1. Hart, H.LA., The Concept of LawGoogle Scholar (Bulloch, P. & Raz, J. eds., 2d ed. 1994)Google Scholar [hereinafter in the text as CL].

2. Hart, H.L.A., American Jurisprudence Through English Eyes: The Nightmare and the Noble Dream, 11 Ga.L. Rev. 969 (1977).Google Scholar

3. It would not be unfair to say that first Dworkin framed the debate in terms of a disagreement about the nature and scope of judicial discretion; next as a debate about the nature of adjudication generally; then as a debate between competing theories of interpretation, as part of an even more general debate about the methods and projects of jurisprudence. There have been common themes throughout, of course, among the most important, but least appreciated, concerning the relative importance of controversy and disagreement to our understanding of legal practice: the nature and scope of agreement necessary to make sense of controversy in law.

4. The editors determined that Hart's observations beyond his response to Dworkin were not sufficiently well developed to be published.

5. In the Postscript, Hart embraces the form of legal positivism that I have been developing since Negative and Positive Positivism. This form of positivism has gone under a variety of names. Wil Waluchow refers to it as “inclusive positivism”; Hart refers to it as “soft positivism”; Dworkin calls it “soft conventionalism.” It is most commonly referred to as Inclusive Legal Positivism, which is designed to emphasize the contrast with Raz's Exclusive Legal Positivism. I prefer Incorporationism, and that is the label I will employ throughout.Coleman, Jules L., Negative and Positive Positivism, 11 J. Legal Stud. 139 (1982)CrossRefGoogle Scholar, reprinted in Ronald Dworkin and Contemporary Jurisprudence 28–48 (Cohen, M. ed., 1984)Google Scholar; also reprinted in Coleman, Jules L., Markets, Morals and the Law 3–27 (1988).Google ScholarWaluchow, Wilfrid, Inclusive Legal Positivism 2–3 (1994).Google ScholarHart, , “Postscript,” supranote 1, at 250–54.Google ScholarDworkin, Ronald, Laws Empire 124–30 (1986).Google Scholar

6. These are my ways of characterizing and describing a family of views about the conditions of law's possibility and normativity.

7. These are very rough characterizations of the differences between Raz and me. For a detailed statement of the most important ones see notes 49, 50, and 54 as well as the accompanying text.

8. That is because Dworkin believes that as it is traditionally understood positivism is subject to the dreaded semantic sting. To save the theory and to render it plausible, Dworkin believes that positivism must be reformulated as an interpretive theory. That interpretive theory is what he calls “conventionalism.”

9. Two tricky points that one has to keep in mind. First, conventionalism plays an enormously important role in positivism, but not the one Dworkin attributes to it. The role he attributes to it is based on his view of it as an interpretive theory of law. He is welcome to treat it that way. Postivists do not, however, and his semantic sting argument offers no reason whatsoever for their having to do so. Second, in saying that conventionalism is important to positivism, one usually has in mind the claim that the rule of recognition is a social convention. This is tricky because Raz, as much a posithist as anyone, never actually invokes the notion of a rule of recognition in developing his theory, though he does when discussing Hart's. For Raz, there just are criteria or standards of legality. He does not believe that it is necessary to introduce the idea of a rule of recognition as that which sets out these criteria of legality. But he does appear to believe that these are the criteria of legal validity just so long as there is a practice among officials of so regarding them. It is not a stretch to think of that practice as constituting a convention among officials: that feature of law, its ultimate conventionality is what I take to be common to all positivists.

10. In Authority and Reason, I took part of Raz's argument from authority in support of the Sources Thesis to rest on the idea that if the law is an authority, it must be uncontroversially identifiable to ordinary citizens. How could law serve to affect the practical reasoning of ordinary citizens if they could not identify the law or determine, at least in general terms, what it required of them? Understood this way, controversial criteria of the sort envisioned by an Incorporationist Rule of Recognition would be ruled out option, since individuals could not easily or reliably identify which norms were law. I am certain that I there misunderstood the connection between Raz's view of authority and the Sources Thesis, and so I want to withdraw that objection. Coleman, Jules L., Authority and ReasonGoogle Scholar, in The Autonomy of Law (George, R. ed., 1997).Google Scholar

11. Shapiro, Scott, The Difference That Rules Make, inGoogle Scholar New Essms in Legal Theory (B. Bix ed.) [forthcoming 1998].

12. In upcoming essays, I offer a reconfigured positivism, including an account of the place within it of the Separability Thesis, as well as an interpretation of the importance of that thesis to the ongoing debate between positivists and natural lawyers—a debate that has been unhelpfully characterized by both positivists and natural lawyers alike.

13. Perry, Stephen, Intepretation and Methodology in Legal Theory, inGoogle ScholarLaw and Interpretation 112–21Google Scholar (Marmor, A. ed., 1995).Google Scholar

14. Hart, , “Postscript,” supra note 1, at 240.Google Scholar

15. Postema, GeraldJurisprudence as Practical Philosophy, 4 Legal Theory (1998).CrossRefGoogle Scholar

16. Dworkin, , Law's Empire, supra note 5, at 45–86.Google Scholar

17. In many ways, Perry's most important contribution to jurisprudence is his sustained development of the idea that the methodology of jurisprudence is an important subject matter in its own right. He is perhaps the leading Dworkinian currently working in jurisprudence. His efforts have been confined, however, to vindicating the interpretive or normative methodology, and he is careful to distinguish his endorsement of methodological interpretivism or normative jurisprudence from an endorsement of the substance of Dworkin's jurisprudence, namely law-as-integrity. Perry's other interesting claim is that Hart, the great substantive positivist, was himself a normative methodologist, his protests to the contrary: thus, Perry's claim that Hart was a pre-Dworkin Dworkinian. I do not find this claim persuasive, and part of my purpose in this section of the paper is to counter it. Perry, supra note 13. See also Perry, , Hart and Holmes: The Bad Man in Legal Theory, inGoogle Scholar The Legacy of Holmes, Oliver Wendell Jr. “The Path of the Law”Google Scholar and Its Influence (S.J. Burton ed.) [forthcoming 1998].

