You are on page 1of 14

Oxford Journal of Legal Studies, Vol. 21, No. 2 (2001), pp.

355–368

Defeasibilism
RICHARD H. S. TUR∗

Abstract—The author suggests that law is best represented, understood, and taught
in the form of open-ended, defeasible, normative, conditional propositions. The
meaning, role, and significance of defeasibility is explained by presenting three
‘canonical forms’ and by distinguishing exceptions and overrides. The role of equity
in the law of contract, as understood by the author, is taken as an exemplar of
override and parallels are drawn with policy in the English law of tort and with
mercy in the criminal law of England and Wales. The third and preferred canonical
form is sequenced and internally differentiated and seeks a satisfactory middle way
between all-or-nothing rules and wild-card discretion and a reconciliation of the
surface demands of legal rules and the deeper values embedded in law. The suggested
approach therefore resists closure and seeks a conceptual restructuring of law which
opens it up to the influence of notions such as justice and equity and locates them
as open-ended conditions of defeasance inhabiting the express or implied ‘unless
. . .’ clause which is taken to be an element in every fully-formed statement of law.
This approach accommodates law as a locus not only of traditional rule-based
reasoning, but also of open-ended internal moral reasoning invoking deeper values
inherent in law and only imperfectly captured and implemented by formal legal
rules.
‘A rule that ends with the word “unless . . .” is still a rule’. (H. L. A. Hart)

In this paper, I seek briefly to present and defend the suggestion that law is best
represented, most clearly understood, and most effectively taught in the form of
open-ended, defeasible, normative, conditional propositions.There is nothing
novel about the Kelsenian idea that law may be understood as conditional
normative propositions in the canonical form: ‘If A is, then B ought to be’1
where A stands for the legally defined facts and B for the legally determined
consequences. Nor is open-endedness a new idea in jurisprudence. Indeed, an
understanding of law as including or being constituted by open-ended standards
has become commonplace through the efforts particularly of the Critical Legal
Studies movement. Thus it is suggested that a greatly increased use of ‘open-
ended standards and general clauses in legislation, administration, and ad-
judication’ is characteristic of ‘modern’ legal systems.2 Be that as it may, as will

∗ Benn Fellow and Senior Law Tutor at Oriel College, Oxford. I wish to thank Nick Barber, Alan Brudner,
Bruce Chapman, John Gardner, Jim Harris, Tony Honoré, Jeremy Horder, Brian Leiter, Neil MacCormick, Bill
Powers, Fred Schauer, Stephen Shute, and Nicos Stavropoulos for helpful comments and encouragement.
1
Hans Kelsen, Introduction to the Problems of Legal Theory (Clarendon Press, 1996 trans Bonnie Litschewski
Paulson and Stanley L. Paulson) at 24; Martin P. Golding, Philosophy of Law (Prentice-Hall, 1975) at 39–43.
2
Roberto Mangabeira Unger, Law in Modern Society (The Free Press, 1976) at 194.
 2001 Oxford University Press
356 Oxford Journal of Legal Studies VOL. 21

emerge in the sequel, this ‘modern’ approach locates open-endedness in the


wrong place.
Even defeasibility—the central and distinctive element in my approach—is
not new. Writing about the legal concept of ‘contract’, Hart expressed the opinion
half a century ago that this ‘concept is irreducibly defeasible in character and to
ignore this is to misrepresent it’.3 What this signified at that time for Hart was
that
. . . it is not usually possible to define a legal concept such as . . . “contract” by specifying
the necessary and sufficient conditions for its application . . . such concepts can only
be explained with the aid of a list of exceptions or negative examples showing where
the concept may not be applied or may only be applied in a weakened form.4

Building on this analysis, Hart plausibly proposed that defeasibility is char-


acteristic of all legal concepts but he also proposed a ‘more controversial thesis
that the concept of a human action is an ascriptive and defeasible one’5 which,
he thought, might solve a whole raft of problems in the philosophy of action.
However this was not to be and 20 years later, in the Preface to a collection of
his essays on Punishment and Responsibility, Hart indicated that he thought that
‘its main contentions no longer seemed . . . defensible, and that the main criticisms
of it . . . are justified’.6 However the two articles cited by Hart (by Geach,7 and
Pitcher8) at no point call into question the first main part of Hart’s original paper
which addressed the defeasibility of legal concepts in general and the concept
of contract in particular.
The idea of defeasibility, but not the word, seemed to me (in an article
published in 19779) evident in the following observation from MacCormick’s
intriguing inaugural lecture in 1974: ‘Even if we were to write out the statutory
rules including in them all the exceptions hitherto imposed by the courts . . . we
could not be confident that we had succeeded in listing the sufficient conditions
for the validity of a determination’10 and it was not until 1995 that MacCormick
himself expressly indicated the parallels between his insights and Hart’s early
interest in the defeasibility of legal concepts.11 The most significant contribution,
however, is Baker’s challenging and insightful article in 1977 on ‘Defeasibility
and Meaning’ in which he rises to the challenge implicit in Hart’s early criticism
of ‘obstinate loyalty to the persuasive but misleading logical ideal that all concepts

