United Kingdom Supreme Court

Michaelmas Term

[2016] UKSC 55

On appeal from: [2015] NICA 34

LORD KERR: (with whom Lord Clarke, Lord Hughes, Lord Toulson and Lord Hodge agree)

Introduction

1.                   On 20 October 2010, following a trial before a judge and jury at Belfast Crown Court, Angeline Mitchell was convicted of the murder on 11 May 2009 of Anthony Robin. Ms Mitchell and Mr Robin had been partners and had lived together for about three years. They had separated some time before May 2009 but they continued occasionally to meet and spend time together. On 10 May 2009 they met at a Belfast hotel at about 3pm. They had something to eat and they drank alcohol there during the remainder of the afternoon and into the evening. They then went to Mr Robin’s flat in Fitzroy Avenue, Belfast, where they continued to drink alcohol.

2.                   A friend of Mr Robin, Michael McGeown, and Mr McGeown’s girlfriend, Jacqueline Cushnan, were also staying at the flat. All four watched a film on television together and then played some music. At about 12.15am on 11 May 2009. Mr Robin received a telephone call from the home of a former partner, Rosena Oliver. She and Mr Robin were the parents of two sons, Anthony junior and Thomas. The sons lived with their mother. After receiving the telephone call, Mr Robin went to Ms Oliver’s home. Ms Mitchell accompanied him. While they were there, Anthony junior was arrested by police officers. Apparently, this was because of a dispute that he had had with his mother. It was then decided that Mr Robin and Ms Mitchell should return to the flat at Fitzroy Avenue and that they should bring the younger son, Thomas, with them.

3.                   On their return to the flat, Mr Robin and Ms Mitchell discussed Anthony junior’s arrest with Mr McGeown and Ms Cushnan for a short time. After that Mr McGeown and Ms Cushnan went to bed. Ms Mitchell continued to talk to Mr Robin about his son’s arrest and repeatedly told him that he should go to the police station to which Anthony junior had been taken. He told her to mind her own business. Exchanges between them became more heated. There is a dispute as to what happened next but it is accepted that, at some stage, Ms Mitchell obtained a knife. It is also accepted that she stabbed Mr Robin. He sustained five knife wounds, one of which caused his death. This was a wound to the left side of the chest which was approximately 20 centimetres deep.

4.                   Other wounds suffered by Mr Robin were to the left side of the scalp, to the right parietal area of the scalp and to the upper thoracic region of the back. Mr Robin’s son, Thomas, claimed to have witnessed the attack on his father. He said that he saw Ms Mitchell “going with a knife” at Mr Robin and that he was crying out as she did so. Thomas said that his father turned away from Ms Mitchell and she continued to attack him.

5.                   When questioned by police at the scene of the killing, Ms Mitchell claimed that there had been a fight between Mr Robin and “a Swedish girl” who had just left the flat. After speaking to others present, however, police officers arrested Ms Mitchell on suspicion of the attempted murder of Mr Robin. She was cautioned. In response she said that she had tried to help Mr Robin and now she was “getting the blame”.

6.                   During her trial, the story about the Swedish girl was not repeated. Ms Mitchell did not dispute that she had stabbed Mr Robin but said that she had acted in self-defence. She also claimed that she had been provoked and that she did not have the intention to kill or cause really serious harm to Mr Robin.

