England and Wales Court of Appeal (Criminal Division) Decisions

IN THE SUPREME COURT OF JUDICATURE
COURT OF APPEAL (CRIMINAL DIVISION)
ON APPEAL FROM CROWN COURT AT WINCHESTER
MRS JUSTICE HALLETT AND A JURY

		Royal Courts of Justice Strand, London, WC2A 2LL
		19th January 2004

Lord Justice Judge:

1. Angela Cannings was born in May 1963, the daughter of Stephen Connolly. After her marriage to Terry Cannings, they had four children, Gemma, born on 14th August 1989, Jason, born on 25th April 1991, Jade, born on 15th January 1996 and Matthew, born on 5th July 1999. Three of these children, Gemma, Jason, and Matthew died in infancy.
2. On 16th April 2002 in the Crown Court at Winchester, before Mrs Justice Hallett and a jury, after a conspicuously fair and balanced summing up, the appellant was convicted of two counts of murder, count one relating to Jason and count two relating to Matthew. It was elicited at trial that the appellant had also been charged with Gemma's murder, but that allegation did not proceed. Before his death Jason, and the surviving child, Jade, too, suffered an "Acute" or "Apparent Life Threatening Event" (ALTE). Before his death, Matthew was thought to have suffered an ALTE, but the preponderance of specialist evidence suggested that this was no more than a troublesome and worrying "episode" rather than a true ALTE.
3. The relevant chronology is straightforward. Gemma died on 14th November 1989. She was then aged 13 weeks. Jason suffered an ALTE when aged 6 weeks, but he died a week or so after discharge from hospital, on 13th June 1991. Jade suffered an ALTE on 1st April 1996, when aged 11 weeks. She made a full recovery. On 3rd November 1999 Matthew, aged 17 weeks, suffered his episode. He was discharged from hospital, but died some nine days later.
4. The Crown's case was that the appellant had smothered both Jason and Matthew, intending to kill or do them really serious bodily harm by obstructing their upper airways. To support that allegation it was suggested that Gemma's death, and each of the ALTEs, were also consequent on smothering by the appellant, and that the deaths of Jason and Matthew formed part of an overall "pattern". The appellant, a woman of good character, described as a loving mother, apparently free of personality disorder or psychiatric condition, consistently denied harming any of her children. It was not suggested that they had been harmed by anyone else. Her case was that the deaths were disastrous, but natural, even if unexplained incidents, to be classified as Sudden Infant Death Syndrome (SIDS), known colloquially as cot death. In expressing the issue in this way we do not overlook the comforting ease of applying such a description to an extremely complicated subject, nor that specialists do not treat SIDS and cot death as synonymous.
5. Without medical evidence about the appellant's mental state, a verdict of infanticide was not open to the jury. Following conviction of murder, sentences of life imprisonment on each count were mandatory. Hallett J expressed her concern at the "kind of injustice that can be caused by mandatory sentences". Honouring, as she had to, the verdicts of the jury, and acknowledging the absence of any medical evidence relevant to the appellant's state of mind, she observed:

"I have no doubt that for a woman like you to have committed the terrible acts of suffocating your own babies there must have been something seriously wrong with you. All the evidence indicates you wanted the children, and apart from these terrible incidents you cherished them, so in my layman's view, it is no coincidence that these events took place within weeks of your giving birth. It can, in my view, be the only explanation for why someone like you could have committed these acts when you have such a loving and supportive family."

6. This is an appeal against both convictions. On 10th December 2003 we quashed them. We shall now give our reasons.

The issue

7. The issue before the jury is easily defined. The deaths of Jason and Matthew were either natural sudden infant deaths, or unnatural sudden infant deaths. In the vast majority of cases of murder, there is no doubt that someone has caused or contributed to the death of the deceased in some way, for example by gunshot wound, or knife, or repeated blows, or indeed a single blow. In a few cases issues may arise whether the defendant's actions caused the death, or whether some other underlying or intervening cause contributed to or was responsible for it. Unusually, but not uniquely, the primary and stark question in the present case was whether either of these children was killed at all, their deaths resulting from deliberate actions by their mother to deprive them of breath, and successfully achieved without revealing any obvious physical manifestation of a killing. We are not blind to the fact that, sadly, these catastrophes happen: mothers, and fathers also, do sometimes kill their infant children. Nevertheless the crucial question here remains whether the deaths of either of these two baby boys were indeed the result of crime.
8. The unavoidable reality is that some infant deaths remain "unexplained" or "unascertained". Professor Berry, a consultant pathologist called by the Crown, identified three categories of death which were encompassed within this last word, "unascertained". It includes deaths which are natural and explained (i.e. a similar mechanism to sudden infant death syndrome, but not fitting the strict case definition), natural and explicable (but the cause was not found) or unnatural (accidental or deliberate). Whatever form of categorisation, and there are several, we must emphasise what is self-evident, that only proof of a death falling within the second half of the last of Professor Berry's categories, an unnatural death resulting from deliberate action, can justify a conviction for murder.
9. The convenient acronym SIDS requires a little amplification, particularly in relation to the last "S", which stands for syndrome. Treating the problem as a syndrome tends to obscure the fact that sudden unexplained infant deaths occur in different circumstances, and some may be multi-factorial, the result of a coincidence of processes which, taken in isolation, would not necessarily cause death. No underlying condition for every death categorised as SIDS has been identified. The critical point of each such death is that it is indeed unexplained, and that its cause or causes, although natural, is or are as yet unknown. SIDS does not apply to deaths, or if already attributed to SIDS, ceases to apply to deaths which are clinically explicable or consequent on demonstrable trauma. In each SIDS case the mechanism of death is the same, apnoea, loss of breath or cessation of breathing. In the true SIDS case we do not know why the particular infant's breathing stopped. All we know is that for some unexplained reason it did. One obvious reason for loss of breath is smothering or some deliberate interference with the infant's normal breathing process. However the same process, with the same result, also occurs naturally. In the present context ALTEs are SIDS in which no death has actually resulted. Again, approaching the issue in layman's terms, ALTEs may be described with reasonable accuracy as sudden infant acute or apparent life threatening events. They represent the consequence of an unexpected and unexplained temporary loss or cessation of breathing, which has occurred naturally. Whether the cause is natural or harmful interference, the body of the infant undergoes great stress. If the event is the result of deliberate interference with the infant's breathing it is not a true ALTE: it is attempted murder or attempted grievous bodily harm.

Two critical problems

10. It would probably be helpful at the outset to encapsulate different possible approaches to cases where three infant deaths have occurred in the same family, each apparently unexplained, and for each of which there is no evidence extraneous to the expert evidence that harm was or must have been inflicted (for example, indications or admissions of violence, or a pattern of ill-treatment). Nowadays such events in the same family are rare, very rare. One approach is to examine each death to see whether it is possible to identify one or other of the known natural causes of infant death. If this cannot be done, the rarity of such incidents in the same family is thought to raise a very powerful inference that the deaths must have resulted from deliberate harm. The alternative approach is to start with the same fact, that three unexplained deaths in the same family are indeed rare, but thereafter to proceed on the basis that if there is nothing to explain them, in our current state of knowledge at any rate, they remain unexplained, and still, despite the known fact that some parents do smother their infant children, possible natural deaths.
11. It will immediately be apparent that much depends on the starting point which is adopted. The first approach is, putting it colloquially, that lightning does not strike three times in the same place. If so, the route to a finding of guilt is wide open. Almost any other piece of evidence can reasonably be interpreted to fit this conclusion. For example, if a mother who has lost three babies behaved or responded oddly, or strangely, or not in accordance with some theoretically "normal" way of behaving when faced with such a disaster, her behaviour might be thought to confirm the conclusion that lightning could not indeed have struck three times. If however the deaths were natural, virtually anything done by the mother on discovering such shattering and repeated disasters would be readily understandable as personal manifestations of profound natural shock and grief. The importance of establishing the correct starting point is sufficiently demonstrated by this example.
12. Before this trial began, this Court, differently constituted, had decided that the fact of three deaths (that is those of Gemma, Jason and Matthew), as well as each of the ALTEs, provided admissible evidence relevant to each count. There could be no denying that the death of three apparently healthy babies in infancy while in the sole care of their mother was, and remains, very rare, rightly giving rise to suspicion and concern and requiring the most exigent investigation. Given the overwhelming consensus of medical evidence, it would indeed have been an affront to common sense to treat the deaths of the three children and the ALTEs as isolated incidents, entirely compartmentalised from each other. All the available relevant evidence had to be examined as a whole. Nevertheless a degree of caution was necessary to avoid what might otherwise have been the hidden trap of taking the wrong starting point. If, for example, at post mortem it was positively established that Matthew's death had resulted from natural causes, the situation reverted to precisely where it stood before he died. The concerns which would have arisen as a result of his death – as the third in the sequence – would have been dissipated. There would have been a positive innocent explanation for the death, which would no longer be a SIDS, and might help to confirm that the earlier deaths were indeed natural deaths. Equally, if there were unequivocal evidence that one of these deaths, or even one of the ALTEs, had resulted from deliberate infliction of harm by the appellant, that would be likely to throw considerable light on the question whether the other deaths, or ALTEs, resulted from natural or unnatural causes. If, after full investigation, the deaths, or ALTEs, continued to be unexplained, and there was nothing to demonstrate that one or other incident had resulted from the deliberate infliction of harm, so far as the criminal process was concerned, the deaths continued properly to be regarded as SIDS, or more accurately, could not properly be treated as resulting from unlawful violence.
13. Reverting to the two possible approaches to the problems posed in a case like this, in a criminal prosecution, we have no doubt that what we have described as the second approach is correct. Whether there are one, two or even three deaths, the exclusion of currently known natural causes of infant death does not establish that the death or deaths resulted from the deliberate infliction of harm. That represents not only the legal principle, which must be applied in any event, but, in addition, as we shall see, at the very least, it appears to us to coincide with the views of a reputable body of expert medical opinion.
14. As we have already emphasised, the question in this case was whether there were any crimes at all, and whether there were, in the legal sense, any victims of crime. As we shall see, there was no direct evidence and very little indirect evidence to suggest that they were, and there was further evidence which tended to suggest that they were not. Mrs Cannings was not contending, for example, that someone else had harmed her babies, or that she had caused their injuries accidentally. If so, the jury might reasonably have thought that three infant deaths as a result of accident, all at the hands of their mother, would be highly improbable, or equally, that her presence at home, on her own, when each of these "accidents" occurred, extinguished any realistic possibility that someone else was responsible. Mrs Cannings' defence was simple: she had done nothing to harm any of her children. Although she was contending that the deaths were natural, notwithstanding specialist evidence called on her behalf at trial, she could not explain them, and she was not seeking to offer an explanation of her own. And, unusually, she was doing so in the very special context that medical specialists, both domestically and internationally, continue to acknowledge that the death of an infant or infants at home can simultaneously be natural and unexplained, even by them.
15. In recent cases which attracted considerable public concern and interest, two mothers were charged, and in one case, Sally Clark, convicted, and the other, Trupti Patel, acquitted of murder, following the deaths of their infants. The verdict in Trupti Patel decided no point of principle: on the evidence the jury was not sure of guilt. Lest anyone seek to read anything deeper into that observation, and imply that we are commenting one way or another on the verdict, we emphasise that that is all that is ever decided by a "not guilty" verdict. Nowadays, we read and hear much about a new concept, "found innocent": that is not a verdict known to the law.
16. As is well known, the conviction of Sally Clark has been quashed. Save superficially, however, this appeal is dissimilar, and raises different issues. Unlike the Court of Appeal Criminal Division in that case, we have not been presented with evidence of apparent misconduct and serious non-disclosure by an expert witness, Dr Williams, called by the Crown, which came to light after conviction. Of itself, that would have been sufficient for the conviction to be quashed. In addition, expert evidence describing statistical probabilities was also severely criticised. That evidence was given by an expert witness of great distinction, if not pre-eminence in this field, Professor Sir Roy Meadow, whose evidence would undoubtedly have carried great weight with the jury which tried Sally Clark. If it were flawed, as it was, the safety of the jury's decision was further called into question. Professor Meadow's evidence in the present case did not extend to the flawed statistical evidence presented to the jury during the trial of Sally Clark. The present convictions therefore cannot be quashed on either or both of the grounds relied on in her appeal, and the observations on the facts in the CACD in that case were case-specific, and not otherwise of general application to the present appeal.
17. Mr Michael Mansfield QC submitted that although Professor Meadow did not expressly give statistical evidence, he offered it to the jury by implication. We shall examine the relevant evidence later in this judgment. On any view however, we must reflect on the likely impact on the verdict in the present case if Mr Mansfield had been able to cross-examine Professor Meadow, and undermine the weight the jury would inevitably attach to his evidence, by exposing that, notwithstanding his pre-eminence, at least part of his evidence in the Sally Clark case was flawed in an important respect. To some extent at least, Professor Meadow's standing as a witness would have been reduced. Therefore the flawed evidence he gave at Sally Clark's trial serves to undermine his high reputation and authority as a witness in the forensic process. It also, and not unimportantly for present purposes, demonstrates not only that in this particular field which we summarise as "cot deaths", even the most distinguished expert can be wrong, but also provides a salutary warning against the possible dangers of an over-dogmatic expert approach.
18. The word "dogma" was used in his evidence by Dr Rushton, a consultant paediatric and perinatal pathologist, to record his unhappiness, and that of some of his colleagues at "a set of rules being laid down", and, because they were felt to be over-restrictive, followed with reluctance. In evidence, Dr Rushton repeated part of the contents of his report. It included this passage:

"However, in a family with a history of this type, current dogma is that an unnatural cause has been established unless it is possible to demonstrate an alternative natural explanation for these events."

That dogma encapsulates what in a criminal case we have described as the first, and we would emphasise, the erroneous approach.

19. By way of linking the word "dogma" with SIDS, Dr Rushton went on:

"The definition of SIDS, for instance, .... says usually the definitions include babies up to one year of age; it used to be two years of age but it has been decided by experts in the field meeting that one year is the limitation. I would suggest that biology does not behave like this and [in] any event, sudden unexplained deaths occur throughout life – they also occur in adults. So there is not a magical cut-off point at one year of age."

20. Professor Golding, professor of paediatric and perinatal epidemiology, made the same point in a different way when she said in evidence:

"I think, to put it in context, there is a fashion nowadays that if you have more than one sudden infant death the next one must have been killed deliberately, and that is something that people within the paediatric profession have taken on board without sufficient evidence. Certainly, obviously, there are cases where it happens, but (in the vast majority) there is no evidence of that."

If that is the fashion, it must now cease.

21. Professor Golding continued:

"... there are a few cases where it (smothering) appears to have happened, but it is by no means clear that the claims that so many families where more than one sudden infant death has occurred are due to smothering. The results haven't been subjected to what I would call an appropriate statistical analysis. They are mostly a hunch that the paediatrician or whoever is looking at it might have but it is not based on any scientific foundation."

