Morality and The Nature of Law (Kenneth Einar Himma)

M O R A L I T Y A N D T H E N AT U R E   O F   L AW
ii
Morality and
the Nature of Law
KENNETH EINAR HIMMA
1
iv
1
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Kenneth Einar Himma 2019
The moral rights of the author have been asserted
First Edition published in 2019
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Crown copyright material is reproduced under Class Licence
Number C01P0000148 with the permission of OPSI
and the Queen’s Printer for Scotland
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2018961313
ISBN 978–0–19–872347–9
Printed and bound by
CPI Group (UK) Ltd, Croydon, CR0 4YY
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
This volume is dedicated to my mom, who has taught me more about
courage, patience, honor, and kindness than anyone else. I have never
known anyone who has faced so much adversity with so much grace
and elegance. She is and will always be an inspiration to me.
vi
Preface
This volume turned out to be quite different from what I had expected. When
I first contemplated writing it, I thought that I would just take a number of
previously published papers and do some minor edits to make them suitable
as chapters of the volume. I figured that I would have to write no more than
two new chapters, one introductory and the other purely expository, as I had
previously published papers on the topics of all but two of the issues I had
planned to address. Then I set about reading those papers with an eye toward
revising them, and quickly realized that I had overestimated the quality of my
published work. Each was riddled with errors that were so glaringly bad that
I was appalled and embarrassed that I had published them.
It is important to be honest about one’s mistakes, so I want to own them
with the kind of public apology one hopes never to have to make: if you read
these papers, I sincerely apologize for having wasted your time. From where
I now sit, I think there were ideas worth developing in all of them, but I let
them go much too quickly; as written, those papers were not worth reading.
While there is nothing I can do to get them back, I believe that I have learned
from my mistakes.
Obviously, I cannot claim that this volume is free from errors. What I can
say is that I have spent much more time with each of these chapters than
I spent on any of my previously published work. I have always felt uncom-
fortable imposing on others for detailed comments for various reasons; that
has not changed and likely will not change. Even so, I am hopeful that the
quality of the work here is significantly better than anything I have produced
up to now.
That said, I would like to acknowledge the following people who have
helped and encouraged me over the years: Matthew Adler, Larry Alexander,
Brian Bix, Evgeny Borisov, David Brink, Carlos Bernal, Luka Burazin,
Thomas Bustamante, André Coelho, Jules Coleman, Jennifer Corns, Jorge
L. Fabra, Kenneth Ehrenberg, Imer B. Flores, John Gardner, Leslie Green,
Mark Greenberg, Douglas Husak, Miodrag Jovanović, Nina Kaneda, Matthew
Kramer, Massimo La Torre, C. Stephen Layman, Brian Leiter, Lucas Miotto,
Adam Moore, Ronald Moore, Mark Murphy, Aleš Novak, Vitaly Ogleznev,
Stephen Perry, Joseph Raz, Andrea Romeo, Corrado Roversi, Frederick Schauer,
Stefan Sciaraffa, Scott Shapiro, Lawrence Solum, Bojan Spaić, Horacio Spector,
Juan Pablo Sterling Casas, Noel Struchiner, Valeriy Surovtsev, William Talbott,
vi
viii Preface
Jeremy Waldron, and Wilfrid Waluchow. In addition, I would like to thank
the following people who have commented on some of the chapters, discussed
some of the ideas with me, or helped me in other ways: Claudia Ceniceros,
Andrew Jordan, Milena Jovanović, Marija Momic, Zrinka Mrkonjić, Juan
Jose Otalvares, Svan Relac, and Ashley Robles. I would also like to acknowl-
edge the following universities for having supported my work in one form
or another over the years: National Autonomous University of Mexico
(UNAM), Seattle Pacific University, Tomsk State University, University of
Washington, University of Belgrade Faculty of Law, and University of Zagreb
Faculty of Law.
Finally, I want to thank Maria Elias Sotirhos for everything that she has
done for me over the nearly thirty years we have been together. She has been
astoundingly supportive, as just about anyone who knows both of us can at-
test, putting up with more than any partner can reasonably be expected to put
up with. Maria has painstakingly proofread every chapter of the book, despite
utterly, unrelentingly, and, at times, loudly despising its topic. I cannot im-
agine how people go through life without the sort of unconditional love and
kindness that she gives so naturally. I adore you, my boo.
Contents
Introduction: What Do You Mean by “Law,” Anyway? 1
. Relationships Between Law and Morality

1 5
1. Three types of inquiry about law 6
2. Natural law theories 9
3. Inclusive legal positivism 15
4. Who cares? 18
4.1 Does a conceptual theory of law have any practical implications? 18
4.2 Should conceptual jurisprudence not be done? 22
2. Rethinking the Traditional Interpretation of Anti-Positivist

Theories: Classical Natural Law Theory and Dworkinian
Interpretivism 29
1. Modest and immodest approaches to conceptual analysis 29
2. Two concepts of law 35
3. Four possible interpretations of a conceptual theory of law 40
4. Legal positivism as assuming MCA to explain the descriptive
concept of law 41
5. The traditional interpretation of natural law theory construed
as a rival to positivism 43
6. Dworkin’s interpretivism construed as a rival to positivism 48
7. Construing classical natural law and interpretivism as
deploying MCA to explain an evaluative concept of law 52
8. What do Finnis and Dworkin say? 55
9. Can ICA ground a viable conceptual methodology? 56
10. Conclusions 58
. Legal Positivism and the Possibility of Moral Criteria of Validity

3 61
1. The Differentiation Thesis 61
2. Conceptual foundations of legal positivism 66
2.1 The Artifact Thesis 66
2.2 The Separability Thesis 69
2.3 The Conventionality Thesis 71
2.4 The relationship between the criteria of validity and the
rule of recognition 76
3. Inclusive legal positivism 82
x
x Contents
. Inclusive Positivism and the Arguments from Authority
4 89
1. Epistemic and practical authority 90
2. Practical authority as grounded in a claim of right 95
3. Power, de facto authority, legitimate authority, and law 96
4. Conceptual relations between law and authority 100
5. The service conception of authority 104
6. Practical authority and the possibility of moral criteria of
validity: The Arguments from Authority 114
6.1 Law’s claim of legitimate authority: The general strategy of the
Arguments from Authority 116
6.2 The Authority Thesis, the Preemption Thesis, and inclusive
positivism 118
6.3 The Authority Thesis, NJT, and inclusive positivism 118
6.4 The directives of practical authority as expressing its view about
what ought to be done 119
7. Looking ahead: Evaluating the Arguments from Authority 120
. Law’s Claim of Legitimate Authority

5 121
1. Understanding the content of a claim of legitimate authority 122
2. Can law make claims? Two possible interpretations 123
3. Deriving law’s claim of legitimacy from the beliefs and claims
of officials 129
4. Do the beliefs and claims of officials imply a claim
of legitimacy? 130
4.1 The use of the language of rights and duties 131
4.2 Officials’ belief that they have a right to impose obligations 134
4.3 Officials’ claim that subjects owe allegiance 137
4.4 Designation of officials as “authorities” 140
4.5 Officials’ claim that subjects have an obligation to obey the law 141
4.6 The claims taken together 144
. Authority, Moral Criteria of Validity, and Conceptual Confusion

6 149
1. Identifying the content of our conceptual practices 150
2. The Identification Thesis and the Arguments from Authority 153
3. Could law’s claim of authority be conceptually confused? 155
4. Do most officials accept the Identification Thesis? 159
7. To Whom the Rules Apply: Norm Guidance and the

Incorporation Thesis 167
1. The guidance function of law 168
2. Motivational and epistemic guidance 174
3. The Practical Difference Thesis, the rule of recognition, and
valid legal norms 175
Contents xi
4. Law and the guidance of non-official behavior 176
4.1 The first Guidance Argument 177
4.2 Can norms valid in virtue of moral merit epistemically guide
non-official behavior? 178
4.3 Can subjects learn of their legal obligations regarding non-
official behavior from the rule of recognition? 179
4.4 Must subjects be able to learn of their legal obligations regarding
non-official behavior from the rule of recognition? 182
4.5 A methodological objection: The argument is illicitly grounded
in contestable claims about morality 184
5. Law and the guidance of official behavior 186
5.1 The second Guidance Argument 186
5.2 Can a valid legal norm governing non-official behavior
motivationally guide a judge deciding a case under that norm? 188
6. Revisiting the Arguments from Authority: To whom the rules
apply 193
. The Conceptual Possibility of Moral Criteria of Legal Validity

8 197
1. General methodological considerations 197
2. Prerequisites for a model vindicating the Incorporation Thesis 200
2.1 The model must describe a world that is nomologically possible 200
2.2 The modeled system must meet Hart’s minimum conditions for
the existence of a legal system 201
2.3 The modeled system must incorporate the minimum content of
natural law 202
2.4 The model must satisfy the service conception of authority 203
2.5 The norms of the system must be metaphysically capable of
motivationally and epistemically guiding the behavior they
govern 203
2.6 The model must be plausibly interpreted as incorporating moral
criteria of validity 205
2.7 The model should be incompatible with an exclusivist
interpretation 205
3. Specification of a model vindicating the Incorporation Thesis 206
3.1 The modeled world is nomologically possible 206
3.2 The subjects are accidentally infallible 206
3.3 A model of an institutional normative system with moral criteria
of validity 209
4. Vindicating the Incorporation Thesis: The modeled system is a
legal system 210
4.1 The system satisfies Hart’s minimum conditions for the existence
of a legal system 211
4.2 The system contains norms that incorporate the minimum
content of natural law 212
xi
xii Contents
4.3 The system satisfies the service conception of authority 212
4.3.1 The system has de facto authority 212
4.3.2 The norms are capable of expressing the authority’s view
about what right reason requires and of replacing the
subjects’ views about what right reason requires 213
4.3.3 Is it metaphysically possible for subjects to better
comply with what right reason requires by following the
authority’s view of what right reason requires than by
following their own views of what right reason requires? 214
4.4 The modeled system is capable of performing law’s conceptual
function of regulation through norm-governance and norm
guidance 216
4.4.1 The rule of recognition is metaphysically capable of
motivationally and epistemically guiding official behavior 216
4.4.2 The norms valid under the rule of recognition are
metaphysically capable of motivationally and epistemically
guiding non-official behavior 218
4.5 The system precludes an exclusivist interpretation 220
5. Why all the fuss? The probability of inclusive legal systems in
our world 221
Index 225
Introduction
What Do You Mean by “Law,” Anyway?
The notion of law that is typically the focus of conceptual jurisprudence

applies to all and only institutional systems of norms; it is clearly part of the
nature of law that it is institutional. As such, the concept-term “law” picks
out paradigms of institutional systems of both municipal law and interna-
tional law. These systems regulate different kinds of behavior: institutional
systems of municipal law, as a conceptual matter, necessarily involve state
regulation of both official and non-official behavior while institutional sys-
tems of international law necessarily involve multinational regulation of only
official behavior on the part of nation-states. While it might be conceptually
possible to have systems of international law that regulate non-official behav
ior, it is not conceptually necessary that they even purport to regulate non-
official behavior.1
But this general notion of law also applies to institutional normative sys-
tems that are like and unlike these legal systems in theoretically significant
ways. Like systems of municipal and international law, a system of religious
law has something fairly characterized as a rule of recognition for recogniz-
ing, applying, or enforcing norms that govern the behavior of members of
the relevant community, as well as a set of authorized sanctions, such as ex-
communication, reasonably contrived to induce compliance among subjects
otherwise disposed to violate norms of the system. Unlike systems of munic-
ipal and international law, a system of religious law is necessarily concerned
1
A federal system that brings together subnational entities into a unified national entity is some-
what more difficult to characterize. Conceived as conceptually independent of the subnational enti-
ties it unites, a federal legal system more resembles systems of international law than it does systems
of municipal law in virtue of necessarily containing only norms governing the official behavior of
the states. But conceived as conceptually dependent upon the subnational entities it brings together,
it incorporates the law of those subnational entities and hence more resembles a municipal system
than it does systems of international law.
Morality and the Nature of Law. © Kenneth Einar Himma 2019. Published 2019 by Oxford
University Press.
2
2 Introduction
with articulating, codifying, and enforcing a shared morality deemed to be
the canon of the faith.
Although there is something to be gained in our understanding of law by
explicating the content of a concept that applies to both systems of religious
law and to systems of municipal and international law, there is also something
to be lost. On the one hand, systems of religious law have much in common
with systems of municipal and international law; that is why the term “law”
can be taken to refer to both types of system. On the other hand, systems of
religious law differ in theoretically significant ways from systems of munic-
ipal and international law. To pursue an explication of a general notion of law
that applies equally to each will of necessity suppress theoretically significant
features of one that distinguish it from the other and will hence result in a
theory that under-explicates both.
Another potential drawback to focusing on a general notion that applies
to both subtypes is that it can lead to confusion with respect to our under-
standing of each subtype. Consider H.L.A. Hart’s puzzling remarks about the
conceptual function of law:
[W]‌hereas Dworkin’s interpretive legal theory in all its forms rests on the presup-
position that the point or purpose of law and legal practice is to justify coercion, it
certainly is not and never has been my view that law has this as its point or purpose.
Like other forms of positivism my theory makes no claim to identify the point or
purpose of law and legal practices as such; so there is nothing in my theory to support
Dworkin’s view, which I certainly do not share that the purpose of law is to justify
the use of coercion. In fact, I think it quite vain to seek any more specific purpose
which law as such serves beyond providing guides to human conduct and standards
of criticism of such conduct.2
These remarks are usually interpreted as asserting that the conceptual func-
tion of law is to guide behavior, but a more careful look indicates that Hart
is skeptical about the very possibility of identifying a distinctive conceptual
function of law. It is not just that he states “[l]‌ike other forms of positivism,
my theory makes no claim to identify the point or purpose of law and legal
practices as such”; it is also that he thinks it “quite vain” to say anything
more about law’s conceptual function than that it “provides guides to human
conduct.”
Notice that the claim that law’s conceptual function is to guide behavior
would not distinguish systems of municipal or international law from systems
of religious law. Although both systems are concerned with doing something
2
H.L.A. Hart, The Concept of Law 3rd Ed. (Oxford: Oxford University Press, 2012), 248–49.
Hereinafter CL.
Introduction 3
that involves guiding behavior through the governance of norms, legal norms
and norms of religious law differ, as a conceptual matter, in content because
they are concerned to achieve different things. In our world, municipal and
international law seek to diminish the likelihood of violent conflicts among
subjects so that they can reap the social benefits of living together in a com-
munity; the norms of religious law seek to enforce moral norms grounded in
theological commitments that help to distinguish one faith tradition from
another. If the concern is to explicate all of these various types of norm, all
that can be said about law’s conceptual function is that it is concerned to reg-
ulate behavior through norm-guidance.
But if the concern is to explicate our conceptual practices with respect to
the nature of law as it applies to systems of municipal and international law,
as I take it to be, the failure to distinguish municipal and international law
from other systems of law is far from innocuous. The claim that the concep-
tual function of an automobile is to transport persons or things from one des-
tination to another is problematic because it fails to distinguish automobiles
from airplanes. Just as any conceptual theory of an automobile that fails to
identify a conceptual function that distinguishes automobiles from airplanes
is problematic for that reason, so is a conceptual theory of law that fails to
identify a conceptual function that distinguishes systems of municipal and
international law from other systems of law.
Pursuing an explication of the most general concept of law can lead to
other confusions. If one refers to the rules of a chess association as “law,” as
the World Chess Federation (WCF) does, then it is utterly uncontentious
that positivism’s Separability Thesis is true; if any system of norms properly
characterized as “law” is artifactual all the way down in the sense that the con-
tent of its norms is fully manufactured by its officials, the norms of a chess as-
sociation are.3 Although there are good reasons to reject the conceptual claim
that there can be no unjust laws, it would be silly to argue that the system of
WCF rules suffices to establish the Separability Thesis and thereby to refute
a view of historic importance that has been traditionally associated with clas-
sical natural law theory.
Although it is perfectly legitimate to pursue a conceptual theory that
applies as much to chess and religious law as to municipal and international
law, this is not the approach I take here. This volume is concerned exclusively
with jurisprudence as the term is used among attorneys, judges, and law stu-
dents. This usage is properly applied only to matters arising within systems
of municipal law and international law. The approach here focuses, like most
3
Handbook of the World Chess Federation, available at https://www.fide.com/component/hand-
book/?id=124&view=article.
4
4 Introduction
scholarly inquiries concerning the nature of law, on the practices associated
with law as it pertains to political entities like states, but it assumes that these
systems have a metaphysical nature that is related to but distinct from the na-
ture of other kinds of systems of law.
None of this should be taken to deny that explicating the general concept
that applies equally to municipal law, religious law, and the rules of a chess
association is a worthwhile endeavor. The linguistic conventions we adopt
define a conceptual framework that not only gives structure to the world of
our experience but also says something important about us. An analysis of the
more general concept of law hence conduces to our collective understanding
of who we are and what we value. That our ordinary usages permit the ap-
plication of the term “law” to systems that are as different from one another
as municipal law, chess law, and religious law tells us something important
about ourselves. But I have no urgent interest in the conceptual nature of
chess law or religious law; what I do care a lot about is the nature of the type
of law studied in law schools and practiced in systems of municipal and inter-
national law—and that is the exclusive concern of this volume.
There is one more feature of the approach adopted here that should be
noted. I am of the view that the substance of a conceptual theory is deeply
conditioned by the methodology that underwrites the analysis. The analysis
is explicitly grounded in the linguistic and legal practices that inform our or-
dinary usage with respect to the term “law” and is exclusively concerned with
that usage. The ultimate touchstone, then, for evaluating the claims I make
about the concept of law is whether they conform to ordinary usage as fleshed
out to expose not only the underlying social practices that define them but
also the philosophical presuppositions that ground those practices.
1
Relationships Between Law and Morality
It is not implausible to think that the project of general jurisprudence was

partly motivated by disputes about conceptual relationships between law and
morality. Jeremy Bentham’s and John Austin’s legal positivism was articulated
partly in response to William Blackstone’s view that it is a conceptually nec-
essary condition for a norm to count as law that its content not conflict with
objective standards of justice. Bentham and Austin denied this claim, arguing
instead that the content of the law is fully determined by the commands of a
sovereign willing and able to back them with the threat of a sanction.1
In this chapter, I distinguish three types of inquiry about law in order to
explain the conceptual project with which this volume is concerned. Then
I articulate the two conceptual views about morality and the nature of law
that comprise the focus of this volume. First, I explain positivist and anti-
positivist views with respect to whether it is a conceptual truth that the crite-
ria of legal validity include moral constraints on the content of law. Second,
I explain the dispute between inclusive and exclusive positivists with respect
to whether it is conceptually possible for a legal system to have content-based
moral criteria of validity.
Finally, as the intellectual legitimacy of the project of conceptual jurispru-
dence has recently come under fire for not having practical consequences or
being “interesting” and hence as not being worth doing, I say something brief
in defense of the project. My defense, such as it is, will be somewhat modest.
I will dispute neither the claim that right answers to conceptual questions
lack significant practical consequences nor the claim that conceptual theoriz-
ing is uninteresting.
What I will do is argue that the claim that conceptual jurisprudence should
not be done is either unclear or false. On the one hand, if the claim that
1
See, generally, Jeremy Bentham (1782). Of Laws in General Ed. H.L.A. Hart (London: Athlone
Press, 1970); and John Austin (1832). The Province of Jurisprudence Determined Ed. Wilfred E.
Rumble (Cambridge: Cambridge University Press, 1995). For an outstanding discussion of the
history of these views, see Brian H. Bix, Jurisprudence: Theory and Context 7th Ed. (Durham,
NC: Carolina Academic Press, 2015), Chapter 5. Hereinafter JTC.
University Press.
6
6 Relationships Between Law and Morality

conceptual jurisprudence should not be done is a moral claim, it is false.
From the standpoint of morality, there are many things any legal scholar can
do that would make the world a much better place than writing articles for
academic journals; whatever difference there is between the moral value of
writing in conceptual jurisprudence and that of writing in other areas of legal
scholarship amounts to little. On the other hand, if it is not a moral claim,
then it is not clear exactly what it amounts to.
1. Three types of inquiry about law

A frequent area of interest to those who theorize about law concerns the var-
ious relationships between law and morality. To understand these relation-
ships, it is helpful to distinguish three kinds of inquiry concerning morality
and law. The first is empirical in the sense that it is concerned with identify-
ing certain contingent relationships in our world having to do with law and
morality. One can ask, for example, whether officials of some existing legal
system usually consider what they believe are moral requirements in making
decisions about how to create, adjudicate, and enforce the law. As this ques-
tion concerns the motivations of officials, addressing it requires going into the
world and observing what officials say and how they behave in discharging
their functions as officials.
The second kind of inquiry is normative in the sense that it is concerned
with determining, as a matter of morality, how officials should behave in
discharging their duties or what content the law should have. It is clear both
that laws should be just and that legal systems should be legitimate; and the
content of the law and legal practices should satisfy the appropriate moral
requirements. Normative inquiry, then, is concerned with determining
whether some existing law, legal system, or legal practice satisfies the relevant
demands of morality.
The third kind of inquiry is conceptual in the sense that it seeks to de-
scribe the content of the relevant concept and hence to explicate the na-
ture of the thing picked out by the concept. Conceptual claims purport to
identify the properties that distinguish things that are members of the ref-
erence class of the concept-term from things that are not. For example, the
concept-term “bachelor” is generally thought to apply only to things that
are unmarried and men. The instantiation of all these properties distin-
guishes things that are bachelors from things that are not bachelors; married
things, for example, do not fall within the reference class of the concept-term
“bachelor.”
Three types of inquiry about law 7
As traditionally conceived, the goal of conceptual analysis is purely descrip-
tive. Thus conceived, the goal is simply to explicate, or describe in a philo-
sophically rigorous way, the content of the relevant concept as it is grounded
in those social practices that determine the application-conditions of the rel-
evant concept-term. Conceptual analysis seeks to describe—rather than to
prescribe—the content of the relevant concept and is hence concerned to
provide an account of what the content of the concept is, and not of what the
content of the concept should be.
Explicating the content of a concept requires identifying those properties
that are conceptually necessary for something to fall under the concept in
the following sense: if p is a conceptually necessary property for being an A,
then it is conceptually impossible to be an A without instantiating p. To ex-
plicate the content of a concept is to identify those properties essential to the
thing picked out by the corresponding concept-term. Being unmarried, for
example, is a conceptually necessary condition for being a bachelor or, other-
wise put, is essential to bachelorhood. Insofar as these conceptually necessary
properties are essential properties of the thing picked out by the concept, they
also define the nature of the thing picked out by the relevant concept.
As is evident from the talk of necessary properties, the traditional method-
ology for conceptual analysis trades in the language of modal logic, utilizing
the modalities of necessity and possibility, along with possible-worlds talk.
For example, the conceptual truth that bachelors are unmarried entails that
there are no conceptually possible worlds in which there is someone who is
both married and a bachelor. This latter claim is logically equivalent to the
claim that in every conceptually possible world all bachelors are unmarried.
Accordingly, conceptual inquiry with respect to the nature of law is con-
cerned with determining what is true of law in some, none, or every concep-
tually possible world.
It is important to note that the character of the relevant modal claims
about the nature of a thing depend on the logical relationship between the
content of the concept and the linguistic conventions governing the use of
the relevant concept-term. If, on the one hand, these modal claims presup-
pose a particular conceptual framework grounded in contingent linguistic
practices, then the character of the resulting modal claims would aptly be
described as “conditionally” necessary, and not as “absolutely” necessary, since
the underlying practices can change over time. If, on the other, these modal
claims are thought not to be grounded in the linguistic practices that estab-
lish the application-conditions for using the relevant concept-term, then the
character of the resulting modal claims would aptly be described as “abso-
lutely” necessary.
8

Both of the accounts given above harmonize with the notion of a priori jus-
tification or knowledge, properly understood. To say that a proposition can be
justified or known a priori does not entail that it can be justified or known
independently of all empirical experience. To say that a proposition can be jus-
tified or known a priori is to say that it can be justified or known independ-
ently of any empirical experience not needed to understand the meaning of the
relevant terms. Accordingly, on the assumption that conceptual jurisprudence is a
priori, its focus on identifying the contingent linguistic practices that determine
the lexical meanings of words coheres with the character of a priori knowledge.
Likewise, the claim that conceptual claims are conditionally necessary in the
sense of being dependent on contingent linguistic practices also coheres with the
idea that they are a priori in the sense that they can be known or justified inde-
pendently of any empirical experience not needed to understand the meaning
of the relevant terms.
It is sometimes thought that conceptual analysis, conceived as being grounded
in the linguistic practices that define the lexical meaning of the relevant concept-
term, involves little more than providing a dictionary definition, but this is
mistaken. Thus conceived, conceptual theorizing begins from core intuitions
about how to use a word that are conditioned by its lexical meaning but goes
much deeper than lexicography insofar as it attempts to identify and theorize the
deeper philosophical commitments these intuitions imply or presuppose. While
this might or might not be a distinctively philosophical enterprise, it goes well
beyond the empirical task of merely identifying shared intuitions or core features
of our linguistic practices, which is the job of lexicographers.
Compare what lexicographers have to say about the word “law” with what
Hart has to say by way of explication of the concept of law. Oxford American
Dictionary defines “law” as follows:
law | noun 1 (often the law) the system of rules that a particular country or community
recognizes as regulating the actions of its members and may enforce by the imposition
of penalties: they were taken to court for breaking the law | a license is required by law |
[as adj.] law enforcement.
• an individual rule as part of such a system: an initiative to tighten up the laws on
pornography.
• such systems as a subject of study or as the basis of the legal profession: he was still
practicing law | [as adj.] a law firm . . . .
• a thing regarded as having the binding force or effect of a formal system of rules: what
he said was law.
Natural law theories 9
Notice how much of what Hart’s theory addresses is omitted by the lexical
definition.2 First, there is no mention here of many pieces central to Hart’s
analysis: the definition says nothing about social practices; the rule of recogni-
tion; secondary and primary rules; legal validity; and many other theoretically
salient features of law. Second, the lexicographer’s job is accomplished in a
few lines whereas Hart took more than two hundred pages to give an analysis
of the nature of law in The Concept of Law. If Hart starts from shared views
about the meaning of “law,” he is doing something radically different from
what lexicographers do—and going much deeper into what law ultimately is.
This book is a piece of traditional conceptual analysis that conceives the
project as grounded in the contingent linguistic practices that determine the
lexical meanings of the relevant concept-terms, but it should not be construed
as disparaging the other types of inquiry concerning law or other methodo-
logical principles that have been deployed in the service of conceptual anal-
ysis. I choose this methodology simply because it is the one that has been
adopted by the theorists whose work I find most interesting and with whose
work I wish to engage.
As the concern of this book is to explicate certain relationships between
the concepts of law and morality, I begin by briefly describing the positions
that are the focus of this volume: (1) the position that it is a conceptual truth
about law that the criteria of validity include moral constraints on the con-
tent of law (i.e. natural law theory); (2) the position that it is a conceptual
truth that the criteria of validity are exhausted by source-based criteria (i.e.
exclusive positivism); and (3) the position that it is conceptually possible for
a rule of recognition to incorporate moral criteria of validity (i.e. inclusive
positivism).
2. Natural law theories

It would be helpful to partition conceptual theories of law into two mutu-
ally exclusive and jointly exhaustive categories: positivist and anti-positivist
theories. Positivists hold that the content of law is an artifact that is wholly
manufactured through certain social activities. Both the norms that regulate
the non-official behavior of subjects and those that regulate the behavior of
officials acting as legislators, judges, and executives are social artifacts. Since
there are thus no conceptually necessary moral constraints on what counts
2
See, generally, H.L.A. Hart, The Concept of Law 3rd. Ed. (Oxford: Oxford University Press,
2012). Hereinafter CL.
10

as valid law, the content of the validity criteria is determined entirely by the
social activities of those who make, adjudicate, and enforce what they char-
acterize as law.
In contrast, so-called anti-positivist theories are traditionally interpreted
to deny that the content of the validity criteria is fully determined by our
practices. Thus construed, natural law theories assert that it is a conceptual
truth that the criteria of validity include moral constraints on the content of
valid law. On this view, it is a conceptually necessary condition for a norm to
count as a valid law that its content conforms to some set of objective moral
standards. Accordingly, the social processes through which people manufac-
ture law do not fully determine the content of the validity criteria; no matter
what people do by way of recognizing, applying, or enforcing what they char-
acterize as “law,” there are moral standards that constrain what really counts
as law in a legal system.
Classical natural law theory is hence traditionally interpreted as asserting
that “an unjust law is no law at all” (lex iniusta non est lex). While natural law
theory goes back at least as far as Cicero, the most influential early advocate
of this view is Aquinas, who appears to endorse it in the following passage:
As Augustine says, “that which is not just seems to be no law at all”; wherefore the
force of a law depends on the extent of its justice. Now, in human affairs a thing is
said to be just from being right according to the rule of reason. But the first rule of
reason is the law of nature, as is clear from what has been stated above. Consequently,
every human law has just so much of the nature of law is it is derived from the law
of nature. But if, in any point, it deflects from the law of nature, it is no longer a law
but a perversion of law.3
Aquinas’s view is subsequently picked up by Blackstone, who expresses it in
language quite similar to Aquinas’s:
This law of nature, being coeval with mankind and dictated by God himself, is of
course superior in obligation to any other. It is binding over all the globe, in all coun-
tries, and at all times: no human laws are of any validity, if contrary to this; and such
of them as are valid derive all their force, and all their authority, mediately or imme-
diately, from this original.4
While the traditional natural law view has been largely construed to assert
that it is conceptually impossible for there to be an unjust law, the passages
3
St. Thomas Aquinas, Summa Theologiae, Prima Secundae Partis Q. 95, Art. 2; available
at: http://www.newadvent.org/summa/2095.htm.
4
Sir William Blackstone, “Of the Nature of Laws in General,” Commentaries on the Laws
of England— Book I: Of the Rights of Persons (Oxford: Oxford University Press, 2016), 35.
Hereinafter COM.
above are ambiguous. Aquinas is concerned with the force of law, which
can plausibly be interpreted as referring to the moral force of law. Similarly,
Blackstone’s references to validity and authority are plausibly construed as
referring to moral validity and moral authority. These passages can be con-
strued as asserting no more than that an unjust norm posited as law would
not generate a moral obligation to obey and might be such that morality
obligates us to disobey it. On this interpretation, the claim would be that an
unjust posited norm is not morally binding, and not that an unjust posited
norm is not legally binding.
There are plausible reasons to think that Blackstone is making only claims
about moral validity and moral authority. The discussion that precedes the
passage quoted above is unambiguously concerned with what we are morally
obligated to do, and not with what we are legally obligated to do. Consider,
for example, the following statement, which precedes the last quoted remark
of Blackstone’s by four paragraphs:
Man, considered as a creature, must necessarily be subject to the laws of his creator,
for he is entirely a dependent being. A being, independent of any other, has no rule to
pursue, but such as he prescribes to himself; but a state of dependence will inevitably
oblige the inferior to take the will of him, on whom he depends, as the rule of his
conduct: not indeed in every particular, but in all those points wherein his depend-
ence consists. This principle therefore has more or less extent and effect, in propor-
tion as the superiority of the one and the dependence of the other is greater or less,
absolute or limited. And consequently as man depends absolutely upon his maker
for everything, it is necessary that he should in all points conform to his maker’s will
(COM 33–34).
The deontic locutions used in this passage (such as “must necessarily be sub-
ject,” “obliged,” and “should in all points conform”) are unquestionably moral
in character. It is clear that Blackstone is not concerned with what is pruden-
tially or legally normative—although one certainly has prudential reasons to
obey God’s will if eternal torment is the consequence of disobedience.
Blackstone does not begin to discuss “human law” for several paragraphs
after the last two quotes:
Upon these two foundations, the law of nature and the law of revelation, depend
all human laws; that is to say, no human laws should be suffered to contradict these
. . . . To instance in the case of murder: this is expressly forbidden by the divine, and
demonstrably by the natural law; and from these prohibitions arises the true unlaw-
fulness of this crime. Those human laws, that annex a punishment to it, do not at all
increase its moral guilt, or super-add any fresh obligation in foro conscientiae to ab-
stain from its perpetration. Nay, if any human law should allow or enjoin us to commit
it, we are bound to transgress that human law, or else we must offend both the natural
and the divine. But with regard to matters that are in themselves indifferent, and are not
12

commanded or forbidden by those superior laws; such, for instance, as exporting of wool
into foreign countries; here the inferior legislature has scope and opportunity to interpose,
and to make that action unlawful which before was not so (COM 35–36; emphasis
added).
The italicized portion suggest that Blackstone, like Aquinas, is concerned
with what is morally binding, rather than with what is legally binding in
virtue of being human-made law. The claim is that what we are obligated
to do all things considered is what morality obligates us to do; if a legal and
moral obligation come into conflict, we must do what the moral obligation
requires us to do.
The traditional interpretation of Blackstone as claiming that an unjust
norm cannot be a positive law, then, is implausible. The most reasonable in-
terpretation of Blackstone’s view, given the totality of his remarks, construes
him as being concerned with what is morally (or, as the matter is sometimes
put, “really”) binding and authoritative—and not with the project of concep-
tual jurisprudence as defined by the work of Austin and Bentham. On this
construction, the natural law view asserts only that it is in the nature of law
that it should be just; an unjust norm might be legally valid, but it is defec-
tive qua law insofar as it fails to instantiate the quality of justice to which it
aspires.
Brian Bix has persuasively argued for this interpretation of classical natural
law theory and explicates it as follows:
A more reasonable interpretation of statements like “an unjust law is no law at all”
is that unjust laws are not laws “in the fullest sense.” As we might say of some pro-
fessional, who had the necessary degrees and credentials, but seemed nonetheless to
lack the necessary ability or judgment: “she’s no lawyer” or “he’s no doctor.” This only
indicates that we do not think that the title in this case carries with it all the lauda-
tory implications that it usually carries. It may well be that for our purposes, knowing
that this doctor is not competent is the most important fact; however, the fact that he
does have the required certification is not thereby negated or made entirely irrelevant.
Similarly, to say that unjust laws are “not really laws” may only be to point out that
they do not carry the same moral force or offer the same reasons for action that come
from laws consistent with “higher law” (JTC 73–74).
Bix convincingly argues that Aquinas and Blackstone hold no stronger view
about the nature of positive law than this and that this view is the most “prob-
able interpretation for nearly all proponents of the position” (JTC 74).
Notice that there is no inconsistency between natural law theory and pos-
itivism if Bix is correct that the natural law position should be interpreted as
asserting only that an unjust law is not law in the fullest sense. Positivism does
not even purport to explain what it means to be law in the fullest sense of the
word “law.” Positivism is concerned only to explicate the existence conditions
for laws and legal systems, as those terms are used in the discourses of lawyers,
officials, and other legal practitioners.
Even so, the fact that this confusion about the natural law view is so wide-
spread and persistent—indeed, it is perpetuated in part by natural law theo-
rists who claim they are opposed to legal positivism—is indicative of a need
to consider the view that it is a conceptual truth that an unjust norm cannot
be law. A more rigorous examination and evaluation of this view will be
undertaken in Chapter 2.
For now, it suffices to note that one can hold that it is a conceptual truth
that the criteria of validity include moral norms without taking the position
that it is a conceptual truth that there are no unjust laws. To see this, it would
be helpful to distinguish two theses concerning a conceptual relationship be-
tween law and morality:
NL1: It is a conceptual truth that the criteria of validity include at least some moral
norms; and
NL2: It is a conceptual truth that the criteria of validity include all the norms of jus-
tice such that an unjust norm cannot be valid.
NL2 logically implies NL1. Given that the norms of justice are also norms of
morality, if it is a conceptual truth that the criteria of validity include all the
norms of justice, then it is also a conceptual truth that the criteria of validity
include at least some moral norms.
But NL1 does not logically imply NL2. The claim that the criteria of va-
lidity include some moral norms does not imply that they contain all the
norms of justice. Indeed, NL1 does not clearly imply that the criteria of legal
validity include any norms of justice; if there are norms of political morality
that are not norms of justice, then NL1 is consistent with the criteria of va-
lidity including only those norms.
Consider norms requiring democratic elections. It is not preposterous to
think that norms requiring procedural democracy are not properly character-
ized as norms of “justice.” If the notion of justice is best construed as ensuring
that people get what they deserve economically and otherwise, then norms
of democracy are not norms of justice insofar as it is not true that people
“deserve” to elect their officials.5 It might be true that people have a moral
5
To say that it is not true that people deserve to elect the officials that govern them is not to say
that people do not deserve to elect the officials that govern them. When people claim that someone
does not deserve some benefit, they usually intend to say that someone is undeserving of the benefit
and should not receive it. That is not what I mean by the claim that it is not true that people deserve
to elect the officials that govern them. What I mean is that the concept of desert is irrelevant with
respect to explaining why people should be allowed to elect those officials. The concept is inapt here
14

right to elect their leaders, but having a moral right to some X is not logi-
cally equivalent to morally deserving X. I might have a moral right to inherit
someone’s estate, but that does not entail that I deserve that person’s estate.
Accordingly, NL1 does not entail NL2.
Accordingly, holding NL2, the traditional natural law theory, is not the
only way to hold NL1. Ronald Dworkin, for example, is commonly thought
to hold NL1 without holding NL2:
We need not deny that the Nazi system was an example of law . . . because there is an
available sense in which it plainly was law. But we have no difficulty in understanding
someone who does say that Nazi law was not really law, or was law in a degenerate
sense, or was less than fully law. For he is not then using “law” in that sense; he is not
making that sort of preinterpretive judgment but a skeptical interpretive judgment
that Nazi law lacked features crucial to flourishing legal systems whose rules and
procedures do justify coercion.6
As Dworkin describes his view, the claim that the Nazis did not have law be-
cause their norms were too wicked to count as law is false if it makes a “prein-
terpretive judgment,” which is the type of judgment he takes to be expressed
by positivism’s Separability Thesis. Accordingly, it would seem that Dworkin
rejects NL2.
But it seems clear that Dworkin holds NL1, at least as his view has tradi-
tionally been interpreted. As Dworkin describes his interpretivist theory of
law, it is “the theory that the law of a community consists not simply in the
discrete statutes and rules that its officials enact but in the general principles
of justice and fairness that these statutes and rules, taken together, presup-
pose by way of implicit justification.”7 Elsewhere, he states that “[t]‌he law
of a community on this account is the scheme of rights and responsibilities
that meet that complex standard: they license coercion because they flow
from past decisions of the right sort . . . [and] are therefore ‘legal’ rights and
responsibilities” (LE 93). His view, thus, seems to be that it is a conceptual
truth that the criteria of validity include those moral principles that morally
justify the existing institutional history of the legal system—which would, if
correct, entail that he accepts NL1.
Chapter 2 considers the plausibility of anti-positivist accounts of the rela-
tionship between law and morality, arguing that classical natural law theory
because it is not true that people deserve to elect their officials and it is not true that they do not
deserve to elect those officials.
6
Ronald Dworkin, Law’s Empire (Cambridge, MA: Harvard University Press, 1986), 103–4.
Hereinafter LE.
7
Ronald Dworkin, “The Law of the Slave Catchers,” The Times Literary Supplement, December
5, 1975, 1437.
Inclusive legal positivism 15
and Dworkin’s interpretivism are best construed as explicating a concept of
law that has evaluative content and is hence distinct from the purely descrip-
tive concept that positivism takes itself to explicate. Although there is reason
to think that many natural law theorists have been misinterpreted as rejecting
the Separability Thesis and holding NL2, I nonetheless evaluate this view.
The rationale is not that any particular theorist holds that view. Rather, the
rationale is that, since so many theorists and students of conceptual juris-
prudence have historically interpreted the natural law view as denying the
Separability Thesis, the view should be evaluated in any project that explores
the conceptual relationships between morality and the criteria of validity.
3. Inclusive legal positivism

As will be discussed in more detail in Chapter 3, positivist theories are typ-
ically thought to be grounded in three core theses: the Separability Thesis,
the Artifact Thesis, and the Conventionality Thesis. The Separability Thesis
denies that it is a conceptual truth that the criteria of validity include moral
norms of any kind and hence denies both NL1 and NL2 above. According to
the Separability Thesis, then, there is a conceptually possible legal system in
which the criteria of validity do not include moral constraints on the content
of law.
According to the Artifact Thesis, law is, by nature, an artifact in the sense
that the content of every norm that counts as law is wholly determined by
certain social activities. This entails that the content of both the second-order
norms empowering and constraining officials in their lawmaking and adjudi-
cative functions (and thereby define the criteria of validity) and the first-order
norms legally valid in any legal system is wholly manufactured by human
beings. If this is correct, there is no legal content that is determined by some-
thing other than the social activities of officials, such as norms of objective
morality. Law is an artifact all the way down.
Finally, according to the Conventionality Thesis, it is a conceptual truth
that the criteria of validity are determined by the content of a conventional
rule of recognition governing how officials should discharge their duties and
how they may exercise their powers in the institutional normative system
their activities help to create and sustain. The content of a social convention
that governs members of a social group is determined by the fact that people
in the group converge (1) in accepting that content as governing their
behavior and (2) in complying with that content. Thus, the Conventionality
Thesis expresses Hart’s view that the rule of recognition is a conventional
16

rule of the system insofar as the officials take the internal point of view to-
ward it (thereby satisfying condition (1) above) and generally conform their
lawmaking and adjudicative acts to its requirements (thereby satisfying con-
dition (2) above).
These theses are agnostic with respect to the question of whether it is con-
ceptually possible for a legal system to have moral criteria of validity. The
Separability Thesis, for example, asserts only that it is conceptually possible
for a legal system to have criteria of validity that do not include moral con-
straints on the content of law; that claim implies nothing with respect to
whether it is conceptually possible for a legal system to have criteria of validity
that include such moral constraints. Similarly, the claims that law is a social
artifact and that the criteria of validity are conventional in character do not
obviously imply anything about whether it is conceptually possible for a legal
system to have moral criteria of validity.
Two positions have emerged on this issue. According to inclusive pos-
itivism, it is conceptually possible for a legal system to incorporate moral
criteria of validity; that is, inclusive positivism implies that there are concep-
tually possible legal systems with moral criteria of validity. In contrast, exclu-
sive positivism denies that there are conceptually possible legal systems with
moral criteria of validity. According to exclusive positivism, it is a conceptual
truth that the criteria of validity are exhausted by source-based criteria of va-
lidity that have to do with the manner in which a norm is promulgated as law.
As it is doubtful that any major legal theorist holds NL1, most of the book
will be devoted to the dispute between inclusive and exclusive positivism.
Chapter 3 begins with the articulation of a methodological assumption that is
the foundation for the project of conceptual jurisprudence. The chapter goes
on to provide a more detailed explication of the core theses of positivism and
of how these core theses are logically related to one another.
The next chapters of the book are concerned with evaluating the most in-
fluential arguments against inclusive positivism. Chapters 4, 5, and 6 address
Joseph Raz’s argument that the nature of law entails a claim of legitimate au-
thority that is incompatible with the conceptual possibility of moral criteria
of validity. Chapter 4 develops Raz’s challenging and nuanced analysis of the
nature of authority and his argument that it is incompatible with moral cri-
teria of validity.
Chapter 5 explains and evaluates Raz’s claim that it is part of the very
nature of law that it claims legitimate authority. I consider a number of dif-
ferent interpretations of this view and challenge each. To begin, for example,
I question how an institutional abstract object like a legal system could make
claims. The difficulty here is roughly analogous to the metaphysical diffi-
culties associated with the idea that the framers of the Constitution have
something that would count as a collective intention; as has frequently been
observed, it is difficult to see how collective entities can have mental states.
I go on to consider other interpretations in terms of the beliefs and acts of
officials and argue that none of these other interpretations adequately support
the idea that every conceptually possible legal system, in any literal sense,
claims legitimate authority.
Chapter 6 endeavors to rebut Raz’s claim that it is part of our concept of
authority that the content of an authoritative directive must be identifiable
without recourse to the dependent reasons it is supposed to reflect, balance,
and replace—a claim that implies that moral criteria of validity cannot be au-
thoritative. I argue that if, as Raz maintains, it is not possible for officials to
be systematically mistaken about the legal concepts that are defined by their
adopted practices, there are empirical reasons for thinking there is nothing in
our concepts of law and authority that would imply that moral criteria of va-
lidity are incompatible with the nature of authority. Insofar as our concept of
law is constructed by our linguistic and legal practices, the best account of our
concept of law includes the thesis that there can be moral criteria of validity.
Chapter 7 is concerned with the different ways in which law might be
thought to guide behavior. Scott Shapiro argues that (1) a judge cannot be
motivationally guided by both an inclusive rule of recognition and a norm
she applies in a dispute that is valid in virtue of moral merit; and (2) an in-
clusive rule of recognition is incapable of informing subjects of what their
non-official obligations are under valid law.8 I argue that both those claims
falsely presuppose that a rule must be capable of either guiding or informing
the behavior of persons it does not purport to govern. Finally, this chapter
argues that Raz’s “Arguments from Authority” likewise falsely presuppose that
the rule of recognition must be capable of informing ordinary citizens of what
they must do under valid legal norms; a rule need inform only those persons
whose behavior it governs of what they are required to do under that rule.
Chapter 8, the final chapter of the book, argues for the claim that inclu-
sive legal systems are conceptually possible. Chapter 8 constructs a model
of an institutional normative system that validates all and only mandatory
moral norms in a possible world that resembles ours in every causal respect
and shows that the system satisfies every condition that might plausibly be
thought to be conceptually necessary for the existence of a legal system,
thereby vindicating the Incorporation Thesis.
8
Scott Shapiro, “On Hart’s Way Out,” Legal Theory, vol. 4, no. 4 (December 1998), 469–507.
18
4. Who cares?
One might wonder why an entire book should be devoted to abstruse debates
about the concept of law. Many persons in the legal academy have developed
an aversion to the rarified concerns of conceptual jurisprudence. The sense is
that conceptual debates about the nature of law occur at such a high level of
abstraction that they have no practical relevance whatsoever and are simply
not worth pursuing.
The complaint is not that there are no good reasons to think conceptual
jurisprudence should be done; it is that there are good reasons to think that it
should not be done. This is a considerably stronger claim: to say that I have no
reason to do something does not obviously entail that I have a reason not to
do it. Skeptics argue that pursuing conceptual jurisprudence is problematic
because a conceptual theory of law tells us nothing about what our laws or
legal practices should be. Conceptual jurisprudence, they proclaim, should
not be done.
This critique rests on two different ideas. The first is that answers to con-
ceptual questions do not have any practical implications with respect to what
our laws and associated practices should be. The second is that a theory about
law should not be pursued unless it has implications with respect to what our
laws and practices should be. In the next subsection, I consider whether the
first claim is true. In the sequel, I consider whether the second is true.
4.1 Does a conceptual theory of law have any practical

implications?
The view that conceptual jurisprudence should not be done is not a new
one. Richard Posner devotes a substantial portion of the first of his 1995
Clarendon Law Lectures at Oxford University to arguing that conceptual
jurisprudence should not be done because it is “futile, distracting, and illus-
trative of the impoverishment of traditional legal theory.”9 He explains the
problem as follows:
I have nothing against philosophical speculation. But one would like it to have some
pay-off; something ought to turn on the answer to the question “What is law?” if the
question is to be worth asking by people who could use their time in other socially
valuable ways. Nothing does turn on it (LLT 3).
9
Richard Posner, Law and Legal Theory (Oxford: Clarendon Press, 1996). Hereinafter LLT.
Who cares? 19
Indeed, Posner goes so far as to argue that “the central task of analytic juris-
prudence is, or at least ought to be, not to answer the question ‘What is law?’
but to show that it should not be asked, because it only confuses matters”
(LLT 3).
Apart from the fact that he does not explain how conceptual theories “con-
fuses matters,” Posner’s argument is problematic. On the one hand, Posner
argues that conceptual jurisprudence has no practical implications about
what our laws and legal practices should be. On the other, he claims that
answers to conceptual questions would only confuse matters. The problem is
that if the only theoretical issues about law that matter are normative issues
concerning what our laws and legal practices should be, then a theory that has
no practical implications whatsoever simply could not create any confusion
with respect to those issues.
That said, there is something to Posner’s claim that answers to conceptual
questions have no practical implications whatsoever. Consider the conceptual
question of whether the Pope is properly characterized as a “bachelor.” At first
glance, the answer seems obvious: since the Pope is an unmarried adult male,
he is, by definition, a bachelor. Yet many people feel uneasy with this response
because the Pope has opted out of the marriage game. The thinking is that
the term “bachelor” applies only to unmarried men who are institutionally
or psychologically eligible for marriage. But notice that, either way, the an-
swer tells us nothing about how we should treat the Pope or how we should
treat bachelors. Conceptual analysis might go well beyond lexicography, but
it is still concerned with drawing out the implications of the social practices
that define the relevant concept-words; and it is hard to see how a conceptual
claim could have any practical implications other than those concerned with
how to use the relevant words.
One might think the situation is different with respect to the law. The idea
is that whether, say, an unjust norm, L, that has been properly enacted in
some legal system counts as valid law depends on which theory of law is cor-
rect. If, on the one hand, it is conceptually true, as NL2 asserts, that an unjust
norm cannot count as law, then L is not a law. If, on the other, it is conceptu-
ally true, as positivists assert, that an unjust norm can count as law, then L is
a law in that system despite being unjust. Surely, on this line of reasoning, it
makes a practical difference whether positivism or NL2 is true.
There are two problems with this reasoning. The first is that, at bottom, it
makes no real practical difference with respect to morally evaluating our legal
practices whether positivism or NL2 is true. What matters with respect to
the moral legitimacy of our practices—and these are the practices we should
be concerned with improving—is whether officials are treating L as law in
the sense of backing it up with coercive enforcement mechanisms, and not
20

whether L is really law according to some abstract general theory. If NL2 turns
out to be true, then the rule that the courts are applying might not properly
be characterized as “law,” but that is no consolation to the unfortunate de-
fendant who is being held liable under the rule. As far as considerations of
political morality are concerned, the legitimacy of our practices with respect
to, for example, enforcement of the Fugitive Slave Act does not turn at all
on whether it counts, on some conceptual theory, as being law. A character-
ization of a norm as law simply cannot carry that kind of normative weight.
The second problem is related to the first. If officials are recognizing, apply-
ing, and enforcing an unjust norm as law, then it counts as “law,” on the ordi-
nary social practices that define the content of the concept of law with which
this volume is concerned. It might be true, as a matter of political morality,
that it should not be law, but that tells us nothing about whether it is prop-
erly characterized, according to ordinary usage, as law; it tells us only that
the officials should not, as a matter of political morality, recognize, apply, or
enforce this norm. What matters morally with respect to the practices of our
courts and legislatures has everything to do with the content of the practices
and nothing to do with whether those practices are properly characterized as
“law.” Moral problems cannot be solved by simply recasting them in different
language.
Even so, it is important to note that resolving conceptual issues can some-
times help clarify notions that must be understood to address other kinds of
issues. Consider the normative issue of whether legal systems should permit
medical uses of marijuana. If one is sympathetic to the idea that criminal
prohibitions of marijuana use are too restrictive and the punishments unjus-
tifiably harsh, one might be tempted by the idea all medical uses should be
permitted; after all, at first blush, if purely recreational use is problematic in
a sense that requires legal prohibition, surely nothing accurately characterized
as a medical use could be.
But the issue is more complicated than might initially appear because the
notion of what counts as a medical use is not clear. Most people, for example,
use alcohol as a “social lubricant” to reduce anxiety in certain social situations
or to reduce anxiety to permit sleep after a long and stressful day. One can
plausibly argue that these uses are therapeutic and hence better characterized
as medical than as recreational. After all, the reasons for these uses resemble
the reasons for using certain prescription medications: Ambien, for example,
is prescribed to induce sleep while Valium and Xanax are prescribed for epi-
sodic anxiety of a kind that resembles the sort experienced by many people in
various social settings. These uses are clearly medical and resemble the ways
in which people use alcohol.
Who cares? 21
Certain purely recreational uses of alcohol would seem to be medically
problematic or even pathological. One recreational use might involve con-
suming alcohol for aesthetic reasons having to do with the taste of the rele-
vant beverage, such as wine. But another might involve consuming alcohol
for the purpose of enjoying the inebriating effects of it—i.e. to get drunk. The
problem with this use is that it suggests at least the beginning of an unhealthy
relationship with alcohol that can ultimately culminate in alcoholism. If so,
what appears to be a purely recreational use of a legal substance is far more
problematic than other uses that appear at first glance to be recreational but
turn out to be medical in character.
Alcohol, I suppose, is one thing and marijuana another, but the concep-
tual issues that arise in connection with uses of alcohol that seem most aptly
characterized as “medical” also arise in connection with the normative issue
of whether legal systems should permit medical use of marijuana. And the an-
swer is: it all depends on what one means by “medical use” and whether one
thinks that all uses that are properly characterized as “medical” are sufficiently
innocuous that they should be permitted by law. Answers to conceptual ques-
tions can thus produce epistemic benefits by facilitating conceptual clarity in
addressing other issues that have significant practical importance.
One might thus be tempted to argue that the pursuit of conceptual inquiry
with respect to the nature of law can be justified on the ground that it enables
us to think more clearly and rigorously about problems that have important
normative dimensions—even if the relevant conceptual theories do not imply
solutions to those problems. Proponents of conceptual analysis who feel the
need (and, for what it is worth, I feel no such need) to justify their interest by
reference to socially useful consequences typically do so by pointing to such
epistemic benefits.
The point of conceptual jurisprudence, on this line of reasoning, is not
that it entails solutions to important substantive problems of law or has im-
mediate practical implications that conduce to the betterment of humanity
by improving the moral quality of legal practice. Rather it is that conceptual
analysis helps us solve those problems by enabling us to formulate them more
clearly. The benefit here is not morally instrumental; rather it is epistemically
instrumental.
I am unpersuaded. Perhaps this is true of some conceptual issues like the
issue of what constitutes a medical use of an intoxicant, but this seems true
of comparatively few issues at best. We do not seem to need a plausible ac-
count of the nature of law to successfully address normative problems of law
any more than we need a plausible account of the nature of a number to
successfully address mathematical problems. In both cases, a pre-theoretical
2

understanding of the relevant concept is enough for us to fruitfully address
the relevant class of problems.
Nor do we need a theoretical understanding of the nature of law to do
everything we want to do with a legal system. It is true that we have to be
able to do a lot of things to have a working institutional system of something
that we typically characterize as “law.” Officials have to agree on a set of pro-
cedures for recognizing, applying, and enforcing law; citizens have to be able
to ascertain what their obligations are under the law for law to be efficacious
in guiding their behavior; and so on. But if our ability to characterize these
norms as “law” matters with respect to ensuring that they can efficaciously
regulate behavior, we do not need a conceptual theory of law to inform our
use of the term.
All we need to ensure this is that we largely agree on a pre-theoretic con-
ception of what law is. Law is, on this shared pre-theoretic conception, what
is enacted by a body that efficaciously functions as a rule-making body (which
we traditionally call a “legislature”) and applied by a body that efficaciously
functions as an adjudicative agency (which we traditionally call a “court”).
It does not matter what we call these functions or whether we are correct
under the right conceptual theory in characterizing these norms as law. A so-
cial group that handles the governance of its subjects in a certain way will
succeed in creating an efficacious institutional system of norm-governance—
regardless of whether the relevant norms are correctly characterized as “law”
under the best conceptual theory of law.
If the foregoing is correct, then a purely descriptive account of the con-
tent of a concept has neither practical implications with respect to how we
should structure our legal practices nor epistemic implications with respect
to how we should understand the practical problems that arise in connection
with how we should structure those practices. This part of Posner’s argument
against doing conceptual jurisprudence seems correct. The issue considered in
the next section is whether this is a good reason for thinking that conceptual
jurisprudence should not be done.
4.2 Should conceptual jurisprudence not be done?

