United Kingdom House of Lords Decisions

You are here: BAILII >> Databases >> United Kingdom House of Lords Decisions >> H & Ors (minors), Re [1995] UKHL 16 (05 April 2000)
URL: http://www.bailii.org/uk/cases/UKHL/1995/16.html
Cite as: [1996] AC 563, [1996] 1 FCR 509, [2006] AC 563, [1996] 2 WLR 8, [1996] Fam Law 74, [1996] 1 FLR 80, [1995] UKHL 16, [1996] 1 All ER 1

[New search] [Buy ICLR report: [1996] 2 WLR 8] [Buy ICLR report: [1996] AC 563] [Help]

[1995] UKHL 16

HOUSE OF LORDS

OPINIONS OF THE LORDS OF APPEAL FOR JUDGMENT

IN THE CAUSE

IN RE H AND OTHERS (MINORS) (A. P.) (RESPONDENTS)

ON THURSDAY 14 DECEMBER 1995

Lord Goff of Chieveley
Lord Browne-Wilkinson
Lord Mustill
Lord Lloyd of Berwick
Lord Nicholls of Birkenhead

LORD GOFF OF CHIEVELEY

My Lords,

I have had the advantage of reading in draft the speech prepared by my noble and learned friend, Lord Nicholls of Birkenhead. For the reasons which he gives I too would dismiss this appeal.

LORD BROWNE-WILKINSON

My Lords.

I have the misfortune to disagree with the view reached by the majority of your Lordships. Although the area of disagreement is small, it is crucial both to the outcome of this appeal and to the extent to which children at risk can be protected by the courts.

I agree with my noble and learned friend Lord Nicholls of Birkenhead that the requirement in section 31(2) of the Children Act, 1989, that the court must be satisfied that the child "is likely to suffer significant harm" does not require the court to find that such harm is more likely than not: it is enough if the occurrence of such harm is a real possibility. I further agree with him that the burden of proving any relevant fact is on the applicant and that the standard of proof is the ordinary civil standard, i.e. balance of probabilities. The point on which I differ is how those principles fall to be applied by a judge faced with the decision whether he is "satisfied" that the child is likely to suffer significant harm. Even on this point, I agree that the judge can only act on evidence and on facts which, so far as relevant, have been proved. He has to be satisfied by the evidence before him that there is a real possibility of serious harm to the child.

Where I part company is in thinking that the facts relevant to an assessment of risk ("is likely to suffer . . . harm") are not the same as the facts relevant to a decision that harm is in fact being suffered. In order to be satisfied that an event has occurred or is occurring the evidence has to show on balance of probabilities that such event did occur or is occurring. But in order to be satisfied that there is a risk of such an occurrence, the ambit of the relevant facts is in my view wider. The combined effect of a number of factors which suggest that a state of affairs, though not proved to exist, may well exist is the normal basis for the assessment of future risk. To be satisfied of the existence of a risk does not require proof of the occurrence of past historical events but proof of facts which are relevant to the making of a prognosis.

Let me give an example, albeit a dated one. Say that in 1940 those responsible for giving air-raid warnings had received five unconfirmed sightings of approaching aircraft which might be enemy bombers. They could not, on balance of probabilities, have reached a conclusion that any one of those sightings was of an enemy aircraft: nor could they logically have put together five non-proven sightings so as to be satisfied that enemy aircraft were in fact approaching. But their task was not simply to decide whether enemy aircraft were approaching but whether there was a risk of an air-raid. The facts relevant to the assessment of such risk were the reports that unconfirmed sightings had been made, not the truth of such reports. They could well, on the basis of those unconfirmed reports, have been satisfied that there was a real possibility of an air-raid and given warning accordingly.

So in the present case, the major issue was whether Dl had been sexually abused (the macro fact). In the course of the hearing before the judge a number of other facts (the micro facts) were established to the judge's satisfaction by the evidence. The judge in his careful judgment summarised these micro-facts: that Dl had been consistent in her story from the time of her first complaint; that her statement was full and detailed showing "a classic unfolding revelation of progressively worse abuse"; that there were opportunities for such abuse by Mr. R and that he had been lying in denying that he had ever been alone either with Dl or with any of the other children; that D2 had made statements which indicated that she had witnessed "inappropriate" behaviour between Mr. R and Dl; that the mother (contrary to her evidence) also suspected that something had been going on between Mr. R and Dl and had sought to dissuade D2 from saying anything to the social workers. The judge also found a number of micro facts pointing the other way. Having summarized all these micro facts pointing each way, he reached his conclusion on the macro fact: "I cannot be sure to the requisite high standard of proof that [Dl's] allegations are true". But he also made further findings (which he thought to be irrelevant in law) on the basis of the micro facts:

"This is far from saying that I am satisfied the child's complaints are untrue. I do not brush them aside as the jury seem to have done. I am, at the least, more than a little suspicious that [Mr. R] has abused her as she says. If it were relevant. I would be prepared to hold that there is a real possibility that her statement and her evidence are true, nor has [Mr. R] by his evidence and demeanour, not only throughout the hearing but the whole of this matter, done anything to dispel those suspicions ..."

That conclusion that there was a real possibility that the evidence of Dl was true was a finding based on evidence and the micro facts that he had found. It was not a mere suspicion as to the risk that Mr. R was an abuser: it was a finding of risk based on facts.

