Voluntary Obligations and Normative Powers

Neil MacCormick; Joseph Raz

Proceedings of the Aristotelian Society, Supplementary Volumes, Vol. 46. (1972), pp. 59-102.

Stable URL:
http://links.jstor.org/sici?sici=0309-7013%281972%2946%3C59%3AVOANP%3E2.0.CO%3B2-W

Proceedings of the Aristotelian Society, Supplementary Volumes is currently published by The Aristotelian Society.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained
prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in
the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/journals/aristotelian.html.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org.

http://www.jstor.org
Thu Mar 20 22:31:44 2008
 VOLUNTARY OBLIGATIONS AND

NORMATIVE POWERS

Neil MacCormick and Joseph Raz

I-Neil MacCormick

Civil lawyers at least since the time of Gaius have been in the
habit of taking 'the law of obligations' as a separate branch of
study; within obligations, the practice of distinguishing
'voluntary obligations' from what Lord Stair called 'obediential
obligations' is also fairly venerable. Voluntary obligations are
of course those which arise from contracts and promises,
obediential include, e.g., those whose incidence is determined
by the law of delict. The point of the distinction is clear
enough; I can choose whether or not to buy your house, and
only if I choose to do so will I incur a n obligation to pay you
the price. If on the other hand I wilfully burn your house
down, I am in breach of an obligation which applies to me
whether I like it or not; and the existence of my consequential
obligation to compensate you is equally independent of my
choice. In the seventeenth century, at least, legal theorists
such as Grotius, Pufendorf and Stair treated it as a very
significant question to inquire into the nature and basis of
voluntary obligation-how can the will obligate itself?
Sometimes such disputes had significant practical results. For
example, according to Grotius, it is of the essence of promises
that they cannot bind without acceptance, whereas Stair took
the opposite view. Grotius' view prevailed in most continental
systems, but in relation for example to the rights of third
parties under contracts, Stair's view is the basis of the modern
Scots law.
I t might be thought, however, that speculation as to the
nature of voluntary obligation is essentially a misconceived
activity, especially when it is supposed to yield solutions to
practical legal problems. To common lawyers, there must
60 I-NEIL MACCORMICK

appear to be a certain futility in the efforts of French doctrinal

writers to settle the question when a contract made by postal
correspondence is to be deemed complete-whether on
declaration of acceptance, posting of it, delivery to the offeror's
address, or actual scrutiny of it by him. Such problems are not
to be solved by theorising as to the true nature of accord de
volonte; but by laying down more or less arbitrary rules.
The point of the objection could be put thus: in law,
contractual and other 'voluntary' obligations presuppose the
existence of power-conferring rules. Because, and only because,
certain rules exist in virtue of which persons may by performing
certain actions place themselves under obligations to perform
further acts specified by themselves, it therefore follows that
performance of the rule-defined actions 'generates' valid legal
obligations. Thus English law contains a rule by virtue of
which persons who sign, seal, and deliver a certain sort of
document called 'a deed' thereby incur a duty to perform
the acts specified in the document. There would be no point
in dedicating ourselves to a study of 'deeds' as such, or of the
acts of will or other mental states of parties who execute
deeds, with a view to discovering what it is in their intrinsic
nature which explains why they generate obligations. The only
answer to that question 'why?' is to point to the power-
conferring rule. A serious question which might then be raised
would be to ask for the justification of that power-conferring
rule; but the answer to that should be in terms of the general
expediency of providing 'facilities' (Professor Hart's term) of
this sort.
I t need scarcely be added that this legal-philosophical thesis
is neatly paralleled currently both in moral philosophy and in
the philosophy of language. Promising, it is said, can be
analysed as a speech act in an intelligible way only if it is
observed that particular promissory utterances essentially
'count as' promises only because they are intended to be and
are recognized as instances of a social 'practice', or 'institution'.
The act presupposes the practice; the practice is rule-defined
in that widely recognized social rules exist under which if any
person S says to another A 'I promise that I shall 4' it becomes
obligatory upon S to 4, unless he can point to some defeasing
or excusing circumstance which either exceptionally negates
 the existence of the promise or excuses him from his obligation
under it. The conditions of defeasance or excuse are themselves
defined by existing social rules, which are no doubt constantly
evolving and being further refined and developed by the usages
of persons in society.
Thus if anyone were so misguided as to ask why promises
generate obligations the answer must be, 'because the rules of
the practice are such that the serious utterance of certain
sentences has a specified illocutionary force and therefore
counts as incurring an obligation'. Then it becomes clear that
what has to be morally justified is the existence of the practice,
not the judgment that performance of this or that promise is
prima facie obligatory. TJtilitarian arguments may well be
advanced at this stage, and Professor Rawls has indeed
suggested that this may rescue utilitarianism from certain of
the traditional objections to it. The practice as a whole is
justified on grounds of utility, but the rightness and wrongness
of particular acts of promise keeping and promise breaking are
to be judged by reference to the rules of the practice, not by
immediate considerations of utility. The utility of the practice
in this view presupposes the existence of the rules which
constitute it.
So the overall picture appears to be very neat and tidy. The
law of voluntary obligations confers upon people a power to
create legally enforceable obligations by making promises
which satisfy certain formal and/or substantial conditions laid
down in the law. The concept of 'promising', which the law
rather takes for granted than defines, is itself to be explained
in terms of a logically prior social practice whose constitutive
rules confer on people a normative power to perform certain
speech acts which 'count as' promises and which give rise to
obligations upon the people who perform such speech acts.
Neat though the picture is, I find it quite unconvincing. My
object in this paper is to show that it is indeed unconvincing
because it is wrong. It is quite unnecessary in the explanation
of promising and of the obligation to keep promises to invoke
the concept of a power-conferring rule enshrined in a social
practice as a necessary condition of the possibility of promising.
If, on the other hand, it is necessary in analyzing the law of
voluntary obligations to make reference to normative powers
62 I-NEIL MACCORMICK

(i.e., powers conferred by rules), that is a consequence not of

the nature of promising but of the legal enforceability of those
promises which constitute legal contracts.

To take the first of these assertions first, promising as a speech

act will here be explained as an utterance of the speaker's
about his own future conduct which is essentially characterized
by the speaker's intending his addressee to take it as being
intended to induce the addressee to rely upon the speaker's
taking the action in question. The relevant concept of 'reliance'
does not itself presuppose the existence of rules under which the
addressee is entitled to rely on the promisor's keeping his word.
And the fact of the addressee's reliance on the promisor is
sufficient ground for asserting that the promisor has an obliga-
tion to keep his word.
If these propositions are well founded, then promises need
n o t be represented as instances of the exercise of normative
powers conferred by rules. But of course it will still be possible
to represent them in terms of a power-conferring rule, thus:
'A person having the following characteristics [age, soundness
of mind] may by the following acts . . . "make a promise";
anyone who so makes a promise has an obligation to perform
the acts mentioned in his promise.' But, so far from being the
only terms in which 'promise-making' and 'promissory
obligation' are explicable, that rule is itself explicable in terms
of the existence of good reasons why promise-breaking should
be regarded as wrongful, given the essential nature of the
activity of promising.
It will be helpful to make clear the point of this suggestion
by drawing a contrast with two other speech acts which are
also 'acts-in-the-law'. In some legal systems, a man can divorce
his wife simply by the act of saying 'I divorce thee, I divorce
thee'.l I n the laws of the U.K., an auctioneer who declares
the subjects of the auction 'sold to Mr. So-and-so', thereby
effects the sale. But the concept 'divorce' can only have
meaning, and an act can only count as 'divorcing', if there
exists a legally defined status of marriage whose commencement,
is effected by certain legally defined acts, whose continuance
entails the applicability of numerous laws granting privileges,
imposing duties, creating powers and so forth, and whose
termination is likewise regulated by the law. So, too, the
concept 'sale', implying as it does the passing of ownership
from one person to another presupposes a concept of 'owner-
ship' which can be accounted for only in terms of laws, that is,
investitive, constitutive and divestitive laws, as Dr. Raz has
pointed out,2 following Bentham. Thus the utterances 'I
divorce thee' and 'Sold to Mr. So-and-so' are incomprehensible
save in the context of such laws of status and of property.
And they can take effect as they are intended to, only if the
particular legal system in question does actually confer the
appropriate power. Does the system recognize talaq divorce ?
Does it recognize ownership, and does it recognize a principle
of agency such that an auctioneer can validly sell goods on
behalf of this owner? If so, these speech acts can be felicitously
performed as meaningful acts-in-the-law, and if not, not.
A full analysis of the talaq divorce or the auction sale as
speech acts would therefore necessarily include a n intention
on the speaker's part to be taken as invoking such laws. The
clear-headed husband or auctioneer addressing a comprehend-
ing audience intends it to be recognized that his utterance is
intended to take effect as 'a divorce' or 'a sale'; the addressee
of the act is clearly aware that the act is done with that
intention, and that he is intended to recognize that intention,
and so on. Necessarily, the act is performed and understood
in the light of 'secondary' social and legal rules.
From such acts promising can be radically differentiated.
It is not necessary to suppose that promisors either intend to
invoke, or to be taken as invoking, any particular moral social
or legal rules in making promises. Contrary to Searle's view,3
promises can be shown to be speech acts explicable in terms of
an intention to bring about a specific perlocutionary effect,
and an intention that that intention be recognized. Searle's
summary dismissal of the possibility of such an explanation is
unconvincing :
"There is no perlocutionary effect of . . . promising
which will distinguish promises from firm statements of
intention and emphatic predictions. All three tend to
64 I-NEIL MACCORMICK

