DATE DOWNLOADED: Sun Sep 15 16:46:27 2024

SOURCE: Content Downloaded from HeinOnline

Citations:
Please note: citations are provided as a general guideline. Users should consult their preferred
citation format's style manual for proper citation formatting.

Bluebook 21st ed.

Jane Stapleton, Law, Causation and Common Sense, 8 OXFORD J. LEGAL STUD. 111 (1988).

ALWD 7th ed.

Jane Stapleton, Law, Causation and Common Sense, 8 Oxford J. Legal Stud. 111 (1988).

APA 7th ed.

Stapleton, Jane. (1988). Law, causation and common sense. Oxford Journal of Legal
Studies, 8(1), 111-131.

Chicago 17th ed.

Jane Stapleton, "Law, Causation and Common Sense," Oxford Journal of Legal Studies 8,
no. 1 (1988): 111-131

McGill Guide 9th ed.

Jane Stapleton, "Law, Causation and Common Sense" (1988) 8:1 Oxford J Legal Stud 111.

AGLC 4th ed.

Jane Stapleton, 'Law, Causation and Common Sense' (1988) 8(1) Oxford Journal of Legal
Studies 111

MLA 9th ed.

Stapleton, Jane. "Law, Causation and Common Sense." Oxford Journal of Legal Studies,
vol. 8, no. 1, 1988, pp. 111-131. HeinOnline.

OSCOLA 4th ed.

Jane Stapleton, 'Law, Causation and Common Sense' (1988) 8 Oxford J Legal Stud 111
Please note: citations are provided as a general guideline. Users should consult
their preferred citation format's style manual for proper citation formatting.

-- Your use of this HeinOnline PDF indicates your acceptance of HeinOnline's Terms and
Conditions of the license agreement available at
https://heinonline.org/HOL/License
-- The search text of this PDF is generated from uncorrected OCR text.
-- To obtain permission to use this article beyond the scope of your license, please use:
Copyright Information
 LAW, CAUSATION AND COMMON SENSE
JANE STAPLETON*

INTRODUCTION

Causation is a topic of central importance in the understanding of legal thought

and case law. Yet relatively little emphasis seems to be put on it in law courses
and it is an area about which the average lawyer knows relatively little. As a
result law students, practitioners and academics often either grossly oversim-
plify or completely neglect causal issues. Those who delve more than super-
ficially into the area may abandon the exercise in exasperation at the apparently
intractable complexity of the subject-matter and principles involved. A con-
certed effort to state these principles as coherently as possible would obviously,
then, be a most worthwhile exercise. This is what Hart and Honor6's book
Causationin the Law,1 the second edition of which has recently been published,
seeks to do. It certainly contains much of value and interest to the general
reader. Here, for example, an undergraduate would find an explanation of such
mysteries as: why the damage deliberately caused by the Borstal boys in the
Dorset Yacht 2 case was taken to be the consequence of the Borstal officials'
carelessness; why a negligent motorist may be said to have caused the additional
injuries suffered by his victim when the latter receives negligent treatment in
hospital (an example of the 'ulterior harm' cases); why Wagon Mound P has not
had the dramatic impact on tort case law which some expected; and how the risk
and foreseeability doctrines relate to each other and to other doctrines advanced
as limits on the scope of legal responsibility. Likewise practitioners and judges
would find useful perspectives, for example on the increasingly important topics
of apportionment and onus of proof.
Nonetheless, although the book contains much of interest to the general
reader it fails to present it in a readily accessible form. Of course, Hart and
Honor6 do not claim to be providing an undergraduate introduction to the topic
of causation nor a practitioners' guide. Still, it is a shame that such a storehouse
of useful theory-written in an easy and elegant style-could not have been
structured in a way which would have facilitated easier access by this wider
audience. The problem is that in presenting such a large work-the book is over
500 pages long-on such a complex topic the authors have understandably been
concerned to expound the detail of and exceptions to their thesis. Unfortunately

*Fellow of Balliol College, Oxford.

1 Causation in the Law 2nd ed (Oxford, Clarendon Press, 1985) (1st ed 1959).
2 Home Office v Dorset Yacht Co Ltd [1970] AC 1004.
3 Overseas Tankship (UK) Ltd v Morts Dock & EngineeringCo Ltd [1961] AC 388.

111
@Oxford University Press 1988 Oxford journal of Legal Studies Vol. 8 No. I
 LAW, CAUSATION AND COMMON SENSE

the result is that the main channel of argument fails to emerge clearly, at least on
a first reading. From the general readers' point of view the principal drawback to
the second edition of Causation in the Law is that, despite a lengthy new
preface, the authors have not taken this opportunity to insert a summary of their
theory. The omission of this simple improvement means that the text remains
relatively inaccessible to the wider readership of lawyers generally. With this
criticism in mind and in order to make specific comments on the text more
readily understandable it will be useful to state in a short compass the basic
structure of the authors' thesis.

THE THESIS

Two aims
In the second edition Hart and Honor6 reiterate the two related aims of the first
edition published in 1959. The first of these (which reflected a dominant
philosophical interest of that earlier time, linguistic analysis) was to analyse the
usage of causal language in everyday life to see if some conceptual framework
could be constructed from such usage. This made obvious sense as a first step in
the consideration of causation in the law-and still does despite the relative
eclipse of linguistic philosophy-because courts often assert that it is the
ordinary person's concept of causation which is to be applied to the particular
case. With frustrating frequency, however, courts then tend to retreat into the
use of vague metaphors, such as 'breaking the chain of causation', to express
their reasoning. Hart and Honor6's initial task, therefore, was to see if the
ordinary person's 'common sense' notions of causation did provide a workable
approach to problems of causation in the law and if so to formulate the extent
and limitations of that approach.
In doing so they hoped to provide themselves with the means to achieve the
second aim: namely, to combat the theories of what they call the 'causal
minimalists' who take these obscure metaphors of the courts to indicate that it is
impossible to isolate any principles of causation except that of sine qua non; and
that, therefore, apart from the bare issue of sine qua non, all issues discussed
under notions of 'proximate cause', 'legal causation' or 'remoteness of damage'
are at least to some degree issues of legal policy. Hart and Honor6s goal was to
formulate the general rules underlying common sense notions of causation, and
in particular to show how these extend beyond the issue of sine qua non; and to
show to what extent the law adopts a similar approach.
Their basic thesis is that courts are correct to assert that it is common sense
notions of cause-rooted in the ordinary person's ideas of when it is fair to
punish or seek compensation-with which the law generally seems to be
concerned. They argue that, far from being a mere screen behind which courts
weigh policy considerations, causal metaphors are often simply an inept attempt
 JANE STAPLETON

to express stateable common sense notions of cause. 4 Hart and Honor6 do not
contend that isolating and stating these principles would provide a code by
which legal issues of causation could or necessarily should be decided,s but it
would, they argue, 'increase our understanding and powers of criticism of the
framework within which legal thought moves and . . . permit the clear
formulation of constantly recurrent factors which count, though not conclus-
ively, for or against decisions'. 6 In particular, in order to meet the claims of the
causal minimalists, Hart and Honor6 want as far as possible to isolate causal
questions from policy issues.