22. These observations serve to highlight the second problem which can arise in this case, and cases like Sally Clark and Trupti Patel. We have read bundles of reports from numerous experts of great distinction in this field, together with transcripts of their evidence. If we have derived an overwhelming and abiding impression from studying this material, it is that a great deal about death in infancy, and its causes, remains as yet unknown and undiscovered. That impression is confirmed by counsel on both sides. Much work by dedicated men and women is devoted to this problem. No doubt one urgent objective is to reduce to an irreducible minimum the tragic waste of life and consequent life-scarring grief suffered by parents. In the process however much will also be learned about those deaths which are not natural, and are indeed the consequence of harmful parental activity. We cannot avoid the thought that some of the honest views expressed with reasonable confidence in the present case (on both sides of the argument) will have to be revised in years to come, when the fruits of continuing medical research, both here and internationally, become available. What may be unexplained today may be perfectly well understood tomorrow. Until then, any tendency to dogmatise should be met with an answering challenge.
23. This troublesome aspect of cases such as this is well illustrated by post-trial material which we have received in evidence. Each paper is awaiting publication. The first, the Sudden Unexplained Infant Death (Europe): findings of the Europe Concerted Action on SIDS (ECAS), has been accepted for publication in The Lancet. It is anticipated that the second, entitled Repeat Sudden Unexpected And Unexplained Infant Deaths: Natural or Unnatural, based on analysis of the Care Of Next Infant Program (CONI) research to determine the probability that a second infant death in the same family may be natural or unnatural, will, subject to minor corrections, shortly be published. Professor Berry described this paper as a "long-awaited" study. Although not yet formally "accepted" we received the study in evidence. One passage, at 9.5.2, provides a clear illustration of the speed with which knowledge in this particular field is developing.

"In the CONI study there were two families in which both deaths were attributed to the same condition (one … VLCAD, and one prolonged QT syndrome). In both families, diagnosis was assisted or confirmed by the birth of a third child identified with the same condition. Rib fractures, attributed to resuscitation, were found in the VLCAD CONI infant. A few years ago these deaths would have been totally unexplained. Both families would probably have had a third unexplained death had the underlying cause not been identified and treated, and at least one of the parents might have been suspected of murder."

24. In this context we should perhaps also note the concerns expressed by Dr Paul Johnson, consultant clinical physiologist and Director of Maternal Infant Healthcare and Tele-monitoring Research Centre at the Women's Centre at John Radcliffe Hospital, Oxford. He has for many years held appointments both in the United States and here which have involved him in research into perinatal pathology, paediatric neonatology, cardiovascular medicine and physiology, as well as developmental neurophysiological disorder, including respiratory and cardiac disorders. The extent of his current work is well-illustrated by the fact that he has developed a unique computerised tele-monitoring system and service for infants, with one of the largest databases for cardio-respiratory function during sleep in infants aged under one year.
25. Dr Johnson accepts without hesitation that parents who appear to be affectionate and caring towards their infants sometimes kill them. He described the impact on him personally of experiencing infanticide "first hand" when he was training in California. He regarded three infant deaths in apparently healthy infants in the same family as "extraordinarily unusual". He did however lay emphasis on the apparent good health of such infants. He examined the information about the deaths and ALTEs in the Cannings family. He was extremely concerned at the paucity of information about these children, without, as far as we can see, criticising those responsible for the post mortems carried out on Gemma, Jason and Matthew.
26. Dr Johnson believed that there was "a wealth of information on development, both before birth and after birth" which would have provided valuable insight into what happened to these three children. He would say, "categorically that once there has been sudden (unexplained) infant death ... you are obliged, when that mother produces a second child or foetus" to do much more than was done with this family. The purpose was to discover whether the child or children may or may not deviate in their normal "developmental processes", some of which were "extraordinarily subtle". He pointed out that unless these investigations took place before and after birth and in the first months of life, they could not be replicated. It was now known that SIDS infants may have been affected by events before birth, although the mechanism remained unknown. These include well-known factors such as maternal malnutrition, and stress, and smoking. Without going further into the details of Dr Johnson's evidence, on the broad issues currently under consideration his position was summarised in a few words, "We know a lot, we don't know enough."
27. Dr Johnson linked this level of ignorance with a concern that what was already known in the United Kingdom was less well applied here than elsewhere. In Belgium, Scandinavia, and some parts of the United States, Dr Johnson believed that the investigation into the development of Jason, Jade, and Matthew, both when their mother was carrying them, and after they were born, would have been much more precise and extensive. We see no reason to assume that these concerns, and criticisms, are misplaced. Again, when monitoring arrangements have improved, we should expect not only a greater degree of insight into the natural causes of infant death, but hopefully and more important a reduction in their number.
28. This is too tragic a case to trivialise any aspect of it. We can however illustrate this problem in a layman's context. Not so long ago, experts were suggesting that new born babies should lie on their tummies. That was advice based on the best-informed analysis. Nowadays, the advice and exhortation is that babies should sleep on their backs – Back to Sleep. This advice is equally drawn from the best possible known sources. It is obvious that these two views cannot both, simultaneously, be right. Towards the end of the hearing, we became aware of research in Australia which suggested that the advice that babies should sleep on their backs had not achieved the improvement in the rate of cot deaths attributed to the modern practice. We do not for one moment comment whether this research is valuable. Paediatricians and other experts will certainly take serious issue with it. Our point however is to highlight the fact that even now contrasting views on what might be thought to have been settled once and for all are current.
29. At a later stage in the judgment, we shall examine the results of further research of immediate importance to the particular problems posed in this case, and the evidence of Professor Meadow and Dr Ward Platt on the rarity of three unexplained infant deaths in the same family, and the "pattern" which was said to have emerged in the Cannings family. For the moment we shall confine ourselves to observing that this research powerfully reinforces the need for caution against the dangers of dogmatism at a time when our knowledge is limited and incomplete.
30. Before turning to the facts of this case in more detail, we shall begin by considering Mrs Cannings' extended family.

The Family Context

31. The parents of Mrs Cannings' father, Stephen Connolly, were Stephen and Harriet. Harriet's sister, Babs, had a number of grandchildren of which her son, Patrick McDermott, was the father. Patrick, who lives in Ireland, had seven children. Two died in infancy, one at 9 weeks, and one at 3 weeks. Both deaths were properly categorised as SIDS. The next two children suffered what appeared to have been genuine ALTEs in their early months, again both natural but unexplained incidents. Patrick McDermott is Mrs Cannings' second cousin. Professor Patton, a consultant clinical geneticist described "a very striking family tree ... with the rare condition occurring in two parts of the extended family". This view was contradicted at trial on the basis of the large number of apparently unaffected members of the common family between the two branches which had suffered infant deaths or near deaths, that is, the McDermott and the Cannings families. An important aspect of the investigation of these issues was whether the pattern of inheritance was autosomal recessive or autosomal dominant. Elucidation of this point was complicated by the information that Mrs Cannings' uncle, her father's brother Johnny, had a child who died suddenly at 8 months, whose death may or may not have been regarded as a SIDS, but which was subsequently attributed to established natural causes.
32. Since the trial further investigation has been carried out into the extended family. Further infant deaths have been identified. However on close examination it is not obvious that they were all relevant. Professor Patton himself did not suggest that any sufficient link had been made between the problem presently under consideration and one male infant, born in Ireland in 1896, who died on the day he was born, of "debility". Mrs Cannings' grandparents, Stephen and Harriet, had two further children who appear to have died in infancy. We say "appear" to have died because although there is a birth certificate showing the birth of Anastasia on 20th January 1926, there is no corresponding death certificate. It seems to us however a reasonable inference to be drawn from the fact that none of the surviving members of the family remember her, and that a younger sibling was given her name, that she must have died young. However, how young, and from what cause, is impossible to say. Anastasia had a brother, Michael. In his case, both his birth and death certificates have been found. Michael died in April 1931, aged 10 months. The cause of death was described as "debility since birth". This description would not now be used, and it is impossible to say what the certifying doctor would have meant by it in 1931.
33. Professor Patton was tentatively inclined to include these two children in the category of SIDS. He may, of course, be right, but we know nothing of the circumstances of Anastasia's death, and the expression "debility since birth", applied to Michael's death, suggests a chronic condition rather than a sudden occurrence. It is therefore far from clear to us that either of these deaths should be regarded as an earlier SIDS in Mrs Cannings' family.
34. There is a more compelling candidate for inclusion. Unbeknown to her at trial, Mrs Cannings has a half-sister, who following these convictions, made herself known to her solicitors. For quite understandable reasons, she wishes not to be identified by name. We shall respect her wishes and refer to her as JM. Stephen Connolly is the father of both these women. They share a quarter of their genes. JM has three children, two of whom suffered what appear from her description to have been possible ALTEs. The first has been excluded from consideration by Professor Patton, following a close examination of the medical records. JM also has identical twin daughters. In March 1997, when four weeks old, one of them, AM, was admitted to hospital following what was described at the time as an "apnoeic episode". This was a true life threatening event, and AM required resuscitation. At the time when it occurred, it is common ground between Professor Patton and Dr Ward Platt that the incident was rightly treated as an ALTE. Professor Patton continues to think that it was. Dr Ward Platt refers to AM's subsequent admission to hospital in May 1997 for what appeared to be a seizure, or febrile convulsions. He suggests that this provides a very plausible explanation for the earlier incident. Although we understand the argument, we are unconvinced that the diagnosis for the second incident necessarily provides the correct diagnosis for the first. In any event, in our current state of knowledge, it would be unwise not to recognise the distinct possibility that, in addition to the McDermott family, Mrs Cannings' niece, therefore the cousin of her children, who had never met, let alone lived with her, suffered an ALTE when she was 4 weeks old. This is important fresh evidence. Taken on its own, it strengthens the view expressed by Professor Patton at trial that this is indeed an unusual family tree, and that the incidence of infant death and ALTEs in this particular family are unlikely to be explained by chance alone. There may well be a genetic cause, as yet unidentified, for the deaths of the Cannings children, manifesting itself in some, but not all of the extended family, through autosomal dominant inheritance with variable penetrance. That would mean that the child in question needed only to inherit the gene from one parent to be liable to develop whatever the genetic mechanism may be.
35. We cannot avoid recording that we have been left with the strong impression that some of the mystery surrounding SIDS is likely to be dissipated when our knowledge of the impact of genes and genetics is greater than it is now. By way of example, Professor Golding described human genetics as a very complicated story, with much research yet to be done. Speaking broadly, she observed:

"Genetics at the moment is such that there are new discoveries all the time. Things that we have no idea about are being revealed every day ... There is a lot of work to be done, and once we have looked at 30,000 genes we should have a clearer idea of what we should be looking at."

In any event, for the purposes of this appeal, we are quite unable to reject the realistic possibility that in the absence of some compelling piece of evidence, whether specialist or extraneous, suggestive of the deliberate infliction of harm, there may have been a genetic cause, as yet unidentified, for the deaths and ALTEs experienced by the Cannings children.

Mrs Cannings' Children

36. We must therefore narrate the circumstances surrounding the deaths and ALTEs in Mrs Cannings' family. The expert evidence runs to several thousand pages of typescript. A mass of expert evidence from witnesses of great distinction in their respective fields was presented to the jury and to us. Listing these witnesses, their many professional qualifications, their practical expertise, and the contributions they have made to areas of research in and connected with this field would itself occupy several pages of this judgment. We shall not repeat everything contributed by each expert. Rather, we shall seek to encapsulate the critical issues in brief summary, so far as possible avoiding technical language.

Gemma

37. In evidence at trial the appellant described feeling "fantastic" when Gemma was born by Caesarean section. She was a good baby, and her development appeared normal. As was then routine, she was laid on her tummy. On 13th November she was given her triple vaccine immunisation. Current research suggests that nothing in that process provides any basis for anticipating what happened to her next day.
38. On 14th November 1989, the history of events leading up to her death, based on the appellant's account, appeared straightforward. The record shows:

"On the night of 13th the baby was restless but then seemed to return to normal and slept through the night. On 14th the mother fed the baby at 9.00 am and went into town. Came home at 10.30 am and the baby was fine. She checked her at 1.00 pm to give her a feed and found her to be lifeless."

39. The cause of death was recorded as "natural, being Sudden Infant Death Syndrome (cot death)". The pathologist, Dr Scott, could find nothing which may have caused or contributed to Gemma's sudden death, no causes which might explain it. In the course of microbiological examination she found the organism, staphylococcus aureus, in Gemma's nose and mouth.
40. The defendant was first interviewed on 16th November 1999, almost exactly ten years after Gemma's death, and a few days after Matthew's death. She said that she had found Gemma lying on her back, looking very, very white. She tried, unsuccessfully, to revive her. She called an ambulance. In a later interview on 8th March 2000, she described events during the day of 14th November 1989. She found Gemma looking white, and not moving. At trial she said that after Gemma's death the shock and trauma meant that she could not remember everything that had happened. In view of the passage of time that was not unreasonable.
41. The Crown's opening at trial was unequivocal.

"Nothing was found on post mortem examination that was suggestive of an unnatural death and there was nothing in the history of Gemma's life or in the circumstances of her death that showed her death to have been unnatural. However, no cause for her death was ever identified. The fact of Gemma's death is a background against which you will have to consider what happened to the two children you are principally concerned with, Jason and Matthew."

42. Gemma's death did indeed provide a good deal of background relevant to subsequent events. For example, as we shall see, when Jason's ALTE occurred, Mrs Cannings greeted the health visitor, "It's happened again". Without knowledge of what had happened to Gemma, that evidence would have been incomprehensible to the jury. In reality, at trial, Gemma's death assumed much greater importance than mere background to Jason's and Matthew's deaths, and Jason's and Jade's ALTEs, and whether they resulted from the deliberate infliction of harm. As we shall see, Professor Meadow attached huge importance to the fact of three infant deaths in the same family. With hesitation, Dr Ward Platt would have been content to accept that, taken on its own, Gemma's death fell within the description SIDS. He did however say:

"Question marks have to hang over Gemma. Again we do not have in clinical terms the evidence to suggest hers was anything other than a natural death but when one steps back from the situation it has been repeatedly found that when everything comes out, in fact the first death turns out not to have been natural."

43. We can illustrate the development of the Crown's case by this passage from the cross-examination of Professor Golding.

"Q. Are you saying then that we should regard these three deaths as possibly being entirely random?

A. I can't see any evidence that they shouldn't be ... by random doesn't mean, you know, a bolt from the blue; it means that this is a normal sort of pattern for repeated cot deaths, or, if you had a cot death and took somebody else's cot death and somebody else's cot death and put them together you would get that sort of pattern.

Q. Well, taking three cot deaths from three different families and putting them together is surely quite a different thing to taking three deaths within one family?