create expectations in the hearer about the future, but

'I promise' does not mean 'I predict' or 'I intend'."*
The tacit supposition that 'creating expectation' is the only
conceivable such effect which might be adduced in explanation
of promises is a false supposition; so that argument doesn't
amount to much. There is an alternative which will both
satisfactorily serve to explain promising and clearly distinguish
it from predicting etc., namely that promises do, and are
calculated to, induce their addressees to rely upon future
performance by the promisor. This involves more than merely
creating expectation. If A believes that B intends to take some
course of action, or to refrain from taking some action, and if
A then himself acts (or refrains from acting) in a manner
which is, or is likely to be, detrimental or at least unprofitable
to himself unless B acts as A believes he will, then A may justly
be said to have acted 'in reliance' on B. So reliance involves
not merely having expectations about others' conduct, but
basing one's own conduct on such expectations, as when a
host takes trouble and expense in preparing dinner for an
expected guest, or when a man deliberately refrains from
catching a train because he believes his neighbour will give
him a lift to work. 'Inducing reliance' means intentionally or
knowingly inducing a person to base his conduct on expectations
about one's own.
This concept of reliance can be well illustrated by references
to the law, for example, to the recent English case of Inwards
v. Baker [1965] 2 Q.B. 29. In 1931 young Mr. Baker was
contemplating buying land to build a house, but the site he
had in mind proved too expensive. His father ('old Baker')
said to him, "Why not put the bungalow on my land and make
the bungalow a little bigger?" So young Baker set about
doing this, with his father's help and encouragement. I n 1933
he completed his bungalow, and there he resided until his
father's death in 1951. After his death, the plaintiffs who were
trustees of his will (old Baker's mistress and her two daughters)
permitted him to continue in residence until 1963 when they
instituted proceedings for possession of the house. They lost
their action. The grounds for the Court of Appeal's decision in
young Baker's favour were :
( I ) young Baker had put himself to trouble and expense in
 VOLUNTARY OBLIG.4TIONS A N D NORMATIVE POLVERS 65

building the house because he believed that his father

would let him stay there as long as he wished
( 2 ) that belief was actively encouraged by his father, who
knowing that such was his belief stood by and let him
go ahead with the building
(3) generally, if any person has encouraged, or knowingly
acquiesced in, another's acting irrevocably to his
detriment in reliance upon expectations as to the
former's future conduct which the former has knowingly
or intentionally induced in him, then the Court must
restrain the former from going back upon the under-
standing
(4) the principle applied equally to the trustees of the will
as successors in title to the land.
To non-lawyers, this no doubt seems a rather roundabout
way ofdoing obvious justice; but the reason whythe roundabout
route was chosen is most instructive. The arrangement made
between old and young Baker was not such as to imply any
agreement to transfer any legal estate recognized by the English
law of property which the Court could enforce by directing
that the appropriate estate be vested in young Baker. Precisely
because the powers of transferring ownership created by
property law had not been invoked b y the parties, they could
not be invoked by the Court. It followed that the son's position
could be protected only under the principle of 'equitable
estoppel'-you may not go back on an arrangement to let
someone occupy your land if you have stood by and let him
spend money on it in the belief, encouraged by you, that you
will let him stay there.
Another type of case relates to action in reliance upon repre-
sentations of present fact rather than of future conduct. T h e
old case of Neville v. Wilkinson ( I 788) I Br0.C .C. 543 arose out
of negotiations for the marriage of the young and aristocratic
but profligate Mr. Neville to the daughter of a rich merchant.
Before he would consent to the marriage, the merchant
insisted upon seeing a statement of all Mr. Neville's debts. The
latter's agent, Mr. Wilkinson, prepared a statement showing
that Mr. Neville owed the sum of i(;18,ooo, but deliberately
excluded from the total the further sum of i(;8,ooo owing to
himself. The merchant paid all those debts, and the marriage
66 I--NEII. MACCORMICK

went ahead, whereupon Mr. Wilkinson, confident that he had

secured his debtor's means, proceeded to sue him for the sum
owing to him. I t was held that he was 'estopped', that is
precluded, from giving the lie to his earlier statement, since the
parties had acted in reliance upon its truth, as he had intended
and foreseen that they would.
Be it noted that these legal doctrines do not depend upon
any concept of intent to incur a legal obligation. Whether or
not the parties intended to act with reference to legal rules
under which they might incur obligations towards other parties
who acted in reliance upon their words is a perfectly indifferent
question at law. What matters is whether one side did in fact
rely upon the other's representations about future conduct or
present fact; and whether that other intended that they should
or knew that they would. Sometimes it is even the case that
persons make statements hoping to evade rather than incur
obligations, as when a contract requires in law to be reduced
into writing, and A falsely assures B that it has been so written;
B performs his part, then A tries to plead as against B that the
contract cannot be enforced for want of writing; but A is
estopped from giving the lie to his earlier assurance, upon
which B has acted. Here A neither intended to incur an
obligation nor to be taken as incurring an obligation; he
intended to evade an obligation, and he intended B to take it
that he had already fulfilled an obligation. But at all events
he intended that B should act irrevocably in reliance on his
assurance, and that will suffice to debar him from now taking
advantage of B by pleading the lack of writing as a defence.j
In all these situations there are two elements: ( I ) the fact
that one person has tz.ken detrimental action in the belief
either that another will take certain action in the future or
that present assurances of the other are true, that is the fact of
reliance; and (2),the fact that the speaker intended the other to
rely upon him, or, we may add with equal force, knew or
thought it likely that he would so rely. I hope we can quickly
agree that the first of these, the bare fact of reliance, is
ordinarily quite indifferent morally, as it is and should be
legally. A student applies for admission to a University because
he wishes to do research with Dr. Jekyll. Dr. Jekyll does not
know that this is the reason for the student's application, and
after recommending its acceptance he himself takes up a post
at a different University before the student comes into residence.
The student did in fact act in reliance on Dr. Jekyll's staying
put, but Dr. Jekyll neither encouraged him to do so nor knew
he had done so. He has therefore no obligation to avoid
disappointing him. It would have made all the difference if
Dr. Jekyll had intentionally or knowingly induced the student
to act in reliance on him. A further illustration will show why
that makes so great a difference.
Suppose that Jones has been swimming from a beach at the
foot of a cliff on a stormy day. He has failed to notice the speed
with which the tide is rising, and is now in such a position
that a desperate dash along the beach will perhaps enable him
to reach the path up to safety before he is cut off by the tide.
Macdonald happens just then to be strolling along the cliff top
carrying a few hundred feet of stout climbing rope. He spots
Jones's predicament and lowers the rope down the cliff-face
to him. Jones sees what he is doing and waits for the rope to
reach himself, thus losing time so that he can no longer
conceivably reach the path before the tide comes in. He then
starts to climb the rope.
These facts entitle us to infer in relation to Macdonald's
state of mind:
either ( I ) He intended Jones to climb up the rope, and
intended Jones to recognize that he intended him to
climb u p the rope in reliance upon Macdonald's
holding it till he reached the top.
or ( 2 ) He knew or thought it likely that Jones would
suppose that Macdonald intended him to climb up the
rope in the belief that Macdonald would hold on till
he reached the top.
Even if Macdonald could, in very exceptional circumstances,
satisfy us that he acted without adverting to the possibility
of Jones' making such suppositions, we should be justified in
saying
(3) Macdonald ought to have realized that Jones
would make the supposition mentioned in ( 2 ) above.
From the moment at which it ceased to be possible for Jones
to reach safety by running, because he had waited for
Macdonald's rope to reach him, as Macdonald intended he
68 I-NEIL MACCORMICK

should or knew he would, we can surely say that it would be

wrong for Macdonald to abandon the enterprise, drop the
rope, and go away-it would be wrong even if he could do
equal good, save another person, even two other persons, by
going and rescuing them from some other predicament. Jones
has acted in a way which is now irrevocable, and which will
be greatly detrimental to him unless Macdonald holds on to
the rope till Jones has climbed up. Jones so acted because
Macdonald intended him to take it that Macdonald intended
him so to act (or knew that he would). Macdonald therefore has
a n obligation to hold on to the rope till Jones reaches the top.
Why do we say that Macdonald has this obligation? Need
we suppose that he belongs to a society which recognizes some
practice or institution according to which persons have a
power, by throwing ropes over cliffs, and such like acts, to
incur obligations? What if in that society it is accepted that
nobody is ever deemed to have made a promise unless he uses
the explicit performative formula 'I promise to +', rather as in
republican Rome stipulations could (according to one widely
held view) only be made by the promisor's replying 'spondeo'
to the stipulator's question 'spondesne . . . ?' I venture to think
not. Is it not sufficient that we should assert and accept the
following principle: if one person acts in a potentially detri-
mental way in reliance upon beliefs about another's future
conduct, and if the latter person has by some act of his inten-
tionally or knowingly induced the former to rely upon him,
then the latter has an obligation not to act in a manner which
will disappoint the other's reliance.
T h e use of the word 'obligation' in that principle signifies
the force and function which the principle is conceived to have.
T h a t is to say, (following but perhaps modifying Professor
Hart6), it is a principle breach of which will justify serious
criticism, as wrongdoing not mere solecism, and conformity
with which deserves to be thought essential to the preservation
of social life and to be supported by strong demands and
pressures, even though conformity may be sometimes be
inconvenient to the obligee.
Since it has often been suggested that promissory obligations
are inexplicable in utilitarian terms, it is worth while to point
out, briefly, that the principle mentioned is certainly justifiable
in utilitarian terms, at least in terms of negative utilitarianism.
If there is a general obligation not to cause preventible harm
to others, it must presumably be taken to imply an obligation
not to cause harm by two successive and mutually related acts.
Contemplate the Jones and Macdonald case in that light:
Jones could perhaps have got to safety if Macdonald had not
thrown down the rope, and if Jones had not delayed himself in
waiting to catch it. If Jones is now killed, by drowning because
Macdonald lets go before he starts to climb, or by falling if
hlacdonald lets go later, his death will be a direct consequence
of his having taken Macdonald's first act of lowering the rope
down the cliff as Macdonald either intended him to take it or
knew he would take it. By the two acts: first, of inducing
reliance intentionally or knowingly; second, of letting go the
rope; Macdonald can do Jones harm which he would not have
done by any one act. Jones's death by Macdonald's second act
will only become possible if the first act has the consequences
in relation to Jones's action which it was intended to have or
thought likely to produce. Mutatis mutandis the act of breaking
a promise causes one's earlier act of promising to have harmful
consequences which were foreseeable at the moment the
promise was made, given that the promise was intended to be
taken by the promisee as giving him grounds to act in reliance
upon the promisor's promised performance.
Even if it so happened that when Jones was half-way up the
rope, Macdonald spotted someone else whose life was in danger,
and whom he could also rescue, the general negative utilitarian
principle which I have adumbrated sho\vs that it would be
wrong for him to let go the rope and rush off to save the other
life. Though the good he can do Jones and the other is at this
moment equal, Jones will suffer harm as a direct and readily
foreseeable result of Macdonald's series of acts; but the harm
which the other will suffer is not a morally significant conse-
quence of Macdonald's actions. I t is not morally significant,
because he did not, when he started to rescue Jones, know of the
other's plight; so he could not foresee that his sodoingwould pre-
vent himgoing to the rescueofanother in equalorgreaterdanger.
So, too, in a situation of promising. You invite me to dinner
with you on Tuesday night, explaining that your nephew, who
wishes to meet me, will be asked also. I accept, and you then
7O I-NEII. MACCORMICK

make arrangements with your nephew for him to come too.

