Hamer, David --- "Chance Would Be a Fine Thing': Proof of Causation and Quantum in an Unpredictable World" [1999] MelbULawRw 24; (1999) 23(3) Melbourne University Law Review 557

'Chance Would Be a Fine Thing': Proof of Causation and Quantum in an Unpredictable World

DAVID HAMER[*]

[The courts have long recognised that, while the past is a matter of historical fact, the future is affected by chance. Past losses, proved on the balance of probabilities, can be recovered in full. Future losses, though proven to lesser degrees of probability, can still be recovered in part. However, chance also affects the causation and quantum of past losses. These elements pose the predictive question, ‘what would have happened had the defendant not committed the breach?’ On occasions, courts have responded by awarding proportional damages. Where the defendant’s breach deprives the plaintiff of a commercial opportunity, the plaintiff may be awarded a proportion of the possible profits, even though the profits would probably not have crystallised. In personal injury actions, courts have been reluctant to award only partial compensation. Where the defendant’s breach exposes the plaintiff to the risk of injury, causation may be presumed, and full compensation awarded, even though the injury was probable in any case.]

INTRODUCTION

It might be thought a safe generalisation that the standard of proof for the elements of civil actions is ‘more probable than not’, otherwise known as the ‘balance of probabilities’.[1] Suppose, for example, the plaintiff is a passenger in a car driven by the defendant when the defendant drives through a red light and collides with a bus. The plaintiff claims damages for a broken leg. The plaintiff will be required to establish (i) that the defendant owed her a duty in tort, contract or under statute[2] to drive at a certain standard; (ii) that the defendant breached that duty; (iii) that the breach caused the plaintiff’s injury; (iv) that the injury is not too remote from the breach; and (v) the nature or quantum of the plaintiff’s loss.[3] The plaintiff should establish each element on the balance of probabilities, and will then be entitled to full compensation. If the plaintiff fails to establish any of these elements on the balance of probabilities, however small the failure, she will receive no compensation. This is the traditional ‘all or nothing’ approach to compensation.

One well-recognised exception or refinement[4] to this traditional approach concerns future or post-trial losses flowing from the defendant’s breach.[5] Suppose, in the example given above, there is a 40 per cent chance of the plaintiff requiring an operation a few years after the trial at a cost of $20,000. The plaintiff will not be denied compensation on the ground that, on the balance of probabilities, the operation will not take place. Since the operation is a future event, proof on the balance of probabilities is not required. Proportional compensation will be awarded, based upon the degree of proof attained. The plaintiff will recover $8,000 — 40 per cent of the operation’s full cost.

This distinction drawn between the past and the future is not limited to the law. As discussed in the next section, philosophers have suggested that our everyday experience of time is asymmetric. The past has already happened and is, in principle, knowable. The future, on the other hand, is a matter of chance, and is a far less certain object of knowledge.

The complications that chance presents for civil proof, however, are not confined to post-trial events. Causation, even in application to past events, has a predictive element. Suppose the plaintiff in the broken leg case was on a business trip. She had employed the defendant’s chauffeur service to drive her to a meeting at which she hoped to clinch a lucrative deal. Because of the accident, she missed the meeting and a competitor won the contract. Can she claim damages for the loss of profits associated with that deal? This is a pre-trial loss, but the traditional standard of proof appears too strict. The causation inquiry is counterfactual, requiring a comparison between what happened following the defendant’s breach and what would otherwise have happened. It is a matter of historical fact that the plaintiff was not awarded the contract, and it seems appropriate that this should be established on the balance of probabilities. However, to determine what would have happened without the defendant’s breach requires a hypothetical prediction from a time immediately prior to the defendant’s breach. The prediction is hypothetical since it is premised on the defendant not committing the breach, which ex hypothesi they did in fact commit. If the balance of probabilities standard with ‘all or nothing’ compensation is inappropriate for future losses, it also appears inappropriate for the predictive component of causation.

The same difficulty arises with the quantum of past losses. Quantum, like causation, is counterfactual.[6] Its goal is to place the plaintiff in the position they would have been in without the defendant’s breach. It is arguable that for the predictive component of pre-trial losses the traditional ‘all or nothing’ approach to quantum is inappropriate, just as it is inappropriate for post-trial losses. The counterfactual nature of causation and quantum is discussed in Part II.

This article examines a number of principles and doctrines that the courts have developed in response to the difficulties of proof arising from the counterfactual nature of causation and quantum. As discussed in Part II(B)(3), in a recent decision of the High Court of Australia, Malec v J C Hutton Pty Ltd,[7] proportional damages were awarded for pre-trial losses to take account of predictive uncertainty. It was found on the balance of probabilities that the defendant had caused the plaintiff’s injury. However, it was probable that the plaintiff would have suffered an injury of that kind prior to the trial without the defendant’s breach, and in view of this certain heads of damages were discounted by more than 50 per cent. In another recent decision, Sellars v Adelaide Petroleum NL,[8] discussed in Part III(A), the High Court awarded proportional compensation where the defendant’s actions deprived the plaintiff of a pre-trial commercial opportunity, even though the opportunity was unlikely to have crystallised into actual profit.

In Malec, the High Court adopted a proportional approach to damages, and in Sellars it adopted a proportional approach to causation. However, while Malec appears to have laid down an approach of general application, the principles developed in Sellars may be limited to certain kinds of loss. Sellars contained statements that the causation concept does not distinguish between different causes of action and different types of loss. The loss of chance approach has been utilised in a number of cases where the negligence of the defendant legal adviser led to the plaintiff client’s claim against an alleged wrongdoer being statute-barred.[9] The approach was extended to medical negligence by a single judge of the Supreme Court of New South Wales in Tran v Lam.[10] More recently, however, in Chappel v Hart[11] and Naxakis v Western General Hospital[12] a number of members of the High Court have suggested that causation is context-sensitive. With the exception of Kirby J in Chappel and Callinan J in Naxakis, little support was expressed for the proposition that a plaintiff patient could recover for the loss of chance of avoiding a personal injury resulting from a defendant doctor’s negligent misdiagnosis or treatment. This suggests that the authority of Sellars may be limited to actions for lost chances of a commercial nature. The loss of chance action and its possible boundaries and exceptions are discussed in Part III.

It cannot be said, however, that courts have ignored the difficulties faced by personal injury plaintiffs in proving causation. In Part IV, a set of principles is discussed which may be more advantageous to plaintiffs than the proportional loss of chance doctrine. In Chappel and Naxakis, the High Court supported the proposition that on proof of breach and injury the court may presume the causal link between them. The evidentiary onus is shifted to the defendant to rebut this presumption and, should the defendant fail, the plaintiff will be entitled to full compensation, subject possibly to a Malec discount on quantum. In Britain, these principles were developed to deal with cases in which the plaintiff contracted a medical condition from exposure to noxious substances from various sources.[13] Though the defendant’s breach lay behind one source of exposure, this ‘guilty’ source of exposure was accompanied by one or more ‘innocent’ sources of possibly greater magnitude. The application of ‘injury within scope of risk’ principles helped the plaintiff overcome the resulting difficulties in proving causation. The defendant had contributed to the risk, and on that basis causation was presumed. The difficulty of proof then became the defendant’s problem. While the House of Lords in Wilsher v Essex Area Health Authority[14] rejected the shifting of the onus as a general proposition, a number of members of the High Court embraced it in Chappel, Naxakis and several earlier cases.[15] What is more, the High Court appears to have extended the operation of the principle beyond the noxious exposure cases to cases of medical negligence[16] and even a failure to provide legal advice.[17]

An underlying issue in this area of law is the potential conflict between considerations of logic[18] and policy. As a matter of logic, the recognition of the existence of chance seems to carry with it a probabilistic conception of causation and quantum. However, the High Court has often observed that the legal concept of causation involves considerations of value and policy as well as fact and logic.[19] The court may feel considerable sympathy for personal injury plaintiffs, particularly where the defendant is a large manufacturer, employer or occupier with liability insurance.[20] Commercial disputes in which the parties are more evenly matched and only money is at stake may invite a more strictly logical approach.

In the remainder of this introduction, I will examine in more detail chance and the asymmetric nature of time which underlie the difficulties of proof of causation and quantum.

A Asymmetric Time

Commentators in a number of fields — history, philosophy of science, law — have noted that our everyday experience of time is asymmetric. While the past appears dead, fixed and closed, the future is seen as living, plastic and open.[21] The future appears governed by chance, but there is no chance about the past.[22] A putative past event[23] has either happened or not happened.[24] Consequently, we may feel certain that it rained yesterday while only having in mind the probability of it raining tomorrow. We can check which horse won the last Melbourne Cup, but can only give the odds of the next winner. Lewis provides a powerful representation of this asymmetric view of time (Figure 1).[25] The past has taken a single path, but at the present time a number of future paths appear open.

This picture seems to capture our everyday experience of time. But it is out of step with the notion of time that motivated classical science. The goal of science is knowledge, and classical science considered many future events to be as knowable as past events.[26] ‘Classical science ... gives a very specific picture of physical existence ... it is a kind of giant clockwork in which there is no direction of time, no distinction between past and future.’[27] The paradigmatic example of classical science is Newtonian mechanics, which has been remarkably successful in predicting the motion of the planets, eclipses and the tides on earth.[28] ‘The future is hidden from us. But does an astronomer think like this when he calculates an eclipse of the sun?’[29]

A number of twentieth-century developments in science have revealed fundamental obstacles to predictive certainty which suggest that our everyday asymmetric view of time may be more realistic than the symmetrical view of classical science. In quantum mechanics, scientists have suggested that subatomic events can only be predicted to a level of probability below certainty.[30] Chaos theorists have investigated systems where future states are so sensitive to variations in the present state that the flutter of a butterfly’s wing can result in a cyclone.[31] Scientists have considered the existence of systems so complex that the largest conceivable computers would be incapable of adequately dealing with all the variables and calculations.[32] These considerations suggest that the predictability of the solar system is the exception rather than the rule.[33] The future is fundamentally unpredictable.

A determinist may object at this point that it is necessary to distinguish between the ontology and the epistemology of time.[34] There is no ontological difference between the future and the past, but only an epistemological difference. The future is just as fixed or determined as the past. The obstacles to prediction are purely epistemic — a lack of sufficiently accurate data or of computing power.[35] Admittedly, it is obstacles of this nature that result in the unpredictability in chaos theory and complexity theory — the models constructed in these fields are actually deterministic.[36] There is some basis to this objection, but it is easy to overstate. The epistemology and ontology of time are, of course, closely related: ‘[E]vidence for determinism will tend to come precisely from our ability to predict the course of events.’[37] And, if that evidence is lacking, an ontological theory about the symmetry of time appears speculative, to say the least. One commentator views the quite widespread conviction in determinism as ‘peculiar in the extreme, because events in the ordinary world are neither sufficiently predictable nor sufficiently under our control that an inference of inviolable regular succession would be warranted from our experiences’.[38]

This article is applied epistemology rather than speculative metaphysics. The following sections examine the principles governing juridical proof of past, future and hypothetical events. More useful than the distinction between the ontology and epistemology of time is the epistemological distinction between objective (un)certainty and subjective (un)certainty. The past is considered fixed, while the future is considered fluid. The past is therefore more knowable than the future. But this is not to say that we invariably have greater certainty about the past than the future. Versions of the past may compete, as do theories about the future.[39] In practice, there is uncertainty both about the past and about the future. The sharp distinction between the past and the future is objective rather than subjective. Subjective uncertainty about the past will often linger because our assessment is subject to limited evidence.[40] We consider, however, that with ‘complete’ evidence certainty would be achievable. The certainty of the past is objective both in the sense of being ‘not dependent on the mind for existence’ and in a second sense of ‘a goal or aim’.[41] Any lingering uncertainty about the past reflects an absence of evidence and a shortcoming in the proof which should perhaps be remedied, resources permitting. Uncertainty about a future event, on the other hand, may be due to incomplete evidence, but it may also reflect the inherent objective uncertainty about the future. The future is a matter of objective chance, and consequently certainty is not a valid objective for the proof of future events.[42]

Time constantly passes. A consequence of this dynamic is that, in one sense, a future event is more knowable than a past event. If an event is in the future, it is necessarily uncertain, but it is also approaching the present. There will be an opportunity for the event to be witnessed or recorded, and for the uncertainty to be resolved.[43] The event is fixed the moment it reaches the present. This of course is the basis for betting — the combination of a necessary uncertainty with an opportunity for the uncertainty to be incontrovertibly resolved. Past events, however, are forever moving into the more distant past. As time passes, the evidence can only decline in quality; as memories fade, footprints get washed away, fingerprints are smudged and documents are put through the shredder.[44]

The law’s differential treatment of pre-trial and post-trial losses appears based on a recognition of the asymmetry of time. Past losses traditionally are required to be proven on the balance of probabilities, which is usually interpreted as ‘more probable than not’ or a mathematical probability of more than 50 per cent.[45] The plaintiff can then obtain full compensation. If the plaintiff is unable to satisfy this standard, they recover nothing. Recovery for past loss is ‘all or nothing’. This reflects the view that it is objectively certain whether the injury has occurred or not. The true probability of the injury is either 100 per cent or zero. The standard of proof of 50 per cent, on the other hand, is a consequence of the lingering subjective uncertainty that often colours our perception of the past. This standard, which is half way along the probability scale, treats the parties even-handedly.[46] Apart from the placement of the onus of proof, the standard is essentially the same for both parties. If the plaintiff can establish a 51 per cent probability that they did sustain the injury, the plaintiff will succeed. If the defendant can establish a 51 per cent probability that the plaintiff did not sustain the injury, the defendant will succeed. Of course, it may not always be appropriate to treat the plaintiff and the defendant equally. In some cases and for some issues, an unbalanced standard may be appropriate.[47]

Future losses are treated quite differently. They need not be proved on the balance of probabilities — the plaintiff is entitled to proportional compensation in accordance with the probability that the loss will be incurred. This conforms to the view that there is objective chance about future events rather than objective certainty. In view of this objective uncertainty, it would be ‘unfair’[48] to demand that the plaintiff establish future loss on the balance of probabilities. Though the risk of the plaintiff’s loss may be less than 50 per cent, it may still be a real risk, an objective chance. If the plaintiff can establish that there is a real risk, they should be compensated for it.[49]

In this section, I have discussed the traditional distinction between the law’s treatment of past and future losses. In the next part, it will be argued that the concepts of causation and quantum have a predictive element even in application to past losses. For consistency with the treatment of future losses, where an uncertainty pertains to the predictive element of causation or quantum, proof requirements should be relaxed and compensation should be proportional. However, courts and commentators are resistant to change. On occasions, expressions of preference for the traditional ‘all or nothing’ approach have a deterministic flavour.[50]

II THE COUNTERFACTUAL PREDICTIVE NATURE OF CAUSATION AND QUANTUM

It is necessary to distinguish between three related issues relating to the plaintiff’s alleged pre-trial or post-trial loss: (i) whether the event alleged to constitute a loss actually occurred (or will occur); (ii) whether this event was caused by (or would be caused by) the defendant’s breach; and (iii) whether this event constitutes (or would constitute) a loss for the plaintiff, and, if so, a loss of what magnitude or quantum. The first of these questions is the most concrete and straightforward. It concerns actual events, whether pre-trial or post-trial. Is the plaintiff’s leg broken? Will the plaintiff have the operation? The second and third questions, concerning causation and quantum of loss, are more abstract and complex. These concepts involve not only actual events, but a comparison between actual events and hypothetical events — what would have happened had the defendant not committed the breach. The construction of this hypothetical sequence of events requires prediction. Even where the hypothetical events are pre-trial, they are predicted from an earlier time, prior to the defendant’s breach.

A Causation: A Counterfactual Concept

Consider first the concept of causation, which continues to generate a great deal of discussion among judges and commentators.[51] My main purpose is to reveal its fundamentally counterfactual nature, but even with this limited purpose I will need to touch upon a number of the concept’s complexities.

1 The ‘But For’ Test: Causation in Fact, ‘Value Judgments’ and ‘Policy Considerations’[52]

The counterfactual nature of the causation issue is apparent in the popular ‘but for’ causation test. In determining whether the defendant’s breach caused the plaintiff’s injury, it is helpful to ask whether the plaintiff would have suffered that injury but for the defendant’s breach. It is assumed that the defendant did commit a breach and that the plaintiff did suffer an injury. By the ‘but for’ test, this actual state of affairs is being compared with a hypothetical or counterfactual state of affairs — what would have happened if the defendant had not committed the breach.

The ‘but for’ test is not a conclusive test of causation. It is often described as a negative test, serving the purpose of ‘screening out and eliminating from further consideration factors which made no difference to the outcome’.[53] Where the ‘but for’ test is satisfied, causation may still not be recognised at law. One exception is where an event subsequent to the defendant’s breach is considered to break the chain of causation. Suppose, for example, party A negligently starts a fire, and later, as it is going out of its own accord, party B deliberately pours petrol on it. The fire then spreads and burns down the plaintiff’s house. The actions of both A and B will satisfy the ‘but for’ test, but a court may treat the action of B as breaking the chain of causation from A’s breach. This is a situation of B

irresponsibly [doing] something which transforms the outcome of A’s conduct into something of far greater consequence, a consequence not readily foreseeable by A. In such a situation, A’s act is not a cause of that consequence, though it was an essential condition of it.[54]

In this scenario, B’s conduct constitutes a novus actus interveniens.

In Haber v Walker, Smith J suggested there are two categories of

occurrence ... sufficient to sever the causal connexion ... (a) human action that is properly to be regarded as voluntary, [and] (b) a causally independent event the conjunction of which with the wrongful act or omission is by ordinary standards so extremely unlikely as to be termed a coincidence ...[55]

The fire hypothetical obviously falls into the first category. The High Court was recently divided in Chappel as to whether they were presented with a novus actus of the second category.[56]

Chappel was a medical negligence case arising out of a throat operation performed by the defendant doctor on the plaintiff. Prior to the operation, the plaintiff inquired whether it carried any risk to her voice. The doctor failed to warn the plaintiff of the slight risk that her oesophagus would be perforated and become infected, with serious consequences for her voice. Unfortunately, when the doctor operated this risk ensued. The defendant’s negligence was not in the performance of the operation, but in his failure to warn of the risk of injury.[57] At trial, the plaintiff established that, had she been warned, she would not have had the operation performed by the defendant doctor, but would have obtained the most expert and experienced doctor available. Her pre-existing condition was such that it would have been unwise to delay the surgery for long. As discussed below, there was a crucial difference of opinion between majority and dissenting judges as to whether surgery by a more experienced doctor would in fact have carried a lower risk to her voice. The defendant claimed that there was an absence of causation between the breach and the injury.

By a majority of three to two,[58] the High Court held that causation was established. The ‘but for’ test was clearly satisfied. For example, McHugh J, in dissent, acknowledged that ‘[w]ithout that failure [to warn the patient of the risk], the injury would not have occurred when it did and, statistically, the chance of it occurring during an operation on another occasion was very small’.[59] However, the dissenting judges were unwilling to make a positive finding of causation. They considered that the defendant’s negligence merely set the scene for the injury rather than being a true cause of the injury. ‘His omission to warn had nothing to do with him perforating the oesophagus on that particular day, except as one of many events that combined to place him in the theatre that day operating on the plaintiff.’[60] The operation had not been performed negligently; even had the plaintiff been warned, she still would have had the operation at an early date, and the dissenting judges accepted that the risk would have been the same had another surgeon performed the operation.[61] The cause of the injury was not the defendant’s negligence but the ‘inherent risk of the procedure’.[62] In the language of Haber v Walker, the plaintiff’s injury was an unfortunate coincidence.

The majority judges also recognised that the ‘but for’ test is not conclusive and that there were difficulties in its application to the present case. Gummow J described it as the ‘central issue ... whether there was adequate reason in logic or policy for refusing to regard the “but for” test as the cause of the injuries sustained by [the plaintiff]’.[63] However, the majority judges considered that on balance causation was established. They emphasised that the injury suffered was precisely the kind of injury that the patient had inquired about and that the doctor failed to warn of.[64] Contrary to the dissenting judges, two of the majority held that the risk would have been lower had the surgery been performed by a more experienced surgeon.[65] The injury was not an unfortunate coincidence; it was the result of the defendant’s failure to warn.

Chappel raised the difficult issue of the precise role of the ‘but for’ test in establishing causation. As Mason CJ noted in March, some commentators

subdivide the issue of causation in a given case into two questions: the question of causation in fact — to be determined by the application of the ‘but for’ test — and the further question whether a defendant is in law responsible for damage which his or her negligence has played some part in producing.[66]

On this view, in the example above, A’s action in starting the fire would be a cause in fact, however B’s action in pouring petrol on the fire just as it was going out would provide sufficient reason to consider A not legally responsible for the damage to the plaintiff’s house.

Such a division was embraced by McHugh J in March.[67] In fact, he suggested that the division should be made more explicit. Legal causation should be equated with factual causation, and, apart from exceptional cases,[68] the ‘but for’ test should be adopted as the test of factual causation. Policy considerations should be kept out of the causation issue and brought under the heading of remoteness of damage. However, the other judges in March[69] agreed with Mason CJ, who criticised this view on two grounds. First, Mason CJ indicated that this view ‘places rather too much weight on the “but for” test to the exclusion of the “common sense” approach which the common law has always favoured’.[70] (In the next section, I will consider some exceptional cases where the ‘but for’ test gives results contrary to common sense, and appears to fail, even as a test of causation-in-fact.) Secondly, Mason CJ questioned the implication that ‘value judgment has, or should have, no part to play in resolving causation as an issue of fact’.[71] Yet, the division between fact and value in causation appears quite well supported. Mason CJ himself appeared to point to a division of a similar nature earlier in his judgment in March, distinguishing between causation in ‘philosophy and science’, which is concerned with the ‘relationship between conditions and occurrences’, and causation in ‘law’, which is concerned with ‘ascertaining or apportioning legal responsibility’[72] — comments echoed by Deane J in March.[73] A number of judgments in Chappel make reference to these passages, apparently confirming the fact–value dichotomy.[74]

Later in the article, I will return to the issue of whether a particular approach to causation or quantum reflects a concern with facts or with values. My analysis in large part is analytic, aimed towards correct factual reasoning about causation. However, I am aware that it is always necessary to keep in mind that, ‘from the standpoint of other important values, an epistemologically inferior technique may on the whole be preferable’.[75] The ‘injury within scope of risk’ principles discussed in Part IV may be open to criticism on logical grounds while being sound in policy.

2 Over-Determination and the ‘NESS Test’

As noted above, the ‘but for’ test is often described as a negative rather than a positive test — if the test has a negative result, there is no causal link, but, even if the test has a positive result, causation may be denied on other grounds. There are, however, recognised though exceptional cases where the ‘but for’ test is not satisfied, but causation is considered to be established, in fact and in law. These are cases of over-determination, ‘where there are two or more acts or events which would each be sufficient to bring about the plaintiff’s injury’.[76] Consider a second fire hypothetical. Two defendants negligently start separate fires, fire X and fire Y, both reaching the plaintiff’s house at the same time, and both of which are independently sufficient to burn it down.[77] Did fire X cause the plaintiff’s house to burn down? Applying the ‘but for’ test, fire X was not a cause, since the house would have burnt down anyway as a result of fire Y. And, by the same reasoning, nor was fire Y a cause. The ‘but for’ test suggests that neither fire caused the house to burn down. But, if there were no cause, why did the house burn down? This result is contrary to common sense and unacceptable at law.

In March, Mason CJ pointed to the ‘unacceptable results’ of the ‘but for’ test in cases of over-determination, and concluded that ‘the results ... must be tempered by the making of value judgments and the infusion of policy considerations’.[78] With respect, however, the over-determination problem is one of fact and philosophy rather than value and policy. To suggest that neither fire caused the destruction of the plaintiff’s house is clearly contrary to common sense, irrespective of considerations of responsibility and compensation. If neither fire had been lit, the house would not have burnt down. Commentators have proposed an alternative counterfactual test which appears to overcome the problem.[79] According to this test, the defendant’s breach is a cause of the plaintiff’s injury if it is a necessary element of a set of conditions sufficient to bring about the plaintiff’s injury (‘NESS test’). It should be noted that the ‘sufficient set’ may be a subset of the actual conditions. In the two-fire situation, fire X is a NESS cause. Excluding fire Y, fire X would be a necessary element of a set of conditions sufficient to burn down the plaintiff’s house. And fire Y is also a NESS cause by the same reasoning.[80]

One way of expressing the difference between the two tests is to say the ‘but for’ test is one of strong necessity while the NESS test is one of strong sufficiency.[81] Neither fire X nor fire Y was necessary for the plaintiff’s house to burn down, but both were sufficient for the plaintiff’s house to burn down. Honoré suggests that, in its origins, the concept of causation was one of sufficiency. It was used to ‘provide recipes’ — in the broad sense of a ‘set of conditions sufficient to produce a given result’.[82] These recipes were learnt over time, and passed on as shared knowledge. ‘To crack a nut, you need a stone of a certain size and weight and you must bring it down on the nut with a certain force.’[83] In the two-fire case, there were two recipes for the plaintiff’s house burning down, and both can therefore be recognised as causes.

It should be noted that in most cases the NESS test will produce the same result as the ‘but for’ test, and for the same reasons may be viewed as a negative rather than a positive test. In the first fire hypothetical posed above, the action of party A in negligently starting the fire is a NESS cause of the plaintiff’s house burning down, as is the action of party B in pouring petrol on the fire as it was going out. It is not possible to create a subset of actual conditions sufficient for the result without including the actions of both A and B. However, the fire would have harmlessly burnt out had party B not poured petrol on it, and, for this reason, A may not be held legally responsible for the destruction of the plaintiff’s house. While as a test of factual causation the NESS test is superior to the ‘but for’ test in dealing with over-determination, the NESS test still needs to be tempered by value judgments and policy considerations, including the notion of novus actus interveniens.

The NESS test has gained academic support as the leading test of factual causation.[84] However, the courts have displayed little awareness or understanding of it.[85] It may be that they are unwilling to commit to a test which is more complex than the simple ‘but for’ test while still failing to be comprehensive. As noted above, the NESS test needs to be supplemented by considerations of policy and value to the same extent as the ‘but for’ test. And, as discussed in the next section, the NESS test shares with the ‘but for’ test a predictive element that presents fundamental difficulties of proof, for which further provision must be made. It is complexities such as these that led to this expression of frustration in Fitzgerald v Penn: ‘If one is once to enter on a philosophic examination of the meaning of “cause and effect”, there is no telling where one ought to stop. ... In truth the conception in question is not susceptible of reduction to a satisfactory formula.’[86] The only formula that has been endorsed with any enthusiasm is the flexible, albeit vague and question-begging[87] one of ‘common sense’.[88]

3 The Predictive Hypothetical Aspect of Causation

The ‘but for’ and NESS tests share a feature that is crucial to the issues discussed in this article. They are both counterfactual tests, involving a comparison between an ‘actual’[89] state of affairs and a counterfactual hypothetical state of affairs.[90] This is shown in Figure 2, a representation that incorporates Lewis’s asymmetric picture of time (see above Figure 1). The ‘but for’ test compares the actual course of events, in which the defendant committed a breach and the plaintiff sustained an injury, with what would have occurred had the defendant not committed the breach. The NESS test draws a comparison between the events that actually followed the sufficient set of conditions and the events that would have followed the sufficient set without the alleged necessary element, the defendant’s breach. ‘It is only by comparing these two sets of facts (one actual and one hypothetical) that the influence or effect of the negligent act or omission can be judged.’[91] In Figure 2, by drawing this comparison it can be said that the defendant’s breach caused events to take one path rather than another.

