To the question--when may a claimant recover substantial damages from a defendant despite being unable to prove that the defendant's wrongful conduct caused her damage?--English tort law gives an incoherent and unjust answer. After briefly outlining that answer in the first part of this article, the second part describes some fundamental aspects of its incoherence and injustice. It traces each aspect to the House of Lords decision in Fairchild v Glenhaven Funeral Services Ltd, (1) the modern source of English tort law's enclave of exceptional proof of causation rules. If the charges of incoherence and injustice against Fairchild are well-founded, it amply deserves a place in this special issue. The third part of the article sets out what the law on proof of causation in tort should have been had it not been distorted by Fairchild and its progeny.
I THE CURRENT LAW
We can state the law on proof of causation in English tort law as follows.
(1) The claimant generally bears the legal burden of proof to demonstrate on the balance of probability that the defendant's wrongful conduct was a cause of its damage in order to recover substantial damages. (2) This will involve demonstrating on the balance of probability that but for the defendant's wrongful conduct, the damage would not have occurred or that the defendant's wrongful conduct materially contributed to the damage. (3) The legal burden of proof dictates that if a finding cannot be made as to the existence or non-existence of some fact according to the governing standard of proof, the party bearing the burden fails to establish the relevant fact. Consequently, as a general rule, if it is impossible to demonstrate that the defendant was a cause of the damage on the balance of probability, the allocation of the burden of proof dictates that the claimant loses on the causal issue.
(2) The claimant is exceptionally relieved of proving that the defendant caused her damage on the balance of probability where the claimant suffers from mesothelioma (an asbestos-induced cancer) and it is impossible, due to the limitations of medical science, to determine on the balance of probability whether or not the defendant's wrongful conduct was a cause of the claimant's mesothelioma and all potential causes of the claimant's mesothelioma would operate to cause mesothelioma in a substantially similar way. In such circumstances, C may establish that D is liable in damages for C's mesothelioma simply by showing that D's wrongful conduct has materially increased the risk of C contracting mesothelioma.
This exception to the general rule was established by the decision in Fairchild. Fairchild involved three claims in each of which an employee had been wrongfully exposed to asbestos dust by successive employers and had either died or suffered from mesothelioma. Each employee had, on the balance of probability, contracted mesothelioma as a result of wrongful conduct, but was unable to prove on the balance of probability which employer's or employers' wrongful conduct had been causative because of limitations of scientific knowledge over the mechanism by which mesothelioma is caused. It was and continues to be unknown whether the disease is caused by a very small number of fibres or by the cumulative effect of many fibres. Consequently, each defendant could argue that the claimant's disease had been caused by a small number of fibres for which another defendant was responsible. The House of Lords allowed the estates of the deceased employees and the only living employee to succeed against each defendant on the basis that each had materially increased the employee's risk of mesothelioma. (4) In the later case of Barker v Corus (UK) Ltd., which involved the same facts as Fairchild except that the claimant's mesothelioma may have been caused by his contributorily negligent exposure to asbestos whilst self-employed, the House of Lords held that D's liability in damages in situations to which the decision in Fairchild applies is proportionate to the risk to which D wrongfully exposed C. (5) According to Barker, if D wrongfully exposed C to a 20% risk of mesothelioma and the mesothelioma caused C 50,000[pounds sterling] of loss, C should recover 10,000[pounds sterling] from D. Parliament very soon after reversed the apportionment of liability made in Barker, but only in cases of mesothelioma caused by asbestos, with Section 3 of the Compensation Act 2006. (6)
(3) The most difficult question is whether the claimant is exceptionally relieved of proving that the defendant caused her damage on the balance of probability in cases where the damage suffered is not mesothelioma. It is likely that there are two further situations in which this will be so under the decision in Fairchild. In both situations the defendant's liability will be proportionate to the level of risk it has wrongfully imposed upon the claimant. This is because the decision in Barker v Corus (UK) Ltd. continues to apply to any non-mesothelioma cases which fall within the decision in Fairchild. The first situation is where D wrongfully exposes C to a toxic agent or wrongfully fails to protect C against a risk posed by a toxic agent, which conduct materially increases C's risk of contracting a disease, C suffers from the disease, and the aetiology of the disease is such that it is impossible to determine whether or not the wrongful risk imposition caused C's disease, where all potential causes of C's disease operate in a substantially similar way, and the defendant is not the National Health Service. (7) The second is where C has been the victim of a tort which has caused C to suffer an injury and it is impossible for C to demonstrate on the balance of probability which of a number of defendants, each of whom, in breach of a duty, has materially increased the risk of that injury to C. Call this last situation the 'indeterminate wrongdoer situation'.
There are at least three reasons to believe that Fairchild will apply in the first situation. First, the House of Lords in Fairchild considered that the earlier House of Lords decision in McGhee v National Coal Board was, properly interpreted, an authority supporting the decision in Fairchild. (8) In McGhee, the claimant could not prove whether his employer's failure to provide showers to wash off brick dust after working in the employer's brick kilns had caused his dermatitis. The uncertainty over the contraction of dermatitis was structurally similar to the uncertainty in Fairchild: the medical evidence did not permit one to determine whether a few dust particles cause the disease or whether a large number of dust particles acting in combination cause the disease. So Fairchild is likely to apply to dermatitis, at least on the assumption that medical understanding has not advanced since McGhee. (9) Secondly, the Court of Appeal has applied Fairchild, albeit unnecessarily, since a finding of but-for causation could probably have been made, in a case involving Vibration White Finger disease. (10) Thirdly, in the recent Supreme Court decision in Sienkiewicz v Greif (UK) Ltd., a mesothelioma case involving a wrongful exposure of an employee to asbestos dust by an employer and non-wrongful exposure due to asbestos present naturally in the environment, Lord Phillips and Lord Rodger, with whom the other seven Justices agreed, considered that the Fairchild decision was justified by the 'rock of uncertainty' faced by mesothelioma claimants, given the available scientific evidence over the aetiology of the disease. (11) This tends to suggest that if similar impenetrable uncertainty surrounded the aetiology of another disease then the Fairchild decision would apply to it. (12) It is impossible, however, to be confident as to this. The inference is only justified so long as it is reasonable to expect judges to decide cases in accordance with an identifiable legal principle. As we shall see, that expectation may not be rational in the proof of causation context.
The approval by the House of Lords in Fairchild of the decisions in the Californian case of Summers v Tice (13) and the Canadian case of Cook v Lewis (14) supports the claim that Fairchild will also apply in the indeterminate wrongdoer situation. Summers and Cook both involved a claimant who had been negligently shot by a hunter but, because another negligent hunter had fired at the same time as the causative shot, it was impossible to determine which hunter had causatively wronged the claimant: both concerned the indeterminate wrongdoer situation. Secondly, Lord Phillips in Sienkiewicz suggested that Fairchild would apply in an indeterminate wrongdoer situation. (15)
II INCOHERENCE AND INJUSTICE IN THE LAW
The current law is incoherent in two main respects. First, the reasons offered by judges to explain the law do not explain it. Secondly, the way in which the law conceptualises exceptions to proof of causation on the balance of probability does not accurately describe the legal nature of those exceptions. It is not necessary to think that 'coherence' in the law is intrinsically valuable to find each of these problematic. (16) Rather, the consequences of the incoherence are (also) regrettable. If the reasons offered for the scope of the law do not explain it, the consequences are that like cases may be treated differently; claims may be denied or refused without the reasons for success or failure withstanding scrutiny; and the law may become difficult to predict. If the law is not accurately conceptualised, there is a danger of errors of analysis being made in future cases. The current law is unjust (beyond the injustice implicit in its incoherence) in so far as it conceives of the requirement of causation as a practical tool for assigning responsibility for an injury, rather than as fundamental to a person's entitlement to damages, and so, to a defendant's liability to pay damages. In this part, these criticisms of the current law will be developed and traced to the decision in Fairchild.
A Rationales Outstripping Rules
The reasons given in Fairchild for making an exception to the rule that the claimant must prove that the defendant's wrongful conduct caused her damage on the balance of probability did not justify the ambit of the legal rule supported by their Lordships in that case. Although some of the reasoning does not go beyond an appeal to intuitive fairness, three main arguments can be discerned in the judgments for altering the orthodox requirement that the claimant prove causation on the balance of probability. (17)
The first was that insistence upon the orthodox legal burden of proof would empty of content the defendants' duty of care in negligence. The duty to take care would be emptied of content because claimants would never be able to prove that a particular defendant's negligence caused its mesothelioma in light of the scientific uncertainty over the mechanism by which mesothelioma is caused: '[i]f liability depends upon proof that the conduct of the defendant was a necessary condition of the injury, it cannot effectively exist'. (18) Call this the 'empty duty' argument.
