The but-for test
Factual causation is most commonly established by the but-for test: the claimant must show that, on the balance of probabilities, the damage would not have occurred but for the defendant's breach. This is a counterfactual enquiry. The court asks: what would have happened if the defendant had discharged the duty of care?
In Barnett v Chelsea and Kensington Hospital Management Committee [1969] 1 QB 428, the casualty officer negligently failed to examine a patient who presented with vomiting; the patient died hours later from arsenic poisoning. Nield J held that, even if the doctor had examined him, the arsenic would have been undetectable and the patient would have died regardless. The breach did not satisfy the but-for test. The outcome may appear harsh—the hospital certainly acted wrongfully—but liability in tort is confined to causally connected harms.
The test is applied on the balance of probabilities (>50%). A claimant who proves a 49% chance that the breach caused the loss will fail entirely: Hotson v East Berkshire Health Authority [1987] AC 750. English law rejects proportionate recovery for loss of chance in personal injury claims, although it permits it in professional negligence: Allied Maples Group Ltd v Simmons & Simmons [1995] 1 WLR 1602.
Multiple sufficient causes
The but-for test fails when two or more factors, each independently sufficient, combine to cause harm. Suppose two defendants negligently light separate fires, each of which would have destroyed the claimant's property. Neither fire is a but-for cause, yet both are intuitively responsible. English law has addressed this through the concept of a material contribution to the damage. In Bonnington Castings Ltd v Wardlaw [1956] AC 613, the House of Lords held that where the defendant's breach made a material (i.e. more than negligible) contribution to an indivisible injury, causation is established even if other factors also contributed.
The test was refined in Bailey v Ministry of Defence [2008] EWCA Civ 883, where Waller LJ held that the claimant need not prove that the breach was the sole or even the principal cause, provided it made a material contribution to the cumulative state that resulted in harm. The line between material contribution to injury and material increase in risk (see Fairchild, below) remains contested.
The Fairchild exception: material increase in risk
In cases of indeterminate causation—where the claimant cannot identify which of several exposures caused the disease—the House of Lords has carved out a narrow exception. Fairchild v Glenhaven Funeral Services Ltd [2003] 1 AC 32 concerned mesothelioma induced by asbestos exposure during employment with multiple successive employers. Medical science cannot identify which exposure triggered the disease; a single fibre may suffice. Strict application of the but-for test would deny recovery. The House of Lords held (Lord Bingham, Lord Nicholls, Lord Hoffmann, Lord Hutton, Lord Rodger) that where (i) the claimant was exposed by the negligence of two or more defendants, (ii) each exposure was capable of causing the disease, and (iii) the disease is indivisible and the precise causal mechanism scientifically indeterminate, proof of a material increase in risk suffices for liability. Each defendant is liable in full (joint and several), subject to contribution proceedings.
Barker v Corus UK Ltd [2006] UKHL 20 initially interpreted Fairchild as imposing several (proportionate) liability, but Parliament reversed that holding within months by enacting the Compensation Act 2006, s 3, which restores joint and several liability in mesothelioma claims.
The Supreme Court has emphasised that Fairchild is not a general relaxation of the but-for test. In Sienkiewicz v Greif (UK) Ltd [2011] UKSC 10, Lord Phillips confirmed that the exception applies only where it is impossible to prove but-for causation and where the claimant has been wrongfully exposed to the relevant risk by the defendant. In Sienkiewicz itself, the deceased had both occupational and environmental (background) exposure to asbestos. The Supreme Court held that the defendant could be liable under Fairchild even though the relative contributions of occupational and environmental exposure were uncertain, provided the occupational exposure materially increased the risk.
Legal causation: remoteness and intervening acts
Even where factual causation is established, the defendant may escape liability if the loss is too remote or if a novus actus interveniens (new intervening act) breaks the chain. These doctrines reflect policy judgments about the fair limits of responsibility.