In asking what would have occurred but for the defendant’s breach, the counterfactual tests pose a predictive question. The fact-finder is required to consider the situation prior to the defendant’s breach and to predict the events that would have followed had the defendant not committed the breach. In this respect, the task of determining causation resembles the task of predicting future loss. Given the existence of objective chance, such predictions present fundamental proof difficulties. As noted in Part I, the proof requirements for the quantum of future loss have been relaxed — proportional compensation may be awarded for even improbable future losses. Traditionally, however, causation must be proved to the traditional standard — on the balance of probabilities — despite the element of prediction. For example, Gaudron J in Bennett declined to accept that,

where questions of causation depend on hypothetical considerations, allowance should be made ... for the possibility that some event would not have occurred. ... [Q]uestions [of causation] are not answered ‘maybe’, or even, ‘more probably than not’. They are answered ‘yes’ or ‘no’ depending on the probabilities for or against. In this respect they are indistinguishable from the question whether an event happened.[92]

Wright, a prominent supporter of the NESS test, has expressed a similar view. He suggests the term ‘counterfactual’ ‘may be inappropriate — and certainly has proven misleading’.[93] He argues that the causation issue is ‘factual’ and ‘determinate’. However, his view appears based upon a deterministic vision of the world. He suggests that ‘[m]any apparently indeterministic processes turn out on closer examination to be deterministic, and as far as we know all apparently indeterministic processes may be deterministic’.[94] ‘Any indeterminacy results solely from defects in our empirical knowledge, rather than from the nonexistent open-ended nature of the counterfactual context.’[95] Honoré apparently endorses Wright’s view. He disagrees with the view that ‘counterfactual propositions cannot be true or false. The tort process ... assumes, in my view rightly, that they can be. We can often know what, on a certain hypothesis, would probably happen in the real world.’[96] However, while not alone,[97] Wright and Honoré are out of step with generally accepted views of time in science and everyday life, explored in Part I(A). Moreover, in its well-established proportional treatment of future loss, the law appears to recognise the indeterminacy, or at least the unpredictability,[98] of the future.

B Quantum of Loss: Restitutio in Integrum

1 Similarities and Differences Between the Causation and Quantum Issues

Having determined that the defendant’s breach has caused the plaintiff’s loss, it becomes necessary to assess the quantum of the plaintiff’s loss. Quantum is a causal issue, and it also entails a comparison between an actual and a hypothetical state of affairs.[99] As a consequence of the defendant’s breach, the plaintiff is in a particular position. The goal of awarding compensation is generally acknowledged to be restitutio in integrum — to put the plaintiff ‘in the position that she would have occupied if the wrong had not been done’.[100] This involves a hypothetical inquiry.[101] Indeed, it is dangerously hypothetical. As Waddams points out, the court is invited to ‘pursue a potentially limitless inquiry into the precise circumstances that would have attended the plaintiff if the wrong had not been done’.[102] Waddams notes a tension between the restitutio goal on the one hand, and, on the other, the pragmatic need for ‘rules that are clear, predictable, workable, fair between one claimant and another in similar circumstances, and reasonably inexpensive of application’.[103]

The quantum issue has a causal nature, however there is a valid distinction between the question of whether a defendant caused the plaintiff’s loss, and the question of what quantum of the plaintiff’s losses the defendant should be held liable for. In many cases, the causation and quantum questions will have answers that correspond to a large degree. The quantum of a defendant’s liability will be reduced to the extent that the plaintiff’s losses can be attributed to some cause other than the defendant’s breach. However, the defendant’s liability may also be reduced on account of other potential causes that the defendant’s breach has rendered causally impotent. To that extent, the causation and quantum questions diverge. The quantum of the defendant’s liability may be less than the damage that was caused by the defendant’s breach.

Consider the problem of simultaneous over-determination illustrated by the two-fire problem discussed above. This can be termed ‘simultaneous’ since both potential causes were operative at the same time. Fire X and fire Y arrived at the plaintiff’s house at the same time. On the ‘but for’ test, neither of the fires would be considered a cause since neither was necessary for the plaintiff’s house to burn down. The NESS test provided a solution. Both fires were sufficient to burn down the house, and so both would be considered NESS causes. However, the problem reappears when quantum is considered. Both defendants can argue that, while their negligence may have NESS-caused the plaintiff’s house to burn down, it did not actually result in a loss to the plaintiff, since the plaintiff’s house would have burnt down at that precise moment anyway. The authorities considered below suggest that, in this situation, this quantum argument would be rejected.[104]

Successive over-determination may raise quantum issues in situations where causation is clear-cut even on the ‘but for’ test. The term ‘successive’ indicates that the potential causes were operative at different times. A good illustration is provided by Fleming’s brief account of an American case, Dillon v Twin State Gas & Electric Co:[105]

[A] boy fell from a bridge to almost certain death on the rocks below but was instead electrocuted on the defendant’s negligently suspended wires. Without denying the causality of that negligence, the court discounted the award to the value of his limited life expectancy at that moment.[106]

The boy’s death was caused by the defendant’s negligently suspended wires, since, but for the defendant’s breach, the plaintiff would not have died at that precise time.[107] The second potential cause, the boy’s collision with the rocks below, had no causal effect because the boy had already died. ‘[I]t seems clear that it is impossible to cause an event that has already occurred.’[108] But the pre-empted potential cause still had an impact on quantum. At the time of the boy’s death on the defendant’s electrical wires, his prospects were pretty grim as he was fast approaching the rocks below. Even without the defendant’s negligence, the boy would have died an instant later. This was reflected in the quantum of damages — only slight compensation was required to put the boy in the position he would have been in but for the defendant’s negligence (see Figure 3).

Over-determination can take many diverse forms, depending upon the number, timing, strength and nature of the potential causes, and the law has become correspondingly complex. Some of this complexity is apparent in the House of Lords case, Jobling v Associated Dairies.[109] Lord Keith indicated that, in cases of successive over-determination where both potential causes were tortious, ‘it would clearly be unjust to reduce the damages awarded for the first tort because of the occurrence of the second tort’.[110] Therefore, ‘in proceedings against the first tortfeasor ... the occurrence of the second tort cannot be successfully relied on by the defendant as reducing the damages he must pay’.[111] And to avoid over-compensation ‘[t]he award against the second tortfeasor cannot ... fail to recognise that the plaintiff whom he injured was already to some extent incapacitated’.[112] In Jobling, the second cause, Y, added to the impact of the first cause, X. Cause Y was therefore not causally impotent as it was in Dillon v Twin State Gas, but brought about a further worsening of the plaintiff’s injuries. This was a case with both causation and quantum over-determination issues.[113] In Figure 4,[114] the quantum of damages payable by the first defendant would be A plus B, and the second defendant would pay damages of quantum C.[115] The plaintiff would receive ‘full compensation’.

On the facts of Jobling, however, only cause X was tortious, while cause Y was a naturally occurring supervening illness. Lord Keith held that, in view of this, cause Y had to be treated as ‘one of the ordinary vicissitudes of life’.[116] The court should ‘recognise that the illness would have overtaken the plaintiff in any event so that it cannot be disregarded in arriving at proper compensation, and no more than proper compensation’.[117] In this situation, the non-tortious second cause would reduce the first defendant’s liability. With reference to Figure 4, the plaintiff would receive damages of quantum A only.[118] Lord Wilberforce said ‘[i]t is not easy to accept a solution by which a partially incapacitated man becomes worse off in terms of damages and benefit through a greater degree of incapacity’.[119] He nevertheless thought this the only viable result.

It is clear that the House of Lords in Jobling was not grappling with purely factual and scientific issues. Lord Wilberforce expressly acknowledged this, commenting that

no general, logical, or universally fair rules can be stated which will cover, in a manner consistent with justice, cases of supervening events whether due to tortious, partially tortious, non-culpable or wholly accidental events. The courts can only deal with each case as best they can in a manner so as to provide just and sufficient but not excessive compensation, taking all factors into account.[120]

This appears to be an area where value and policy implications turn on fine distinctions, making legal complexity unavoidable.[121]

It should be noted that the problems discussed in this section arose notwithstanding that the actual and potential causal impacts could be gauged with virtual certainty. In the next section, I return to the main theme of this article, the logical difficulties of proof of causation and quantum presented by the unpredictability of chance.

2 Quantum of Future Loss

Just as the predictive hypothetical aspect of causation, discussed above, invites a relaxation of the traditional ‘all or nothing’ approach to proof and recovery, so too does the predictive hypothetical aspect of quantum. This is the case for both post-trial and pre-trial loss. However, while proof standards have long been relaxed in calculating the quantum of post-trial losses, the proportional approach has only recently been extended in certain jurisdictions to pre-trial losses.

Actually, as Atiyah notes, in calculating the quantum of future loss,

the judge often has to make two wholly different sets of predictions at the date of trial ... First, he has to predict what would have happened to the plaintiff if he had not been injured, a prediction which obviously cannot be verified or falsified by subsequent events. Secondly, the judge may have to predict what is now likely to happen to the plaintiff.[122]

In the broken-leg hypothetical discussed in the introduction, there was uncertainty about the actual future events, but virtual certainty about hypothetical future events. The fact-finder was uncertain whether the plaintiff’s leg would require an operation in the future, giving this event a probability of only 40 per cent. However, although not expressly stated, the fact-finder was quite certain that this operation would not have been required had the defendant not negligently broken the plaintiff’s leg.

In other cases, there is a high level of certainty about actual future events, and the fact-finder’s doubt is instead about hypothetical future events — what would have occurred had the defendant not committed the breach. A recent example is Wynn.[123] The plaintiff had suffered injuries in a motor vehicle accident in 1986 and, at trial in 1992, received substantial damages, a large proportion of which was for future economic loss. At the time of the accident, the plaintiff was vigorously and successfully pursuing a business career. Following the accident, it seemed almost certain that her career path was closed off. On appeal to the New South Wales Court of Appeal, damages for future economic loss were almost cut in half, and the plaintiff appealed to the High Court seeking the restoration of the original award. The plaintiff was partially successful.

In the High Court, the majority’s[124] starting point was to note that, in calculating the future component of damages for loss of earning capacity, some allowance is made for the possibility that without the accident some other misfortune may have visited the plaintiff with a similar impact. The majority observed that ‘the practice in New South Wales is to proceed on the basis that a 15% discount is generally appropriate, subject to adjustment up or down to take account of the plaintiff’s particular circumstances’.[125] They noted, however, that ‘[a]ll “contingencies” are not adverse: all “vicissitudes” are not harmful’.[126] In the present case, there was a ‘real possibility’[127] that the plaintiff would have been promoted, and that her earnings would have increased above their level at the time of the accident. That possibility had to be balanced against contingencies that may have reduced her earning capacity.

One such contingency was that the plaintiff would have needed to take maternity leave or sacrifice her career on other family grounds. The trial judge had imposed a discount of five per cent on this basis, which the Court of Appeal had increased to eight per cent. The High Court preferred the figure of the trial judge. The Court of Appeal had said that there was a prospect that the plaintiff

‘would at some stage [choose] or [be] forced to accept a less demanding job’ because she ‘would be unable or unwilling to remain in her job which placed such heavy demands on her time, energy and health and the love and patience of her husband’.[128]

In response to this, the majority of the High Court observed that ‘there is nothing in the evidence to suggest that the appellant was any less able than any other career-oriented person, whether male or female, to successfully combine a demanding career and family responsibilities’.[129]

There was another matter for which the plaintiff’s damages should be discounted. The plaintiff had suffered spinal injuries in a previous accident in 1972. She received surgery in 1974 and subsequently was able to resume an active life. The trial judge had found it unlikely that the injury would have affected the plaintiff’s future employment. However, as the majority indicated, ‘it was still necessary for his Honour to consider whether it might have done so. This he did not do.’[130] Taking account of this possibility, the majority increased the discount to 12.5 per cent. Following Jobling, discussed in the previous section, the majority indicated that it was not necessary to ‘have regard to the possibility of further tortious injury’.[131]

The final quantum issue was one on which Brennan CJ’s reasoning differed from that of the majority, though he agreed in the result. The High Court held that the Court of Appeal was wrong to impose a further discount on the basis that, as a consequence of the defendant’s breach, the plaintiff would no longer incur the costs of domestic help and child care which she may have otherwise incurred. The majority noted that, according to Sharman v Evans, such deductions should only be made for ‘outgoings necessary for the realization of [earning] capacity’.[132] They appeared to advance a general rule that domestic help and child care do not fit within the Sharman v Evans principle. Brennan CJ agreed with the result on the facts, but interpreted the Sharman v Evans principle differently. Where a plaintiff

can reasonably be expected to dispense with or to provide a service that had previously been provided for the plaintiff at a cost in order to facilitate the earning of the plaintiff’s income, the plaintiff’s capacity to provide the service and to avoid the incurring of the cost must be taken into account in assessing the plaintiff’s loss.[133]

Brennan CJ considered that there may be cases where the costs of both domestic help and child care are covered by the Sharman v Evans principle. However, such an argument would need to be supported by evidence, and in the present case Brennan CJ considered that there was no evidence on point.[134]

The difference between Brennan CJ and the majority may reflect the conflicting objectives of the quantum determination, as identified by Waddams.[135] Brennan CJ’s approach is more in line with the ideal of restitutio in integrum. The majority’s approach, while perhaps departing from the restitutio ideal, is far easier to implement and offers the advantages of being predictable, cheap and efficient. There is another reason why the majority may not have applied the restitutio principle too rigorously. Unlike most quantum issues, this concerned a potential saving for the plaintiff flowing from the defendant’s breach, not a potential loss. To borrow a maxim of equity, the court may consider that the defendant is not entitled to such a discount because of their lack of ‘clean hands’.[136] In the difference between the approaches of Brennan CJ and the majority, there are again signs of values and policies impacting upon the calculation of quantum.

3 Quantum of Past Loss

In calculating the quantum of pre-trial loss, a comparison is drawn between the actual sequence of pre-trial events, and what would have occurred to the plaintiff had the defendant not committed the breach. Only one predictive inquiry is conducted, rather than two in the case of future loss. But there is still the potential for chance to create a fundamental uncertainty. After all, in Wynn, all of the issues relating to quantum arose from an uncertainty, not about actual future events, but about hypothetical future events. The distinction between hypothetical future events and hypothetical past events appears irrelevant since neither is actual. In Wynn, the court dealt with uncertainty about hypothetical future events by providing for proportional recovery. However, there are different views as to whether uncertainty about hypothetical past events should also result in proportional compensation. On this point, the High Court of Australia has taken a different path from the House of Lords, and the Court of Appeal of New South Wales recently followed the House of Lords rather than the High Court.

The High Court authority is Malec.[137] The plaintiff had been employed by the defendant as a labourer in their meatworks. At some point between 1975 and 1977, the plaintiff contracted acute brucellosis as a result of the defendant’s negligence with respect to the work environment. In 1979, the plaintiff was diagnosed with a neurotic illness, which had been brought on by the brucellosis. As a consequence of his illness, the plaintiff ceased work in April 1980. By June 1983, the plaintiff had recovered from the brucellosis, however he remained unemployable. The prognosis for the neurotic illness was extremely grim, and it appeared that the plaintiff would never be able to work again. At trial in September 1987, the plaintiff claimed for loss of earning capacity and pain and suffering, and made a Griffiths v Kerkemeyer[138] claim for the care and attention that he now required from his wife.

The case was one of successive over-determination, broadly similar to Wynn. There was no doubt that the defendant’s breach caused the plaintiff’s losses. However, the Full Court of the Supreme Court of Queensland considered it probable that, even had the defendant not committed the breach, the plaintiff would have developed his neurotic condition and lost his earning capacity from an unrelated cause — a spinal condition developed prior to 1982.[139] The difference from Wynn was that the Supreme Court in Malec considered that, in the absence of the defendant’s breach, the plaintiff would probably have suffered equivalent losses pre-trial. Since they were past losses, the majority of the Supreme Court apparently required that the plaintiff prove them on the balance of probabilities. The plaintiff was unable to satisfy this standard. It was considered more probable than not that the plaintiff would have lost his earning capacity by 1982 without the defendant’s negligence, and damages for economic loss were only awarded for the period up until May 1982. For the same reason, the majority rejected the Griffiths v Kerkemeyer claim. However, ‘somewhat surprisingly’,[140] damages were awarded for pain and suffering with both pre-trial and post-trial components.

The High Court[141] held that the majority of the Supreme Court had assessed all heads of damages incorrectly. The High Court preferred the approach of the dissenting judge, Carter J, who held that the plaintiff was entitled to damages for his continuing condition, caused by the brucellosis, discounted to take account of the contribution made by the unrelated back condition.

In Malec, the High Court made very interesting comments on the nature of time and its implications for proof and quantum.[142] These are worth quoting at length. The majority said:

When liability has been established and a common law court has to assess damages, its approach to events that allegedly would have occurred, but cannot now occur, or that allegedly might occur, is different from its approach to events which allegedly have occurred. A common law court determines on the balance of probabilities whether an event has occurred. If the probability of the event having occurred is greater than it not having occurred, the occurrence of the event is treated as certain; if the probability of it having occurred is less than it not having occurred, it is treated as not having occurred. Hence, in respect of events which have or have not occurred, damages are assessed on an all or nothing approach. But in the case of an event which it is alleged would or would not have occurred, or might or might not yet occur, the approach of the court is different. The future may be predicted and the hypothetical may be conjectured. But questions as to the future or hypothetical effect of physical injury or degeneration are not commonly susceptible of scientific demonstration or proof. If the law is to take account of future or hypothetical events in assessing damages, it can only do so in terms of the degree of probability of those events occurring. The probability may be very high — 99.9 per cent — or very low — 0.1 per cent. But unless the chance is so low as to be regarded as speculative — say less than 1 per cent — or so high as to be practically certain — say over 99 per cent — the court will take that chance into account in assessing the damages. Where proof is necessarily unattainable, it would be unfair to treat as certain a prediction which has a 51 per cent probability of occurring, but to ignore altogether a prediction which has a 49 per cent probability of occurring. Thus, the court assesses the degree of probability that an event would have occurred, or might occur, and adjusts its award of damages to reflect the degree of probability. ... The approach is the same whether it is alleged that the event would have occurred before or might occur after the assessment of damages takes place.[143]

Applying these principles to the facts of Malec, the majority indicated that the plaintiff should be entitled to recover damages for his loss of earning capacity resulting from the ongoing neurotic condition that had been caused by the defendant’s negligence. On the basis of the finding that the plaintiff probably would have lost his earning capacity from his back condition, independently of the defendant’s negligence, damages should be discounted by ‘somewhere between 51 per cent and 99 per cent’.[144] The majority considered that, if the plaintiff had developed a neurotic condition in any event, this would not have been as a direct result of his back condition, but rather as a product of his unemployability. Being the product of two contingencies, rather than just one, the discount for the pain and suffering should be less than the discount for loss of earning capacity.[145] The majority considered that the plaintiff’s need for his wife’s care and attention was a result of his neurotic condition rather than the back condition alone, and accordingly the latter, smaller discount was appropriate for Griffiths v Kerkemeyer damages also.

The minority discussed the nature of time and its implications for proof in similar terms to the majority:

Hypothetical situations of the past are analogous to future possibilities: in one case the court must form an estimate of the likelihood that the hypothetical situation would have occurred, in the other the court must form an estimate of the likelihood that the possibility will occur. Both are to be distinguished from events which are alleged to have actually occurred in the past.[146]

However, the minority appeared a little uneasy with the explicit probabilistic calculations of the majority. They agreed with the ‘general thrust of the reasoning on this point’, but thought it ‘undesirable for damages to be assessed on the footing of an evaluation expressed as a percentage’.[147] One wonders, however, how else uncertainty could be accounted for in calculating quantum.

In Malec, it was held that past losses should be discounted to take account of predictive uncertainty, just as future losses are. On the unusual facts of the case, the consequence of the recognition of the Malec ‘discount’ was actually to increase the damages of the personal injury plaintiff. It is interesting to speculate whether the High Court would have reasoned in the same way had the probability of the plaintiff incurring the same kinds of losses without the defendant’s negligence been less than 50 per cent. In this case, past quantum would have been established on the balance of probabilities, and it would have been not the personal injury plaintiff but the defendant employer arguing for proportional rather than ‘all or nothing’ compensation. The defendant would have been arguing for the reduction of the damages award, rather than the personal injury plaintiff seeking an increase in damages. The logic would be the same in either case, but the policy considerations would clearly differ.[148]

Predictive uncertainty affects both past and future quantum. Past losses are conceptually more certain, as there is only one source of predictive uncertainty rather than two. The event constituting the past injury is a matter of historical fact, and is treated on an ‘all or nothing’ basis. The predictive uncertainty is about whether the loss would have occurred without the defendant’s breach. With future losses, predictive uncertainty affects both the question whether the loss will now occur and whether it would have occurred without the defendant’s breach.

In September 1998 in Chappel, Gaudron J said ‘[i]t is well settled that an award of damages must take account of the probability that some or all of the damage suffered by the plaintiff would have occurred in any event’.[149] However, it appears equally well established in England that no such discount applies to the quantum of past losses, even where they are affected by predictive uncertainty. The judgments in Malec made reference to a passage in Lord Diplock’s judgment in Mallett v McMonagle[150] where his Lordship also discussed the asymmetric nature of time. However, Lord Diplock drew a distinction between ‘what did happen in the past’, and ‘what will happen in the future or would have happened in the future’.[151] He made no reference to past hypothetical events.[152]

More recently, in Hotson v East Berkshire Health Authority[153] Lord Bridge said:

[I]f the plaintiff had proved on a balance of probabilities that the [defendant hospital’s negligent misdiagnosis] had materially contributed to the development of the [plaintiff’s condition], I know of no principle of English law which would have entitled the [defendant] to a discount from the full measure of damage to reflect the chance that, even given prompt treatment, [the condition] might well still have developed.[154]

Lord Ackner made similar comments, saying that in another case a judge’s decision to ‘[discount] an award for a stillbirth, because there was a five per cent risk that the plaintiff would have had a stillborn child even if the hospital had not been negligent, was clearly wrong’.[155]

It seems that Australian law and British law on the quantum of past loss have diverged.[156] The New South Wales Court of Appeal was incorrect in Commonwealth v McLean[157] in holding that Hotson had application rather than Malec, on the basis that that case concerned ‘causation ... of an actual event, and not causation of an hypothetical event’.[158] The plaintiff had developed cancer of the throat and alleged this was a consequence of the consumption of alcohol and tobacco which was brought on by a stress condition caused by the defendant employer’s negligence. It was inconsistent with Malec for the court to hold that ‘the defendant was not entitled to any deduction for the chance that the plaintiff [a smoker prior to the defendant’s alleged negligence] might have contracted throat cancer in any event’.[159] This was a predictive uncertainty, for which Malec clearly stated damages should be discounted. The only ground on which Malec is distinguishable is that in Commonwealth v McLean it would have been the defendant employer who would have benefited from a reduction in damages, rather than a personal injury plaintiff benefiting from an increase in damages. The policy considerations are different in the two cases, but the logic is the same.

C Conclusion on Causation and Quantum Concepts

The discussion in this part has demonstrated the counterfactual nature of both causation and quantum. Both concepts involve a comparison between what actually did happen following the defendant’s breach and what hypothetically would have happened had the defendant not committed the breach. In requiring the courts to conduct a predictive hypothetical inquiry into what would have been the case had the defendant’s breach not occurred, the causation and quantum issues present fundamental proof difficulties. The past is objectively certain, and it appears fair to require the plaintiff to prove their version of past events on the balance of probabilities. However, the future is affected by objective chance. It therefore appears unfair to require the plaintiff to ‘predict’ actual future events or hypothetical past or future events to any particular standard of probability.

In some cases, the quantum inquiry will have a broader temporal focus than that of causation. Causation may be established on the basis that the plaintiff would not have suffered the damage at that precise time but for the defendant’s breach. However, quantum may still be reduced on the basis that the plaintiff would have suffered that damage at some later time as a result of some other cause. Causation can be visualised as a fork in the road, while quantum is the area between two paths (compare Figure 2 with Figures 3 and 4).

The broader temporal focus of the quantum inquiry may explain why courts have been quicker to recognise the fundamental proof difficulty in connection with quantum than they have in connection with causation. A probabilistic approach to post-trial quantum is well established. In Australia, though not in Britain, this has recently been extended to pre-trial quantum where the uncertainty concerns hypothetical events.[160] However, broadly speaking, causation continues to be regarded as an ‘all or nothing’ affair.

In this part, I have considered the impact of chance on causation and quantum in broad conceptual terms. The following parts focus on specific doctrines developed by the courts to deal with plaintiffs’ proof difficulties arising out of the uncertainty of chance. The loss of chance doctrine, discussed in Part III, constitutes a probabilistic conception of causation with consequent proportional compensation. However, it appears to operate only within narrow and unclearly defined boundaries. The ‘injury within scope of risk’ principles, discussed in Part IV, provide the plaintiff with the benefit of a presumption of causation, which may result in their achieving full compensation within the traditional ‘all or nothing’ framework.

III THE LOSS OF CHANCE DOCTRINE

Consider the following scenario. A plaintiff buys one of 1000 lottery tickets, with a prize of $10,000. The defendant puts all the tickets in a hat, and pulls out the winning ticket, which is not the plaintiff’s. But then it is discovered that the plaintiff’s ticket was never put in the hat in the first place. The defendant, in breach of contract, deprived the plaintiff of the chance to win $10,000. The plaintiff has only lost the chance of winning. If the plaintiff’s ticket had been put in the hat, the plaintiff was unlikely to have won anyway. However, the plaintiff paid the defendant for this chance, and should be compensated for its loss. Had the plaintiff won, they would have received $10,000, however they only had a one-in-1000 chance of winning. Their lost opportunity should therefore be valued at $10.[161]

This is a classic situation of loss of chance.[162] The Supreme Court of Canada in Laferrière recently questioned the recognition of lost chances in other than such ‘exceptional classical cases’,[163] refusing to apply the doctrine to a case of medical negligence where it was argued the defendant doctor’s negligence denied the plaintiff patient her chance of recovery. In 1971, the plaintiff had consulted the defendant about lumps in her right breast. The defendant excised the lumps and performed a biopsy but failed to inform the plaintiff of the results, and the plaintiff had no further treatment at that time. In fact, the tests indicated that the lumps were cancerous, and by 1975 the cancer had spread throughout her body. The plaintiff commenced litigation against the doctor in November 1975. After her death on 27 January 1978, her executor continued the action.

The plaintiff faced considerable difficulties in proving causation. Given ‘the insidious and unforeseeable nature of the development of a cancer’, the trial judge was unable to conclude ‘that it was probable in 1971 that [the patient’s] chances of survival would have been greater if she had undergone a treatment other than that given by the defendant’.[164] However, a majority of the Court of Appeal for Quebec allowed recovery on the basis of the loss of chance doctrine. While the plaintiff may not have been able to establish that she would have had a greater than 50 per cent chance of surviving with proper treatment, Jacques JA indicated that she would have had ‘a real and serious chance to benefit from proper medical care’ and that ‘the extent of the damage is to be determined according to the likelihood of success of such proper medical care’.[165]

On appeal, by a majority, the Supreme Court of Canada declined to apply the doctrine of loss of chance to this situation. Echoing the deterministic statements of Wright, discussed above,[166] the Court said ‘[w]hen one speaks of a lottery ticket there is little or nothing other than probability in play’.[167] However, this is not the case with medical conditions such as cancer.

