That the tort of negligence is a mess goes almost without saying.” So said David Ibbetson in How the Romans Did for Us: Ancient Roots of the Tort of Negligence (2003) 26 UNSWLJ 475.

As James Plunkett put it in The Duty of Care in Negligence (Hart Publishing, 2018), p. 76: 

Since Donoghue v Stevenson was handed down in 1932, the courts have endured a continual struggle in their attempts to articulate a precise formula that can be used to determine when a duty of care does or does not exist; in particular, a formula that is wide enough to capture the existing duties, but not so wide as to permit recovery in a raft of previously unrecognised situations; narrow enough to explain the no-duty cases, but not so narrow as to prevent the recognition of a duty in deserving cases.

It is little surprise, then, that the courts have rejected the notion of a single test for the imposition of a duty of care in favour of an incremental approach. In Robinson v Chief Constable of West Yorkshire Police [2018] AC 736 at [21], Lord Reed said this:

The proposition that there is a Caparo test [see Caparo Industries plc v Dickman [1990] 2 AC 605, 617-618] which applies to all claims in the modern law of negligence, and that in consequence the court will only impose a duty of care where it considers it fair, just and reasonable to do so on the particular facts, is mistaken. As Lord Toulson pointed out in his landmark judgment in Michael v Chief Constable of South Wales Police (Refuge and others intervening) [2015] UKSC 2; [2015] AC 1732, para 106, that understanding of the case mistakes the whole point of Caparo, which was to repudiate the idea that there is a single test which can be applied in all cases in order to determine whether a duty of care exists, and instead to adopt an approach based, in the manner characteristic of the common law, on precedent, and on the development of the law incrementally and by analogy with established authorities.

Only if a duty of care falls outside any established category – that is, if it is a truly novel duty of care – will the court undertake the three-stage Caparo analysis: Robinson at [27].

The boundaries of the duty of care in negligence are thus ever-expanding (James Plunkett refers to the “Staggering March of Negligence”) and, indeed, are never closed (Michael at [101] per Lord Toulson).

In terms of current areas of development, in Begum v Maran (UK) Ltd [2021] EWCA Civ 326 at [54], Coulson LJ observed that there is “a growing trend of claims in negligence where there has been an intervention of some kind by a third party, such as claims against public bodies and local authorities based on the acts of others.

This blog looks at the Court of Appeal’s approach to the duty of care issue in Begum.

The Relevant Facts

The Claimant in Begum is the widow of a Bangladeshi shipbreaker working in a yard in Chattogram (formerly known as Chittagong) in Bangladesh. On 30 March 2018, her husband fell to his death whilst working to dismantle an oil tanker. It is alleged that the working conditions at the Chittagong shipyards were notoriously extremely hazardous, with little or no personal protective equipment and no regard for safe working practices. Proceedings were issued against the former managers of the tanker, an English company, which had arranged for the sale of the tanker to a third party, which had in turn sold it on to the shipyard where it was beached for demolition.

The Defendant applied for strike out and/or summary judgment on the basis that it did not owe a duty of care to the Claimant’s husband. It was common ground that the Defendant was not involved with the tanker after sale and that it had no control over, or interest in, the shipyard. It was, however, assumed for the purposes of the Defendant’s application that the Defendant was aware that the demolition would take place in Bangladesh rather than in a more reputable – and thus safer – yard in China or Turkey.

The Defendant’s application failed before Jay J, who granted the Defendant permission to appeal.

Other issues arose as to limitation, which are beyond the scope of this article.

Two Routes to a Duty of Care

The Claimant put forward two alternative arguments as to the existence of a duty of care:

  • that the Defendant is liable in negligence arising from a known danger applying the principles in Donoghue v Stevenson [1932] AC 562 (“Route 1”); or
  • that the Defendant is liable for damage caused by third parties, viz. the purchaser and/or the shipyard owner and/or the Claimant’s husband’s employer, on the grounds that it created the source of the danger (“Route 2”).

Duty of Care: Route 1

At [42]-[49], Coulson LJ identified the difficulties with the Claimant’s argument as follows:

[Counsel for the Claimant] said that there were two questions for the court to answer. The first was: if the [Defendant] had sold a dangerous product direct to the yard with full knowledge of its unsafe practices, was the [Defendant’s] relationship with the deceased sufficiently proximate to establish a duty of care? He said the answer to that first question was Yes. He said the second question was whether that duty could then be negated as a result of the involvement of third parties. He said the answer to that second question was No.

The potential difficulty inherent in [the] first question is that it assumes that the vessel was a dangerous product. That seems to me to be far from certain. If its danger lay in its size, that was hardly something that was hidden, in contrast to the snail in the ginger beer bottle. Moreover, it is arguable that if, say, the vessel was in a dry dock in China, surrounded by cranes and safety infrastructure, it would not itself be inherently dangerous. When I put that to [counsel for the Claimant] during the course of argument, he said that the vessel was dangerous because its demolition was an inherently dangerous activity. That answer potentially moves this case outside the scope of ordinary Donoghue v Stevenson principles, not least because the dangerous activity – namely the demolition – was not being arranged, supervised or performed by the [Defendant].

As to [the] second question, namely the effect, if any, of the involvement of third parties, [counsel for the Claimant] said the judge had been wrong to say that such involvement could even arguably negate the duty. He said that this was because neither [the buyer] nor [the yard] had acted in any way other than anticipated. He said that they acted in the way determined by the [Defendant’s] original decision to sell to [the buyer]. …

In reliance on the proposition that the existence of a duty could not be extinguished by the involvement of third parties, if those third parties acted in the way that was or could be anticipated, [counsel for the Claimant] relied on the well-known decision of the House of Lords in Home Office v Dorset Yacht Co Limited [1970] AC 1004 …

… I am not persuaded that Dorset Yacht is of very much assistance in the present case. There, as [counsel for the Defendant] rightly pointed out, the basis for the finding of liability against the Home Office was that the Borstal trainees who caused the damage were “under the supervision and control of three Borstal officers”. It was that significant element of supervision and control of the third parties which gave rise to the finding of liability on the part of the Home Office.