A. It is different only that it is from one family.

Q. Are you saying, therefore, that these deaths may not be linked in some way by some common cause?

A. No. What I am saying is that the fact that they are from one family does suggest that there is something else happening that would be responsible for putting that family at higher risk. ...

Q. It is not just the three deaths that have to be looked at when considering the history of the four Cannings children, is it?

A. No, everybody has put acute life-threatening events together with them.

Q. Yes. Do you?

A. I see them as part of the pattern certainly, and definitely knowing that there is good evidence that children who have apparent life-threatening events are at greater risk of sudden infant death syndrome. ...

Q. In this family we have a total of six events: three deaths, three non-fatal events?

A. Yes.

Q. Affecting all four children. If it is right that no cause has been found for any of them, they are six unexplained events?

A. I don't want to comment on whether it is right or not that no cause has been found ... but these children seem particularly susceptible."

44. This cross-examination was perfectly legitimate. If we may say so in passing, the responses were equally robust. The passage we have quoted, and others to be found in the transcript, do however illustrate that Gemma's death was no longer being viewed simply as "background": it encapsulated the fundamental basis of the Crown's case against Mrs Cannings, dependent on the "pattern" of events in this family, and the extreme rarity of three infant SIDS in the same family. Our earlier warnings about the danger of taking the wrong starting point or making inappropriate assumptions of guilt do not mean that the sequence of events was irrelevant. As we have said, the deaths and ALTEs were not and could not be treated as "isolated incidents, entirely compartmentalised" from each other. Defence as well as prosecution experts were troubled by the sequence. That is why each case had to be investigated with great care. Gemma's death formed part of this troublesome sequence, and so far as practicable, was properly investigated at trial. We can perceive no ground for complaint.
45. In the event, we must record that there was not a scintilla of direct objective evidence that Gemma's death was anything other than SIDS. The Crown's opening was correct.
46. Mrs Cannings and her husband did not think about the possibility of another baby for some little while because of her sense of shock at Gemma's death.

Jason

47. When she became pregnant with Jason, the appellant was given instructions in resuscitation techniques and provided with an apnoea alarm. This is a well known device, which detects the motion of a baby's breathing. When none is detected, after a pre-set period, the alarm sounds. Without diminishing its potential value, it is not, as some think, a machine which prevents an infant death.
48. On 25th April 1991 Jason was born by Caesarean section. At birth his hips were dislocated. The Aberdeen splint required to manage this condition meant that, as was in any event customary at the time, he was laid on his tummy rather than his back.
49. In her evidence, the appellant described Jason as a "beautiful" baby. She cared for him, following her mother's example, among other things making sure that windows were appropriately opened to air the home.
50. The appellant was expecting a visit from the health visitor, Mrs Peacock, during the morning of 4th June 1991. When Mrs Peacock arrived, the appellant answered the door, saying, "It's happened again". Mrs Peacock followed her into the flat. Apart from the appellant herself, no-one else was present, and Jason was lying on the double bed "white and apparently lifeless". The health visitor resuscitated him, telling the appellant to telephone for an ambulance and for her husband. Before resuscitating him, she had checked that the baby's mouth was clear. She found no fluid or vomit in it. After a while, Jason gasped and seemed to respond. She picked him up, and patted his back, and stimulated his breathing. Thereafter, she had to keep breathing air into him whenever he seemed to be drifting off. She did so successfully.
51. She took the baby to his mother who seemed very shocked, and was sobbing. At one stage she heard the appellant in the bathroom, retching or vomiting.
52. In due course, Mr Cannings returned home. The ambulance arrived, and the baby was taken to hospital, followed by his mother and father. On arrival at the accident and emergency department this account of the incident was recorded:

"10.15 am. Apnoea alarm – mother finds baby pale and limp. Health visitor … arrived coincidentally at the same time. Resuscitated the baby."

53. Jason's condition on arrival was noted as "crying healthily – good movement and tone. Breathing spontaneously – no stridor/wheeze. Pale. Shut down. Mottled. Cool peripherally." His breathing was irregular, and described as "grunting", suggesting but not conclusive of a respiratory problem. Part of the base of the right lung was collapsed. His fingers and toes, legs and arms were very cold, and there was poor blood flow to the baby's skin. His temperature was 35^o; significantly lower than normal. His temperature dropped a little further, but then started to recover.
54. Investigation revealed a raised white cell blood count, raised platelet count, and raised blood glucose. It was common ground that these findings in Jason were consistent with a baby who had been under stress, or unwell, but they were non-specific or unspecific on the question whether or not the stress was the result of the deliberate infliction of harm. The first urine test showed the presence of blood and protein, again suggestive of kidney stress, but again unspecific as to cause. Some glucose was also found, but the subsequent urine sample was normal for glucose. In the umbilical swab staphylococcus aureus was discovered, and an antibiotic prescribed. Bio-chemical changes in Jason, potentially important to the possible involvement of the vagus nerve, were also noted.
55. Full infection screening was carried out. Nothing untoward was found. The baby's treatment at hospital followed conventional lines. By 4.00 pm, his condition had stabilised. A physical examination by the consultant paediatrician, Dr Marshall, revealed nothing significant. To stabilise circulation, a drip was used. When it was noted that Jason had passed some rather loose motions, his feeding was withheld for a short time until he was settled. By 6th June, Jason was well, and he was discharged.
56. Hospital staff who observed Mr and Mrs Cannings described them as loving and caring towards Jason, and handling him well. When Jason was discharged, the consultant paediatrician spoke to them, explaining how to give mouth to mouth resuscitation, and cardiac massage. He thought that Mr Cannings understood these instructions, but was unsure whether the appellant was fully taking in everything he said. A care plan was drawn up. A note was made that Mr and Mrs Cannings were very upset and anxious about Jason because of a previous cot death.
57. When interviewed on 16th November 1999, the appellant explained that she was in a state when the health visitor arrived because she had discovered Jason:

"I think on one of the occasions I had gone downstairs because we had a garden out the back … I had been putting some washing out and when I came back upstairs the alarm was going off and I didn't have the walkie-talkie thing, and I just went in, the alarm was going off."

The health visitor was "literally arriving as I'd found him." She had run to the door to open it to her. On 8th March, in her further interview, she said that she was in the room with Jason when she heard the car. She had not heard the apnoea alarm start up, but heard the alarm sound.

58. In evidence at trial she described hearing the sound of the alarm when she was in the garden. She had run to Jason. When she arrived he did not look very good, and he was only moving slightly. She turned the alarm off. She put him on the bed. She panicked. She was distressed and shocked. She tried to remember what she had been told about resuscitation but did not try it. She saw the health visitor arriving and thought "Thank God". She thought it was a pretty serious situation, and Jason in need of urgent help. She said to the health visitor, when she opened the door, "It's happened again".
59. These accounts of events immediately before Jason's ALTE were closely analysed by Mr Paul Dunkels QC, before the jury, and again before us. The Crown's contention was that the appellant smothered Jason, and that the smothering began once the appellant knew that Mrs Peacock was arriving at her home, in an attempt to evoke sympathy. We had difficulty following this suggestion. We cannot understand why the appellant should have done so, unless to draw attention to herself, a manifestation of factitious disorder by proxy, a condition which, in her case, was excluded. If however she had no intention of being caught, or indeed intended to kill Jason, she did not have to wait until the moment when the health visitor was about to come into the house and interrupt her, or carry out the attempted smothering process in such a way that no sign of it would be apparent. The process of airways obstruction could have been started earlier, or later, after Mrs Peacock had left. Furthermore, if this suggestion were correct, this particular ALTE (in which, on the Crown's case, Mrs Cannings was apparently hoping to be caught, rather than in each of the other cases, seeking to evade detection) was quite inconsistent with the establishment of the "pattern" which represented such an important strand in the Crown's case.
60. The appellant stayed with Jason for a couple of days in hospital. When he was discharged she thought that he was not as alert as he had been before, but she was reassured by the midwife. She had therefore felt no need to take Jason to the doctor.
61. On 13th June 1991 Jason died. He was due to be taken to hospital that morning for a consultant paediatrician to check the progress of his dislocated hips. Mr Cannings went to work. He telephoned home, to be told that Jason had stopped breathing again, and he called the emergency services. The appellant's account of what had happened was recorded in reported speech, as follows:

"Jason stopped breathing on 4th June but was resuscitated by a health visitor. Since that time he had been fine, feeding well and no chest or breathing problems. At 3.30 am on 13th June he had feed (bottle) no problems, and was put back into his cot with baby alarm turned on, mum went back to bed herself in the same room. At 7.45 am mother checked baby, he was ok, she went to get her own breakfast and the baby alarm went off at 9.00 am. She went back to the room and found him still and white, resuscitation tried with no success."

62. Before Jason arrived at hospital, in response to a 999 call, the emergency services arrived at the appellant's home. A paramedic said that as he arrived, the appellant was coming up from a kneeling position on the floor where Jason was lying. Although attempts were made at resuscitation, Jason was probably already dead. One paramedic, using specialist instruments, thought she had found some vomit in Jason's lower airway, but it was agreed at trial, that if there was any vomit it was entirely consistent with the baby's stomach contents having been released after he died.
63. In interview on 16th November 1999, eight years after Jason's death, the appellant explained that she had felt that Jason had not been quite right, or quite the same, after he was discharged from hospital. She thought she had heard the apnoea alarm after it had started. She went in to the baby. She found that Jason was white. She had tried to revive him. In her interviews on 8th March 2000, she repeated that she did not think that the baby had been himself on his return from hospital. She thought she had not heard the alarm starting, but became aware of its sound. She went straight to Jason, but found that he was very pale, fairly cold and lifeless. She was asked how it was that she had not heard the sound of the alarm earlier. The basis for that question was that if Jason felt cold by the time the appellant reached him, then he must have stopped breathing for some time before she went to him. The appellant said that the bedroom door was shut or nearly shut because of the family dog.
64. In evidence at trial the appellant described how Jason seemed "alright" at 7.45 am. Then at 9.00 she realised that the alarm was sounding. She found him still and white. He was not breathing. She tried to resuscitate him, without success. She thought her husband may have phoned at the time when she was trying to resuscitate Jason, and she told him that there was a crisis. An ambulance was called. When she realised Jason had died, neither she nor her husband could believe that it was happening again. She said that she blamed herself and felt that she "had failed".
65. The paramedics all described Mrs Cannings as upset, sobbing and distressed, and they tried to comfort her. The post mortem examination on Jason was carried out by Dr Scott who had performed the post mortem on Gemma. As with Gemma, she found nothing that could have caused or contributed to Jason's death. She carried out precisely the same tests as she had before, and also an x-ray, which was negative. She looked for any "feature of abnormality" and found none. The child's brain was looked at by a paediatric pathologist. It, too, was normal. Dr Scott was also aware of Jason's ALTE nine days earlier. She appreciated that she should bear in mind the possibility of smothering, or something similar. These concerns led her to check her findings with a consultant neuropathologist, Dr Isabelle Moore. At the end of her investigation Dr Scott's opinion of the cause of Jason's death was "Sudden Infant Death Syndrome (cot death)".
66. When the health visitor returned from her holiday, she visited Mr and Mrs Cannings. She gave evidence that throughout her dealings with the appellant and Gemma and Jason, she thought they were properly cared for by their parents. She introduced them to Professor Emery, the distinguished paediatric pathologist, to some of whose research into cot death we shall draw attention later, and Dr Scott herself later had contact with Professor Emery, who has now died. She showed him all her reports. He conducted a microscopic examination of the lungs. She understood that he agreed with her conclusions that both these deaths were natural, but that the causes were not yet defined. Dr Barnes confirmed that in Professor Emery's mind there was nothing suspicious about these two deaths. When Dr Scott carried out this post mortem, in accordance with practice at the time, she did not look for intra-alveolar haemorrhaging and iron-laden macrophages. The possible relevance of such material only became apparent in 1997, another mark of the constant development of knowledge.
67. After Matthew's death these findings were reviewed by Professor Berry, a consultant paediatric pathologist at the Bristol Royal Infirmary. He read all the records for Mrs Cannings and all the children, reviewed Dr Scott's post mortems on Gemma and Jason (speaking favourably of the quality of Dr Scott's work) and conducted the post mortem on Matthew. On examining the slides of Jason's lung tissue he found evidence of intra-alveolar macrophages, consistent with bleeding into Jason's lungs, which had taken place on an occasion earlier than his death. In view of the quantity, this evidence was consistent with an indication of obstruction of his airways at the time of the ALTE. However, on any view, there was no copious recent bleeding into the lungs. So there was no evidence of fresh bleeding, and as we shall see, a critical dispute developed whether and if so to what extent, there was evidence of old bleeding.
68. Professor Berry explained macrophages as scavenger cells found in the lungs. When blood is present in the intra-alveolar cells, the macrophages are unable to consume its iron content. Perl's (stain) test is used to detect siderophages, cells which themselves scavenge haemosiderin, the iron left behind in the lungs from a prior bleed, and which are well described as iron-laden macrophages. Pathologists can be confident that there has been prior bleeding, but the existence of siderophages, even if significant, is not diagnostic of smothering. Professor Berry expected that haemosiderin would develop within 36 to 48 hours and that the siderophages would be cleared, by absorption into the blood stream, coughing or swallowing within weeks.
69. Dr Rushton took a different view from Professor Berry about the quantity of haemosiderin visible on the slides. Using a microscope in court, he explained what he was seeing on them, and his view that it was possible to describe haemosiderin as present in only two of the seven pieces of lung which were examined, and then only in parts of those two pieces. He explained that the cells were not "all through the lung with a wide distribution; they are in quite small areas of the lung, in fact very small, microscopic areas of the lung". He disputed Professor Berry's observations that macrophages were present in exceptional numbers, "because in fact most of the lung doesn't contain any of these cells. They are localised in the sections to two of the small pieces of tissue and only in parts of those tissues." In his opinion the amount of haemosiderin to be seen was not exceptional, and based on his experience he concluded that, at the highest, it was only marginally in excess of the amount occasionally seen in cot death.
70. Dr Rushton postulated various ways in which bleeding in the lungs such as that suffered by Jason might occur, including blood inhaled at birth (which was excluded by Professor Berry), and damage to the lung from foreign bodies such as vomit or stomach juices, or from the inhalation of toxic gas, or infection, or from bleeding or circulatory disorders, or even from resuscitation procedures. He was not in the slightest degree critical of the health visitor, or her efforts at resuscitation on 4th June, which were described as gentle. It was suggested that if so, attempts at gentle resuscitation would not be likely to have caused any bleeding. His response, in effect, was that there was no way of knowing "because obviously most babies that are resuscitated don't die, and therefore we can only judge from the ones that die. It may be that in fact lung haemorrhages are very common in resuscitation. Unfortunately, that's evidence that you can't acquire." He also noted that resuscitation was carried out in the ambulance on the way to hospital. Again, although he implied no criticism whatever of the paramedics, the same point remained. If the effort were too gentle, the lung would not expand and resuscitation would not take place: if the resuscitation were too hard, then some of the thin membranes with blood vessels in them might be torn and produce local bleeds. Dr Rushton agreed that the blood could have resulted from the ALTE which had taken place nine days earlier, but felt it was difficult to estimate the time taken for the blood to leave the lungs. This affected the question whether the appearance of the macrophages was more consistent with an incident at the time of the ALTE rather than on delivery at birth. Dr Rushton's concern was that the length of time taken for blood to disappear from the lungs was not known. He pointed out that "Some people believe it can disappear extremely rapidly while other people believe it may hang around for some considerable time."
71. Although we have explained the areas of dispute between these experts, the real question dividing them was whether there was a sufficiently reliable way of assessing the amount of haemosiderin to enable them to offer assistance to the jury on whether the quantity found in Jason was sufficient to be indicative of, or consistent with smothering, or SIDS. Despite the disputed problem of haemosiderin in Jason's lung tissue, Professor Berry made clear at the outset that he was not "for a moment" suggesting that what he found was "diagnostic of imposed upper airway obstruction". He did however regard it as a "warning" and "extremely worrying", and although he thought that the blood found in Jason's lungs had not resulted from efforts at resuscitation, he could not "absolutely rule this out". As a consultant paediatric pathologist, Professor Berry believed and even now remains of the view that the cause of Jason's death, and Matthew's, too, remains "unascertained". Dr Rushton did not suggest that his overall analysis of the pathological data provided any positive alternative explanation to that advanced by the Crown that there was an unnatural cause for the deaths and ALTEs. He also recognised that any assessment of the amount of haemosiderin had a subjective element, namely the experience and judgment of the individual examining the relevant slides. That said, he did not resile from his fundamental view that the amount of haemosiderin to be seen was not inconsistent with a natural event, or efforts at resuscitation after it. In our judgment, the presence of haemosiderin on Jason's lung slides was an ascertained fact. It was not however possible to conclude that it represented the result of smothering.
72. We shall reflect on the vaso-vagal issue, as it may relate to Jason's ALTE, when dealing with the ALTE subsequently suffered by Jade. We also note in passing that an issue canvassed at trial related to the proximity of the house to Porton Down, and to the deaths of two other babies living in the same vicinity in 1990 and 1991. This issue was not pursued before us, and it plays no part in our decision.
73. We must now return to the narrative: after meeting Professor Emery, Mr and Mrs Cannings were willing to help him in his research into cot deaths. They changed their lifestyle. They gave up smoking. They moved house. Mr Cannings returned to work. After a while the appellant felt "empty". She wanted another child.

