1. Jules Coleman, The Practice of Principle: In Defence of a Pragmatist Approach to Legal Theory (2001), at 121.

2. Henceforth, references to “authority” are to legitimate practical authority unless otherwise indicated. Raz has presented his account of authority on numerous occasions: see especially Joseph Raz, The Morality of Freedom (1986), chs. 2–3; Raz, Joseph, The Problem of Authority: Revisiting the Service Conception, 90 Minn. L. Rev. 1003–1044 (2006)Google Scholar.

3. The fullest presentation of the argument is in Raz, Joseph, Authority, Law, and Morality, in Ethics in the Public Domain: Essays in the Morality of Law and Politics 210–237 (rev. ed. 1994)Google Scholar.

4. See, e.g., Hurd, Heidi, Challenging Authority, 100 Yale L.J. 1611–1677 (1991)CrossRefGoogle Scholar; Moore, Michael S., Authority, Law, and Razian Reasons, 62 S. Cal. L. Rev. 827–896 (1989)Google Scholar; Perry, Stephen, Second-Order Reasons, Uncertainty and Legal Theory, 62 S. Cal. L. Rev. 913–994 (1989)Google Scholar; Fred Schauer, Playing by the Rules: An Examination of Rule-Based Decision-Making in Law and in Life (1991), at 88–91.

5. See, e.g., Hershovitz, Scott, Legitimacy, Democracy, and Razian Authority, 9 Legal Theory 201–220 (2003)CrossRefGoogle Scholar; Himma, Kenneth Einar, Just ‘Cause You're Smarter Than Me Doesn't Give You a Right to Tell Me What to Do: Legitimate Authority and the Normal Justification Thesis, 27 Oxford J. Legal Stud. 121–150 (2007)CrossRefGoogle Scholar; Shapiro, Scott J., Authority, in The Oxford Handbook of Jurisprudence & Philosophy of Law 382–439 (Coleman, J. & Shapiro, S. eds., 2002)Google Scholar; Waldron, Jeremy, Authority for Officials, in Rights, Culture, and the Law: Themes from the Legal and Political Philosophy of Joseph Raz 45 (Meyer, L.H. et al. eds., 2003)CrossRefGoogle Scholar.

6. See, e.g., Gardner, John, Law's Aims in Law's Empire, in Exploring Law's Empire: The Jurisprudence of Ronald Dworkin 207–223 (Hershovitz, S. ed., 2006)Google Scholar; and Dworkin, Ronald, Response, in Exploring Law's Empire: The Jurisprudence of Ronald Dworkin 291–311 (Hershovitz, S. ed., 2006)Google Scholar. See also Himma, Kenneth Einar, Law's Claim of Legitimate Authority, in Hart's Postscript: Essays on the Postscript to The Concept of Law 271–309 (Coleman, J. ed., 2001)CrossRefGoogle Scholar.

7. Coleman, supra note 1, at 120–133. Arguably, some of Dworkin's and Waluchow's criticisms of the argument from authority are also of this type; see Ronald Dworkin, Justice in Robes (2006), at 203–211; W.J. Waluchow, Inclusive Legal Positivism (1994), at 132–140.

8. Raz, Authority, supra note 3, at 215.

9. Raz sometimes refers to the views expressed by legal institutions, rather than by legal officials. For our purposes, nothing turns on this.

10. Id. at 215–216.

11. Though the first fact (that legal institutions are officially designated as “authorities”) appears to be an exception.

12. Id. at 215.

13. Id. As presented in this passage, the second prerequisite requires only that it be possible to identify the fact that the directive was issued by the alleged authority without considering one's dependent reasons. However, elsewhere (see e.g., id. at 219, 234), Raz states that it must also be possible to identify the content of the directive without considering one's dependent reasons. This is clearly the better view. Neither the normal justification thesis nor the preemption thesis can be satisfied unless both conditions are met. A directive can replace our dependent reasons in our practical reasoning, and we can better conform to those reasons by following that directive, only if we can both identify the existence of the directive and determine what it requires without considering our dependent reasons.

14. While the second prerequisite can be derived from either the normal justification thesis or the preemption thesis, the first prerequisite is derived from the dependence thesis. A directive could not be based on reasons that apply to us anyway if that directive was not at least presented as someone's view of how we ought to behave; id. at 218–219.

