1. [2011] UKSC 10.

2. Fairchild v Glenhaven Funeral Services Ltd; Fox v Spousal (Midlands) Ltd; Matthews v Associated Portland Cement Manufacturers (1978) Ltd and others[2002] UKHL 22.

3. Barker v Corus (UK) plc (formerly Saint Gobain Pipelines plc); Murray v British Shipbuilders (Hydrodynamics) Ltd; Patterson v Smiths Dock Ltd and others[2006] UKHL 20.

4. Lord Brown, above n 1, para 174.

5. Bailey, SH ‘Causation in negligence: what is a material contribution?’ (2010) 30 LS 167 Google Scholar.

6. [2009] EWCA Civ 1159.

7. Willmore v Knowsley Metropolitan Borough Council[2011] UKSC 10.

8. This suspicion usually pleads in aid that most hoary of old chestnuts, the paradox of the ‘Blue Bus Company’– and Sienkiewicz was no exception, see text, below n 85.

9. Above n 5, at 184.

10. Hart, Hla and Honoré, A Causation in the Law (Oxford: Clarendon, 1985)CrossRefGoogle Scholar.

11. Wright, RW ‘Causation in tort law’ (1985) 73 CLR 1735 CrossRefGoogle Scholar. For an introduction to the NESS test, which takes issue with some of its critics as well as, occasionally, Wright himself, see Miller, C ‘NESS for beginners’ in Goldberg, R (ed) Perspectives on Causation (Oxford: Hart Publishing, 2011)Google Scholar.

12. This paper makes no reference to liability for ‘pure risk’, that is, near misses which, by good fortune, do not result in injury. Although advocated by some academics, this proposal represents a digression in an area which is already difficult enough.

13. [1956] 1 All ER 615.

14. Above n 5, at 184.

15. Per Lord Keith, above n 13, at 627.

16. Ibid, at 626.

17. Most of the criticism of NESS tends to be focused on its treatment of a more complex form of over-determination, viz pre-emption; see for instance Fumerton, R and Kress, K ‘Causation and the law: preemption, lawful sufficiency and causal sufficiency’ (2001) 64 Law & Contemporary Problems 83 CrossRefGoogle Scholar and also D Fischer ‘Causation in fact in omission cases’[1992] Utah L Rev 1335. For a demonstration of the ability of NESS to counter criticisms of its treatment of pre-emption without inventing what R Wright calls ‘causal priority’, see C Miller, above n 11.

18. A reviewer of an earlier draft has suggested that a better illustration of Bonnington over-determination would consist of one conspirator severing a major artery while another simply opens a small vein. I gratefully accept that suggestion.

19. Above n 2.

20. Above n 3.

21. [1973] 1 WLR 1.

22. Ibid, at 4.

23. Ibid.

24. Ibid.

25. Ibid.

26. If McGhee creates a genus of ‘material contribution’, Bonnington turns out to be a rather primitive species. In fact, the occupational deafness case of Thompson v Smiths Shiprepairers (North Shields), below n 42, is a more interesting exemplar because, although deterministic, its bare skeleton closely mirrors McGhee's: a period of non-tortious exposure to an occupational agent (noise in this instance) of injury is followed by exposure to a tortious source of the same agent, which would have been less harmful had control measures not been negligently absent.

27. Above n 21, at 5.

28. Why a premium bond rather than a lottery ticket? Because a premium bond retains its capacity to ‘win’ a prize from one draw to the next. An asbestos fibre once absorbed into the body similarly retains, over time, its mysterious role in the mutations that culminate in mesothelioma.

29. Per Lord Hoffman in Barker, above n 3, para 13, where he claimed that in determining McGhee, the House had treated it ‘as an application avant la lettre of the Fairchild exception’. Given its greater generality as well as its seniority, McGhee arguably has a better claim to be eponymous.

30. Perhaps the most insightful commentary is to be found in Stapleton, J ‘Lords a leaping evidentiary gaps’ (2002) 10 Torts LJ 276 Google Scholar.

31. Above n 2, paras 1–35.

32. Lord Hoffmann's speech in Barker is discussed below, see text at n 62.

33. Above n 2, para 2.

34. [1988] 1 All ER 871.

35. Above n 2, para 22.

36. Above n 34.

37. Above n 3, para 24, his position, on the ‘single agent’ criterion, now seems indistinguishable from that which Lord Rodger formulated in Fairchild, viz a ‘substantially similar’ criterion or one in which risk is increased ‘if not by exactly the same agency [. . .] at least by an agency that operated in substantially the same way’; above n 2, para 170.

38. He may well be correct in assuming that the mechanisms by which asbestos and tobacco smoke lead to lung cancer are dissimilar but they could have a multiplicative or synergistic effect, as Lord Phillips suggested in Sienkiewicz, above n 1, para 75. This example of synergy was judicially considered in Amaca Pty Ltd v Ellis[2009] HCA Trans 77.

