New facts raise more doubt: did Sally Clark really kill her babies?

Strong evidence that may point to the innocence of Sally Clark, the solicitor serving a life sentence for killing her babies, was not disclosed in court during her infamous "73 million to one" double murder trial.

The development highlights growing fears of a reversal of the presumption of innocence – a cornerstone of the English legal system – in cases where mothers are accused of killing their babies.

Clark's husband, Steve, said: "We have discovered new medical evidence which wasn't disclosed to the defence at the time of the trial which appears to reveal a natural cause of death for one of our babies."

Tests were carried out on the blood and tissue of the second baby to die, Harry, on samples taken by a Home Office pathologist, Dr Alan Williams. The report containing the results – compiled in January 1998, a few days after Harry died – was not used in the trial, held in October 1999.

Mr Clark tells BBC Radio 4's File on Four programme, to be broadcast tonight: "We found about it almost by accident really, hidden away in a huge pile of papers that we finally managed to extract from Macclesfield Hospital late last year after almost two years of asking for them, and the scary thing is that it could still be there and no one would know."

The report is believed to provide compelling evidence that Harry died a natural death. If the conviction for Harry's murder was undermined, it would automatically raise doubts about Clark's conviction for the murder of her first son, Christopher.

The Criminal Cases Review Commission is looking at the case, and the Clark family expects it to be returned to the Court of Appeal.

Mr Clark remains astonished that the crucial material was not seen by the defence: "The report was not disclosed to the defence at the time of the trial, even when the jury asked a specific question about that area," he said.

At Clark's trial in October 1999 the results of the tests Dr Williams commissioned on baby Harry were not disclosed to the court.

Mr Clark said: "When I saw the report I didn't know what its significance was, but we passed it to some of the doctors who have been helping us behind the scenes and they almost said 'eureka!'. We now have statements from two eminent British pathologists in this field, stating that the second baby died from natural causes.

"I have always known that my wife was innocent and have been determined to find justice for her. After more than four years of struggle, I hope that the system will now move quickly to give it to us.

"Sally's first reaction, when I told her, was not 'thank goodness, I'm coming home', but to burst into tears and then ask, 'Harry didn't suffer, did he?'. As is so typical of the kindest and gentlest person I have ever known, her first thought was not for herself, but for her baby.

"We still have not been able to grieve properly for our little boys, and should never have been put in this appalling situation where it is we who have had to find out the true cause of death."

Mr Clark added that Dr Williams should not have been relied upon because he was not a paediatric pathologist and he pointed out that the prosecution admitted a fundamental error was made by both Dr Williams and another pathologist in one of the post-mortem examinations. In addition, Dr Williams was found to have made mistakes in another case.

The vital question now is why evidence that suggested Clark was innocent was never put to the jury. Dr Williams has declined to comment.

Professor Sir Roy Meadow, the star witness for the prosecution, testified that the two deaths were unnatural. He told the jury that there was a "one in 73 million chance" that both deaths had occurred naturally.

Sir Roy added that there were specific features in the case which made the likelihood of innocent death even more remote, saying "it could only happen once every 100 years".

The "73 million to one" figure was a statistical smoking gun. In one soundbite the jury had a compelling case against Sally Clark. The jury, some of whom were openly weeping, convicted her by a majority of 10-2, after two days of deliberations.

But the statistic was grossly inaccurate. It is more, rather than less, likely that a mother who has suffered one cot death will suffer another. The true odds were not one in 73 million but one in 60.

Despite that, the Court of Appeal ruled that the mistake did not have a significant impact on the jury's decision.

The campaign to establish Clark's innocence was bolstered when the Solicitors' Disciplinary Tribunal made the unprecedented decision not to strike her off the Law Society roll, signalling that many in the legal world were not convinced of the safety of her conviction.

For that tribunal she had prepared a video, in which she said of her ordeal: "I now suffer the minute-by-minute torture of life imprisonment, knowing, as I accept only I could know, that I did not harm my little boys and did nothing but love them."

Had the jury known about the test results that suggested Harry died of natural causes, they might not have convicted her. Other prisoners are notoriously cruel to child killers, and Mrs Clark suffered threats and some violence when she was first locked up at Styal prison near Manchester.

Mr Clark said: "It is virtually impossible for innocent mothers who have endured cot deaths to defend themselves from the charge of murder.

"The first question to my wife by the prosecution at the trial was: 'You were never cut out to be a mother, were you?' The burden of proof has been reversed in these cases. Mothers who have suffered cot death are assumed to be guilty until they can prove they are innocent.

"It has taken us, with all our resources and support, nearly three years to prove Sally's innocence; yet still she is in prison, separated from me and our surviving child, who badly needs his mummy."

There are also grave concerns about the safety of the conviction of Angela Cannings, who, like Sally Clark, was found guilty of double murder and sentenced to life imprisonment in April this year. Sir Roy Meadow's published view in such cases is that, unless proven otherwise, one cot death is a tragedy, two is suspicious and three should be seen as murder. Cannings suffered two cot deaths – and then her third baby died.

At her trial at Winchester Crown Court, one of the main prosecution witnesses, who told the jury that the deaths were not natural, was Sir Roy. Cannings continues to protest her innocence.

Terry Cannings, who, like Mr Clark, is convinced his wife is innocent, said: "Matthew was pronounced dead on 12 November 1999. The police officer knelt down, held Angela's hand and said: 'I'm sorry but I've got to arrest you on the suspicion of three deaths – Gemma, Jason and Matthew.' It was like she was shot in the kneecaps."

Dr David Drucker is a microbiologist at the University of Manchester. A colleague of his performed tests on Matthew's body, which showed that the baby's immune system was not working properly.

Dr Drucker said: "That baby had so few antibodies of an entire class of antibody that it was just about almost below the limit of detection of the tests that were being used. And I remember commenting to someone at the time if I'd had so little antibody, I would expect to be dead or very seriously ill.

"I wasn't a jury member, so I don't know what went through their heads, but I don't know how they could ignore a finding that was so important."

The latest discoveries in genetic science, Dr Drucker said, suggested that Sir Roy's view on multiple cot deaths does not make scientific sense.

With colleagues at Manchester University, he has discovered what they believe to be a "cot death gene". In layman's terms, the cot death gene is a faulty switch – it fails to switch on the baby's defence system against killer bugs. There is a gap between the mother's defence system fading out as the baby grows to around eight weeks old and the baby's own defences kicking in. And it is through that gap that a bug or infection can strike.

Sir Roy does not agree. "There is no evidence that cot deaths run in families, but there is plenty of evidence that child abuse does," he has said.

File On Four is on BBC Radio 4 at 8pm tonight.