15. Id. at 211.

16. Id. at 231. X may hold the view that one should act morally, but (Raz claims) we can determine that this is X's view without engaging in moral reasoning. Obviously, we cannot determine what it is to act morally without engaging in moral reasoning, but Raz would deny that one must determine this in order to determine what X's view is, since he denies that whatever is entailed by a particular statement or judgment is to be regarded as part of that statement or judgment; id. at 227–229.

17. Some inclusive legal positivists deny that this follows. They argue that the moral considerations relevant to identifying the existence and content of legal directives may differ from the moral considerations that feature among our dependent reasons: see, e.g., Coleman, supra note 1, at 126–127; Waluchow, supra note 7, at 139–140. However, even if they are right, it gives them only part of what they want; see Coleman, supra note 1, at 127.

18. Raz, Authority, supra note 3, at 217.

19. Id.

20. I take “confusion” to include not only mistakes about but also ignorance of the relevant aspects of the concept, since either error or ignorance on the part of officials could undermine the inference that the law must be capable of possessing authority. (The references to “aspects” of a concept may appear obscure. I seek to give this notion more concrete content below.)

21. Raz, Joseph, Two Views of the Nature of the Theory of Law: A Partial Comparison, in Hart's Postscript: Essays on the Postscript to The Concept of Law 1–37 (Coleman, J. ed., 2001)Google Scholar, at 8.

22. Raz, Authority, supra note 3, at 217, suggests that someone who claims that X has authority where X is in fact incapable of having authority must 1) be insincere, 2) misunderstand the nature of authority, or 3) misunderstand X. The supporting argument addresses the first two possibilities but not the third.

23. Raz, Morality, supra note 2, at 63 (emphasis added). See also Raz, Authority, supra note 3, at 220 (stating that his service conception of authority, as expressed in the dependence and normal justification theses, “correctly represents our concept of authority”).

24. Indeed, all three theses are normative; Raz, Morality, supra note 2, at 63.

25. Id.

26. See Murphy, Liam, Razian Concepts, 6 Am. Phil. Ass'n Newsl. on Phil. & Law 27, 28 (2007)Google Scholar for a similar explanation of how the normal justification thesis forms part of an explanation of the concept of authority even though it is a normative thesis about when a purported authority is morally legitimate.

27. Raz, Joseph, Can There Be a Theory of Law?, in The Blackwell Guide to the Philosophy of Law and Legal Theory 324–342 (Golding, M.P. & Edmundson, W.A. eds., 2005), at 326Google Scholar.

28. Raz suggests that having a basic understanding of a concept involves being able to use it correctly in normal circumstances: id. at 327; Raz, Two Views, supra note 21, at 9. If he means all normal circumstances, then it is doubtful whether someone who is confused about the key theses could satisfy this test. However, there is reason to think that this test is overly demanding. I have a basic understanding of the concept of DNA, but I doubt very much that I could use that concept correctly in all normal circumstances.

29. I am not suggesting that when legal officials claim that the law possesses authority, they have the key theses in mind. For the reasons given in Subsection III.B, this will often not be the case.

30. Raz, Two Views, supra note 21, at 12 (footnote omitted). In the omitted footnote, Raz acknowledges that there is a circularity in this characterization of a criterial explanation but claims that it can be easily eliminated.

31. Id. at 16–17 (“those that are generally believed to be the correct criteria are the correct criteria.”). There is some suggestion that the reason this is so is that each individual intends to use the concept in accordance with the criteria that are generally accepted within his or her language community.

32. Nor does he say much about what he means by criteria for the correct use of a concept. Indeed, he speaks more often of explanations of a concept. It may be that we can translate between these modes of speech on the basis that the central theses in an explanation of a concept provide criteria for the correct use of that concept. On this view, the dependence, normal justification, and preemption theses provide criteria for the correct use of the concept of authority.