39. [2007] EWCA Civ 1261.

40. Ibid, para 74. It is not clear whether a 30% reduction of damages was applied in Novartis, as perhaps it should have been following Barker and, as a cancer other than mesothelioma, falling outside the remit of the Compensation Act 2006.

41. Holtby v Brigham & Cowan (Hull) Ltd[2000] 3 All ER 421.

42. Thompson v Smiths Shiprepairers (North Shields)[1984] 1 QB 405.

43. Allen v British Rail Engineering Ltd[2001] EWCA Civ 242.

44. Bailey v Ministry of Defence[2008] EWCA Civ 883.

45. Ibid, para 47.

46. Above n 41.

47. Above n 42.

48. Hatton v Sutherland[2002] 2 All ER 1, at 20.

49. Re British Coal Respiratory Disease Litigation QBD 23/01/98 unreported.

50. Ibid, para 97.

51. Ibid.

52. See below n 81.

53. One of the many merits of Lord Reid's ‘risk increased’ formulation of the rule in McGhee (see text at n 27) is that it is applicable whether the uncertainty, which gives rise to that increase, is ‘epistemic’, that is, derives from our incomplete understanding of the processes causing dermatitis, or whether that uncertainty is ‘objective’, that is, it derives from an inherent randomness in the processes themselves; for a further discussion of the different approaches to probability, see Mellor, DH The Matter of Chance (Cambridge: Cambridge University Press, 1971)Google Scholar. A protracted philosophical debate on the differences might have obviated Lord Reid's pragmatic solution.

54. Above n 39.

55. Above n 1, at para 59.

56. Ibid.

57. Above n 6, para 9. I have referred to these numerical estimates as derived from ‘modelling’ rather than epidemiological studies. This view is supported by Lord Mance in the Supreme Court who wrote, rather disparagingly: ‘[t]he “epidemiological evidence” which was adduced consisted of a series of assumptions and speculations rather than actual data which could be related to the experience of those who developed mesothelioma. What the testimony amounted to was the promotion of a theory rather than the establishment of facts and it did not constitute evidence on which reliable conclusions could be reached,’ above n 1, para 205.

58. Above n 6, para 34.

59. Compensation Act 2006 (ch 29), s 3 of which restored, to mesothelioma victims, the joint and several liability which Fairchild had imposed but which, in Barker, had been replaced by damages factored by the estimate of each defendant's contribution to the increased risk.

60. Above n 6, para 55.

61. Above n 2, para 61.

62. Above n 3, para 17.

63. Above n 1, para 160.

64. This is the question posed by S Steel in his comment on Sienkiewicz in the Court of Appeal, ‘Uncertainty over causal uncertainty: Karen Sienkiewicz (Administratrix of the Estate of Enid Costello Deceased) v Greif (UK) Ltd’ 73 (2010) MLR 631.

65. This is not to assume, or to argue for, a reversal of the burden of proof. It simply recognises that, since the claimant's burden is considerably eased when the exception applies, the defendant has every reason to question its application.

66. Imagine that, in the future, the science of dermatology is so advanced that the cause of the type of dermatitis suffered by James McGhee can be determined with a reliability comparable to that of DNA testing today. At what point between then and now (or rather, 1972) can a ‘rock of uncertainty’ be said no longer to exist in regard to dermatitis and occupational exposure to brick dust? It would be the point at which the fact-finder accepts on BOP the claimant's (defendant's) causal account of the injury which holds that the delay in washing was (was not) a necessary element. By relying on BOP, the demise of the exception is signalled by a familiar legal criterion rather than a scientific one.

67. Above n 1, para 91.

68. Ibid, para 162. It is not entirely clear whether or not Smith LJ was confining her comment to cases in which the exception applied. But Lord Phillips, ibid, para 121, believed that she had ‘made a very general statement about the approach which courts should adopt to issues of causation’ and Lord Rodger referenced this paragraph when stating his opposition.

69. Ibid, para 170.

70. Ibid, para 164.

71. Gold, S ‘Causation in toxic torts: burdens of proof, standards of persuasion, and statistical evidence’ (1986) 96 Yale LJ 376 CrossRefGoogle Scholar.

72. Above n 1, para 156.

73. Ibid, para 163.

74. Hill, A Bradford ‘the environment and disease: association or causation?’ (1965) 58 Proc R Soc Med 295 Google ScholarPubMed.

75. Above n 1, para 160.

76. In any event, they would be of little help in this case where the risk estimates came, not so much from an epidemiological study published in a peer-reviewed journal, but from ad hoc calculations; see the comments of Lord Mance, above n 56.

77. Above n 1, para 179.

78. [1987] AC 750.

