1. Hart, H.L.A., The Concept of Law (1961) esp. ch. 5.Google Scholar

2. Hart further assumed that in each and every legal system, eventually one rule of recognition can be identified, which subsumes, in a hierarchical ordering, all the rest. This further assumption we need not share here, and I shall henceforth assume that nothing substantial is lost if we hold the more plausible assumpdon that the rules of recognition are not necessarily hierarchically structured.

3. Dworkin, R. M., Taking Rights Seriously (1977).Google Scholar

4. Hart, , supra note 1.Google Scholar

5. Hart, H.L.A., The Concept of Law (J. Raz & P.A Bulloch eds., 2d ed. 1994), Postscript.Google Scholar

6. Dworkin, , supra note 3Google Scholar, and Raz, J., Practical Reason and Norms (1975)Google Scholar. There is some lack of clarity about the question whether Hart thought that the practice theory of rules applies to morality as well. In any case, it is clear enough that it cannot apply to moral rules. Rules of morals and similar domains need not be practiced in order to be valid. Hart's requirement for a regularity of behavior as a basis for the rule's existence simply does not apply to such spheres as morality. We often hold moral rules to be valid despite the fact that they are not practiced in the pertinent community, or anywhere.

7. “Following a reason” may sound rather awkward, and perhaps it is. I use this unusual term (instead of, for example, “acting on a reason”) not only to emphasize symmetry, but also because I intend to refer to those cases where people act according to a reason with some generality. Generality is actually twofold, referring both to the kind of reasons and to the occasions in which those reasons are taken to apply.

8. The distinction, as well as the game example, was first suggested by Warnock, G. J., The Object of Morality 43–46 (1971).Google Scholar

9. Raz, Joseph, supra note 6Google Scholar. In his famous analysis of mandatory rules in Practical Reason and Norms, Raz claimed that such rules are there to replace the balance of the first-order reasons, which would have applied to the circumstances. Mandatory rules, Raz claims, are a combination of first-order reasons for action and exclusionary reasons, namely reasons not to act on certain types or excluded reasons. Although I tend to agree with this analysis as far as mandatory rules are concerned, I do not think that this account is adequate to explain the uniqueness of other types of rules, such as “rules of thumb” and, perhaps more importantly, constitutive rules.

10. Hart, , Postscript, supra note 5, at 256.Google Scholar

11. Ronald Dworkin, Hart's Posthumous Reply (unpublished manuscript).

12. Dworkin, Ronald, Laws Empire 91 (1986).Google Scholar

13. I am indebted to Ben Zipursky for mentioning this possibility to me.

14. Lewis, David, Convention: A Philosophical Study (1969).Google Scholar

15. It is worth keeping in mind that the solution of a coordination problem—unlike other types of game-theory-situations—does not call for a sanction-based solution.

16. Postema, G. J., Coordination and Convention at the Foundations of Law, 11 J. Legal Stud. 165 (1982)CrossRefGoogle Scholar. See also Raz, J., ‘Facing up: A Reply,’ 62 S. Cal. L. Rev. 1153 (1989).Google Scholar

17. It should be admitted that the analysis of coordination problems has advanced considerably in the last few decades, and a great deal of complexity and richness has been added to Lewis's original account. Nor should we assume that coordination problems are typically simple or of one particular kind. Complexities notwithstanding, the basic idea of a coordination problem remains the same: securing some form of uniformity of action where such uniformity is in the best interest of the parties involved, and it is recognized by them as such. For reasons to be clarified later on, I do not believe that this is the rationale of most of the rules of recognition.

18. I am not assuming here that conventional rules necessarily serve a particular purpose. My doubts about such functionalist explanations of conventions will be spelled out in the sequel, mainly the last section. For the time being, however, Lewis's framework should suffice for the illustration of the arbitrariness of conventions.

19. See, e.g., Gilbert, M., On Social Facts 340–41 (1989).Google Scholar

20. I have presented and defended this definition in my article On Convention, 107 Synthese 349 (1996)Google Scholar. There is however, one clarification that I should mention here: the reason[s], A, which motivates people to comply with the rule R in the first place, need not be independent of the conventions themselves. I may have no reason whatsoever to play chess if there is no such a conventional game to play. If I do have a reason to play chess, surely it depends on the fact that there is such a game that I can actually play. The importance of this point (which goes against Lewis's definition) will only become apparent later on.

21. See Lewis, , supra note 14, at 76–80Google Scholar, for an account of imperfect coordination problems.

22. See id. at.75.

23. Burge, T., On Knowledge and Convention, 84 Phil. Rev. 249 (1975).CrossRefGoogle Scholar

24. Id. at 250.

25. I have already presented the basic argument about constitutive conventions. See supra note 2.

26. The systematic nature of constitutive rules was first observed by J. Searle. See Searle, J., Speech Acts 33–42 (1969)Google Scholar. Although I certainly draw here quite heavily on Searle's distinction between constitutive and regulative rules, I am not sure that I want to subscribe to his basic idea that these are two separate types of rules. Others have claimed, and perhaps they are right about this, that all rules have both constitutive and regulative functions.

27. On this particular point, see the very intriguing article of Schwyzer, Hubert, Rules and Practices, 78 Phil. Rev. 451 (1969).CrossRefGoogle Scholar

28. Wittgenstein, L., Philosophical Investigations (1953), sec. 373.Google Scholar

29. It is true, of course, that if the coordination convention does not instantiate the optimal solution to the relevant coordination problem, a certain pressure may build up to move toward the optimal solution. But once the optimal solution has been reached, there would be no further inherent pressure for change.

30. As every amateur legal historian knows, the roots of the Continental system's decline to follow the practice of precedent arc in the French Revolution and the rigid doctrine of separation of powers it engendered.

31. But not the content, of course. I am certainly not trying to suggest that a whole legal community can be wrong about the content of the rules of recognition. On the contrary; I have elsewhere argued at some length that the conventional foundations of law make it impossible for a whole community of lawyers to misidentify their law. See Marmor, Andrei, Interpretation and Legal Theory (1992), ch. 5.Google Scholar

32. Green, L., The Concept of Law Revisited, 94 Mich. L. Rev., 1687, 1697 (1996).CrossRefGoogle Scholar

33. The beginning of such an explanation can be found in Marmor, A., An Essay on the Objectivity of Law, in Bix, B. (ed.) Analyzing Law, Oxford U. Press (forthcoming).Google Scholar

34. Green, , supra note 32, at 1698 ff.Google Scholar