Even though our understanding of medical matters is often limited, I am not prepared to conclude that particular medical conditions should be treated for purposes of causation as the equivalent of diffuse elements of pure chance, analogous to the non-specific factors of fate or fortune which influence the outcome of a lottery.[168]

The majority quoted from the French commentator, Savatier: ‘What is incorrectly called chance is insufficient information on the part of the judge about how destiny took its course’.[169] However, the quotation was not entirely apposite. The uncertainty was rather about how destiny would have taken its course had the defendant not committed his breach, which, as discussed in the previous part, is a predictive rather than a postdictive uncertainty. While the Court did not concede that the defendant had deprived the plaintiff of a chance of recovery, the Court awarded damages for the plaintiff’s increased suffering, which flowed from the defendant’s negligent treatment.[170]

Opponents of the loss of chance doctrine, such as the majority in Laferrière, seem to consider the future to be just as determined as the past. A case in point is the commentator Hill. He questions whether a plaintiff in a Laferrière kind of case could be said to have lost a chance. ‘If 100 patients were all in such a position ... only 15 would survive in the long term “but for” the defendant’s negligence. It is the task of the courts to assess into which group the plaintiff fell. This is a question of past fact.’[171] Hill displays the most extreme form of determinism in suggesting that an individual’s future is ‘a question of past fact’. Would Hill deny individuals any control over their destiny?[172]

The High Court of Australia and the English Court of Appeal have recently recognised that objective chance and predictive uncertainty extend beyond the classic games of chance. However, this recognition has been accompanied by endorsements of traditional ‘all or nothing’ causation, and it remains unclear how widely the doctrine extends and, in particular, whether it covers medical negligence and other personal injury cases. It is too early to say whether the loss of chance doctrine constitutes a broad reconceptualisation of causation in probabilistic terms, as appears to have occurred with quantum in Malec.[173]

A Sellars: Loss of a Commercial Opportunity

The High Court recently upheld an award of damages for the loss of a commercial opportunity in Sellars.[174] The plaintiff company was seeking an injection of working capital. It entered into negotiations with both the defendant company, Poseidon, and a third company, Pagini, in the hope of selling a parcel of the directors’ shareholdings. The plaintiff came close to agreement on a draft contract with the third party (‘draft Pagini contract’), but then received what seemed to be a better offer from the defendant. It ceased negotiations with the third party, and signed a contract with the defendant (‘Poseidon contract’). Shortly afterwards, however, the defendant claimed that their negotiator had exceeded his authority, and refused to comply with all the terms of the Poseidon contract. The plaintiff treated that as a repudiation of the contract, which it accepted. The plaintiff then resumed negotiations with the third party and reached agreement (‘second Pagini contract’), but on less favourable terms than the earlier draft Pagini contract. The plaintiff claimed that the defendant and their negotiator, in entering and then repudiating the Poseidon contract, had made misrepresentations in contravention of s 52 of the Trade Practices Act 1974 (Cth) (‘TPA’) and that, as a consequence, the plaintiff had lost the commercial opportunity of concluding the earlier draft Pagini contract. The plaintiff sought damages under s 82 of the TPA.

Contravention of s 52 of the TPA was established. However, it was far from certain that the draft Pagini contract would have been completed to the plaintiff’s benefit, even without the defendant’s breach. The trial judge considered it more likely than not that the contract would have been entered into, but the completion of the contract was then subject to no fewer than seven conditions precedent.[175] To allow for this uncertainty, the trial judge ‘discounted the [estimated benefits of the draft Pagini contract] to 40 per cent to allow for the probability that the agreement would not have proceeded’.[176] The trial judge acknowledged that it could be questioned whether ‘the loss of that benefit has been proved to the required standard’.[177] His Honour, however, considered that the plaintiff was entitled to compensation for ‘the loss of chance represented by the decision not to proceed with the [draft Pagini contract]’.[178]

The basis of an appeal to the Federal Court and then to the High Court was that damages for loss of chance were not available under s 82 of the TPA. The defendant argued that the plaintiff had failed to establish on the balance of probabilities that it would have obtained actual benefits from the draft Pagini contract.[179] The majority[180] in the High Court reviewed not entirely consistent authorities[181] from contract and tort on proof of future events and past hypothetical events, placing special reliance on Malec,[182] and concluded that the doctrine of loss of chance should be recognised.

The principle recognized in Malec was based on a consideration of the peculiar difficulties associated with the proof and evaluation of future possibilities ... as contrasted with proof of historical facts. Once that is accepted, there is no secure foundation for confining the principle to cases of any particular kind.[183]

The High Court affirmed the award of compensation for the loss of the opportunity to receive the benefits of the draft Pagini contract, quantified at 40 per cent of the full value of the contract. As in Malec, the discount reflected the Court’s recognition of a fundamental predictive uncertainty about hypothetical pre-trial events. The contingencies affecting the completion of the draft Pagini contract were hypothetical events since, following the defendant’s contravention, negotiations on the draft Pagini contract ceased. In relation to the defendant’s contravention, they were future events, however they were pre-trial — by the time of the trial, the draft Pagini contract, if the parties had pursued it, would have been completed or abandoned as the case may be.

Sellars differed from Malec in that the proof difficulty went to causation and not merely to quantum. In Malec, the Court accepted that, but for the defendant’s negligence, the plaintiff would not have developed the neurotic condition in 1979. The ‘but for’ test was satisfied on the balance of probabilities. The difficulty was in proving quantum, since it seemed likely that the plaintiff would have developed a similar condition by 1982, without the defendant’s negligence. In Sellars, however, the ‘but for’ test was not satisfied on the balance of probabilities. Had the defendant not contravened the TPA, there would have been only a 40 per cent probability that the plaintiff would have enjoyed the benefits of the draft Pagini contract. In Sellars, by the doctrine of loss of chance, a probabilistic conception of causation displaced the traditional ‘all or nothing’ conception.

It is worth noting also that the chance recognised by the High Court in Sellars was far more complex than the narrow notion of classical chance discussed by the Supreme Court of Canada in Laferrière. According to Gonthier J, an identifying characteristic of a genuine chance is that there is ‘no factual context in which to evaluate the likely result ... Effectively, the pool of factual evidence regarding the various eventualities in the particular case is dry’.[184] The chance recognised by the High Court in Sellars was grounded in a complex commercial transaction with no fewer than seven contingencies. The chances were not comparable to a simple roll of the dice.[185]

The recognition of the existence of the objective uncertainty of the future provided a solid foundation for the Court’s recognition of the loss of chance doctrine. Unfortunately, the majority judgment in Sellars also invoked a far weaker argument for the loss of chance doctrine. ‘The approach results in fair compensation whereas the all or nothing outcome produced by the civil standard of proof would result in the vast majority of cases in over-compensation or under-compensation’.[186] The possibility of under-compensation and over-compensation is an inevitable consequence of uncertainty, and cannot be avoided by making proportional awards. In fact, while ‘all or nothing’ awards are right at least some of the time, proportional awards are ‘bound to be wrong’,[187] as Atiyah points out with regard to proportional awards for possible future medical complications for personal injury plaintiffs. ‘If the risk eventuates, the amount awarded will be too little; if it does not, too much.’[188] Only two years earlier, a majority of the High Court in Wardley Australia Ltd v Western Australia[189] recognised that proportional awards also carry this risk.

Once it is recognised that the risk of under-compensation or over-compensation cannot be averted, the question becomes which party should bear the risk. As a matter of policy, the answer to this question will depend upon the nature of the parties. The logic of loss of chance distributes the risk equally, which may be appropriate for commercial litigation. However, as discussed in Part IV, in personal injury litigation, there may be good policy grounds for making the deep-pocketed or insured defendant bear the greater share of the risk, particularly where they have been shown to be in breach.

B The Reality of Chance

In Sellars, while the Court relaxed the traditional proof requirements for causation, there were also a number of statements endorsing the traditional approach. For example, the majority said that the plaintiff ‘can only recover compensation for actual loss or damage incurred, as distinct from potential or likely damage’,[190] and that ‘the general standard of proof in civil actions will ordinarily govern the issue of causation and the issue whether the applicant has sustained loss or damage’.[191] Brennan J asked, ‘what is the standard of proof in cases where the issue of causation depends on competing hypotheses?’ His answer was ‘the balance of probabilities’.[192]

Unless it can be predicated of an hypothesis in favour of causation of a loss that it is more probable than competing hypotheses denying causation, it cannot be said that the plaintiff has satisfied the court that the conduct of the defendant caused the loss.[193]

The loss of chance doctrine was reconciled with the traditional proof requirements by focusing on the chance of the benefit, rather than the actual benefit. The majority said: ‘The prejudice or disadvantage which the respondents suffered in the present case was the loss of the opportunity or chance of securing commercial benefits.’[194] The plaintiff was required to establish causation of the loss of the chance of the benefit on the balance of probabilities.[195] The value of the chance is then ‘ascertained by reference to the degree of probabilities or possibilities’.[196]

1 First- and Second-Order Probabilities

The loss of chance action requires the court to make two probability assessments. To successfully establish an action for loss of chance, the plaintiff has to prove on the balance of probabilities that they had the chance of some benefit.[197] The court’s uncertainty has two dimensions, both of which can be measured on the probability scale. The chance itself can be labelled a first-order probability, and the fact-finder’s partial belief in the existence of the chance can be labelled a second-order probability.[198]

The defendant may argue that it is not essential to make two assessments, and that these should be collapsed into a single figure using the calculus of probability. Suppose, for example, that the court in Sellars considered there to be a second-order probability of 70 per cent that the plaintiff had a first-order chance of 40 per cent of completing the draft Pagini contract. By the mathematical conjunction rule, these two probability figures can be collapsed into a single figure of 28 per cent.[199] If the benefits of the draft Pagini contract were valued at $1,000,000 perhaps the plaintiff should be awarded $280,000 rather than $400,000.

The plaintiff may present a competing argument. It could be said that there are sets of first- and second-order probabilities that are equivalent, in that they would collapse to the same single-order figure. The plaintiff would select first- and second-order probabilities so as to maximise their claim for damages. To continue the same example, the plaintiff would argue that there was a 51 per cent second-order probability that they had had a 55 per cent first-order chance of receiving the benefits of the contract. These collapse into the same single-order figure, 28 per cent. The second-order probability has been minimised, while still satisfying the balance of probabilities. The first-order probability and the plaintiff’s damages have been maximised. On this basis, the plaintiff would recover $550,000.

However, these strategies of collapsing and reconstituting first- and second-order probabilities are inconsistent with the recognition of the objective reality of first-order chance. The reality of chance is well illustrated by the House of Lords decision in Hotson.[200] The plaintiff in Hotson was a boy who, at age 13, fell while climbing a tree, severely fracturing his hip. He attended the defendant’s hospital, but the injury was not correctly diagnosed. It was not until five days later that the injury received correct diagnosis and treatment. However, by that time the plaintiff had suffered avascular necrosis in his hip. The blood vessels in his leg were so damaged that no blood was being carried to his injured hip in order to repair the damage, and the plaintiff’s disability had become permanent. There was a possibility that this condition could have been avoided had the plaintiff received correct initial diagnosis and treatment. The defendant admitted breach of duty, and conceded liability for the plaintiff’s five days of avoidable pain and suffering following the initial misdiagnosis. However, it argued that causation of the permanent disability had not been established on the balance of probabilities. Even if the plaintiff had received correct treatment initially, the permanent disability would probably have followed in any case. The plaintiff responded by raising the loss of chance doctrine. The plaintiff sought compensation for being deprived of the chance of avoiding the permanent disability.

At trial, the plaintiff was successful. The plaintiff was compensated for the loss of a 25 per cent chance of complete recovery, and received a quarter of the damages that would have constituted full compensation for the permanent disability. An appeal to the Court of Appeal was dismissed, the Court accepting the loss of chance doctrine in similar terms to the High Court in Sellars.[201] For example, Donaldson MR endorsed the traditional proof requirements: ‘[T]he plaintiff has to prove some loss or damage and must do so on the balance of probabilities.’[202] However, he went on to agree with the trial judge that the plaintiff’s injury was not the avascular necrosis itself, which probably would have ensued in any case. The plaintiff’s injury was being denied ‘the chance of avoiding avascular necrosis and its consequences’.[203]

On appeal to the House of Lords, the defendant was successful. The House of Lords left open the question of whether the loss of chance doctrine in application to medical negligence was a part of English law.[204] It was held that the plaintiff had failed to prove on the balance of probabilities that there was a chance of recovery, and therefore the doctrine would not apply anyway.[205] According to the House of Lords, the trial judge had not found on the balance of probabilities that the plaintiff had had a 25 per cent chance of recovery. The trial judge had found that there was a probability of 25 per cent that the plaintiff had had a chance of recovery. Expressed another way, the House of Lords held that the trial judge and the Court of Appeal had mistakenly treated a second-order probability — a subjective level of belief — as though it were a first-order probability — an objective chance.

In order to have had a chance of recovery following the accident, it was necessary that a sufficient number of blood vessels were left intact to maintain the blood supply to the injured hip. Incorrect treatment would then close off this chance since ‘the bleeding of ruptured blood vessels ... [would] compress and thus block the intact but distorted remaining vessels’.[206] The trial judge had held that the defendant’s ‘delay in diagnosis denied the plaintiff the 25 per cent chance that, given immediate treatment, ... the plaintiff would have made a very nearly full recovery’.[207] This passage suggests that the trial judge viewed the 25 per cent figure as a first-order chance. However, the House of Lords focused on another finding, inconsistent with the first, that it was ‘possible but improbable’ that ‘the fall [had] left intact sufficient vessels’[208] for a full recovery to be achievable. This suggests that the 25 per cent figure was not a first-order chance, but a second-order level of belief in the existence of the chance. The question of the existence of the chance was identical to the question of whether sufficient blood vessels were left intact by the fall. This was merely ‘some relevant past event,’[209] ‘a matter of past fact’,[210] and the plaintiff had failed to establish this on the balance of probabilities.

The House of Lords clearly recognised the objective reality of the plaintiff’s first-order chance of recovery, and distinguished this dimension of uncertainty from the second-order partial belief in the existence of the chance. The court clearly considered that the two orders of probability are not interchangeable. The trial judge and Court of Appeal judgments had made an error in confusing a chance with a level of belief.[211]

The reasoning of the House of Lords in Hotson has recently been misunderstood in two New South Wales Supreme Court decisions.[212] In Daniels v Anderson,[213] the Court of Appeal said that the House of Lords in Hotson denied recovery for the loss of chance. Their Honours suggested that the House of Lords had ‘proceeded upon the basis that a 75 per cent chance that the plaintiff would suffer from the disabling disease without the intervention of any negligence should be treated as a certainty’.[214] Like the trial judge and the Court of Appeal in Hotson, the New South Wales Court of Appeal failed to distinguish between a first-order chance and a second-order partial belief. In Tran v Lam,[215] Badgery-Parker J appears to have made the same mistake. In this case, his Honour awarded a personal injury plaintiff damages for the chance of attaining a better prognosis, lost as a result of the defendant doctor’s negligent medical treatment. However, as in Hotson, his Honour’s findings indicate that, as a matter of historical fact, on the balance of probabilities this chance did not exist.

The plaintiff in Tran v Lam consulted the defendant general practitioner on 16 December 1994 about a lump in her breast. The defendant was negligent in her diagnosis and treatment. The lump was in fact a malignant cancer which, by the time the plaintiff received proper treatment in April 1995, had metastasised and become incurable. Had the defendant not been negligent, the plaintiff’s breast cancer would have been diagnosed and treated by the end of January 1995. The question, therefore, was whether treatment at that time would have given the plaintiff the chance of a better prognosis. The answer to this turned upon the time at which metastasis occurred. ‘Her situation seems to be that if metastasis to the brain had occurred before the end of January 1995 the plaintiff’s chance of survival was no different than that which is now known’.[216] His Honour’s findings imply that on the balance of probabilities metastasis had already occurred by the end of January. Badgery-Parker J held: ‘It appears to me that the chance that metastasis had occurred by 17 January 1995 is rather greater than the chance that it had not occurred.’[217] His Honour’s use of the term ‘chance’ is inappropriate. The plaintiff’s condition at that point in time is a matter of historical fact, and there is no chance about the past. His Honour was referring to a second-order degree of belief, rather than a first-order chance. The finding that metastasis had probably occurred by the end of January should have led his Honour to find, on the balance of probabilities, that the defendant’s negligent treatment had not deprived the plaintiff of the chance of a better prognosis. However, Badgery-Parker J conflated the first- and second-order probabilities and held:

I am satisfied on the balance of probabilities that there was a significant chance that [had] the cancer ... been diagnosed as it should have been on or about 17 January 1995 and treated appropriately and properly thereafter, that it would have been cured or at the very least, that the period before its recurrence would have been extended and the plaintiff’s lifespan accordingly prolonged.[218]

2 Speculative Chances

The loss of chance doctrine was recently applied in commercial litigation by the English Court of Appeal in Allied Maples.[219] The judgments offer further insights into the reality of chance. The plaintiff company had sought the advice of the defendant firm of solicitors on the purchase of certain businesses belonging to the vendor. The defendant advised that, as these businesses held leases and approvals that were personal to them, the plaintiff would need to buy the vendor’s shares in the company running the businesses in order to acquire the businesses. The plaintiff followed this advice, but then discovered that it had also acquired certain liabilities of the company. The plaintiff argued that the defendants had been negligent in failing to advise it of the risk of such liabilities, depriving it of the chance to protect itself with a warranty from the vendor. The defendant conceded its advice had been negligent. However, it argued that the plaintiff had failed to establish causation. Even had the plaintiff been advised of the risk of liability, it was not clear that the vendor would have granted the warranty, and the plaintiff may have incurred the liabilities in any case. The ‘but for’ test was not satisfied on the balance of probabilities.

The majority of the Court of Appeal pointed out that the benefit that the plaintiff had been denied by the defendant’s breach was a hypothetical benefit, not an actual benefit. Therefore, the benefit did not have to be proved on the balance of probabilities. The plaintiff merely had to show that it had a ‘real or substantial chance [of obtaining the benefit] as opposed to a speculative one’.[220] Once the existence of the chance is established, then ‘the judge is free on the quantum [issue] to assess the chance of successful negotiation as greater or less than 50 per cent’.[221]

Millett LJ dissented. He agreed that the case required the application of the loss of chance doctrine. ‘Since the question depended on what an independent third party would have done in a hypothetical situation, the plaintiffs did not need to establish it on a balance of probabilities.’[222] However, he considered that

the evidence was not ... sufficient to justify the inference that there was any real or substantial chance that [the vendor] would have acceded to the plaintiffs’ request. Whether they would or would not have done is, on the evidence so far adduced, a matter of pure speculation.[223]

By his reference to ‘speculation’, Millett LJ meant not that the chance was too small,[224] but that it was too difficult to estimate. In part, the problem was that the vendor had not given evidence as to whether they would have provided a warranty had it been requested. But, even had the vendor given evidence, it would have been difficult to satisfy Millett LJ’s criteria for a ‘real chance’. He indicated it would not have helped the plaintiffs if the vendor had said:

‘We might have [provided a warranty]; we might not. It is really not possible to say.’ That does not lead to the conclusion that there was a 50 per cent chance or any substantial chance: it is still all speculation. Nor will it avail the plaintiffs to obtain apparent admissions from [the defendants] that: ‘Of course we might have yielded; so there was a chance,’ and then to drive the witness into hazarding a percentage by asking: ‘A 50 per cent chance? 40 per cent? 30 per cent?’ until he reluctantly accepts a figure. That would not be inviting the witness to give evidence, but to speculate.[225]

The majority judges considered this point, but did not agree that the chance was

arbitrary and wholly unpredictable. Those with experience of commercial negotiation are able, with a reasonable degree of accuracy, to form a view of what can be achieved by such negotiation. ... It is possible to make an informed judgment of what the chances were of achieving certain results.[226]

The approach of the majority is to be preferred. Millett LJ’s demand for certainty in the precise quantification of the chance is misguided.[227] Even in cases where the chance is capable of precise quantification, as in the classic lottery case, the chance is still perfectly consistent both with the chance being realised and with the chance not being realised. Chance is, after all, a measure of uncertainty. Millett LJ appeared to agree with the majority that, had the defendant not been negligent, there was some chance of the plaintiff obtaining the contract with the warranty. If the court denied the plaintiff relief on account of the difficulty in estimating the plaintiff’s chance, the court would be failing in its duty. ‘Many cases illustrate that uncertainty in the quantification of damage, either in cases of contract or tort, does not prevent an assessment; provided that some broad estimate can be made.’[228]

Commonwealth v Amann Aviation Pty Ltd[229] raised the issue of what the plaintiff had lost when the defendant wrongfully repudiated a contract with the plaintiff. The chances of the plaintiff obtaining profits from the contract were unclear. Deane J provided an analysis of the loss of chance, in which he recognised the existence of

extreme cases in which curial procedures are simply inadequate to determine whether there was any real or significant chance that an alleged benefit would actually have been obtained ... The profit one experienced commercial person may see as lying at the end of some commercial undertaking might be seen as an inevitable and disastrous loss by another. ... The nature of what would have been obtained ... may be so completely speculative that ‘it is quite impossible to place any value’ upon it.[230]

However, in such a case, the plaintiff’s difficulty in quantifying the chance of obtaining a benefit should not lead a court to award only nominal damages. Deane J indicated that the plaintiff’s difficulty can be overcome by a ‘presumption that the value of those benefits would have been at least equal to the total detriment [sustained] by the plaintiff in [procuring and performing] the contract’.[231] In this kind of case, where the court finds the future absolutely unpredictable, it should look back at the past and award reliance damages in substitution for expectation damages.

C Restrictions on the Loss of Chance Doctrine?

In Commonwealth v Amann Aviation Pty Ltd, Deane J noted

the assessment of damages ... ‘is a pragmatic subject ... [which] does not lend itself to hard-and-fast rules’. ... [I]t is neither desirable nor practicable to seek to formulate an exhaustive comprehensive rule defining the circumstances in which it is appropriate for a court to assess damages on the basis that what has been lost ... is the probability or possibility of a benefit ... as distinct from the benefit ... itself.[232]

The High Court recently applied the doctrine to a situation of lost commercial opportunity in Sellars, however it appears too early to announce the demise of ‘all or nothing’ causation. Even while the loss of chance doctrine was being embraced, the traditional proof requirements and ‘all or nothing’ recovery were being endorsed.[233] The boundaries of the loss of chance doctrine remain unclear. In the sections following, I comment on a number of proposed delimitations and exceptions to the loss of chance doctrine, some of which appear stronger in authority and reasoning than others.

1 Omissions Rather than Commissions

A number of judges and commentators have conjectured that the difficulty of proof of causation and quantum associated with the ‘hypothetical “what if” inquiry ... is greater in cases of omission than of commission, because in the latter we know at least what happened, whereas in the former we can only speculate what would have happened’.[234] This suggests that the loss of chance doctrine and Malec principle may have a more natural application to cases of omission than those of commission.

The reasoning behind the distinction is not transparent. The ‘hypothetical “what if”’[235] inquiry that leads to the difficulties of proof is made in every case, not only cases of omission. The rationale for the distinction appears to be that, in a case of commission, the court will often be able to assume that, but for the defendant’s breach, the ante quo position would have continued. ‘[I]n many cases it is clear that the loss would not have befallen the plaintiff if the defendant had not been guilty of contravening or tortious conduct.’[236] On the basis of this assumption, the court compares what actually occurred after the commission, with what actually was the case before the commission. There is the illusion that causation simply requires a before–after comparison (see Figure 6(a)). ‘In those cases, there is no need to advert to hypotheses when determining the issue of causation of loss or damage: the chain of causation will appear, if at all, from the historical facts.’[237] If, for example, the plaintiff is claiming damages for a broken leg sustained in a motor vehicle accident, the causation comparison appears to be between the actual broken leg following the negligent act and the actual unbroken leg prior to the negligent act. However, this perception of undiluted actuality is an illusion. A hypothesis is in fact being adopted. ‘[C]ausation of the personal injury is proved only by the adoption of an hypothesis that the injury would not have befallen the plaintiff but for the negligence of the defendant.’[238]

In cases of omission, however, it is often the case that the ante quo position continues or develops according to the natural course of things. The defendant should have intervened to give the plaintiff a chance of improving their position — for example, a medical recovery or winning a lottery — but failed to do so. In this situation, the ante quo state of affairs offers little guidance on what exactly would have happened had the intervention occurred (see Figure 6(b)).

The distinction between cases of omission and cases of commission offers insights into the nature of the causation comparison, but it is clear that commissions can present predictive proof difficulties.[239] Sellars itself was a case of a contravening commission — the making of a misrepresentation. However, the High Court was nevertheless struck by the difficulty of predicting what would have occurred had the misrepresentation not been made, and held that the loss of chance doctrine was applicable.

2 The Plaintiff’s Own Conduct: Choice, Not Chance

There is a body of authority in support of the proposition that, where the hypothetical inquiry concerns the plaintiff’s own conduct, the traditional proof requirement should apply. What the plaintiff would have done is considered to have been ‘his choice, not the choice of fate’.[240] ‘Although the question is a hypothetical one, it is well established that the plaintiff must prove on balance of probability that he would have taken action to obtain the benefit or avoid the risk.’[241]

In applying the loss of chance doctrine in Sellars, the High Court sought to distinguish two decisions that would otherwise have appeared inconsistent with the loss of chance doctrine.[242] In these cases, the defendant’s breach consisted of giving the plaintiff the false impression that they had insurance cover for a certain risk. When that risk ensued, the plaintiff found that they had no cover and sought compensation from the defendant. The plaintiff failed on the issue of causation. The plaintiff was unable to establish on the balance of probabilities that, had the defendant not given the false impression, the plaintiff would have been able to obtain the insurance that covered that risk. In Sellars, the majority indicated that these cases may constitute an exception to the loss of chance doctrine. Whether the plaintiff would have obtained this cover was dependent in part on the plaintiff’s own actions. ‘When the issue of causation turns on what the plaintiff would have done, there is no particular reason for departing from proof on the balance of probabilities notwithstanding that the question is hypothetical.’[243] Of course, obtaining the cover would also depend on other factors beyond the plaintiff’s control, and these matters would, perhaps, not require proof on the balance of probabilities. To the extent that these cases could not be reconciled with recovery for the loss of commercial opportunity, the majority indicated disapproval of them.[244]

The plaintiff’s hypothetical conduct will commonly be at issue in workplace injury cases where the defendant’s breach is a failure to provide certain safety equipment.[245] The plaintiff is required to establish that, had the safety equipment been provided, they would have used it and the injury would have been averted. In Duyvelshaff v Cathcart & Ritchie Ltd,[246] the plaintiff was a plumber employed by the defendant. The plaintiff had fallen from a plank on which he was sitting while joining a PVC pipe to a fixed copper pipe, and injured himself. The plaintiff argued that the defendant should have provided a safety belt to use with a ladder provided by the defendant. Walsh J said that the plaintiff ‘must establish ... that his injuries were caused by the [defendant’s] omission to furnish a safety belt. This means that he must show that it is probable that he would have used the belt if it had been furnished.’[247] Gibbs J said: ‘In all the circumstances it is impossible to conclude on the balance of probabilities that if the employer had provided a safety belt the appellant would have used the ladder and the belt instead of doing what he in fact did.’[248] The court applied the traditional standard of proof to the hypothetical events that would have followed had the defendant not been in breach of duty. However, this can be understood as an exception to the loss of chance doctrine. The hypothetical facts turned on the plaintiff’s own conduct, which is a matter of choice, not chance.