For these reasons … I am doubtful whether, at trial, a court would follow [counsel for the Claimant] down route 1 to reach his intended destination, namely the finding of a duty of care in the present case on Donoghue v Stevenson principles. I am inclined to agree with the judge’s conclusion at [37] that this case “does not fit comfortably” within such principles.

Notwithstanding this scepticism, however, Coulson LJ declined to strike out or enter summary judgment as regards Route 1 on the basis that the argument was not “so fanciful that it should be struck out”: see [50].

Duty of Care: Route 2

In relation to Route 2, Coulson LJ reviewed various authorities dealing with the ‘creation of danger’ exception, starting with Smith v Littlewoods Organisation Ltd [1997] 1 AC 241 (Lord Goff) and including Attorney General v Hartwell (British Virgin Islands) [2004] 1 WLR 1273, Mitchell v Glasgow City Council (Scotland) [2009] 1 AC 874, Michael (supra), Robinson (supra) and Poole Borough Council v GN [2020] AC 780.

Coulson LJ made two important observations regarding the authorities at [59]-[61] (emphasis added):

Two general points should be made about these authorities. First, they show that it may be unnecessary and potentially distracting to insist that claims in tort arising from the intervention of third parties must be corralled into a pen marked “pure omissions”. That may have been how Lord Goff treated such claims in Smith, but that language is not always helpful and can appear outdated. It can be unhelpful because it can lead to interminable debates (including in the present case) about whether the facts of a particular case could be said to amount to “pure omissions”, or whether there was an act or acts which might somehow make all the difference to the application of the exception. …

[Lord Reed’s judgment in Poole] explains why the language of “pure omissions” might now be regarded as somewhat outdated. He instead drew the distinction between causing harm (making things worse), on the one hand, and failing to confer a benefit (not making things better), on the other. He noted that the law often imposes a duty of care not to make things worse, but rarely imposed a duty to make things better. That is a much clearer description of where the dividing line might be drawn in any particular case.

The second general point to be made about these authorities is that they demonstrate that one of the most fast-developing areas of the law of negligence at present concerns the scope and extent of this and other exceptions to the general rule that there is no liability in tort for harm caused by the intervention of third parties. That is a point also emphasised in Lord Reed’s judgment in Poole.”

Coulson LJ then opined as follows at [63]-[66]:

This second way of putting the duty of care is again not straightforward. The quartet of decisions in Hartwell, Mitchell, Michael and Robinson demonstrate the restricted circumstances in which a defendant will be liable in tort for damage caused by the intervention of a third party, and that it will only be in a relatively extreme case that the ‘creation of danger’ exception will operate. Much will turn on the precise nature and extent of the danger said to have been created. …

However, whilst I do not underestimate the hurdles that [Route 2] will face at trial, I consider that this way of putting the claim is arguable, and not fanciful. The ‘creation of danger’ is a recognised exception to the usual rule as to the intervention of third parties which may give rise to a duty of care. I consider that [the agreed factual assumptions] are capable of triggering that exception: the [Defendant] arguably played an active role by sending the vessel to Bangladesh, knowingly exposing workers (such as the deceased) to the significant dangers which working on this large vessel in Chattogram entailed. That might render the vessel and its demolition dangerous in a way that the large empty cinema in Smith was not. The [yard’s] failure to provide any safety harnesses or any other proper equipment, and the tragic consequences of their not doing so, were entirely predictable …

This may properly be described as an unusual extension of an existing category of cases where a duty has been found, but it would not be an entirely new basis of tortious liability. Moreover, the [Claimant’s] case, when reduced to its basic ingredients, is that, not only is there the necessary foreseeability, but there is also a relationship of proximity between the [Defendant] and those who might foreseeably be injured by the danger created by (or inherent in) the [Defendant’s] decision to sell to [the buyer]. In my view, by reference to the authorities noted above, such a case is at least arguable.

I have some sympathy with [counsel for the Defendant’s] submissions on the practicalities of all this (that the [Defendant] cannot be held to be the shipbreaking world’s policeman and that, if there was a duty of the kind alleged here, the ramifications might be extremely far-reaching), but it seems to me that they go to a consideration of whether, if the alleged duty is found to fall outside any established category, it would still be fair, just and reasonable to find the existence of the duty in this case. The authorities make plain that such an issue is not to be decided at an interim stage like this.

The appeal on the duty of care point was accordingly dismissed. There will now be a preliminary trial on the question of limitation (for the reasons in sections 8-10 of Coulson LJ’s judgment).


In a short concurring judgment in Begum, Males LJ said this at [133] (a passage expressly endorsed by Bean LJ at [141]):

… it may turn out that the Claimant is unable to prove the factual averments which she makes. But if she is able to do so, it might be thought that it would be a poor system of justice which gave her no remedy against the Defendant.

The Court of Appeal’s refusal to strike out the Claimant’s duty of care arguments ensures that exceptions to the general rule that there is no liability in tort for harm caused by the intervention of third parties will continue to generate argument. Indeed, one very recent example is the case of DFX v Coventry City Council [2021] EWHC 1382 (QB) – the first ‘failure to remove’ claim to be tried following the Supreme Court’s decision in Poole.