Posner’s claim that conceptual jurisprudence should not be done because it
squanders time that could be used in “socially valuable ways” seems to en-
tail that it is wrong to spend time on doing conceptual jurisprudence. First,
wasting a socially valuable resource is usually thought to be wrong in some
sense (whether prudentially or morally); if doing conceptual jurisprudence
squanders a valuable resource, it would be wrong in the same way. Second,
Who cares? 23
as a general matter, to say that something should not be done according to
some set of standards is to say that doing it violates those standards and is
hence wrong according to those standards. Doing something that should not
be done might not amount to a particularly egregious wrong but doing some-
thing that should not be done is wrong under whatever standards dictate that
it should not be done. Posner’s complaint, then, is that it is wrong for theo-
rists to do conceptual jurisprudence.
This is a puzzling complaint. To begin, the sense in which conceptual juris-
prudence should not be done is not clear. On the most natural interpretation,
Posner is making a moral claim, but it seems implausible to think that it is
morally wrong to work in conceptual jurisprudence. If conceptual jurispru-
dents were arguing for wicked conclusions (e.g., that people should torture
babies), that would be a reason to think that they should not do what they are
doing. But the claim that law is a social artifact grounded in a conventional
rule of recognition practiced by officials seems pretty clearly to be a morally
innocuous one—which, of course, is partly explained by the fact that it is
purely descriptive.
If Posner is properly construed as making a moral claim, perhaps a more
plausible interpretation is that it would be morally better for legal theorists
to devote their energies to more practical issues. But the claim that one act is
morally better than another does not entail that the latter should not be done.
As a general matter, we do not say that one act is better than another unless
both acts are morally permissible. Someone who says giving to charity is mor-
ally better than committing murder is most plausibly construed as making
a philosophically nuanced (but not very funny) joke that calls attention to
vagueness with respect to the expression “morally better.” We generally use
that locution only to compare options that are morally permissible in the
sense that neither option is either prohibited or required. As far as ordinary
usage is concerned, to say that it is morally better that I give $25 to charity
than it is that I give $20 to charity neither asserts nor implies that I should
not give $20 to charity.10
But putting aside that concern, this construction of Posner’s claim remains
problematic. It is surely true that conceptual jurisprudents could find mor-
ally better things to do with their time, but so can most people who write in
the legal academy on other issues. Perhaps what other legal theorists do with
their research interests is morally better than what conceptual jurisprudents
do with theirs. But if so, the difference is not enough to make a focus on con-
ceptual jurisprudence worth complaining about. Indeed, if it is true that we
10
The situation changes if one adds the word “just.” It seems true that I should not give just $20
to charity.
24

should all attempt to work on issues likely to maximize moral value, the vast
majority of us working on the kinds of issues that fill the pages of law reviews
would do better to either give up our theoretical pursuits to focus on practical
pursuits involving law or to leave law entirely to study medicine or something
else more likely to produce social benefits than law.
None of this should be construed to disparage the efforts of those who
work in these areas. Every theorist, on my view, is entitled to pursue those
issues she finds most interesting. We are all entitled to considerable moral
latitude to do what we need to do to make our lives meaningful; life is, after
all, nasty, brutish, and short. For this reason, in the absence of special circum-
stances, it is as misguided to claim that legal theorists should do something
morally better with their lives as it is to claim that people in any other legiti-
mate occupation should do so.
But if the relevant sense of “should” is not moral, it is just not clear what
class of norms would support the claim that conceptual jurisprudence should
not be done. It makes little sense, as far as I can tell, to argue that the relevant
norms are epistemic; the idea would have to be that pursuing conceptual ju-
risprudence is likely to result in our having false beliefs about something—a
claim that is completely lacking in plausible support. Nor does it seem to
make much sense to think that the relevant norms have anything to do with
what is prudentially rational. As far as I can tell, Hart and Dworkin were
made famous by their contributions to conceptual jurisprudence; if so, that
work certainly conduced to their interests. If the norms of practical ration-
ality encompass some other kind of normativity, that needs to be specified in
order to begin to make out Posner’s claim.
Even so, it is helpful to consider how implausible remarks like Posner’s
would be if made about other disciplines. Though a great deal of work
in pure mathematics has been used to create technologies that improve
our lives, one of the most celebrated mathematical accomplishments of
recent years is not thought to have such applications. In 1995, Andrew
Wiles of Princeton University published a successful proof of Fermat’s Last
Theorem, which asserts that there are no positive integers x, y, z, and n >
2 such that the equation xn + yn = zn is true. I would be surprised if any of
the mathematicians who devoted countless hours to trying to prove this
theorem were motivated by an expectation it would produce a significant
improvement in the human condition—or, for that matter, even believed
it would have such applications.
Imagine the reaction in the mathematical community to someone whose
stature in mathematics is comparable to Posner’s stature in law publishing the
following remarks in response to the efforts to prove Fermat’s Last Theorem:
Who cares? 25
I have nothing against mathematical speculation. But one would like it to have some
pay-off; something ought to turn on the answer to the question “Is Fermat’s Last
Theorem true?” if the question is to be worth asking by people who could use their
time in other socially valuable ways. Nothing does turn on it.
The point here is not to deny that Wiles’s proof might turn out to have ben-
eficial practical applications; highly abstract mathematical results sometimes
have such applications. For example, Kurt Gödel proved the Incompleteness
Theorem long before there were computers, but some of the mathematics he
invented along the way wound up having many practical applications in the
field of computer science. Seemingly purely theoretical advances in abstract
mathematics have sometimes had unexpected technological applications.
Rather, the point is that there is no reason to think that the pursuit of
knowledge, mathematical or otherwise, is justified (or justifiably motivated)
only insofar as it has direct practical benefits apart from the joy of the en-
deavor and the knowledge it makes possible. Mathematicians recognize,
as they should, that the pursuit of knowledge need not be justified by in-
strumental concerns and hence that a problem need not have instrumental
implications to be worth discussing. Pursuit of knowledge, mathematical and
otherwise, can be justified by the value of knowing for its own sake.
It is not surprising that solutions to problems in conceptual jurisprudence
lack practical applications. Concept-words are used to group acts, events, and
entities into certain categories and do no more than pick out particular classes
of acts, events, and entities. It seems clear that we cannot solve any interesting
moral problems merely by altering our conceptual characterization of some
act, event, or entity. Whether hitting someone in the face is morally wrong
cannot turn on whether it is properly characterized as a gift; if it is properly
characterized as a gift, then we will have to rethink our moral views about the
permissibility of gift-giving from the ground up. What substantive normative
qualities any particular act or event has cannot turn on how it is grouped
through our linguistic practices with other acts or events. To decide those
questions, we need the help of propositions that express the relevant values.
But theoretical pursuits can be justified by other reasons than those having
to do with their practical applications. If we think that knowledge is both in-
trinsically and instrumentally valuable, a theoretical pursuit can be justified
by the intrinsic value of knowing the truth about the matter. Even if it turns
out that neither Fermat’s Last Theorem nor its proof have practical applica-
tions, we are surely made better off as a community—in some sense relevant
to human flourishing—in virtue of having a proof of the theorem because
(1) we know now that the theorem is true and (2) knowing the truth about a
matter, other things being equal, is valuable for its own sake. Likewise, getting
26

a handle on the nature of law, our most important normative institution, and
one that contributes significantly to our collective self-understanding, is also
valuable for its own sake.
Mathematics is not the only area of intellectual activity that is justified, at
least in part, by the intrinsic value of pursuing it. No sensible person thinks
that art must be justified by its practical applications; both the pursuit of art
and its products are valuable for their own sakes. It seems reasonable to think
that the same is true of philosophy, even if some areas of philosophical specu-
lation, such as substantive moral theory, also have instrumental benefits. The
pursuit of knowledge or truth on non-trivial matters (and law is certainly a
non-trivial matter) is an intrinsically valuable activity.
If one needs more to justify conceptual inquiry than this, then it might
help to note that a theoretical pursuit can also be justified on the ground
that it presents special intellectual difficulties. Part of the reason that Wiles’s
achievement has been so widely celebrated is that finding a proof for Fermat’s
Last Theorem was so difficult that it eluded mathematicians for hundreds
of years. This, of course, is no less true of non-intellectual pursuits: one can
justifiably feel proud of having scaled Mount Everest precisely because of the
daunting challenges it presents. Regardless of whether there is any practical
payoff to the pursuit of conceptual jurisprudence, this much cannot plau-
sibly be denied by anyone who understands the work of Hart, Dworkin, and
Raz: conceptual issues present intellectual difficulties unique among issues in
law and legal theory precisely because they are so much more abstract than
other issues.
This, of course, is not to suggest that the issues involved in conceptual
inquiry are more difficult than the issues involved in empirical or norma-
tive inquiry any more than the claim that climbing Mount Everest presents
a unique challenge suggests that climbing Mount Everest is more difficult
than hitting a 95-mph fastball. Such activities present challenges that are so
different that there is no obvious standard by which the difficulties can be
compared and ranked.
In this connection, it is worth considering how many other topics of abiding
interest in philosophy have no direct practical or technological applications.
As far as I can tell, there are few topics in metaphysics, one of philosophy’s
core areas of inquiry, that have direct practical or technological applications.
It is hard to see any practical implications or technological applications as-
sociated with solving the philosophical problems associated with the nature
of time, personal identity, universals and particulars, the ontology of the
universe, the nature of modality, aesthetics (or the nature of beauty), and
even the nature of the mind. Similarly, there are few topics in philosophy of
language, another core area of philosophical inquiry, that have obvious
Who cares? 27
practical or technological applications. As was true of questions of meta-
physics, it is hard to see how solutions for these philosophical puzzles could
have any practical implications or applications.
The justification for pursuing these highly abstract issues has to do with the
intrinsic value of both the process and the results of such pursuits. Indeed,
if one pursues philosophical research unconnected with normative issues of
morality, the motivation and justification must have something to do with
the intrinsic value of doing so. Given that we are all entitled to significant
moral latitude to decide what to devote the working hours of our lives to, it
seems a little puerile to single out conceptual jurisprudents for this kind of
criticism. Live and let live.
In any event, I hope that even readers who think it somewhat extravagant
to expend the kind of time and effort I have squandered on these issues find
something in what follows that engages their interest. I find much of interest
and beauty in the work of my colleagues and believe that I have learned much
from reading their work that transcends the admittedly narrow concerns of
conceptual jurisprudence. I hope that what I have produced in the following
chapters reflects both the value of what I have learned from them and the skill
with which they have imparted that value to me.
28
2
Rethinking the Traditional Interpretation
of Anti-Positivist Theories
Classical Natural Law Theory
and Dworkinian Interpretivism
This chapter challenges the traditional interpretation of classical natural law

theories and Dworkinian interpretivism. Insofar as these theories, construed
as rivals to positivism, are either straightforwardly false or must be grounded
in methodological assumptions that call into question the epistemological vi-
ability of the project of explicating the nature of law, they are best construed
as explicating a different concept of law than the one positivism seeks to ex-
plicate. The concept that positivism seeks to explicate is a purely descriptive
one that applies to any norm that has been recognized, applied, or enforced
in something that counts, on ordinary linguistic practices, as a legal system.
In contrast, the concept that classical natural law theories and interpretivism
seek to explicate is more aptly construed as grounded in the descriptive con-
cept that positivism seeks to explicate but also has evaluative content that
applies only to valid norms that can be characterized as law “in the fullest
sense.” Thus construed, these theories complement rather than rival posi-
tivism and are hence misleadingly characterized as “anti-positivist.”
1. Modest and immodest approaches

to conceptual analysis
Fundamental to the traditional project of analyzing a concept is the task of
identifying those properties jointly instantiated by all and only things that,
properly understood, fall under the concept. The underlying assumption is
that two things fall under a concept C if and only if they both instantiate
properties that distinguish things that are C from things that are not C. These
properties explain why something that falls under C is properly characterized
University Press.
30
30 Reinterpreting Anti-Positivist Theories

by the corresponding concept-term in the following sense: something is prop-
erly characterized as a C because it instantiates the properties that constitute
anything that has them as C.
A couple of examples should suffice to secure the point. If all and only
things that instantiate the properties of being unmarried and being a man
are properly characterized by the concept-term “bachelor,” then the proper-
ties of being unmarried and being a man distinguish things that are bachelors
from things that are not bachelors; if all and only things that are unmarried
men are, therefore, bachelors, then instantiating the properties of being un-
married and of being a man constitutes anything that instantiates them as
bachelors. Similarly, if all and only things that instantiate the compound
property of being a floating mass of water vapor are properly character-
ized by the concept-term “cloud,” then the compound property of being a
floating mass of water vapor distinguishes things that are clouds from things
that are not clouds; if all and only floating masses of water vapor are, there-
fore, clouds, then instantiating the compound property of being a floating
mass of water vapor constitutes anything that instantiates that property
as a cloud.
It is crucial to note that the relationship between having the properties that
are jointly instantiated by all and only things picked out by the concept-term
and being a referent of that concept-term is that of constitution—and not that
of causation. Being an unmarried man might have many causal effects—for
example, loneliness. But being an unmarried man does not cause a person to
be a bachelor; whatever it is that explains why that man is unmarried also
explains why he is a bachelor. Being unmarried constitutes a man as a bach-
elor in the sense that he falls under the concept of bachelor (i.e. is a bachelor)
wholly in virtue of being unmarried. Likewise, being a mass of floating water
vapor does not cause that mass to be a cloud; whatever it is that explains why
something is a mass of floating water vapor also explains why it is a cloud
Being a mass of floating water vapor constitutes something as a cloud in the
sense that it falls under the concept of cloud (i.e. is a cloud) wholly in virtue
of being a mass of floating water vapor.
Conceptual analysis, as traditionally conceived, is thus concerned with the
properties that distinguish things that fall under the relevant concept C from
things that do not fall under C and hence with the properties that consti-
tute something that instantiates them as C. The methodology is frequently
described as concerned with identifying those properties that are essential to
something in the following sense:
A property p is essential to C if and only if it is not possible for a thing to be C without

also instantiating p.
Approaches to conceptual analysis 31
A complete list of all the properties essential to C exhausts the nature of the
thing picked out by C in the following sense: something that is properly
characterized as C falls under that concept—i.e. is a C—only and wholly in
virtue of instantiating all of the essential properties of C. The nature of a
thing is nothing more than the essence of a thing, and the essence of a thing is
fully explained by a complete description of all its essential properties. If, for
example, the essential properties of bachelorhood are exhausted by the com-
pound property of being an unmarried man, then this compound property
fully defines the nature of bachelorhood.
The language of essences and essential properties is somewhat misleading
insofar as it suggests that a thing picked out by the relevant concept-term has
properties that define its nature independently of any social practices deter-
mining the application-conditions of the relevant concept-term; on this view,
an unmarried man would fall under the concept of bachelor regardless of
whether we use the term “bachelor” to pick out unmarried men. One might
take that position, but one need not do so. Given that describing the relevant
properties as “essential” could be construed as suggesting a substantive com-
mitment to that view, it is potentially misleading.
It is best to speak instead of properties that are conceptually necessary to
being in the reference class of the relevant concept-term. While the relevant
terminology should not assume that the nature of a thing is fully determined
by our conceptual practices, it should not rule that out either; whether or not
the relevant properties are essential in the sense described above is properly
conditioned by one’s choice of methodology. The term “conceptually neces-
sary” is thus preferable to the term “essential” insofar as it connotes agnosti-
cism with respect to the idea that the content of a concept and the nature of
the relevant thing can change over time.
As discussed in Chapter 1, conceptual analysis is traditionally regarded
as concerned with claims that can be justified a priori—i.e. those that can
be justified with no more empirical observation than is needed to learn the
meanings of the relevant terms expressing the claims. Insofar as no further
empirical observation than that needed to understand the meaning of a sen-
tence is needed to confirm or disconfirm it, the considerations that confirm
or disconfirm the truth of that claim are the same in every possible world.
Claims about the nature of a thing are therefore metaphysical and expressed
in terms of claims that are either necessarily or possibly true.
Conceived as a descriptive metaphysical enterprise, conceptual analysis
requires a different methodology from those deployed in normative and em-
pirical inquiries. To this end, Frank Jackson distinguishes a modest from an
immodest approach to conceptual analysis. Jackson believes that, although
metaphysics is about what the world is like, the relevant questions must be
32

framed in a language, and this gives rise to an important methodological
constraint:
[T]‌hus we need to attend to what the users of the language mean by the words they em-
ploy to ask their question. When bounty hunters go searching, they are searching for a
person and not a handbill. But they will not get very far if they fail to attend to the repre-
sentational properties of the handbill on the wanted person. Those properties give them
their target, or, if you like, define the subject of their search. Likewise, metaphysicians
will not get very far with questions like: Are there Ks? Are Ks nothing over and above Js?
and, Is the K way the world is fully determined by the J way the world is? in the absence
of some conception of what counts as a K, and what counts as a J. 1
Since the goal of the modest approach is hence “the elucidation of the pos-
sible situations covered by the words we use to ask our questions” (FEM 33),
it would be appropriate to begin from “serious opinion polls on people’s
responses to various cases” (FEM 36–37)—or from something like lexico-
graphical surveys.
Conceptual analysis, on the modest approach, purports to give us insight
into what the world is like as we structure it through the conceptual frame-
work we impose on it through our linguistic practices; that is, conceptual
analysis gives us insight into what the world is like as we view it through a lens
defined by the language we adopt to describe and make sense of our experi-
ence of the world. A modest approach to conceptual analysis thus purports to
give us insight into the nature of a thing as it is determined by our linguistic
practices.
The immodest approach to conceptual analysis can be defined as simply
the negation of the modest approach. Jackson does not say much by way of
directly explicating the immodest approach, illustrating it instead with a well-
known line of criticism of the idea that time constitutes a fourth dimension
on par with the three spatial dimensions represented by the x, y, and z axes
of a graph:
[M]‌any have taken this kind of consideration to show that four-dimensionalism qua
thesis about what our world is like is false. They, in effect, argue as follows:
Pr. 1 Different things (temporal parts or whatever) having different properties is
not change. (Conceptual claim illustrated in the case of temperatures)2
1
Frank Jackson, From Metaphysics to Ethics: A Defense of Conceptual Analysis (Oxford: Oxford
University Press, 1998), 30–31. Hereinafter FEM.
2
As Geach describes this case with temperatures: “[T]‌he variation of a poker’s temperature with
time would simply mean that there were different temperatures at different positions along the
poker’s time axis. But this, as McTaggart remarked, would no more be a change in temperature than
a variation of temperature along the poker’s length would be. Similarly for other sorts of change.”
P.T. Geach, “Some Problems about Time,” in Geach, Logic Matters (Oxford: Basil Blackwell, 1972),
Approaches to conceptual analysis 33
Pr. 2 Things change. (Moorean fact)
Conc. Four-dimensionalism is false. (Claim about the nature of our world)
We now have an example of conceptual analysis in what I call its immodest role.
For it is being given a major role in an argument concerning what the world is like
(FEM 42–43).
The idea is that four-dimensionalism does not accurately describe our world
as it is independent of our conceptual practices because it is inconsistent with
the mind-independent fact that things change. This criticism presupposes an
immodest approach to metaphysical analysis insofar as it purports to identify
features of the world as it really is independent of the conceptual framework
we impose on the world to understand it.
The two approaches, then, differ with respect to the proper object of study.
The modest approach (MCA) seeks to understand the nature of a thing as
it is defined by the conceptual framework we impose on the world through
shared linguistic practices. The immodest approach (ICA), in contrast, seeks
to understand the nature of a thing as it really is independent of any such
conceptual framework that we impose on the world.3
The methodologies of MCA and ICA are both partly empirical in character
insofar as each adopts “ordinary” intuitions as the starting point of concep-
tual analysis. Which intuitions are ordinary is a matter of which intuitions
are commonly shared among people in the relevant population; and that is an
empirically observable feature of the world.
302–18, 304. The idea is that a difference in temperature between one point on the poker and an-
other is not properly characterized as a change in temperature; it is just a difference between the
temperature at one point and the temperature at another point, such as would be the case if one
end of the poker had been placed in fire and the other end in a freezer. If, however, reality is four-
dimensional in the sense that time is a fourth dimension on par with the three dimensions of space,
then there cannot be any changes in temperature from one moment in time to the next. What
appears to us in a three-dimensional world as a change in the temperature of the poker would simply
be a variation of the temperature at “different positions along the poker’s time axis.”
3
Here is another way to understand the distinction. Kant distinguished between things
as they appear to us mediated through the categories of space and time through which we pro-
cess all sense perceptions and things as they are in themselves. Immanuel Kant, Critique of Pure
Reason (Cambridge: Cambridge University Press, 1998) in Paul Guyer and Allen Wood (eds.), The
Cambridge Edition of the Works of Immanuel Kant (Cambridge: Cambridge University Press, 1992–
98). The categories with which Kant is concerned, of course, are not concepts picked out by some
concept-term; rather, they involve the innate ability of the mind to structure experiences so that
the world appears to us as having spatial dimensions and temporal qualities. Even so, it would be
helpful throughout this chapter to express some of the ideas in the Kantian distinction between the
world as it is and the world as it appears to us. To put it in Kantian terms, the immodest approach
to conceptual analysis purports to give us knowledge of things as they are in themselves independent
of the conceptual frameworks we use to make sense of the world.
34

If the two approaches agree on starting from ordinary intuitions, they disa-
gree on why we should start there. Insofar as MCA seeks to reconcile conceptual
theories with ordinary talk, it requires that we consider ordinary talk as a touch-
stone for evaluating theories about the nature of the relevant thing, and ordinary
talk reflects ordinary intuitions conditioned by the shared social conventions for
using the corresponding term. Accordingly, MCA begins from an understanding
of the lexical meanings of words; these meanings form the starting point for an
investigation into what deeper philosophical commitments these ordinary prac-
tices imply. In contrast, there is nothing in ICA that would logically require that
we take ordinary talk as a starting point and hence nothing that grounds ICA
in the shared practices defining the lexical meanings of the relevant concept-
terms; insofar as we seek to understand what the world is like independent of the
concepts picked out by our linguistic practices, there is no reason to think that
recourse to ordinary intuitions is even helpful, much less necessary. Indeed, there
is nothing in the idea that conceptual analysis seeks to articulate the nature of
things as they are independent of our thoughts and practices that even gestures
in the direction of starting from the lexical meanings of the relevant concept-
terms or the intuitions that the corresponding social practices condition in us.
Not surprisingly, the two approaches differ according to how much epi-
stemic weight should be assigned to the corresponding intuitions. MCA takes
these intuitions as providing the ultimate standard for evaluating a concep-
tual theory because MCA assumes that the object of conceptual analysis is
to uncover the nature of the world as we define it through the conceptual
frameworks that our linguistic practices impose on it. ICA, in contrast, takes
ordinary intuitions to be nothing more than a guide to understanding the na-
ture of a thing; our intuitions or ordinary talk are not conceived as defining
the nature of the thing.
Accordingly, there are certain kinds of errors that would warrant the rejec-
tion of a conceptual theory deploying MCA that would not obviously warrant
the rejection of a conceptual theory deploying ICA. Insofar as our ordinary
talk defines the nature of a thing, our conceptual theory of the thing must
harmonize with ordinary talk; failure to do so is a fatal error for a conceptual
theory under MCA. Insofar as ICA presupposes that ordinary talk does not
define the nature of some thing T, any or all of our ordinary intuitions about
T can be false.
To describe this difference using a term coined by J.L. Mackie, MCA
cannot, while ICA can, result in an error theory—i.e. a theory that purports to
show that everyday thought, or what has the status of being the “folk theory”
in some area, is so deeply and widely in error as to warrant its rejection.4 An
4
J.L. Mackie, Ethics: Inventing Right and Wrong (London: Penguin Books, 1977), 35.
Two concepts of law 35
error theory about a thing, then, is a theory that implies that our folk theory
about that thing is systematically in error and should be rejected. Insofar as
our folk theory comprises our ordinary intuitions, an error theory implies
that the relevant ordinary intuitions should also be rejected. Since, in effect,
MCA makes the folk theory the touchstone for evaluating conceptual theo-
ries, MCA cannot result in an error theory. Since ICA assumes that ordinary
intuitions are merely a guide and not a touchstone, ICA can result in an error
theory.
2. Two concepts of law

There are two different concepts of law that might be the focus of a conceptual
theory—a purely descriptive concept and an evaluative concept. The former
is purely descriptive in the sense that whether or not something counts as law
does not depend on whether it satisfies any moral standards that are not in-
corporated into the criteria of validity. The evaluative concept is grounded in
the descriptive concept but also has evaluative content. Whether something
that counts as law on the purely descriptive concept also counts as law on the
evaluative concept depends on whether it is a good example of its kind ac-
cording to certain moral standards that express values that should be satisfied
by anything that antecedently counts as law on the descriptive concept. The
evaluative concept thus expresses the idea of what counts as law in its fullest
or ideal sense—i.e. what the law should be given some understanding of law’s
point or aspirations.
It might be thought that this account of the descriptive concept of law
begs the question against the traditional interpretation of classical natural
law theory, but this is incorrect. Whether a norm counts as law on the de-
scriptive concept in some system might depend on whether it satisfies certain
moral norms. If so, then those moral norms are part of the system’s criteria of
validity and are hence, so to speak, internal to the system in this sense: they
determine what counts as law in that particular system but not what counts as
good law or law in its fullest sense. If it is true that every conceptually possible
legal system incorporates these moral norms as criteria of validity, then those
norms are also internal to the existence conditions for law as defined by the
descriptive concept.
Notice that the claim that a legal norm is consistent with certain moral
requirements does not imply that it is a morally good law or that it is hence
law in its fullest sense. For example, there might be laws imposing taxes
on citizens that are consistent with some set of moral standards of fairness
36

incorporated into the system’s criteria of validity but are not morally ideal ac-
cording to other standards. The claim that something satisfies a set of moral
standards does not imply that it is morally good; my not having ever com-
mitted murder satisfies the moral prohibition against murder, but it would be
odd to characterize my having refrained from murder as morally good. If the
traditional interpretation of natural law theory is the correct account of the
descriptive concept, then there are standards of morality internal to law be-
cause incorporated into the criteria of validity of every conceptually possible
legal system. But there can still be additional moral standards that are external
to a system’s criteria of validity and to the existence conditions of law in the
sense that they distinguish law that is morally ideal from law that is morally
acceptable but not morally ideal.
Both concepts are grounded in convergent patterns of ordinary lexical
usage. The purely descriptive concept might be the more familiar one, as it is
the one more commonly deployed in questions about what the law requires
or permits. For example, if I seek a lawyer’s advice on whether I may take a
certain deduction in calculating my tax liability, my question deploys a de-
scriptive concept grounded in convergent patterns of ordinary lexical usage.
What I care about on this usage is what is required by the tax statutes that are
enforced by the IRS because I want to avoid having to serve a prison sentence
for tax fraud. Knowing which tax statutes count as law “in the fullest sense”
will not help me ascertain how much I need to pay to avoid such liability.
Indeed, paying too much attention to which tax statutes count as law in the
fullest sense might tempt me to pay less than is required by all the tax statutes
taken together and hence to do something that would subject me to exactly
the liability I wish to avoid. When I am attempting to plan my behavior to
avoid criminal or civil liability, the relevant usage of “law” is the one that
picks out the purely descriptive concept grounded in convergent patterns of
ordinary usage.
The evaluative concept is also grounded in convergent patterns of ordinary
lexical usage. Opponents of abortion rights argue that the U.S. Constitution
does not define a right to reproductive privacy that requires states to legalize
abortion during the first trimester. As the claim is sometimes put, “there is no
constitutional right to abortion”—which, properly fleshed out, expresses the
claim that there is nothing in the law defined by the Constitution that entails
a legal right to abortion. If understood as employing a purely descriptive
usage of the concept-term “law” that is conventionally defined, as all words
are, by certain social practices, the claim is simply confused. If understood
as employing an evaluative usage of “law” that seeks to distinguish what is
just or legitimate from what is unjust or illegitimate, the claim might be false
but at least it expresses a plausible position: thus construed, the claim is that,
while technically (or descriptively) “law,” the holding that established such a
right is illegitimate because inconsistent with how the relevant provisions of
the Constitution should, as a matter of political morality, be interpreted.
The concept of law resembles the concept of art in this respect. As is true
of law, there are two concepts of art, one descriptive and the other evaluative.
According to the institutional conception of the purely descriptive concept of
art, the term “art” is properly applied to any artifact that is presented to a cer-
tain community for the purpose of creating an aesthetic experience. On this
usage, any painting displayed in a museum counts as a piece of art regardless
of quality. According to the evaluative usage, the term “art” is properly ap-
plied to those objects that count as art in the purely descriptive sense but also
satisfy certain standards of aesthetic quality.
It is not uncommon for people to alternate between the two concepts of art
depending on what the circumstances dictate as appropriate. If I am moved
by a Caravaggio and exclaim “Wow, that’s art,” it should be clear that, absent
unusual circumstances, I am not using the term to apply the purely descrip-
tive concept. It is unnecessary to say of a Caravaggio in a museum that it is art
in this sense; a Caravaggio is so paradigmatically art in the descriptive sense
that an exclamation of its status as such says nothing anyone would have any
reason to say. What I am claiming is that it is an exemplary piece of art or is
art in the fullest sense—what art is at its aesthetic best.
In contrast, if I come across a pair of eyeglasses on the floor in an empty
corner of a museum exhibit of conceptual art and ask an employee whether
it is art, I am deploying the descriptive concept. In particular, I might be
wondering whether it is part of the exhibit, or I might be wondering whether
someone has deliberately left it there to make an ironic comment on the
quality of the exhibited art. Either way, I am asking whether the eyeglasses are
part of the exhibit and hence constitute a piece of art in the purely descriptive
sense of the term.
It can sometimes be difficult to determine whether a concept-term is
being used in a purely descriptive sense or whether it is being used in an
evaluative sense. If, for example, I ask someone who does not know what
motivates the question whether the pair of eyeglasses is a piece of art, she
might take the question to be about the applicability of the evaluative con-
cept. Thus construed, my question would not be whether the eyeglasses piece
falls within the reference class of the descriptive concept-term; it would be
whether the piece is art in the fullest sense and hence falls within the refer-
ence class of the evaluative concept-term. Alternatively, she might construe
the question to be about the applicability of the descriptive concept. Thus
construed, the question would be whether the pair of eyeglasses is art in the
descriptive sense.
38

Complicating the interpretive issue is that the content of the evaluative
concept is not always clear. Suppose that I notice a card posted over the eye-
glasses with the name of an artist and the title “Eyeglasses” on it and ask
someone, “Is this art or bullshit?” Construed as a question employing an
evaluative concept, I am using the term to deploy a different evaluative con-
cept from the one that I deployed when exclaiming “Wow, that’s art” of a
Caravaggio painting. In that example, I was expressing the judgment that the
painting was art in the fullest sense of the term—a model for what art is at its
best. In the case where I ask of the eyeglasses whether the piece is art or bull-
shit, I am asking whether it meets some minimum standard for counting as
art on any reasonable conception of an evaluative sense of the term. In both
cases, I am deploying an evaluative concept, but one picks out what is art in
a normatively fullest sense while the other picks out what is art in the norma-
tively minimal sense.
One lesson that emerges from this discussion is that it is important to be
clear about which concept is being explicated by a theory of any concept-term
that is ambiguous between a descriptive use and an evaluative use. It matters
a great deal whether the relevant usage picks out a purely descriptive concept
or whether it picks out a concept that is grounded in the descriptive concept
but also has evaluative content. A failure to be clear about which concept is
the relevant one can lead to systematic confusion about the character of the
phenomenon under study.
Another lesson that emerges here is that there is no necessary inconsistency
between a theory of a descriptive concept of a thing and a theory of an eval-
uative concept of the same thing that give different answers to the question
of whether some particular item falls under the relevant concept. There is no
necessary inconsistency between a conceptual theory that implies that some
street mural is art in a descriptive sense and a conceptual theory that implies
the mural is not art in an evaluative sense. They might be inconsistent, but
they need not be. If the two theories agree on the content of the descriptive
content of art that is presupposed by the account of the evaluative concept,
there is no inconsistency between the two judgments. If the two theories dis-
agree on the content of the relevant descriptive content and the evaluative
theory denies the status of art to the mural on the ground that it is not art on
the underlying descriptive concept, then the two judgments are inconsistent.
One could, I suppose, ask whether the evaluative concept better cap-
tures the real nature of the thing under study, but it is not clear how one
could answer this question. If we consider this question in connection with
the concepts of art, there are two ways to construe this question, and both
are problematic. First, one might construe the question as concerned with
which concept better picks out what art really is, as defined independently of
anything we might think about it or independently of the conceptual frame-
work we impose on the world through our linguistic practices regarding
such objects. But it is not clear either what this question is really asking or
what kind of evidence to which we have access that would help to answer it.
Perhaps, the question is concerned with what an omniscient being would re-
gard as “art.” If this is what is intended by the question, it is simply not clear
how we could go about answering it. Apart from having it directly answered
by such a being, which I am guessing is not likely to happen soon, it is just
not clear what we could do to determine which concept better captures the
nature of a thing that is determined in a way that is completely independent
of our thoughts, preferences, and linguistic practices. If asking such a ques-
tion is coherent, there is no real point to doing so.
Second, one might construe the question as concerned with which concept
better picks out what we intend to pick out with our conceptual practices
concerning the use of the term “art.” The problem, however, is that we have
two sets of conceptual practices that are contrived to pick out different, al-
beit closely related, phenomena. The evaluative concept is generally deployed
to pick out a proper subset of objects to which the descriptive concept-term
applies. No competent speaker of English can dispute that a Caravaggio is
a piece of art regardless of whether we are using the descriptive or evalua-
tive concept. In contrast, competent speakers of English could disagree on
whether a ready-made by Duchamp is “really” art under either the descriptive
or evaluative concepts; I do not think that there are any real issues there, but
the question can coherently be asked. Either way, there is no one set of phe-
nomena that would tell us that one concept better picks out what we intend
to pick out by “art” than the other because we use the term differently to pick
out two different sets of phenomena.
The question of which use of “art” better captures the real nature of art
makes no more sense than the question of which use of “bank”—the one
referring to financial institutions or the one referring to the natural barri-
ers defining the boundaries of a river—better captures what a bank really is.
There is no single right answer to that question because, for all practical pur-
poses, there are two logically independent concepts of bank that are picked
out by two different usage patterns of the concept-term “bank.” The same is
true of the question of which use of “art” better captures the real nature of art.
There is no single right answer to that question because the two concepts of
art are likewise logically independent of one another. To say that something
is descriptively art does not imply that it is a good piece of art. Conversely,
and somewhat surprisingly, to say that something is art on the evaluative
usage does not imply that it is art on the descriptive usage; one might charac-
terize an exceptionally clever criminal act as “art” despite the fact that it is not
40

presented to an audience for the purpose of producing an enjoyable aesthetic
experience.
The fact that we use the same word to refer to phenomena that might be re-
lated in some way does not imply that one concept better captures the nature
of the thing picked out than the other. We could just as well use a different
term to express the evaluative sense of the term “art” from the one that we use
to express the descriptive sense of the term. Although doing so might result
in a loss of some non-cognitive expressive force (which would not be true if
we used different words to express the two senses of the word “bank”), such
a change would result in no significant difference in the cognitive content of
the term.
For this reason, conceptual theories that purport to explain a purely de-
scriptive use of a term do not engage conceptual theories that purport to ex-
plain an evaluative use of the same term—except in unusual cases that likely
involve some confusion about the relevant notions. Insofar as a theory of an
evaluative concept of a thing is properly grounded in assumptions that pick
out the correct descriptive concept of a thing, there could be no logical con-
flict between them. Any criticism of a theory of an evaluative concept would
have to be directed at the normative standards that express the evaluative con-
tent of the concept. Conversely, it makes even less sense to criticize a theory
of the relevant descriptive concept on the ground that it does not jibe with
the normative content of the evaluative concept because a purely descriptive
concept has no normative content.
In articulating a conceptual theory of law, one must therefore be clear
about which concept the theory is intended to explicate. On the one hand,
an explication of the purely descriptive concept of law will not directly engage
an explication of the evaluative concept of law. On the other, an explication
of the evaluative concept of law will not directly engage an explication of the
descriptive concept of law. If this is correct, then it is a mistake to think that
a conceptual theory of the evaluative concept is an object-level rival to a con-
ceptual theory of the descriptive concept.
3. Four possible interpretations of a conceptual

theory of law
The discussions in the last two sections call attention to a couple of inter-
pretive difficulties in assessing conceptual theories of law. The first difficulty
has to do with which usage or concept the relevant conceptual theory pur-
ports to explain. Since there are two concepts of law that a theory might
Legal positivism as assuming MCA 41
purport to explicate, there are two possible ways to interpret a conceptual
theory of law: one interprets the theory as concerned to explicate the descrip-
tive concept of law while the other interprets it as concerned to explicate the
evaluative concept of law.
The second difficulty has to do with the character of what determines the
content of the relevant concept of law. A conceptual theory of law can be
interpreted as presupposing MCA and hence as purporting to explicate the
relevant concept as it is determined by our shared linguistic and legal prac-
tices; thus construed, the theory purports to explicate a concept of law that is
ours in the sense that our linguistic and legal practices determine its content.
Alternatively, the theory can be interpreted as presupposing ICA and hence
as purporting to explicate the nature of law as it is independent of any of our
linguistic and legal practices; thus construed, the theory purports to tell us
something about the extra-linguistic world as it pertains to the nature of law
as it really is.
Accordingly, there are four possible interpretations of a conceptual theory
of law: a conceptual theory can be construed (1) as an explication of the
content of the descriptive concept of law as it is determined by our shared
linguistic and legal practices and hence as presupposing MCA; (2) as an ex-
plication of the content of the descriptive concept of law as it is determined
by objective features of the world independently of our shared linguistic and
legal practices and hence as presupposing ICA; (3) as an explication of the
content of the evaluative concept of law as it is determined by our shared
linguistic and legal practices and hence as presupposing MCA; and (4) as an
explication of the content of the evaluative concept of law as it is determined
by objective features of the world independently of our shared linguistic and
legal practices and hence as presupposing ICA. I argue below that positivism
is best construed according to option (1) and that classical natural law theory
and Dworkin’s interpretivism are best construed according to option (3).
4. Legal positivism as assuming MCA to explain

the descriptive concept of law
Positivism purports to explain the nature of law as such—and this requires an
account of the existence conditions for legal systems, as well as for the valid
legal norms that furnish a legal system and help bring it into existence. Since
positivism is a conceptual theory of law, it is a metaphysical theory insofar as
it purports to explain the nature of law by (1) identifying conceptually neces-
sary properties that (2) constitute any norm or system as one of law. Insofar
42

as positivism does not purport to tell us anything about which properties a
norm must have to be a law that is good, it purports to explain the purely
descriptive concept of law.
H.L.A. Hart provides such an account of the existence conditions for legal
norms and legal systems.5 First, on Hart’s view, there is a legal system in L if
and only if (1) there is a rule of recognition defining the criteria of validity
that is accepted and practiced by those who serve as officials in L; and (2) the
behavior of people in L generally conforms to the norms valid under the rule
of recognition. Second, a norm n is a law in a legal system L if and only if
there exists a legal system L in which either (1) n is one of the recognition
norms practiced in L or (2) n is valid under the criteria of validity defined by
the recognition norms practiced in L.
Hart understands himself to be explicating descriptive concepts picked out
by ordinary linguistic and legal practices. As Hart describes his project, it is
to identify those features of law that define its nature as we ordinarily un-
derstand it. As he puts it, “The starting-point for this clarificatory task is the
widespread common knowledge of the salient features of a modern municipal
legal system which . . . I attribute to any educated man” (CL 239). Insofar
as the common knowledge that comprises the starting point for conceptual
theorizing about law is grounded in an ordinary understanding of how the
concept-term “law” is used, the descriptive concept of law that Hart purports
to explicate is the one grounded in social practices that are both linguistic and
legal in character.
Joseph Raz is more explicit in affirming that the descriptive concept his
theory is concerned to explicate is grounded in our linguistic and legal prac-
tices. On his view, officials cannot be systematically confused about the na-
ture of authority because it is their claims, conceptions, and practices that
construct our concept of authority as it functions in legal practice:
[W]‌hile [legal officials and institutions] can be occasionally [confused,] they cannot
be systematically confused. For given the centrality of legal institutions in our struc-
tures of authority, their claims and conceptions are formed by and contribute to our
concept of authority. It is what it is in part as a result of the claims and conceptions
of legal institutions. (ALM 217; emphasis added)
The concept of authority that is therefore the focus of his theoretical
inquiries—i.e. the one that matters—is a concept of authority that is ours
because it is the one presupposed, and hence constructed, by our linguistic
and legal practices.
5
H.L.A. Hart, The Concept of Law 3rd Ed. (Oxford: Oxford University Press, 2012).
Hereinafter CL.
The traditional interpretation of natural law theory 43
Contemporary positivism is thus properly understood as presupposing
MCA to explicate the purely descriptive concept of law. MCA attempts to
explicate the nature of a thing as it is determined by ordinary patterns of
linguistic usage with respect to characterizing things of that kind and hence
takes the relevant social practices that pick out those things and condition our
ordinary intuitions about them as the standard for understanding it. While
ICA is also grounded in intuitions that are ordinary in some sense, posi-
tivism begins from those intuitions because they express the ordinary conven-
tions we have adopted for using the terms that construct our concepts of law
and authority. On a modest approach, those conventions, which inform our
common intuitions, provide the touchstone for evaluating a theory of the
concept of law.
What this means, given the arguments of the last section, is that legal pos-
itivism cannot result in an error theory since it presupposes MCA. Insofar
as our core linguistic and legal practices determine or construct the content
of the relevant concepts, we cannot—absent extraordinary confusion about
what our core linguistic and legal practices are—be systematically mistaken
in understanding law. Thus, the adoption of MCA by positivists presupposes
that the intuitions informed by these core practices are largely correct and
hence that an explication of the relevant concepts cannot result in an error
theory of law.
5. The traditional interpretation of natural law theory

construed as a rival to positivism
Classical natural law theory is traditionally construed as inconsistent with
legal positivism and hence as purporting to explicate the same concept of
law that positivism purports to explicate. The content of the descriptive
concept of law that positivism purports to explicate is determined by con-
vergent patterns of linguistic usage that are informed by practices under-
stood to be paradigmatically legal; positivism thus purports to explicate the
nature of law as it is determined by the core legal and linguistic practices
that define the application-conditions for using our descriptive concept-
term “law.” The traditional interpretation of natural law theory construes it
as a rival to positivism and hence as explicating the same concept of law
as positivism. The traditional interpretation of classical natural law theory,
then, presupposes that it purports to explicate a descriptive concept of law
that is picked out by our core legal and linguistic practices and is hence our
concept of law.
4

The problem with the traditional interpretation is that it implies that un-
just norms can never be properly characterized as “law,” as we use the term in
its descriptive sense—a claim that is straightforwardly inconsistent with the
ordinary practices that construct the content of the descriptive concept that
positivism purports to explicate. It is clear, to begin, that there can be unjust
laws and legal systems as far as our ordinary linguistic practices are concerned.
Consider, again, a representative lexical definition of the concept-term “law”:
law (noun): the system of rules that a particular country or community recognizes as
regulating the actions of its members and may enforce by the imposition of penalties.6
There is nothing in this definition that even remotely suggests that unjust
laws are impossible. If the relevant community recognizes and enforces rules
that are unjust as part of “the system of rules recognized and enforced [as
law] by [that] particular community,” then the law can include rules that are
unjust. Indeed, if this were not true, then the claim that there can be unjust
laws or legal systems would be, from the standpoint of the lexical definition,
as much a contradiction in terms as the claim that there can be married bach-
elors. Whatever else might be true of the claim that there can be unjust laws,
it is quite clearly not a contradiction in terms.
The definition of “law” on this point should be contrasted with the defini-
tion of “bachelor,” which is as follows:
bachelor (noun): a man who is not and has never been married.7
It should be clear that the definition of the concept-term “bachelor” explicitly
precludes the conceptual possibility of a bachelor who is married. Unlike the
claim that there can be just laws, the claim that there can be married bach-
elors is quite clearly a contradiction in terms.
Further, it is clear, as far as ordinary linguistic usage is concerned, that
laws and legal systems can be illegitimate for a variety of reasons that in-
clude the unjustness of the norms. The legal system of apartheid in South
Africa is universally condemned among right-minded people as illegitimate
because of the wicked quality of its valid norms, which mandated pervasive
racial segregation in that country; those unjust norms, as a matter of ordinary
linguistic usage, were laws. If the linguistic conventions defining the lexical
definition of “law” are the touchstone for ordinary usage, there can clearly be
unjust laws.
6
“Law,” Oxford Online Dictionary; available at: https://en.oxforddictionaries.com/definition/
law.
7
“Bachelor,” Oxford Online Dictionary; available at: https://en.oxforddictionaries.com/defini-
tion/bachelor.
Ordinary legal practice in many legal systems also straightforwardly presup-
poses that there can be unjust laws. What is properly promulgated as law in
these legal systems is recognized and enforced by courts as laws that legally jus-
tify imposition of sanctions on subjects who violate them. A court has the legal
authority (i.e. authority conferred by the convergent recognition practices of
officials) to hold a subject liable under a properly promulgated norm regardless
of whether the norm is unjust or is simply considered unjust. This of course does
not mean that the court must do so or will do so; a court might also have legal
authority to decline to enforce—or even to change—an unjust norm. The point,
however, that it is clearly false as an empirically verifiable matter of ordinary legal
practice that unjust norms never count as law for purposes of the practices that
determine the content of the concept of law that positivists take themselves to
be explicating.
Courts in Anglo-American legal systems explicitly concede that there can be
unjust or morally problematic laws. U.S. courts acknowledge that they lack au-
thority to revise properly promulgated laws on the ground that they are morally
problematic; in cases where unjust norms have been properly enacted by the leg-
islature and violate no constitutional protections, courts are—as a matter of what
is recognized by the courts as settled law—required to defer to the judgment of
the legislators. Consider the following remarks from the U.S. Supreme Court:
Whether embodied in the Fourteenth Amendment or inferred from the Fifth, equal
protection is not a license for courts to judge the wisdom, fairness, or logic of legis-
lative choices . . . . [A]‌legislative choice is not subject to courtroom factfinding and
may be based on rational speculation unsupported by evidence or empirical data . . . .
Only by faithful adherence to this guiding principle of judicial review of legislation
is it possible to preserve to the legislative branch its rightful independence and its
ability to function.8
Insofar as this principle requires federal courts, other things being equal, to defer
to the judgment of the legislature in cases regardless of whether the court believes
it is “fair” or “wise” to do so, it acknowledges that there can be unjust federal laws.9
8
FCC v. Beach Communications, Inc., 508 U.S. 307 (1993). Similarly, in Williamson v. Lee
Optical, 348 U.S. 483 (1955), the Court stated that “the day is gone when this Court uses the Due
Process Clause to strike down state laws, regulatory of business and industrial conditions, because
they may be unwise, improvident, or out of harmony with a particular school of thought.” The
Court indicated that in cases where a challenged statute implicates no fundamental constitutional
rights it must uphold the law unless there is no conceivable reason for enacting the bill. In such cases,
the only ground for striking down the law is that it is utterly irrational. Even if the challenged law is
morally problematic, it must be upheld if a coherent rationale can be given for its enactment. I am
grateful to Todd Shaw, Ashley Robles, and David Brink for these and other examples.
9
Similar comments can be found in judicial opinions in cases from the U.K. In Madzimbamuto
v. Lardner-Burke [1969] 1 AC 645, 723, the court argued: “It is often said that it would be
46

The same is true of state law. State courts not uncommonly indicate that
they lack authority to modify statutory enactments on the ground that they
are unjust or otherwise ill-advised. As the Washington Supreme Court unam-
biguously explains:
[H]‌owever much members of this court may think that a statute should be rewritten,
it is imperative that we not rewrite statutes to express what we think the law should
be. We simply have no such authority.10
The claim by the court that it lacks authority to modify problematic statutes
implicitly acknowledges the possibility that a statute with the status of law
might be unjust. Whether unjust or unwise, courts in Washington state are
legally bound to defer to the judgment of the legislature.
Indeed, the rules of civil procedure governing both the federal system and
many state systems preclude bringing lawsuits on purely moral grounds.
Federal Rule of Civil Procedure 11(b)(2) provides as follows:
By presenting to the court a pleading, written motion, or other paper—whether
by signing, filing, submitting, or later advocating it—an attorney or unrepresented
party certifies that to the best of the person’s knowledge, information, and belief,
formed after an inquiry reasonable under the circumstances . . . the claims, defenses,
and other legal contentions are warranted by existing law or by a nonfrivolous ar-
gument for extending, modifying, or reversing existing law or for establishing
new law.
Washington Court Rule 11(a)(2) similarly provides that an attorney’s signa-
ture on a pleading constitutes an affirmation that the action “is warranted by
existing law or a good faith argument for the extension, modification, or re-
versal of existing law or the establishment of new law.” While these rules permit
grounding an action to change an existing law in moral considerations,11 this
unconstitutional for the UK Parliament to do certain things, meaning that the moral, political and
other reasons against doing them are so strong that most people would regard it as highly improper if
Parliament did these things. But that does not mean that it is beyond the power of Parliament to do
such things. If Parliament chose to do any of them the courts could not hold the Act of Parliament
invalid.” See also R (Countryside Alliance) v. Attorney General [2008] 1 AC 719, para. 45. (“The dem-
ocratic process is liable to be subverted if, on a question of moral and political judgment, opponents
of the Act achieve through the courts what they could not achieve in Parliament.”) I am grateful to
Kenneth Ehrenberg for these examples.
10
State v. Groom, 133 Wash.2d 679, 689, 947 P.2d 240 (1997).
11
There is no distinction that matters here between changing existing law and creating new law.
It is true that the rules permit grounding an action to “establish[] new law” in moral argument, the
act of creating a new law is simply a certain kind of change in existing law. A new law that imputes
a duty to subjects that they did not have because there was no legal norm imputing such a duty has
the effect of changing existing law to make impermissible behaviors that were formerly permissible
under the law.
presupposes that the relevant rule both has the status of law and is arguably
and hence possibly unjust. If what is regarded by ordinary citizens and legal
practitioners as paradigmatic legal practice in legal systems like that of the
U.S. is taken as defining the relevant concept of law, it is clear that there can
be unjust laws.
Note that the problem with interpreting classical natural theory as a rival
theory to positivism is worse than that it is just false; the problem is that it
implies an error theory of law. Thus construed, the traditional interpretation
of classical natural law theory is inconsistent with the claims that (1) a legal
system exists wherever the appropriate institutions, practices, and normative
outputs can be found without regard to the moral quality of these institu-
tions, practices, or normative outputs; (2) a norm is a law whenever it is prop-
erly promulgated (which might or might not include moral constraints on
the content of law); and (3) every mandatory legal norm gives rise to a legal
obligation. It seems clear that (1), (2), and (3) express obvious truths about
the content of a descriptive concept of law that is defined by practices that
are pre-theoretically characterized as legal and form the touchstone against
which a conceptual theory of law must be evaluated. Insofar as positivism is
grounded in these assumptions, then, this interpretation of traditional nat-
ural law theory implies an error theory of law.
Given that a conceptual theory of law employing MCA cannot result in
an error theory, the traditional interpretation of classical natural law theory
as a rival to positivism should be rejected. Whatever else classical natural law
theory might be, it cannot charitably be construed as a rival to positivism that
directly conflicts with it. Given the ambiguities in Aquinas’s and Blackstone’s
description of their natural law position, it is most charitably and plausibly
construed as doing something other than as adopting MCA to explicate the
descriptive concept of law that positivism purports to explicate.
There remain three options for interpreting classical natural law theory: clas-
sical natural law theory can be construed as either (1) adopting ICA to expli-
cate the real nature of law in a descriptive sense determined independently of
the linguistic practices that construct the content of our descriptive concept
of law; (2) adopting ICA to explicate the real nature of law in an evaluative
sense (i.e. law in its fullest sense) that is determined independently of the
linguistic practices that determine the content of our evaluative concept of
law; or (3) adopting MCA to explicate an evaluative concept of law that is
grounded in the descriptive concept positivism seeks to explain but has nor-
mative content as well. Which option is the best one depends on whether it is
plausible to think that officials, legal theorists, and ordinary citizens could be
systematically mistaken about what counts as law in legal systems like that of
the U.S.—i.e. whether there are any reasons to adopt an error theory, either
48

as a theory of the descriptive concept of law or as a theory of the evaluative
concept of law.
6. Dworkin’s interpretivism construed as a rival

to positivism
The core thesis of Dworkin’s interpretivism is expressed in the claim that “the
law of a community consists not simply in the discrete statutes and rules that
its officials enact but in the general principles of justice and fairness that these
statutes and rules, taken together, presuppose by way of implicit justifica-
tion.”12 Whether he is explicating the content of a descriptive concept of law
or is explicating the content of an evaluative concept, then, Dworkin’s view
can be expressed as follows:
Dworkinian Interpretivism: A norm n is a law in L if and only if either n is duly
promulgated by a court or legislature or n can be rationally derived from the general
moral principles of justice and fairness that show the totality of the law in L in its
best moral light.
Construed as a rival to positivism that adopts MCA to explicate a purely
descriptive theory of law defined by ordinary linguistic and legal practices,
Dworkin’s interpretivism is straightforwardly inconsistent with two elements
of institutional practices that are paradigmatically legal in character according
to both ordinary linguistic usage and the unambiguous practices of courts
and other officials in Anglo-American legal systems.
The first element of judicial practice with which Dworkin’s theory, con-
strued as a rival to positivism, is inconsistent has to do with the putatively
legal authority of courts to create new norms that legally bind other officials
of the legal system. Thus construed, Dworkin’s theory denies that judges have
anything plausibly characterized as a quasi-lawmaking authority—a denial
that is inconsistent with the convergent practices of judges and officials with
respect to court holdings and orders. As far as the ordinary legal practices of
common law adjudication in the U.S. are concerned, legislatures have im-
plicitly delegated authority to common law courts to develop and construct
certain areas of law, such as contract law and the law of torts.
It is true that legislatures in legal systems like that of the U.S. have au-
thority to take control over any area of the law that they have delegated to
judges to develop—and they have sometimes done exactly that. Sales of
12
Ronald Dworkin, “The Law of the Slave Catchers,” The Times Literary Supplement, December
5, 1975, 1437.
Dworkin’s interpretivism as a rival to positivism 49
certain goods, which were formerly governed by the common law of contracts
in the U.S., are now governed by statutory law in most states.13 Liability for
defective design or manufacture of products, which was formerly governed by
the common law of torts in the U.S., is now also governed by statutory law in
most states.14 The legal authority of courts in the U.S. to make and change the
common law is impliedly delegated by the legislature,15 but it is clear that offi-
cials in the U.S. converge on recognizing that courts have a quasi-legislative
discretion to change the content of the common law until such time as the
legislature enacts statutes that bring the relevant area of law under legislative
control. If what counts as law is determined by the convergent practices of
officials, then it is clear that judges in the U.S. have some legal authority to
make and change the content of the law.
There is another way to see this. Insofar as a common law court over-
rules or otherwise departs in a holding from a line of precedents, the court’s
holding has the effect of changing, to some extent, the norms of the common
law that were recognized as binding. Insofar as a common law court has legal
authority to do this (i.e. can bind lower courts with such a holding), it has
the authority to change norms of the common law. There is simply no way
to explain the fact that a common law court can bind itself and lower courts
with either of two conflicting holdings other than to acknowledge that courts
have a limited legal authority to modify the common law. If Dworkin’s theory
is intended to explicate the content of a descriptive concept of law that is
grounded in ordinary linguistic and legal practices, it is simply false; officials
in the U.S. clearly converge on recognizing judicial holdings that either de-
part from existing precedent or decide novel issues of law as establishing the
content of what counts as law in the purely descriptive sense of the term that
positivists seek to explicate.
But this quasi-legislative authority of the courts to make law in this de-
scriptive sense is not limited to changing the content of the common law. In
many cases, courts also have legal authority to modify statutes by exercising
a quasi-legislative discretion to make new law to fill gaps in the coverage of
statutes. As one Washington appellate court put it:
13
See, e.g., Revised Code of Washington (RCW) 62A.2-100 et seq.; available at: http://app.leg.
wa.gov/rcw/default.aspx?cite=62A.2-106.
14
See, e.g., RCW 7.72 et seq.; available at: http://app.leg.wa.gov/rcw/default.aspx?cite=7.72.
15
I say “impliedly delegated” here because, as far as I know, there is no statutory enactment that
explicitly delegates responsibility for various areas of law to the courts. The authority of common law
courts over such areas derives from practices of common law courts in the U.K. that were adopted
by officials in the U.S. without an explicit statutory delegation of authority.
50

Washington courts have also construed this statute to permit the adaptation of the
common law to address gaps in existing statutory enactments, providing that the
common law may serve to “fill interstices that legislative enactments do not cover.”16
Judicial authority to modify statutory law is clearly conferred upon courts
in the relevant jurisdictions by conventional recognition norms having the
status of law under the descriptive concept constructed by ordinary linguistic
and legal practices. It might be true that, depending on what one means by
“really,” that courts do not really have such authority to change law. What is
indisputable, however, is that, according to the empirically observable prac-
tices of officials that reflect their understanding of what counts as law in the
U.S., common law courts have a limited quasi-legislative authority to modify
the content of both statutory and common law.
The second element of judicial practice with which Dworkin’s theory,
construed as a rival to positivism, is inconsistent has to do with the status
of certain norms that are paradigmatically law under the descriptive con-
cept that positivism purports to explicate. Here is the problem. If Dworkin
is correctly construed as a rival to positivism, then either Plessy v. Ferguson,
which held that public school race-based discrimination does not violate the
Equal Protection Clause of the Fourteenth Amendment, or Brown v. Board of
Education, which held that such discrimination violates the Equal Protection
Clause, is mistaken in the sense that it is inconsistent with the principles
showing the existing legal history in its best moral light. Accordingly, one of
those holdings—presumably Plessy, decided nearly sixty years before Brown—
is mistaken in this sense: only one of them could be derived from the general
moral principles showing the legal history in the best light; the other is logi-
cally inconsistent with those principles. If consistency with moral principles
showing the legal history in the best light is a conceptually necessary condi-
tion for being law, then one of those two holdings cannot count as law.
Construed as purporting to explicate the content of the descriptive concept
of law grounded in ordinary linguistic and legal practices, Dworkin’s view that
one of these holdings does not count as establishing the content of the law
in the U.S. is clearly false. Even dissenting officials converged on recognizing
and treating Plessy as establishing the content of the law for nearly one hun-
dred years. Statutes permitting or mandating public school segregation were
enacted with the expectation that they would be applied and enforced by the
courts. The courts met those expectations, applying and enforcing those stat-
utes in ways that significantly changed the lives of millions of people. If, by
“law,” Dworkin intends the purely descriptive concept that is determined by
16
Dep’t of Soc. & Health Servs. v. State Pers. Bd., 61 Wn.App. 778, 783–84, 812 P.2d 500 (1991).
Dworkin’s interpretivism as a rival to positivism 51
what is pre-theoretically treated as law by the officials of the legal system, then
it is clearly false that the Plessy holding was not law.
Accordingly, if Dworkin’s claims about what counts as law are construed
as explicating the content of the same purely descriptive concept that posi-
tivism purports to explicate, then they are inconsistent with legal practices in
the U.S. that are characterized, according to the legal practices that construct
the content of that concept, as paradigmatically legal in character. To begin,
officials, as a matter of standard practice, regard judicial holdings that result
in new common law rules as being legally binding. Additionally, judges, law-
yers, and officials, also as a matter of standard practice, regard statutes upheld
by the courts as law and treat them as such regardless of whether they are
consistent with the moral norms showing the existing legal history in its best
moral light.
As was true of the traditional interpretation of natural law theory, the
problem with construing Dworkin’s interpretivism as a rival to positivism is
not just that it is false; the problem is that it entails an error theory of law.
Otherwise put, the problem is not just that Dworkin’s view that judges lack
quasi-lawmaking authority is inconsistent with official practices regarding
morally problematic holdings like Plessy; it is rather that, thus construed,
Dworkin’s view potentially calls into question whether too many of what are
treated by officials as legal norms are really “law.” For all we know, we all might
be mistaken about whether many legal norms are consistent with the objec-
tive moral norms that Dworkin believes constrain the content of the law. It
is certainly reasonable to hypothesize that many people at the time Plessy was
decided believed that the separate-but-equal doctrine was consistent with the
relevant standards of objective morality. For all we know, then, officials, legal
theorists, and ordinary citizens could all be systematically mistaken about
what counts as law in any legal system. Thus construed, Dworkin’s theory
entails an error theory insofar as it entails that, for all we know, we could all
be mistaken about what constitutes even the settled law in any paradigmatic
instance of a legal system.17
Given that a conceptual theory that presupposes MCA cannot result in
an error theory, Dworkin’s theory, like classical natural law theory, must be
17
Dworkin’s theory does not, strictly speaking, entail that we are systematically mistaken about
the law; since we do not have infallible access to what an objective morality requires, I am not
justified in claiming that his theory entails that we are mistaken about whether the relevant legal
norms conform to the relevant standards of an objective political morality. But it is enough that we
might, for all we know, be mistaken about the status of those norms. It is thus more accurate to say
that Dworkin’s theory might, for all we know, entail an error theory than it is to say that Dworkin’s
theory does entail an error theory of law. Either way, the result is problematic, and for the same
reasons.
52

construed either as employing ICA to explicate what is really the content of
the descriptive concept of law positivism purports to explicate or as explicat-
ing a concept of law that is grounded in the descriptive concept but also
has evaluative content. Which option is the better one depends in part on
whether it is plausible to think that officials, legal theorists, and ordinary
citizens could be systematically mistaken about what counts as law in legal
systems like that of the U.S. In other words, which interpretive option is
preferable depends on whether we have plausible reason to believe that the
presuppositions of ordinary legal practice entail an error theory of law.
7. Construing classical natural law and interpretivism

as deploying MCA to explain an evaluative concept of law
The fact that ICA can, while MCA cannot, result in an error theory has im-
portant implications with respect to evaluating a conceptual theory. Error
theories attribute a special kind of mistake to commonly held views about the
nature of something that are quite stubborn because they seem self-evident.
From an ordinary point of view, the law consists of those rules that the state
promulgates in some official way and enforces with its police power. An error
theory of law entails that these seemingly obvious views are systematically in
error and should be rejected.
Insofar as the “folk” views challenged by an error theory enjoy a strong pre-
sumption of being correct, an error theory requires a strong showing to justify
its acceptance. Consider, for example, the eliminative materialist’s view that the
furniture of the world contains no substances, properties, functions, or states
that have the properties thought to distinguish the mental from the physical.
As Paul Churchland describes the problem with our folk psychology: “[O]‌ur
common-sense psychological framework is a false and radically misleading
conception of the causes of human behavior and the nature of cognitive ac-
tivity.” Thus, the eliminative materialist holds that our folk psychology, which
is grounded in the idea that people have beliefs, thoughts, desires, and other
mental states, is systematically in error and should be rejected. Like the term
“phlogiston,” the terms “belief,” “thoughts,” “desires,” and “perceptions” refer,
on the eliminative materialist’s view, to nothing that really exists; there are no
such things as beliefs, thoughts, desires, or perceptions.
The intuitive implausibility of eliminative materialism calls attention to
the issue of what must be shown to justify the acceptance of an error theory.
Although it is not clear what would be needed to justify accepting such a
theory, what is clear is that the burden that must be met to justify accepting
MCA to explain an evaluative concept of law 53
an error theory is quite demanding. What is needed is a line of reasoning that
is sufficiently compelling to outweigh the very stubborn intuitive confidence
we have in the folk views challenged by the theory in question. For example,
a justification for eliminative materialism can succeed only insofar as the ev-
idence for accepting an error theory of mind strikes us as more compelling
than the evidence provided by our own experiences of having beliefs, desires,
and perceptions, which we take to exist in some real sense.
This is where eliminative materialism falls short. Eliminative materialists
point, for example, to what they take to be failures on the part of our folk
psychology to explain phenomena it must explain to be justifiably accepted.
For example, eliminative materialists argue that the hypothesis that we have
mental states contributes nothing to an explanation of why we need sleep,
how we catch fly balls, or why some people have mental illnesses. Insofar as
the hypothesis that we have mental states is not needed to explain such phe-
nomena, the principle of Ockham’s Razor dictates that the hypothesis should
be rejected.
These considerations strike most theorists and laypersons as inadequate to
justify accepting an error theory of mind. It might be true that the hypothesis
that we have mental states does not figure into a causal explanation of why
we need sleep or how we catch fly balls. But there is no way to make sense of
our conscious experiences without assuming that we have mental states and
that these states have some kind of existence in the world—even if their exist-
ence is causally dependent on the brain states that produce them, and even if
these states are epiphenomenal in the sense that they do not cause any acts or
behaviors. At bottom, this argument lacks the resources to convince people
to accept an error theory of mind because most people have far more intuitive
confidence in the claim that they really have beliefs, desires, and perceptions
than in any claim that the eliminative materialist could marshal in support of
the idea that there are no such things.
It is important to be clear on what is being claimed here with respect to the
viability of eliminative materialism. The claim is not that we have any conclu-
sive reason to accept the idea that mental states “really” exist. Rather, the claim
is that the arguments adduced by the eliminative materialist do not give us
adequate reason, given that it implies an error theory, to reject the folk the-
oretic view that mental states exist in some sense. While a folk theory might
enjoy some special presumption of correctness, the claim that the proponent
of an error theory fails to meet the argumentative burden for justifying an
error theory neither implies that the error theory is false nor that the folk
theory is true. It merely implies that the proponent has not given adequate
reason for rejecting the folk theory in favor of the error theory.
54

The difficulties in giving persuasive reasons to accept an error theory mil-
itate decisively against construing classical natural law theory and Dworkin’s
interpretivism as conceptual theories of law that directly engage positivism.
It is simply not clear what kind of reasons one could give that would jus-
tify accepting the view that what we regard as paradigmatic instances of law
might not really be law in the descriptive sense of the term. I have far more
intuitive confidence, given ordinary linguistic and legal practices, in the claim
that the normative outputs of courts and legislatures characteristically result
in something properly characterized as “law” in the descriptive sense than in
anything a proponent of an error theory of law could adduce in support of it.
Accordingly, there is little to be gained in construing classical natural
law theory and Dworkin’s interpretivism as deploying ICA to explicate the
real nature of law as it is defined independently of our conceptual practices.
Although ICA can, while MCA cannot, result in an error theory of law, it
does nothing to make those theories more plausible to understand them as
deploying ICA, rather than MCA because it does nothing to meet the burden
that must be met to justify accepting an error theory of law. Positivism pur-
ports to explicate the content of a purely descriptive concept of law that is
grounded entirely in ordinary linguistic and legal practices. To justify reject-
ing positivism on the ground that these ordinary practices tell us nothing
about what law really is, the proponent of an error theory needs to articulate
a compelling reason that we should think that the practices that construct
our shared concept of law are systematically misleading with respect to what
law really is, properly understood. In the absence of a sufficiently compelling
reason to accept such an interpretation of these putatively “anti-positivist”
theories, the better option is to construe them as deploying MCA to explicate
a different, but related, concept of law—a concept that is grounded in the
descriptive concept that positivism purports to explicate but has evaluative
content.
Thus construed, positivism, classical natural law theory, and interpretivism
converge on adopting MCA in attempting to give an analysis of the nature
of law but diverge with respect to which concept of law they are attempt-
ing to explicate. Whereas the positivist adopts a methodology that assumes
MCA in order to explicate the content of a purely descriptive concept of law
grounded in ordinary usage patterns, classical natural law and interpretivism
adopt, like the positivist, a methodology that assumes MCA but, unlike the
positivist, deploy it to explicate the content of an evaluative concept that is
also grounded in ordinary usage patterns. On this interpretation, then, all
three theories converge in rejecting ICA as a meta-methodological principle
but diverge with respect to which usage of the term “law” they are attempt-
ing to explicate. The positivist attempts to flesh out the deeper philosophical
What do Finnis and Dworkin say? 55
commitments of the descriptive usage of “law” as it picks out those norms
that are treated and characterized as law in practices that are pre-theoretically
regarded as legal. Classical natural law theories and Dworkin’s interpretivism
attempt to flesh out the deeper philosophical commitments of the evaluative
usage of “law” as it picks out those norms that are law not only in a purely
descriptive sense but also in a morally normative fullest sense of the term.
8. What do Finnis and Dworkin say?