If I subsequently decide to go somewhere else on Tuesday
night, the inconvenience and disappointment which you and
your nephew suffer is directly caused by my two acts, of accept-
ing the invitation in the first place and of subsequently failing
or refusing to come. If, conversely, you make an arrangement
with your nephew for dinner on Tuesday in the hope that I
can come, then telephone inviting me, whereupon I refuse the
invitation, both of you may be disappointed and incon-
venienced. But I cannot here be blamed for disappointing an
expectation which was brought about by no act of mine.
I t is thus reasonable to claim that a negative utilitarian
principle can justify the principle that we must not so act as to
disappoint the reliance of others when we have intentionally
or knowingly induced them to rely upon us. What has been
assumed rather than demonstrated so far is that the case of
promising itself falls under that second principle of not dis-
appointing reliance. But that is easy enough to show. I propose
the following partial definition :
The utterance 'I shall +" addressed by S to A is n
+
promise to only if:
either ( I ) S intended A to take it as signifying that S
+
intended to and intended A to rely upon S 4-ing, by
making plans or taking actions in the expectation that
+
S will
or ( 2 ) S knew or thought it likely that A would take it as
signifying that S intended A to rely upon S+-ing, in the
sense mentioned in ( I ) .
In order that the promise should be successfully made, it is
essential that A should realize that S intends him so to take his
utterance 'I shall 4'.I t is not essential to the constitution of
the promise that A should subsequently take any irrevocable
action in reliance upon it. So soon as an utterance is niade by
a speaker intending knowing or thinking it likely that his
addressee will understand it in the manner mentioned, and
provided the speaker does so understand it, a promise has been
made. The extent to which the addressee then acts, by making
plans or proceeding to overt acts or omissions, relates not to
the existence of the promise but to the seriousness of breaches
of it. (It is, for example, obviously worse to cancel a dinner
 VOLUNTARY OB1,IGATIONS AND NORMATIVE POWERS 71

engagement when the meal is already half-cooked, than half

a n hour after accepting the invitation and three weeks in
advance of the date fixed.) So the definition given looks only
to the speech act and the intention with which it is uttered;
if the definition is correct, any such speech act clearly falls
within the ambit of the principle against disappointing reliance
wilfully or knowingly induced. Hence the obligation to keep
promises is derivable from that more general principle.
T h e definition seems to me to be preferable to Searle's
suggestion that that promising essentially requires an intention
to incur an obligation, and an intention to be taken as under-
taking an obligation. T h e trouble with Searle's approach is
that it is either incomplete or wrong. Let me show why. How
could we make sense of the doctrine that a promisor essentially
intends to incur an obligation? Only by supposing a logically
prior rule or principle under \\hich some obligation arises,
which rule the prornisor intends to invoke. What is the rule?
Presumably, it is the rule or principle that it is obligatory to
keep promises. Thus far, the definition is incomplete, becausc
the concept 'promise' in this rule has yet to be explained.
O n e possible way of going on to explain it is wrong: supposc
the rule to be, 'Whoever says "I promise to +" with the intell-
tion to be taken as invoking this rule, has an obligation to +'.
This would keep us out of the vicious circle, but at the expense
of leading us straight into nonsense. For it hould follow that
there might be numerous occasions on which S says to A ,
'I promise to 4' afterwards claiming that he had no intention
to be taken as invoking that or any other rule, whereupon it
would follow that he had neither made a promise nor incurred
an obligation. And that will not do. We must not "provide
Hippolytus with a let-out, the bigamist with an excuse for his
'I do' and the welsher for his 'I bet'."8 Moreover, it would
make it very hard to understand how any utterance not using
the explicit performative could possibly count as a promise.
T h e thesis which I have been advancing does indeed enable
us to give a particularly clear account of the function of
explicit performatives such as 'I promise . . . '. Only if I intend
you to take it that I intend you to rely on me, or know etc.
that you will take it so, do I promise when I say 'I shall I$'.
Usually the context makes it clear enough whether I have such
72 I-NEIL MACCORMICK

intention or knowledge. But sometimes doubt may arise.

Because the word 'promise' has the meaning it has, someone
who says 'I promise to 4' necessarily settles all doubt, if he
knows the language. We need not trouble about his intentions;
knowing the language, he must know that his addressee will
suppose that he intends him to rely upon the act being done.
He is therefore 'estopped' from denying that he promised,
and whether or not he has the least intention actually to do +,
and whether or not he regards it as obligatory to 4, we need
have no doubt that he has in fact promised to +, and that now
he has indeed a duty to 4, whatever morally eccentric views he
nlay hold.
I t would be silly to deny, and I do not deny, that conscious-
ness of the obligatoriness of promise keeping, in a society in
which the obligation is clearly recognised, in itself reinforces
the practice of promising in a variety of ways. A promisor who
makes a promise in the knowledge that his associates will take
a very serious view of his misconduct if he does not keep it, is
thereby the more likely to weigh seriously whether or not to
make such and such a promise at all in the first place, and
whether or not subsequently to break it. The promisee is more
likely to take the promise seriously, and go ahead to lay plans
and so on in the unquestioning assumption that the promise
will be kept. The more likely it is that the promisee will
seriously rely on the promise being kept, the more important
it is that the promisor should keep it. Thus observance of the
rule reinforces the practice which in turn reinforces the rule,
in an entirely virtuous circle. In such circumstances it is indeed
likely to be the case that people who assure others that they
will 4, with the intention that those others take it that they are
to rely upon that performance, do in standard cases have a
conscious intention to incur an obligation. In stable societies,
people who promise probably do consciously intend to incur
obligations by promising. (The difference between Searle's
position and mine is that the last sentence is necessarily
tautologous on his view; on mine it is synthetic, which I take
to be a merit.) It does not follow that we need stipulate 'inten-
tion to incur obligation' as a necessary condition of promising;
'intention to induce reliance' is surely better, if only because
it provides us with intelligible reasons why promises should
 VOLUNTARY OBLIGATIONS AND NORMATIVE POWERS 73

generate obligations, and therefore why it may indeed be the

case that promisors do in fact sometimes consciously intend
to create them.
Thus the concept 'promising' can be explained as a speech
act in terms simply of an intention on the speaker's part that
his addressee should take the utterance as one upon whose
performance he is intended to rely. If a 'promise' in that sense
is made, whatever may be the form of words used, whether or
not the explicit performative is used, then if the promisee has
properly understood the utterance, and if he does proceed to
take actions or make plans in reliance upon the promisor's
keeping his word, the promisor plainly does wrong should he
fail to perform as promised. That is to say, the promisor has an
obligation to keep his word if it has been taken seriously by the
promisee. That he has this obligation, and that we should have
a rule to the effect that promises must be kept, is explicable in
general terms of the wrongfulness of disappointing people
whom we have wilfully induced to take action on the faith of
assurances of ours about present facts or future acts. And that
general principle is itself further explicable and justifiable in
terms of a negative utilitarian principle. We do not have to
presuppose any socially recognized 'secondary rule' about the
making of promises as a necessary condition of the intelligi-
bility of promissory obligations, or indeed 'voluntary obliga-
tions' in general.
In this respect, therefore, promising is radically distinguish-
able from such concepts as 'marriage' or 'divorce' or 'convey-
ance' or 'sale' which are inexplicable save by reference to rules
which define the status of marriage, and provide for its incep-
tion and dissolution, or to rules which define the concept of
property and provide for its investment and divestment; that
is, for the 'transfer' of 'property in things'. For this reason a
parallel distinction exists between the performative utterances
essential to promising and to marrying, divorcing (in some
systems), selling, and conveying.

Turning now to the law of voluntary obligations, we should

recall a point made rather earlier, that questions of intention
74 I-NEIL MACCORMICK