18. In this essay, I will sometimes use descriptive and conceptual jurisprudence interchangeably for ease of exposition. They are importantly different, however. What Perry and others object to is descriptive jurisprudence, not conceptual analysis as such. In fact, the architecture of the categories is quite complex. The important differences among philosophers are between descriptive and normative analysis on the one hand, and between apriori and aposteriori analysis on the other. Fundamentally, Hart, Raz, and I, for example, engage in descriptive, apriori conceptual analysis. Perry, Dworkin, and perhaps Postema embrace normative, apriori, conceptual analysis. Others like Brian Leiter, so-called philosophical naturalists, reject or limit the role of apriori conceptual analysis of whatever sort, embracing instead a kind at aposteriori conceptual analysis. So Perry's real concerns are not with conceptual analysis as such, nor with the ability to engage in it apriori;, his objections are to its claims to normative neutrality. See generally Leiter, Brian, Naturalism and Naturalized Jurisprudence, inGoogle Scholar New Essays in Legal Theory (B. Bix ed.) [forthcoming 1998].

19. See Raz, Joseph, Two Views of the Nature of the Theory of Law: A Partial Comparison, 4 Legal Theory 249 (1998)CrossRefGoogle Scholar; Endicott, Timothy A.O., Herbert Hart and the Semantic Sting, 4 Legal Theory 283 (1998).CrossRefGoogle Scholar

20. A related but equally familiar mistake is to infer from the in-principle reusability of what, at any time, we take to be the core of a concept, the claim that the concept has no core. The difference between reusability and stability is often missed in jurisprudence. The core can change; it is subject to revision. That does not mean that it is always being revised or rethought or reconsidered: nor, of course, does it mean that it is empty. This is one of the most important tenets of pragmatism, a corrollary, in many ways, of the belief/doubt principle. See Hart, , supra note 1, at 12, 123, 147–54.Google Scholar

21. This view reminds me of one of my favorite graduate school stories. The late, great Wilfred Scllars was a Visiting Professor at Rockefeller University while I was a student there. He was co-teaching a course on Kant with Margaret Wilson. On the first day, he proclaimed his affection for Kant as an early and particularly talented Sellarsian. It took most of the students a week to figure out what he was saving, and a semester to appreciate the extent of the compliment he took himself to be conferring.

22. No one denies that inquiry, conceptual or other, is regulated by norms. This is as true in the hard and social sciences as it is in law. In this sense, all inquiry is normative. The descriptive jurisprudent denies the much stronger claim that we cannot identify law's necessary features without also defending a particular normative conception of law's function.

23. Stephen Perry accuses Hart of failing to countenance that there are many “insider's” points of view. Perry argues that Hart privileges one kind of “insider's point of view”: the point of view of the person who accepts the law as legitimate, and reason giving accordingly: in other words, his version of the “internal point of view.” Perry argues that Hart must defend his privileging this particular insider's perspective. Perry then claims that such a defense presupposes both a particular conception of the person (a normative one), and a particular view of law's purpose (to guide conduct by offering reasons). Thus, Perry's claim that Hart is a pre-Dworkin Dworkinian.

Briefly, just as Hart does not deny that there are many possible motivations for compliance, he does not deny that there are many possible insider's perspectives. He does not identify the insider's perspective with the internal point of view, and then choose one of many such perspectives. The insider's perspective is the insider's perspective. The internal point of view is one kind of insider's perspective. It is the perspective of the person who sees the law as reason-giving or legitimate in that sense. Hart never says that this is the preferred perspective or the only perspective insiders can adopt. Wliat he says is this: that (as a conceptual, not a normative matter) it is impossible to understand our concept of law without attributing the internal point of view as the perspective of at least some individuals (i.e., officials) toward at least some rules (i.e., secondary rules, especially the rule of recognition). In other words, Hart does not privilege a particular conception of the internal point of view so much as he demonstrates the very different point that our concept of law presupposes the conception he offers of it.

Perry's extremely interesting, subtle, but, I believe, ultimately unconvincing argument is developed in Perry, Hart and Holmes, supra note.17 For an artful and, by my lights, convincing response to Perry's arguments, see Scott Shapiro, Hart v. Holmes: The Bad Stan in Legal Theory, in The Legacy of Oliver Holmes, Wendell Jr.: “The Path of the Law”Google Scholar and Its Influence (S.J. Burton ed.) [forthcoming 1998].

24. Having emphasized the conceptual or descriptive projects of jurisprudence, several points should be noted. The descriptive project of jurisprudence is to identify the essential or necessary features of our concept of law. No serious analytic legal philosopher—positivist or interpretivist—believes that the prevailing concept of law is in any sense necessary: that no other concept is logically or otherwise possible. Nor do we believe that our concept of law can never be subject to revision. Quite the contrary. Technology may someday require us to reuse our concept in any number of ways. Still, there is a difference between the claim that a particular concept is necessary and die claim that there are necessary features of an admittedly contingent concept.