79. Per Lord Rodger in Barker, above n 3, para 64 and in Gregg v Scott[2005] 2 AC 176 passim. However, Lord Rodger in Sienkiewicz observes that ‘there is [following Fairchild] no room, however, for Lord Mackay's rule in cases of that kind in English or Scots law’. Above n 1, para 159.

80. See J Stapleton ‘The gist of negligence’[1988] LQR 389; Miller, C ‘Liability for negligently increased risk: the repercussions of Barker v Corus Uk (plc)’ (2009) 8 Law, Probability and Risk 39 CrossRefGoogle Scholar.

81. I have borrowed the term ‘control’ from Stapleton, above n 30. She sees it as one of six characteristics shared by McGhee and Fairchild. Initially, and when referring to Lord Bingham's fifth condition (see text, above n 32), she describes the shared factor as ‘all relevant exposures to risk occurred in the victims' workplace’, above n 30, at 285. When later explaining why Wilsher lay outside the exception, she writes that it was because ‘not all sources of risk were under the control of the defendant’, ibid, at 287. This latter construction must surely be preferable if the Fairchild exception is not to be confined to occupational injury. It is possible to argue that, in Barker, the condition of ‘all sources of risk under the control of the defendants’ was not necessarily infringed: Mr Barker, when self-employed, was equivalent to any of the other (defendant) employers in that this exposure (now non-negligent but not environmental) was no less under his control.

82. This point is developed in more detail in a discussion of XYZ v Schering Health Care Ltd[2002] EWHC 1420 (QB) in Miller, C ‘Causation in personal injury: legal or epidemiological common sense?’ (2006) 26 LS 544 Google Scholar.

83. Above n 1, para 98.

84. Ibid, para 221.

85. Smith v Rapid Transit, Inc 317 Mass 469; 58 N.E.2d 754 (1945).

86. See for example, Tillers, P ‘If wishes were horses: discursive comments on attempts to prevent individuals from being unfairly burdened by their reference classes’ (2005) 4 Law, Probability and Risk 33 at 48 CrossRefGoogle Scholar.

87. Beever, A Rediscovering the Law of Negligence (Oxford: Hart Publishing, 2007) p 477 Google Scholar.

88. If there are n mutually exclusive possibilities and we have no further information about them, then we should assign an equal probability (1/n) to each of them.

89. If there are m sources each with a probability of 1/2 of being a ‘hit’ and 1/2 being a ‘miss’, there are 2^m combinations of which any one source (A) will feature as a ‘hit’ in 2^m−1. We know ex post the diagnosis of the tumour that the ‘total miss’ possibility did not arise; hence the probability of A having been involved is: P(A)=2^m-1/(2^m-1)>1/2 for all m≥1

90. According to I Hacking, the indifference principle can be traced back to Leibniz in the seventeenth century; see The Emergence of Probability (Cambridge: Cambridge University Press, 1975) p 122. The principle remains a source of dispute among philosophers.

91. Fischer, above n 17, at 1347.

92. Above n 1, para 186.

93. In Margereson v J W Roberts[1996] PIQR P154, the claimants were able to secure damages for mesothelioma against a local asbestos employer, for whom they had never worked. Their exposure was ‘environmental’ in the sense that they had lived for some time, and played as children, near the factory where control over asbestos discharge to the immediate locality was minimal. The defendant's subsequent appeal – that there was no breach of duty because mesothelioma had only recently become a foreseeable consequence of asbestos – was readily dismissed. However, the particular location, Armley near Leeds, appears to have been exceptional; see Tweedale, G Magic Mineral to Killer Dust: Turner & Newall and the Asbestos Hazard (Oxford: OUP, 2000)Google Scholar. It is now hard to imagine an industrial defendant like Greif (UK) not contesting their responsibility for all environmental asbestos exposure in their vicinity – even though, following the Compensation Act 2006, this will not assist them in mesothelioma litigation.

94. See above n 66.

95. This is essentially the point made by Lord Dyson, above n 1, para 208.

96. Health and Safety Executive Occupational, Domestic and Environmental Mesothelioma Risks in Britain: Research Report 696 (London: HSE Books, 2009).

97. Ibid.

98. Greenland, S ‘Relation of probability of causation to relative risk and doubling dose: a methodologic error that has become a social problem’ (1999) 89 (8) American Journal of Public Health 1166 CrossRefGoogle Scholar.

99. For a more formal derivation of this inequality, see A Philip Dawid ‘The role of scientific and statistical evidence in assessing causality’ in Goldberg, above n 11, p 143.

100. Above n 1, para 75.

101. See, for example, Siemiatycki, J and Thomas, D ‘Biological models and statistical interactions: an example from multistage carcinogenesis’ (1981) 10 (4) International Journal of Epidemiology 383 CrossRefGoogle Scholar.

* D Lewis Philosophical Papers Volume II (Oxford, 1986) pp 83–113.

103. Above n 1, para 149.

104. Ibid, para 140.

105. Ibid, para 181.