The second argument was that there was a particular unfairness in the fact that if only one defendant had breached a duty of care to the claimant that the claimant would succeed in establishing a claim for damages because that defendant's wrongful conduct could be identified as causative but because there were multiple defendants in breach of duty, the claimant could not recover because he could not point to the causative defendant. The addition of a wrongful defendant causes one to lose one's claim for damages. As Lord Rodger put it: 'the greater the risk that the men have run at the hands of successive negligent employers, the smaller the claimants' chances of obtaining damages'. (19) Call this the 'multiple wrongdoer' argument.
The third argument was that a false positive was not as unjust as a false negative. In other words: holding a negligent, but causally innocent, defendant liable is less unjust than denying a wrongfully injured claimant a remedy, where a choice must be made between the two: '... such injustice as may be involved in imposing liability on a duty-breaking employer in these circumstances is heavily outweighed by the injustice of denying redress to a victim'. (20) Call this the 'relative injustice' argument.
None of these reasons justifies the major limitations placed upon the exception to the general rule recognised in Fairchild. The limitations were that the exception should only operate where the claimant has been exposed to sources of risk which are all of the same kind or which operate in a substantially similar way and that the exception should only operate where the claimant's inability to prove causation is a result of deficiencies in scientific knowledge.
The former requirement is met in mesothelioma cases because the only known cause of mesothelioma is asbestos: all potential causes of it are therefore the same (except to the extent that there are different kinds of asbestos dust with different causal potencies (21)). Lord Bingham said that: 'it is one thing to treat an increase in risk as equivalent to the making of a material contribution where a single noxious agent is involved, but quite another where any one of a number of noxious agents may equally probably have caused the damage'. (22) Similarly Lord Hutton restricted the decision in Fairchild to situations 'where there is only one causative agent'. (23) Less exigent was Lord Rodger's formulation that the injury be caused 'by an agency that operated in substantially the same way'. (24)
None of the justifications of the decision in Fairchild compels or suggests either restriction. (25) The defendants' duty of care would equally be without content where proof is recurrently impossible for reasons independent of the absence of scientific knowledge and there are multiple different agents or agents which operated in substantially different ways. If the concern addressed by the 'empty duty' argument is that defendants are not incentivised to adhere to their duties of care--that they are able to ignore them without fear of sanction--the concern applies wherever there is a recurrent evidential uncertainty over causation which is reasonably predictable ex ante by defendants.
It may be correct that a defendant can readily predict severe difficulty in proving causation against it where the claimant has been or will likely be exposed to similar risk sources and there is impenetrable scientific uncertainty surrounding the mechanism by which a risk source operates. In such circumstances it may be rational to believe that because of the existence of multiple similar risk sources, it will be difficult to pinpoint the causative source. (26) But even if the idea of a similar risk source and absence of scientific knowledge picks out a relatively sound rule of thumb for one situation in which a defendant's duty will be empty of content, (27) it cannot provide a necessary condition for such situations. There are many sources of recurrent and so predictable uncertainty over causation. For example, lengthy latency periods between exposure to a toxic agent and disease manifestation give rise to substantial difficulties of causal attribution because of the diminution of evidence over long periods of time. (28) This has nothing to do with the lack of scientific knowledge. Moreover, in any activity in which there is a background risk of n% of some harm occurring, D may wrongfully expose C to a risk of that harm up to n% and if the only evidence is purely statistical, then it will be impossible to prove that the defendant's wrongful conduct caused any claimant's damage. (29)
The intuitive unfairness in the idea that the claimant's legal position becomes worse the more she is the victim of wrongful conduct by successive persons--i.e. the less her chances of being able to sue for damages--is also, obviously, independent of the kinds of causal agents used by those wrongfully subjecting her to risks of injury or the kind of uncertainty at stake. And, finally, it is obscure how the relative injustices of the situation change by reference to the types of causal agents used by wrongfully acting defendants or the type of uncertainty in question.
The difficulty, then, with the substantially similar agent requirement and the absence of scientific knowledge over the causal mechanism requirement is the jar created with the underlying reasons adduced to motivate the exception to proof of causation in the first place. This has only been mildly mitigated by Lord Phillips' suggestion in Sienkiewicz that Fairchild will apply in an indeterminate wrongdoer situation involving purely factual uncertainty. (30) The substantially similar agent requirement remains and the requirement of scientific uncertainty continues to apply beyond the indeterminate wrongdoer situation. The result is a persisting tension between the underlying principles and their authoritative statement as a legal rule. (31)
The consequence of this is that claims are rejected for reasons which do not adequately justify their rejection. Consider the facts of AB v Ministry of Defence (32) The claimants sought to recover damages for cancer which they alleged had been caused by their employer's having negligently exposed them to radiation during their employment. They could not establish causation on the balance of probability but, at most, that the defendants had wrongfully materially increased their risk of cancer. The claim was struck out by the Court of Appeal (whose decision was approved by the Supreme Court) on the ground that there were multiple possible causes of the claimant's cancer and so the substantially similar agent requirement under Fairchild was not satisfied. But since this requirement does not follow from any of the reasoning in Fairchild, and has no other obvious normative force, it clearly provides insufficient reason to reject such a claim.
B Analytical Problems
The reasoning of three members of the majority in Barker v Corus led to the law on proof of causation becoming analytically confused. The seeds of that confusion are traceable to Fairchild.
In Barker, three of the majority judges held that the damage for which defendants are held liable under the rule in Fairchild is not mesothelioma but the risk of mesothelioma. (33) The consequence of this was that each defendant was only liable in proportion to the extent that it had wrongfully exposed the claimant to the risk of mesothelioma.
This was unnecessary, undesirable, and legally unsustainable. (34) It was unnecessary because it is clear that the only aim sought to be achieved was proportional liability in situations to which Fairchild applied. It should have been possible simply to state that liability should be in proportion to the risk that the defendant caused the damage (as a 'trade-off' for the injustice of holding defendants liable without proof of causation on the balance of probability), rather than holding that the 'risk' was itself damage. It was undesirable because, as Lord Rodger observed, it created obvious inconsistency with the rejection elsewhere in the law of tort of the claim that merely subjecting a person to an increase in risk could serve as 'damage'. It was legally unsustainable because two of the three claimants in Fairchild and all three of the claimants in Barker were bringing claims under the Fatal Accidents Act 1976 for damages in respect of the deaths of their husbands. The Act requires that the death in respect of which the claimant is suing be 'caused by any wrongful act, neglect or default'. (35) It is 'death' not the risk of death which must be shown under the Act. All of these claims should have failed if the jurisprudential basis of the Fairchild rule were that the defendants were liable for causing the risk of mesothelioma.
These problems can hardly be attributed entirely to the House of Lords in Fairchild. But there are two respects in which Fairchild had encouraged them. The first is that some of the analysis in the judgments in Fairchild suggests that the exception to the orthodox burden of proof on causation being created is to be conceived of as an alteration in the substantive law of causation. Lord Nicholls stated that a defendant's causing a material increase in the risk of mesothelioma 'should be regarded as a sufficient degree of causal connection'. (36) Similarly, Lord Hoffmann stated that a material increase in risk was being 'treated ... as sufficient in the circumstances to satisfy the causal requirements for liability'. (37) This coheres with Lord Hoffmann's argument, expressed extra-judicially, that the substantive causal requirements for liability depend upon the purpose of the legal rule being interpreted--from which it follows, given that different rules may have different purposes, that there is no univocal legal concept of causation--rather the meaning of the word 'causation' depends upon interpreting the scope of the particular legal rule at stake. (38) If the substantive requirements of causation can be modified to reflect the underlying normative considerations of a legal rule (like the 'empty duty' argument), then why not also the concept of damage? In this way, there is a continuity between the normative analysis of causation in Fairchild and the normative analysis of 'damage' in Barker. (39) Second, the judgments in Fairchild had placed considerable justificatory weight upon the idea that the defendants had increased the employees' risk of mesothelioma. This was problematic since it is difficult to understand what justificatory role the fact of risk imposition plays beyond the role already implicit in saying that the defendants have been negligent--since all negligent conduct involves a risk imposition. So if the idea of 'risk' were to play a non-redundant role in justifying the Fairchild exception, it could have been thought that its role had to be as forming the damage which the defendants had caused the claimants. For these two reasons, the problematic analysis in Barker can fairly be attributed in part to Fairchild.
