Justice Without Science Isn't Justice.
By Frank Ward (Version 17/9/2003)

(Click on the links to read the appendices, use your back button to return. Editor)

When mathematicians or similar scientists say something is proven, there is no doubt it is so. The requirement is that it is always so found. When a court delivers a guilty verdict, there is often no such certainty, even though the language used often implies there is. The purpose of this article is to encourage further research into establishing why innocent people are on occasions found guilty and how widespread the problem might be.

My personal interest in the subject was first aroused with the Lindy Chamberlain case in Australia. The injustice, suspected by many people at the time, was only corrected by the courts on a chance finding of minor evidence several years later. While she was compensated financially, there was never any mention in the press of an investigation into how the initial erroneous verdict occurred. That seems a strangely cavalier attitude, since the justice system's prime purpose is to protect people from the wrong-doing of others and when someone is wrongly convicted, the system is violating its own charter. >From a scientific viewpoint therefore, if not a legal one, the causes of such errors deserve a very public exploration.

I have little knowledge of legal matters and it may therefore be considered presumptuous to speak out on such things. On the other hand, failures in the justice system could affect any one of us and as an actuary and an analyst, I can claim some knowledge that is relevant to the matter. The science of chance and probability is at the heart of actuarial science and clearly it should feature much in weighing up judicial evidence.

A much more recent civil case, involving the Equitable Life Office in Britain, made me realise that the Lindy Chamberlain saga was not perhaps a one-off, exceptional mistake. Law Lords, at the peak of their profession, allowed a wealth of factual evidence to be overridden by claims that in scientific terms were mere hypothesis and then compounded their error by making decisions that lacked consistency, a most important requirement of scientific logic (see Appendix 1).

This has greatly increased my concern about the quality of legal methodology and its apparent failure to embrace quite rudimentary scientific processes. So, naturally, I have been taking more interest in legal decisions that are reported in the press. Two other cases involving mothers charged with murdering their babies soon came to light. Some details of the first one are given in Appendix 2. While the judge at sentencing showed some concern for Mrs Angela Cannings by suggesting she must be ill, her honour naively fails to grasp that she is responsible for the apparently foolish verdict. Clearly a jury of butchers, bakers and candle-stick makers needs a lot of advice about what "beyond reasonable doubt" means, about the widely differing values to be given to factual and speculative evidence, about the bias inherent in expert witnesses chosen for their viewpoint, about the science of chance, about the need for consistency in all the evidence and about the greater seriousness of convicting an innocent person compared with acquitting a guilty one.

With practically everyone saying how pleasant and helpful the defendant was, I wonder how the jury in finding her guilty explained why such a talented actress is working as a supermarket assistant. It might hint at yet another bias in the legal process, with imaginative circumstantial evidence commonly being used against a defendant, but rarely to support them. It is also hard to understand how the jury could ignore the fact that Prof. Jean Golding was first called as a prosecution expert witness but on seeing the pre-trial evidence became a witness for the defence instead. She also negated the evidence of Sir Roy Meadow. This leads to another issue that is scientifically unsound and I think is legally unethical. Sir Roy lectures judges on how to interpret medical evidence and then gives evidence before judges in particular cases. This could well be a contributory factor in the judge's handling of this case.

There is a quirk of the legal system that introduces very definite bias against good people demonstrated in this case. If perchance this mother had been tried and convicted of murdering her first born back in 1989, legal practice would have required that event to be kept hidden from this jury. With no such black mark against her, the prosecution is allowed to make specific mention of that first child's death and use it as leverage, despite the pre-trial hearing finding that there was no case to answer in respect of that first death.

Thanks to the Internet, it is now easy to find related information and I soon found another report that provides a ray of enlightened sunshine that fellow actuaries and similar scientists will appreciate. It is a follow up to another case a few years ago where another mother, Mrs Sally Clark, was similarly convicted for losing two babies in undetermined circumstances (see Appendix 3A).

In the case of Mrs Clark, without knowing the jurors' deliberations, the judges at her first appeal were willing to say the acknowledged major statistical mistake in prosecution evidence wouldn't have changed the original jurors' opinion. The original verdict was in fact a 10-2 split which makes the appeal judges decision even more reckless. Nor does their decision, or even the original verdict, sit comfortably with the fact that the defendant was offered a lesser charge of manslaughter avoiding any prison term (which she refused), showing the prosecution wasn't wholeheartedly behind the charge of murder. Mr Stephen Clark, the defendant's husband, said the appeal judges had been "intellectually dishonest" and although Mr Clark is certainly not an independent party to the proceedings, any fair minded person would have to agree that is an accurate assessment. A selection of the judges' arguments is examined in Appendix 3B.

These and a few other cases I've explored highlight several prominent problems from a scientific viewpoint. The first is judges don't appear to appreciate the extra seriousness of convicting an innocent person (see Appendix 4). The second is the misleading nature of evidence from "expert witnesses" who have been selected by the prosecution and the defence (see Appendix 5). The third is the failure to distinguish clearly fact from opinion and hypothesis. Disturbingly, judge's opinions are also often relayed as fact (see Appendix 6 for some illustrations).

Such errors are then compounded by several unsound legal procedures that have crept into the legal system over the centuries, some introducing deliberate bias. These include jury selection, exclusion of evidence and the double-jeopardy rule. These are explored in more detail in Appendix 7.

