1. On Bentham, see particularly Gerald Postema, Bentham and the Common Law Tradition 328–36 (1986).Google Scholar

2. Moore, Michael S., Law as a Functional Kind, in Natural Law Theory 188Google Scholar (George, Robert P. ed., 1992)Google Scholar. But see infra, note 32.Google Scholar

3. Hart, H.L.A., The Concept of Law (1961)Google Scholar. In this essay, all citations to this work are to the second edition, and are given by parenthetical page references in the text H.L.A. Hart, The Concept of Law (Bulloch, P. & Raz, J. eds., 2d ed., 1994).Google Scholar

4. As other commentators have observed, Hart draws the distinction between primary and secondary rules in at least two different ways. See Hacker, P.M.S., Hart's Philosophy of Law, in Law. Morality and Society: Essays in Honour of H.L.A. Hart 1, 19–21Google Scholar (Hacker, P.M.S. & Raz, J. eds., 1977)Google Scholar. At one point Hart draws the distinction in terms of the rules' supposed normative function: primary rules are said to impose duties, whereas secondary rules are said to confer powers (81). At another point he writes that secondary rules “may all be said to be on a different level from the primary rules, for they are all about such rules” (94). This second distinction is a logical one: secondary rules are in effect characterized as second-order rules that in some way operate on, or permit operations with, primary rules; primary rules are then simply the rules at the first level in a logical hierarchy. It seems clear that the second of these distinctions, based on logical type, is the more appropriate way to understand the difference between primary and secondary rules. (Even so, it is necessary to regard the notion of one rule being “about” another in a fairly loose sense.) The most important secondary rule, the rule of recognition, is duty-imposing rather than power-conferring, and the primary rules it identifies as valid could be either power-conferring or duty-imposing. A second ambiguity still remains, however. Hart sometimes uses the term “primary rule” to refer to a rule that is, according to some rule of recognition, valid. A primary rule in this sense could, presumably, be either duty-imposing or power-conferring. But sometimes Hart uses the term “primary rule” to refer to a duty-imposing social rule—meaning a customary rule—that either stands alone or forms part of a “regime” of primary rules (92). “Regime” here simply means a non-systematic set of duty-imposing social rules that exist together in a community that does not have any secondary rules. In this essay I will follow Hart in using the term “primary rule” in both these senses; the context should always make clear which is intended.

5. Cf. John Finnis, Natural Law and Natural Rights 1–6 (1980). My thinking about methodology in legal theory has greatly benefitted from Finnis's general discussion of this topic, and in particular from his illuminating critique of Hart. See id. at 1–22.

6. See, e.g., Waluchow, W.J., Inclusive Legal Positivism 19–29 (1994)Google Scholar. Waluchow clearly believes that Hart is a proponent of the descriptive-explanatory method. Id. at 14–15. In Perry, Stephen R., The Varieties of legal Positivism, 9 Can. J. L. & Juris. 361 (1996)CrossRefGoogle Scholar, I criticize Waluchow's application of the descriptive-explanatory approach to jurisprudence in general. In Section II of the present essay I argue that Hart cannot be understood as employing that approach, at least in anything like a pure form.

7. Raz, Joseph, The Authority of Law 50 (1979).Google Scholar

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

9. Id.

10. Id. at 69–70.

11. In speaking of regimes of primary social rules that are backed by physical sanctions administered by the community at large rather than by officials. Hart says that “we shall be inclined to classify [such] rules as a primitive or rudimentary form of law” (86). The characterization of such a regime as “primitive” is pretheoretical and potentially question-begging. By itself, it cannot bear the weight of a fundamental theoretical demarcation between the simpler social arrangement and systems based on a union of primary and secondary rules. It is for that reason that Hart requires the argument that a regime of primary rules is in certain respects defective.