My Lords, I am anxious that the decision of the House in this case may establish the law in an unworkable form to the detriment of many children at risk. Child abuse, particularly sex abuse, is notoriously difficult to prove in a court of law. The relevant facts are extremely sensitive and emotive. They are often known only to the child and to the alleged abuser. If legal proof of actual abuse is a prerequisite to a finding that a child is at risk of abuse, the court will be powerless to intervene to protect children in relation to whom there are the gravest suspicions of actual abuse but the necessary evidence legally to prove such abuse is lacking. Take the present case. Say that the proceedings had related to Dl, the complainant, herself. After a long hearing a judge has reached the conclusion on evidence that there is a "real possibility" that her evidence is true, i.e. that she has in fact been gravely abused. Can Parliament really have intended that neither the court nor anyone else should have jurisdiction to intervene so as to protect Dl from any abuse which she may well have been enduring? I venture to think not.

My Lords, for those reasons and those given by my noble and learned friend Lord Lloyd of Berwick I would allow the appeal.

LORD MUSTILL

My Lords,

I have had the advantage of reading in draft the speech prepared by my noble and learned friend, Lord Nicholls of Birkenhead. For the reasons which he gives I would dismiss this appeal.

LORD LLOYD OF BERWICK

My Lords,

In this case we are concerned with two sisters and two half-sisters. In September 1993 the eldest sister, C, then aged 15, gave a detailed written statement to the police in which she alleged that she had been subject to sexual abuse by her step-father since the age of 7 or 8, culminating in four acts of rape. The step-father (whom I shall refer to as "the father") was arrested and charged. C gave evidence at his trial. In October 1994 he was acquitted on all six counts. The jury took only 14 minutes to reach their verdict.

Meanwhile in February 1994 Nottinghamshire County Council applied for care orders in respect of the three younger sisters. It was decided not to apply for a care order in respect of C, since she had been living with foster-parents since November 1993, following a short period under police protection, and the placement appeared to be satisfactory.

The hearing took place before His Honour Judge Davidson Q.C. in November 1994. It lasted seven days. The question he had to decide was whether the threshold criteria set out in section 31(2) of the Children Act 1989 were satisfied. That subsection provides:

"A court may only make a care order or supervision order if it is satisfied (a) that the child concerned is suffering, or is likely to suffer, significant harm; and (b) that the harm, or likelihood of harm, is attributable to - (i) the care given to the child, or likely to be given to him if the order were not made, not being what it would be reasonable to expect a parent to give to him; or (ii) the child's being beyond parental control."

Since it was not suggested that the three younger sisters had suffered or were suffering any harm, the question was whether, on the evidence before the court, the judge was satisfied that they were likely to suffer significant harm in the future.

The judge heard from the mother (who came from prison to give her evidence) as well as the father. As to the mother, he found that he could not rely on her evidence, since she had been untruthful in at least three respects in the witness box. As to the father, he said that he had seldom been less impressed by a witness. But, as he went on to point out, the fact that the mother and the father told material lies in the witness box did not mean that C was necessarily telling the truth.

As to C herself, the judge set out carefully and comprehensively the factors which told for or against her evidence. It was clearly a most anxious case. But in the event he found that he could not be sure to the "requisite high standard of proof" that C's allegations were true. Accordingly he held that he had no jurisdiction to make a care order. The threshold test was not met. But the judge did not leave it there. He went on to say that he was far from satisfied that C's complaint was untrue.

"I am at the least more than a little suspicious that the [father] has abused her as she says. If it were relevant, I would be prepared to hold that there is a real possibility that her statement and her evidence are true, nor has the [father] by his evidence and demeanour, not only throughout the hearing but the whole of this matter, done anything to dispel those suspicions ..."

The Court of Appeal dismissed the local authority's appeal. Kennedy L.J. dissenting. There is now an appeal to your Lordships' House. The parties have helpfully identified five points for decision. The first three are of general importance. I will take them in the reverse order, since it is only the third which gives rise to any difficulty.

(1) In order to establish that a child is "likely" to suffer significant harm in the future, is it necessary to establish the likelihood of such harm on a balance of probabilities, i.e. to establish that it is more likely than not that the child will suffer such harm in the future, or is it enough that there is a "substantial" as opposed to a "speculative" risk?

The word "likely" in ordinary language may mean probable, in the sense of more likely than not; or it may include what might well happen. Thus in Davies v. Taylor [1974] A.C. 207, one of the questions was whether the judge had applied the correct test in a case under the Fatal Accidents Acts 1846-1959. In the course of his judgment he had used the word "likely" to indicate the test which he was applying. Lord Cross of Chelsea said, at p. 222:

"The word 'likely' which occurs in the last two of the three passages from the judgment which I have quoted above, may be used in different senses. Sometimes it may be used to mean 'more likely than not' at other times to mean 'quite likely' or 'not improbably' though less likely than not."

Similarly, in Dunning v. United Liverpool Hospitals' Board of Governors [1973] 1 W.L.R. 586, the question was whether a claim in respect of personal injuries was "likely to be made" for the purposes of section 31 of the Administration of Justice Act 1970. Lord Denning M.R. said, at p. 590 that "likely to be made" should be construed as meaning "may" or "may well be made." James L.J. said that he would construe "likely" as meaning a "reasonable prospect."

Coming to section 31(2)(a) of the Act itself, the Court of Appeal in Newham London Borough Council v. A.G. [1993] 1 F.L.R. 281, rejected an argument that "likely to suffer significant harm" was to be equated with "on the balance of probabilities". In In re A (A Minor) (Care Proceedings) [1993] 1 F.L.R. 824, it was again argued that "likely" meant more probable than not. Thorpe J. held that the argument was not open to the appellants in the light of Newham London Borough Council v. A.G., a decision which he regarded as of great importance.

In the present case the Court of Appeal have held unanimously, in line with the Newham case, that the threshold test is satisfied if, in the court's view, there is a real or substantial risk of significant harm in the future.