3
H. L. A. Hart, ‘The Ascription of Responsibility and Rights’ in XLIX Proceedings of the Aristotelian Society
(New Series) (1948–49) 171–194 at 176.
4
Ibid at 174.
5
Ibid at 187.
6
H. L. A. Hart, Punishment and Responsibility (Claredon Press, 1968), Preface.
7
P. T. Geach, ‘Ascriptivism’ in LXIX The Philosophical Review (1960) at 221–225.
8
George Pitcher, ‘Hart on Action and Responsibility’ in LXIX The Philosophical Review (1960) at 226–235.
9
Richard H. S. Tur, ‘Positivism, Principles, and Rules’, in Elspeth Attwooll (ed), Perspectives in Jurisprudence
(Aberdeen University Press, 1977) at 42–78, n 46, where I suggested, on the basis of MacCormick’s analysis, that
‘. . . legal formulae are always defeasible’.
10
Neil MacCormick, ‘Law as an Institutional Fact’ (1974) 90 LQR 102 at 125.
11
Neil MacCormick, ‘Defeasibility in Law and Logic’ in Zenon Bankowski, Ian White, and Ulrike Hanh (eds),
Informatics and the Foundation of Legal Reasoning (Kluwer, 1995) 99–117 at 99.
SUMMER 2001 Defeasibilism 357
must be capable of definition through a set of necessary and sufficient conditions’12
by offering a ‘radically different’13 kind of semantics which he calls ‘con-
structivism’14 because ‘. . . defeasibility can be harvested only as the fruit of a
new species of semantics’.15 It is unnecessary to test the merits of the theory of
meaning which Baker proposes because it may not be directly, if at all, relevant
to the species of defeasibilism which I wish to expound and defend, and in any
event I am concerned more with the structure of the legal formula16 than with
the meaning of legal concepts. With respect, neither Hart nor MacCormick saw
as far as they might have done in the illumination of their separate insights. Both
appear to limit their notions of defeasibility primarily to exceptions and not
expressly, if at all, to overrides, in a sense shortly to be explained.
Although none of the ingredients are themselves new, the approach which I
seek to expound and defend combines them in what appears to me to be a
unique and novel way which may greatly assist in the delivery of legal knowledge
and the teaching of law. This approach belongs, speaking loosely, to the rules-
based and purportedly value neutral tradition in jurisprudence and it is, in fact,
a refinement and extension of my understanding of salient aspects of Hart’s
project and Kelsen’s, but it goes beyond these authors in an effort suitably to
locate ideas such as ‘justice’ within its sophisticated version of the canonical
form of a norm. To bring the approach into sharper focus, it is necessary to
distinguish three versions of rule-like formulae.
First, ‘If A is, then B ought to be’. This is what Golding felicitously called the
canonical form of a [Kelsenian] norm’.17 Here ‘A’—the antecedent clause (or
factual protasis)—represents a fact or, more usually, a cluster of facts (a + b + c)
and ‘B’ ( the normative apodosis) represents the legally determined consequence,
attributed, ascribed or imputed to these facts. Kelsen summarized this as
‘Condition, copula, consequence’. The legally determined consequence may be
ultimate or intermediate. For Kelsen, the ultimate legal consequence is a coercive
act but in my view representing, understanding, and teaching law often requires
that attention be paid not to the ultimate but to an intermediate legal consequence.
Thus in the law of contract we may be much more interested in establishing the
intermediate consequence that, given certain facts, a contract ought to be
recognized as having come into being than we may be in the ultimate con-
sequences for failure to remedy breach as ordered by a court if for no other reason

12
H. L. A. Hart, ‘The Ascription of Responsibility and Rights’ in XLIX Proceedings of the Aristotelian Society(New
Series) (1948–49) 171–194 at 178.
13
G. P. Baker, ‘Defeasibility and Meaning’ in P. M. S. Hacker and J. Raz (eds), Law, Morality, and Society:
Essays in Honour of H. L. A. Hart (Claredon Press, 1977) 26–57 at 57.
14
Ibid at 50.
15
Ibid at 57.
16
Frederick Schauer is critical of theses similar to the one advanced in this paper, suggesting that they are ‘. . .
partly confused, partly superfluous, and partly wrong’; ‘On the Supposed Defeasibility of Legal Rules’ (1998) 51
Current Legal Problems 223 at 224; though in fairness to Schauer and to myself, there is very substantial agreement
between us as is revealed particularly by his closing paragraphs. I have also been very greatly assisted by the same
author’s ‘Exceptions’ 58 The University of Chicago Law Review 871–899 (1991).
17
M. P. Golding, ‘Kelsen and the Concept of “Legal System”’ in Robert S. Summers (ed), More Essays in
Jurisprudence (Blackwell, 1971) 69–100 at 88.
358 Oxford Journal of Legal Studies VOL. 21

