RULES AND SOCIAL FACTS
JULES L. COLEMAN*
Ronald Dworkin has identified H.L.A. Hart with the view that
law consists in rules. 1 That attribution is partially understandable, if ultimately unwarranted. Hart does claim that law consists
in rules, but he also explicitly acknowledges that customary
norms can constitute part of a community's law though they are
not rules. Even if Dworkin overstates the point, it is true that
rules are essential both to Hart's jurisprudence and to his theory of adjudication. Why is it that law, for Hart, is primarily a
matter of rules?
I.
HART AND
AUSTIN
In The Concept of Law,2 Hart develops his own position by
contrasting it with Austin's. My view is that Hart is constrained
by his methodology, which is to develop his view in the light of
the particular shortcomings he identifies in Austin's jurisprudence. Nowhere is this clearer than in his development and articulation of the view that law consists in social rules. Before
turning to the way in which Hart is drawn to identify law with
social rules, it is useful to look at another example of the way in
which Hart develops his view as a response to, and ultimately as
an extension of, Austin's.
Hart correctly argues that Austin's view of law as the order of
a sovereign, backed by a threat of sanction, can explain neither
(1) the fact that the commands of dead and departed sovereigns continue to be law, nor (2) the fact that the first command
of a nascent sovereign is law in spite of the fact that it has not
yet secured the requisite habit of obedience from those it
orders.
The flaw in Austin's logic lies in his narrow conception of the
nature of law as consisting in liberty-limiting or constraining
* John A. Garver Professor of Jurisprudence and Philosophy, Yale Law School.
B.A., 1968, Brooklyn College; M.S.L., 1976, Yale University; Ph.D., 1972, Rockefeller
University.
1. See R. DWORKIN, LAw's EMPIRE 34-35 (1986).
2. H.L.A. HART, THE CONCEPT OF LAW (1961).
3. See id at 60-76.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 703 1991
704
HarvardJournal of Law & Public Policy
[Vol. 14
principles or norms.4 By introducing into the concept of law
the idea that some legal norms enhance the scope of liberty by
empowering individuals, the problem Hart poses for Austin's
account is solved. The sovereign at any given time is someone
who occupies a legal office. Occupants of that office have authority of a certain sort; their commands are law. Thus, they are
empowered with legal authority by the rules that define their
office.
Were all positive law liberty-limiting, then the concept of an
office of this sort would be impossible. Such offices are defined
by legal rules, but these rules do not constrain. Rather, these
rules empower. Were there no such rules, there could be no
offices. Were there no office of the sovereign, there would be
only particular sovereigns whose identities as such would depend on having secured and maintained the habit of obedience.
Thus, the problems of persistence and continuity to which Hart
draws our attention would remain. They can be resolved only
by notions of sovereignty and the office from which the sovereign governs. Such offices, in turn, are constituted by rules that
empower, not by norms that constrain.
Because offices require power-conferring or enabling norms,
it comes as no surprise that in identifying the core of his position, Hart claims that law consists in norms of two sorts: those
that constrain, consistent with Austin's analysis, and those that
enable.5
Hart's introduction of rules into the concept of law has a similar genesis. He begins by noting a problem in Austin's account
and demonstrates that the source of the problem is Austin's
identification of law with particular commands rather than with
6
norms of sufficient generality and normativity (that is, rules).
Jurisprudential theories have traditionally attempted to answer
two distinct but related questions. First, what is law, and second, why is it binding? The first of these questions is analytic,
hence the concept of analytic jurisprudence. The second question is normative. Many scholars have taken the first question
to be an invitation to provide an account of the meaning or the
proper or ordinary use of the term "law." These are the theo4. See J. AuSTIN, THE PROVINCE OF JURISPRUDENCE DETERMINED (H.L.A. Hart ed.
1954).
5. See H.L.A. HART, supra note 2, at 77-79.
6. See id at 78-96.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 704 1991
No. 3]
Rules and Social Facts
705
ries-semantic theories-that Dworkin believes are particularly
subject to what he calls the "semantic sting. ' 7
The more interesting question concerns the law's normativity or authority. If law is the command of a sovereign, how is it
that law can be binding or thought to create obligations in
those to whom it is directed? Austin's answer to this question
relies on the fact that as a result of the sovereign's ability to
offer credible threats, individuals develop a habit of obedience
directed toward the sovereign and his commands. Hart objects
that the picture Austin paints is one more suitable to the relationship between gunman and hostage, where we are more
likely to speak of those to whom commands are directed as being obliged to behave in certain ways, rather than as their being
obligated so to act. As such, Austin does not account for the
law's ability to impose obligations upon its citizens. The shorthand way of characterizing this objection is to say that "habits
of obedience express what people do as a rule," and, as a consequence, are mere descriptions lacking the normative dimension
necessary to explain the normative force of law.
Having identified the problem, it remains for Hart to provide
the solution. In this case, the solution is provided by the concept of rules. Law is not formed through the marriage of commands with habits of obedience. Instead, Hart argues that law
consists in rules, and, in particular, law consists in social rules.
What are social rules? Social rules have two dimensions. In one
respect, they are descriptions and characterizations of what
people do as a rule. As such, they correspond closely to Austin's
"habits of obedience." Habits of obedience, however, lack a
normative dimension. By conceiving of these habits as motivated by the credible threats of the sovereign, Austin fails to
address this second, normative dimension.
