House of Lords · Unanimous; Lord Radcliffe gave the principal speech
Facts
Bairstow and Harrison, the taxpayers, bought a complete spinning plant from a company in liquidation. They had no connection with the spinning industry, and they sold the plant in separate lots at a profit after a relatively short period. The Inspector of Taxes, Edwards, assessed the profit to income tax as arising from an adventure in the nature of trade. The General Commissioners found that the transaction was not an adventure in the nature of trade, and the Inspector appealed by way of case stated.
Issue
Whether the question whether a transaction is an adventure in the nature of trade is one of fact or of law, and in what circumstances a court may set aside the Commissioners' determination on a case stated.
Held
Appeal allowed. The Commissioners' determination that there was no adventure in the nature of trade could not be supported on the facts they had found, and it was set aside.
Ratio
- Whether the facts found amount to trade is a mixed question: the meaning of the statutory term is a matter of law, but its application to primary facts is usually for the fact-finder.
- A court may set aside the fact-finder's determination if it shows a misdirection in law, or if the facts found are such that no person acting judicially and properly instructed as to the relevant law could have reached it, that is, where the true and only reasonable conclusion contradicts the determination.
The judgments
Lord Radcliffe (lead)
Held that where the Commissioners have found the primary facts, the court may intervene if their conclusion is one that the facts cannot reasonably support, since this shows that they must have misunderstood the law. Applying this, he found that the features of the transaction (purchase for resale, no use of the plant, breaking up into lots, short holding period) pointed only towards trade.
Key reasoning
- The statutory expression adventure in the nature of trade has a legal meaning, so the Commissioners' conclusion depends on their having applied the correct legal concept to the facts.
- Where the Commissioners give no reasons beyond a bare conclusion, the court can test that conclusion against the primary facts found, and if those facts admit of only one conclusion, the court may infer an error of law.
- On the facts found, the plant was acquired solely to be resold at a profit, was not used or enjoyed by the taxpayers, and was disposed of piecemeal, so the only reasonable conclusion was that the transaction was an adventure in the nature of trade.
Where it fits in the module
In the Administrative Law FHS topic on questions of law and fact, Edwards v Bairstow shows the line between a decision-maker's primary findings, which are largely immune, and the application of a legal standard to those findings, which is open to scrutiny. It sits alongside Anisminic [1969] 2 AC 147 (the later collapse of the distinction between jurisdictional and non-jurisdictional error) and E v SSHD [2004] EWCA Civ 49 (mistake of fact as a ground of review). It also frames the debate over deference to expert tribunals, which later courts emphasised in cases such as AH (Sudan) v SSHD [2007] UKHL 49. The tutorial question is whether the supposed distinction between fact and law is principled or merely a label for the intensity of review a court chooses to apply.
Criticisms
- The true and only reasonable conclusion test is in substance irrationality review, so it blurs the line between an appeal on law and judicial review of unreasonableness.
- The distinction between law and fact is unstable, because a question is treated as law largely when the court wishes to control the result; the label conceals the real policy choice about which body should decide.
- Scrutiny of inferences from primary facts invites re-argument of the merits and undermines the position of specialist fact-finders who are better placed to evaluate the evidence.
- The reasoning allows a court to infer an error of law from the outcome alone, which makes the test difficult to apply predictably and gives little guidance on how strong the inference must be.
Written with AI (claude-sonnet-5.5 (Claude Code writer)) on 1 November 2026, reviewed for errors by a second model (claude-sonnet-5.5 (Claude Code checker)). Read the case itself at [1956] AC 14. Spotted an error? Tell us.