than that ordinarily parties discharge their contractual obligations. Unpacking the
canonical form, where A=a + b + c which stand, respectively, for offer,
acceptance and consideration (the necessary and sufficient definitional elements
of a contract according to one widely held view of English common law18) and
B=a contract, this aspect of the law of contracts may be represented as If a +
b+ c exist, then a contract ought to be recognized to exist.
Two additional points must be made about my conception of this basic
canonical form. First, no element of value should be allowed to infect the
antecedent clause. For example, rape should not be defined as unlawful non-
consensual sexual intercourse, knowing that the woman is not consenting or
being (subjectively) reckless as to whether she is consenting. The inclusion of
‘unlawful’ among the definitional elements leads to circularity unless ‘unlawful’
is taken descriptively to mean ‘contrary to current social morality’ which would
give it descriptive sense and reference. In English criminal law there are competing
opinions as to whether ‘unlawful’ or ‘without lawful excuse’ add anything to the
definition of a crime. In R v R both Lord Lane CJ in the Court of Appeal and
Lord Keith of Kinkel in the House of Lords expressed the opinion that the word
was ‘surplusage’.19 However some years earlier in the Court of Appeal, Lawton
LJ denounced the view that the word ‘unlawful’ is tautologous, stating that ‘. . .
the word “unlawful” does import an essential element into the offence’20 and
that was approved by a differently constituted Court of Appeal in the same
year.21 Clearly in the face of such differing opinions, the law teacher, or legal
theorist is compelled to make a choice. It cannot be that ‘unlawful’ is both
‘surplusage’ and an ‘essential ingredient’. On the approach advanced, defended,
and illustrated here, including ‘unlawful’ in the protasis involves locating value
issues in the wrong place. Such a choice implies some reconstruction of the
data. The approach I favour seeks to reconstruct the data in a structured,
conceptually sequenced manner.22
Second, this first, basic canonical form is wholly closed, absolutely attributing
the legally defined consequence to the fact-cluster in an all-or-nothing fashion.
Such closure is consistent with extreme formalism but the certainty thereby
purchased is at the expense of justice. Although nothing of significance turns on
it for the purposes of the present discussion, I do not think that Hart or Kelsen,
or even Austin, can plausibly be read as formalists, committed only to law in
the basic canonical form, although, perhaps both Blackstone and Bentham are
plausible candidates. Hart certainly understood that ‘. . . a rule that ends with
18
This is not how I, myself, would conceptualize contract. As a defeasibilist, I do not seek necessary and
sufficient conditions. So I treat offer and acceptance as evidence of agreement and consideration as evidence of
an intention to be bound. Accordingly, the presence of these elements, although ordinarily very highly persuasive,
do not necessitate the conclusion that there is a contract, nor is their absence fatal to a claim that there is contract.
This, I submit, is wholly consistent with the reasoning in Gibson v Manchester City Council [1979] 1 W L R 294
where the House of Lords expressly left open the latter possibility.
19
[1991] 2 All ER 257 at 265; [1991] 4 All ER 481 at 489.
20
R v Kimber [1983] 3 All ER 316 at 320.
21
Gladstone Williams (1984) 78 Cr App R 276.
22
Bruce Chapman, ‘Law Incommensurability, and Conceptually Sequenced Argument’ 146 University of
Pennsylvania Law Review 1487–1528 (1998).
SUMMER 2001 Defeasibilism 359
the word “unless . . .” is still a rule’23 and this insight (though neglected by his
critics such as Simpson and Dworkin) leads on to what I call the ‘second
canonical form’: If A is, then B ought to be, unless . . . This can be illustrated by
the relation in English criminal law between definitional elements and defences.
Where a definitional element is missing, the prosecution ‘withers on the bough’.
Defences, properly-so-called, come into play, if at all, only once all definitional
elements are present. For example, if all the definitional elements of assault are
present, the defendant ought to be convicted unless there is a defence that
negatives the standard legal attribution of criminal liability to those elements.
Notice, in passing, that this structure works as well for the law of contracts as
it does for criminal law. If ‘defences’ such as duress, fraud or mistake are
represented, respectively, by x, y, and z, the unpacked version of the second
canonical form may be spelled out as follows: If a + b+ c exist, then a contract
ought to be recognized to exist, unless x, or y, or z exists.
It is important to note, however, that the second form is logically equivalent
to the first form in that all the excepting conditions can be incorporated into
the antecedent clause, without remainder, thus: If a + b+ c exist, and neither z,
y, or z is present, then a contract ought to be recognized to exist. It follows that the
second canonical form is as formalist or closed as is the first, although there
may be didactic or practical reasons to present the data in the second rather
than the first form as, indeed, the usual structure of textbooks on criminal law
or the law of contract, and the ebb and flow of court argument bears out.24 It
is for this reason that I suggest that neither Hart, nor MacCormick, take
defeasibility far enough. What is interesting and very much worth exploring is
not only the relationship of a rule and its exceptions but the question whether
a rule, taken together with all its exceptions, might nonetheless be disapplied,
or nullified, or overridden in a particular case on grounds such as mercy, justice
or equity, purpose, or rights. ‘Defeasibilism’ is simply a shorthand way of referring
to a cluster of salient features of at least some legal systems which impact
negatively upon the normal attribution of legally determined consequence to
legally defined conditions. In some cases, the defeating consideration impacts
only upon the result but not on the rule. Thus a rule of the law of contract
remains intact and available to be applied again and again even though trumped
or defeated by equitable considerations in a particular case. In general, the
defeasibility of the normal attribution of legally defined consequences to legally
defined conditions means that the attribution ‘. . . is subject to termination or
defeat in the face of certain contingencies and it survives intact if no such
contingencies mature’.25
However, on the analysis outlined above, the invocation or invention of an
exception logically entails alteration of the rule in that the cluster of positive and