The Equitable Life Office case mentioned earlier also illustrates a quite different aspect of court practice that is scientifically questionable - that the outcome is forced into being black or white. There were three court hearings, one found for the claimants, one found for the defendants and one found for the claimants in a 2-1 majority verdict. That is more like roulette than a just conclusion. Sadly, as Appendix 1 shows, in this particular case the verdict had no correlation with the evidence and of course a wrong decision is even worse than an arbitrary one.

The case of Mrs Sally Clark has just been reheard (29/1/2003) in a specially allowed second appeal and she has been exonerated. The justification was that evidence given by a second expert witness for the prosecution was shown to be faulty. It has taken the court system three years to come to the conclusion that practically everyone else acquainted with the case "knew" almost from the start. The defendant, her husband and colleagues had to prove her innocence by tracking down this error themselves. While Sally Clark was innocent until found guilty, the opposite also applied, so a shadow is thrown on yet another maxim of the law. It also highlights the inequity of pitting the resources of the state against the resources of the individual. However, putting the sole blame, or even the main blame, on the erroneous evidence is a very superficial analysis of the errors in this case.

Sally Clark would have been JUST AS INNOCENT if this faulty evidence hadn't been tracked down or if in fact the test it referred to had never been made. The real problem is clearly much deeper than these individual pieces of evidence. It is the same faults that appear in all the cases I've referred to, convicting someone when all the evidence was not CONSISTENT and then giving undue weight to selected information. Science requires evidence to be consistent to prove a proposition and however wise judges are they cannot ignore that. The scenario in the Sally Clark case is this. We know Sally & her husband were an apparently normal, law-abiding couple, we know it is the combative role, or at least the practice, of the prosecution to put a sinister connotation on the innocuous day-to-day events that happen to us all, we know the particular prosecution expert evidence was chosen to support their case and so involved deliberate bias and not being expert themselves, the judge and jury don't know how significant that bias was. That the judge and jury should let the unknown part overwhelm the known part, highlights the speciousness of the process. The same thing happened in the Chamberlain case, the Equitable Life Office case and Mrs Cannings case. Is it possible this error occurs in every case?

Mrs Angela Cannings has languished in gaol well over a year but at last an appeal has been scheduled for November 2003. The inconsistency of the evidence in her case is perhaps even more self-evident than Sally Clark's as Appendix 2 shows. Judges need to be more than just umpires of court procedure to ensure justice is done. In Sally Clark's case twice justice wasn't done and the judge's failure to highlight the inconsistencies in Mrs Cannings case has surely contributed to the unreasonable verdict.

This is quite an incredible indictment of the legal process. So much of what they claim in words and maxims is not born out in fact and the wisdom of the judiciary is questioned even more by their failure to react to such occurrences as to the occurrences themselves.

There is yet another case I have looked at briefly, involving a Mr Derek Christian, that appears riddled with inconsistent evidence. It also highlights again that customary habit for a prosecution to search out quite innocuous actions and put a sinister twist on them, just as was done in Mrs Cannings' and Mrs Clark's cases. Such drama, rather than cool analysis, cannot be expected to help bring out an intelligent decision from a jury. It is clearly a tactic aimed at capturing the minds of the jurors rather than informing them. Nor is it fair between educated, quick-witted people and those uncomfortable with erudite dissertations in an unusual and archaic environment.

Another questionable situation is that Sally Clark, prior to being charged with murder, was offered the lesser charge of manslaughter which would not have warranted a prison term. Now either she was suspected of murder or she wasn't, so what explanation is there for the prosecution to introduce bargaining. There is no scientific honesty in that type of manoeuvre. Arm-twisting and other bullying tactics are completely unbecoming a serious attempt at revealing the truth. It suggests the judicial system is not transparently honest and plea bargaining is yet another biased tactic that assists the truly guilty but does nothing for the innocent.

A small chance find gave both Lindy Chamberlain and Sally Clark back their innocence. Surely there must be at least one wise judge that recognises that justice by such a tenuous thread is not satisfactory and that something is seriously wrong when ordinary people reading news reports can see things more clearly than the courts. This latter situation could be partly explained by a failure within a court environment to see the forest for the trees and as a result giving undue weight to selective minutiae in the evidence, by being influenced more by manner than substance.

There is also the problem of the human mind being suspicious without due cause. It is recognised in Darwinism that this is a healthy trait and helps with survival and most animals exhibit the tendency - even cool-headed scientists. A recent case of alleged cheating in the "Who Wants To Be A Millionaire" program highlights the situation. The fallacy in the human mind arises because it does not distinguish between the significance of evidence coming before the suspicion and vice versa. If you tell me you are going to toss a coin six times in succession and each time it will come down heads and then you do it, I would be suspicious of the coin or your technique, since everything being fair, it should only happen once in sixty-four times. If however you told me you had just had six heads in a row, I'd be foolish to jump to such a conclusion. Many people are tossing coins all round the world. One in sixty-four of those would give six heads in a row and why shouldn't you be one of them? Practically all cases of circumstantial evidence involve the suspicion preceding the evidence and therefore are liable to this error. The current justice system fails because it UNDERVALUES the very real importance of all the evidence being consistent to establish guilt and OVERVALUES apparent coincidence in circumstantial evidence.

While the independence of the judiciary is a jealously guarded right, it should be limited to independence from government, not independence from the God-given rules of science.