Social rules can have normative force in that they have a prescriptive or reason-giving dimension. Whether they provide
reasons for action depends on citizens accepting them from an
internal point of view. Social rules that are constituted by convergent social behavior accepted from an internal point of view
7. See R. DWORKIN, supra note 1, at 45-46. My own view is that there is no such thing
as a semantic sting. Whatever objection Dworkin intends to put forward under this
label, it is considerably less troublesome than Dworkin thinks it is. Nor do I believe that
any of the people Dworkin seeks to critidcize, particularly positivists like Raz or myself,
see jurisprudence as an effort to provide definitions in the form of necessary and sufficient conditions.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 705 1991
706
HarvardJournal of Law & Public Policy
[Vol. 14
provide reasons for compliance with their demands and
grounds for criticizing the noncompliance of others. As such,
social rules possess normative force. Only if laws are social
rules in this sense can law's normativity be understood in a
manner consistent with positivism's identification of law with
social fact. In part, this is because acceptance from the internal
point of view is itself characterized in terms of the behavior of
members of the community. Rules are accepted from an internal point of view when the actors treat the rules as providing
them with reasons for acting and with grounds for criticizing
the behavior of those who fail to comply with the rules'
requirements.
Rules are introduced into the concept of law by Hart as a way
of solving the problem of accounting for law's obligatory or
reason-giving nature. It is important that the notion of a rule
Hart introduces is that of a social rule, a rule that is constituted
by convergent social practices, and whose authority as law is
also a matter of social fact. The notion of law as social rule is
important for two reasons. The first, and most important, is
that the concept of authority implicit in law is rooted in behavior, and ultimately in concrete social fact. The second, and
most relevant for my current purpose, is that Hart's notion of a
social rule is built up from and incorporates Austin's "habits of
obedience." A social rule is constructed from convergent social
practices; its content, in other words, is given by what people
do as a rule. To make all well with Austin's account, we can
keep the identification of law with convergent social practices.
We simply jettison the view that such practices are normative
because they have a causal history of a certain sort-that is,
they arise from the sovereign's capacity to render credible
threats-and replace it with the idea that such practices are aspects of social rules. Once again, we see how Hart's position is
built up from his objections to Austin's.
II.
SocIAL RULES AND LEGAL RULES
Introducing this conception of rules into Hart's jurisprudence is not without its share of problems. Law is binding because it consists in social rules. Social rules require convergent
%socialpractices. In fact, law can have authority, even in the absence of convergent behavior with respect to them. In some
cases, legal rules create a convergent practice where none had
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 706 1991
No. 3]
Rules and Social Facts
707
previously existed. In other cases, legal rules adjudicate among
conflicting and competing practices. Rules can be law, and
binding if law is binding, even if they are not social rules and
lack the convergent social practice associated with them. It cannot be the case that the authority of a law depends on the existence of a convergent social practice.
In fact, Hart gives up this account of law's authority. He does
not ultimately require that, in order that law be binding, law
consist in social rules. In the end, Hart's view is something like
this: The authority of a norm as law depends on its relationship
to a master rule-a rule of recognition. Law is authoritative if it
is valid under a rule of recognition. The rule of recognition sets
forth the conditions of legal validity. For each norm
subordinate to the rule of recognition, its authority does not
depend on it being a social rule. Social rules depend on convergent social practices, and in the case of many particular legal
rules, there is no social practice of the relevant sort. Therefore,
we find in Hart an inconsistency that is largely a function of the
development of his theory entirely as a response to the shortcomings he identifies in Austin.
If Hart cannot consistendy advance the view that law's authority depends on law consisting in social rules, what view of
law's authority does he ultimately come to? Hart has a two-part
answer. For rules subordinate to the rule of recognition, legal
authority is a matter of legal validity. Particular norms are binding provided they are valid under a rule of recognition, but
what makes the rule of recognition authoritative?
There are only three possibilities. First, the authority of the
rule of recognition may itself be a matter of its validity under
some other rule. This is hardly a satisfactory solution because it
renders this latter rule the true rule of recognition. Instead of
answering the question, it merely postpones it. Alternatively,
the authority of the rule of recognition may depend on its morality; that is, the rule of recognition is itself ultimately a normative rule whose authority is a matter of its truth as a
principle in some defensible critical morality. This solution will
not work for the positivist for the simple reason that it reduces
positivism to a form of natural law theory. Finally, the authority
of the rule of recognition may itself consist in its being a social
rule, constituted by a social practice among the relevant officials, a rule that they accept from the internal point of view.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 707 1991
708
HarvardJournal of Law & Public Policy
[Vol. 14
This last solution is in fact Hart's position.' The claim that all
laws are binding because they are social rules disappears. Instead, laws are authoritative, provided they are valid under
some other rule, the rule of recognition. Thus, the authority of
particular norms as law is a function of their having a formal
relationship of validity with the rule of recognition. But that
rule's (the rule of recognition's) authority cannot be a matter
either of its formal validity or its substantive defensibility. Its
authority must be a matter of socialfact. Thus, Hart saves the
social rule account of legal obligation by abandoning the claim
that all legal rules are social rules. He replaces that account
with the view that a norm has legal authority only if it is valid
under a rule of recognition that is itself authoritative because it
is a social rule.9
The rule of recognition is a social rule, but a very special one.
8. See H.L.A.
HART,
supra note 2, at 144-50.
9. Two brief points should be made here. First, in the same way that Hart's view is
constrained by the fact that he develops it largely as a response to the shortcomings he
finds in Austin's account, Dworkin's early work, Taking Rights Seriously, is constrained by
the fact that his view is developed largely as a response to the features of Hart's view
that he finds objectionable, especially Hart's theory of adjudication. Thus, it comes as
no surprise that Dworkin develops his theory of law by beginning with a theory of
adjudication. See R. DWORKIN, supranote 1, at 1-44, passim; R. DWORIN, TAKING Rirrs
SERIOUSLY 14-45, passim (1977). Like Hart, whose work is defined by the parameters
Austin's account provides, Dworkin is imprisoned by his connection to Hart's account.
The second point concerns the rule of recognition and is more interesting and controversial. Two possible conceptions of the claim that the rule of recognition is a social
rule should be distinguished. The first is that. the rule of recognition is itself a social
rule in the sense that its content is determined by a convergent social practice accepted
from the internal point of view. The substance or content of the rule is given by the
practice.
The second interpretation of the claim that the rule of recognition is a social rule
recognizes that the rule itself may not be constituted by convergent social practices.