are frequently difficult to settle, so that it may be a difficult

question to answer whether a man who said that he would do
x meant to be taken as promising to do x , or merely as pre-
dicting that he would or stating an intention to do it, or what-
ever. Such uncertainty is an unavoidable inconvenience of
social life, and perhaps not much can be or need be done
about it-after all, someone to whom such an equivocal utter-
ance is made, can there and then ask 'Is that a promise?' and
the speaker's prompt reply one way or the other is necessarily
conclusive. In the law, however, certainty one way or another
is highly desirable if the machinery of state coercion is to be
used in enforcing voluntary obligations. I t is in this light
interesting to recall how in the earlier Roman law, no promise
as such was enforceable unless couched in a formal answer to
a formal question: 'Spondesne . . . ?' 'Spondeo.' This require-
ment only died out when it had become universal practice to
reduce stipulations to writing; the fondness for writing has
descended to modern civilian systems; Scots law requires proof
by writ or oath of all gratuitous promises, and of most private
non-commercial transactions, and French law likewise requires
all private contracts of a non-commercial type to be proven
by an appropriate written instrument, or a t least to be proven
in part by writing, in all save a few exceptional cases where it
is thought unconscionable for the promisor to plead the want
of writing as a defence. I n English law, the only way in which
a gratuitous promise may be made enforceable at law is by
couching it in a solemn written instrument known as a 'deed'
which takes effect when signed, sealed, and delivered. In cases
of contracts properly so called in which consideration 'moves
from' all the parties, it is interesting to observe that the ques-
tions whether a promise was made, and what promise was
made, are settled by what lawyers call an 'objective test'.
After proof of what the parties actually said and did, the court
concerns itself only with asking what 'a reasonable man'
would in that light have assumed to be the intention of the
parties; not what it actually was.
Another difficulty which we noticed earlier was the difficulty
in which a promisor may find himself if he does not know that
the promisee has properly understood him, nor whether or
not he wishes to 'take him up' on the promise. Again one way
 VOLUNTARY OBLIGATIONS AND NORMATIVE POWERS 75
in which we can tackle that problem in ordinary life is by
asking the promisee, and again in the nature of the case his
answer is conclusive. To be on the safe side, one can even
expressly make it a condition of one's promise that it must be
expressly accepted. The law can clarify such problems, either
by ruling that the matter is, so to say, at the promisor's risk
unless he expressly makes acceptance a condition of his being
bound; Lord Stair favoured this rule, and it found its way into
Scots law. I n French law on the other hand, acceptance is, as
a requirement of law, essential to the completion of an3, contract,
even a gratuitous one. In all systems, of course, bilateral
contracts, which essentially consist in an exchange of promises,
necessarily require some form of acceptance. I n such cases,
further problems can arise as to the time of completion of the
contract, or to the place-for example when dealings are by
post. In this, it may be convenient that the law should lay
down a rebuttable rule that acceptance takes place, for
example, at the moment of posting of a letter of acceptance.
I t is then up to the parties to prove that a different condition
was stipulated by the offeror, or the presumption contained in
the rebuttable rule stands.
One way to represent legal requirements of the kind which
I have discussed is as follows: they are designed to provide
safeguards relating to the legal enforcement of promissory
obligations. Certain possible areas of doubt and difficulty
relating to whether or not a promise was genuinely made, or
genuinely 'taken up' by the promisee, and to when it was 'taken
up', can be clarified by stipulating certain procedural and
evidentiary rules as conditions, not of the existence of a promise
as a matter of social fact, but of the enforceability of promises
at law. This implies that the concepts of 'promises' or 'agree-
ments' are pre-legal and extra-legal. The law adopts them and
enforces the obligations arising from them, but only if certain
clearly established conditions are fulfilled with regard to the
mode of making promises. Whether or not such conditions are
said by lawyers to go to the 'constitution' or only to the 'proof'
of voluntary obligations is comparatively unimportant. The
function which they essentially fulfil is to secure, as far as may
be, that promises are not enforced unless it is clear that the
'promisor' really intended to be taken seriously, and had
76 I-NEIL MACCORMICK

grounds to suppose that the promisee really took him seriously.

They are rules laying down procedural prerequisites for the
enforceability of voluntary obligations, not 'parts' of power
conferring laws which are constitutive of the concept 'volun-
tary obligation'; this way of representing the matter seeks to
represent the law of voluntary obligations as being a special
case of, rather than a radically different set of norms from, the
social practice of promising, and the social norm as to the
obligatoriness of keeping promises.
But that mode of explaining the law, useful though it is for
some purposes, depends crucially on an unexplained notion
of the 'legal enforceability' of promises and agreements. And
so soon as we try to explicate that concept we find that refer-
ence to power-conferring rules is irreducibly essential in any
proper explanation of the law of voluntary obligations. Let us
recall that the concept of an 'enforceable' contract or promise,
is not of a promise which Courts may in their discretion enforce
but rather of one which they must enforce, which they have
an obligation virtute oflcii to enforce, whether by injunction,
specific performance, or award of damages. Certainly, this
obligation is itself conditional upon one or other party's choos-
ing to litigate, and proving an actual or anticipated breach of
the other's obligation. But if those conditions are fulfilled, and
if the other party cannot prove some excusing or vitiating
circumstance, such as mistake, misrepresentation, frustration or
illegality, the Court has no choice but to enforce a properly
made contract. I t may, of course, be said that if trial judges
and appellate judges choose to flout the law, no one can stop
them; but that is as irrelevant to the existence of an obligation
on their part to enforce the law as is the manifest truth that
murders are committed to the equally plain truth that citizens
have an obligation in law not to murder each other.
Thus the rule, 'A person may make a gratuitous promise
legally enforceable by executing a deed expressing it, that is,
by taking the following steps . . . ' could be restated in the
following terms :
'If any promisor expresses his promise in a deed executed
as follows, it must be enforced by the Courts at the
instance of the promisee.'
Now we may pose the question: is that rule a power-conferring
 VOLUNTARY OBLIGATIONS AND NORMATIVE POWERS 77

law or a law of obligation? Is it a P-law or a D-law? To that

the answer depends essentially upon one's point of view: for
judges, and other officials such as inspectors of taxes it is
essentially a D-law; for private citizens it is essentially a P-law,
since it makes available to them the option of so acting that
Courts will have to enforce their promises, or of refraining
from so acting, neither the act nor the refrainment being in
its own character legally wrongful. Similarly, in the case of
mutually onerous agreements, the parties are free to express
an intention not to create legal relations, whereupon the
Courts may not enforce their agreement, and vice versa (the
presumption being that, apart from such exceptions as domestic
agreements and till recently industrial collective bargains,
onerous agreements are intended 'to create legal relations').
Now it might well be argued that what distinguishes the
legal concept of a contract from the plain man's conception of
a promise is, precisely, that the former is legally enforceable
in the sense just explained. I t is indeed clear that legal enforce-
ability is a concept which necessarily involves reference to a
type of rule which may most appropriately be classified as a
power conferring rule: a rule under which one person may by
a specified act, whose performance and non-performance by
him are equally permissible bring it about that another has an
obligation whose content is partly specified by the former's
act. The existence of such a rule is certainly logically prior to
the concept of an 'enforceable contract'-indeed, whenever it
is the case that one person's act directly or indirectly engenders
an obligation upon another party, we have an instance of an
irreducible power-conferring rule, but only if the act which is
the condition of the actualisation of the other person's obliga-
tion is itself an optional act whose performance and non-
performance are equally permissible. However, given that
their legal enforceability is what distinguishes the sub-class of
legal contracts from the larger class of promises, that conclusion
rather confirms than negates the thesis that promises and
voluntary obligations are in their essential nature explicable
without reference to power-conferring rules.
Perhaps one final reflection upon power-conferring laws
may be permitted. I t has earlier been pointed out that the laws
rendering certain sorts of promise made or evidenced in certain
78 I-NEIL MACCORMICK

ways legally enforceable are laws which present two faces:

from one point of view, they are conditional D-laws, the
condition of the concretisation of the official's duty being the
voluntary performance of an optional act by a citizen or
citizens. From another point of view, the citizen's, they thus
present the character of P-laws. If this observation were capable
of being generalised, it would follow that all P-laws are in
essence merely a sub-class of D-laws. D-laws which impose
obligations upon one person or group conditionally upon
another person or group's voluntarily performing some optional
act of a specified kind, are P-laws. There is meat for further
investigation here, especially in relation to powers to create
powers, and powers to abrogate obligations, but it would
neither surprise nor distress me if this suggestion proved true.
Certainly, all forms of legislative power can be represented in
terms of conditional obligations of courts and other law-
applying organs: "If the legislature [defined] enacts a statute
by undertaking the following acts . . . , which it is free to
perform or not to perform as it chooses, then the Courts
[defined] are obligated to apply the statute thus enacted."
Kelsen may not have been so far wrong after all. But that's
another story.

REFERENCES
Cf. Qureshi v. Qureshi [1g71] I All E.R. 325.
J. Raz, The Concept of a Legal System (Oxford, lgjo), pp. 175-183.
J. Searle, Speech Acts (Cambridge, 1969) passim (esp. at pp. 33-72,
I75- 188).

Id. p. 46.

Cf. Wakeham v. Mackenrie [1g68] 2 All E.R. 783.

The Concept of Law, pp. 84-85. Cf. J. Raz, The Concept of a Legal System,
PP. '47-156.
Or any other utterance expressly or implicitly referring to the future
conduct of S.
Cf.J. L. Austin, How to Do Things with Words (Oxford, 1962), p., 10.
I should like to acknowledge with gratitude helpful advice and criticism
received during the preparation of this paper from Dr. A. J. P. Kenny, Dr.
A. F. Rodger and Mr. L. C. Holborrow. Dr T. D. Campbell's recent Adam
Smith's Science of Morals (London, 1971)'at p. 95 reminds me that the view I
expound about promising resembles, and perhaps supports, that espoused
by Adam Smith (Lectures on Justice, etc., ed. Cannan, Oxford, 1896, pp,
I30f).
 VOLUNTARY OBLIGATIONS AND
NORMATIVE POWERS
Neil MacCormick and Joseph Raz

The main thesis for which I shall argue is that a satisfactory

analysis of voluntary obligations must be in terms of normative
powers. To substantiate this claim it will be necessary to
examine in some detail the concepts of a normative power and
of a power conferring norm (or P-norm). I shall start with an
analysis of legal powers and then proceed to examine the
applicability of the concept outside the law.
I shall refer throughout to "norms" rather than "rules".
Norms, unlike rules, can be particular as well as general. Some
principles as well as rules are norms, but not all the types of
rules and principles are norms. Technical rules, e.g., are not
norms. The sense in which "norms" is being used will be
explained later.