7.                   Before the trial began, the prosecution had intimated an intention to apply to the judge for permission to lead evidence of Ms Mitchell’s previous bad character. This was said to be for the purpose of showing that she had a propensity to use knives in order to threaten and attack others. None of the episodes to which the proposed evidence related had resulted in a conviction. In relation to two of the incidents, it was agreed between the prosecution and the defence that statements settled between them should be read to the jury. The first of these involved an incident in 2003. The agreed statement contained the following sentence, “During the course of a dispute about mobile telephones, Angeline Mitchell chased Andrew McAuley and James People with two knives and tried to stab them”. The second agreed statement related to an incident in December 2007 and was in these terms:

8.                   As well as allowing the agreed statements to be read to the jury, the trial judge permitted the prosecution to adduce evidence of five other incidents involving Ms Mitchell. Two of these related to attempts to attack Michael McGeown with a knife. Two concerned the concealment of knives to prevent possible use of them by her. Evidence was also given about a conversation between Ms Mitchell and Mr McGeown in which she was alleged to have told him that she was going to stab Mr Robin and that before she did so, she would stab Mr McGeown because she knew that he would try to intervene and go to Mr Robin’s aid.

9.                   Despite having authorised her legal advisers to agree that the statements be read to the jury, in the course of giving evidence, Ms Mitchell did not accept that events had occurred in the way described in the statements. The judge dealt with this part of her evidence in the following passage from his charge to the jury:

10.              On the question of how the jury should treat the bad character evidence, it is agreed that the judge did not direct them on whether they required to be satisfied of the truth of the evidence. Nor did he indicate to the jury that they had to be satisfied that the bad character evidence had established the particular propensity on the part of Ms Mitchell which the prosecution had alleged. It appears that counsel who appeared for the prosecution and the defence at the trial did not invite the judge to give any particular form of direction on these topics. Perhaps understandably, therefore, his charge contained no reference to these matters. This is what he said to the jury about the evidence:

11.              Some three years and three months after her trial, Ms Mitchell applied for leave to appeal against her conviction. An extension of time within which to appeal was granted on 6 March 2014 and subsequently Maguire J gave Ms Mitchell permission to appeal on one ground only viz that the trial judge had failed to direct the jury properly on “the purpose of the bad character evidence or the standard of proof to which the jury had to be satisfied before any member of the jury could take the bad character evidence into account in any way”. She was refused leave to appeal on other grounds. Ms Mitchell renewed her application for leave to appeal on those grounds but that application was refused by the Court of Appeal in Northern Ireland (Girvan, Coghlin and Gillen LJJ) on 30 April 2015. The grounds on which leave to appeal was refused have not featured in the appeal before this court and nothing need be said about them.

12.              The Court of Appeal allowed Ms Mitchell’s appeal on the single ground on which she had been given leave to appeal and quashed her conviction. Subsequently they ordered that she should face a retrial. This took place in April 2016. Ms Mitchell pleaded not guilty to murder on the ground of diminished responsibility and pleaded guilty to manslaughter. She was acquitted of murder and sentenced to ten years’ imprisonment for the manslaughter of Mr Robin.

13.              Gillen LJ, who delivered the judgment of the Court of Appeal, said, at para 50 of his judgment, that the correct legal position had been stated by the authors of the 2015 edition of Archbold, Criminal Pleading Procedure and Practice at para 13-68:

“Where non-conviction evidence is being relied on to establish propensity and the evidence is disputed, the jury must be directed not to rely on it unless they are sure of its truth: R v Lafayette [2009] Crim LR 809 and R v Campbell [2009] Crim LR 822, CA.”

14.              This passage does not distinguish between two distinct but potentially overlapping issues: the need, on the one hand, to establish to the requisite criminal standard the facts said to provide evidence of propensity and, on the other, the existence of such a propensity. Should the jury be directed that they have to be satisfied beyond reasonable doubt of the veracity and accuracy of the individual facts which, it is claimed, provide evidence of misconduct on the part of the defendant? Alternatively, is the real issue not this: what requires to be proved is that the defendant did have a propensity? Or must both issues be addressed?

15.              Subsidiary questions arise which will be discussed later in this judgment. These include: should a preliminary evaluation be carried out by the jury of the truth and accuracy of the matters alleged before the question of the existence of propensity is examined or is the proper approach to consider all the evidence about the various instances of misconduct before deciding whether (i) they have been individually established; and (ii) particular instances of misconduct can serve as supportive of evidence in relation to other incidents.