The idea that the defendants are liable for the risk of mesothelioma (where the risk is the damage) under the Fairchild exception seems to have been laid to rest by the recent Supreme Court decision in The Trigger Litigation. (40) But the analytical problems in Fairchild which encouraged the analysis in Barker have not. In the Trigger Litigation, which concerned whether employers could rely upon Fairchild to establish that their employee had 'sustained' or 'contracted' mesothelioma due to the employer's negligent exposure (so as to claim under liability insurance policies), Lord Mance considered that it was 'entirely natural' to view Fairchild as imposing a 'responsibility for the [employee's] mesothelioma, based on a 'weak' or 'broad' view of the 'causal requirements' or 'causal link' appropriate in the particular context to ground liability for the mesothelioma'. (41) The idea of different types of causal connection comes from Hart and Honore (42) (cited by Lord Mance) who distinguished a central concept of causation (cases of voluntary human interventions or abnormal natural events bringing about results which would not normally (relative to some viewpoint) occur without those interventions and without the intervention of a free, deliberate, voluntary action or abnormal natural event) (43) and a weaker concept of causation (providing an opportunity or means by which another person's voluntary action causes--in the core sense--an outcome). Whatever the merits of this pluralist view, these authors were giving an analysis of the substantive concept or concepts of causation. They were obviously not considering how to conceptualise a legal exception to the proof of causation. Nor would they have accepted the incorrect thesis that 'materially increasing the risk' is a species of causal connection which could sit alongside their other types of causal connection. Materially increasing the risk of a harm is conceptually nothing like causing it. Rather, materially increasing the risk of harm can either be fictionally equated with causation of harm for a particular normative purpose or it can serve as a first step in an evidential inference of causation. Instead of accepting the fictional route, a better view may be that the law ought to conceive the Fairchild rule as a rule which reverses the burden of proof on the question of whether the defendant was a but-for or materially contributory cause of its damage when the claimant has proven on the balance of probability that the defendant materially increased the risk of its damage. Failure to discharge this burden of proof establishes liability. This liability is then (in non-mesothelioma cases) made proportionate to reflect the fact that the allocation of the burden of proof diverges from the orthodox allocation. Designating the rule firmly as an evidential rather than substantive rule is a truer reflection of the mischief--evidential uncertainty--to which the rule responds.
The current law is unjust because it holds defendants liable without their being shown to have caused the claimant damage on the balance of probability without good reason. I will advance a weaker and a stronger version of this claim. The weaker claim is that the 'empty duty' argument for departing from the orthodox legal burden of proof on causation is unconvincing in mesothelioma cases. The stronger claim is that the 'empty duty' argument is unconvincing in most imaginable cases.
The 'empty duty' argument is basically a deterrence argument. It says that defendants are insufficiently incentivised to adhere to their duties of care in the tort of negligence because there will be no sanction for the causative breach of those duties. This arises as a result of claimants' recurrent inability to prove causation. In other words: defendants are insufficiently deterred against breaching their duties of care. (44)
Here are three reasons why the argument is unconvincing in mesothelioma cases. First, even if the law of tort provides no sanction for breaches of duties to take care not to expose persons to asbestos dust, the criminal law often does. (45) Secondly, as others have pointed out, even if it is true that claimants cannot establish on the balance of probability that a particular defendant has caused their mesothelioma, defendants continue to have incentives provided by the law of tort to conform to their duties of care. (46) This is because breach of those duties may lead to other asbestos-related diseases such as asbestosis in respect of which claimants will generally be able to establish causation on the balance of probability. As asbestosis is worsened by cumulative exposures, it will generally be possible to establish that a particular defendant has caused some part of the claimant's overall condition. (47) Thirdly, contrary to what is often assumed, it is simply not the case that a person can never prove that a particular defendant's wrongful conduct has caused her mesothelioma on the balance of probability. If a person is wrongfully exposed to asbestos for 29 years by D1 and 1 year by D2 and the level of environmental exposure is far smaller than either D1's or D2's exposure, then it should be possible to say on the balance of probability that D1 has contributed to C's mesothelioma. This is because the uncertainty over the causative mechanism only prevents us from ruling out the possible causative impact of a defendant's wrongful conduct: as it may be the case that mesothelioma can be caused by a very small number of asbestos fibres, even a small exposure may have caused a person's mesothelioma. However, in D1's case, it is probable that D1 has contributed to the mesothelioma regardless of the nature of the causative mechanism. It can be argued that D1 either contributed by supplying a few fibres sufficient in themselves to cause the mesothelioma or by supplying fibres which in combination with later fibres caused the mesothelioma. Whichever the mechanism, it is likely that D1 contributed to it. (48)
Now consider five points which tell against the 'empty duty' argument quite generally. First, there is the simple point that it may well be true that beyond the mesothelioma context there are incentives not to breach duties of care provided by alternative (criminal) institutions (49) and by virtue of the fact that wrongful conduct may risk more than one type of harm, where at least some of those risked harms will be provably attributable to defendants. If this is true, then (further) incentives provided by tort law will be unnecessary. If it is objectionable to hold defendants liable without proof of causation on the balance of probability, this should not be endorsed if it is unnecessary. (50)
Secondly, it must be questioned how much weight one should give to deterrence arguments when we have such poor evidence as to the concrete effects of tort liability upon behaviour in particular contexts. (51) Even if we assign a speculative empirical argument some weight, it is hard to accept that that weight is sufficient to outweigh the principle that the defendant ought only to be made liable if it has been proven on the balance of probability to have caused the claimant's damage. If we think that causation is normatively very important to grounding the defendant's liability to pay damages to the claimant, then commitment to that position requires more than a speculative empirical argument to justify a departure. (52) The point here is not that consequentialist arguments are never relevant to the question of whether a person ought to bear a liability--the monist corrective justice position (53)--but rather that, without greater empirical substantiation, they do not outweigh other important normative considerations.
Compare the following situation involving an empirical uncertainty as to the effects of liability. Suppose that Doctor negligently fails to diagnose Patient with the (undisputed) result that Patient suffers a brain injury. If Doctor argued that Patient should not be compensated by him because there is a chance that the imposition of negligence liability upon doctors for failures to diagnose would lead doctors generally to practise defensive medicine (with substantial negative effects upon patients' welfare), most would not find this argument compelling. The courts obviously do not accept it. (54) But if we find appeal to empirical speculation problematic in rejecting an otherwise established liability, (55) there is some reason to think that appealing to empirical speculation in establishing an otherwise not established liability is also problematic.
Thirdly, relying upon the 'empty duty' argument to depart from the balance of probability rule creates a highly unstable exception to the general rule. It is unstable because there are no available criteria with which to assess with any confidence whether the exception ought to apply or not in a particular factual situation. This is because (as we have already said) it is, on the one hand, difficult to tell whether sufficient incentives to conform to tortious duties already exist outside of the law of tort and, on the other, it is only possible to conjecture as to whether the imposition of tort liability in situations of recurrent causal uncertainty will lead to improvements in the overall number of harms or wrongs since we can only conjecture over what effect the imposition of liability will have in a situation of recurrent causal uncertainty. Courts are invited under the 'empty duty' argument to assess the impossible question of whether tort liability will induce incentives beyond those already provided by morality, regulatory regimes, criminal sanctions, market forces and the fact that conduct often risks multiple harms some of which will be provably attributable to defendants.
Fourthly, even if we assumed the empirical validity of the 'empty duty' argument, accepting it creates distributive injustice between classes of claimants (and classes of defendants). (56) Some claimants will receive awards of damages because of the contingency that they may have been injured by a defendant whose activity is prone to recurrent causal uncertainty, whilst others will receive nothing because their receiving compensation does not serve the social goal of deterrence. This injustice is exacerbated by the fact that, even if we assumed the empirical validity of the 'empty duty' argument, it does not necessarily provide an argument for compensating the claimant rather than another person. It may be the case that allowing private individuals who have been wrongfully exposed to a risk by the defendants to claim damages against those defendants for materially increasing their risk of harm in cases where there is recurrently predictable causal uncertainty reduces the overall level of future harms or wrongs. But it may be the case that alternative mechanisms could better perform this function. For example, allowing companies to compete in the market for licences to sue wrongful risk-increasers may be better at deterring wrongful conduct. The reason one claimant recovers and the other does not, then, is that it is possibly necessary for the social goal of deterrence that the former recovers.
A slightly different version of the last point is that, assuming the empirical validity of the 'empty duty' argument, and supposing that we had a non-arbitrary way of applying it in particular cases, then accepting the argument will result in a 'checkerboard' law of torts. It will be akin to Dworkin's statute which says that only persons who are born in even years can have abortions in order to reach a compromise between different views on the morality of abortion. (57) Lawyer-economists who support liability in cases of recurrent causal uncertainty despite the absence of proof of causation on the balance of probability give examples of situations where liability would be mandated on the 'empty duty argument' which suggest that the law of tort would become highly 'dappled' or checkerboard. Porat, for example, suggests the introduction of proportionate liability in cases where there is a 'downward bias' in the probability of causation distribution of cases to which a certain activity gives rise. (58) That is: in activities where the probability of D's having caused the claimant's damage is frequently below 0.5. He gives the example of an emergency treatment ward where patients on average have chances of survival of less than 30% such that none can ever establish causation on the balance of probability in the event of negligence. If we do think that 'integrity' in the law is a distinctive value or a value closely animated by justice, (59) then this kind of unequal treatment of defendants (and claimants) is problematic.
