By Per Laleng, Senior Lecturer in Law
The Facts
On 10 February 2019, a few hours after being released from police custody, Alexander Lewis-Ranwell bludgeoned three elderly men to death in their homes in Exeter with a hammer. He believed they were paedophiles. At his criminal trial, three psychiatrists agreed that he had been suffering from severe paranoid schizophrenia and did not know that what he was doing was wrong. The jury found him not guilty by reason of insanity. He was detained at Broadmoor Hospital under a hospital order with restrictions.
So far, a tragic story of severe mental illness and its consequences. But in 2020 Lewis-Ranwell issued civil proceedings against the organisations responsible for his mental health care in the lead-up to the killings. The police had arrested him on 8th and 9th February for separate offences, including causing grievous bodily harm to an elderly man with a saw. His case was stark: had the defendant organisations assessed him properly and detained him under the Mental Health Act when they should have done, he would not have killed three men shortly after being released from custody. He sought substantial compensation — for psychiatric injury, loss of liberty, lost earnings, cost of care, and damaged reputation — and, notably, an indemnity against any claims the victims’ families might bring against him.
Three of the four defendants applied to strike out the claims in negligence using the illegality defence. Almost seven years later, on 21 January 2026, the Supreme Court unanimously held that the illegality defence succeeded and the claims in negligence were barred.
Why this case was novel
The illegality defence — known by the Latin maxim ex turpi causa non oritur actio, meaning no right of action arises from a dishonourable cause — is a rule of public policy that prevents claimants from recovering compensation for the consequences of their own illegal or immoral conduct. Its modern framework was presumptively settled in Patel v Mirza [2016] UKSC 42, where a nine-justice bench replaced a rigid rules-based approach with a structured three-stage inquiry. Courts should ask whether: (a) denying the claim would serve the purpose of the underlying prohibition; (b) any countervailing policies weigh against denial; and (c) denial would be proportionate. The answers determine whether allowing the claim would damage the integrity of the legal system and would be contrary to the public interest.
The complication in Lewis-Ranwell was that all previous cases applying the illegality defence in the context of unlawful killing – Gray v Thames Trains, Henderson v Dorset Healthcare, and Clunis v Camden – had involved claimants convicted of manslaughter by reason of diminished responsibility. Lewis-Ranwell had been acquitted. He bore no criminal responsibility whatsoever. Both the High Court and the Court of Appeal majority held that this distinction was decisive: the criminal law had exonerated him, and the civil law should not treat him as it would a convicted killer. Andrews LJ dissented: civil liability in battery persisted regardless of insanity, and the policy arguments were no less compelling ( for a full analysis of these issues, see my article here).
The threshold question: a new rule
Before applying the Patel framework, the Supreme Court asked a prior question: is the illegality defence even engaged? Not every unlawful act, however trivial, triggers the need for a Patel assessment [see para 113, judgment]. The novelty of Lewis-Ranwell’s case required the Court to articulate when the defence comes into play at all.
The Court of Appeal’s answer was effectively: when the claimant bears criminal responsibility. Criminal responsibility was treated as a necessary condition. The Supreme Court rejected this. Criminal responsibility may be sufficient to engage the defence – the analogous trio of cases demonstrates that – but it is not necessary. The relevant test, synthesised from dicta of Lord Mansfield in Holman v Johnson and Lord Sumption JSC in Apotex, is this: the illegality defence is in play if the claimant’s unlawful conduct engaged and has engaged the interests of the State or the public interest [Lewis-Ranwell [131], [134]].
The intellectual manoeuvre here is subtle. Lord Mansfield spoke of courts refusing to assist claimants who found their claims on an “immoral or illegal” act. Lord Sumption in Apotex explained that this meant acts engaging the public interest – which would include criminal or quasi-criminal conduct, but not purely civil wrongs where corrective justice suffices. Since Lord Mansfield’s formula included immoral acts, the threshold does not require criminal responsibility. And since what matters is whether the public interest is engaged, the absence of moral culpability is similarly downgraded: the question is not whether the claimant is blameworthy, but whether their conduct is of the kind the State has an interest in condemning.