Blackstone’s classical natural theory has traditionally been interpreted as a
rival theory to positivism that claims that an unjust norm cannot be a posi-
tive law. On this interpretation, Blackstone’s theory purports to deploy MCA
to explicate the descriptive concept of law defined by our linguistic and legal
practices. This interpretation can be challenged, as discussed in Chapter 1, on
the ground that a more careful study of the relevant texts suggests that he is
concerned with what is morally (or “really”) binding and authoritative—and
not with the project of conceptual jurisprudence as defined by the work of
Austin and Bentham.
There are similarly persuasive textual reasons to think that contemporary
natural law theorists and Dworkin are concerned to deploy MCA to explicate
the evaluative concept of law. John Finnis, the most influential contemporary
natural law theorist, endorses the Separability Thesis and hence denies the
view that it is conceptually impossible for there to be an unjust legal norm
and rejects the attribution of that view to Aquinas.:
“There is no necessary or conceptual connection between positive law and morality.”
True, for there are immoral positive laws; “there are two broad categories (with many
sub-classes) of unjust laws . . .”. And a conceptual distinction or disconnection is
effortlessly established by the move made in the Summa, of taking human positive
law as a subject for consideration in its own right (and its own name), a topic readily
identifiable and identified prior to any question about its relation to morality . . . .
“The identification of the existence and content of law does not require resort to any
moral argument.” True, for how else could one identify wicked laws such as Israel’s
prophet denounced in words so often quoted by Aquinas: “Woe to those who make
unfair laws [leges iniquas] who draw up instruments imposing injustice [iniustitiam],
and who give judgments oppressing the poor”?18
18
John Finnis, “The Truth in Legal Positivism,” in Robert P. George (ed.), The Autonomy of Law
(Oxford: Clarendon Press, 1996), 203, 204.
56

Similarly, while it is true that Dworkin took himself in his early work to be
articulating a view inconsistent with positivism, Dworkin is clear in Law’s
Empire that he is explicating an evaluative concept of law—and indeed
acknowledges that his theory is consistent with positivism:
We need not deny that the Nazi system was an example of law . . . because there is an
available sense in which it plainly was law. But we have no difficulty in understanding
someone who does say that Nazi law was not really law, or was law in a degenerate
sense, or was less than fully law. For he is not then using “law” in that sense; he is not
making that sort of preinterpretive judgment but a skeptical interpretive judgment
that Nazi law lacked features crucial to flourishing legal systems whose rules and
procedures do justify coercion.19
Dworkin concedes, as he must, that in a purely descriptive sense (the “prein-
terpretive” sense) the Nazis “plainly” had a system of law, but this is the sense
of the term that is expressed by the purely descriptive concept that positivism
purports to explicate. What the Nazis did not have, on his view, was some-
thing that was “really” or “fully” law in the interpretive—or evaluative—sense
of the term.
It turns out, then, that the traditional interpretation of classical natural law
theory and Dworkin’s interpretivism as inconsistent with positivism does not
cohere with what these theorists have had to say about the matter. Properly
construed, those theories are not inconsistent with positivism because they
are not rival accounts of the same concept. While positivism is concerned to
explicate the content of a purely descriptive concept of law, classical natural
law theory and Dworkin’s interpretivism are concerned to explicate the con-
tent of an evaluative concept that does not engage a positivist theory of law.
9. Can ICA ground a viable conceptual methodology?

One might nonetheless still be tempted at this point to reject the arguments
given above and to attempt to rescue an interpretation of Dworkin and clas-
sical natural law theories as anti-positivist. On this line of argument, classical
natural law theorists and Dworkin are best construed, despite any ostensible
evidence to the contrary, as deploying ICA to explicate a descriptive concept
of law the content of which is determined independently of our shared lin-
guistic and legal practices.
19
Ronald Dworkin, Law’s Empire (Cambridge, MA: Harvard University Press, 1986), 103–04.
Hereinafter LE.
Can ICA ground a viable conceptual methodology? 57
The problem with this line of interpretation is that ICA is just not a
methodological approach that beings like us could adopt with respect to
explicating the nature of law. If ICA requires being able to understand the
nature of law independent of the lexical meanings that fix our concept of
it, it seems clear we are not equipped to do this; the only epistemic access
that we could have to the nature of law is through the conventions that de-
fine the application-conditions for using the term “law.” To put the point in
Kantian terms, we can apprehend the nature of things only as they appear to
us mediated through the concepts we deploy to organize and make sense of
our experience; we simply have no way to apprehend things as they are utterly
independent of the concepts through which we organize the materials of our
experience. We do not have reliable epistemic access to what the nature of law
is if determined by considerations utterly independent of the practices that
define the lexical meaning of the term.
It is true that ICA starts from the lexical meaning of the term “law” in
attempting to identify the unconstructed real nature of law but it is utterly
mysterious as to how such an approach could produce a theory that we have
any reason to accept. As Jackson describes it, ICA purports, like MCA, to
begin from ordinary intuitions, but there is no reason to think that ordinary
intuitions would be reliable in explicating a concept or the nature of a thing
that is not even partly determined by our linguistic practices. Ordinary intu-
itions are reliable in identifying the content of concepts defined by our lin-
guistic practices because those intuitions are conditioned by those practices.
I have the strong intuition that only unmarried people can be bachelors only
because our linguistic practices converge on defining the term “bachelor” as
“unmarried adult male.” In contrast, if I have any intuitions about what a
bachelor is that are independent of our linguistic conventions for using the
term, I have no clue how to go about identifying them. ICA simply cannot
ground an epistemically accessible conceptual methodology.
There are hence two reasons for thinking that anti-positivist theories are
most charitably construed as explicating a different concept of law than the
purely descriptive concept that positivism purports to explicate. First, as dis-
cussed in the last section, construing anti-positivist theories as explicating the
purely descriptive concept of law that positivism purports to explicate implies
an error theory of law that we have no reason to accept. In the absence of
an argument that meets the difficult standard for justifying an error theory,
any theory of law that implies an error theory should summarily be rejected.
Construing anti-positivist theories as deploying ICA to explain the descrip-
tive concept does nothing to meet the requisite standard for justifying an error
theory. But, second, given that it is clear that we have no way of knowing how
we could identify the nature of law as defined utterly independently of our
58

conceptual practices, it is more plausible to construe anti-positivist theories
as deploying MCA to explain an evaluative concept of law that is grounded
in the descriptive concept with which positivism is concerned, but also has
evaluative content—namely that which expresses what counts as law in its
morally fullest sense or what counts as law in its morally best light.
10. Conclusions
The idea that one theory X entails that another theory Y is an error theory
presupposes that the two theories are attempting to explain exactly the same
phenomenon. It is reasonable to think that only a theory of the descriptive
concept of law can entail that another theory of the same concept is an error
theory. Construed as rival accounts that purport to explicate the content of a
descriptive concept of law fixed by ordinary legal and linguistic practices, pos-
itivist theories and putatively anti-positivist theories, like interpretivism and
the traditional interpretation of natural law theory, would both be candidates
for error theories.
But, as discussed above, there is another way to construe the disagreement
between positivism and so-called anti-positivist views—namely, as explicat-
ing different concepts of law that are both grounded in ordinary linguistic
and legal practices. On this view, positivists are explicating a purely descrip-
tive concept of law while natural law theory and interpretivism are explicat-
ing a concept that is grounded in the descriptive concept positivism purports
to explicate but also has evaluative content. On this construction of natural
law theory and interpretivism, a morally problematic norm that is properly
promulgated is properly characterized as a “law” in a purely descriptive sense
but not in the fullest evaluative sense of the term. Just as the evaluative con-
cept of art incorporates aesthetic standards that are external to the existence
conditions for art in its descriptive sense, the evaluative concept of law incor-
porates moral standards that are external to the existence conditions for law
in its descriptive sense. Positivists and theorists traditionally characterized as
anti-positivist are, thus construed, answering different questions about the
nature of law.
This should not be thought to disparage either conceptual project rela-
tive to the other. As discussed in Chapter 1, we need to approach the in-
stitution of law from empirical, normative, and conceptual points of view
to fully understand it. But, even as far as conceptual jurisprudence is con-
cerned, it is not enough to explicate one purely descriptive concept of law
to fully understand the nature of the institution of law. In addition, there
Conclusions 59
are the concepts of obligation, normativity, norm-guidance, and a host of
other related concepts—including the evaluative concept that is frequently
used, for example, to express disapproval of court holdings interpreting the
U.S. Constitution. If understanding the purely descriptive concept of law is
the foundation for a comprehensive conceptual understanding of law, un-
derstanding the evaluative concept is equally crucial. Thus construed, posi-
tivism, classical natural law theory, and Dworkinian interpretivism all have
something crucial to contribute to a fully comprehensive understanding of
the institution of law.
60
3
Legal Positivism and the Possibility
of Moral Criteria of Validity
This chapter provides an overview of the theses associated with the various
forms of legal positivism. It begins with a discussion of the assumption,
which grounds the project of conceptual jurisprudence, that legal norms and
systems have certain properties that distinguish them from other norms and
systems that purport to govern or regulate the behavior of subjects. It contin-
ues with a description of the core theses of legal positivism that are intended
to identify the conceptually necessary properties that constitute something as
law and distinguish things that count as law from things that do not count
as law. The chapter closes with an explication of inclusive and exclusive pos-
itivism, which comprise the focus for the remaining chapters of the volume.
1. The Differentiation Thesis

The project of conceptual jurisprudence is concerned with explicating the
content of those concepts that figure most prominently in practices pre-
theoretically regarded as legal in character; these concepts include the con-
cepts of law, validity, authority, and a legal system. As traditionally described,
the point of a conceptual theory of law is to explicate the content of each
of these concepts, locating them among a general conceptual framework
determined by the linguistic and legal practices that define the application-
conditions for using the corresponding concept-terms.
There are a couple of methodological approaches to the project of concep-
tual jurisprudence but the most common, and only epistemically viable, ap-
proach grounds the content of our concepts in the ordinary linguistic practices
that define the lexical meanings of the corresponding concept-terms. Given
that the lexical meaning of a term fixes the content of the corresponding con-
cept, any viable approach to explicating the content of a concept must begin
with the lexical meaning of the corresponding concept-term.
University Press.
62
62 Possibility of Moral Criteria of Validity

This does not imply, as discussed in Chapter 1, that there is no significant
difference between the methodologies of conceptual analysis and empirical
lexicography. The lexical meaning of a word contributes to fixing the content
of a concept but need not exhaust it. Insofar as the social practices that fix the
lexical meaning of a word are indeterminate with respect to the application
of the word in some theoretically interesting case, the lexical meaning fails to
exhaust the content of the corresponding concept. In this case, conceptual
analysis is needed to eliminate the indeterminacy by identifying the deeper
philosophical commitments grounding the relevant practices. The method-
ology of conceptual analysis, unlike the purely sociological methodology
of lexicography, is concerned to identify these philosophical commitments
presupposed by the relevant social practices and is hence philosophical in
character.
Consider, again, the term “bachelor.” While the lexical meaning of the
term “bachelor” is reported in dictionaries as “unmarried man,” the meaning
seems to be indeterminate with respect to whether the Pope is properly char-
acterized as a bachelor. Many people have the intuition that the term “bach-
elor” connotes a formal eligibility for marriage or a psychological openness
that is lacking in the case of the Pope and take the view that the Pope is not
properly characterized as a bachelor. Others have the intuition that being an
unmarried man exhausts the notion of a bachelor and hence take the view
that the Pope is as paradigmatically a bachelor as any other unmarried man.
The existence of conflicting intuitions on the applicability of the term to
the Pope gestures in the direction of thinking that the lexical meaning of
“bachelor” is indeterminate with respect to the Pope but is not enough to es-
tablish that it is indeterminate with respect to this application. The problem is
that we do not know what percentage of speakers would affirm that the Pope
is a bachelor and what percentage of speakers would deny that the Pope is a
bachelor. If, on the one hand, there were an even split on the matter, then the
lexical meaning of the term would be indeterminate with respect to whether
the Pope is a bachelor because there is no linguistic convention with respect
to whether the Pope is properly characterized as a bachelor. If, on the other
hand, 99 percent of speakers hold that the Pope is a bachelor while 1 percent
deny this, then there is a convention that equates being a bachelor with being
an unmarried man; the dissenting 1 percent are confused about the lexical
meaning of the term. Since we do not know the extent of the split among
speakers on this case, we do not know whether the lexical meaning of “bach-
elor” is indeterminate with respect to the Pope.
This tells us something about how a conceptual issue requiring philosoph-
ical explication arises. When there is a significant split among speakers about
the application of a term to a case that has some theoretical importance,
The Differentiation Thesis 63
the issue cannot be resolved by purely empirical means because an empirical
analysis can disclose only what people believe about the meaning of a term
and cannot provide reasons for resolving an issue of indeterminacy one way
or another. In the case of a 99–1 split on the application of “bachelor” to
the Pope, the term clearly applies to the Pope. In the case of the 50–50 split
above, there is an indeterminacy with respect to the lexical meaning that
requires resolution. While any indeterminacy in meaning can be resolved by
brute stipulation, this will not usually be regarded as satisfactory insofar as
competent speakers believe that there is an objectively correct answer; this
objectively correct answer would be determined by the deeper shared com-
mitments grounding the convergent patterns of usage.
It is important to note that these deeper commitments are philosophical in
character—even if people in the relevant community of speakers would not
think to characterize them as such. To begin, if one thinks that the Pope is
not a bachelor, it is not because the lexical definition rules it out; the Pope is
clearly a bachelor, on this definition. In this case, one is expressing skepticism
that the lexical definition fully expresses what a bachelor ultimately is, and
what a bachelor ultimately is defines the nature of bachelorhood.
Further, someone who accepts the claim that the Pope is a bachelor also
does so on the strength of a philosophical intuition concerning the nature
of bachelorhood. Insofar as skepticism about whether the Pope is a bachelor
is grounded in philosophical intuitions about the nature of a bachelor, so is
the intuition that the Pope is as much a bachelor as any other. In this case,
the view is that the lexical definition of a bachelor as an unmarried man fully
expresses the nature of a bachelor.
Ordinary speakers might not be likely to describe their intuitions about
whether the Pope is properly characterized as a bachelor as philosophical, but
the underlying intuition in each case is metaphysical insofar as it concerns the
nature of the thing picked out by the relevant concept-term. Claims about
the nature of a thing are claims about what properties constitute something as
belonging to the reference class of the concept-term picking out the thing and
are hence claims about the essential properties of the thing. As such, they are
properly understood as metaphysical and hence philosophical in character.
Conceptual analysis, then, purports to identify these deeper philosophical
commitments shared by a community of speakers that would resolve theoret-
ically significant issues of lexical indeterminacy. These deeper commitments
are implicitly grounded in the conventions for using the term but do not usu-
ally rise to the surface except in the form of puzzles, such as is posed by the
question of whether the Pope is a bachelor or, more to the point, by the ques-
tion of whether there can be objective moral constraints on the content of law
if it is a social artifact. Because these puzzles are ultimately metaphysical in
64

character, none can be resolved wholly on the strength of empirical observa-
tions having to do with convergent patterns of usage; these puzzles arise only
insofar as there is no convergence with respect to the usage in question. What
is needed to resolve the indeterminacy is an analysis that is explicitly philo-
sophical in character but one that is also grounded in the empirical patterns
of usage that fix the lexical meaning of the term.
The methodology of conceptual analysis attempts to resolve lexical inde-
terminacy by deploying various logical tools and the methodology of possible
cases to identify deeper intuitions about the nature of the thing picked out
by the concept-term in order to explicate these intuitions in the form of a
conceptual theory of the thing. Just as ethical theorizing proceeds by eliciting
intuitions on possible cases in order to extract deeper commitments in the
form of general ethical norms, conceptual analysis proceeds in a similar way
by eliciting intuitions on possible cases to extract deeper commitments with
respect to the nature of the thing picked out by the relevant concept-term.
Insofar as an analysis of a concept is concerned to disclose the nature of the
thing referred to by the corresponding concept-word, it is expressed in claims
that purport to be necessarily true. The nature of a thing is constituted by a
list of those properties that a thing must have in order to be properly charac-
terized as falling within the reference class of the relevant concept-term. It is
hence a necessary truth that a thing properly picked out by some concept-
term has every property defining the nature of the thing picked out by the
concept-term; it is, for example, a necessary truth that every bachelor has the
compound property of being an unmarried man.
A conceptual analysis of law, then, is intended to explicate the nature of
law by displaying all of the properties that make something that has them
an instance of law, rather than an instance of something else. As discussed
in Chapter 2, these properties do not merely distinguish things that are law
from things that are not law; they also explain why something that counts
as an instance of law has that status. In other words, the relevant properties
constitute something that is law as law in this sense: something that is law has
that status wholly in virtue of instantiating each of the relevant properties.
Just as, for example, instantiating the property of being an unmarried man
constitutes something as a bachelor, instantiation of all the conceptually nec-
essary properties of law constitutes something as law.
The idea that conceptual jurisprudence is concerned to identify properties
that distinguish things that are law from things that are not law presupposes
the claim that in every conceptually possible legal system there exist neces-
sary and sufficient conditions for a norm to count as law. Since the law of a
system includes both the norms that are legally valid and recognition norms
that define the criteria of validity, the relevant membership criteria include
The Differentiation Thesis 65
not only the conditions something must satisfy to be a valid legal norm but
also include the conditions something must satisfy to be a legal recognition
norm. Accordingly, the project of conceptual jurisprudence presupposes the
following:
The Differentiation Thesis: In every conceptually possible legal system S, there is a
set of criteria of legality (as opposed to legal validity) such that, for every norm n, n is
a law in S at time t if and only if n satisfies the criteria of legality at t.
The Differentiation Thesis is a metaphysical thesis and not an epistemological

thesis. The Differentiation Thesis neither presupposes nor implies any claims
about the extent to which the criteria of legality can be identified in any pos-
sible legal system. Although the Differentiation Thesis implies that there is
a set of common properties that distinguishes things that count as law from
things that do not, the claim that there exists such a set of properties implies
nothing with respect to our ability to reliably determine whether a particular
thing instantiates these properties or not. The former claim is a purely on-
tological one while claims concerning the latter issue are epistemological in
character. Whatever views a theorist takes on the epistemological qualities of
the criteria of legality—or, for that matter, the criteria of validity—must be
grounded in other commitments that rest, in part, on considerations of nor-
mative epistemology and, in part, on observable facts about legal practice in
the legal system of interest.
Further, the Differentiation Thesis implies nothing about the nature or
moral character of the law in any or every conceptually possible legal system.
As far as the Differentiation Thesis is concerned, legally valid content might—
or might not—be necessarily constrained by moral principles or by the inher-
ently interpretive character of law. Since the Differentiation Thesis is agnostic
on its face with respect to such claims, it is consistent with any non-skeptical
theory of the nature of law.
The Differentiation Thesis simply expresses the truism that, in every
conceptually possible society with a legal system, legal norms have certain
properties that distinguish them, as a conceptual matter, from moral norms
and from other social norms that might be used to appraise the behavior
of subjects in that society. Every conceptual theory of law has assumed the
Differentiation Thesis as a starting point because it implies that there exists a
set of properties that distinguish norms of law from norms of other kinds and
that constitute a norm with those properties as being a legal norm. As such,
the Differentiation Thesis entails that there is a unique set of existence condi-
tions that constitute something as law—and hence implies that our practices
define a concept of law that can be explicated in the form of necessary and
sufficient membership conditions.
6
2. Conceptual foundations of legal positivism
2.1 The Artifact Thesis
The most fundamental of positivism’s core commitments is the Artifact Thesis,
which asserts that law is, in essence, a social creation and artifact. What distin-
guishes law from non-law, according to this thesis, is the occurrence of some
contingent social fact (or facts) that constitutes a norm or system as one of
law. It is the occurrence of the relevant social fact (or facts) that constitutes any
norm or system that has the status of law as having this status.
The relevant facts constituting something as a norm or system of law can
be expressed in terms of some social property instantiated by the relevant
thing. For example, if the relevant social fact that must occur for a norm to
count as law is that it be officially promulgated, then a norm counts as law
in virtue of instantiating the social property of having been officially promul-
gated. According to the Artifact Thesis, then, anything that counts as law in
any possible legal system has that status wholly in virtue of instantiating the
relevant social properties. It is the occurrence of the relevant social facts or,
otherwise put, the instantiation of the relevant social properties that brings a
legal norm into existence as one of law and sustains it as having that status.
The occurrence of the relevant social facts manufactures the norm as one
of law.
The Artifact Thesis implies that both valid legal norms and the rule of
recognition have the status of law in virtue of instantiating some contingent
social property. The Artifact Thesis holds that law is wholly a social creation,
and this means that every norm that has the status of law is an artifact. Since
both the rule of recognition and the norms it validates all have the status of
law, the content of the rule of recognition, the legal system to which it gives
rise, and every norm it validates can be explained in terms of the relevant so-
cial facts or properties. Law is an artifact all the way down
The type of social fact or social property that explains, or manufactures, the
content of the rule of recognition and hence the criteria of validity will be dif-
ferent from the type of social fact that explains, or manufactures, the content
of the norms that are validated by the rule of recognition. The type of social
fact that explains the content of the valid norms of the system can differ from
one legal system to another because the content of the rule of recognition can
differ from one legal system to another; since the content a rule of recogni-
tion defining the properties a norm must have to be valid can differ from one
legal system to the next, so can the properties a norm must have to be valid.
Conceptual foundations of legal positivism 67
In contrast, the general type of social fact that explains the existence and
content of the rule of recognition is the same in every conceptually possible
legal system; and the relevant social fact must have something to do with
what certain people—i.e. those who serve as officials of the system—believe,
say, and do. What these people believe, say, and do manufactures the law and
determines its content. If there are no people doing things like promulgat-
ing, applying, and enforcing norms that tell people what to do, then there is
nothing that could count as law. Law requires the existence of someone in a
community who is doing something that purports to make certain behaviors
non-optional on the part of other persons in the community.
The issue, then, with respect to applying the Artifact Thesis to the rule of
recognition is to determine the type of social fact having to do with the beliefs
and behavior of officials that would manufacture in every conceptually pos-
sible legal system the content of the rule of recognition. While all positivists,
of course, are committed to the Artifact Thesis, they differ with respect to
which type of social fact manufactures the rule of recognition—and hence
with respect to which social fact manufactures the criteria that determine
what counts as valid law.
John Austin fleshes out the Artifact Thesis in terms of a community of
people who habitually obey the commands of a sovereign that are backed
with the threat of a sanction. On Austin’s view, the distinguishing feature of
a legal system is the presence of a sovereign who is habitually obeyed by most
people in the society but who is not in the habit of obeying anyone else. On
Austin’s view, a norm n is legally valid in a society S if and only if (1) n is
the command of the sovereign in S; and (2) n is backed up by the threat of a
sanction. The social fact that explains the content of the rule of recognition
that validates all and only commands of a sovereign is the existence of a com-
munity of people who habitually obey those commands, while the social fact
that explains the content of a valid legal norm is that it is commanded by the
sovereign. In both cases, the content of law is artifactual in virtue of being
manufactured by social facts.
Hart rejects Austin’s view that it is a conceptually necessary condition for
law that the norms be backed by coercive sanctions for noncompliance. Since
Hart denies that being enforced by a sovereign is a conceptually necessary fea-
ture of either a legal system or a legal norm, he must also deny that being en-
forced by a sovereign is one of the properties that distinguishes legal systems
or norms from other systems or norms. Insofar as he thereby denies that the
property of being enforced by a sovereign constitutes something as law, Hart
rejects Austin’s view of the social property that manufactures law.
On Hart’s view, the social fact that manufactures the legally authorita-
tive rule of recognition is a convergent practice of officials with respect to
68

what is recognized, applied, and enforced as law in a community in which
citizens generally satisfy the standards that officials recognize, apply, and en-
force as law. It is the behavior of officials with respect to what they recog-
nize, apply, and enforce as law—and not the disposition of citizens to obey
certain norms—that constitutes a norm that has the status of law as having
this status. Thus, on Hart’s well-known view, the social fact that establishes
the rule of recognition as legally authoritative is that officials converge in
practicing a norm that governs official behavior in recognizing, applying, and
enforcing law and thereby defines the content of the criteria of validity in that
system. The social fact that constitutes a norm as a valid law of the system is
that it is recognized, applied, or enforced in a manner that conforms to the
content of the rule of recognition practiced by the officials in the system.
Hart, like Austin, conceives of law as wholly artifactual in character.
Either way, the Artifact Thesis explains the content of the rule defining the
system’s criteria of validity and hence the authority of valid legal norms in
terms of some set of social facts and thereby conceptualizes law as an artifact
all the way down. On Hart’s version of the thesis, the relevant social fact is
the convergence of officials in practicing a rule of recognition that defines the
criteria of validity in the system; on Austin’s version, the relevant social fact is
the sovereign’s ability to coerce compliance. But, in either case, since the con-
tent of every legal norm is authoritative wholly in virtue of instantiating some
social property, the legal system and valid legal norms to which they give rise
are social creations. According to the Artifact Thesis, then, it is a conceptual
truth that law is a social artifact all the way down in the sense that the legal
system, rule of recognition, and valid legal norms are legally authoritative in
virtue of the occurrence of some contingent social fact or the instantiation by
the relevant entity of some contingent social property. The content of any law,
then, is wholly manufactured by the occurrence of the relevant social fact.
The Artifact Thesis should be distinguished from the so-called Sources
Thesis, according to which it is a conceptual truth that the existence and
content of law are fully determined by social sources. To understand the sub-
stantive difference between the two theses, it is helpful to note the logical
relationship between the Sources Thesis and Artifact Thesis. To begin, the
Sources Thesis logically implies the Artifact Thesis. If it is a conceptual truth
that the content of every law is fully determined by its sources, then it follows
that it is a conceptual truth that the content of every law is fully determined
by social facts that manufacture that content. If the Sources Thesis is true,
then law is an artifact all the way down.
In contrast, the Artifact Thesis does not imply the Sources Thesis. The
claim that the content of every law is manufactured by certain social facts
does not imply anything about what the content of the criteria of validity
must be; it implies only that what content the criteria of validity have they
have wholly in virtue of the occurrence of the relevant social facts. Since the
Artifact Thesis makes a claim only about what makes the rule that defines the
criteria of validity legally authoritative as a general matter, it does not imply
that the artifactual criteria of validity cannot incorporate constraints on the
content of law. The Artifact Thesis implies only that if the rule of recognition
incorporates constraints on the content of valid law, those constraints are le-
gally authoritative in virtue of the same general social facts that confer such
authority upon source-based criteria of validity.
The Artifact Thesis is hence related to the Sources Thesis but distinct from
it. Whereas the Sources Thesis states a conceptual constraint on the content of
the rule of recognition that determines what counts as valid law in the system,
the Artifact Thesis states only that the rule of recognition is authoritative
in virtue of instantiating some general social property that manufactures its
content and hence its existence as law. The Artifact Thesis, in essence, states
a claim about what accounts for the existence and authority of the rule of
recognition as a legal norm; in contrast, the Sources Thesis limits the content
of the criteria of validity defined by an authoritative rule of recognition to
source-based considerations.
The Artifact Thesis states a defining thesis of legal positivism, while the
Sources Thesis does not. One can, as inclusive positivists do, reject the Sources
Thesis and still be a positivist, but one cannot reject the Artifact Thesis and
still be a positivist. Unlike the Sources Thesis, the Artifact Thesis is part of the
shared foundation that distinguishes legal positivism from other conceptual
theories of law and is thus accepted by all positivists.
2.2 The Separability Thesis
The Separability Thesis can be understood at the most general level as simply
the negation of the distinguishing thesis of classical natural law theory as tra-
ditionally interpreted. Classical natural law theory asserts, on this construc-
tion, that it is a conceptual truth that the criteria of validity incorporate moral
constraints on the content of law. The Separability Thesis denies this claim
as it applies to the concept of law constructed by our ordinary linguistic and
legal practices and thereby acknowledges the conceptual possibility of wicked
laws and legal systems.
Thus construed, the Separability Thesis asserts no more than that it is con-
ceptually possible for a legal system to have criteria of validity that consist
entirely of considerations having to do with the source and manner in which
a norm is promulgated or recognized by officials. In such a legal system,
70

whether or not a norm counts as law is wholly determined by whether it has
been properly promulgated, which would include its having its source in the
appropriate official agency.
There is little logical distance between the Separability Thesis and the
Artifact Thesis. To begin, the Separability Thesis seems to be a straightforward
logical consequence of the Artifact Thesis. If it is a conceptual truth that the
content of every legal norm is fully manufactured by certain contingent social
processes, it seems to follow that there are no conceptually necessary content-
based restrictions on what can count as a valid legal norm; since these social
processes can change with respect to any element, so can the content that is
determined by that particular element. But insofar as a moral constraint on
the content of law would be a content-based restriction on what can count as
a valid law, it follows that it is not a conceptually necessary feature of law that
its content conforms to moral standards and hence that there are no concep-
tually necessary moral criteria of validity.
Conversely, the Artifact Thesis can be derived from the Separability Thesis.
If there are no conceptually necessary moral constraints on the content of
valid law, then it would seem to follow that the content of every valid legal
norm, as a conceptual matter, is fully determined by the occurrence of some
contingent social fact. The only determinant of legal content that could ob-
tain independently of any contingent fact would have to be content-based
in character. But if, as the Separability Thesis asserts, there are no necessary
moral constraints on the content of law, assuming moral constraints are the
only plausible candidates for necessary content-based constraints on the con-
tent of valid law, then the only other possible contingent determinants of
legal content would have to be social in character—and the only plausible
candidate for the contingent fact that determines the content of a valid legal
norm is some social fact involving the acts and dispositions of certain persons
in the community. Since the content of every legal norm is manufactured, it
follows that the legal system is manufactured and hence that law is an artifact
all the way down.
The important point, for our purposes, is this: while it is typically thought
that the core of positivism can be fully expressed only by reference to both
the Artifact and Separability Theses, the Artifact Thesis is sufficient to express
the distinguishing core of positivism since it implies the Separability Thesis.
At its foundation, then, positivism is nothing more than the Artifact Thesis.
Lending further support to the idea that the Artifact Thesis fully expresses
the core of legal positivism is the fact that the logical relationship between the
Separability Thesis and the Sources Thesis parallels that between the Artifact
Thesis and Sources Thesis. To begin, if there are no conceptually possible legal
systems with moral criteria of validity, then every conceptually possible legal
system lacks moral criteria of validity. But if there is at least one conceptually
possible legal system and there are no conceptually possible legal systems with
moral criteria of validity, then there is at least one conceptually possible legal
system without moral criteria of validity. The Sources Thesis, then, implies
the Separability Thesis—as it does the Artifact Thesis.
In contrast, as was also true of the logical relationship between the Artifact
and Sources Theses, the Separability Thesis does not imply the Sources Thesis.
The claim that there can be legal systems without moral criteria of validity
does not imply that there cannot be legal systems with moral criteria of va-
lidity; the claim that there can be families without children, for example, does
not imply that there cannot be families with children. Otherwise put, the
claim that there can be legal systems without moral criteria of validity is con-
sistent with the claim that there can be a legal system with moral criteria of
validity. Since it is therefore possible for the Separability Thesis to be true and
the Sources Thesis false, the Separability Thesis does not imply the Sources
Thesis.
Inclusive and exclusive positivism are hence both prima facie options for a
positivist theory of law. Although the Sources Thesis logically implies both the
Artifact Thesis and Separability Thesis logically implies the Sources Thesis, it
is logically implied by neither the Artifact Thesis nor the Separability Thesis.
This entails that a positivist can, at least at first glance, consistently hold that
it is conceptually possible for a legal system to incorporate moral criteria of
validity and hence that there is nothing in the core commitments of legal
positivism as expressed by the Artifact Thesis or the Separability Thesis that
is facially inconsistent with the conceptual possibility of moral criteria of
validity.
2.3 The Conventionality Thesis
If the Artifact Thesis is true, as seems plausible, of the purely descriptive con-
cept of law defined by ordinary usage, then there are no other determinants
of the content of the validity criteria than what those persons who serve
as officials of the system believe, say, and do with respect to what counts
as law. What we characterize as “law,” according to ordinary usage, in any
conceptually possible legal system is fully defined by what officials converge
in characterizing as law in the course of performing their functions as offi-
cials. The only determinants of the content of the criteria of validity have
to do, then, with what officials recognize, apply, and enforce as valid law in
the system.
Subjects of valid law, which include every person within the relevant juris-
diction, must generally acquiesce to official actions in the following minimal
72

respect: their behavior must conform sufficiently to the norms that officials
recognize, apply, and enforce as valid law in the community. This implies that
subjects of valid law acquiesce with respect to what officials treat and charac-
terize as “law.” But apart from this minimal efficacy condition, what counts
as law in any stable community that is plausibly characterized as governed by
law is fully determined by what officials recognize, apply, and enforce as law.
This implies that officials must largely agree with respect to what they
recognize, apply, and enforce as law. The existence of a stable legal system
depends on there being some minimal degree of consistency and clarity with
respect to what norms will be applied and enforced against citizens as law,
and this could not be achieved without officials converging in their behavior
with respect to what counts as law in the system. Further, this convergence of
behavior must be mutually conditioned in the following sense: part of what
explains why one official treats a norm as law with respect to her behavior as
an official of the system must have something to do with the fact that she is
aware that the other officials converge in treating that norm as law.1
What this means, then, is that the content of the criteria of validity is
wholly determined by the mutually responsive practices of officials with re-
spect to what they recognize, apply, and enforce as law in the following re-
spect: insofar as the content of the criteria of validity is determinate with
respect to whether some norm counts as law, its content is determined by and
only by the convergent practices of officials.
The relevant official practices might be indeterminate with respect to
whether some norm counts as valid law, but that is not a problem for the
Artifact Thesis. The Artifact Thesis asserts only that what content the law has
it has wholly in virtue of the occurrence of certain social facts that manufac-
ture it as law. By itself, the claim that all legal content is an artifact implies
nothing whatsoever about the substantive coverage of that content. It asserts
no more than that whatever content the criteria of validity have in any in-
stance is wholly manufactured by the convergent practices of officials with
respect to what they recognize, apply, and enforce as law.
This raises an issue with respect to how to characterize the relationship
between the relevant official practice and the rule of recognition. There are
two possibilities. The relationship between the rule of recognition and the
relevant official practice could be characterized as one of identity; thus con-
ceived, the rule of recognition would be identical with the practice in the
1
It might be that, as a purely conceptual matter, one cannot rule out the possibility of a system
in which officials always converge accidentally on what they recognize, apply, and enforce as law. But
if that is a conceptual or nomological possibility, it picks out a fringe state of affairs that tells us little
of significance about the nature of law as it is defined by our practices.
sense that the term picking out the rule of recognition and the term picking
out the relevant practices refer to exactly the same object. Alternatively, the
relationship between the two could be characterized as one of determination;
thus conceived, the rule of recognition would be determined by the practice
but would not be the same object as the practice.
Hart is not clear on this issue but seems to think that the relationship be-
tween the rule of recognition and the social practice that determines its con-
tent is one of identity:
The account I have given of [social rules] has become known as “the practice theory”
of rules because it treats the social rules of a group as constituted by a form of social
practice comprising both patterns of conduct regularly followed by most members of the
group and a distinctive normative attitude to such patterns of conduct which I have called
“acceptance”.2
To say that these patterns of conduct “constitute” the social rules of a group
seems to say something stronger than just that the patterns of conduct fix
or determine the content of the rule; it seems to say that they comprise the
rule. Insofar as these patterns of conduct constitute the rule in the sense of
comprising it, the aggregate of these practices is ontologically identical with
the rule; that is, the term referring to the rule of recognition and the term
referring to the practice pick out the same object.
This is a much stronger claim than the claim that the practices fix or de-
termine the content of the rule. Since, on the former view, everything that
can correctly be said of the rule can correctly be said of the aggregate of these
practices and conversely, it follows that the rule and the aggregate of these
practices have every property in common and hence do not differ from one
another in any respect. Since, on the latter, the rule is fixed or determined by
the aggregate of the practices but is not the same as those practices, the con-
tent of the rule merely supervenes on the content of the practices in the sense
that there can be no changes in the content of the rule without there being
corresponding changes in the content of the practices.
In any event, whether he believes that the rule of recognition is identical
with, or merely determined by, the aggregate of the practices, Hart accepts
the following claim:
The Conventionality Thesis: In every conceptually possible legal system, the con-
tent of the criteria of validity is fully determined by a rule of recognition that is con-
ventional in character.
2
H.L.A. Hart, The Concept of Law 3rd Ed. (Oxford: Oxford University Press, 2012), 255.
Emphasis added. Hereinafter CL.
74

The Conventionality Thesis, as is evident, asserts two conceptual claims about
the rule of recognition and the criteria of validity. It asserts, first, that the
rule of recognition is a conventional rule and, second, that there are no other
determinants of the content of the criteria of validity than the content of the
conventional rule of recognition.
The distinguishing feature of a conventional rule is that it is adopted and
followed by members of the group as a standard that governs their behavior.
It is a rule for the group because and only because they have converged on
accepting and following it. While the group might converge on accepting and
following it because they take a favorable attitude toward its content, it is not
a rule for members of the group in virtue of its having this favored content; it
is a rule for members of the group in virtue of their converging on accepting
the rule and conforming their behavior to its content.
This distinguishes conventional norms from other norms that govern a
group in virtue of their content. If, for example, moral norms are objective in
character, then it is true that what morality requires of people in any given in-
stance does not depend on what people believe about what morality requires
of them. But it is also true that these requirements govern their behavior
regardless of whether they accept those moral norms; objective moral norms
apply to rational subjects and govern their behavior because of the content
of those norms. In contrast, conventional norms do not apply to subjects in
virtue of what content they have; they apply to a group only insofar as mem-
bers of the group converge in accepting and following them.
To situate the Conventionality Thesis relative to positivism, it is worth
noting that it logically implies each of the distinguishing theses of positivism.
First, the Conventionality Thesis logically implies the Artifact Thesis. To
begin, the content of the rule of recognition is artifactual if fully determined
by a conventional rule. If the rule of recognition is a conventional rule, then
its content is wholly determined by the practices of officials, who converge
on accepting it and satisfying its requirements. These practices bring the rule
into existence as a legal norm and hence manufacture its content; accordingly,
if the rule of recognition is a conventional rule, its content is artifactual in
character.
The content of valid law is also artifactual in character if the Conventionality
Thesis is true. If the content of the rule of recognition is fully determined by
a contingent conventional rule, the content of the criteria of validity is also
fully determined by a contingent conventional rule. Since there are thus no
necessary constraints on the content of valid law, the content of valid law is
fully determined by the contingent occurrence of the social facts that, ac-
cording to the criteria of validity, constitute a norm as legally valid. Insofar
as the occurrence of the relevant social facts constitutes a piece of content as
legally valid, those social facts manufacture the content of valid law. If the
Conventionality Thesis is true, law is an artifact all the way down.
Second, the Conventionality Thesis logically implies the Separability
Thesis. If the rule of recognition is a conventional rule, then its content is
wholly determined by the contingent practices of officials. But this latter
claim implies that there are no conceptually necessary moral criteria of va-
lidity. Since (1) the content of the criteria of validity in any conceptually
possible legal system is fully determined by the content of the conventional
rule of recognition in that system and (2) the content of the rule of recogni-
tion is fully determined by the contingent practices of officials in that system,
the content of the rule of recognition and the content of the validity criteria
it determines are both contingent and hence can vary with respect to all par-
ticulars from one possible legal system to the next. Thus, there are no concep-
tually necessary criteria of validity—moral or otherwise.
The Conventionality Thesis thus bears the following relationship to pos-
itivism. Insofar as the Conventionality Thesis implies both the Separability
Thesis and the Artifact Thesis, any theory that logically presupposes or
implies the Conventionality Thesis is also a positivist theory. Assuming
the Artifact Thesis does not imply the Separability Thesis, those two theses
would, taken together, define the core distinguishing content of positivism;
since the Conventionality Thesis logically implies both, one cannot hold the
Conventionality Thesis and reject positivism as a theory of law. The adoption
of the Conventionality Thesis constitutes any coherent theory as positivist in
character.
The converse is not true: a theory can reject the Conventionality Thesis and
still be a positivist. Scott Shapiro, for example, rejects the idea that the rule
of recognition is a conventional rule, holding instead that the rule of recog-
nition is a plan:
The existence conditions for law are the same as those for plans because the funda-
mental rules of legal systems are plans. Their function is to structure legal activity so
that participants can work together and thereby achieve goods and realize values that
would otherwise be unattainable.3
Despite rejecting the Conventionality Thesis, Shapiro’s theory is clearly pos-
itivist in character. Note, to begin, that he accepts the Separability Thesis:
My strategy is to show that there is another realm whose norms can only be dis-
covered through social, not moral, observation, namely, the realm of planning. The
proper way to establish the existence of plans, as I argue below, is simply to point to
3
Scott Shapiro, Legality (Cambridge, MA: Harvard University Press, 2011), 119.
Hereinafter LEG.
76

the fact of their adoption and acceptance. Whether I have a plan to go to the store
today, or we have a plan to cook dinner together tonight, depends not on the desir-
ability of those plans but simply on whether we have in fact adopted (and not yet
rejected) them. In other words, positivism is trivially and uncontroversially true in
the case of plans: the existence of a plan is one thing, its merits or demerits quite an-
other (LEG 199).4
Further, he also accepts the Artifact Thesis: “[p]‌ositivists disagree with one
another about the nature of [the relevant] social facts, [but] . . . all legal facts
are ultimately determined by social facts alone” (LEG 27). Shapiro’s planning
theory of law is thus a positivist theory in virtue of accepting the Separability
Thesis and Artifact Thesis.
The Conventionality Thesis should not be thought of as a defining
thesis of positivism. The Conventionality Thesis is just one way to flesh
out the Artifact Thesis as it applies to the content of the rule of recogni-
tion and criteria of validity; that is, the Conventionality Thesis expresses
just one way in which the rule of recognition and criteria of validity can
be explained as social artifacts. It is true, as we have seen, that anyone who
accepts the Conventionality Thesis is a positivist, but that is because the
rule of recognition and valid legal norms would be social artifacts if the
rule of recognition is a conventional rule. But it is not true that anyone
who accepts the Artifact Thesis and Separability Thesis—and who is hence
a positivist—thereby accepts the Conventionality Thesis; accepting the
Conventionality Thesis is sufficient to constitute a view as positivist, but it
is not necessary.
2.4 The relationship between the criteria of validity and

the rule of recognition
The terms “rule of recognition” and “criteria of validity” differ in both
meaning and reference. The rule of recognition purports, by nature, to be
normative insofar as it has the form of a rule. It is a conceptual truth that an-
ything with the form of a rule, which incorporates deontic operators such as
“should” or “must,” purports to be normative; that is, in part, the work that
deontic operators necessarily do.
Something with the form of a rule might not, in fact, be normative but it is
purportedly normative. For example, the deontic proposition expressed by the
sentence “people should kill their first-born child” purports to be normative
4
This would also be true of conventions. The existence of a convention is one thing; its merit
is another. If so, then there is nothing in Shapiro’s plan positivism that makes positivism more
plausible than it would be if the rule of recognition is, as the Conventionality Thesis asserts, a
conventional rule.
given that it incorporates deontic operators that purport to prescribe an act
and hence to tell people what they should do. The proposition merely claims,
so to speak, to have some kind of normative force in virtue of its form; that
claim might be true, or it might not be. The rule of recognition, then, is pur-
portedly normative by its very nature insofar as it takes the form of a norm
and purports to tell officials how they should behave with respect to what
they recognize, apply, and enforce as law.
The rule of recognition is also subjectively and intersubjectively normative.
It is subjectively normative for a particular official insofar as she takes the in-
ternal point of view toward it and accepts it as governing her behavior. It is
intersubjectively normative for the group of officials insofar as officials con-
verge in adopting the rule of recognition as a norm that governs their behav
ior as officials and hence converge in accepting it as a norm governing their
official behavior. The rule of recognition, then, is purportedly normative in
virtue of having the form of a norm but is subjectively and intersubjectively
normative in virtue of the social practice that, as a conceptual matter, brings
it into existence.
A rule of recognition might not actually be normative from the standpoint
of objective right reason. A rule of recognition that, by nature, is purport-
edly, subjectively, and intersubjectively normative might not be objectively
normative because its authority over officials is not morally legitimate; in
that case, the rule of recognition has legal, but not legitimate, authority over
them. A rule of recognition that lacks objective normativity might justifiably
govern the behavior of officials from the standpoint of what the law requires
or permits but would not justifiably do so from the standpoint of what mo-
rality requires. In such a case, its requirements might create obligations that
are legal in character, but they would not necessarily create obligations that
are moral in character.
A morally illegitimate rule of recognition could thus legally bind officials
without morally binding them. Insofar as apartheid South Africa had a system
of law in the purely descriptive sense of the term defined by ordinary lin-
guistic practices, it had a rule of recognition that was purportedly, subjec-
tively, and intersubjectively normative and hence that legally bound officials.
But insofar as a rule of recognition that requires judges to apply and enforce
racially discriminatory rules could not be morally legitimate, the rule of rec-
ognition in apartheid South Africa was not objectively normative and did
not morally bind officials. By nature, then, a rule of recognition, qua rule, is
purportedly normative and, qua rule that is authoritative in virtue of being
accepted by officials, is subjectively and intersubjectively normative; however,
a rule of recognition is not, by nature, objectively normative.
78

To say that a rule of recognition is objectively normative is not neces-
sarily to say that the legal system to which it gives rise is morally legitimate.
Whether a legal system is morally legitimate surely depends on more than just
whether the rule of recognition that defines how law is recognized, applied,
and enforced in the system satisfies the relevant standards of political mo-
rality; whether a legal system is morally legitimate would also depend, in part,
on whether citizens largely consent to its authority and, in part, on whether
the content of its valid laws conforms to that of other moral standards. Even
so, it is not unreasonable to think that it is a necessary condition for a legal
system to be legitimate that its rule of recognition is objectively normative.
In contrast, the criteria of validity cannot be purportedly, subjectively, in-
tersubjectively, or objectively normative because the content of the criteria
of validity simply reports the properties that distinguish legally valid norms
from all other norms. Consider a legal system S in which all and only com-
mands of a sovereign are legally valid. The content of the criteria of validity
in S does no more than assert that what distinguishes legally valid norms
in S from all other norms is that each of the former and none of the latter
instantiates the property of having been commanded by the sovereign in S.
Since the criteria of validity for any legal system merely describe the proper-
ties that distinguish valid laws from other norms in the system—rather than
prescribe what properties ought to distinguish valid laws from other norms
in the system—the content of the criteria of validity is purely descriptive in
character.
This is true even if the criteria of validity include moral constraints on the
content of law. If, for example, it is conceptually possible for there to be a
legal system S in which the rule of recognition validates all and only com-
mands of a sovereign that are morally just, then the properties that distin-
guish content that is legally valid in S from content that is not legally valid
in S include not only procedural properties but also the substantive property
of being morally just. Either way, a statement of the criteria of validity in S
merely describes—rather than prescribes—the properties that all and only le-
gally valid norms have and that explain why norms in S that are legally valid
have that status.
The criteria of validity are as purely descriptive in character as the semantic
criteria for applying the term “bachelor.” Something counts as a bachelor, ac-
cording to the lexical definition, if and only if it instantiates the compound
social property of being an unmarried man. Similarly, something counts as
a valid law in a system if and only if it has all of the social properties that
officials converge in accepting as constituting something as valid law. The
criteria of validity, like semantic criteria governing the application of terms,
are purely descriptive in content.
The content of the criteria of validity resembles the content of semantic
criteria governing the application of a term in two additional respects. First,
both the criteria for applying the term “bachelor” and the criteria of validity
state existence conditions: the former state the existence conditions for bach-
elorhood while the latter state the existence conditions for valid law. In the
former case, a person’s being an unmarried man wholly explains his existence
as a bachelor; in the latter case, a norm’s satisfying the criteria of validity
wholly explains its existence as a valid legal norm of the system.
Second, both the content of the criteria for applying the term “bachelor”
and the content of the criteria of validity are determined by a convergence
among people in the relevant social group on accepting and conforming their
behavior to the relevant norm. The content of the semantic criteria for apply-
ing the term “bachelor” is determined by the content of a semantic norm
practiced by speakers that defines a standard of correctness for using the term;
according to this norm, “bachelor” should be used to refer only to unmarried
men. The content of the criteria of validity is determined by the content of a
rule of recognition practiced by officials that defines standards for how officials
may recognize, apply, and enforce valid legal norms; according to this norm,
“valid law” should be used to refer to only norms properly recognized, ap-
plied, or enforced under the rule of recognition.
While the content of both the semantic criteria and the criteria of va-
lidity are determined by standards that purport to be normative and are in-
tersubjectively accepted as subjectively normative, the content of each kind
of criteria is purely descriptive; such purely descriptive content does not even
purport to be normative because it lacks the form of a norm and hence cannot
be subjectively, intersubjectively, or objectively normative. Only something
that can be subjectively, intersubjectively, or objectively normative can pur-
port to be normative; and only something that has the form of a norm in
virtue of deploying deontic operators can be subjectively, intersubjectively, or
objectively normative.
Descriptive content can be incorporated into a norm that can be purport-
edly, subjectively, intersubjectively, and objectively normative but that does
not entail that such content is itself normative. Insofar as one treats, for ex-
ample, the lexical meaning of a term as defining a standard for using the term
to which one ought to conform, one has adopted a norm that requires one to
conform one’s use of a term to the lexical meaning and hence that incorpo-
rates in the relevant way the descriptive lexical meaning of the term. Similarly,
insofar as officials converge in treating properties defined by certain descrip-
tive content as criteria of validity, they have adopted a rule of recognition that
incorporates that descriptive content and thereby makes it normative. But be-
cause the content of semantic criteria or the content of the criteria of validity
80

can be fully expressed without the use of deontic operators or other norma-
tive language, such content is purely descriptive in character and hence, by
itself, lacks the resources to be normative in any respect. That is, such criteria
are neither purportedly, subjectively, intersubjectively, nor objectively norma-
tive because they can be fully expressed without deontic operators and hence,
properly expressed, do not have the form of a rule.
The existence conditions defined by the relevant criteria can nonetheless be
deduced from the corresponding norm. It is true that people regard the rel-
evant relationship between being a bachelor and being an unmarried man as
defining the lexical standard for using the term that is normative in every rel-
evant respect. While this relationship defines or presupposes a lexical standard
for applying the term, the relationship itself is not the standard. The standard
is a norm that warrants applying the term “bachelor” to all and only un-
married men. The existence conditions can be deduced from the norm, but
they do not fully comprise or constitute a norm—and could not because, as
discussed above, they can be completely expressed without the use of deontic
language.
The same, of course, is true for the criteria of validity, which define the
existence conditions for valid law: the existence conditions can be deduced
from the rule of recognition that is, by nature, purportedly, subjectively, and
intersubjectively normative, but the existence conditions do not comprise or
constitute a norm—and are hence not themselves normative in any sense. The
purely descriptive criteria of validity, which distinguish norms that count as
legally valid from those that do not count as legally valid and which state the
properties that constitute something as legally valid, can hence be expressed
without using any deontic operators according to the following schema:
Criteria of Validity Schema: A norm x is a valid law in S if and only if x satisfies the
conditions described in the proposition P, where P expresses a set of properties nec-
essary and sufficient for a norm to count as valid law in the system.
The foundations for law and language resemble one another in another the-
oretically significant respect. Consider, for example, a recognition norm that
requires officials to recognize, apply, and enforce as valid law any norm enacted
by the majority of legislators in the legislature. Such a norm empowers the
legislature to make law but does not impose any duties on the legislature to
enact any laws and is hence not categorically normative. But it is normative
in the sense that it defines a procedural requirement for something to be
recognized, applied, and enforced as valid law—namely, the norm must be
promulgated by the vote of a majority of legislators. Although the norm does
not provide a reason to enact any laws, it provides a reason for any legislator
with some antecedent reason to enact a law to do certain things to introduce
the bill to the legislature and to secure a majority vote. Such a norm is thus
fairly characterized as conditionally or hypothetically normative.
In this respect, such a recognition norm resembles those defining standards
for the correct use of words. A norm that requires that “bachelor” be used
only to refer to unmarried men provides no reason whatsoever to use the
term; there are many words I will never use despite my knowing the semantic
norms that define the correct application conditions. But insofar as I have
an antecedent reason to use some word correctly, the norms that accurately
state the application-conditions for the word give rise to a reason to use the
word in a manner that conforms to those conditions. The semantic norm that
prescribes using the term “bachelor” only to refer to unmarried men gives me
no reason ever to use the word “bachelor,” but it does give me a conditional
reason to conform my use of the term to that norm—at least when I want to
use the term correctly in order to communicate some thought. Both semantic
norms and recognition norms defining powers are conditionally normative in
precisely the same way because both explicitly make normative certain exist-
ence conditions.
A rule of recognition is likely to be comprised of a number of recogni-
tion norms that specifically define either how officials of the system must
be selected (e.g., by election or lineage) or how official functions of various
types (e.g., legislative, judicial, and executive) must be performed with respect
to recognizing, applying, and enforcing valid legal norms. We can call the
former Selection Norms and the latter Substantive Norms. The general form
of each can be expressed as follows:
Selection Norm Schema: To count as an official of some type (legislative, judicial,
or executive), selection of that person must be preceded by the occurrence of social
facts, F1, F2, . . . ,  Fn.
Substantive Norm Schema: To be legally authoritative, officials (i.e. those persons
who satisfy the relevant Selection Norm) must conform their legislative, judicial, and
executive acts to conditions C1, C2, . . . ,  Cm.
Both the Selection Norm and Substantive Norm Schemas define powers and
duties that are associated with recognizing, applying, and enforcing law and
can be expressed in the form of a norm that is, by nature, purportedly nor-
mative and that can be subjectively, intersubjectively, and objectively norma-
tive. To properly select an official, the selecting agency must behave such as
to ensure the occurrence of the social facts F1, F2, . . . , Fn. To ensure that
an official act with respect to recognizing, applying, and enforcing valid law
is authoritative, she must ensure that her behavior conforms to the social
conditions expressed by C1, C2, . . . , Cm. Accordingly, the rule of recogni-
tion, which is comprised of all the various selection and substantive norms
82

practiced by officials of the system, derives its purported, intersubjective, and
subjective normativity from the purported, subjective, and intersubjective
normativity of all the selection and substantive norms that fully comprise
its content. Unlike the criteria of validity, then, the rule of recognition has
the logical resources in the form of incorporated deontic notions—such as is
picked out by the term “must” in the above schemas—to define and express
duties.
Insofar as the purely descriptive criteria of validity for a society must
be extrapolated from the rule of recognition that is accepted and prac-
ticed by officials, the recognition norm is metaphysically basic while the
criteria of validity are derived from the recognition norm. In this respect,
the relationship between the two parallels the relationship between a se-
mantic norm that defines the application-conditions for using a word
and the lexical definition of a word; it is the convergence of a community
of speakers on satisfying a semantic norm that defines the descriptive
criteria for applying the word and, so to speak, brings into existence the
thing to which the relevant word applies. Likewise, it is the convergence
of officials on satisfying a rule of recognition that defines the descriptive
criteria of validity and brings valid law into existence in a system where
citizens generally conform their behavior to norms validated by the rule
of recognition.
3. Inclusive legal positivism