Instead, its authority as a rule of recognition depends on there being a social practice
of accepting it as authoritative.
The difference is important and complex. First, to claim that the rule of recognition
is a social rule is to say that its content cannot be specified other than as a description
of a prevailing practice. In the second interpretation, the one that I favor, the rule of
recognition, in principle at least, can be specified independently of the existence of a
social practice. Rather, the social practice is a complex of behaviors of officials oriented
toward the rule of recognition. There can be a rule of recognition in the second sense
independent of the practice, though there may be epistemic barriers to determining its
content. The first, traditional interpretation, which views the rule of recognition itself
as a social rule, maintains that the content of the rule is given by the practice, whereas
the authority of the rule is given by its acceptance from an internal point of view. In the
second interpretation, it is not a necessary feature of the rule of recognition that its
content be given by a convergent practice. It is this distinction that threatens to reduce
positivism to a form of realism. What makes the second interpretation of the rule of
recognition consistent with positivism is the claim that the authority of the rule depends on the existence of a social practice among officials oriented toward it, a practice
of accepting a rule accepted from the internal point of view. In both interpretations,
there can be no rule of recognition without a convergent practice. The difference is
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 708 1991
No. 3]
709
Rules and Social Facts
It serves two functions in Hart's jurisprudence: One of these is
epistemic, the other ontological or semantic. " The rule of recognition serves an epistemic function to the extent it specifies
conditions of identification, validity, and authority. The rule of
recognition serves an ontological or semantic function1 to the
1
extent that it specifies existence and truth conditions.
Once there is a rule of recognition, the authority of a particular rule need not depend upon its being a social rule. If norms
can have legal authority without being social rules, must they
be rules at all? What constraints does positivism impose on the
sorts of norms that can be law or sources of law? To the extent
that law is a matter of rules, what sort of rules must law be? The
account that Hart gives in the first few chapters of The Concept of
Law suggests that law must consist in rules and that otherwise
law's authority cannot be explained.1 2 Once Hart recognizes
the plain fact that law can be authoritative even if it does not
consist in rules, however, he can no longer require that law
consist in social rules. Does this mean he has to give up the
claim that law consists in rules? Or does it mean that if he remains committed to the view that law consists in rules, he
needs another argument in support of it? If he does not require
that law consist in rules alone or primarily, what else might law
consist in? Does legal positivism have a stake in the extent to
which law is necessarily or primarily a matter of rules?
III.
SCHAUER AND THE POSITIVIST TRADION
Hart does not specifically address any of these questions.
They are, however, among the questions that Professor Frederick Schauer takes up both in this Symposium"3 and in the booklength manuscript, PlayingBy The Rules. 4 Schauer sets for himthat in the first interpretation, but not the second, the rule is constituted by the
practice.
10. SeeJ.
COLEMAN,
Negative and Positive Positivism, in MARKErs, MoAs.s
AND THE
LAW
1, 5 (1988).
11. One of the problems in Raz's version of positivism is the fact that he invariably
AUTHORrrY OF LAw: ESSAYS ON LAw
AND MoRALrrY 212 (1979). So, for example, he claims that the sources thesis applies to
both law's identification and existence conditions. The conditions that must be satisfied
to identify law, however, need not coincide with those that are necessary either to establish law's existence or its justifiability.
12. See H.L.A. HART, supra note 2, at 1-76.
13. See Schauer, Rules and the Rule of Law, 14 HAR. J.L. & PuB. POL'Y 645 (1991).
14. F. SCHAUER, PLAYING By THE RuLEs (forthcoming 1991). In the essay in this volume, Schauer restricts himself primarily to the question: What is the relationship be-
runs these two functions together. SeeJ. RAz, THE
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 709 1991
710
HarvardJournal of Law & Public Policy
[Vol. 14
self the task of analyzing the nature of rules, and then exploring their relationship to law. In his contribution to this volume,
he is most interested in the under- and over-inclusiveness of
rules.
We can distinguish between rules and the sets of factors that
support them. These factors include considerations of justice,
social policy, morality, and efficiency. In some cases, a decision
based on rules will not further the goals or aspirations that support the rule as well as might a decision based directly upon
more particularistic and contextual considerations. Rules are
incapable of being perfectly fine-tuned. Sometimes rules will
include within their domain cases that fall outside their set of
background reasons, and they will exclude others that fall
within that set. These characteristics are a function of rule generality. Thus, rules are necessarily under- and over-inclusive
with respect to the sets of reasons that support or ground them.
Because rules are designed to promote certain background
aims and ambitions, and because rules are both under- and
over-inclusive with respect to those goals and principles, it is a
fair question to ask why law should emphasize rules. Why not
appeal directly to the reasons and leave rules out altogether?
This question presupposes that a particular jurisprudence is a
matter of choice, that a community can choose whether its law
should consist in rules or in the principles that inform rules.
Finally, this raises the question in analytic jurisprudence
whether law is necessarily a matter of rules, for if law is necessarily a matter of rules, then communities are not free to formulate law in terms of background reasons alone--or even
primarily in those terms.
We can distinguish between an analytic question of general
jurisprudence and a question about a particular jurisprudence,
namely, positivism. The question of general jurisprudence is
whether law necessarily is a matter of rules. Is there something
about the nature or essence of law that requires that legal
norms consist in rules? The question about positivism is
whether it is committed to the view that law necessarily consists
in rules. Certainly, Dworkin believes that positivism is committween law and rules? In the book-length manuscript, he provides an analysis of the
nature of rules more generally, and of the way rules figure into our social and normative lives.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 710 1991
No. 3]
Rules and Social Facts
ted to the view that law consists in rules. 5 That assumption is
one of the reasons that in his early articles he concludes that he
has defeated positivism by showing that principles and policies
can be legal norms though they are not rules.
Hart is committed to the view that law consists in rules as a
consequence of his belief that the reason-giving capacity of law
depends on law consisting in social rules. Once the social rule
argument evaporates, so too does the only argument for lawas-rules in The Concept of Law. So if positivism is committed to
the model of rules, there is no argument for that position in
Hart. My view is that positivism is not in fact committed to the
view that law consists in rules.