Mr. Levy, for the respondents, submitted that the Newham case was wrongly decided. He pointed out that both halves of section 31(2)(a) are governed by the words "if [the court] is satisfied." Since, as was common ground, the court has to be satisfied on a balance of probabilities under the first half of the clause, i.e. that the child is suffering harm, it must follow (so he argued) that the court must be satisfied on a balance of probabilities that the child will suffer harm under the second half of the clause. Therefore "likely" in the second half of the clause must mean more likely than not. But this is a non sequitur. It confuses what has to be proved under the second half of the clause, i.e. the likelihood of significant harm, with the standard of proof required under the first half of the clause. There is no necessary connection between the two.

As for the word "satisfied" on which Mr. Levy placed reliance, it does not throw any light on the degree of risk contemplated by the second half of the clause. It is a word with a range of meanings covering the criminal burden of proof ("satisfied so as to be sure") through the civil burden of proof ("satisfied on a balance of probabilities") to a synonym for "conclude" or "determine." In Blyth v. Blyth [1966] A.C. 643, the House had to consider section 4(2) of the Matrimonial Causes Act 1950. That subsection provided:

"If the court is satisfied on the evidence that - (a) the case for the petition has been proved; and (b) where the ground of the petition is adultery, the petitioner has not in any manner . . . condoned, the adultery ... the court shall pronounce a decree of divorce ..."

Lord Pearson said, at p. 676:

"The phrase 'is satisfied' means, in my view, simply 'makes up its mind'; the court on the evidence comes to a conclusion which, in conjunction with other conclusions, will lead to the judicial decision."

Lord Pearce, at p. 672, regarded "satisfied" as a neutral word allowing of proof to a different degree in relation to the two halves of the subsection, i.e. proof of adultery and proof that the petitioner has not condoned the adultery. So here, the word "satisfied" in section 31(2)(a) is neutral. It means that the court must make up its mind. It thus bears the same meaning in relation to both halves of the clause, but. as I have said, throws no light on the meaning of "likely."

I therefore conclude that the decision of the Court of Appeal as to the meaning of "likely to suffer significant harm" was correct.

In so far as it is either relevant or necessary in proceedings under the Act to prove an allegation of sexual abuse, is the standard of proof required (i) a standard higher than the ordinary civil standard though falling short of the criminal standard of proof, (ii) the balance of probabilities, but so that the more serious the allegation the more convincing is the evidence needed to tip the balance in respect of it. or (iii) the simple balance of probabilities.

All three counsel were agreed that the correct answer to the above question should be (ii). As a result there was no argument as to the theoretical difference between the three possible answers, or. perhaps more important, the practical consequences. Nor was there any citation of earlier authority on the point, of which there is a great deal.

In the course of his judgment at first instance Judge Davidson referred to and followed the headnote in In re W (Minors) (Sexual Abuse: Standard of Proof) [1994] 1 F.L.R. 419, 420 which reads:

"Charges of sexual abuse in civil proceedings must be proved to a standard beyond a mere balance of probability, but not necessarily a standard as demanding as the criminal standard."

In other words, the judge favoured solution (i). In the Court of Appeal [1995] 1 F.L.R. 643 Millett L.J. said, at p. 659E, that in all civil cases there is only one standard of proof, namely, proof on the balance of probabilities, and that, contempt of court apart, it is never necessary to prove facts to a standard beyond the balance of probabilities. Since we have heard no argument on the point, I am not for my part prepared to endorse so wide a proposition. It will have to await a future occasion when authorities in other branches of the civil law, including decisions of your Lordships' House, can be considered. So I propose to confine what I am about to say to the standard of proof under section 31(2) of the Act.

In my view the standard of proof under that subsection ought to be the simple balance of probability however serious the allegations involved. I have reached that view for a number of reasons, but mainly because section 31(2) provides only the threshold criteria for making a care order. It by no means follows that an order will be made even if the threshold criteria are satisfied. The court must then go on to consider the statutory checklist in section 1(3) of the Act. But if the threshold criteria are not met, the local authority can do nothing, however grave the anticipated injury to the child, or however serious the apprehended consequences. This seems to me to be a strong argument in favour of making the threshold lower rather than higher. It would be a bizarre result if the more serious the anticipated injury, whether physical or sexual, the more difficult it became for the local authority to satisfy the initial burden of proof, and thereby ultimately, if the welfare test is satisfied, secure protection for the child.

Another indirect pointer may be found in section 26 of the Family Law Reform Act 1969. At common law the presumption of legitimacy could only be rebutted by proof beyond reasonable doubt. This was one of the considerations which led the House to its conclusion in Preston-Jones v. Preston-Jones [1951] A.C. 391. By section 26 of the Act of 1969 the presumption can now be rebutted on a simple balance of probabilities. Although in Serio v. Serio [1983] 4 F.L.R. 756, 763 the Court of Appeal held that the standard of proof should be "commensurate with the seriousness of the issue involved" (in other words, that it might require more than a mere balance of probabilities), this seems to read words into the statute which are not there. If the legislature has ordained that the presumption of legitimacy can be rebutted on a simple balance of probabilities, I have no great difficulty in concluding that section 31(2) requires a simple balance of probabilities, and no more, even when there is a serious allegation of sexual abuse.