Jade

74. After an unremarkable pregnancy, Jade was born by Caesarean section on 15th January 1996. Before her birth Mr and Mrs Cannings had accepted Mrs Kenny's suggestion that they should join the CONI scheme to which reference has already been made. Mrs Kenny was a new health visitor covering the area of their new home, and like Mrs Peacock, she was very experienced. With advice from Dr Barnes, a consultant paediatrician with a special interest in these problems, they did so in December 1995.
75. A neo-natal check shortly after Jade's birth revealed no problems. Mr Cannings himself was off work, sick, for some six weeks. Mrs Kenny made her first visit to the appellant's home some two and a half weeks after Jade's birth. Jade was very well. Mrs Kenny continued to visit on a regular basis, a total of fifteen times before 1st April. Nothing was noticed by Mrs Kenny during her visits to raise any cause for anxiety. Jade was also seen during the post-natal check and a six-week check, which revealed "a perfectly normal healthy baby". The General Practitioner also recorded that Mr and Mrs Cannings seemed to be very well. The appellant herself had no concerns about the little girl.
76. On 1st April 1996, in response to a call by the appellant, her general practitioner, Dr Meader, called at the appellant's home. She found the appellant distressed, because Jade had woken that morning with breathing difficulties and her condition was similar to that of the two babies that had died previously. On examination, the doctor found that Jade was not in fact "desperately ill", but she was "floppy, heavy breathing, vomiting ++". Her chest was clear. She nevertheless took the appellant, together with Jade, to hospital where Jade was admitted at 8.45 am. The appellant's account of events in reported speech was:

"Baby fed at 6.00 am – floppy, laid back down as thought baby tired. Turned apnoea alarm off as thought baby would wake. Mother fell asleep again. Woke at 7.30 am went into her bedroom and noticed white colour, eyes closed, breathing gasping and laboured. Shook baby, called GP, baby began crying, breathing still laboured. Apnoea alarm went off two weeks ago when disconnected."

The evidence also showed that Jade had vomited twice at home and again in the ambulance on the way to hospital. The appellant added that until this incident Jade had been well, but she reported that the baby had been lethargic during the previous day, and had suffered two bouts of diarrhoea. There was a dispute at trial between experts, which summarising it simply, was whether this was a true ALTE, or simply a consequence of a bout of gastro-enteritis. If the latter, of course, it ceased to be relevant to the issue before the jury: the baby was suffering a normal unremarkable illness.

77. When interviewed on 16th November 1999, the appellant told interviewing officers that after her husband had gone to work, and Jade had woken and finished her feed, she had put Jade back down to let her sleep. She forgot to put the apnoea alarm on, and went back to bed herself. She woke with a start, and went to Jade, who was very pale and white. She rang for the doctor. Jade started vomiting when the doctor arrived. In interviews on 8th March, she repeated this explanation, saying that she had forgotten to put Jade's apnoea alarm back on when she put her down after feeding her, saying, "Maybe I was still tired myself from getting up and giving her a feed."
78. In evidence at trial, the appellant described the incident in very similar terms. She said that she had forgotten to put the apnoea alarm back on because she was tired, adding that this was the only time she had forgotten, and she did not forgive herself for a long time afterwards. She had telephoned the doctor because Jade was very pale, gasping for breath, with her eyes closed. She denied having shaken Jade, but added that it seemed that what had happened to Jade was similar to what had happened to Jason nine days before he died. She described an overwhelming sense of panic. She went on to describe events at the hospital, and then her own increasing confidence as Jade flourished.
79. On examination at hospital Jade was found to be very pale and her hands and feet were cold and mottled. Her fontanelle appeared sunken, and it was noted that there was "slight wheezing" in her chest. Treatment by an intravenous drip, containing albumin and salt solution, and later intravenous antibiotics was arranged. A broad range of tests was carried out. Her glucose level, white blood cell count and platelet count were moderately but not significantly raised. Analysis of organic acids in a urine sample taken four hours after admission suggested that her kidneys had become "unhappy". It also showed increased levels of lactate, or lactic acid, pyruvate, alanine and hydroxybutyrate. This would be consistent with a stress response to a severe incident of acute deprivation of oxygen. Cerebral ultrasound of the brain was normal, and a disorder of the body chemistry was ruled out, although an increase in IgM was noted. At the time gastro-enteritis was not diagnosed. At trial the defence case was that the non-diagnosis was wrong.
80. Jade remained in hospital until 13th April. While there she continued to have episodes of vomiting and diarrhoea. There were no problems with her breathing. Taking this in a little more detail, by 3rd April, the problem of diarrhoea and vomiting appeared to have been alleviated. The next day, it was felt that Jade was well enough to go home, but her parents decided that she should remain as an in-patient. There were further episodes of vomiting, which continued into 5th April. A substitute baby milk was introduced and then reduced. On 6th April there was a single episode of loose stools, but no vomiting, and a spike of temperature. It was noted that she "has passed urine now with some diarrhoea". Progestamol was discontinued lest it was causing the vomiting. She was given some water to drink, and that afternoon was "more alert." On 7th April Jade was tolerating her food, and said to be clinically well. On 8th April observations continued. The baby was feeding well and was more comfortable. Her temperature was normal. On 9th April the progress had continued. The baby had a very good morning, and was again tolerating all food and her temperature was normal.
81. Jade was due to have second stage immunisations, and so the appellant wanted her to be observed in hospital. She therefore remained for another 3 days. The immunisations had no side effects. Tests were carried out which excluded any possible metabolic problems. A twenty-four hour heart recording excluded cardiac problems. Special immunological tests were carried out. These proved negative. As with Jason, bio-chemical changes in Jade, of potential relevance to this particular ALTE as a vaso-vagal incident, were noted.
82. Jade left hospital with her mother on 13th April.
83. As already noted, there was a dispute at trial about the correct diagnosis. Professor Meadow and Dr Ward Platt disagreed with the contention that Jade had suffered an episode of acute gastro-enteritis. She was not displaying sufficient symptoms to enable that diagnosis to be made. Professor Meadow, while conceding that paediatric gastroenterology was not his speciality, nevertheless believed that this was a topic well within his sphere of expertise, and noted that Jade had taken a feed normally at 6 am on the morning of her admission, which was inconsistent with severe gastro-enteritis, and it was "inconceivable" that gastro-enteritis could explain the state in which she was found shortly afterwards. Dr Ward Platt believed that Jade's reduced temperature, as noted on admission, would have been unusual, and the episode of loose stools and vomiting prior to admission was insufficient to justify this diagnosis. He suggested that this would require some days of loose stools or vomiting, together with an inability to take fluid, and that a severe attack of gastro-enteritis would have resulted in a significant reduction in the baby's body weight as measured on admission and set against her weight on discharge. He did not accept that the baby was dehydrated. The sunken anterior fontanelle which, he agreed, would have been a sign of dehydration, was, he believed, misleading, and one which was often misinterpreted by junior doctors. The term had not been used by the more senior clinician who saw the baby later. The use of the intravenous drip had been unnecessary. With Professor Meadow he also believed that the good urine output noted while Jade was in hospital militated against the diagnosis of dehydration, and thus of severe gastro-enteritis. He could accept mild, but would reject severe gastro-enteritis. His final position was that Jade had an illness consistent with gastro-enteritis, but that after an incident of smothering, a "knock-on" effect upon the gut can occur.
84. Dr Ward Platt found no evidence of an active infection in Jade's chest x-rays, and ruled out any chemical disorder, believing her to be a normal child developmentally, who had endured a striking and severe stress response. He believed that the episode was respiratory in nature, characteristic of oxygen deprivation, and having eliminated the other possibilities, he concluded that this ALTE had resulted from smothering.
85. Professor Milla, a professor of and specialist in paediatric gastroenterology and nutrition, with twenty years' clinical experience, gave evidence for the defence. His belief was that Jade had suffered a circulatory collapse (rather than a respiratory condition) the cause of which was not clear, but which, taken together with her pre-admission symptoms, suggested gastro-enteritis. On analysis of the case records he believed that:

"She had a circulatory collapse which resulted in her being admitted to hospital. The cause of that circulatory collapse is not entirely clear but she had symptoms both prior to the collapse and following the collapse suggestive of gastro-enteritis."

86. He rejected the idea that diarrhoea was the sort of thing to be expected in a young baby, commenting:

"Neither you nor I nor infants have diarrhoea for no reason. Diarrhoea is caused by malfunction of the gastro-intestinal tract and that may be caused by a whole host of different disorders but most commonly it will be an acute infection of the gastro-intestinal tract, and in this country viral infections are the commonest cause of them."

87. He addressed the continuing history of diarrhoea and vomit, as described in the notes, observing that the history suggested that this was not a simple "one off" situation. He agreed that the elevation in the levels of lactic acid, pyruvate, and hydroxybutyrate were indicators of non-specific episodes of stress, but suggested that they were equally consistent with gastro-enteritis. As Jade had a diarrhoeal illness, her body fluids may have shifted from one part of her body to another, causing a drop in the sodium level. This was consistent with her notes which showed a low salt level. Her low temperature, sunken fontanelle, mottled skin, and the description of her as "floppy" suggested a circulatory shut-down. It was a factor consistent with dehydration that Jade had not passed urine until the afternoon after her admission. He was fortified in this view by the estimation of her fluid loss by the clinicians who treated her, assessed as between 6 and 10%. He viewed the figures for weight on admission and discharge as without relevance, since her normal daily growth would have been about 30g. Taking into account that she had been given fluid during her period of admission, he deduced that Jade had not grown at all while in hospital.
88. Professor Milla cautioned that the level of diarrhoea endured by a baby can be masked by the efficacy of modern nappies. Toxins can be fast acting, and the physical signs suggested a peripheral shut-down in the circulatory system, consistent with the passing of a lot of watery diarrhoea. The history given was neither uncommon nor inconsistent with gastro-enteritis, and though her hospital notes were insufficiently informative on the subject, he felt that a number of causes considered by the treating clinicians plainly included gastro-enteritis. He pointed out that Jade's history would not have been "that uncommon" in a child who had gone on to have a severe problem, and that "Something clearly was carrying on for quite a while after her admission into hospital." He ended his evidence by repeating that Jade had signs and symptoms consistent with gastro-entero infection, and that the history of her time in hospital was consistent with that diagnosis, and nothing inconsistent with it had happened.
89. In summary, if Jade's admission to hospital on 1st April resulted from circulatory rather than respiratory problems, nothing unnatural had occurred. We take note of the fact that a leading expert in this field was unshaken in his belief that Jade's problems were consistent with gastro-enteritis, and that they may have been circulatory in nature. If so, this was not or may not have been an ALTE.
90. Whether Jade suffered severe gastro-enteritis or not, fairly mild gastro-enteritis or mild dehydration might have provided a trigger for an inappropriate vaso-vagal response. As we have already noted, the same possibility was said to apply to Jason's ALTE, and it is convenient to deal with the point in relation to both infants together. This phenomenon was described by Dr Barnes as a safety response against reflux, and the possibility of inhaling vomit or regurgitated substances. Professor Berry acknowledged that as a pathologist he knew, "Babies can die very suddenly by another mechanism we don't quite understand, so I think it is also possible the baby could die almost instantaneously – almost a kind of shock effect." He agreed that it would be something like a vagal response, or kind of nervous response. Professor Meadow rejected the idea that a vaso-vagal attack was a significant factor in a sudden infant death. Dr Ward Platt rejected it both generally, and with particular reference to this case. He accepted the possible existence of an abnormal vaso-vagal response in an infant, but held the view that a baby would recover very quickly from such an incident.
91. Dr Johnson, a consultant clinical physiologist, whose concerns about the lack of detailed investigation and monitoring of the children born to Mrs Cannings after Gemma's death have already been narrated, drew attention to the way in which the vagus nerve operates. The nerve serves many of the organs of the body, and not only carries information to the brain, but also controls the heart rate and the opening of the airways. It is "a major controlling nervous pathway". He described difficulties which babies have in the first few months of life in breathing through their mouths, and the way in which chemo-receptors will react to foreign agents by going into spasm, causing a temporary cessation of breath.
92. Dr Johnson is researching the question why some babies over-react by way of vaso-vagal response, and shut down inappropriately. He described obstruction of the airways and how it could occur, and yet remain undetected at autopsy. He believes that existing research shows that a vaso-vagal response may be one of the causes of ALTEs, and his concern that in these two cases there had not been the investigation which could, and in his professional opinion should, have been undertaken. He was not asserting that a vaso-vagal response definitely caused either of the two ALTEs, nor was he ruling out smothering or unnatural causes, which, like a vaso-vagal response, were high on his list of possibilities. He referred to research from Brussels, based on an investigation between 1983 and 1990 of 3,799 young children. This, he said, suggested that a significant problem with gastro-oesophageal reflux and an exaggerated vaso-vagal response might bring about changes to defend the baby's airway. The lack of proper investigation led him to conclude that an alternative explanation to smothering had not been excluded. We cannot resolve this important and developing issue. Once again, however, we record a responsible opinion suggesting uncertainty about the true causes of these two incidents.
93. What is certain is that Jade survived. The appellant described her as "the light of our lives". She and her husband wanted her to have a brother or sister, and when the appellant became pregnant, Jade was very excited. After her ALTE, Jade was seen regularly by Mrs Kenny. She made good progress. In July she was admitted to hospital suffering from what was believed to be some form of gastro-enteritis, as well as upper respiratory tract infection and urinary tract infection. She attended at clinic, and the health visitor returned to see her at home on a regular basis until she was one year old. There were seventy-seven such contacts during this period. It was pointed out by the Crown that the family and extended family made considerable efforts to ensure that during this period the appellant was not left alone. The inference to be drawn was presumably that this reduced the opportunity for the appellant to take any further steps likely to harm Jade. If so, we reject the suggestion that she would have lacked any appropriate opportunity to have caused harm to her baby if she had wished to do so. In truth, Jade's subsequent survival and healthy development after the single ALTE is inconsistent with the pattern sought to be established by the Crown: putting it simply, both Jason's ALTE, and Matthew's "episode", as we shall see, were followed shortly afterwards by sudden death.
94. Mrs Kenny expressed the belief that "Mrs Cannings had bonded very well with Jade, she seemed to be a well cared for and loved baby. Nothing at any time gave me any cause for any kind of concern in the way in which either parent treated Jade."
95. Before long, Mr and Mrs Cannings moved house again, and Mrs Peacock resumed her responsibilities for Jade. Happily she has continued to grow and develop uneventfully. Apart from normal childhood illnesses, there have been no further problems, and no further ALTE.