For his part, Schauer simply assumes that positivism is committed to the view that law consists in rules. 6 Like Dworkin, he
identifies positivism with the jurisprudence of rules and attempts to provide an explication of what rules are. Assuming
that positivism is the jurisprudence of rules, he begins his inquiry with the problem of general jurisprudence: Is law necessarily a matter of rules or not? Schauer's answer to the problem
of general jurisprudence is that law is neither necessarily a matter of rules exclusively, nor is it necessarily a matter of reasons
exclusively. This is the easy answer, but it is an answer to an
easy question as well. The more difficult question is whether
law must be in some part, large or small, a matter of rules, even
if it is not necessary to the concept of law that all law consist in
rules.
Following Schauer, I accept that law need not be exclusively
a matter either of rules or of reasons. It can be a mixture of
both, or it can, in a particular community, be one or the other.
Which mix of rules and reasons is expressed by American jurisprudence? That is a question of particular jurisprudence, and
at first blush, appears to be a question in sociology, not one in
analytic or normative jurisprudence. To the extent that answering it requires a constructive interpretation of our existing
practice, however, it is as much a problem in jurisprudence, analytic and normative, as it is a problem in sociology.
Like Hart, who constructs his account in response to the failings he identifies in Austin, and like Dworkin, who develops his
theory of law as a response to the shortcomings he notes in
15. See R. DWORKIN, supra note 1, at 33-35.
16. See Schauer, supra note 13, at 662-63.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 711 1991
712
HarvardJournalof Law & Public Policy
[Vol. 14
what he takes to be Hart's theory of adjudication, Schauer
stakes out the ground between Hart and Dworkin, who represent the jurisprudence of rules and of reasons, respectively. He
accepts Dworkin's characterization of positivism as the model
of rules. 1 7 At the same time, he characterizes Dworkin as arguing for the centrality of reasons and sources of law, not for
rules of law as such. 8 Rules merely give expression to a set of
principles that provide the best explanation of them. Jurispru-
dence begins with the rights of litigants. These rights derive
from the best theory of the law. Rules are the data from which
the theory is to be constructed, but the judge's task is to enforce the relevant rights, not to apply the rules. Thus, Schauer
identifies Hart's positivism with the model of rules and Dworkin's position with the model of reasons. 9 In advancing the
view that law is neither a matter necessarily of rules nor of reasons alone, Schauer intends to offer an alternative to both
Hart's positivism as well as to Dworkinian rights theory. He
calls his thesis "presumptive positivism," emphasizing the centrality of rules to jurisprudence and the importance of departures from rules to the reasons that support them.
A. Jurisprudenceand Reasons
In fact, as Schauer correctly notes, Dworkin's theory is only
one of a family of possible alternatives to rule-based jurisprudence. Let us call all members of this family "all-things-considered" theories, by which we mean that in a particular case, a
judge should reach the result that is best, all things considered.
The contrasting view is that in particular cases, judges should
reach those decisions that are dictated by the rules, whenever
rules dictate decisions, whether or not the decision is the best,
all things considered.2 ° In this regard, three points need to be
emphasized.
17. See id at 668-71.
18. See id.
19. See id at 665-74.
20. It is unclear exactly the sense in which rules dictate particular results, in
Schauer's view. In one sense, we can say that the rule has particular linguistic indications. For example, suppose the relevant rule of law prohibits motor cars in the park on
weekends. A Ford or Chevrolet automobile is a motor car, and if such a car is found in
the park on the weekend, it stands in violation of the law. On the other hand, we might
say that a rule has decisive normative indications if it turns out that a particular case is an
instance of the policies, principles, and goals that support the relevant rule. Linguistic
indications of a rule need not coincide with its normative indications.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 712 1991
Rules and Social Facts
No. 3]
713
1. Rules, Reasons, and Open Texture
The first point concerns the open texture of rules. A theory
of law that emphasizes rules in the way Schauer believes that
legal positivism does recognizes cases in which rules neither
dictate nor indicate a particular result. A positivist may claim
that judges should decide such cases in ways that produce the
best result, all things considered. In spite of this feature of rulebased theories, an important difference between them and allthings-considered theories remains. This difference surfaces in
those cases in which the rule departs from what would be best,
all things considered. In such cases, all-things-considered theories claim that judges should impose the outcome that is the
best, all things considered, even if the rules dictate a different
result. And this is precisely what someone who believes in the
centrality of rules means to deny. Thus, the open texture of
rules provides an area of overlap between rule- and non-rulebased theories, but it does not imply that rule-based theories
collapse into all-things-considered theories.
2.
Rules, Reasons, and Adjudication
The second point concerns the theory of adjudication. In allthings-considered theories, adjudication is nothing more than
determining what, in a particular context, is the best result. It is
central to Schauer's analysis that rule-based theories can give
different results than all-things-considered theories. This claim
requires that Schauer produce a theory of adjudication-an account, in other words-of the ways in which judges apply and
follow rules. In particular, he argues that rules dictate or indicate results. 21 This is precisely what realists deny. I do not deny
that rules dictate results, but it is incumbent upon anyone who
wants to emphasize a distinction between rule-based and other
theories to show how rules generate results in ways that do not
collapse rule-based theories into all-things-considered
theories.
3. Reasons and Constraints
The third point to note is that there is an enormous variety of
all-things-considered theories that differ in a number of dimensions. First, there can be forward- and backward-looking theo21. See Schauer, supra note 13, at 671-74.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 713 1991
HarvardJournal of Law & Public Policy
[Vol. 14
ries. Economic instrumentalism is a version of the former;
Dworkin's rights thesis is a version of the latter. According to
economic instrumentalism, judges decide cases so as to minimize costs and maximize welfare in thefuture. Legal disputes enable us to more finely hone the incentives that legal rules put in
place; those incentives are, or should be, wealth-maximizing.