There remains the question whether anything should be said about the cogency of the evidence needed to "tip the balance". For my part I do not find these words helpful, since they are little more than a statement of the obvious; and there is a danger that the repeated use of the words will harden into a formula, which, like other formulas (especially those based on a metaphor) may lead to misunderstanding. The formula seems to owe its origin to the need to qualify or explain Denning L.J.'s judgment in Bater v. Eater [1951] p. 35 and Hornal v. Neuberger Products Ltd. [1957] 1 Q.B. 247. But once it is accepted that the standard of proof under section 31(2) is the simple balance of probabilities, and that the subsection does not require a degree of probability commensurate with the seriousness of the allegation, then the need for the qualification disappears. Despite the unanimity of counsel's preference for answer (ii) to the second question, I would prefer (iii), and leave the rest to the good sense of judges and magistrates. They will be well aware of, and pay full regard to, the factual context in which they must reach their difficult decisions.

As for the present case, I can have no doubt that the judge applied a higher than ordinary standard of proof. It is what he says in plain terms. Sir Stephen Brown P. said that the judge may nevertheless have applied the right test and drew attention to the reference in his judgment to In re W (Minors) (Sexual Abuse: Standard of Proof) [1994] 1 F.L.R. 419. But In re W is the very case in which the Court of Appeal held that a higher than ordinary standard is required. Millett L.J. also thought that the judge had applied the correct test, despite what the judge said. I fear that in this respect the majority of the Court of Appeal were being over-generous to the judge.

Where the allegation that a child is "likely to suffer significant harm" within the meaning of the second limb of section 31(2)(a) of the Act arises solely out of alleged sexual abuse in the past, is it first necessary to prove to the appropriate standard of proof (see (2) above) that such abuse has in fact occurred?

The third question is the one that gives rise to difficulty. The problem can be stated quite simply. The case has been fought on the basis that the sole cause for concern are the allegations of sexual abuse made by C. It may be that in that respect the case is unusual, and that in many, if not most cases, a local authority applying for a care order will rely on a number of contributing factors. It is only when the local authority relies, as here, on a single incident or series of incidents relating to the same child, that the problem arises in a stark form. If the court finds on the balance of probabilities that the incidents did not occur, how can it go on to hold that by reason of those incidents there is a real or substantial risk of significant harm in the future?

Before giving my answer to that conundrum, it is helpful to look at the background to section 31 of the Act. A feature of the 1970s and 1980s was the sudden and very rapid increase in the number of applications for wardship in the High Court, mainly due to the increased use of wardship by local authorities: see The Government White Paper on The Law on Child Care and Family Services (1987) (Cm. 62) p. 15, para. 59 published in January 1987, and the table set out in Bromley's Family Law, 8th ed., (1992) p. 477. One of the purposes of the Act of 1989, as I understand it, was to abrogate the power of the High Court to place a ward of court in care, (see section 100 of the Act, which repealed section 7 of the Family Law Reform Act 1969 and placed tight restrictions on the High Court's inherent jurisdiction), while at the same time making the benefits of the old wardship jurisdiction more generally available. As Butler-Sloss L.J. said in In re B (Minors) (Termination of Contact: Paramount Consideration) [1993] Fam. 301, at 310, the Act "aims to incorporate the best of the wardship jurisdiction within the statutory framework". The consequence was that Part IV and V of the Act became the only route open to a local authority for protecting children at risk.

A number of cases prior to 1991 (when the Act of 1989 came into force) illustrate the old wardship approach. Thus in H. v. H. (Minor) (Child Abuse: Evidence) [1990] Fam. 86, a case concerning access, Butler-Sloss L.J. said, at p. 101:

"[The judge] may have found individual facts, such as inappropriate knowledge or behaviour, which constitute a high degree of concern about the child without being able to say on the test that they amount to actual abuse. They are, however, relevant to the exercise of the discretion. He may have sufficient evidence of concern about the past care of the child to be satisfied that the child was in a potentially abusing situation without having sufficient evidence to be satisfied as to the extent of the abuse in the past or the identity of the abuser."

Stuart-Smith L.J. said, at p. 121:

"In the type of case with which we are concerned in these appeals there may be insufficient evidence upon which the judge can conclude that the father has sexually abused his children, nevertheless there may be sufficient evidence to show that there is a real chance, possibility or probability that he will do so in the future if granted access."

In In re W (Minors) (Wardship: Evidence) [1990] 1 F.L.R. 203, another wardship case, Butler-Sloss L.J. said, at p. 215:

"It is not necessary to make a finding of sexual abuse against a named person in order for the judge to assess the risks to the child of returned to that environment. He is engaged in a different exercise, that of the assessment of the possibilities for the future."

Neill L.J. said, at p. 228:

"There may also be cases, however, where the court may not be in a position to make a positive finding on the evidence as to what has happened in the past, but may nevertheless come to the conclusion that a child may be at risk for the future."

Although these cases were decided in wardship, and not under the Children Act, they underline a general point. Evidence which is insufficient to establish the truth of an allegation to a required standard of proof, nevertheless remains evidence in the case. It need not be disregarded. The point will be familiar to anyone who has taken part in a criminal trial. It is not uncommon for defence counsel to tell the jury that unless they are sure that a particular witness is telling the truth, they must reject his evidence altogether. But this is quite wrong. The witness's evidence remains evidence in the case. The jury is entitled to take it into account in deciding whether on all the evidence they are sure of the defendant's guilt.

I now return to the second half of section 31(2)(a). It requires the court to be satisfied that the child is likely to suffer significant harm in the future. There is nothing in the second half of the subsection which requires the court to make a finding about anything in the past or present. The finding of future risk must, of course, be based on evidence. It cannot be based on hunch. If there is no evidence to support a finding of risk, the finding will be set aside. But if there is such evidence, then a finding may be made, even though the same evidence is insufficient to support a finding of past fact. In the present case the judge was not persuaded by C.'s evidence that she had been sexually abused. But that does not mean that he rejected her evidence as worthless. On the contrary, he went out of his way to find that she might well have been telling the truth. He was prepared to hold that this was a real possibility.