25. Whereas I find Hart's descriptive jurisprudence basically convincing, I do not find his particular account of the kinds of reasons the law creates plausible. Hart claims that the concepts of right and obligation in law have a different meaning or sense than they do in morality, and this line of argument runs into serious trouble. His introduction of content-independent reasons for action is extremely valuable and one of the great contributions to legal philosophy, but his particular use of it is not persuasive.

26. Much of the argument against descriptive jurisprudence takes the form of purporting to show that Hart, contrary to his own reflections on the matter, did not engage in or faithfully execute a descriptivist methodology. Such arguments are extremely interesting and genuinely important, whether ultimately sound or not Even if they are sound, however, they fall short of undermining the very possibility of a descriptive jurisprudence, focusing so much, as they do, on Hart.

27. In asking “How is law possible?” I do not mean to invite a causal or historical answer. The question is conceptual. It asks: “What are the existence conditions for law?”

28. Part of what distinguishes positivism from natural law is the answer each provides. Very roughly, we might say that natural lawyers explain the possibility of law by deriving or embedding it in morality (political morality or natural law). So understood, law is the institutional embodiment of an important aspect of morality: namely political morality. There are several problems with this kind of approach, not the least of which is that it assumes rather than explains law's most salient feature—that is, its institutionality. In an obvious sense, therefore, this line of argument does not really explain law without invoking law.

29. Stephen Perry argues that Hart's confining the necessity of acceptance from the internal point of view to officials, and not extending it to the public at large, is an embarrassment to his position. Hart's point is that acceptance from the internal point of view by officials is necessary to explain the possibility of law, whereas acceptance by the bulk of the populace does not bear on any of the fundamental questions of jurisprudence. Hart may be wrong to think that acceptance from an internal point of view by the bulk of the population is unnecessary from the point of view of answering the fundamental questions of jurisprudence, but he was right to think that it was irrelevant to answering any of the fundamental questions in which he was interested. Frankly, I think he was right about the more general claim as well. In any case, he has good reason to focus on officials in his discussion of the internal point of view while ignoring the common folk. Thus, it is unfair to view his failure to do so as an embarrassment. See Perry, , Hart and Holmes, supra note 17.Google Scholar For an objection to this line of argument similar to mine, see Shapiro, , supra note 23.Google Scholar

30. This is the point of Negative and Positive Positivism, Coleman, , supra note 5.Google Scholar

31. The Conventionality Thesis leaves open whether the relationship between rule and practice is reduction or supervenience, or some other kind of relationship which would need further explanation. One reason for thinking that Hart might have understood the relationship as supervenience is that supervenience in ethics was in the air at Oxford during the period in which CL was written. Ever since the publication of CL many philosophers, in both moral philosophy and the philosophy of mind, have offered reductive accounts. Hart rejects reductive accounts. That much is unarguable. I reject such accounts as well. I do not mean to suggest, however, that acceptance of the Conventionality Thesis entails rejection of reductionism. One could be a reductivist about the relationship between the rule of recognition and the relevant practices among officials. Neither Hart nor I are.

32. Marmor, Andrei, Legal ConventionalismGoogle Scholar, this issue.

33. Marmor argues that coordination conventions are incapable of explaining tlie sense that many people in most communities have that their rule of recognition is correct, desirable, just, especially reasonable or sensible. No one develops any special attachment to the convention of driving on the right or the left. Coordination conventions do not seem to explain what we might think of as “attachment” to the rule of recognition. It would seem just a bit perverse for Americans to feel a special pride about the practice of driving on the right in contrast to the British practice of driving on the left, whereas it would be surprising if Americans did not take pride in the rule of recognition.

Despite its superficial attractiveness, this argument is unpersuasive. Marmor's own position is also subject to precisely the same objection. First, people can and do develop attachments to arbitrary conventions. This implies the more fundamental point that there is no reason to suppose that the fact that a rule solves a coordination problem must provide the explanation of the attachment to it that might develop. The grounds of attachment may be completely different. The rule exists and survives because it solves a coordination problem, and represents a Nash equilibrium. Whether people develop an attachment to it or not depends on a range of other factors. It is simply naive to suppose that the explanation of attachment must have the same grounds as the explanation of its existence or stability. Marmor's objection is based on ignoring the distinction between the causal explanation of attachment to the rule and normative justification of it The rule isjustified because it solves a coordination problem. People get attached to it for all other sorts of reasons—if they get attached to it at all.

Were this a good objection to the coordination convention conception of the rule of recognition, it would be an equally powerful objection to the constitutive convention conception. After all, what is it about the fact that certain rules constitute or define games that explains the attachment people have to those rules? In the end, Marmor notes that the explanation of the attachment some feel to the rule of recognition needs some other explanation.

34. I do not suppose that conventions provide reasons, only that they are capable of doing so. The question is, Under what conditions are conventions capable of doing so?

35. One of my objections to Marmor's argument is that he asks us to abandon the coordination convention conception of the rule of recognition in favor of the constitutive convention conception when the ways in which the former give rise to reasons for acting are well understood, whereas the ways in which constitutive conventions do is mysterious. To reduce the mystery, Marmor offers analogies. The conventions governing certain activities, like opera, create norms and values internal to the activity. There is no reason why this shouldn't be true of law. The problem is that we need an account of the ways in which the conventions in opera and elsewhere actually create reasons or values; otherwise the analogy may prove unhelpful.