12. The point here is that Hart claims that law as a general type of institution is identifiable solely by reference to various kinds of social fact; thus, the general concept of law is non-thick. But, as Hart makes clear in the Postscript, he does not think that each valid law within a legal system must be identifiable solely in social terms; the rule of recognition can adopt moral as well as social criteria of validity. This soft or incorporationist version of positivism is to be contrasted with the sources thesis defended by Raz, which holds that all laws must be identifiable as such solely by reference to social facts and without resort to moral argument. It will not be necessary, for purposes of this essay, to decide between these two versions of positivism. For the best recent discussion of incorporationism, see Coleman, Jules L., Reason and Authority, inGoogle Scholar The Autonomy of Law: Essays on Legal Positivism 287 (George, R. ed., 1995)Google Scholar. On the sources thesis, see Raz, Joseph, Authority, Law and Morality, 68 The Monist 295 (1985).CrossRefGoogle Scholar

13. Hart, H.L.A., Essays in Jurisprudence and Philosophy 13 (1983).Google Scholar

14. Peter Winch, The Idea of a Social Science (1958). Hart cites Winch as holding a similar view to Hart's on the internal aspect of rules (289).

15. “For the observer may, without accepting the rules himself, assert that the group accepts the rules, and thus may from the outside refer to the way in which they are concerned with them from the internal point of view” (89).

16. Hart would claim this because he holds (i) that the term “modern municipal legal system” has a generally agreed-upon pretheoretical meaning and extension, and (ii) that statements about the existence of a rule of recognition are simply statements of fact. I shall suggest in Section V, however, that to support the claim that even a particular legal system possesses a rule of recognition requires normative argument; a fortiori, the same would be true of the claim that all modern municipal legal systems possess such a rule. A less controversial example of how external observation could produce new knowledge might be the empirical demonstration or refutation of the claim that, say, all modern municipal legal systems permit greater penalties to be imposed for completed crimes, as opposed to attempts.

17. This point is obvious in the case of the last two issues. It holds true of the first issue as well because Hart's concern there is with Austin's reductive analysis of the concept of legal obligation (i.e., with the analysis of legal obligation in terms of orders backed by threats).

18. See Dworkin, Ronald, The Model of Rules II, in Taking Rights Seriously 46 (rev. ed. 1977)Google Scholar. In a previous article I interpreted Hart as making the claim in The Concept of Law that social rules do create a special form of non-moral obligation, but Jeremy Waldron has convinced me in conversation that this reading is probably mistaken. See Perry, Stephen R., Interpretation and Methodology in Legal Theory, in Law and Interpretation 97, 105, 115–16, 122Google Scholar (Marmor, Andrei ed., 1995)Google Scholar. It is, however, true that in later work Hart speaks of the rule of recognition as giving judges what he calls “an authoritative legal reason.” H.L.A. Hart, Essays on Bentham: Studies in Jurisprudence and Political theory 160 (1982). The general tenor of the discussion in that work suggests that Hart thinks judges really do have such a reason and not simply that they regard themselves as having one. If Hart did believe that social rules give rise to actual (non-moral) reasons for action, that would be, for reasons that will become clearer below, the first step toward a theory of law advanced from the internal point of view. But until normative argument were offered to support the otherwise mysterious claim that actual reasons for action are created, the theory would remain incomplete.

It is worth remarking that the claim that social rules necessarily give rise to actual (non-moral) reasons for action would yield a very strong form of internal theory. A weaker internal theory would argue that the concept of law, properly analyzed, sets out the conditions under which law is in principle capable of giving rise to new reasons for action, but without claiming that such reasons are necessarily always (or even ever) created in fact. See further die discussion in Section VII below. As I note later in the text of the present section, Hart begins in the Postscript to offer a conventionalist account of law that can plausibly be interpreted as an internal theory in this weaker sense. In other work Hart has, of course, offered an account of political obligation that centers on the so-called principle of fair play. See Hart, H.L.A., Are There any Natural Rights? in Theories of Rights 77Google Scholar (Waldron, J. ed., 1984)Google Scholar. If that account were correct, it would show that certain kinds of joint enterprise give rise to obligations that people would not otherwise have. But, as Waldron has recently emphasized, Hart does not think that instances of law are necessarily joint enterprises in the requisite sense. See Waldron, Jeremy, All We Like SheepGoogle Scholar, Can. J. L. & Juris. (forthcoming). Going beyond this empirical point, it is clear that Hart also rejects the idea that our concept of law implicates die principle of fair play. This rules out the possibility of an internal theory—even a weak internal theory—that takes that principle as its starting-point.