In those circumstances it would, I think, have been open to him to find, on C.'s evidence, that there was a real possibility of one or more of C.'s sisters suffering significant harm so as to satisfy the threshold test. But the judge never asked himself that question. He adopted what has been called the two-stage approach. Once he had decided not to make a finding of sexual abuse on the balance of probabilities, he never asked himself the vital question posed by the second half of the subsection, whether, on such evidence as there was, he should make a finding of serious risk for the future. I quote from his judgment:

"Bearing in mind all these factors ... I find myself in the position that I cannot be sure to the requisite high standard of proof that C.'s allegations are true. It must follow that the statutory criteria for the making of a care order are not made out."

With great respect this does not follow. The fact that the first half of section 31(2)(a) is not satisfied on the balance of probabilities does not mean that the second half may not be satisfied. The two halves of the subsection are not interlinked, logically or linguistically. They could as well have been contained in separate sub-paragraphs.

Sir Stephen Brown P. [1995] 1 F.L.R. 641, 652, and Milieu L.J., at p. 657, upheld and adopted the judge's two-stage approach. Milieu L.J. said, at p. 658:

"If the likelihood of the child suffering harm in the future depends upon the truth of disputed allegations, the court must investigate the allegations and determine, on the balance of probabilities, whether they are true or false. It is not sufficient that there is a real possibility that the allegations may be true if the probability is that they are not."

The fallacy, if I may respectfully say so, lies in the protasis. The likelihood of future harm does not depend on proof that disputed allegations are true. It depends on the evidence. If the evidence in support of the disputed allegations is such as to give rise to a real or substantial risk of significant harm in the future, then the truth of the disputed allegation need not be proved.

In another passage p. 658 Millett L.J. refers to the two different factual situations covered by section 31(2)(a).

"In the first it is plain that the court must be satisfied, on a balance of probabilities, that the child is suffering significant harm. It is not enough for the court to conclude that there is a real possibility that the child may be suffering significant harm. The same test must be applied to the second factual situation."

I have difficulty with the last sentence for two reasons. In the first place, there is nothing in the subsection which requires the same test to be applied to both halves of the subsection: see Blyth v. Blyth [1966] A.C. 643. The argument accepted by the majority in the court below is very similar, if not identical, to the argument rejected by the majority of the House in that case. Secondly, I am not clear in what sense the future, which is the subject matter of the second half of the subsection, can ever be said to be a "factual situation" subject to proof in the same way as past or present fact: see Davies v. Taylor [1974] A.C. 207, 212, per Lord Reid at p. 212.

I confess that I much prefer the simpler one-stage approach adopted by Kennedy L.J. Although a two-stage approach may come naturally to lawyers, the same cannot necessarily be said for magistrates, social workers, and others who have got to understand and apply section 31. It may not be too difficult when there is an isolated issue of fact on which alone the outcome depends. But this will seldom, if ever, be the case in practice. Facts are always surrounded by other facts. Macro facts, to adopt my noble and learned friend Lord Browne-Wilkinson's vivid terminology, are surrounded by micro facts. In the usual case, there will be a number of interlocking considerations, all of which will give rise to separate issues of fact, and on all of which, if the Court of Appeal be right, the court would have to make separate findings on the balance of probabilities before proceeding to the second stage. Suppose, for example, there are three or four matters for concern which have led the social services to the belief that a child is at risk, on each of which there is credible evidence, supported, it may be, by evidence from a child psychiatrist, but suppose the evidence is insufficient on any of them to justify a finding that the child has been abused. Is the court powerless to proceed to the second stage? This is not what Parliament has said, and I do not think it is what Parliament intended. Parliament has asked a simple question: Is the court satisfied that there is a serious risk of significant harm in the future? This question should be capable of being answered without too much over-analysis.

In an unusual case such as the present, which has been fought on the basis of a single issue of past fact, it will no doubt make sense for the court to start by deciding whether that issue has been proved to its satisfaction, or not. But this is only the beginning. Even if the evidence falls short of proof of the fact in issue, the court must go on to evaluate the evidence on that issue, together with all the other evidence in the case, and ask itself the critical question as to future risk.

In Newham London Borough Council v. A.G. [1993] 1 F.L.R. 281, 289 Sir Stephen Brown P. said at page 289:

"I very much hope that in approaching cases under the Children Act 1989 courts will not be invited to perform in every case a strict legalistic analysis of the statutory meaning of section 31."

The editors of Clarke Hall & Morrison on Children, 10th ed. (Loose-leaf), vol. 1, para. 612, commented: "This strongly suggests that the court should be flexible in interpreting section 31 and then exercise its discretion in the light of sections 1 and 8." I agree.

I have left to the end an argument which Millett L.J. regarded as confirming his approach. Under section 43 of the Act, a court may make an assessment order if it is satisfied that the local authority has "reasonable cause to suspect" that a child is likely to suffer significant harm. Under section 44 the court may make an emergency order if it is satisfied that there is "reasonable cause to believe" that the child is likely to suffer significant harm. Similarly, under section 38 it may make an interim care order if it is satisfied that there are "reasonable grounds for believing" that a child is likely to suffer significant harm. Finally, under section 31(2) the court may make an order if it is satisfied that the child is likely to suffer significant harm.

These sections represent progressive stages on the road to the making of a care order, from "cause to suspect" through "ground for belief" to the substantive finding. Little evidence suffices at the early stages. Much more evidence is required at the later stages. But it will be noticed that at all the stages the court has to be "satisfied" on whatever evidence there is. So the use of the word "satisfied" at the final stage does not, I think, point a contrast with the earlier stages; nor does it show, as Millett L.J. thought, that the likelihood of significant harm has to be proved on a balance of probabilities before a care order can be made. For the reasons which I have attempted to state in answer to the first question "satisfied" is a neutral word which means no more than conclude or determine or decide.