The Separability Thesis asserts that it is conceptually possible for a legal system
to have criteria of validity that do not include any moral norms constraining
the content of what counts as valid law in that system; that is, it asserts that
there is a conceptually possible legal system in which the criteria of validity
are exhausted by considerations having to do with the source and manner in
which a norm is recognized or promulgated as law. In such a legal system, it
is neither a necessary nor a sufficient condition for a norm to be legally valid
that its content satisfies some set of moral norms.
In contrast, the Incorporation Thesis asserts that it is conceptually possible
for a legal system to have criteria of validity that include moral constraints
on the content of valid law in the system. The Incorporation Thesis implies,
then, that there is a conceptually possible legal system in which the criteria of
validity are not exhausted by considerations having to do with the source of a
norm or the manner in which it is recognized, applied, or enforced as law. In
such a legal system, whether a norm counts as valid depends, at least in part,
on its moral merit.5
There are a number of different claims that one might associate with the
Incorporation Thesis:
(1) it is conceptually possible for there to be a legal system in which officials
converge in believing they are practicing a norm that requires them to rec-
ognize, apply, and enforce as valid law only properly promulgated norms
that satisfy certain moral standards;
converge in actually practicing a norm that requires them to recognize,
apply, and enforce as valid law only properly promulgated norms that
satisfy certain moral standards; and
converge in practicing a recognition norm that succeeds in incorporating
moral criteria of validity.
The first is a claim about what is possible for officials to believe about the rec-
ognition norm they are practicing; that is, the first is a claim about whether
officials can believe they are practicing a recognition norm that defines moral
5
There are two ways in which the validity of a norm could depend on the moral merit of its con-
tent. First, it might be a sufficient condition in a legal system for a norm to be legally valid that it
reproduces the content of some moral norm. In this case, a norm would be valid wholly in virtue of
the moral merit of its content—and not even partly in virtue of having its source in some author-
itative act of promulgation. Second, it might be a necessary condition in a legal system for a norm
to be legally valid that its content be consistent with some set of moral norms. In this case, morality
would function as a constraint on legislative or judicial authority to promulgate law by enactment
or declaration in the course of adjudication.
The relevant logical relation will generally differ depending on whether moral merit is a necessary
or sufficient condition for validity. While the relevant relation with respect to necessary conditions
will usually be the consistency relation, the relevant notion with respect to sufficient conditions will
usually be the conformity relation. One could not plausibly use the consistency relation in express-
ing sufficient conditions because it would validate inconsistent norms; there are many propositions
P such that P and ~P are each consistent with morality. A law that requires drivers to drive on the
right side of the road is no less consistent with morality than a law that requires drivers to drive on
the left side. Likewise, as a practical matter, one could not plausibly use the conformity relation in
expressing necessary conditions because it would result in too few norms—at least in modern legal
systems like ours. Many laws are intended as solutions to coordination problems and hence do not
reproduce the content of some moral norm.
This should not be construed to imply that it is conceptually impossible for a system to make it
a necessary condition for a properly promulgated norm to count as valid law that it reproduces the
content of a moral standard. If officials properly promulgate the minimum content of the natural
law and the peace is largely kept in the resulting system, then it arguably constitutes a legal system.
The point here is not that such a system is conceptually impossible; it is that it is not likely to be
practicable in worlds like ours given the features of our psychology and the features of the world of
scarcity in which we live.
84

constraints on the content of valid law. The second is a claim about what is
possible for officials to do with respect to the moral quality of the recogni-
tion norm they are practicing; that is, the second is a claim about whether
officials can converge on correctly applying some standard that purports to
define moral constraints on the content of valid law. The third is a claim
about whether it is conceptually possible for there to be a legal system in
which officials practice a recognition norm that succeeds in defining moral
criteria of validity; that is, the third claim is just another way of expressing the
Incorporation Thesis.
Claims (1) and (2) are logically independent. Claim (1) does not imply
claim (2). Officials might believe they are practicing a recognition norm that
requires them to recognize, apply, and enforce as valid law only properly
promulgated norms that satisfy certain moral standards but be mistaken be-
cause their practices are so patently immoral that they could not plausibly be
described as conforming to standards that are moral in character; in this case,
officials know which recognition norm they are practicing but are mistaken
about its moral quality.
Conversely, claim (2) does not imply claim (1). Officials might in fact be
practicing a recognition norm that requires them to recognize, apply, and en-
force as valid law only properly promulgated norms that satisfy certain moral
standards without believing that the relevant standards are moral in character;
in this case, they are practicing a norm that incorporates moral standards as
criteria of validity without realizing that the relevant standards are moral in
character. Since the relevant beliefs and actions of officials can come apart in
both directions, claims (1) and (2) are logically independent.
Claims (1) and (3)—i.e. the Incorporation Thesis—are also logically inde-
pendent. Claim (1) does not imply claim (3). The claim that officials believe
they are practicing a recognition norm that requires them to recognize, apply,
and enforce as valid law only properly promulgated norms that conform to
certain moral standards does not logically imply the claim that the recogni-
tion norms succeed in incorporating moral criteria of validity. Officials could
believe they are practicing a recognition norm that implicates certain moral
standards when the recognition norm they are practicing fails to incorporate
moral criteria because they are mistaken in believing the relevant standards
are standards of morality.
Conversely, claim (3) does not imply claim (1). The claim that the rec-
ognition norms officials are practicing incorporate moral criteria of validity
does not logically imply the claim that officials believe that the recognition
norm they are practicing requires them to recognize, apply, and enforce only
properly promulgated laws that satisfy standards that are moral in char-
acter. Officials could mistakenly believe that the recognition norm they are
practicing does not implicate standards that are moral in character when that
norm does implicate such standards and succeeds in incorporating them into
the criteria of validity of the system. What officials believe about whether they
are practicing a norm that implicates moral standards in the relevant way is
logically independent of whether the norm they are practicing succeeds in
incorporating moral criteria of validity into the system.
In contrast, claims (2) and (3) are not logically independent. It is helpful
to note that claim (3) logically implies claim (2), on a positivist theory of law,
because it is not possible for there to be a legal system with moral criteria of
validity unless officials succeed in practicing a recognition norm that defines
moral constraints on the content of law; the criteria of validity in every con-
ceptually possible legal system are fully determined, on a positivist theory,
by the convergent practices of officials with respect to what they recognize,
apply, and enforce as law.
But claim (2) does not obviously logically imply claim (3). And the problem
here is not just the one mentioned above, namely that officials might have ob-
jectively mistaken beliefs about the moral quality of their official actions. The
problem, rather, is that the existence of moral criteria of validity might be in-
consistent with some feature thought to be a conceptually necessary property
of a legal system, such as a claim to legitimate authority. If so, then the claim
that officials in an institutional normative system converge in practicing a
recognition norm that requires them to treat as valid legal norms only norms
satisfying some moral principle does not imply the claim that the criteria of
validity in that system incorporate moral constraints on the content of law.
Officials could succeed in practicing a recognition norm that defines moral
constraints on the content of valid norms but that nonetheless fails to actually
incorporate moral criteria of legal validity because such a system would lack
some conceptually necessary feature of a legal system.
Assuming that the possibility of moral criteria of validity is incompatible
with some conceptually necessary feature of a legal system, there are two ways
to characterize an institutional normative system in which officials practice a
recognition norm that putatively defines moral constraints on the content of
valid law. It might be that the system is not properly characterized as a legal
system because it fails to instantiate some conceptually necessary feature of
a legal system. Alternatively, it might be that the system is a legal system but
that the criteria of validity are best characterized, despite the beliefs of offi-
cials, as exclusively source-based in character because the system instantiates
all the conceptually necessary features of a legal system thought to be incon-
sistent with the existence of moral criteria of validity.
In essence, the Incorporation Thesis asserts that there are conceptually pos-
sible legal systems in which the legal validity of a norm depends on the moral
86

merits of its content. To say that there is a conceptually possible legal system
with moral criteria of validity is simply to say that there is a conceptually pos-
sible legal system in which the properties that distinguish valid legal norms
from other norms and that constitute a norm as legally valid include whether
a norm satisfies some specified moral standard. In such systems, the validity
of a norm depends on the moral merit of its content.
Exclusive positivists deny the Incorporation Thesis and instead hold the
Sources Thesis, according to which it is a conceptual truth that the existence
and content of law are fully determined by social sources. If the Sources Thesis
is true, then the criteria of validity are fully defined in terms of criteria hav-
ing to do with a norm’s being properly promulgated by an authorized official
act. It follows, then, that it is not conceptually possible to have content-based
criteria of validity of any kind. Since the only possible criteria of validity, on
the Sources Thesis, are those that define certain procedures for manufacturing
law, the Sources Thesis implies that the Incorporation Thesis is false.
Even so, it is important to note that the Sources Thesis is not logically equiv-
alent to the negation of the Incorporation Thesis. While, as just shown above,
it is true that the Sources Thesis implies the negation of the Incorporation
Thesis, it is not true that the negation of the Incorporation Thesis implies the
Sources Thesis. The negation of the Incorporation Thesis implies that there
are no conceptually possible legal systems with content-based constraints on
what counts as valid law that are moral in character, but the Sources Thesis
implies that there can be no content-based constraints of any kind on what
counts as valid law. Thus, the Sources Thesis and the Incorporation Thesis are
mutually exclusive but not jointly exhaustive with respect to what kinds of
criteria of validity are conceptually possible for a legal system.
Although the Sources Thesis is typically expressed in terms of a norm’s
having the appropriate sources, it is more perspicuously expressed in terms
of certain procedural constraints. To say that a norm is valid in virtue of its
source is to say something about the procedures by which the appropriate
officials recognize or promulgate it as law. To say that all and only commands
of a sovereign are law is to indicate a required source (i.e. that the norm must
originate with the sovereign) and a procedural constraint (i.e. the norm must
be commanded by the sovereign). Insofar as the notion of an appropriate
source can be understood in terms of authority conferred by something that
can roughly be characterized as procedural, the Sources Thesis is better under-
stood as claiming that the only possible criteria of validity are those having to
do with the procedures by which a norm is recognized, applied, and enforced,
which would include procedures for selecting the appropriate officials. The
only possible criteria of validity, on the Sources Thesis, are those that define
certain procedures for manufacturing law and for selecting persons author-
ized to do so.
Either way, exclusive positivists do not—and need not—deny either claim
(1) or (2) above. It should be clear, to begin, that exclusive positivism has
nothing to say whatsoever about the issue, addressed by claim (1), of whether
it is conceptually possible for officials to believe that they are practicing a
recognition norm that requires them to recognize, apply, or enforce as valid
law only norms that are consistent with some set of moral requirements—or
whether it is possible, for that matter, for officials to have the more explicitly
theoretical but related belief that they are practicing a rule of recognition that
incorporates moral criteria of validity. Although there is no plausible reason
to doubt that it is possible for officials to have either belief, there is nothing
in any conceptual theory of law that would have anything to say about what
kinds of beliefs an official could have about such matters.
Nor does exclusive positivism entail a position with respect to the issue
addressed by claim (2) of whether it is conceptually possible for officials to
converge on practicing a recognition norm that requires them to recognize,
apply, and enforce as valid laws only norms that satisfy certain moral stan-
dards. For example, the officials of a legal system might converge in recog-
nizing, applying, and treating as valid law only those properly promulgated
norms that satisfy certain moral standards of justice. What an exclusive pos-
itivist must deny is that these recognition practices succeed in establishing
conformity to some moral standard as a necessary or sufficient condition for
a norm to count as a valid law of the system. It might be true that exclusive
positivists typically deny claim (2) but, strictly speaking, the Sources Thesis,
like the Incorporation Thesis, is agnostic with respect to the issues addressed
by claim (1) and claim (2).
What the exclusive positivist denies is claim (3) that it is conceptually pos-
sible for a legal system to have moral criteria of validity, which is the distin-
guishing thesis of inclusive positivism. On this view, the only conceptually
possible criteria of legal validity consist exclusively of source-based consider-
ations having to do with the procedures by which officials are selected and the
procedural norms to which they must conform in executing their powers and
duties in making, changing, and adjudicating valid law. Although the exclu-
sive positivist view precludes all types of content-based criteria of validity, it
directly engages the Incorporation Thesis insofar as a moral constraint on the
content of valid law is also a content-based constraint on the content of valid
law. The Sources Thesis is fairly, if not fully, characterized in terms of denying
the Incorporation Thesis.
The claim that it is conceptually impossible for a legal system to have moral
criteria of (legal) validity, again, does not imply that it is conceptually impossible
8

for an institutional normative system of some other kind to have moral criteria
of (institutional) validity. Since, by its own terms, the Incorporation Thesis
is limited to stating a thesis concerning what is conceptually possible for an
institutional normative system that counts as a system of law, its negation is
limited to stating a thesis about the existence conditions only for legal sys-
tems. The Sources Thesis is a thesis that is limited by its own terms to the
existence conditions for systems of law.
Exclusive positivism, as will be discussed below, denies the conceptual pos-
sibility of a legal system with moral criteria of validity on the ground that
any institutional normative system with content-based criteria of validity
lacks some conceptually necessary feature of a legal system. Chapters 4, 5,
and 6 consider the argument that the existence of content-based criteria of
(legal) validity is inconsistent with the nature and justification of authority.
Chapter 7 considers the argument that an institutional normative system
with moral criteria of (institutional) validity is incapable of guiding subject
behavior in a way that it must be able to do in order to count as a legal system.
Chapter 8, the final chapter of this volume, provides a positive argument for
the Incorporation Thesis by defining a model of an institutional normative
system validating all and only mandatory moral norms that satisfies every
condition that is plausibly thought conceptually necessary for the existence
of a legal system.
4
Inclusive Positivism and the Arguments
from Authority
All positivists accept the Separability Thesis, according to which there can
be legal systems without moral criteria of validity, but positivists disagree on
whether there can be legal systems with moral criteria of validity. Exclusive
positivists hold that it is not conceptually possible for there to be a legal
system with moral criteria of validity; on this view, the criteria of validity in
every conceptually possible legal system are exhausted by considerations hav-
ing to do with how and by whom norms are promulgated. Inclusive positivists
accept the Incorporation Thesis, according to which there is a conceptually
possible legal system with moral criteria of validity; in such a legal system, the
properties that constitute a norm as legally valid include properties having to
do with whether its content satisfies certain moral standards.
Critics of inclusive positivism argue that the conceptual possibility of
moral criteria of validity is inconsistent with the nature and justification of
practical authority. On this line of reasoning, the possibility of moral criteria
of validity is inconsistent with law’s conceptual claim to authority because
an institutional normative system with moral criteria of validity is incapable
of instantiating all the conceptually necessary features of authority. Joseph
Raz argues that an institutional normative system with moral criteria of va-
lidity cannot be legitimately authoritative because the norms of such a system
cannot replace the judgments of its subjects in their practical deliberations
about what they should do and hence cannot be legitimate.
This chapter explores the various tensions that might be thought to arise
between the Incorporation Thesis and practical authority. It begins with a
general discussion of the differences between practical and epistemic au-
thority. It then articulates the various theses associated with the so-called
service conception of authority and concludes with a summary of the argu-
ments for the claim that this conception of authority is inconsistent with the
Incorporation Thesis (henceforth the “Arguments from Authority”).
University Press.
90
90 Inclusive Positivism and Authority
1. Epistemic and practical authority

There are two kinds of authority: epistemic and practical authority. P has
epistemic authority over Q with respect to a proposition x if and only if P
sincerely asserting that x is true is a reason for Q to believe that x is true. An
example of an epistemic authority with respect to cancer would be an oncol-
ogist; an oncologist sincerely informing a patient that she has cancer provides
the patient with a new reason to believe she has cancer. In contrast, P has
practical authority over Q with respect to the performance of an act x if and
only if P sincerely directing Q to perform x provides Q with a reason to do x.
An example of a practical authority would be a police officer; a police officer’s
directive to drive slower provides the driver with a new reason to drive slower.
These two forms of authority are logically independent of each other. The
claim that P has practical authority over Q does not imply that P has epi-
stemic authority with respect to Q over the domain of relevant propositions;
a police officer might still have practical authority over someone who knows
better what the relevant traffic laws and considerations of public safety re-
quire drivers to do. Conversely, the claim that P has epistemic authority with
respect to Q does not imply that P has practical authority over Q over the
domain of relevant propositions; a moral theorist might have epistemic au-
thority with respect to what someone is obligated to do, but this does not
make her a practical authority with respect to what she is obligated to do.
It is true, of course, that if P is an epistemic authority with respect to what
Q is obligated to do, P’s expressed judgments about what Q is obligated to
do will give Q a reason to believe what P has claimed Q should do, but those
judgments do not, simply in virtue of being expressed by P, provide Q with
a new reason to do what P has claimed Q is obligated to do. An attorney cor-
rectly informing me that I am legally obligated to pay a $100 fine does not
give me a new reason to pay the fine; whatever reason I have to pay the fine is
fully determined by the content of the applicable legal norms. The two forms
of authority, as well as the reasons to which they give rise, are sufficiently
different in character that there are no obvious logical connections between
them; reasons to believe and reasons for action do not bear any obvious rela-
tionships of logical entailment to one another.
The foregoing gestures in the direction of an important similarity with re-
spect to the character of the reasons that each form of authority provides.
Although different with respect to the type of reasons they provide, both kinds
of authority provide reasons that are content-independent in the following
sense: it is the source—and not the content—of the directive that provides the
Epistemic and practical authority 91
subject with a new reason. The reason to believe that the oncologist’s opinion
provides is content-independent; had the oncologist sincerely asserted that the
patient does not have cancer, that would be a reason for the patient to be-
lieve she does not have cancer, as it is the oncologist’s expertise that makes
her opinion trustworthy, and not the content of her opinion. Likewise, the
new reason for action that the police officer’s directive provides is content-
independent; had the police officer directed the driver to drive faster, that
would be a new reason to drive faster, as it is the police officer’s status as a
practical authority that makes her directive binding, and not the content of
the directive.
Contrast this with the reasons that moral norms would provide if morality
is objective. It is the content of the norm that it is wrong to intentionally
harm innocent persons that provides a reason not to intentionally harm in-
nocent persons—and not the source. Even if objectively true moral norms
have a source in divine promulgation, the content of those norms would still
provide reasons to comply with them. If the norm that it is wrong to inten-
tionally harm innocent persons has an authoritative source that gives rise to a
source-based reason to comply, the content of the norm would also give rise
to a reason not to intentionally harm innocent persons—namely, that it is
wrong to do so. Objective moral norms are the kind of thing, if anything is,
that provide content-based reasons to conform one’s behavior to those norms.
There is a conceptual issue as to whether, on the ordinary usages with
which this volume is concerned, a proposition can be authoritative in virtue
of its content. On the one hand, the notion of having authority—and hence
the notion of being authoritative—seems conceptually connected to the no-
tion of having an author in virtue of the two terms being etymologically
related: only norms that have an author can have authority or be authorita-
tive. If so, the norms of an objective morality, assuming they lack an author,
would be binding and would provide content-independent reasons for action
without being authoritative.
On the other hand, ordinary usage seems to permit characterizing binding
norms as authoritative even if they are not grounded in the directives of some
personal practical authority. It is not utterly counterintuitive to distinguish, as
some philosophers of religion have, between the authority of reason and the
authority of faith; if it is true that the authority of faith is source-based, the
authority of reason is not. Whether a norm has authority in virtue of being
derivable from reason—and not in virtue of having an appropriate source
in an author—would seem to depend on whether its content is authorita-
tive. Similarly, it does not seem implausible, as far as ordinary usage goes, to
characterize the norms of an objective morality as authoritative, regardless of
92

whether they have an author since mandatory moral norms function the same
way that any other authoritative norm does.
The reported lexical meanings seem indeterminate with respect to the issue
of whether a proposition can be authoritative only if it has a source in the
directives of something with practical authority. Merriam-Webster, for ex-
ample, defines the notion as follows:
Authoritative—1: having, marked by, or proceeding from authority; 2: possessing
recognized or evident authority: clearly accurate or knowledgeable.1
Whereas the first definition seems to suggest that a norm or proposition can
be authoritative only in virtue of having a source, the second seems to allow
for the possibility that a norm can be authoritative if its content is clearly
accurate. The lexical meanings thus appear consistent with the possibility of
norms that are authoritative in virtue of source and with the possibility of
norms that are authoritative in virtue of content.
Either way, the notions of epistemic authority and practical authority seem
to function differently with respect to propositions. If propositions can be
either epistemically or practically authoritative, the considerations that de-
termine whether a proposition is epistemically authoritative will be different
from those that determine whether a proposition is practically authoritative
for the following reason: a proposition can give rise to a reason for action
solely in virtue of its content but cannot give rise to a reason for belief solely
in virtue of its content. The content of the proposition expressed by “it is
morally wrong to set living human infants on fire” would appear to provide
a reason to refrain from setting living human infants on fire; the fact that it
is wrong to set living human infants on fire, by itself, provides a reason not
to do so. In contrast, the content of the proposition expressed by “2 + 2 = 4”
does not appear to provide a reason to believe it; the fact that the sum of 2
and 2 is 4 does not, by itself, appear to provide a reason to believe that 2 +
2 = 4; otherwise, we would have a reason to believe every proposition that is
true simply in virtue of its being true.
One might think that the self-evident quality of the proposition expressed
by “2 + 2 = 4” is what gives rise to its epistemic authority but that has more
to do with the abilities of rational knowers than with any inherent quality of
the content. The claim that a proposition is self-evidently true merely asserts
that its truth can reliably be discerned from a cursory inspection of its content
1
“Authoritative,” Merriam- Webster Online Dictionary; available at: https://www.merriam-
webster.com/dictionary/authoritative. The notion of authority being defined should be construed,
given that it is a general definition of “authoritative,” as the general one of which epistemic authority
and practical authority are subspecies.
Epistemic and practical authority 93
by a rational agent with certain abilities; however, the fact that its truth can
be discerned in such a manner is not enough by itself to make the proposi-
tion epistemically authoritative for any particular agent. That some particular
agent has, in fact, reliably discerned from a cursory inspection of its content
that the truth of the proposition is self-evident might make that proposition
epistemically authoritative for her, but it would not make the proposition au-
thoritative for any person who has not discerned its self-evidence.
There is another important way in which the notion of authority seems to
apply differently in epistemic contexts than it does in practical contexts. The
directives of practical authority bind subjects in the sense that they create an
obligation on the part of subjects to conform their behavior to the require-
ments of the authority as expressed in its directives. This obligation might
be moral in character, but it need not be. The authoritative rules of a chess
club, for example, bind in virtue of defining an obligation to comply that is
institutional, rather than moral, in character. It might also be true, of course,
that the directives of a chess club are morally obligatory but the obligation
to which its directives give rise in virtue of having an authoritative source in
the chess club is institutional. The precise character of the relevant type of in-
stitutional obligation and the mechanism by which it binds are not clear but
there is no obvious prima facie reason to think that it is a conceptual truth
that authoritative directives of every kind create or purport to create moral
obligations.
The capacity to create obligations distinguishes the directives of a prac-
tical authority from the opinions of an epistemic authority. It is commonly
thought that legitimate (or morally justified) practical authorities have the
capacity to bind subjects by providing reasons in the form of a moral obliga-
tion to comply. If one can freely choose one’s acts, then one can freely choose
to act, other things being equal, in a manner that conforms to the relevant
moral obligation. Of course, it might not be true that we freely choose our
acts; however, the idea that our behavior is conditioned by choices that are
responsive to reasons is presumptively plausible insofar as it conforms to our
experience of ourselves.
In contrast, the idea that epistemic authority gives rise to reasons that ob-
ligate the subject cannot be reconciled with our experience with respect to
belief formation. Consider, for example, the proposition expressed by “2 +
2 = 5.” There is nothing I could do to make myself genuinely believe that
proposition. There are many acts I could perform with respect to that prop-
osition: I could balance my checkbook adding 2 to 2 as if the sum were 5;
I could tell people that this proposition is true; perhaps I could even make
an effort to bring it about somehow that I come to believe it is true. But one
thing I cannot do is bring it about that I believe that 2 + 2 = 5 merely by
94

choosing to believe it. My belief that the sum of 2 and 2 is 4, and not 5, is
not a matter of choice.
The same is true of propositions that are not necessarily true. There is
nothing I could do to bring it about that, despite the sun shining brightly
through my window as I write this sentence, I believe it is now raining out-
side. What I believe is conditioned by how the evidence strikes me; while
I can freely choose to seek out evidence that confirms what I want to believe
and to avoid evidence that disconfirms what I want to believe, how the ev-
idence strikes me is not within my direct volitional control. Insofar as it is
clear that we do not choose what to believe, we cannot be obligated to believe
any particular proposition. The idea that epistemic authority can give rise to
an obligation of any kind simply does not cohere with certain obvious facts
of our experience.
Nevertheless, if it makes sense to think that we have obligations to believe
that are epistemic in character, there is an important respect in which these
obligations would differ from other paradigmatic forms of obligation: prac-
tical obligations are almost always owed to other people, but it is implausible
to think that one owes an obligation to believe what an epistemic authority
says. Suppose, for example, that a patient, for whatever reason, does not be-
lieve her oncologist’s opinion that she has cancer. It might make sense to char-
acterize her as irrational, although even that seems too strong. But it seems
implausible to characterize her as having wronged the oncologist by violating
some sort of obligation she has to believe the oncologist’s opinions.
This is not to say that the notion of obligation has no application with re-
spect to issues of what or whom we should believe; it is rather to say that the
notion of obligation, assuming that the same notion applies in both contexts,
has different properties when applied in the context of theoretical rationality
(i.e. epistemically) than when applied in the context of practical rationality.
Insofar as the concepts of authority and obligation are conceptually linked
across all contexts, this is enough to show that an opinion from an epistemic
authority would function quite differently in a subject’s reasoning than a di-
rective by a practical authority.
While it is nevertheless true that the notion of epistemic authority is rele-
vant in addressing many issues that arise in connection with law, the concept
of authority that figures most prominently in law is that of practical authority.
Law is not primarily contrived, by nature, to create epistemic obligations
or reasons to believe something. Law is primarily contrived, by nature, to
regulate behavior by issuing directives that create practical obligations that
provide new reasons to do what the directives require. It might be true that
the law must ensure that people know what the law requires to be efficacious,
but that function is subordinate to, and derives from, its primary function
Practical authority grounded in a claim of right 95
of creating new reasons to do certain things. Insofar as law has an epistemic
function by its nature, the epistemic function is secondary to its practical
function of regulating the outward expression of certain mental states in the
form of acts or behaviors.
2. Practical authority as grounded in a claim of right

Practical authority is typically, if not as a matter of conceptual necessity,
asserted or exercised under a claim of right of some kind; that is, practical au-
thority is characteristically grounded in, or justified by, some sort of right on
the part of the authority to tell the subject what to do. The right that grounds
a claim to, or exercise of, authority might be moral in character, as in the case
of parental authority, or it might be legal in character, as in the case of the
authority of the police to order me to park my car elsewhere.
The fact that practical authority is asserted under a claim of right
explains why authoritative directives obligate the subject. Insofar as prac-
tical authority regulates the behavior of subjects by means of directives
issued under some claim of right, authoritative directives obligate subjects
in some sense that depends on the character of the right claimed by the
authority. If a person’s authority over another is grounded in a legal right,
then directives that fall within the scope of that authority will create a legal
obligation on the part of the latter to comply. If, in contrast, a person’s
authority over another is grounded in a moral right, then directives that
fall within the scope of that authority will create a moral obligation on the
part of the latter to comply. It would be hard to make sense of the idea
that I have a legal/moral right to tell you what you must do with respect to
doing p if my directive that you do p did not bind you by creating a legal/
moral obligation to do p.
The directives of practical authority hence purport, at least, to provide the
kind of reasons generally associated with obligations, but the strength of the
reasons will depend on what grounds the relevant claim of authority. Insofar
as one person’s practical authority over another is grounded in a claim of legal
right, the novel reasons to which her directives give rise will be of the sort to
which legal obligations give rise; insofar as one person’s practical authority
over another is grounded in a claim of moral right, the novel reasons to which
her directives give rise will be of the sort to which moral obligations give rise.
If moral obligations trump conflicting legal obligations, the directives of a
moral authority will win over the directives of a merely legal authority in cases
where they conflict.
96

The claim of right that entitles a practical authority P to tell Q what she
must do allows P to direct Q’s behavior only within certain largely determi-
nable limits. P’s authority over Q might be general in the sense that P’s claim
of right embodies a wide range of behaviors and contexts within which Q’s
behavior is subject to being regulated by the former; the political authority of
a legislature might be an example of such general authority. In contrast, P’s
authority over Q might be limited with respect to both behaviors and con-
texts; a teacher’s practical authority over a student is limited to what goes on
in the classroom during school hours and is limited with respect to what she
can require of students during those hours.
3. Power, de facto authority, legitimate authority,

and law
Raz distinguishes between de facto authority and legitimate authority (or
authority per se). A de facto authority “either claims to be legitimate or is
believed to be so, and is effective in imposing its will on many over whom it
claims authority.”2 Not every de facto authority is legitimate; some practical
authorities might claim legitimacy without being legitimate, as is the case
with corrupt totalitarian regimes that systematically violate the moral rights
of citizens. A de facto authority, on Raz’s view, is legitimate—or is an authority
per se—when either its claim that it has legitimate authority is true or, what
comes to the same thing, its subjects’ beliefs that it has legitimate authority
is true.
The notion of legitimacy is concerned with the moral justification of prac-
tical authority. To say that a practical authority is legitimate is to say that the
authority is morally justified in issuing directives that tell subjects what to do.
As Raz puts the point:
[Authorities] claim . . . a right [to rule], i.e. they are de facto authorities because they
claim a right to rule as well as because they succeed in establishing and maintaining
their rule. They have legitimate authority only if and to the extent that their claim is
justified.3
The relevant form of justification is moral justification: “If [a legal system]
lacks the moral attributes required to endow it with legitimate authority then
2
Joseph Raz, Ethics in the Public Domain (Oxford: Oxford University Press, 1994), 211.
Hereinafter EPD.
3
Joseph Raz, The Morality of Freedom (Oxford: Oxford University Press, 1988), 26; emphasis
added. Hereinafter MF.
Power, authority, and law 97
it has none . . . . To claim authority it must be . . . a system of a kind which is
capable in principle of possessing the requisite moral properties of authority”
(EPD 215). A legitimate practical authority is, then, one that is morally jus-
tified in “imposing its will” on subjects by issuing directives that tell them
what they must do.
Both de facto authority and legitimate authority should be distinguished
from power. A person can have power over another person without having ei-
ther de facto or legitimate authority. P can be a de facto authority over Q only
insofar as P either claims legitimate authority over Q or is generally accepted
as a legitimate authority in the relevant community. All that is needed for P
to have power over Q is that P has some reliable means for coercively induc-
ing Q to comply with P’s commands. Someone who points a loaded gun at
me and demands my money has power over me but does not have de facto
authority over me.
The notions of having de facto authority and having power are logically in-
dependent. To begin, if someone has the kind of compromising information
on me that could, if disclosed, result in my being incarcerated for an extended
period of time, then, in the absence of some sort of claim of right to direct my
behavior, that person has power over me without having de facto authority.
Conversely, assuming that the one can be effective in “imposing one’s will”
over another person without having a reliable means to coercively induce
her to do what one directs, a person can have de facto authority over another
person without having power over that person. If, for example, I accept as an
authority someone with no reliable coercive means of inducing me to comply
with her directives because I believe she has legitimate authority over me, she
has de facto authority over me without having power.
Similar things can be said about the relationship between power and legit-
imate authority. Assuming, again, that one can be effective in imposing one’s
will over another person without having a reliable means to coercively induce
her to do what one directs, one can have legitimate authority over someone
without having power over her, as might be true of someone without any
means to coercively induce my compliance whom I accept as an authority
and whose commands always accurately express the requirements of right
reason. Alternatively, one can have morally legitimate authority with respect
to someone over whom one has coercive power. Depending on (1) whether
the authority of the U.S. government is legitimate with respect to citizens and
(2) whether the enforcement mechanisms authorized for violations of civil
and criminal laws rise to the level of being reliably coercive, the U.S. govern-
ment might be an example of a morally legitimate authority with coercive
power over its subjects.
98

The above shows that it is possible for a legal system to have power, de facto
authority, and legitimate authority over its subjects. If it is possible for there
to be a legitimate legal system with coercive power over its subjects, then it
is surely possible for there to be a legitimate legal system with coercive power
over its citizens that is believed by subjects to be legitimate.
Although it is clear that there can be a legal system that has de facto au-
thority without having morally legitimate authority, there is a question as
to whether there can be a morally legitimate authority without de facto au-
thority. As Raz defines the notion, a de facto authority “either claims to be
legitimate or is believed to be so” (EPD 211). The question is thus whether
there could be a morally justified authority who neither claims to be legiti-
mate nor is believed by subjects to be so.
The answer depends on both moral and conceptual considerations.
Consider the role that consent might play in legitimizing practical authority.
To begin, if we assume that it is possible, as a matter of political morality, for
P to have legitimate practical authority over Q without Q giving meaningful
consent, then it would seem that something could be a legitimate authority
without also being a de facto authority. Suppose that some putative authority,
P, who succeeds in imposing her will on subjects and does nothing that would
imply a claim of morally justified authority, satisfies all the moral conditions
for being a morally justified authority. Suppose, further, that subjects do not
consent to P’s authority and do not believe P’s authority is morally justified.
Then P would be an example of someone who has morally legitimate au-
thority without having de facto authority.
But if we assume that it is not possible, as a matter of political morality,
for P to have legitimate practical authority over Q unless Q gives meaningful
consent, then it would appear that it is not possible for something to be a le-
gitimate authority without also being a de facto authority. If one cannot give
meaningful consent to being subject to an authority without believing that
the authority is morally legitimate, then there could not be a morally legiti-
mate authority that is not also a de facto authority.
The assumption that one cannot give meaningful consent to an authority
without believing the authority is legitimate seems reasonable both as a nor-
mative matter of morality and as a descriptive matter of human psychology.
As a normative moral matter, consent must be meaningful to be effective,
and it is reasonable to think that consent is not meaningful in the absence
of a minimally informed belief that the authority is legitimate. As a descrip-
tive psychological matter, it seems reasonable to think that, in the absence of
duress or other factors that would problematize consent from a moral stand-
point, people would not give consent to be subject to an authority unless they
believe the authority is legitimate. If so, then there could not be a morally
Power, authority, and law 99
legitimate authority that is not a de facto authority—on the assumption that
one can have legitimate authority only over subjects who give meaningful
consent to authority.
The notion of legitimate authority is both descriptive and evaluative while
the notions of power and de facto authority are purely descriptive. To deter-
mine whether P has legitimate authority over Q, one must determine whether
P satisfies the morally normative conditions for having morally justified au-
thority over Q; and this requires an analysis that has both normative and em-
pirically descriptive elements. First, one must identify the appropriate moral
norms for evaluating whether a putative authority is legitimate and determine
what observable properties a putative authority must have to be legitimate;
and both of these tasks require morally normative analysis. Second, one must
determine whether the putative authority instantiates the relevant observable
properties; and this involves empirically descriptive analysis. If, for example,
a morally normative analysis correctly concludes that it is both necessary and
sufficient for a putative authority to be legitimate that the subject expresses
meaningful consent to the authority, then one will have to determine whether
the subject has, as an empirically observable matter, expressed consent and
whether that consent is, as a morally normative matter, meaningful.
In contrast, one can determine whether P has de facto authority over Q by
purely empirical means. To determine whether P has de facto authority over
Q, one must determine whether (1) P is generally successful in imposing her
will on Q; (2) P claims legitimate authority over Q; and (3) Q believes that P
has legitimate authority over Q. If (1) is true and either (2) or (3) is true, then
P has de facto authority over Q.
But notice that one can determine whether each of these conditions is
true by purely empirical means. Whether, first, P is generally successful in
imposing her will on Q can be determined by empirical means that require
determining whether Q’s behavior generally conforms to P’s commands.
Whether, second, P claims legitimate authority over Q can be addressed by
empirical means that require determining whether any of P’s behaviors imply
a claim of legitimate authority. Whether, third, Q believes that P has legiti-
mate authority over Q can be determined by asking Q, assuming Q answers
sincerely. Since the satisfaction of each of these three conditions can be deter-
mined by purely empirical means, the issue of whether one person has de facto
authority over another person is purely descriptive.
Similarly, the issue of whether P has power over Q is purely descriptive.
To determine whether P has power over Q, one must determine whether a
number of empirically observable conditions are satisfied. One must deter-
mine whether (1) P issues directives to Q; (2) whether Q generally complies
with these directives; (3) whether P has a means of coercing Q to comply with
10

P’s directives and is willing to use these means, if necessary, to secure Q’s com-
pliance; and (4) whether Q is inclined to comply with P’s directives, when
otherwise disinclined to do so, solely to avoid the application of such coercive
means by P. Since whether each of these conditions have been satisfied can be
determined by purely empirical means, the issue of whether one person has
power over another person is a purely descriptive issue.
4. Conceptual relations between law and authority

As a conceptual matter, legal systems are authoritative in the minimal sense
that they issue directives in the form of mandatory legal norms that tell
people what they must do and that are treated by officials of the system
as binding against subjects. These mandatory legal norms might be cate-
gorical as with the norms of the criminal law, which tell people what they
must not do regardless of any desires they might have; these norms are
categorical in the sense that they purport to provide reasons for action that
are not conditional on other reasons the subject might have. Alternatively,
they might be hypothetical as with the norms of contract law, which tell
people what they must do to create a contract that can be enforced by the
court; these norms are hypothetical in the sense that they purport to pro-
vide reasons that are conditional on other reasons that a subject might or
might not have.
Valid legal norms tell subjects what they must do, whether categorically or
hypothetically, and thereby define obligations that are treated by officials of
the system as binding. Valid legal norms might sometimes seem merely to en-
courage certain behaviors but even in those cases they define legal obligations.
A law giving a tax break to those who purchase automobiles with optional
emissions-reduction technologies defines obligations on the part of officials
to honor the promised reduction in taxes. While it might be conceptually
possible for a legal norm merely to encourage an act, legal systems do not
typically trade in norms that encourage some act; they characteristically trade
in norms that require acts.
Similarly, the rule of recognition that regulates official behavior is also au-
thoritative in the sense that it tells officials what they must do to make, apply,
and enforce law and is treated by officials as binding them. It seems to be a
conceptual requirement for the existence of a legal system that the rule of rec-
ognition includes some norms that are categorical in character; it is not im-
plausible to think that every conceptually possible legal system must include
a norm that requires judges to apply valid legal norms in deciding disputes
among citizens. In contrast, it is not obviously a conceptual requirement for
Conceptual relations between law and authority 101
the existence of a legal system that the rule of recognition contains some
norms that are hypothetical. It might be, if the Incorporation Thesis is true,
that there could be a legal system with a rule of recognition that validates all
and only mandatory moral norms; these norms would be valid in virtue of
content—without formal promulgation or recognition.
But if the Incorporation Thesis is false, it would appear to be a conceptual
requirement for the existence of a legal system that the rule of recognition
contains some hypothetical norms governing how valid legal norms can be
promulgated; if the only criteria of validity are source-based, then all valid
legal norms would have to have an authoritative source in official promul-
gation. Since the relevant recognition norms would have to be hypothetical
because norms that merely define the permissible procedures for making valid
law cannot provide any reasons for officials to make law, they are capable only
of providing conditional reasons that apply to officials with an antecedent
desire to make law. Even so, these norms also tell officials what they must do
provided that they have certain antecedent reasons to recognize, apply, or
enforce law.
While it should be clear that it is a conceptual truth that legal systems
have something that counts, according to ordinary usage, as practical au-
thority over subjects, it should also be clear that it is not a conceptual truth
that legal systems have legitimate practical authority over subjects—at least,
not on the descriptive concept of a legal system defined by ordinary usage.
It is straightforwardly false to think that, on that purely descriptive usage,
every law or legal system is morally justified; given that we commonly speak
of wicked laws and legal systems, it is clear that ordinary linguistic practices
do not require that something be morally justified to count as either a law or
a legal system.
The issue of whether a legal system necessarily instantiates power or de
facto authority is somewhat more complicated. Depending on one’s views
concerning whether sanctions are a conceptually necessary feature of a legal
system, it might or might not be true that legal systems necessarily have
power. If, on the one hand, sanctions are not a conceptually necessary feature
of a legal system, there could be a legal system that does not have power be-
cause, in the absence of available sanctions, it lacks a reliable means to coerce
compliance. If, on the other, sanctions are a conceptually necessary feature of
a legal system, then it is at least prima facie plausible to think that it is a con-
ceptual truth that legal systems have power.
Even so, the claim that a legal system authorizes sanctions as a response
to non-compliance does not imply that it has power because it does not
imply that those mechanisms are sufficient to reliably coerce compliance. It
might be that (1) such mechanisms are inadequate to produce compliance in
102

instances in which people are disposed not to comply; and (2) people in the
system are inclined to comply in the vast majority of cases without regard to
whether there are sanctions authorized for non-compliance. For this reason,
it is probably not true that it is a conceptual truth that legal systems have
power—even if it is a conceptual truth that legal systems authorize sanctions
for violations of mandatory legal norms.
The issue of whether a legal system has de facto authority is similarly
nuanced. A legal system has de facto authority if and only if (1) it is generally
successful in imposing its will on subjects; and either (2) claims legitimate au-
thority over subjects or (3) is believed by subjects to have legitimate authority
over them. To begin, it is not entirely clear whether it is a conceptual truth that
legal systems are generally successful in “imposing their will” on subjects. It is,
of course, a conceptually necessary condition for the existence of a legal system
that the behavior of citizens generally satisfies the requirements of valid legal
norms, but this does not imply that what explains their conforming behavior
is that the legal system is generally successful in imposing its will on them. P is
accurately characterized as imposing her will on Q with respect to doing r only
insofar as Q’s doing r is motivated in large measure by it being P’s will that Q
do r—at least in instances in which Q is inclined to do otherwise. If subjects’
conforming behavior is not generally motivated by its being the collective will
of the officials that they comply with valid legal requirements, then something
could be a legal system without being a de facto authority.
But regardless of whether a legal system is properly characterized as gen-
erally successful in imposing its will on citizens, it is clear that legal systems
exercise practical authority under a claim of right conferred by law insofar as
they tell citizens what to do and treat valid legal norms as creating obligations
on the part of citizens. As far as our ordinary linguistic and legal practices are
concerned, this seems so clearly true that no plausible theory of the descrip-
tive concepts of law or authority could deny this on the modest approach to
conceptual analysis adopted in this volume.
Further, it is not clear that it is a conceptually necessary condition of a legal
system that either it claims legitimate authority over subjects or is believed
by subjects to have legitimate authority over them. Certainly, it is not a con-
ceptually necessary condition for the existence of a legal system that citizens
accept it as legitimate. I know, for example, that the U.S. has something
properly characterized as a legal system on the relevant usage, but I do not
know whether citizens generally believe this system is morally legitimate—or
even whether most understand the notion of legitimacy well enough to have
a coherent belief about the legitimacy of the U.S. legal system. They might,
for all I justifiably believe, or they might not. As far as citizens are concerned,
Conceptual relations between law and authority 103
all that is conceptually necessary for the existence of a legal system is that
their behavior generally satisfies valid legal norms; it might be that citizens
are naturally inclined to refrain from the kinds of socially disruptive acts that
are prohibited by the valid legal norms of the system. Accordingly, for the
disjunction in the topic sentence of this paragraph to be true, it must be the
case that law claims legitimate authority in every conceptually possible legal
system in which citizens do not generally believe the legal system has legitimate
authority over them.
One way this disjunction could be conceptually true is, of course, if either
of the disjuncts is conceptually true; as a matter of logic, if p is conceptu-
ally true, then so is p or q. If it is true that every conceptually possible legal
system claims legitimate authority, then it is trivially true that every concep-
tually possible legal system either claims legitimate authority or is believed by
citizens to have legitimate authority. But if it is not a conceptual truth that
law claims legitimate authority in every conceptually possible legal system
in which citizens do not generally believe the legal system has legitimate au-
thority over them, then the issue of whether any particular legal system has
de facto authority is a contingent one that requires empirical investigation to
resolve.
This suggests that there might be more to the taxonomy of practical au-
thority than is captured by the notions of de facto and legitimate authority.
It is clearly a conceptual truth, if ordinary usage is the touchstone, that legal
systems have something properly characterized as practical authority and
that this authority is necessarily legal, as opposed to legitimate, in character.
Insofar as it is not a conceptual truth that legal systems have either de facto
authority or legitimate authority, the concepts of de facto authority and legit-
imate authority do not exhaust the relevant notions of practical authority as
they apply to legal systems.
Either way, there is nothing in any of this that seems obviously incon-
sistent with the possibility of moral criteria of validity. Positivists who reject
inclusive positivism believe that the possibility of moral criteria of validity
is incompatible with a legal system’s having the kind of authority that every
legal system can, as a conceptual matter, be presumed to have. Insofar as
there is some kind of tension between inclusive positivism and the concep-
tual claim that law is an authority of some kind, it will have to be found
in deeper and more specific commitments with respect to the nature and
justification of practical authority, such as are expressed below in the service
conception of authority.
104
5. The service conception of authority

Raz articulates his service conception of authority in the form of several
claims about the nature and justification of practical authority. It is these
more specific commitments that are thought to be inconsistent with the con-
ceptual possibility of moral criteria of legal validity. Insofar as an institutional
normative system has moral criteria of validity, it lacks, according to exclu-
sive positivists, some property that is, according to the service conception, a
conceptually necessary element of practical authority. Since it is a conceptual
truth that legal systems instantiate practical authority and that institutional
normative systems with moral criteria of validity do not, it is not conceptually
possible for there to be a legal system with moral criteria of validity.
According to the service conception of authority, the conceptual point or
function of authority is to resolve disputes among subjects by determining
what they should do in a manner that takes into account and weighs all of the
applicable reasons that antecedently apply to the subjects. Otherwise put, the
point of authority is to serve subjects by deciding for them what they should
do given the reasons that apply to them in the relevant situations. Authority
serves its subjects by mediating between them and the reasons that anteced-
ently apply to them by providing mandatory directives properly grounded in
those reasons.
But to minimally perform this mediating function of authority, an au-
thoritative directive must be capable of replacing the subject’s assessment of
the applicable reasons in her practical deliberations. If it cannot replace the
subject’s own assessment of the reasons, then the subject must assess them to
come to a decision about what she should do. In this case, the directive fails
to mediate between the subject and the reasons that apply to her because it
does not in any way change the way the subject must think about the reasons;
there is simply nothing done by the directive that is plausibly characterized
as mediating between the subject and those reasons—or, for that matter, as
serving its subjects.
The following, then, expresses a conceptual truth about practical authority,
on the service conception of authority:
The Preemption Thesis: The conceptual point of practical authority is to tell subjects
what they must do by issuing mandatory directives that can replace the subjects’ own
assessments of the applicable reasons in their deliberations about what to do.
While the service conception of authority does not imply that authoritative
directives must actually replace the subjects’ judgments in their practical rea-
soning, it does imply that authoritative directives must be capable of replacing
The service conception of authority 105
their judgments in their practical reasoning. Any directive that cannot replace
these judgments cannot do anything that is properly characterized as serving
subjects by mediating between them and the relevant reasons.
The Preemption Thesis implies that authoritative directives must be capable
of providing the subject with a first-order reason to do what the authority
directs her to do. Insofar as the Preemption Thesis asserts that mandatory au-
thoritative directives must be capable of functioning to replace the subject’s
own assessments of what right reason requires her to do, they must be ca-
pable of providing first-order reasons for action that are of the same kind that
the subject’s own assessments of the balance of applicable first-order reasons
would yield.
Further, authoritative directives must also be capable of providing the sub-
ject with a second-order reason not to act on her own assessment of the appli-
cable first-order reasons. Insofar as authoritative directives must be capable of
precluding subjects from acting on their own assessments of the applicable
first-order reasons, they must provide some kind of reason for subjects not to
act on their assessments of those reasons. Insofar as they provide a reason for
subjects not to act on their own assessments of the applicable first-order rea-
sons, they provide a reason for action that is about reasons and that is hence
second-order in character.
There is nothing in the Preemption Thesis that implies that a subject of an
authoritative directive is precluded from assessing for herself what the balance
of reasons objectively requires her to do; it merely implies that the subject of
an authoritative directive may not act on her assessment of the balance of
reasons. Preemptive reasons, unlike what Hart calls “peremptory” reasons,
do not preclude, or even purport to preclude, a subject from considering
what her own assessment of the underlying reasons would require.4 As far as
the Preemption Thesis is concerned, there is nothing in the character of an
authoritative directive that would bar subjects from deliberating, debating,
or arguing about what the balance of first-order reasons requires of them; an
authoritative directive purports, at most, to bar subjects from acting on their
own assessment of the reasons.5
4
On Hart’s view, commands are intended to “cut[] off deliberation, debate, or argument” and
hence provide second-order peremptory reasons that bar subjects from even considering what the bal-
ance of first-order reasons require them to do. H.L.A. Hart, Essays on Bentham (Oxford: Clarendon
Press, 1982), 254.
5
Raz’s claim that it is the nature of authority to issue directives that bar subjects from acting on
their own assessments of the balance of relevant first-order reasons seems too strong. There is no
reason to think in the case where the subject’s assessment of the balance of reasons agrees with the
authority’s assessment that the subject is barred from acting on her assessment when it dictates that
she should do what the authority has directed her to do. If I refrain from committing murder only
because it is morally wrong and not because it is illegal, I have done nothing to either violate the
106

Both the first-order and second-order reasons that an authoritative direc-
tive purports to provide are content-independent. Propositions that are au-
thoritative in virtue of being directed by a practical authority are authoritative
in virtue of source rather than content. Since the normativity of such proposi-
tions is thus source-based rather than content-based, the first-order reason to
act as the authority prescribes and the second-order reason for the subject not
to act on her own judgment of what is required by the balance of first-order
reasons are both content-independent. An authoritative proposition purports
to bind subjects to do what the proposition directs regardless of what their
own assessments of the reasons might yield and regardless of what the prop-
osition directs.
The Preemption Thesis is purely descriptive. It states a thesis about the
point or function of practical authority and hence makes no normative claims
about authority. To say, for example, that the function of an automobile is
to reliably transport people from one place to another place states a purely
descriptive claim about an automobile. Knowing the point or function of a
thing can help us to determine whether it is a good thing of its kind; an auto-
mobile that can, because of its design, transport persons only short distances
at very slow speeds is not a good example of its kind. But the idea that one
can evaluate the quality of an automobile in terms of how well it performs its
function does not entail that statements identifying its function are norma-
tive or evaluative. Insofar as the Preemption Thesis merely describes the point
or function of practical authority, it is purely descriptive.
In contrast, the two remaining theses fleshing out Raz’s service conception
of authority are normative in character. The first is a normative thesis with
respect to the character of the reasons on which an authoritative directive
should be based. Given that the point of practical authority is to serve subjects
by mediating between them and the reasons that antecedently apply to them,
an authority’s directives should be grounded in, and reflect, the authority’s
assessment of what those reasons dictate. Raz expresses this thesis as follows:
The Dependence Thesis: All authoritative directives should be based, among other
factors, on reasons which apply to the subjects of those directives and which bear on
the circumstances covered by the directives.
The Dependence Thesis expresses a normative claim about how practical
authorities should go about performing this mediating function. A practical
authority performs this mediating function, in part, by determining for the
subjects what the balance of relevant reasons dictates with respect to how they
legally authoritative directive prohibiting murder or to impugn the (legal) authority of the officials,
system, or norm.
should behave. If the function of authority is to issue directives that provide
subjects with both first-order reasons for doing what the authority requires
and second-order reasons not to act on their own assessments of the reasons
that antecedently apply to them, then the directives of an authority should be
based on and reflect the authority’s assessment of those reasons as they apply
to the subjects. There is no meaningful sense in which an authority could re-
alize this function of serving its subjects by mediating between them and the
reasons that antecedently apply to them if the authority’s directives do not
reflect the authority’s assessment of those reasons.
Though normative, the Dependence Thesis does not express a morally nor-
mative claim. Properly construed, the Dependence Thesis does not assert that
authoritative directives should, as a matter of political morality, be based on
and reflect the authority’s assessment of the reasons that antecedently apply
to the subjects—although that claim might also be true. The Dependence
Thesis expresses a claim that is functionally normative: given that the func-
tion of authority is to serve subjects by determining for them what the bal-
ance of reasons requires them to do, authoritative directives should be based
on and reflect the relevant reasons. While it might also be true that authorities
should, as a matter of political morality, base their directives on the balance
of the relevant reasons, the point is that authorities could not do what they
are supposed to do, given their function, without grounding their directives
in an assessment of the reasons that antecedently apply to the subjects. The
Dependence Thesis, as Raz intends it, expresses a claim that is functionally
normative, as opposed to morally normative.
Similar sorts of functionally normative claims apply to automobiles. Once
the function of an automobile is known, it is not difficult to extrapolate claims
that purport to explicate what an automobile must do in order to be capable
of even minimally performing its function. Given that the function of an
automobile is to reliably transport people from one place to another, an auto-
mobile’s engine must, as a functionally normative matter, be capable of being
started. The point here is not that an automobile that cannot be started will
not perform its function well; it is rather that an automobile cannot perform
its function unless it can be started. Both this thesis and the Dependence
Thesis state normative claims that are purely functional in character.
Thus construed, the Dependence Thesis expresses a thesis about the na-
ture of authority, and not a thesis about the justification of authority. If a
directive that does not reflect the underlying reasons is thereby rendered
incapable of performing the mediating conceptual function of practical au-
thority, then such a directive is no more plausibly characterized as authori-
tative than a vehicle that lacks any mechanism to start its engine is plausibly
characterized as an automobile. It is part of the nature of an automobile, given
108

its conceptual function, that there is some mechanism by which it can be
started. An automobile with a broken starter is a broken automobile but it
is still an automobile; something that utterly lacks a mechanism to turn its
engine on is not an automobile at all—except in, as Fuller might put it, “the
Pickwickian sense in which a void contract is a contract.”6
The service conception includes not only theses that purport to explicate
the nature of practical authority; it also includes a thesis about the justification
of practical authority:
The Normal Justification Thesis (NJT): An authority is justified to the extent that
its subjects are more likely to comply with what the balance of reasons requires if they
follow the authority’s directives than if they follow their own assessments of what the
balance of reasons requires.
Practical authority, then, is justified over a subject only insofar as the subject
“is likely to better comply with reasons which apply to him (other than the
alleged authoritative directives) if he accepts the directives of the alleged au-
thority as authoritatively binding, and tries to follow them, than if he tries to
follow the reasons which apply to him directly” (EPD 214).
There are a number of observations that should be made here about NJT.
First, like the Dependence Thesis, NJT expresses a normative thesis about
authority; however, unlike the Dependence Thesis, NJT is morally norma-
tive in character. Raz characterizes NJT as “a moral thesis about the type of
argument which could be used to establish the legitimacy of authority”7 and
locates NJT among other theories of moral legitimacy:
As John Bell pointed out to me, if there is a common theme to liberal political
theorizing on authority it is that the legitimacy of authority rests on the duty to sup-
port and uphold just institutions, as, following Rawls, the duty is now called. But
that duty is of course dependent on a prior understanding of which institutions are
just. The account here offered is meant as a beginning of an answer to that question.
(AJ 28)
The legitimacy of authority, on Raz’s view, turns on whether it has certain
moral properties: “If [a legal system] lacks the moral attributes required to
endow it with legitimate authority then it has none” (EPD 215; emphasis
added). The Dependence Thesis, construed to express a morally normative
thesis, might also be true of authority but is intended to express a thesis that is
only functionally normative. In contrast, NJT clearly expresses a moral thesis
about the conditions an authority must satisfy to be legitimate.
6
See Lon Fuller, The Morality of Law, Rev.ed. (New Haven, CT: Yale University Press, 1969), 39.
7
Joseph Raz, “Authority and Justification,” Philosophy and Public Affairs, vol. 14, no. 1 (Winter
1985), 18; emphasis added. Hereinafter AJ.
Second, unlike the Dependence Thesis, NJT does not express a func-
tionally normative claim about what is minimally required for authority to
be capable of performing its function of mediating between subjects and
the reasons that antecedently apply to them. A practical authority that fails
to satisfy NJT is not thereby rendered incapable of serving subjects in this
way; all the authority must do to succeed in serving them is to make better
decisions about what subjects should do based on the reasons that anteced-
ently apply to them. An authority that fails to satisfy the Dependence Thesis
cannot minimally serve its subjects by mediating between them and the rea-
sons that antecedently apply to them because such an authority cannot do
what is conceptually required to mediate between subjects and the reasons
that apply to them. Whereas the Dependence Thesis is functionally norma-
tive and not morally normative, NJT is morally normative and not function-
ally normative.
Finally, although there is no conceptual link between epistemic and prac-
tical authority (apart from their being two different species of the genus au-
thority), NJT is a normative theory of authority that connects the two in the
following way. If NJT is true, then whether or not a de facto authority, P, has
morally legitimate practical authority over Q depends in the “normal” case
on whether P is an epistemic authority with respect to Q on the matters that
fall within the scope of P’s practical authority. P has epistemic authority over
Q with respect to a class of issues only if P is more likely than Q to be correct
about those issues. If NJT is true, then P has morally justified practical au-
thority over Q with respect to a class of acts in the normal case only if P has
epistemic authority over Q with respect to whether acts in that class comport
with the balance of first-order reasons that antecedently apply to Q.
The Preemption Thesis, NJT, and the Dependence Thesis also apply to
advice. Insofar as the point of advice, like that of authority, is to serve the sub-
ject by mediating between the subject and the reasons that antecedently apply
to her, something like the Dependence Thesis is also true of advice: any piece
of advice “should be based, among other factors, on reasons which apply to
the subjects of [that advice] and which bear on the circumstances covered by
the [advice]” (EPD 214).
The relationship between a piece of advice and the underlying reasons
appears tighter than it is between an authoritative directive and the under-
lying reasons. It is true that a piece of “advice” that does not even purport to
be grounded in an assessment of the reasons that antecedently apply to the
subject is simply not worth considering by the advisee because there is no
reason to think that it is likely to be correct. But a suggestion that does not even
purport to be grounded in an assessment of the underlying reasons is just not
plausibly characterized as advice. If you ask me for my advice on whether you
10

should do p and I make my suggestion based on the outcome of a coinflip,
I have not given you anything that would count as advice. According to or-
dinary linguistic conventions, someone who asks me for advice is implicitly
asking me to weigh all the relevant reasons and make a recommendation
based on the outcome of my assessment of those reasons.
In contrast, it is not obvious that a directive that counts as authoritative
cannot, as a conceptual matter, be grounded in a coinflip. If a judge in a case
orders me to do something I am otherwise unwilling to do on the strength of
a coinflip and the decision survives appeal, then it seems reasonable to char-
acterize the order as authoritative. Notwithstanding that a judge’s directive
in that case would be illegitimate, such a directive would still arguably be
authoritative, according to ordinary usage because (1) it would still be issued
by the judge under a claim of legal right; (2) it tells me what I must do; (3) it
would still be treated as binding on me; and (4) it would succeed in imposing
the will of the judge on me. Given that a piece of advice cannot, as a concep-
tual matter, be grounded in a coinflip, the above suggests that the mediating
role of advice, as a conceptual matter, requires a tighter connection between
the prescription and the underlying reasons than is required by the mediating
role of authority.
Something like NJT is also true of advice. It makes sense for us to seek out
and take advice only from people we have some reason to think are likely to
provide advice that is sound in the sense that its content prescribes the act
best supported by the applicable reasons. Accordingly, it is plausible to think
that a subject is justified in seeking and taking advice from another person
only insofar as the subject is “better likely to comply with the reasons which
apply to him (other than the alleged [piece of advice]) if he accepts the [ad-
vice] . . . and tries to follow [it] than if he tries to follow the reasons which
apply to him directly” (EPD 214).
There is, however, this difference between NJT as applied to authority and
NJT as applied to advice. NJT is clearly morally normative in the former case
in the sense that it defines the conditions of morally legitimate practical au-
thority, but it is not obviously morally normative in the latter. It seems, to my
mind, odd to think that someone needs a moral justification in the ordinary
case either to seek or take advice. In essence, seeking advice amounts to no
more than asking someone’s opinion, and taking advice amounts to no more
than acting on someone else’s opinion; in neither case does there seem to be
even a prima facie issue of moral justification.
Nor do there seem to be any moral norms governing giving advice that do
not also govern most ordinary speech acts that would hence tell us anything
in particular about the nature of advice. It is true that one has a moral duty
to give advice in good faith or to be sincere in offering advice, but that duty
seems to apply to most types of communicative acts. It is also true that the
duties of sincerity and good faith seem to entail that a piece of advice should
be grounded in one’s best assessment of the applicable reasons. But once that
requirement is satisfied, it seems odd to think that one can justifiably give
advice only if one is—or has reason to think one is—more likely, as a general
matter, to get the relevant issues correct. An authoritative directive, after all,
requires an act while a piece of advice merely recommends it.
As it pertains to advice, NJT is instead more plausibly construed as stating
the conditions under which one is prudentially justified in seeking or taking
advice from another. People frequently seek out advice for purely prudential
reasons. If I am seeking advice on the best way to invest my savings to max-
imize my gains, I am looking for someone whose advice is most likely to en-
able me to realize my prudential goals of maximizing my savings. If a person
is prudentially justified in doing something insofar as doing that something
conduces to the satisfaction of her interests, then NJT seems quite plausibly
construed as a principle of prudential justification.
There is thus some dissimilarity with respect to the applicability of NJT to
practical authority and its applicability to advice. If the foregoing is correct,
then NJT applies differently to advice than it does to practical authority. But
this construction of NJT as applied to advice, like the intended construction
of NJT as applied to authority, seems to state a plausible principle of prac-
tical reasoning; principles of prudential justification are no less principles of
practical reasoning than are principles of moral justification. If the parallel
between advice and authority is a bit weaker here than in the other respects
considered above, it remains nonetheless conspicuous.
Something like the Preemption Thesis is also true of advice. If we assume
that an advisor is someone who plausibly holds herself out as having expertise
with respect to whether certain kinds of actions should be performed, then
“[t]‌he fact that an [advisor suggests] performance of an action is a reason for
its performance, which is not to be added to all other relevant reasons when
assessing what to do, but should replace some of them” (EPD 214). Insofar as
the advisor is enlisted to make a suggestion that is based on her evaluation of
the reasons that antecedently apply to the subject, the advice should express
that evaluation in the form of a prescription that can replace the subject’s
own assessment of the reasons. Insofar as advice can and should replace the
subject’s own assessment of the reasons, it should be possible for advice to pre-
clude a subject’s acting on her own assessment of the reasons; if she accepts
and follows the advice, then it will have succeeded in precluding the subject
from acting on her own assessment of the reasons.
As applied to advice, the Preemption Thesis entails that a piece of advice
purports to provide the subject with both a first-and second-order reason
12

to comply with the advice. First, insofar as a piece of advice is intended to
replace the subject’s own assessment of the applicable reasons in her delibera-
tion, it must, as was true of authoritative directives, provide the same kind of
first-order reason for doing something that her own assessment of the reasons
would provide. Second, insofar as a piece of advice is intended to replace the
subject’s assessment of the reasons and to preclude her from acting on her
own assessment of the reasons, it purports to provide a second-order reason
for the subject not to act on her own assessment of the applicable reasons.
Further, as is true of authoritative directives, the first-and second-order
reasons that a piece of advice purports to provide are content-independent.
To begin, NJT as it pertains to advice implies that whatever reasons a piece
of advice purports to provide are content-independent. Insofar as seeking or
taking a piece of advice is justified, whether morally or prudentially, on the
strength of the advisor’s superior expertise with respect to determining what
the applicable reasons require, the piece of advice is justified on the basis of
its source, and not its content. This means that whatever reasons a piece of
advice provides, they purport to be content-independent. To begin, the rel-
evant first-order reasons are content-independent: if, on the one hand, the
advice is to do p, it purports to provide a first-order reason to do p; if, on the
other, it is not to do p, it purports to provide a first-order reason not to do p.
Further, the second-order reasons are also content-independent: regardless of
whether the advisor recommends doing p or not doing p, the advice aims to
preclude a subject from acting on her own assessment of the applicable first-
order reasons.
Advice and authority have another conceptual feature in common. The
Preemption Thesis implies that the content of an authoritative directive must
be identifiable by the subject without her having to weigh the applicable
reasons for herself; that is, the Preemption Thesis implies the following thesis:
The Identification Thesis: It must always be possible for a subject to identify the
existence and content of an authoritative directive without her having to assess for
herself what the balance of applicable first-order reasons dictates that she should do.
If the Preemption Thesis is true, then an authoritative directive must be ca-

pable of replacing the subject’s own assessment of the applicable reasons in
her deliberations about what to do. But a directive that cannot be identified
by the subject without having to assess the applicable reasons is incapable
of replacing the subject’s judgments about the balance of reasons because
the subject must assess those reasons in order to ascertain the content of the
directive—and hence ends up acting on her own assessment, rather than
the putative authority’s, of the relevant reasons. For example, if an authority
directs the subject simply to do the right thing, the subject must figure out for
herself what the right thing to do is in order to determine what the directive
requires. If she then acts on that judgment, she is acting on her own assess-
ment of what the right thing to do is rather than on the assessment of the
authority.
Insofar as advice cannot be distinguished from authority on the basis of
the Preemption Thesis, the Dependence Thesis, and NJT, something like the
Identification Thesis would also have to be true of advice—and it is. The
conceptual point of advice is for the advisor to make a suggestion grounded
in the advisor’s assessment of the balance of reasons that can replace in the
advisee’s deliberations her own assessment of what the applicable reasons dic-
tate that she should do. But a piece of advice that has content that cannot be
identified by the advisee without having to assess for herself what the appli-
cable reasons require her to do cannot replace the advisee’s own assessment of
the applicable reasons in her deliberations about what to do.
Consider the suggestion expressed by the sentence “you should do what
you think is best.” One could, I suppose, argue that someone who offers this
thought as a piece of advice has merely given bad advice, but this seems mis-
taken. If I ask someone what she thinks I should do and she responds with
this sentence, I would interpret it as expressing a refusal to give me advice—
perhaps for the reason that she did not want to put herself in the position of
having given me bad advice on which I acted to my detriment. If this analysis
is correct, then the content of a piece of advice, like that of an authoritative
directive, must be identifiable by the subject without requiring her to assess
the balance of reasons that antecedently apply to her.
What ultimately succeeds in distinguishing a piece of advice from an au-
thoritative directive, of course, is that the latter is a directive and is hence
mandatory in character. Something that is plausibly characterized as a di-
rective purports to bind the subject thereby making the prescribed behavior
non-optional in some relevant sense; something that is merely advice does
not purport to make the suggested behavior non-optional in this or, for that
matter, any other ordinary sense of the word.
Law purports to make certain acts mandatory, binding, and non-optional;
and this is why it makes sense to think that law is authoritative in some sense
that must harmonize with a plausible account of authority. Insofar as the
valid norms of law incorporate the minimum content of natural law, they are
incorporating not only the content of the relevant norms but also their man-
datory character. Insofar as the rule of recognition imposes duties on officials,
it makes certain behaviors mandatory in the way that authoritative directives
purport to do. Insofar as the legal powers created by valid legal norms and
recognition norms define, at the very least, obligations of non-interference on
the part of other persons, they purport to make certain behaviors mandatory.
14

