1 Bonnington Castings v Wardlaw[1956] AC 613.

2 New York: Baker, Voorhis & Co; London: Stevens and Haynes, 3rd edn, 1874. This was a book on the common law of negligence, published in the USA and the UK, and citing authorities from both countries.

3 At pp 9–10. In Gordon, WE and Griffith, WH Addison's Treatise on the Law of Torts (London: Stevens and Sons, 8th edn, 1906)Google Scholar, there are discussions of remoteness of damage and contributory negligence (pp 51–60; 772–778) but no separate discussion of causation in fact; the but-for test is mentioned as part of the latter (p 772). Clerk, JF and Lindsell, Whb The Law of Torts (Wyatt Paine, 3rd edn, 1904) pp 130–138, 468–484Google Scholar, includes at pp 468–471 cases that establish that ‘Plaintiff must show that defendant's negligence was the cause of the damage’, including Metropolitan Railway Co v Jackson (1877) 3 App Cas 198 (the House held that the defendant's negligence in allowing overcrowding in a railway carriage at one station was not a cause of an accident where the plaintiff's thumb was jammed in the door at a later station; whether this was an application of the but-for test (see Lord O'Hagen at 206) or a ruling on remoteness (see Lord Blackburn at 209–210) was unclear.

4 At pp 30–31.

5 At p 30, fn 4.

6 At paras 595–596.

7 The developments in legal analysis whereby a clear distinction came to be drawn between causation in fact and causation in law are considered in Hart, Hla and Honoré, A Causation in the Law (Oxford: Clarendon Press, 2nd edn, 1985) ch IV.CrossRefGoogle Scholar

8 Except to counter a defence of contributory negligence at common law: see Wakelin v London and South Western Railway Co[1896] 1 QB 189, per Brett MR at 190. Standard illustrations of the but-for test are McWilliams v Sir William Arrol[1962] 1 WLR 295 (HL), Barnett v Kensington and Chelsea Hospital Management Committee[1969] 1 QB 428 and Robinson v The Post Office[1974] 1 WLR 1176 (CA). Where an injury could have been caused by either tortious Risk A or non-tortious Risk B, but-for causation is established in respect of Risk A if it is shown that it is more than doubled the risk that arises from Risk B: see the discussion in Sienkiewicz v Greif (UK) Ltd[2009] EWCA Civ 1159 (holding that a claimant that can bring a case within the Fairchild exception (see nn 55 and 56 below) is entitled to prove causation by showing a material increase in risk and is not required where quantification of the increase in risk can be made to establish causation by showing that the risk has been more than doubled).

9 Cf Glanville Williams's unqualified statement in 1951: ‘If a fact is not the scientific cause of an event, it cannot be the proximate cause’: Joint Torts and Contributory Negligence (London: Stevens & Sons Ltd, 1951) p 240.

10 See Corey v Havener (1902) 182 Mass. 250 (concurrent sufficient causes); Baker v Willoughby[1970] AC 467 (successive sufficient causes); Kuwait Airways Corp v Iraqi Airways Co (Nos 4 and 5)[2002] UKHL 19, [2002] 2 AC 883 per Lord Nicholls at paras [73] and [74]; and see generally on multiple sufficient causes (termed by the authors ‘additional causes’), Hart and Honoré, above n 7, pp 235–249. These cases will not be analysed further in this article.

11 [1973] 1 WLR 1. See Lord Hope of Craighead ‘James McGhee – a second Mrs Donoghue’[2003] CLJ 587. Lord Hope was junior counsel for the defenders.

12 [2002] UKHL 22, [2003] 1 AC 32.

13 See Anderson v Minneapoplis, St P & S St M Ry Co (1920) 146 Minn 430, 179 NW 45.

14 See Restatement of the Law (Third) of Torts: Liability for Physical Harm (Proposed Final Draft, April 6, 2005); J Stapleton ‘Legal cause: cause-in-fact and the scope of liability for consequences’ (2001) 54 Vanderbilt LR 941.

15 [1956] AC 613.

16 1955 SC 320.

17 Ibid.

18 Lord Carmont at 342.

19 At 339.

20 See at 616–618.

21 (1886) 12 App Cas 41 at 47.

22 See counsel at 617 and 618: ‘The onus was on [the pursuer] to show that more dust came from the swing grinders…On the facts the true cause of the disease was the dust from the hammer’.