33. I am grateful to an anonymous referee for pressing me to address this issue.

34. Raz, Authority, supra note 3, at 219–220.

35. Raz, Problem, supra note 2, at 1008 n.10.

36. Raz may distinguish between being systematically confused about a concept and having only a partial understanding of that concept. (Indeed, he denies that people are guilty of conceptual confusion just because their understanding of a concept is only partial; id. at 1006–1007.) He could then argue that the fact that users of a concept may be unaware of or mistaken about many aspects of the concept shows only that they may have a partial understanding of the concept, not that they may be systematically confused. This would not enable him to meet the objection in the text, however. If officials’ partial understanding of the concept of authority means that they are unaware of or mistaken about the key theses, then it still seems that we are unable to infer that the law is capable of possessing authority from the fact that officials claim that it possesses authority.

37. This is an adaptation of Sally Haslanger's distinction between manifest and operative concepts; Haslanger, Sally, What Are We Talking About? The Semantics and Politics of Social Kinds, 20 Hypatia 10–26 (2005)CrossRefGoogle Scholar, at 14. I call the criteria we actually use in practice “operative beliefs” so as to retain Raz's formulation that the correct criteria are those that are generally believed to be correct. One might contest this use of the word “belief,” but the important thing is to distinguish between the criteria we consciously believe to be correct and the criteria we actually use in practice, not to debate whether the latter should be labeled “beliefs.”

38. The failure to draw this distinction was one shortcoming of my earlier discussion of Raz's criterialism in Smith, Dale, Has Raz Drawn the Semantic Sting?, 28 Law and Philosophy 291–325 (2009)CrossRefGoogle Scholar.

39. Gilbert Harman suggests several other ways in which our use of the concept of bachelorhood does not conform to the criterion of an unmarried man; Gilbert Harman, Doubts about Conceptual Analysis, in Reasoning, Meaning, and Mind 138–143 (1999), at 140.

40. This is not to deny that there may be overlap between our manifest and operative beliefs or even that the former may partly determine the latter. It is, however, to deny that our consciously held beliefs about the correct criteria for using a concept always capture perfectly the way in which we use that concept. Moreover, this is not just because we sometimes misuse the concept. It is not necessarily a misuse of the concept of bachelorhood to deny that the Pope is a bachelor.

41. Raz, Two Views, supra note 21, at 12 (footnote omitted; emphasis added).

42. Admittedly, Raz states that what the correct criteria are depends on what people think they are (id. at 16). The most natural reading of this statement is that it is our manifest beliefs that determine which criteria are correct. However, not only does this sit uneasily with his focus on our use of concepts, but, for the reasons given below, his position is much stronger if he is interpreted as claiming that it is our operative beliefs that determine which criteria are correct.

43. Id. at 16, 20. The example of the concept of bachelorhood discussed above could be taken to illustrate this point.

44. It was suggested by an anonymous referee that Raz believes that when officials claim that the law possesses authority, they assert that the law has the power to impose obligations on others by its say-so, not that the normal justification thesis is satisfied. The normal justification thesis describes how things must be in order for the claim to be true but does not feature in the content of the claim. Whatever Raz's view is on this point, the account in the text is an account of what it is for the key theses to feature among officials’ operative beliefs, not of what is asserted when officials state that X has authority.

45. I consider Raz's claim about the centrality of legal institutions in our structures of authority in Section IV.

46. Raz, Problem, supra note 2, at 1010.

47. Id.

48. Id. at 1011 (“I keep referring to ‘our’ concept of authority. But is there such a thing? Are there not several concepts, all of them descending from the very same ancestors? Quite possibly so.”).

49. Id. Raz adds that if multiple concepts of authority exist within a single society, there may well be competition between them.

50. For commentators who reject the normal justification thesis, see supra note 5. For more general rejections of Raz's account of authority, see, e.g., Coleman, supra note 1, at 133; Dworkin, supra note 7, at 206–211.

51. At least, not among legal philosophers. Raz could argue that there is a generally accepted view within the broader community. However, he offers no reason to think that this is so. Even if he did, it would not be an adequate response to the objection for the reason given below.

52. The point could equally be made using the preemption thesis, since there is also widespread disagreement regarding that thesis; see supra note 4.

53. Strictly speaking, Raz's criterialism requires only that there be generally shared operative beliefs, not that the normal justification thesis feature among those beliefs. However, if that thesis does not feature among those beliefs, Raz must choose between his criterialism and his account of authority.