Matthew

96. Matthew was born by Caesarean section on 5th July 1999. Thereafter the appellant was sterilised. The birth was uncomplicated, but unfortunately the appellant suffered from Bell's palsy. At birth Matthew's Apgar score was 9 out of 10, very healthy. Because of the two previous deaths extensive investigations were carried out. MCAD deficiency was excluded. Blood plasma was checked and no abnormality found. No cardiac abnormality was detected by electrocardiogram (ECG), but the measurement of the QT interval (which we shall describe later) was drawn to the attention of the paediatric cardiologist. A further ECG was carried out in early August. The information revealed in these tests was re-examined after Matthew's death, and questions were subsequently raised whether Matthew and his family were susceptible to Long QT syndrome. In that context ECG tests were also performed on Jade and Mr and Mrs Cannings. In the meantime, the appellant was taken through a paediatric life support course to teach her advanced resuscitation techniques.
97. On their return home, Matthew seemed well, but the appellant was concerned because her treatment for Bell's palsy involved taking steroids, and in due course this meant she had to give up breast feeding him. The appellant apprehended that the apnoea alarm meant that it was not unsafe for Matthew to sleep on his tummy, which was his natural preference anyway. Matthew's progress was continually monitored. No concerns were expressed about him. Indeed all the checks, both at clinic, and at home, were normal, and the appellant herself did not report any problems to the health visitor. Everything therefore seemed to be going well. His MMR vaccination was given on 1st November. The appellant's confidence grew, and she said she felt able to leave him in the house, with the front door open, or, later, in the car, while she chatted to one of her neighbours.
98. On the morning of 3rd November, the appellant made a 999 call to the ambulance service. The transcript is available. She said that Matthew was "breathing, I think, but he has been sick everywhere" and after being urged to calm herself down, said that he was definitely breathing, but he had been sick and "his hands are pale". She told the ambulance control of the two prior cot deaths. She said "he looks like he's trying to, trying to get his breath … he just looks as if he's not with us you know, he, his eyes are open. He is breathing but he is sucking in everything, it is horrible". She was asked whether he was breathing and said, "yes, but he is very laboured, he is on his front at the moment".
99. The paramedics responded to the call. The problem was differently described as an ALTE, or simply as an "episode", not life threatening, indeed not of great gravity, which appears to have been the conclusion of Professor Meadow and Dr Ward Platt. When the paramedics arrived, the baby was found in a cot. The appellant told them that his apnoea alarmed had worked. One of them tried to assess the child, who seemed conscious and appeared normal and alert. Indeed although his limbs were slightly pale, the baby felt quite warm and there appeared to be no major problems. The appellant was standing on the opposite side of the room.
100. Given the previous history, it was thought sensible to take the baby to hospital. The report of the appellant's account reads:

"Mum fed him his breakfast this morning and dad put him to bed – 9.00 am. Apnoea mattress was on. About 20 minutes later the mattress was alarm so mum went to investigate. Matthew had been sick – breakfast and some clear fluid. Fighting for breath. Pale, not blue. Mum describes him as being distant. Phoned 999 for ambulance."

101. When first examined, Matthew was pale, but conscious and alert, breathing was satisfactory and temperature normal. He was said to be bright and alert. The paediatric registrar noticed some crusted milk around his nose, and a "slightly inflamed looking throat". Otherwise the examination was unremarkable, and the record on file suggested that this was not a true apnoeic event. Arrangements were made for Matthew to be kept in hospital overnight, for observation. He seemed very well, feeding properly, alert and smiling. Neither the oxygen monitor nor the heart monitor, gave any cause for alarm. By next morning Matthew was happy and well. He was discharged.
102. At hospital it was noted that the appellant was distressed and had been crying and her husband was in a similar condition. Her worries for Matthew appeared genuine.
103. In her interview on 16th November 1999 the appellant told the police that Matthew was put down on his front to sleep because he never settled on his back. She understood that because an apnoea monitor was available, it was acceptable not to sleep him on his back. She described hearing the alarm when she was downstairs. She ran upstairs to the bedroom and turned the alarm off. She turned Matthew over. He was being sick. His condition was such that she ran downstairs and telephoned 999. In the interviews on 8th March 2000 she gave a similar account. She said she had not heard the start of the alarm, but she heard it sounding. The noise of the alarm was relayed downstairs through a walkie-talkie system. When she went upstairs Matthew was breathing but gasping for breath.
104. Two features attracted particular interest at trial. First, on the basis of the answers in interview, there was said to have been a lengthy gap between the moment when the appellant found Matthew, and the 999 call, precisely the opposite situation to that relied on by the Crown in relation to Jason's ALTE. Second, if the apnoea alarm was working correctly, and this was no more than an "episode", then it should have cut out once Matthew started to breathe again, unless the baby was having some sort of fit and breathing very rapidly indeed. That of course begged the question whether this was "any more than an episode". If it was no more than that, it was inconsistent with a pattern of events relied on by the Crown.
105. A few days later the appellant spoke to her neighbour, Mrs Aldous, about the incident. On this occasion, after they had had a chat about the circumstances in which the ambulance had come to her home on 3rd November, the appellant produced Matthew from her car, rather proudly showing him off. Mrs Aldous thought he looked fine and well, but estimated that on a very cold day Matthew had been alone in the car for at least ten minutes. The Crown seemed to attach some importance to evidence of this kind, the suggestion being that it was inappropriate for the appellant to have left Matthew alone in the car without the safeguard of his apnoea alarm. We do not see the basis for criticism: the appellant was nearby, and this behaviour seems unremarkable.
106. At 11.04 am on 12th November, the ambulance service responded to a 999 call from the appellant's address. This call was made by Mr Cannings, who had himself received a call from his wife while he was at work telling him that "It's happened again". The appellant was upstairs, on the floor with Matthew, performing mouth-to-mouth resuscitation. The baby was very limp, and cyanosed. There were no obvious signs of breathing or cardiac output. He appeared to be in cardiac arrest. The paramedic believed that the baby's airway was clear, without vomit or any other residue, either in the mouth or the airways themselves.
107. Matthew was taken straight to hospital. On the way resuscitation continued in the ambulance. In fact, it was too late. He was in cardiac arrest, not breathing, cool and unresponsive, and centrally cyanosed. There was no evidence of any blood or trauma to the baby's face or his head. There was a little, unremarkable, vomit around the mouth area, and traces of fluid that smelt like partly digested milk, again in the context, unremarkable. The condition of his immune system became a subject for discussion, and disagreement between experts. The baby was pronounced dead at 11.05 am.
108. At hospital, the appellant was tearful, and her husband extremely distressed. A staff nurse heard from the appellant that the baby had been in bed when she heard the alarm. She had turned it off and ran upstairs. She turned the baby over and saw that he was purple and blotchy. Her husband asked her in the presence of the staff nurse why she had called him before she had called an ambulance, as indeed she had. She was quiet for a few minutes, and then told her husband that she had panicked. He asked her again why she had not called the ambulance first and she repeated that she had panicked. She also spoke to Dr Barnes, the consultant paediatrician who it will be remembered, had taken a particular interest in this family, and arranged for its inclusion in the CONI project.
109. In interview on 16th November 1999, the appellant described how Matthew was in bed upstairs, with the apnoea alarm on and the walkie-talkie in use. She went into the kitchen, and put on the dishwasher. She was working in the kitchen with the television on. When she left the kitchen she heard the alarm. She ran upstairs and turned it off. She turned Matthew over and saw that he was purple, with his face in blotches. She ran to the telephone in the other bedroom and called her husband asking him to come home because she thought that they had lost Matthew. She said that initially she had spoken to a receptionist at her husband's place of work, but when there was no answer she had rung the bakery there, asking the person who answered the phone to get her husband, urgently. She ran back to Matthew's room, grabbed him and put him on a bed and tried resuscitation. On her husband's return he asked if an ambulance had been called. She said not.
110. In interview on 8th March 2000 she was again asked why she had telephoned her husband rather than an ambulance and said "because I wanted Terry to be there". She was asked why she had done so and she said that she wanted his help, so that he could see Matthew and see what he was like. This issue was pursued by the Crown on the basis that, given Matthew's condition, it was odd that the appellant had not herself directly and immediately sought help either from the emergency services or indeed from neighbours, at least one of whom was a nurse who had offered to help.
111. The apnoea alarm was working properly. Its audibility, through the walkie-talkie system, was tested by sound engineers. Assuming the alarm and transceiver were operating correctly (which they were) they concluded that the signal would have been audible in the kitchen, if both the dishwasher and television were in use in the way described by the appellant.
112. In her evidence the appellant said that as soon as she had heard the alarm sounding, she turned it off and went upstairs to Matthew, doing what she could to resuscitate him. She telephoned for her husband, needing his help. She "couldn't believe the way he was". She wanted "Terry to be there to support me. I had always been on my own". She added that after Matthew's death "she was still trying to take in what happened." She was "numb and in shock". When she was asked about some of the answers she had given in interview, she said that she had not really felt up to being questioned.
113. Dr Barnes, the consultant paediatrician, who knew this family, considered whether in all the circumstances the fact that there had been three deaths led him to conclude that the cause was filicide. On 2nd December 1999 he wrote "although the number of deaths in this case must arouse suspicion I can say that I did not have suspicions of that nature from observing Mr and Mrs Cannings' behaviour during the time that I have known them". Dr Ward Platt agreed that taking both the episode and Matthew's death together, and looked at in isolation from the other events, there was nothing to suggest the deliberate infliction of harm on him.
114. Three days after his death, the post mortem examination of Matthew was carried out by Professor Berry. Focal pulmonary haemorrhage and oedema in the baby's lungs were found, both unremarkable and not diagnostic of unnatural death. Matthew had inhaled a small amount of stomach content, and a small quantity of aspirated gastric contents was found in his airways. No relevant haemosiderin was observed. The post mortem examination showed a very mild tracheo-bronchitis, but no natural cause for his death, and no unnatural cause for it either. Professor Berry expressed the opinion that,

"In view of the extreme rarity of three deaths without explanation occurring in the same family I have given the cause of death as unascertained pending further investigations."