I. Legal Powers
Legal writers, judges and lawyers do not use the term "legal
power" as often as do legal theorists. They speak of the powers
of the police, of courts or ministers, but seldom refer to thc
power of a person to make contracts or wills, etc., where terms
like "competence" or "capacity" are more often used. Legal
theorists have given "legal power" a wider and more technical
meaning, not because they sought to elucidate the ordinary
meaning of the words, but because they saw the need for a
general concept to draw attention to important similarities
between otherwise heterogeneous phenomena. The same must
be true of an investigation of the r61e the concept should play
in general normative theory. T o be sure, the term is not often
employed in ordinary discourse concerning voluntary obliga-
tions. The question is whether we shall gain in clarity and
understanding from its employment to describe certain non-
legal normative phenomcna. I have no doubt that we shall.
80 11-JOSEPH RAZ

Most of the text-book definitions of a legal power are derived

from Hohfeld's famous analysis of the concept (Fundamental
Legal Conceptions, ~ g ~ gthough
), by far the most thorough and
penetrating analysis is Bentham'sl Not wishing to engage in a
comparative study of various definitions of the concept, I shall
take Hohfeld's analysis as my starting point.
"A change in a given legal relation", he says, "may result ( I )
from some superadded fact.. . not under the volitional control of
a human being.. .; or ( 2 ) from some superadded fact or group of
facts which are under the volitional control of one or more
human beings. As regards the second class of cases, the person
(or persons) whose volitional control is paramount may be said
to have the (legal) power to effect the particular change of legal
relations that is involved" (pp. 50-51). Following Hohfeld
most text books define legal power as the ability to effect a
legal change, a change in the legal situation, by a voluntary act.
Such definitions though helpful and in the main correct are not
precise and may mislead. Let me add four points clarifying and
modifying the traditional definition.
I . The relation between tfze act and its legal consequences: A legal
power is normatively, not causally, effective. I t is not the
ability "to cause certain desired legal effects". (Ross, On Law
andJustice, 1958, p. 166.) I may be able to bring it about that
my wife will make a gift of all her property to me, but it is she
not I, who has the power to make a gift of her property. I may
be able to make all the voters of my ward vote for me in every
local election, but I have the power to cast just one vote in each
election. I have a legal power only if it is my act which is
recognized by law as effecting a legal change.
2. The power-exercising act: Hohfeld's and most of the other
explanations of a legal power are too sweeping. Not every
voluntary act which effects a legal change is an exercise of a
legal power. Every offence or civil wrong changes the legal
situation. I t usually makes, e.g., the offender liable to a
sanction, and imposes on the wrong-doer an obligation to com-
pensate. There are many other acts which effect a legal change
but are not the exercise of legal powers. By changing my
residence from one town to another, or from one country to
another, I change my rights and duties, but I do not have a
legal power to effect these changes by such action. Salmond on
Jurisprudence ( I 2th ed. edited by P. J. Fitzgerald, 1966, p. 2 2 0 )
seems to meet the problem by defining power "as ability con-
ferred upon a person by the law to alter, by his own will
directed at that end," the legal situation. Unfortunately the added
condition makes his definition both too wide and too restricted.
I t does not avoid the objection I raised, for people have been
known to commit an offence in order to be punished, or to
change residence in order to change their legal position. O n
the other hand it is possible to exercise power and to make a
contract or some other legal transaction with no intention to do
so, if one does not correctly appreciate the legal consequences
of one's action.
The solution of the difficulty lies not in the intention wit11
which power-exercising acts are performed but in the reasons
for which they are recognized as effecting a legal change. An
action is the exercise of a legal power only if one of the law's
reasons for acknowledging that it effects a legal change is
that it is of a type such that it is reasonable to expect that actions
of that type will, if they are recognized to have certain legal
consequences, standardly be performed only if the person
concerned wants to secure these legal consequences.
This is a most important feature of legal powers, which is all
too often neglected. It explains why they are exercised either by
special formal and ceremonial acts as in making a deed or
getting married, or by ordinary actions whose legal consequen-
ces approximate to their non-legal and obvious consequences,
as in making a contract. I t also explains why most legal powers
are exercised by acts with only negligible non-normative
consequences, like signing, so that there are few reasons for or
against doing them apart from their legal or other normative
consequences.
3. The Nature of the legal consequences: Hohfeld thought that
every legal change is a change in someone's rights and duties.
However, a legal change may consist in change of status or in
the creation or liquidation of a legal person, and Hohfeld was
mistaken in thinking that these are no more than bundles of
rights and duties. A legal change should therefore be defined as
a change in the existence, content or application of a law or
legally recognized norm. T h e distinction between laws and
other legally recognized norms will be discussed below.
82 11-JOSEPH RAZ

Powers of legislationandof making contracts are, respectively,

the paradigmatic examples of powers determining the content
and existence of laws and of legally recognized norms. By their
exercise a law is created or is modified or repealed, or a contract
is created or modified or rescinded. Such powers I shall call
norm-creating powers. (Not every law is a norm, but those
which are not are logically related to legal norms and therefore
there is no harm in calling powers determining their content
and existence "norm-creating" powers.) The powers to appoint
a judge or transfer ownership do not determine the content of
laws or of legally recognized norms. They govern the applica-
tion of such norms. T h e rights and duties of a judge or of a
property owner are determined by laws, the validity and con-
tent of which are unaffected by the appointment or the transfer
of title. T h e latter merely invest a person with these rights and
duties and, in the case of change of ownership, divest some
other person of them. I shall call such powers regulative
powers.
4. Powers and Rights: I t is commonly thought that powers are
a species of rights. This is a mistake. Though powers are essen-
tial to the explanation of rights, they are not in themselves rights.
I t is possible to speak of a power-right only if one is a t liberty
to use or not to use the power at ,wille3But often one has a
power which one is obliged either to use or not to use, with-
out having any choice in the matter. In Israel, Jews and
Muslims have power to contract polygamous marriages. If
they perform the marriage ceremony their polygamous marriage
is valid. But they have no right to contract such marriages
and indeed they commit a criminal offence if they do so. In
many countries a thief has the power, but not the right, to sell
stolen property in the open market. Under the law of agency,
an agent often has powers which he is under a duty to the
principal not to exercise, or to exercise only under certain
conditions. Public officials often have mandatory powers which
they are under a legal obligation to use only when certain
conditions obtain and, therefore, their powers do not amount
to rights. Company directors are often in a similar position.

11. Legal Powers and Legal Norms

Let us agree, then, that an act is the exercise of a legal p o ~ e r
only if it is recognized in law as effecting a legal change and if it
is so recognized because, among other things, it is an action of a
type which it is reasonable to expect to be performed for the
most part only when the person concerned wants to bring about
the legal change. A person has a legal power to produce a
certain legal change if and only if when he behaves in a certain
way he produces this legal change by exercising a legal
power. Let us now turn to the problem of the relations between
legal powers and legal norms. Clearly legal powers can be
conferred by legal norms and they can be powers the exercise
of which affects legal norms. But must all legal powers be
related in both these ways to legal norms? Two questions
should be examined :
( I ) Are all legal powers necessarily conferred by legal
norms ?
( 2 ) Does the exercise of every legal power determine
either the existence or application of legal norms?
Mr. MacCormick presupposes a positive answer to the first
question: If there is a certain power there must be a rule
conferring it. I t is certainly true that most legal powers are
conferred by laws. But it is interesting to notice that the analysis
of the concept of a legal power suggested above did not include
any reference to power-conferring laws. There are, no doubt,
strong policy reasons to minimize legal recognition of legal
powers not conferred by laws, but there is nothing in the
concept of a legal power to necessitate an affirmative answer to
the first question.
Let us first note as, I hope, an uncontroversial fact that a
legal system may recognize powers which are not conferred by
the laws of the system but by some other laws. Private interna-
tional law, e.g., includes laws whose purpose is to define the
conditions under which such powers are to be recognized.
Similarly, powers granted by social norms may be recognized by
a certain legal system. A Muslim country, e.g., may recognize
the validity of Taleq divorce performed in a foreign country
where such a divorce is not recognized and may even be
illegal. (Whether we should call such powers "legal powers" or
"legally recognized powers" need not concerrl us here.) I wish,
however, to argue for a stronger thesis. Namely, that legal
84 II-JOSEPH RAZ

powers may exist even if not conferred by any norm, whether

legal or non-legal. I t is worth our while to examine two cases of
legal powers not conferred by norms in order to see the reasons
for admitting the existence of such powers.
Austin's sovereign obviously has powers of legislation, but
these are not powers conferred by a law. No law can apply to
the sovereign. The fundamental reason for this was explained by
M a r k b ~The
. ~ powers of an Austinian sovereign are illimitable.
All his general commands are law. His particular commands are
not law, not because they are not binding, but because Austin
erroneously thought that laws are necessarily general. There
can be no law purporting to confer power on the sovereign,
not because a person cannot grant powers to himself-this
is possible and does occasionally happen-but because such
law would not guide the sovereign's behaviour, and would be
empty and pointless. The sovereign has these powers regardless
of any law purporting to grant them to him.
A second example. The first Israeli law, embodied in the
Israeli declaration of independence, was made by a body called
The People's Council. In it this council granted to itself
unlimited legislative powers and changed its name to the
"Provisional Council of State". The Provisional Council of
State had legislative powers derived first from this and later
from other laws. The People's Council on the other hand was
never granted powers by any law. Nevertheless it had and did
exercise legislative powers by enacting the first law. The reason
why its powers were not granted by law is obvious. There was
no authority which could enact such a P-law at the time, and
there was no point in enacting a retroactive authorizing law
later. I t could make no difference. It could neither guide the
Council's actions nor affect their legality. Some might argue that
in such cases it is the rule of recognition which, on Hart's
theory at any rate, confers these powers. But, as I have argued
at length elsewhere's the rule of recognition is a law of a dif-
ferent kind. It imposes duties and cannot be interpreted as
conferring powers. It is arguable whether or not there was in
Palestine at the time a social norm which conferred powers on
ihe People's Council. I doubt whether there was such a norm.
The point is that the existence of such a norm does not affect
the issue. It is possible for a legal power to be recognized in some
such circumstances even if it is not conferred by a pre-existing
social norm.
T o conclude. A legal power may exist even if it was not
conferred by any norm at all. Whenever the exercise of a power
cannot be guided by a norm we should regard it as not con-
ferred by one.
What about the second question posed above? Does the
exercise of every legal power determine the existence or
application of a legal norm? I shall use the expression "a power
affects a norm" to refer both to cases where a norm is created,
abrogated or modified by the exercise of a power and to cases
where the exercise of a power regulates the application of a
norm. Is it necessary that all the powers recognized by a legal
system S, affect in one of these ways the laws of S,? I t seems to
me that the answer is a qualified Yes.
The reason for the affirmative answer lies in the concept of
legal recognition. A power is legal only if it is recognized by a
legal system. As we have seen, to be recognized in S, does not
mean to be granted by a law of S,. I t means that the courts of
S, have a duty to apply the norm affected by the power
subject to the changes in its content or application made by the
use of the power. Thus, the notion of a legally recognized power
is parasitic on that of a legally recognized norm. A power is
recognized only if the norms it affects are recognized as
affected by it. A norm is recognized only if the courts are under a
duty to apply it. I t follows that if a power affects a recognized
norm it affects also the law imposing a duty to apply that
norm. Consequently, to be a legally recognized power entails
being a power regulating laws of the system which determine
which norms the courts ought to apply. In other words, every
legally recognized power is a regulative power. The legislative
powers of Parliament, e.g., are also regulative powers. They
regulate the duty of the courts to apply parliamentary legis-
lation. Similarly the power to make a gift is regulative both bj.
regulating the application of property laws and by regulating
the application of the duty of the judges to apply these laws.
One should distinguish between the direct and the indirect
character of legal powers. All legal powers are indirectly
regulative for if they are recognized in law they regulate the
application of laws imposing duties on courts to apply certain
86 11-JOSEPH RAZ