Law seeks to, when necessary, impose its will on subjects—and this is why a
conceptual account of law must be consistent with a plausible account of the
nature and justification of practical authority.
6. Practical authority and the possibility of moral criteria

of validity: The Arguments from Authority
There is nothing in the most general features of practical authority, which
are discussed above in the first four sections of this chapter, that is obviously
inconsistent with the possibility of moral criteria of validity. At this level of
generality, authority merely purports, under a claim of right, to impose its
will on subjects by telling them what they must do in a way that, in some
sense, binds them. An authoritative directive, then, purports to make cer-
tain behaviors non-optional in a manner that provides subjects with a novel
content-independent reason contrived to persuade otherwise disinclined sub-
jects to do as the directive requires. There is little reason, on the basis of these
general considerations, to think that the directives of an institutional norma-
tive system with moral criteria of validity could not perform these general
functions of practical authority.
At first blush, it seems easy to think of an institutional normative system
with moral criteria of validity that is authoritative with respect to its subjects.
Suppose, for example, that the following propositions are true: (1) the God
of classical theism exists; (2) God manufactures an objective morality through
the social act of issuing directives in the form of the Ten Commandments;
and (3) God punishes those who do not conform their behavior, for the most
part, to the Ten Commandments with an eternity of torment unmatched in
suffering by anything that can otherwise be experienced. Suppose, further,
that the vast majority of people believe that (1), (2), and (3) are true and, in
consequence, generally conform their behavior to the Ten Commandments—
enough to make it possible for people to peacefully live together in a so-
ciety, as well as for them to cooperate with one another for mutual prudential
advantage.
It seems clear, as a matter of ordinary usage, that God satisfies the most
general existence conditions for being a practical authority and that God’s
commands are authoritative with respect to God’s subjects. God purports,
under a claim of right deriving from God’s status as the ultimate sovereign
of the universe, to tell people what they must do in a manner that provides
reasons for doing what God commands: the threat of an eternity of torment
is about as good a reason, if one believes such things are true, as one could
The arguments from authority 115
possibly have for complying with any directive.8 Further, these reasons are
content-independent in the sense that had God commanded subjects to do
otherwise, the prospect of eternal torment would give subjects an equally
strong first-order reason to do otherwise and an equally persuasive second-
order reason not to act on their own assessment of the reasons. Finally, each
of the Ten Commandments purports to make certain acts non-optional
and binding on subjects: the threat of an eternity of torment for violating a
commandment makes an act about as non-optional and binding as any act
could be.
It would appear, at first blush, that the institutional normative system
described above satisfies the minimum conditions for a legal system and hence
is properly characterized as a system of law. To begin, God (who serves as the
only official of the system) seems to be practicing a rule of recognition that
requires that God recognize, apply, and enforce as valid norms of the system
only those norms that reproduce the content of the moral norms defined by
the Ten Commandments; the Ten Commandments thus define moral con-
straints on the content of valid law and hence define moral criteria of validity.
Further, insofar as subjects generally comply with the Ten Commandments,
which express the minimum content of natural law, the valid norms of the
system are as efficacious as the norms of any legal system need to be.
There is nothing in the conceptual features of practical authority discussed
in the first four sections that would preclude characterizing the institutional
normative system described above as a legal system—despite the fact that it
seems to incorporate moral criteria of validity. If this system is conceptually
disqualified from being a legal system on the ground that it has moral criteria
of validity inconsistent with law’s being authoritative, it would have to be on
the strength of the theses that flesh out Raz’s service conception of authority.
As will be seen below, the idea is that an institutional normative system with
moral criteria of validity is incapable of serving its subjects by mediating in
the right sort of way between them and the reasons that antecedently apply
to them.
8
As described above, an eternity of torment is not the punishment for any one sin; it is the
punishment for living a life that does not “for the most part” conform to the Ten Commandments.
Nonetheless, a subject never knows what “for the most part” amounts to in God’s judgment, so any
violation of a commandment might be sufficient to ensure an eternity of torment for the disobedient
subject. Clearly, a system in which (1) the overall quality of one’s life is judged by its conformity
to the Ten Commandments and (2) one is subject to eternal torment for insufficient conformity to
those norms can provide a strong, if not conclusive, reason for complying in every instance with
the Ten Commandments. Every sin, on this conception, matters with respect to what one’s ultimate
fate will feel like.
16

6.1 Law’s claim of legitimate authority: The general strategy
of the Arguments from Authority
Raz argues that the conceptual possibility of moral criteria of validity is incon-
sistent with the nature and justification of practical authority. On this general
line of argument, any institutional normative system with moral criteria of
validity lacks the kind of authority that any institutional normative system
must have, as a conceptual matter, to count conceptually as a legal system.
This does not show that it is conceptually impossible for there to be an insti-
tutional normative system with moral criteria of validity; what it shows is that
any such system lacks a conceptually necessary feature of a legal system and thus
does not count as a system of law. But this line of reasoning is compatible with
the conceptual possibility of institutional normative systems of other kinds
that have moral criteria of institutional validity. For example, as far as this
line of argument goes, a chess association could have a normative system with
moral criteria of validity; that system just would not have practical authority.
The claim here is merely that any type of institutional normative system that
is authoritative cannot have moral criteria of validity because a system with
moral criteria of validity cannot be authoritative. Since legal systems are au-
thoritative or claim to be such, they cannot have moral criteria of validity.
If such reasoning is sound, then it would imply that the institutional nor-
mative system that validates all and only norms reproducing the content of
the mandatory moral norms defined by the Ten Commandments is not prop-
erly characterized as a legal system. Since it incorporates moral criteria of
validity, it cannot be authoritative because it lacks a conceptually necessary
feature of practical authority. At best, on such reasoning, this institutional
normative system would be properly characterized as pre-legal despite de-
fining a structure of norms that is perfectly isomorphic to that of anything
properly characterized as a legal system.
Raz gives, in essence, three different arguments intended to show the con-
ceptual impossibility of a legal system with moral criteria of validity, but each
depends on the following thesis:
The Authority Thesis: Every conceptually possible legal system claims legitimate
authority.
According to the Authority Thesis, then, an institutional normative system
that does not claim legitimate authority is lacking a conceptually necessary
feature of a legal system and hence cannot be a system of law.
The Authority Thesis implies that, as a conceptual matter, law must be
capable of mediating between law’s subjects and the right reasons that apply
to them. To begin, on Raz’s view, if it is a conceptual truth that law claims
legitimate authority, then it must be conceptually possible for a system of
law to actually have legitimate authority: “If the claim to authority is part
of the nature of law, then whatever else the law is it must be capable of
possessing authority” (EPD 215). But, further, to be capable of possessing
authority, on Raz’s view, the law must be capable of serving its subjects by
“mediat[ing] between [them] and the right reasons that apply to them”
(EPD 214).
Raz’s service conception of practical authority grounds three main lines
of argument against inclusive positivism but all rely on the idea that an
institutional normative system with moral criteria of validity is incapable
of instantiating legitimate authority and hence lacks a conceptually nec-
essary feature of a legal system. The directives of a legitimate authority,
on Raz’s view, (1) must, under the Preemption Thesis, be able to replace
the judgments of subjects of what the balance of reasons requires them to
do; (2) must, under NJT, be such that subjects are more likely to comply
with the requirements of right reason by following the authority’s judg-
ment of what right reason requires than by following their own judgments;
and (3) must express the authority’s view about what specifically subjects
ought to do.
The basic strategy of these arguments against the Incorporation Thesis
is to show that the directives of an institutional normative system with
moral criteria of validity cannot be law because they cannot satisfy any of
the three necessary conditions for legitimate authority above. The first ar-
gument attempts to show that the directives of an institutional normative
system with moral criteria of validity cannot satisfy (1) because a subject
cannot identify the content of those directives without reading her own
assessments of the balance of reasons into it. The second attempts to show
that the directives of an institutional normative system with moral criteria
of validity cannot satisfy (2) because subjects must follow their own judg-
ments, rather than those of the authority, with respect to what they should
do. The third attempts to show that the directives of an institutional norma-
tive system with moral criteria of validity cannot satisfy (3) because subjects
can identify the content of those directives only by reading into them their
own views about what they should do according to right reason. Since, on
each of these lines of argument, an institutional normative system with
moral criteria of validity is incapable of instantiating legitimate authority,
it lacks a conceptually necessary feature of a legal system and cannot be a
system of law.
18

6.2 The Authority Thesis, the Preemption Thesis, and
inclusive positivism
Raz argues that the Preemption Thesis and Authority Thesis are inconsistent
with inclusive positivism. The argument is straightforward. Since the concep-
tual point of authority is to provide directives that can replace the subject’s
own assessment of the balance of reasons in her deliberations about what she
should do, it must always be possible for a subject to determine what an au-
thoritative directive requires of her without having to assess those reasons for
herself; that is, the Preemption Thesis implies the Identification Thesis. Any
directive the content of which cannot be identified by the subject without
assessing the balance of reasons that antecedently apply to her is conceptually
disqualified, under the Identification Thesis, as being authoritative.
The problem this poses for inclusive positivism is as follows. The
Identification Thesis implies that the content of the law must be identifiable
by the subject without having to assess the balance of reasons with respect to
what she should do. But a subject cannot identify the content of a norm valid
in virtue of moral merit without assessing the underlying reasons with respect
to what she should do. To determine what the law requires of a subject under
a recognition rule that validates only enacted norms consistent with some
mandatory moral norms, the subject must be able to identify the content of
the relevant mandatory moral norms that ultimately determines or constrains
what the law requires of her. But a subject cannot determine the validity of,
say, an enacted norm prohibiting the killing of “innocent persons” without
assessing the balance of reasons with respect to what she ought to do, as a
matter of morality; in particular, a subject must assess the balance of appli-
cable reasons to ascertain who counts as an innocent person under the law.
This implies, on Raz’s view, that a rule of recognition that incorporates moral
criteria of validity cannot be legitimately authoritative and hence cannot give
rise to law because the law qua authority is supposed to settle for subjects
which persons would count as “innocent” under the relevant moral norm.
6.3 The Authority Thesis, NJT, and inclusive positivism

The second line of argument against inclusive positivism is grounded in the
Authority Thesis and NJT. NJT asserts that authority is morally legitimate
only if the subject is more likely to do what right reason requires if she follows
the authority’s judgment about what right reason requires than if she follows
her own judgment about what right reason requires. But an institutional
normative system with moral criteria of validity is not capable of being le-
gitimate under NJT because a subject cannot identify the content of valid
norms without assessing for herself what right reason requires of her. Insofar
as the subject must identify the content of a valid norm by assessing for her-
self what right reason requires, she is, of necessity, following her own judgment
about what right reason requires, and not the judgment of the authority.
Consider an institutional normative system with a rule of recognition that
validates all and only mandatory moral norms, and suppose that a subject
must decide whether an institutionally valid norm requires her to do p. To
determine whether the valid norm requires that she does p, she must assess
for herself whether the balance of reasons requires that she do p. If she is
motivated to do what she thinks the valid norm requires with respect to doing
p, then she will do what she decides the balance of reasons requires with re-
spect to doing p: if she decides that the norm requires her to do p, then she
will do p; if she decides that the norm requires her not to do p, she will re-
frain from doing p. Since, either way, she cannot avoid following her own
judgment of what the balance of reasons requires, it is not possible for her
to follow the authority’s judgment of what the balance of reasons requires
and hence it is not possible for her to better comply with right reason by fol-
lowing the authority’s judgment of what the balance of reasons requires than
by following her own judgment. For this reason, an institutional normative
system with moral criteria of validity cannot instantiate legitimate authority
and hence is not properly characterized as a legal system
6.4 The directives of practical authority as expressing its view

about what ought to be done
Raz argues that it is a necessary condition for a norm to be legitimately au-
thoritative that “it must be, or be presented as, someone’s view on what the
subjects ought to do” (EPD 221). To be capable of legitimate authority, the
directives of a normative system must express the authority’s view about which
specific act the balance of reasons dictates that subjects must perform. Since
a subject can determine what is required by a norm in an inclusive system
only by deciding for herself what the balance of reasons requires that she do,
she is simply identifying her own view about what she ought to do: again,
if she decides that the norm requires her to do p, then she will do p; if she
decides that the norm requires her not to do p, she will refrain from doing
p. Since an inclusive rule of recognition requires her to decide for herself, for
each act p, whether the balance of applicable reasons requires her to do p or
refrain from doing p, it is incapable of expressing the authority’s view about
what specific act she ought to perform and is hence incapable of being a le-
gitimate authority. Since it is a conceptual truth that law claims legitimate
120

authority and hence is capable of legitimate authority, an institutional nor-
mative system grounded in an inclusive rule of recognition is conceptually
incapable of being a legal system.
7. Looking ahead: Evaluating the Arguments

from Authority
The next three chapters evaluate the Arguments from Authority. Chapters 5
and 6 are concerned with evaluating two theses on which each of the
Arguments from Authority depends. Chapter 5 is concerned with chal-
lenging the Authority Thesis that it is part of the very nature of law that it
claims legitimate authority; I argue that there are no conceptually necessary
features of a legal system that are properly interpreted as expressing a claim of
legitimate authority. Chapter 6 is concerned with the Identification Thesis;
in particular, I argue that there is nothing in the ordinary linguistic or legal
practices defining the use of the relevant concept-terms that entails that the
nature of our concept of authority requires that the content of an authorita-
tive directive be identifiable without recourse to the dependent reasons it is
supposed to reflect, balance, and preempt. Insofar as the Identification Thesis
is a logical consequence of the Preemption Thesis, the argument of Chapter 6
also purports to refute, albeit indirectly, the Preemption Thesis.
Finally, Chapter 7 attempts to show that the Arguments from Authority
incorrectly presuppose that the rule of recognition must be capable of inform-
ing subjects of what their non-official obligations are under valid law. The
problem is that there is no reason to think that a rule that does not govern
non-official behavior must be capable of informing subjects of what other
rules that do govern non-official behavior require with respect to such be-
havior; the rule of recognition defines standards that govern only the law-
making, adjudicative, and enforcement acts of officials of the system. Insofar
as the Arguments from Authority presuppose that non-officials must be able
to learn of their obligations from the rule of recognition, the arguments fail.
5
Law’s Claim of Legitimate Authority
At the foundation of each of the Arguments from Authority is the idea that
it is a conceptual truth that law claims legitimate authority (the Authority
Thesis) and hence that it is a conceptual truth that law is capable of being le-
gitimate. The problem with inclusive positivism, on these arguments, is that
an institutional normative system with moral criteria of validity is incapable
of being legitimately authoritative because the subject in such a system cannot
determine what the relevant norms require without assessing for herself what
she ought to do according to right reason. Since such a system cannot hence
vindicate law’s claim of legitimate authority, it cannot be a system of law.
There are three related problems that arise in connection with an institu-
tional normative system with moral criteria of validity. First, insofar as the
subject of such a system cannot determine what she must do under the norms
without assessing the balance of reasons for herself, those norms are incapable
of replacing her own judgments in her deliberations about what she ought to
do according to right reason. Second, insofar as the subject of such a system
cannot determine what she must do under the norms without assessing for
herself what right reason requires, it is impossible for her to follow the author-
ity’s assessments of what right reason requires and is hence impossible for her
to better comply with right reason by following the authority’s judgment
than by following her own. Third, insofar as the subject of such a system
cannot determine what she ought to do under the norms without deciding
what she thinks she ought to do according to right reason, the norms cannot
express the authority’s view about what specifically she ought to do according
to right reason. In each case, an institutional normative system with moral
criteria of validity is incapable of being legitimately authoritative because it
cannot perform some function that something must be able to perform to be
authoritative.
Strictly speaking, the claim that causes the problem for inclusive positivism
is the claim that it is a conceptually necessary condition for something to
count as a system of law that it be capable of legitimate authority—and not
the Authority Thesis. The Authority Thesis figures into the argument only
University Press.
12
122 Law’s Claim of Legitimate Authority

insofar as it entails that something must be metaphysically capable of being
legitimate to count as a legal system. But since it seems obvious, on any plau-
sible conceptual theory of law, that something must be capable of legitimate
authority to count as a legal system, the Authority Thesis, contra what is usu-
ally thought, plays no essential role in the Arguments from Authority.
Even so, it is important to evaluate the Authority Thesis because it purports
to state a fact about the nature of a legal system. One point of this volume
is to determine whether the arguments against the Incorporation Thesis suc-
ceed, but its overriding point is to facilitate a deeper understanding of the
conceptual relationships between law and morality. Insofar as the notion of
legitimate authority is a concept with morally normative content, it is impor-
tant to understand whether it is part of law’s nature that it claims legitimate
authority. For this reason, this chapter attempts to determine whether the
Authority Thesis is true.
1. Understanding the content of a claim

of legitimate authority
The Arguments from Authority, as Raz articulates them, rest on the Authority
Thesis, according to which it is a conceptual truth that “every legal system
claims that it possesses legitimate authority.”1 Insofar as it is conceptual
truth that an authority is legitimate if and only if its directives give rise to
content-independent moral obligations to obey, the Authority Thesis implies
that every conceptually possible legal system claims that its subjects have a
content-independent moral obligation to obey legal directives in virtue of
having the status of law. It is hence a conceptual truth, according to the
Authority Thesis, that law claims that every behavior that is legally obligatory
is also morally obligatory in virtue of being legally obligatory.
The Authority Thesis asserts a stronger claim than might initially appear.
Insofar as it asserts that a claim of legitimate authority is a conceptually nec-
essary feature of a legal system, it implies that Hart’s minimum conditions
for the existence of a legal system do not exhaust the existence conditions
for a legal system. If the Authority Thesis is true, it is also a conceptually
necessary condition for an institutional normative system to count as a legal
system that some official or institution does something that expresses a claim
of legitimacy.
1
Joseph Raz, “Authority, Law and Morality,” The Monist (1985), reprinted in Raz (ed.), Ethics in
the Public Domain (Oxford: Oxford University Press, 1994), 215. Hereinafter EPD.
Can law make claims? 123
2. Can law make claims? Two possible interpretations

While the Authority Thesis might seem plausible at first glance, it is impor-
tant to note that it depends crucially on a metaphysical assumption that is far
from intuitive—namely, the assumption that legal systems can, like human
beings, perform acts that make claims. The problem is that it is not clear that
legal systems possess the requisite characteristics that are metaphysically or
conceptually necessary to make claims. If they do not possess these character-
istics and hence cannot make claims, then the Authority Thesis is false.
The worry here arises because of the kind of thing that a legal system is by
nature. A legal system has many pieces, which include, most conspicuously,
laws; every conceptually possible legal system has a rule of recognition that
defines the criteria of validity for that system as well as a set of legal norms
that are valid under the rule of recognition. But a legal system also seems to
include, as constituents, the various institutions defined by those criteria of
validity; these institutions would have to include, as a conceptual matter,
judicial institutions but they also would arguably have to include distinct
legislative, executive, and enforcement institutions. Each of these institutions
is composed, in part, of officials who have duties and powers defined by cer-
tain legal norms, which might include those defining validity criteria and
first-order  norms.
A legal system, as a conceptual matter, appears thus to be an aggregate of
these norms, institutions, and officials. If there are no norms, there is no legal
system. If the system contains none of the conceptually necessary institutions,
there is no legal system. If there are no officials, there is no legal system. Since
each of these is a conceptually necessary element of a legal system and a legal
system is not identical with any one of them, a legal system must, as an on-
tological matter, be identical with a set that includes as members all of these
objects. Whatever else it might be, then, a legal system is a composite object.
The question is whether and how something that has the defining quali-
ties of a legal system could do something accurately characterized as making
claims. Although it is certainly reasonable to question whether whatever
claims a legal system makes include a claim of legitimate authority, the initial
question is a metaphysical one: is it conceptually possible for a legal system to
make claims in a sense that would imply the crucial claim that a legal system
must be capable of instantiating legitimate authority?
It is clear, of course, that some of the conceptually necessary pieces of a legal
system can make claims. Individual officials can, and sometimes do, make
claims about the content of something that is or ought to be law: legislators
124

can, and sometimes do, make claims about the costs or benefits that would
result from the enactment of some proposed bill; judges can, and frequently
do, make claims about the duties of the parties in legal disputes. Anyone
who serves as an official of any kind can make claims partly in virtue of
being able to perform various linguistic acts that make statements which bear
truth-value.
But some of these conceptually necessary pieces do not seem capable of
making claims—at least not on any ordinary understanding of that notion.
While it is true that a legal norm can connote the kind of propositional con-
tent that might be expressed by a claim, a legal norm is just not the kind of
thing capable of making a claim in any relevant sense of the phrase.
Notice that the idea that L expresses a claim does not imply the idea that L
thereby makes a claim. Sentences express claims but do not make claims; the
sentence “capital punishment is unjust” expresses a claim that capital pun-
ishment is unjust, but it does not make that claim. Speakers make claims by
using sentences that express the content of those claims; I use the sentence
“capital punishment is unjust” to make the claim that capital punishment is
unjust. I can make claims, in part, because I can perform linguistic acts that
express the propositional content I wish to assert as a claim; in contrast, legal
norms can express propositional content but cannot perform acts of any kind.
Claim-making of the sort asserted by the Authority Thesis, if intended in any
sense that conforms to ordinary usage, is an act and as such can be accom-
plished only by things—such as judges, presidents, and legislators—capable
of doing something that would count as an act.
But even if it is plausible to think that the propositions expressed by legal
norms constitute claims made by those norms, it would still be problematic
to think that legal norms make any claims that would warrant attributing a
claim of legitimate authority to the legal system. The problem is that legal
norms do not appear to express any content properly construed as implying a
claim of legitimate authority. For example, the content expressed by the legal
norm prohibiting murder is that murder is prohibited by the system of legal
norms. Assuming this content expresses a claim, it should be clear that the
claim that murder is legally prohibited does not imply the claim that murder
is morally prohibited—much less does it imply the claim that the norm is
legitimately authoritative or that murder is morally prohibited in virtue of
its being prohibited by a legal norm. There is simply nothing in the norm
expressing that murder is legally prohibited that entails any claim, on the
ordinary usages that are the topic of this volume, about the moral properties
of either the prohibited act, the legal norm, or the legal system to which the
norm belongs. The law prohibiting murder lacks the right kind of content
even to express such a claim.
Nor would it help to aggregate all the legal norms of a system as a means
of deriving such a claim. It is simply not plausible to think that all the legal
norms of any conceptually possible legal system, taken together, must imply
a claim of legitimate authority. All that the legal norms of any legal system,
taken together, must express is that the various acts they govern are either
legally required, legally prohibited, or legally permitted: if A1, A2, A3, . . . ,
An is a complete list of the acts regulated by the legal norms of the system,
then all that the set of a system’s legal norms, taken together, must express is
that acts A1, A2, A3, . . . , An are either legally required, legally prohibited,
or legally permitted in that system. But the claim that acts A1, A2, A3, . . . ,
An are either legally required, legally prohibited, or legally permitted nei-
ther asserts nor implies that those acts are either morally required, morally
prohibited, or morally permitted—much less that they are morally required/
prohibited/permitted in virtue of being legally required/prohibited/permit-
ted, which is what would be needed to express a claim of legitimate authority.
If a claim of legitimate authority is properly attributed to every conceptual
legal system, it cannot be derived from anything necessarily expressed by the
system’s legal norms.
But these observations are not germane because the Authority Thesis is a
thesis about what is claimed by a legal system, and no one of these elements
constitutes a legal system. Indeed, to the extent that a legal system is properly
conceived to be a unified institution, it must be characterized as a set contain-
ing all of these various elements—and a set is a non-propositional abstract ob-
ject, rather than a concrete physical object. But even if we omit the language
of sets and abstract objects, it should be clear that the legal system could not
be a physical object. While one can see, hear, smell, touch, or taste various
elements of a legal system, it should be clear that one cannot see, hear, smell,
taste, touch, or taste the legal system itself.
The idea that a non-physical abstract object of this kind can claim anything
is metaphysically problematic. To begin, it is clear that a non-physical non-
propositional object, unlike physical and propositional objects, cannot even
express claims. A sentence is a propositional object that can express claims in
virtue of being an aggregate of various linguistic entities that bear meaning.
But a set that consists of sets of norms, sets of institutions, and sets of per-
sons is not an aggregate of purely linguistic entities; sets of non-physical
propositional objects (i.e. norms) are aggregated with sets of non-physical
non-propositional objects (i.e. institutions) and sets of physical persons (i.e.
officials) to comprise the set constituting a legal system. Even if it were true
that every entity contained in one of the sets making up a legal system can
express or make claims, there is no way to aggregate those sets to express a
proposition.
126

Further, entities that can make claims in any literal sense can do so only in
virtue of possessing certain properties. What makes it possible for us to make
claims is that we are rational intentional agents who can perform acts using
language to communicate content to other similarly abled agents. And each
of these properties is necessary for being able to make claims in the relevant
ordinary sense. The ability to make claims in this sense requires linguistic
competence. Rationality is necessary because the ability to use language pre-
supposes the ability to grasp and use concepts. Intentionality is necessary be-
cause claims always express content about something. The absence of any one
of these properties is enough to render a being incapable of making claims.
Since legal systems, qua abstract object, lack all of these capacities, it is not
conceptually possible for legal systems to make claims in this sense.
The idea that a legal system can either make or express claims involves a
conceptual confusion that rises to the level of a category mistake. As Oxford
Online Dictionary defines the notion, a category mistake makes “[t]‌he error
of assigning to something a quality or action which can only properly be
assigned to things of another category.”2 The idea that the object denoted by
the numeral “2” can make or express claims involves a conceptual confusion
that rises to the level of a category mistake; someone who sincerely makes
this claim must be confused either about the nature of a number or about
the nature of making a claim. But it seems no less conceptually confused to
think that a legal system can make claims in some ordinary literal sense than
it is to think that the object denoted by the numeral “2” can make claims in
the same sense.
The confusion might be somewhat more difficult to see in the former case,
but there is nothing in the notion of a category mistake that implies that cat-
egory mistakes are easy to discern. It is clear that a plant can neither make nor
express claims, but it is not clear whether the idea that a plant can make or
express a claim involves a conceptual confusion that rises to the level of a cat-
egory mistake.3 People frequently attribute certain interests and even mental
states to plants that they do not have. One might say on observing a plant’s
brown leaves that the plant “wants” to be watered. This kind of claim expresses
2
“Category mistake,” Oxford Online Dictionary; available at: https://en.oxforddictionaries.com/
definition/category_mistake.
3
Indeed, it is not clear whether it even involves a conceptual confusion. One might be clear on
all the relevant notions but think that some plants have the relevant capacities to make claims. That
idea might be false, but it is not obviously false in virtue of being conceptually confused; there is
nothing in the linguistic conventions for using the term “plant” that entails that plants necessarily
lack these capacities. To determine this, one must know more about the world than just the def-
inition of “plant” or the underlying philosophical presuppositions; one would also have to know
presumably contingent facts about the empirical properties of everything that counts as a plant.
an error that might rise to the level of a category mistake but whether it does is
not clear. Whether it does will depend on conceptual considerations having to
do with what counts as having a want, as well as metaphysical considerations
having to do with what properties something must have to have wants. If, for
this reason, the idea that a plant can make or express claims involves a category
mistake, that mistake is not obvious.
One might argue that it is no more obvious that the Authority Thesis
involves a category mistake than it is that a plant can make or express claims
does. Perhaps the Authority Thesis involves no greater conceptual confusion
than the idea that a plant can make or express claims—although a plant is
a living physical object while a legal system is a non-propositional abstract
object; however, it should be clear that a legal system is simply not the kind
of thing that can make or express claims in any ordinary sense of the terms.
Whether it involves a category mistake or not, the Authority Thesis is ob-
viously false if construed as asserting that legal systems claim legitimate au-
thority in any literal sense of the term “claim” as ordinary usage defines it.
There are two potential responses here. First, one might think that the ob-
jection incorrectly presupposes a literal interpretation of the Authority Thesis.
On this line of defense, the Authority Thesis should be interpreted metaphor-
ically to attribute claims to the legal system in the same way that claims are
frequently attributed to other types of abstract objects in ordinary and legal
practice. The law treats for example, corporations as persons who can sue and
be sued because it is useful to do so; to facilitate its aims, the law will attribute
claims made by corporate agents to the abstract corporation itself. As long as
we respect its intuitive boundaries, there is no decisive reason to reject the use
of a device that attributes, in some non-literal metaphorical sense, claims to
an abstract object like a legal system.
There are, of course, no obvious objections to using a metaphorical device
to describe some element of a legal system and doing so might well yield serv-
iceable insights about the nature of law. One could think of the legal system
in metaphorical terms as an author to help us to understand the nature of
interpretation as it functions or should function in judicial practices. There is
no plausible reason to think that attributing to law an ability to make claims
in some metaphorical sense might not disclose something of value about the
nature of law.
The problem is that, assuming the Authority Thesis is needed to derive the
idea that law must be capable of instantiating authority, it would not imply
that idea unless law is capable of making claims in some literal sense. Raz
argues that if the law sincerely claims authority, then it must be capable of
instantiating authority because the legal practices that construct our concept
of authority cannot be conceptually confused about that notion. But this
128

argument is sound only if the relevant sense of claim-making is literal: it is
one thing to infer that the law must be capable of being legitimate from the
idea that it can and does claim in a literal sense to be legitimate; it is quite
another to infer that law must be capable of being legitimate from the idea
that law sincerely claims in some purely metaphorical sense to be legitimate.
Either way, a claim of legitimate authority can sensibly be attributed to
law only on the strength of the behaviors of its officials. These behaviors
might involve deliberately expressive behaviors that make claims in a literal
or metaphorical sense, but they need not involve such behaviors. There is
no a priori reason to rule out inferences that might be made on the basis of
non-expressive behaviors; we are still dealing, after all, with beings capable
of making and expressing claims. But whatever behaviors are identified as
forming the basis for such claims, they must occur in every conceptually pos-
sible legal system since the Authority Thesis attributes such claims to law as a
matter of conceptual necessity.
The idea that there are certain behavioral requirements included in the
existence conditions for a legal system is not problematic. After all, on a posi-
tivist view, the existence conditions for a legal system are at least partly defined
by two behavioral requirements. First, it is a necessary condition for the exist-
ence of a legal system in S that officials in S behave in a manner that amounts
to their practicing a social rule of recognition that governs the promulgation
and adjudication of law in S. Second, it is a necessary condition for the exist-
ence of a legal system in S that people in S behave in a manner that generally
conforms to the norms validated by the rule of recognition in S.
Although each condition states a behavioral requirement, it is plausible to
think that the existence conditions for a legal system are exhausted by such
requirements because a legal system is a relational object that obtains in some
society relative to some population of people whose behavior it regulates. If,
on the one hand, there is no one in S who practices a rule of recognition gov-
erning the promulgation and adjudication of law in S, then there is no legal
system in S because an institutional normative system of the appropriate kind
does not exist in S. If, on the other hand, there is such a rule of recognition
but people in S do not obey the norms validated by the rule of recognition,
then there is no legal system in S because the institutions that the rule pur-
ports to bring into existence in S do not stand in the appropriate relation to
people in S.
While there can and must be behavioral requirements among the existence
conditions for a legal system, the evidence for thinking that the officials must
behave in a way that implies a claim of legitimacy must be satisfactory. In par-
ticular, there must be a non-question-begging reason for denying the status
of law to an institution in which the norms valid under a rule of recognition
Deriving law’s claim of legitimacy 129
practiced by officials efficaciously guide non-official behavior but in which
officials exhibit none of the requisite behavioral elements that imply a claim
of legitimate authority.
I argue below that Raz’s argument for the Authority Thesis fails to meet
this evidentiary burden. I argue that, on the assumption that a legal system is
the kind of thing to which a claim can plausibly be attributed, the behaviors
he cites as supporting the Authority Thesis are neither conceptually necessary
conditions for the existence of a legal system nor expressive of a claim of legit-
imate authority. This does not show that the Authority Thesis is false because
it does not rule out the possibility that there are other conceptually necessary
features of a legal system that imply a claim of legitimate authority. At most,
the argument below shows that even if we assume that legal systems are ca-
pable of making claims in some relevant sense, we have no reason to believe
that they claim legitimate authority.
3. Deriving law’s claim of legitimacy from the beliefs and

claims of officials
Raz attributes a claim of legitimate authority to the legal system on the
strength of certain beliefs and behaviors on the part of officials:
The claims the law makes for itself are evident from the language it adopts and from
the opinions expressed by its spokesmen, i.e. by the institutions of the law. The law’s
claim to authority is manifested by the fact that legal institutions are officially des-
ignated as “authorities,” by the fact that they regard themselves as having the right
to impose obligations on their subjects, by their claims that their subjects owe them
allegiance, and that their subjects ought to obey the law as it requires to be obeyed
(i.e. in all cases except those in which some legal doctrine justifies breach of duty).
Even a bad law, is the inevitable official doctrine, should be obeyed for as long as it is
in force, while lawful action is taken to try and bring about its amendment or repeal
(EPD 215–16).
There are five practices, then, that imply a claim of legitimate authority that
should be attributed to law: (1) the use in the law of such terms as “right” and
“duty”; (2) the official designation of legal institutions as “authorities”; (3) the
claims of officials that subjects “owe” allegiance to officials; (4) the claims of
officials that subjects “ought to obey the law”; and (5) the beliefs of officials
that they have a right to impose obligations on their subjects.
Raz does not explicitly attribute these claims and practices to officials
of the legal system; rather, he attributes them to unspecified “institutions
of law.” But if one doubts that a legal system is metaphysically capable of
130

making claims because of its abstract character, one will also doubt that a legal
institution is metaphysically capable of making claims because of its abstract
character. A legal institution is no more a physical object than a legal system
is; a legislature or judiciary is made up, in part, of physical objects but it is
not a physical object. One can see, hear, smell, touch, and taste judges and
legislators, but one cannot see, hear, smell, touch, or taste a legislative or ju-
dicial institution.
Even so, we can make sense of the idea that the beliefs, claims, and prac-
tices of officials can assert or imply a claim of legitimacy that can metaphori-
cally be attributed to the legal system. Officials are coherently conceived of as
representatives of the institutions in which they serve and hence as authorized
to speak on behalf the other members of these institutions.4 Insofar as the
beliefs, claims, and practices of those officials who represent the legal system
and speak on behalf of other officials of the system imply a claim of legiti-
macy, it can metaphorically be attributed to the system itself.
4. Do the beliefs and claims of officials imply a claim

of legitimacy?
The Authority Thesis asserts that every conceptually possible legal system
includes features that make a claim of legitimate authority. Insofar as a legal
system’s claim of legitimate authority can be derived from (or otherwise
explained only in terms of ) some set of beliefs, claims, or practices instanti-
ated by officials in the system, it follows that officials in any conceptually pos-
sible legal system must instantiate some subset of the relevant beliefs, claims,
or practices.
This does not mean that the relevant beliefs, claims, or practices that imply
the claim to legitimate authority in one conceptually possible legal system
must be exactly the same as those that imply that claim in every other concep-
tually possible legal system. It might be that there are a number of properties,
P1, P2, . . . , Pn of which instantiation of any one is sufficient to imply such a
claim. If so, it is consistent with the Authority Thesis that the claim of legit-
imate authority is fully constituted in one conceptually possible legal system
by the instantiation of P1 and in another conceptually possible legal system
4
The idea that one can speak on behalf of an abstract object in a literal sense is no less metaphys-
ically problematic than the idea that an abstract object can make claims. One can speak on behalf
of only things that have some capacity to speak; as non-propositional non-physical objects, legal
institutions and systems cannot speak for themselves in any literal sense. Strictly speaking, a repre-
sentative of a group can speak on behalf of only other members of the group.
The beliefs and claims of officials 131
by the instantiation of P2. The Authority Thesis is agnostic with respect to
whether the claim of legitimate authority is constituted by the instantiation
of exactly the same beliefs, claims, and practices in every conceptually pos-
sible legal system.
I argue below that the relevant features identified by Raz as implying a
claim of legitimate authority do not warrant accepting the Authority Thesis.
First, none of these features implies a claim of legitimate authority because
the relevant claims, beliefs, and practices imply nothing at all about the moral
properties of the system. Second, there is no reason to think that official acts
in every conceptually possible legal system must exhibit one of these features.
While it might be true that a claim of legitimate authority can justifiably be
attributed to some conceptually possible legal systems, Raz has not given any
reason to believe that such a claim must be attributed to all conceptually pos-
sible legal systems.
4.1 The use of the language of rights and duties

The first feature of law cited by Raz as implying a claim of legitimate au-
thority is the “language it adopts” (EPD 215). Like moral norms, legal
norms define protections that we characterize as “rights,” as well as require-
ments that we characterize as “duties.” The U.S. Constitution, for example,
includes fifteen instances of the term “right,” which include its appearance
in the First, Second, Fourth, Sixth, Seventh, and Ninth Amendments, and
the law of torts prototypically imputes to potential defendants a “duty” of
reasonable care.
The use of such language is clearly not coincidental. Legal requirements
purport to bind subjects in a manner resembling that in which moral require-
ments bind moral agents. In law and in morality, the existence of something
properly characterized as a “duty” entails that the relevant behavior is required
or non-optional. In both cases, the norm states a requirement that is exclu-
sionary in the sense that a violation of a norm cannot be justified by the sub-
ject’s prudential interests. It is in this sense that a duty of either kind “binds”
subjects of the directive creating the duty.
This is a conceptual truth about duties. It is part of the very nature of any
kind of duty that it is binding on a subject in the following sense: subjects
must do what they have a duty to do regardless of what their preferences
might be; failure to satisfy a duty constitutes a wrong for which the subject
is properly held accountable. If giving to charity is morally required, then
someone who fails to do so has committed a moral wrong for which she
can properly be held accountable under the norms of morality. Similarly, if
132

paying taxes is legally required, then someone who fails to do so has com-
mitted a legal wrong for which she can properly be held accountable under
the law.
This is the hallmark of a duty in the relevant sense: an agent A has a duty to
do p under a class of rules R if and only if A’s non-performance of p is a wrong
under R for which she can properly be held accountable under the rules of R.
Given that legal norms purport to make specified acts required in a sense that
subjects persons to liability for non-performance, it is quite natural that legal
practice would adopt the language of duties in expressing its requirements.
It is equally natural that legal practice would adopt the language of rights
in expressing certain requirements of law insofar as the notion of a right is
conceptually related to the notion of a duty. Duties correlate, as a conceptual
matter, with rights in at least this respect: it is a conceptual truth that P has a
right against Q with respect to act a only if Q owes a duty to P with respect
to a. It might be that Q’s owing a duty to P with respect to a is not conceptu-
ally sufficient for P’s having a right against Q, but it is conceptually necessary.
Either way, it is perfectly natural—and, indeed, predictable—that the law
would use the language of rights to describe protections that function the way
rights do under moral norms.
No more than these similarities are needed to explain why the law uses the
same language to express its requirements that we use to express the require-
ments of morality. The use in law of such terms as “obligation,” “duty,” and
“right” signifies no more than a collective recognition on the part of lawmak-
ers, judges, and ordinary citizens that the requirements of law define prohi-
bitions that are exclusionary in the same sense that moral prohibitions are
exclusionary and that are hence applied and enforced in a way that expresses
that exclusionary character.
Law-talk is not unique in co-opting the normative language of morality.
We commonly speak of obligations and rights in a host of non-moral con-
texts that have nothing to do with the law. In addition to legal obligations and
rights, we speak of social, institutional, and cultural obligations and rights;
the idea that the very use of these terms warrants attributing a belief to speak-
ers that the relevant norms give rise to obligations and rights that are moral in
character would imply that members of any club, no matter how trifling its
purpose, believe that the club rules defining their obligations and rights also
give rise to moral obligations and moral rights. Perhaps that is true in most
or even every existing case; however, this is clearly not necessarily true, as a
metaphysical, conceptual, or psychological matter.
Further, there is nothing in either ordinary linguistic or legal practices
that would warrant the attribution of such a belief to officials on the ground
that law-talk uses some of the same language as morality-talk. As discussed
in Chapter 2, courts recognize the possibility that a mandatory legal norm
might be sufficiently wicked that it gives rise to no moral obligation to obey;
thus, there is nothing in the lexical meanings of the terms that would entail
that laws must be just and hence that legal obligations and rights give rise to
moral obligations and rights. Indeed, there is nothing conceptually confused
in the idea that what the law requires might be so wicked that citizens have
a moral obligation to disobey it. As far as ordinary usage is concerned, there
is nothing in the meaning of the relevant terms that would warrant the attri-
bution of such a belief to people simply in virtue of using the relevant terms.
The idea that one can attribute a belief to officials that legal obligations
and legal rights define or give rise to moral obligations and moral rights on
just the strength of their use in both contexts of “obligations” and “rights”
is problematic not only because there is nothing in the meanings of the two
terms that entails any moral content; it is also problematic in virtue of attrib-
uting an error theory of law to every speaker of the language. As discussed
in Chapter 2, the strong natural law claim that it is a necessary condition
for something to count as a valid legal norm that its content be morally just
implies an error theory of law.5 Since the claim that legal obligations and legal
rights necessarily give rise to moral obligations and moral rights presupposes
this strong natural law claim, it also implies an error theory of law.
In the absence of a sociological survey that shows that most speakers of
the language hold this view, it is uncharitable to attribute an error theory of
law to speakers of the language simply on the basis of practicing a linguistic
convention regarding the use of the terms “obligation” and “right.” Without
such sociological verification, the use of such language provides no support
for the Authority Thesis.
5
For what it is worth, the idea that some claim C implies an error theory and the idea that C
commits a category mistake are logically independent. To begin, the idea that C commits a category
mistake does not entail the idea that C implies an error theory; the idea that law claims legitimate
authority seems to commit a category mistake but does not imply an error theory of anything.
Conversely, the claim that C implies an error theory does not imply that C commits a category mis-
take. The claim that morality is not objective implies an error theory of morality in the sense that
most of our folk moral judgments would be false if that claim is true, but it does not imply that our
folk moral judgments involve a category mistake. The problem is not that there is some conceptual
confusion with respect to our use of the relevant terms; it is rather that those terms do not pick out
some moral property that exists in the universe independently of our own beliefs and practices.
134

4.2 Officials’ belief that they have a right to impose
obligations
The idea that a claim of legitimate authority can be attributed to a legal system
on the strength of the language used by law enjoys a prima facie plausibility
because the claim of legal authority and the language used by the law share
an important property. A claim of legitimate authority is a public claim that
could, in principle, be derived from other public features of law. Insofar as it is
a conceptually necessary condition of law that it is publicly promulgated and
hence is a public entity, the language in which it is expressed is also public.
It is quite natural to think that a public claim of legitimate authority can be
inferred from the public language in which law is framed.
One somewhat surprising property from which Raz attempts to derive a
public claim of legitimate authority seems to be a private mental state un-
connected with some public act on the part of officials. As Raz puts the
point, law’s claim of legitimate authority “is manifested [in part] by the fact
that . . . [officials] regard themselves as having the right to impose obligations
on their subjects” (EPD 215–16; emphasis added).6
The property of regarding oneself as having a right of some kind to impose
obligations on citizens is an attitude constituted by purely internal beliefs
and dispositions. To say that officials regard themselves as having such a right
expresses no more than that (1) they believe that they have such a right; and
(2) they have certain normative dispositions to behave in certain ways. Beliefs
and dispositions are purely mental states that might be expressed in acts,
but they need not be. I have many beliefs and dispositions that I have never
acted on because I have never found myself in circumstances in which both
would be activated in a way that would culminate in some act on my part. On
this line of reasoning, then, a claim of legitimate authority can reasonably be
attributed to the legal system on the strength of beliefs and dispositions that
need not be acted on or publicly expressed.
6
There are a couple of points about the character of the relevant rights and obligations that
should be made. First, the right officials believe they have to impose obligations on their subjects
must be construed as a moral right. A belief on the part of officials that they have a merely legal right
to impose such obligations would not provide any support for attributing a claim of moral legiti-
macy to law. Second, the character of the obligation they believe they have a right to impose is most
plausibly construed as legal; thus construed, the idea is that they believe they have a moral right to
impose legal obligations on subjects. It is not just that those are the only obligations that are neces-
sarily imposed by a mandatory legal norm; it is also that it is implausible to think that anyone has a
moral right to impose moral obligations on subjects. When someone comes into my home, she has
a moral obligation not to start smashing my property. This is an obligation that I can waive, but it
misdescribes the situation to think that I have imposed that obligation on her in the exercise of some
general moral right to impose moral obligations.
One might think that, given the nature of what officials are and must
do, it is not conceptually possible for someone to be properly characterized
as an official without sometimes doing something official that expresses the
relevant attitudes. Officials are officials, on this line of argument, in virtue of
what they do in contexts in which law is recognized, applied, and enforced;
a judge is only a judge insofar as she entertains and adjudicates disputes be-
tween citizens.
But this is not obviously true. It is arguably possible to conceive of a legal
system, like a “society of angels,” in which judges never have to decide cases
because citizens never have the kinds of conflicts that give rise to legal dis-
putes requiring judicial adjudication. In this world, the citizens and the law
would have some properties that they lack in the world of our experience: (1)
the law would have to be sufficiently clear that disagreement among citizens
about what the law requires never occurs; (2) citizens would have to be suffi-
ciently motivated always to do what the law requires; and (3) it would have to
be clear to all citizens that the behavior of every other citizen always complies
with the law. In such a system, it would still be necessary to designate certain
persons as judges just in case there are disputes. Those persons designated as
judges would have to have the relevant beliefs and dispositions but would
never have to do anything that expressed those beliefs and dispositions.
If such a state of affairs is conceptually possible, then the idea would have
to be that the instantiation by officials of just the relevant mental states
(which include the belief and associated dispositions) warrants attributing a
claim of legitimate authority to the legal system. On this line of reasoning,
then, the fact that officials have the belief that they have a right to impose
legal obligations on subjects along with the associated dispositions implies a
claim of legitimate authority.
There are two problems with this view. First, there is nothing in the claim
that officials believe they have a right to impose legal obligations on citizens
that would warrant attributing a claim of legitimate authority to the legal
system. Unstated beliefs and dispositions are just not the kinds of thing that
can express or make claims, either by themselves or together. Public claims
can reliably be attributed to people only on the strength of public behaviors
that are partly expressive in character.
Second, there is no reason to think that officials typically believe that they
have a moral right to impose legal obligations on citizens. If the idea is that
such a belief can be inferred from the public acts of officials, it is problematic.
The most that can reliably be inferred from the adjudicative acts of judges is
that they believe they have a legal right to impose legal obligations on citi-
zens because their powers and duties as officials are defined by the law itself.
Judges, for example, have authority to decide legal disputes because the law
136

affords them that authority. There is nothing in any of this that would entail
either that they have a moral right to impose legal obligations on citizens or
that they believe they have such a moral right.
It might be true that many judges believe they have some sort of moral
right to impose legal obligations on citizens because they believe the system
is morally legitimate, but it cannot be assumed that judges, as a matter of
conceptual necessity, have such a belief. The idea that a judge might impose
legal obligations on subjects without believing that she is morally justified in
doing so is neither conceptually incoherent nor psychologically impossible.
Whether all or even most officials in a legal system have this belief is an em-
pirical issue that can be resolved only by sociological investigation.
This is not to suggest that one can never reliably infer a mental state express-
ing a motive from an act of some kind, but there will always be more than
one such mental state that can plausibly be inferred. For example, it might be
true that one can usually infer that a person is hungry from the fact that she
is eating. But there are other mental states that would explain why a person is
eating: a person might be eating when she is not hungry to maintain a healthy
weight or to make sure that she is getting proper nutrition. To suggest that
there are difficulties in inferring beliefs and dispositions from public acts is
not to claim that beliefs and dispositions can never reliably be inferred from
public acts; it is merely to claim that such inferences, even in the most favor-
able of contexts, are subject to some uncertainty.
Inferring an official’s beliefs and dispositions from the fact that she is apply-
ing or promulgating some norm as law is subject to no less uncertainty than
inferring someone’s beliefs and dispositions from the fact that she is eating.
Assuming the act of applying or promulgating a norm R as law entails support
for recognizing it as law, a judge or legislator might support recognizing R as
law because she believes that R is morally legitimate, or she might support R
because she believes that she needs to in order to maintain some desired level
of support from her constituency. There are just as many beliefs and sets of
related dispositions compatible with an official’s applying or promulgating
a norm as law as there are beliefs and sets of dispositions compatible with
someone’s eating something.
Accordingly, there are two problems with the argument here. First, there is
no reason to think that it is a conceptual truth that officials in a legal system
regard themselves as having a moral right to impose legal obligations on their
subjects. Second, even if officials necessarily regard themselves as having such
a moral right, a public claim to legitimate authority cannot be inferred from
the purely private unexpressed mental states of officials. The claim, then, that
officials regard themselves as having a right to impose obligations on subjects
provides no support for the Authority Thesis.
4.3 Officials’ claim that subjects owe allegiance
A third line of reasoning attempts to derive a claim of legitimate authority
from claims on the part of officials that subjects owe a duty of “allegiance.”
As Raz puts it, “law’s claim to authority is manifested by . . . [officials’] claims
that their subjects owe them allegiance” (EPD 215–16; emphasis added).7
The idea here is that one can derive law’s claim of legitimate authority from a
claim made by officials that subjects owe a moral obligation of loyalty.
At the outset, it is important to note that a moral obligation of loyalty
could not be owed to a legal system. It is far more plausible to think that we
owe moral obligations to plants than it is to think we owe moral obligations
to legal systems because it is far more plausible to think that plants have inter-
ests that can be harmed than it is to think that legal systems have interests that
can be harmed. On the one hand, it seems reasonable to think that one can
harm a plant because plants, as living things, are plausibly thought of as hav-
ing interests in getting enough sun and water. On the other hand, it is hard
to see how one could harm a legal system because non-living, non-sentient
non-physical artifacts do not have interests of any kind.
Consider the claim that legal systems have an interest in justice. It is as
silly to claim that legal systems have an interest in justice as it is to claim that
legal systems want to be just; they are simply the wrong kind of thing to have
interests or desires. Plants are non-sentient objects that lack any state that
could plausibly be characterized as a desire, but they are living things that are
reasonably thought to have interests and this is why it makes sense to think
that they can be harmed instead of merely damaged. It might be true that
legal systems and legal institutions, as non-living non-sentient non-physical
abstract artifacts, can be damaged but they can no more be harmed than an
automobile can.
The problem with the idea that these objects can be owed moral obliga-
tions is not just that it is false; it is rather that it rises to the level of a category
mistake if ordinary usage is the touchstone. The lexical meaning of “category
mistake,” as Oxford Online Dictionary reports it, is particularly telling in this
regard:
7
The relevant obligation would have to be a moral obligation. First, there is no law in legal sys-
tems like that of the U.S. that requires “allegiance” of subjects and hence no law that would impute
a legal obligation of allegiance. Second, the claim that subjects owe a legal obligation of allegiance
would not assert or imply anything that would support attributing a claim of morally legitimate au-
thority to the legal system.
138

Category Mistake: (n) The error of assigning to something a quality or action which
can only properly be assigned to things of another category, for example treating ab-
stract concepts as though they had a physical location.8
The example in the definition is directly on point: the idea that an abstract
object can be owed a moral obligation seems as much and as clearly a category
mistake as “treating abstract concepts as though they had a physical location.”
It is no less conceptually confused to think that such things could be owed
a moral obligation or could be harmed as it is to think they have a physical
location.
The same is true of the claim that citizens owe a moral duty of loyalty to
the legal system, and this poses a problem for the idea that this claim can
plausibly be attributed to officials in every conceptually possible legal system.
Perhaps one can argue without conspicuous absurdity that officials of every
conceptually possible legal system make a claim that is false. But it is implau-
sible to suppose that it is a conceptual truth that officials make a claim that
expresses a category mistake because this would imply that an institutional
normative system in which officials do not commit this category mistake is
not properly characterized as a legal system for that reason.
The only remotely plausible interpretation of the view is that officials claim
that subjects owe a moral duty of loyalty to the officials, and not that they
claim that subjects owe a duty of loyalty to the legal system itself. Officials
have interests that can be harmed in virtue of being persons who are alive,
sentient, and rational—unlike legal systems, which are abstract objects like
the object denoted by the numeral “2” and are hence the wrong kind of thing
to which interests can be coherently attributed.
It is certainly possible for officials to believe or claim that citizens owe them
a moral duty of loyalty since that idea is neither an obvious category mistake
nor obviously false in all conceptually possible legal systems, but it is prob-
lematic to assume that all or some officials in a legal system necessarily believe
or claim this. First, it might be plausible to think that citizens owe a moral
duty of loyalty to some officials, but it seems clearly implausible to think that
citizens owe a moral duty of loyalty to everyone who is properly characterized
as an official of the legal system. It is difficult to make sense of what would be
required by a moral duty of loyalty to the President; it is considerably more
difficult to make sense of what would be required by a moral duty of loyalty
to officials at the Internal Revenue Service.
8
“Category mistake,” Oxford Online Dictionary; available at: https://en.oxforddictionaries.com/
definition/category_mistake.
Second, it is implausible to think that officials in democratic systems like
that of the U.S. necessarily believe or claim that citizens owe them a moral
duty of loyalty. In a democracy, citizens are not conceived of as subjects of the
officials who recognize, apply, and enforce law; officials are instead conceived
of as public servants who owe a duty to their constituents to serve them.
While there might be some plausible metaphorical construction of the claim
that citizens should be loyal to their country, such a construction would nei-
ther assert nor imply that citizens owe a moral duty of loyalty to officials. If
the claim that citizens owe a moral duty of loyalty to officials is true in some
conceptually possible legal systems, it is not obviously true in democratic
legal systems in which officials are conceived of as public servants.
It might be true that some officials in the U.S. are confused about this
and believe that citizens owe them a moral duty of loyalty. It might even be
true, although even this sufficiently strains credulity to require empirical ev-
idence, that most officials in every existing legal system believe such a claim.
But it is simply false that all or some officials in every conceptually possible
legal system believe that citizens owe them a moral duty of loyalty; we can
easily conceive of a legal system in which officials reject such self-important
nonsense.
Similarly, it might be true that some officials in the U.S. do something
that asserts or implies that citizens owe them a moral duty of loyalty. It might
even be true that most officials in every existing legal system do something
that asserts or implies that citizens owe them a moral duty of loyalty. But it
is false that some or all officials in every conceptually possible legal system
do something that asserts or implies that citizens owe them a moral duty of
loyalty; there is simply nothing in what must be included in the job descrip-
tion of a judge or legislator that would require doing anything that asserts or
implies such a claim.
There is nothing in ordinary patterns of usage that even suggests that it is a
conceptually necessary condition for the existence of a legal system that even
one official believes that citizens owe officials a duty of loyalty. There is simply
nothing in the nature of law, in the nature of being an official, in the content
of the typical job descriptions for legislators or judges, or in the psychology
of human beings that entails that someone who serves as an official of a legal
system must believe—much less claim—that citizens owe them a duty of
loyalty. The idea, then, that officials claim that citizens owe them a duty of
allegiance provides no support for the Authority Thesis.
140