The third and fourth arguments are to my mind sufficient to dismiss the 'empty duty' argument. The first and second arguments, however, rely upon an as yet undefended premise. (60) Each of these arguments is of the form: 'given the importance of proof of causation to a person's entitlement to damages from the defendant, the force of the empty duty argument is insufficient to outweigh that importance'. But nothing has yet been said to support the importance of causation to the claimant's entitlement to recover from the defendant in respect of its damage. Establishing that importance will serve to vindicate the first and second arguments by making out a crucial premise as well as providing a further, independent, point against the 'empty duty' argument.
The importance of causation to establishing that the defendant should compensate the claimant for its loss may be defended in at least four ways. From a consequentialist perspective, it may be argued that requiring the claimant to prove that the defendant caused its injury reduces the efficiency-reducing effect of judicial errors in setting the standard of care in negligence-based torts and reduces overall administrative costs of the tort system by reducing the number of claimants who can sue. (61) If this account of the importance of causation were accepted, then the 'empty duty' argument (if it had empirical credentials) would have force. On the consequentialist account, the normative force of causation is simply a product of whatever good consequences it brings about (fewer harms or wrongs or economic costs). The 'empty duty' argument says that the usual benefits which insisting upon causation brings are outweighed by the negative consequence that, in a situation of recurrent causal uncertainty, there is no incentive to take care provided by the tort system. (62) On this account, causation is simply a practical tool to maximise welfare (or wealth or some good consequence).
Clearly, then, if the first and second arguments are to be successful, they must rely upon another, more robust, account of the importance of proof of causation. (63) One such account is what we can call the wrong-based account. According to this account, it is necessary to prove that the defendant caused the claimant's damage because the defendant has only wronged the claimant in the event that the defendant has caused the claimant's damage. This provides a strong reason not to find defendants liable for substantial damages absent proof at least on the balance of probability that the defendant caused the claimant's damage. To hold otherwise would be to presume that the defendant is a wrongdoer, when surely the defendant has some entitlement not to be presumed to have committed a wrong against the claimant. The wrong-based account has some force so far as the tort of negligence is concerned since many people think that a constituent element of the legal and moral wrong in negligence is causation of damage. (64) This would also provide a reason for insisting upon proof of causation on the balance of probability in other contexts where the only plausible way in which the defendant could be said to have wronged the claimant is through having caused the claimant damage.
But to rely solely upon the wrong-based account would be problematic in two respects. First, it can plausibly be contended that the defendant legally and morally wrongs the claimant when it negligently increases the risk of, say, a physical injury to the claimant by its actions. (65) If so, then wrongly finding that the defendant has caused the claimant physical injury in circumstances where the defendant has been shown to have negligently risked the claimant physical injury on the balance of probability does not involve holding that a wholly innocent (66) person has committed a legal and moral wrong, but rather that a person who has breached a legal and moral duty to the claimant (not negligently to increase C's risk of physical injury) has committed a further legal and moral wrong (not negligently to cause C physical injury). This is perhaps a less serious matter. Secondly, the wrong-based account cannot explain why we should insist (as surely we ought) that the claimant prove that the defendant caused her loss in cases of non-causal wrongs (the torts actionable per se).
In addition to the wrong-based account, then, we would also point to the normative importance of finding that the defendant is responsible for the claimant's damage in order to hold the defendant liable to pay compensatory damages in respect of that damage. We tend to think that it is unjust to make a private individual bear another's loss unless the individual is responsible for that loss. The fact of causation is an essential element of individual responsibility for a loss (outcome responsibility). (67)
The importance of causal responsibility to the legitimacy of making an individual liable to bear some loss is evident in our judgments about the permissibility of acting in self-defence against another person. (68) Suppose that Alex's car is hurtling unstoppably towards Ben and will kill him unless Ben pushes Carrie, who is trying to kill David with a gun which, unbeknownst to her but to the knowledge of Ben, is unloaded, into the path of Alex's car. It does not seem permissible for Ben to push Carrie in front of the car, even although Carrie is seriously culpable, because Carrie is not causally responsible for the threat to Ben. A more humdrum example is where Evan is negligently shot by Fred having, only seconds before, been negligently risked (but not caused) the same injury by Greg's firing a gun. Fred is bankrupt and has no insurance. It would be a serious injustice to Greg to hold that he should be made to pay for Evan's injury when he has not caused it, even although Evan will not be compensated for his wrongfully inflicted injury. Greg, though he has behaved wrongfully, is simply not responsible for the injury.
A final reason for thinking that proof of causation is particularly important in justifying the imposition of a duty to pay substantial damages to compensate for a loss is what can be called the duty-based account. It is arguable that the only reason for the state to intervene on behalf of an individual to require another individual to pay substantial damages is where that person can be said to have a moral duty to the other to pay those damages. (69) It is not enough to say that imposing liability would make the claimant better off and not affect the defendant very much because he is well off. Stronger reason is required. It may be that this stronger reason is only found where a person can be said to have a moral duty to do what the State is ordering that person to do. In cases where the defendant has not caused the claimant loss it is hard to identify a duty which could justify the payment of substantial damages to the claimant. The only duty which looks as if it could play that role is the secondary duty to compensate the claimant which would arise out of a breach of the primary obligation not wrongfully to cause damage to the claimant. (70) But without causation of damage, that primary duty is not breached.
The wrong-based, responsibility-based and duty-based accounts of the importance of causation together suffice to make out the premise that causation is of significant normative importance in establishing that the defendant ought to be liable to pay substantial damages to the claimant. Given this importance, it is legitimate to insist that claimants at least prove that the defendant probably caused their damage. And given this importance, it is unacceptable to depart from this rule on the basis of a speculative empirical argument in circumstances where it may be unnecessary to do so. (71)
This section has argued that the current law is unjust because the 'empty duty' argument relied upon by the House of Lords in Fairchild does not provide good reason to depart from the rule that the claimant prove causation on the balance of probability in mesothelioma cases and in others. This naturally raises the question whether the law can ultimately be justified upon some other ground even if the 'empty duty' argument is unsuccessful. In the next part, I suggest that there is a justification for an exception to proof of causation on the balance of probability in the indeterminate wrongdoer situation for reasons akin to the second and third arguments made by the House of Lords in Fairchild. (72) This justification is limited, however, to indeterminate wrongdoer cases and could not expand beyond them.
Let us briefly consider two significant alternative justifications for the current law's position that claimants ought to obtain proportional damages against defendants in circumstances where they have not shown that it is even probable that the defendant's wrongful conduct caused their damage (and where it is not shown that the claimant has suffered their loss as a result of wrongful conduct). (73) The first is that increasing the risk of harm is itself to cause harm. If this were true, proving that the defendant wrongfully increased the risk of harm would itself be to have proven that it has caused damage. But, as others have pointed out, it is highly problematic to say that increasing the risk of harm is itself harming. (74) It is hard to accept that a person who is exposed to a risk which does not materialise and who is never aware of having been placed at risk has been harmed. If they have, then every time a person is exposed to another's negligent driving they have been harmed. Moreover, even if we were willing to accept that risk imposition is itself harmful, it is doubtful that the kind of harm suffered by the claimant would merit a claim for substantial damages. (75)
The second argument is offered by Professors Porat and Stein. They suggest that in some cases 'accuracy in fact-finding is undermined by circumstances for which no person can be blamed' but that 'in other cases ... the existing uncertainty and the consequent inability of the court to determine the facts accurately result from a person's wrongful conduct'. (76) In such cases, they say that the defendant has inflicted 'evidential damage' upon the claimant. Evidential damage is inflicted whenever the defendant 'impair[s] the plaintiff's ability or reduce[s] his chances to establish the facts necessary for prevailing in a ... lawsuit'. (77) It is then argued that wrongful infliction of evidential damage should give rise to damages in proportion to the probability that the evidence would have sustained the legal claim.
It seems clear that the authors would hold that the defendants in Barker and Sienkiewicz inflicted evidential damage upon the claimant. They hold, for example, that in cases like Hotson v East Berkshire Health Authority, (78) where it was unclear whether a doctor's negligent failure to diagnose the claimant timeously had caused his hip injury, that the doctor's negligence inflicted evidential damage. The basic problem with these claims is that it is entirely unclear what distinguishes the defendant's conduct in cases like Hotson, Barker and Sienkiewicz from any other cases of negligent conduct. In other words: if the defendant has inflicted evidential damage in Hotson and Sienkiewicz, then there seem to be no cases in which a negligent defendant will not be said to have inflicted evidential damage. Porat and Stein could respond that, in fact, in every case where a defendant has been negligent, evidential damage has been inflicted in the sense that: had the defendant not been negligent, the claimant would not be asking the question of whether that negligence caused its damage. In that sense, the defendant is a but-for cause of the claimant's uncertainty over causation. This would conflate two propositions, however: causation of uncertainty and causing there to be less evidence of something. The defendant doctor in Hotson was a but-for cause of the claimant's being uncertain over whether his negligence caused the hip injury. But it does not follow that the defendant deprived the claimant of the evidence necessary to substantiate that claim. Only the latter could realistically justify some alteration in the general proof rules and in most cases not involving destruction or failures to maintain information-storing items, this is not the case.