According to a version of Dworkin's thesis, judges enforce the
preexisting rights of litigants. In both cases, the role of preexisting rules is of contingent and derivative significance. In the
instrumentalist account, rules are (at best) approximations of
efficient solutions to resource allocation problems. Moreover,
in determining what would be efficient, judges should consider
the extent to which departures from the rule threaten to upset
the coordination effects of well-entrenched rules.
In Dworkin's view, the rules are presumed to be an articulation or expression of underlying principles of a political morality. In every case, even those that fit the core of a rule, the
judge must take recourse to those principles to determine the
litigant's preexisting rights under the law, not under the rules.
Thus, economic instrumentalism is forward-looking; Dworkin's
rights-based thesis is backward-looking. In both cases, explicit
legal rules are understood in the context of approximations of
the background policies and principles. In the former case,
legal rules are efficient solutions to resource allocation
problems. In the latter case, legal rules operate as expressions
of the underlying principles ofjustice and fairness characteristic of constitutional democracies.
IV.
PosrrVsmI
AND PRESUMPTIVE POSMVISM
Schauer articulates a theory located between rule-based positivism and Dworkin's all-things-considered theories: a theory
that lies between rules and reasons. Schauer identifies his position as presumptive positivism, indicating the centrality of rules
in the first instance and the flexibility to consider the background reasons. He attempts to develop his theory in such a
way so as to be conceptually unassailable, descriptively accurate, and normatively attractive.
There are many ways one might seek to criticize Schauer's
position. One might ask whether he correctly describes or characterizes the positions with which he seeks to contrast his own.
On the other hand, he might have these accounts right, but his
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 714 1991
Rules and Social Facts
No. 3]
715
account of Anglo-American jurisprudence may be descriptively
inaccurate.Finally, his account of law in constitutional democracies may be descriptively accurate, but it may be,normatively
unattractive. That is, he may over- or under-emphasize the value
of rules in law.
There are some problems with all of these aspects of
Schauer's thesis, but in an essay of this sort I am in no position
to explore them all. Therefore, I will focus on problems of the
first sort. Frankly, as one of the positivists Schauer discusses, I
don't think he has positivism right. I want to lay out my version
of positivism and contrast it both with Schauer's and Raz's,
with which he periodically aligns himself. I then want to explain
why the version of positivism I advance is descriptively accurate
in ways that Schauer suggests it cannot be. Ultimately, my account is descriptively and conceptually more accurate than
Schauer's.
A.
Negative and Positive Positivism
It is important to distinguish between two forms of positivism. I label these negative and positive positivism.2 2 Both theories attempt to identify the core claims of positivism: just what
positivism as a legal theory seeks to assert or to deny; what its
essential elements are; what every positivist must assert is true
of law necessarily. In "Positivism and the Separation of Law
and Morals," 2 Hart argues for a version of positivism that
merely denies a necessary connection between law and morals.
There is 'no necessary connection between what the law is and
what it ought to be. In other words, the fact that a legal norm is
valid in a particular community does not imply that it is a morally desirable norm of behavior. On the other hand, as Raz
notes in passing, it does not follow that the validity of a norm
as law somehow precludes a norm's moral desirability, even to
the extent that it is compatible with this claim of positivism that
every valid norm of a legal system turns out to be morally desirable. 24 The core claim of positivism is simply that the morality
of a legal norm is analytically distinct from its legality.
Positivists and their critics refer to this core claim of positiv22. SeeJ. COLEMAN,supra note 10.
23. Hart, Positivism and the Separationof Law and Morals, 71 HARv.L. REv. 593 (1958).
24. SeeJ. RAz, supra note 11, at 212. Raz makes the same argument in a more recent
article. See Raz, Legal Rights, 4 OxFoRDJ. LEGAL S-rUD. 1, 10 (1984).
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 715 1991
716
HarvardJournal of Law & Public Policy
[Vol. 14
ism as the separability thesis. Though no proponent or critic of
legal positivism denies that positivism is committed to the separability thesis, there is far less agreement about what it means
and its implications for positivism. Hart characterizes the separability thesis as the claim that there is no necessary connection
between the law as it is and as it ought to be. I understand this
to mean that positivism is committed to the view that the morality of a norm is not necessarily a condition of its legality.
Ultimately, however, positivism does not rule out the possibility that, in some legal regimes, the legality of a norm can depend on its morality. An important difference, in my view, is
that positivism denies that the morality of a norm is a necessary
condition of its legality. The separability thesis can be characterized in other equally defensible ways. For example, it can be
understood as the claim that even in those communities in
which law and morality are coextensive, they are not cointensional. Alternatively, it might be understood as claiming that
what makes a norm a matter of the community's law is distinguishable from what makes it a part of that community's
morality.
All versions of positivism that are characterized entirely in
terms of the constraints imposed by the separability thesis
alone I call negative positivism to draw attention to the sort of
claim that they make, namely, a negative one. Instead of articulating some truth about all law everywhere, negative positivism
simply denies that morality is necessarily a condition of legality
for all possible legal systems.
Given a proper interpretation of the separability thesis, I
think negative positivism is conceptually unassailable and descriptively accurate. There is no logical or conceptual contradiction in asserting that there exists a possible world in which
there is law and in which what makes something law is not a
matter of its morality. So long as such assertions are not contradictory, negative positivism will be unassailable. It is descriptively accurate largely because it makes no descriptive
claim. It does not say that such-and-such is true of law in the
United States, England, or the like. It simply says what is not
necessarily true of law, period. So what is true of law in the
United States, or anywhere else, for that matter, will be perfectly compatible with positivism, so understood. Therefore,
while I am convinced that negative positivism is both concep-
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 716 1991
No. 3]
Rules and Social Facts
717
tually unassailable and descriptively accurate, it is also true that
I do not think it a very interesting thesis-and neither should
you.