16.              The Court of Appeal in this case referred to two decisions of the English Court of Appeal which, it was suggested, supported the statement in Archbold. The first was R v Ngyuen [2008] EWCA Crim 585; [2008] 2 Cr App R 9. In that case only one incident of previous misconduct was involved, so that proof of propensity might be said to have merged with proof of the facts involved in that incident. The second case was R v O’Dowd [2009] EWCA Crim 905; [2009] 2 Cr App R 16. Three alleged incidents of previous misconduct were involved in O’Dowd. The Court of Appeal in that case, at para 65, paraphrased with approval a statement by Moses LJ in R v DM [2008] EWCA Crim 1544; [2009] 1 Cr App R 10 that the jury would need “to consider with as much detail and concentration all the facts” in relation to the three allegations as they would in relation to the offences of which the appellant was charged. It concluded, at para 83, that because it had proved necessary “to scrutinise the evidence concerning the three disputed allegations” the trial had been unnecessarily prolonged and rendered unduly complex. The convictions were therefore quashed. This was essentially a decision prompted by the view that the trial judge should not have admitted the evidence because allowing it to be given made the jury’s task too difficult. Beatson J’s judgment certainly assumed  that the jury was required to examine separately the question whether facts relevant to each incident of misconduct had been established But to say that the jury must examine the evidence of each previous incident relied upon does not answer the question asked in this case, namely whether it must consider in independent compartments whether each such incident has been proved to the criminal standard, or whether it is enough that, having examined all the evidence, it is satisfied that a propensity to behave as charged has been established.

17.              Both Ngyuen and O’Dowd will be considered in greater detail below. It is sufficient to say at this stage that the Court of Appeal’s endorsement of them in this case would appear to indicate their approval of an approach which involves the individual consideration of the facts of each specific instance of misconduct said to establish a propensity on the part of the defendant. The correctness of that approach will also be considered.

18.              The Court of Appeal was asked by the Crown to give permission to appeal to this court and to certify a question for this court’s opinion. Permission to appeal was refused but the following question was certified:

19.              Mr McCollum QC, who appeared for the prosecution (the appellant before us), submitted that evidence in relation to propensity did not call for any special examination by the jury. In particular, it should not be placed in a special compartment requiring consideration in isolation from other evidence in the case. It was in the nature of this form of trial that all relevant evidence should be assessed by the jury so as to allow them to determine whether they had been brought to the point of conviction of the defendant’s guilt. It was inimical to that fundamental aspect of jury trial that a particular issue be segregated from the generality of the evidence and a pre-emptive decision be made in relation to that issue, before the question of the guilt or innocence of the accused was tackled.

20.              Dealing with all the evidence together was, Mr McCollum said, not only the principled way to proceed, it was necessary in order to avoid the unsatisfactory situation encountered in the O’Dowd case where a number of incidents required the jury to conduct what was in effect a series of individual trials. Moreover, there was nothing in the legislation authorising the admission of bad character evidence (in this case the Criminal Justice (Evidence) (Northern Ireland) Order 2004 (SI 2004/1501)) which expressly or by necessary implication required the incidents which are said to have constituted evidence of propensity to be proved beyond reasonable doubt.

21.              For Ms Mitchell (the respondent in this court) Mr O’Donoghue QC claimed that before the enactment of the 2004 Order the common law rule was that where evidence of bad character was disputed, it had to be proved beyond a reasonable doubt before it could be taken into account by a court or jury. That rule was not abolished by the 2004 Order, he suggested; to the contrary, the Order clearly contemplated that it was the function of the jury to evaluate the evidence of bad character in a conventional way. Facts supporting the claim that the defendant had a particular propensity had to be proved beyond reasonable doubt. It was inconceivable, Mr O’Donoghue argued, that a jury could entertain a reasonable doubt as to the accuracy or veracity of the evidence said to underpin such a propensity and, nevertheless, accept that evidence as sufficient to establish its presence.