36. Davidson, Donald, Actions, Reasons and Causes, 60 J. Phil. 685, 690–700 (1963).CrossRefGoogle Scholar

37. I am indebted to Scott Shapiro for helping me formulate this argument and especially for bringing its possibility to light. Shapiro is also responsible for characterizing the project of legal positivism as showing how certain kinds of rule-governed behavior (law) are made possible only by understanding the connection to rule-guided behavior.

38. The person who looks to the law for reasons sees the law motivationally. The fact that something is the law motivates him: he does it because it is the law. The person who looks to the law for information corresponds to Hart's “puzzled man.” The puzzled man needs no motivation; he just needs information. The law can guide his conduct not by offering him reasons different from what he already has but by offering him information about what those reasons actually require of him.

39. There are several possible formulations of the separability thesis. Only under the formulation according to which morality could never be a condition of legality would Dworkin's argument, if sound, undermine the separability thesis. I have defended a different formulation of the separability thesis, according to which morality is not necessarily a condition of legality, in which case, even if Dworkin's argument were sound, it would not entail that the separability thesis is false. One of the more interesting disputes among legal positivists is the disagreement Raz and I have about where the modal operator should be placed in characterizing the separability thesis.

40. Dworkin rightly notes that positivists do not understand discretion to be a privilege or power to resolve disputes by any standard or preference that strikes a judge's fancy. Discretion is a power to appeal to extralegal standards to resolve legal disputes. But the appeal must be regulated by norms of rationality and reasonableness and is subject to evaluation and criticism if it is not Though regulated by norms and subject to criticism, Dworkin nevertheless conceives of discretion for positivists as being some kind of freedom from legal constraint. He is not alone in holding this view, but it does not strike me (as a positivist) as the correct characterization of discretion.

The positivist believes that even when and where law runs out, the judge is often obligated to appeal to certain moral principles and not to others. Discretion for many positivists is not primarily a power or privilege or freedom. It is not the absence of legal constraint Rather, it is the legal duty to appeal to certain principles that are not law in order to resolve the issue at hand. Discretion is a claim about the status of the principles to which the judge must appeal.

We can distinguish two pictures of discretion: one focuses on the freedom of the judge; the other on the status of the standards to which the judge appeals. Dworkin paints a picture of the first sort On his view, positivism is committed to seeing judges as having a kind of unavoidable, but regulated, freedom. In fact, most positivists view the judge as having no such freedom. Instead, the judge has a legal duty to appeal to certain principles and is bound by the conclusions drawn from them. It is just that these principles or standards are extralegal; they are not part of the community's law. Contra Dworkin, the positivist believes that in what he would call hard cases a judge is directed to appeal to appropriate moral principles: ones he has no freedom to ignore. He is bound by aspects of his role to appeal to such standards. The positivist's point is that even if those standards are binding on officials it is not because they are law. They are not the law of the relevant community, even if they are binding on the judge. Thus, the important distinction Raz has drawn between standards that are binding on judges and those that are part of the law of the community, and binding for that reason.

On the other hand, one could object on behalf of the Dworkinian that if this is what a positivist means by discretion, all the apparently interesting and important differences between positivists and Dworkin evaporate into semantic ones. Both believe that the judge is bound to appeal to moral standards in certain cases; the only difference is that the positivist believes that these standards are not law, though binding on officials, whereas Dworkin believes that the standards are binding because they are law. One might ask, Why notjust accept the Dworkinian characterization, which has the advantage of offering an elegant and simple explanation of why moral standards are binding on officials? They are binding because they are law.

This solution is plausible, but it misunderstands or underappreciates the point of the positivist's refusal to treat these moral standards as binding law: binding, but not law. The reason is that positivists, especially Raz, are concerned primarily to vindicate law's claim to authority. For reasons I cannot delve into at this point, law's claim to authority—especially for Raz—is tied up with restricting the category “law” to norms that have social sources. Moral principles that are not source-based cannot therefore count as part of the community's law. But that does not mean that such standards cannot be binding on officials.

This is one way in which my position differs from Raz's. I am not moved primarily by a desire to vindicate law's claim to practical authority. As a result, I have less trouble with Dworkin's characterization of these moral standards as binding on officials because they are law. For me, as will become clear, it is not important whether we treat the moral standards as binding on officials because they are law, what is extremely important is that if we treat them as binding law, we do so because they satisfy conditions set forth in a rule of recognition. Right here we have the differences among inclusive and exclusive positivists and Dworkin, and some appreciation of the implicit motivations for each.

41. Sartorious, Rolf E., The Enforcement of Morality, 81 Yale L.J. 891 (1972).CrossRefGoogle Scholar

42. Of course, even as discretion owing to the availability of legally binding resources decreases (as the set of available resources increases), the possibility of discretion owing to vagueness increases (as the set of controversial moral predicates legally binding on officials increases). The same predicates that reduce discretion on one front increase it on another: a consequence of Dworkin's own view that moral predicates are controversial and contestable. To defeat the positivist view of discretion Dworkin has to do more than show that the set of available legal resources on which a judge can draw is much greater than the positivist who limits the resources to a subset of rules has. Of course, no positivist really limits the set of resources binding on officials in the way Dworkin claims. The debate has to be resolved on grounds having more to do with the structure of argument and legal justification than with the set of binding legal sources.