54. This point could also be made using the preemption thesis. We can imagine a society in which people are split roughly fifty-fifty between those who, once they recognize X as having authority, typically treat its directives as replacing their dependent reasons and those who do not.

55. Raz, Problem, supra note 2, at 1008. I use this example because Raz appeals to the existence of multiple concepts of authority when analyzing the example; see infra.

56. Id.

57. See Joseph Raz, On the Nature of Law, in Between Authority and Interpretation: On the Theory of Law and Practical Reason 91–125 (2009), at 91 (stating that if a jurisprudential theory is true, its theses “apply universally, that is they speak of all law, of all legal systems; of those that exist, or that will exist, and even of those that can exist though they never will”).

58. Raz, Authority, supra note 3, at 217. The full quotation is set out in Section II.

59. The “centrality of legal institutions in our structures of authority” is also a contingent fact, at least on the understanding of this phrase suggested in Section IV.

60. See text accompanying notes 16 and 17supra.

61. Raz, Problem, supra note 2, at 1010–1011.

62. Raz does argue that the normal justification thesis is “supple” enough to account for a number of apparent counterexamples; id. at 1030–1031. This might be taken to suggest that the normal justification thesis is true according to most concepts of authority. However, Raz would need to show that with regard to the remaining concepts of authority for which that thesis is not true, one of the other key theses is true. Moreover, as Scott Hershovitz argues, interpreting the normal justification thesis in this “supple” way renders it “nearly empty”; Hershovitz, Scott, The Role of Authority, 11 Philosophers’ Imprint 1–19 (2011), at 5–6Google Scholar.

63. See, e.g., Raz, Morality, supra note 2, at 55 (“The dependence and the normal justification theses are mutually reinforcing.”), and 57 (“From the dependence and normal justification theses it is but a short step to the pre-emption thesis.”).

64. The authority facts are quoted in full in the extract accompanying note 10supra.

65. Raz, Authority, supra note 3, at 215.

66. On Raz's account, obligations are necessarily preemptive in nature. However, for the same reasons that he recognizes multiple concepts of authority, he is likely to recognize multiple concepts of obligation. If so, it is unlikely that officials in all possible legal systems use a concept of obligation according to which obligations are preemptive in nature.

67. I am grateful to an anonymous referee for very helpful comments on this issue.

68. Stavropoulos’ semantic realism is set out in Nicos Stavropoulos, Objectivity in Law (1996). Limitations of space prevent me from considering more than one alternative to Raz's criterialism. However, while Stavropoulos's semantic realism provides the basis for his antipositivism, it is clearly not the case that all positivists must accept Raz's criterialism. Thus, although the ensuing discussion represents a critique of Raz's criterialism and hence the argument from authority, it is not meant as a critique of positivism per se. It is even controversial whether semantic realism, when applied to law, favors antipositivism. Stavropoulos argues that it does, and I am inclined to agree. However, if there are natural kinds like water and moral kinds like authority, perhaps there are also social kinds, where the extension of the relevant concept is fixed by sociological rather than moral considerations. Perhaps legal concepts are of this sort. (I am grateful to Patrick Emerton for helpful discussions of this issue.)

69. The classic presentation of those arguments is in Saul Kripke, Naming and Necessity (1980); and Putnam, Hilary, The Meaning of “Meaning,” in Language, Mind, and Knowledge 131–193 (Gunderson, K. ed., 1975)Google Scholar.

70. In fact, I think that Raz believes he can accommodate Putnam's insights about natural kind terms within his criterialist approach; see Raz, Two Views, supra note 21, at 19.

71. See Thagard, Paul R., The Best Explanation: Criteria for Theory Choice, 75 J. Phil. 76–92 (1978), at 86–88CrossRefGoogle Scholar. While it was developed with scientific explanation in mind, Thagard's approach to theory choice has been influential among philosophers in general (see, e.g., Lycan, William G., Epistemic Value, 64 Synthese 137–164 (1985)CrossRefGoogle Scholar, at 159 n.11) and legal philosophers in particular (see, e.g., Leiter, Brian, Explaining Theoretical Disagreement, 76 U. Chi. L. Rev. 1215–1250 (2009)Google Scholar, at 1239 n.101; Moore, Michael, Metaphysics, Epistemology, and Legal Theory, 60 S. Cal. L. Rev. 453–506 (1986)Google Scholar, at 481 n.111).

