When Angela and Ian Gay were acquitted of poisoning with salt the three-year-old boy they were adopting, it marked the end of a nightmare that began with Christian Blewitt’s death in December 2002 and the couple’s subsequent arrest.

At their first trial, which began at the end of 2004, they were acquitted of murder but convicted of manslaughter. The basis for the conviction was hypernatraemia caused by forcible ingestion of salt — at least four teaspoonfuls in one dose. There was no realistic explanation of how such an amount could have been administered.

They were sentenced to five years’ imprisonment. Like others incarcerated for crimes against children, they suffered humiliation and abuse. After 15 months the Court of Appeal quashed the conviction, ordered a retrial and both of the Gays were released on bail.

After their acquittal at the recent retrial, their solicitor, William Bache, observed: “I do not believe that cases of this sort should be brought to trial when so much doubt remained about the science involved.”

Over the past few years, this has become a disappointingly familiar tale which has recurred in the cases of Sally Clark, Angela Cannings, Donna Anthony and Trupti Patel (sudden infant death syndrome); Lorraine Harris and Michael Faulder (shaken baby syndrome); and Marianne Williams (salt poisoning). In each, the defendant parent was ultimately vindicated.

There are several failings common to many of these cases, although not all. Frequently, a prosecution is embarked upon when it is clear from the start that there is a serious division of opinion between reputable experts about the cause of particular injuries. Sometimes the dispute is as basic as what the original findings amount to. In other words, what can actually be discerned from scans, the X-rays, the histology and so forth.

The process nearly always involves at least these two stages — the findings and then interpretation; often with the added complication that the original samples have been used up, or lost and, occasionally, that the appropriate examinations were not all done in the first place.

What remains, therefore, tends to be the opinion of one expert against another. This provides a nearly impossible situation for any fact-finding tribunal whether jury and judge, or judge alone. On what basis can a nonscientist arrive at a certain conclusion (guilt) when the choice perhaps comes down to a preference in presentation? Rarely will there be exhibits that can be pored over and, even if there are, they are often unintelligible.

In the Cannings case, the Court of Appeal provided some extremely sensible guidance: Sir Igor Judge suggested that where an investigation into two or more sudden unexplained infant deaths was followed by a serious disagreement between reputable experts, then the prosecution should not be started or continued — without additional cogent evidence.

Unfortunately, this advice has not been heeded. Cases proceed in the absence of corroborative evidence. Even when experts have been discredited, cases proceed by substituting other experts who will come up with the same theory or a new one, or who stray beyond their own expertise.

It is time that the criminal justice system took a firm grip of the situation by establishing a clear, regulatory structure with obligatory protocols and guidelines. This has been raised endlessly in professional debate, parliamentary reports, reviews and Royal Commissions. But there persists an extraordinary reluctance to act.

What is needed is a national institute of forensic science to lay down standards and guidelines for experts who give evidence in court. The newly established Council for the Registration of Forensic Practitioners is not comprehensive and membership is only voluntary. There should also be a national and obligatory register of accreditation. In all cases where forensic science is involved there needs to be — as in the United States — a preliminary hearing where the trial judge assesses the admissibility of expert evidence (the Daubert Test). Essentially, the courts exercise a form of quality control. The judge decides which experts are capable of giving evidence and the areas where they are qualified to comment. This ensures that certain basic thresholds are achieved, that the theory or technique employed can or has been tested, subjected to peer-review or publication, scrutinised for any known rate of error and identifies whether there is a consensus within the scientific community.

The present deficiencies have been given a recent and stark illustration by the conviction of a so-called expert forensic psychologist. Gene Morrison was recently convicted of 20 offences involving a deception he had practised for 27 years. His qualifications had been bought from a website — he had no genuine academic or forensic psychology skills. Up to 700 cases may now have to be reviewed because of the bogus opinions he proffered.

Besides wondering why forensic scientists, police officers, solicitors, barristers and judges never asked a single fundamental question that would have exposed this individual, what the case really illustrates is systemic failure on a grand scale. Obligatory accreditation and hearings before the admission of such evidence would have weeded Morrison out 27 years ago.

An improvement in standards would also help to prevent future cases where the expert evidence is disputed from going to trial without giving effect to Lord Justice Judge’s wishes.

The author is head of Tooks Chambers and represented Angela Cannings, Lorraine Harris, Michael Faulder and Ian and Angela Gay on appeal

Explore Law

• Columnists

• Law Panel

• Law Reports

• Student Law

Times Recommends

• The best legal yarns this Christmas
• Legal school 'earned £1m by exploiting bar course'
• Axe may fall on 5,000 more City lawyers