Indeed, that is the line of argument Dworkin ultimately takes. His best argument is that positivists don't have a plausible account of the nature of adjudication (regardless of the size of the set of resources on which judges can draw). I think he is absolutely right about this, though the claim is a bit general, and does not apply to all positivists. I, for one, am a positivist with a view of the structure of adjudication and justification in law that is very different from the one Hart presents in Chapter 7 of The Concept of Law. I am a meaning holist whose views on the structure of legal discourse is very close to Dworkin's. Both of our views owe much to Quine. It is just that in my view, there are a range of issues having to do with law's institutionality, possibility, and the individuation of legal systems that require the kind of conventionalist answer the theory of the rule of recognition provides.

A theory of the structure of justification in legal discourse is not a theory of law: essential questions of the sort just mentioned are left unanswered by it. Like David Lewis in the philosophy of language, I believe that features of the relevant practice require a conventionalist answer. In philosophy of language, these issues include how terms get meanings (i.e., by conventions). Nevertheless, we have to draw a distinction between the question How do bearers of semantic content get meanings? and the question What is the meaning a bearer of semantic content has? We answer the former question by invoking the idea of linguistic conventions. We do not (or need not) answer the second question in that way. Certainly a meaning holist would not Lewis, well known for his possible world semantics, does not either.

There is an important and overlooked analogy between the role of conventions in the philosophy of language and in jurisprudence. In my view, the convention (the rule of recognition) is necessary because it helps to explain how norms secure their status as law. But we do not continually refer back to the convention in order to determine the validity of particular legal claims. Unlike other positivists, I do not believe that we resolve disputes about what the law is on a particular matter by referring back to the conventions that make law possible. Such a view would be incompatible with the meaning holist picture to which I am drawn.

This is another aspect of Incorporationism as I develop it that is distinct from the traditional forms of legal positivism. I have no reason to think that other Incorporationists who accept the basic structure of the position I set out on tlie rule of recognition do or would accept the account of the structure of legal discourse that I do. It is clear from Chapter 7 of The Concept of Law that Hart has something else in mind altogether. I have no reason to think that Dworkin has conceived of the issues in this way, separating questions of the structure of justification in law from questions about the nature of law. On my reading of the body of his work, I do not think that he does. Indeed, most people read Dworkin as identifying the theory of law with the theory of adjudication. At the very least, one can plausibly read him as deriving the theory of law from the theory of adjudication. In any case, I am inclined to separate the two. The debate between us ultimately is whether there are interesting and important questions in law or about law that need to be answered or which are left unanswered by a theory of the structure of justification in legal discourse. I think there are, and I think the best explanation of them is conventionalist. This is one of the main issues I take up in the third Clarendon Lecture.

43. See especially Coleman, , supra note 5Google Scholar; Coleman, Jules L., Second Thoughts and Other First ImpressionsGoogle Scholar, in New Essays in Legal Theory 401–505Google Scholar (B. Bix ed.) [forthcoming 1998]; Coleman, Jules L. & Leiter, Brian, Legal PositivismGoogle Scholar, in A Companion to Philosophy of Law and Legal Theory 241–60Google Scholar (Patterson, Dennis ed., 1996).Google Scholar

44. Even though Incorporationism makes clear both that morality can sometimes be a condition of validity and that it cannot necessarily be a condition of legality, some have associated Incorporationism with an altogether different claim, namely that law and morality must (at least in one possible legal system) diverge. This view doubly misunderstands Incorporationism. First, it confuses Incorporationism with Negative Positivism. Second, it misunderstands Negative Positivism. Let us begin with the latter confusion first.

Negative Positivism is the view that there must be one logically possible legal system in which morality is not a condition of legality. This follows from the Separability Thesis, the claim that morality is not a necessary condition of legality. It hardly follows from this that there must be a possible legal system in which morality and legality diverge. The latter advocates a form of extensional divergence, whereas Negative Positivism advocates a form of intentional divergence. Negative Positivism is not Incorporationism as the former claims that there is no necessary connection between law and morality, whereas the latter claims that sometimes there can be a connection between them, but there needn't be. The two are obviously consistent They are not equivalent, of course, because the latter entails the former, but not vice versa.

45. We need a theory of law that merges the idea of a rule of recognition with the view that moral principles can sometimes be binding on officials. Virtually everyone accepts the latter claim. Dworkin thinks that because moral principles, if binding, are so in virtue of their substantive merits, one has to abandon the former. Not only do I think he is mistaken; without the former, several key features of law are left inadequately explicated, if explained at all: in particular, law's possibility, its institutionality, and our capacity to individuate legal systems. Positivists who accept a rule of recognition offer a variety of different views about how to accommodate moral principles. Some treat them as not binding. Most treat them as binding, but not law. Some treat them as binding law but binding in virtue of their social source; and some, Incorporationists, treat them as binding in virtue of their merits. Which position one takes on this issue will depend, as I have suggested, on one's other motivations: in particular, on one's view about the importance of vindicating law's claim to authority, as well as what is required in order to do so.

Part of what distinguishes my Incorporationism from others is that I not only advance the view that moral principles can be legally binding in virtue of their merits, but I also adopt something like a Dworkinian view about the way in which such principles and other binding sources figure in the legal practice of justification. On the other hand, I believe that a theory like Dworkin's, which has no place for a conventionalist rule of recognition, lacks the resources to provide a plausible account of law's possibility, institutionality, and individuation.