4.4 Designation of officials as “authorities”
Raz argues that a claim of legitimate authority can be derived from the fact
that “legal institutions are officially designated as ‘authorities’ ” (EPD 215;
emphasis added). The idea seems to be that a claim of legitimacy can be
attributed to law on the ground that the law itself characterizes officials as
authorities; after all, the only way that legal institutions could be officially
designated as “authorities” is through some official act that has the status of
law. If the law itself specifically designates these institutions as “authorities”
and that term connotes that these institutions are morally legitimate, then
law must be understood to claim that it is legitimate.
There are two problems with this line of argument. First, and most con-
spicuously, the claim that “legal institutions are officially designated as
‘authorities’ ” is not true of even all existing legal systems. Nowhere in the
U.S. Constitution are executives, legislators, or judges officially designated
as “authorities.” Likewise, if the Supreme Court ever refers to Congress as
a body of legislative “authorities,” such references would not amount to an
official designation of Congress as an authority; what would be involved in
designating something as an authority is not clear, but it is clear that only
something with a certain kind of authority that the Court lacks could of-
ficially designate Congress as an authority. If it is not true of the U.S. legal
system, then it is not true of all conceptually possible legal systems.
Second, even if legal institutions have been officially designated as “author-
ities,” that fact is not enough to warrant attributing a claim of legitimate
authority to a legal system. The problem is that, as far as ordinary usage is
concerned, one can have practical authority over a class of persons without
its being morally justified or its being believed to be justified. A police officer,
P, might have something properly characterized as practical authority over
another person, Q, that is neither morally justified or is believed by either P
or Q to be morally justified.
There is nothing obviously problematic from the standpoint of our or-
dinary linguistic conventions with claiming that Nazi courts had authority
over those who appeared before them. Although we might also clarify that
such authority was not legitimate, there is nothing in these ordinary linguistic
conventions that would require that we characterize such authority as merely
de facto. Most competent native speakers of English would not understand
what I meant if I said that the Nazi courts had merely de facto authority over
citizens. But if they did understand the statement, they would have inferred
what I meant from the context of the utterance based on shared assumptions
about the wickedness of the Nazi system. The inference would be grounded
in certain facts and shared moral judgments about the Nazis—and not in or-
dinary linguistic conventions for using the term “authority.”
Further, there is nothing in our linguistic conventions that allows for the
use of the term “authority” to describe an illegitimate institution only if cit-
izens generally believe that the institution is legitimate. As far as ordinary
usage is concerned, Nazi courts are properly characterized as having authority
over subjects regardless of whether the subjects believed that those courts or
judges had morally legitimate authority over them.
The problem to which this gives rise is that one cannot infer a claim of legit-
imate authority from the claim that legal institutions are officially designated
as “authorities.” If what people mean when they use the term “authority” to
refer to someone does not, on our linguistic conventions, connote that the
person’s authority is morally legitimate, then a claim of morally legitimate au-
thority cannot validly be inferred from just the use of the term “authority.”
Since our linguistic conventions allow use of the term “authority” to refer to
authorities that are neither morally legitimate nor generally believed to be so,
there is nothing in the “designation” of legal institutions as authorities that
would imply a belief or assertion that their authority is morally legitimate.
To summarize: a claim of legitimate authority must be grounded in beliefs
or acts that imply that the law is morally justified; and given the ordinary
meanings of the term, there is nothing in the use of “authority” that would as-
sert or imply that the authority of legal institutions is morally justified. Since,
as far as our linguistic conventions are concerned, there can be authority that
is legal in character without being morally justified, any official designation of
legal institutions as “authorities” entails no more than that these institutions
have legal authority under the laws of the system; such a designation would
not imply that what legal authority those institutions have is morally justified
and hence would not warrant attributing a claim of legitimate authority to
the legal system.
4.5 Officials’ claim that subjects have an obligation to obey

the law
Perhaps the most promising ground for the Authority Thesis is the idea that
officials claim that subjects owe an obligation to obey the law:
The law’s claim to authority is manifested by . . . [officials’] claims . . . that their sub-
jects ought to obey the law as it requires to be obeyed (i.e., in all cases except those in
which some legal doctrine justifies breach of duty). Even a bad law, is the inevitable
official doctrine, should be obeyed for as long as it is in force, while lawful action is
taken to try to bring about its amendment or repeal (EPD 215–16; emphasis added).
142

The idea is that officials of the legal system express its claim of legitimate au-
thority by insisting that subjects have a content-independent obligation to
comply with valid legal norms regardless of what they require until those laws
are repealed or are otherwise properly changed.
It is reasonable to think that officials believe that citizens have a content-
independent obligation to comply with the law. Judges are charged with recog-
nizing, applying, and enforcing as law anything that has the appropriate source
in legislative or judicial acts regardless of content. Legislators enact law with the
expectation that it will be enforced by judges in an exclusionary manner regardless
of content. Further, official acts of any kind must satisfy the rules of law defining
the duties and powers of officials regardless of the content of those rules. It is
hard to understand how someone could competently serve as an official without
believing that subjects have a content-independent obligation to comply with
the law; while someone who lacked such a belief might not be conceptually dis-
qualified from serving as an official, it is hard to see how they could perform their
duties in a competent manner without believing something like this.
But there is no reason to think that officials must believe that the relevant ob-
ligation is moral in character; it would also be sufficient for them to competently
perform their functions that they believe the only relevant obligations are legal.
It is true that judges could not apply and enforce whatever norms happen to
have the status of law against citizens regardless of their content unless they be-
lieve (or act as if they believe) that those norms give rise to content-independent
obligation of some kind; to apply and enforce a norm in an exclusionary manner
is to treat it as if it gives rise to a content-independent obligation on the part of
the subject to comply. Since legal norms give rise to legal obligations that are
content-independent in character, all that is necessary to explain how judges can
competently perform their duties is that they believe (or act as if they believe)
that law gives rise to content-independent legal obligations.
The problem here is not just that it is unnecessary to impute to judges a
belief that the law gives rise to content-independent moral obligations to ex-
plain how they can competently perform their duties; it is rather that such a
belief would contribute nothing to explaining how judges can competently
do what they do. To carry out their legal duties, judges must apply and en-
force legal norms against subjects in an exclusionary manner that is also de-
fined by law. Courts are not charged, legally or otherwise, with deciding
disputes under norms of morality; indeed, courts are frequently barred by
law from deciding purely moral disputes and from deciding legal disputes
by recourse to moral norms that do not have the status of law.9 The idea
9
As discussed in Chapter 2, Federal Rule of Civil Procedure 11(b)(2) provides as follows: “By
presenting to the court a pleading, written motion, or other paper—whether by signing, filing,
that officials believe that law gives rise to content-independent moral obliga-
tions cannot explain a judge’s ability to competently perform her function
because a judge’s job description requires only that she decide disputes that
arise under legal norms.
Perhaps the thought is that officials could not, as a psychological matter,
do things that have the effect of coercively restricting the freedom of others
in a manner that subjects them to the threat of incarceration unless they
believed that legal norms give rise to a content-independent moral obligation
to comply. The idea here is that, for example, a judge could not, as a psycho-
logical matter, issue an order incarcerating a defendant unless she believed
she was morally justified in doing so. On this line of reasoning, the only ob-
vious moral justification for incarcerating a defendant who failed to comply
with the law is that she violated a content-independent moral obligation to
comply.
There is little reason to suppose that officials could not, as a psycholog-
ical matter, perform such functions unless they believe that citizens have a
content-independent moral obligation to comply with the law. Even the
most morally virtuous people are psychologically capable of doing things
they believe are morally wrong; no one is free of wrongdoing. While one
can hope that most officials are motivated by moral considerations, we know
from experience that officials are as capable as anyone else of acting on selfish
and even wicked motivations. It seems clear that it is not a conceptually or
psychologically necessary condition for officials to be capable of performing
their official functions that they believe law gives rise to content-independent
moral obligations to comply.
Nor is there any reason to think that there is anything in any official’s job
description that cannot be done unless she publicly claims that the law gives
rise to content-independent moral obligations to comply. It is true that some
officials, such as judges, must occasionally in the course of performing their
job responsibilities make claims about what citizens are legally obligated to
do; however, there is no obvious reason to think that there is anything that
submitting, or later advocating it—an attorney or unrepresented party certifies that to the best
of the person’s knowledge, information, and belief, formed after an inquiry reasonable under the
circumstances . . . the claims, defenses, and other legal contentions are warranted by existing law or
by a nonfrivolous argument for extending, modifying, or reversing existing law or for establishing
new law.” In essence, Rule 11(b)(2) prohibits bringing an action simply on the ground that it is
necessary to remedy moral injustice in the law. It is true that an action can be properly grounded in
a good-faith argument to change the law for moral reasons; but that presupposes that there can be
laws that are morally problematic and that the court has legal authority to change the law for that
reason. See Note 16, Chapter 2.
14

any official must do that would imply a claim about what citizens are mor-
ally obligated to do. Legislators are not typically required to do anything that
would imply any claim about what citizens are obligated to do—either legally
or morally. Judges must, in the course of deciding such disputes, make claims
about the legal obligations of the parties, but they need not make claims
about what morality requires. A judge can apply and enforce a valid legal
norm against a citizen without doing anything that asserts or entails a claim
about what citizens are morally obligated to do.
This argument for the Authority Thesis fails, then, because there is no
reason to think that it is a conceptual truth that even some officials must
do something that would entail a claim that citizens have a general content-
independent moral obligation to comply with the law. Since the only claims
any officials need make about the obligations of citizens concern only their
legal obligations under the valid laws of the system, the idea that officials
claim that “subjects ought to obey the law as it requires to be obeyed (i.e. in
all cases except those in which some legal doctrine justifies breach of duty)”
provides no support for the Authority Thesis.
4.6 The claims taken together

One might concede that none of the features discussed above warrants attrib-
uting a claim of legitimate authority to law but argue that some combination
of them does. For example, one might think that (1) the instantiation of any
three of the above five features is sufficient to warrant attributing a claim of
legitimate authority to an institutional normative system and that (2) every
conceptually possible legal system exhibits at least three of these features. If
(1) and (2) are true, then it follows that every conceptually possible legal
system claims legitimate authority. Instantiation of each of the relevant fea-
tures would be conceptually necessary for a legal system to claim legitimate
authority, while instantiation of all of them would be conceptually sufficient
for a legal system to claim legitimate authority.
The problem is that we can conceive of something lacking all these features
that seems to be a paradigmatic instance of a legal system. Suppose there is
a society, S, that is as much like that of the U.S. except that all citizens in
S deny that law could be morally legitimate and hence deny that law could
give rise to a content-independent moral obligation to comply on the part
of subjects. Recognizing the need to keep the peace, they all agree to be gov-
erned by a normative system that authorizes coercive enforcement mechanisms
for violations of the system’s obligations and explicitly contract with one another
to conform to those obligations. The officials accept a conventional recognition
norm that provides recipes for recognizing, applying, and enforcing the
norms of the system, while citizens generally obey the valid norms of the
system, which incorporate the minimum content of natural law. But being
skeptics about the possibility of morally legitimate systems of law, officials
refrain from using the potentially misleading terms “authority,” “duty,” “ob-
ligation,” and “right,” relying instead on terms like “official,” “required,”
“mandatory,” “non-optional,” and “permitted” in recognizing, applying, and
enforcing, and citizens follow this practice. In addition, officials deny that
they are owed any duty of allegiance or loyalty. Despite such skepticism about
whether any system could be morally legitimate, these institutional norms are
just and justly administered.
Regardless of whether this imagined normative system is properly charac-
terized as “law,” it seems to be a morally legitimate system of regulation. All
citizens promise to comply with the norms of the system in exchange for the
promises of all others to do the same. The norms of the system are just, justly
administered, and protect citizens from morally culpable assaults on their
persons and property. If we further suppose that the system also contains
every norm that any legal system must contain to be morally legitimate, then
it seems to satisfy all the standards that are required for a legal system to count
as morally legitimate—assuming that their views about moral legitimacy are
false. Whether or not the system counts as one of law, it is indisputably a mor-
ally legitimate system of governance, if any is.
Further, the system in S seems to have all the features that are concep-
tually necessary for something to count as being authoritative. Citizens all
take the internal point of view toward the rule of recognition and the valid
norms of the system and hence regard the norms of the system as providing
content-independent exclusionary reasons for action. The norms tell citizens
what they must do; they are applied and enforced in an exclusionary manner
that reflects the exclusionary character of the reasons the citizens believe such
norms and orders provide. It is true that citizens refrain from using the term
“authority” to avoid a suggestion that the system is morally legitimate and
do not believe it is legitimate; but it is not a requirement for something to
count as an authority on any plausible conception of the term that citizens
use that particular word to describe things with authority. It seems clear that
the norms and officials do everything that something must do to be properly
characterized as an authority according to ordinary usage. If Nazi courts had
authority on this ordinary usage, then so does the system in S.
The existence conditions for a legal system are satisfied under any plausible
conceptual theory of law. From the standpoint of Hart’s minimum condi-
tions for the existence of a legal system, the imagined system is a legal system
for the following three reasons: (1) the officials of S adopt the internal point
146

of view toward the putative rule of recognition; (2) citizens of S generally
comply with the norms validated by the putative rule of recognition; and
(3) S incorporates the minimum content of natural law. But since, by hypoth
esis, all the norms are just and justly administered, the imagined system would
also count as one of law under the traditional interpretation of natural law
theory. If we suppose, further, that the system also incorporates those moral
principles that justify the application of coercive enforcement mechanisms
and shows the existing institutional history in its best moral light, it would
satisfy Dworkin’s third theory of law. Finally, assume that everyone refers to
the norms of the system as “law.”
What plausible, non-question-begging reason could there be to deny that
this system of rules is a legal system? The only salient difference between the
legal system of the U.S. and the institutional normative system in S is that
officials in the U.S. typically believe (assuming they believe this) that the
system is legitimate, while officials in S lack such a belief—a difference that
seems irrelevant to the characterization of the latter as a system of law. All
the major institutions are there: the rule of recognition defines institutions
that promulgate, apply, and enforce law. All the citizens of S accept the deter-
minations of the officials as reasons for action. The rules of S are obeyed to
precisely the same extent as the rules of law are in the U.S. It is plausible to
characterize S as having a legal system because it has all the pieces necessary
for efficacious state regulation of behavior even though there is nothing that
could be construed as entailing an institutional claim to legitimate authority.
Raz anticipates the possibility of constructing a normative system some-
thing like this one as a counterexample to the Authority Thesis:
[T]‌ry to imagine a situation in which the political authorities of a country do not
claim that the inhabitants are bound to obey them, but in which the population does
acquiesce in their rule. We are to imagine courts imprisoning people without finding
them guilty of any offense; damages are ordered, but no one has a duty to pay them.
The legislature never claims to impose duties of care or of contribution to common
services. It is not merely ordinary people who are not subjected to duties by the leg-
islature: courts, policemen, civil servants, and other public officials are not subjected
by it to any duties in the exercise of their official functions.10
Raz’s response misses the mark. The issue is not whether there could be an
authoritative system that arbitrarily imposes penalties and damages without
there being a violation of something that counts as a norm stating a require-
ment that is applied and enforced in an exclusionary fashion; there could not
10
Joseph Raz, The Morality of Freedom (Oxford: Clarendon Press, 1986); emphasis added.
Hereinafter MF.
be. But that is not what is going on in S. People who are punished are being
punished for violating a norm, but officials and citizens simply refrain from
using “offense” to refer to the violation. People who are being ordered to pay
damages are required to pay them under a norm that makes people liable for
negligently causing injury.
The issue is, rather, whether it is conceptually possible for there to be a
legal system that does not claim legitimate authority in virtue of using cer-
tain language and hence whether it is a conceptually possible for there to be a
legal system that does not use terms like “offense” and duty” in characterizing
the violations and requirements of norms; that is a different issue. It seems
clear from the example above that there could be. There is nothing either in
our conceptual practices or our moral views that would require the use of
“offense” in characterizing violations of the law that are subject to punitive
liability or the use of “duty” to describe requirements of the law that will be
enforced against violating subjects in an exclusionary manner; it is enough
that we can speak in terms of “violations” and “requirements” and that these
terms pick out notions that function in practical reason the way the notions
picked out by the terms “offense” and “duty” normally do.
On the strength of this foundation, Raz goes on to argue that the imag-
ined system lacks authority and is hence not properly characterized as a legal
system:
Two things stand out when contemplating a political system of this kind. First,
it is unlikely that any such society ever existed. Societies we know about are in-
variably subject to institutions claiming a right to bind their subjects, and when
they survive this is in part because at least some of their subjects accept their
claim. Secondly, if such a society were to exist we would not regard it as being
governed by authority. It is too unlike the political institutions we normally regard
as authorities (MF 27).
There are two problems here—one less and the other more important. The
less important problem is that the claim that “it is unlikely that any such so-
ciety ever existed” (MF 27) is an empirical claim and is hence not relevant;
what matters is whether it is conceptually possible for such a legal system to
exist, and not whether the existence of such a system is empirically probable.
The more important problem is that Raz’s argument asserts without evidence
that “we would not regard [such a society] as being governed by authority.”
To refute the idea that the system like the one constructed above is a system
of law, one must provide a plausible reason grounded in ordinary usage to be-
lieve that it is not properly characterized using the terms “law” or “authority.”
Raz’s argument fails to do so because it begs the question against the relevant
line of argument.
148

That said, there is nothing in the argument of this section that would en-
tail the falsity of the Authority Thesis. Assuming it is conceptually possible
for a legal system to make claims, it might be that there is some other con-
ceptually necessary feature of a legal system that warrants attributing to it a
claim of legitimate authority; if so, the Authority Thesis is true. What it does
show, however, is that the considerations typically adduced in support of the
Authority Thesis provide insufficient support for it. But if it could be shown,
as seems plausible, that these features are the only ones that imply a claim of
legitimate authority that can be attributed to the legal system as a whole, the
arguments of this section, like the argument of Section 1 if it is sound, would
be enough to falsify the Authority Thesis.
6
Authority, Moral Criteria of Validity, and
Conceptual Confusion
There are three Arguments from Authority. The first argues that the norms
of an institutional normative system S with moral criteria of validity cannot
replace the judgments of subjects about what right reason requires because
subjects cannot determine what its norms require without figuring out
for themselves what right reason requires; since S cannot hence satisfy the
Preemption Thesis, it cannot be legitimate. The second argues that subjects
in S cannot better comply with what right reason requires by following the
authority’s judgments about what right reason requires than by following their
own judgments because subjects can determine what its norms require only
by judging for themselves what right reason requires and hence can follow
only their own judgments; since S cannot hence satisfy Normal Justification
Thesis (NJT), it cannot be legitimate. The third argues that the norms of S
cannot express the authority’s specific view about what the subject should do
because the subject can determine what those norms express only by reading
her own views about what she should do into the norms; since the rules hence
cannot express what directives must express to be authoritative, S cannot be
legitimate.
Each of these arguments depends on the idea that our conceptual practices
with respect to the term “authority” entail that an institutional normative
system cannot be legitimate unless it is conceptually possible for a subject to
determine what its norms require without having to decide for herself what
she ought to do (the Identification Thesis). This thesis, together with the con-
ceptual truism that a legal system must be capable of instantiating legitimate
authority (the Instantiation Thesis), entails the Sources Thesis. Since (1) an
institutional normative system with moral criteria of validity cannot be le-
gitimate because its directives cannot be identified by the subject without
judging for herself what she ought to do and (2) only institutional normative
systems that can be legitimate are legal systems, an institutional normative
system with moral criteria of validity cannot be a legal system.
University Press.
150
150 Is Inclusive Positivism Conceptually Confused?

This chapter challenges the Identification Thesis. If our conceptual prac-
tices with respect to using the terms “law” and “authority” are determined by
what competent speakers typically believe and say about law and authority,
then those ordinary linguistic practices allow for the possibility of a legiti-
mate legal system with moral criteria of validity; our conceptual practices are
thus inconsistent with the Identification Thesis. As the Identification Thesis
expresses the core of the service conception of authority, the service concep-
tion is inconsistent with the concept of authority as it is defined by our con-
ceptual practices.
1. Identifying the content of our conceptual practices

The content of any concept plausibly characterized as ours is determined by
the ordinary linguistic practices that construct the corresponding concept-
term’s lexical meaning. While logical space contains all possible concepts
and hence includes concepts with which we are unfamiliar, our concepts are
picked out by the conventions that we converge in practicing to express cer-
tain content. The content of our concept of bachelor equates bachelorhood
with being an unmarried man because and only because we have converged
on practicing a convention that dictates that the term “bachelor” is correctly
used to refer only to unmarried men. Had we converged on practicing a con-
vention that dictates that the word “bachelor” refers only to plants, it would
not be true that all bachelors are unmarried men.
This is not to suggest that there is any ontological difference between con-
cepts that are ours and concepts that are not ours. If logical space contains
every piece of content that could be associated with some concept-term, it
includes the content of concepts that are not properly characterized as ours.
Our linguistic practices do not bring any concepts into existence. Our lin-
guistic practices determine which of these pre-existing concepts are associated
with our use of the relevant concept-term; if we had adopted a different con-
vention for using the term “law,” that term would pick out a different one of
those pre-existing concepts than the one that it currently picks out.
Our convergent linguistic practices define a mapping of the terms of our
language to the pieces of content that we use the terms to express; these pieces
of content define the only concepts that are properly characterized as ours.
Insofar as the project of conceptual analysis is concerned to identify the con-
tent of our concepts and the content of our concepts is determined by our
convergent linguistic practices, the contents of the relevant concepts of law
and authority are defined by those practices for using the concept-terms “law”
Identifying the content of our conceptual practices 151
and “authority.” Any conceptual claims about law or authority must hence be
evaluated on the basis of whether they conform to the linguistic conventions
we practice for using the terms “law” and “authority.”
Although the Arguments from Authority are conceptual and hence met-
aphysical in character, each relies on empirical claims about the content of
our linguistic conventions with respect to the use of the terms “law” and “au-
thority.” As discussed in Chapter 2, modest conceptual analysis is concerned
with identifying the deeper philosophical commitments presupposed by our
ordinary linguistic practices with respect to using the relevant concept-term.
The crucial claim is—and must be if the Arguments from Authority are prop-
erly construed as presupposing a modest approach to conceptual analysis—
that speakers of the language converge on practicing linguistic conventions
that ultimately preclude the application of the terms “law” and “authority” to
an institutional normative system with moral criteria of validity.
This is an empirical claim about the content of our linguistic conventions,
and it is stronger than might initially appear. Insofar as our ordinary lin-
guistic conventions determine the lexical meanings of the words “law” and
“authority,” each of the Arguments from Authority ultimately depends on the
claim that the lexical meanings of the terms “law” and “authority” preclude
their application to an institutional normative system with moral criteria of
validity.
Some caution is needed here. The claim that the lexical meaning of a
term precludes its application to something does not imply that the lexical
meaning is transparent with respect to the relevant application. As we have
seen, it is not clear whether the term “bachelor” properly applies to the Pope
even though dictionaries universally report its lexical meaning as “an unmar-
ried man.” The problem is that many competent speakers would decline to
characterize the Pope as a bachelor on the ground that he is not institution-
ally eligible for marriage; if most speakers practice a convention that entails
that only unmarried men institutionally eligible for marriage are bachelors,
then the lexical meaning of the term precludes characterizing the Pope as a
bachelor.
In this case, what is reported as the meaning of “bachelor” is not fully ac-
curate because the linguistic conventions that determine the lexical meaning
of the term “bachelor” do not entail that every unmarried man is a bachelor,
but that is of no importance here. What matters is that the issue of whether
the Pope is properly characterized as a bachelor according to the lexical mean-
ings is an empirical issue that requires sociological lexical analysis to resolve.
There are three possibilities here: (1) the term “bachelor” might, according
to the conventions that determine its lexical meaning, properly apply to the
Pope; (2) the term “bachelor” might not, according to these conventions,
152

properly apply to the Pope; or (3) the conventions that determine the lex-
ical meaning of “bachelor” might be indeterminate with respect to whether
the term “bachelor” properly applies to the Pope. In the case where the
majority of speakers apply “bachelor” to the Pope, the conventions dictate
that the Pope is properly characterized as a bachelor; in the case where the
majority of speakers reject the application of the term to the Pope, the
conventions dictate that the Pope is not properly characterized as a bach-
elor; and in the case where there is a significant split, the conventions are
indeterminate with respect to whether the Pope is properly characterized
as a bachelor.
Philosophical analysis is needed only when the lexical conventions are in-
determinate. Such an analysis in the case of the term “bachelor” purports to
identify common views about the nature of bachelorhood; in particular, it
would purport to determine whether the nature of bachelorhood is exhausted
by the compound property of being an unmarried man or whether instead
the nature of bachelor includes institutional eligibility for marriage. As claims
about the nature of a thing are metaphysical in character, an analysis that
purports to clarify the nature of bachelorhood is philosophical in character.
A philosophically grounded modest conceptual analysis attempts to explicate
the deeper philosophical commitments that ground the lexical convention
governing the use of “bachelor” in order to determine whether they resolve
the indeterminacy
The same is true of whether the lexical meanings of the terms “law”
and “authority” preclude the application of these terms to an institutional
normative system with moral criteria of validity. If most speakers use these
terms in a way that precludes their application to an institutional norma-
tive system with moral criteria of validity, then the lexical meanings are
determinate with respect to the issue and entail that there cannot be a
legal system with moral criteria of validity. If most speakers use the terms
in way that allows their application to an institutional normative system
with moral criteria of validity, then the lexical meanings are determinate
with respect to the issue and entail that there can be a legal system with
moral criteria of validity. If there is a significant split on the issue among
speakers, then the lexical meanings are indeterminate with respect to the
issue—and a philosophical analysis is needed to attempt to resolve the
indeterminacy.
A modest conceptual analysis of law and authority begins with the defini-
tions of the terms that are reported in dictionaries. One representative and
highly respected dictionary, Oxford Online Dictionary, reports their lexical
meanings as follows:
The Identification Thesis 153
Law: The system of rules which a particular country or community recognizes as
regulating the actions of its members and which it may enforce by the imposition of
penalties; and
Authority: The power or right to give orders, make decisions, and enforce obedience.
Notice that, assuming these reports are completely accurate, there can be
legal systems with moral criteria of validity since there is nothing in the lex-
ical meanings that precludes applying the terms “law” and “authority” to an
institutional normative system with moral criteria of validity. Just as there
was nothing in the dictionary definition of “bachelor” that precludes its ap-
plication to the Pope, there is nothing in the dictionary definitions of “law”
and “authority” that precludes their application to an institutional norma-
tive system with moral criteria of validity. If the reports above accurately and
completely state the application-conditions for the terms and the Arguments
from Authority are grounded in the ordinary lexical meanings of these terms,
then there is nothing in these meanings that would entail the falsity of the
Incorporation Thesis.
It might be, of course, that the reports of the lexical meaning of the terms
“law” and “authority” are not fully accurate because the relevant polling
failed to inquire about the relevant cases. Just as the dictionary definition of
“bachelor” would be inaccurate if the Pope is not a bachelor, the dictionary
definitions of “law” and “authority” would be inaccurate if they falsely allow
for the possibility of a legal system with moral criteria of validity. In exactly
the same way that the definition of “bachelor” would be problematic if it
incorrectly fails to preclude the Pope’s being a bachelor, the definitions of
“law” and “authority” would be problematic if they incorrectly fail to pre-
clude the possibility of an authoritative legal system with moral criteria of
validity.
2. The Identification Thesis and the Arguments

from Authority
The Arguments from Authority, notwithstanding the differences among
them, all depend on three principal ideas. The first is the Authority Thesis,
according to which it is a conceptual truth that every legal system sincerely
claims legitimate authority.1 Insofar as “the claim to authority is part of the
nature of law” (EPD 215), the idea of a legal system that does not sincerely
claim authority is conceptually incoherent.
1
Joseph Raz, “Authority, Law, and Morality,” in Raz, Ethics in the Public Domain (Oxford:
Clarendon Press, 1994). Hereinafter EPD.
154

The second is the proposition that “[i]‌f the claim to authority is part of the
nature of law, then whatever else the law is it must be capable of possessing
authority” (EPD 215). This conditional claim, together with the idea that
every conceptually possible legal system sincerely claims to be a legitimate
authority, implies:
The Instantiation Thesis: It is a conceptually necessary condition for something to
count as a legal system that it is capable of legitimate authority.
For if every conceptually possible legal system sincerely claims authority and
only an institutional normative system capable of being legitimate can sin-
cerely claim authority, then any institutional normative system incapable of
legitimacy is conceptually disqualified from being a legal system.2
The third idea articulates one of the necessary conditions for an institu-
tional normative system of any kind to be legitimate:
The Identification Thesis: It is a conceptually necessary condition for an institu-
tional normative system to be legitimate that subjects can identify the content of its
norms without having to assess for themselves what right reason requires of them.
The Arguments from Authority employ a template that can be summarized
as follows. If an institutional normative system cannot be legitimate unless
the content of its norms can be identified by the subject without having to
determine for herself what right reason requires, then an institutional norma-
tive system with moral criteria of validity cannot be legitimate because the
subject cannot identify the content of its norms without determining for her-
self what right reason requires. But if an institutional normative system with
moral criteria of validity is incapable of being legitimate and any institutional
normative system incapable of being legitimate is conceptually disqualified
from being a legal system, then an institutional normative system with moral
criteria of validity is conceptually disqualified from being a legal system.
All the claims in this argument template are conceptual and hence meta-
physical in character. The claim is that it is conceptually and hence metaphys-
ically impossible for a subject of an institutional normative system S with
moral criteria of validity to determine what she ought to do under its norms
without assessing what right reason dictates that she should do. It is therefore
conceptually and hence metaphysically impossible for S to be legitimately au-
thoritative. The idea that an institutional normative system with moral crite-
ria of validity can be legitimate is hence grounded in a conceptual confusion
2
Raz thus derives the Instantiation Thesis from the Authority Thesis. This is unnecessary, as
discussed in Chapter 5, because the Instantiation Thesis cannot plausibly be denied.
Can claim of authority be confused? 155
about the nature of legitimate authority, as is the idea that such a system can
be a legal system.
The Identification Thesis, unlike the Instantiation Thesis, is far from self-ev-
ident. The Instantiation Thesis expresses a truism about law that can be derived
from an obvious conceptual truth about artifacts: only those things capable of
instantiating all conceptually necessary features of an A can be an A. If this latter
claim is true and it is a conceptually necessary feature of law that it can be legiti-
mate, then the Instantiation Thesis must also be true. The Identification Thesis,
in contrast, is neither a truism about law nor immediately derivable from ob-
vious conceptual truths about artifacts. Whatever else might be true of the
Identification Thesis, it expresses a claim that is neither trivial nor self-evident.
The issue is whether we have any reason to believe that the Identification
Thesis is true. Insofar as it purports to be a conceptual truth, the Identification
Thesis must be grounded in the linguistic practices that determine the content
of the relevant concepts. This, again, does not mean that the Identification
Thesis must be transparently entailed by the lexical meanings of the terms
in the way that the claim that every bachelor is unmarried is transpar-
ently entailed by the lexical meanings of “bachelor” and “unmarried.” The
Identification Thesis might be entailed by deeper philosophical commitments
that are presupposed by our ordinary linguistic practices, but it must still be
compatible with the lexical meanings that are determined by those practices.
3. Could law’s claim of authority

be conceptually confused?
The inclusive positivist might argue that officials who claim legitimate au-
thority and accept the Identification Thesis are confused, but Raz rejects this
line of defense:
[W]‌hile [legal officials and institutions] can be occasionally [confused,] they cannot
be systematically confused. For given the centrality of legal institutions in our struc-
tures of authority, their claims and conceptions are formed by and contribute to our
concept of authority. It is what it is in part as a result of the claims and conceptions
of legal institutions (EPD 217).3
3
Raz’s use of “systematic” suggests that the relevant confusion has both a vertical and a hori-
zontal dimension. The horizontal dimension requires that the conceptual confusion be widespread.
One official’s conceptual confusion, no matter how extensive or deep, is not enough to consti-
tute systematic confusion. The vertical dimension requires that the confusion pervade a plurality
of core practices regarding the concept. Widespread conceptual confusion about purely peripheral
issues is not enough to constitute systematic confusion. Conceptual confusion that has both these
156

It is helpful to note that there are two ways in which a claim of legitimate
authority might be false. If, on the one hand, S is an institutional normative
system or official, the statement that S is legitimate can be true or false. If it
is true that S is legitimate, it is because S has the requisite moral properties,
which will differ depending on whether S is an institutional normative system
or S is an official. If it is false that S is legitimate, it is because S lacks the requi-
site moral properties. Either way, it is a contingent matter whether S is legiti-
mate because whether or not S has the requisite moral properties depends on
properties of S that are contingent: if S is an institutional normative system,
then whether S is legitimate depends on the contingent content of its rules; if
S is an official, then whether S is legitimate depends, among other things, on
whether S was properly selected to serve as an official. Thus, if S is either an
institutional normative system or an official lacking legitimate authority, the
statement that S is a legitimate authority is contingently false.
But if, on the other hand, S is not an institutional normative system or an
official, then the statement that S has legitimate authority is necessarily false.
If S is a building or a novel, it is clearly false that S has authority. One need
not know anything more about S than that S is one of those things to know
that it does not have authority. We do not need to know what contingent
features a building has to know whether it has authority. It does not because
it could not; a building is not the kind of thing that, as a conceptual matter,
could have authority. We likewise do not need to know what contingent fea-
tures a novel has to know whether it has authority. It does not because it
could not; a novel is not the kind of thing that, as a conceptual matter, could
have authority. The claim that a building or novel has legitimate authority is
necessarily false because it is conceptually confused.
In the case where S cannot have legitimate authority as a conceptual
matter—regardless of whether S is a building or a novel—the claim that S has
legitimate authority expresses a category mistake. The problem in each case
is that S is not the kind of thing to which the category of authority applies.
The category of authority applies only to things that are rational free agents
or that are either norms or systems of norms. It makes no more sense to char-
acterize a building as having legitimate authority than it does to characterize
a number as making claims or as having a physical location.
The Identification Thesis implies that the claim that an institutional nor-
mative system with moral criteria of validity can be legitimate expresses a
dimensions is system-wide and hence “systematic” in the relevant sense: it afflicts a large percentage
of officials and pervades the entire structure of beliefs regarding the concept. Officials can, on Raz’s
view, be conceptually confused as long as this confusion lacks either the horizontal dimension or
the vertical dimension.
Can claim of authority be confused? 157
category mistake. Insofar as it is true that an institutional normative system
with moral criteria of validity is conceptually incapable of being legitimate, it
is as much the wrong kind of thing to instantiate authority as a building or
a novel. It makes sense to ask of two legal systems which is closer to being
legitimate; it makes no sense to ask of a novel and a book which is closer to
being legitimate. Something that is conceptually incapable of legitimacy is
neither close to nor far from being legitimate; it is just not the kind of thing
to which legitimacy is sensibly attributed. If institutional normative systems
with moral criteria of validity are conceptually incapable of legitimacy, it is as
much a category mistake to attribute authority to them as it is to attribute a
claim or physical location to a number.
There are circumstances in which a person might knowingly state claims
about authority that express category mistakes. A person might say some-
thing that expresses a category mistake about authority for comedic effect.
Assuming that it is a category mistake to think that a dog can have authority,
an excessively doting dog owner might suggest that the animal has authority
as a humorous way of expressing that the owner’s life revolves around the
dog’s desires. Or a person might claim the dog has authority to deceive an
alien in circumstances where she believes the alien poses a threat that can be
neutralized with such a falsehood.
Alternatively, a person might say something expressing a category mistake
without knowing that it expresses a category mistake. Assuming that it is a
category mistake to think that a dog can have authority, someone could assert
that a dog has authority without knowing that it expresses a category mistake;
if one does not know that such a claim is, as a conceptual matter, necessarily
false, one can sincerely assert it without knowing that it expresses a category
mistake.
But there are certain kinds of category mistakes about authority that cannot
be made. If a person (1) understands the nature of authority, (2) under-
stands the nature of the thing about which a claim of authority is made, and
(3) intends to make a literal claim attributing authority to the thing, then she
cannot sincerely claim that something conceptually incapable of authority ac-
tually has authority. Thus, if officials sincerely claim authority on behalf of the
legal system and understand the nature of authority, then a legal system can
fail to have authority because it is unjust but not because it is the wrong kind
of thing to have authority. If officials sincerely claim that law has authority,
then that claim cannot express a category mistake: law must be the right kind
of thing to have authority.
This reasoning seems facially incontrovertible. After all, sincerity obviously
precludes the deliberate making of false claims. In ordinary circumstances, a
person who claims that a building or novel has authority might instantiate
158

any number of mental states that involve insincerity. Such a person might
intend to deceive or to make a joke. But she cannot be characterized as being
sincere in the absence of extreme confusion about the concept of authority.
The instantiation of sincerity by a person in ordinary circumstances pre-
cludes deliberately making conceptually false claims about authority, but it
does not preclude unintentionally making such claims. A person might not
realize that some entity S is conceptually incapable of authority and might
sincerely claim that S has authority. If so, there is nothing precluding an offi-
cial’s being conceptually confused about authority and sincerely attributing it
to something conceptually incapable of having it.
Raz allows that some officials can be confused about the nature of au-
thority but denies that most officials could be confused about it because “their
claims and conceptions are formed by and contribute to our concept of au-
thority” (EPD 217). Since the content of the concept is determined by the
convergence of officials in applying the corresponding concept-term, there
cannot be widespread conceptual confusion about authority among officials.
Accordingly, while some officials can exhibit conceptual confusions about
authority that express category mistakes, it is not possible for most officials to
exhibit such conceptual confusions.
Even so, it seems premature to dismiss the possibility of widespread con-
ceptual confusion among officials because some conceptual confusions are
more difficult to discern than others. It is certainly true that some mistakes
are highly unlikely to occur except under the most unusual circumstances. As
an empirical matter, it is extremely unlikely that a competent speaker of the
language would think that a building or novel has authority because the con-
ceptual confusion is transparent.
But not all such confusion is so transparent. It was widely believed for
many years that space-time is defined by certain Euclidean constraints. In
particular, it was widely thought necessary that, given a line L and a point p
off of L, there is one and only one line passing through p parallel to L; that is,
it was thought that the very concept of space-time was Euclidean in character.
Despite the continuing intuitive appeal of this view, the general theory of rel-
ativity presupposes a non-Euclidean concept of space-time in which there are
an infinite number of lines passing through p parallel to L. If the confusion
about our universe was partly empirical in character, it was also partly con-
ceptual in character.
The possibility of conceptual confusion about whether space- time is
Euclidean can arise among competent speakers precisely because there is
nothing in our ordinary linguistic practices that would transparently decide
the issue of how many lines pass through p that are parallel to L. Space-time
appears to be something that could, as a conceptual matter, be Euclidean in
Do officials accept the Identification Thesis? 159
character. The claim that space-time is not Euclidean depends, in part, on
empirical considerations about the universe that have nothing to do with our
ordinary linguistic conventions for using the relevant concept-terms. Insofar
as the claim that space-time is Euclidean involves a conceptual confusion, it
is not one that is obvious.
But systematic conceptual confusion about authority among competent
speakers of the language is not possible because the nature of authority is de-
fined by the ordinary linguistic conventions that are practiced by such speak-
ers. A competent user of a term cannot be systematically mistaken about
its application because its application-conditions are defined by the way
such a speaker uses it. Just as a competent user of the terms “bachelor” and
“unmarried” cannot be confused about whether bachelors are unmarried, a
competent user of the term “authority” cannot be confused about whether a
building can have authority.
Law’s claim of authority cannot be conceptually confused, then, because it
is constituted by the beliefs, claims, and practices of officials that determine
the meaning of the term “authority.” If the Incorporation Thesis is incon-
sistent with the concept of authority as it is defined by the beliefs, claims,
and practices of officials, it must be rejected; that thesis cannot be rescued
by claiming that most officials are conceptually confused about the nature of
authority.
4. Do most officials accept the Identification Thesis?

Raz believes that law’s claim of authority cannot be conceptually confused,
but officials in legal systems like ours can clearly be conceptually confused.
Any official who would include moral language in the terms of a constitution
believing that such language incorporates a moral constraint on the content
of law would be conceptually confused on this view.
Constitutions often include provisions with language that seems to in-
corporate moral norms into the criteria of validity. The Eighth Amendment
states that “cruel and unusual punishments [shall not] be inflicted.” The no-
tion of cruelty has some descriptive content, but it also has morally normative
content: to say that a punishment is cruel is to say, among other things, that
it inflicts suffering that is morally excessive given the gravity of the offense. It
is certainly conceivable that an official might include such language in a con-
stitution in the belief that the relevant moral standard is thereby incorporated
into the criteria that determine what counts as valid law in the system.
160

It is likewise conceivable that a judge might apply what she takes to be a
legal rule expressing a criterion of validity as though it incorporates a moral
principle. A judge might interpret a constitutional provision prohibiting
“cruel” punishments as if it incorporates the relevant moral standard into the
criteria that determine what counts as valid law. On this reading, the con-
stitution would invalidate official acts that inflict morally excessive punitive
detriment on someone who violates the law.
These are not only possible motivations for including putatively moral lan-
guage in a constitution or applying it in a particular way; they are also the
most probable motivations for these acts. It is true that, if Raz is correct, the
Eighth Amendment cannot incorporate moral content into the criteria of va-
lidity, but there is no reason to suppose that officials would know this. Given
that many legal philosophers deny the Sources Thesis, there is no reason to
suppose that philosophical laypersons would believe, much less know, that
it is conceptually impossible to incorporate moral norms into the criteria of
validity.
One can argue that any person who has this belief is conceptually con-
fused, but this is a move that is plausible only up to a point. The more officials
and citizens there are who have this belief, the less plausible it is to think that
most officials and citizens are practicing a linguistic convention with respect
to the term “authority” that entails the Identification Thesis. If the majority of
officials and citizens believe that the constitution authoritatively incorporates
moral standards that constrain the content of what counts as law and act on
this belief, they could not be practicing a linguistic convention regarding “au-
thority” that implies the Identification Thesis.
The problem is not just that it is conceivable that officials believe it is pos-
sible to incorporate moral norms into the criteria of legal validity; the problem
is that there are judges in existing legal systems who have this belief and act
on it. There is a split among judges about how to interpret those clauses of
the Bill of Rights that are expressed in moral language. Originalist Justices
like Antonin Scalia believe that a moral term appearing in the Constitution
should be construed in accordance with the common understanding of its
meaning at the time the relevant provision was ratified. Others like William
Brennan believe that recourse to moral argument is necessary to identify the
moral content of such terms.
These views about constitutional interpretation reflect views about what
the meaning of the constitutional language really is. Insofar as originalists
believe that the putatively moral language of the Constitution should be
interpreted in accordance with the original understanding, it is because they
believe that the original understanding of that language conforms to what it
really means. Insofar as judges like Brennan believe that the putatively moral
language of the Constitution should be interpreted as incorporating moral
content, it is because they believe that the meanings of that language incor-
porate such content. But to the extent that their views of constitutional inter-
pretation are grounded in the belief that they correctly identify the meanings
of the relevant language, it follows that judges who hold a view like Brennan’s
are conceptually confused.
Even worse, every official, citizen, and legal theorist who holds a theory of
constitutional interpretation like Brennan’s would be conceptually confused
on Raz’s view. It is not clear what percentages of judges, citizens, lawyers,
and legal theorists hold the respective views because no rigorous sociological
surveys have been done. But it certainly seems possible that most officials, cit-
izens, and legal theorists in our world side with Brennan over Scalia.
This poses a problem for the Identification Thesis. If it is possible that most
people in our world accept Brennan’s view, that is because there is no obvious
contradiction between that view and our linguistic conventions for using the
term “authority.” It is not possible that most competent speakers of English
accept that a married man can be a bachelor because there is an obvious con-
tradiction between that view and the linguistic conventions for using the
term “bachelor” that they understand themselves to be practicing.
The idea that these officials, citizens, and theorists can hold such a view
only if they are conceptually confused is problematic. For if, given our lin-
guistic practices regarding the term “authority,” it is possible for some linguis-
tically competent citizens, officials, and theorists to reject the Identification
Thesis, then it is possible for all of them to do so. On this line of reasoning,
if a state of affairs in which some linguistically competent judges believe the
Constitution incorporates moral norms is consistent, given our linguistic
practices, with its defining a legal system, then so is a state of affairs in which
they all do. There is no reason to think, for example, that there could not be
a legal system, compatible with our linguistic practices, in which every lin-
guistically competent person holds Brennan’s view on the possibility that the
Constitution incorporates moral norms as criteria of legal validity.
But one cannot consistently hold (1) that it is possible, given our lin-
guistic practices, for all such competent speakers to reject the Identification
Thesis and (2) that anyone who rejects the Identification Thesis is conceptu-
ally confused. Our linguistic practices regarding the term “authority” imply
the Identification Thesis only insofar as most competent speakers do not re-
ject that thesis. If it is possible for most competent speakers to reject the
Identification Thesis, then they could not be practicing a linguistic conven-
tion that implies or presupposes that thesis.
The Arguments from Authority, then, presuppose that most speakers
cannot be conceptually confused about the Identification Thesis because they
162

are practicing a linguistic convention that implies this thesis, but Raz offers
no more support for this claim than the following remarks about arbitration:
Suppose that an arbitrator, asked to decide what is fair in a situation, has given a
correct decision . . . . Suppose that the parties to the dispute are told only that about
his decision, i.e., that he gave the only correct decision. They will feel that they know
little more of what the decision is than they did before. They were given a uniquely
identifying description of the decision and yet it is an entirely unhelpful description
(EPD 219).
If this is intended to justify a claim about what people believe about the
concept of authority, the argument is problematic. It is probably true that
most people would feel that an arbitrator who gave her decision in such terms
had not performed her function, but this tells us more about the nature of
arbitration than it does about the nature of either law or authority. An arbi-
trator who tells the disputants only that they should do the right thing has
clearly not done anything properly characterized as arbitrating the dispute,
but that is because the point of arbitration is to resolve disputes about what is
the right thing to do by specifically identifying what the right thing to do is in
the circumstances. Given this, it is clear that an arbitrator cannot perform the
arbitration’s conceptual function of “resolving” a dispute by telling the dispu-
tants only that they ought to do what is right. The conceptual function of an
arbitrator is both practical and epistemic—namely, to resolve a disagreement
by specifying what exactly disputants must do.
It is true that the conceptual function of law, like the conceptual function
of arbitration, is both practical and epistemic, but the conceptual function of
law—as opposed to, say, the conceptual function of adjudication—is not to
resolve disputes about which of two specific acts must be performed by some
particular person. It is rather to promulgate and apply general norms that
provide enough guidance to make it possible for people to live together in
society. As far as our linguistic practices are concerned, an enacted norm that
prohibits killing “innocent persons” is no less law or authoritative because it
does not indicate with specificity who counts as “innocent” under that norm.
The same is true of a constitutional norm prohibiting “cruel punishment.”
As far as our linguistic practices are concerned, the Eighth Amendment is no
less authoritative or legal in character in virtue of not specifying exactly what
would count as cruel. The rule has always been legally binding and authorita-
tive in the U.S. despite the fact that, prior to any court precedent specifying
what counts as cruel, the only way that a subject could determine what it
prohibits is by assessing the balance of reasons with respect to cruelty. As far as
ordinary usage is concerned, the Eighth Amendment is and has been legally
binding and authoritative since its ratification.
It is partly because the functions of law and arbitration are different
that each institution is needed. As Lon Fuller puts it, the point of law is to
“achiev[e]‌ . . . [social] order . . . through subjecting people’s conduct to the
guidance of general rules.”4 Law could not succeed in keeping the peace if it
characteristically provided guidance that is as specific as an arbitrator’s deci-
sions; it is not nomologically possible for a human authority to comprehen-
sively regulate the behavior of human beings by specifying every act that the
authority seeks to regulate. Law is an institution that is contrived to keep
the peace among beings like us, and that can only be done if behavior is
largely regulated through the recognition, application, and enforcement of
general rules.
Raz’s arbitration argument fails to establish the Identification Thesis, then,
because it incorrectly presupposes that law and arbitration perform the same
conceptual function. Law is needed to provide general rules while arbitration is
needed to settle disputes that arise under those general rules, and our linguistic
practices with respect to the terms “law” and “arbitration” straightforwardly
track that difference. Indeed, it is because law must regulate behavior by such
means that a judicial agency—which arbitrates legal disputes—is needed to
resolve disputes about what those authoritative legal norms require. As far as
our linguistic practices go, law and arbitration have different but related func-
tions. Both are authoritative, but each has a different epistemic function. Our
ordinary linguistic practices distinguish the two because they have different
functions; as far as our linguistic conventions are concerned, law and arbitra-
tion are not sufficiently analogous to preclude the possibility of an institutional
normative system with moral criteria that is both law and authoritative. There
is simply nothing in our linguistic practices that entails a commitment to the
Identification Thesis.
Further, there is nothing in our linguistic practices that would entail a
commitment to the Preemption Thesis, which expresses the core of the serv
ice conception of authority, because the Identification Thesis is logically
equivalent to the Preemption Thesis. The proof is trivial. If, on the one hand,
4
Lon L. Fuller, “A Reply to Professors Cohen and Dworkin,” vol. 10 Villanova Law Review
(1965), 655, 657. Emphasis added. Similarly, John Austin writes that “[a]‌law . . . may be defined
as a rule laid down for the guidance of an intelligent being by an intelligent being having power
over him.” John Austin, Lectures on Jurisprudence or the Philosophy of Positive Law (St. Clair Shores,
MI: Scholarly Press, 1977), 5. See also H.L.A. Hart, “Postscript” in The Concept of Law 2nd Ed.
(Oxford: Clarendon Press, 1994).
164

an institutional normative system cannot be legitimate unless its directives
can replace the subject’s assessment of the balance of reasons in her delibera-
tions, then an institutional normative system cannot be legitimate unless the
content of its norms can be identified by the subject without having to assess
the balance of reasons; thus, the Preemption Thesis implies the Identification
Thesis.5 If, on the other hand, an institutional normative system cannot be
legitimate unless the content of its norms can be identified by the subject
without having to assess the balance of reasons, then an institutional norma-
tive system cannot be legitimate unless its directives can replace the subject’s
assessment of the balance of reasons in her deliberations about what to do;
thus, the Identification Thesis implies the Preemption Thesis.
The foregoing empirical considerations refute the service conception of
authority. Insofar as the Identification Thesis is logically equivalent to the
Preemption Thesis, the above reasoning would also refute the Preemption
Thesis. If there is nothing in our ordinary linguistic practices that implies the
Identification Thesis, then there is nothing in those practices that implies the
Preemption Thesis. If the service conception of authority is intended to ex-
plicate the concept of authority as it is determined by our ordinary linguistic
practices, then it fails insofar as there is nothing in those practices that implies
the Preemption Thesis.
These considerations also refute exclusive positivism if conceptual juris-
prudence is concerned to explicate the concept of law as it is determined by
our ordinary linguistic practices. Insofar as our ordinary linguistic practices
allow for the conceptual possibility of a legal system with moral criteria of
5
It is worth noting that the Identification Thesis can also be derived from NJT, which is a
substantive moral claim about the justification of authority and not a conceptual claim about the
nature of authority. NJT asserts that an authority is legitimate insofar as the subject is more likely
to better comply with right reason if she follows the authority’s assessments than if she follows her
own. If NJT is true, then an institutional normative system cannot be legitimate if it is not possible
for a subject to better comply with right reason by following the authority’s assessments than by fol-
lowing her own. But it is not possible for a subject to better comply with right reason by following
the authority’s assessments if the authority’s directives cannot be identified, as would be true of an
institutional normative system with moral criteria of validity, by the subject without having to assess
the balance of reasons. Thus, if NJT is true, then an institutional normative system cannot be legiti-
mately authoritative unless the content of its norms can be identified by the subject without having
to assess the balance of reasons.
One might think that the above derivation of the Identification Thesis from NJT is problematic
because one cannot derive conceptual claims from substantive moral claims, but the above deriva-
tion does not rest only on substantive moral claims. The conceptual claim that an institutional nor-
mative system cannot be legitimately authoritative unless the content of its norms can be identified
by the subject without having to assess the balance of reasons depends in part on NJT. But it also
depends on the conceptual claim that it is not possible for a subject to better comply with right
reason by following an authority if its directives cannot be identified without the subject’s having to
assess the balance of reasons.
validity, those practices are inconsistent with the Sources Thesis. Given that
the Incorporation Thesis is merely the negation of the Sources Thesis, our
ordinary linguistic practices seem to presuppose the Incorporation Thesis.
If there are no other compelling reasons to think that it is conceptually
impossible for a legal system to incorporate moral criteria of validity, the
Incorporation Thesis would seem to be true as far as our concept of law is
concerned.
16
7
To Whom the Rules Apply
Norm Guidance and the Incorporation Thesis
A legal system, like any other artifact, is an object manufactured to achieve

some characteristic purpose that corresponds to its conceptual function.
Given that a legal system is, by nature, comprised of norms that are recog-
nized, applied, and enforced as law against subjects, it seems clear that the
conceptual function of law involves regulating behavior through the guid-
ance of norms. Since it is a necessary condition for something to count as
an instance of a type of artifact that it is capable of performing that artifact’s
conceptual function, it is a necessary condition for an institutional normative
system to count as a system of law that its norms are capable of guiding the
behavior of its subjects.
This chapter is concerned with two arguments for the claim that the norms
of an institutional normative system with moral criteria of validity are inca-
pable of guiding behavior (the Guidance Arguments). The problem, on this
line of reasoning, is that neither a rule of recognition that validates norms
on the basis of moral merit nor a norm that is valid in virtue of moral merit
can properly guide the persons they must be able to guide to perform law’s
conceptual function. To begin, since non-officials cannot identify what their
obligations are under valid law by consulting a rule of recognition that vali-
dates all or only morally meritorious norms without determining for them-
selves what the morally meritorious thing is to do, the valid norms cannot
properly guide non-official behavior.1 Further, since a judge who follows such
a rule of recognition is already motivated to do the morally meritorious thing
in deciding a case, the judge cannot also follow the applicable valid norm
because she will make the same decision regardless of whether she consults
the valid norm. In both cases, an institutional normative system with moral
1
A rule of recognition that validates all morally meritorious norms makes moral merit a suffi-
cient condition for validity, while a rule of recognition that validates only morally meritorious norms
makes moral merit a necessary condition of validity. A rule of recognition that validates all and only
morally meritorious norms makes moral merit a necessary and sufficient condition for validity.
University Press.
168
168 Norm Guidance and the Incorporation Thesis

criteria of validity is incapable of performing law’s conceptual guidance func-
tion and hence cannot be a legal system.
This chapter challenges the Guidance Arguments. It argues that the
guidance function of law does not imply that every legal norm must be
capable of guiding or informing the behavior of every person; it implies
only that every legal norm must be capable of guiding or informing the
behavior of every person whose behavior it governs. Since the rule of rec-
ognition governs only official behavior, there is no reason to think that
subjects must be able to determine what their non-official obligations are
under valid law by consulting the rule of recognition. Since a valid legal
norm that a judge applies in deciding a dispute involving non-official be-
havior defines no judicial obligations or powers and hence does not govern
judicial behavior, it is not possible for the judge to do anything in deciding
the dispute that would count as following the valid norm. The Guidance
Arguments are problematic, then, insofar as they assume that a rule that
does not govern someone’s behavior must nonetheless be capable of guid-
ing or informing her behavior.
The Arguments from Authority are problematic for a similar reason. It
might be true that a rule of recognition defining moral criteria of validity
cannot replace the judgment of a subject with respect to what her non-official
behavior should be in her practical deliberations about what to do, but a legal
norm must be capable of replacing the judgments of only those persons whose
behavior it governs. Since the rule of recognition, by definition, governs only
official behavior, any norm that governs non-official behavior cannot be a rec-
ognition norm in the same sense that a married woman cannot be a bachelor.
Insofar as the Arguments from Authority presuppose that the rule of recog-
nition must be able to replace the judgments of subjects with respect to what
their non-official behavior should be, it falsely presupposes that the rule of
recognition must be capable of guiding persons whose behavior it does not
govern. Like the Guidance Arguments, the Arguments from Authority fail to
refute the Incorporation Thesis.
1. The guidance function of law