'COMMON SENSE' NOTIONS OF CAUSAL CONNECTION

What, then, are these common sense notions? Early chapters isolate three causal
connections which these notions seem to centre upon. The first central concept
is that of cause and effect, of 'causing' an effect (e.g. of 'doing' or 'causing harm')
by one's own act. (Hereafter I will call this concept 'causal connection (i)'.)
Among the factors 'but for' which an effect would not have occurred, only those
which interfere with the ordinary course of events and thereby account for the
result seem to qualify as 'causes' in ordinary speech. The others, such as the
presence of oxygen in a forest fire, are mere conditions. According to Hart and
Honor6, ordinary common sense notions of causation identify both deliberate
(or what they term 'voluntary') human acts and abnormal contingencies 7 as the
sort of 'but for' factors which may qualify as causes in this way. Moreover,
deliberate 'voluntary' human acts and abnormal contingencies intervening
between the event and what might otherwise be characterized as the cause are
identified as factors which negative this central type of causal connection. 8
Hart and Honor6 argue that in ordinary speech the verb 'to cause' and the
noun 'a cause' are usually only employed in this first sense. But they acknowl-
edge that the concept of causation as used in ordinary life is not a unitary one,
confined to this central case. They describe, for example, a second type of causal
connection found in ordinary speech-that of X 'providing reasons' for Y to do
something. (Hereafter I will call this concept 'causal connection (ii)'.) In such

4 Pp xxxiii-xxxiv, xlix, 33, 73-4, 96-8, 131.

5 In the second edition the authors emphasize that they do not argue that questions of responsibility
should be settled solely by reference to causal criteria (the approach of causal maximalists like
Epstein). They acknowledge and accept the role of policy but seek to separate out its influence for
the sake of clarity: xxxv, xlii-lv, lxxvi, 306-7. (For Epstein's theory see citations on lxxiii of Hart
and Honor6.)
6 P3.
7 Including non-deliberate human acts which were so unusual an interference with the ordinary
course of events that they explain the result.
8 P 71. Hence the notion of 'breaking the chain of cause' etc because they break the analogy with the
simplest case of 'doing harm' where all intermediate factors satisfying the but-for relation are mere
conditions, 73.
 LAW, CAUSATION AND COMMON SENSE

'interpersonal transactions' this may amount to no more than advice, but when it
amounts to 'inducing' Y to act there are often direct analogies in rules of legal
responsibility.
A third class of causal connection is where X provides Y (or another factor)
with the opportunity or means necessary to do something. (Hereafter I will call
this concept 'causal connection (iii)'.) In this class of cases, which Hart and
Honor6 call 'occasioning' harm, X may act unintentionally, for example where
X carelessly leaves the door of a friend's house unlocked, thereby providing the
9
opportunity for a thief to burgle the house.
What does all this mean for the law? It is clear that legal responsibility does
not necessarily entail that harm has occurred (e.g. inchoate crimes) or, where it
has that there is some type of causal connection between that harm and conduct
of the person responsible (e.g. vicarious liability of employers). But a very
common ground for legal responsibility (with or without added grounds such as
fault) is that there is a particular causal connection between the defendant's
conduct and harm. 10 Where this is so, the authors argue, the case law reveals
that over a major part of the law courts have sought to apply notions of causal
connection which draw their meaning and force from common sense notions of
causation. 11 Thus where the legal rule of attribution of responsibility invokes
the requirement that X 'cause' harm (i.e. causal connection (i)) Hart and
Honor6 argue that courts emphasize subsequent intervening deliberate 'volun-
tary' human acts or abnormal coincidences as negativing or 'breaking the chain'
of the requisite causal connection. In other words events subsequent to these
intervening factors are not attributed to the antecedent conduct of the defendant
as its consequence even though these subsequent events would not have
occurred but for that conduct."1 In short, courts seem to adopt the common
sense notions of causation reflected in causal connection (i).

Causalminimalism
The review of the case law which Hart and Honor6 carry out to establish their
first aim is also used to advance their second aim: to argue that the cases do not
support the view of some modern American writers-the 'causal minimalists'-
that the threshold question of sine qua non ('would the harm have occurred but
for the defendant's conduct?') is the only factual question in the area called 'legal
causation'. According to these writers, after that question is dealt with, the
remaining issues of 'legal causation', 'proximate cause' or 'remoteness of. damage'
(even if expressed in terms of causation), are inextricably tied up with questions
of policy: when should the defendant be liable. The idea of the minimalists is
that the limits on liability imposed in addition to sine qua non are those required

9 P 59. See n 27, below.

10 P 66.
11 Pp xxxv, 130.
12 P 133.
 JANE STAPLETON

by the scope, policy or purpose of the particular legal rule. The argument is
two-fold: first that this is how, despite recourse to causal language, courts in fact
do decide such issues; and secondly, that this is how courts should decide them,
either by exercising the sort of case-by-case judicial discretion advocated by
Green, or by application of a policy formula such as risk or foreseeability. 13 The
minimalists support their first argument by pointing to the courts' use of
apparently incomprehensible causal metaphors and the fact that many courts
treat legal cause as a unitary issue and one influenced by policy.
Hart and Honor6 on the other hand, while conceding that the minimalists'
emphasis on policy has some limited advantages, 14 argue that once the threshold
question of sine qua non has been answered in the affirmative, 5 the remaining
issues of legal causation can be separated in two distinct issues, one of causation,
one of policy. These different techniques for limiting liability are, they argue,
radically different from one another. 16 First there are factual issues of causation:
given the sine qua non test is satisfied, is the causal connection required by the
relevant legal rule present or not? The determination of this issue is, according
to Hart and HonorS, influenced by factors derived from common sense notions
of causation and not by inventions of legal policy dominated by the judges'
arbitrary sense of fair play. 17 Secondly, there is the question of whether, as a
matter of policy, the law ought in the particular case to enlarge or restrict
liability independently of causal connection, by the use of what the authors term
'scope rules'. 18
An example can be used to illustrate the dichotomy Hart and Honor6 put
forward. Say X negligently starts a fire and as it is about to flicker out Y
deliberately fans the flames so that it destroys neighbouring houses. Here X will
not be held liable for the damage even though it would not have occurred but for
X's act. 19 According to Hart and Honor6, this is because the remaining causal
question of fact after sine qua non 'was X the cause of the damage?' (the criteria
for answering which reside in common sense notions of causation)-has been
answered in the negative. Compare this case to one where, for example, X
negligently starts a fire in New York which burns down a line of houses. In New
York it is a legal ('scope') rule that if a fire negligently started spreads to
buildings, damages may only be recovered in respect of the first of the
buildings. Here again there will be no liability (except for the first house) but
this is because of a question of legal policy, a question of law.
13 Pp 5, 131. Green's theory can be formulated as a loose type of risk theory, 285. (For Green's
theory see citations on 291-9 of Hart and Honore.)
14 P 109.
15 And see Chapter 5 on why sine qua non usually pinpoints causally relevant factors.
16 P 307.
17 Even in borderline cases, 92.
18 On how these are formulated see 304-7.
19 Assuming nothing but satisfaction of causal connection (i) will satisfy the legal rule. Compare
Ioccasioning harm' below.
 LAW, CAUSATION AND COMMON SENSE

