I do not believe Angela Cannings murdered her babies. Not the two whose murders she was charged with, nor the first, Gemma, recorded as a cot death. Yet I do think Cannings, whose lawyers filed papers for an appeal last Friday, probably caused her babies' deaths. It's the word "murder" I take issue with. The most obvious explanation - one which was never put before the judge and jury - is that she acted while in the grip of a severe post-partum depression. Mrs Justice Hallett, the judge who presided over her trial last month, seems to have had little doubt as to what lay behind the smothering of seven-week-old Jason in 1991 and four-month-old Matthew in 1999.

Left with no choice but to jail Cannings for life once the jury delivered its murder verdict, the judge made her views clear: "There was no evidence before the court that suggested there was anything wrong with you when you killed your children. I have no doubt that for a woman like you to suffocate these babies there must have been something seriously wrong with you. You wanted these babies and you cherished them. It's no coincidence, in my view, that you committed these acts in the weeks after their births. It is not my decision when you will be released, but I intend to make it known in my remarks that in my view you will never be a threat to anyone in future."

Hallett was saying she thought Cannings should have been convicted of the lesser crime of infanticide - the killing by a mother of her own child under the age of one year. The Infanticide Act 1938 lays down (and there are similar laws in at least 29 countries) that a woman who kills her child "while the balance of her mind was disturbed by reason of the fact that she had not fully recovered from the effect of giving birth" should not be found guilty of murder - which, in 1938, carried a death sentence. Women convicted of infanticide are rarely jailed. Judges usually put them on probation with a condition that they undergo treatment.

The Court of Appeal was told in 1989 that of 59 women convicted of infanticide between 1979 and 1988, none had been imprisoned; all were given probation, supervision or hospital orders. Of 46 women convicted between 1989 and 2000, only two were imprisoned. But what if the woman is adamant, as Cannings was, that she did not kill her babies? Then the court never gets to hear evidence from psychiatrists that the balance of her mind was disturbed at the time.

Helena Kennedy, the Labour peer and QC, who has defended several women charged with killing their children, says: "The problem in these cases is that unless a woman says, 'I did it', you can't mount a psychiatric defence. You can't go behind her back and get a psychiatric report. The problem is these women can't admit it to themselves, and they certainly can't admit it to their partners. I had a lovely woman in one case, a schoolteacher. It was about somehow feeling she couldn't ever be a good enough mother. In another case, a woman with sickle-cell anaemia drowned her baby in the bath.

"Deep in their hearts, these women feel they can't be good mothers and the child would be better off not being raised by them, because they're going to fail them. The problem for the lawyer is getting the woman to acknowledge it herself. It's complex and cuts across all classes."

Psychiatrists say the effect of the hormonal upset that comes with pregnancy and childbirth can trigger a range of mood disorders, from the mild "baby blues" which many experience, through (in women who are susceptible) the more severe post-partum depression, to the rare post-partum psychosis, in which a woman can lose her grip on reality.

Defence lawyers for Andrea Yates, the Texas mother who drowned her five children last year, argued that she was suffering from post-partum psychosis after the birth of her fifth baby when she held all five under the bathwater and then lined their bodies up on a bed. Texas has no infanticide statute and Yates, who was convicted of murder in March, faces 40 years in prison before she can apply for parole.

Severe post-partum depression is thought to have been the trigger that drove Suzanne Killinger-Johnson, a Toronto GP and psychotherapist, to make three attempts in August 2000 to jump in front of an underground train while cradling her six-month-old son, Cuyler, in her arms. The third attempt succeeded; her son was killed instantly and she died nine days later. Friends and co-workers said she seemed to "have it all" and showed no signs of depression. But she was believed to have discontinued the anti-depressants she had been taking, for fear of affecting the baby through her breast milk.

Post-partum depression captured the attention of the US and Canadian media in the wake of the Yates and Killinger-Johnson cases. In Britain, women who kill their children have been a talking point since solicitor Sally Clark was convicted in 1999 of the murder of two of her babies. Her failed appeal a year later, and the campaign to prove her innocence, along with Cannings' trial and conviction, have kept the issue in the media spotlight.

Last week an unnamed woman was bailed on charges of murdering three of her children and the attempted murder of a fourth. Yet the media debate seems to focus on only two possibilities: murder or cot death. That may have been understandable in the context of the Clark and Cannings cases, because neither raised infanticide as a defence. Both insisted they had not killed their babies. But now that the verdicts have been delivered, their murder convictions - the only ones open to the juries if they were sure the babies died at their mothers' hands - should not obscure the fact that there is another alternative: a loving mother can kill her baby and yet not be responsible for that baby's death.

I express no view on who or what caused the deaths of Sally Clark's babies, 11-week-old Christopher and eight-week-old Harry. But it's particularly strange, in her case, that the debate should be polarised between murder and cot death, since, as the trial went on, it became clear that even the experts called for the defence were not classifying the deaths as sudden infant death syndrome (SIDS), the technical term for cot death. As the appeal court judges put it: "The defence doctors were to agree that neither of these deaths was a SIDS because of the suspicious circumstances surrounding them - the children suffered injuries which were not sufficiently explained."

The media have made much of the unfortunate and misleading statistic cited by prosecution expert Professor Roy Meadow that the chances of two cot deaths in a family like the Clarks' was 1 in 73 million. Juries are notoriously prone to prejudice, and that statistic may well have prejudiced the jury against Clark. But how relevant was it when even the defence experts accepted that neither death was a true SIDS death?

In any event, the trial judge warned the jury, "We do not convict people in these courts on statistics," and told them they must not take into account the circumstances of the second death in deciding guilt unless they were first convinced that Clark had killed one baby. And he pointed out that a distinguished professor called by the defence had made the point that the statistic cited by Meadow did not take account of the possibility of familial factors influencing a first cot death which might make a second one more likely.

Disregarding the Meadow statistic altogether, the appeal court concluded "there was overwhelming evidence of the guilt of the appellant [Clark] on each count". Guilty of the killings she may or may not be. But if so, is she guilty of murder? We may never know. While the mandatory life sentence for murder remains, judges have no choice but to pass life sentences on mothers who deny killing their babies but are convicted of murder. Maybe a woman denies it because she genuinely didn't do it, or maybe she can't bring herself to admit it. Either way, it is a miscarriage of justice.

See also:
Angela Joyce Cannings: Another miscarriage of justice - Excellent article by the Portia campaign