23
H. L. A. Hart, The Concept of Law (Claredon Press, 1961) at 136.
24
Bruce Chapman, ‘More Easily Done than Said: Rules, Reasons and Rational Social Choice’ (1998) 19 OJLS
293–329.
25
Ibid at 323.
360 Oxford Journal of Legal Studies VOL. 21

negative conditions constituting the fully stated protasis is refined by the inclusion
of the new element. In the course of such refining, over time, the exceptions
may come to outweigh the rule and a restatement of the rule is then appropriate.
This was the reasoning process favoured by Lord Lane CJ in R v R.26 But the
conceptual structure remains the unsequenced logic of closure and the intellectual
aspiration remains reconstructing the legal data as a series or system of formal
legal rules stipulating the necessary and sufficient conditions for the normative
attribution of the legally defined consequence.
What is striking in Hart’s long list of conditions defeating contract is the
absence of any reference to equity. Of course Lloyd’s Bank v Bundy27 could not
have been known to Hart in 1948 but London Property Trust v High Trees House
Ltd28 might have triggered an interest in the defeat or qualification of a legal rule
by a judicial invocation of equity. Perhaps it was regarded as an aberration at
the time. Defeasibilism as an approach to law is warranted, if warranted at all,
specifically by cases like Riggs29 or Henningsen30 which are not cases of old or
newly minted exceptions but cases of equitable or justice override. In his inaugural
lecture MacCormick was well aware of this: ‘In Riggs the argument on principle
is taken as justifying the opposite conclusion that a will may be invalid even
though all the expressed requirements of the law have been met’.31 But in his
own subsequent contribution to the defeasibily debate he seems to lose sight of
this important point and ‘reduces’ the notion to the ‘pragmatics of legal practice’.32
There is, of course, a serious question whether defeasibility merely reflects the
content of some positive laws (in some systems) and is not really relevant to
legal theory at all which is the implication of Feinberg’s conclusion that, ‘The
notion of defeasibility then is inextricably tied up with an adversary system of
litigation and its complex and diverse (procedural) rules . . . ’.33 I think this is
incorrect, or too swift. It is possible that defeasibilism is of the essence of law
and legal systems and not merely an accidental feature of some laws in some
systems. If this intuition can be sustained it appears to follow that theories of
law that express or imply closure do a kind of conceptual mischief to the subject
matter.
What I want to attempt is to transcend closure without embracing absolute
(or scorer’s) discretion. I seek to accommodate justice, and like notions, at some
point within the concept of law, without jeopardising certainty and com-
munication to an unacceptable degree. What I am seeking is ‘A structure that

26
[1991] 2 All ER 257 at 265.
27
[1975] QB 326.
28
[1947] KB 130.
29
22 N E 188 (NY 1889).
30
(1960) 32 NJ 356 (Supreme Court of New Jersey).
31
Neil MacCormick, ‘Law as an Institutional Fact’ (1974) 90 LQR 102 at 124.
32
Neil MacCormick, “Defeasibility in Law and Logic” in Zenon Bankowski, Ian White, and Ulrike Hanh (eds),
Informatics and the Foundation of Legal Reasoning (Kluwer, 1995) at 104.
33
Joel Feinberg, ‘Action and Responsibility’ in Alan R White (ed), The Philosophy of Action (Oxford University
Press, 1968) at 97.
SUMMER 2001 Defeasibilism 361
needs to accommodate both regularity and (an independent) particularity’.34
That leads on to the third canonical form which explicitly resists the closure
exhibited by the first two forms. The difference may be brought out in verse:
If all be true that I do think
There are five reasons I do drink
Good food, good wine, or being dry,
Or lest I should be, by and by;
Or any other reason why . . . .

Notice the shift in what might be called legislative technique between the fourth
and fifth lines. One moves from a list of specific reasons, to a general catch-all
residual category which unlike a list is not closed and cannot be closed. The
idea which this sophisticated canonical form seeks to accommodate is that of
Keble J to the effect that ‘All laws admit of cases of just excuse’.35 What is in
point is a conceptual niche for considerations such as justice and equity, and
the like.
Legal theory has not accommodated equity comfortably, if at all. Yet two great
historic systems of law, the common law and Roman law, exhibit equity at work
through the Chancellor and the praetor. For some theorists and textbook writers
equity is dead, long since incorporated, as a body of closed rules, into the general
body of law also conceived in this way. However, if one resists this ‘fusion
fallacy’, not least because legal rules are general and prospective whereas equity
is always particular and retrospective, equity presents an embarrassing problem
for legal theory. How can this unpredictable, subjective, contextual category of
practical thought and decision supplement the law, understood as certain,
objective, and logical, without wholly subverting it? Equity, this ‘roguish thing’
thus appears as law’s ‘dangerous supplement’ but it is ‘. . . the office of equity,
to support and protect the common law from shifts and crafty contrivances
against the justice of the law. Equity therefore does not destroy the law, nor
create it, but assist it’.36 This insight poses a problem for the form of the legal
proposition. Unless ossified, and denatured, equity resists inclusion in either of
the first two canonical forms. And what is true here of equity is true also of
justice, mercy, rights, purpose and policy.
What is required is neither the all-or-nothing, closed rules of the aspirational
absolute formalist, nor the wholly unstructured discretion of a platonic guardian
doing ineffable justice but a means of bringing these two, apparently conflicting,
notions into some harmonious form. This, I submit, is what the third canonical
form achieves. What it does not and cannot achieve, of course, are answers to
our moral and practical problems that can be read off as from an ethical blueprint.
The third canonical form merely locates the site of such problems and reveals
how justice and the like conceptually interfaces with rules of law. This logical
34
Bruce Chapman, ‘Law Games: Defeasible Rules and Revisable Rationality’ (1998) 17 Law and Philosophy at
480.
35
Moore v Hussey 1609 Hob 93 at 96 per Keble J.
36
Dudley v Dudley (1705) Pr Ch 241at 244 per Lord Cowper.
362 Oxford Journal of Legal Studies VOL. 21