The conclusion to be drawn from this section, then, is that the House of Lords in Fairchild justified an exception to proof of causation on the balance of probability with an argument which was hardly tenable on the facts of the case and which is deeply problematic as a matter of principle. Although there may be other potential justifications not discussed here for departing from proof of causation on the balance of probability, the discussion here suggests that it is unlikely that the scope of the current law can be justified on alternative grounds.
III A PATH NOT TAKEN
This part argues that the decision itself in Fairchild was correct. However, it was possible to reach that decision without reliance upon the problematic 'empty duty' argument. The result could have been justified on a ground of fairness. The justification is that we ought not to allow a wrongdoer, whose liability to the claimant is established, to benefit by relying upon another person's wrongful conduct. In Fairchild, one of the defendants had wrongfully injured the claimant. Rejecting the claim would be allowing that defendant to avoid liability due to the presence of another person behaving wrongfully. This is similar to the second argument relied upon by the House of Lords that the claimant ought not to be prejudiced by the presence of an additional wrongful actor. If we accepted this as a condition for departing from the orthodox burden of proof on causation, only Fairchild would be correct. In Barker and Sienkiewicz, the court did not know that it would be allowing a defendant, who had causally wronged the claimant, to benefit from another's wrongful conduct were it not to award damages since the claimant's injury may have been caused by a non-wrongful cause. The first section below introduces the argument with an analogy. The second section below defends it against objections.
A An Analogy
The justification for the principle being proposed is that it is problematic to allow the defendant (unidentified though he may be) who has wrongfully caused the claimant's damage in Fairchild to rely upon the fact another defendant has behaved wrongfully to avoid liability. (79)
We find this idea implicit in other parts of the law on causation in tort. Consider Baker v Willoughby. (80) In that case the defendant negligently ran the claimant down on September 12 1964, permanently injuring the claimant's left leg. The courts thought that 1,600[pounds sterling] would fairly compensate him for the loss of effective use of his leg for the rest of his life. On November 29 1967, that same left leg was shot by armed robbers and had to be amputated. The issue in Baker was whether what happened in 1967 should have an effect on the defendant's liability to the claimant. The defendant argued that it should on the ground that the claimant would have lost the effective use of his leg after November 29 1967 anyway and so compensation should only be due for the period between September 12 1964 and November 29 1967. The claimant argued that this would be seriously unjust since he would not be compensated for the loss of the use of his leg from November 29 1967 onwards. If the claimant sued the robbers he would only have obtained minimal damages since his leg was already effectively useless by the time of the robbery. The House of Lords held that defendant's liability was not to be reduced by the subsequent independent wrongful conduct of another person.
Had, however, the claimant in Baker been totally incapacitated on November 29 1967 by a dormant spinal disease caused by no one's wrongdoing, the defendant's liability would have been reduced. (81) The defendant is allowed to point to a subsequent independent non-wrongful event, but not to another's wrongdoing to reduce his or her liability to the claimant. In other words, there is a difference between a case where all of the causes of the claimant's loss are wrongful and a case where one of the causes is non-wrongful.
Now consider a modified Baker v Willoughby. D1 negligently shoots X in the head. Independently of D1, D2 negligently shoots X in the head at the same time. Each shot is independently sufficient to cause X's death. In this case, it seems highly likely that English law would hold both D1 and D2 liable in solidum in respect of X's death. This result would follow even although neither D1 nor D2's negligence could be said to be a but-for cause of X's loss. (82) That is to say: although but-for causation is shown against neither defendant, X's estate would succeed in full against both in respect of the entirety of its losses. And everyone seems to agree that this is the right result despite the fact that neither defendant has made the claimant any worse off. (83) Call this case OverdeterminedDeath, because X's death was causally overdetermined.
In Fairchild, one of the defendants is seeking to rely upon the uncertainty created by another person's wrongful conduct so as to reduce his liability to the claimant. Just as in Overdetermined Death, neither defendant is shown to have been a but-for cause of the claimant's damage. The result of rejecting liability is that the claimant 'falls between two stools': the causative wrongdoer and the person who has negligently risked damage to the claimant. Although the defendant who is seeking to do this is unidentified, it still seems objectionable to allow the defendant, whoever he or she may be, to take advantage of the uncertainty created by the other person's wrongful conduct.
The natural objection to this is that even if it is objectionable to allow a wrongdoer to reduce partially or entirely their liability to the claimant by relying upon another person's wrongful conduct, this does not justify imposing liability upon the wrongdoer who has not caused any damage. After all, the robbers in Baker v Willoughby would not have been liable for the loss which had already occurred in that case even although they would have been, in a sense, relying upon the earlier wrongdoing of the defendant. (84)
The response is that, in Fairchild, the court is not knowingly violating the defendant's entitlement to be free of liability for damage which it has not caused (as it would be if it held the robbers liable in Baker). Rather, it is risking the possibility that the defendant has not caused the damage in circumstances where the defendant may have wrongfully caused the damage in order to preclude reliance upon another's wrongdoing. There is a difference between knowingly using the defendant as a means to compensate another person for a wrong for which it is not responsible and risking that possibility both in order to preclude reliance upon wrongdoing and where there is still the possibility of correcting the injustice inflicted upon the claimant by holding the party responsible for the loss liable (as there is not with the robbers in Baker).
There is, however, a significant normative difference between cases like Fairchild and cases like Overdetermined Death in virtue of the fact that one of the defendants has not been shown even to have contributed to the claimant's damage, but only wrongfully to have risked that damage. To recognise this difference between the cases, there is a plausible argument for holding (as the current law does in non-mesothelioma cases) that the defendant's liability should only be proportionate to the risk to which it has exposed the claimant. This is a reflection of the fact that some inroad has been made upon the defendant's claim not to be subject to a liability absent proof of causation on the balance of probability.
Before turning to some further objections to this argument, we may note the virtues which an emphasis upon the normative consideration of precluding reliance upon wrongdoing would bring to the current law. First, it avoids the 'instability' objection to the 'empty duty' argument. Courts would not be asked to assess an impossible empirical question concerning the incentives provided by the imposition of liability but may rather rest upon the wholly normative consideration of precluding reliance upon another's wrong to the detriment of the claimant. Secondly, it avoids the 'distributive injustice' objection. It may seem odd that a claimant who cannot prove causation against a negligent defendant but whose damage may have been caused by a non-wrongful cause does not succeed under the present argument but a claimant who has been the victim of a wrong but who cannot identify the causative wrongdoer does succeed. But, as we noted above, the law already accepts a distinction between cases where a person is seeking to rely upon the wrongdoing of another and cases where a person is seeking to point to a non-wrongful event. Moreover, it is possible to justify differential treatment by tort law of these persons precisely because the second person has been shown to have suffered a tortious wrong. (85) Thirdly, it would bring the law into line with the position taken by most American States, Canada, Germany and France--all of which allow a special limited exception to proof of causation on the balance of probability in indeterminate wrongdoer cases. (86)
A first objection is made by Professor Stevens. The result of the rule being proposed is that 'an employer who employed the claimant for one day during a 40-year career in working with asbestos with multiple employers would be held liable in full for the injury, although the overwhelming probability must be that he was not its cause'. (87) A first point is that, if the defendant is able to prove on the balance of probability it was not in fact a cause, it should not be liable. Fairchild should only apply where one cannot ascertain on the balance of probability who caused the damage. A second point is that, under the proposed rule, liability would be in proportion to the risk so this intuitively unfair result would be substantially mitigated.
A second objection is that the argument involves illegitimately looking beyond the bipolar relationship of the defendant to the claimant and so violates the principle of 'correlativity' which states that the only normative considerations relevant to adjudicating between two individuals are considerations which link the two parties together in the one 'normative relationship'. (88) On this objection, the fact that the defendant is relying upon another person's wrong is not sufficiently related to the relationship of the defendant to the claimant. A first response is that the principle of correlativity must have limits. If a third party will kill one hundred innocent people unless a judge wrongly holds a defendant liable in damages, the judge should hold the defendant liable in damages. There are disputes as to where these limits lie. The argument made here suggests another limit which is already recognised in other parts of the law on causation in tort. A second response is that the 'correlativity' principle will lead to very strange results in much more ordinary cases. Consider the following case. D1 and D2 independently wrongfully cause a minor explosion near where C1 and C2's cars are parked. Each explosion destroyed one car but it is impossible to say which explosion caused which car to be destroyed. It seems extremely unjust to say that C1 should not be able to sue either D1 or D2. Both defendants have wrongfully caused damage to one of the claimants and may have caused that damage to the particular claimant, who is certainly the victim of a wrong at the hands of one of the defendants. (89)
A third objection to the argument is that, just like the 'empty duty' argument', it works an injustice against the causally innocent defendant in Fairchild. Whereas the 'empty duty' argument uses the causally innocent defendant as a means to pursue the socially beneficial goal of deterring future wrongs or harms (and so commits an injustice), the 'reliance upon another's wrong' argument uses the causally innocent defendant as a means to compensate the claimant in respect of a wrong for which the defendant is not responsible. Either way, the defendant is being held responsible for loss which it did not cause--what does it matter (to the defendant's responsibility) that that claimant's loss is a result of another person's wrongdoing?