Hart and Honor6 argue, therefore, that a division of issues of 'proximate' or

'legal' cause into these two classes, one causal and the other non-causal and
policy-related, is possible and useful in analysing case law. Moreover, it is
preferable as a matter of principle. An understanding of the principles of legal
responsibility requires causal limitations (to a great degree the same in many
areas of law) to be separated from scope rules which will vary from area to area
according to the particular policy issues thought relevant. 2 0 It is particularly
important for the practitioner where jury trial requires the division of the issues
between judge and jury. More generally it is important for the theorist to
21
separate them where the legal rule itself merges the two.
The authors see the metaphors to which the causal minimalists object as
simply an inept attempt by judges to express common sense notions of cause
which had hitherto not been adequately formulated. Similarly they explain why
some courts appear to treat legal causation as a unitary question: although a
legal system may distinguish the two types of legal causation question-for
example, in negligence the duty of care issue can serve to separate off some
policy issues from the causal connection issues-the two may remain fused in a
particular legal rule. For example, if a legal rule simply says the defendant is
liable if he causes harm, the courts will want to be able to limit liability in
particular cases for obvious policy reasons.
If courts wish to do justice, and yet not openly flout the authority of rules stated in bald
causal terms, one expedient is to take such matters into account under the heading of
causation and to extend this term to cover the limitations of policy thought desirable ...
Inevitably if this is done the colour flows back: questions of policy are called questions of
'proximate cause' and questions of proximate cause cease to be distinguished from
22
questions of policy.

Evidence from case law

The authors analyse the case law and, following most writing on legal causation,
they focus particularly on the law of torts. 23 They attempt to show how the case
law is consistent with the three-step analysis they suggest (sine qua non, causal
connection, scope limitations) and in particular how the outcome of the cases in
general accords with that suggested by common sense notions of causation,
except where policy factors intervene. In an impressive section Hart and
Honor6 are able to show that over a very wide area of tort law where an element
in responsibility is the central type of causal connection, 'the decisions of the
courts have been controlled by the principle that this connection is negatived if
the factors required, in addition to the wrongful act, for the production of the

20 Pp 132, 305-7. See also pp xxxv-xxxvi.

21 Pp 132, 429.
22 P 94.
23 Pp 84, 308. And this review article also concentrates on this area but see Chapters 11 (contract),
12-14 (criminal taw) and Part III on continental theories of causation.
 JANE STAPLETON

harm include a voluntary human action on an abnormal occurrence'., 4 This is so

even though the language of other factors such as 'foreseeability' is used.2"
Their thesis appears at its most powerful in explaining the outcome in the
so-called 'ulterior harm' cases. These are cases where the requisite causal
connection between the defendant's wrongful conduct and subsequent harm is
held not to be negatived by an intermediate third factor which also satisfies a
but-for relation to the harm. For example, X negligently swerves his car onto a
footpath; Y jumps out of the way and in doing so falls into a trench injuring
himself; Y attempts to rescue someone in X's path and is injured; or Y is hit
directly but suffers added injuries due to negligent medical treatment. In such
cases courts consistently hold X liable for the ulterior harm. The problem with
trying to explain this outcome in terms of merely the scope or policy or purpose
of the legal liability rule which made X's initial conduct wrongful is that legal
rules can rarely be so detailed as to provide an answer to the question of whether
such ulterior harm was within the scope of the rule. I will return to this below.
Hart and Honor6's theory seems at first sight to fit these cases well. They
argue that these cases reflect situations where particular factors prevent the
intervening conduct being considered by common sense as fully 'voluntary' or in
other words as negativing causal connection (i) in ordinary speech. According to
the authors, while common sense recognizes that the free, deliberate and
informed (i.e. what they call the 'voluntary') conduct of a person intended to
exploit the situation created by the defendant negatives causal connection (i),
acts prompted by, for example, self-preservation or moral obligation (e.g. to
rescue), or the negligent acts of a third party,2 6 are not sufficient to negative
causal connection (i), either in common sense or over a considerable area of tort
law.

Occasioningharm: risk orforeseeabilityas guide to the imposition of liability

Despite Hart and Honor6's emphasis on the causal connection of 'causing harm'
and how it seems to be negatived by intervening voluntary human acts and
abnormal occurrences, they have to concede that there are a number of cases
where the law does not recognize such factors as negativing the relevant causal
connection between the defendant's conduct and the harm. It is here that the
argument of the causal minimalists that the best explanation of the cases is some
form of policy seems strongest, be it in the form of arbitrary decisions by judges
as to what is fair or some formula such as risk or foreseeability. Hart and Honor6
examine these cases (which they divide into cases of causal connection
(ii)-interpersonal transactions-and causal connection (iii)-occasioning
harm) and refine their theory in order to counter the claims of the minimalists.
In particular they seek to counter the major proposed substitutes for causal tests
24 P 133.
25 P 284.
26 Unless they are so grossly reckless or negligent as to do so on the grounds of abnormality.
 LAW, CAUSATION AND COMMON SENSE

to limit responsibility, namely foresecability and risk, which at first appear to be

much more workable explanations of the outcome in these cases than 'common
sense causation'.
The most illuminating arena for this conflict of theories appears to be
provided by cases of occasioning harm. In negligence (and breach of statutory
duty) it is becoming increasingly recognized by courts that in certain types of
cases liability may be based on the fact that the defendant negligently provided
an opportunity for other persons to do harm deliberately or for coincidental
natural phenomena to do harm.2 7 For example, a house guest leaves the front
door of the house unlocked. Since the opportunity so provided is one commonly
exploited by thieves it is now recognized that there may be a legal duty not to
provide it. Here Hart and Honors acknowledge that once sine qua non is
satisfied, the attribution of consequences in causal statements is controlled not
by the notion of intervening voluntary acts or abnormal events but only by the
notion of the risk created by the wrongful act which causes the act to be
considered wrongful .8 One could call this 'occasioning harm within the risk'.
Thus the fact that the thief's act was an intervening deliberate voluntary act
does not prevent liability; after all, it is the very sort of contingency which made
the house guest's conduct wrongful. On this basis it is clear why the intervention
of the deliberate acts of the Borstal boys in the Dorset Yacht case did not
negative the causal connection between the official's negligence and the damage.
In all these cases the imposition of liability can be clearly predicted by the risk
theory. According to the more stringent form of that theory, liability should
extend to those types of harm the risk of which formed a reason for the
imposition of liability, i.e. to all 'harm within that risk'. In the case of the tort of
negligence, where it is agreed that foreseeability of harm is relevant to whether
the defendant was 'negligent in fact', this risk theory is expressed in terms of
foresecability: that liability in the tort of negligence should extend to all harm
the foreseeability of which formed the reason for the imposition of liability in
negligence.2 9 In other words, in a negligence action where a voluntary act or
abnormal occurrence has intervened liability may extend beyond the harm
which the wrongful conduct would, on common sense principles, be said to
have 'caused' (that is, beyond causal connection (i) above), to encompass certain
foreseeable harm of which the defendant's conduct was a sine qua non.
Hart and Honor6 deal with this by recognizing occasioning harm as a species
of common sense causal connection, albeit a less demanding one than 'causing
harm' (i.e. causal connection (i)) and one which in particular limited circum-
stances has begun to be recognized as sufficient to ground liability. Although the
defendant cannot be said to have caused the harm, he occasioned it and this