Of course, even accepting the notion of human autonomy,[249] and that an individual has genuine control over their actions, it does not necessarily follow that an individual would be able to predict their own actions in a hypothetical scenario.[250] And then it is to take a further step from the assumption of human autonomy to demand that the plaintiff prove their hypothetical actions to a court on the traditional standard of proof.[251]

3 The Opportunity to Pursue Litigation

The weight of authority is that the loss of chance doctrine has application to cases where the negligence of a defendant solicitor has caused the plaintiff’s action against a third party to become statute-barred. For example, in Commonwealth v Amann Aviation Pty Ltd Deane J said that in this situation the plaintiff ‘may recover damages by reference to the court’s assessment of what the chance of success ... would have been even though that assessment is 50 per cent or less’.[252] Similar statements were made by the High Court in Nikolaou v Papasavas, Phillips & Co[253] and Johnson v Perez.[254] The High Court has not actually had occasion to discount the plaintiff’s damages for loss of the opportunity to litigate, as the plaintiffs in Nikolaou and Johnson v Perez were considered certain of success.[255] However, State Supreme Courts have. For example, in Murphy v Miller[256] the New South Wales Court of Appeal approved a trial judge’s award of damages incorporating a discount of 50 per cent, not only for the chance of success in the original action, but also for the chance of judgment being satisfied, given that the potential defendants were outside of the jurisdiction. In Golec v Scott,[257] a majority of the New South Wales Court of Appeal increased the plaintiff’s damages on the basis that they gauged the plaintiff’s chance of success against the original wrongdoer 10 percentage points higher than did the trial judge.[258] Priestley JA dissented, since he did ‘not feel competent to draw so fine a distinction as between 75 per cent and 85 per cent in a matter so difficult to quantify precisely as the plaintiff’s chances of success in his case’.[259] In Williams v Bodewes,[260] there was no discount since the trial judge had found that the ‘likelihood of [the plaintiff] succeeding ... was “a virtual certainty”’.[261]

However, there is not a total consensus on this application of the loss of chance doctrine. Brennan J, dissenting in Johnson v Perez, suggested that in this kind of case the task of the court is to ‘determine as best it can on the balance of probabilities whether the plaintiff would have succeeded (and if so to what extent) or failed’.[262] Even if litigation is recognised as something of a lottery,[263] it is arguable that the plaintiff’s chance of winning has been preserved. The plaintiff has lost the chance to bring their action against the third party, however the plaintiff can be given the opportunity to prove the same case in their action against the defendant solicitor. The court can ‘conduct a trial within a trial to determine what the cause of action would have produced. That is what the cause of action was worth to the plaintiff.’[264]

However, the dominant view, rather than merging the two actions, emphasises the differences between them. And so, for example, the plaintiff in Nikolaou was not able to recover fully for the nervous condition that was caused by the breach of the original wrongdoer. He could only recover for the ‘possibility’ that this condition would develop, since it was only a possibility at the time the original claim would have been brought.[265] In the action against the solicitor, the court was not able to take the benefit of hindsight and the knowledge that ‘[w]hat was a possibility originally has matured into a positive fact’.[266]

4 Contract, Tort and the TPA; Economic Loss and Personal Injury

Some courts and commentators have suggested that the loss of chance action should be limited to certain causes of action or to certain kinds of damage. In Sellars, the High Court of Australia appeared to reject such distinctions. As discussed above, this was a case in which the plaintiff sought damages for the loss of a commercial opportunity under s 82 of the TPA. Brennan J noted that the clearest loss of chance case would be ‘for breach of a contractual promise to afford the plaintiff an opportunity to acquire a benefit’.[267] The plaintiff would be entitled to nominal damages for the defendant’s merest breach of the contract. Actions in tort and actions for damages under s 82 of the TPA are ‘in a different category [since] damage is the gist of the cause of action’.[268] However, he ultimately concluded that this should not be a basis for denying damages for loss of chance in tort and s 82 TPA cases. ‘There is no rational basis for distinguishing between a loss for which more than nominal damages may be awarded in contract and a loss for the purposes of s 82(1) of the Act and the law of torts.’[269] The majority made similar comments:

[D]amages for deprivation of a commercial opportunity, whether the deprivation occurred by reason of breach of contract, tort or contravention of s 52(1), should be ascertained by reference to the court’s assessment of the prospects of success of that opportunity had it been pursued.[270]

As Croom-Johnson LJ noted in Hotson, ‘[t]here would be no sense in a distinction which allowed damages to be recovered from a private doctor but not against a National Health Service doctor’.[271]

A similar distinction can be drawn between economic loss and personal injury. Writing before the recent landmark decisions of Malec[272] and Sellars,[273] Coote suggested that the doctrine of loss of chance was limited to cases of ‘loss of a chance of financial gain’ because ‘the chance itself has an economic value’, while ‘the chance of physical injury or the loss of a chance of physical recovery’ is not a ‘form of injury in itself’.[274] Financial gains are measured in dollars and cents, and are clearly divisible. Personal injuries, for example a broken leg or lung cancer, obviously have financial implications, but the injuries themselves are indivisible. Proportional compensation appears more applicable to commercial litigation than to personal injury litigation.

Some support for Coote’s view may have been derived from Lord Bridge’s judgment in Hotson, in which he indicated that he saw ‘formidable difficulties in the way of accepting the analogy’ between economic loss of chance cases and ‘the principle of awarding for the lost chance of avoiding personal injury or, in medical negligence cases, for the lost chance of a better medical result’.[275] It is unclear whether the distinction has support in the High Court. While Sellars involved economic loss, it was preceded by Malec, a case of personal injury.[276] The High Court in Sellars invoked Malec in support of the recognition of the loss of a commercial opportunity. ‘Neither in logic nor in the nature of things is there any reason for confining the approach taken in Malec concerning the proof of future possibilities and past hypothetical situations to the assessment of damages for personal injuries.’[277]

More recently, however, in Chappel[278] and Naxakis[279] the High Court has shown less enthusiasm for the loss of chance doctrine in medical negligence cases. In Chappel, discussed above,[280] the plaintiff sought compensation for injury to her voice following throat surgery. The doctor’s negligence was not in the performance of the operation, but in failing to warn the plaintiff of the risk of this injury, despite a pointed inquiry. The plaintiff argued that, had she been warned, she would have employed a more experienced surgeon than the defendant, with perhaps a lower risk of the operation going wrong. It was late in proceedings that the loss of chance perspective was introduced, by the defendant rather than the plaintiff,[281] and the court did not give it detailed consideration. By a majority, the court held that the plaintiff was entitled to full compensation for the actual injury to her voice, and not merely a proportional award for a lost or reduced chance of avoiding the injury.[282] However, a number of the judgments contained comments on the loss of chance doctrine.

In Chappel, Kirby J was most positive about the loss of chance action.[283] Kirby J referred to his earlier judgment as Acting Chief Justice of the New South Wales Court of Appeal in CES v Superclinics (Australia) Pty Ltd,[284] where he ‘indicated [his] attraction to this approach as a more rational and just way of calculating damages caused by established medical negligence’.[285] The other judges were far less enthusiastic in their comments. Gaudron J said ‘clearly, the damage sustained by Mrs Hart was not the loss of a chance — valuable or otherwise — but the physical injury which she, in fact, sustained’.[286] Gummow J referred to Gonthier J’s severe restriction of the loss of chance doctrine in Laferrière with apparent approval, quoting: ‘As far as possible, the court must consider the question of responsibility with the particular facts of the case in mind, as they relate concretely to the fault, causation and actual damage alleged in the case.’[287] Hayne J adverted to practical difficulties with the application of the doctrine in quantifying and valuing the chance, concluding that ‘the loss of chance analysis is flawed and should not be adopted’.[288] Note, however, that a number of the judges nevertheless indicated that damages for past loss should be subject to a Malec discount.[289]

While in Sellars the judgments promoted a common conception of causation for different causes of action, the High Court has since preferred a context-sensitive conception. In Chappel, Gaudron J said that ‘[q]uestions of causation are not answered in a legal vacuum. Rather they are answered in the legal framework in which they arise. For present purposes, that framework is the law of negligence.’[290] In separate judgments, Gummow, Kirby and Hayne JJ referred to a recent House of Lords decision, Empress Car Co (Abertillery) Ltd v National Rivers Authority, in which Lord Hoffman said: ‘[O]ne cannot give a common sense answer to a question of causation for the purpose of attributing responsibility under some rule without knowing the purpose and scope of that rule’.[291]

Similar contextual considerations are apparent in the Naxakis decision. Like Chappel, this was a case in which loss of chance was raised late in the proceedings, was not fully argued[292] and was discussed only briefly in some judgments. The plaintiff was admitted to the defendant hospital following a collapse, and was treated by the defendant neurosurgeon. The plaintiff alleged that the defendants were negligent in failing to diagnose his aneurism. The plaintiff was discharged from hospital, the aneurism burst, and the plaintiff suffered permanent physical and intellectual impairment. The trial judge withdrew the case from the jury on the basis that there was insufficient evidence that the defendants were negligent in failing to test for the possibility of an aneurism. The Court of Appeal of the Supreme Court of Victoria dismissed the plaintiff’s appeal. The High Court allowed a further appeal, and ordered that a new trial be held. The major issues in the case were the test for the sufficiency of the evidence and the principles governing medical negligence.

Loss of chance was raised by the plaintiff in the High Court because the trial judge and the Court of Appeal apparently doubted it would have been possible to operate to avert the burst aneurism, even had it been properly diagnosed.[293] On this basis, it would have been difficult for the plaintiff to establish causation on the balance of probabilities. The plaintiff argued as an alternative that the misdiagnosis had deprived him of the chance of having the operation and avoiding the harm. It may be that the plaintiff would nevertheless face difficulties in establishing the existence of the chance as a matter of historical fact, as in Hotson and Tran v Lam discussed in Part III(B)(1) above.

Gaudron J noted that the High Court had recently applied the loss of chance doctrine in a commercial context in Sellars. Her Honour suggested that ‘philosophical or logical analysis’ may indicate that the doctrine is ‘strictly correct’ for cases ‘involving the failure to diagnose or treat a pre-existing condition’.[294] However, ‘different considerations apply where, as here, the risk eventuates and physical injury ensues’.[295] Gaudron J suggested that ‘all or nothing’ causation then applies,[296] referring with approval to Laferrière.[297] The plaintiff should establish that they are one of the statistical percentage ‘who would have recovered if properly treated, [in which case] his or her loss is not merely the loss of a chance, but the injury or disability that eventuated’.[298]

Gaudron J observed that the loss of chance approach is ‘not one that necessarily works to the benefit of the individual plaintiff’[299] in that the plaintiff only obtains damages for a fraction of their injuries. Furthermore, Gaudron J suggested that ‘[t]here is ... a tendency to exaggerate the difficulties associated with proof of causation’.[300] Her Honour adverted to the ‘injury within scope of risk’ principles that I discuss in the next part.[301] Where the defendant’s breach increases the risk of the plaintiff’s injury, and the plaintiff suffers that injury, causation is presumed.[302] Unless the defendant can rebut the presumption, the plaintiff will be entitled to full compensation. Gaudron J indicated that, while the cause of a plaintiff’s injuries may be uncertain according to science and philosophy, this ‘does not mean that their cause is uncertain in law [on a] commonsense approach’.[303]

Likewise, Kirby J indicated that causation could be presumed from breach and damage, and that there was therefore no need to consider loss of chance.[304] Callinan J suggested that the presumption may be invoked,[305] however his Honour considered the loss of chance doctrine in case, on retrial, causation was still not proven on the balance of probabilities.[306] Callinan J considered that the doctrine should be open in medical negligence cases. The plaintiff should not ‘suffer’ on account of the difficulty in ‘saying what the outcome would have been had the [appropriate] procedure or treatment ... been performed’, since that difficulty arises from the defendant’s non-performance.[307] However, his Honour indicated that the loss of chance doctrine is ‘not without its difficulties’.[308] Like Gaudron J, his Honour appeared concerned about the plaintiff not receiving full compensation for their personal injuries. He suggested that if the plaintiff shows that ‘the chance that has been lost is a 51% or greater chance ... [the plaintiff] should recover his or her damages in full’.[309]

D Conclusion on Loss of Chance

Traditionally, each element of the plaintiff’s case needs to be proven on the balance of probabilities, and compensation is ‘all or nothing’. A recognised exception is future losses. The future is fundamentally uncertain, and so the plaintiff is allowed compensation for even improbable future losses, proportional to the probability of the loss. In Malec,[310] the High Court recognised that the same kind of predictive uncertainty affects pre-trial losses, and allowed proportional recovery even though the plaintiff probably would have suffered the loss without the defendant’s breach. In Sellars,[311] the High Court noted the difficulty of confining the Malec principle. The same kind of predictive uncertainty is presented by the concept of causation. In Sellars, proportional recovery was allowed, via the loss of chance doctrine, for the loss of potential profits even though it was improbable that the plaintiff would ever have enjoyed those profits.

Logically, it appears the proportional approach should be adopted wherever predictive uncertainty presents itself. On the issue of future quantum, it should be used to deal both with uncertainty about what will happen and what would have happened. On the issues of past quantum and causation, subject perhaps to one or two exceptions noted above, proportional compensation should be used to deal with uncertainty about what would have happened. What did happen is a matter of historical fact, for which the traditional ‘all or nothing’ approach seems appropriate.

However, causation in the context of law raises considerations of value as well as logic.[312] In Chappel[313] and Naxakis,[314] there was little support for the application of loss of chance to a medical negligence case. The judgments contained strong indications that personal injury plaintiffs experiencing difficulties in the proof of causation would be assisted by a different set of principles. These are the ‘injury within scope of risk’ principles considered in the next part. They operate within the traditional ‘all or nothing’ approach to causation, not just fulfilling a fraction of the injured plaintiff’s needs, but affording full compensation.

IV ‘INJURY WITHIN SCOPE OF RISK’: THE PRESUMPTION OF CAUSATION FROM BREACH

In Chappel and Naxakis, the High Court showed some reluctance to adopt a probabilistic notion of causation in the personal injury context. Instead, some support was given to an alternative response to the plaintiff’s difficulty in proving causation, one which does not involve a departure from the traditional ‘all or nothing’ approach. Essentially, this consists of the judicial recognition of a strong inference or presumption of causation, based on the defendant’s proven breach in creating a risk of injury of the kind suffered. The effect of the inference is to shift the evidentiary onus of proof of causation to the defendant, and, if the defendant cannot discharge this onus, causation is taken to be established on the balance of probabilities. Full compensation is then recoverable, subject to a possible Malec discount.

In Britain, the doctrine developed in the toxic tort cases, Bonnington Castings[315] and McGhee.[316] The defendant employer in each case was in breach in exposing the plaintiff to a toxic substance which carried a risk of injury, and the plaintiff subsequently developed an injury within the scope of that risk. However, causation was still difficult to prove, since others who had been exposed to the toxic substance did not develop the injury, and there were also other known causes of the injury. The plaintiffs’ difficulties in proving causation were overcome by the development of the presumption of causation from breach and injury.

Loss of chance reasoning does appear applicable to such toxic tort cases. Suppose that the defendant negligently exposed a group of 100 people to a toxic substance.[317] Ten of the group suffer a medical condition which is a known risk of that toxic substance, however statistical data suggest that six of them would have developed the condition from other causes without the defendant’s breach.[318] It appears arguable that the plaintiff’s injuries were probably caused not by the defendant’s breach, but by other causes to which the majority of these injuries may be attributed. As discussed below,[319] if the ‘guilty’ and ‘innocent’ causes operate cumulatively rather than independently, there is an opposing argument that the defendant’s breach made a material contribution to the plaintiff’s injury, albeit a lesser contribution than the other causes.

Loss of chance reasoning is applicable to these facts. The defendant’s breach decreased their chance of avoiding the condition by four per cent, or, equivalently, the defendant increased their risk of contracting the condition by four per cent.[320] On this basis, however, the plaintiffs would receive very inadequate compensation — only four per cent of the actual measure of their harm. In addition, the defendant would pay for only a small portion of the damage they actually caused. Suppose that full compensation for each plaintiff was $100,000. The defendant would pay $4,000 to 10 plaintiffs, a total of $40,000. However, it seems the defendant was responsible for four additional cases of the medical condition, with losses of $400,000.

It could perhaps be argued that all 100 members of the group would be able to bring an ‘exposure to risk’ action, even those for whom the risk did not ensue, in which case the defendant would pay the full measure of the damage they had caused, a total of $400,000.[321] This has a certain logic to it, since the action is for the risk or chance and not the actual damage. However, plaintiffs who suffered no actual damage are unlikely to be successful. Perhaps the ‘exposure to risk’ action could be modified, so that each of the injured plaintiffs recovers not four per cent but 40 per cent of the full amount of loss.[322] This may be justified on the basis that the risk did ensue, and roughly 40 per cent of the risk came from the defendant’s breach. This would also result in the defendant paying a full measure of damages. However, the plaintiffs would still receive less than half of their full losses.

Consider now the results that would follow from a presumption of causation on proof of breach and injury. Assuming the defendant failed to rebut the presumption, the defendant would pay each of the 10 injured plaintiffs $100,000, the actual measure of each plaintiff’s loss. The defendant may be entitled to a Malec discount of six per cent for the probability that the plaintiff would have suffered the injury even without the defendant’s breach, in which case each award would be $94,000. This amount is still far closer to the true costs of the medical condition. However, the defendant would pay a total of $1,000,000 or $940,000 — more than twice the actual damage that it caused.

As in all cases where certainty is unattainable, the risk of under-compensation or over-compensation is unavoidable.[323] The ‘injury within scope of risk’ principles clearly shift the risk from the plaintiff to the defendant. As a consequence, the defendant’s liability may appear to include a punitive as well as a compensatory component. This may be justified on the basis that the defendant was at fault. They did create the risk of the plaintiff’s injury. And perhaps it is preferable that the defendant pay too much, rather than that the personal injury plaintiff receive too little.

A The Presumption of Causation from Breach and Injury

1 The High Court and the House of Lords

In Betts v Whittingslowe, Dixon J said

breach of duty coupled with an accident of the kind that might thereby be caused is enough to justify an inference, in the absence of any sufficient reason to the contrary, that in fact the accident did occur owing to the act or omission amounting to the breach.[324]

This passage received recent High Court approval in separate judgments by Gaudron and Kirby JJ in Chappel.[325] And McHugh and Gummow JJ in separate judgments made similar statements.[326] An alternative formulation of the principle, adopted in separate judgments by McHugh and Kirby JJ, is that, on proof of breach, the ‘evidentiary onus’ on the causation issue shifts to the defendant.[327] Further support was given to this line of reasoning by Gaudron, Kirby and Callinan JJ in separate judgments in Naxakis.[328]

Gaudron J provided a rationale for this evidential presumption[329] in Bennett.[330] In order to find that the defendant’s breach was the cause of the plaintiff’s injury, it is necessary to find that the injury would not have occurred without the defendant’s breach. However, the finding that the defendant’s conduct constituted a breach in itself implies a general finding that the defendant’s conduct was sufficient to cause the injury: ‘a precaution is not classified as “reasonable” unless it can be said that its performance would, in the ordinary course of events, avert the risk that called it into existence’.[331] Palmer is right to question the correctness of this statement. ‘Sometimes the common law requires a person to take precautions which would merely reduce, rather than eliminate, a particular risk.’[332] Gaudron J did acknowledge that a finding of breach implies only the ‘general sufficiency’ of the precaution that the defendant failed to take, and is not necessarily conclusive.[333] It is sometimes necessary for the court to consider ‘what would or would not have happened in the particular circumstances of the case’.[334]

These recent High Court discussions of the relationship between breach and causation have drawn upon a series of British decisions, in which similar principles were discussed.[335] In the landmark English case, Bonnington Castings,[336] the plaintiff had contracted pneumoconiosis from the inhalation of silica dust while working for the defendant. The dust came from two sources, a pneumatic hammer and a swing grinder. The first of these was ‘innocent’ in that the defendant was not in breach in failing to protect the plaintiff from dust from this source. However, the second source was ‘guilty’. The defendant was in breach of a statutory duty in failing to contain the dust from this source. It appeared, however, that exposure to dust from the innocent source was greater than that from the guilty source.[337] On these facts, the plaintiff was facing considerable difficulties in establishing that, but for the guilty exposure, he would not have contracted the disease. This was a case where the precaution required of the defendant did not eliminate the risk of injury but merely reduced it somewhat.[338] It is far from clear that the finding of breach supported an inference that would establish causation on the balance of probabilities. Given the greater magnitude of the innocent exposure, it appeared at least arguable that the plaintiff would probably have contracted the disease without the guilty exposure. This argument is explored in the next section. Nevertheless, the court equated risk and material contribution,[339] and the plaintiff was successful.

This decision was followed by the House of Lords in McGhee, in which Lord Reid said:

From a broad and practical viewpoint I can see no substantial difference between saying that what the defender did materially increased the risk of injury to the pursuer and saying that what the defender did made a material contribution to his injury.[340]

More recently, however, the existence of a general presumption of causation from breach and injury was rejected by the House of Lords in Wilsher.[341] Lord Bridge indicated it was quite wrong to suggest that McGhee ‘modifies, as a matter of law, the nature of the burden of proof on causation which a plaintiff or pursuer must discharge once he has established a relevant breach of duty’.[342] According to Lord Bridge, the majority in McGhee merely drew a ‘legitimate inference’ of causation which was open on the ‘undisputed primary facts of the case’.[343] However, there is only a narrow range of cases in which the inference is legitimate. In McGhee, the defendant’s breach exposed the plaintiff to a toxic substance with the potential to combine in a cumulative way with other risks to produce the plaintiff’s injury. In Wilsher, however, the defendant’s negligence exposed the plaintiff to a toxic substance with the potential to operate as an independent cause of the plaintiff’s injury, and the inference of causation was not open. The distinction between cumulative and independent causes is discussed further below.[344]

It appears that High Court authority may support a stronger, more general relationship between findings of breach and causation than does current House of Lords authority. The High Court has applied the presumption beyond just the cumulative toxic tort cases.[345] Bennett[346] concerned the loss of the opportunity to litigate. The plaintiff, a youth in the care of the government department administered by the defendant, had been injured while using the department’s equipment with inadequate supervision. The department paid the plaintiff an amount of compensation according to the workers’ compensation scale, and the plaintiff failed to bring a common law action within the limitations period. The plaintiff argued that the defendant had breached its duty by failing to provide him with legal advice regarding his entitlement to common law damages.[347] The defendant presented two arguments against causation.

The defendant argued first that ‘but for’ causation was not established.[348] Even had the plaintiff been provided with legal advice, the advice may have been wrong, and thus the common law action may not have been pursued. The majority dismissed this argument in a couple of sentences, indicating that

[t]he director’s duty was to obtain legal advice from a competent legal practitioner based on adequate and accurate instructions. The minister does not, and could not, contend that advice sought from such a source and based on such instructions would not have acknowledged or asserted the existence of a cause of action.[349]

The defendant’s second line of attack was to attempt to break the chain of causation with a novus actus interveniens. After leaving the department’s care, the plaintiff obtained independent legal advice which failed to inform him of his right to compensation at common law. However, this argument was also dismissed by the majority, primarily for the reason that, had the defendant fulfilled its duty in the first place, there would have been no need for the plaintiff to seek this independent legal advice.[350] The majority therefore indicated it did not need to consider

the view that there is no real distinction between breach of duty and causation ... as well as the question whether a failure to take steps which would bring about a material reduction of the risk amounts to a material contribution to the injury [and] a possible shift in the onus of proof ... [I]t seems that the problem still awaits final resolution.[351]

Gaudron J apparently took the first argument of the plaintiff more seriously, and it was in this connection that she considered the breach–causation connection. She indicated that

[t]he finding that there was a duty [on the part of the Director to obtain independent legal advice] could only have been made on the basis that his Honour was satisfied, on the balance of probabilities, that in the ordinary course of events and by reason of that advice, [the plaintiff] would have been made aware of his right to recover damages ...[352]

On the facts of Bennett, Gaudron J’s presumption of causation from breach appears quite justified. As McHugh J noted, had the director fulfilled his duty, this would ‘almost certainly’[353] have resulted in advice that common law proceedings could have been commenced.

The presumption has also been considered by the High Court in the context of medical negligence. In Chappel, it is arguable that the presumption did not have to do much work. As discussed above, it was applied by the majority merely to confirm the positive result of the ‘but for’ test that had already been obtained.[354] However, in Naxakis[355] it was suggested that the presumption could be used to overcome genuine difficulties in establishing ‘but for’ causation. Even had the plaintiff’s condition been properly diagnosed, there were doubts about whether it would have been possible to improve the plaintiff’s prognosis.[356] The High Court indicated that the presumption of causation was open from breach and injury. It should be noted however that these comments were in the context of a decision that the trial judge was in error in taking the case from the jury. The High Court was therefore suggesting not that the jury should infer causation on these facts, but that a jury could infer causation.[357]

2 State Supreme Courts

Several State supreme courts have considered the breach–causation relationship in toxic tort cases resembling the British cases, with diverse results. A majority of the Full Court of the Supreme Court of South Australia supported a strong inference of causation from breach in Birkholz v R J Gilbertson Pty Ltd.[358] The plaintiff had contracted brucellosis while working in the defendant’s abattoir. There was evidence that the infection may have come from either guilty or innocent sources. On appeal, the Supreme Court held that the plaintiff had proved on the balance of probabilities that the infection came from a guilty source. However, King CJ added: ‘I do not think that the [plaintiff’s] inability to prove on the balance of probabilities that the infection occurred [from a guilty source] would be fatal to the success of his case.’[359] Relying upon McGhee, King CJ advanced this broad obiter dictum:

Where a defendant is under a legal duty to take precautions to protect the plaintiff from the risk of contracting disease, and, by omitting those precautions, he substantially increases the risk of the plaintiff contracting that disease, the law treats that increase in risk as a sufficient basis, in the absence of evidence showing how the infection occurred, for an inference that the omission of the precautions materially contributed to the contracting of the disease.[360]

Bollen J agreed with the Chief Justice. However, Matheson J rejected this broad principle, questioning whether such a strong inference was logically consistent with the precise facts of the case. Anticipating the House of Lords decision in Wilsher, he suggested that the inference of causation was only open where the innocent and guilty exposures operated cumulatively, whereas in this case they operated independently. His Honour’s analysis of cumulative and independent causes is considered further below.