geography of justice and law in turn assists in explaining not merely the trite
saw that ‘law’ is not or not always mechanical application of rules but also at
what point and how less formal considerations enter into the concept of law.
The third canonical form is: If A is, then B ought to be, unless there is an
overriding reason to the contrary. Applied to the law of contracts, for example, the
unpacked version of the second canonical form is restructured as follows: If a
+ b+ c exist, and neither x, y, or z is present, then a contract ought to be recognized
to exist, unless it would be unconscionable (or otherwise intolerably unjust) to do so.
This is, as it appears to me, exactly the form of sequenced argument in English
contract cases where, once having asked the question of law and answered it by
application of the common law rules, one then should also ask, ‘Is there any
equitable override?’. An Australian Contract Code presented for discussion by
the ill-fated Law Reform Commission of Victoria in 1992 concludes with Article
27, the ‘overriding article’ which states that a person may not assert a right or
deny an obligation to the extent that it would be ‘unconscionable to do so’.37
Critics assert that such a clause gives too much power to judges but those
responsible for drafting the code claim that this clause merely reflects established
judicial practice.
In 1523, Christopher St German stated in his Doctor and Student, ‘Wherefore
in some cases it is necessary to leave the words of the law [i.e. the strict meaning]
and to follow that [which] reason and justice requireth, and to that intent equity
is ordained; that is to say, to temper and mitigate the rigor of the law’. Some
four-and-a-half centuries later Lord Denning was to echo this in Re Vandervell’s
Trust (No 2): ‘Every unjust decision is a reproach to the law or the judge who
administers it. If the law should be in danger of doing injustice, then equity
should be called in to remedy it. Equity was introduced to mitigate the rigour
of the law’.38 Lord Denning, too, is associated with the observation that ‘Equity
is not past the age of child-bearing’39 and this observation is crucial to a full
understanding of the third canonical form’s resistance to closure.
A similar conclusion as to the utility of the third canonical form can be
established from a consideration of the English law of tort, with policy—an
unruly horse—replacing equity as the open-ended category of defeasance. There
is a significant overlap between what judges refer to as ‘policy’ and equity in the
history of English law. What was clearly an equitable principle in Riggs—that
none should profit from their own wrong—was treated as ‘policy’ by the English
court in ex parte Connor.40
Classically, tort liability for negligence involves a breach of a duty of care
causing damage. Still classically, Lord Atkins’s famous ‘neighbour principle’
delimits the duty of care:
You must [he says] take reasonable care to avoid acts or omissions which you can
37
Law Reform Commission of Victoria, An Australian Contract Code Discussion Paper No 27, September 1992.
38
[1974] 1 Ch 269, 322.
39
Eves v Eves [1975] 3 All ER 768 at 771.
40
R v National Insurance Commissioner, ex parte Connor [1981] 1 All ER 770.
SUMMER 2001 Defeasibilism 363
reasonably foresee would be likely to injure your neighbour. Who, then, in law is my
neighbour? [This receives a restrictive answer] . . . persons who are so closely and
directly affected by my act that I ought reasonably to have them in contemplation as
being so affected when I am directing my mind to the acts or omissions which are
called into question.41