23 Counsel at 617.

24 Ibid.

25 See above.

26 Stimpson v Standard Telephones and Cables Ltd[1940] 1 KB 342.

27 Vyner v Waldenberg Brothers Ltd[1946] KB 50

28 At 619.

29 At 620.

30 At 621.

31 At 622.

32 At 623 (emphasis added). Compare Lord Reid's comments about Bonnington Castings in McGhee, below n 43.

33 At 623.

34 At 618 and 627.

35 At 626.

36 Ibid. Cf Lord Reid's statement (n 32 above) that the tortious dust ‘did help to produce the disease’.

37 Eg X carefully installs a gas fire but it develops a leak; D negligently tries to discover the source of the leak by using a lighted match. Cf Burrows v March Gas Co (1872) LR 7 Ex 96 (the same facts except X was in breach of a contractual duty).

38 1955 SLT 225 at 228.

39 Applying the principle subsequently recognised in Jobling v Associated Dairies[1982] AC 794.

40 The addition of the words at the end of the dictum ‘and might not have developed it at all’ does not seem to add anything of significance. If the pursuer would probably (but not certainly) have developed the disease anyway at a later date from the non-tortious dust then that leaves open as a possibility that he would have never ever have developed the disease without the tortious dust. But Lord Keith does not seem to be suggesting that it is sufficient for a claimant to establish that the defendant possibly caused the damage.

41 [1973] 1 WLR 1. It is also consistent with the approach of the Court of Appeal in Page v Smith (No2)[1996] 1 WLR 855, where Bingham LJ stated that the correct test for determining whether the claimant's symptoms (ME) were caused by the defendant's negligence was ‘whether the accident did in fact have that result’ (at 858). The trial judge's formulation that it was sufficient for the claimant to show that the accident ‘did…cause or materially contribute or materially increase the risk of’ the development of ME was held to be wrong insofar as it included the words ‘or materially increase the risk of’ (ibid). On the facts, the judge had been entitled to find causation established on the correct test, having accepted medical evidence that the claimant's recovery (from previous ME) ‘would probably have continued but for the accident’ (at 858). There is again no suggestion of any middle position between application of the but-for test and a material increase of risk test.

42 As interpreted by the House of Lords in Fairchild v Glenhaven Funeral Services Ltd[2002] UKHL 22, [2003] 1 AC 32 and Barker v Corus UK Ltd[2006] UKHL 20, [2006] AC 572.

43 At 4. Lord Wilberforce expressed a similar view at 6–7. Lords Kilbrandon and Salmon did not analyse Bonnington Castings. In Stapley v Gypsum Mines Ltd[1953] AC 663, Lord Reid at 681 clearly regarded but-for causation as a standard requirement in establishing cause. In his analysis of McGhee (n 11 above), Lord Hope contrasts the orthodox test, for him illustrated by Bonnington Castings, that the claimant must show that the defendant's negligence was a necessary, albeit not the sole cause of the damage (at 596–597), with the novel principle established by McGhee that in some cases it is sufficient to show that the defendant's negligence materially increased the risk of harm (at 599–600). There was no suggestion of any third principle lying somewhere between these two.

44 This point is noted as an objection to the outcome of Bonnington Castings by Stapleton, Jane in ‘The gist of negligence part Ii’ (1988) 104 LQR 389 at 404–405.Google Scholar

45 A number of textbooks express this view: Dugdale, AM and Jones, MA Clerk and Lindsell on Torts (London: Sweet & Maxwell, 19th edn, 2006) para 2-26Google Scholar; Todd, S The Law of Torts in New Zealand (Wellington: Brookers, 5th edn, 2009) para 20-2-02Google Scholar. This seems to be based on the fact that full damages were awarded.

46 [1973] 1 WLR 1 at 8.

47 A similar objection applies to the interpretation of Bonnington Castings by Khoury, Lara Uncertain Causation in Medical Liability (Oxford: Hart Publishing, 2007) pp 21–22 Google Scholar, as an exception to the but-for test; Khoury states that it was not possible to conclude on the evidence that but-for the guilty dust, W would have suffered injury. It is submitted that is not right; on the evidence the tortious dust was clearly a but-for cause of the onset of the pneumoconiosis. This will generally be the case where similar tortious and non-tortious causes operate cumulatively.

48 [2006] UKHL 20, [2006] AC 572 at [72]. See also Bailey v Ministry of Defence, below n 75.

49 At [127]–[133].

50 (1859) 1 E & E 385.

51 At 392.

52 (1886) 12 App Cas 41.

53 See RB Stevens Law and Politics (University of North Carolina Press, 1979) pp 341–354: ‘During the time that the House was dominated by Simonds, its purpose, intellectually and practically, remained the preservation of the status quo’ (p 342). Stevens does not mention Bonnington Castings.