norms. But they have this indirect character only because they
affect norms directly either by making, changing or repealing
them or by regulating their application. To the question
whether every power recognized in S, affects a law of S,, the
answer is affirmative, in so far as its indirect character is con-
cerned. T o be recognized in S, it must indirectly regulate laws
of S, which impose duties on courts to apply certain
norms.
But now we are faced with another question: Are the norms
directly affected by the power also always laws of S,? I agree
with the implication in Mr. MacCormick's paper that they need
not be. As I have argued e l s e ~ h e r ethe
, ~ view that every norm
which it is the duty of the courts to enforce is part of the system
is mistaken. I t is part of the function of'a legal system to ensure
the enforcement of some norms not belonging to it. Con-
sequently, legal systems recognize powers which directly
affect norms which do not belong to these systems.
Powers to directly affect norms which are not part of the
system S , may be recognized in S, for two different types of
reason. They may be recognized in S, because they are recog-
nized by some other legal system or by social conventions or
practices which S, wishes to respect and enforce. When this is
the reason for recognizing these powers one can talk of deriva-
tive recognition.' But some powers not directly affecting the
laws of S, may be recognized in S, for different reasons. The
law may simply intend to create ways in which individuals may
achieve certain results even though these ways are not recog-
nized by some other norms. The powers to make and amend the
regulations of private companies are an example of this kind of
recognition. They are normative powers because they affect the
company's regulations, which are norms. They are recognized
in law since they indirectly affect the duties of the courts to
enforce these regulations. But a company's regulations are not
part of the legal system, nor are they recognized in law because
they are conventional social rules. These powers are powers
directly affecting norms which are not laws of the system but
they are given original rather than derivative recognition.
Powers which enjoy only derivative recognition in S, are not
usually called legal powers in S, though they may properly be
called powers recognized in S,. All the powers which enjoy
original recognition in S, are called legal powers even if they
are not powers directly affecting the laws of S,.
Concerning the legal status of powers to make contracts, Mr.
MacCormick's view- formulated in this terminology-is that
they are not really legal powers but, since they are derivatively
recognized, merely powers recognized in law. They are recog-
nized only because they are recognized according to sound
moral principles. I agree with him that they are not powers
directly affecting the laws of the system in which they are
recognized. Contracts are not laws; they are merely legally
recognized norms. But I disagree with his contention that they
are recognized derivatively. Though the fact that many con-
tracts are morally binding, and more important still the fact
that they are socially accepted as binding, are part of the
reasons for their legal recognition, ultimately they are recognized
in law because, regardless of moral and social norms, they are
considered as a desirable means to achieve some legitimate
ends. I t may be that the facts which contribute to the argument
for regarding promises as morally binding are also among the
facts which make it desirable to make contracts binding in laic.
But it is these facts which count and not their results in morality.

111. Power-Conferring laws

The concept of a legally recognized power is logically connected
with that of a norm because every legally recognized power is a
power affecting one or more legally recognized norms, not
because all such powers are conferred by laws. But most legal
powers are conferred by laws and something must be said
about the nature of power-conferring laws (P-laws).What is the
logical character of a P-law? Are all P-laws also duty-imposing
(D-laws) ? The answers to these questions depend on the
principles of the individuation of laws, for they determine what
counts as one complete law. Legal practitioners are not nor-
mally concerned with the doctrine of individuation and ordinary
legal discourse embodies only a very vague notion of what is one
law. The problem is of considerable importance in juris-
prudence, however, since it underlies any attempt to classify
laws into distinct logical types; and this is a major juris-
prudential concern. Let me explain briefly the nature of the
problem and the kind of considerations involved
88 11-JOSEPH RAZ

I t is obvious that not every true legal statement describes one

complete law. Of the three following statements
(I) Every male has a duty not to trespass;
( 2 ) Every female has a duty not to trespass;
(3) Every person has a duty not to trespass;
only the third describes a law while the others are deductions from
it. I t might be thought that the formulation of the law in the stat-
ute or judgment which created it determines which statement
describes one complete law and which does not. But this is a
mistake as is explained by Bentham: "What is a law? What
the parts of a law? The subject of these questions, it is to be
observed, is the logical, the ideal, the intellectual whole, not the
physical one: the law, and not the statute. An inquiry, directed
to the latter sort of object, could neither admit of difficulty nor
afford instruction. I n this sense. . . So much as was embraced
by one and the same act of authentication, so much as received
the touch of the sceptre at one stroke, is one law.. . and nothing
more. A statute of George I1 made to substitute an or instead of
an and in a former statute is a complete law; a statute containing
an entire body of laws, perfect in all its parts, would not be
more so. By the word law then . . . is meant that ideal object, of
which the part, the whole, or the multiple, or an assemblage
of parts, wholes and multiples mixed together, is exhibited by
a ~ t a t u t e " .In
~ a word, we want the concept of a law to be
determined by rational principles not by the vagaries of legis-
lative technique or judicial style.
An acceptable doctrine of individuation has to satisfy in the
best possible way a set of partly conflicting desiderata. Three of
these are important for the present discussions: ( I ) The laws
individuated by the principles of individuation should be
relatively simple. (2) I t is desirable that every generic act that
is guided by law should be the core of a separate law. (3) T h e
laws individuated by the principles of individuation should, as
far as possible, display important connexions between various
parts of a legal system. Consider the application of these
desiderata to the following legal statements :
( I ) A person has power to make a gratuitous promise
binding in law by executing a Deed.
(2) The promisee of a Deed has a power to claim its
performance.
 (3) The courts ought to enforce the performance of
duties at the instance of those who have a claim to
their performance.
(4) any promisor expresses his promise in a Deed, it
If
must be enforced by the courts at the instance
of the promisee.
This is a much simplified and somewhat distorted statement of
the law, but it will do for our purposes. Statement (4) is
entailed by statements ( I ) - ( 3 ) . I t entails statements ( I ) and
( 2 ) but not (3), which ascribes a more general duty to the
courts than the duty to enforce deeds mentioned in statement
(4). Mr. MacCormick claims that statement (4) describes one
complete law and, presumably, he regards the other statements
as conclusions of law derived from it and (in the case of (3))
other laws. From this premiss he rightly concludes that the law
described in (4) can be seen as both conferring powers and
imposing duties.
That some laws can be both power-conferring and duty-
imposing is not a very startling conclusion. Every statement of
the form (5), for example, entails and is entailed by a state-
ment of the form (6) :
(5) " X has power to enact D-laws applying to members
of population y."
(6) "Members o f y ought to obey the enactments of
x."
Every statement ascribing powers to enact D-laws entails and
is entailed by a statement ascribing a duty to obey. If one
member of such a pair of statements describes one complete
law so does the other member of the pair, and they both
describe one and the same law. Such laws both confer powers
and impose duties. It is, however, important to remember that
most laws conferring norm-creating powers authorize the
enactment of P-laws as well as of D-laws. When the same rules
govern the exercise of these powers we usually have sufficient
reason to regard them as conferred by one law and such a law
cannot be completely described without distortion, by a
statement ascribing a duty to obey. Consequently, most laws
conferring norm-creating powers can only be properly de-
scribed as P-laws, even though their existence entails as a
conclusion of law that certain persons have a duty to obey
go 11-JOSEPH RAZ

D-laws enacted by the exercise of these powers. Furthermore,

even when laws confer only powers to enact D-laws we may still
have reasons for regarding them as primarily power-con-
ferring laws and to regard statements of the form (5) as their
proper description and statements of the form (6) as conclusions
of law from them. This interpretation fits the ordinary way of
talking about them better, and the laws can be said to guide the
behaviour of the legislators more directly than that of the
ordinary citizen, which is affected only if and after the legis-
lator has exercised his legislative powers. But the arguments for
this view are by no means overwhelming, nor do they always
apply outside the law. We certainly refer more often to the
child's duty to obey his parents then to the parents' power to
command.
Our judgments concerning the proper description of a
norm depend in such cases on pragmatic considerations which
may vary from one context to another. In any case, they do not
affect the conclusion that laws conferring powers to enact
D-laws impose duties as well.
But Mr. MacCormick suggests not only that some laws may
both confer powers and impose duties, but that all the laws
which confer powers impose duties as well. In this he is
following Bentham and even more closely, as he himself said,
Kelsen. For Mr. MacCormick is tempted by the idea that all
the laws relating to voluntary obligations are directed to the
courts and instruct them to perform certain acts if certain
conditions obtain. Let me admit straight away that if Mr.
MacCormick is right in thinking that statement (4) des-
cribes one complete norm then he is right in suggesting that
P-laws are merely a subclass of D-1aws.For Bentham demons-
trated long ago that the complete content of a legal system
can be described by statements each one of which asserts that a
duty exists if certain conditions are fulfilled, using the same
technique by which (4) is derived from ( I ) - ( 3 ) . Furthermore,
since it is the duty of the courts to enforce all legal obligations,
provided that certain conditions are satisfied, it is possible to
describe the complete content of a legal system by a set
of statements every one of which describes an obligation
imposed on the courts. The possibility of providing such
descriptions is not in doubt. The only problem is whether each
 VOLUNTARY OBLIGATIONS AND NORMATIVE POWERS 91
statement in the set really describes one complete law or "an
assemblage of parts wholes and multiples mixed together",
to use Bentham's felicitous expression.
T h a t the mere possibility of providing a complete de-
scription of a legal system by means of such statements does not
entail that each of them describes one complete law is a
conclusion forced on us by the fact that not every true legal
statement describes one complete law. T o settle the question we
need to resort to the doctrine of individuation. And the
fundamental problem facing us is crystallized in our example of
statements ( I ) to (4). If statement (4) describes one complete
law then M r MacCormick and Kelsen are right, but if state-
ments ( I ) to (3) each describes one complete law then they
are wrong and we must conclude that power-conferring laws
are a distinct type of law.
There can be no doubt that according to the three require-
ments of a doctrine of individuation mentioned above every one
of the statements (I)-(3)describes a law whereas statement (4)
describes two laws and part of the content of a third. Statements
( I ) - ( 3 ) are simpler than (4). They each refer to one act guided
by law, whereas statement (4) refers to three such acts (the
act of the promisor making the Deed, the act of the promisee
lodging a claim against the promisor who did not keep his
promise and the action of the court in enforcing the Deed).
Statement (3) has the further advantage of satisfying the third
requirement better. Since it refers to the court's duty to enforce
many other types of duties apart from thosearising out ofDeeds,
it draws attention to the fact that it is the function of courts to
enforce not only the laws regulating Deeds but many other laws
as well at the instance of a person having a claim that the duty
shall be performed. Statement (3) thereby directs our attention
to a n important general function of courts and to a special legal
technique of making the enforcement of a whole class of duties
subject to the will of certain individuals.
I t is true of course, that the fact that according to these
requirements of individuation statement (4) does not describe
one complete law may be a n argument against accepting these
requirements. But this is so only if there are good reasons for
thinking that statement (4) does describe one complete law. Mr.
MacCormick provides no such arguments and he fails to answer
g2 11-JOSEPI 1 RAZ

the criticism of Bentham's and Kelsen's theories made by Prof.