19. Cf. Raz, Joseph, The Morality of Freedom 63–64 (1986).Google Scholar

20. Id. at 63.

21. Some of these are discussed very briefly in the following section.

22. Dworkin, , supra note 18, at 53–54, 59–61.Google Scholar

23. See supra, note 4.

24. Dworkin, , supra note 18, at 53.Google Scholar

25. Id. at 60–61.

26. Jules Coleman recognizes that a jurisprudential theory must address the problem of the normativity of law, and that doing so requires, among other things, that the theory furnish an account of the concept of authority. He further recognizes that such an account must show whether and under what circumstances the law's claim to authority is justified, meaning it must show whether and under what circumstances the law creates the reasons for action it claims to create. Coleman, , supra note 12, at 296–305Google Scholar. Coleman holds that Hart's social rule theory “claims that the internal point of view transforms what would otherwise be a non-normative description of a convergent practice… into a reason-giving practice….” Id. at 299. Although I once thought this view of Hart's intent was correct (see supra note 18), I now have my doubts. It seems to me that Hart is primarily concerned with describing the practice of law from the outside, and that in consequence he simply sets aside the problem of the normativity of law with which he begins. In any event, Coleman correctly argues that even for officials a rule of recognition cannot be authoritative simply in virtue of being a social rule, because acceptance from the internal point of view does not by itself give anyone a reason for action. He does suggest, however, that such acceptance might be a reliable indicator that the practice has independent normative force. This is compatible with its being a concurrent normative practice in Dworkin's sense. Coleman also suggests, as Hart eventually came to accept in the Postscript, that the rule of recognition might be the basis of a conventional social practice. Coleman, , supra note 12, at 300–2.Google Scholar

27. See Coleman, , supra note 12, at 300–2.Google Scholar

28. Cf. Raz, , supra note 7, at 247–48Google Scholar. Of course, in this type of case general conformity is only a necessary condition for the existence of a reason for action, and not a sufficient condition; the activity must also be morally worthwhile for independent reasons.

29. Cf. Coleman, , supra note 12, at 298, 302Google Scholar. See also Coleman, Jules L. & Leiter, Brian, Legal Positivism, inGoogle Scholar A Companion to Philosophy and Legal Theory, 241, 247–48 (Patterson, D. ed., 1996).Google Scholar

30. Coleman, , supra note 12, at 298.Google Scholar

31. As was noted in Sections IV and V, Hart begins in the Postscript to offer the outline of an internal theory, based on the idea that the rule of recognition creates convention-based reasons for action. This theory is incomplete and, for the reasons noted at the end of Section V, it is also problematic in its own terms. The important point for present purposes, however, is that this is a theoretical undertaking quite different from Hart's stated aim of simply describing, from the outside, the acceptance of social rules and other, related, social phenomena. The aim now is not simply to describe people's beliefs that law gives them reasons for action, but to show how and why it in fact gives them reasons for action. This new theoretical undertaking involves a complete abandonment of methodological positivism and would, if carried to completion, yield an internal theory of law.

32. Cf. Perry, Stephen R., Holmes versus Hart: The Bad Man in Legal Theory, inGoogle Scholar The Legacy of Oliver Wendell Holmes: The Path of the Law and its Influence (S.J. Burton ed., forthcoming).