While I have concentrated on two particular and similar cases, the validity of the arguments I have made apply to many others. Just recently (July 2003), Michael Shirley was found innocent of a murder for which he had spent sixteen years in prison. Appendix 8 gives more details and shows once again the slack way circumstantial evidence is evaluated and the medieval nature of judicial practices.

The police force and medical professionals embrace scientific advance with great success and there is no justification for the judiciary to remain aloof and resist change by brazenly claiming their methods have served us well for centuries. Quite clearly they haven't and it is time to trade in their creaking charabanc for a more reliable vehicle.

No modern day thesis is complete without a dabble in the realm of conspiracy theory. As light relief, Appendix 9 explores just that and with more seriousness suggests the kinds of improvements most needed by the judicial system.

The science-based failings identified in this article are summarised in Appendix 10. Some articles by other authors have been collected together as Extra Appendices and are available on request, but these can also be accessed via the Internet at the time of writing.

I welcome any criticism of this treatise and any further ideas for it. My sole personal interest in anything referred to is just the Christian ethic of caring about one's fellows.

Frank Ward (fcamward@scs.brisnet.org.au)

Appendix 1 - Equitable Life Office.

Equitable Life Assurance v. Certain Policyholders

The complainants were policyholders that had guaranteed annuity rate options exercisable at maturity (abbreviated G-holders below). They were claiming they were entitled to the identical bonuses as those policyholders without the guarantee. A summary of the case can be found in "Actuary Australia March 2001 Page 12" , a magazine of the Institute of Actuaries of Australia. The following is a precis of the most significant parts of letters from me published in "Actuary Australia" and in "The Actuary" (a UK publication) in subsequent months.

Somewhat dubiously, the judges chose to let assessments of claimed expectations override clear contractual statements. Surprisingly, they also made monetary decisions without considering where the money comes from and the impact this has on other innocent parties (The Equitable is a mutual life office totally owned by the policyholders alone). But most amazingly, although they gave case-deciding weight to the expectations of the complainants, they paid no heed at all to the expectations of all the other policyholders. Scientists know that an inconsistency is God's way of saying a conclusion is invalid and I'm sure he applies the same rules to judge's conclusions.

Here's a list of salient questions with what I believe are nigh indisputable answers.

1. Should non-G-holders contribute to a benefit they can't receive? No.

2. Should guaranteeing an annuity rate be considered to be guaranteeing the bonus too? No.

3. Should a judgment be made looking at just one aspect of the situation, ignoring both where the money comes from and the interests of the other policy holders? No.

4. What bonus can a policy holder claim to expect? Something hopefully, but nothing specifically, other than the same as any other identically placed policyholder.

5. Has any one of the G-holders claimed they were misled by anyone in particular, intentionally or otherwise? I don't believe so.

6. Is the purpose of a bonus to return to the policyholder any excess earnings from his contributions after meeting the contractual entitlements of his contract? Yes.

7. Should an unsubstantiable a posteriori claim "that the G-holders expected to get the same bonus as non-G-holders" be more significant than the substantive contractual statement "that the directors have absolute discretion in deciding each policy's allocation of bonus"? No.

8. Even if there were an adjudged wrong, does it make sense to give a decision that creates an even more definite wrong? No.

Answers to all the above questions are clearly opposite to the judgment made. Try as I can, I cannot think of any salient question that would favour the judgment.

There was no contradiction of any consequence from other writers in response to the points made above.

For another, INDEPENDENT assessment, see www.cookham.com/community/equitable/billdavies.htm (or Extra Appendix D). It is quite a good summary, though I don't see the evidence to attribute any fault to the Equitable managers. I do question one sentence it includes, namely "The Lords made a serious mistake in not considering more than the narrow legality of the case". While scientific logic is both definable and testable, the concept of legal logic has no such qualities. It has fictitious precision and in this instance is merely opinionated interpretation where the judges deem there is an IMPLIED condition and let it override an unquestionable right. We all know cases where judges have done just the opposite and used minor factual details to thwart the intention of a law.

The mathematical concept of "reductio ad absurdum" is a very powerful tool for exposing faulty logic. If all the Equitable policy holders had been G-holders it is clear how much bonus could be allotted to them. Why then should they get more bonus because there are other different policy holders as well?

Appendix 2 - Mother Accused of Murdering Her Babies.

The following is an abridged version, for space reasons, of a report that can be viewed in full at news.telegraph.co.uk for 17/4/2002. While the prosecutors in this case haven't been as inventive as those in the Lindy Chamberlain one, there is much similarity, with a noticeable lack of credit being given to the wholesome good character of the defendant.

Experts have no idea why a doting mother would kill her babies. By Stewart Payne (Filed: 17/04/2002)

AS Angela Cannings began a life sentence for the murder of her two sons last night, police and welfare experts admitted that they were no nearer discovering why she killed them.

No motive for the killings was offered during the six-week trial of Cannings, who was said to be a doting mother.

The jury listened to complex and conflicting evidence from dozens of expert witnesses. The prosecution argued that she had suffocated seven-week-old Jason in 1991 and 18-week-old Matthew in 1999, while the defence put forward several alternative possibilities, ranging from cot death to accidental chemical poisoning.

Their first-born child, Gemma, also died, aged 12 weeks, in 1989. Cannings was charged with her murder too but the charge was dropped before the trial. Paul Dunkels, QC, told the jury to regard Gemma's death as the "backdrop" to the case.

It was not disputed in court that all who knew Cannings, including health visitors, believed her to be a good mother.

