​Sherlock Holmes, 221B Baker St, London, famously relied on logic to solve his cases, beginning with the principle that once you have eliminated the impossible, “whatever remains, however improbable, must be the truth”.

But the UK Court of Appeal in Graves v Brouwer1 has unanimously agreed that it is not always appropriate to apply Sherlock Holmes’ reasoning to establish causation.

The case of the flaming embers

The defendant, Mr Brouwer had set light to two A4 size pieces of paper and two A4 size pieces of cardboard in a narrow alleyway which separated his property from his neighbour’s, Miss Graves.  Thirty minutes later, Miss Graves’ house caught fire.

Miss Graves brought proceedings against Mr Brouwer, alleging that embers from his fire had floated up and ignited combustible material in her roof eaves. 

There was no forensic investigation of all the potential causes of the fire, such as arson, smoking materials, cooking/heating materials, or anting (birds picking up discarded cigarettes and returning them to their nest).  The parties’ experts acknowledged that the lack of forensic evidence hampered the efficient determination of the cause of the fire.

The High Court found that the damage at Miss Graves’ house had been caused by Mr Brouwer.  The Judge placed significant weight on the acceptance by the defendant’s experts during cross examination that “if the court takes the view that arson did not cause the fire then, on the balance of probabilities, [embers from the fire] must be the cause even if improbable.”

The decision

In allowing Mr Brouwer’s appeal, the Court of Appeal held that the process of reasoning which led the High Court Judge to conclude that the claimant had succeeded on causation was flawed for three reasons:

  • in the absence of satisfactory evidence, the only just course for the court to take is to decide the case on the burden of proof
  • Sherlock Holmes’ deductive logic only applies when all relevant facts are known, so that all possible explanations, except a single extremely improbable one, can properly be eliminated.  Here, Holmes’ deductive reasoning did not apply because there had not been a thorough forensic investigation into all possible causes of the fire to eliminate all other possible outcomes; and
  • the High Court Judge had failed to stand back and ask herself the ultimate question whether she was satisfied that the suggested cause of fire was more likely than not to be true.  If a judge concludes, on a whole series of cogent grounds, that the cause of the event is extremely improbable, a finding by the judge that the cause will satisfy the “more likely to have occurred than not” standard, does not accord with common sense.

Chapman Tripp comments

The decision, although described by the Court of Appeal as “by any standards a bizarre case”, contains several important points in relation to causation, in particular:

  • the fact that one possible cause of an event was discounted does not automatically mean that another possible cause, however implausible, will satisfy the causation test, and
  • there are cases, on the facts, in which a claimant will fail to establish causation on the balance of probabilities if there remains significant doubt as to a range of possible causes – even if the plaintiff’s account is the most likely amongst those. 

The Court of Appeals of England and Wales’ rejection of easy simplifications in the causation analysis has resonance with the recent judgment in Vero Insurance New Zealand Ltd v Morrison2  where the New Zealand Court of Appeal upheld an insurers’ challenge to the High Court’s decision to rely on modelling evidence to allocate damage, and remediation costs, between different earthquake events. 

The Court concluded that the inherent limitations in the model meant that it could not be relied upon to this extent.  There could be no assurance that all of the factors which might have created the damage had been accounted for and the model did not clearly link the damage to the cost of remediation.

Taken together, the cases reiterate that causation in fact is an intensely factual inquiry, in which simplifications – whether by experts or otherwise – are unlikely to succeed, even if supported by reference to famous literary detectives.

Or as the long-suffering Dr Watson might have said were he here: “Not so elementary, My Dear Holmes.  Not so elementary”.