It is a truism that law has a conceptual function. If law is an artifact, it must
have a conceptual function that expresses its purpose as such. Artifacts are
objects, as a conceptual matter, that are manufactured, adapted, or adopted
by intentional beings to serve some characteristic purpose; an object that is
not intended to do something might be artificial, but it is not artifactual.
The guidance function of law 169
What distinguishes an object that is merely artificial from one that is also
artifactual is that the latter is a non-natural object in virtue of being manu-
factured or adapted by intentional beings to do something while the former
is simply a non-natural object. If I throw a bunch of fallen tree branches into
a heap, the heap is a non-natural, and hence artificial, object in virtue of my
having modified the various fallen branches from the natural state in which
I found them by grouping them into a heap. But it is not an artifact unless
I have manufactured or adapted it to serve some purpose for which it can
generally be used. If I have manufactured the heap in order to present it to
an audience to induce an aesthetic experience, the heap is an art object and
hence an artifact. Everything that is artifactual is artificial, but not everything
artificial is artifactual.
An artificial object must be capable of performing the conceptual function
of some artifactual kind to be an artifact of that kind. An object that cannot
perform the conceptual function of an automobile unless it is transformed
into some other kind of object is not properly characterized as an automo-
bile. It might be possible to make an automobile out of a cardboard box with
a sufficient number of modifications, but an unmodified box is not capable
of doing what something must be able to do to be an automobile; whatever
modifications are needed would transform the object from something that
was just a cardboard box into something that is not a cardboard box. If law is
an artifact, then an object incapable of performing law’s conceptual function
is not properly characterized as “law.”
Like any other artifact, something that is properly characterized as law
has a conceptual function, but the term “law” is ambiguous as between two
purely descriptive usages. The first picks out or describes an individual norm
(a law) that is recognized, applied, or enforced by officials in some system,
while the second picks out or describes the system (the law) of which those
laws are constituents. A law is a part of the law.
The functions of the two are related in the same way that the function of
some part of any other artifact is related to the function of the artifact itself.
Every essential part of an automobile does something that makes it possible
for the automobile to transport persons or things to some intended desti-
nation. The function of a steering wheel is to enable the driver to direct the
automobile by manipulating its position, while the screws holding the various
pieces of a steering wheel together facilitate the performance of the steering
wheel’s function by preserving its structural integrity; if the parts making
up the steering wheel come loose, the steering wheel will come apart and be
thereby rendered unable to perform its conceptual function.
The conceptual function of a legal system is to regulate behavior through
the guidance of legal norms; these norms purport to guide behavior in a way
170

that makes it possible for subjects to reap the social benefits of living and
working together. Since it is not possible to reap such benefits unless socially
disruptive conflicts are reduced to a point where something properly charac-
terized as a community is possible, the conceptual function of a legal system
is to regulate behavior in a manner that keeps the peace enough to permit the
existence of a community among its subjects.
There are two classes of legal norms comprising a legal system. Some of
these norms govern legislative, adjudicative, and enforcement activities and
hence govern official behavior; these norms construct the rule of recogni-
tion and thereby define the criteria that distinguish norms that officials must
recognize, apply, or enforce as law from those that they may not recognize,
apply, or enforce as law. The remainder of these norms govern the non-official
behavior of everyone within the jurisdiction of the system, including officials
in their non-official capacities. These norms, along with the institutions they
construct and the officials who staff those institutions, are all proper parts of
a legal system.
As proper parts of a legal system, legal norms are contrived to perform a
function that enables the system to regulate the behavior of those within its
jurisdiction. But since a complete specification of the conceptual function of
law qua norm includes that it guides behavior, it is a conceptually necessary
condition for something to count as a legal norm that it is capable of guiding
behavior. A box is not a legal norm because it is incapable of guiding behav
ior. A purely descriptive preamble to a law is likewise not a legal norm because
purely descriptive content is incapable of guiding behavior.
It might be true that it is a necessary condition for something to count as a
law that it is capable of guiding behavior, but it is not in virtue of being a legal
norm that something is capable of guiding behavior.2 Consider the sentence
“citizens must pay taxes with six copper disks shaped like a square circle.”
Despite the otiose occurrence of the deontic operator “must” in the sentence,
it fails to express something properly characterized as a norm insofar as it pur-
ports to require someone to do something that is metaphysically impossible.
2
If we assume that the conceptual function of law qua norm distinguishes it from other kinds
of norms, the conceptual function of law must involve something more than just guiding subject
behavior; after all, the conceptual function of anything that counts as a norm is to guide behavior.
While legal norms and norms of etiquette are specific subclasses of the general category of artifactual
norms, the nature of the subcategory also contributes something to defining the conceptual function
of the more specific type of norm; the conceptual function of even a legal norm that reproduces the
content of a norm of etiquette will be different from the conceptual function of the norm of eti-
quette the content of which it reproduces. Part of the difference in the conceptual functions of the
two subclasses of norms will be explained by the differences in the content of the respective rules,
but part of the difference will have to do with the institutional character of law qua system and its
conceptual function.
The sentence has the logical form necessary to express a norm in virtue of
picking out a class of subjects, describing an act, and featuring a deontic op-
erator that purports to require members of the former to do the latter. But the
sentence fails to express something that would count, as a conceptual matter,
as a norm because it is metaphysically impossible for even an omnipotent
being to do what the sentence purports to require. There could not be a legal
norm that requires citizens to pay taxes with copper disks in the shape of a
square circle because there could not be a norm of any kind that requires this.3
Anything properly characterized as a norm, whether legal or otherwise, must
be capable of guiding behavior in some relevant sense.
Although something must be metaphysically capable of guiding behavior to
be a norm, it need not be nomologically capable of guiding our behavior to be
a norm. The sentence “all subjects must jump twenty feet off the ground once
a day” expresses something that is properly characterized as a norm insofar
as the existence of rational beings that can do that is nomologically possible;
such beings might be stronger than we could be or might live on a planet
where the pull of gravity is weaker. Either way, the norm it expresses cannot
govern our behavior because, at this point in time, we are nomologically in-
capable of jumping that high; a norm that requires what is nomologically
impossible from persons is metaphysically incapable of governing their behav
ior. Only norms nomologically capable of guiding the behavior of a class of
beings are metaphysically capable of governing their behavior. Norm govern-
ance presupposes norm guidance.
While every legal norm must be metaphysically capable of guiding be-
havior, not everything capable of guiding behavior counts as a legal norm.
Moral norms are capable of guiding behavior, but that does not make them
legal norms. Moral norms, like legal norms, can guide behavior in virtue of
being norms. But they have some property that conceptually distinguishes
them from legal norms insofar as there can be a moral norm that is not also
a legal norm. It is hence a necessary, but not a sufficient, condition for some-
thing to count as a legal norm that it is capable of guiding behavior.
More difficult is the issue of whether two norms requiring conflicting
actions can be properly characterized, according to ordinary usage, as being
norms of the same system. There are some normative systems that clearly
cannot contain norms with inconsistent requirements. If morality is objective
3
It is conceptually possible for officials in something that counts as a legal system to “enforce”
content requiring subjects to do what is nomologically impossible, but that would not constitute
such content as a norm. If norms must be metaphysically capable of guiding behavior and every con-
ceptually possible law is a norm, then content requiring what is nomologically impossible could not
be a law. The enforcement of such content by officials is, as a conceptual matter, a form of oppression
that has nothing to do with applying or enforcing norms that have the status of law.
172

in character and there are two norms requiring conflicting actions of the same
persons in the same circumstances, then at least one of those norms is not
a moral norm. But it is not as clear that there can be no systems contain-
ing norms with inconsistent requirements. Suppose Congress enacts a norm
requiring subjects in circumstance C to do a and subsequently enacts another
norm requiring subjects in circumstance C to abstain from a without stating
that the newer norm supersedes the older one. Suppose, further, that it is
nomologically possible for subjects to do a and that it is nomologically pos-
sible for subjects to abstain from a. Under these assumptions, it is clear that
each piece of content is nomologically capable of governing our behavior.
But given that it is metaphysically impossible for someone to both do a and
abstain from doing a in C, the question is whether both norms are properly
characterized as laws of the same legal system at time t.
There is nothing in ordinary usage that obviously precludes there being
two inconsistent laws in the same legal system. Each is properly characterized
as a norm insofar as each has the requisite logical form and is metaphysically
capable of guiding behavior; similarly, each norm is metaphysically capable
of governing our behavior in virtue of requiring something that is nomolog-
ically possible to do. It is true that they cannot both be satisfied at the same
time, but there is nothing in the concept of a law, as defined by our linguistic
practices, that clearly dictates that a requirement of one norm of the system
must be consistent with those of every other norm of the system. At the very
least, the statement that one law requires doing something that another law
prohibits is not obviously self-contradictory in the same sense that the claim
that there can be married bachelors is obviously self-contradictory.
This is not to deny that a legal system should not require inconsistent acts.
In the absence of exceptional circumstances, it would be morally wrong for
a legal system to do so because it is generally unfair to require subjects to
do what they cannot do. Moreover, a legal system that frequently required
inconsistent acts is less likely to keep the peace among subjects because the
application and enforcement of inconsistent requirements will likely lead to
the kind of widespread dissatisfaction that expresses itself in socially disrup-
tive ways. But this says nothing about whether it is conceptually possible for
a legal system to require inconsistent acts; it goes only to the likelihood that
the system will succeed in efficaciously regulating behavior in a manner that
keeps the peace and thereby facilitates social cooperation.
It is for these reasons that most legal systems have laws resolving these
inconsistencies by giving legal effect to only the most recently enacted law,
but the fact that laws are needed to resolve these problems entails that it is
conceptually possible for there to be two laws with inconsistent requirements.
If it were conceptually impossible for norms stating inconsistent requirements
to count as laws of the same jurisdiction, it would not be necessary to adopt
a legal norm stipulating which of two conflicting “laws” may be applied or
enforced in the system.
Further, there is nothing in the conceptual function of a legal system that
straightforwardly precludes the possibility of a legal system with inconsistent
laws because the existence of two inconsistent laws does not render the system
incapable of regulating behavior in a manner that keeps the peace. There have
surely been legal systems with norms defining inconsistent requirements that
have nonetheless succeeded in keeping the peace; if one of those norms does
not count as law, it is not in virtue of any obvious facts about the relevant
usages. Although the inclusion of too many inconsistent requirements in an
institutional normative system might render the system incapable of realizing
its conceptual purpose, there is nothing in the concepts of a legal norm and
a legal system that obviously entails that there can be no inconsistent legal
requirements in a legal system.
The conceptual function of a legal system involves the regulation of behav
ior through norms that can guide subject behavior, but that does not imply
that everything that counts as a legal norm must be efficacious in guiding
behavior for a legal system to perform its function. Suppose Congress enacts
a sentence that coherently expresses a norm, but people cannot discern, for
contingent reasons, from that sentence which norm is expressed by the sen-
tence and what that norm requires of them. Such an enactment would likely
create confusion and discontent; however, the legal system would not thereby
be rendered incapable of performing its conceptual function of regulating
behavior through the guidance of norms—regardless of whether the enact-
ment is properly characterized as creating a law. It is typically enough for a
legal system to achieve its function that people can determine, one way or
another, enough of what the norms of the system require to enable them to
live together in peace. To efficaciously and fairly regulate anyone’s behavior in
a world like ours, a legal system must provide enough information to inform
subjects of what law requires of them; a subject cannot obey a law if she does
not know what it requires.4
The notion of obedience is different from the notion of compliance in the following way. To
4
say that a subject S obeys a norm n is to say that (1) n governs some act of S; (2) it is nomologi-
cally possible for S’s behavior to violate n; (3) S’s behavior conforms to n; (4) S believes that her
behavior conforms to n; and (5) S’s conforming behavior was motivated by a desire to satisfy n’s
requirements—either because n is a norm or because n will be enforced. To say that S complies
with n asserts no more than that (1) n governs some act of S; (2) it is nomologically possible for S’s
behavior to violate n; and (3) S’s behavior conforms to n. A subject can comply with a law without
knowing what is required, but she cannot obey a law without knowing what it requires.
The notion of obeying a norm is also conceptually distinct from the notion of following a norm,
as it is used below. To say that S follows n is to say that (1) S obeys n; and (2) S obeys n because n
174

Legal norms are distinct from other norms in virtue of the unique role they
play in facilitating the realization of the conceptual function of a legal system.
What distinguishes legal norms from other norms is not that they are meta-
physically capable of guiding behavior; every norm is metaphysically capable
of guiding behavior. What distinguishes legal norms from every other norm
is that the officials of a system of law recognize, apply, or enforce them as le-
gally binding against those persons to whom they are properly applied under
some recognition norm of the system. Legal norms, then, are those proposi-
tions nomologically, and hence metaphysically, capable of guiding behavior
that will be applied and enforced by officials against subjects; those norms
are properly said to govern the relevant behaviors of subjects insofar as they
may, under a recognition norm of the system, be applied or enforced against
subjects who are nomologically capable of satisfying and violating them.
2. Motivational and epistemic guidance

There are two types of norm-guidance that a legal system must provide to be
capable of regulating the behavior of law-subjects through the governance of
legal norms. First, the law must provide norms metaphysically capable of mo-
tivationally guiding behavior in the following sense: a norm n motivationally
guides a person P if and only if P’s conformity to n is motivated by the fact
that n is a valid norm of the system. To be motivationally guided by n is to
conform to n because n is a norm that governs one’s conduct. Second, the law
must provide norms metaphysically capable of epistemically guiding behavior
in the following sense: n epistemically guides P if and only if P learns of her
obligations under n by consulting n and satisfies those obligations under n.
Both types of norm-guidance require conforming behavior. P is motiva-
tionally guided by n if and only if (1) P’s behavior conforms to n and (2) P
conforms her behavior to n because it is a norm of the system. P is epistemi-
cally guided by n if and only if (1) P learns of her obligations under n by con-
sulting it and (2) P’s behavior conforms to n. If P’s behavior does not conform
to n after learning what it requires, P’s behavior is neither motivationally nor
epistemically guided by n.
Consider a legal norm that establishes a maximum speed limit of seventy
miles per hour on Interstate 5. I am motivationally guided by that norm only
insofar as I maintain a speed of under seventy miles per hour on Interstate 5
is a norm of the relevant system. Following a norm entails obeying it but obeying a norm does not
entail following it; that is, it is not conceptually possible to follow a norm without obeying it, but it
is conceptually possible to obey a norm without following it.
PDT and inclusive positivism 175
because the rule requires it; insofar as I maintain that speed only to avoid the
fine authorized for violations, I am not motivationally guided by the norm.
I am epistemically guided by the speed limit norm insofar as I learn of the
speed limit by consulting the norm and, in consequence, maintain a speed
of under seventy miles per hour while driving on Interstate 5. In both cases,
I am guided by the norm only insofar as I maintain a speed of under seventy
miles per hour on Interstate 5 and thereby conform my behavior to the norm.
These two types of norm-guidance are logically independent. On the one
hand, one can be epistemically guided by a norm without being motivation-
ally guided by it; if I consult the norm establishing a speed limit of seventy
miles per hour and conform my behavior to the norm to avoid the fine for
speeding, I am epistemically guided by the norm without being motivation-
ally guided by it. On the other hand, one can be motivationally guided by a
norm without being epistemically guided by it; if I follow the example of the
other drivers and drive at the same speed at which they are driving to ensure
that I comply with the speed limit norm because it is a norm and not because
it authorizes a sanction for violations, then my conforming behavior is moti-
vationally guided by the rule without being epistemically guided by it.5
3. The Practical Difference Thesis, the rule

of recognition, and valid legal norms
Scott Shapiro argues that the Incorporation Thesis is inconsistent with the
idea that law’s conceptual function is to guide behavior.6 On his view, the
norms of an institutional normative system with moral criteria of validity are
metaphysically incapable of performing law’s guidance function. Since some-
thing must be metaphysically capable of performing an artifact’s function to
be properly characterized as that kind of artifact and law is an artifact, the
norms of an institutional normative system with moral criteria of validity are
not properly characterized as norms of law insofar as they are metaphysically
incapable of performing law’s conceptual function.
The Guidance Arguments are grounded in the conceptual truism that
something must be capable of guiding behavior to be a norm and hence to
be a law. Since a norm can provide either epistemic or motivational guidance,
This assumes that my gauging the speed of other drivers to attempt to learn of the speed limit
5
does not constitute consulting the rule.

6
Scott Shapiro, “On Hart’s Way Out,” Legal Theory, vol. 4, no. 4 (1998), 469–507, 490; em-
phasis added. Hereinafter HWO.
176

the idea that the conceptual function of law is to regulate behavior through
the guidance of norms implies the following thesis about the nature of law:
Practical Difference Thesis (PDT): Every conceptually possible legal norm
must be metaphysically capable of either epistemically or motivationally guiding
subjects.
PDT is an eminently plausible way to flesh out the conceptual truism that
every norm must be capable of guiding behavior as it distinctively applies to
law. It is conceptually possible for a piece of propositional content, n, to per-
form law’s function of guiding subjects only if n is metaphysically capable of
changing either what the subject believes about what she should do or what
the subject does. If n is metaphysically incapable of changing either what the
subject believes about what she should do or what she does, it is metaphys-
ically incapable of making the kind of practical difference that something
must be able to make to be properly characterized as a legal norm that governs
the subject’s behavior. Thus, if n is metaphysically incapable of either motiva-
tionally guiding or epistemically guiding a subject’s behavior, n cannot make
the kind of practical difference that a piece of content must be able to make
to be a law.
The two classes of legal norms (i.e. those that govern official behavior and
those that govern non-official behavior), as Shapiro interprets Hart, guide be-
havior in different ways. Since Hart’s minimum conditions for the existence
of a legal system require that subjects generally comply with laws governing
non-official behavior but do not require a specific motivation, the valid legal
norms of the system must be metaphysically capable of epistemically guiding
non-official behavior. But since the minimum conditions for the existence of
a legal system require that officials take the internal point of view toward the
rule of recognition governing official behavior, the rule of recognition must
be metaphysically capable of motivationally guiding official behavior.
4. Law and the guidance of non-official behavior

Shapiro articulates two Guidance Arguments purporting to show that the
norms of an institutional normative system with moral criteria of validity
cannot satisfy PDT. The first attempts to show that the norms of an insti-
tutional normative system with moral criteria of validity cannot guide non-
official behavior in the requisite manner. The second attempts to show that
the norms of an institutional normative system with moral criteria of validity
cannot guide official behavior in the requisite manner. Either one of these
arguments, if sound, is sufficient to refute the Incorporation Thesis.
Law and the guidance of non-official behavior 177
4.1 The first Guidance Argument
The first of the two Guidance Arguments attempts to show that it is not con-
ceptually possible for the norms of an institutional normative system with
moral criteria of validity to epistemically guide non-official behavior:
It is hard to see, however, how the law can serve this [guidance] function with respect
to rules that are valid in virtue of their moral content . . . . Marks of authority are sup-
posed to eliminate the problems associated with people distinguishing for themselves
between legitimate and illegitimate norms. However, a mark that can be identified
only by resolving the very question that the mark is supposed to resolve is useless.
Therefore, a norm that bears such a trivial mark . . . is unable to discharge its epistemic
duties. (HWO 494–5)7
This entails, on Shapiro’s view, that the norms of an institutional norma-
tive system with moral criteria of validity cannot guide non-official behavior
in a way that would satisfy PDT. The norms of an institutional normative
system valid in virtue of moral merit cannot epistemically guide non-official
behavior because “[t]‌elling people that they should act on the rules that they
should act on is not telling them anything” (HWO 494). If it is conceptually
impossible for an institutional normative system with moral criteria of va-
lidity to satisfy PDT and it is conceptually impossible for there to be a legal
system that does not satisfy PDT, then it is conceptually impossible for there
to be a legal system with moral criteria of validity.
7
In Legality, Shapiro makes a similar argument in terms that reflect his view of law as consisting
of shared plans: “If law is to guide behavior in the manner of plans, then it follows that its existence
and content cannot be determined by facts whose existence the law aims to settle. For if the existence
or content of law were determined in such a manner, then the proper way to ascertain its existence or
content would be to deliberate about the merits of different courses of action. But the point of hav-
ing plans is to obviate this very activity. It would be self-defeating, in other words, to have the plans
do the thinking for us if the right way to discover their existence or content required us to do the
thinking ourselves . . . . The problem with inclusive positivism is that it . . . violates [this principle]. If
the point of having law is to settle matters about what morality requires so that members of the com-
munity can realize certain goals and values, then legal norms would be useless if the way to discover
their existence is to engage in moral reasoning.” Scott Shapiro, Legality (Cambridge, MA: Harvard
University Press, 2011), 275. The criticisms of the Guidance Argument articulated below are equally
applicable to the version of the argument expressed in Legality.
It is worth noting that Shapiro’s characterization of law’s conceptual function seems to conflict
with the Separability Thesis. The only conceptually possible way to settle a dispute about what mo-
rality requires is to recognize, apply, and enforce the mandatory norm that reproduces the content of
the applicable mandatory moral norm. But this entails that only norms capable of settling disputes
about what morality requires can count as valid law and hence that the criteria of legal validity
necessarily include moral norms constraining the content of law, which is inconsistent with the
Separability Thesis.
178

4.2 Can norms valid in virtue of moral merit epistemically
guide non-official behavior?
The claim that it is not possible for norms valid in virtue of moral merit to
epistemically guide non-official behavior seems vulnerable to a straightfor-
ward objection: if it is possible, as is presupposed by our ordinary practices,
for mandatory moral norms to epistemically guide non-official behavior, it
would also have to be possible for legal norms valid in virtue of moral merit
to epistemically guide non-official behavior. It is hard to see how an offi-
cial act of recognizing, applying, or enforcing a mandatory moral norm that
is antecedently capable of epistemically guiding non-official behavior would
render that norm incapable of doing so simply because officials treat it as a
mandatory legal norm.
Consider an institutional normative system that validates all and only
mandatory moral norms. The valid norms of the system would presumably
include (1) a prohibition on intentionally killing innocent persons; (2) a pro-
hibition on theft; and (3) a prohibition on engaging recklessly in activities
highly likely to cause life-threatening injuries to other persons. Each one of
these valid norms reproduces the content of the corresponding moral norm
and expresses exactly as much content with respect to what subjects must do
as the corresponding moral norm does.
This seems to entail that it is conceptually possible for an institutional
norm valid in virtue of moral merit to epistemically guide non-official behav
ior. If we are morally accountable for our behavior, it must be possible for
us both to discern what a mandatory moral norm requires by consulting the
norm and to conform our behavior to that norm. If it is therefore possible for
a subject to be epistemically guided by mandatory moral norms governing
non-official behavior, then it must also be possible for a subject to be episte-
mically guided by an institutional norm governing non-official behavior that
is valid in virtue of its reproducing the content of one of these mandatory
moral norms. Simply giving a norm a new name and treating it as a member
of a different class of norms cannot change what subjects can learn about its
requirements by consulting it.
Thus construed, the first Guidance Argument misfires immediately.
Assuming there are mandatory moral norms governing non-official behavior,
an institutional norm valid in virtue of moral merit can do everything it must
be able to do to be properly characterized as a law governing non-official
behavior. To be properly characterized as a norm, a piece of content must be
metaphysically capable of epistemically guiding the behavior of subjects; to
be properly characterized as a norm that governs the behavior of some class
of subjects, it must be nomologically possible for subjects to obey that norm.
An institutional norm valid in virtue of moral merit is therefore metaphysi-
cally capable of epistemically guiding non-official behavior if moral norms are
metaphysically capable of doing so.
4.3 Can subjects learn of their legal obligations regarding

non-official behavior from the rule of recognition?
One might argue that the foregoing criticism of the first Guidance Argument
overlooks that a putative legal system with moral criteria of validity cannot
properly inform subjects of which norms governing non-official behavior are
legally valid. On this line of response, it is metaphysically possible for norms
governing non-official behavior to epistemically guide non-official behavior
only if subjects can discern from the rule of recognition which norms are valid.
But a rule of recognition that validates all and only moral norms tells subjects
no more than that they “should act on the rules that they should act on” is
not telling them anything that would help them to determine which norms
governing non-official behavior are valid.
The underlying idea is plausible. It is hard to understand, as a practical
matter, how a subject S could learn of her non-official obligations under a
norm n that belongs to a system Δ by consulting n if S has no reliable way to
determine that n is a member of Δ without having to figure out whether n
should be a member of Δ. Practically speaking, there is no way for a subject
to determine what she must do according to a norm governing non-official
behavior that belongs to the system unless she can determine that something
is a norm of the system without having to deduce that it is a norm of the
system on the strength of her judgment that it should be a norm of the system.
Otherwise, she is relying on her own substantive judgment to identify a norm
that is supposed to be identifiable by its authoritative “mark.”
The claim here is not that it must be conceptually possible for an inclusive
rule of recognition to epistemically guide non-official behavior. That claim
is clearly false. It is not conceptually possible for a rule of recognition of any
kind, source-based or otherwise, to epistemically guide non-official behavior
because the rule of recognition governs only official behavior. It might be
true that officials must be able to learn of their obligations under the rule of
recognition since it governs official behavior. But there is no reason to think
that non-officials must be able to learn of their obligations under the rule of
recognition because they have none; by definition, the rule of recognition
governs only official behavior.
Consider the First Amendment to the U.S. Constitution, which provides
that “Congress shall make no law . . . abridging the freedom of speech.” Since
180

I am not Congress and I cannot make law, it is conceptually impossible for
anything I do to be motivationally guided or epistemically guided by the
First Amendment. Since the First Amendment, by its own terms, defines no
obligations that bind me as a non-official, my behavior can neither violate
nor comply with it and hence can neither be motivationally nor epistemically
guided by it. It is simply not conceptually possible for a norm to motiva-
tionally or epistemically guide behavior it does not govern. Norm guidance
presupposes norm governance.
Insofar as the first Guidance Argument presupposes, as it must, this con-
ceptual truism, the operative claim is that it must be conceptually possible
for subjects of a legal system to identify what valid law requires with respect
to non-official behavior by consulting the rule of recognition. In order for
subjects to determine what valid law requires of them, they must be able to
determine which norms are valid, and this can be done only insofar as they
can accurately distinguish norms that are valid from norms that are not by
consulting the rule of recognition. Since it is not possible for the subject to
determine which norms are valid by consulting an inclusive rule of recogni-
tion, it follows, on this line of reasoning, that there cannot be a legal system
with moral criteria of validity.
The claim that it must be conceptually possible for subjects of valid norms
to determine which norms are valid by consulting the rule of recognition
does nothing to rescue the first Guidance Argument. If we are morally ac-
countable for our behavior, then it must be true not only that we can be
epistemically guided by mandatory moral norms, but also that we can deter-
mine which norms are mandatory moral norms. We could not be morally ac-
countable for our behavior were we not capable of determining which norms
correctly express moral requirements—i.e. which norms are members of the
set of mandatory moral norms. To “know the difference between right and
wrong,” we must be able to determine (1) which norms belong to the set of
mandatory moral norms that govern our behavior and (2) what those norms
require of us.
But if a person can determine which mandatory norms are moral norms
and hence what her moral obligations are under a system of mandatory moral
norms, then she can also learn which norms are valid and what her legal
obligations are under valid law from a rule of recognition that validates all
and only mandatory moral norms. A rule of recognition that validates all and
only mandatory moral norms tells competent subjects that mandatory moral
norms, which those subjects can and should know govern their behavior, will
be recognized by judges as valid law and will hence be applied and enforced
against them as such with respect to their non-official behavior. If the claim is
that it must be possible for subjects of a legal system to learn which norms are
valid and what their legal obligations are with respect to non-official behav-
iors from the rule of recognition, there is no reason to think that a rule of
recognition that defines moral criteria of validity cannot inform subjects of
their non-official obligations to the degree necessary for law to perform its
guidance function.
Consider, again, a rule of recognition in an institutional normative system
that validates all and only mandatory moral norms; and suppose that the
valid norms of the system are backed with coercive enforcement mechanisms.
Suppose further that everyone within the relevant jurisdiction is contingently,
but not necessarily, infallible with respect to what morality requires of her.8 If
the epistemic issue faced by subjects in this conceptually possible world is to
determine which acts are required or prohibited by valid institutional norms,
then knowing that all and only mandatory moral norms are valid can tell
them exactly what they must do to comply with the valid norms of the system.
Notice that someone can be infallible with respect to what morality
requires only if she can infallibly determine not only what every mandatory
moral norm requires but also which norms belong to the system of manda-
tory moral norms. Subjects in this conceptually possible world can determine
from the rule of recognition that every norm they antecedently know belongs
to the system of mandatory moral norms is also legally valid: such a rule of
recognition tells subjects that their legal obligations governing non-official
behavior require them to do what they already know mandatory moral norms
require them to do.
The claim that people cannot learn of their obligations with respect to
non-official behavior from an inclusive rule of recognition, then, mistakenly
assumes that they can learn no more from an inclusive rule of recognition
than that “they should act on the rules they should act on.” An institutional
normative system that tells citizens no more about their non-official obliga-
tions than that would be no more informative than a judge’s decision that
tells litigants that they should do what they should do. But a rule of recog-
nition that validates norms in virtue of moral merit conveys something far
more useful than that: it informs citizens that moral requirements will be ap-
plied against them as law and possibly enforced against them with the police
power of the state. Again, if mandatory moral norms are capable of informing
subjects what they must do to satisfy their moral obligations, then a rule of
recognition that validates legal norms in virtue of their moral merits is like-
wise capable of informing subjects what they must do to satisfy their legal
obligations under valid law with respect to their non-official behavior.
8
See Chapter 8, Section 3.2, for a straightforward proof that such a world is not only conceptu-
ally possible but is also nomologically possible.
182

It is unlikely, of course, that the system described above could be effica-
cious in a world like ours because we are not contingently infallible about
moral issues. But the claim being defended here is not that a system like this
could be successful enough in regulating our behavior to ensure that we can
live and cooperate in peace. The claim is, rather, that there is a conceptually
possible world in which such a system can make a practical difference with re-
spect to the non-official behavior of its subjects. Although it is highly unlikely
that such a state of affairs would be realized among beings with our abilities
and disabilities, it is conceptually possible—and that is all that is needed,
strictly speaking, to respond to this line of argument. This construction of the
first Guidance Argument fails.
4.4 Must subjects be able to learn of their legal obligations

regarding non-official behavior from the rule
of recognition?
Notice that this construction of the first Guidance Argument depends on the
claim that the conceptual function of the rule of recognition is, in part, to in-
form subjects of their non-official obligations under valid legal norms. Insofar
as the rule of recognition must therefore be capable of informing subjects of
their non-official obligations under valid laws, a norm that is incapable of
informing subjects of their non-official obligations under the valid norms of
the system is not properly characterized as part of a rule of recognition. Since,
on this line of reasoning, a putative recognition norm incorporating moral
criteria of validity is metaphysically incapable of performing this function, it
cannot be part of any conceptually possible rule of recognition that succeeds
in giving rise to a legal system.
But it is implausible to suppose that subjects must, as a conceptual matter,
be able to learn of their legal obligations with respect to non-official behavior
by consulting the rule of recognition. The conceptual function of a rule of
recognition is to govern the behavior of officials in their legislative, adjudica-
tive, and enforcement activities. It might be true that subjects must be able to
learn what their official obligations are from the rule of recognition because
it defines those obligations, but there is nothing in any plausible conceptual
theory of law that would entail that subjects must be able to learn what their
non-official obligations are from the rule of recognition because it cannot, by
definition, govern non-official behavior. There is simply no reason to think
that subjects must, as a conceptual matter, be able to learn from a norm that
cannot govern their behavior what their obligations are under other norms that
do govern their behavior.
It is, of course, a conceptually necessary condition for the existence of a
legal system that the behavior of subjects generally conforms to the valid
norms of the system, but it is clearly possible for a legal system to efficaciously
guide behavior without subjects being able to learn from the rule of recognition
what their obligations are under valid law. On the one hand, it is unlikely,
given the depressingly low rate of constitutional literacy in the U.S., that
many law subjects know much about what the Constitution says or under-
stand how its requirements figure into determining what counts as valid law.9
On the other hand, the institutional normative system that the Constitution
brings into existence is clearly efficacious enough in guiding non-official be-
havior that it is beyond dispute that this system is a system of law. Regardless
of whether subjects can identify their non-official obligations by consulting
the rule of recognition, it is clear that the U.S. not only has a legal system but
one that sufficiently exemplifies the nature of law that it constitutes a para-
digm instance of law.
Surprisingly, it is not obviously a conceptually necessary condition for the
existence of a legal system that most subjects have even a clue about what
valid law requires with respect to non-official behavior. If their behavior al-
ways happens to conform to the valid institutional norms that would be ap-
plied or enforced against them if they behaved differently, the fact that they
do not know what those norms require does not clearly preclude character-
izing either the valid norms or the system itself as “law”; most of us, after all,
are sufficiently disinclined to commit murder, fraud, theft, and a host of other
anti-social acts that we would abstain from committing those acts even if they
were not prohibited by law and even if we were not aware that these acts are
prohibited by law.
One could not plausibly deny that, in conceptually possible worlds like
ours in which intellectually limited self-interested beings like us live in condi-
tions of material scarcity, an institutional normative system in which subjects
are better informed about the content of their legal obligations is more likely
to keep the peace than one in which subjects are less informed about the con-
tent of their legal obligations. But this is relevant only with respect to whether
it is probable that some given institutional normative system will succeed in
keeping the peace in our world. This tells us nothing about whether it is
9
Hart puts this obvious point as follows: “[I]‌n a modern state it would be absurd to think of
the mass of the population, however law-abiding, as having any clear realization of the rules specify-
ing the qualification of a continually changing body of persons entitled to legislate . . . . We would
only require such an understanding of the officials or experts of the system; the courts, which are
charged with the responsibility of determining what the law is; and the lawyers whom the ordinary
citizen consults when he wants to know what the law is.” H.L.A. Hart, The Concept of Law, 3rd Ed.
(Oxford: Clarendon Press, 2012), 60. Hereinafter CL.
184

conceptually possible for an institutional normative system in which subjects
know little about the content of their legal obligations can succeed in keeping
the peace enough to count as a system of law.
The state of affairs in which an institutional normative system succeeds
in keeping the peace in our world without informing subjects of what the
norms require is, admittedly, a fringe state of affairs; however, to say that a
state of affairs is fringe is to say only that it is a highly improbable state of af-
fairs in our world. Again, the issue with respect to whether the Incorporation
Thesis is true is whether it is possible for an institutional normative system
with moral criteria of validity to do everything it must do to be properly
characterized as a system of law—and not whether it is probable that there
is a legal system with moral criteria of validity in our world. It is compatible
with the Incorporation Thesis that there never has been and never will be a
legal system in our world with moral criteria of validity.
It is not even true, as a purely practical matter, that a legal system in our
world could not be efficacious unless subjects are able to ascertain by con-
sulting the rule of recognition which norms are legally valid or what these
norms require by way of non-official behavior. Most subjects in existing legal
systems have enough of an idea of what valid law requires to steer clear of
problems with the law. I knew long before I ever studied criminal law, con-
tract law, and tort law that I am legally obligated not to kill or otherwise
assault people, to comply with the terms of my contracts, to refrain from
fraud and tax evasion, and to be reasonably careful not to cause injury to
other persons. In retrospect, I cannot recall how I knew; maybe I learned
from television that the law tends to frown on gratuitous violence, breach of
contract, fraud, tax evasion, and negligence. But I know that I, and every sen-
sible person over the age of sixteen I have ever known, knew as much about
the law as I did prior to studying it—without having a clue what the rule of
recognition is. There are clearly other ways for subjects to learn of what valid
law requires with respect to non-official behavior than by consulting the rule
of recognition.
4.5 A methodological objection: The argument is illicitly

grounded in contestable claims about morality
One might argue that the foregoing analysis is methodologically problem-
atic insofar as it grounds a claim about the nature of law in a contestable
claim about morality—namely that we are morally accountable for our
behavior. This is problematic because, as Hart puts it, “legal theory should
avoid commitment to controversial philosophical theories of the general
status of moral judgments” (CL 253–4). To avoid committing his theory to
such theories, Hart made his commitment to the Incorporation Thesis explic-
itly conditional on the assumption that moral norms are objective and “left
open . . . the general question of whether they have what Dworkin calls ‘ob-
jective standing’ ” (CL 253). Insofar as the argument above is grounded in the
contestable claim that we are morally accountable for our behavior, it fails to
avoid a commitment to controversial philosophical theories of morality and
is problematic for this reason.
As it turns out, however, there is nothing methodologically problematic
with assuming that we are morally accountable for our behavior. As discussed
in Chapter 2, a modest approach to conceptual analysis purports to give us
insight into what the world is like as we structure it through the conceptual
framework we impose on the world through our linguistic and other social
practices; that is, conceptual analysis purports to explicate what the world is
like as we view it through the social practices we adopt to describe, structure,
and make sense of our experience of the world. Given that we converge on a
set of conceptual practices that presuppose that we are morally accountable
for our behavior, it is perfectly legitimate for a modest conceptual theory of
law to take that into account.
This is why it is true, as a general methodological matter, that an explication
of the descriptive concept of law defined by our conceptual practices “should
avoid commitment to controversial philosophical theories of the general status
of moral judgments.” A claim is controversial in virtue of being disputed by a
large enough segment of the population that it cannot be plausibly character-
ized as capturing our conceptual practices, which are ours in virtue of being
widely practiced by us.
These conceptual practices would be false if it turns out that there is
nothing in the world—as it “really” is independent of the conceptual frame-
work we impose on it—that constitutes a system of morality that governs
our behavior, but that is not relevant on a modest approach; what is relevant
is whether our conceptual practices presuppose that there is such a system.
It would be methodologically illegitimate for a conceptual theory of law to
presuppose that morality is objective in character because that claim is not
presupposed by our conceptual practices; after all, many people in the rel-
evant community of speakers believe that morality is conventional or “rel-
ative to culture.” But it is no more problematic for a conceptual theory of
law to presuppose that we are morally accountable for our behavior than it
is to presuppose that we are not being deceived by a Cartesian evil demon;
our conceptual practices clearly presuppose that the material world exists and
resembles what our senses tell us about it and that our behavior is governed by
a system of moral norms, regardless of whether they are objective in character.
186

As it turns out, the first Guidance Argument and the Arguments from
Authority also depend on the assumption that we are morally accountable
for our behavior. The first Guidance Argument is grounded in the claim that
“[m]‌arks of authority are supposed to eliminate the problems associated with
people distinguishing for themselves between legitimate and illegitimate
norms” and hence presupposes that there are norms that legitimately govern
our behavior; however, a norm can legitimately govern our behavior only if
we are legitimately held accountable under it. The Arguments from Authority
are likewise grounded in the claim that the conceptual function of authority
is to mediate between subjects and “right reason,” which is “right” in virtue
of conforming to a correct assessment of what objective norms require, which
include norms of morality under which we are properly held accountable.
These arguments, then, all rise or fall with the counterarguments given
above. If it were methodologically problematic for a conceptual theory of law
to presuppose that there exists a system of mandatory moral norms governing
our behavior, then the counterarguments to the first Guidance Argument
given above would fail to refute it. But if that assumption is methodologically
problematic, the counterarguments above would fail, but so would the first
Guidance Argument and the Arguments from Authority for the very same
reason.
5. Law and the guidance of official behavior
5.1 The second Guidance Argument

Successfully rebutting the first Guidance Argument is not enough to vin-
dicate inclusive positivism because there is another that purports to iden-
tify a second inconsistency between the Incorporation Thesis and PDT. The
problem, on this line of reasoning, is that the possibility of moral criteria of
validity is inconsistent with the claim that all the relevant norms in a legal
dispute are capable of making a practical difference in the deliberations of the
judge. As Shapiro explains:
Consider, therefore, the following inclusive rule of recognition: “In hard cases, act ac-
cording to the principles of morality.” In Riggs, judges guided by this inclusive rule of
recognition would conform with the principles of morality when deciding whether to
invalidate the will. Let us further assume that the only relevant principle of morality
is that people should not profit from their own wrongs and that the majority in that
case believed this to be so . . . . [T]‌he principle that no man should profit from his
wrongs cannot itself make a practical difference as a legal norm. For if the judge were
guided by the inclusive rule of recognition, but did not appeal to the moral principle,
Law and the guidance of official behavior 187
he or she would still end up invalidating the will . . . . The moral principle, therefore,
can make no practical difference once the rule of recognition makes a practical differ-
ence, because the judge will act in exactly the same way whether he or she personally
consults the moral principle or not. Guidance by the inclusive rule of recognition by
itself is always sufficient to give the judge the right legal answer. (HWO 496)
The problem, on this line of reasoning, is that a rule of recognition that vali-
dates all and only mandatory moral norms requires the judge to do exactly
the same thing that is required of her by the applicable norm valid in virtue
of reproducing the content of the relevant mandatory moral norm. A rule
of recognition directing the judge to apply mandatory moral norms in hard
cases dictates that the judge not allow a murderer to take under the will of
her victim. But this is exactly the same result dictated by the Riggs principle,
which is valid in virtue of being a mandatory moral norm, that no person
should profit from her own wrong. Once a judge is motivated to conform
her behavior to the rule of recognition because it is the rule of recognition,
she cannot also be motivated to conform her behavior to the Riggs principle
because it is a valid norm; thus, it is not possible for both norms to make a
practical difference in the deliberations of the judge. Since PDT implies that
every legal norm must be capable of making a practical difference, the norms
of an institutional normative system valid in virtue of moral merit fail to sat-
isfy PDT and hence, according to the second Guidance Argument, are not
properly characterized as law.10
The second Guidance Argument seems to refute the Incorporation Thesis.
Insofar as an institutional normative system with moral criteria of validity
lacks any source-based criteria of validity, every valid norm of the system
would be valid in virtue of moral merit. This means, however, that none of
the norms valid under its rule of recognition would be capable of motiva-
tionally guiding the official behavior of a judge who is motivationally guided
10
Strictly speaking, what this implies is that norms valid only in virtue of moral merit, like the
Riggs principle, are not norms of law; it does not imply that the system itself is not a system of law.
If there are a sufficient number of norms in the system valid partly in virtue of having an official
source, then it might be that there remain enough norms in the system to warrant characterizing it
as a legal system. In that case, the system would, despite any appearances to the contrary, be a legal
system that has only source-based criteria of validity. If, for example, a legislature decided to codify
the minimum content of natural law, which would also be valid in virtue of moral merit, the norms
that express that content might still be properly characterized as norms of law. While norms valid
only in virtue of moral merit would not be properly characterized as legal norms, there would still
be enough content valid in virtue of source to satisfy all the conceptually necessary conditions for
the existence of a legal system. But if there are not a sufficient number of norms in the system that
have an appropriate source, there would not be enough valid content in the system to satisfy all the
conceptually necessary conditions for the existence of a legal system; in that case, the relevant social
practices would not give rise to something properly characterized as a legal system.
18

by the rule of recognition to apply those norms. Since no valid norm of the
system is capable of making a practical difference in the deliberations of a
judge motivationally guided by the rule of recognition, none would be prop-
erly characterized as a legal norm. The system would thereby have a putative
rule of recognition without having any valid legal norms and would hence
clearly not be properly characterized as a legal system. There can be no law
without laws.
But insofar as an institutional normative system with moral criteria of
validity also incorporates source-based criteria of validity as a necessary
condition for a norm to be valid in the system, only the norms of the
system that are valid partly in virtue of source can make a practical differ-
ence in the deliberations of a judge motivationally guided by the rule of
recognition. This means, however, that every norm putatively valid wholly
in virtue of moral merit would not be a legal norm under PDT. Since,
in this case, there are no norms valid in virtue of moral merit that count
as law, the system would have, despite appearances to the contrary, only
source-based criteria of validity.
Either way, the system is not properly characterized as a legal system with
moral criteria of validity. If, on the one hand, all the norms expressing, say,
the minimum content of natural law have not been officially promulgated
and are putatively valid only in virtue of moral merit, then they are not legal
norms and the system fails, for that reason, to be a legal system. If, on the
other, all the norms expressing the minimum content of natural law have
been officially promulgated and are hence valid in virtue of source, then the
system is a legal system; those norms would be valid in virtue of having a
proper source. But since every norm of the institutional system that is puta-
tively valid wholly in virtue of moral merit would not be a legal norm, the
legal system to which that institutional system gives rise would lack moral
criteria of legal validity.
5.2 Can a valid legal norm governing non-official

behavior motivationally guide a judge deciding a case
under that norm?
To evaluate the second Guidance Argument, it is crucial to note that there
is an ambiguity with respect to the Riggs principle. It can be construed as
stating either a recognition norm that governs the official behavior of judges
or as stating a valid legal norm that governs non-official behavior. On the
first construction, the Riggs principle defines an obligation on the part of
judges not to allow someone to profit from her own wrong; on the second,
it prohibits a subject from doing something that amounts to profiting from
her own wrong.
The second construction is problematic because the Riggs principle is not
plausibly treated as a norm governing non-official behavior. To begin, if the
opinion of the Riggs court is any indication, the Riggs principle is a recogni-
tion norm governing the official behavior of judges—and not a valid norm
governing non-official behavior. As the court explains it, the Riggs principle is
being treated by the court as governing its authority to enforce wills:
Besides, all laws as well as all contracts may be controlled in their operation and effect
by general, fundamental maxims of the common law. No one shall be permitted to
profit by his own fraud, or to take advantage of his own wrong, or to found any claim
upon his own iniquity, or to acquire property by his own crime . . . . These maxims,
without any statute giving them force or operation, frequently control the effect and
nullify the language of wills. A will procured by fraud and deception, like any other
instrument, may be decreed void and set aside, and so a particular portion of a will may
be excluded from probate or held inoperative if induced by the fraud or undue influence
of the person in whose favor it is . . . . So a will may contain provisions which are im-
moral, irreligious or against public policy, and they will be held void. 11
Insofar as the court takes the Riggs principle to govern how it may decide the
case, the court is clearly treating the Riggs principle as a norm governing its of-
ficial behavior, and not as a norm governing non-official behavior. Thus con-
strued, the Riggs principle that “no one shall be permitted to profit by . . . his
own wrong” is not a valid legal norm that governs non-official behavior; it is
a norm that governs judicial behavior and is hence part of the rule of recogni-
tion in virtue of being practiced by the officials of the system.
Further, the Riggs principle is just not plausibly interpreted as a norm gov-
erning non-official behavior. Construed as such, the Riggs principle defines a
legal obligation on the part of a subject not to profit from her own wrongful
non-official behavior. Thus construed, a subject would violate the Riggs prin-
ciple not by engaging in a wrongful non-official behavior from which she
profits, but rather by attempting to profit from the wrongful non-official
behavior.
This construction might seem intuitively plausible at first blush, as bur-
glary, theft, and robbery would all violate the Riggs principle, but it is cru-
cial to note that what this interpretation would prohibit in the case of wills
is not murdering a testator to prevent her from changing her will. Rather,
this construction of the Riggs principle would prohibit a person from
attempting to claim under the will of someone she murdered. Moreover, on
11
Riggs v. Palmer, 115 NY 506 (1889), at 512. Emphasis added.
190

this construction, its application would not punish or fine the attempt of
the murderer to take under the will of her victim; it would simply require
a judge to nullify the will with respect to the murderer’s gift under the will.
It is, of course, conceptually possible for something like the Riggs principle
to govern non-official behavior, but this construction is so conspicuously
odd and alien to our experience with law that the court never considers it.
But if the Riggs principle is construed as governing official behavior, the
second Guidance Argument immediately misfires because the Riggs principle
would simply be a logical implication of a rule of recognition validating all
and only mandatory moral norms. Either the Riggs principle is properly char-
acterized as a legal norm or it is not. If it is properly characterized as a legal
norm despite being incapable of motivationally guiding a judge who is moti-
vationally guided by the rule of recognition, then PDT, as it functions in this
argument, is false; it is not a conceptually necessary condition for a norm to
count as a valid legal norm that it is capable of making a practical difference
in the deliberations of a judge. If, however, it is not properly characterized as
a legal norm, it is because our conceptual practices dictate that only first prin-
ciples of law can count as norms. In this case, the Riggs principle, construed as
governing official behavior, would be a logical corollary of the first principle
that the judge should “act according to the principles of morality” (HWO
496). But the reason that the Riggs principle is not properly characterized as a
legal norm would then be that our more general conceptual practices dictate
that the more general term “norm” be applied only to first principles—and
not because our practices regarding the concepts of law or legal system are in-
consistent with the conceptual possibility of moral criteria of legal validity. If
the second Guidance Argument construes the Riggs principle as a recognition
norm, which is what it seems clearly to be, it is a non-starter.
The second Guidance Argument fares no better if the Riggs principle is con-
strued as a valid norm governing non-official behavior. Construed as a valid
norm governing non-official behavior, the Riggs principle merely prohibits
profiting from a murder. On this construction, the Riggs principle was clearly
metaphysically capable of motivationally and epistemically guiding Elmer,
the defendant in Riggs, who murdered his grandfather to prevent him from
changing his will. Insofar as the Riggs principle, on this construction, defines
a legal obligation on his part not to profit from the murder of his grandfather,
Elmer could have both (1) determined by consulting it that he should refrain
from attempting to claim under his grandfather’s will and (2) refrained from
doing so because it was required by the Riggs principle.
But the problem with this painfully contorted construction is that a valid
norm governing non-official behavior is metaphysically incapable of moti-
vationally or epistemically guiding the judge’s behavior because a norm that
does not govern a person’s behavior is metaphysically incapable of guiding
it. Even if the Riggs principle is valid in virtue of having a proper source, the
judge cannot learn what she is obligated to do as a judge by consulting that
principle because it governs nothing the judge can do to decide the case. Nor
can the judge conform her official behavior to the Riggs principle because it
does not govern that behavior.
It is true that the judge is applying the Riggs principle to evaluate Elmer’s
behavior, but that changes nothing. The judge does not thereby conform her
official behavior to the Riggs principle by evaluating Elmer’s behavior because
one can conform to the Riggs principle, construed as governing non-official
behavior, only by refraining from claiming under the will of one’s murder
victim.
That, of course, is something the judge can do at any time, including while
she is deciding the case, but there is nothing that the judge can do in deciding
the case that would violate that norm because that norm does not govern the
judge’s official behavior. If the judge, while conducting a hearing in the Riggs
case, were to claim under the will of someone she murdered, the judge would
be violating the Riggs principle on this interpretation of it; however, the act
of claiming under the will of someone the judge murdered would no more be
properly characterized as an official act because done during the hearing than
a sneeze would be if the judge sneezed during the hearing. While the judge
can be both motivationally and epistemically guided by the Riggs principle,
she can be guided by it only with respect to behavior that is not properly
characterized as official behavior.
It might be helpful to consider an example that does not involve a legal
norm that is ambiguous as to whether it governs official or non-official behav
ior. Consider an institutional normative system that validates all and only
mandatory moral norms, and suppose that a dispute arises under a norm,
valid in virtue of moral merit, that holds automobile manufacturers strictly
liable for injuries proximately caused by manufacturing defects. Notice that
the strict liability norm does not say anything about what the judge must do
to adjudicate disputes under the norm and hence defines no judicial duties;
it is the rule of recognition, which defines the powers and duties of officials
and validates all and only mandatory moral norms, that requires the judge to
apply the strict liability norm when relevant.
The reason is that, since judges are not automobile manufacturers, the rule
does not apply to anything a judge could do in deciding a case. It is the rule
of recognition that requires that the judge apply the valid norm to resolve a
dispute between an automobile manufacturer and someone who claims that an
injury was proximately caused by a manufacturing defect; the rule of recogni-
tion thereby defines what the judge must do to decide the case—and not the
192

strict liability norm. The only duty that the relevant valid legal norm defines,
by its own terms, is that of the defendant. In deciding the dispute, then, the
judge can follow the rule of recognition by applying the relevant valid legal
norm, but the judge cannot follow the relevant valid legal norm by applying
it to decide the case.
There is a conceptual difference between following a norm and applying
a norm. A person P follows a norm n by doing α, in the relevant sense, if
and only if (1) n requires that P do α; (2) P does α; and (3) P does α because
P believes that n requires that P do α. In contrast, P applies a norm n that
requires someone to do α, in the relevant sense, if and only if P evaluates the
behavior of someone whose behavior n governs (which might include P) to
determine whether the subject has satisfied n by doing α.
These two notions are logically independent. One can apply a norm without
following it. An arbitrator P might, for example, authoritatively decide a con-
tract dispute between Q and R by deciding whether R has performed her
duties under the terms of the contract without being, or thereby becoming,
a party to the contract. To do that, P must apply the relevant valid norms of
contract law to R’s behavior to determine whether it conforms to the terms of
the contract as required by contract law. But, in so doing, P does not thereby
follow either the terms of the contract or the valid legal norm that requires
parties to a contract to perform their contractual obligations. The claim that
P applies n does not logically imply that P follows n.
Conversely, one can follow a norm without applying it. I might follow
a norm that prohibits killing an innocent person simply by refraining al-
ways from killing anyone because I sensibly believe that it is better not to kill
someone the norm allows me to kill than to kill someone the norm does not
allow me to kill. But my refraining from killing some particular person does
not involve applying the norm to my own behavior because I am not evalu-
ating whether my behavior has satisfied the norm. If I beat someone nearly
to death and subsequently attempt to evaluate my behavior under the norm
prohibiting killing innocent persons, I am applying the norm. But my fol-
lowing the norm prohibiting killing in some particular instance involves an
altogether different act from my evaluating whether my behavior satisfies that
norm in that instance. The claim, then, that P follows n neither implies the
claim that P applies n nor is implied by that claim. Knowing that a person
has followed a norm tells us absolutely nothing about whether she has applied
it, and conversely.
The reason is that the behaviors that constitute one act, as a conceptual
matter, do not constitute the other. There is no conceptually possible norm
that one can simultaneously apply and follow by doing exactly the same
thing. The acts that constitute P’s following a norm, n1, that requires P to
Revisiting the Arguments from Authority 193
apply some other norm, n2, to Q’s behavior do not constitute applying n1.
On the one hand, to follow n1, P must evaluate Q’s behavior by applying n2;
on the other, to apply n1, P must determine whether someone else whose
behavior is governed by n1 has properly applied n2 to Q’s behavior. It might
be possible for someone to simultaneously follow and apply a norm, but they
will be doing two distinct things at the same time—one of which constitutes
following the norm and the other of which constitutes applying the norm.
It is, thus, conceptually impossible for the judge to follow a norm simply
in virtue of applying it to evaluate someone else’s behavior—regardless of
whether that norm governs her behavior. If the norm does not govern the
judge’s behavior, the act of applying the norm to evaluate someone else’s
behavior cannot satisfy the requirements of the norm as they pertain to the
judge; since the norm does not govern the judge’s behavior, it requires nothing
of her. But if the norm does govern the judge’s behavior, then the judge does
not conform her behavior to that norm in virtue of applying it because the
acts that constitute applying a norm are different from the acts that constitute
following it. Following and applying a norm are different notions that have
nothing more salient in common than that the corresponding concept-terms
involve doing something with norms.
The problem with this second Guidance Argument is, in essence, the same
as the problem with the first—namely that it presupposes that someone’s acts
must be guided, in some intuitive sense, by a norm that does not govern the
relevant acts. There is nothing in any plausible conceptual theory of law that
entails that it must be possible for the rule of recognition to epistemically
guide or inform non-official behavior because the rule of recognition does not
govern that behavior. Similarly, there is nothing in any plausible conceptual
theory of law that entails that the valid legal norms that a judge applies to a
case involving non-official behavior must be capable of motivationally guid-
ing the judge’s official behavior because those norms do not govern the judge’s
official behavior. Since neither strand of the Guidance Argument shows that
the norms of an institutional normative system with moral criteria of validity
cannot properly guide someone whose behavior it must be capable of guid-
ing, the Guidance Arguments fail to refute the Incorporation Thesis.
6. Revisiting the Arguments from Authority: To whom

the rules apply
The Arguments from Authority are vulnerable to a similar objection. The
Arguments from Authority are grounded in the idea that the conceptual
194

function of authority is to provide directives that (1) express the authority’s
specific view about what the subject should do according to right reason;
(2) are capable of replacing the subject’s judgment in her deliberations about
what she should do according to right reason; and (3) enable the subject to
better comply with what right reason requires by following the authority’s
view than by following her own view.
The Arguments from Authority purport to show that the norms of an in-
stitutional normative system with moral criteria of validity can do none of
these things. Since a subject cannot identify what the relevant norms require
except by discerning what she believes right reason requires, those norms are
incapable both of replacing her judgment about what she should do in her
deliberations about what to do and of expressing the authority’s view about
what she should do. Likewise, since a subject cannot determine what she
should do under the relevant norms without discerning what she believes she
should do according to right reason, it is conceptually impossible for her to
better comply with right reason by following the authority’s judgment than
by following her own; since the relevant norms cannot express the authority’s
view, it is conceptually impossible for her to follow the authority’s view.
The similarities between the Guidance Arguments and the Arguments from
Authority are conspicuous. An institutional normative system that validates
all and only mandatory moral norms cannot make the practical difference
in the deliberations of subjects that something must be able to make to be
properly characterized as a legal system. As Shapiro describes the problem, “a
mark that can be identified only by resolving the very question that the mark
is supposed to resolve is useless” (HWO 495). As Raz describes the problem,
“[a]‌decision is serviceable only if it can be identified by means other than
the considerations the weight and outcome of which it was meant to settle”.12
The reason for these similarities is that both strategies of argument con-
ceive the notion of practical authority partly in terms of its being capable of
playing an epistemic role in “mediating” between subjects and norms. The
point of authority, as Shapiro conceives it, is to mediate between “competing
standards” by informing subjects which of competing standards they must
satisfy. The point of authority, as Raz conceives it, is to mediate between sub-
jects and the reasons that antecedently apply to them by providing directives
that can replace their judgments about what they ought to do in their prac-
tical deliberations. In both cases, the authoritative character of law requires
that law be metaphysically capable of playing these mediating roles in the
practical deliberations of subjects.
12
Joseph Raz, Ethics in the Public Domain (Oxford: Oxford University Press, 1994), 219.
Revisiting the Arguments from Authority 195
This similarity, by itself, is anodyne as far as the Arguments from Authority
are concerned, but those arguments share a second feature with the Guidance
Arguments that is fatal to each. The Arguments from Authority, like the
Guidance Arguments, presuppose that it is a conceptually necessary condi-
tion for the existence of law that the rule of recognition is metaphysically
capable of informing subjects of their legal obligations with respect to non-
official behavior. Since a rule of recognition, according to the Arguments
from Authority, validating all or only mandatory moral norms is incapable of
informing subjects of what their obligations are with respect to non-official
behavior, it cannot be authoritative and hence cannot ground an institutional
normative system that is properly characterized as a system of law.
The problem with both lines of reasoning is, for this reason, exactly the
same: there is simply no reason to think, as the Guidance Arguments and
Arguments from Authority assume, that one conceptual function of the rule
of recognition is to inform subjects of their non-official obligations under the
valid norms of the system. If the rule of recognition must, as a conceptual
matter, play an epistemic role in informing anyone of what her obligations
are or which of competing standards she must follow, it need inform only
officials of what they ought to do with respect to recognizing, applying, and
enforcing law. Insofar as each of these arguments depends on the claim that
the rule of recognition must be able to inform subjects of their non-official
obligations, neither gets out of the blocks and neither hence succeeds in refut-
ing the Incorporation Thesis.
196
8
The Conceptual Possibility of Moral Criteria
of Legal Validity
This volume has been concerned with the engagement between the concepts
of law and morality as defined by our ordinary linguistic and legal practices.
There are only three possibilities with respect to this engagement as it pertains
to the criteria of legal validity. The descriptive anti-positivist claims that the
criteria of validity must include moral constraints on the content of law. The
inclusive positivist claims that the criteria of validity can include moral con-
straints on the content of valid law. The exclusive positivist claims that the cri-
teria of validity cannot include moral constraints on the content of valid law.
These three positions exhaust the logical space of conceptual claims regarding
whether the criteria of validity include moral norms: the criteria of validity
either must, can, or cannot contain moral constraints on the content of law.
This chapter completes the defense of the Incorporation Thesis undertaken
in the last two chapters with a positive argument for the claim that the criteria
of validity can incorporate moral constraints on the content of law. The argu-
ment constructs a model of an institutional normative system that validates
all and only mandatory moral norms in a possible world that resembles ours
in every causal respect and shows that the system satisfies every condition
plausibly thought conceptually necessary for the existence of a legal system.
1. General methodological considerations