Det Sgt Rob Findlay, who interviewed Cannings, said: "There was nothing in her behaviour which aroused suspicion. She was nice and helpful and, as was said in court, she was a caring mother, which makes it more difficult to understand."

© Copyright of Telegraph Group Limited 2002.

Appendix 3A - Criticism of Expert Evidence Given by a Prosecution Witness Against Another Mother Accused of Murdering Her Babies.

'Statistical error' in child murder trial By Celia Hall (news.telegraph.co.uk for 31 December 1999)

THE conviction of a solicitor for the murder of her two infant sons is unsafe because of an error in statistics given in evidence and a lack of understanding of probability theory, a doctor claims today.

Dr Stephen Watkins, a public health expert, says guidelines are urgently needed for judges in criminal cases when probability theory forms part of the evidence. Sally Clark was found guilty in November of smothering her sons Christopher, 11 weeks, and Harry, eight weeks, a year later.

An expert witness for the prosecution said the chance of a double cot death in her family was one in 73 million. The witness said the chance was "vanishingly small". This evidence may have played a significant part in the outcome of the case. Clark, from Wilmslow, Cheshire, was found guilty of double murder in a 10 to two verdict. She protests her innocence and plans an appeal.

Dr Watkins, director of Public Health at Stockport Health Authority, writes in the British Medical Journal: "It is speculation whether Sally Clark would have been acquitted without this evidence. But with this mathematical error prominent, the conviction is unsafe." He argues that the real chance of a double cot death in a family is one in 8,500 but says that even this figure is meaningless.

Dr Watkins said yesterday: "If statistics like this continue to be used in criminal cases, miscarriages of justice will occur and this may have been one. I can say that this particular evidence was wrong." Dr Watkins said understanding and interpreting statistics was an essential part of his job as a director of public health. He said: "This case concerned me particularly. We do not even fully understand cot deaths."

He says in the journal that from whole population data it has been calculated that the square of the population risk of cot death is one in 2.75 million. He said: "There are 378,000 second or subsequent births each year in England. So if cot deaths are random events two cot deaths will occur in the same family somewhere in England once every seven years."

But cot deaths were not random events since there have been several studies of recurrence. The studies put recurrence rates of cot death at about five times the general rate of cot death, "implying recurrence somewhere in England about once every year and a half. Two studies have shown higher rates. He said: "The fact that studies of recurrence rates have been done means that this event is not vanishingly rare."

He says that in the Clark case the prosecution used the figure of one in 73 million rather than one in 2.75 million because of the family's affluence. But he argues that individuals do not necessarily have the attributes of their social group. He said: "The basic principles [of probability theory] are not difficult to understand and judges could be trained to recognise and rule out the kind of misunderstanding that arose in this case. Never again must mathematical error be allowed to conflict with mathematical fact as if each were a legitimate expert view."

Dr Watkins says medical evidence is trusted in court, a situation that must be retained. He said: "It is possible to be an extremely good doctor without being numerate and not every eminent physician is best placed to give epidemiological evidence. When errors occur we expect them to be admitted, learned from and corrected. Expert witnesses can hold a substantial part of defendants' lives in their hands. Defendants deserve the same protection as patients."

Copyright of Telegraph Group Limited 1999.

If judges' analysis of cases matched the scientific clarity of this one by Dr Watkins, there is good reason to believe there would be far fewer miscarriages of justice.

Appendix 3B - The First Appeal for Mrs Sally Clark.

I recommend you read the whole transcript of the judges findings (available in Extra Appendices, on request) to appreciate how pervasive their bias is, but I will examine here the points they stated particularly indicated guilt.

(1) The first was Mr Clark's mis-stating the time he got home on the night Harry died. They labelled this "of the greatest significance" (see para 13 of Extra Appendix B). Compare this finding with the judges attitude to the fact that Dr Williams twice changed his opinion on the children's cause of death (see paras 3, 11 & 26). You might say the difference is one of motive. As the article by Dr Theodore Dalrymple (Extra Appendix C) clearly shows there is significant motive for expert witnesses to please their employing counsel. What is more Mr Clark would only have a misleading motive if he thought the defendant were guilty, whereas Dr Williams has such a motive in ALL circumstances. That significance is one that scientific analysis measures but the undisciplined human mind so readily overlooks.

(2) Suspicion of foul play is considered because of an old fracture of the second rib (Para 8 of Extra Appendix B), yet at the same time suspicion of foul play is considered because both boys were "healthy" when they died (Paras 2, 12, 34 & 43). This type of double standard, where it is impossible for the defendant to get credit whether the children are sick or well, also highlights the "intellectual dishonesty" of the appeal judges as well as the quality of the original trial judgment.

(3) This is the transcript of Para 255 of the appeal judges' findings.

255. Taken separately there was a very strong case on each count. Taken together we conclude that the evidence was overwhelming having regard to the identified similarities:

a) the babies died at the same age;

b) they were both found by the appellant and both, according to one version of the appellant, in a bouncy chair;

c) they were found dead at almost exactly the same time of evening, having been well, having taken a feed successfully and at a time when the appellant admitted tiredness in coping;

d) on each occasion the appellant was alone with the baby when it was found lifeless;

e) on each occasion the appellant's husband was away from home, or about to go away from home;

f) in each case there was evidence of previous abuse: for Christopher an attempted smothering; for Harry an old rib fracture;

g) in each case there was evidence of deliberate injury recently inflicted: for Christopher bruising and a torn frenulum; for Harry hypoxic damage, petechial haemorrhages in the eyelid and fresh bleedings of the spine and swelling of the spinal cord;

h) the rarity of two natural deaths in one family with the first five features above present, and the extraordinary coincidence, if both deaths were natural, of finding evidence of old and recent abuse."