Hart and myself.1° Nor does he furnish any other grounds to
question the soundness of the requirements. The justification of
the requirements seems to me to be very obvious. The main
purpose of dividing the enormous amount of legal material
constituting one legal system into laws is to create simple units
for easy reference. And it is desirable to direct our attention to
acts guided by law by making each one of them the subject
of a separate law. Our understanding of the working of the
law is further promoted if we individuate laws in a way which
draws attention to important connexions between groups of
laws.
Power-conferring laws form, therefore, a distinct type of
law which is different from duty-imposing laws.

IV. Normative powers

Can the concept of a normative power be usefully applied out-
side the law? There can be little doubt that it can. Legally
recognized powers are powers to affect legally recognized
norms. To the extent that one can talk of non-legal norms which
can be affected in the appropriate way by human action one
can also talk of normative powers not recognized in law. There
is indeed little difficulty in applying the concept to a discussion
of the rules and principles of voluntary associations or to rules
and principles which have become the practices of a community.
Nor does Mr. MacCormick find fault with the use of the concept
of a normative power in such contexts. He does, however,
object to its application to norms which are recognized neither
by an existing legal system nor by an existing association and
which are not practices of a community. The problem we face,
therefore, concerns the possibility and fruitfiilness of giving a
general account of the concept of a normative power which
makes it applicable in a general theory of norms, regardless of
whether these are the actual practices of any community,
institution or organization. Let us consider first the feasibility of
such an account and defer to the next section consideration of
its fruitfulness.
There are two ways of challenging the feasibility of such an
account of normative powers. I t might be claimed that the
concept of a norm can only be used to refer to actual practices of
 VOLUNTARY OBLIGATIONS AND NORMATIVE POWERS 93
groups or institutions. Alternatively, admitting the applicability
of the notion of a norm in other contexts as well (e.g.,to moral or
prudential reasoning in general), one may claim that norms
which are not practices cannot be affected by human acts in a
way which allows one to regard these acts as the exercise of
normative powers. I have tried elsewhere" to meet the first
challenge and will not deal with it here. Instead I shall try to
show how one type of norm, namely mandatory norms, can be
affected by the exercise of powers.
Mandatory norms are norms requiring that an agent
behave in a certain way in certain circumstances. Duty-
imposing norms, both legal and non-legal, are the most import-
ant species of mandatory norms. Given an account of power
to affect mandatory norms, there is no difficulty in generalizing
it to provide a general account of normative powers, i.e., powers
to affect any type of norm be it mandatory, power-conferring or
permissive.
An act is the exercise of a legal power only if it effects a
legal change-i.e., creates, extinguishes or modifies the applica-
tion of a legal norm. I t is obvious that non-legal norms are
similarly affected by human action. If I intentionally slap John
in the face, my action affects the application of the norm pre-
scribing that one should apologize and compensate for in-
tentionally inflicting harm on others and I, therefore, incur a
duty to compensate and apologize.12 If I move to a new house
my action affects the application of the norms obliging me to
certain modes of behaviour towards my neighbours. I con-
sequently incur new obligations to my new neighbours and am
relieved of my obligations to my old neighbours. When I
promise to do A, I create a new norm requiring that I do A, and
consequently I incur an obligation to do A . However, we have
seen that not every act affecting a legal norm is an exercise of a
legal power and it is equally clear that not every act affecting
non-legal norms is an exercise of a non-legal normative
power.
Two conditions must be fulfilled if an act affecting a legal
norm is to be regarded as an exercise of a legal power. First, it
must be normatively and not merely causally effective. Second,
it must be acknowledged in law as normatively effective for
reasons of a special kind. If non-legal norms can be affected by
94 II-JOSEPH RAZ

the exercise of normative powers, then analogous conditions

must be fulfilled. Let us examine each of them in turn.
An act is an exercise of a normative power only if it affects a
norm normatively and not causally. The distinction turns on
the distinction between the results and the consequences of
acts. "The result of an act", explains Dr. Kenny, "is the end
state of the change by which the act is defined. When the
world changes in a certain way there may follow certain other
changes . . . In that case we may say that the second trans-
formation is a consequence of the first and of the act which
brought the first about. The relation between an act and its
result is an intrinsic relation, and that between an act and its
consequences is a causal relation".13
That John wakes up is the consequence of my turning on the
light, but the result of my waking John which I may have done
by turning on the light. An act affects a norm causally if its
consequences regulate the application of a norm. It affects a
norm normatively if the act itself or its result affects the existence
or application of the norm. Raising my hand with the intention
of slapping John in the face and with theconsequence that heis hit,
causally affects the norm concerning intentionally inflicted
harm. Slapping John affects the same norm normatively.
Similarly, bullying somebody to promise to do A with the
consequence that he does make the promise is causally effective
in creating the norm that he ought to do A, but only his act of
promising affects this norm normatively.
As is clear from my examples not every act which norma-
tively affects a norm is an exercise of a normative power.
Intentionally inflicting harm on a person and changing
residence are examples of acts which normatively affect norms
but which are not the exercise of powers. T o be an exercise of a
normative power the act must be recognized as affecting a
norm for reasons of a special sort. Whose reasons and what
kind of reasons are these? I n the case of powers to affect norms
which are the practices of institutions or groups, the answer to
the first question is simple enough: The reasons of the in-
stitution or the group for recognizing the normative effects of
the act determine its character as a power-exercising act.
When dealing with norms which are not practices the relevant
reasons are those because of which the norm is binding, the
 VOLUNTARY OBLIGATIONS A N D NORMATIVE POLVEKS 95
reasons which justify the norm, bccause of which one
should respect and follow the norm. If a norm which can be
affected by human action is to be respected then the reasons
for respecting it reveal also why it ought to be respected as
created and regulated by human action. I t is the nature of
the reasons justifying the norm which determine whether acts
affecting its existence or application are power exercising
acts.
Arguments justifying norms are of different types. Some are
based on reasons establishing the desirablity of performing the
norm-act. These are content-dependent justifications. They
establish that the norm-act is desirable because its performance
secures desirable consequences or contributes to the realization
of some value. For the justification to be complete it has to do
more than that. But it is the fact that the justification is based on
the desirablity of the norm-act which determines that it is a
content-dependent justification. Some justifications are con-
tent-independent. Sometimes there are reasons justifying a
norm which do not bear directly on the desirablity of the norm-
act. The justification of the mandates of authority are of this
nature. They depend not on the desirability of acting as
commanded or advised in the particular instance but on the
fact that the instruction was issued by authority.
Jeremy, John's 'teenage son, is told by his father to return
home before midnight. There may or may not be content-
dependent justification for this command, but in any case if it is
true that parents have authority in such matters over their
children then there is a content-independent justification for
the command. It does not justify the making of the command
but it explains why Jeremy ought to follow it once made. I t is
important to notice that the content-independent argument
justifies the particular norm-John's command that Jeremy
be home bv midnight. The reasons for his general authority may
also be content-independent (God gave parents authority over
their children etc.) or they may be content-dependent (acting
on parents' instructions will, on the whole, have best conse-
quences etc.) But this is a separate question. One should
further notice that the identification of the reasons for the
various norms does not necessarily presuppose that the norms
are followed or believed to be justified either by John or
g6 11-JOSEPH RAZ

Jeremy or by the community at large. I t depends solely on the

availability of the justifying reasons.
Content-independent justification is to be distinguished from
conservative justification. A conservative justification establishes
that even though there may not have been adequate reasons
justifying a norm in the first place, since it is foilowed and has
become a practice it should not be changed, for changing it
will have worse consequences than preserving it. A conservative
justification applies only to norms which are practices. Content-
independent justifications apply both to norms which are
practices and to those which are not.
Given these distinctions and clarifications our task is com-
plete. An act is the exercise of a norm-creating power if and only
if it normatively affects the content or existence of a norm which
is justified by content-independent arguments. An act is the
exercise of a regulative power if and only if it normatively
affects the application of a norm and there are content-
independent arguments for respecting the norm as affected by
such acts. This explanation does not presuppose that normative
powers are themselves conferred by norms. But it makes plain
that sometimes they are. The problem of the conditions under
which a normative power should be regarded as conferred by
a norm raises interesting issues which cannot be examined here.
The account proposed of the concept of a normative power is
over-simplified because it overlooks the possibility of justifica-
tions of norms which combine content-independent and content-
dependent arguments. But this further complication does not
raise new problems of principle and need not detain us here.