It is trite law that foreseeability of likely harm is not in itself enough. Some
further ingredient is necessary and classically that further ingredient is found in
the notion of proximity. The question then is whether these two notions, (a)
foresight of harm and (b) proximity together are ordinarily necessary and sufficient
conditions for the assertion of a duty of care? I would like to say (with Lord
Reid and the classical tendency) that they are and that together with (c) breach,
(d) causation, and (e) damage or loss, we have the ‘A’ or antecedent clause for
a formulation of the law: If A (i.e. foresight, proximity, breach, causation, and
damage) occur together, then B (the legally determined remedy, i.e compensation)
ought to be awarded, unless . . . . So what ‘unless . . .’ might there be?
In Dorset Yacht,42 Lord Reid remarked that Lord Atkin’s famous foreseeability
test was not a statutory formula but a statement of general principle which
establishes a duty subject to any judicial reasons to warrant its exclusion. This
produces the complete formulation: ‘If A (i.e. foresight, proximity, breach,
causation, and damage) occur together, then B (the legally determined remedy,
i.e compensation) ought to be awarded, unless there is some justification or valid
ground for its exclusion’. This has the effect of presuming for the remedy unless
it is excluded by argument. On this approach, that there be a remedy where all five
ingredients co-exist is the default position. In Anns v Merton,43 Lord Wilberforce
indicated that one should first apply Lord Atkin’s test and then weigh policy
considerations (if any). His Lordship advocated asking first about proximity and
foresight and if that produced an affirmative answer then ‘a prima facie duty of
care arises’. The second stage involves asking whether there ‘are any con-
siderations that negative, reduce or limit the scope of the duty or the class of
person to whom it is owed . . .’. Incorporating this produces the formulation, ‘If
A (i.e. foresight, proximity, breach, causation, and damage) occur together, then
B (the legally determined remedy, i.e compensation) ought to be awarded, unless
there are any considerations that negative, reduce or limit the scope of the duty
or the class of person to whom it is owed and hence exclude, limit or reduce
the remedy’. Controversy arises with this potentially very wide approach in
relation to some functions of the police and other emergency services, the Crown
Prosecution Service, professional bodies such as the Law Society, local authority
social services for children, and in-court lawyers. In many disputed areas, the
absence of sufficient proximity would limit the scope of the potentially very wide
classical approach, and where proximity is established, policy override remains
available to limit its scope.
41
M’Alister (or Donoghue) v Stevenson [1932] AC 562 at 580.
42
[1970] AC 1004 at 1027.
43
[1978] AC 728 at 751–2.
364 Oxford Journal of Legal Studies VOL. 21

More recently there has been a fulfilled tendency—in my view theoretically


unsupportable—to substitute for the earlier two-stage approach of (1) fore-
seeability and proximity and (2) considerations that negative duty, a compendious
one-stage analysis conjoining foreseeability and proximity with whether it is ‘just
and reasonable’ to impose a duty.44 On the structured and sequenced approach
canvassed in this paper, this gets any relevant question of justice or reason
defeating the normal attribution of legal consequence to the legally determined
fact-cluster in the wrong place and generates greater uncertainty and judicial
license than would obtain under the earlier, more structured analysis and
presentation. Contemporary revisionism has its roots in a powerful dictum of
Brennan J in the High Court of Australia in Sutherland Shire Council v Heyman:
It is preferable in my view, that the law should develop novel categories of negligence
incrementally and by analogy with established categories, rather than by a massive
extension of a prima facie duty of care restrained only by indefinable “considerations
which ought to negative, or to reduce or to limit the scope of the duty or the class of
person to whom it is owed”.45

This passage is quoted with approval by Lord Bridge in Caparo v Dickman,46


where he adds to the classical ingredients of foresight and proximity, a new
element, namely that it is ‘fair, just and reasonable that the court should impose
a duty of a given scope on the one party for the benefit of the other’. Lord
Bridge continues to the effect that the ingredients of proximity and fairness ‘are
not susceptible of any such precise definition as would be necessary to give them
utility as practical tests’. They are
little more than convenient labels to attach to the features of different specific situations
which, on a detailed examination of all the circumstances, the law recognises prag-
matically as giving rise to a duty of care of a given scope.

Five years earlier, Lord Keith of Kinkel in Peabody Trust v Parkinson Ltd,47
whilst acknowledging Lord Wilberforce’s ‘two stage’ test and endorsing the
requirement of ‘proximity’, indicated that the scope of the resultant duty of care
was dependant upon circumstances. He cited references in Lord Atkin’s judgment
in M’Alister (or Donoghue) v Stevenson invoking what is fair and reasonable and
suggesting that policy need not be invoked where reason and good sense point
in the same direction. Lord Keith concludes from this that in determining
whether or not there is a duty of care the courts should take account of what is
‘fair and reasonable’. In Yuen Kun Yeu v A-G of Hong Kong48 (which raised a
question about liability for negligence of a commissioner of deposit-holding
companies) Lord Keith suggested that policy override would be ‘rare’ though
44
Sutherland Shire Council v Heyman (1985) 60 ALR 1 at 43–44; Caparo v Dickman [1990] 2 AC 605 at
617–618; Peabody Trust v Parkinson Ltd [1985] AC 210 at 240–241; and Yuen Kun Yeu v A-G of Hong Kong [1988]
AC 175 at 193.
45
(1985) 60 ALR 1 at 43–44
46
[1990] 2 AC 605 at 617–618
47
[1985] AC 210 at 240–24.
48
[1988] AC 175 at 193.
SUMMER 2001 Defeasibilism 365
he does acknowledge Rondel v Worsley,49 in my view correctly, and Hill v Chief
Constable of West Yorkshire,50 in my view questionably, as examples. Because
override will be rare, the two stage approach is not to be regarded as in all
circumstances a suitable guide to the existence of a duty of care’. The implication
seems to be that even though foresight of harm and proximity are present, and
there is no obvious policy override, a court may deny that a duty of care arises,
and therefore deny the plaintiff any remedy. In this case, Lord Keith preferred
to deny a remedy on the basis of a denial of a duty of care, rather than on public
policy, though he also thought that there was ‘much force in these arguments’.
The contemporary revisionist position seems to me unsatisfactory in a number
of ways. First, the addition of a requirement that it be fair and reasonable to
impose a duty of care presumes against such remedy unless it is established by
argument that there be no remedy even where all five original ingredients co-
exist, unless it is also fair and reasonable to impose it. That restricts, perhaps
quite significantly, the availability of remedies. Second, contemporary revisionism
has the effect of embedding and therefore obscuring what are essentially policy
decisions in the definitional elements. A claimed virtue of the structure that I
propose is that the policy issue is always highly visible, rather than hidden. Third,
what is ‘fair and reasonable’ appears not to be a question of fact, but one of
value. On the structure of the legal proposition that I favour, such value questions
ought not to be couched in positive terms and located in the definitional clause
but should be located in the ‘unless clause’ and should be couched negatively.
This would have the effect of assimilating Lord Wilberforce’s ‘considerations
that negative . . .’ and Lord Reid’s ‘justification for exclusion . . .’ on the one
hand, with Lord Keith’s ‘fair and reasonable to impose’ requirement, on the
other. In other words, that it was not fair and reasonable to impose a duty of
care would be relocated conceptually as a policy or justice override, rather than
feature as an (extraordinarily vague and imprecise) definitional element. The
effect of Lord Keith’s location of the question of value is to give the courts an
opportunity to refuse a remedy, where prima facie apt, without having to invoke
‘policy’ arguments increasingly perceived as themselves dubious and unfair and
which expose the judiciary to criticisms from the public and the European Court
of Human Rights alike.
In Hill v Chief Constable of West Yorkshire51 there is room to doubt whether
there was sufficient proximity to establish a duty of care. If there was not, then
it is not illustrative of policy override but merely of the absence of a definitional
element. I believe that it is at least arguable that Hill’s case is an instance of
policy override52 but even if I am wrong on that, clearly Osman v Ferguson53 was