This para 255 is so full of statistical howlers about coincidence that it discredits the whole idea of allowing judges to be recruited from the Humanities with no concept of elementary Science. Any scientist making these statements would be laughed out of court - well, out of the lab.

Item (a). One child was 11 weeks old and the other 8 weeks when they died. Alright that's near enough the same age. The obvious question to expose the fallacy, is what difference in age is necessary to support innocence rather than guilt and just why? Indeed, it isn't clear to me why the similar age doesn't suggest more a common problem in the children's physical development rather than a common mental problem of the mother. One causal interpretation is ignored, whereas the other, more conjectural one, is readily interpreted as guilt.

Item (b). Well I'm a typical family man and if anything had happened to my children when young, it is statistically nigh certain my wife would have been first on the scene. Not satisfied with conjuring up guilt from this, they largely repeat the same point as Item (d).

Item (c). Well the appellant feeling tired I suppose has some negative connotations, but does feeling tired really add great weight to the idea that she killed her babies. It equates to saying every mother is a potential killer on that score. The similarity in time of course has little statistical significance since it has only been considered after the event. Babies are often fed at regular times, yet the judges want to add a sinister connotation to this in addition to the time being the same. Such multiple counting of the same point is a common indication of trying to make the evidence fit a pre-conceived conclusion.

Item (d). This old chestnut is always raising its ugly head. Aren't there enormous amounts of time when mothers are alone with their babies? If something happens when they are not, it can help in their defence, but the number of babies that any mother would need to have die, without anyone else being present when they find it so, for it to be statistically unexpected is a lot more than two.

Item (e). This statement is only marginally different from saying one time he was home and the other time he wasn't. Now you don't have to be a scientist to appreciate the blatant dishonesty in suggesting that confirms anything at all.

Item (f). Dr Williams first found Christopher had died of a respiratory tract infection and only changed his opinion to smothering a year later, after Harry had also died. Dr Williams then identified an old rib fracture in Harry that should have caused him trauma and discomfort. The parents claimed no knowledge of this and no one else, health workers included, had any knowledge of such distress either. The judges saw only one conclusion, the parents were hiding something even though Professor Berry said that the fracture of the rib had not been confirmed (Para 76). The judges even make up their own time for this fracture (compare Paras 243 & 250 and Dr Williams statement in Para 61). The way circumstantial evidence is selectively viewed like this is at heart of many of the judiciary's mistakes.

Item (g). There is a wealth of technical medical evidence and the most consistent thing about it is that the experts agree on very little. Dr Williams (who changes his mind) and Sir Roy Meadow (who uses misleading statistics) are the only two who are willing to talk in certainties. Are the judges who have made such errors in simpler matters fit to come to conclusions that the medical experts themselves cannot?

Item (h). The jump from unknown cause of death to not natural death is clearly another deliberate act of bias by the judges. The value of the rest of this paragraph needs no further comment.

(See Extra Appendix A or www.jspubs.com/Experts, Sally Clark paragraph headed "Appeal" for more, independent information)

Appendix 4 - Seriousness of Convicting an Innocent Person.

Not only do the examples I've given show how readily opinion and circumstantial evidence is allowed to outrank fact, but they also show a failure to appreciate the greater significance of convicting an innocent person versus failing to convict a guilty one. The much vaunted phrase "beyond reasonable doubt" requires a thorough scientific definition. In the cases examined in this article, it appears to encourage the jurors to jump to a conclusion rather than be absolutely sure of the defendant's guilt. To be consistent with the other tenet of justice that "everyone is considered innocent until proven guilty", it should be "beyond any doubt". Where all the evidence is circumstantial and speculative, a more consistent directive would be "if there is a clear possibility that the accused is innocent, that should be the verdict". From a scientific viewpoint unless ALL the evidence is consistent with guilt, the case is not proved. Consistency is a very fundamental requirement in scientific proof. It would be insincere to argue that this would allow the guilty to escape justice, since, as shown in various places through this article, the present system already includes several cases of bias and every one acts AGAINST "good" people.

Appendix 5 - Expert Witnesses.

Consistent with the combative nature of the courtroom process, expert witnesses are chosen by the prosecutor and the defendant and deliberately selected to support the viewpoint required. The scientific effect of this is that they are drawn from the extremes of their expertise and are inclined not to be representative of the main body of such experts. In mathematical statistics, it is the tails of a distribution that are the first to be discounted and it would be false to think a useful interpolated value could be obtained from the two extremes. For those not comfortable with scientific terminology, it means that, like in newspaper reporting, an over exposure is given to more extremely opinionated people, or in streetwise banter, cranks! Selectively chosen expert witnesses seem to have been prominent in almost all the cases I've referred to. Scientifically the lack of independence in their selection requires the factual value of their testimony as experts to be much downvalued. I think there is probably good evidence to show that it takes a scientist to detect rogue science and judges need to expand this aspect of their knowledge to advise the jury capably about the weight to be applied to such evidence.