Birkholz was decided prior to the House of Lords contraction of the ‘injury within scope of risk’ principles in Wilsher.[361] Following Wilsher, the Full Court of the Supreme Court of Victoria took a more limited view of the breach–causation relationship in Wintle v Conaust (Vic) Pty Ltd.[362] The plaintiff had been negligently exposed to asbestos while working as a stevedore on separate occasions for each of the five defendants. The plaintiff contracted mesothelioma, for which exposure to asbestos is the only known cause. Each defendant was a guilty source of exposure. But, from each defendant’s perspective, there were four additional innocent sources — the other defendants. The plaintiff brought his action out of the limitations period, and needed to demonstrate that ‘there is evidence to establish the cause of action’.[363] The majority[364] considered the British authorities, particularly the narrow approach taken by the House of Lords in Wilsher, and considered that the plaintiff’s case had a fatal weakness. It was not clear that asbestos operated cumulatively to cause mesothelioma,[365] and therefore the plaintiff was unable to establish ‘a foundation for proof of a relevant causal connection with any one particular [defendant]’.[366] Phillips J dissented, considering that, even in view of the views expressed by the House of Lords in Wilsher, the reasoning in McGhee was applicable to the present case.[367]

The New South Wales Court of Appeal has recently decided a series of appeals from the Dust Diseases Tribunal involving asbestos exposure and mesothelioma. In the most recent of these, E M Baldwin & Son Pty Ltd v Plane,[368] Fitzgerald AJA said:

The differences of opinion evident in some of the cases are, for the most part, less related to questions of principle and more attributable to conflicting views of the primary facts, and, in some instances at least, the conclusions to be drawn from expert evidence.[369]

However, while it is true that the expert witnesses did support different aetiological theories at different times, the judgments also display different principles concerning the inference of causation from breach and injury. In Baldwin, the plaintiff was successful against both defendants whose negligence had exposed the plaintiff to asbestos. Assisting the plaintiff, there was stronger evidence than had previously been available that asbestos fibres operated cumulatively in causing mesothelioma rather than independently.[370] However, the Court also quoted at length from the decision of the High Court in Chappel[371] where the presumption of causation from breach and injury received considerable support.[372]

In the decision of Bendix Mintex Pty Ltd v Barnes,[373] which preceded Chappel, the plaintiff was unable to obtain compensation from any of the parties that had exposed him to asbestos. The majority considered that the disease had been caused by asbestos exposure while the plaintiff was employed with the Royal Navy,[374] however the plaintiff was refused leave to proceed against the Ministry of Defence of the United Kingdom.[375] The majority held that the plaintiff had not established that the asbestos fibres operated cumulatively to cause mesothelioma, and the plaintiff failed to establish causation against the remaining defendants.[376] While the evidence in favour of the cumulative theory of mesothelioma may have been weaker than it was in Baldwin,[377] it is also apparent that the plaintiff was made to bear a heavier burden of proof on the causation issue. On this matter, the majority made statements that appear quite inconsistent with the presumption of causation from breach and injury as now formulated by the High Court. Beazley JA referred inter alia to Wintle v Conaust (Vic) Pty Ltd and the House of Lords decision in Wilsher, and held that ‘the clear effect of these cases is that the onus remains on the plaintiff to prove causation on the balance of probabilities. The onus is not discharged by establishing that a particular matter cannot be excluded as a cause of the injury.’[378] Mason P quoted with approval a statement from a case note on the House of Lords decision in Wilsher: ‘The law does not equate the situation where the defendant had materially increased the risk of injury with one where he had materially contributed to the injury.’[379] While his Honour referred to Gaudron J’s broader statement in Bennett,[380] he did not consider this authority clear enough to override the House of Lords decision in Wilsher.[381]

Dissenting in Bendix, Stein JA applied principles more consistent with the current High Court authorities. His Honour relied inter alia upon King CJ’s judgment in Birkholz[382] and Lord Reid’s judgment in McGhee,[383] and indicated ‘[t]here is no difference between materially increasing the risk of injury and making a material contribution to the injury’.[384] While acknowledging the doubts about the aetiology of the disease, his Honour held that the ‘repeated series of risks ... over a continuous and lengthy exposure should be treated as a single cumulative risk of indivisible outcome, which is the onset of disease’.[385]

In view of these decisions of the New South Wales Court of Appeal, it may be surprising that a majority of the court endorsed the presumption of causation from breach and injury prior to Chappel and Naxakis in Commonwealth v McLean.[386] The plaintiff contracted throat cancer for which he sought to hold the defendant liable. The plaintiff had been a naval seaman on board HMAS Melbourne when it collided with HMAS Voyager in February 1964, a collision for which the defendant was alleged to be vicariously liable. The plaintiff claimed that his throat cancer had been caused by his high level of consumption of alcohol and cigarettes, which was caused by a stress disorder suffered as a result of the collision. At trial, a jury held the defendant liable, and it appealed on many grounds. One was the summing up of the trial judge on proof of causation. The defendant’s counsel in his final address suggested to the jury that there was a risk that the plaintiff would have suffered throat cancer without the defendant’s negligence, since he was smoking four or five cigarettes a day before the collision. In responding to this, the trial judge referred to Bonnington Castings[387] and told the jury:

[W]here the defendant’s breach of duty contributes to a potentially harmful situation, thereby increasing the risk of injury, you are entitled to find that the injury resulted from the defendant’s breach of duty if the defendant’s contribution was, in your judgment significant.[388]

The Court of Appeal ultimately ordered a new trial, however it held that a direction of this nature was warranted.[389] Their Honours acknowledged that the House of Lords in Wilsher limited the scope of the earlier decisions in Britain, however since then Gaudron J had made broad statements on the breach–causation relationship in Bennett.[390] The court pointed out that Gaudron J’s comments on the inference of causation from breach did not directly cover the present case. The defendant owed a ‘general [duty] to take care not to injure others’, while the plaintiff’s throat cancer was ‘indirect or consequential damage suffered as a result of initial injuries’.[391] However, the principle still assisted the plaintiff in establishing the second link in the chain of causation. ‘[T]his principle is relevant when deciding whether the increased consumption of alcohol and tobacco due to the tort caused the throat cancer. It is not relevant in deciding whether the tort had caused that increase in consumption.’[392] The principle establishes ‘responsibility where injury results from a noxious substance or situation where only some of the exposure was caused by the tort’.[393]

3 The Logic and Policy of Presumptions

An underlying issue in this area of law is whether the inference, presumption or onus shift is justified as a matter of science and logic, or whether it is the product of policy. Of the English decisions, the judgment of Lord Wilberforce in McGhee was the most open in addressing this issue. In Chappel, Kirby J quoted a lengthy passage from Lord Wilberforce’s judgment, in which his Lordship acknowledged that ‘there is an appearance of logic in the view that the pursuer, on whom the onus lies, should fail’. However, Lord Wilberforce indicated that there are ‘further considerations of importance’. Where there is an inherent evidential difficulty attending the proof of causation, Lord Wilberforce suggested that ‘as a matter of policy or justice ... it is the creator of the risk ... who should bear its consequences’. [394] Kirby J indicated that he found ‘Lord Wilberforce’s exposition compelling’,[395] even though this passage had been singled out for disapproval by the House of Lords in Wilsher as a ‘minority opinion’ with ‘no support’.[396]

The question whether a principle is motivated by logic or by policy often arises with presumptions. Some presumptions are merely formalisations of ‘natural inferences’,[397] facilitating conclusions that are probably correct. An example is the presumption of the accuracy of scientific instruments.[398] Other presumptions, however, reflect concerns beyond accurate factual reasoning. They reflect ‘additional extraneous policy considerations. For example illegitimacy is (or has been) cause for social stigma and therefore should be clearly established by the party asserting it.’[399] Given the policy consideration, the presumption that a child born during lawful marriage is legitimate may require strong evidence to rebut. Such policy considerations no doubt led to the introduction of legislation in some jurisdictions making the presumption of legitimacy irrebuttable.[400] Irrebuttable presumptions are clearly a matter of policy and values rather than logic and fact, since ‘what is true in the eyes of the law is not directly linked to what may be true in “fact” ... The conditions of expressing the truth no longer depend on “objective reality”’.[401]

A presumption of particular relevance in the present context is that of res ipsa loquitur. This evidential presumption[402] assists the personal injury plaintiff where they face difficulties in the proof of breach. The defendant’s breach will be presumed where ‘the thing [causing the plaintiff’s injury] is shewn to be under the management of the defendant or his servants’.[403] The question arises again whether the presumption is a matter of logic or of policy. There is, of course, High Court authority that ‘the so-called “doctrine” is no more than a process of logic’.[404] But it should be noted that the presumption also serves policy goals. In Fleming’s discussion of the doctrine, he notes that ‘[o]ver the years the general trend has undoubtedly become sympathetic to plaintiffs’.[405] He observes that courts should not be too demanding in weighing the plaintiff’s evidence, ‘lest its aid be withheld when it is most needed’.[406] And, in the area of product liability, the effect of the presumption was that

[t]he standard of responsibility demanded from manufacturers quickly assumed some characteristics of strict liability ... The practical result of thus shifting the onus of proof is to make the manufacturer virtually an insurer against manufacturing defects ...[407]

Essentially the same policy considerations apply to the presumption of causation from breach in personal injury cases as apply to the res ipsa loquitur presumption of breach from injury. Broadly speaking, both kinds of cases are similar in their context, parties, facts and injuries.

The recognition of the role played by policy in the construction of presumptions helps to solve something of a mystery regarding the onus of proof. If strength of proof is to be measured on the scale of mathematical probability, the only situation in which the location of the onus of proof matters is where the probability favouring the plaintiff is precisely equal to the standard of proof,[408] usually interpreted as a probability of 50 per cent.[409] In that situation, the party bearing the onus would fail. If the probability level is 49 per cent, the plaintiff would fail, and, if the probability level is 51 per cent, the plaintiff would succeed, wherever the onus lies. Yet discussions of presumptions and the onus of proof suggest that these principles have significance far beyond cases that are perfectly balanced.[410]

The importance placed on presumptions and shifting burdens reflects the difficulty of allocating a probability value to a body of evidence.[411] Even statistical evidence may be open to conflicting probabilistic interpretations, as demonstrated by the discussion of cumulative and independent causes in the next section.[412] If the fact-finder is left in doubt about which interpretation to adopt, the party bearing the onus may fail. The importance placed on presumptions may also reveal that they are a matter of policy as well as probability, and may have the effect not only of shifting the onus of proof, but also of raising the standard of proof. Some presumptions are stronger than others.[413] Irrebuttable presumptions raise the standard to a level that is impossible to meet.

The precise nature of the presumption of causation given proof of breach is unclear. But there is an identifiable policy driving the presumption, and its effect may be not merely to shift the onus to the defendant, but also to ‘raise the bar’ for the defendant.[414] It is well established that, while the civil standard is the ‘balance of probabilities’, the precise meaning of this expression depends upon the circumstances of the case and, in particular, what is at stake for each of the parties.[415] The variability of the civil standard is all the more important given that, in many of the toxic tort cases where the presumption of causation has application, there may be statistical data available which make explicit probabilistic calculations feasible. The probabilistic analysis of such data is considered in the next section.

B Cumulative and Independent Causes and Material Contribution

In cases where the plaintiff’s injuries result from exposure to some toxic or noxious substance, there may be statistical data available correlating rates of exposure with rates of disease. This statistical data is open to probabilistic interpretation. For example, if there is frequency data that 10 per cent of people who are exposed to the toxic substance contract the disease, this suggests that the probability of disease (D) given exposure (E) is also 10 per cent — P(D|E) = 0.10. In this section, I consider some of the complexities that arise in analysing data of disease rates from both ‘guilty’ (G) and ‘innocent’ (I) exposures.

1 The Probabilistic ‘But For’ Test

Of course, if the exposure from the defendant’s breach, E, was the only exposure and the only known cause of the disease, then the plaintiff would probably be considered to have satisfied the ‘but for’ test and would receive full compensation whatever the rate of disease. The chance of the exposure causing the disease may be low, but this must be the explanation because there is no other possibility. The difficulty in the multiple exposure cases is that, as well as the guilty source, G, there is an innocent source of exposure, I. If this is a lesser source — say, P(D|I) < P(D|G) = 0.20 — again, the court may consider the ‘but for’ test established on the balance of probabilities.[416] Suppose that P(D|I) = 0.10. Then the court might consider that the guilty source, G, was twice as likely to be the ‘but for’ cause as the innocent source, I. Since the only two possible causes are G and I, the probability of G being the cause would be almost 67 per cent, which satisfies the traditional standard of proof.

If, however, the probability of disease given the innocent exposure is greater than the probability of disease given the guilty exposure — P(D|I) > P(D|G) — then this probabilistic ‘but for’ test would not be satisfied. These figures appear to support an argument that the innocent exposure is more likely to be the ‘but for’ cause of the disease than the guilty exposure, and the plaintiff would fail to satisfy the traditional standard of proof.[417] However, in Bonnington Castings the plaintiff succeeded, even though ‘much the greater proportion of the noxious dust’[418] came from the innocent rather than the guilty source. In Hotson, in the Court of Appeal, Croom-Johnson LJ referred to Bonnington Castings, and said ‘if there is more than one cause the contribution may be less than half so long as it is material’.[419] This appears inconsistent with the probabilistic interpretation of the ‘but for’ test. Perhaps these decisions can be understood to reflect policy considerations rather than logic. However, a logical justification for the Bonnington Castings principle has been advanced, which is considered in the next section.

2 The Distinction Between Independent and Cumulative Causes

In a number of judgments, as noted above, a distinction has been drawn between multiple possible causes that operate independently and those that operate cumulatively. It has been suggested that the Bonnington Castings principle is applicable to the latter situation, while the probabilistic ‘but for’ test is applicable to the former.

Lord Reid adverted to this distinction in McGhee,[420] a case in which the plaintiff was claiming damages for dermatitis caused by exposure to dust at work. As in Bonnington Castings, there was both a guilty and an innocent source of exposure.[421] Lord Reid suggested that the dust particles may operate cumulatively or independently: ‘It may be that an accumulation of minor abrasions of the ... skin is a necessary precondition for the onset of the disease. Or it may be that the disease starts at one particular abrasion and then spreads’.[422]

In the case of independent operation, one or other of the exposures would be the cause, and the probabilistic ‘but for’ reasoning appears an appropriate way of choosing between them. In the case of cumulative operation, however, the guilty and innocent exposures may have both operated in combination. In this case, the guilty exposure may still be considered a cause, though a lesser cause than the innocent exposure, on the basis that it may have ‘tipped the scale’.[423]

As discussed above, these were competing theories with regard to the aetiology of mesothelioma. As Phillips J explained in Bryce v Swan Hunter Group,

[i]t is possible that risk rises with exposure simply because each fibre has a chance of initiating the fatal mutation so that the more fibres that are inhaled the greater are the chances that one will succeed in so doing. It may be, however, that, as the quantity of fibres in the lung increases, they also have a cumulative effect in damaging or overwhelming the body’s defence mechanism so as to facilitate the occurrence of malignancy.[424]

In the former case, the causes would be operating independently, while in the latter case they would be operating cumulatively.

Given the uncertainty that attends the aetiology of many medical conditions, it will be difficult to tell whether multiple possible causes operate cumulatively or independently. Ideally, it would be apparent from statistical epidemiological data. Suppose P(D|I) = 0.20, and P(D|G) = 0.10. If the two exposures operated independently, it would be expected that the probability of disease, given both exposures, would be P(D|I&G) = 0.28. Out of 100 cases of joint exposure, the innocent source would account for 20 cases of the disease. And then the guilty source would be expected to result in disease in 10 per cent of the remaining 80 cases, giving 28 cases of disease in total. In general, for independent causes, we would expect

P(D|I&G) = P(D|I) + P(D|G) . (1 – P(D|I)) ...(1)

If the disease rate in cases of joint exposure were greater than 28 per cent, this would suggest that the sources operated cumulatively rather than independently. For cumulative causes,

P(D|I&G) > P(D|I) + P(D|G) . (1 – P(D|I)) ...(2)

If (2) is satisfied rather than (1), this suggests guilty exposure did not merely operate independently in the 80 cases of non-disease remaining after the effect of the innocent exposure. Among those 80 cases, the guilty exposure also combined with the innocent exposure, ‘tipping the scale’ in cases where the innocent exposure was insufficient by itself to cause disease. And where the two sources operate cumulatively in this way, it becomes impossible to attribute any individual case of disease to one source rather than the other. The probabilistic ‘but for’ test appears inapplicable.[425] The two sources must be considered to have both contributed to each case of disease.

The inference of causation in Bonnington Castings[426] and McGhee[427] can be justified on the narrow ground that the two exposures operated cumulatively rather than independently. In Wilsher,[428] the House of Lords overturned the decision of the Court of Appeal, which had followed the earlier cases. The House of Lords distinguished Bonnington Castings and McGhee, and held that the facts in Wilsher were such that the various possible causes were not cumulative. In Wilsher, the infant plaintiff was born nearly three months premature, and, since his lungs were not fully developed, extra oxygen was administered. The hospital was negligent in monitoring the oxygen, and the plaintiff was given too high a dose. The plaintiff developed an incurable condition, retrolental fibroplasia (‘RLF’), which severely impaired his vision. He claimed compensation from the health authority on the basis that this was caused by the negligent administration of excessive levels of oxygen. However, he faced a difficulty establishing causation because there was a number of other possible causes of the condition. The condition was a known risk for premature babies, even where there was no over-administration of oxygen.[429]

The trial judge held in favour of the plaintiff. The Court of Appeal upheld the verdict on the basis that proof of the breach gave rise to a presumption of causation which the defendant had failed to rebut. However, as discussed above, the House of Lords denied that the onus of proof on the causation issue shifts to the defendant.[430] The inference of causation was open in Bonnington Castings and McGhee, however the present case was distinguishable. In delivering the leading judgment, Lord Bridge expressed approval of the dissenting judgment of Browne-Wilkinson V-C in the Court of Appeal, in which the Vice-Chancellor appeared to have found that the multiple possible causes operated independently. The Vice-Chancellor said:

There are a number of different agents which could have caused the RLF. Excess oxygen was one of them. ... A failure to take preventative measures against one out of five possible causes is no evidence as to which of those five caused the injury.[431]

An interesting case in which expert evidence suggested independent rather than cumulative multiple causes is Birkholz,[432] the majority judgment of which was discussed in the previous section. The plaintiff contracted brucellosis while working in the defendant’s abattoir. On appeal, the court held that the plaintiff had established that the source of the infection was ‘guilty’ rather than ‘innocent’. However, their Honours went on to consider whether the reasoning in McGhee had application. The majority held that it did.[433] However, Matheson J, in the minority on this point, indicated that he thought that the case was distinguishable, as the various exposures operated independently. There was evidence that repeated small doses, rather than cumulatively increasing the risk of serious infection, would, on the contrary, build up the subject’s immunity to infection.[434] For this reason, the different sources of exposure constituted ‘alternative causes’[435] rather than cumulative causes as in Bonnington Castings and McGhee.[436] The plaintiff ‘contracted brucellosis as a result of an infection incurred on one occasion, and not as a result of an accumulation of organisms over a period of time’.[437] The probabilistic ‘but for’ test would have application rather than the Bonnington Castings principle. If the innocent exposure were greater in magnitude than the guilty exposure, it would appear more probable than not that the innocent exposure was the ‘but for’ cause.

3 Background Exposure and the Probabilistic ‘But For’ Test

In many cases, the innocent source will be background exposure. In this situation, it will be difficult to get figures for the guilty exposure alone — P(D|G). Rather than having figures for the risk given guilty exposure and the risk given innocent exposure — P(D|G) and P(D|I) — the available data will be for the risk given guilty and innocent exposure together, and the risk given innocent exposure alone — P(D|I&G) and P(D|I). If this is the case, the probabilistic test (2) for distinguishing between independent and cumulative causes will be of no assistance.

Commentators have suggested that, with figures for background exposure and for background and guilty exposure, the guilty exposure will satisfy the probabilistic ‘but for’ test if the risk given guilty and background exposure is greater than twice the risk given background exposure alone.[438] However, this is not quite correct. As argued above, the probabilistic ‘but for’ reasoning only appears appropriate if the two sources are independent rather than cumulative. In that case, the relationship is not, as these commentators imply,

P(D|I&G) = P(D|I) + P(D|G) ...(3)

which gives the incorrect test,

P(D|I&G) > 2 . P(D|I) ...(4)

The relationship is

P(D|I&G) = P(D|I) + P(D|G) . (1 – P(D|I)) ...(5)

which gives

P(D|I&G) > 2 . P(D|I) – P(D|I)2 ...(6)

This formula is not so easily expressed in words.

The difference between the incorrect test (4) and the correct test (6) is quite small for low risk exposures. For example, if P(D|I) = 0.01, then the incorrect threshold for P(D|I&G) suggested by (4) is 0.02, while the correct threshold suggested by (6) is 0.0199, a difference of 0.0001 — only 0.5 cent of the correct value. However, the difference between the tests becomes greater for higher risk exposures. If, for example, P(D|I) = 0.4, the incorrect threshold for P(D|I&G) suggested by (4) is 0.8 while the correct threshold suggested by (6) is 0.64, a difference of 0.16 — 25 per cent of the correct value.

V CONCLUSION

In Part IV, the ‘injury within scope of risk’ principles were discussed. In Wilsher,[439] the House of Lords narrowed the operation of these principles.[440] It appears English courts may now draw the inference of causation from breach only in toxic tort cases where the plaintiff’s exposure to the toxic substance flowing from the defendant’s breach is greater than other exposures, or where innocent and guilty exposures operate cumulatively rather than independently. A number of members of the High Court supported a far broader principle in Bennett,[441] Chappel[442] and Naxakis.[443] Australian courts may presume causation from breach in a wide range of personal injury cases. The presumption applies in medical negligence cases as well as toxic tort cases.

It could be argued that the English approach has greater validity in terms of logic and probability theory. The defendant may have committed a breach by failing to reduce a risk. However, even had the defendant complied with the duty, a considerable risk would have remained. It may appear probable, therefore, that the injury would have occurred without the defendant’s breach, in which case a finding of breach does not lead to a finding of ‘but for’ causation. However, the appeal to logic is out of place. The ‘injury within scope of risk’ principles operate within a broader context of ‘all or nothing’ causation. And, once the existence of objective chance is recognised, which has apparently been the case for many years in the traditional principles governing proof and recovery of future loss, then ‘all or nothing’ causation itself appears illogical.

The High Court has recognised the logic of chance in its treatment of quantum.[444] The High Court has acknowledged that objective chance presents difficulties not only for future quantum, but also for the predictive element of past quantum. Would the plaintiff have suffered the same pre-trial loss, albeit slightly later in time, without the defendant’s breach? This is a predictive inquiry, about which certainty is not achievable, and, in Australia, a Malec discount may be imposed to make allowance for this possibility.[445] The House of Lords in Hotson, illogically, opposed a proportional approach to past quantum.[446]

It is clear that the loss of chance doctrine applies to commercial cases both in Australia[447] and in England.[448] The plaintiff may recover for a lost commercial opportunity or the lost opportunity to litigate, even though the court has doubts about whether the opportunity would have crystallised into a concrete actual benefit. Compensation is in proportion to the probability that the plaintiff would have obtained an actual benefit. This constitutes a probabilistic conception of causation. In both England and Australia, however, it is doubtful whether the loss of chance doctrine applies to medical negligence or toxic tort personal injury cases.[449]

In Australia, though no longer in England, there is support for the illogical but humane ‘injury within scope of risk’ principles to help the plaintiff overcome the inherent difficulty of proving causation. Once the defendant’s breach is proved, causation is presumed. Unless the defendant can rebut the presumption, and this may not be easy to do, the plaintiff will recover full compensation, subject possibly to a slight Malec discount. These principles may treat the plaintiff and the defendant unequally, but then, in many personal injury cases, the equities are unequal.[450] The injured plaintiff has more at stake. Moreover, as Lord Loreburn LC commented almost a century ago, ‘a Court should be very indulgent and always bear in mind who was to blame’.[451]

[*] BSc, LLB (Hons) (ANU); currently completing PhD dissertation, The University of Melbourne; Lecturer, School of Law, The University of New England, <dhamer2@metz.une.edu.au>. I acknowledge the helpful comments of Richard Johnstone, Edwin Coleman, Harry Geddes, Anne Deegan and Geoff Hamer on an earlier draft. I am grateful to the anonymous referee for their detailed comments, and for drawing to my attention a number of pertinent authorities.

[1] ‘That degree [of cogency required in a civil case] is well settled. It must carry a reasonable degree of probability ... If the evidence is such that the tribunal can say: “We think it more probable than not,” the burden is discharged, but, if the probabilities are equal, it is not’: Miller v Minister of Pensions [1947] 2 All ER 372, 374 (Denning J), quoted in J D Heydon, Cross on Evidence (5th Aust ed, 1996) 235. Cf the Evidence Act 1995 (Cth) s 140(1), but note s 140(2). See also below nn 414–415 and accompanying text.

[2] See below Part III(C)(4) for discussion of whether the concept of causation varies for different causes of action and different types of injuries.

[3] The authorities are not easy to reconcile, however it appears that, while the plaintiff bears a burden to prove to the traditional standard the nature of their injury, eg, ‘physical loss’, the burden of proof is lighter with respect to ‘the value in money terms of that loss’: Harold Luntz, Assessment of Damages for Personal Injury and Death (3rd ed, 1990) 98. In Commonwealth v Amann Aviation Pty Ltd [1991] HCA 54; (1991) 174 CLR 64, 80, Mason CJ and Dawson J observed that ‘[t]he onus of proving damages sustained lies on a plaintiff’. However, a little later in their judgment they added that ‘[m]ere difficulty in estimating damages does not relieve a court from the responsibility of estimating them as best it can’: at 83. Cf Deane J at 118, 125. The apparent inconsistency between these two statements can be resolved by reference to Toohey J’s observation that ‘to say as a general proposition that it is for the plaintiff to prove his damages is not to say that, in some instances, damage may not be inferred or presumed’: at 138. Toohey J’s comment also has application to Watts v Rake [1960] HCA 58; (1960) 108 CLR 158, 159, where Dixon CJ suggests that ‘[t]he law of course places upon a plaintiff ... the burden of satisfying the tribunal of fact of the damages he has suffered both special and general and of the quantification in money that should be adopted in the sum awarded’. That case stands as authority for a presumptio hominis discussed further below that actually assists the personal injury plaintiff in obtaining the full quantum of damages for their injuries: see below n 324. JLW (Vic) Pty Ltd v Tsiloglou [1994] VicRp 16; [1994] 1 VR 237 is an unusual case in which the court recognised that the plaintiff had suffered a loss, but held no damages were recoverable in view of the lack of credible evidence as to quantum: at 248 (Brooking J), 250–1 (Tadgell and J D Phillips JJ). However, two particular features of the case should be noted. First, the loss was of a kind that the plaintiff should have been able to quantify with precision — goods stolen from a shop. Secondly, the plaintiff had actually led evidence of the quantum of the loss, but the trial judge had considered this evidence to have been fabricated. With reference to the first point, there is authority that the plaintiff is ‘required to prove its loss with as much certainty and particularity as [is] reasonable’: from the head note of Ray Teese Pty Ltd v Syntex Australia Limited [1998] 1 Qd R 104, 104, see 109–10 (Pincus JA; Fitzgerald P and Thomas J agreeing); JLW (Vic) Pty Ltd v Tsiloglou [1994] VicRp 16; [1994] 1 VR 237, 241–3 (Brooking J); Commonwealth v Amann Aviation Pty Ltd [1991] HCA 54; (1991) 174 CLR 64, 138 (Toohey J); Ratcliffe v Evans [1892] UKLawRpKQB 131; [1892] 2 QB 524, 532–3 (Bowen LJ). However, Pincus JA in Ray Teese Pty Ltd v Syntex Australia Ltd [1998] 1 Qd R 104, 110 questioned whether it was appropriate in JLW (Vic) Pty Ltd v Tsiloglou to award no damages given the court’s view that the plaintiff had suffered some loss. While Ray Teese Pty Ltd v Syntex Australia Ltd was also a case of an appeal court overturning a trial judge’s award of damages, this was on the ground of lack of proof of loss per se rather than lack of proof of a specific quantum. For a discussion of this issue in relation to loss of chance, see below Part III(B)(2).