I can summarise my views as follows:

(1) "Likely" in section 31(2)(a) means that there is a serious risk or real possibility that the child will suffer significant harm.

(2) Where it is claimed that the child has suffered or is suffering significant harm the standard of proof is the simple balance of probabilities, no matter how serious the underlying allegation.

(3) Where it is claimed that the child is likely to suffer significant harm, the simple one-stage approach suffices. The question is whether, on all the evidence, the court considers that there is a real possibility of the child's suffering significant harm in the future. If so, the threshold criterion is satisfied. The court does not have to be satisfied on the balance of probabilities that the child has in fact suffered significant harm in the past, whether by sexual abuse or otherwise, even where the allegation of abuse is the foundation of the local authority's case for a care order.

It follows that the judge fell into error in two respects. First he applied a standard of proof in respect of C. 's allegation of sexual abuse which was manifestly too high. Secondly, he never asked himself the right question about significant harm in the future. He was misled by the two-stage approach, as a consequence of which he held that the second and vital question did not arise.

For the reasons which I have given, as well as those given by Kennedy L.J., I would allow this appeal. If I have not quoted at length from Kennedy L.J.'s judgment, it is only because I have read it with admiration, and agree with every word.

I would therefore have remitted the case to the judge for a further hearing if there be continuing cause for concern. But as a majority of your Lordships take a different view, this will not be necessary.

LORD NICHOLLS OF BIRKENHEAD

My Lords,

The subject of this appeal is the care of children. Section 31 of the Children Act 1989 empowers the court to make an order placing a child in the care of a local authority or putting a child under the supervision of a local authority or a probation officer. Section 31(2) provides that a court may only make such an order:

"if it is satisfied - (a) that the child concerned is suffering, or is likely to suffer, significant harm; and (b) that the harm, or likelihood of harm, is attributable to - (i) the care given to the child, or likely to be given him if the order were not made, not being what it would be reasonable to expect a parent to give to him; or (ii) the child's being beyond parental control."

In short, the court must be satisfied of the existence or likelihood of harm attributable either to the care the child is receiving or likely to receive or to the child being beyond parental control. Harm means ill-treatment or impairment of health or development: see section 31(9). This appeal concerns the need for the court to be "satisfied" that the child is suffering significant harm or is "likely" to do so.

The facts are set out in the judgment of Sir Stephen Brown P sitting in the Court of Appeal, reported at [1995] F.L.R. 643. For present purposes I can summarise them shortly. The mother has four children, all girls. Dl and D2 were children of her marriage to Mr. H in 1979. Dl was born in June 1978 and D2 in August 1981. Mr. H and the mother then separated. In 1984 she commenced living with Mr. R and they had two children: D3, born in March 1985, and D4, born in April 1992.

In September 1993, when she was 15, Dl made a statement to the police. She said she had been sexually abused by Mr. R ever since she was 7 or 8 years old. She was then accommodated with foster-parents, and Mr. R was charged with having raped her. In February 1994 the local authority applied for care orders in respect of the three younger girls. Interim care orders were made, followed by interim supervision orders.

In October 1994 Mr. R was tried on an indictment containing four counts of rape of Dl. Dl was the principal witness for the Crown. The jury acquitted Mr. R on all counts after a very short retirement. Despite this the local authority proceeded with the applications for care orders in respect of D2, D3 and D4. These girls were then aged 13, 8 and 2 years. The local authority's case, and this is an important feature of these proceedings, was based solely on the alleged sexual abuse of Dl by Mr. R. Relying on the different standard of proof applicable in civil and criminal matters, the local authority asked the Judge still to find that Mr. R had sexually abused Dl, or at least that there was a substantial risk he had done so, thereby, so it was said, satisfying the section 31(2) conditions for the making of a care order in respect of the three younger girls.

The applications were heard by Judge Davidson Q.C. sitting in the Nottingham County Court. On 23 November, after a hearing lasting seven days, he dismissed the applications. He was not impressed by the evidence of Mr. R or of the mother. Nevertheless he concluded he could not be sure "to the requisite high standard of proof" that Dl's allegations were true. He added:

"It must follow that the statutory criteria for the making of a care order are not made out. This is far from saying that I am satisfied the child's complaints are untrue. I do not brush them aside as the jury seem to have done. I am, at the least, more than a little suspicious that [Mr R] has abused her as she says. If it were relevant, I would be prepared to hold that there is a real possibility that her statement and her evidence are true, nor has [Mr R] by his evidence and demeanour, not only throughout the hearing but the whole of this matter, done anything to dispel those suspicions, but this in the circumstances is nihil ad rem."

By a majority, comprising the President and Millett L.J., the Court of Appeal dismissed an appeal by the local authority. Kennedy L.J. disagreed.

"Likely" to suffer harm

I shall consider first the meaning of "likely" in the expression "likely to suffer significant harm" in section 31. In your Lordships' House Mr. Levy Q.C. advanced an argument not open in the courts below. He submitted that likely means probable, and that the decision of the Court of Appeal to the contrary in Newham London Borough Council v. AG [1993] 1 F.L.R. 281 was wrong. I cannot accept this contention.

In everyday usage one meaning of the word likely, perhaps its primary meaning, is probable, in the sense of more likely than not. This is not its only meaning. If I am going walking on Kinder Scout and ask whether it is likely to rain, I am using likely in a different sense. I am enquiring whether there is a real risk of rain, a risk that ought not to be ignored. In which sense is likely being used in this subsection?