Remoteness: the Wagon Mound test
In The Wagon Mound (No 1) [1961] AC 388, the Privy Council held that a defendant is liable only for damage of a kind or type that was reasonably foreseeable at the time of breach. The charterers of a ship negligently discharged oil into Sydney Harbour; the oil ignited two days later when welding sparks fell onto debris. Although some damage to the wharf (fouling) was foreseeable, fire damage was not. The Privy Council (Lord Simonds) held that the fire damage was too remote and therefore not recoverable.
Wagon Mound replaced the earlier directness test articulated in Re Polemis and Furness, Withy & Co Ltd [1921] 3 KB 560, which had imposed liability for all direct consequences, however unforeseeable. The modern test aligns remoteness with the scope of the duty: the defendant is expected to guard only against reasonably foreseeable kinds of harm.
Importantly, the claimant need not show that the extent or precise manner of harm was foreseeable. In Hughes v Lord Advocate [1963] AC 837, workmen left an open manhole covered by a tent and surrounded by paraffin lamps. A child knocked a lamp into the hole, causing an explosion. Although explosion was unforeseeable, burns from the lamp were foreseeable, and the House of Lords held the harm was not too remote: the injury was of a foreseeable type, even if it occurred in an unforeseeable way. This principle—that the precise mechanics need not be foreseen—is often called the "thin skull" or "egg-shell skull" rule by analogy, though that rule properly concerns extent rather than type.
Intervening acts (novus actus interveniens)
An intervening act may break the chain of causation if it is so independent and potent that it becomes the sole effective cause of the damage. The question is whether the intervening act was reasonably foreseeable or whether it was so extraordinary as to eclipse the defendant's breach.
Third-party interventions: In Knightley v Johns [1982] 1 WLR 349, the defendant negligently caused a traffic accident in a tunnel. The police inspector then negligently ordered officers to ride back against the traffic flow; one officer was injured. The Court of Appeal (Stephenson LJ) held that the inspector's negligent order was a novus actus: it was not a natural or probable consequence of the original collision and broke the chain.
By contrast, in Scott v Shepherd (discussed above) and Rouse v Squires [1973] QB 889, the intervening act was held to be within the range of foreseeable responses to the defendant's breach and did not break the chain. The test is whether the intervention was reasonably foreseeable, not whether it was itself negligent.
Acts of the claimant: A claimant's unreasonable conduct may also break the chain. The defendant remains liable, however, if the claimant's response was reasonable or instinctive. In McKew v Holland & Hannen & Cubitts (Scotland) Ltd [1969] 3 All ER 1621, the pursuer suffered a leg injury that occasionally gave way without warning. He later attempted to descend a steep staircase without a handrail, his leg gave way, and he suffered further injury. The House of Lords held this was unreasonable and broke the chain. Contrast Wieland v Cyril Lord Carpets Ltd [1969] 3 All ER 1006, where the claimant, forced to wear a neck brace after the defendant's negligence, fell down stairs because her bifocals and the brace limited her vision. The Court of Appeal held the fall was a natural consequence and not a novus actus.
Natural events: A subsequent natural event will rarely break the chain unless it is truly extraordinary. In Carslogie Steamship Co Ltd v Royal Norwegian Government [1952] AC 292, a ship damaged by the defendant's negligence was further damaged by a storm during the voyage to repair. The House of Lords held both defendants (the original tortfeasor and any party responsible for the storm damage) jointly liable; the storm did not break the chain.
Scope of duty and the SAAMCO principle
Recent authority has emphasised that questions of legal causation often reduce to the scope of the duty assumed or imposed. This is particularly clear in professional negligence. In South Australia Asset Management Corp v York Montague Ltd (SAAMCO) [1997] AC 191, the House of Lords (Lord Hoffmann) held that a valuer who negligently overvalues property is liable only for losses flowing from the inaccuracy of the valuation, not for all losses incurred in the transaction. The scope of the duty defines the risk against which the defendant must protect the claimant. This principle has been applied in clinical negligence (Chester v Afshar [2004] UKHL 41) and financial mis-selling cases, and it represents a shift from mechanical causal tests toward a more openly normative enquiry.