115. No viruses were detected, but a blood culture showed two types of coliform bacilli and enterococcus. The former was detected in the cerebro-spinal fluid sample, and in a culture. This information was not known at the time of trial. In our pre-reading of the papers it seemed that there might be serious dispute about the impact of such evidence, if it had been before the jury. In the result, neither side addressed this issue in any depth, and we did not believe we should derive any benefit from analysing it for ourselves.
116. We must return to Matthew's ECGs, and the complex problem of the prolonged QT interval syndrome, the hereditary form of which is usually known as Congenital Long QT Syndrome. Taking it as untechnically as we may, the interval between the Q wave and the end of the T wave is a measure of time in milli-seconds (ms) taken for the pumping mechanism of the heart, to contract and relax. This, the QT interval, varies with the heart rate, so that a formula is used which expresses the interval as if the heart rate were 60. Bazett's formula is the QT divided by the square root of the RR interval (the time between two heart beats). The formula gives the corrected QT interval expressed as QTc. Good practice suggests that the QT is measured in the ECG leads which show the longest interval, often leads 2 and V5, also called C5. A long QTc indicates that the relaxation time of the pumping chambers is lengthened and this can be an indicator for sudden death. The QTc can vary from day to day and on the same tracing, but Long QT can be diagnosed when the QTc is over 440 ms and with some certainty when it is over 480 ms. Problems in measuring the QT interval include difficulty in telling where the T wave ends, and in particular it can be difficult in infants in part due to their fast heart rate.
117. Long QT syndrome is an inherited disorder. A number of genes may be responsible for this genetic defect, usually autosomal dominant, occasionally autosomal recessive, and some of the genes are as yet unascertained. As we understand it, this may be of relevance to Professor Patton's evidence about events in Mrs Cannings' extended family, and the pattern of possible inheritance. Its importance can be readily explained: there is research which suggests that some cases described as SIDS may be due to an "abnormal arrhythmia related to the abnormalities" inherent in the Long QT syndrome. Accordingly, the QTs of Jade, Matthew and Mr and Mrs Cannings were all considered. Each expert who measured them achieved different results. In summary, Dr Morgan, a consultant cardiologist who was involved in the care for Jade, felt unable to exclude Long QT as a factor in the infant deaths suffered within the Cannings family, but the Crown's expert, Dr Wren, was able to do so. A fair summary of his opinion was that although he acknowledged that the concerns raised by Dr Morgan were legitimate, the readings did not cross the divide which suggested that Long QT may have provided a reasonable explanation for Matthew's death.
118. The weight of the expert evidence was that in the first week of life, the QT reading, quite apart from all the other difficulties, is in any event very variable. The first ECG for Matthew, taken when he was three days old was measured by Dr Stratton, the first of the paediatricians involved in this work, who measured the readings in such a way that the readings exceeded the expected, that is between 350 and 430 milli-seconds. Both he and Dr Kinnaird, another paediatrician, felt that the reading was towards the upper limit of and perhaps outside the normal range. So the cardiology team at Southampton was consulted. The team measured the reading at 440/450, a little lower than the reading made by Dr Stratton at 480, but not so as to exclude Long QT. So a paediatric cardiologist, Dr Keaton, assessed it. He was unconvinced that the reading was abnormal, but was not prepared to disregard the findings, so a second ECG was arranged for Matthew in August. Dr Gnanapragasam, a consultant paediatric cardiologist, read Matthew's July ECG as borderline, at the upper limit of normal, and the August ECG as normal, at 370 ms. Dr Wren agreed that Matthew's July ECG was borderline normal, and that in August most definitely normal or lower than normal.
119. Dr Wren reviewed the ECGs for Matthew, Jade, and for Mr and Mrs Cannings. He found no evidence to support a diagnosis of Long QT in any of these family members. He did however explain the difficulties of identifying with confidence what can be described as "normal", and what is not, because in practical terms, at present at any rate, it is not possible definitively to establish where the upper limit rests. He attributed this problem to the development of the Schwartz criteria, which required the assessment of a number of aspects of the health of the patients to be assessed on a points basis. This was a reference to research conducted between 1976 and 1994 by Dr Schwartz in which he used records of more than 34,000 infants aged between 3 and 4 days, following them up a year later, in an investigation whether there was any link between Long QT and SIDS. His conclusion (Schwartz et al, Prolongation of the QT interval and the SIDS, New England Journal of Medicine 1998 Vol 338 1709-1714) was that the presence of a QTc longer than 440 in the first week of life increased by a factor of 41 the risk of SIDS. An examining clinician using the Schwartz criteria would look first for a Long QT interval, and score it between 1 and 3. A female would not score at all unless there were a reading of at least 450. If the appearance of the T wave were odd, or if there were any change of rhythm, or a history of deafness, or a collapse upon exercise, or a relevant family history, points under the scheme might be credited. Dr Wren felt that one major criticism of the Schwartz research was that it was not "blind", but Dr Morgan corrected what seems to have been a misunderstanding. Dr Wren acknowledged that notwithstanding controversy about the Schwartz research, the condition of a congenital Long QT was an established phenomenon, and where it existed, it caused sudden infant death. Professor Golding described the Schwartz paper, and its results, as "exciting" and her opinion was that, so far as epidemiology was concerned, this risk factor identified by Schwartz was striking.
120. Dr Wren described children with Long QT who had collapsed suddenly and unexpectedly, and who had eventually made an almost complete recovery. He also emphasised that there were different ways in which individual clinicians may read the same ECG trace, and he warned against the dangers of adopting too dogmatic an approach to assessing whether a patient was or was not outside the normal range. He himself considered that Matthew's July ECG was outside the Schwartz criteria, there being a temporary difference in his QT level, and observed that the doctors involved regarded Matthew's August ECG as normal. Applying the Schwartz criteria, he was unable to find evidence that any family member came even close to Long QT syndrome.
121. Dr Morgan examined all the ECGs available for Jade, Jason and Matthew. He was less confident than others about the accuracy of the machines, and less inclined to pay attention to the results from them. Jason's measurements, which he was able to extract from the ECG machine only with difficulty, suggested that he did not have Long QT. He felt that Jade's ECG in the first twenty-four hours of life was not relevant, and he devoted some three and a half hours to reading the eleven ECGs taken when she was six years old, in January 2002. He did not feel able to exclude the reasonable possibility that Jade suffered from Long QT. He agreed that Matthew's ECG for July, almost immediately after his birth, showed a QT interval at or beyond the normal range, but was not prepared to make a diagnosis of Long QT on that single reading, made so close after birth. As we read pages 2048 and 2049 of the evidence, he remained concerned because, in relation to the research by Schwartz, Matthew's early ECG was "clearly abnormal on the third day of his life". He accepted that the August ECG reading for Matthew was normal, but he was not prepared to exclude any underlying condition. He had concerns about the readings for Mrs Cannings herself, agreeing with Dr Wren that those were at the upper limit of normal.
122. His final position was a high degree of suspicion, but falling short of the diagnosis, that Matthew and Jade had Long QT. He did not exclude the possibility of Long QT within the family, which would be a marker of an underlying abnormality, which could result in death. He also pointed out that if an infant had died of Long QT, its first presentation was at death and also that measurements apparently within the normal range did not exclude individuals having a Long QT.
123. In summary, the Crown's position on Long QT was that it was a "non-starter", effectively ruled out. The defence suggested that on the basis of Dr Morgan's evidence, there was a real possibility that in this particular family, Long QT represented at least one factor which may have contributed to the natural deaths of Jason and Matthew. For the Crown to succeed, this possibility, like others, had to be excluded.
124. A separate area of specialism was directed to the question whether Matthew suffered from an immunodeficiency. It was not suggested that the deficiency could, by itself, have caused Matthew's death, but, if it were established, it provided what was described as a "gateway" through which other potential dangers, for example infection, might pass.
125. One of the problems with this area of the evidence was the poor quality of the blood sample taken from Matthew post mortem. The second sample too, was deficient, and like the first, contaminated with blood proteins. If the child had been alive, these samples would have been rejected. Dr Unsworth, who gave evidence for the Prosecution, tested for the presence of immunoglobulins. From the tests he was able to carry out, and the other material in the case, including what he believed was an absence of acute infection at post mortem, he came to the conclusion that there was no convincing evidence of immunodeficiency, nor evidence in Matthew's blood of the kind he would have expected if the collapse had been caused by infection. The results produced "normal levels for IgM; very low, basically undetectable levels for IgA, and then a low level for IgG." He concluded that it was not possible to say whether or not immunoglobulin was present in normal or abnormal amounts. This view was supported by Professor Berry from his findings at post mortem. His examination of the relevant slides led him to believe that the part of Matthew's system responsible for antibodies was "completely normal", and using a specific test, he found that the cells for making immunoglobulin, too, were normal.
126. Professor Hutchinson, an expert in immunology called by the defence, tried a series of different tests which he said enabled him to identify only very low levels of IgG, suggestive of a low level of protective antibody in Matthew's blood which made him susceptible to infection. These, he suggested, were robust tests, appropriate to cope with the inadequacies of the two blood samples. He believed that the level of antibody in Matthew's blood system was "probably less than 10% of what one would hope to see. It may even be lower than that". IgG was present, but in rather low amounts. He suggested that, compared with normal amounts of immunoglobulin, "there was a vanishingly small amount of immunoglobulin in Matthew's sample", but conversely, a normal level of IgM.
127. The consequence was that Matthew was susceptible to various kinds of infections, many of which would be completely harmless to adults with properly developed protective antibodies, but potentially damaging to the health of a baby whose protective immunity was inadequate. Professor Hutchinson pointed out that the peak level of vulnerability comes at about five to six months, when the infant is gradually losing the immunity inherited from his mother, without having fully developed his own immune system. He challenged Professor Berry's conclusions that the structures of the immune system were normal, pointing out that whether that was so or not, it would not determine whether the immune system was functioning normally. "The structure of the organs ... doesn't actually tell you how the cells within those organs are functioning." He accepted that cells were present making IgG, which was compatible with a small amount of IgG in the circulation, but questioned whether they had matured sufficiently to produce a normal level of IgG. Professor Hutchinson was unable to explain to the jury how Matthew had died. All he was able to say was that according to his tests, Matthew lacked protective antibody. Dr Rushton supported the view that although the structure of the relevant organs might appear normal, it did not follow that they were functioning normally. Dr Drucker, a reader in microbiology, specialising particularly in researching microbial causes of SIDS, pointed out that Professor Hutchinson had tested Matthew's sample time and again, because the antibody level was "so unbelievably low", so low that it was virtually below the limit of detection. He explained that without antibodies to provide protection, organisms which would otherwise be relatively harmless in an adult would have a very serious effect in a baby which lacked appropriate immunity.
128. Matthew's death triggered the investigation which culminated in these convictions.

The essential features of the Crown's case

129. We must now stand back from the detail of this vast body of evidence and note some of the main further features highlighted by the Crown and developed in evidence, and in submissions before us. The context, it will be remembered, is that our study of the details so far has not demonstrated any single piece of evidence conclusive of guilt. The matters which have created concern among some of the specialists have to be considered against the evidence of other reputable specialists which indicate possible natural explanations for the events with which we are concerned. In reality, the Crown's case therefore depended on specialist evidence about the conclusions to be drawn from the history of three infant deaths and further ALTEs in the same family.
130. The first of the specialists called by the Crown was Professor Meadow. He was particularly concerned by the extreme rarity of a third infant death in the same family, coupled with two earlier ALTEs, involving one of the children who subsequently died, and a fourth child, who did not. No natural cause was identified. In addition, the deaths or ALTEs occurred very soon after the baby in question appeared to be fit and well.
131. More specifically, in relation to Jason, when asked to take an overview of events on 13th and 14th June, Professor Meadow noted that he was a mature healthy baby who, without any significant previous incident, had suffered an unexplained ALTE, from which he made a very rapid recovery, and then died suddenly a week later, shortly after having been seen well.

"That means that on that day, he hadn't got any serious infection or disease going on, he appeared well. So something very sudden happened on that day."

He went on to note that:

"The fact that a previous child had died in the family is relevant because that combination of circumstances, that sort of story is one that is very typical of a child who has died as a result of smothering. So my medical diagnosis there would be probable smothering."

Professor Meadow also took account of the subsequent post mortem findings in relation to haemosiderin in Jason's lungs, which was "one pointer to previous smothering".

132. When asked to address Jade's ALTE, Professor Meadow stressed that the event at the age of 3 months was "very unusual". Jade had suffered a genuine ALTE without any apparent cause from which she had made a remarkably rapid recovery. The event was unexplained.

"In the context of the family as a whole it is of importance, because one of the reasons for such an event as this is smothering or (a word used) airways obstruction could cause a bout like this. And certainly that would come into diagnostic probability for a paediatrician reading these notes in the light of what has happened to other children in this family and reviewing those records."

133. Finally he was asked to turn to Matthew. As we have already recorded, he did not regard Matthew's first admission to hospital as an ALTE, although the episode was "worrying". It had significance because of the way in which Mrs Cannings had described it, and it was said to have arisen very shortly after a feed. When he was asked to address Matthew's death nine days later, he highlighted that the event was a terrible tragedy, and that a third death in the family was a "rare event, very rare". He noted that Matthew's death had followed very shortly after he was seen to be well, and that the time that had elapsed was very unusual. He was asked, "As a clinician what opinion do you have about Matthew's death and any possible cause, please?". He replied:

"... Firstly, the investigations and the pathologists did not find a reason for him dying. For me, the unusual feature is death so soon after being seen well, the fact that there had been previous deaths in the family and the fact that he had had an episode of some sort only nine days before he died that caused him to be assessed in hospital, because those features are ones that are found really quite commonly in children who have been smothered by their mothers. So the diagnosis for me, the clinical diagnosis, would be this was characteristic of smothering. ... One then goes on to say 'Well, is it possible it is a condition that is not yet understood by doctors or described by them?', and that must always be a possibility, but nevertheless as a doctor of children I am saying these features are those of smothering."

134. We have referred to Dr Ward Platt's views in relation to Gemma and Matthew. In summary he was hesitant to accept that Gemma's death fell within the SIDS category, but he was content that that "may be" the correct diagnosis. Nevertheless, for reasons of "pattern within the family", although there was no clinical evidence to suggest that Gemma's death was anything other than natural, serious question marks hung over it. "It has been repeatedly found that when everything comes out that in fact even the first death turned out not to be natural". Looking at Matthew's "episode" and later death, he agreed that looking at them entirely individually, there was nothing to suggest smothering. However his belief in the overwhelming likelihood that the death was a smothering event related to the "pattern in the family".
135. Jason's ALTE was respiratory in origin, and not primarily a circulatory collapse, and the evidence relating to haemosiderin provided strong support for this view. Jason's rapid recovery itself suggested that his earlier problem had resulted from lack of oxygen. This was, he said, "a very severe event ... primarily in my opinion respiratory, we have the evidence of the lungs and we have a previously well child. As a clinician this adds up to me to a child in whom there has been smothering". There was clear evidence that Jason had been under significant stress.
136. The same applied to Jade, a view reinforced by the increased levels of lactate, pyruvate and alanine, and hydroxybutyrate, "the body is becoming short of oxygen and finding other ways to get hold of it". This severe response showed characteristics of deprivation of oxygen. "Having ruled out all other possible conditions I was left with a strong diagnosis this was most likely to be a smothering episode", and as we have indicated, he rejected any significant gastro-enteritis.
137. In relation to both Jason and Matthew, he believed that the earlier admissions to hospital were themselves "highly suggestive of a baby who has been killed", adding that this was described in the literature. In the final analysis, as we read it, the main thrust of his opinion was that although the rarity of three infant deaths in the same family was not of itself a compelling reason to conclude that harm was deliberately inflicted, the pattern revealed by the history as a whole was compelling.