In section 31(2) Parliament has stated the prerequisites which must exist before the court has power to make a care order. These prerequisites mark the boundary line drawn by Parliament between the differing interests. On one side are the interests of parents in caring for their own child, a course which prima facie is also in the interests of the child. On the other side there will be circumstances in which the interests of the child may dictate a need for his care to be entrusted to others. In section 31(2) Parliament has stated the minimum conditions which must be present before the court can look more widely at all the circumstances and decide whether the child's welfare requires that a local authority shall receive the child into their care and have parental responsibility for him. The court must be satisfied that the child is already suffering significant harm. Or the court must be satisfied that, looking ahead, although the child may not yet be suffering such harm, he or she is likely to do so in the future. The court may make a care order if, but only if, it is satisfied in one or other of these respects.

In this context Parliament cannot have been using likely in the sense of more likely than not. If the word likely were given this meaning, it would have the effect of leaving outside the scope of care and supervision orders cases where the court is satisfied there is a real possibility of significant harm to the child in the future but that possibility falls short of being more likely than not. Strictly, if this were the correct reading of the Act, a care or supervision order would not be available even in a case where the risk of significant harm is as likely as not. Nothing would suffice short of proof that the child will probably suffer significant harm.

The difficulty with this interpretation of section 31(2)(a) is that it would draw the boundary line at an altogether inapposite point. What is in issue is the prospect, or risk, of the child suffering significant harm. When exposed to this risk a child may need protection just as much when the risk is considered to be less than fifty-fifty as when the risk is of a higher order. Conversely, so far as the parents are concerned, there is no particular magic in a threshold test based on a probability of significant harm as distinct from a real possibility. It is otherwise if there is no real possibility. It is eminently understandable that Parliament should provide that where there is no real possibility of significant harm, parental responsibility should remain solely with the parents. That makes sense as a threshold in the interests of the parents and the child in a way that a higher threshold, based on probability, would not.

In my view, therefore, the context shows that in section 31(2)(a) likely is being used in the sense of a real possibility, a possibility that cannot sensibly be ignored having regard to the nature and gravity of the feared harm in the particular case. By parity of reasoning the expression likely to suffer significant harm bears the same meaning elsewhere in the Act; for instance, in sections 43, 44 and 46. Likely also bears a similar meaning, for a similar reason, in the requirement in section 3l(2)(b) that the harm or likelihood of harm must be attributable to the care given to the child or "likely" to be given him if the order were not made.

The burden of proof

The power of the court to make a care or supervision order only arises if the court is "satisfied" that the criteria stated in section 31(2) exist. The expression "if the court is satisfied", here and elsewhere in the Act, envisages that the court must be judicially satisfied on proper material. There is also inherent in the expression an indication of the need for the subject matter to be affirmatively proved. If the court is left in a state of indecision the matter has not been established to the level, or standard, needed for the court to be "satisfied". Thus in section 31(2), in order for the threshold to be crossed, the conditions set out in paragraphs (a) and (b) must be affirmatively established to the satisfaction of the court.

The legal burden of establishing the existence of these conditions rests on the applicant for a care order. The general principle is that he who asserts must prove. Generally, although there are exceptions, a plaintiff or applicant must establish the existence of all the pre-conditions and other facts entitling him to the order he seeks. There is nothing in the language or context of section 31(2) to suggest that the normal principle should not apply to the threshold conditions.

The standard of proof

Where the matters in issue are facts the standard of proof required in non-criminal proceedings is the preponderance of probability, usually referred to as the balance of probability. This is the established general principle. There are exceptions such as contempt of court applications, but I can see no reason for thinking that family proceedings are, or should be, an exception. By family proceedings I mean proceedings so described in the Act of 1989, sections 105 and 8(3). Despite their special features, family proceedings remain essentially a form of civil proceedings. Family proceedings often raise very serious issues, but so do other forms of civil proceedings.

The balance of probability standard means that a court is satisfied an event occurred if the court considers that, on the evidence, the occurrence of the event was more likely than not. When assessing the probabilities the court will have in mind as a factor, to whatever extent is appropriate in the particular case, that the more serious the allegation the less likely it is that the event occurred and, hence, the stronger should be the evidence before the court concludes that the allegation is established on the balance of probability. Fraud is usually less likely than negligence. Deliberate physical injury is usually less likely than accidental physical injury. A stepfather is usually less likely to have repeatedly raped and had non-consensual oral sex with his under age stepdaughter than on some occasion to have lost his temper and slapped her. Built into the preponderance of probability standard is a generous degree of flexibility in respect of the seriousness of the allegation.

Although the result is much the same, this does not mean that where a serious allegation is in issue the standard of proof required is higher. It means only that the inherent probability or improbability of an event is itself a matter to be taken into account when weighing the probabilities and deciding whether, on balance, the event occurred. The more improbable the event, the stronger must be the evidence that it did occur before, on the balance of probability, its occurrence will be established. Ungoed-Thomas J. expressed this neatly in In re Dellow's Will Trusts [1964] 1 W.L.R. 451, 455:

"The more serious the allegation the more cogent is the evidence required to overcome the unlikelihood of what is alleged and thus to prove it."

This substantially accords with the approach adopted in authorities such as the well-known judgment of Morris L.J. in Hornal v. Neuberger Products Ltd. [1957] 1 Q.B. 247, 266. This approach also provides a means by which the balance of probability standard can accommodate one's instinctive feeling that even in civil proceedings a court should be more sure before finding serious allegations proved than when deciding less serious or trivial matters.