27 Pp 133, 194.
28 P81.
29 P 286.
 JANE STAPLETON

causal connection is sufficient for liability in certain circumstances because the

harm which occurred (e.g. the theft by the burglar) was the very sort the risk of
which made the defendant's conduct wrongful according to some legal rule.
Hart and Honor6 concede that the risk (or in negligence, the foreseeability)
terminology in which the judgments in such cases are often couched explains the
imposition of liability in these cases well. Nonetheless, they argue that on
looking at the negligence case law the conclusion must be that while there are
particular instances of a duty to guard against certain voluntary interventions of
others, and although this ground of liability may be becoming more common, it
is by no means a general principle of liability. 30 It follows that the risk and
foreseeability doctrines do not provide a general guide to the imposition of
liability. 3 1 While admitting that no single principle of causal connection can be
enunciated to explain all the circumstances where liability for negligence is
imposed, Hart and Honors argue that if there is a central case at the present
time in the case law it is not that of 'occasioning harm' but that of 'causing harm'
(including ulterior harm which the risk and foresceability doctrines have
difficulty explaining-see below). 32 Here the courts apply causal notions
according to which subsequent voluntary acts (even if foreseeable) and
abnormal events negative liability.
Hart and Honor6 also reject the argument that the risk (and foreseeability)
doctrines, even if they do not do so yet, should provide a general basis of tort
liability. They do so on the basis of fairness. 33 Where the harm would not have
resulted from the defendant's conduct but from the voluntary intervention of a
third party, it is one thing to allow recovery where the possibility of such
intervention, even if amounting to a criminal act, was sufficient to make the
defendant's conduct wrongful (e.g. the risk of theft by a burglar in the earlier
example), quite another to allow it where there is not the relation (e.g. where, in
the above example, the burglar commits arson). They argue that the latter
regime, which recognizes a general liability for creating the opportunity for
others deliberately to do harm (of any sort), is contrary to our intuitive sense of
fairness. This, they say, dictates that a wrongdoer is entitled to shift responsi-
bility to a subsequent voluntary agent who exploits the opportunity created by
the wrongdoer even if this exploitation is foreseeable, unless the risk of such
exploitation was what made the initial act wrongful. The argument seems to be
that ordinary notions of fairness confine the latter cases to instances of specific
30 P 281. They argue that in at least one case the House of Lords based their decision on this very
point: that in negligence a wrongdoer may not be responsible for all the foreseeable harm of
which his conduct was a necessary condition because satisfaction of causal connection (iii),
'occasioning harm', is not sufficient for liability in the particular case and because liability based
on 'causing harm' (i) was negatived by an intervening voluntary act, albeit a foreseeable one
(283). But see n 57 below.
31 P 204.
32 Pp 204, 284, 289.
33 Pp 276-7.
 LAW, CAUSATION AND COMMON SENSE

and finite types of harm (e.g. theft by a burglar) and do not extend to cases of
making an act wrongful because of the opportunity it creates for the infliction of
any harm. Hart and Honors reject as unfair the isolated cases which appear to
adopt such a basis of liability and oppose any future extensions of such
open-ended liability for occasioning harm. 34 It may, incidentally, be worth
noting at this point that it is dubious whether many risk or foreseeability
supporters advocate liability of the wide sort which Hart and Honors reject
here.

Risk orforeseeabilityas a guide to limits of liability: the problem of ulterior

harm
Hart and Honor6 accept that the risk (and, in negligence, foreseeability) theory
accords with the cases in as much as it predicts the imposition of liability if the
harm is 'within the risk'. This is so despite the possible presence of factors such
as a 'voluntary' human act which would negative the central type of causal
connection (i) (the burglar stealing example). But, relying particularly on the
phenomenon of recovery for ulterior harm (where the defendant is liable despite
intervening 'non-voluntary' human acts) they reject the second limb of the
theory. This asserts that the extent of recovery is limited to harm the risk of
which was among the reasons for holding the conduct wrongful. 35 What this
second limb translates into in the case of negligence is that foreseeability of harm
is not only relevant to whether the defendant was 'negligent in fact' (i.e.
culpability) as everyone concedes, but also the same test of foreseeability
governs the extent of responsibility once negligence in fact is shown (i.e.
compensation). This was the theory propounded by Goodhart and apparently
accepted by the Privy Council in Wagon Mound I. It has the superficial
attraction that it draws no distinction between culpability and compensation-
both are to be decided by the same test: foreseeability. The defendant is
responsible for all foreseeable harm but only for harm that is foreseeable. This
view is at variance with the traditional view-still advanced by Hart and
Honor6-that while foreseeability is relevant to culpability in negligence, it does
not govern the extent of the defendant's responsibility. This depends, prin-
cipally, they argue, on whether the harm for which recovery is sought was
caused by his wrongful conduct and is of a type which is within the scope of the
rule. Recovery for unforeseeable harm is not barred on this view.
Hart and Honor6 attack the Goodhart view on a number of grounds, the
strongest of which is that it does not in fact provide the common test for
culpability and compensation that is claimed for it (despite the approach of