One might impose an adequacy condition on theories of gen-
eral jurisprudence that they make a claim about what, if anything, is true about all possible legal systems. That is a
reasonable constraint on anything that purports to be a general
jurisprudence, as opposed, say, to an American jurisprudence-a subject about which I for one have considerably less
interest. Now it is perfectly compatible with all that I have said
so far that there is nothing that is true of law as such: that all
features of legal systems other than that captured by negative
positivism are contingent features of them. This is a perfectly
plausible thesis, and perhaps it is the "positive" claim advocates of negative positivism are prepared to put forward.
B.
UnderstandingLaw's Sociology
Still, I think a positivist can say more, and what that something else is is captured by a form of what I call positive positivism. Theories are instances of positive.positivism if they make a
claim about what is true of all legal systems necessarilya-as part
of the veiry concept of law or legality-claims that are consistent with and motivated by the underlying commitments of a
positivist jurisprudence. What are those commitments? So far,
following Hart, we have identified positivism with the separability thesis. Is there anything other than the separability thesis
to which positivism is committed? Raz, for one, identifies positivism with the claim that law is essentially or fundamentally
institutional in nature. 25 I agree, but the claim needs to be clarified further. Both Raz and I want to express the institutional
nature of law in terms of the claim that law is ultimately a matter of social fact. In my view, law is ultimately a matter of social
fact in the sense that the authority of the rule of recognition is
itself a matter of social convention. I want to explain what I
mean by this claim. In doing so, I shall contrast my position
with Raz's, identify where Schauer thinks I run into trouble,
and respond to his objections.
Let me begin by noting that the view I advance, that law is
ultimately a matter of sociology or social fact, is not original
25. SeeJ. RAz, supra note 11, at 191.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 717 1991
HarvardJournal of Law & Public Policy
[Vol. 14
with me. In the introduction to The Concept of Law, Hart claims
that the book can be viewed as an inquiry into analytic jurisprudence or descriptive sociology. 26 This is a deep point. For I
believe the point of positivist jurisprudence is to demonstrate
exactly how thin the concept of law is; how few are the substantive inferences that can be drawn from it; how minimal its
moral content is. If one wants to know anything particularly
substantive about a legal system or culture, one has to go beyond analytic philosophy to sociology, to an account of the
ongoing practices that constitute a community's legal culture,
an account that is, in a variety of ways, internal to the culture
27
and to its practices.
It is also worth noting how deep the commitment to jurisprudence-as-sociology runs in The Concept of Law. First, when discussing how law can be binding, Hart notes that the answer lies
in the fact that law consists in rules. What kind of rules? Social
rules!2 ' And what exactly are social rules? They are constructed
from existing social practices. Moreover, what makes them
binding or authoritative? The fact that they are accepted from
an internal point of view. But how does Hart analyze acceptance from an internal point of view? In terms of social behavior, that's how. Rules are accepted from an internal point of
view if individuals use them in a certain way, if they characteristically appeal to the rules to provide grounds for criticism and
29
reasons for action.
Next, when Hart abandons the view that a law is binding only
if it is a social rule, what does he put in its place? Two related
notions. First, that the legality of a norm depends on some fact
about it. What fact? That it satisfies the conditions of legality
set forth in the rule of recognition. That is either true of a
norm or it is not, and whether it is depends on the criterion of
legality set forth in the rule of recognition. Whether or not it is
true of a norm is a social fact about it. Second, the authority of
the rule of recognition itself is a matter of social fact. Hart puts
26. See H.L.A. HART, supra note 2, at v-vi.
27. One cannot doubt that the central positive claim of positivism is that law is ultimately a matter of social fact. Returning to Austin, one again notes that the law is the
order of a sovereign, but what makes someone the sovereign? Two social facts provide
the answer: First, she has the habit of obedience from subordinates, and second, she is
not herself in the habit of obeying anyone. Raz and the sources thesis also shed light on
this inquiry. SeeJ. RAZ, supra note 11, at 210-16.
28. See H.L.A. HART, supra note 2, at 86-96.
29. See id at 55-56.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 718 1991
No. 3]
Rules and Social Facts
719
this somewhat differently. He says the rule of recognition is itself a social rule. It is not valid or in some other sense correct;
it just is. In either case, legality is a matter of social fact.
There are a lot of ways of understanding these claims, but
Dworkin does not seem fully to appreciate them. The claim that
law is a matter of social fact is sometimes treated by Dworkin as
the claim that law is a matter of plain fact or historical fact or
uncontroversial fact or else it is a matter of pedigree. s ° It is
evident that none of these claims is equivalent to or otherwise
entailed by the claim that law is a matter of social fact. Social
facts may or may not be plain, simple, historical, or
uncontroversial.
Schauer follows Dworkin in ascribing to positivism the view
that the legality of norms is amatter of their pedigree. 3 I fail to
see how either the separability thesis or the claim that law is a
matter of social fact entails this requirement. Let's return to the
distinction I have mentioned between the epistemic and semantic senses of the rule of recognition. Now if the rule of recognition is a semantic rule-as I have argued it is-there is no
reason why the conditions of legality set forth in it should rely
entirely on pedigree.3 2 On the other hand, suppose the rule of
recognition is an epistemic rule, that is, it specifies criteria by
which individuals can come to know what the law of their community is. In that case, the claim that the rule of recognition
must set out a pedigree test of legality is understandable, but it
is still incorrect.
The basic idea is this. Let's contrast a pedigree standard with
a moral standard. A pedigree standard will look something like
this: (x) (x is a proposition of law iff x has been passed by a
legislature and signed by the relevant executive, or x is an established judicial precedent). A moral standard may look like
this: (x) (x is a proposition of law iff x represents a dimension of
justice or morality within the best moral theory). In order to
determine whether something is part of the community's law
under the typical pedigree standard, one need only look it up.