Infant deaths in the same family

138. This leads us to address the critical questions which arise when three unexplained infant deaths occur in the same family. In addition to the concerns already noted about an excessively dogmatic approach, we must immediately note a substantial body of research, not before the jury, and received by us in evidence, suggesting that such deaths can and do occur naturally, even when they are unexplained.
139. The Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI) of Sudden Unexpected Death in Infancy (SUDI) was published in 2000. The list of contributors included professional witnesses of distinction, some of whom gave evidence or were more generally involved in this case.
140. Described in the Preface, written by Professor Meadow, as one of "the most comprehensive scientific paediatric research studies to come from the UK", it is pointed out that although the problem of the unexpected death of a baby appears to received its first mention in the context of the judgment of King Solomon, it is only in the "last fifty years or so" that medical interest and research has focused on it. In 1971 Sudden Infant Death Syndrome (SIDS) became a registrable cause of death in England and Wales. It was applied to an acknowledged category of infant death, caused naturally, and for which no blame could be attached to bereaved parents. In 1997, by when the registration system was believed to be fully effective, 27% of post-neonatal (that is between the ages of one to twelve months) infant death was attributed to SIDS. The other proportions were congenital anomalies 23%, infections 19%, immaturity related conditions 7%, external conditions 5%, other specific conditions 3%, other conditions 15%, asphyxia anoxia or trauma 1%. Those figures were not fixed. It was recognised that a proportion of deaths registered as SIDS variously estimated at between 2% and 10% of those registered as such, might well have resulted from harm inflicted by a parent. Even so that left a substantial proportion of infant deaths which did not. In other words, the possibility that parents were not involved in the sudden deaths of their children was, as Mr Mansfield graphically described it, not a remote academic possibility. Although the precise proportions may vary from study to study, these findings continue to reflect the broad view of expert evidence in the profession, both here and abroad. The jury at trial was told by Dr Ward Platt, the paediatric clinician called by the Crown, that he subscribed to the view "that the overwhelming majority of SIDS deaths are entirely natural". On the basis of the CESDI SUDI-study he believed that about 6% of SIDS were "extremely likely" to have been the result of maltreatment, and a further 8% or so were ones where maltreatment was considered to have played an important part in the chain of events leading to the infant's death.
141. The same remote academic possibility is absent in the case where there have been two sudden infant deaths in the same family. The background to the CONI paper was work done with 6373 families which had lost a baby in "cot death" circumstances. It claimed to be, and we have no reason to doubt that it was "by far the largest follow-up study of families who have had a cot death". It is sufficient to repeat the conclusion which was, "that the occurrence of a second unexpected infant death within a family is not a rare event and is usually from natural causes". Our inexpert analysis of the study does not suggest that the conclusion is unsupported by the research.
142. Nowadays, and we may add, happily reflecting huge advances in our knowledge about and ability properly to care for infants, instances of three infant deaths in the same family are very rare indeed. The scope for research is therefore limited. This stems not from a lack of professional interest in the problem, but from the absence of sufficient opportunity, for as will be readily understood, healthy living infants cannot be subjected to ill-treatment for the purposes of experimentation. Nevertheless some recent research demonstrates that three sudden, apparently unexplained infant deaths in the same family, do not invariably result from unnatural causes. Thus, Professor Carpenter referred us to a study by Emery and Wolkind (this is the same Professor Emery whose research took him to meet Mr and Mrs Cannings, and later Dr Scott, and whose work in this field before his death was greatly respected) entitled Recurrence of Unexpected Infant Death in 1993. This study found three families in which there had been three unexpected infant deaths. Professor Carpenter summarised Professor Emery's notes as follows: "For one family permission to make enquiries was refused; these three deaths were certified as natural" (this was a reference to the coroner's verdict). Returning to his report, "In the second family careful pathological investigation revealed a congenital anomaly which was present in all three infants. The third family was fully investigated. The final case conference concluded that all three deaths were unexplained SIDS. The notes indicate that the possibility of filicide was discussed at this conference, but that the evidence did not support this explanation."
143. It is noteworthy that in his evidence at trial, and assuming he was referring to the same study by Emery and Wolkind, Professor Meadow appears to have reported the results of this research. He said that "They found that – I think they only found one family, one family, in which there were two deaths that they considered were genuinely unexplained and the other cases they either found a genuine disease (which was not found in these children) or they found circumstances that made them believe the child had been killed." Professor Carpenter, plainly assuming that Professor Meadow was referring to the 1993 study by Emery and Wolkind described this evidence as a "travesty". Having been supplied with a copy of the study and read it for ourselves after the hearing was concluded, it seems that Professor Meadow must have been referring to a passage in the study which reads:

"However, only five (9%) of our total series were assessed as being true or idiopathic cot deaths and in only one family were both deaths in this category, suggesting that the chance of recurrence is very small and probably no greater than the general occurrence of such deaths ... only five (9%) were found to be true or idiopathic cot deaths (SIDS)."

144. The study itself refers to three families which had experienced three infant deaths, without any detailed analysis. We do not know whether Professor Meadow had the advantage of reading Professor Emery's notes about these three cases. If he did, his evidence on this point was indeed a travesty: if not, it would be unfair to criticise him for knowing less about the three families than we have discovered from Professor Carpenter's analysis of the notes themselves. In any event our prime concern is the correct analysis of Professor Emery's notes, and the problems of three infant deaths in the same family.
145. The CONI study itself also referred to a number of cases where three infant deaths had taken place in the same family. Some of these deaths had taken place before the CONI study began, and it was not possible to do more than record contemporaneous findings. The study itself identified what were described as three families with three unexplained infant deaths and one where the deaths were due to "covert homicide". Recognising that not all these cases were fully investigated, according to Professor Carpenter, the study was left with one fully investigated case of three unexplained deaths (in which the possibility that one or more of the deaths might have been unnatural was specifically considered) and one family in which the deaths were due to triple homicide.
146. There is a further area of information addressed by Professor Berry in his commentary on the evidence prepared by Professor P.J. Fleming, for possible use in the course of this appeal. Professor Fleming was not called to give evidence, but Professor Berry has studied his research. He noted that Professor Fleming had been "involved in the care and investigation of three families over a period of twenty five years who have lost three infants suddenly and unexpectedly of natural causes". Professor Berry would have wished to know whether there had been full post mortem examination, and whether the deaths were treated as SIDS, strictly comparable to the present case, or whether they were to some extent explained. We therefore approach this evidence with a degree of caution. If it had stood alone, and without implying that the reported opinion of Professor Fleming was in some way inadequate, we should not have found it possible, acting on Professor Berry's report of it alone to quash the convictions, and indeed if that had been the single point available to Mr Mansfield, he would no doubt have pursued his application to call Professor Fleming. In the circumstances, our approach is that this material is entirely consistent with the conclusions reached by Emery and Wolkind, as revealed by Professor Emery's notes, and the results of the CONI research as explained to us by Professor Carpenter, and helps to fortify our analysis.
147. For present purposes, we should also note that, in the course of studying Sudden Infant Death Syndrome in Oxfordshire, between ten and twenty years ago, Professor Golding herself came across and studied two other families where there were three sudden infant deaths in the families. It seems clear from the context of her evidence, that she concluded that they were all properly categorised as SIDS. From her professional point of view, she was hugely interested in the phenomenon of four infant children suffering death or ALTEs within Mrs Cannings' extended, wider family. This evidence, as well as her evidence about new discoveries in genetics (see paragraph 35), illustrate two of the distinct threads with which this judgment began, first, the identification of the correct starting point where there is more than one unexplained infant death or ALTE, and second, a specific area of continuing research, putting it broadly, genetics, in which it is reasonable to hope that limitations on our present knowledge will be removed.
148. We need not labour these points any further. What is abundantly clear is that in our present state of knowledge, it does not necessarily follow that three sudden unexplained infant deaths in the same family leads to the inexorable conclusion that they must have resulted from the deliberate infliction of harm. There is acceptable evidence that even three infant deaths in the same family may be natural, and may indeed all properly be described as SIDS. At the risk of repetition, we emphasise the passage in the CONI study (see paragraph 23), where the third birth helped to establish not that the earlier deaths resulted from deliberate harm but that they were natural.
149. In fairness to Dr Ward Platt, we must record his opinion was that the history of this family provided "strong grounds for considering that they (the deaths and the ALTEs) may all be the direct result of imposed airway obstruction by their mother" (our emphasis). In his written statement before us, Dr Ward Platt pointed out that he had alluded to literature published by the date of trial which suggested that about half of the "recurrent SIDS" appeared to be accounted for by unnatural death. He relied in particular on the combined results of papers published by A.J. Waite, Repeat Cot Deaths in Families Enrolled Onto A Support Programme, published in 1996, and the study by Wolkind and others, to which reference has already been made, Recurrence of Unexpected Infant Death. This leads him to the conclusion that once there is more than one cot death in the family there is "a strong possibility that the cause may not be natural". He went on to point out that the CONI information suggested that where three sudden and unexpected infant deaths occurred in the same family, "infanticide must be given very serious consideration as a possible explanation". We do not disagree, but add that the use of language like this tends to confirm not merely that triple infant deaths in the same family are very rare, but also, perhaps at the risk of repetition, that material from reputable sources reflects at the very lowest, an acceptable view of our current state of knowledge, that where three such deaths occur the causes may indeed be natural.
150. Next we must address the brief interval between the time when the infant was last seen well, and the onset of death and near-death. Professor Meadow told the jury that he set store by evidence that the interval between the infant appearing to be well, and then found close to death, or dead, was very significant. "Anyone who has dealt with ill children knows it takes time for the symptoms to develop", and he continued "This very fact of there being a very short time interval between being seen well and perhaps taking feed and being seen dead means it is something quite extraordinary that has happened . … This is the big issue of it."
151. This point was specifically addressed in relation to Jason's death. "He died suddenly a week later (that is after the ALTE) an hour and a half after having been seen well. That means that on that day, he hadn't got any serious infection or disease going on, he appeared well. So something very sudden happened on that day." In relation to Matthew's death, much the same point was made. "There are very few things that cause someone to die very, very fast in 20, 25 minutes. So that interval is an unusual one … it is a very very brief period … I think of the known described conditions for children becoming ill so suddenly and dying within minutes of being seen well are really very, very few indeed and I don't find a likely diagnosis of a naturally known condition."
152. On this issue, again, we now know more than we did. Professor Berry notes the further analysis by Professor Fleming based on the CESDI SUDI data in relation to the interval between the baby last being seen well, and being found dead. This analysis demonstrates that 9% of the babies who died during the day (as these three babies did) were apparently well fewer than ten minutes before they were found dead. This is described by Professor Berry as "helpful" clarification. Whether or not it is accurate to say that this information was not "previously generally available" (some of our reading suggests that it is not) it was, in any event, not before the jury. If it had been, it would inevitably have resulted in a difference, if only of emphasis, on an aspect of the evidence to which the Crown attached great importance.
153. The third feature arose from the very fact that Jason and Jade had suffered ALTEs, although in Matthew's case, the occasion of his visit to hospital on 3rd November, was not so regarded. According to this argument, experience and research both showed that it was "very common" for an infant who had been smothered to have had an ALTE, at the age of two or three months onwards, and after being rushed to hospital and resuscitated, some seven to ten days later for a further ALTE to happen, or worse, for the infant to be brought back to hospital moribund or dead. The question that was begged was whether it was possible to conclude that the infant in question had indeed been smothered. The defence, through Professor Golding, accepted that infants who had suffered ALTEs were indeed at greater risk of SIDS, or, in other words, that a child who had suffered a natural ALTE appeared to be at greater risk of suffering an unexplained natural death. That, obviously, is a dramatically different view.
154. For the moment we shall ignore the question of the pattern of these events, and simply ask ourselves why the death of a child who has suffered a previous ALTE should lead to the conclusion that the death itself resulted from the deliberate infliction of harm. Professor Meadow himself described a "lot" of research on following up children who have different sorts of episode or episodes. He continued "There is surprisingly little firm evidence now that preceding apnoea or acute life-threatening events are a predecessor of natural death. At one time it was thought to be much more important than it is now." The theme of constant research, and changing views is well echoed in this observation.
155. The CESDI SUDI study asked parents whose infants had died whether there had ever been "an episode in which he or she became lifeless?" 11.7% of parents of infants who subsequently died of SIDS said that this had happened. These reports suggested episodes of cessation of breathing or change in colour, and approximately half of them were sufficiently serious for the parents to have taken the baby to hospital. Compared with the infants used for the purposes of control, who were healthy infants taken from the list of patients of the midwife who had delivered the baby that died, so far as possible to match that baby who had died, a much greater proportion of infants whose deaths were regarded as SIDS had suffered in this way.
156. In our view this material serves yet again to demonstrate two constant themes of this judgment, now of direct and specific application to the issues which arise for decision here. While the speed of research is gratifying, one unintended consequence is that it sometimes creates doubt about what were once thought to be certainties. And what was confidently presented to the jury as virtually overwhelming expert evidence providing the necessary proof that Jason and Matthew's deaths resulted from the infliction of deliberate harm, should now be approached with a degree of healthy scepticism. On three specific issues, first, the rarity of three natural unexplained infant deaths in the same family, second, the interval between the infant's death, or near-death, and the last time when that infant appeared to be well, and, third, the possible significance of an ALTE preceding death, the evidence before us presents a different picture, and one more favourable to the appellant than that which was before the jury.
157. At trial, understandably, the Prosecution relied on a number of additional features of the evidence to support the case that the deaths were unnatural. The six incidents were linked. Each of the appellant's four children suffered serious ALTE or death, or both, while small babies. They relied on the "pattern of events", suggesting that this "pattern" was reinforced by the fact that all the relevant events, the deaths and the ALTEs, occurred when the baby was in the sole care of the appellant. By contrast, when the babies were in hospital, or being cared for by both the appellant and her husband there were no such events. Factually, that was correct, but of course, with her husband at work, it was inevitable that the appellant would be in sole charge of the children at the time of day when research establishes that many SIDS in fact happen, or are discovered. We acknowledge the force of the argument that all these incidents occurred when Mrs Cannings was alone with her children, and that no incidents occurred when anyone else was present, but its force is less than overwhelming, and as we shall see fails to establish the significant pattern relied on by the experts called by the Crown. It was also suggested by the Crown that the appellant had not told the full truth about the workings of the apnoea alarm. It was strange that in relation to the incidents involving Jason and Matthew, she had not heard the alarm start to sound, or in relation to the Jade ALTE, that she had forgotten to turn it on. This meant that she did not reach the baby until he was either extremely ill, or indeed dead. Moreover, it was suggested that on the morning of Matthew's death, the alarm should have been clearly audible as soon as it started. In relation to Matthew's ALTE there was said to be a lengthy unexplained gap before the appellant made the 999 call, and on the occasion when he died, it was surprising that the appellant had sought assistance from her husband before the ambulance was called. We have addressed these questions, and a number of other matters specific to the individual cases in the course of the earlier narrative, and shall not repeat ourselves.
158. In any event, we doubt the aptness of the description, "pattern". It would be more accurate to describe the events in the Cannings family as a "series" rather than a "pattern" of events. In Gemma's case, her death was not preceded by an ALTE. There was none. In Jason's case, there was one ALTE, which if the Crown was right followed the deliberate infliction of harm by Mrs Cannings, quite uncharacteristically at just the moment when the health visitor was arriving. This was followed a few days later by his death. In Jade's case, however, an ALTE which was said by the Crown to have been a virtual death, was not preceded by an earlier ALTE, nor followed either by an ALTE or death. In Matthew's case, the first incident was not regarded by the Crown's medical experts as an ALTE at all. So his death was preceded by what was no more than an "episode" which had led his mother to call for help although his life was not in apparent danger. On this analysis, the history of each child was different from every other child.
159. In relation to patterns, we must revert to epidemiology, which is particularly concerned with the search for pattern. Professor Golding adds the necessary element of professional expertise to what we recognise is our own lay reaction. She did not believe that there was a "pattern" in the sense which Dr Ward Platt thought gave rise to suspicion. She pointed out that if you were looking at three random cot deaths you would get the sort of pattern which was established here. That would not give rise to a suspicion that the set of deaths was unnatural. This was a normal sort of pattern for repeated cot deaths, and the fact that these deaths happened in the same family suggested that there might be a single factor running through the events, but that was not determinative of the question whether the common factor should be ascribed to the deliberate actions of the children's mother, or to some other undiagnosed natural cause. She added that although this history required a close study to be made of the family, she could find "none of the patterns that one would normally see in the background of somebody who was out to kill their children." In our judgment the mere fact that specific natural causes were not established for any of the deaths did not lead to the inference that the infants had been smothered, or deliberately harmed, but rather left open the possibility that SIDS should not be excluded.
160. As against the factors extraneous to the expert evidence relied on by the Prosecution, the appellant invited us to consider a number of other factors which, again in a common sense way, were relevant. No-one doubted that each of these babies was a wanted child, blessed with love and affection and care from both parents. There was no suggestion of ill temper, inappropriate behaviour, ill-treatment, let alone violence, at any time with any one of the four children. Although three of them died very young, Jade did not, and, no such suggestions have ever been made. Of itself, when four infants are said to have been deliberately harmed, that is an unusual feature. It is distinct from the additional fact that if it was indeed right that the appellant had tried to kill each of her four children, during each of these six incidents, there was an absence of the slightest evidence of physical interference which might support the allegation that she had deliberately harmed them. It is of course possible to smother a baby without leaving any physical signs discernible on medical examination or at post mortem. Nevertheless, given that all four children were said by the Crown to have been subjected to violence sufficient to cause death, the absence of any physical signs of injury was somewhat surprising. There was no fresh copious bleeding in the lungs of the dead children, and no petechial haemorrhage. There were no pressure marks to show as reddening in the area of the mouth and nose, nor blood or bloodstained fluid in the nose. No bruises were discovered on the outer skin surface, or indeed subcutaneously. The fraenulum, in each case, was undamaged.
161. We must address criticisms about the appellant's behaviour. In our view by the time Jason, and then Jade, and then Matthew were born, particularly when troubles enveloped Matthew, the appellant was faced with recurring disasters which made comprehensible any form of response which, on cold forensic analysis, would otherwise appear strange. We also understand the argument that the appellant would not have killed the children (as the jury found that she had) unless she was suffering from some form of personality disorder or psychiatric condition. There was no evidence to sustain any such diagnosis: indeed it was to the contrary. To a layman, it did not make much sense that the appellant killed the babies whom, in the judge's words, she "cherished", unless her state of mind was, to some degree at any rate, abnormal when she did. We recognise that this is a factor of some importance, again noting that of itself the absence of such a diagnosis does not preclude baby killing. Mr Mansfield submitted that if the appellant had indeed murdered her children, and subjected each of them to life threatening incidents, the logical approach to her repeated pregnancies was that she was having babies in order to kill the baby she was carrying after it was born. In its modified, less dramatic, but not less forceful version, Mr Mansfield suggested that these facts meant that the appellant must have been becoming pregnant knowing, at the very least, that there was a serious risk that she would try and kill them. That predicated an extraordinary state of mind, completely out of character, contradicted by the evidence of both her family and outsiders about the love and care she bestowed on her children, and undetected by the distinguished psychiatrist who examined her. Given the absence of any indication of ill-temper or ill-treatment of any child at any time, we acknowledge the force of this argument.