So when Dworkin argues that a nonpedigree criterion of legality cannot be a conventional rule, he is claiming that the particular kind of theory that I have been developing is impossible. My motivation in arguing that moral principles can be legally binding as law is that I am inclined to merge a general, holist view about justification with a conventionalist view about the conditions that make law possible. I want, in effect, to bolster, rather than to undermine, what I take to be the valuable and plausible parts of Dworkin's view.

There is a real question of a deep and philosophically interesting sort about whether, and in what ways, one can merge pragmatism and conventionalism. I think one can (in bodi philosophy of language and philosophy of law). Again, this is an issue I address in the third Clarendon Lecture.

It is not obvious that Dworkin appreciated my motivation for allowing moral principles to figure in legal practice when he argued against me. That doesn't matter. His arguments are not persuasive, and that is the issue I want to address in the next section.

46. Dworkin offered a counter-response, but not to my mind a particularly persuasive one. A little chronology is in order. Long before Negative and Positive Positivism other positivists noted that one way of responding to Dworkin's objections in MOR I would be to abandon the pedigree requirement (I take no credit for discovering tills way out, though I have been more worried than most about the implications for positivism of taking it.) Dworkin responded in MOR II that a rule of recognition that allowed morality to be a condition of legality could not be a social rule (or what I am calling a conventional rule). That is because reference to morality would lead to disagreement incompatible with the rule of recognition's being a social or conventional rule. Dworkin was the party offering the objections to positivism. The burden on positivism is not to show that controversy engendered by a rule of recognition is impossible or that Dworkin was mistaken in interpreting controversy in a way that made it compatible with his overall view. The burden on positivism was to show somehow that controversy engendered by the rule of recognition did not imply Dworkin's conclusion that the rule of recognition could not be a conventional or social rule. The burden was to show the compatibility of controversy and convention: no more, but no less.

This is where Negative and Positive Positivism enters the picture. I advanced two main theses: first, and foremost, that for a positivist, the most important feature of law is that it rests on a social fact; that in Hart's and my versions of positivism, that fact is expressed in a rule tliat is itself a social fact—the social convention Hart refers to as a rule of recognition. Second, the rule of recognition could be conceived of as a conventional rule even in the face of controversy. My argument drew a distinction between disagreement about content and disagreement about application. My claim was that many controversies calling for or inviting controversial moral argument are understandable as disagreements about the application of the rule rather than as disagreements about the rule's content Moral disagreement about the content of the rule would be incompatible with the claim that the rule is a conventional or social one. Disagreement about what the rule requires in a particular case is compatible with the rule of recognition being a social convention. That is because, as I argued, morality cannot settle what the convention is, but it can help us to determine what the convention requires in a particular case. This is a familiar and important distinction. Indeed, Raz draws on a similar distinction in his account of authority: for him, law's claim to authority precludes looking into a law's justifying reasons to determine its identity or content (what it is); but one can investigate the moral reasons that would justify the law in determining what the law requires in a particular case.

Though inclusive positivists rely on my argument, Dworkin was unpersuaded. What was surprising was that he began his counterattack by labelling the distinction between content and application “doubtful,” but it is hard to understand what he could have meant: the distinction itself is so familiar. Sometimes you and I disagree about what the rule we are supposed to be following is. Other times we know full well what the rule is, but we disagree about what it requires of us. The only counter to this obvious point he offers is that we can always describe the same disagreement either way: as a disagreement of content or application. But even saying that affirms, rather than contests, the legitimacy of the distinction. Saying that either description is possible may be his way of saying that the distinction can do no work. But this too is false. Even if it did turn out to be logically possible to describe any disagreement about a rule as a disagreement about the rule's content or its application, it hardly follows that either description is equally apt all the time.

Sometimes the context will reveal to us which description is apt and why. Sometimes there may be no way of knowing independently of offering theoretical considerations on behalf of one interpretation or the other. For my part, I nowhere claim that all disagreements are best understood as disagreements in the application of the rule of recognition. I would never say, and did not say, that the disagreement in the rule of recognition evidenced by Marbury v. Madison was a disagreement in the application of the rule. It is a disagreement in how the rule should be extended or developed. It is a disagreement in the rule's content

Law's conventionality does not require that every disagreement be a disagreement in application. Positivists do recognize discretion, and some disagreements, which are disagreements in content, are best thought of as disagreements about how the law should be extended. I know that in Dworkin's view they are always disagreements about what the law is. But again, the choice between these two views has to be made on theoretical grounds. Dworkin and I simply have conflicting but internally consistent characterizations of controversy. My burden was to show the consistency of a positivist's characterization; and, as far as I can tell, Dworkin is the only legal theorist of note who thinks that the argument fails.

Ironically, faced with the same data and the same interpretive options, Dworkin invariably interprets the disagreements as disagreements of content He does that because at the time he was defending the view that the rule of recognition is a normative and not a social rule. That thesis requires treating disagreements in the rule of recognition as normative disagreements of content So not only did Dworkin rely on a distinction he labelled “doubtful,” he chose the characterization favorable to his position without ever offering anything like a theoretical reason for doing so. That would have been fine had he been in my position: that is, in the position of trying to show that an objection did not undermine his position. In other words, if I had criticized his view by saying that disagreements of law are invariably disagreements about the application of the rule of recognition, it would have been perfectly fair for him to respond that the same disagreements could be characterized as disagreements of content If they are disagreements of content, then his view is not in jeopardy.