33. I noted at the beginning of this essay that Michael Moore has adopted what could be regarded as a variant of methodological positivism. See Moore, , supra note 2Google Scholar. Moore says that the task of jurisprudence is “descriptive,” but he clearly includes moral facts about a social practice as among the attributes that can figure in a description. Moreover, he sees the task of legal theory as showing law to be a “functional kind,” that is, a type of institution which necessarily, in a metaphysical and not just a conceptual sense, uniquely serves some good. It might be better, therefore, to see Moore as a theorist who is seeking an external theory of the kind described in the text, rather than as a methodological positivist His characterization of the debate among proponents of external theories would then be that each is putting forward a different understanding of the metaphysical essence of law. This is a coherent methodological view, but, as I have argued elsewhere, there does not seem to be any good reason to regard law as a functional kind in the way that gold and water are claimed by some philosophers to be natural kinds; there is no reason, that is, to view law as a type of entity having an essential nature. See Perry, , supra note 18, at 124 n. 62.Google Scholar

34. See Raz, , supra note 12, at 321.Google Scholar

35. Id.

36. Id. at 321–22.

37. For example, a Hobbesian theory of law, under an internal rather than an external interpretation, will adopt a conception of the person as a rational utility maximizer. By contrast, theories that argue that law does or is capable of giving rise to moral reasons for action must suppose that people are capable of acting upon such reasons. See further Perry, , supra note 32.Google Scholar

38. Cf. Finnis, , supra note 5, at 14–15Google Scholar: “If there is a point of view in which legal obligation is treated as at least presumptively a moral obligation…, a viewpoint in which the establishment and maintenance of legal as distinct from discretionary or statically customary order is regarded as a moral ideal if not a compelling demand of justice, then such a viewpoint will constitute the central case of the legal viewpoint.”

39. See Perry, , supra note 18, at 121–35.Google Scholar

40. Dworkin, , supra note 8, at 47.Google Scholar

41. Cf. Coleman, Jules L., Risks and Wrongs 401–5 (1992).Google Scholar

42. Dworkin, , supra note 8, at 197–202Google Scholar. The theory of law as integrity is meant to address what Dworkin calls “the puzzle of legitimacy.” Id. at 190–95. It is possible to imagine “external” accounts of law's legitimacy, which could well be associated with the kind of external philosophical theory that was discussed in the preceding section. These would argue for the moral legitimacy of state coercion without supposing that those subject to coercion have an obligation to comply. See, e.g., Ladenson, Robert, In Defence of a Hobbesian Conception of Law, 9 Phil. & Pub. Aff. 134 (1980)Google Scholar. But for Dworkin legitimacy is intimately concerned with the question, “Do citizens have genuine moral obligations just in virtue of law?” Dworkin, , supra note 8, at 191Google Scholar. Dworkin's substantive theory of law can thus be interpreted as an attempt to outline the conditions under which the law's conceptual claim to authority is justified. For Dworkin, however, that claim must be construed in a broader sense than I have construed it elsewhere in this essay. Essentially following Raz, I have supposed that the law's claim to authority is a claim that citizens are obligated by (and only by) the specific acts of those in authority. But Dworkin must suppose that law's claim to obligate is broader than this, as his substantive theory argues that citizens are obligated not only by the specific acts of authorities but by the best justification of the settled law. Indeed, it is precisely because Dworkin maintains that the concept of law involves the moral idea of “best justification” that his substantive theory is not a positivist one. (He does not become a positivist simply because, recognizing that the conditions of integrity might not be met in practice, he accepts that actual legal systems do not necessarily obligate.)

43. Marmor, Andrei, Interpretation and Legal Theory 52–53 (1992).Google Scholar

44. Cf. Perry, , supra note 18, at 125–31.Google Scholar

45. Raz, , supra note 7, at 50–52.Google Scholar

46. See Raz, , supra note 12, at 295–305Google Scholar; Raz, , supra note 19, at 38–105.Google Scholar

47. Raz, , supra note 19, at 78.Google Scholar

48. See Postema, , supra note 1, at 323–27.Google Scholar