[4] See below n 49 and accompanying text.

[5] See, eg, Wynn v NSW Insurance Ministerial Corp [1995] HCA 53; (1995) 184 CLR 485 (‘Wynn’), discussed in text accompanying below nn 123–134.

[6] Quantum has a causal nature, however there is a valid distinction between it and causation proper. See below Part II(B)(1).

[7] (1990) 169 CLR 638 (‘Malec’). See below Part II(B)(3).

[8] (1994) 179 CLR 332 (‘Sellars’). See below Part III(A).

[9] See below Part III(C)(3).

[10] (Unreported, Supreme Court of New South Wales, Badgery-Parker J, 20 June 1997) <http://online.butterworths.com.au> at 11 June 1999, discussed in text accompanying below n 215–218.

[11] (1998) 156 ALR 517 (‘Chappel’). See below Part III(C)(4).

[12] (1999) 162 ALR 540 (‘Naxakis’). See below Part III(C)(4).

[13] Bonnington Castings Ltd v Wardlaw [1956] UKHL 1; [1956] AC 613 (‘Bonnington Castings’); McGhee v National Coal Board [1972] UKHL 7; [1973] 1 WLR 1 (‘McGhee’). See below Part IV(A)(1).

[14] [1987] UKHL 11; [1988] AC 1074 (‘Wilsher’).

[15] See below Part IV(A)(1).

[16] Chappel (1998) 156 ALR 517; Naxakis (1999) 162 ALR 540.

[17] Bennett v Minister of Community Welfare [1992] HCA 27; (1992) 176 CLR 408 (‘Bennett’).

[18] In this article, the term ‘logic’ is used in the broad, everyday sense of valid or correct reasoning. This extends beyond deductive reasoning to inductive probabilistic reasoning and the appropriate use of empirical data.

[19] See below Part II(A)(1).

[20] See below Part IV(A)(3).

[21] Arthur Danto, Narration and Knowledge (revised ed, 1985) 144; David Lewis, Philosophical Papers (1986) vol 2, 93–5, 114–15. Cf Karl Popper, A World of Propensities (1990) 18.

[22] Lewis, above n 21, 93; D H Mellor, The Matter of Chance (1971) 15.

[23] For simplicity of terminology, I will hereafter use the term ‘event’ to mean ‘event or non-event’.

[24] Mark Kelman, ‘A Rejoinder to Cass R Sunstein’ in James Chandler, Arnold Davidson and Harry Harootunian (eds), Questions of Evidence: Proof, Practice and Persuasion across the Disciplines (1994) 199.

[25] Lewis, above n 21, 94.

[26] Determinism, of course, appears inconsistent with human autonomy: see, eg, Hans Kelsen, What Is Justice: Justice, Law, and Politics in the Mirror of Science (1957) 334. See also below nn 96, 249–251. Prigogine refers to the temporal symmetry of ‘physical existence’: Ilya Prigogine, ‘The Meaning of Entropy’ in Werner Callebaut and Rik Pinxten (eds), Evolutionary Epistemology (1987) 57.

[27] Prigogine, above n 26, 58. Cf Edith Sylla, ‘Political, Moral and Economic Decisions and the Origins of the Mathematical Theory of Probability: The Case of Jacob Bernoulli’s The Art of Conjecturing’ in George von Furstenberg (ed), Acting under Uncertainty: Multidisciplinary Conceptions (1990) 19, 37; Theodore Porter, ‘The Quantification of Uncertainty after 1700: Statistics Socially Constructed?’ in George von Furstenberg (ed), Acting under Uncertainty: Multidisciplinary Conceptions (1990) 43, 47; Shlomith Rimmon-Kenan, Narrative Fiction: Contemporary Poetics (1983) 43.

[28] John Dupré, The Disorder of Things: Metaphysical Foundations of the Disunity of Science (1993) 186–7; Fritjof Capra, The Tao of Physics (2nd ed, 1985) 55–61.

[29] Ludwig Wittgenstein, Philosophical Investigations (G E M Anscombe trans, 1953) [223e], quoted in Danto, above n 21, 172.

[30] Michael Polanyi, Personal Knowledge (revised ed, 1962) 145; Capra, above n 28, 144.

[31] See Dupré, above n 28, 186.

[32] Ibid 95–6.

[33] Karl Popper, The Poverty of Historicism (2nd ed, 1961) 139 (emphasis in original):

In general it is only by the use of artificial experimental isolation that we can predict physical events. (The solar system is an exceptional case — one of natural, not of artificial isolation; once its isolation is destroyed by the intrusion of a foreign body of sufficient size, all our forecasts are liable to break down.) We are very far from being able to predict, even in physics, the precise results of a concrete situation, such as a thunderstorm, or a fire.

This may lead to the suggestion that isolated deterministic systems may exist, within which time is symmetrical. Dupré, above n 28, 190 points out the difficulty with this view: ‘[If] it is conceded that any of these domains are truly indeterministic, the determinism of all other domains is threatened by interactions’ (emphasis in original).

[34] See Paul Humphreys, The Chances of Explanation: Causal Explanation in the Social, Medical, and Physical Sciences (1989) 3.

[35] See Dupré, above n 28, 95–6, 174–5, 186.

[36] ‘[A] deterministic world [is] one in which everything that happens is fully necessitated by antecedent circumstances’: ibid 171.

[37] Ibid 175 (emphasis in original).

[38] Humphreys, above n 34, 17. Dupré, above n 28, 184 notes that determinism ‘seems almost entirely, or perhaps entirely, devoid of empirical support’.

[39] Consider, for example, lingering doubts about the events leading up to the death of John F Kennedy:

The Warren Commission on the Assassination of President Kennedy had a superb group of triers, a large staff of skilled lawyers, all the investigative resources of the United States at its command, and ample time. As Professor Goodhart pointed out, ‘no other inquiry in the whole of legal history ever approached the Warren Commission in the extent and detail of its researches.’ Yet even this enormous fact-finding machine left some lingering doubts.

Jack Weinstein, ‘Some Difficulties in Devising Rules for Determining Truth in Judicial Trials’ (1966) 66 Columbia Law Review 223, 242 (citations omitted).

[40] In many cases, assessments of evidence will also be ‘subjective’ in the sense of being relative to the assessor’s personal viewpoint, which is influenced by their cultural background and life experiences: see, eg, Kim Lane Scheppele, ‘Foreword: Telling Stories’ (1989) 87 Michigan Law Review 2073.

[41] The Concise Oxford Dictionary (10th ed, 1999) 981.

[42] The distinction between objective and subjective certainty helps to reconcile what would otherwise be contradictory statements in Twining’s ‘common assumptions’ of ‘rationalist theories of evidence’. His first assumption is ‘[k]nowledge about particular past events is possible’. This appears to contradict the fourth assumption that ‘[t]he establishment of the truth of alleged facts in adjudication is typically a matter of probabilities, short of absolute certainty’: William Twining, ‘Rationality and Scepticism in Judicial Proof: Some Signposts’ (1989) 2 International Journal for the Semiotics of Law 69, 72. The contradiction is resolved by recognising that the first assumption makes reference to objective certainty, while the fourth assumption refers to subjective uncertainty.

[43] Cf Danto, above n 21, 172.

[44] Cf L Jonathan Cohen, ‘Subjective Probability and the Paradox of the Gatecrasher’ [1981] Arizona State Law Journal 627, 632.

[45] The civil standard is frequently interpreted as ‘more probable than not’. See, eg, Utting v Luhtala [1948] ArgusLawRp 65; (1984) 55 ALR 261, 264 (Gibbs CJ; Murphy, Wilson, Brennan and Deane JJ agreeing). This suggests a mathematical interpretation of a probability of 50 per cent: Davies v Taylor [1974] AC 207, 219 (Lord Simon). See also Murphy J’s judgments in West v Government Insurance Office of NSW [1981] HCA 38; (1981) 148 CLR 62, 69–70 and TNT Management Pty Ltd v Brooks (1979) 23 ALR 345, 351–4. In a few exceptional cases, where scant evidence is available to the court, some judges have expressed doubts about the mathematical interpretation, suggesting an ‘actual persuasion’ is required: see, eg, Briginshaw v Briginshaw [1938] HCA 34; (1938) 60 CLR 336, 361 (Dixon J). See generally David Hamer, ‘The Civil Standard of Proof Uncertainty: Probability, Belief and Justice’ [1994] SydLawRw 37; (1994) 16 Sydney Law Review 506, 509–12.

[46] David Kaye, ‘Book Review: Naked Statistical Evidence’ (1980) 89 Yale Law Journal 601, 605; Laurence Tribe, ‘Trial by Mathematics: Precision and Ritual in the Legal Process’ (1971) 84 Harvard Law Review 1329, 1378–86.

[47] See also below nn 414–415 and accompanying text.

[48] Malec (1990) 169 CLR 638, 643 (Deane, Gaudron and McHugh JJ).

[49] It might be argued that the traditional civil standard does have application to future losses. However, it is the existence of the chance that must be established on the balance of probabilities. The chance itself may be less than 50 per cent. An analogous view has been taken in loss of chance cases. See below nn 190–197 and accompanying text.

[50] See below nn 92–97, 168–172 and accompanying text.

[51] ‘Causation has plagued courts and scholars more than any other topic in the law of torts’: John Fleming, The Law of Torts (9th ed, 1998) 218. See also March v E & M H Stramare Pty Ltd [1991] HCA 12; (1991) 171 CLR 506 (‘March’); Chappel (1998) 156 ALR 517, 540–1 (Kirby J). Causation was the primary issue in Chappel. Each of the five members of the court gave a separate judgment, and they split 3:2 (Gaudron, Gummow and Kirby JJ; McHugh and Hayne JJ).

[52] See below n 74.

[53] Fleming, The Law of Torts, above n 51, 220. See Bennett [1992] HCA 27; (1992) 176 CLR 408, 413 (Mason CJ, Deane and Toohey JJ); March [1991] HCA 12; (1991) 171 CLR 506, 515 (Mason CJ); Chappel (1998) 156 ALR 517, 555–6 (Hayne J).

[54] March [1991] HCA 12; (1991) 171 CLR 506, 517 (Mason CJ), cf 523 (Deane J) and 534 (McHugh J).

[55] [1963] VicRp 51; [1963] VR 339, 358.

[56] Surprisingly, Haber v Walker was not referred to in Chappel. See also Tony Honoré, ‘Medical Non-Disclosure, Causation and Risk: Chappel v Hart’ (1999) 7 Torts Law Journal 1, 4, 7.

[57] The plaintiff alleged breach of contract, however to recover more than nominal damages the plaintiff was required to ‘establish an element of causation analogous to that in tort’: Chappel (1998) 156 ALR 517, 533 (Gummow J).

[58] Gaudron, Gummow and Kirby JJ; McHugh and Hayne JJ.

[59] Chappel (1998) 156 ALR 517, 524, cf 555 (Hayne J).

[60] Ibid 531 (McHugh J), cf 556 (Hayne J).

[61] Ibid 529 (McHugh J), 561 (Hayne J).

[62] Ibid 529 (McHugh J), cf 558–9 (Hayne J). Kirby J was ‘[f]or a time ... attracted to [this argument]’: at 550.

[63] Ibid 536.

[64] Ibid 520 (Gaudron J), 535 (Gummow J), cf 524–5 (McHugh J). The ‘injury within scope of risk’ principles are discussed in below Part IV.

[65] Ibid 522 (Gaudron J), 551–2 (Kirby J). Gummow J did not express a view. Honoré, ‘Medical Non-Disclosure’, above n 56, 8 considers that the defendant should have been held liable even if the defendant’s negligence had not increased the risk of injury, since the plaintiff’s injury was ‘the type against which he should have warned her [and the defendant] did cause the harm that [the plaintiff] suffered, though not by the advice he failed to give her’ (emphasis in original).

[66] [1991] HCA 12; (1991) 171 CLR 506, 515, referring to John Fleming, The Law of Torts (7th ed, 1987) 172–3; H L A Hart and Tony Honoré, Causation in the Law (2nd ed, 1985) 110. Cf Fleming, The Law of Torts, above n 51, 218; Tony Honoré, ‘Necessary and Sufficient Conditions in Tort Law’ in David Owen (ed), Philosophical Foundations of Tort Law (1995) 363, 363: ‘There is, I contend, no special “legal” meaning of causation’; cf 375–6.

[67] [1991] HCA 12; (1991) 171 CLR 506, 533–4.

[68] See below Part II(A)(2).

[69] [1991] HCA 12; (1991) 171 CLR 506, 522 (Deane J), 524 (Toohey J), 525 (Gaudron J).

[70] Ibid 515.

[71] Ibid (emphasis added).

[72] Ibid 509.

[73] Ibid 522.

[74] In Chappel (1998) 156 ALR 517, 523, McHugh J conceded that ‘the court has refused to regard the concept of remoteness of damage as the appropriate mechanism for determining the extent to which policy considerations should limit the consequences of causation-in-fact’. However, he quoted the passage from March in which Mason CJ drew the distinction between scientific causation and legal causation: see text accompanying above n 66. Gummow J also quoted this passage: Chappel (1998) 156 ALR 517, 534. Kirby J in Chappel drew upon the statements of Mason CJ and Deane J in March and suggested that, if the ‘but for’ test is satisfied, causation ‘generally speaking, [will] be established’. The effect of recent discussions of the ‘but for’ test is merely to ‘temper the results thereby produced with “value judgments” and “policy considerations”’: at 545. Hayne J also suggested that a positive result to the ‘but for’ test may be disregarded on the basis of ‘value judgments [and] policy considerations’: at 556.

[75] Mirjan Damaška, ‘Presentation of Evidence and Factfinding Precision’ (1975) 123 University of Pennsylvania Law Review 1083, 1103. Damaška is talking about criminal proof, however his comments appear pertinent to civil matters. See also Simon Blackburn, ‘Review: L Jonathan Cohen, The Probable and the Provable’ (1980) 44 Synthese 149, 149; Jonathan Koehler and Daniel Shaviro, ‘Veridical Verdicts: Increasing Verdict Accuracy Through the Use of Overtly Probabilistic Evidence and Methods’ (1990) 75 Cornell Law Review 247, 248; Robert Thompson, ‘Decision, Disciplined Inferences and the Adversary Process’ (1991) 13 Cardozo Law Review 725, 728–9.

[76] March [1991] HCA 12; (1991) 171 CLR 506, 516 (Mason CJ), see also 523 (Deane J), 534 (McHugh J); Chappel (1998) 156 ALR 517, 555 (Hayne J). The over-determination problem is even greater for quantum issues: see below Part II(B)(1).

[77] Richard Wright, ‘Causation, Responsibility, Risk, Probability, Naked Statistics, and Proof: Pruning the Bramble Bush by Clarifying the Concepts’ (1988) 73 Iowa Law Review 1001, 1018.

[78] March [1991] HCA 12; (1991) 171 CLR 506, 516–17.

[79] Wright, above n 77, 1019; Hart and Honoré, above n 66, 110–14; Fleming, The Law of Torts, above n 51, 219, 222; Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66, 364. Cf Mackie’s INUS test — insufficient but necessary element of an unnecessary but sufficient set of conditions — which appears to be essentially the same as the NESS test: John Mackie, The Cement of the Universe: A Study in Causation (1974) 59–63, discussed by Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66, 365. Mackie, however, favours the ‘but for’ test. See Wright, above n 77, 1023.

[80] Wright, above n 77, 1022.

[81] Cf ibid 1020; Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66, 363–4.

[82] Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66, 375.

[83] Ibid.

[84] See above n 77.

[85] See also Elizabeth Adeney, ‘The Challenge of Medical Uncertainty: Factual Causation in Anglo-Australian Toxic Tort Litigation’ [1993] MonashULawRw 2; (1993) 19 Monash University Law Review 23, 27. In March [1991] HCA 12; (1991) 171 CLR 506, the judgments note the difficulty that the ‘but for’ test has with over-determination. While reference is made to supporters of the NESS test, such as Hart, Honoré and Wright, the judgments do not discuss the NESS solution: see above n 69. McHugh J provides a formulation of the NESS test — ‘every necessary member of the set of conditions which is sufficient to produce the relevant damage is a cause of that damage’ — but then describes this as ‘the basis for the but for test’ and suggests that it is subject to ‘some exceptional cases’, apparently unaware that it overcomes the difficulty of over-determination: at 529–30. Palmer rehearses McHugh J’s mistake: Andrew Palmer, ‘Causation in the High Court’ (1993) 1 Torts Law Journal 9, 10.

[86] [1954] HCA 74; (1954) 91 CLR 268, 277–8 (Dixon CJ, Fullagar and Kitto JJ), the latter part recently quoted in March [1991] HCA 12; (1991) 171 CLR 506, 515 and Chappel (1998) 156 ALR 517, 544 (Kirby J). Note that in Fitzgerald v Penn, unlike more recent cases, the court was concerned with the appropriate jury instruction on causation, and the risk of confusion.

[87] As McHugh J noted in March [1991] HCA 12; (1991) 171 CLR 506, 532, ‘it is doubtful whether there is any consistent commonsense notion of what constitutes a “cause”. ... “[I]n general talking about the ordinary man, common sense, and everyday judgments appears somewhat hazardous”’ (quoting Sally Lloyd-Bostock, ‘The Ordinary Man, and the Psychology of Attributing Causes and Responsibility’ (1979) 42 Modern Law Review 143, 167). See also Haber v Walker [1963] VicRp 51; [1963] VR 339, 357–8 (Smith J) (FC), quoted in Harold Luntz and David Hambly, Torts: Cases and Commentary (4th ed, 1995) 278.

[88] March [1991] HCA 12; (1991) 171 CLR 506 (see above nn 69–70 and accompanying text); Chappel (1998) 156 ALR 517, 519 (Gaudron J), 523 (McHugh J), 534 (Gummow J), 544–5 (Kirby J), 557 (Hayne J). In Chappel, Gummow, Kirby and Hayne JJ discussed the recent House of Lords authority which endorsed the common sense approach to causation, Empress Car Co (Abertillery) Ltd v National Rivers Authority [1998] UKHL 5; [1998] 1 All ER 481, 486 (Lord Hoffman).

[89] As noted in the previous section, the ‘sufficient set’ of NESS conditions is not perfectly actual. It may exclude certain actual conditions. In the two-fire case, fire X is found to be a NESS cause by excluding fire Y from the sufficient set.

[90] Cf Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66, 370 (emphasis in original): ‘[W]hichever theory we adopt, a hypothetical question must be put and answered. This involves a counterfactual proposition, namely that if a condition that in fact occurred had not occurred, the outcome would have been so-and-so.’

[91] Chappel (1998) 156 ALR 517, 555 (Hayne J).

[92] [1992] HCA 27; (1992) 176 CLR 408, 422–3, quoted with approval by Brennan J in Sellars (1994) 179 CLR 332, 367. Cf Bennett [1992] HCA 27; (1992) 176 CLR 408, 428 (McHugh J): ‘Whether or not a causal connection exists between a breach of duty and any harm suffered by the person to whom the duty is owed is a question of fact to be decided on the balance of probabilities.’ Luntz, above n 3, 81–2 (citations omitted):

[Courts] generally regard the cause of the plaintiff’s injury ... as theoretically capable of certain proof, so that if such a link is established on a balance of probabilities the plaintiff recovers for that loss in full, whereas if the proof falls short of such a balance, the plaintiff recovers nothing at all ...

Cf Luntz and Hambly, above n 87, 278.

[93] Wright, above n 77, 1042.

[94] Ibid 1029.

[95] Ibid 1041.

[96] Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66, 373 (emphasis added). Cf Honoré’s observation that counterfactual statements are ‘in principle still true or false’ even though ‘we cannot tell for certain which it is’: at 374. While he acknowledges that quantum mechanics is indeterministic, he suggests that this is partitioned off from the world in which we operate. ‘As Wright neatly observes, indeterminacy at this micro-level ... does not affect the macro-phenomenon which are the law’s concern’: at 382. But this ignores the possibility of interaction. See above n 33. He questions whether indeterministic candy machines, hypothesised by Mackie, are genuinely indeterministic, suggesting their ‘erratic’ behaviour could be explained: at 381. The only example of indeterminism that Honoré is prepared to concede is human conduct, based upon a belief in human autonomy, and even this may merely be an example of ‘unpredictability’: at 382. See also nn 26, 249–251.

[97] Robertson and Vignaux make a similar suggestion that it is a ‘fallacy’ to consider ‘probability as a property of objects and processes in the real world rather than a measure of our own uncertainty’: Bernard Robertson and G A Vignaux, ‘Probability — The Logic of the Law’ (1993) 13 Oxford Journal of Legal Studies 457, 460. They argue that ‘[t]here are always causes; it may simply be that the process is beyond our current understanding’: at 470. See also below nn 168–172.

[98] See above nn 34–38 and accompanying text and above Part II(B)(2).

[99] See S M Waddams, ‘Principles of Compensation’ in Paul Finn (ed), Essays on Damages (1992) 1, 11.

[100] Ibid 1 (emphasis added).

[101] Luntz, above n 3, 80:

In accordance with the basic principle that the plaintiff is to be put back into the position in which he or she would have been but for the defendant’s wrong, it is necessary for the court to consider what that position would have been. Once, however, the wrong has actually been committed, it can never be known for certain what that position would have been and the court is necessarily thrown back on evaluating the chances.

[102] Waddams, above n 99, 1 (emphasis added).

[103] Ibid.

[104] See below nn 110–111 and accompanying text.

[105] 163 A 111 (NH, 1932).

[106] Fleming, The Law of Torts, above n 51, 225.

[107] Wright considers that the ‘but for’ test does not adequately deal with successive over-determination, but that the NESS test does. In the two-fires case, if fire X arrived first, destroying the house before fire Y’s arrival, Wright suggests that neither would be a ‘but for’ cause. He claims, however, that the NESS test would identify fire X as the cause, but not fire Y.

[T]here was only one actually sufficient set — the one containing fire X but not fire Y. The set containing fire Y but not fire X was hypothetically, but not actually, sufficient. It would have been sufficient if fire X had not existed, since then the house would still have been standing when fire Y arrived — but this was not an actual condition. The earlier arrival of fire X pre-empted its occurrence.

Wright, above n 77, 1022. However, Wright’s analysis appears unnecessarily complex. Suppose fire X arrived at 1pm, destroying the house by 2pm with fire Y arriving at 3pm. If at 2pm we had asked, ‘would the house be standing but for fire X?’, the answer would be ‘yes’, indicating that fire X is a ‘but for’ cause. If at 4pm we had asked, ‘would the house be standing but for fire Y?’, the answer would be ‘no’. Fire X had already destroyed it. Fire Y is not a ‘but for’ cause. To reject this analysis, Wright must be applying the ‘but for’ test with no temporal component. However, this would be inconsistent, since in applying the NESS test Wright does take account of time. He recognises that fire X ‘pre-empted’ fire Y.

[108] Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66, 379. Cf Luntz and Hambly, above n 87, 289. Performance Cars Ltd v Abraham [1961] EWCA Civ 3; [1962] 1 QB 33 is a case of successive over-determination in which the second potential cause did not add to the impact of the first cause. The plaintiff’s Rolls Royce sustained similar damage in two minor collisions within the space of a fortnight. The defendant who had caused the second collision was successful in appealing against an award of damages for respraying. ‘[T]he necessity for respraying was not the result of the defendant’s wrongdoing because that necessity already existed’: at 39 (Lord Evershed MR). Luntz and Hambly, above n 87, 289 (emphasis in original), suggest that the defendant should have been held liable despite the absence of causation:

Although where it is known that the damage has already occurred, we cannot truly say that the second tortfeasor has caused it [but] there seems to be no sound reason of policy why we should not attribute responsibility to someone whose negligence would have caused the same harm if it had not already happened ...

[109] [1981] UKHL 3; [1982] AC 794 (‘Jobling’).

[110] Ibid 815. Cf Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66, 379–80.

[111] Jobling [1981] UKHL 3; [1982] AC 794, 815. See also Baker v Willoughby [1969] UKHL 8; [1970] AC 467 (partially overruled in Jobling); Wynn [1995] HCA 53; (1995) 184 CLR 485, 498 (Dawson, Toohey, Gaudron and Gummow JJ) (see below n 131 and accompanying text). On this basis, presumably both defendants in the two-fire case will be held jointly liable.

[112] Jobling [1981] UKHL 3; [1982] AC 794, 815.

[113] Cf Watts v Rake [1960] HCA 58; (1960) 108 CLR 158, a ‘bad back’ case which also raised both causation and quantum issues. The defendant argued

first that [the plaintiff] was predisposed to many or at least some of the arthritic and other conditions which have so seriously and rapidly developed as a consequence of the accident considered at all events as a precipitating cause; second, that part of his present condition is traceable to causes other than the accident, and thirdly, that had there been no accident he would eventually and prematurely have been incapacitated by the seeds of disability within him.

[1960] HCA 58; (1960) 108 CLR 158, 160 (Dixon CJ). The three arguments overlap, however it is possible with a little artificiality to separate them. The first argument is that the plaintiff’s predisposition and the defendant’s breach in combination contributed to the plaintiff’s injuries at the time of the accident. According to the so-called ‘thin skull’ rule, ‘a negligent defendant must take his victim as he finds him’, and damages will not be reduced: at 164 (Menzies J), 160 (Dixon CJ). The second argument also involves causation and quantum issues. The plaintiff was already suffering back problems which the defendant’s breach merely added to. Both potential causes had impact. The defendant should not pay damages for problems already in existence. There is no causation. The third argument recognises that the defendant’s breach caused the plaintiff’s back problems to occur at the time of the accident, but submits that the plaintiff would have suffered them in any event at a later date. Damages should be discounted accordingly. This argument appears to be directed purely at quantum. The defendant appears to concede that the plaintiff’s pre-disposition will now have no causal effect. It has been causally pre-empted. The back condition that would have occurred later has been ‘accelerated’ by the defendant’s breach: at 163 (Menzies J). The second and third arguments are valid, however the plaintiff has the advantage of a presumptio hominis, and the evidential burden shifts to the defendant: at 163 (Menzies J), 159–60 (Dixon J); Purkess v Crittenden [1965] HCA 34; (1965) 114 CLR 164, 168 (Barwick CJ, Kitto and Taylor JJ), 171 (Windeyer J). See also below n 324.

[114] Cf the figure appearing in Luntz and Hambly, above n 87, 295, adapted from A I Ogus, The Law of Damages (1973) 65.

[115] See Luntz and Hambly, above n 87, 295.

[116] Jobling [1981] UKHL 3; [1982] AC 794, 815.

[117] Ibid. See also Wynn [1995] HCA 53; (1995) 184 CLR 485, discussed in the text accompanying below n 130.

[118] See Luntz and Hambly, above n 87, 295.

[119] Jobling [1981] UKHL 3; [1982] AC 794, 804.

[120] Ibid.

[121] Luntz and Hambly, above n 87, 289 suggest that a further distinction may be drawn between property damage and physical injury, and that, on the property damage side, a series of shipping cases involving successive collisions or a collision followed by bad weather is ‘not easy to reconcile’.

[122] P S Atiyah, Accidents, Compensation and the Law (3rd ed, 1980) 176.

[123] [1995] HCA 53; (1995) 184 CLR 485.

[124] Dawson, Toohey, Gaudron and Gummow JJ gave a joint judgment; Brennan CJ agreed subject to some comments on future economic loss for domestic help, discussed in below nn 132–136 and accompanying text.

[125] Wynn [1995] HCA 53; (1995) 184 CLR 485, 497–8.