There is certainly force in this objection. But it has its own problems. It mischaracterises the injustice inflicted in cases like Fairchild. The injustice inflicted by the court in a Fairchild case is not constituted by the fact that one of the defendants is held liable for a harm for which it is not responsible. This injustice occurs whenever a person is held liable for a harm which it did not cause. But so long as we are confident on the balance of probability that the defendant caused the harm, then (even if the defendant did not cause the harm in reality) the injustice is acceptable. (90) The (potentially unacceptable) injustice is rather that the court is wrongfully imposing too great a risk of an incorrect verdict against the defendant. If the defendant is entitled minimally to proof on the balance of probability of causation, then by insisting upon something less, the court is taking too great a risk of a false positive on the causal question. The pertinent question, then, is really: why should the presence of another wrongful actor affect the defendant's entitlement that a case be proved against it on the balance of probability?
The response is that whilst it may be impermissible to knowingly risk a false positive where the risk is greater than 50% (imagine the wrong committed by holding D liable where it is shown on the balance of probability that D did not cause the damage) it is not so obvious that taking the risk of a false positive against the defendant is impermissible where the risk of a false positive is unknown and unknowable. In Fairchild cases it is impossible to assess the probability that the defendant caused the damage. In cases where it is impossible to decide the matter according to the substantive law according to the governing standard of proof (a burden of proof situation), it becomes more plausible to say that other normative considerations can outweigh the defendant's general entitlement not to be liable absent proof of causation. (91) One such consideration is the fact that the causative defendant is relying upon another's wrong to evade liability. (92)
This article has been highly critical of the decision in Fairchild. It created explanatory and analytical incoherence in the law. It led to unjust developments in the law by its failure to restrict the exceptional rule being created to the normatively singular indeterminate wrongdoer situation where the causative defendant is relying upon another's wrongful conduct to evade liability.
(1)  1 AC 32 (Fairchild).
(2) Cf. e.g. Sienkiewicz v Greif (UK) Ltd.  UKSC 10 (Sienkiewicz): 'It is a basic principle of the law of tort that the claimant will only have a cause of action if he can prove, on the balance of probabilities, that the defendant's tortious conduct caused the damage in respect of which compensation is claimed.'  per Lord Phillips; see also  per Lord Rodger (describing this as the 'usual' rule);  per Lord Brown and  per Lord Kerr (both describing it as the 'normal' rule).
(3) Some cases use the language of 'material contribution' in circumstances where it would be perfectly possible to say that but for D's wrongful conduct, (part of) C's damage would not have occurred. In such cases, the language of 'material contribution' is unnecessary and obfuscating. See, on this: S H Bailey, 'Causation in Negligence: What is a Material Contribution?' (2010) 30 Legal Studies 167. There is, however, a non-redundant concept of 'material contribution' according to which D materially contributes to C's damage if and only if D's wrongful conduct played a more than minimal role in a mechanism which was causally sufficient for the claimant's damage. For a more precise analysis of the non-redundant concept, see S Steel and D J Ibbetson, 'More Grief on Uncertain Causation in Tort'  70 Cambridge Law Journal 451.
(4) Fairchild v Glenhaven Funeral Services Ltd  1 AC 32. Although the employees in Fairchild were accepted to have been the victims of a complete tort on the balance of probability (i.e. one or more defendants had wrongfully caused the employee's mesothelioma) and so all the potential causes of the employee's mesothelioma were wrongful, the decision was not limited to such situations. Rather, Fairchild was premised upon the earlier decision of McGhee v National Coal Board, 1973 SC (HL) 37;  1 WLR 1, where C suffered dermatitis either through dust attacking his skin non-wrongfully whilst at work or through D employer's wrongful failure to provide showers. Already, then, from the decision in Fairchild, it was difficult to argue that it could be restricted to cases where all potential causes of C's damage were wrongful. See further, below, pp 245-246.
(5) Barker v Corus (UK) Ltd.  2 AC 572.
(6) Section 3 of the Compensation Act 2006 entitled 'Mesothelioma: Damages' states at Section 3(1) that 'This section applies where--(a) a person ('the responsible person') has negligently or in breach of statutory duty caused or permitted another person ('the victim') to be exposed to asbestos, (b) the victim has contracted mesothelioma as a result of exposure to asbestos, (c) because of the nature of mesothelioma and the state of medical science, it is not possible to determine with certainty whether it was the exposure mentioned in paragraph (a) or another exposure which caused the victim to become ill, and (d) the responsible person is liable in tort, by virtue of the exposure mentioned in paragraph (a), in connection with damage caused to the victim by the disease (whether by reason of having materially increased a risk or for any other reason).' Section 3(2) then provides that: 'The responsible person shall be liable--(a) in respect of the whole of the damage caused to the victim by the disease (irrespective of whether the victim was also exposed to asbestos--(i) other than by the responsible person, whether or not in circumstances in which another person has liability in tort, or (ii) by the responsible person in circumstances in which he has no liability in tort), and (b) jointly and severally with any other responsible person'.
(7) It seems clear that Fairchild cannot apply in the medical negligence context in claims against the N.H.S: see Gregg v Scott  2 AC 176 at ,  per Lord Hoffmann and at  per Lady Hale; Barker v Corus  2 AC 572 at  per Lord Hoffmann.
(8) Fairchild,  1 AC 32 ,  per Lord Bingham.
(9) McGhee was criticised in obiter remarks in Sienkiewicz, above n 2, by Lord Phillips  and Lord Brown . These obiter remarks do not alter the fact that McGhee was treated as an important authority for the decision in Fairchild.
(10) Brown v Corus (UK) Ltd.  PIQR P30,  EWCA Civ 374. See also the earlier obiter dicta in Transco v Griggs  EWCA 564 at  per Hale LJ that Fairchild applied in a Vibratory White Finger case and the view of Smith LJ in Novartis (Grimsby) Ltd. v Cookson  EWCA Civ 1261 that Fairchild could not be applied to bladder cancer because it was not 'analogous to mesothelioma'  (implicitly accepting, then, that an analogy could be made to mesothelioma in cases of other diseases). It is true that Fairchild has also not been applied in cases which do not involve mesothelioma. But these cases are readily explicable that in Fairchild was not applicable for some reason other than simply that the disease was not mesothelioma. This is true of: Sanderson v Hull  EWCA Civ 1211 (the uncertainty over causation was not due to the lack of scientific knowledge); Clough v First Choice Holidays and Flights Ltd.  EWCA Civ 15 (the same as Sanderson); AB v Ministry of Defence  UKSC 9 (there were multiple different possible causative agents).
(11) Sienkiewicz  UKSC 10  per Lord Rodger; at ,  per Lord Phillips;  per Lord Kerr.
(12) Admittedly, Lord Brown attempted to restrict Fairchild to mesothelioma cases. His Lordship said that 'Save only for mesothelioma cases, claimants should henceforth expect little flexibility from the courts in their approach to causation': Sienkiewicz, above, n.2, at . None of the other Justices spoke in these very restrictive terms however. For a similar view of the effect of Sienkiewicz, see P Laleng, 'Sienkiewicz v Grief (UK) Ltd. and Willmore v Knowsley Metropolitan Borough Council: A Material Contribution to Uncertainty?' (2011) 74 Modern Law Review 777, 786-787.
(13) Summers v Tice, 33 Cal 2d 80; 199 P 2d 1 (California Supreme Court, 1948).
(14) Cook v Lewis  SCR 830.
(15) Sienkiewicz  UKSC 10 .
(16) Some passages in Ernest Weinrib, The Idea of Private Law (Harvard University Press, 1995) would tend to suggest that the author held this view of the value of coherence.
(17) For criticism of Fairchild on the ground that it hardly goes beyond an appeal to very broad notions of fairness, see Jonathan Morgan, 'Causation, Politics and Law: Reflections upon the English--and Scottish--Asbestos Saga' in R Goldberg (ed), Perspectives on Causation (Hart, 2011).
(18) Fairchild,  per Lord Hoffmann. See also Fairchild at  per Lord Bingham and at  per Lord Rodger: 'The substantive duty of care would be emptied of all practical content so far as victims are concerned'. A similar consideration had been advanced in McGhee.
(19) Fairchild, . Cf similarly: Fairchild, at  per Lord Bingham.
(20) Fairchild,  per Lord Bingham.
(21) This fact will not cause a claim to fall foul of the requirement: Barker, at  per Lord Scott.