It is important to note the character of the claim that the Incorporation Thesis
makes about moral criteria of validity. To begin, it neither asserts nor implies
that there are, have been, or ever will be legal systems with moral criteria of
validity in our world. Insofar as the Incorporation Thesis makes a claim only
about what is conceptually possible, it entails nothing about whether there
are, have been, or will be, legal systems in our world with moral criteria of
validity. Just as the claim that it is possible that Bigfoot exists tells us nothing
University Press.
198
198 The Possibility of Moral Criteria of Validity

about whether Bigfoot does exist, the claim that it is possible that there are
legal systems with moral criteria of validity tells us nothing about whether
there are legal systems with moral criteria of validity.
Further, insofar as the Incorporation Thesis makes a claim only about what
is conceptually possible, it entails nothing about whether such a system is
nomologically possible—i.e. possible in worlds with the same laws of na-
ture that obtain in our world. It is obviously true that what is nomologically
possible is conceptually possible, but it is not true that what is conceptually
possible is nomologically possible. A state of affairs might be realized only in
conceptually possible worlds with causal properties so different from those of
our world that their inhabitants have some characteristic that we could not
have because we and our world are just not built that way.
Consider the idea that it is conceptually possible for there to be a legal
system without sanctions in a world where beings are always conclusively
motivated by altruistic love. It is not clear that such a claim tells us anything
about whether it is nomologically possible for there to be such a system be-
cause it is not clear whether it is nomologically possible for inherently self-
interested beings like us who live in conditions of material scarcity always to
put aside our own interests in the service of altruistic concern for others. If,
on the one hand, (1) it is nomologically possible for us always to do this and
(2) there are no other features of this conceptually possible world that are
nomologically impossible, then this world is both conceptually and nomolog-
ically possible. But if, on the other hand, (i) it is not possible for us always to
do this and (ii) the regulatory efficacy of the system depends critically on this
feature of the subjects’ psychology, then such a legal system is conceptually
but not nomologically possible.
But even if we know that some state of affairs S is nomologically possible,
that tells us next to nothing about whether it is likely to be realized in our
world. The claim that S is nomologically possible entails only that the proba-
bility of its being realized in our world is greater than zero. This tells us little
about whether S is likely to be realized in our world because, for all we can
infer from the nomological possibility of S, it might be that the probability of
realizing S in our world is .99, or it might be that the probability of realizing
S in our world is .0000000000000001.
The probability that S occurs in a nomologically possible world will vary
from one world to the next. What is highly improbable in our world because
of some nomologically contingent feature of ours might be highly probable
in another nomologically possible world that lacks that feature. If the proba-
bility that S occurs is exactly the same in every nomologically possible world,
it is because the feature of beings that fully determines the probability is a
non-contingent feature of beings that resemble us in all causally relevant ways
General methodological considerations 199
and is hence a nomologically necessary feature of ours. To get a reliable sense
for the extent to which moral criteria of legal validity are probable in our
world, we would have to know considerably more about the particular fea-
tures of nomologically possible worlds with moral criteria of legal validity and
the extent to which those features are likely to occur in our world.
The foregoing calls attention to a methodological problem with respect to
grounding claims about the conceptual possibility of moral criteria of validity
in empirical investigation: we can confirm the Incorporation Thesis on the
basis of empirical experience, but we cannot disconfirm it. If, for example,
we can correctly identify a legal system with moral criteria of validity in our
world, that would establish the truth of the Incorporation Thesis. The claim
that there is a legal system with moral criteria of validity trivially implies
the claim that there could be a legal system with moral criteria of validity.
But, again, it would not follow from the claim that there are no actual legal
systems with moral criteria of validity that the Incorporation Thesis is false
because the claim that there is no legal system with moral criteria of validity
in our world does not imply that there could not be a legal system with moral
criteria of validity. For this reason, a search of past and present legal systems
might be inconclusive with respect to determining whether the Incorporation
Thesis is true.
But there is an epistemic difficulty that diminishes our ability even to con-
firm the Incorporation Thesis on the strength of ostensibly favorable empir-
ical observations. It is always possible for theorists to agree on certain acts that
constitute a legal practice in the courts but disagree on how to characterize
that practice. Something that counts as a legal practice must be interpreted to
characterize it, and it is always possible to disagree on the relevant interpretive
claims even when there is no dispute with respect to what the uninterpreted
facts are that constitute the practice. Two theorists can look at the same set
of acts and agree that they constitute a legal practice but disagree with respect
to how to characterize those acts in other respects; an inclusive and exclusive
positivist might look at a set of judicial practices and agree that these practices
are legal but disagree about whether the practices are properly characterized as
defining moral criteria of validity.1
1
Wilfrid Waluchow identifies certain judicial practices in Canada that he interprets as incorpo-
rating moral criteria of validity. Wilfrid Waluchow, Inclusive Legal Positivism (Oxford: Clarendon
Press, 1994). Stephen Perry responds by arguing that those practices are as plausibly interpreted as
implicating only source-based criteria of validity. See, e.g., Stephen Perry, “Two Varieties of Legal
Positivism” (Critical notice of Waluchow’s Inclusive Legal Positivism), Canadian Journal of Law and
Jurisprudence, vol. 9 (1996).
20
2. Prerequisites for a model vindicating

the Incorporation Thesis
The above discussion suggests that one must do two things to show the con-
ceptual possibility of a legal system with moral criteria of validity. First, one
must produce a model of an institutional normative system in a world like
ours that can plausibly be interpreted as having moral criteria of validity that
clearly satisfies every condition plausibly thought to be necessary for the ex-
istence of law. Second, to ensure that the model establishes the Incorporation
Thesis, it should be incompatible with an exclusivist interpretation.
2.1 The model must describe a world that is

nomologically possible
The subjects in a model defining a conceptually possible legal system with
moral criteria of validity should resemble human beings with respect to those
psychological and physical features that explain why we need law to do what
it does in our world. Law is contrived by us to regulate the behavior of beings
like us, and one can sensibly worry that a model that is proposed as a defense
of some conceptual claim about law defines a world that is too far removed
from ours to tell us anything about law as our conceptual practices define
the notion. A model that purports to describe a conceptually possible legal
system that regulates the behavior of all-powerful gods is problematic insofar
as it is not clear that, given our conceptual practices, the concept-term “law”
properly applies to such an institutional normative system.
Consider the society-of-angels model that purports to refute the claim that
every conceptually possible legal system authorizes sanctions for violations of
valid laws. In describing the model, Raz takes care to specify assumptions that
ensure that the angels’ world resembles ours in theoretically salient respects:
[W]‌e can imagine . . . rational beings who may be subject to law, who have, and who
would acknowledge that they have, more than enough reasons to obey the law regard-
less of sanctions. Perhaps even human beings may be transformed to become such
creatures. It is reasonable to suppose that in such a society the legislator would not
bother to enact sanctions since they would be unnecessary and superfluous. If such a
normative system has all the features of a legal system described above, then it would
be recognized as one by all despite its lack of sanctions.2
2
Joseph Raz, Practical Reason and Norms (Princeton, NJ: Princeton University Press, 1990),
159–60. Emphasis added. Hereinafter PRN.
Prerequisites for a vindicating model 201
These assumptions are contrived to ensure that the world of the “angels” is one
that is possible for beings like us. The angels are contrived to resemble human
beings enough to ensure that “human beings [might] be transformed to be-
come such creatures.” Human beings could not be “transformed” into such
creatures except through the standard material and sociological processes that
explain how we come to develop certain physical and psychological features.
Insofar as we are rationally self-interested and vulnerable to certain kinds of
injury for which we need the protection of a legal system, so are the angels.
The one salient respect in which the angels are different from us is that
they are conclusively motivated always to conform their behavior to the law
without any need for the prudential incentives needed in our world. In every
other salient respect, the psychological and physical characteristics of the
angels are the same as ours. Insofar as we could be transformed into beings
like angels, it is nomologically possible, if improbable, for us to evolve and
develop the motivational qualities that distinguish the angels from what we
are at present.
The society-of-angels argument is problematic in one respect that is par-
ticularly salient with respect to the project of vindicating the Incorporation
Thesis. The problem is that it is just not clear whether the world of the angels
is nomologically possible because it is just not clear that self-interested beings
like us who live in a world of material scarcity could ever become like the
angels. If our conceptual practices assume that law is a distinctively human
institution and such a world is not nomologically possible, the model would
not succeed in showing the existence of a legal system without sanctions.
To avoid similar concerns about its relevance, a model that purports to vin-
dicate the Incorporation Thesis should describe a world that is nomologically
possible. If such a model fails to vindicate the Incorporation Thesis, it will not
be because the institutional normative system it describes is inconsistent with
law’s distinctively human character. It will be because the system it describes
fails to satisfy some conceptually necessary condition of law.
2.2 The modeled system must meet Hart’s minimum

conditions for the existence of a legal system
The model must satisfy Hart’s minimum conditions for the existence of a
legal system and hence must satisfy the following conditions: (1) the officials
in the modeled world must practice a rule of recognition that governs offi-
cial behavior with respect to recognizing, applying, and enforcing law; and
(2) non-official behavior in the model must generally conform to the norms
valid under the rule of recognition. Although these two conditions do not
20

exhaust the necessary conditions for law and hence do not define conditions
that are jointly sufficient for the existence of a legal system, nothing that fails
to satisfy these two conditions is properly characterized as a legal system.
2.3 The modeled system must incorporate the minimum

content of natural law
Hart takes the position that it is a “naturally” necessary condition for the
existence of a legal system that it incorporates the minimum content of
natural law:
Reflections on some very obvious generalizations— indeed truisms— concerning
human nature and the world in which men live, show that as long as these hold
good, there are certain rules of conduct which any social organization must contain
if it is to be viable. Such rules do in fact constitute a common element in the law and
conventional morality of all societies which have progressed to the point where these
are distinguished as different forms of social control . . . . Such universally recognized
principles of conduct which have a basis in elementary truths concerning human
beings, their natural environment, and aims, may be considered the minimum con-
tent of Natural Law . . . . We can say, given the setting of natural facts and aims,
which make sanctions both possible and necessary in a municipal system, that this is
a natural necessity.3
There are five features “concerning human nature and the world in which men
live” that, on Hart’s view, necessitate the incorporation of the minimum con-
tent of natural law into any humanly possible system of municipal law4: (1)
we are vulnerable to physical and psychological harm; (2) we are largely equal
with respect to the physical and psychological qualities that enable us to dom-
inate others; (3) our capacity to be motivated by altruistic considerations is
limited given that we are inherently self-interested and experience our wants
and needs with a persuasive urgency; (4) the material resources of the world
are subject to conditions of scarcity; and (5) we are limited with respect to our
intellectual capacities and our strength of will (CL 193–7).
But insofar as these five properties are instantiated in every possible world
that can tell us something about the nature of law as our practices construct
it, they also express what is distinctive about our world that gives rise to our
need for a system of law. If law, as we conceive it, is distinctively human in the
3
H.L.A. Hart, The Concept of Law 3rd Ed. (Oxford: Oxford University Press, 2012), 192–3;
underlined emphasis added. Hereinafter CL. The term “natural necessity,” as Hart uses it, is most
plausibly construed as referring to nomological necessity.
4
This requirement applies only to systems of municipal law because systems of international law
purport to regulate only the official behavior of nation-states.
sense that “law” applies only to institutions that regulate behavior in worlds
with each of these five features, then it is not just a nomologically necessary
condition for the existence of a legal system regulating non-official behavior
that it includes the minimum content of natural law; it is also a conceptually
necessary condition.
To ensure that the model below satisfies every condition plausibly thought
of as conceptually necessary for the existence of law, it will be defined to
ensure that the norms governing non-official behavior incorporate the min-
imum content of natural law. The model will thus be structured to include
the minimum content of natural law to ensure that it succeeds in vindicating
the Incorporation Thesis—regardless of whether law is properly character-
ized, as a conceptual matter, as an institutional normative system that regu-
lates only the behavior of beings like us in worlds like ours.
2.4 The model must satisfy the service conception

of authority
Insofar as it is a truism that every conceptually possible legal system is met-
aphysically capable of instantiating legitimate authority, the model must be
defined so that the institutional normative system is metaphysically capable
of instantiating legitimate authority. Since the most widely accepted analysis
of the nature and justification of authority among conceptual jurisprudents
is the service conception of authority, the model should be specified in such
a manner that the system it defines is metaphysically capable of instantiating
legitimate authority according to the service conception of authority.
This requires that the norms of the constructed system (1) are capable of
expressing the authority’s view about what subjects ought to do according to
right reason; (2) are capable of replacing the judgment of subjects in their prac-
tical deliberations about what they should do according to right reason; and
(3) are framed in a manner that can be legitimate under Normal Justification
Thesis (NJT). Although earlier chapters have raised concerns with the service
conception of authority, the modeled legal system will be constructed so as to
satisfy the service conception to avoid begging any questions about authority.
2.5 The norms of the system must be metaphysically capable

of motivationally and epistemically guiding the behavior
they govern
The modeled legal system should be constructed to ensure that it can per-
form law’s conceptual function of regulating the behavior of subjects through
204

norms metaphysically capable of guiding the acts they govern. As discussed
above, there are two classes of legal norms that govern different types of
behavior: (1) the rule of recognition, which governs official behavior; and
(2) the norms valid under the rule of recognition, which govern non-official
behavior.
It is a conceptual requirement for something to count as a norm of any
kind that governs a subject’s behavior that it is metaphysically capable of both
epistemically and motivationally guiding her behavior. To begin, a piece of
propositional content n can count as a norm governing the behavior of mem-
bers of some class, C, of subjects only if that content expresses what n requires
of them in a manner that members are metaphysically capable of discerning
by consulting n. Any piece of propositional content metaphysically incapable
of epistemically guiding a class of subjects because, as a nomological matter,
they characteristically lack the ability to discern what the norm requires of
them by consulting it cannot, as a conceptual matter, be a norm that gov-
erns the behavior of members of that class. Any norm that governs members
of some class C is thus metaphysically capable of epistemically guiding their
behavior.
There are two ways to see that any norm that governs members of C must
be metaphysically capable of motivationally guiding their behavior. First, the
claim that a norm, n, is metaphysically capable of epistemically guiding the
behavior of members of C implies the claim that n is metaphysically capable
of motivationally guiding their behavior. If n is metaphysically capable of
epistemically guiding the behavior of members of C, then it is nomologi-
cally possible for members of C both to learn from n what it requires and to
satisfy n after learning what n requires. But if members of C can conform to
n partly in virtue of having learned from n what it requires, then they can
conform to n because n is a norm; if learning what n requires can play some
role in a subject’s decision to conform to n, then the subject can conform to n
for any rational subjective reason—including that n is a norm of the system.
Accordingly, any piece of propositional content metaphysically capable of ep-
istemically guiding behavior is also metaphysically capable of motivationally
guiding the same behavior.
Second, a piece of propositional content n can be a norm that governs the
behavior of S only if n is nomologically, and hence metaphysically capable,
of motivationally guiding the behavior of S. A piece of propositional con-
tent n is properly characterized as a norm governing the behavior of S if and
only if (1) it is nomologically possible for S to satisfy n’s requirements and
(2) it is nomologically possible for S to violate n’s requirements. Insofar as it
is nomologically possible for S to satisfy n and it is nomologically possible
for S to violate n, it is nomologically possible for S to obey n. Insofar as it is
nomologically possible for S to obey n, it is nomologically possible for S to
conform her behavior to n because n is a norm that governs her behavior and
hence to be motivationally guided by n; if a subject can obey a norm, she can
obey it for any subjective reason that is minimally rational. Since everything
that is nomologically possible is also metaphysically possible, it is a concep-
tual truth that every norm governing S’s behavior is metaphysically capable of
motivationally guiding S.
This has the following implications for the concept of a legal norm. Since
(1) it is a conceptual truth that every legal norm is a norm and (2) it is a
conceptual truth that something cannot be a norm governing the behavior
of some class of subjects unless it is metaphysically capable of both motiva-
tionally and epistemically guiding members of that class, it follows that it is
a conceptual truth that every legal norm governing the behavior of members
of a class is metaphysically capable of both motivationally and epistemically
guiding them.
2.6 The model must be plausibly interpreted as incorporating

moral criteria of validity
To vindicate the Incorporation Thesis, the modeled system must be plausibly
interpreted as incorporating moral criteria of validity. This means that the
model must be specified so that two conditions are satisfied. First, it must be
clear that judges believe they are practicing a rule of recognition that incor-
porates moral criteria of validity. Second, it must also be clear that judges are
correctly applying those moral standards more often than not; otherwise, the
most that can be confidently asserted about the judges in the model is that
they believe they are applying standards of morality when they make decisions
about whether a particular promulgated norm is legally valid.
2.7 The model should be incompatible with an exclusivist

interpretation
Finally, the model should be incompatible with an exclusivist interpretation.
Although the fact that a model can plausibly be interpreted as defining an
inclusive legal system suggests that the idea of moral criteria of legal validity
is coherent and hence that a legal system with moral criteria of validity is
conceptually possible, it would be preferable to produce a model that can be
interpreted as a legal system only insofar as it is construed as incorporating
moral criteria of validity. This leaves the exclusive positivist with the com-
paratively undesirable option of denying the status of law to an institutional
206

normative system that satisfies every condition plausibly thought to be con-
ceptually necessary for the existence of law.
3. Specification of a model vindicating

the Incorporation Thesis
3.1 The modeled world is nomologically possible

To ensure that the modeled world can tell us something about law on the
assumption that it is a distinctively human institution, the model defines a
world that resembles ours both with respect to our causal laws and with re-
spect to the features that give rise to our need for regulation by law. In partic-
ular, the subjects in this world are self-interested, rational, and vulnerable to
the same kinds of psychological and physical harms to which we are. Further,
there are not enough material resources in this world immediately available
to satisfy all of their wants and needs. Insofar as the model includes those
features of our world that explain our need for law, it defines a nomologically
possible world.
3.2 The subjects are accidentally infallible

The subjects of this nomologically possible world have intellectual abilities
that are limited in the same ways as ours, but they differ from us with respect
to their beliefs about morality in the following respects. First, the subjects
always agree on what morality requires. Second, the beliefs of the subjects
with respect to what morality requires always happen to be correct—whether
morality is properly conceived as objective or conventional in character; while
any subject of this world could have been wrong on any and even every moral
issue, they just happen, as a contingent matter, never to be wrong. Third,
their beliefs with respect to what morality requires always happen to be epi-
stemically justified; the subjects always stumble onto a sound argument that
justifies their beliefs and are hence in cognitive possession of an epistemic
justification for each of their beliefs.
Although these stipulations describe a state of affairs that is improbable for
beings like us, such a state of affairs is nonetheless possible for beings like us.
First, it is obviously nomologically possible for any two agents to agree on
some particular moral issue; we frequently disagree in our world with other
people on some issues but also agree with them on others. Given that it is
nomologically possible for any two agents to agree on one moral issue, it is
nomologically possible for any two agents to agree on every moral issue. But
Specification of a vindicating model 207
if it is nomologically possible for any two agents to agree on every moral issue,
it is nomologically possible for all agents to agree on every moral issue; there
is no reason to think that beings like us could not agree on what morality
requires on every issue.
Second, there is a nomologically possible world in which every person al-
ways happens to be correct about what morality requires. Since it nomologi-
cally possible for me to be correct about what morality requires on one issue,
it is nomologically possible for me to be correct about what morality requires
on every issue. But since what is nomologically possible for one agent in this
regard is nomologically possible for all agents, it is nomologically possible for
all agents to be correct about what morality requires on every issue.
Third, there is a nomologically possible world in which every person knows
what morality requires on every issue. Since it is nomologically possible for
me to be epistemically justified and correct with respect to what I believe mo-
rality requires on one issue, it is nomologically possible for me to be epistemi-
cally justified and correct with respect to what I believe morality requires on
every issue. But since what is nomologically possible for one agent to correctly
believe and be justified in believing about morality is nomologically possible
for all agents, it is nomologically possible for all agents to believe what is
correct about what morality requires on every issue and to be epistemically
justified with respect to every such belief.
The beings in the model are thus conceived to be accidentally infallible
with respect to morality—and not essentially or necessarily infallible—in the
sense that they always know what morality requires. The difference is as fol-
lows. Since they always have true beliefs about what morality requires and
are always in cognitive possession of an argument that epistemically justifies
their beliefs, they always know what morality requires and are hence infallible
with respect to what it requires. But this infallibility is contingent or acci-
dental in the following respect: although the subjects in this world happen to
be correct on every issue, they could have been incorrect with respect to any
issue. It is not, to reiterate, that they are smarter or have more reliable access
to the requirements of morality; they are as intellectually limited as we are
but just happen to stumble on to the right views and the right arguments.
Unlike an essentially omniscient God who cannot, either as a conceptual or
nomological matter, ever be mistaken about anything, the subjects of this
nomologically possible world, like us, could be mistaken on any—or even
every—moral issue concerning what they ought or ought not to do.
The assumption that subjects of this world could be mistaken on any moral
issue is necessary to ensure that they resemble us in all relevant respects and
hence to ensure that their world is nomologically accessible to ours. If not
for the fact that the subjects could be wrong, like us, on any moral issue, the
208

resulting world would be too distant from ours to tell us anything about the dis-
tinctively human nature of law as our practices, properties, and vulnerabilities
define it. The practices that give rise to what we characterize as law are shaped
by both our vulnerabilities and our disabilities, as well as by the content of the
norms we adopt to regulate the behavior of beings with those vulnerabilities
and disabilities.
The beings in this world resemble us in another salient and unfortunate
respect that explains why they need law: they often commit, as we do, so-
cially disruptive acts they believe are morally wrong. Insofar as these persons,
like us, are strongly self-interested and are commonly moved to act by strong
emotions, they frequently act in ways they believe to be morally wrong—or
would if they were to consider the issue in the heat of the moment—to pro-
mote what they urgently feel to be in their best interests. Given that the mate-
rial resources of the world are scarce, there will frequently be violent conflicts
in this world that breach the peace.
How often these conflicts occur in that world relative to how often they
occur in ours is not clear, but it is reasonable to hypothesize that they occur
less frequently. The reason is that, in our world, we commonly wind up doing
something morally wrong to achieve our own interests because we mistakenly
believe the act to be morally permissible; this kind of confusion about mo-
rality does not occur in the world I have described—at least not insofar as the
subjects dispassionately consider the moral issue. If violent conflicts in the
modeled world do not arise because of moral confusion, there are likely to be
fewer violent conflicts in that world than there are in ours.
But it should be clear that these socially disruptive conflicts will none-
theless occur in this world with sufficient frequency that something like law
is needed to keep the peace. Beings like us frequently commit “crimes of
passion” in the heat of a moment in which we are overcome by importunate
desires or needs. Most wrongful assaults, for example, are spontaneous crimes
of passion that result from the frustration of desires experienced as urgent.5
Since subjects in the model are as prone to committing violent acts in the heat
of passion as we are and since the scarcity of material resources (including ro-
mantic and sexual resources) ensure that there will be occasions for conflicts
among subjects, there is clearly a need for something like law in their world.
5
For example, 37.5% of female murder victims are killed by their spouses, and 91% are killed
by people they knew. See “Crime in the U.S.,” Uniform Crime Reporting; available at: https://ucr.fbi.
gov/crime-in-the-u.s/2010/crime-in-the-u.s.-2010/offenses-known-to-law-enforcement/expanded/
expandhomicidemain; and “29 Intriguing Crimes of Passion Statistics,” Brandon Gaille (May 23,
2017); https://brandongaille.com/27-intriguing-crimes-of-passion-statistics/.
Specification of a vindicating model 209
3.3 A model of an institutional normative system with moral
criteria of validity
Given that subjects in this world will often have conflicts that breach the
peace, there is a need for something resembling a system of law to help min-
imize the frequency of such conflicts. Such a system, properly equipped, will
operate to reduce the number of conflicts enough that people will acqui-
esce to, if not actively solicit, this kind of regulation of their behavior. While
people in this world differ from us in that they always have correct beliefs
about what morality requires, they are sufficiently like us in emotional make-
up that their world looks nothing like the Garden of Eden. Some sort of in-
stitutional regulation with authorized sanctions for non-compliance will be
welcomed as a necessary evil.
These sanctions have the same character that they do in our world. They
will sometimes be expressly punitive, as is true of the sanctions intended to
punish criminal wrongdoing in our world, but sometimes the point will
simply be, as is true of the sanctions associated with tort law in our world, to
compensate individuals for injuries wrongly caused by the acts or omissions
of others. Sometimes the agencies assigned to adjudicate these norms will
decline to enforce certain instruments, like contracts or wills, that are not
properly formed as a means of providing what is reasonably contrived to be
a prudential incentive to induce subjects to ensure that their instruments are
properly formed.
As to the rule of recognition, the officials in this institutional normative
system converge on recognizing, applying, and enforcing as rules of the system
all and only mandatory moral norms. Moral norms of goodness that merely
encourage behaviors as good and do not require them as obligatory are not
recognized as valid “legal” norms. For example, the system would include as
law those moral norms that prohibit intentionally killing an innocent person
but not those that encourage helping others insofar as the relevant acts are
morally good but not required. Thus, the officials of the system converge on
treating as “law” only those moral norms that specify acts that one “shall” or
“must” do—and not those that specify acts that one “should” do.
Further, the non-official behavior of people in this nomologically possible
world conforms as much to what is required by valid law as our non-official
behavior does in our world. In most cases, the conforming behavior of people,
as is true of us, is not motivated by a desire to conform to the law or by knowl
edge of what the law requires; they are as emotionally predisposed to refrain
from acts of murder and theft as we are. In some cases, their conforming be-
havior, as might be true of some of us, is motivated by a desire to avoid the
sanctions authorized for violations of valid law. But there are always cases, as
210

is true in our world, in which a subject undeterred by the sanctions commits
a socially disruptive act that violates the law. Even so, subjects conform to the
valid norms enough to permit them to live and work peacefully together in
something properly characterized as a community.
Finally, subjects in this world generally believe that the norms governing
their behavior are morally justified as a necessary but regrettable means of
ensuring that they can live together in comparative peace so as to secure the
benefits of social cooperation. This applies to the subjects of every class of
legal norm: officials generally regard those norms governing official behavior
as morally justified, while all subjects within the jurisdiction of the system re-
gard the valid legal norms governing their behavior as morally justified.
As described, the model defines a nomologically possible world with an
efficacious institutional normative system that validates all and only manda-
tory moral norms. It should be clear that an institutional normative system
that validates all and only mandatory moral norms is conceptually possible;
such a system might not count as a system of law, but the existence of an effi-
cacious institutional normative system that validates all and only mandatory
moral norms is clearly conceptually possible. Insofar as it is nomologically
possible for the self-interested, intellectually limited, vulnerable subjects of
such a world to be accidentally infallible, the existence of an efficacious insti-
tutional normative system validating all and only mandatory moral norms is
also possible for us.
4. Vindicating the Incorporation Thesis: The modeled

system is a legal system
There are two issues with respect to whether the model succeeds in vindi-
cating the Incorporation Thesis. The first issue is whether such a system
is properly characterized, according to the ordinary usage with which this
volume is concerned, as a system of law. To be properly characterized as
law, the system must (1) satisfy Hart’s minimum conditions for the exist-
ence of a legal system; (2) contain norms governing non-official behavior
that incorporate the minimum content of natural law; (3) satisfy the service
conception of authority; and (4) regulate behavior by the recognition, appli-
cation, and enforcement of norms metaphysically capable of motivationally
and epistemically guiding behavior. The second issue is whether the system,
if properly characterized as law, is more plausibly interpreted as having an
inclusive rule of recognition than as having an exclusively source-based rule
of recognition.
The modeled system is a legal system 211
4.1 The system satisfies Hart’s minimum conditions for the
existence of a legal system
The institutional normative system sketched above clearly satisfies Hart’s
minimum conditions for the existence of a legal system. First, there is a con-
ventional rule of recognition that requires judges to apply and enforce all
and only mandatory moral norms against subjects of the system. Insofar as
the officials of the system converge on taking the internal point of view to-
ward the rule of recognition and following it in all relevant circumstances,
the norm is a conventional norm, and one that binds the officials in virtue of
their shared commitments.6
Second, subjects in the system generally conform their behavior to the
valid norms of the system. Although Hart describes this condition as requir-
ing obedience,7 this is too strong. The claim that a subject has obeyed a rule
implies that the motivation for acting as the rule requires has something to
do with the fact that the rule requires it; it might be that the subject is mo-
tivationally guided by the rule or it might be that the subject’s conformity
is motivated by a desire to avoid the authorized sanctions. But there is no
reason to think that, as a conceptual matter, the conforming behavior of a
subject must always be explained by reference to the rule or its authorized
sanctions; the fact that I have never killed a person has nothing to do with
the fact that the law prohibits or punishes it. It is sufficient that subjects
are motivated to conform to the norm in enough of those circumstances
in which they might be tempted to do otherwise that the system succeeds
in keeping the peace enough to ensure that life in the community is pre-
dictably stable. That requirement is also clearly satisfied by the system con-
structed above; it is not necessary to inquire into the motivations of citizens
that would explain their conforming behavior. Accordingly, both of Hart’s
minimum conditions are satisfied by the institutional normative system
described here.
6
This neither presupposes nor endorses the Conventionality Thesis. It assumes only that there
is a nomologically possible world in which the rule of recognition is a conventional rule. But if the
reader finds this characterization objectionable, the rule can be characterized as whatever kind of rule
she believes the rule of recognition is most plausibly thought to be. See Chapter 3 for a discussion
of whether the Conventionality Thesis is a core thesis of positivism.
7
As Hart puts it, “those rules of behavior which are valid according to the system’s ultimate
criteria of validity must be generally obeyed” (CL 113). Emphasis added.
21

4.2 The system contains norms that incorporate
the minimum content of natural law
The system contains what Hart calls the minimum content of natural law,
which Hart defines as “certain rules of conduct which [given certain salient
facts about human nature] any social organization must contain if it is to be
viable” (CL 193). These rules of conduct include prohibitions on theft, vio-
lence, and other acts that would threaten “the minimum purpose of survival
which men have in associating with each other” (CL 193). As these rules of
conduct are all mandatory moral norms, they are included in an institutional
normative system that validates all and only mandatory moral norms. If it is
a conceptually necessary feature of municipal law in worlds nomologically
accessible to ours that it contains the minimum content of natural law, then
the model of an institutional normative system in this world satisfies that
additional condition.
4.3 The system satisfies the service conception of authority

4.3.1 The system has de facto authority
The system has de facto authority over subjects with respect to both official
and non-official behavior. As will be recalled, a de facto authority “either [1]‌
claims to be legitimate or is believed to be so, and [2] is effective in imposing
its will on many over whom it claims authority”.8
The system satisfies both conditions for instantiating de facto authority. To
begin, subjects in the modeled world all believe the system is morally legit-
imate in virtue of being a necessary evil to make the social benefits of com-
munity living possible. Further, the institutional normative system succeeds
in “imposing its will on many” subjects by making certain behaviors non-
optional and deterring those behaviors with the threat of authorized sanc-
tions. Insofar as the system is needed, by hypothesis, to keep the peace and
succeeds in keeping the peace, it does so only by deterring violence that would
otherwise have occurred but for its efficacious regulation. It thus succeeds in
imposing its will on those compliant subjects who would otherwise disobey
without a threat of sanctions.
8
Joseph Raz, Ethics in the Public Domain (Oxford: Oxford University Press, 1994), 211.
Hereinafter EPD.
4.3.2 The norms are capable of expressing the authority’s view about
what right reason requires and of replacing the subjects’ views
about what right reason requires
All the norms of the system are metaphysically capable of expressing the
authority’s view of what right reason requires and of replacing the judgments
of subjects in their practical deliberations about what right reason requires.
The rule of recognition expresses the view, implicit in the convergent practices
of officials and hence accepted by officials as authoritative, that officials must
recognize, apply, and enforce as law against subjects only mandatory moral
norms; it thereby also expresses the authority’s view that individual officials
must refrain from recognizing, applying, and enforcing as law against subjects
any other norms they might believe (or might have believed if their views
were different) right reason requires them to recognize, apply, or enforce.
This view is capable of providing officials with a first-order reason to rec-
ognize, apply, and enforce only mandatory moral norms, as well as a second-
order reason to refrain from acting on any views they might have with respect
to what right reason requires. Insofar as officials take the internal point of
view toward the rule of recognition, it provides both a first-order reason to
conform their official behavior to the requirements of the rule of recognition
and a second-order reason to refrain from acting on their own assessments
of what should be recognized, applied, or enforced as law. Although officials
converge in treating all and only mandatory moral norms as law, individual
officials might occasionally be tempted to depart from the requirements of
the rule of recognition on the strength of self-interested reasons or on the
strength of transiently biased assessments of what right reason requires. The
rule of recognition is thus metaphysically capable of replacing officials’ judg-
ments of what right reason requires with respect to their official behavior.
It is true that a rule of recognition that validates only mandatory moral
norms does not tell a judge which mandatory moral norm is the applicable
one in a dispute, but there is nothing in either the concept of law or the
service conception of authority that entails that an authoritative recognition
rule must distinguish which valid norm is applicable in some dispute from
all other valid norms; a rule of recognition that defines purely source-based
criteria of validity, after all, is no better equipped to do that work for a judge.
There is nothing in any plausible construction of the service conception of
authority that entails a denial of the self-evident fact that judges must use
their judgment in attempting to determine what valid legal norms are rele-
vant in a dispute and how the relevant legal norms should be applied in de-
ciding it. Adjudication is demanding work.
214

The valid legal norms of the system are also metaphysically capable of
expressing an authoritative view about what right reason requires. The valid
legal norms require, as a matter of law, that subjects do what they are mor-
ally required to do, and subjects understand that these norms will be applied
and enforced against them as law. These valid legal norms thus express the
authority’s view that the law requires subjects to comply, or be subject to
legal sanctions, with mandatory moral norms when those norms conflict with
their prudential interests.
The valid legal norms of the system are also metaphysically capable of pro-
viding subjects with both a first-order reason to do what the norms require
and a second-order reason not to act on their own judgments of what right
reason requires with respect to non-official behavior. To begin, given that it is
backed by authorized sanctions, the authority’s view that subjects are legally
obligated to comply with all and only mandatory moral norms is clearly ca-
pable of providing a first-order reason for subjects to comply instead of doing
something else they might antecedently be inclined to do. But if one rejects
the idea that the relevant first-order reasons can make reference to prudential
reasons having to do with a desire to avoid sanctions, the subjects all accept
the authority of officials as legitimate and hence rationally treat their direc-
tives as providing novel first-order reasons to comply.
Further, the valid legal norms are metaphysically capable of providing sub-
jects with a second-order reason not to act on their own judgments of what
right reason requires with respect to non-official behavior. It is clearly meta-
physically possible for subjects to rationally treat the fact that law requires a
non-official behavior as a second-order reason not to act on their own judg-
ments; since they regard the directives as morally legitimate, those directives
are as capable of providing the relevant second-order reasons as the under-
lying mandatory moral norms are. Thus, the valid legal norms are capable of
replacing subjects’ own judgments of what right reason requires with respect
to non-official behavior.
4.3.3 Is it metaphysically possible for subjects to better comply

with what right reason requires by following the authority’s
view of what right reason requires than by following their
own views of what right reason requires?
At first blush, the model seems to run afoul of NJT. The problem is as follows.
According to NJT, authority is morally justified only insofar as subjects are
likely to better comply with respect to right reason by following the authority’s
view of what it requires than by following their own views. If moral norms
take into account prudential interests and dictate what subjects should do
all things considered according to right reason, then it is not metaphysically
possible for subjects to better comply with right reason by following the
authority’s view than by following their own views because they will always
arrive at exactly the same result with respect to what right reason requires since
the moral standards reflect the balance of all the applicable reasons, including
the prudential reasons.
One possible response, of course, is to reject the model on the ground that
it does not satisfy the service conception, but this is problematic. Subjects in
this world are morally infallible but not morally impeccable; they are no less
prone to acting out in socially disruptive ways in virtue of being self-interested
than we are and hence are as motivated as we are to act in ways that would
breach the peace. It is precisely for that reason that an authoritative system of
norms is needed to provide an additional first-order reason to comply with
the requirements of right reason. But insofar as the system sometimes deters
such behavior, subjects are more likely to better comply with right reason by
complying with the law than by acting on the basis of their own transient
impulses, inclinations, and subjectively corrupted assessments of what right
reason requires.
But this seems to call NJT into question if construed as stating a neces-
sary condition for the legitimacy of a legal system. As described, the system
appears to be morally legitimate: the norms (1) are morally just and justly
enforced; (2) succeed in keeping the peace; (3) distribute the resources of the
society in the manner required by morality; (4) enjoy the consent of subjects;
and (5) are such that subjects will better comply with right reason by fol-
lowing them than by following their inclinations, impulses, and corrupted
judgments. Although this system does not satisfy NJT, there could not be
a system that is morally preferable in the modeled world. If any legal system
authorizing sanctions for violations of law is legitimate in this nomologically
possible world, this system is.
Nevertheless, NJT is not properly construed as articulating a necessary
condition for the existence of a legitimate authority:
The normal justification thesis: The normal and primary way to establish that a person
should be acknowledged to have authority over another person involves showing that
the alleged subject is likely better to comply with reasons which apply to him (other
than the alleged authoritative directives) if he accepts the directives of the alleged au-
thority as authoritatively binding, and tries to follow them, than if he tries to follow
the reasons which apply to him directly (EPD 214; emphasis added).
Raz here scrupulously avoids characterizing NJT as providing either neces-
sary or sufficient conditions for legitimacy; NJT expresses the “primary” and
“normal” way to show that authority is justified. Insofar as he refrains from
216

using such language, it is presumably because he is aware of the problems
that arise if NJT is construed as defining a necessary or sufficient condition
for legitimacy.
The most plausible construction of NJT, for reasons of both theoretical
substance and interpretive accuracy, is to construe it as compatible with the
legitimacy of the system constructed in the model. Thus construed, although
the “normal” condition is not satisfied, the system is nonetheless legitimate
in virtue of the distinctive properties of the system and its subjects. Since the
most salient features of the system are highly improbable in our world, the
existence of a legitimate system that does not satisfy NJT does not impugn its
status as the “normal” way to justify authority. Insofar as NJT gives the cor-
rect answer in the vast majority of cases likely to be encountered in our world,
it would remain, if correct, the “normal” strategy for justifying authority.9 On
this construction, the modeled system does not satisfy NJT, but it does not
run afoul of it and is hence compatible with the most plausible construction
of the service conception of authority.
4.4 The modeled system is capable of performing law’s

conceptual function of regulation through
norm-governance and norm guidance
To show that the modeled system is metaphysically capable of performing
law’s regulatory function, it must be shown that (1) the valid norms can mo-
tivationally and epistemically guide non-official behavior and (2) the rule of
recognition can motivationally and epistemically guide official behavior. If
both these claims are shown, it follows that the modeled system can do eve-
rything it needs to do to perform law’s regulatory function.
4.4.1 The rule of recognition is metaphysically capable of motivationally

and epistemically guiding official behavior
A rule of recognition that validates all and only mandatory moral norms is
metaphysically capable of motivationally guiding official behavior. To begin, a
rule of recognition that validates all and only mandatory moral norms defines
a legal obligation on the part of a judge to apply the relevant mandatory
9
Even so, the system defined above suggests that NJT would be more plausible if reconfigured
to allow for the possibility of a legitimate authority in a world where the authority is not less likely
than the subjects to correctly discern the requirements of right reason. One potential problem with
justifying democratic systems under NJT is that, if the views of officials are perfectly representative
of those of the subjects, the officials are no more and no less likely than subjects to correctly discern
the requirements of right reason.
moral norms as law in adjudicating legal disputes. It is obviously possible for
a judge to violate this obligation by applying some norm that does not have
the status of law in virtue of moral merit, but it is as obviously possible for
a judge to satisfy this obligation by applying the relevant norm that has the
status of law in virtue of moral merit.
But insofar as a judge can knowingly conform her official behavior to a rule
of recognition, R, that requires her to apply all and only mandatory moral
norms, she can knowingly conform her official behavior to R for any reason
that rationally strikes her as persuasive. She can conform her official behavior
to R because she thinks she must in order to retain her seat on the bench, or
she can conform her official behavior to R because R is the rule of recognition.
It should be clear—without recourse to any of the features that distinguish
the modeled world from ours—that R is metaphysically capable of motiva-
tionally guiding judicial behavior.
R is also capable of epistemically guiding judicial behavior. In a world in
which subjects are accidentally infallible about their moral obligations, the
judge can learn which norms she must apply in deciding a dispute by con-
sulting R; the judge who is confused about her legal obligations can learn
from R that it requires her to apply and enforce a subject’s moral obligations
in disputes concerning non-official behavior. Since the judge can always cor-
rectly identify the content of the applicable legal norm, she can learn what
R requires of her and satisfy her legal obligations under R by consulting R.
Insofar as the judge would apply different norms if R required it, R is capable
of making a practical difference in the deliberations of a judge.
It is true that the judge must decide for herself how a legal norm valid in
virtue of moral merit should be applied to a dispute regarding someone’s non-
official behavior, but that is also true of a legal norm valid in virtue of having
an authoritative source. The judge will have to do exactly the same thing to
decide a case that requires applying a valid norm that prohibits recklessly per-
forming acts highly likely to cause life-threatening injuries to other persons
regardless of whether it is valid in virtue of moral merit or valid in virtue of
source. It does not matter what kind of criteria of validity a rule of recogni-
tion defines; the judge will still have to figure out for herself how any relevant
valid norm applies in any given case. It is no part of what something must do
to be a rule of recognition that it defines a decision procedure that produces
right answers to every dispute under the law.
There is another way to see that R is capable of epistemically guiding ju-
dicial behavior that does not require recourse to the particulars of the model
described above. If we are morally accountable for our behavior, then we can
be epistemically guided by mandatory moral norms governing non-official be-
havior. Insofar as this implies we can determine which norms are mandatory
218

moral norms that govern non-official behavior, we can distinguish mandatory
moral norms that apply to the relevant behavior from all other norms. But
this implies that a judge can determine from a rule of recognition validating
all and only mandatory moral norms which norm she is legally obligated to
apply in deciding a dispute and hence can conform her judicial behavior to
the rule of recognition by applying the relevant valid norm to decide the dis-
pute. Thus, judges in our world can be epistemically guided by a rule of rec-
ognition validating all and only mandatory moral norms, on the assumption
that we are morally accountable for our behavior.
A rule of recognition validating all and only mandatory moral norms is
thus metaphysically capable of both motivationally and epistemically guiding
judicial behavior and hence of performing its guidance function with respect
to official behavior. This vindicates the Incorporation Thesis with respect to
the guidance function of the rule of recognition.
4.4.2 The norms valid under the rule of recognition are metaphysically

capable of motivationally and epistemically guiding
non-official behavior
The norms valid under the modeled rule of recognition are metaphysically
capable of doing everything they must do to enable law to perform its reg-
ulatory function. Since subjects in this world are accidentally infallible with
respect to what mandatory moral norms require of them, they are also acci-
dentally infallible with respect to which norms are mandatory moral norms.
If they know that the judges are recognizing, applying, and enforcing all and
only mandatory moral norms as law, then they have enough information to
determine which norms are legally valid (i.e. those that reproduce the content
of a mandatory moral norm); what those valid norms require of them with
respect to non-official behavior (i.e. exactly the same thing that the corre-
sponding mandatory moral norm requires of them); and can conform their
behavior to that mandatory moral norm. Thus, every legal norm that is valid
in virtue of reproducing the content of a mandatory moral norm is metaphys-
ically capable of epistemically guiding the non-official behavior of subjects in
this nomologically possible world.
Likewise, every legal norm valid in virtue of reproducing the content of a
mandatory moral norm is metaphysically capable of motivationally guiding
the non-official behavior of subjects in this world. Insofar as it is metaphysi-
cally possible for subjects to knowingly conform their non-official behavior to
a mandatory moral norm, it is metaphysically possible for subjects to know-
ingly conform their non-official behavior to that norm for the reason that
it is a valid legal norm of the system. It follows that every norm governing
non-official behavior valid in virtue of reproducing the content of a man-
datory moral norm is metaphysically capable of motivationally guiding the
non-official behavior of the subjects of the model.
As was true in the last subsection, there is another way to see that the valid
norms of this system can motivationally and epistemically guide non-official
behavior without recourse to the particulars of the model. First, if we are
morally accountable for our behavior under mandatory moral norms, then
it must be metaphysically possible for us to knowingly satisfy those moral
norms. But if it is metaphysically possible for us to knowingly satisfy a man-
datory moral norm, then it must also be metaphysically possible for us to
satisfy that norm because it is a moral norm. Thus, if we are morally account-
able for our behavior, mandatory moral norms are metaphysically capable of
motivationally guiding our behavior.
But if it is metaphysically possible for a mandatory moral norm to motiva-
tionally guide behavior, then it is also metaphysically possible for a legal norm
valid in virtue of reproducing its content to motivationally guide behavior. If
it is possible for us to knowingly satisfy a mandatory moral norm because it is
a mandatory moral norm, then it must also be possible for us to satisfy a legal
norm valid in virtue of reproducing the content of that mandatory moral
norm because it is a legally valid norm; simply giving the norm a new name
cannot render us incapable of knowingly satisfying a norm that we can oth-
erwise knowingly satisfy because it belongs to this or that normative system.
It is, then, every bit as possible for us to be motivationally guided by a legal
norm valid in virtue of reproducing the content of a mandatory moral norm
as it is to be motivationally guided by the mandatory moral norm itself.
Similarly, it is metaphysically possible for us to be epistemically guided
by legal norms valid in virtue of reproducing the content of a mandatory
moral norm, assuming that we are morally accountable for our behavior. We
are morally accountable for our behavior only insofar as it is nomologically
possible for us to determine what our moral obligations are and to know-
ingly satisfy those obligations. This requires that we be nomologically capable
of (1) distinguishing mandatory moral norms from other norms; (2) deter-
mining what a particular mandatory moral norm requires with respect to
non-official behavior; and (3) knowingly satisfying mandatory moral norms
governing non-official behavior. But if we have the capacities described in
(1), (2), and (3) and we know that judges are applying and enforcing as law
all and only mandatory moral norms, then we are nomologically capable of
satisfying a legal norm valid in virtue of reproducing the content of a manda-
tory moral norm and we are nomologically capable of determining what is re-
quired of non-official behavior by a legal norm valid in virtue of reproducing
the content of a mandatory moral norm by consulting the valid legal norm.
20

This suffices to show that every legal norm governing non-official behavior
valid in virtue of moral merit is metaphysically capable of both motivation-
ally and epistemically guiding the governed behavior and hence suffices to
vindicate the Incorporation Thesis with respect to the guidance function of
legally valid norms.
4.5 The system precludes an exclusivist interpretation

The system in this world was constructed to preclude its being plausibly inter-
preted as an exclusivist legal system. While the rule of recognition validating
all and only mandatory moral norms is a legal norm in virtue of its being
practiced by officials, none of the system’s norms can be considered valid in
virtue of some kind of official promulgation.
One might argue that the institutional normative system described in
the model is a legal system but none of the norms of the system are valid,
despite appearances to the contrary, in virtue of reproducing the content
of a mandatory moral norm. On this line of reasoning, judicial acts apply-
ing and enforcing these norms necessarily involve acts of source-based
promulgation; for this reason, every valid legal norm is valid in virtue of
being applied and enforced by judges in the course of deciding some legal
dispute.
This line of reasoning is problematic insofar as it assumes that judges nec-
essarily have occasion to apply and enforce these norms because citizens nec-
essarily violate them at least some of the time. It is true that citizens might
frequently violate these norms but there is no reason to assume that they must
do so: if an institutional normative system in a Razian society of angels is a
system of law, then it is an example of a conceptually possible legal system in
which subjects never violate the law; if, further, it is true that we could evolve
to become like angels, then it is an example of a nomologically possible legal
system in which subjects never violate the law. As long as subjects are aware
that there is a legal system that validates norms capable of providing some-
thing that subjects can rationally treat as a reason to comply, that is enough
to ensure that the appropriate regulatory relationship exists between the legal
system and its putative subjects.
Insofar as the conceptual function of a legal system is to regulate behav
ior so as to permit people to live and work together in peace, the case in
which subjects never violate the law would be a case in which the system most
completely realizes its conceptual function. The idea that it is a conceptually
necessary condition for the existence of a legal system that subjects some-
times violate the law is simply false; there is no plausible reason to think that
an institutional normative system that completely realizes law’s conceptual
Why all the fuss? 221
function of keeping the peace is conceptually disqualified from being a legal
system.10
An exclusive positivist could, I suppose, deny that the institutional norma-
tive system in the model constitutes a legal system, but this is implausible.
The system (1) satisfies the minimum conditions for the existence of a legal
system; (2) incorporates the minimum content of natural law; (3) satisfies
the service conception of authority; (4) is capable of performing law’s regula-
tory function by providing norms capable of motivationally and epistemically
guiding the behavior they govern; (5) authorizes sanctions as a response to
violations; and (6) contains only morally just norms and is morally legiti-
mate. Given that it satisfies every plausible candidate for a conceptually nec-
essary feature of law on both positivist and anti-positivist theories, there is no
plausible reason to think it is not a legal system.
5. Why all the fuss? The probability of inclusive legal

systems in our world
One common motivation for accepting the Incorporation Thesis is the
thought that there are many existing legal systems that incorporate moral
criteria of validity. This seems to have been Hart’s motivation for accepting
the Incorporation Thesis:
[Dworkin] treats my doctrine of the rule of recognition as requiring that the criteria
which it provides for the identification of law must consist only of historical facts
and so as an example of ‘plain-fact positivism’. But though my main examples of the
criteria provided by the rule of recognition are matters of what Dworkin has called
‘pedigree’, concerned only with the manner in which laws are adopted or created by
legal institutions and not with their content, I expressly state both in this book and
in my earlier article on ‘Positivism and the Separation of Law and Morals’ that in
some systems of law, as in the United States, the ultimate criteria of legal validity might
explicitly incorporate besides pedigree, principles of justice or substantive moral values,
and these may form the content of legal constitutional restraints (CL 250; emphasis
added).
10
Nor is there any reason to think that it is nomologically impossible for a legal system to produce
perfect compliance in worlds like ours regulating the behavior of beings like us. A system of law with
sufficiently stringent sanctions and a highly accurate mechanism for detecting and attributing viola-
tions of law could, in principle, produce perfect compliance. There is comparatively little crime, for
example, in societies like North Korea because of the harsh sanctions and the pervasively intrusive
character of the available detection mechanisms. A totalitarian legal system is no less a legal system
in virtue of being totalitarian.
2

Justifiably characterizing an existing legal system as inclusive is more compli-
cated than might initially appear. As discussed above, there are prohibitive
epistemic difficulties with attempting to vindicate the Incorporation Thesis
on the basis of empirical observations—which, in essence, is what is done by
someone who accepts it on the belief that there are existing legal systems with
moral criteria of validity. The problem is that any existing legal system that can
be interpreted as inclusive can be as plausibly interpreted as exclusive. A va-
lidity decision grounded in the application of some putatively moral norm
need not be characterized as involving the application of some authoritative
moral criterion of legal validity; it is as plausibly characterized as the result of
the judicial exercise of a quasi-legislative discretion to create new law in hard
cases. This is why vindicating the Incorporation Thesis requires construct-
ing a world with properties that are possible but vanishingly improbable in
our world.
Further, the highly idealized character of the modeled world, though nom-
ologically possible, might confirm the Incorporation Thesis, but it tells us
nothing about the character of existing legal systems. There are two features
of these highly artificial vindicating models that account for their irrelevance
in explaining existing legal practices. The first is that judges in those worlds
always agree on which norms are valid in virtue of moral merit. The problem
is that it is quite unlikely that judges in any legal system as complex as the
modern municipal legal systems that motivate conceptual theorizing about
law will converge in agreeing on the moral merits of any putative legal norm.
Judges in legal systems like that of the U.S. frequently disagree on how
to interpret putatively moral language in constitutions in making validity
decisions. Some judges adopt an originalist approach to making validity deci-
sions under the constitution while others adopt a conflicting approach that
allows them to interpret that language in a manner that conforms to some
set of conventional or critical moral standards; and this disagreement extends
to every provision of those constitutions plausibly interpreted as defining a
moral constraint on the content of valid law. If it is a necessary condition for
the existence of an inclusive legal system that judges converge with respect to
their views about which norms are valid in virtue of moral merits, there are no
inclusive legal systems in our world—nor are there likely to be.
The second is that these vindicating models must be constructed so that
we are epistemically justified in believing that judges are always correct about
what morality requires with respect to non-official behavior. The problem
here is not just that we are epistemically limited with respect to identifying
the substantive requirements of whatever moral norms legitimately govern
our behavior. It is also that we are epistemically limited with respect to de-
termining whether there are any moral norms that legitimately govern our
Why all the fuss? 223
behavior and, if so, whether those norms are objective or intersubjective in
character. If there are correct answers to these questions that are utterly inde-
pendent of our social practices, there is simply no epistemically reliable way
to tell what they are.
It might be true that we are epistemically justified in believing that our
practices presuppose that morality objectively governs our behavior, but it is
not clear how we could be epistemically justified in believing that this presup-
position is true. We might be justified with respect to our beliefs about how
the world appears to us, but it is hard to see, as argued in Chapter 2, how we
could be justified with respect to our beliefs about how the world really is in-
dependently of how it appears mediated through the conceptual framework
we impose on it. All we have to go on in figuring out how to get around in
our world, with the possible exception of the laws of logic and some trivial
arithmetical claims, is how the world appears to us through the filter of the
relevant conceptual practices. Meta-ethical nihilism might, for all we know,
be objectively true.11
This is why the model (1) assumes that morality legitimately governs our
behavior and (2) is constructed so that subjects with our limitations always
happen to be correct about what morality requires—regardless of whether it
is objective or intersubjective. The former was assumed because whether the
Incorporation Thesis is true depends on whether there are moral norms that
legitimately govern our behavior; if not, the Incorporation Thesis is trivially
false. The latter was assumed because we are not in an epistemic position to
ground a model of an inclusive legal system in practices that conform to some
specified set of norms that we know correctly report the requirements of mo-
rality. Even so, if the model is coherent, it establishes the ontological claim
that inclusive legal systems can exist in worlds like ours.
But the ontological claim that there can be inclusive legal systems in worlds
like ours implies nothing with respect to whether we are in an epistemic po-
sition to accurately determine that some existing system succeeds in incor-
porating moral criteria of validity. It is clear that we can verify whether a
system affords courts with authority to bind the other officials with morally
mistaken decisions and hence that we are in an epistemic position to reliably
rule out that some existing system is inclusive. What we cannot do, however,
is infer that a legal system succeeds in incorporating moral criteria of validity
from the fact that the officials always agree on the relevant moral issues. If the
relevant moral views are incorrect enough of the time—and what counts as
11
Meta-ethical nihilism, as I use the term, asserts that there are no moral norms that, as an
objective matter, legitimately govern our behavior. As the point is sometimes put, there is no right
or wrong.
24

enough of the time is not at all clear—then the system is not properly charac-
terized as inclusive. Insofar as we are not in an epistemic position to reliably
decide these matters, we are not in an epistemic position to reliably charac-
terize some existing legal system as inclusive.
The fact that neither the Incorporation Thesis nor the supporting argu-
ments offered in this volume entails anything that would explain legal prac-
tice in any existing legal system might seem to validate the views of those
who believe conceptual jurisprudence should not be done. The intuition
motivating that view is that it is wrong to devote limited resources to solving
problems when the solutions have no practical applications with respect to
existing legal practices—especially when those resources can be as produc-
tively devoted to solving problems that tell us something about what our legal
practices should be.
But there is nothing surprising about any of this. It was conceded in
Chapter 1 that conceptual claims about law lack both normative implications
with respect to what our legal practices should be and practical applications
that would enable us to explain existing legal systems. Conceptual theories of
law can be quite pretty, but they are otherwise useless.
The uselessness of such theorizing is what is behind the complaint, dis-
cussed in Chapter 1, that conceptual jurisprudence should not be done; on
this oddly anti-philosophical view, the only justification for pursuing a line
of theoretical inquiry is that it is likely to produce something of instrumental
value. I argued in Chapter 1 that a line of theoretical inquiry need not be
justified by its instrumental value; it is enough to justify pursuing any line of
inquiry that it adds to our store of knowledge.
The fact that an inquiry conduces to knowledge does not make it inter-
esting; there are many eminently uninteresting and forgettable truths about
the world. Perhaps the truths of conceptual jurisprudence are among them.
Either way, I have no illusions about what I have done here: assuming I have
somehow gotten things largely right, I am certain that I have said nothing in
this volume that should, in some objective sense, elicit an interest in the topic
where one was antecedently lacking.
That being conceded, I hope that those who have taken the time to sludge
through this volume have found something to reward their efforts. If I did
not think that there was something valuable in this endeavor, I would not
have devoted so many hours to it—and I certainly would not have spent
more time on such issues than was absolutely necessary to reap whatever pru-
dential benefits I thought might make it worthwhile. I developed my taste
for this topic from the beauty I saw in the work of other conceptual jurispru-
dents; I hope that others might see a little something here that approximates
that beauty.
Index
Aquinas, Thomas 10–11, 12, 47, 55 Hart, Herbert Lionel Adolphus 2, 9, 15–16,

Augustine of Hippo 10 24, 26, 42, 67–68, 73, 105, 122, 145–46,
Austin, John 5, 12, 55, 67–68, 163 163, 176, 183, 184–85, 201–2, 210,
211–12,  221
Bell, John 108
Bentham, Jeremy 5, 12, 55 Jackson, Frank 31–32,  57
Bix, Brian H. 5,  12–13
Blackstone, William 5, 10–11, 12, 47, 55 Kant, Immanuel 33, 57
Brennan, William 160–61
Brink, David 45 Mackie, John Leslie 34–35
McTaggart, John McTaggart Ellis 32
Caravaggio, Michelangelo Merisi da 37, 38, 39
Churchland, Paul 52 Posner, Richard 18, 19, 22–24
Cicero, Marcus Tullius 10
Rawls, John 108
Descartes, René 185 Raz, Joseph 16–17, 26, 42, 89, 96, 98,
Duchamp, Marcel 39 104, 105, 106, 107, 108, 115–17, 118,
Dworkin, Ronald 2, 14–15, 24, 26, 29, 41, 119–20, 122, 127–28, 129–30, 131,
48, 49, 50–52, 54–55, 56, 58–59, 145–46, 134, 137, 140, 146–47, 153, 154, 155,
184–85,  221 158, 159, 160–62, 163, 194, 200, 212,
215–16,  220
Ehrenberg, Kenneth 45 Robles, Ashley 45
Euclid of Alexandria 158–59
Scalia, Antonin 160, 161
Fermat, Pierre de 24–26 Shapiro, Scott 17, 75–76, 175, 176, 177,
Finnis, John 55 186–87,  194
Fuller, Lon 107–8,  163 Shaw, Todd 45
Geach, Peter 32 Waluchow, Wilfrid 199

Gödel, Kurt 25 Wiles, Andrew 24, 25, 26
26
28

Morality and The Nature of Law (Kenneth Einar Himma)

Uploaded by

Copyright:

Available Formats

Morality and The Nature of Law (Kenneth Einar Himma)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Morality and The Nature of Law (Kenneth Einar Himma)

Uploaded by

Copyright:

Available Formats

M O R A L I T Y A N D T H E N AT U R E O F L AW

Introduction: What Do You Mean by “Law,” Anyway? 1

. Relationships Between Law and Morality

2. Rethinking the Traditional Interpretation of Anti-​Positivist

. Legal Positivism and the Possibility of Moral Criteria of Validity

. Law’s Claim of Legitimate Authority

. Authority, Moral Criteria of Validity, and Conceptual Confusion

7. To Whom the Rules Apply: Norm Guidance and the

. The Conceptual Possibility of Moral Criteria of Legal Validity

The notion of law that is typically the focus of conceptual jurisprudence

It is not implausible to think that the project of general jurisprudence was

6 Relationships Between Law and Morality

1. Three types of inquiry about law

8 Relationships Between Law and Morality

2. Natural law theories

10 Relationships Between Law and Morality

12 Relationships Between Law and Morality

14 Relationships Between Law and Morality

3. Inclusive legal positivism

16 Relationships Between Law and Morality

18 Relationships Between Law and Morality

4.1 Does a conceptual theory of law have any practical

20 Relationships Between Law and Morality

22 Relationships Between Law and Morality

4.2 Should conceptual jurisprudence not be done?

24 Relationships Between Law and Morality

26 Relationships Between Law and Morality

This chapter challenges the traditional interpretation of classical natural law

1. Modest and immodest approaches

30 Reinterpreting Anti-Positivist Theories

A property p is essential to C if and only if it is not possible for a thing to be C without

32 Reinterpreting Anti-Positivist Theories

34 Reinterpreting Anti-Positivist Theories

2. Two concepts of law

36 Reinterpreting Anti-Positivist Theories

38 Reinterpreting Anti-Positivist Theories

40 Reinterpreting Anti-Positivist Theories

3. Four possible interpretations of a conceptual

4. Legal positivism as assuming MCA to explain

42 Reinterpreting Anti-Positivist Theories

5. The traditional interpretation of natural law theory

44 Reinterpreting Anti-Positivist Theories

46 Reinterpreting Anti-Positivist Theories

48 Reinterpreting Anti-Positivist Theories

6. Dworkin’s interpretivism construed as a rival

50 Reinterpreting Anti-Positivist Theories

52 Reinterpreting Anti-Positivist Theories

7. Construing classical natural law and interpretivism

54 Reinterpreting Anti-Positivist Theories

8. What do Finnis and Dworkin say?

56 Reinterpreting Anti-Positivist Theories

9. Can ICA ground a viable conceptual methodology?

58 Reinterpreting Anti-Positivist Theories

1. The Differentiation Thesis

62 Possibility of Moral Criteria of Validity

64 Possibility of Moral Criteria of Validity

The Differentiation Thesis is a metaphysical thesis and not an epistemological

66 Possibility of Moral Criteria of Validity

2. Conceptual foundations of legal positivism

M O R A L I T Y A N D T H E N AT U R E   O F   L AW

2. Rethinking the Traditional Interpretation of Anti-Positivist

1. Three types of inquiry about law

4.1 Does a conceptual theory of law have any practical

4.2 Should conceptual jurisprudence not be done?

2. Two concepts of law

8. What do Finnis and Dworkin say?

1. The Differentiation Thesis

2.4 The relationship between the criteria of validity and

6.3 The Authority Thesis, NJT, and inclusive positivism

6.4 The directives of practical authority as expressing its view

4.1 The use of the language of rights and duties

4.5 Officials’ claim that subjects have an obligation to obey

4.6 The claims taken together