Sir Roy Meadow who gave the faulty statistical evidence at Sally Clark's trial also gave evidence against Mrs Cannings and more recently (May 2003) against a Mrs Patel. This habit of just a few members of a profession regularly acting as expert witnesses is another indication that such evidence is not truly representative and not broadly based. Indeed there seems to be a dangerously close relationship between prosecutors and their regular expert witnesses and in Sir Roy Meadow's case this close relationship extended to his lecturing judges on his opinions. The danger is particularly blatant in Sir Roy's case because he likes to use the catch phrases "One sudden infant death is a tragedy, two is suspicious and three is murder unless proved otherwise." and "There is no evidence that cot deaths run in families, but there is plenty of evidence that child abuse does". That immediately identifies him as a pseudo scientist and as we saw in Appendix 3A, there is some evidence that cot deaths do run in families.

I recall an expert - a surveyor I believe - once claiming that the system was faulty because unless you gave the evidence required you were never called as an expert witness again. The criticism of the system was quashed by a judge in a condescending manner claiming he found the system worked satisfactorily and offering no evidence to substantiate that opinion. For a profession that proclaims the importance of proof, it is surprisingly reticent to debate, let alone prove, the merits of its own practices.

A very recent article, "Expert witnesses aren't what they seem - and I should know" by Dr Theodore Dalrymple (Filed: 02/02/2003) Daily Telegraph, contains more information on the subject. It is reproduced as Extra Appendix C.

Appendix 6 - Opinion and Fact Not Clearly Distinguished.

Firstly, here are a couple of trivial illustrations to show how pervasive the problem is.

A recent example of a judge being swayed more by opinion than scientific reason was the case of two people imprisoned for disobeying a court order not to cut down a hedge (reported in various UK newspapers 3/8/2002). A neighbour had obtained the temporary order on the basis that he thought some part of it might have been on his land. Assuming the law's purpose is to protect citizens from wrongdoing and to be their servant not their master, the coolheaded reasoned approach would have been to wait and see if any part of the hedge was on the neighbour's land. Then a factual rather than opinionated decision of wrongdoing could be made.

On 28th February 2001, there was a traffic accident when a car driver named Gary Hart lost control of his car and trailer and the final consequence was several people killed in a train crash. The main evidence against him at his trial in December 2001 was that he hadn't had much sleep the night before, which he had acknowledged right from the start and never tried to conceal. The judge castigated the driver for not admitting he fell asleep, without any qualification that this must be just an opinion, and without acknowledging that what followed showed that the driver must have recovered alertness very quickly. He was found guilty by a majority verdict and I am not questioning the verdict, but the judge's immodest behaviour. The judge then castigated him again because a stress counsellor reported the defendant was shocked and angry at the verdict. He then sentenced him to five years in jail. Now every driver could say that there but for the grace of God go I. There was no suggestion of evil intent, but the judge then insisted the driver was solely responsible for the full catastrophe, despite the string of chance happenings that led to it. There is a strong indication that with a less opinionated judge, the defendant could have received a more minimal penalty, especially with the split verdict. Such chance outcome is because judges are not required to be scientific in their analysis. Now if the law says intent has little bearing on the matter and you are responsible for the full outcome of your mistakes, then to be consistent, there is a case against the professional road designer who failed to provide a barrier at this clearly dangerous spot (a barrier started further along) and against the railway safety officials who failed to do something similar at the railway tracks and even against the railway designers who created closely aligned rail tracks for opposing trains to rush past each other at incredible speeds, which is what finally caused the fatalities some considerable way from the original incident. A more realistic measure of the driver's guilt to society would be his driving record, not a one-off chance outcome. This suggests the justice system has difficulty differentiating between malevolent acts and adjudged carelessness, the same way it does between fact and opinion.

Recently there have been quite a few lawyers stating that the double-jeopardy rule has served us well for over 800 years but they don't actually explain what evidence supports that statement. This too indicates a distinct lack of training in scientific thinking within the legal profession - since the last thing a scientist would do is portray an immeasurable opinion as a fact.

There is no surety that anyone found guilty on indefinite evidence is in fact guilty. The use of DNA testing has been a recent source for highlighting that past "opinion-based" judgments and poor legal methodology leave something to be desired. That is probably a classic British understatement. The more prominent reports of this I've seen to date seem to be American and that is probably due to their greater openness in these matters (for example, Los Angeles Times 18SEP2002 "Illinois ..." and Prof. Barry Scheck's testimony to the USA Senate, www.criminaljustice.org/public.nsf/testimony/2000jun13?OpenDocument). Collecting together information on this, showing how many pre-DNA cases have retained suitable DNA material, how many have had it tested and what proportion of them contradict the original verdict, could be a good preliminary indication of how often verdicts are wrong. Terry Gilbert, an American defense attorney, has estimated that 5% of all convicted felons in the United States are innocent (that is 100,000 people). Since many court cases would involve clearcut factual evidence, the error rate for cases involving only circumstantial evidence must be much higher. What is more, a higher standard of "proof" is required in criminal law than civil law, so perhaps we shouldn't be surprised at the nonsense outcome in the Equitable Life Office case.

Even admission of guilt is not definite evidence. There is a particularly sinister practice at the prison stage of pressuring prisoners to admit guilt to get early release. It could be seen as a desperate attempt to add credibility to the veracity of the total process. A Scotsman, Mr Kenny Richey, on death row in America even came within an hour of execution with the pressure to admit guilt and live ("Best Chance for death row Scot", BBC News Online 10/5/2003, available in Extra Appendices). Lauding a system that behaves like this is condoning the evil. It isn't a matter of pragmatism over theory, it is a matter of honesty over deception.