(22) Fairchild,  per Lord Bingham.
(23) Fairchild, .
(24) Fairchild, ; Lord Hoffmann did not endorse the single agent requirement in Fairchild but later did in Barker v Corus Ltd.  2 AC 572, 587. For an obiter approval of Lord Rodger's formulation: Novartis Grimsby Ltd. v Cookson  EWCA Civ 1261 at .
(25) Similarly: Chris Miller, 'Causation in personal injury: legal or epidemiological common sense', 26 Legal Studies 533, 560-563; Robert Stevens, Torts and Rights (Oxford University Press, 2007) 51; Allan Beever, Rediscovering the Law of Negligence (Hart, 2007) 475; Morgan, above, n 17, 61-62. That Fairchild (at least where there is a potential non-wrongful cause) is limited to cases of scientific uncertainty was confirmed by the Court of Appeal in Sanderson above n 10.
(26) Another example outside of the mesothelioma context is the mass manufacture of generic defective drugs.
(27) This is unclear. See below, at pp.253-256.
(28) Cf. the facts of Sindell v Abbott Laboratories 607 P 2d 924 (Cal 1980).
(29) See David Kaye, 'The Limits of the Preponderance of Evidence Standard: Justifiably Naked Statistics and Multiple Causation' (1982) American Bar Foundation Research Journal, 487. It is arguable that cases like Wilsher v Essex Area Health Authority  AC 1074 also display sufficient ex ante predictable causal uncertainty so as to diminish D's incentives to conform to its duty of care. The claimant baby suffered retrolental fibroplasia which caused him to be blind. One possible cause of his condition was the defendant doctor's negligent over-administration of oxygen. However, a number of other possible (non-wrongful) conditions from which he likely suffered may have caused it. In Wilsher, the fact of multiple different agents made no difference to the inability to identify the causative agent.
(30) Above, p.246. Lord Phillips' suggested rationalisation of the single agent rule in Sienkiewicz at  cannot be accepted. His Lordship stated that: 'The possibility that mesothelioma may be caused as the result of the cumulative effect of exposure to asbestos dust provides a justification, even if it was not the reason, for restricting the Fairchild/Barker rule to cases where the same agent, or an agent acting in the same causative way, has caused the disease, for this possibility will not exist in respect of rival causes that do not act in the same causative way.' From the context in which this suggestion appears, its argument seems to be that it is the possibility of a cumulative mechanism which makes proof by statistical evidence impossible. That is partly true (cf below, text at n 48). It remains the case, however, that this cannot be the only situation in which proof is predictably impossible to establish--which is what is of normative significance for the 'empty duty' argument.
(31) As Jonathan Morgan has observed, this aspect of the reasoning in Fairchild reads as if it were a 'legislative attempt to carve out an exceptional category', in so far as its emphasis upon the particular is akin to the statutory law's ability 'to enunciate precise, isolated legal rules which apply strictly to the situations intended by Parliament, but have no wider effect': Morgan, above n 17, 58, 65.
(32)  UKSC 9.
(33) Perhaps the clearest passages to this effect are Barker, - (Lord Hoffmann); ,  (Lord Scott);  (Lord Walker).
(34) One might also add 'logically incoherent': if it is impossible to prove causation on the balance of probability, then one cannot assess the probability that the defendant caused the damage. But, then, how can one assess the risk (the probability of harm) to which the defendant exposed the claimant?
(35) Section 1(1).
(36) Fairchild , .
(37) Fairchild .
(38) Lord Hoffmann, 'Causation', in Richard Goldberg (ed), Perspectives on Causation (Hart, 2011) 9.
(39) See above, p.250.
(40)  UKSC 14. For a discussion of this aspect of the decision, see Nicholas McBride and Sandy Steel, 'The Trigger Litigation' (2012) 28 Professional Negligence 285.
(41) Trigger . Cf. Lord Clarke, who similarly stated that Fairchild imposes liability upon 'the employer who can fairly be said to have caused the disease' at .
(42) HLA Hart and AM Honore, Causation in the Law (Oxford, 2nd ed, 1985).
(43) For an elegant simplification of this idea of causes as making a difference to normal states of affairs: Alex Broadbent, 'The Difference Between Cause and Condition' (2008)108 Proceedings of the Aristotelian Society 355.
(44) Where the argument may differ in part from a standard deterrence argument is in what the argument considers to be the undesirable consequence which grounds the need to deter certain kinds of conduct. Normally, deterrence arguments assume that certain kinds of conduct need to be deterred, where the reason for deterring that conduct is ultimately to reduce the overall level of harm in the world (or to maximise the level of wealth in the world). It is possible that the House of Lords in Fairchild thought that part of the reason for deterring breaches of duties to take care was that it would reduce the overall number of wrongs (breaches of duties of care) committed in the future, as well as reducing the number of harms (asbestos-related cancers).
(45) E.g. under Sections 2 and 3 of the Health and Safety at Work Act 1974. Note the recent fine given to the retailer Marks and Spencer for breach of these provisions: 'Marks and Spencer hit with 1m[pounds sterling] asbestos fine', The Telegraph, 27 September 2011.
(46) E.g. J Stapleton, 'Lords a'leaping evidentiary gaps' (2002) 10 Torts Law Journal 1.
(47) Cf. Holtby v Brigham & Cowan (Hull) Ltd.  3 All ER 421.
(48) Cf. J Stapleton, 'Factual Causation, Mesothelioma and Statistical Validity' (2012) 128 Law Quarterly Review 221, 226: 'Suppose that bladder cancer can only be caused by a particular agent (amines) and a victim had been, for exactly the same period of time, simultaneously exposed to only two sources of that agent: occupational amine exposure and amines contained within cigarette smoke. Suppose further that the volume of the agent from occupational exposure was considerably more than double that from the smoking exposure. In such circumstances, it would be possible to infer that it was probable that the dominant source of the agent, occupational exposure, contributed to the disease: this is because whatever the mechanism of the cancer, it is probable that the occupational exposure was involved--for example by being the probable source of the single unit (if that were the mechanism) or by probably contributing to the threshold (if that were the mechanism) and so on. (Similar reasoning might be used to help establish that one source of 'enormous quantities' of asbestos relative to other simultaneous sources probably contributed to a claimant's mesothelioma.) But it is not a corollary of this proposition that the minor source of risk probably did not contribute: for example, if the mechanism involved a threshold, that source would have contributed to it.'
(49) And by the 'natural' incentives provided by morality. Cf. Stevens, above n 25, 322: 'The vast majority of drivers take care not to run over other people because they consider it morally wrong to carelessly injure others.'
(50) Polinsky and Shavell argue that it is often the case that sufficient incentives are provided by market forces and strong regulatory regimes in the context of manufacturers of consumer products. They go so far as to argue that, in those situations, product manufacturers should not be subject to any tort liability. It may be true, then, that there are sufficient incentives provided by alternative institutions in other contexts. See A. Mitchell Polinsky and Steven Shavell, 'The Uneasy Case for Product Liability' (2010) 123 Harvard Law Review 1437.
(51) The empirical evidence available as to the effect of tort liability upon the level of accidents is 'far from unequivocal': P Cane, 'Tort Law as Regulation' (2002) 31 Common Law World Review 305, 310. Similarly: Stevens, above n 25, 322-323: 'The evidence available for the deterrent impact of the law of torts is thin, and the results are, at best, equivocal'. Perhaps the most famous empirical study is: D Dewes, D Duff, M Trebilcock, Exploring the Domain of Accident Law (Oxford University Press, 1996) whose authors often offer agnostic conclusions in the face of equivocal results (e.g. 205).
(52) This objection may seem inconsistent with the last objection. The last objection said that it may not be necessary to provide incentives to conform to duties under tort law in cases of recurrent causal uncertainty because of the existence of alternative institutional incentives. The second objection says that there are not particularly strong empirical credentials to the 'empty duty' argument. But if we know little about the empirical effects of tort law, how can we object that there are or may be alternative incentives--is that not to enter into the empirical fray without any empirical evidence? Not so. The first objection takes the incentive argument on its own merits--accepting the various (empirical) assumptions that defendants are rational, engage in cost-benefit analyses, that judges do not misapply the standard of care (etc.) and says that even given those assumptions, the argument may not work. The second objection questions the weight one should give to those empirical assumptions, given our uncertainty.
(53) The position most often associated with Weinrib, above n 16. Others who exclude consequentialist considerations are: Beever, above n 25.
(54) Though cf. s 3(1) Compensation Act 2006.
(55) Compare, however, the (inconsistent and problematic) use of such arguments particularly in cases involving public authorities: Smith v Chief Constable of Sussex Police  UKHL 50.
