, especially when the norms in question are very demanding.

32 This description of the point of the practice draws on Schwartz, W.F. Baxter, K. and Ryan, D. ‘The Duel: Can These Gentlemen be Acting Efficiently?’ Journal of Legal Studies 13 (1984) 321–55CrossRefGoogle Scholar. I have adapted their account to my own perspective: they explain the practice by its social function, of which the participants need not be aware; I point to reasons that might make duelling acceptable to members of a military or administrative elite. Such reasons must be available for the people concerned.

33 For the Bavarian army, tribunals were made possible in 1823, and for the Prussian army in 1843. See Ute Frevert, Ehrenm ãnner: Das Duell in der b Ürgerlichen Gesellschaft (Munich: C.H. Beck 1991), 105f. On the arguments for and against, see also 53-4.

34 Vries, Anne de Panokko (Nijkerk: CallenbachGoogle Scholar, n.d.)

35 For comments on earlier versions of this paper I am indebted to Veit Bader, Gijs van Donselaar, Charles Goossens, Govert den Hartogh, Pieter Pekelharing, Bruno Verbeek and several anonymous reviewers. The first plans for the paper were discussed with Trudy van Asperen, who should have supervised the project; because of her untimely death (at the age of 52) in October 1993, the final version must do without the benefit of her wisdom and perspicacity.