Perhaps the most obvious example of opinion being valued over fact is the appeal process as shown in the first appeal by Sally Clark. Without knowing the jurors' deliberations, merely a guilty verdict, the three judges had to decide whether the original jury would have had a different opinion if Sir Roy Meadow hadn't given the faulty prosecution evidence. Doesn't such crystal-ball gazing discredit the present appeal procedure? Indeed there is much about judicial methodology that suggests its DNA would show closer kinship to astrology than to science. Doesn't it help show why it is so difficult to overturn an initial faulty verdict? I haven't found one comment on that particular appeal decision that is complimentary, such was the quality of the judges' reasoning (see Appendix 3B).

In the trial of Mrs Cannings the defence had elicited specific testimony from BOTH defence and prosecution expert witnesses that Mrs Cannings had no behavioural disorder, yet Mrs Justice Hallett in her sentencing statement said "But I have no doubt that for a woman like you to have committed these terrible acts of suffocating your own babies there must have been something seriously wrong with you.". Does the judge believe her personal opinion should override the trial evidence of medical experts? If not, there is of course only one logical consequence of her statement, Mrs Cannings DIDN'T commit these terrible acts. Science can be beautifully simple and clear. Judges really need to embrace it.

The most obvious drawback to an opinion is that it is derived from the sum total of what we know. When you realise that what each of us doesn't know is infinite in comparison, it makes a human opinion a very poor second to an unambiguous fact. It is an awareness of what we don't know that measures wisdom.

Appendix 7 - Unscientific Legal Procedures.

One legal procedure that contradicts scientific rules is jury selection because of the ability of prosecutors and defendants to interfere with the selection. No scientist would be allowed to help get the result they want by tampering with the dice. While it is fair to the participants from a combatant point of view, since both sides have the same opportunities, it scientifically decreases the veracity of the finding. Indeed the whole approach of jury deliberation is a scientific flaw. If twelve people are selected to give a balanced independent view of the evidence, that decision should be made by them immediately after the judge's summing up and independently of knowing what their colleagues are thinking. Instead of that the judicial process deliberately sends them away to discuss it amongst themselves and to agree on a consensus verdict. Highly opinionated and obstinate people aren't going to change their own minds, so the pressure is on more reasonable folk to give in to them. It is the only way to be allowed to go home. This essentially nullifies the benefit of the independent opinions and the effect is to make the value of the outcome scientifically unintelligible. In scientific statistical terms the tails of the distribution of opinion that should be the first to be ignored actually dominate the process; that translates to "the tail wagging the dog", the very same error that happens with expert witnesses. In Mrs Cannings trial the verdict was unanimous and the jurors' deliberations lasted over nine hours. Now if all twelve thought she was guilty when they entered the jury room, what were they doing after the first ten minutes?

A dabble with a little mathematics helps to discipline human thinking. Where a case depends solely on circumstantial evidence, if the evidence is truly evenly balanced, the opinion amounts to just a guess. Each person's chance of being right is then 50% and the striking thing is that a jury of twelve would fail to give a unanimous verdict 99.95% of the time. It only reduces a little, to 96%, when the judge accepts a 10-2 or better verdict. If there is some room for considered assessment and we say a juror's opinion is right 2 times out of 3, a hung jury would still occur 99% of the time. Even if a juror's opinion were right 4 times out of 5, the jury would still be hung 93%, but if a 10-2 verdict is accepted, it drops to 44%. While hung juries do occur, they happen far more rarely than this, so either each person's independent opinion is right nearly all the time (which would make for a very agreeable world, but doesn't match with empirical evidence) or many jurors must be changing their vote to achieve the required collective agreement. If in practice agreed verdicts are achieved in 90% of the cases involving only circumstantial evidence, if each juror's opinion is right 4 times out of 5, on average four people must change their vote for unanimity and two for a 10-2 or better verdict. If each juror's opinion is only right 2 times out of 3, all five would have to change their vote to achieve unanimity. Clearly this pressure to reach an agreed verdict greatly reduces its authenticity.

Information in a case is also deliberately tampered with by such court rules as previous convictions must be kept from a jury. While hiding past convictions has the scientific quality of preserving independence, tinkering with basic rules only creates other anomalies and the distinct bias this creates against a "good" person is very clearly demonstrated in Mrs Cannings case. My legal knowledge is very limited, but I believe there are several other well-intentioned rules about excluding evidence and my guess is they too wouldn't stand up to scientific scrutiny.

Another such tampering by "human opinion" with scientific fact, arises with the double jeopardy rule, referred to earlier. While it helps possibly guilty people, it does nothing for innocent ones. It might even act against innocent ones in forcing prosecution cases to be extra belligerent since they only have one shot. An even clearer bias against good people resulting from the "double jeopardy" rule is demonstrated by the fact that (1) where further prosecution evidence is discovered after someone is acquitted, a re-trial isn't allowed at all, which is to the benefit of a "crook", whereas (2) as illustrated by Mrs Sally Clark's first appeal, if further defence evidence is discovered after someone is found guilty, a re-trial isn't readily allowed, which disadvantages a "good" person.