72. The central thesis of Raz's criterialism is set out in full in Subsection III.B.

73. It will sometimes be controversial as to what counts as a central claim of a theory and what counts as an additional hypothesis. However, it seems plausible to treat Raz's canonical statement of his criterialism as representing the central claim of that theory and so to regard his contention that there are multiple concepts of authority as an additional hypothesis needed to explain widespread disagreement and error.

74. Putnam, supra note 69, at 144.

75. Stavropoulos, supra note 68, at 46.

76. Id. at 15.

77. See, e.g., Thagard, supra note 71, at 79–80. Clearly, there is much more to be said here. However, this very rough characterization of explanatory power is sufficient for my purposes.

78. While there are important differences between them, this objection bears obvious similarities to Dworkin's treatment of the semantic sting argument; Ronald Dworkin, Law's Empire (1986), at 43–46. Note also that the problem for Raz is not confined to concepts of authority, since he recognizes a multiplicity of other concepts as well: see, e.g., Raz, Can There Be, supra note 27, at 331 (recognizing a multiplicity of concepts of law).

79. Raz denies that the umbrella concept is the “real” concept of authority; see Subsection III.C.

80. Raz, Two Views, supra note 21, at 15.

81. Cf. Bix, Brian, Raz, Authority, and Conceptual Analysis, 50 Am. J. Juris. 311–316 (2005)CrossRefGoogle Scholar, at 314 (arguing that the fact that we do not talk at cross-purposes when discussing the law does not necessarily show that we all hold the same concept of law, since that fact might instead be explained on the basis that there is overlap in the various concepts of law that different people hold).

82. The similarities between these two concepts are not limited to the fact that both are picked out using the word “bank.” There are also similarities of content—e.g., both refer to physical objects.

83. I am grateful to an anonymous referee for suggesting this possibility.

84. Is this because most discussions of authority occur between people who share a concept of authority? Raz cannot assume so because he allows that there may be no single concept of authority that is “ours.”

85. For example, it is controversial whether semantic realism can account for nondenoting terms such as “unicorn.” (For Stavropoulos’ attempt to do so, see Stavropoulos, supra note 68, at 45.) Moreover, some antiobjectivists about morality might regard the concept of (morally legitimate) authority as a nondenoting term. Conversely, however, if one is a moral objectivist, one should be wary about adopting a criterialist approach according to which the correct criteria for applying moral concepts like authority are the criteria that are generally used in practice.

86. It does, however, draw on Raz's claim that an explanation of a concept includes an account of that which it is a concept of, since this is necessary to explain why he is concerned with the possibility of confusion about the concept of authority; see Subsection III.A.

87. Raz, Authority, supra note 3, at 217. The full quotations are presented in Section II in the text accompanying note 18 and the extract accompanying note 19.

88. Waluchow, supra note 7, at 123.

89. Raz, On the Nature, supra note 57, at 96–97.

90. But cf. Raz, Problem, supra note 2, at 1006 (“The account [of authority he offers] is not about what people think it is like to have authority or to be subject to it, but of what it is to have it or be subject to it.”). I also leave to one side the question of whether the thesis that explaining the concept of authority involves explaining our self-understanding is plausible only if we accept Raz's criterialism.

91. I am grateful to an anonymous referee for pressing me to address this possibility.

92. Indeed, Raz suggests that, within the one society, its concepts may change as its self-understanding changes; Raz, On the Nature, supra note 57, at 98.

93. See, e.g., Raz, Joseph, Incorporation by Law, 10 Legal Theory 1–17 (2004)CrossRefGoogle Scholar.

94. See, e.g., Shapiro, Scott J., On Hart's Way Out, in Hart's Postscript: Essays on the Postscript to The Concept of Law 149–191 (Coleman, J. ed., 2001)Google Scholar; Marmor, Andrei, Exclusive Legal Positivism, in The Oxford Handbook of Jurisprudence & Philosophy of Law 104–124 (Coleman, J. & Shapiro, S. eds., 2002)Google Scholar, at 106–108.