[126] Ibid 497, quoting from Windeyer J in Bresatz v Przibilla [1962] HCA 54; (1962) 108 CLR 541, 544.

[127] Wynn [1995] HCA 53; (1995) 184 CLR 485, 498, citing, inter alia, Malec (1990) 169 CLR 638, 639–40; Sellars (1994) 179 CLR 332, 365–7 (Brennan J); Mallett v McMonagle [1970] AC 166, 176 (Lord Diplock). See the text accompanying below n 146, below Part III(A) and the text accompanying below n 150 respectively.

[128] Wynn [1995] HCA 53; (1995) 184 CLR 485, 494 (Dawson, Toohey, Gaudron and Gummow JJ), citing NSW Insurance Ministerial Corp v Wynn [1994] Aust Torts Reports 81-304, 61,742.

[129] Wynn [1995] HCA 53; (1995) 184 CLR 485, 494 (Dawson, Toohey, Gaudron and Gummow JJ).

[130] Ibid 499.

[131] Ibid 498. See above n 111.

[132] Wynn [1995] HCA 53; (1995) 184 CLR 485, 495, quoting from Sharman v Evans [1977] HCA 8; (1977) 138 CLR 563, 577 (Gibbs and Stephen JJ) (emphasis added).

[133] Wynn [1995] HCA 53; (1995) 184 CLR 485, 491.

[134] Ibid.

[135] Waddams, above n 99.

[136] See, eg, Kettles & Gas Appliances Ltd v Anthony Hordern & Sons Ltd [1934] NSWStRp 51; (1934) 35 SR (NSW) 108, 130–1 (Long Innes J).

[137] (1990) 169 CLR 638.

[138] [1977] HCA 45; (1977) 139 CLR 161.

[139] Malec (1990) 169 CLR 638, 641 (Deane, Gaudron and McHugh JJ).

[140] Ibid 642 (Deane, Gaudron and McHugh JJ).

[141] The majority judgment was given by Deane, Gaudron and McHugh JJ. Brennan and Dawson JJ, subject to some brief observations, agreed with the majority.

[142] See also Commonwealth v Amann Aviation Pty Ltd [1991] HCA 54; (1991) 174 CLR 64, 118–26 (Deane J).

[143] Malec (1990) 169 CLR 638, 642–3 (Deane, Gaudron and McHugh JJ) (emphasis added).

[144] Ibid 644.

[145] This is a result of the mathematical conjunction rule — the probability of the back condition leading to unemployability should be multiplied by the probability of the unemployability leading to the neurotic condition. The product is necessarily less than (or equal to) the first factor. See also below n 199 and accompanying text on the mathematical conjunction rule.

[146] Malec (1990) 169 CLR 638, 639–40 (Brennan and Dawson JJ).

[147] Ibid 640.

[148] Cf the comments of Callinan J in Naxakis (1999) 162 ALR 540, 575 in the text accompanying below n 309.

[149] (1998) 156 ALR 517, 521–2 (citations omitted) (emphasis added), cf 549–50 (Kirby J), 539–40 (Gummow J). In addition to Malec, Kirby J cited an English Court of Appeal authority, Allied Maples Group Ltd v Simmons & Simmons (A Firm) [1995] EWCA Civ 17; [1995] 1 WLR 1602, 1609–10 (‘Allied Maples’). However, while the Court of Appeal recognised the loss of chance doctrine (discussed in the text accompanying below nn 219–226), the pages cited by Kirby J do not endorse the discounting of past quantum. Stuart-Smith LJ drew the traditional distinction between ‘historical fact’ — ‘[o]nce established on balance of probability ... the plaintiff recovers his damage in full’ — and ‘future uncertain events’ — ‘these questions are not decided on a balance of probability, but rather on the court’s assessment, often expressed in percentage terms’: at 1609–10. See discussion of the English approach to past quantum in the text.

[150] [1970] AC 166.

[151] Ibid 176.

[152] Although, as Coote notes, ‘while the second part of Lord Diplock’s dictum refers to what would have happened in the future, his initial statement, about the distinction between what was on the one hand and what will be or would have been on the other, is not so confined’: Brian Coote, ‘Chance and the Burden of Proof in Contract and Tort’ (1988) 62 Australian Law Journal 761, 767.

[153] [1988] UKHL 1; [1987] AC 750 (‘Hotson’).

[154] Ibid 783.

[155] Ibid 793, referring to Bagley v North Herts HA (1986) 136 New Law Journal 1014.

[156] Cf Adeney, above n 85, 37.

[157] [1996] NSWSC 657; (1996) 41 NSWLR 389, discussed in the text accompanying below nn 386–393.

[158] Ibid 411 (Handley and Beazley JJA).

[159] Ibid.

[160] Doubts may remain whether the proportional approach to pre-trial quantum will operate to reduce a plaintiff’s damages below full compensation, a result that was rejected in Commonwealth v McLean [1996] NSWSC 657; (1996) 41 NSWLR 389, or whether the proportional approach will only operate to increase the plaintiff’s damages from nothing, which essentially was the result in Malec (1990) 169 CLR 638.

[161] Given the modus operandi of the gambling industry, it is likely that the plaintiff would have paid more than $10 for the ticket, in which case the plaintiff would be better off seeking a refund for breach of condition, rather than damages for loss of chance. Note also that the chance in this case was only 0.1 per cent, which, according to the majority in Malec, may make it too ‘speculative’ to take into account: see the text accompanying above n 143. Nevertheless, this remains the paradigmatic loss of chance case: see below n 162.

[162] Sellars (1994) 179 CLR 332, 349 (Mason CJ, Dawson, Toohey and Gaudron JJ); Commonwealth v Amann Aviation Pty Ltd [1991] HCA 54; (1991) 174 CLR 64, 120 (Deane J); Laferrière v Lawson [1991] 1 SCR 541, 605; 78 DLR (4th) 609, 654 (the majority judgment of Lamer CJ, L’Heureux-Dubé, Sopinka, Gonthier, Cory and McLachlin JJ was delivered by Gonthier J; La Forest J dissented) (‘Laferrière’).

[163] [1991] 1 SCR 541, 605; 78 DLR (4th) 609, 656.

[164] Ibid 549; 613.

[165] Ibid 550–1; 614.

[166] See the text accompanying above nn 93–95.

[167] Laferrière [1991] 1 SCR 541, 600–1; 78 DLR (4th) 609, 652.

[168] Ibid 605; 656.

[169] Ibid 574; 632, quoting with approval from the French commentator, René Savatier, Une faute peut-elle engendrer la responsabilité d’un dommage sans l’avoir causé (1970).

[170] Laferrière [1991] 1 SCR 541, 610–12; 78 DLR (4th) 609, 660–1.

[171] Timothy Hill, ‘A Lost Chance for Compensation in the Tort of Negligence by the House of Lords’ (1991) 54 Modern Law Review 511, 512. Cf the views of Wright, above n 77; Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66; and Robertson and Vignaux, above n 97. See above nn 92–97.

[172] See above nn 26, 96 and below nn 249–251.

[173] (1990) 169 CLR 638.

[174] (1994) 179 CLR 332.

[175] Ibid 345–6 (Mason CJ, Dawson, Toohey and Gaudron JJ), 358 (Brennan J).

[176] Ibid 347.

[177] Adelaide Petroleum NL v Poseidon Ltd (1990) 98 ALR 431, 531 (French J), quoted in ibid 344.

[178] Adelaide Petroleum NL v Poseidon Ltd (1990) 98 ALR 431, 531 (French J), quoted in Sellars (1994) 179 CLR 332, 343.

[179] Sellars (1994) 179 CLR 332, 347.

[180] Mason CJ, Dawson, Toohey and Gaudron JJ. Brennan J delivered a separate judgment along broadly similar lines.

[181] Sellars (1994) 179 CLR 332, 349–55. The most difficult case to reconcile appeared to be Norwest Refrigeration Services Pty Ltd v Bain Dawes (WA) Pty Ltd [1984] HCA 59; (1984) 157 CLR 149 (‘Norwest’). See below Part III(C)(2).

[182] (1990) 169 CLR 638.

[183] (1994) 179 CLR 332, 355, cf 350.

[184] Laferrière [1991] 1 SCR 541, 603; 78 DLR (4th) 609, 654. See above nn 163–170.

[185] It is tempting to present an argument, directly opposed to that in Laferrière, that the uncertainty of an event increases with the number of contingencies governing its occurrence. But, on reflection, the generalisation is inconsistent with experience. A passenger getting on a cross-Atlantic flight in New York can be quite certain that they will arrive safely in London, even though there are thousands of things that could go wrong with the plane.

[186] Sellars (1994) 179 CLR 332, 355. Cf Adeney, above n 85, 62, who refers to the ‘statistical certainty that mistakes will be made and that many plaintiffs will either be left uncompensated or will make windfall gains at the expense of “innocent” defendants. But this is not a problem with the evidence, it is a problem with the all-or-nothing rule.’ Cf also Fleming, The Law of Torts, above n 51, 229: ‘[D]enial [of the loss of chance doctrine] under-deters the guilty while continuing to (arguably) overcompensate plaintiffs who succeed in mounting the 50 per cent hurdle.’

[187] Atiyah, above n 122, 177.

[188] Ibid. Cf Hill, above n 171, 518 (emphasis in original): ‘[I]f the courts are minded to compensate for such a perceived loss of “chance” in this way, an unfair result will always be produced.’ Cf Waddams, above n 99, 11:

The implications of these decisions are ... not necessarily beneficial to plaintiffs, because it follows that a plaintiff who proves a 60 per cent probability of causation will recover only 60 per cent of the loss, not, as formerly, 100 per cent. This means that, even though the plaintiff proves that the defendant probably caused the loss, the plaintiff will be compensated for only a part of the loss — a case in which it was formerly thought that justice entitled the plaintiff to full compensation.

Cf Adeney, above n 85, 65: ‘Should one be concerned that under this system plaintiffs whose condition was fully attributable to the defendant’s breach would recover only partial damages or that all those whose disease was in fact not caused by the defendant’s breach would automatically recover?’

[189] [1992] HCA 55; (1992) 175 CLR 514. In situations where a court is ‘forced to estimate damages on the basis of likelihood or probability instead of assessing damages by reference to established events ... there would be an ever-present risk of undercompensation or overcompensation, the risk of the former being the greater’: at 527 (Mason CJ, Dawson, Gaudron and McHugh JJ).

[190] Sellars (1994) 179 CLR 332, 348 (Mason CJ, Dawson, Toohey and Gaudron JJ), citing Wardley Australia Ltd v Western Australia [1992] HCA 55; (1992) 175 CLR 514, 526.

[191] Sellars (1994) 179 CLR 332, 355 (Mason CJ, Dawson, Toohey and Gaudron JJ).

[192] Ibid 367.

[193] Ibid 367–8, approving a statement of Gaudron J in Bennett [1992] HCA 27; (1992) 176 CLR 408, 422–3.

[194] (1994) 179 CLR 332, 348 (emphasis added), see also 355.

[195] Ibid 368 (Brennan J):

Where a loss is alleged to be a lost opportunity to acquire a benefit, a plaintiff who bears the onus of proving that a loss was caused by the conduct of the defendant discharges that onus by establishing a chain of causation that continues up to the point when there is a substantial prospect of acquiring the benefit sought by the plaintiff. Up to that point, the plaintiff must establish both the historical facts and any necessary hypothesis on the balance of probabilities.

[196] Ibid 355 (Mason CJ, Dawson, Toohey and Gaudron JJ), cf 368 (Brennan J).

[197] Cf Coote, above n 152, 768: ‘[I]t is perhaps hardly necessary to add that the existence of the chance must itself be established on the balance of probabilities.’ Presumably, the civil standard applies in the same way to the chance of future losses. See above n 49.

[198] See, eg, Brian Skyrms, ‘Higher Order Degrees of Belief’ in D H Mellor (ed), Prospects for Pragmatism (1980) 109.

[199] The probability of the conjunction of two independent events is the product of the probability of the first event occurring and the probability of the second event occurring. In this instance, the first event is the existence of the chance, and the second event is the fruition of the chance. See also above n 145. The chance had a 70 per cent probability of existing and a 40 per cent probability of coming to fruition. The calculation can be checked by ensuring that the probability of the negation is complementary, that is, that the sum of the probability of the event and its negation sum to one. There are two paths by which the plaintiff may not have enjoyed the benefits of the draft Pagini contract. First, there may have been no chance of the benefits being enjoyed. This has a probability of 30 per cent. Secondly, the chance may have existed (70 per cent), but not have come to fruition (60 per cent). The probability of this conjunction is 42 per cent. By the addition rule, the probability of either of these no-benefits paths being taken is 72 per cent, which is the complement of the probability of the benefits being enjoyed.

[200] [1988] UKHL 1; [1987] AC 750 (CA), 770.

[201] (1994) 179 CLR 332.

[202] Hotson [1988] UKHL 1; [1987] AC 750, 760.

[203] Ibid, cf 764 (Dillon LJ), 768 (Croom-Johnson LJ).

[204] Lord Mackay noted that the Court of Appeal took a different view of the facts, and said ‘[m]uch of the judgment of the Court of Appeal will remain for consideration in the future’: ibid 790. Lord Bridge, however, said there were ‘formidable difficulties in the way of accepting the analogy’ between the economic loss of chance cases ‘and the situation of personal injury or medical negligence’: at 782.

[205] See Coote, above n 152, 765; Luntz, above n 3, 83. Hill’s assessment of the decision, that ‘the House of Lords refused to take the step from a “statistical” to a “personal” chance’ (Hill, above n 171, 515), is unjustified. But see Hotson [1988] UKHL 1; [1987] AC 750, 769 (Croom-Johnson LJ) (CA), quoted in Naxakis (1999) 162 ALR 540, 549 (Gaudron J). Indeed, there is no sharp distinction between a ‘statistical’ chance and a ‘personal’ chance. The distinction is, as Nesson notes, ‘a matter of degree’: Charles Nesson, ‘Agent Orange Meets the Blue Bus: Factfinding at the Frontier of Knowledge’ (1986) 66 Boston University Law Review 521, 534.

[206] Hotson v Fitzgerald [1985] 1 WLR 1041 (Simon Brown J) (QBD).

[207] Ibid.

[208] Ibid.

[209] Hotson [1988] UKHL 1; [1987] AC 750, 782 (Lord Bridge).

[210] Ibid 785 (Lord Mackay).

[211] Coote, above n 152, 769 discusses a New Zealand case, Takaro Properties v Rowling [1986] NZCA 27; [1986] 1 NZLR 22, which also concerned the distinction between a subjective certainty and an objective chance. The plaintiff claimed damages for the loss of a commercial opportunity arising from the defendant’s negligent denial of approval for a project to go ahead. The trial judge held in favour of the defendant on the basis that the plaintiff had not proved on the balance of probabilities that the project would have been profitable. The Court of Appeal allowed the plaintiff’s appeal. The trial judge had adopted an ‘approach which fails to distinguish between the difficulty of assessing the level of an admitted chance with evidence that shows the chance was actually present’: at 64. On the correct approach, the plaintiff was required to prove ‘that the project would have had the chance to trade its way out of trouble once the added share capital had been injected’, but not ‘that the chance was a better one than 50%’: at 64 (Woodhouse P). An appeal to the Privy Council was allowed on another ground. However, as Coote, above n 152, 770 notes, there is a suggestion in the Privy Council judgment that they considered that, ‘on the findings in the High Court, there had been no chance in the first place’. In this case, unlike in Hotson, the chance was not readily identifiable with a concrete state of affairs. In such cases, it is difficult to discern the difference between a high probability of the existence of a slight chance, and a slight probability of the existence of a considerable chance. And yet the difference determines whether the plaintiff would succeed or not.

[212] See also above n 205.

[213] (1995) 37 NSWLR 438.

[214] Ibid 526–7 (Clarke and Sheller JJA; Powell JA agreed on this point).

[215] (Unreported, Supreme Court of New South Wales, Badgery-Parker J, 20 June 1997) <http://

online.butterworths.com.au> at 11 June 1999.

[216] Ibid 43.

[217] Ibid 44 (emphasis added); see also 46.

[218] Ibid 41.

[219] [1995] EWCA Civ 17; [1995] 1 WLR 1602 (Stuart-Smith and Hobhouse LJJ; Millett LJ dissenting).

[220] Ibid 1614.

[221] Ibid.

[222] Ibid 1623.

[223] Ibid (emphasis added).

[224] Cf text accompanying above n 143.

[225] Allied Maples [1995] EWCA Civ 17; [1995] 1 WLR 1602, 1626.

[226] Ibid 1620 (Hobhouse LJ).

[227] Cf Luntz, above n 3, 91–2.

[228] Callaghan v Wm C Lynch Pty Ltd [1962] NSWR 871, 877 (Evatt CJ, Herron and Sugerman JJ) (FC). Cf Yorkshire Electricity Board v Naylor [1968] AC 529, 548 (Lord Devlin): ‘[I]n the law of damages ... difficulty in calculation is not ordinarily taken as a ground either for reducing or for increasing the award’; Luntz, above n 3, 100.

[229] [1991] HCA 54; (1991) 174 CLR 64.

[230] Ibid 125–6, quoting from McRae v Commonwealth Disposals Commission [1951] HCA 79; (1951) 84 CLR 377, 411 (Dixon and Fullagar JJ).

[231] Commonwealth v Amann Aviation Pty Ltd [1991] HCA 54; (1991) 174 CLR 64, 126. See also above n 3.

[232] Ibid 119–20, quoting from Takaro Properties Ltd v Rowling [1986] NZCA 27; [1986] 1 NZLR 22, 69 (Cooke J).

[233] See the text accompanying above nn 190–196.

[234] Fleming, The Law of Torts, above n 51, 221. Cf Bennett [1992] HCA 27; (1992) 176 CLR 408, 420 (Gaudron J):

[A] case based on omission ... will, in certain respects, fall for analysis in a way that differs from that appropriate for a case based on a positive act. Thus, in the case of a positive act, questions of causation are answered by reference to what, in fact, happened. In the case of an omission, they are answered by reference to what would or would not have happened had the act occurred.

See also Allied Maples [1995] EWCA Civ 17; [1995] 1 WLR 1602, 1609–10 (Stuart-Smith LJ); Luntz and Hambly, above n 87, 279.

[235] Fleming, The Law of Torts, above n 51, 221.

[236] Sellars (1994) 179 CLR 332, 367 (Brennan J).

[237] Ibid.

[238] Ibid. In Brennan J’s discussion, this quotation preceded the two referred to in the previous notes. The point in the text does not come out as strongly as it appears in Brennan J’s judgment.

[239] In Laferrière [1991] 1 SCR 541, 570–1; 78 DLR (4th) 609, 629–30, Gonthier J referred to the suggestion that loss of chance actions are limited to cases of omission. However, he saw ‘no basis for treating acts and omissions differently. Accordingly, there is no theoretical imperative directing courts to abandon traditional causal analysis and to adopt instead an essentially artificial loss of chance analysis’: at 605; 655.

[240] Hotson [1988] UKHL 1; [1987] AC 750, 762 (Donaldson MR).

[241] Allied Maples [1995] EWCA Civ 17; [1995] 1 WLR 1602, 1610 (Stuart-Smith LJ). Cf Coote, above n 152, 768.

[242] Norwest [1984] HCA 59; (1984) 157 CLR 149 and Gates v City Mutual Life Assurance Society Ltd [1986] HCA 3; (1986) 160 CLR 1.

[243] Sellars (1994) 179 CLR 332, 353 (Mason CJ, Dawson, Toohey and Gaudron JJ).

[244] Ibid 355. In Daniels v Anderson (1995) 37 NSWLR 438, 530 (Clarke and Sheller JJA; Powell JA agreeing with this part of the judgment), the court indicated that

there is something to be said for the view that the dictum in Sellars ... is inconsistent with the holding in Norwest. However, the majority in Sellars expressly said that the comments which are cited above were not critical to the decision in the case. Upon this basis it is open to this Court, in our view to apply the latter dicta.

As indicated in the text, I consider that the two can be reconciled. In Daniels v Anderson, however, it was appropriate for the court to follow Norwest. The plaintiff corporation suffered huge losses on foreign exchange investments and brought an action against their auditors. In that case, the hypothetical question was how the directors of the plaintiff would have acted had the defendants not been negligent in failing to advise them of gross deficiencies in the company’s records and internal controls. The court held that the directors would have taken appropriate steps, and so causation was established: at 539. There was still an element of chance as to whether some loss would nevertheless have been sustained, and the court indicated that a proportional award of damages could be made to take account of this: at 539. Fortunately, however, these ‘elements of uncertainty ... were removed by the agreement of the experts’: at 564. The court said that it was prepared to disregard the element of chance since ‘there is no basis for preferring adverse contingencies to beneficial ones’: at 564.

[245] The category covers all situations where the defendant is alleged to have failed to give an adequate warning, for example, occupiers’ liability for concealed hazards or medical negligence cases where the negligence consists of failing to warn of a risk associated with a treatment. Cf Chappel (1998) 156 ALR 517, 520 (Gaudron J), 527–8 (McHugh J); Laferrière [1991] 1 SCR 541, 605; 78 DLR (4th) 609, 630; Hotson [1988] UKHL 1; [1987] AC 750, 772.

[246] [1895] ArgusLawRp 104; (1973) 1 ALR 125.

[247] Ibid 134 (emphasis added); cf 127 (McTiernan ACJ agreeing).

[248] Ibid 138–9 (emphasis added). See also Quigley v Commonwealth (1981) 35 ALR 537; Luntz, above n 3, 82.

[249] See above n 26. As discussed in above n 96, Honoré is something of a determinist. However, he recognises that human conduct, given human autonomy, may be indeterminate, or at least unpredictable: Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66, 382. It may appear paradoxical that this recognised category of unpredictability is nevertheless subject to the traditional stricter proof requirements. The explanation is that the source of unpredictability is free will — the choice of the relevant individual. The individual’s choice is not pre-determined. If it were, it would not be a genuine choice. This is not to say that their choice is random like the throw of a die. See Dupré, above n 28, 214–15. Cf Kelsen, above n 26.

[250] Honoré supports the notion of human autonomy: see above n 249. However, in discussing Chappel, he suggests that ‘[t]he patient’s evidence of what she would have done in circumstances which did not in fact arise needs to be carefully scrutinised since she cannot know for certain what her reaction would have been’: Honoré, ‘Medical Non-Disclosure’, above n 56, 2.

[251] Quigley v Commonwealth (1981) 35 ALR 537 was a case where the plaintiff was injured at work and sued his employer for failing to provide adequate safeguards. The employer argued that, even had these been provided, the plaintiff had not established that they would have been utilised. The majority applied the traditional standard of proof. Murphy J dissented, objecting that this ‘puts the plaintiff to proof, not of what happened, but of what would have happened if circumstances had been different’: at 546 (citations omitted). In Public Trustee As Administrator of the Estate of the Late Peter Saroukas v Sutherland Shire Council [1992] Aust Torts Reports 81-149, the plaintiff, who was injured diving into a bay, sued the council for failing to put up adequate warning signs. Again, causation raised the hypothetical issue whether, had there been such signs, the plaintiff would have dived anyway. The plaintiff was an experienced swimmer and had some knowledge of the bay. The trial judge adverted to human autonomy in describing the plaintiff as ‘very substantially the author of his own misfortune’: at 61,142. However, he recognised the ‘possibility that a pictogram sign may have been seen, and may have influenced the conduct of the plaintiff’: at 61,138. The case was argued and decided on the basis of traditional ‘all or nothing’ causation, however the trial judge allowed for this possibility or chance by finding the defendant liable but reducing damages by 85 per cent for the plaintiff’s contributory negligence: at 61,141. The defendant’s appeal to the New South Wales Court of Appeal was successful. With Priestley and Handley JJA agreeing, Gleeson CJ pointed out that the finding on contributory negligence amounted to a finding that causation had not been established on the balance of probabilities: at 61,142. Even though the hypothetical issue concerned the plaintiff’s own conduct, Gleeson CJ suggested that a loss of chance approach may have been appropriate. His Honour left this question open, however, as the parties had agreed that the case should be dealt with on the traditional ‘all or nothing’ basis: at 61,141.

[252] [1991] HCA 54; (1991) 174 CLR 64, 119.

[253] [1989] HCA 11; (1988) 166 CLR 394, 400 (Mason CJ), 404 (Wilson, Dawson, Toohey and Gaudron JJ) (‘Nikolaou’).

[254] [1988] HCA 64; (1988) 166 CLR 351, 360 (Mason CJ), 366 (Wilson, Toohey and Gaudron JJ), 389–90 (Dawson J).

[255] See, eg, Nikolaou [1989] HCA 11; (1988) 166 CLR 394, 405; Johnson v Perez [1988] HCA 64; (1988) 166 CLR 351, 360 (Mason CJ). See also Bennett [1992] HCA 27; (1992) 176 CLR 408 (see below n 347).

[256] (Unreported, New South Wales Court of Appeal, Mason P, Priestley and Beazley JJA, 16 October 1998) <http://online.butterworths.com.au> at 11 June 1999.

[257] (1995) 38 NSWLR 168 (Priestley, Meagher and Cole JJA).

[258] Ibid 171 (Meagher JA; Cole JA agreeing).

[259] Ibid 169.

[260] [1997] Aust Torts Reports 81-449.

[261] Ibid 64,577.

[262] [1988] HCA 64; (1988) 166 CLR 351, 372. In Sellars (1994) 179 CLR 332, 353–4, the majority reviewed the inconsistent authorities without expressing a preference, while Brennan J (at 368) referred to his statement in Johnson v Perez.

[263] I find it a little surprising that judges are so quick to acknowledge ‘the unpredictable hazards of the forensic process’: Wilsher [1987] UKHL 11; [1988] AC 1074, 1092 (Lord Bridge), quoted in Wintle v Conaust (Vic) Pty Ltd [1989] VicRp 84; [1989] VR 951, 955–6 (Crockett and Gray JJ) and Luntz and Hambly, above n 87, 281. ‘Lottery’ may be the wrong word. The contingencies in litigation are clearly more numerous and complex than in a classical game of chance. Cf the text accompanying above nn 167, 184–185.

[264] Johnson v Perez [1988] HCA 64; (1988) 166 CLR 351, 371 (Brennan J).

[265] Different views were expressed in Nikolaou [1989] HCA 11; (1988) 166 CLR 394 as to the date at which damages should be assessed. One view is that the ‘appropriate date ... is the date the personal injury action should have been determined’: at 398 (Mason CJ); see also 405 (Brennan J). Wilson, Dawson, Toohey and Gaudron JJ offered some support for this view (at 404), but also said ‘it is necessary to speak of the date when the cause of action against the solicitors arose rather than the notional date of trial of the [original] action ... because the damage flowing from the negligence of the solicitors crystallized at that date’: at 403. Deane J differed considerably, holding that the trial judge should assess damages ‘as at the date of the hearing before him rather than as at the date when the action for personal injury against [the original wrongdoer] would probably have been heard if the [defendant] solicitors had acted with due diligence’: at 407. Their Honours expressed a similar range of views in the slightly earlier case, Johnson v Perez [1988] HCA 64; (1988) 166 CLR 351, 360–1 (Mason CJ), 366–7 (Wilson, Toohey and Gaudron JJ), 373 (Brennan J), 384–5 (Deane J), 389–90 (Dawson J). I have suggested in the text that there is a connection between the issue of the date at which damages should be assessed and the issue whether the recovery should be proportional or ‘all or nothing’. However, on this view Brennan J’s position on the two issues appears inconsistent, as does Deane J’s.