54 Eg a road accident caused entirely by the negligence of one driver.

55 See McGhee and Fairchild.

56 See Fairchild; Barker; Clough v First Choice Holidays[2006] EWCA Civ 15, [2006] PIQR 325.

57 See Lord Bingham in Fairchild at paras [21] and [35].

58 See below, section 4.

59 [1957] 1 WLR 613.

60 See Viscount Simonds at 615.

61 Ibid. Lords Oaksey and Morton of Henryton agreed with Viscount Simonds.

62 Ibid at 618.

63 At 620–624.

64 At 627.

65 This reflects the basic point that ‘In tort causal questions are usually relevant to both the existence of liability and to its extent’: Hart and Honoré, above n 7, p 85.

66 [1984] 1 QB 405.

67 At 437–438.

68 [1984] 1 QB 405 at 441. It is submitted that this interpretation of Bonnington Castings overstates the significance of that decision, for the reasons given above.

69 [2000] 3 All ER 421.

70 Tort Law (Oxford: Oxford University Press, 2002) pp 69–70. Weir has accepted that where indivisible harm is incrementally caused ‘this novel doctrine may possibly be entertained, on the view that a defendant should not be liable for already existing harm to which he did not contribute’:[2001] CLJ 237 at 239.

71 C McCaul ‘Holtby and the end game’[2006] Journal of Personal Injury Law 1 at 6–11.

72 This issue can cause difficulty, there being an increasing willingness to hold injuries to be divisible: see, eg, Rogers, Wvh Winfield & Jolowicz on Tort (London: Sweet & Maxwell, 17th edn, 2006) p 923 Google Scholar, citing Rahman v Arearose Ltd[2001] QB 351 and Hatton v Sutherland[2002] EWCA Civ 76, [2002] ICR 613 at [41]. In Dickens v O2 plc[2008] EWCA Civ 1144, [2009] IRLR 58, Smith LJ at [43]–[46] doubted a view attributed to Hale LJ in Hatton that apportionment was possible even in respect of an indivisible injury; however, Hale LJ's summary of the principles in Hatton at [43] confines apportionment to cases of divisible injuries. Tony Weir criticised the decisions in Rahman and Holtby at [2001] CLJ 237, the former on the ground that the court had distinguished between different aspects of the claimant's overall psychiatric condition and their respective causes; there was no scientific basis for this: ‘his mania is aetiologically indiscerptible as when grief and shock combine to wreck the life of a parent who witnesses the death of her children’ (at 238).

73 [1970] AC 467.

74 As stated by May LJ in Environment Agency v Ellis[2008] EWCA Civ 1117 at [39].

75 [2008] EWCA Civ 883, 103 BMLR 134. See also Dickens v O2 plc[2008] EWCA Civ 1144, [2009] IRLR 58, where Smith LJ stated obiter at [42]–[47] that apportionment was not appropriate in any case of an indivisible injury, even, it seems, in a case falling within Bailey.

76 At [46].

77 [2007] EWHC 2913 (QB).

78 At [28]–[30].

79 At [54].

80 Ibid (original emphasis).

81 At [60].

82 At [31]–[34].

83 At [36]. This echoed a comment of Foskett J at first instance at [57] where he cited Lord Rodger's dictum in Fairchild at [129] that the material contribution rule does not involve application of the but-for test. Cf nn 48 and 49 above.

84 This mismatch between facts and law may explain the differing interpretations of Bailey by commentators. G Turton ((2009) Med L Rev 140) regards the material contribution concept as involving application of the but-for test and rightly argues that there is a need for greater clarity in the use of causal concepts; S Green ((2009) 125 LQR 44) regards the case as applying the material contribution test, and argues that this involves requiring but-for causation to be established in respect of ‘part of the ultimate injury’ (a view the present author does not share for the reasons set out in this paper); M Stauch ([2009] CLJ 27) regards the case as involving a departure from the but-for test and rightly criticises such an extension to the law.

85 Wilsher v Essex Area Health Authority[1988] AC 1090.

86 Ibid.

87 See the discussion by Lord Hope, above n 11.

88 The extent to which Fairchild did so is discussed by Steele, J in ‘“Breach of duty causing harm?” Recent encounters between negligence and risk’ (2007) 60 CLP 296 at 324–326.Google Scholar

89 Reversal of the burden of proof in cases of breach of statutory duties designed to protect workers from the very kind of injury or disease that develops, the approach rejected in Bonnington Castings, in fact has a lot to commend it; if it had been adopted, some of the difficulties in McGhee, Fairchild and Barker might have been avoided. Note also the decision of the Court of Appeal of New Zealand in Accident Compensation Corporation v Ambros[2007] NZCA 304 declining to apply a modified approach to causation in medical negligence cases under the New Zealand accident compensation legislation.