Law of Torts
Causation in Tort: Factual and Legal Causation, the But For Test, and the Supplements the Law Has Built Around It
Causation is two questions asked in order. Factual causation asks whether the defendant's breach made any difference at all, and the standard test is the but for test: but for the breach, would the harm have occurred? Legal causation asks a different question, whether the law will treat the breach as the cause for the purpose of imposing liability, and it filters out what is too remote or insufficiently operative. The Latin vocabulary marks the same divide: causa causans is the immediate and effective cause, causa sine qua non a cause without which the harm would not have happened but which is not the operative one. The but for test fails in three recognised situations, and the law has built supplements to meet them.
The two questions, where the but for test fails, and the supplements the law has built
1. Factual Causation: The But For Test
- The test. But for the defendant's breach, would the plaintiff have suffered the harm he did? If he would have suffered it anyway, the breach caused nothing.
- It is a question of fact, decided on the balance of probabilities like any other question of fact.
- It is a test of exclusion. It tells you what did not cause the harm. Passing it does not establish that the defendant is liable, only that he is not excluded.
- It is applied to the harm actually suffered, and not to some more general risk.
- A defendant who fails it wins outright, however serious his breach was.
📖 Barnett v. Chelsea and Kensington Hospital Management Committee [1969] 1 QB 428 Facts Three night watchmen went to the defendants' casualty department in the early morning complaining of vomiting after drinking tea. The nurse telephoned the casualty officer, who did not examine them and told her to send them home to see their own doctors. One of them died some hours later of arsenic poisoning. The evidence was that even if he had been admitted and treated, he would have died in any event, the antidote not being capable of being administered in time. Held The hospital had been negligent in failing to examine him, but was not liable for his death. Ratio The plaintiff must prove that the breach caused the damage. Where the harm would have occurred even if the defendant had discharged his duty, the but for test is not satisfied, and the breach, however clear, founds no liability for that harm. |
2. Where the But For Test Fails
The situation | Why the test breaks down |
Multiple sufficient causes | Two fires, each on its own enough to destroy the house, converge on it. But for either, the house still burns. The test acquits both defendants, which cannot be right |
Cumulative causes | The harm results from the combined effect of several contributions, some innocent and some tortious, and none of them alone would have produced it |
Evidential gaps in science | Where medical science cannot say which of several exposures triggered the disease, the plaintiff can never discharge the but for burden |
Successive causes | A second event supervenes and would have produced the same result in any case, raising the question how far the first defendant's liability runs |
3. The First Supplement: Material Contribution to the Harm
📖 Bonnington Castings Ltd. v. Wardlaw [1956] AC 613 Facts The plaintiff contracted pneumoconiosis from inhaling silica dust at the defendants' steel foundry. The dust came from two sources: pneumatic hammers, where no known precaution was available and the defendants were not in breach, and swing grinders, where dust extraction equipment should have been maintained and was not. It was impossible to say what proportion came from each. Held The defendants were liable. Ratio Where a disease is caused by the cumulative effect of a noxious agent, it is enough for the plaintiff to prove that the defendant's breach made a material contribution to the injury. It need not be the sole, or even the main, cause. Anything that is not de minimis is a material contribution. |
4. The Second Supplement: Material Increase in Risk
📖 McGhee v. National Coal Board [1973] 1 WLR 1 Facts The plaintiff worked in brick kilns and was exposed to abrasive brick dust. The employers were not in breach in exposing him during work, but were in breach in failing to provide washing facilities, so that he cycled home caked in dust. He contracted dermatitis. Medical science could not say whether the additional exposure on the journey home had caused the disease or merely added to a risk already present. Held The employers were liable. Ratio Where the state of scientific knowledge makes it impossible to prove that the breach caused the injury, proof that the breach materially increased the risk of the injury may be treated as sufficient. The principle was reaffirmed and extended in Fairchild v. Glenhaven Funeral Services Ltd. [2003] 1 AC 32, where a mesothelioma victim had been exposed to asbestos by several employers and could not show which exposure triggered the disease. |
- The supplements are exceptional and are kept so. They exist because the ordinary rule would produce injustice where science cannot supply the answer, not because the burden of proof has generally been relaxed.
- Material contribution is to the harm; material increase is to the risk. The first is a modification of the but for test, the second a departure from it.
- Loss of a chance raises the same difficulty in a different form. Where the plaintiff cannot show that proper treatment would probably have saved him, but can show that it would have improved his prospects, the English courts have been reluctant to allow recovery for the lost chance in personal injury, while allowing it in economic loss cases.
5. Legal Causation
- Passing the factual test is not enough. Every event has an infinite chain of causes behind it, and the law must select.
- Causa causans is the immediate, effective and operative cause: the one the law fixes on and attaches liability to.
- Causa sine qua non is a cause without which the harm would not have happened, but which is not the effective cause. Everything in the chain back to the defendant's birth qualifies, which is why the test needs a filter.
- The two filters the law uses are novus actus interveniens, which asks whether a later event broke the chain, and remoteness of damage, which asks whether the harm was of a foreseeable kind.
- Legal causation is a question of law and policy, not of fact, and that is why reasonable judges can differ on it where they could not differ on the but for question.
6. The Order of the Questions
Step | The question | The consequence of failure |
1 | But for the breach, would the harm have occurred? | If it would, the claim fails: Barnett |
2 | If but for cannot be applied, did the breach materially contribute to the harm? | If not, the claim fails: Bonnington Castings |
3 | If science cannot say, did the breach materially increase the risk? | If not, the claim fails: McGhee; Fairchild |
4 | Did a later event break the chain of causation? | If it did, liability stops there: novus actus interveniens |
5 | Was the damage of a kind reasonably foreseeable? | If not, it is too remote: The Wagon Mound |
7. Where the Law Keeps the Line
⚠ The supplements do not reverse the burden of proof It is easy to read Bonnington Castings and McGhee as having relaxed the plaintiff's burden generally. They have not. The burden of proving causation remains on the plaintiff throughout, and it remains a burden to prove causation on the balance of probabilities. What the two cases do is recognise that in a particular class of case, where the harm is produced cumulatively or where science cannot identify which exposure triggered a disease, insisting on the but for test would mean that no claimant could ever succeed against a defendant who was plainly in breach. The exceptions are confined to that class. Outside it, a plaintiff who cannot show that the breach probably made a difference loses, however sympathetic his position and however clear the breach, and Barnett is the standing illustration. |
8. The Position Stated Shortly
1. Causation is two questions: factual causation, and legal causation.
2. The but for test is the standard test of factual causation: but for the breach, would the harm have occurred?
3. Barnett v. Chelsea and Kensington Hospital is the leading illustration: a negligent failure to examine a man who would have died anyway founded no liability for his death.
4. The test fails where there are multiple sufficient causes, where causes are cumulative, and where science cannot identify which exposure caused the harm.
5. Bonnington Castings v. Wardlaw supplies the first supplement: a material contribution to the harm is enough.
6. McGhee v. National Coal Board, extended in Fairchild, supplies the second: a material increase in the risk may suffice where science cannot say more.
7. The supplements are exceptional, and the burden of proving causation remains on the plaintiff throughout.
8. Legal causation asks whether the law will treat the breach as the cause, and it is a question of law and policy.
9. Causa causans is the effective and operative cause; causa sine qua non is a necessary but not operative one.
10. The two filters the law uses are novus actus interveniens and remoteness of damage.