34 E.g. cases allowing recovery against a negligent motorist for the theft of a wallet from his
unconscious victim by a third parson, simply on the basis that it would not have occurred but for
the defendant's negligent conduct and the harm was foreseeable, 281. See also 284.
35 Although they concede at least one way in which the limiting aspect of the theory can be useful,
262.
 JANE STAPLETON

some courts) because the claim rests on an ambiguous use of the term
'foreseeable'. 36 Culpability depends on many practical factors of which fore-
seeability of harm is merely one. These other factors, such as the utility of the
defendant's conduct, influence the degree of foreseeability required for liability
in the particular case. So, the defendant is liable for harm which is 'foreseeable'
in a practical sense, i.e. such that a reasonable person would take precautions to
avoid it in the context of these other factors. Now, as Hart and Honor6 argue, it
is conceivable that a legal system might restrict the responsibility of the
negligent defendant to harm which is foreseeable in this practical sense (i.e.
have a common test); but this is clearly not how the law of negligence currently
operates because it indisputably extends to cases of ulterior harm. Here the
plaintiff recovers for harm the risk of which was not a reason for holding the
defendant's conduct negligent (so the risk theory would not predict recovery for
this harm 'outside the risk'). 37 In foreseeability terms, the plaintiff recovers for
harm which was not necessarily 'foreseeable' in the practical sense used to decide
the issue of culpability.
One area where this restrictive limb of the foreseeability doctrine-i.e. no
recovery for unforeseeable harm-has been applied, and for which Hart and
Honor6 give some support, is in cases of initial harm where the causal processes
which have occurred are of a radically different type from those reasonably
expected. 3 8 For example, fire, rather than mechanical damage, results from the
impact of a falling plank. Hart and Honor6 argue that this is virtually the only
area where the limiting limb of the foreseeability doctrine has been used to
truncate liability. In other cases, they argue, the doctrine of an identical test of
foreseeability for both culpability and compensation apparently accepted in
Wagon Mound I has not been adopted. 39 Hart and Honor6 argue that once
culpability is established according to the requirements of the particular legal
rule (including cases where this is satisfied by the 'occasioning of harm with the
risk'), the extent of liability is generally limited not by notions of risk but by the
factors negativing the central type of causal connection (i). In other words it
extends to all harm (including ulterior harm) which the defendant's conduct was
sufficient to produce without the intervention of voluntary human acts or
abnormal contingencies. But even where liability is based on causal connection
(iii), if a voluntary act intervenes other than that which the defendant had a duty

36 Pp 262-6. They also ask the simple and obvious question of why, merely because negligence
presupposes the likelihood of certain harm, it should follow that the responsibility of the
defendant must be limited to harm of that sort.
37 Pp 266 n 47, 288.
38 Pp li, lxvii, 269-270, 274. It would be hard for them to argue against such an outcome since
liability for freak results hardly accords with common sense notions of cause, see Williams [1961]
CL] 62, 82-3.
39 P 256, n 10.
 LAW, CAUSATION AND COMMON SENSE

to guard against (e.g. the burglar commits arson), the defendant is not liable for
the resultant harm.
To sum up: Hart and Honor6 acknowledge that, as a test for explaining the
imposition of liability in cases where causal connection (i) is not satisfied, the
risk and foreseeability doctrines can be supported both in principle and from
certain cases. But even here they are of limited usefulness because 'occasioning
harm within the risk' is not yet anything like a general or even a central ground
of liability. Moreover, as tests for limiting responsibility the doctrines are
inadequate because they do not adequately explain cases of recovery for ulterior
harm. They are therefore not useful as a guide to general principles of legal
causation as they operate in the cases, nor do they provide, Hart and Honor6
argue, an adequate substitute for the more centrally important factors isolated
40
from an analysis of common sense notions of causation.

SOME CRITICISMS

The Preface
Having traced the bare bones of the Hart and Honor6 thesis it is now possible to
make a few comments. The lengthy (48 pages) Preface to the second edition
makes no attempt to sketch the authors' thesis (a serious drawback). Instead it is
principally concerned to discuss the major contributions to the causation debate
since the first edition, including a succinct yet powerful overview of the
approach of economic theorists. 4 1 The Preface does, however, attempt to clarify
one important dimension of the theory, and this is accomplished well. 42 The
first edition had been criticized for placing too little emphasis on the role of
policy as a factor in the decisions of courts. The authors therefore in the second
edition set out the four points at which they recognize that legal (legislative or
judicial) policy can impinge on the determination of causal issues as they affect
legal responsibility.
First, policy determines the choice of the grounds of legal responsibility. The
decision to include a causal connection requirement, the form that requirement
takes, and therefore what causal issues arise in the particular case are matters of
policy. Second, the fixing of scope rules truncating liability more narrowly than
that produced by mere satisfaction of such a designated causal requirement, if
any, is also a matter of legal policy. Third, the formulation of causal issues,
merging them with non-causal issues under headings such as remoteness of
damage, may result in the infection of causal issues with policy issues in lawyers'

40 P 204.
41 For citations to the work of these contradictions see Hart and Honor6: Mackie (at xxxvii), Becht
and Miller (xxxv), Malone (lvii), Keeton (lxii) and Williams (lxvi), and for those in the law and
economics debate, Calabresi (at lxxi), Posner (lxxv), Epstein (lxxiii).
42 Pp xlii-lv.
 JANE STAPLETON

thinking. Finally, rules about the incidence of the burden of proof and
presumptions may embody legal policy choices where, for example, evidence
one way or the other on a point of causation is unavailable because they will then
be determinative of that issue. Much of the explanation of these points is
illuminating and it is a pity they were not also worked into the text of the book
more thoroughly.

'Common sense' usage and empiricalevidence

As to the revised text of the book itself, clearly 'common sense notions of
causation' remains the pivotal idea. But what does it mean? Hart and Honors
acknowledge that such notions are not hard-edged and may not provide clear
answers in borderline cases. 43 Yet despite this they assert that there is a 'central
core of commonly agreed meaning' 44 which they go on to enunciate and analyse
at length. The somewhat surprising aspect of all this is that, despite their
emphasis on ordinary linguistic usage and the 'common sense' views of the
ordinary person on causation, the authors pay little, if any, attention to
empirical work concerned with these phenomena. 45 They simply state that 'the
ordinary person' uses words in such and such a way, according a particular
causal connection in such and such circumstances, as if these were established
facts.
At a practical level, the reader will often find the assertions easy to accept, but
the technique is deceptive: is it really the case that when a defendant negligently
injures V (say by breaking his arm) and V later suffers ulterior harm (say,
paralysis) due to negligent medical treatment he receives for the broken arm, the
ordinary person would say the defendant 'caused' or was 'a cause' of the
paralysis? 46 Is not this assertion dubious at best? In fact, could not most cases of
ulterior harm, which play such an important role in Hart and Honor6's
argument, be placed in the borderline area where 'common sense' notions of
causation are not at all self-evident? Moreover, there are unequivocal examples
of judicial usage which are at odds with the authors' assumptions about common
sense causal usage. For example, in Dorset Yacht Lord Reid describes the Home
Office as 'causing' the loss produced by the deliberate voluntary acts of the
Borstal boys. It is very surprising that no attempt is made to deal with these
counter examples to their thesis.