Applying this to the facts, the Supreme Court found the threshold crossed. Lewis-Ranwell had committed the actus reus of murder. His acquittal by reason of insanity did not negate the mens rea – he had intended to kill, even if he did not appreciate the wrongfulness of his act. His conduct was not justified (as self-defence would be) but merely excused: the criminal law exempted him personally from responsibility, but the killings themselves remained legally and morally wrongful. He had breached a fundamental moral norm – “you shall not kill” – which applies even to those who bear no criminal responsibility. And crucially, his conduct had engaged the public interest concretely: criminal proceedings were brought, and the result was lawful detention for the protection of the public.
Two senses of ‘public interest’: a structural ambiguity
The threshold rule raises several questions that Lewis-Ranwell does not fully answer, and the most fundamental concerns the concept of the “public interest”. The judgment deploys it in two quite different senses without explicitly distinguishing them.
At the threshold stage, “public interest” refers to the State’s interest in condemning and responding to certain forms of conduct: a conduct-based idea. At the Patel assessment stage, “public interest” refers to the systemic interest in maintaining the coherence and integrity of the legal system: a quite different, structural idea. The Court uses the same phrase for both without acknowledging the shift. The result is that the two stages of the inquiry appear to ask the same question when they do not.
This is not merely a terminological complaint. The Supreme Court announced that coherence was “of central importance in determining the threshold,”[Lewis-Ranwell [127]] but the actual threshold reasoning rests on the actus reus, the mens rea, the moral norm, and the hospital order, none of which are straightforwardly coherence arguments. Coherence is invoked rhetorically to displace the Court of Appeal’s approach, but a conduct-based conception of public interest does the actual analytical work. The Court uses the language of coherence and the logic of state condemnation – and does not explain how they relate, if indeed they do. This is the judgment’s deepest structural difficulty.
Further difficulties: degree, triviality, and open questions
A second problem is logical. The Supreme Court argues that the difference between insanity and diminished responsibility is one of degree rather than kind, citing the sentencing judge’s observation in Henderson that the claimant there bore only a “very low level of criminal responsibility” [Lewis-Ranwell [126]]. But if criminal responsibility is binary – you either have it or you do not – then comparing “very low” with “none” is comparing two categorically different states, not two points on a spectrum. The Court assumes a gradualist picture of criminal responsibility without defending it.
Third, the threshold rule is poorly specified at its margins. The Court acknowledges that trivial unlawful acts do not trigger the Patel assessment but does not explain where trivial becomes non-trivial. The hospital order in Lewis-Ranwell provides a concrete marker of State engagement – but it may be doing more work than the Court acknowledges, functioning not merely as evidence of public interest engagement but as its practical criterion. The stated rule is that public interest engagement is what matters. The application of that rule, however, suggests criminal proceedings (even without conviction) are its principal indicator. What happens when there are no criminal proceedings at all? What if the police investigate but do not prosecute? What if the claimant is granted immunity? These are questions the threshold rule raises but does not answer.
There is also a broader implication for future cases. Lord Sumption’s formulation in Apotex drew a line between criminal or quasi-criminal conduct (which engages the public interest) and purely civil wrongs (which do not). Lewis-Ranwell endorses that line but also extends it: acts sufficiently immoral to engage the public interest can cross the threshold even without criminal responsibility. The precise scope of that extension – which immoral acts, and how seriously immoral? – is left open. Cases like Patel, Stoffel, and Hounga as well as the analogous trio and indeed Lewis-Ranwell involved straightforwardly criminal underlying conduct, so the threshold would be crossed without difficulty. The harder questions will arise in future cases where the conduct is “serious” but not criminal.
Looking ahead
The Supreme Court’s answer to the threshold question extends the reach of the illegality defence into territory it had not previously covered: acquitted defendants, cases without criminal responsibility, and conduct engaging the State’s interest through means other than conviction. Whether that extension is doctrinally justified, and whether the analytical costs – in terms of precision and conceptual clarity – are acceptable, are questions that only become fully visible when the Patel assessment is examined. That is the subject of Part II.
Per Laleng’s teaching, publications and contact details can be accessed here