No doubt it is this feeling which prompts judicial comment from time to time that grave issues call for proof to a standard higher than the preponderance of probability. Similar suggestions have been made recently regarding proof of allegations of sexual abuse of children: see In re G (A Minor) (Child Abuse: Standards of Proof) [1987] 1 W.L.R 1461,1466, and In re W (Minors) (Sexual Abuse: Standard of Proof} [1994] 1 F.L.R. 419,429. So I must pursue this a little further. The law looks for probability, not certainty. Certainty is seldom attainable. But probability is an unsatisfactorily vague criterion because there are degrees of probability. In establishing principles regarding the standard of proof, therefore, the law seeks to define the degree of probability appropriate for different types of proceedings. Proof beyond reasonable doubt, in whatever form of words expressed, is one standard. Proof on a preponderance of probability is another, a lower standard having the in-built flexibility already mentioned. If the balance of probability standard were departed from, and a third standard were substituted in some civil cases, it would be necessary to identify what the standard is and when it would apply. Herein lies a difficulty. If the standard were to be higher than the balance of probability but lower than the criminal standard of proof beyond reasonable doubt, what would it be? The only alternative which suggests itself is that the standard should be commensurate with the gravity of the allegation and the seriousness of the consequences. A formula to this effect has its attraction. But I doubt whether in practice it would add much to the present test in civil cases, and it would risk causing confusion and uncertainty. As at present advised I think it is better to stick to the existing, established law on this subject. I can see no compelling need for a change.

I therefore agree with the recent decisions of the Court of Appeal in several cases involving the care of children, to the effect that the standard of proof is the ordinary civil standard of balance of probability: see H v. H (Minor) (Child Abuse: Evidence) [1990] Fam. 86, 94, 100, In re M (A Minor) (Appeal) (No 2) [1994] 1 F.L.R. 59,67, and In re W (minors) (Sexual Abuse: Standard of Proof) [1994] 1 F.L.R. 419,424, per Balcombe L.J. The Court of Appeal were of the same view in the present case. It follows that the contrary observations already mentioned, in In re G (A Minor) ( Child Abuse: Standard of Proof) [1987] 1 W.L.R. 1461,1466, and In re W (Minors) (Sexual Abuse: Standard of Proof) [1994] 1 F.L.R. 419,429, are not an accurate statement of the law.

The threshold conditions

There is no difficulty in applying this standard to the threshold conditions. The first limb of section 31(2)(a) predicates an existing state of affairs: that the child is suffering significant harm. The relevant time for this purpose is the date of the care order application or, if temporary protective arrangements have been continuously in place from an earlier date, the date when those arrangements were initiated. This was decided by your Lordships' House in In re M (A Minor) (Care Orders: Threshold Conditions) [1994] 2 A.C. 424. Whether at that time the child was suffering significant harm is an issue to be decided by the court on the basis of the facts admitted or proved before it. The balance of probability standard applies to proof of the facts.

The same approach applies to the second limb of section 31(2)(a). This is concerned with evaluating the risk of something happening in the future: aye or no, is there a real possibility that the child will suffer significant harm? Having heard and considered the evidence, and decided any disputed questions of relevant fact upon the balance of probability, the court must reach a decision on how highly it evaluates the risk of significant harm befalling the child, always remembering upon whom the burden of proof rests.

Suspicion and the threshold conditions

This brings me to the most difficult part of the appeal. The problem is presented in stark form by the facts in this case. The local authority do not suggest that the first limb of section 31(2)(a) is satisfied in respect of D2, D3 or D4. They do not seek a finding that any of the three younger girls is suffering harm. Their case for the making of a care order is based exclusively on the second limb. In support of the allegation that D2, D3 and D4 are likely to suffer significant harm, the local authority rely solely upon the allegation that over many years Dl was subject to repeated sexual abuse by Mr. R.

The judge held that the latter allegation was not made out. Mr. R did not establish that abuse did not occur. The outcome on this disputed serious allegation of fact was that the local authority, upon whom the burden of proof rested, failed to establish that abuse did occur. However, the judge remained suspicious and, had it been relevant, he would have held there was a reasonable possibility that Dl's allegations were true. The question arising from these conclusions can be expressed thus: when a local authority assert but fail to prove past misconduct, can the judge's suspicions or lingering doubts on that issue form the basis for concluding that the second limb of section 31(2)(a) has been established?

In many instances where misconduct is alleged but not proved this question will not arise. Other allegations may be proved. The matters proved may suffice to show a likelihood of future harm. However, the present case is not unique. In re P (A Minor) (Care: Evidence) [1994] 2 F.L.R. 751 is another instance where the same problem arose. There the only matter relied upon was the death of the child's baby brother while in the care of the parents. Douglas Brown J. held that it was for the local authority to prove that the death was non-accidental and that, since they failed to do so, there was no factual basis for a finding of likelihood of harm to the surviving child.

In the Court of Appeal in the present case [1995] 1 F.L.R. 643 the President adopted the same approach (at p. 652). Since the judge rejected the only allegation which gave rise to the applications for care orders, it was not then open to him to go on and consider the likelihood of harm to the children. Millett LJ. agreed. He said, at p. 657:

"... where the risk of harm depends on the truth of disputed allegations, the court must investigate them and determine whether they are true or false. Unless it finds that they are true, it cannot be satisfied that the child is likely to suffer significant harm if the order is not made."

Kennedy L.J. reached a different conclusion. To satisfy the second limb there must be acceptable evidence of a real risk that significant harm will be sustained, but he added, at p. 654:

"I . . . do not accept that if the evidence relates to alleged misconduct . . . that misconduct must itself be proved on a balance of probabilities before the evidence can be used to satisfy the threshold criteria in section 31(2)(a)."

A conclusion based on facts