It can be found by nearly everyone, at least in principle, and so
everyone can determine by the use of the relevant rule of recognition what their rights and duties under the law are. On the
30. See R. DWORKIN, supra note 1, at 33-35.
31. See Schauer, supra note 13, at 666;R. DWORKIN, supra note 9, at 17.
32. See supra p. 709.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 719 1991
HarvardJournal of Law & Public Policy
[Vol. 14
other hand, determining what the law is under the moral criterion is no easy matter. It requires substantive moral argument.
Controversy about one's rights and duties will be inevitable,
and individuals guided by such a rule of recognition will be unlikely to ascertain adequately, let alone fully, the scope of their
legal rights and obligations. Only if one identifies positivism
with Hart's version of it does it make sense to characterize positivism as committed to a pedigree standard of legality.
But there is no reason to identify every claim Hart advances
with an essential commitment of positivism more generally.
Moreover, epistemic adequacy does not require pedigree. Pedigree standards are among those standards that in principle can
be epistemically adequate. In principle, any non-contentful criterion of legality can be epistemically adequate. The real problem surrounds the question whether contentful standards of
legality can be epistemically adequate. To answer this question
completely, we will need a good analysis of what it means for a
standard to be contentful as well as a conception of epistemic
adequacy. In the context of this paper, we shall settle for something less. Let's say that a standard is contentful if determining
whether something is law under it requires an evaluation of the
norm's value. On the other hand, let's say that something is
epistemically adequate if average citizens can reliably call upon
it to determine with an acceptable degree of confidence what
their legal rights and responsibilities are.
It follows that some contentful standards of legality can be
epistemically adequate. Substantive consideration of a norm's
value need not stir controversy nor need a criterion that requires such assessments as a condition of legality prove epistemically unmanageable. Certainly in communities that share
some fundamental set of values, not only in the abstract, but in
concrete particulars as well, reference to a norm's value as a
condition of its legality need not render the rule epistemically
inadequate or essentially controversial.
Of course, some possible rules of recognition will be less
than epistemically adequate in this sense. The rule that a norm
is a legal norm iff it is part of the correct morality may well be
such a rule. But even here notice how much is being assumed
about the background conditions of the community. When critics like Dworkin deny that any such rule can be a rule of recognition in the positivist's sense, they are presupposing that any
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 720 1991
No. 3]
Rules and Social Facts
such rule will be a source of substantial controversy. If it is a
source of substantial controversy, it cannot be a social rule nor
can it be epistemically adequate. It cannot be a social rule because social rules are constructed in part from convergent practices and such rules lead to divergence, not convergence. 3 3 It
cannot be an epistemic rule because individuals cannot appeal
to it in a reliable way to determine their legal rights and duties.
But neither of these claims really follow, do they? In my view,
as I noted earlier, the rule of recognition is not itself a social
rule. It is authoritative only if there is a social practice in regard
to it among relevant officials. Therefore, it need not be constructed out of convergent practices. And, in particular communities, there may be so much agreement about the demands
of a correct morality that the rule of recognition is in fact epistemically adequate in that community. In other words, the
claim that such a rule cannot be epistemically adequate has as
its background condition a view of societies like our own-extremely heterogenous and pluralistic. That is a contingent feature of social organizations, not a necessary one.
Moreover, I deny that the rule of recognition must be epistemically adequate in this sense. My claim is that an analysis of
the concept of law is fundamentally a metaphysical inquiry; and,
as such, the rule of recognition must specify ontological or semantic conditions only. Is the analysis fundamentally a metaphysical one?
To see where the issue lies, consider the difference between
Raz and me. Recall that we are discussing my claim that positivists are committed to the view that law is in some sense a
matter of sociology or social fact. Raz and I differ on what the
proper interpretation of this claim is or should be. For Raz, it
entails the view that what makes any norm a matter of law must
be a social fact about it: This is his famous Sources Thesis. 4
Thus, even if the norms themselves need not be social rules or
facts, the conditions of legality must be social facts, as opposed,
say, to moral criteria. That, for Raz, is what it means for the law
to consist in social facts. And it is a position he believes is true
of both law's epistemic and existence conditions. Thus, in Raz's
view, "(x) (x is a legal norm iff it is a dimension of justice or
morality in the best critical moral theory)" could never be a
33. SeeJ.
COLEMAN, supra note 10, at 12-20.
34. SeeJ. RAz, supra note 11, at 212.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 721 1991
HarvardJournal of Law & Public Policy
[Vol. 14
rule of recognition, for such a rule would identify the grounds
or sources of law with moral arguments, not social facts.
In my view, there is no reason why such a rule cannot be a
rule of recognition, at least in principle. Therefore, I must have
something else in mind when I claim that the law is ultimately a
matter of sociology or social fact. I do. My claim is not that the
conditions of legality must specify only social facts about
norms; rather, the claim is that a particular rule of recognition
that is the rule of recognition in a particular community is a
social fact about that community. For Raz, the social fact thesis
is a constraint on the content of the rule of recognition itself. In
my view, neither the separability thesis nor the view of the rule
of recognition as a semantic rule imposes any constraints on
the substantive content of a rule of recognition. Virtually anything can in principle be a rule of recognition. That a particular
norm is the rule of recognition in a community is in my view a
social fact about that community; it is not, for example, a claim
in critical morality, and therefore, its truth does not depend on
substantive moral argument. That in my view is the cash value
of the claim that law is a matter of social fact. In short, determining which norms are part of a community's law may well
involve substantive moral argument given a particular rule of
recognition; that that rule is the rule of recognition, however, is
a social fact about the community that does not require moral
argument for its truth.
V. THE LIMITED-DOMAIN THESIS
Properly understood, the separability thesis does not restrict
the content of a rule of recognition. Thus, in some community,
the rule of recognition may be: x is law in the community iffx is
a principle of a correct or defensible morality. In that case, a
community's law will be a proper subset of morality. This is a
logical consequence of my view. Is there anything wrong with
it? Both Schauer and Raz think so. Schauer asserts that positivism insists on a distinction between "law" and "non-law," and
he cites a variety of people to that effect.8 5 And since my view
entails the possibility that in a particularcommunity, there may
be no difference between the two, either my view is incorrect or
it is not positivism. This objection is groundless, however.