49
[1961] 1 AC 191.
50
[1987] 1 All ER 1173 (CA);[1988] 2 All ER 238 (HL).
51
[1988] 2 All ER 238.
52
Like Lord Keith, Lord Oliver in Caparo v Dickman [1990] AC 605 at 633 sees both Rondel v Worsley [1969]
1 AC 191 and Hill v Chief Constable of West Yorkshire [1989] AC 53 as instances of policy override, although there
may well be no duty of care for want of sufficient proximity.
53
[1993] 4 All ER 344.
366 Oxford Journal of Legal Studies VOL. 21

(and was subsequently subjected to fatal criticism in the European Court of


Human Rights on that ground because the policy which had been invoked as
an override was unprincipled, or not in accordance with law). Other illustrations
include Giannerelli v Wraith54 in Australia and Rondel v Worsley55 in England and
now overtaken by the House of Lords in Hall v Simons on 20 July 2000.56
Like conclusions may be advanced in criminal law with mercy in addition to
fairness as an open-ended category of defeasance. At least from the time of
Christ, the role of mercy in relation to law was well understood: ‘Mercy has
freedom, and sentences not in accordance with the letter of the law but in
accordance with what is fair’.57 Oliver Goldsmith, writing in the 18th century
thought that ‘a spirit of mercy breathes through the law of England’58 and
apparently Thomas Fuller may have observed a few years earlier that ‘all are not
hanged that are condemned’.
The interface of the rules of criminal law and mercy and the conceptually
sequenced reasoning process is perfectly illustrated in Dudley and Stephens59
where sailors convicted of murder over one hundred years ago had their sentences
commuted to six months imprisonment by operation of the royal prerogative of
mercy. Here there is not only conceptual sequencing but institutional separation
which contributes both to the certainty and integrity of the rules of law and to
reducing the incidence of unacceptable outcomes in rare cases where the ap-
plication of a rule of law on its own terms delivers inappropriate results. Part of
the point of reconceptualising law as I seek to do is to open up to critical ethical
consideration its standard attributions of legally determined consequences to
legally defined conditions. This is informed in part by a notion that even the
best rules fall short of perfection and may be both over- and under-inclusive.
These defects entail that on some configurations of facts the application of the
rule according to its meaning will work injustice. One of the objectives of the
analysis offered in this paper is to provide the intellectual tools with which to
reduce the incidence of dubious outcomes without exposing the law to un-
acceptable levels of subversive wild-card discretion.
Is it appropriate that the defendants in cases such as Thornton60 (spousal
homicide in circumstances of sustained abuse, not amounting to provocation
nor diminished responsibility) or in Cocker61 (domestic euthanasia by agreement
with spouse) should serve life imprisonment, that is, about 10 years which is 20
times the murderous cannibals’ actual sentence or should, as I would argue,
some mercy be shown in such tragic cases? Certainly, use of the prerogative of
mercy seems to have withered away although some courts are now showing
considerably more interest in how Ministers exercise their very considerable
54
(1988) 165 CLR 543.
55
[1969] 1 AC 191.
56
[2000] 3 All ER 673.
57
Seneca (4 BC - 65 AD) De clementia Bk II ch III.
58
Citizen of the World (1762) Letter LXXX.
59
(1884) 14 QBD 273.
60
[1992] Crim LR 54.
61
[1989] Crim LR 740.
SUMMER 2001 Defeasibilism 367
discretion as in ex parte Bentley.62 In New Zealand, the Court of Appeal asserted
in Burt v Governor General that
. . . it must be right to exclude any lingering thought that the prerogative of mercy is
no more than an arbitrary monarchial right of grace and favour. As developed it has
become an integral element in the criminal justice system, a constitutional safeguard
against mistakes.63

