The Appeal Court's decision to release Donna Anthony has once again highlighted the controversy over the science used to take parents of cot death victims to court.
Some families may have a higher risk of cot death
BBC News looks at the issues facing scientists on this emotive issue.
Millions of people in this country are prepared to go into a shop, and part with a pound because they believe that it's possible to win the National Lottery - despite odds of many millions to one against success.
Yet a jury is asked to believe that potentially shorter odds - that natural cot death can strike three times in one family - are impossible.
So says "Meadow's Law", espoused by Crown expert Sir Roy Meadow, which says that three deaths "must be murder - unless proved otherwise".
Sir Roy gave evidence at Donna Anthony's original trial, suggesting that her two children, Jordan, aged 11 months, and her son Michael, aged four months, had been smothered. She has always said her they had died from cot death.
Finding the needle
In this country, approximately 600 children each year die suddenly and unexpectedly at some time between their first week of life, and their first birthday.
In half of these cases, a clear medical reason for the death is found at post mortem - which leaves 300 sets of parents with no explanation, just the term "sudden infant death syndrome" (SIDS) to cling to.
However, concealed among this distraught group are a handful of parents who know exactly why their baby died - their abuse or maltreatment.
Somehow the police and legal system is expected to root these out and bring the offending parent to justice.
Among the thousands of families in the UK who have suffered a cot death, those parents who have suffered a second - or perhaps even a third stick out like a sore thumb.
As such, they draw special attention from the medical profession - and the police.
If these cases are brought to court, part of the argument advanced by the Crown is that since cot death is a relatively rare event, the chances of a subsequent, or even third cot death are so rare as to point towards abuse.
The presumption of innocence becomes one of guilty.
Cot death genes
Some experts say it is perfectly possible that lightning can strike two, or even three, times in the same family.
There may well be cot death genes, as yet undiscovered, they say, which mean that in some families, the chance of repeated cot deaths are much higher.
Professor Jean Golding, from the University of Bristol - who has examined how genetic factors play a role in infant death, said that much remained to be learned about cot death.
"We are very much in the infancy of understanding how the genes for cot death work.
"What the mother has been asked to do is prove her innocence without the scientific means to do it."
In the now infamous case of Sally Clark, a solicitor whose conviction for killing two baby sons was later quashed on appeal, prosecution expert Sir Roy Meadow quoted the chances of a second cot death in this middle class family as one in 73 million.
The chances of this being a natural cot death, he said, were "vanishingly small".
This provoked a storm of protest from statisticians - whose calculations put the true figure far lower.
One made a simple analogy. If an archer fires a number of arrows randomly at a wall, the chances of twice hitting a pre-drawn bullseye on a target on that wall are very small.
However, if the archer fires his arrows at a blank wall, and then the bullseye was drawn on later around two arrows that happened to be close together, there would be far more chance of hitting the bullseye.
That, he said, was exactly what had happened here - in a collection of cot deaths, police had sought out two in the same family and drawn the bullseye around them.
One medical expert said that there may be another reason why figures giving the probability of multiple cot deaths may be misleading.
If you take the whole of the UK population, simple maths tells you that one baby in every 1,600 born will die from cot death.
But, unless you believe that everyone in Britain is exactly the same, things are not as simple as this.
For example, if you looked at the risk of a cot death among couples in which the mother is under 27, has more than one previous child, with both parents unemployed and are smokers, it increases to approximately one in 200.
If none of these factors apply, then the chance falls to one in 8,500.
However, even if you break people down into broad stereotypes, it's still impossible to reliably calculate the risk to any one individual.
Dr Richard Wilson, a paediatrician and trustee of the Foundation for the Study of Infant Deaths, said: "You simply can't take a group risk and apply it to the individual family. It's just wrong."
Dr Wilson told BBC News Online that while studies showed that families who lost one baby to cot death were at increased risk of a recurrence, it still was not entirely clear why this might be happening.
He said data showed that in families which had lost one child to cot death, subsequent babies were at an increased risk of dying from the same cause, or from a variety of natural causes, such as meningitis.
"There are still some unknown factors out there that haven't been found."
It is difficult to prove - for prosecution or defence - that, in the face of these uncertainties, an individual cot death was either definitely due to abuse, or definitely not.
Call for change
Part of the problem for the defence lies within the current coroner's system, says Dr Wilson.
If a second or third cot death arouses suspicions, it is often hard to prove anything about the first or second death - because the standard post-mortem is not designed to spot obscure natural causes of death.
"We need to change the whole coroner's system - they are not set up to find a natural cause of death.
"The majority of these deaths are through natural causes."
Juries in all three recent cases were told that the deaths of three children consecutively effectively ruled out natural causes.
But, each year in the UK, a handful of families who lose a baby to cot death lose another child in a similar fashion.
Researchers believe that roughly a quarter of these are due to abuse, about a half to an identifiable natural cause, and the final quarter unexplained - a natural cause that can't be discerned.
It is hard for a jury to grasp that such a tragic event could happen not just once, but twice, and then three times to one family - with no abuse.
But leading experts say that this cannot be ruled out.