[266] Nikolaou [1989] HCA 11; (1988) 166 CLR 394, 398 (Mason CJ), 402 (Wilson, Dawson, Toohey and Gaudron JJ), quoting Judge Harris of the County Court of Victoria.

[267] Sellars (1994) 179 CLR 332, 359, citing Chaplin v Hicks [1911] UKLawRpKQB 104; [1911] 2 KB 786 and McRae v Commonwealth Disposals Commission [1951] HCA 79; (1951) 84 CLR 377, 412 (Dixon and Fullagar JJ). Cf Sellars (1994) 179 CLR 332, 349 (Mason CJ, Dawson, Toohey and Gaudron JJ), citing Commonwealth v Amann Aviation Pty Ltd [1991] HCA 54; (1991) 174 CLR 64: ‘And there can be no doubt that a contract to provide a commercial advantage or opportunity, if breached, enables the innocent party to bring an action for damages for the loss of that advantage or opportunity.’

[268] Sellars (1994) 179 CLR 332, 359.

[269] Ibid 364.

[270] Ibid 355 (Mason CJ, Dawson, Toohey, and Gaudron JJ). In Hotson [1988] UKHL 1; [1987] AC 750, the English Court of Appeal also rejected the distinction between recovery in contract and tort for the purposes of the loss of chance doctrine: at 760 (Lord Donaldson), 764 (Dillon LJ), 768 (Croom-Johnson LJ). The House of Lords overturned the decision on other grounds, however Lord Bridge briefly indicated that he saw ‘formidable difficulties in accepting the analogy’ between Chaplin v Hicks [1911] UKLawRpKQB 104; [1911] 2 KB 786 and Kitchen v Royal Air Force Association [1958] 1 WLR 563 and a claim for medical negligence: Hotson [1988] UKHL 1; [1987] AC 750, 782.

[271] Hotson [1988] UKHL 1; [1987] AC 750, 768. Strangely, Coote, above n 152, 771 appears comfortable with this anomaly.

[272] (1990) 169 CLR 638.

[273] (1994) 179 CLR 332.

[274] Coote, above n 152, 772. Cf Hill, above n 171, 519.

[275] Hotson [1988] UKHL 1; [1987] AC 750, 782.

[276] Even though Malec was concerned with quantum of past loss rather than the loss of chance doctrine and causation, the two areas are closely related. Waddams, above n 99, 11 describes Malec in terms of loss of chance: ‘The High Court held that the plaintiff was entitled to compensation for loss of the chance that he might have retained his earning capacity if he had not been injured.’

[277] Sellars (1994) 179 CLR 332, 350 (Mason CJ, Dawson, Toohey, Gaudron JJ), cf 366–7 (Brennan J).

[278] (1998) 156 ALR 517.

[279] (1999) 162 ALR 540.

[280] See above Part II(A)(1).

[281] See Chappel (1998) 156 ALR 517, 560 (Hayne J).

[282] See ibid 519 (Gaudron J).

[283] Freckelton gives a misleading impression of this aspect of the judgment. He suggests that Kirby J, with McHugh J, provided the ‘most substantial [analysis] of the law of causation’ and that Kirby J’s decision ‘is likely to be highly influential’. In summarising Kirby J’s analysis, Freckelton indicates that ‘[i]n some circumstances a plaintiff may recover as damages the value of a lost chance caused by the negligent act or omission’. However, Freckelton fails to note that the other judgments in Chappel were far less positive about the loss of chance action: Ian Freckelton, ‘The New Duty to Warn’ [1999] AltLawJl 4; (1999) 24 Alternative Law Journal 17, 19.

[284] (1995) 38 NSWLR 47, 56–7.

[285] Chappel (1998) 156 ALR 517, 549.

[286] Ibid 519.

[287] Ibid 539, quoting from Laferrière [1991] 1 SCR 541, 606; 78 DLR (4th) 609, 656.

[288] Chappel (1998) 156 ALR 517, 560–1.

[289] See above Part II(B)(3).

[290] Chappel (1998) 156 ALR 517, 519–20.

[291] [1998] UKHL 5; [1998] 1 All ER 481, 488. See Chappel (1998) 156 ALR 517, 534 (Gummow J), 550–1 (Kirby J), 557 (Hayne J). The High Court’s support for the ‘common sense’ formula was discussed in above nn 69–70, 87–88 and accompanying text.

[292] Naxakis (1999) 162 ALR 540, 542 (Gleeson CJ).

[293] Ibid 546 (Gaudron J), 564–5, 571 (Callinan J).

[294] Ibid 547.

[295] Ibid.

[296] Ibid 549.

[297] [1991] 1 SCR 541, 573–4; 78 DLR (4th) 609, 632.

[298] Naxakis (1999) 162 ALR 540, 549.

[299] Ibid 548.

[300] Ibid 547.

[301] Ibid.

[302] Ibid 561 (Kirby J).

[303] Ibid 549.

[304] Ibid 561.

[305] Ibid 574.

[306] Ibid 574–5.

[307] Ibid 575.

[308] Ibid.

[309] Ibid.

[310] (1990) 169 CLR 638.

[311] (1994) 179 CLR 332.

[312] See above Part II(A)(1).

[313] (1998) 156 ALR 517.

[314] (1999) 162 ALR 540.

[315] [1956] UKHL 1; [1956] AC 613.

[316] [1972] UKHL 7; [1973] 1 WLR 1.

[317] The calculations in this example are rough and ready. A more careful analysis is provided in below Part IV(B).

[318] These are not multiple unnecessary but sufficient causes, as in the two-fire cases discussed in Part II. The difficulty in the multiple exposure cases is that it is not clear that any of the sources of exposure are individually necessary or sufficient. The difficulty is not over-determination, but indeterminacy. Cf Honoré, ‘Necessary and Sufficient Conditions in Tort Law’, above n 66, 380–1.

[319] See below Part IV(B)(2).

[320] ‘“Loss of chance” has its mirror image in the proposition that exposure to risk should itself be compensable’: Fleming, The Law of Torts, above n 51, 229.

[321] See Hotson [1988] UKHL 1; [1987] AC 750, 762 (Donaldson MR), 771 (the defendant’s submission to the House of Lords); Wardley Australia Ltd v Western Australia [1992] HCA 55; (1992) 175 CLR 514, 545 (Deane J).

[322] Cf Chappel (1998) 156 ALR 517, 561 (Hayne J).

[323] See above nn 186–189.

[324] [1945] HCA 31; (1945) 71 CLR 637, 649. In Quigley v Commonwealth (1981) 35 ALR 537, 546, Murphy J (dissenting) referred to this passage, and suggested that the plaintiff’s task of establishing the elements of his case on the balance of probabilities can be facilitated by the court ‘easily inferring causation from a breach of a duty imposed to avoid the very kind of accident which occurred’. See also above n 251. Cf Sutherland Shire Council v Heyman [1985] HCA 41; (1985) 157 CLR 424, 467 (Mason J):

When there is a duty to take a precaution against damage occurring to others through the default of third parties or through accident, breach of the duty may be regarded as materially causing or materially contributing to that damage, should it occur, subject of course to the question whether performance of the duty would have averted the harm.

Cf the presumptio hominis discussed in Watts v Rake [1960] HCA 58; (1960) 108 CLR 158, 159–60 (Dixon J), 163–4 (Menzies J) and Purkess v Crittenden [1965] HCA 34; (1965) 114 CLR 164, 168 (Barwick CJ, Kitto and Taylor JJ), 171 (Windeyer J).

The plaintiff in his own case need do no more than rely upon the conclusion to be drawn from the basic facts that before the accident he was not disabled: that after it he was disabled in a way that could be a consequence of the hurt he sustained. Whether one calls such a conclusion an inference, a presumption of fact or a presumptio hominis matters not. It is an inference which any tribunal of fact would ordinarily draw until the defendant had ... provided some material to overcome it ...

Ibid 171 (Windeyer J). This latter presumption may, however, go to quantum rather than causation. See the discussion of this distinction in above Part II(B)(1).

[325] (1998) 156 ALR 517, 520 (Gaudron J), 548 (Kirby J). See also Freckelton, above n 283, 21, 29.

[326] ‘If a wrongful act or omission results in an increased risk of injury to the plaintiff and that risk eventuates, the defendant’s conduct has materially contributed to the injury that the plaintiff suffers whether or not other factors also contributed to that injury occurring’: Chappel (1998) 156 ALR 517, 525 (McHugh J). ‘Here, the injury ... occurred within an area of foreseeable risk. In the absence of evidence that the breach had no effect or that the injury would have occurred even if [the defendant] had warned her of the risk ... the breach of duty will be taken to have caused the injury’: at 535 (Gummow J) (citations omitted).

[327] Ibid 527 (McHugh J):

[T]he onus of proving that the failure to warn was causally connected with the plaintiff’s harm lies on the plaintiff. However, once the plaintiff proves that the defendant breached a duty to warn of a risk and that the risk eventuated and caused harm to the plaintiff, the plaintiff has made out a prima facie case of causal connection. An evidentiary onus then rests on the defendant to point to other evidence suggesting that no causal connection exists.

‘Once a plaintiff demonstrates that a breach of duty has occurred which is closely followed by damage, a prima facie causal connection will have been established’: at 548 (Kirby J).

[328] (1999) 162 ALR 540, 547 (Gaudron J), 561 (Kirby J), 574 (Callinan J), discussed in above Part III(C)(4).

[329] The terminology relating to presumptions and shifting burdens of proof can lead to ‘feelings of hopelessness, despair, perplexity and confusion’: J J Bray, ‘The Increasing Vulnerability of the Presumption of Legitimacy: An Historical Survey’ in Enid Campbell and Louis Waller (eds), Well and Truly Tried (1982) 10, 10. I think it is more useful to talk in terms of the strength of presumptions than of whether they are evidential, persuasive or legal. Cf Australian Law Reform Commission, Aspects of Proof, Research Paper No 14 (1983) 73. See below nn 413–415 and accompanying text. It appears clear that ‘evidential’ presumptions, which leave the legal onus unchanged, can nevertheless carry considerable strength. Consider the presumptio hominis of Watts v Rake [1960] HCA 58; (1960) 108 CLR 158 and Purkess v Crittenden [1965] HCA 34; (1965) 114 CLR 164: see above n 324.

[330] [1992] HCA 27; (1992) 176 CLR 408. In a joint judgment, Mason CJ, Deane and Toohey JJ mentioned this issue without expressing a view: see the text accompanying below n 351. Their Honours referred to Canadian authorities that have considered ‘a possible shift in the onus of proof’: at 416. The fifth judge in Bennett, McHugh J, made no mention of it. Gaudron J’s analysis in Bennett was relied on by Gummow and Kirby JJ in Chappel (see above nn 325–327) and Gaudron J in Naxakis (see above n 328). The connection between breach and causation has also been considered in industrial injury cases where the plaintiff employee alleges that the defendant employer’s breach was a failure to take some safety measure. In Quigley v Commonwealth (1981) 35 ALR 537, 541–3 Mason and Aickin JJ considered that there was a lack of causation. Even if the safety measure had been provided by the defendant, the plaintiff had failed to establish that he would have utilised it. Stephen J, however, applied similar reasoning to find that there had been no breach. If the plaintiff would not have used the safety measure had it been provided, then the defendant had committed no breach of duty in not providing the safety measure: at 539–40. See also above Part III(C)(2).

[331] Bennett [1992] HCA 27; (1992) 176 CLR 408, 422.

[332] Palmer, above n 85, 21.

[333] Bennett [1992] HCA 27; (1992) 176 CLR 408, 421.

[334] Ibid.

[335] See Kirby and Callinan JJ in Naxakis (1999) 162 ALR 540 (see above n 328); Kirby J in Chappel (1998) 156 ALR 517, 548 (see above nn 325, 327); Gaudron J in Bennett [1992] HCA 27; (1992) 176 CLR 408 (see above nn 331–334). Cf Mason CJ, Deane and Toohey JJ in Bennett (see below n 351).

[336] [1956] UKHL 1; [1956] AC 613.

[337] Ibid 622 (Lord Reid).

[338] See text accompanying above n 332.

[339] [1956] UKHL 1; [1956] AC 613, 621–2 (Lord Reid), 623 (Lord Tucker), cf 626 (Lord Keith).

[340] [1972] UKHL 7; [1973] 1 WLR 1, 5, cf 12–13 (Lord Salmon).

[341] [1987] UKHL 11; [1988] AC 1074.

[342] Ibid 1090.

[343] Ibid.

[344] See below Part IV(B)(2).

[345] See also Fitzgerald v Lane [1987] QB 781, in which the English Court of Appeal applied the McGhee reasoning to the situation of a plaintiff pedestrian who had been hit successively by two cars, and was having difficulty proving which car had caused his injuries. Note that this was prior to the House of Lords decision in Wilsher [1987] UKHL 11; [1988] AC 1074.

[346] [1992] HCA 27; (1992) 176 CLR 408.

[347] The Court considered that, had the plaintiff been informed of his rights at common law, he would have pursued them: Bennett [1992] HCA 27; (1992) 176 CLR 408, 413 (Mason CJ, Deane and Toohey JJ), 423–4 (Gaudron J), 429 (McHugh J). See above Part III(C)(3).

[348] See Palmer, above n 85, 20.

[349] Bennett [1992] HCA 27; (1992) 176 CLR 408, 413, cf 429 (McHugh J).

[350] Ibid 414 (Mason CJ, Deane and Toohey JJ), 423 (Gaudron J), 429–30 (McHugh J). It also appeared that the department was the source of the erroneous ‘independent’ advice provided by the Legal Assistance-funded barrister: at 414 (Mason CJ, Deane and Toohey JJ).

[351] Ibid 416.

[352] Ibid 422.

[353] Ibid 429.

[354] See above Part II(A)(1).

[355] (1999) 162 ALR 540.

[356] See above Part III(C)(4).

[357] See, eg, Naxakis (1999) 162 ALR 540, 544 (Gaudron J).

[358] (1985) 38 SASR 121 (‘Birkholz’).

[359] Ibid 130 (Bollen J agreeing).

[360] Ibid.

[361] See the text accompanying above nn 341–343.

[362] [1989] VicRp 84; [1989] VR 951.

[363] Limitation of Actions Act 1958 (Vic) s 23A.

[364] Crockett and Gray JJ.

[365] Wintle v Conaust (Vic) Pty Ltd [1989] VicRp 84; [1989] VR 951, 955 (Crockett and Gray JJ).

[366] Ibid 952.

[367] Ibid 963.

[368] [1999] Aust Torts Reports 81-499 (‘Baldwin’).

[369] Ibid 65,639; cf 65,622 (Beazley JA). Meagher and Beazley JJA agreed with Fitzgerald AJA generally, though Meagher JA dissented on apportionment: at 65,622. The trial judge apportioned damages 95 per cent against Baldwins and five per cent against Jsekarb. Fitzgerald AJA, while acknowledging that ‘[r]egard must be had to the “relative importance of the acts of the parties in causing the damage”’ (at 65,649, quoting Wynbergen v Hoyts Corp Pty Ltd [1997] HCA 52; (1997) 149 ALR 25, 29 (Hayne J)), indicated that ‘the evidence provides no rational basis for the division of causation [and that it is therefore] impossible to establish any apportionment is “just and equitable” other than an equal apportionment’: Baldwin [1999] Aust Torts Reports 81-499, 65,649–50, referring to Broken Hill Pty Co Ltd v Duffy (1943) 16 Australian Law Journal 374, 376. However, as Meagher JA pointed out, there were clear grounds for the trial judge’s apportionment. ‘Baldwins for many decades had exposed [the plaintiff] to the ... more deadly asbestos, whilst Jsekarb had at worst exposed [the plaintiff] to the ... milder ... form ... for a period of no more than 10–15 years’: Baldwin [1999] Aust Torts Reports 81-499, 65,622.

[370] Baldwin [1999] Aust Torts Reports 81-499, 65,642 (Fitzgerald AJA).

[371] (1998) 156 ALR 517.

[372] Baldwin [1999] Aust Torts Reports 81-499, 65,632–6. See above nn 325–327.

[373] (1997) 42 NSWLR 307 (‘Bendix’).

[374] Ibid 319 (Mason P).

[375] Ibid 311.

[376] Ibid 320 (Mason P), 339–44 (Beazley JA).

[377] See, eg, ibid 320 (Mason P), 325 (Beazley JA).

[378] Ibid 339 (Beazley JA). Cf Wallaby Grip (BAE) Pty Ltd (in liq) v Macleay Area Health Service (Unreported, Supreme Court of New South Wales, Court of Appeal, Sheller, Beazley JJA and Sheppard AJA, 18 December 1998) <http://online.butterworths.com.au> at 11 June 1999, 7–9 (Beazley JA; Sheller JA and Sheppard AJA agreeing). Cf ICI Australia Operations Pty Ltd v Walsh [1997] Aust Torts Reports 81-452, 64,596, 64,601 (Mason P; Handley JA agreeing).

[379] Bendix (1997) 42 NSWLR 307, 316, quoting A Grubb, ‘Causation and Medical Negligence’ [1988] Cambridge Law Journal 350, 350–1.

[380] [1992] HCA 27; (1992) 176 CLR 408. See above nn 329–334.

[381] Bendix (1997) 42 NSWLR 307, 315–16. See above nn 341–343.

[382] (1985) 38 SASR 121. See above n 360.

[383] [1972] UKHL 7; [1973] 1 WLR 1. See above n 340.

[384] Bendix (1997) 42 NSWLR 307, 348. Cf ICI Australia Operations Pty Ltd v Walsh [1997] Aust Torts Reports 81-452, 64,603 (Meagher JA dissenting).

[385] Bendix (1997) 42 NSWLR 307, 349.

[386] [1996] NSWSC 657; (1996) 41 NSWLR 389 (CA). See also above n 145 and accompanying text.

[387] [1956] UKHL 1; [1956] AC 613.

[388] Commonwealth v McLean [1996] NSWSC 657; (1996) 41 NSWLR 389, 408 (Sperling J).

[389] Ibid 411 (Handley and Beazley JJA), 412 (Santow AJA agreeing).

[390] Ibid 409 (Handley and Beazley JJA).

[391] Ibid.

[392] Ibid 411.

[393] Ibid 410.

[394] McGhee [1972] UKHL 7; [1973] 1 WLR 1, 6.

[395] Chappel (1998) 156 ALR 517, 548. The majority in Birkholz (1985) 38 SASR 121, 132 also drew upon Lord Wilberforce’s judgment in McGhee and said:

[H]owever precisely the disease was contracted, the substantial increase in the risk of [the plaintiff’s] contracting it resulting from the [defendant’s] failure to discharge its legal obligations, should be regarded in a realistic, practical and legal sense, if not in a strictly logical sense, as materially contributing to the contracting of the disease.

See also Bendix (1997) 42 NSWLR 307, 345 (Stein JA dissenting).

[396] [1987] UKHL 11; [1988] AC 1074, 1087. See above nn 341–343 and accompanying text.

[397] Heydon, above n 1, 209, quoting from J R Gulson, The Philosophy of Proof in Its Relation to the English Law of Judicial Evidence (2nd ed, 1923) 371. Cf Andrew Ligertwood, Australian Evidence (3rd ed, 1998) 391–2.

[398] Ligertwood, above n 397, 392.

[399] Ibid 316. See discussion in David Byrne and John Heydon, Cross on Evidence (4th Aust ed, 1991) 234–8.

[400] Eg, Children (Equality of Status) Act 1976 (NSW) s 10(1), referred to by Byrne and Heydon, above n 399, 239–40. This Act was repealed by s 37 of the Status of Children Act 1996 (NSW). The latter Act makes the presumption of legitimacy irrebuttable in situations where children are born following fertilisation procedures, and where a court has made orders to that effect, but not otherwise. See ss 12(1), 14(1), 14(3) and 15.

[401] Eric Landowski, ‘Truth and Veridiction in Law’ (1989) 2 International Journal for the Semiotics of Law 29, 35–6. Similarly, Byrne and Heydon, above n 399, 226 note that the law ‘sometimes prohibits all rational investigation into the existence of a fact’. Adeney, above n 85, 48 suggests that the presumption of causation from breach is something of a legal fiction:

Whether the creation of legal fictions in common law is desirable is questionable. They tend to arise where the law as it exists is seen to be out of step with conventional notions of justice. Thus, in this context, they are the creatures of policy and are interlopers in a system whose fundamental principle is that of reason or logic.

There is some truth to what she says. Perhaps the courts should be open in addressing the policy underlying illogical principles. On the other hand, ambiguity has some value in providing the courts with flexibility.

[402] See above n 329.

[403] Scott v The London & St Katherine Docks Co [1865] EngR 220; (1865) 3 H & C 596; 159 ER 665, 667 (Erle CJ).

[404] Government Insurance Office of NSW v Fredrichberg [1968] HCA 54; (1968) 118 CLR 403, 413 (Barwick CJ; Kitto J agreeing).

[405] Fleming, The Law of Torts, above n 51, 354.

[406] Ibid.

[407] Ibid 539–40. Another example where it is unclear whether policy or logic is the motivating force is the presumption of advancement in property law. Contrast the judgments of Gibbs CJ and Murphy J in Calverley v Green [1984] HCA 81; (1984) 155 CLR 242, 245–53 (Gibbs CJ), 264–5 (Murphy J).

[408] Neil Cohen, ‘Confidence in Probability: Burdens of Persuasion in a World of Imperfect Knowledge’ (1985) 60 New York University Law Review 385, 418.

[409] See, eg, Davies v Taylor [1974] AC 207, 219 (Lord Simon). See generally Hamer, above n 45, 509.

[410] See Neil MacCormick, Legal Reasoning and Legal Theory (1978) 45; C R Williams, ‘Placing the Burden of Proof’ in Enid Campbell and Louis Waller (eds), Well and Truly Tried (1982) 271.

[411] Some commentators suggest that the evident importance of presumptions and the onus of proof is an indication that proof should not be measured by a simple probability value: eg, Neil Cohen, above n 408. This argument arose in the ‘naked statistical evidence’ debates. See Hamer, above n 45, and references therein.

[412] See also Glenn Shafer, ‘The Construction of Probability Arguments’ (1986) 66 Boston University Law Review 799.

[413] ‘Since the standards of proof are variable depending on the nature of the case, the weight of evidence required to satisfy the standard allocated by the presumption ... will vary’: ALRC, above n 329, 67 (citations omitted).

[414] Fleming, The Law of Torts, above n 51, 226 suggests that by the presumption of causation the bar is ‘raised’ for plaintiffs, but this must be an editorial error.

[415] Where the plaintiff’s allegations have moral overtones, as in fraud or adultery cases, the civil standard may be higher than where mere negligence or breach of contract is alleged. The defendant may have the benefit of something in the nature of the presumption of innocence: Briginshaw v Briginshaw [1938] HCA 34; (1938) 60 CLR 336, 363 (Dixon J). Cf the Evidence Act 1995 (Cth) s 140(1); Hamer, above n 45, 512–14. In Neat Holdings Pty Ltd v Karajan Holdings Pty Ltd [1992] HCA 66; (1992) 110 ALR 449, a majority of the High Court indicated that a ‘balanced’ standard was appropriate even though fraud allegations were involved. However, that may reflect the fact that each party was alleging fraud of the other. The majority indicated that ‘the trial judge should be conscious of the gravity of allegations on both sides’: at 451 (Mason CJ, Brennan, Deane and Gaudron JJ; Toohey J dissenting).

[416] Often in multiple exposure cases, the innocent exposure is background exposure. In this case, it will be hard to eliminate the innocent exposure to arrive at a figure for P(D|G). At best, figures will be available for P(D|I) and P(D|I&G). The probabilistic ‘but for’ test for these figures is provided in below Part IV(B)(3).

[417] See the text accompanying above nn 317–319, 394.

[418] [1956] UKHL 1; [1956] AC 613, 622 (Lord Reid).

[419] [1988] UKHL 1; [1987] AC 750, 765.

[420] [1972] UKHL 7; [1973] 1 WLR 1.

[421] The facts were slightly different from those in Bonnington Castings. In McGhee, the innocent and guilty exposures were successive in time. The defendant was not negligent in failing to prevent the plaintiff from being exposed to the dust. However, the defendant was negligent in failing to provide showers. As a consequence, the dust stayed on the plaintiff’s skin longer than it should have. The guilty exposure was this extended period of exposure. See the discussion in Wilsher [1987] UKHL 11; [1988] AC 1074, 1087–8.

[422] McGhee [1972] UKHL 7; [1973] 1 WLR 1, 3–4. See also Adeney, above n 85, 53.

[423] Wilsher v Essex Area Health Authority [1987] QB 730, 752; Hotson [1988] UKHL 1; [1987] AC 750, 773; Adeney, above n 85, 36; Fleming, The Law of Torts, above n 51, 226.

[424] [1988] 1 All ER 659, 665 (emphasis added).

[425] As Fitzgerald AJA held in Baldwin [1999] Aust Torts Reports 81-499, 65,642 accepting the cumulative theory, the asbestos fibres ‘from Jsekarb brake blocks inhaled in combination with other [fibres] probably materially contributed to [the plaintiff’s] mesothelioma although [the exposure from this source] would itself probably not have caused mesothelioma in the absence of inhalation of other ... asbestos fibre’.

[426] [1956] UKHL 1; [1956] AC 613.

[427] [1972] UKHL 7; [1973] 1 WLR 1.

[428] [1987] UKHL 11; [1988] AC 1074.

[429] Ibid 1081 (Lord Bridge).

[430] See above nn 341–343 and accompanying text.

[431] [1987] UKHL 11; [1988] AC 1074, 1090–1 (HL); Wilsher v Essex Health Authority [1987] QB 730, 779 (CA).

[432] (1985) 38 SASR 121.

[433] See text accompanying above n 360.

[434] Matheson J quoted from the testimony of Dr Philpot (not very well reported): ‘[R]epeated small exposure is more likely to immunise the person better rather than to give them more disease’: Birkholz (1985) 38 SASR 121, 137.

[435] Ibid 142.

[436] Ibid 139, 141–2.

[437] Ibid 142–3.

[438] Luntz, above n 3, 85; Mary Dant, ‘Gambling on the Truth: The Use of Purely Statistical Evidence as a Basis for Civil Liability’ (1988) 22 Columbia Journal of Law and Social Problems 31, 62. Cf Adeney, above n 85, 50; Hotson [1988] UKHL 1; [1987] AC 750, 786 (Lord Mackay).

[439] [1987] UKHL 11; [1988] AC 1074.

[440] See above nn 341–343, 431 and accompanying text.

[441] [1992] HCA 27; (1992) 176 CLR 408.

[442] (1998) 156 ALR 517.

[443] (1999) 162 ALR 540. See the text accompanying above nn 324–334.

[444] See above Parts II(B)(2) and II(B)(3).

[445] See above Part II(B)(3).

[446] See the text accompanying above nn 153–155.

[447] See above Part III(A).

[448] See the text accompanying above nn 219–226.

[449] See above Parts III(C)(4) and IV.

[450] John Fleming, ‘Probabilistic Causation in Tort Law’ (1989) 68 Canadian Bar Review 661, 669.

[451] Lodge Holes Colliery Co Ltd v Wednesday Corp [1908] UKLawRpAC 40; [1908] AC 323, 325, quoted by Waddams, above n 99, 10. See also Fleming, The Law of Torts, above n 51, 221, who notes that ‘resolution of causal uncertainty is apt to be influenced by the strength of disapproval of the defendant’s fault’.