Encouraging as such isolated insights are, what is problematic about some of


the less sophisticated versions of legal positivism is that they squeeze mercy and
like considerations out of the equation. Just as a narrow formalist positivism has
marginalized equity in the law of contract, and attempted to constrain it within
closed formulations, so it has almost totally eliminated the very idea of mercy
from the criminal justice system. Although conceptually sequenced and separated,
and institutionally distinct, mercy can be more readily understood as an integral
element in the criminal justice system on the basis of the third canonical form
which provides a conceptual niche, without exposing the criminal law to any
self-destructive uncertainty.
So I conclude this paper with the claim that my conception of law as open-
ended, defeasible, normative, conditional propositions is well illustrated and
supported by a scrutiny of the data of contract, tort and crime even though that
data must exhibit, to a degree difficult to assess, the causal effect of the intellectual
hegemony of closed, formalist positivism and whilst there is very much more to
be said about this conception of law (including its fit with ‘purpose’ in statutory
interpretation and ‘rights’ in constitutional adjudication), I hope that I have
sufficiently developed these ideas to justify the conclusion that lawyering always
involves the judgment either that the consequence normally attributed by law
to specified fact should or should not be overridden on the particular facts of
the instant case on grounds of equity, justice, policy, mercy, (purpose, or rights).
It seems to follow that lawyering necessarily involves a dispositive, decisional,
or judgmental element and that the legal formulation is always defeasible. What
is particularly important here is that the normal legal consequence attributed by
the legal norms in the event of the legally determined fact-cluster occurring is
defeated not by an exception (which can be incorporated by reformulation into
a closed legal norm) but by equitable and other override.
Devlin, more interested than Hart in the judicial process, distinguishes between the
rules of law and the justice of the case. He recognizes a tension between them,
recognizes that judges sometimes “stretch the law” to do substantive justice—and does
not disapprove of this practice, provided the judges do not acknowledge that they are
stretching the law. Devlin cannot admit a place in the concept of law for the moral
feelings that shape our responses to the ‘equities’ of a legal dispute.64

62
[1993] 4 All ER 442.
63
1992 3 NZLR 672 at 681.
64
Richard A. Posner, Law and Legal Theory in England and America (Claredon Press, 1996) at 17, citing Patrick
Devlin, The Judge (Oxford University Press, 1979) at 90–93.
368 Oxford Journal of Legal Studies VOL. 21

But whether or not a judge can ‘admit’ this in the sense of confession, legal
theory must acknowledge the presence of equitable and other overrides in legal
systems and, accordingly, must also acknowledge that only the third canonical
form is sufficiently comprehensive to accommodate the data, and in so doing it
must also ‘admit’, in the sense of permitting entry to, ideas such as justice and
equity as part of law.
In skeletal form, the conception of law as open-ended, defeasible, normative
conditional propositionscan be summarized as follows: If A (legally defined facts)
is, then B (legally determined consequence) ought to be, unless there is EITHER
(1) an operative exception, being (i) a known and established exception or (ii)
an exception yet to be established; OR (2) an overriding consideration, including
(iii) equity and/or justice, (iv) policy, (v) mercy, (vi) purpose, (vii) rights, or
(viii) a residual category of ‘damn good reason’65 or ‘compelling objection’.66
In a well-structured and mature legal system, many of the rules of law will
not be so substantially out of line with the contemporary requirements of justice
and the like as to call conditions of defeasance into play. Just as the bulk of the
law of contract complies in its application sufficiently with the requirements of
equity, so the bulk of the general law complies in its application with the
requirement of justice, policy, mercy, (purpose, or rights). Occasionally one
encounters perplexing circumstances in which the application of the rule leads
to great absurdity or gross injustice. In such circumstances the application of
the rule defeats the objectives of the rule.
There certainly are cases (of comparatively rare occurrence) wherein the specific
considerations balance or outweigh the general: cases which (in the language of Bacon)
are “immersed in matter”: cases perplexed with peculiarities from which it were
dangerous to abstract them . . .’.67

In such cases there is a choice. Either the rule rules or it is susceptible of override.
Closed formalist positivism tends to the first option. I advocate the second. In
so doing I am also advocating an approach to legal discourse which calls for
constant monitoring of outcomes against a range of side-constraints in order to
assess whether the legally mandated outcome is acceptable. With this extension,
or subversion, legal positivism can accommodate a conception of law as a locus
not only of traditional, closed rule-based reasoning, but also of open-ended
applied internal moral reasoning. The values deployed in the override phase are
not values external to the law but the deeper values inherent in the law which
the legal rules can only imperfectly capture and implement.

65
Mortimer R. Kadish and Sanford H. Kadish, Discretion to Disobey: A Study of Lawful Departure from Legal
Rules (Stanford University Press, 1973) at 22.
66
Bruce Chapman ‘Law, Incommensurability, and Conceptually Sequenced Argument’ 146 University of
Pennsylvania Law Review 1487 (1998) at 1527.
67
John Austin, The Province of Jurisprudence Determined and the Uses of the Study of Jurisprudence (Weidenfeld &
Nicolson, 1968) at 52.

You might also like