V . Normative Powers and Voluntary Obligations

Having attempted to provide the outline of a general analysis
of the concept of a normative power, it is time to demonstrate
the importance and usefulness of the concept. I t has been
suggested above that the concepts of a norm and of a normative
power provide the key to the notion of authority. A person has
authority to the extent that he has power to affect norms.
Instead of developing this suggestion, let us turn to the problem
of voluntary obligations.
I agree with Mr. MacCormick's main thesis, namely, that
the analysis of the concept of a volu~ltaryobligation does not
entail that only social practices can give rise to binding volun-
tary obligations. Mr. MacCormick is, however, mistaken in
thinking that this entails that voluntary obligations can be
explained without reference to normative powers.
Mr. ~MacCormick restricts his attention to promises, but
promises are not the only source of voluntary obligations. Among
the other sources of voluntary obligations which some people
regard as binding one should mention the private vow or oath.
I do not wish to consider to what extent and on what grounds
private vows, promises or other sources of voluntary obligations
are to be regarded as binding. My point is that any adequate
account of the concept of a voluntary obligation should be
wide enough to explain all the sources of voluntary obligations,
regardless of whether one believes that they ought to be recog-
nizedas binding. A man does not necessarily make a private vow
intending to induce others to rely upon it, or knowing that his
act betrays such an intention. Consequently, so far as the obli-
gatoriness of a vow can be justified, this must be in a manner
different from the justification of promises. Promises do,
whereas private vows do not, create claims for other
people on the behaviour of the persons under obligation.
This explains the difference in the way these voluntary obliga-
tions are created and justified. Since Mr. MacCormick has
confined himself to an examination of promises we must con-
clude that he has not provided us with a general explanation of
the concept of a voluntary obligation.
Not every obligation created for a person by his own volun-
tary action is a voluntary obligation. By locking John up Tom
has imposed upon himself an obligation to release John, to
explain his behaviour, apologize and, sometimes, compensate
John. But his is not a voluntary obligation. The only way in
which voluntary obligations can be distinguished from other
obligations which a person imposes on himself by his oivn action
is by reference to their justification. Voluntary obligations are
characterized by having content-independent justification.
The agent is under a voluntary obligation to perform an act
only if the reasons because of which the act is obligatory bear
not on the desirability of performing the act but on the manner
in which the obligation was incurred.
Obligations imposed by authority and voluntary obligations
g8 11-JOSEPH RAZ

are the two main types of obligation justified by content-

independent arguments. This leads to the conclusion that the
concept of a normative power applies to voluntary obligations.
Those who are entitled to bind themselves by voluntary
obligations have normative power to do so. Since we regard
every person as having equal power to bind himself we do not
have much need to refer to these powers. If we believed that
only certain categories of persons had these powers, we should
have more reason to refer to them. But this does not mean that
the concept does not apply or that its examination does not
clarify the structure of practical reasoning.
Because all types of voluntary obligations are charac-
terized by being mandatory norms with content-
independent justification, they are justified by the justification
of the general norm that promises, or that private vows, ought
to be respected; they are not justified by giving reasons for the
desirability of each obligatory act in its particular circumstances.
Mr. MacCormick is, therefore, wrong in rejecting the two-level
procedure of justification-first justifying the obligatoriness of
the act in terms of the general norm that promises etc. ought to
be respected, and then justifying that norm itself. The two-level
justification does not presuppose, as Mr. MacCormick thinks it
does, that the general norm is a social norm.

VI. Normative Powers and Promises

It follows from the discussion above that to the extent that
promises are a source of voluntary obligations they are made by
the exercise of normative powers. The obligatoriness of many
promises can no doubt be explained on other grounds which
do not depend on the fact that promises yield voluntary obliga-
tions. But such explanations, correct and useful as they are, miss
the essential point in the common conception of promises.
I t may be that the common conception is morally indefensible,
and that we should abandon or reform the common belief
which regards promises as a source of voluntary obligation and
justified as such. I t is important, however, not to confuse
reform with elucidation.
I t seems to me that Mr. hlaccormick, while professing to
elucidate and defend the common conception of promises, is
actually proposing to rejectit, and to elucidatea different type of
obligation which is not a voluntary obligation at all. That this
 VOLUNTARY OBLIGATIONS AND NORMATIVE POWERS 99
is the effect of his account is intimated in his explanation of
the promising act and is made clear by his justification of the
binding force of promises. I shall comment on each of these
points in turn.
Strictly speaking there is nothing in Mr. MacCormick's
account of the promising act which conflicts with the explana-
tion given above of the r81e of normative powers in the analysis
of voluntary obligations. If his account is correct we need
only conclude that acts intended to induce reliance constitute
the exercise of the normative power which creates a binding
promise. Nevertheless, I doubt whether intentionally inducing
reliance or acting in a way which suggests such a n intention is
either necessary or sufficient for promising.
Can one intentionally induce reliance without promising?
Imagine that John wants to know whether he can rely on
Harry giving him a lift to town tomorrow. Harry tells him: "I
am almost certain to offer you a lift to town tomorrow. I n the
circumstances it would be far wiser for you to rely on me
rather than make alternative arrangements, but remember-
I do not promise anything, I am merely advising you."
Harry is intentionally inducing John to rely on him but he
does not promise anything. Promising is surely more than
inducing reliance, by promising I bind myself and confer a
right on the promisee. Harry in my story makes clear that he
does not wish to bind himself or to give John any rights.
Can one promise without intending to induce reliance or
acting in a way which suggests such an intention? Suppose that
Colin tells David: "I promise to make you a gift of a million
pounds on your 2 1st birthday, but I beg of you: Until the time
comes act as if I had notpromised. It would ruin your character
if you should now start behaving in anticipation of your gift." We
may well think that Colin was ill-advised to make the promise,
because once made, David may have reasonable grounds for
relying on it and may disregard Colin's advice. But we cannot say
either that Colin intended David to rely on his promise or that
he behaved in a way which created the impression that he had
such an intention.
I have followed Mr. MacCormick in interpreting "inducing
reliance'' as meaning encouraging action in reliance upon the
promised act. One may interpret the expression as including the
I00 11-JOSEPH RAZ

inducement of a belief. However, the only belief that the pro-

misor need always intend to induce or be taken to have the
intention of inducing is the belief that he (the promisor)
intends to undertake an obligation and to confer a right on the
promisee. It is, indeed, both a necessary a n d a sufficient condition
of making a promise that the promisor beha1.e in a way which
is either intcnded or can reasonably be taken to indicate the
promisor's intention to bind himself to perform a n act and to
give the promisee a right against the promisor.
O n this account of the promising act it is plain that in normal
circumstances acting with an intention to induce reliance is
making a promise, but the account also allows for special
circumstances in which this is not so. This explanation of the
nature of the promising act also helps to clarify one aspect of
the relevance of social convention to promising. I agree with
1Mr. MacCormick that the concept of a promise does not
presuppose that of a social practice of regarding promises as
binding. I t may be, though I do not wish to express an opinion
on the matter, that there are adequate reasons for regarding
promises as binding even in a society which does not have any
such practice. But it is clear that social convention may none-
theless determine what acts can reasonably be talten to express
an intention to undertake an obligation and confer a right. I n
our culture, communicating an intention to act in a certain
way to a person who is known to be interested in the action is
conventionally regarded as an act expressing an intention to
undertake an obligation to act in that way and to confer a right
on the person concerned. Since this is our social convention
anyone wishing to express such an intention and yet not
wishing to undertake the obligation must make his intentions
clear by saying that he is not promising or words to that effect.
But for this convention,communicatin,a an intention to perform
an act in these circumstances would not normally amount to
promising, though there would still be other ways of promising.
T h e fact that M r MacCormick thinks that communicating
a n intention to induce reliance is logically, and not merely
contingently, related to promising, suggests that he does not
regard promises as a case of voluntary obligation a t all. This
is borne out by his justification for the obligation to keep
promises. Mr. MacCormick derives the obligation to keep a
 L'OLUNTARY 0BI.IGATIONS A N D K O R M A T I V E POIVERS 101

promise from the principle that one "must not act as to dis-
appoint the reliance of others when we intentionally or know-
ingly induce them to rely on us".
But this principle in itself is not suficient to explain the binding
force of promises. As illustrated above, it fails to distinguish
them from certain cases of advising or even merely informing,
and more important still, it does not explain why promises once
made and understood are binding even if not relied upon. Mr.
MacCormick mentions this feature of promises but fails to
account for it. I n fact he regards promises as a case of
estoppel. T h e doctrine of estoppel does rest on the principle he
mentions, but duties arising from estoppel are not voluntary
obligations. Though many cases are both cases of estoppel and
of promising, the two notions are distinct. This is clear from the
fact that estoppel applies only if the person concerned relied on
the other person, whereas promises are binding even in the
absence of reliance. T h e reason for this difference is that
promises are binding because it is desirable to make it possible
for people to bind themselves and give rights to others if they
so wish. I t is desirable, in other words, to have a method of
giving grounds for reasonable reliance in a special way, not
necessarily by intending to induce reliance, but by intending to
bind oneself. I t is because of this feature that we regard pro-
misesasasourceof voluntary obligations, and it is because of this
feature that making a promise or undertaking any other volun-
tary obligation is exercising a normative power.

REFERENCES
In Of Laws in General, 1970. Bentham's analysis is discussed in detail by
H. L. A. Hart, "Bentham on Legal Powers", Yule, L.J., 1972.
Lawyers have established procedures for determining what are the law's
intentions or reasons. These need not be discussed here.
On this point cf.Hart: "Bentham on Legal Rights", in A. W. B. Simpsor~
Oxford Essays in Jurisprudence, 2nd series, O.U.P. 1972, in which some
exceptions to this rule are discussed.
* Elements o f Law, 5th ed. 1896, pp. 93-4.
"The identity of Legal Systems", 59 CulSfornia Lazo Review (1971).
Id., P. 795;'
' Cf. Hart: Kelsen's Doctrine of the Unity of Law" in Ethics and Social
Justice, edited by H. E. Kiefer and hl. K. Munitz, 1970, at p. 196.
For a detailed discussion see Raz, The Concept of a Legal System, (1970) pp.
90-92 and 140-7.
I02 11-JOSEPH RAZ

From the note ending An Introduction to the Principles of Morals and

Legislation.
lo See Hart, The Concept of Lnw, 196I , pp. 35-41 ; Raz, The Concept of a
Legal System. pp. 70-92, 109-120.
11 "Reasons for Action, Decisions and Norms" (unpublished).
laThis and other norms mentioned in the sequel are meant as illustrative
examples. My case does not depend on agreement that they are binding
moral or prudential norms.
l 3 A. Kenny, "Intention and purpose in Law", Essays in Legal Philosophy
ed. R. S. Summers, 1968, p. 150. See also Von Wright, Norm and Action,
1963, PP. 39-42.