43 Pp 68, 75, 92. Although they argue that even in these cases courts are not forced back onto pure
policy or discretion but can and do refer in their analysis to the degree to which the borderline
case differs from standard forms of causal connection recognized by common sense: 92, 165-6.
44 P 92.
45 For a review of psychological research on how people perceive causality see S. Lloyd-Bostock
(1979) 42 MLR 143.
46 Yet this is what the authors imply since they classify such negligent acts of third parties as
'involuntary' (152) and argue that in common sense notions such acts do not negative the causal
connection of 'causing harm', 44.
 LAW, CAUSATION AND COMMON SENSE

Bald assertions of what the ordinary person recognizes as causal connection

are also objectionable in theory. The authors do not provide a discussion of the
work of social psychologists who have attempted to examine empirically the
attribution of causal connection by ordinary people. We do not, for example, get
a discussion of modern empirical findings that the ordinary person's notions of
when another person has been at fault and when someone should be liable to pay
compensation can be profoundly affected by the pattern of existing effective
liability rules. 47 Could the same be said about the ordinary person's attitudes to
causal connection? Are these also liable to be influenced by changing patterns of
legal liability? This is not as fanciful as it might at first sound. Hart and Honors
themselves concede that the pivotal notion of a 'voluntary' human act (which,
they argue, according to common sense negatives the central causal connection
of 'causing harm') rests on evaluative issues. They say that these reduce to a
policy judgment: did the value of the interest threatened justify the risk the
actor took? 48 If yes-as in cases of rescue or medical assistance to the victim of
initial harm-then the intervention does not negative the connection of 'causing
harm'. Hart and Honor6 would classify the act as 'involuntary'. Yet it is not
difficult to think of possible situations where a change in the pattern of effective
legal liability might have an effect on this 'common sense' judgment. In other
words, because the change incorporates a different valuation of the intervening
act, ordinary people might in time adapt their 'common sense' valuation with a
concomitant change in their notions of causal connection. Of course, this raises
the even wider unanswered question of how dynamic common sense notions of
causation may be, and how they change, if at all. The point to be made here is
that whatever value the empirical work of psychologists on ordinary people's
ideas of cause may or may not have for the law, it seems a sufficiently important
and central issue to be raised by Hart and Honor6.
The mention of the central notion of a 'voluntary' act leads on to a related and
major difficulty with the authors' thesis. They stress the desirability of keeping
the causal issues of proximate cause separate from the scope rules with which
they are frequently merged. Their resultant analysis is convincing in the case
where liability is truncated by an intervening abnormal occurrence, but it is
flawed in the case of intervening human acts. This is because, although the
authors choose to divide such acts into the relatively objective sounding
categories of 'voluntary' acts (which negative causal connection (i)) and
'involuntary' acts, they define these terms in a very odd way. 49 Thus the
definition of 'involuntary' is so wide that it covers virtually all but deliberate
acts. For example, it covers acts done under moral or legal obligation, with the
result that the intervention of a doctor to treat the victim of a road accident, for
47 Lloyd-Bostock (1979) 42MLR 143; Lloyd-Bostock (1980) Insur L 331; Lloyd-Bostock (1981)
20 BrJ Soc Psych 161.
48 P 157.
49 P 41 esp n 12; 136-8.
 JANE STAPLETON

example, is classed as 'involuntary'. Quite apart from the fact that this is a
remarkable departure from ordinary usage for authors committed to analysis of
the accurate use of plain language, it serves to disguise the fact that the division
between those acts which negative causal connection (i) and those that do not
appears, as we have seen, to be value-based, that is, to depend on the evaluation
of the interest served by the intervening act.
If this is so, and if policy enters into this central notion of what is or is not a
'voluntary' act, then the authors' desire to keep the causal and scope realms
separate appears much more problematical. The dilemma this presents to the
authors is accorded only a slight mention, and their attempt to resolve it does
50
not adequately meet the criticism and is not convincing. One is left with the
uneasy feeling thereafter that the superficially workable pivotal test of, 'did a
voluntary human act intervene?', is in fact an artificial vehicle for ex post facto
rationalization. The definition of 'voluntary' is narrowed from its ordinary
meaning in order to fit the cases and in particular in order to exclude the acts
involved in cases where recovery for ulterior harm is allowed. It would have
been better to use a more descriptive word and to have admitted the degree to
which certain policy considerations infect the first question of proximate cause,
thereby providing an area of overlap with the second ('scope') question of
proximate cause where these and other policy factors operate.

A central inconsistency of terms

While readers may accept that authors committed to the accurate use of
language may yet, in defining a term, depart from ordinary usage for conceptual
reasons, it is a much greater surprise to find internal inconsistencies of usage. In
Causationin the Law there appears to be a major inconsistency of usage which
dulls the analysis at many crucial points. From the sketch of the authors' thesis
given earlier it is clear that the notion of 'occasioning harm' is a crucial
battleground for the proponents of the risk (or foreseeability) doctrines. In
refuting their claims, however, Hart and Honor6 fail to make clear whether they
see this concept as a form of causal connection or not. The confusion stems from
their inconsistent use of the terms 'causal' and 'causal connection'. Initially,
those terms are used to describe not only the central notion of 'causing harm' but
also the concepts of 'inducing' or 'occasioning' harm.51 Yet later, in describing
situations of 'occasioning harm', Hart and Honor6 repeatedly assert that in this
context the risk theory has supplanted older notions of causal connection so that
liability extends 'beyond the point at which the older causal tests would cut it
off', 'dispensing with causal tests when the harm that occurs is within the risk

50 They admit the distinction is an evaluative question but stress that it is not open-ended and is
anyway one that is mirrored in the reasons underlying common sense usage of causal terms (i.e.
the idea the doctor or rescuer 'had no choice'), 156-7.
51 See e.g. 62, 81. See also 195. This is the usage adopted in this article.
126 LAW, CAUSATION AND COMMON SENSE

which led to the imposition of liability'. 5 Thus, despite 'the intervention of a

free voluntary act or of an abnormal or coincidental event which would on causal
principles negative responsibility, [the] defendant is still ... responsible for the
harm'. 53 This only makes sense if the authors have at these points switched to a
second narrower meaning of 'causal notion'/'causal connection', one that is
confined to the causal connection of 'causing harm'. This inconsistent use of
these central terms emerges clearly in a few places from the context, 54 but in
many more the ambiguity is confusing and damages the flow of the argument.
The description in the Preface of bases of responsibility such as occasioning
5s
harm as 'quasi-causal' only adds to the problem.