(56) Of course it might be objected that tort law is inherently distributively unjust. It allows the victims of wrongfully caused injury damages against defendants but it does not award damages to those who suffer serious injury without anyone's fault. The objection is misguided. It is defensible that a person who has not been wrongfully injured by a defendant cannot successfully sue that defendant whilst a person who has been wrongfully caused injury by that person can. But this restriction becomes difficult to defend if the reason the defendant is being made liable is to encourage the reduction of overall wrongs or harms. There is then no reason (beyond a debatable reason of efficiency) to restrict the claim to those who have wrongfully suffered. For a similar point, see Arthur Ripstein and Benjamin Zipursky, 'Corrective Justice in an Age of Mass Torts' in Gerald Postema (ed), Philosophy and the Law of Torts, (Cambridge University Press, 2001) 230.
(57) Ronald Dworkin, Law's Empire (Harvard University Press, 1986) 178.
(58) Ariel Porat, 'Misalignments in Tort Law' (2011) 121 Yale Law Journal 82, 108.
(59) For discussion of the value of 'integrity', see TRS Allan, 'Law, Justice, Integrity: The Paradox of Wicked Laws' (2009) 29 Oxford Journal of Legal Studies 705.
(60) The arguments made in the text should not be confused with another type of objection to the 'empty duty' argument. We can call this the reductio objection. The reductio objection says that if we follow the 'empty duty' argument to its logical conclusion, there is no need to insist upon causation in any case. This is because optimal deterrence would be achieved by a system of liability for pure risk creation. For this objection: D P Nolan, 'Causation and the Goals of Tort Law' in Andrew Robertson and Tang Hang Wu (eds), The Goals of Private Law (Hart, 2009) 187-190. The problem with this objection is (i) that not all economists would agree that this is so (see e.g. C Grechenig and A Stremitzer, 'Der Einwand rechtmaBigen Alternativverhaltens--Rechtsvergleich, okonomische Analyse und Implicationen fur die Proportionalhaftung' (2009) 73 Rabel Journal of Comparative and International Private Law 336, 358-359); and (ii) it might be argued that tort law should be reformed better to represent this consequentialist model.
(61) E.g. Marcel Kahan, 'Causation and the Incentives to Take Care under the Negligence Rule', (1989) 18 Journal of Legal Studies 427.
(62) Grechenig and Stremitzer, above n.60, 358-359.
(63) For an early attempt at explaining why causation matters: Judith Jarvis Thomson, 'Remarks on Causation and Liability' (1984) 13 Philosophy & Public Affairs 101. For doubts as to the importance of causation as a necessary condition of being subject to a liability to compensate someone: Christopher Schroeder, 'Causation, Compensation and Moral Responsibility' and Jeremy Waldron, 'Moments of Carelessness and Massive Loss', both in David Owen, (ed) Philosophical Foundations of Tort Law (Oxford University Press, 1995).
(64) E.g. John Gardner, 'Obligations and Outcomes in the Law of Torts' in Peter Cane and John Gardner (eds), Relating to Responsibility: Essays for Tony Honore (Hart, 2001).
(65) For this claim about the law: Nicholas McBride, 'Duties of Care: Do They Really Exist?' (2004) 24 Oxford Journal of Legal Studies 417.
(66) C.f. Nolan, above n.60, 175; Beever, above n 25, 446.
(67) Nils Jansen, Die Struktur des Haftungsrecht (Tubingen, 2003) 132.
(68) See Jeff McMahan, 'The Basis of Moral Liability to Defensive Killing' (2005) 15 Philosophical Issues 386. For the view that causation is important but not determinative, see: Victor Tadros, The Ends of Harm (Oxford University Press, 2011) 191-196.
(69) It is only arguable as there are obvious potential counter-examples. It might be said that liability in unjust enrichment does not enforce a moral duty to repay the enrichment, for example. (See, e.g., Stephen Smith, 'Duties, Liabilities and Damages' (2012) 125 Harvard Law Review, 1727. Looking more generally to state interference, it is difficult to justify punishment as enforcing a moral duty. For an attempt to do so, see Tadros, above n 68.
(70) For a plausible explanation of the sense in which secondary obligations derive from breach of primary obligations, see John Gardner, 'What is Tort Law For? Part 1: The Place of Corrective Justice' (2011) 30 Law and Philosophy 1.
(71) See the first two objections, above, p.253-254.
(72) See above, p.247-248.
(73) Other potential justifications are discussed in Chapters 2 and 3 of Sandy Steel, Proof of Causation in Tort Law (Cambridge University Press, forthcoming).
(74) E.g. Stephen Perry, 'Risk, Harm and Responsibility' in David Owen (ed), Philosophical Foundations of Tort Law (Oxford University Press, 1995) 334-336.
(75) See, at length, Sandy Steel, 'Rationalising Loss of a Chance in Tort', in E Chamberlain, J. Neyers, S. Pitel (eds) Challenging Orthodoxy in Tort Law (Hart, forthcoming 2013).
(76) Ariel Porat and Alex Stein, Tort Liability Under Uncertainty (Oxford University Press, 2001) 160.
(77) Ibid 161.
(78)  1 AC 750.
(79) For related suggestions: Mark Geistfeld, 'The Doctrinal Unity of Alternative Liability and Market Share Liability' (2006) 155 University of Pennsylvania Law Review 447 (who endorses a similar type of argument); Jonathan Morgan, 'Lost Causes in the House of Lords' (2003) 66 Modern Law Review 277. C.f. Stevens, above n 49, 147 (who rejects it).
(80)  AC 467.
(81) Jobling v Associated Dairies Ltd.  AC 794.
(82) Although each defendant is a NESS cause of the death (a necessary element of a set of conditions sufficient for the death), it is still the case that damages ought generally not to be awarded in cases where the defendant has not made the defendant worse off. So even if C shows that D is a NESS cause of C's loss, it remains for C to show that C is worse off as a result of D's wrongful conduct. Separating this question from the causal question, see Jane Stapleton, 'Cause-in-Fact and The Scope of Liability for Consequences' (2003) 119 Law Quarterly Review 388, 412-417.
(83) It will be objected that the claimant has a right to be free from the negligent infliction of injury and that therefore the appropriate counterfactual world in assessing whether each defendant has made the claimant 'worse off' is one in which the claimant does not suffer physical injury through the negligent wrongdoing of anyone. But the claimant has no such right as against each defendant. The defendant owes the claimant a duty not negligently to cause the claimant physical injury, not a duty to protect the claimant from the wrongdoing of third parties. C.f. Arthur Ripstein, 'As If It Had Never Happened' (2007) 48 William and Mary Law Review 1957.
(84) It could be argued that the robbers would have been liable on the basis that they destroyed the claimant's ability to sue the defendant for losses which would have happened anyway and that this was not too remote a loss because the defendants had behaved intentionally. But the decision in Baker precludes this argument. It is in any event seriously problematic: see Nicholas McBride and Sandy Steel, 'Suing for the Loss of the Right to Sue: Why Wright is Wrong' (2012) 26 Professional Negligence 27.
(85) '[T]here seems to be a significant leap between making whole a plaintiff who has suffered a tortious injury and allowing recovery to one whose injury might not otherwise fall within the scope of tort law': Ernest Weinrib, 'A Step Forward in Factual Causation' (1975) 38 Modern Law Review 518, 525.
(86) For the United States, France, and Germany, see Sandy Steel, 'Sienkiewicz v Greif and Exceptional Doctrines of Natural Causation' (2011) 2 Journal of European Tort Law 294. The Canadian position is now reflected by Clements v Clements 2012 SCC 32 which accepts a limitation to indeterminate wrongdoer cases.
(87) Stevens, above n 49, 147.
(88) On 'correlativity', see generally Ernest Weinrib, 'Correlativity, Personality, and the Emerging Consensus on Corrective Justice' (2001) 2 Theoretical Inquiries in Law 107. For reliance upon 'correlativity' in the context of an argument against liability in the indeterminate wrongdoer situation, see Nolan, above n 60, 174-176.
(89) The only way around this case for those committed to 'correlativity' seems to be along the following lines. In the example of two cars, each D knows that it has causally wronged someone and therefore that it ought to compensate someone. Given this knowledge, each D cannot simply do nothing. Rather, D's secondary obligation arising out of the commission of a wrong is to make an agreement with the other defendant whereby they jointly authorise each other to compensate the claimants in respect of their wrong. If the defendants fail to conform to this secondary obligation, perhaps the courts can hold them liable for that. I develop this as a potential solution to mass tort cases in which D is known to have wrongfully caused damage in Steel, above n 73, Chapter 2.
(90) For a helpful discussion of the morality of standards of proof, see Ronald Dworkin, A Matter of Principle (Harvard University Press, 1985) Chapter 3.
(91) C.f. Ripstein and Zipursky, above n 56, 242.
(92) Another, peculiar to Fairchild itself, is that it was extremely probable that an injustice would be committed against the claimant in Fairchild but only possible that an injustice would be committed against both defendants. If the mechanism of mesothelioma is cumulative, both defendants may have played a role in its occurrence.
Sandy Steel, Lecturer in Law, King's College London. Email: firstname.lastname@example.org.…