That would have been fair had his been the position under attack. But it wasn't He was on the offensive. Hart's position was the one under attack, and I defended it in a perfectly plausible and sensible way by showing that the objection could be rendered harmless under some plausible interpretation of it.

The dispute about how to understand controversy engendered by a rule of recognition has to be resolved by appeal to theoretical considerations: elegance, simplicity, explanatory power, and the like. For my part, I have tried to show how the explanation of law's possibility, institutionality, and individuation requires a conventionalist account of law; and therefore that understanding disagreement in the rule of recognition in a way that is compatible with law's conventionality could be defended on theoretical grounds. Dworkin could rightly argue that we need to understand controversy in a way that is true to the structure of the practice of justification in legal discourse. I agree. My view is twofold: (1) such an interpretation is not incompatible with a social convention at law's core; (2) the existence of such a convention is necessary to explain law's possibility, institutionality, and individuation.

47. One might try to turn the practical objection to a controversial rule of recognition into a conceptual one in the following way. One might argue as Stephen Perry does that a conception of law's function is necessarily part of a theorist's conception of law. If a positivist is committed to the view that law's function is to guide conduct then the function will constrain the rest of the theory. In order to be a rule of recognition in the positivist's sense, the rule must be capable at least in principle of satisfying the relevant function. A controversial rule of recognition, for example, one that embraces morality conditions of legality, would then be incompatible with positivism. The problem with this line of argument, I have suggested, is that the concept of law presupposes no controversial conception of law's function or purpose.

48. I want to detail what I take to be the subtle but significant differences between Razians and me. As I mentioned earlier, Raz himself rarely talks about law in terms of a rule of recognition. Instead, he talks entirely in terms of criteria or conditions of legality. So instead of asking what are the constraints on the rule of recognition to which the exclusive positivist is committed, we should ask, what constraints, if any, does exclusive positivism impose on the conditions of legality? The answer depends on the theory of authority. On my reading, Uie theory of law for the exclusive positivist is driven by the theory of authority.

For the exclusive positivist all criteria of legality must operate through social sources. The legality of a norm depends on its social source. But this does not imply that moral principles cannot be conditions of legality. This is an important and subtle point. Moral principles can be conditions of legality provided the way in which they operate as conditions of legality is by directing officials to social sources. Let me explain.

Take the following moral principle: individual citizens ought to follow the enactments of duly elected legislatures only. Treat this as a principle of political morality. It might also constitute a condition of legality provided it operates by directing citizens and officials to social sources. Only those enactments passed by duly elected legislatures can be legally binding on citizens.

So here is a case of a moral principle that can be a condition of legality because it operates as a condition of legal authority by directing officials and citizens to social sources only. It operates through social sources.

In contrast, inclusive positivists do not require that in order to be a condition of legality a moral principle must operate through social sources. According to the inclusive postivist the principles of morality can operate directly in the sense of requiring that a norm be evaluated on its substantive (moral) merits. Thus, die inclusive positivist can allow morality to be a condition of legality. Morality can only operate as a condition of legality by directing officials to the substantive moral merits of particular norms. That is an important difference between inclusive and exclusive positivists.

Inclusive positivists (incorporationists in my sense) can allow morality to be a condition of legality provided the rule of recognition sets morality forth as a legality condition. Morality' then operates by evaluating norms directly on the basis of their merits. Exclusive positivists can allow moral principles to operate as conditions of legality only if they direct officials to social sources.

An inclusive positivist can object to the exclusive positivist in either of two ways. The exclusive positivist account of the conditions of legality is animated by the theory of authority. Therefore, an inclusive positivist can reject the theory of authority or argue that the theory of authority does not require the constraints on the conditions of legality the exclusive positivist claims it does. In this section, I adopt the second approach, arguing that the Sources Thesis as a constraint on legality is not required by the Razian theory of authority.

49. In Second Thoughts and Other First Impressions, I consider possible responses to my argument There I say the following.

One might respond by claiming that all considerations of fairness are part of the underlying justification of any rule—including the prohibition against murder. All fairness considerations are part of the dependent reasons that are replaced by accepting the rule as binding authority. So, if I ask myself whether I ought to accept the rule against murdering others as authoritative, I have to ask whether the demands of right reason, including those of fairness, will be better satisfied by following it than by following the first order reasons that counsel against murder anyway. And if the rule is unfairly administered, if it does not give individuals sufficient opportunity to appeal, etc. then the demands of right reason—including those of fairness—are not better served by accepting the rule as binding authority. I would do better by simply complying with the demands of right reason directly.

This response appears to blur the distinction between the question of whether I will do better by accepting a rule as an authority over me and the very different question about whether it would be better in general to have such a rule. The two are merged only if we understand the demands of right reason applicable to me to be, in effect, to see to it that no rules are adopted that violate fairness: that, in other words, the kinds of considerations that ought to figure in the deliberations of others who have to face the question about whether to accept such a rule as binding on them given that they will be administering it are reasons that necessarily apply to me, that I have to consider, even though I do not administer it. The argument appears, in other words, to slip in a form of agent-generality or agent-neutrality that I had always taken to be absent from it.

In fact, I do not see how a Razian could advance this line of response. Raz claims that law can be an authority over me—or I can accept it as an authority—without having to accept all law as having authority over me. Similarly, one would think, the same law can be an authority over me, but not over everyone else. And that is just another way of saying that the relevant demands of right reason, and our capacity to satisfy them without relying on legal directives, can differ among us. There is, in other words, a kind of agent-specificity in the way the Razian account of authority is presented that appears inconsistent with the kind of agent neutrality to which the response seems committed.