The discussion of case law

A major feature of the first edition of Causationin the Law was the authors' use
of a wide selection of case law to illustrate and support their text. In the second
edition, they can claim to have greatly strengthened this aspect by the inclusion
of a wealth of new cases. Not only is there a discussion of major landmark cases
such as Wagon Mound I but also of 'novelty' cases such as the 'market share'
cases. 56 Unfortunately there are occasions where the reader's confidence in the
accuracy of such citations is shaken. The following are examples drawn from the
57
cited case law on multiple causation, although there are others in the book.
In discussing cases of 'additional causation', such as that where A and B
simultaneously shoot at C, each shot being sufficient to kill C, Hart and Honor6
argue that each act is properly described as the 'cause' of the death so that both
A and B are criminally and civilly liable for C's death. They then cite as an
example from the case law, Basko v Sterling Drug Co, 58 in which case we are
told the 'plaintiff took two drugs, each sufficient to damage his retina and
defendant was responsible for failing to warn of the danger of one of them, it
was held that he could be held liable for the whole damage'. 5 9 Clearly Basko
could only qualify as an example of additional causation if one makes the added
assumption that the retina was ruined and that each drug was sufficient to
achieve that result. If this was not the case, then Basko represents quite a
different type of case-multiple cumulative causation-like cases of dust

52 See 6-7, 289 resp. Other clear examples include xliv, 136, 162 and 208.
53 P 276.
54 See e.g. 'causal connection' on line 2, 133, must bear this narrower meaning, see following
paragraph. See also lines 21-3, 202, lines 20-26, 283.
55 P lxxx.
56 Pp liv, 102, 424.
57 A particularly noteworthy example of which is the important discussion of McKew's case (282-3)
which seems internally inconsistent., If the conduct of a plaintiff negatived causal connection
how could apportionment have been a more appropriate outcome as suggested?
58 (1969) 416 F. 2d 417.
59 Pp 235-6.
 JANE STAPLETON

60
diseases of the lung where each exposure to dust adds to the disease.
An example of the latter is Bonnington Castings v Wardlaw where the victim's
pneumoconiosis was caused by the cumulative contributions of dust from both
'innocent' and 'guilty' sources. 61 The narrow ratio of that case was simply the
reiteration of the traditional point that the burden of proof on the issue of
causation is on the plaintiff or pursuer. He must prove that the guilty source had
caused at least some material damage (i.e. more than de minimis). The point
was unremarkable. What is remarkable is the misplaced reliance placed on the
case by the House of Lords in McGhee v National Coal Board,62 a case not of
multiple cumulative sources of damage but of multiple possible sources. Here
the victim contracted dermatitis from industrial dust. There were two relevant
sources of contact with the dust, exposure to the dust while at work (an
'innocent' source because it was not the subject of a successful claim of
negligence against the employer/defendant) and exposure to the dust caked on
his skin while he travelled home (a 'guilty' source because it had been negligent
not to provide washing facilities after work).
Now had the medical evidence been that the dermatitis, like the
pneumoconiosis in Bonnington Castings, was a cumulative disease with each
exposure to dust contributing to the disease the case would have raised
comparable causation issues to those in that earlier case. But the evidence was
that the disease may have resulted from a 'triggering' incident of exposure and
there was no way of showing that it was more likely than not to have been
triggered by the guilty source of risk. There was no way, therefore, for the
pursuer to prove on the balance of probabilities that the guilty source had
caused the damage-i.e. that is was a sine qua non of it-and therefore on the
basis of Bonnington Castings itself, he could not win. But the House of Lords,
citing that earlier case as support, held that by creating the guilty source of risk
the defendants had materially contributed to the risk and this was sufficient to
discharge the pursuer's burden of proof. While Hart and Honor6 correctly state
the ratio of McGhee in terms of material contribution to risk, they not only
classify it as a case where the damaging process had a cumulative effect, which it
did not-that was the basis of the case's novelty-but more importantly they
63
argue that McGhee is an application of the material contribution 'doctrine'.
What is this doctrine? If it is the 'doctrine' in Bonnington Castings, is it simply
the conventional point that a causal connection to damage (damage greater than
de minimis) must be shown? McGhee is palpably inconsistent with such a
doctrine.

60 The type of case mentioned in lines 12-17, 226.

61 [1956] AC 613. 'Guilty' source is the one which is the subject of the complaint.
62 [1972] 3 All ER 1008.
63 P 410, see also liii.
 LAW, CAUSATION AND COMMON SENSE

Another complaint about the way in which the authors deal with these House
of Lords' decisions is that they fail to discuss an odd but important aspect of the
outcomes. In Bonnington Castings the pursuer recovered the full amount of his
damage with no deduction being made for the contribution to the cumulative
damage done from the innocent source of dust. As this practical result of the
case is entirely inconsistent with the ratio-the pursuer had not shown a causal
connection to that 'innocent' part of the damage-it deserves a mention. The
point is also worth pursuing in the McGhee context. 64 Indeed the whole idea of
managing certain increasingly problematical areas of the proof of sine qua non
by the use of apportionment could have received much more attention from the
authors, particularly in pursuit of their goal of making Causation in the Law
65
more useful to practitioners.
In one section the authors do discuss the problem of apportionment in those
multiple cause cases where (unlike multiple possible cause cases such as
McGhee) two factors, one of which is the defendant's wrongful conduct, both
clearly satisfy a sine qua non relation to some damage but the available evidence
is inadequate to quantify how much damage was due to each. Here Hart and
Honor6 make the remarkable comment that...
It is absurd that a defendant who has clearly done harm should escape altogether and it is
submitted that, if there is no evidence how much of the total is caused by his act, the onus
of proof should be transferred to the defendant, who would thus be liable for the whole
66
harm unless he can prove what part of it was caused by someone else.
To shift the burden of proof to the defendant on an issue impossible of proof,
once a sine qua non relation to some damage has been shown, certainly
67
rationalizes the recovery of full damages in such cases as Bonnington Castings.
Moreover, it is arguably a fair result where the remaining factors contributing to
the overall damage were under the defendant's control. But as a general
proposition, surely it is too extravagantly pro-plaintiff? Say V suffers from a
stress-related heart disease. Should his or her employer be held liable for the
whole damage merely because V's employment is wrongfully stressful and the
contribution to the disease made by the work stress cannot be quantified?

64 Although here a deduction for the possibility that an innocent source of risk was the actual cause
was impossible because experts could not give even crude estimates of the relative contributions
to risk. The only mention of the point by Hart and Honor6 is at n 48, 102.
65 P xxxvi. A particularly important area of application concerns cases of wrongful omissions e.g.
would the employee-plaintiff have used the safety device his employer negligently failed to
provide?
66 P 228.
67 Although it should be recognized that it is tantamount to a shift in the grounds of responsibility
away from a requirement of proving causal connection (i), (see above). See also, liii-lv, lxii, 102,
424.
 JANE STAPLETON

THE THESIS AS AN ANALYTICAL TOOL

Finally let us look at the major potential strengths of the text for the general
reader. How much does its conceptual framework improve the reader's under-
standing of some complex legal issues involving causation?