35. See Schauer, supra note 13, at 666 & n.41.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 722 1991
No. 3]
Rules and Social Facts
723
Suppose that, in every possible legal system, the community's law completely coincided with what was morally correct.
This would show only that law and morality are everywhere coextensive, not that they are cointensional. As I understand positivism, it is the intensional (conceptual) relationship between
law and morality that counts. Even if law and morality coincide,
what makes something a moral principle is the fact that it is so in
the correct theory of morality; and what makes something law is
the fact that it satisfies the conditions of legality in the rule of
recognition. The ontological and semantic criteria of law and
morality continue to differ. Generally, if law and morality coincide in a particular community because the criterion of legality
is x is law only if it satisfies the criterion of morality, the distinction between law and morality would remain. There is a criterion of morality, and there is the separately identifiable
criterion of legality.
It wouldbe a very different thing if it were not merely a contingent social fact about all legal systems that their rules of recognition made morality a condition of legality, if, for example,
it could not be otherwise, that the concept of law required that
law be a matter of morality. Then, the relationship between legality and morality would be intensional. Only then would
there be no separation between law and morality of the sort
envisioned by the separability thesis.
There is an important sense in which anyone who accepts the
separability thesis believes that there is a difference between
the categories of "law" and "non-law." The difference is one of
criteria of meaning, not extension.
Schauer appears to treat the claim that there is a difference
between law and non-law as equivalent to what he calls the limited-domain thesis, by which he means that in any community the
36
law must be only a part of that community's stock of norms.
The limited-domain thesis is not entailed by the law/non-law
distinction. The former is a claim about concepts; the latter is a
claim about norms in particular communities. As I demonstrated above, in some community, the law may coincide completely with its morality, and yet there will be the analytical or
conceptual difference between law and non-law.
Because the limited-domain thesis is not entailed by the sep36. See id at 667.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 723 1991
724
HarvardJournalof Law & Public Policy
[Vol. 14
arability thesis, it must stand on its own. The limited-domain
thesis's truth is supposed to present a problem for my view.
Because my view allows anything to be a rule of recognition, it
allows law to be unlimited, whereas in fact law everywhere is
limited. Therefore, my form of positivism, even if it is conceptually unassailable, is descriptively inaccurate. Nothing could be
further from the truth. My view is not that law will invariably be
unlimited, only that in principle it can be; it all depends on the
particular rule of recognition in effect. Very likely, the law of
most communities will be limited by the rule of recognition in
just the way Schauer thinks our law is limited and for just the
sorts of reasons Schauer has in mind. My thesis, remember, is
that legal theory is ultimately a matter of descriptive sociology,
not conceptual analysis. So how can its conceptual claim, which
after all is just, law is sociology, render it descriptively
inaccurate?
It is my view that the law can in a particular community be
unlimited. Perhaps Schauer's point is that because of the limited-domain thesis, law can never be unlimited. I hold that law
can in principle be unlimited. But the limited-domain thesis denies that it can be. Then my theory fails, not because it is descriptively inaccurate, but because it is conceptually
inadequate. Positivism is committed to both the separability
and social fact theses. Neither of these entail the limited-domain thesis. Therefore, there are no grounds for rejecting my
version of positivism because it violates the limited-domain
thesis, which is itself unmotivated by the concerns that animate
positivism in the first place.
In short, positivism as I understand it does not claim that law
is necessarily a matter of rules. Nor does it claim that law necessarily has a limited domain. It does claim that law and morality
are distinct and that law is a social fact. I have cashed these out
in the following way. The difference between law and morality
is conceptual; it has to do with the criteria of meaning of the
terms. Terms with different meanings can have the same extensions, and so the law of a community can completely overlap
with its morality. The criterion of meaning of law is the rule of
recognition; and what that rule is in a particular community is a
matter of social fact; and that is the social fact aspect of law.
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 724 1991
No. 3]
Rules and Social Facts
VI.
725
CONCLUSION
I have argued that the limited-domain thesis is itself unmotivated, but this claim is contentious. I want to close by sketching
a version of positivism that grounds the limited-domain thesis
in an account of the relationship between law and practical reason. I attribute this view to Joseph Raz. The theory of practical
reasoning is at the heart of Raz's work. 37 What is interesting
about law for Raz is that it can provide reasons for acting, reasons of a special kind, peremptory or exclusionary reasons.
Such reasons are imposed by authorities. Law must claim for
itself the possibility of being such an authority. This is a necessary feature of law. In other words, the concept of law entails
the possibility of authority. If law can be the sort of thing that
has authority, then it must be the sort of thing that provides
reasons of the sort authorities do. These are exclusionary reasons. Now it is part of Raz's view that norms that provide such
reasons cannot be ftindamentally controversial, because if they
are, they cannot guide action in the appropriate way. Thus,
legal norms cannot be substantive in the way in which my theory says they can be. For such norms cannot be authoritative in
the way in which they must be if they are to provide reasons for
action.
Raz's view then requires the version of the social fact thesis
he advocates, and it precludes mine. It also entails a dichotomy
between law and non-law in every community in which there is
law. But notice that in Raz's case, the limited-domain thesis is
motivated in a way in which it is not in Schauer's case. To evaluate whether Raz or I get the better of this, we would need to
determine first whether the concept of law is connected to the
concept of authority in the way in which Raz says it is; whether
the concept of authority is to be understood in the way Raz says
it must be; and finally, whether analytic jurisprudence is itself
part of the theory of practical reasoning. In fact, I want to deny
each of the claims Raz either explicitly or implicitly makes on
behalf of his thesis, but that is an argument for another
occasion.
37. See, e.g., Perry, Second-Order Reasons, Uncertainty, and Legal Theory, 62 S. CAL. L.
REV. 918 (1989).
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 725 1991
HeinOnline -- 14 Harv. J. L. & Pub. Pol'y 726 1991