Similar miscarriage of justice to Sally Clark case “could still occur”


Feeny: Reliability of trials involving scientific evidence needs to be improved

A miscarriage of justice “analogous” to the wrongful conviction of solicitor Sally Clark for the murder of her two baby boys in 1999 “could still occur”, a barrister and the director of an expert evidence service provider have argued.

Charles Feeny and Carolina Stamboulid said an “exact repetition” of the Clark case was improbable, but changes since 1999 had not removed “the structural risk” exposed by the case.

“An analogous failure could still occur where expert evidence appears authoritative, where the reasoning depends on specialist assumptions not readily visible to lawyers or jurors, where probabilistic or scientific evidence is not properly tested before trial and where impressions of credibility or demeanour reinforce an apparently coherent prosecution narrative.

“The necessary safeguard is therefore a legal process capable of identifying when the evidential basis of expert opinion requires specialist scrutiny before it is placed before the jury, rather than reliance on any single rule, training requirement or checklist.”

Mr Feeny, who practises in Liverpool, is also a director of Complete Counsel, a digital support business for barristers, and Ms Stamboulid is director of Liverpool-based Episteme Scientific Consulting, which provides scientific analysis of complex medical evidence in court cases.

Sally Clark was convicted of murder in November 1999. Her first appeal was rejected by the Court of Appeal in October 2000, but her second appeal succeeded in January 2003. However, she died in 2007 from alcohol poisoning.

Both boys were subject to post-mortem examinations carried out by Home Office pathologist Dr Williams. Dr Williams failed to mention, both in the post-mortems and during the trial, a microbiology report which suggested that one of the boys could have died from natural causes.

The jury also heard from Professor Roy Meadows, a paediatrician called by the prosecution, who infamously stated that the probability of two sudden infant deaths occurring naturally within a family such as the Clarks was approximately one in 73m – a view subsequently discredited and for which he was struck off the medical register.

The Court of Appeal overturned the conviction on the basis of the undisclosed tests but said Professor Meadows’ statistical error would have sufficed.

In an article for Forensic Science International, as part of an issue on the science behind wrongful convictions, Mr Feeny and Ms Stamboulid said section 7 of the 2023 Criminal Practice Direction addressed the admissibility of expert evidence by reference to relevance, necessity, competence and sufficient reliability, and went on to identify 26 factors relevant to the assessment of reliability.

The practice direction “requires the court to be astute to identify potential flaws” in expert evidence, they wrote.

“The difficulty is that discernment in this context depends on a technical knowledge base that judges and advocates may not possess.”

The authors said specialist training in the giving of expert evidence “may assist, as may the disclosure of previous adverse judicial comment required by the Criminal Practice Directions 2023, but neither removes the need to consider whether the expert is qualified to give the particular opinion advanced”.

The article recounted how some contemporary and later accounts of the trial commented that Ms Clark’s demeanour remained calm almost throughout.

“She was a qualified solicitor and the daughter of a senior police officer, and her composure may therefore have reflected restraint, shock or professional self-control. The concern is that it may nevertheless have been capable of being read adversely, giving the impression of emotional coldness.”

In this “limited sense”, her case “invites comparison with the fate of the anti-hero Meursault in Albert Camus’ classic novel The Outsider”, sentenced to death after a trial “which involved, amongst other things, the prosecution producing evidence of his lack of emotion at his mother’s funeral”.

The comparison was useful because it captured a “familiar forensic danger, namely that a person’s apparent lack of expected emotion may be treated as if it reveals character, guilt or moral abnormality.

“That danger is intensified where it forms part of a wider evidential failure. In Clark, the jury were not considering demeanour in a vacuum. They were considering it against a background of disputed medical evidence, confident expert opinion, statistical rarity and a prosecution case into which the prosecutor’s fallacy had been incorporated.

“In that context, apparent composure could take on a significance far beyond its true evidential value and assist in giving apparent coherence to a case which already appeared to carry scientific force.”

The authors said the Clark case demonstrated “the risk inherent in a system in which confident expert opinion can be advanced, admitted and relied upon without sufficiently structured scrutiny of the assumptions on which it rests.

“The prevention of similar injustices depends on both expert integrity and legal process. Expert witnesses must remain within the limits of their expertise and courts, in turn, must have mechanisms capable of testing the evidential foundation of expert opinion before it acquires forensic force before a jury.”

Mr Feeny wrote on his website that they hoped the article “contributes to the continuing discussion about improving the reliability of criminal trials involving scientific and medical expertise”.




Blog


The AI question law firms should answer before a client asks it

Most law firms we speak to are already using AI, whether or not they have approved it, because fee-earners are doing it anyway. The tool people settle on is often Claude.


From chat to Cowork: preparing your law firm for agentic AI

AI is evolving quickly. For law firms still working out how to get consistent value from the tools they already have, the conversation has moved on again to agentic AI.


Why good governance matters when managing Law Society forms

Everyday legal transactions rely on Law Society forms. But they are often treated as static templates rather than what they really are: living documents that change over time.


Loading animation