A practicalapplication
Take an example that is exercising the minds of personal injury litigation
lawyers at present: the potential liability of tobacco companies for the damage
done to health by their products. In this area one of the major difficulties for
plaintiffs appears to be the risk that the defence of volenti will be raised against
them. 68 We can now put this in Hart and Honor6's causal terms: the
defendant-manufacturer's act in marketing the dangerous product was not a
,cause' of the resultant ill health in the user because the intervention of the
voluntary act of the smoker negatived the central type of causal connection,
'causing harm'. Hart and Honor6's analysis also suggests to us an alternative
argument on the smoker's behalf: even if we assume the smoker's initial choice
to smoke is a deliberate 'voluntary' act, the context of the marketing of a highly
dangerous product likely to precipitate disease in users (killing a quarter of them
prematurely 69 ) is, arguably, one that calls for the recognition of a duty of care to
avoid providing an occasion for the victim to harm himself by his own deliberate
act. In other words even if liability based on the first causal paradigm cannot be
established, liability based on the third paradigm may be imposed. This line of
argument is, of course, all the more powerful in the light of the fact that the
product involved is not only highly dangerous but addictive. Indeed, the
addictiveness of the drug allows the plaintiff to claim that damage done by
smoking after a point in time when he or she had become dependent on tobacco,
was caused by the manufacturer (i.e. causal connection (i) is satisfied): could it
not be argued that the choice to smoke by a dependent smoker is not sufficiently
free to negative the central causal connection to the manufacturer's conduct?

Formulatinga new legal rule

Another example of the usefulness of Hart and Honor6's framework is the way
in which it illuminates the formulation of a new liability rule. Take products
liability. In recent years many have advocated a wider liability for the
manufacturer of products. What form could this take? Drawing on the authors'
analysis the following initial comments might be made.
First, legal responsibility need not require a causal connection between the
defendant's conduct and the harm. 70 For example, a car manufacturer might be
made liable (as, in effect, an insurer) for any personal injury suffered by an
68 See e.g. 'Note: Plaintiff's Conduct as a Defence to Claims against Cigarette Manufacturers'
(1986) 99 Harv LR 809.
69 R. Peto (1980) 39 Health EducJ 45-46.
70 P xxxv.
 LAW, CAUSATION AND COMMON SENSE

occupant of one of its cars. No one advocates this drastic form of products
liability reform.
Second, where, as is usually the case, a causal connection is an element of
legal responsibility satisfaction of a but-for relationship is insufficient to
establish the connection according to Hart and Honor6. This agrees with the
general consensus of reformers that a producer ought not to be liable merely
because the injury would not have occurred but for (the use of) its product.
Liability must be drawn more narrowly than this,
Third, strict liability (for products as for the general case) is based on proof of
causal connection, although this need not be between the defendant's conduct
and the harm. For example, as Hart and Honor6 say, in 'Rylands v Fletcher,
once it is proved that the defendant has accumulated a noxious substance on his
land, he is liable if it escapes' resulting in harm. 7 1 In accord with the general
Hart and Honor6 thesis, the causal connection between escape and harm may be
negatived by the act of a third party or an abnormal event such as an act of God.
Applying this to strict liability for products we can now formulate a possible
strict liability rule: once it is proved that the defendant put the product into
circulation for a particular use, he is liable if that product, used in that way,
results in harm i.e. if its use is a sine qua non of the harm and no relevant act of a
third party or abnormal event has intervened between the product use and the
harm to negative the causal connection. This liability formula, the width of
which is intermediate between the unacceptably wide 'but-for' test (see above)
and the unnecessarily restrictive 'defect'-based formulas resorted to by the great
majority of reformers, is arguably an important missing link in the products
liability debate. 72 Hart and Honor6's framework both suggests and illuminates
this important option.

Thin skulls
It is a strange but well-settled maxim that a tortfeasor takes his victim as he finds
him. Thus if D negligently inflicts a scratch on V which, due to V's abnormal
susceptibility, results in cancer, D is liable not only for the foreseeable scratch
but for the cancer, even if the latter was completely unforeseeable. Hart and
Honor6 discuss the rule 73 but fail to allude to the startling practical dis-
continuity in liability it produces. Suppose the victim contracts cancer due to
his abnormal susceptibility to the defendant/employer's work conditions; the
susceptibility will be irrelevant if V can point to a trauma (e.g. a scratch) which

71 Pp 85-6, see also xliv-xlvii, 134, 164, 287.

72 J. Stapleton (1986) 6 OxfordJ Legal Stud 392. One could also formulate a still wider strict
liability based on occasioning harm to cover special cases. For example, for policy reasons one
might impose liability on a manufacturer for harm done to the (as yet not dependent) users of its
addictive product or on a car owner for personal injuries caused by the thief of his car (in the
latter case on the policy ground that the owner has insurance cover for this risk).
73 Pp 79-80, 172-6, 274.
 JANE STAPLETON

triggered his cancer (and which scratch the defendant ought to have taken
precautions to avoid). If there is no such trauma, the victim's abnormal
susceptibility is not 'taken as it is found' at all but will be a major factor
militating against a finding of negligence in fact (because the susceptibility is
likely to be unforeseeable). In a large number of cases, therefore, liability will
depend on the fortuitousness of there being a foreseeable, even if minor, wound
on which to hang the maxim. This is certainly a sufficiently anomalous and
important practical result to have warranted a mention by the authors.
Hart and Honors seem to accept the thin skull rule even though they admit
that its scope is unclear. They argue that it reflects well-entrenched common
sense notions. But this is unconvincing. Suppose that X has an abnormal
susceptibility to cancer like the plaintiff in Smith v Leech Brain. This is
triggered when his hairdresser carelessly scratches his scalp. Would the ordinary
person really say that the hairdresser caused the resultant cancer? Yet this is the
conclusion one is led to by the authors' reasoning that the abnormal suscep-
tibility, being part of the stage-set, does not negative the common sense notion
of causal connection between the hairdresser's act and the harm. Ironically,
Hart and Honor6 fail to exploit the elements of a more powerful rationale of the
rule which they present elsewhere. Might not a notion of 'occasioning harm' be a
more appropriate basis for the maxim? That is, we attribute the harm to the
defendant in these cases (by regarding as sufficient the less demanding causal
connection of 'occasioning harm') because he has provided the conditions, the
trigger, for harm attributable to the abnormal susceptibility of the plaintiff and,
as a matter of legal policy in cases of abnormal susceptibility, we choose as
between a completely innocent victim and a careless defendant to shift the loss
to the careless defendant.

CONCLUSION
Whether or not a general reader is persuaded by Hart and Honor6's thesis, those
who persevere with it will find that it throws invaluable light on causal concepts
and on the complex debate about which theory of 'legal causation' is more
attractive in principle, and which most effective in explaining the case law. The
complex argument is expressed in an easy style from which there are few lapses,
and it is supported by wide-ranging examples with detailed footnoting. In the
context of the breathtaking scope of this scholarship, some of the earlier
criticisms of detail may appear unimportant. More worrying are doubts about
pivotal concepts such as that of 'voluntary' conduct and the ordinary person's
'common sense' notions of causal connection. Nonetheless, even if Hart and
Honor6's thesis ultimately fails to convince some readers, their book remains the
outstanding and central text in the area.