The Appeal

162. After this lengthy discussion, we can deal relatively briefly with the issues raised in the appeal. In response to Mr Mansfield's main argument that these convictions are unsafe, Mr Dunkels reminded us of the constitutional primacy of the jury. This jury returned its verdict after a long trial during which it had demonstrated a keen and alert interest in the issues, and had the opportunity to evaluate not only the evidence of the experts on both sides, but that of the appellant herself, and a number of other civilian witnesses. In effect, he submitted in careful and measured terms, that the convictions were safe, and that no sufficient ground had been demonstrated for interfering with them. We reflected closely on these submissions before concluding that, for the reasons set out in this lengthy judgment, the convictions are unsafe.
163. We therefore need not deal with all Mr Mansfield's submissions, some of which we have already addressed. We shall deal with the most important arguments in their logical order. As our analysis of the evidence called by the Prosecution demonstrates, there plainly was a case for Mrs Cannings to answer. It would therefore have been wrong for Hallett J to have withdrawn it from the jury. Looking at the evidence at the end of the trial, we can discern no basis for interfering with her decision that the issues should not be withdrawn from the jury. This is an extremely rare course for a judge to take, and Hallett J was entitled to decline to take it here.
164. Mr Mansfield drew our attention to the summing up by Jack J at the trial of Trupti Patel, and his directions about the approach to be taken to the deaths of three infants in the same family. In doing so, Mr Mansfield was not seeking to criticise Hallett J's directions at the time when she gave them, but suggested that Jack J was reflecting the greater understanding of and insight into these problems following Sally Clark's successful appeal to the CACD, in which Hallett J was a member of the court. Equally, he did not submit that such a direction would be compulsory. In essence, he suggested that in this kind of case, such a direction would normally be appropriate.
165. Jack J directed the jury:

"I said that I would have a direction for you on the subject of there being three deaths. You have heard from some of the Prosecution witnesses the idea that the fact of three deaths makes it more likely that the cause was unnatural. Certainly with three deaths one must be suspicious and look the more carefully, for it is potentially a very serious situation. But I am going to ask you to put out of your minds the idea that because there are three that makes it more likely that the causes are unnatural: that is asphyxiation by Trupti Patel. I think that would be a dangerous approach in this case for two reasons.

The first is this: suppose that something happens and there is only one possible event as the cause for it. However rare or common that event may be, it must be the cause: straightforward. If it is rare the unexpected has happened. Suppose, though, that there are two possible events as the cause. One is a common event and one rare. It can then be said that the common event is the more likely cause. Suppose, however, that the two events are both rare; perhaps very rare. They are nonetheless equally likely as the cause even though they are rare, because they are competing with each other to be the cause.

So it is not enough to say that an event is rare so it is unlikely to be the cause of something. One has to look at the likelihood of the other possible cause, or other possible causes. That is the danger with what may be happening here in saying that three SIDS deaths in a family would be very unusual, therefore the deaths are unnatural. How rare would three asphyxiations be, particularly where, as is the case here, the mother loved her children and was immediately distraught and regretful? We simply do not know. We have not had any evidence about that. It is hardly common is it? That is obvious. That is the competing cause of the deaths and nobody has evaluated its likelihood."

166. In the present trial Hallett J's directions required the jury to consider the case "for and against" Mrs Cannings in relation to the counts of murder of Jason and Matthew separately. The evidence, she pointed out, was not the same. She added, rightly, that it did not mean that all the other evidence had to be ignored when the jury were deciding either case, because, as she explained, "You have heard all the evidence to help you in reaching your verdicts, and the evidence in relation to one child may assist you when considering the count in relation to another child."
167. She reminded the jury that both sides were inviting them to look at all the evidence. She then went on to deal with Gemma.

"You have heard of course about Gemma. You know that Gemma is no longer the subject of a murder charge, although the defence elicited that Mrs Cannings was once charged with Gemma's murder ... be careful how you approach Gemma's death. It was a long time ago. We do not of course have the kind of results and tissues still available that Mr Mansfield has been referring to that would help you in knowing from either side's point of view anything more about Gemma's death. You have had to hear about Gemma's death because obviously it is part of the background and it is relevant. It may, for example, be relevant as to whether or not there is a genetic defect. But be very wary how you approach Gemma's death. You know the pathologists carried out a very careful post mortem and decided that the death effectively was SIDS, or cot death, and no suggestion of maltreatment."

168. She ended this part of her directions in unequivocal terms:

"You have not heard about Gemma's death to justify the kind of approach referred to by Mr Mansfield; the Lady Bracknell approach. This is not a case whereby you could say "to lose one baby is misfortune, two carelessness, three murder". As you will appreciate, members of the jury, that is just inappropriate – totally."

169. Hallett J then went on to illustrate how the function of the burden of proof in the context of the exploration by the defence at trial of a number of possibilities which might have accounted for the death of one or other or all of the children, or indeed each of the ALTEs. She said,

"Do not think that when Mr Mansfield called an expert before you he is under any kind of duty to prove that expert is right. He does not have to establish that any particular incident was natural in causation or that it was due to as yet unknown or unidentified causes. The possibilities are put before you because firstly, we know that babies do sadly die of natural causes, as yet possibly unknown or unidentified, but also there may be many contributory factors as to why a baby may die. So when you hear the evidence called by the defence, very much bear in mind the submissions made by Mr Mansfield, and which I wholeheartedly endorse, that he does not have to prove that any of the theories of the experts he has called are correct."

170. She then continued to illustrate the point using Jade's admission to hospital as an example.

"It is not for the defence to prove that Jade was suffering from severe gastro-enteritis. Jade's admission into hospital can only assist the Crown if the Crown can prove that it was due to smothering ... always bear in mind, members of the jury, the burden of proof and who must prove what."

We have some sympathy for the jury. We have to reflect an anxiety which has struck us throughout our own deliberations, whether notwithstanding these clear directions, the whole course of the trial, the sheer number of experts called by the defence, and the complex specialist fields in which these distinguished men and women worked, the jury may not, inadvertently, unconsciously, have thought to itself that if between them all, none could offer a definitive or specific explanation for these deaths, the Crown's case must be right.

171. In any event, in our judgment, Hallett J's directions were sufficient. In any future case, at any rate while so much about Sudden Infant Death remains unknown, a trial judge will no doubt wish to take account of our analysis of the fundamental approach to cases of this kind, as well as Jack J's directions in the case of Trupti Patel. In the result, however, Hallett J's directions, or such absence of directions as Mr Mansfield was able to identify, do not undermine the safety of these convictions.
172. We must acknowledge an interesting debate on what we shall for convenience summarise as the "reasons" argument, that is the problem that may arise from the stark fact that in our criminal justice system the jury delivers its verdict without providing its reasons. Arguing from authority, both domestic and in the European Court of Human Rights, Mr Mansfield suggested that these provided support for the extra-judicial observations of Sir Robin Auld in his Review of the Criminal Courts of England and Wales, 2001. In particular he focused on this part of Sir Robin's conclusion:

"... The time has come for the trial judge in each case to give the jury a series of written factual questions, tailored to the law as he knows it to be and to the issues in evidence in the case. The answers to these questions should logically lead only to a verdict of guilty and not guilty."

173. This raises an issue which may have to be addressed in another appeal, but does not arise here. We shall therefore confine ourselves to three brief observations. First, Sir Robin's proposal goes much further than the now traditional and appropriate use of a series of written questions provided by the judge to enable the jury to work its route through potentially complicated verdicts. This is most frequently exemplified in cases where more than one defendant is charged with murder, and the combination of lack of intent, provocation, self-defence, and even diminished responsibility is simultaneously before the jury. Second, it should not be assumed that Sir Robin's views on this topic necessarily reflect the views of the judiciary as a whole, or were supported in the judiciary's response to his proposals. Third, the proposal, and Mr Mansfield's submission, overlook the principle that although each member of the jury participates in the verdict, each must arrive at his or her conclusion by a conscientious personal examination of the evidence in the context of the legal principles which have been defined by the trial judge. Each member may be convinced, or doubtful, of guilt for reasons which are different to those of each of the other members who nevertheless, for their own conscientious reasons, are agreed on the result.
174. These are issues of vast complexity, and huge potential importance to the community as a whole. Any changes of this kind should be decided in Parliament, which, notwithstanding current reforms of the criminal justice system, after Sir Robin Auld's Review, has not adopted his proposals on this topic.

Conclusion

175. We have now given our reasons for concluding that these convictions are unsafe. We have received significant and persuasive fresh evidence, which was not before the jury, some of it the result of further research, or research published post trial, into the problem of SIDS generally, and some specific to Mrs Cannings and her extended family. The expert evidence was absolutely critical to these convictions. In our judgment the fundamental basis of the Crown's case, based on the extreme rarity of three separate infant deaths in the same family, and the pattern of events in this particular family is, for the reasons we have given, demonstrably undermined. What is more we are satisfied that there is a realistic, albeit as yet undefined, possibility of a genetic problem within this family, which may serve to explain these tragic events. For the moment therefore we cannot be sure either that these deaths were not true SIDS, or that, although properly categorised as SIDS at present, they may not come in due course to be regarded as natural deaths resulting from an explicable, possibly genetic cause. In view of these conclusions, we need not do more than record a further concern which troubled us during our deliberations, which is whether the multi-factorial aspect of each of these incidents was sufficiently addressed. These concerns did not contribute to the quashing of the convictions, and as they do not affect the result, we need not deal with them further. In subsequent cases, this issue may arise more starkly than it does here.
176. In view of the fact that, although not identical to each other, this is the third case of its type to come before the courts in 2003, we must add these further observations.
177. We recognise that the occurrence of three sudden and unexpected infant deaths in the same family is very rare, or very rare indeed, and therefore demands an investigation into their causes. Nevertheless the fact that such deaths have occurred does not identify, let alone prescribe, the deliberate infliction of harm as the cause of death. Throughout the process great care must be taken not to allow the rarity of these sad events, standing on their own, to be subsumed into an assumption or virtual assumption that the dead infants were deliberately killed, or consciously or unconsciously to regard the inability of the defendant to produce some convincing explanation for these deaths as providing a measure of support for the Prosecution's case. If on examination of all the evidence every possible known cause has been excluded, the cause remains unknown.
178. The trial, and this appeal, have proceeded in a most unusual context. Experts in many fields will acknowledge the possibility that later research may undermine the accepted wisdom of today. "Never say never" is a phrase which we have heard in many different contexts from expert witnesses. That does not normally provide a basis for rejecting the expert evidence, or indeed for conjuring up fanciful doubts about the possible impact of later research. With unexplained infant deaths, however, as this judgment has demonstrated, in many important respects we are still at the frontiers of knowledge. Necessarily, further research is needed, and fortunately, thanks to the dedication of the medical profession, it is continuing. All this suggests that, for the time being, where a full investigation into two or more sudden unexplained infant deaths in the same family is followed by a serious disagreement between reputable experts about the cause of death, and a body of such expert opinion concludes that natural causes, whether explained or unexplained, cannot be excluded as a reasonable (and not a fanciful) possibility, the prosecution of a parent or parents for murder should not be started, or continued, unless there is additional cogent evidence, extraneous to the expert evidence, (such as we have exemplified in paragraph 10) which tends to support the conclusion that the infant, or where there is more than one death, one of the infants, was deliberately harmed. In cases like the present, if the outcome of the trial depends exclusively or almost exclusively on a serious disagreement between distinguished and reputable experts, it will often be unwise, and therefore unsafe, to proceed.
179. In expressing ourselves in this way we recognise that justice may not be done in a small number of cases where in truth a mother has deliberately killed her baby without leaving any identifiable evidence of the crime. That is an undesirable result, which however avoids a worse one. If murder cannot be proved, the conviction cannot be safe. In a criminal case, it is simply not enough to be able to establish even a high probability of guilt. Unless we are sure of guilt the dreadful possibility always remains that a mother, already brutally scarred by the unexplained death or deaths of her babies, may find herself in prison for life for killing them when she should not be there at all. In our community, and in any civilised community, that is abhorrent.