The secrecy of a jury's deliberations is also a questionable part of legal methodology if the aim is to get as honest a result as possible. It certainly limits the ability of an appeal court to make an informed decision and analysts to identify flaws in the process. The defendant has far more at stake than a juror, so protecting a juror's position at the expense of the defendant's is not a soundly weighted argument. A vote by the jurors without consultation amongst themselves would give a scientifically sounder result and if that is secret, as has been deemed best in democratic elections, it could lessen concern about interference by unscrupulous defendants.

The miscarriages of justice I've referred to would be less likely to occur if all the cards are put on the table, but the jury directive is tightened to "beyond any doubt". It is not a very intelligent society that reveres a system that so readily convicts innocent people and at the same time assists perennial crooks to escape justice.

Appendix 8 - Sixteen Years In Jail For A Crime He Did Not Commit.

The evidence against Michael Shirley was that he owned a brand of shoes like the killers, he was in Portsmouth when the offence took place there and he had the same blood type as the killer and so did ten million other British men. Incredibly the judicial system allows that to be sufficient information to find someone guilty under the guise of circumstantial evidence. Added to that affront, he served sixteen years in jail, one more than he would have if he had been willing to admit his guilt, despite the DNA evidence that established his innocence being known three years before his release. Using the same slack way courts interpret circumstantial evidence, that could well be interpreted as a deliberate attempt to break his spirit and cover up the system's failings. The website www.innocence.org.uk identifies many other such results from a system devoid of science and tainted with medieval practices.

Appendix 9 - Vested Interests and Some Simple Solutions.

My scientific background of course makes me uncomfortable with the concepts of conspiracy theory. It is the realm of manipulation and innuendo, rather than objective argument. However, such methods, like painting innocuous acts with a sinister brush, are regularly used by prosecutors and regularly permitted by the judiciary, so none of those should object. Why did the judge at the sentencing of Mrs Cannings question the politically sensitive subject of mandatory sentencing, but fail to question the equally obvious and more fundamental subject of whether the verdict of her court was sound? It could be a case of St.Luke 6.41, but it would be more serious if Mrs Cannings' anguish is increased because she is a pawn in a battle between government and the judiciary. There must be some explanation why Mrs Cannings still languishes in jail despite the comments from all those closely involved - ranging from not believing the verdict to the judges own statement that she doesn't need to be in jail. Sentencing was in April 2002 and her trauma clearly started long before then. With an appeal only recently scheduled, the process, including the Criminal Cases Review Commission (CCRC) process, is clearly failing the victims of the judiciary's methods.

The CCRC was set up in 1997 to review cases where people claimed to have been wrongly convicted. Several people have had retrials as a result of this and a number of those were exonerated (see www.innocence.org.uk). That the Sally Clark and Angela Cannings convictions have arisen since 1997 shows that it hasn't changed the fundamental, underlying problems of the judicial system. The CCRC looks to be only more of the same and a more truly progressive move would be the appointment of an ombudsman who is totally independent of legal paraphernalia and who can order a QUICK retrial if the jury's verdict is not compatible with ALL the evidence. Scientists understand exactly why it is impossible to pull oneself up by one's bootlaces and to get momentum for change, the judiciary needs the external pressure of accountability from outside its own sphere of control.

Potential mistakes occurring wouldn't be as bad if there were a quick mechanism for redressing, but it is a slow drawn out procedure getting to trial and an even slower process to get a re-trial. The drain on the defendant's resources - psychological and financial - must be quite enormous. A quicker trial with more readily available re-trials would be much more socially considerate, especially as the present long drawn out affair doesn't provide better quality results. Indeed, short, sharp trials would encourage everyone to keep their eye on the ball and avoid the soft evidence sullying the hard.

Appendix 10 - Synopsis of the Errors in the Justice System Identified Above.

The unscientific procedures identified include

1. Weighting of evidence, where the strength of fully factual information is overridden by evidence that has an element of doubt.

2. Bias, and in every instance against "good people",

2a. The "double jeopardy" rule, since if further evidence becomes available, a "bad" person cannot suffer a re-trial but as demonstrated in Sally Clark's first appeal hearing, a "good" person cannot readily benefit by getting a re-trial.

2b. The limiting of evidence of past convictions, as specifically identified in the case of Mrs Angela Cannings.

2c. By choosing expert witnesses for their given viewpoint and ignoring the significance of that.

2d. Possibly in the use of circumstantial evidence, particularly the inventive kind, more as a prosecution device than a defence device.

2e. Plea bargaining.

3. Interference in jury selection.

4. Interference by limiting evidence and the inconsistency of rules that hide certain information.

5. Interference of the "double jeopardy" rule.

6. The very process of jurors being forced to modify their own position and arrive at a common decision which has far less scientific value than twelve independent conclusions.

7. The use of secrecy in those jury deliberations. The defendant has far more at stake than a juror, so protecting a juror's position at the expense of the defendant's is not a soundly weighted argument.

8. Ignoring how important CONSISTENCY of evidence is before you can say something is proven.

9. Failure to recognise the EXTRA seriousness of convicting an innocent person.

10. A medical expert lecturing judges on how to interpret medical evidence and then giving evidence before judges in particular cases.

11. The portrayal of opinion as fact. It is this tendency of many lawyers that reveals the lack of training the profession has in scientific reasoning. Another particularly obvious example is the nonsense of an appeal court assessing whether a fault in prosecution evidence would have changed a jury's verdict when they have no information whatsoever on the jurors' deliberations.

12. The pressure for a convicted person to admit their guilt in exchange for leniency such as early parole or even commuting a death sentence. This surely shows that medieval practices still pervade the system.