My point is that at least some evaluative (substantive or moral) criteria of legality need not direct us to the underlying reasons that justify the law. So not every evaluative or substantive criteria of legality is inconsistent with the claim to authority. This is especially true where the evaluative or moral criteria are employed to specify necessary or validity-negating legality conditions.

50. For ease of exposition, let us always mean by the phrase “without recourse to moral argument” “without recourse to moral argument about what the law should be.” Surely, if any kind of moral argument would vitiate law's claim to authority, it would be argument about what the law should be as a condition of determining what the law is. Notice that the Razian view of authority is spelled out in terms of mediation between persons and reasons. It may well turn out that the dispute between exclusive and inclusive legal positivists turns on this controversial conception of practical authority. See infra note 53.

51. The argument from authority concerns the criteria by which citizens determine what the law is; it does not concern the criteria by which the law is made determinate. In other words, the argument from authority bears on matters of identification, not validation. The rule of recognition, in contrast, sets forth criteria of validity. This is just the difference between the criteria that make something determinate and those that bear on determining that something is of that sort: the difference between existence and epistemic conditions.

52. I am not sure that this is the kind of argument that Raz would advance in response. It was suggested to me in a very illuminating, helpful, and fruitful exchange with Ruth Chang, and she believes that it is the kind of response Raz would make. I just do not know whether it is. It is, however, a particularly interesting argument, one worth pursuing on its own merits, and very revealing. I take it up for those reasons.

53. Scott Shapiro has brought one final line of response to my attention. Take a rule like, “The law is whatever is just.” My argument in this section is that, in the end, everyone to whom the law is directed can identify the rules validated by this rule of recognition without recourse to the dreaded form of moral argument by consulting the notorious Swede who is the moral/legal expert. The law is not directed toward the Swede, so the law is the sort of thing that is capable of being identified without recourse to moral argument about what it should be by every member of the community to whom the law is directed. A Razian could say that in the case, as I have constructed it, the rule “The law is whatever is just” does not perform the mediation function between reasons and persons that authorities do. For those of us who learn the law from the Swede, he is the authority. And it is not just because we learn the law from him. Rather, it is his judgment that mediates between reason and persons.

The same does not hold in the case of a source-based rule of recognition. Even if each of us learns what the law is from our lawyer, and not from the rule of recognition, it is the rule of recognition and the law valid under it that mediates for us. The lawyer just passes on that information. He does not mediate. He simply reports.

Someone drawn to exclusive positivism from a commitment to the Razian account of authority might claim that an Incorporationist rule of recognition could not in principle play the requisite mediating role between persons and reasons that the theory of authority claims law purports to. Only source-based conditions of legality can.

I have not made up my mind about this response, but a couple of things come to mind. As I have always claimed, though others have disagreed, if this is what the argument against Incorporationism comes to, its basis is an epistemic, not a metaphysical claim about the nature of law. Second, and more importantly, the argument rests on accepting the Razian account of authority as mediating between reasons and persons. This is a controversial view, and one with which I am not in agreement Working out that difference must await another ocassion. Still, it would be very interesting to know whether in the end the objection to Incorporationism that exclusive positivists have is based entirely on a theory of authority that sees law, and authority generally, as mediating between persons and reasons, something, it appears, an Incorporationist rule of recognition may preclude.

Note, however, that some exclusive positivists object to Incorporationism on other grounds. For example, Shapiro, like me, understands authority as mediating between (or among) options—in his case, by making some infeasible; in my case, by making one salient. Nevertheless, he objects to Incorporationism on the grounds that it is incompatible with the PDT. I take up Shapiro's argument in the next section.

54. Shapiro, , supra note 11.Google Scholar

55. Other legal rules could help specify more precisely the content of what the rule of recognition requires, but they would add no additional reasons for acting of their own. This is Shapiro's key point.

56. For examples of how the argument can be generalized for any such “incorporationist” clause, see Shapiro, Scott, On Hart's Way Out, this issue.Google Scholar

57. Shapiro also has an objection to that distinction at least as it applies to officials. On his reading, for officials the rule of recognition typically serves both epistemic and validation functions. He has a neater way of putting the point, but I have not made my mind up about the merits of the argument just yet. His objection I do consider here is powerful enough. See Shapiro, , supra note 11.Google Scholar

58. Hart, H. L. A., Commands and Authoritative Reasons, in Essays on Bentham (1982).Google Scholar

59. I need to develop this suggestion further, but not here. The general idea is that legality always must purport to imply the possibility of a practical difference. Positivists typically associate this difference with instrumental guidance: enabling individuals to pursue their interests or satisfy the requirements of right reason, or the like. Alternatively, legal rules need not be seen entirely in terms of the role they play in furthering interests and guiding action accordingly. A rule's legality allows it to play a role in the justification of a decision or the resolution of a dispute that, in the absence of its having that status, it could not play. I take it that there is a sense in which the Dworkinian project involves the generalization of this idea. That, in turn, marks the difference between seeing law as guiding action—as positivists typically see it—or as adjudicating disputes according to accessible, defensible standards—as Dworkin typically sees it. My claim is that neither conception of law expresses a conceptual truth about law. In a sense, that is why, at bottom, I do not see the PDT as asserting a conceptual truth about law.