22.              At common law, as a general rule, evidence of the bad character of an accused person was not admissible in a criminal trial. There were exceptions to this. One of these was similar fact evidence. An early and notable example of the admission of this type of evidence was the case of Makin v Attorney General for New South Wales [1894] AC 57. In that case, the Lord Chancellor, Lord Herschell, while acknowledging the general rule that the prosecution may not adduce evidence that tended to show that an accused person was guilty of criminal acts other than those with which he had been charged, stated that, in appropriate circumstances it was legitimate to allow evidence to be admitted which was relevant to an issue in the case. Thus, in Makin where the accused had been charged with the murder of an infant who had been given into their care by the child’s mother after payment of a fee, evidence that several other infants had been received by the accused persons from other mothers and that their bodies were found buried in gardens of houses occupied by the prisoners, was admissible.

23.              Makin, together with the later cases of R v Kilbourne [1973] AC 729, R v Boardman [1975] AC 421 and Director of Public Prosecutions v P [1991] 2 AC 447, established the common law rule that, in order to be admissible, similar fact evidence had to go beyond simply demonstrating a criminal tendency (or propensity). It had to show sufficient pattern of behaviour, underlying unity or nexus to exclude coincidence and thus have probative force in proving the indicted allegation. In Scotland the same distinction was long recognised: see Moorov v HM Advocate 1930 JC 68. Clearly, the evidence in Makin was relevant in that, if accepted, it had, at least, the potential to show that the defendants were more likely to have killed the child. The decision in that case does not address the issue which is central to this appeal, however, since the question of how evidence of similar facts, if properly admitted, should be treated, did not arise. The case is of interest only as part of the background to the exception to the general common law rule that evidence of antecedent misconduct is not admissible unless shown to be directly relevant to an issue in the trial. Since, as I shall discuss below, evidence of propensity or similar fact evidence is, essentially, extraneous to that which is directly probative of the accused’s guilt of the charges on which he stands trial, the case can be made that it should be subject to the conventional criminal standard requirement of proof beyond reasonable doubt. And, it may be said, this is especially so where the claims in relation to similar fact evidence or propensity are disputed.

24.              In R v Armstrong [1922] 2 KB 555 the accused, a solicitor, was charged with the murder of his wife by giving her arsenic. His defence was that he had not administered the poison, although he admitted that he had arsenic which, he said, he used as a weed killer. He claimed that his wife had either committed suicide or had taken the arsenic by accident. The prosecution was permitted to call evidence that another solicitor, a Mr Martin, had visited the accused’s home eight months after his wife’s death and had suffered an episode of arsenic poisoning that evening. The purpose of calling evidence about the attempt to poison Mr Martin was, the prosecution said, to rebut the suggestion that Mrs Armstrong had either committed suicide or taken the arsenic by accident. The trial judge, Darling J. directed the jury that, unless it was proved that Armstrong had given arsenic to Martin with intent to injure him, the evidence had “no bearing whatever upon this case”. That direction was not disapproved by the Court of Appeal when it affirmed the accused’s conviction.

25.              It was argued on behalf of the present respondent, therefore, that this case supported the proposition that the common law rule was that the prosecution had to prove to the criminal standard “the truth of all similar fact evidence irrespective of the purpose for which it was admitted”. This claim must be viewed against the background that Armstrong involved a single instance of similar fact evidence. If there was doubt about the truth or accuracy of the evidence relating to the Martin episode, one can readily understand why it would have to be disregarded by the jury. The purpose of introducing the evidence was to show that Armstrong was prepared to use arsenic on another individual. If the jury was not convinced that he had administered poison to Mr Martin, self-evidently, that incident could not be used as proof of a propensity on the part of Armstrong to use arsenic to poison others.

26.              What of the situation where there are several disparate instances of alleged antecedent conduct, said to demonstrate propensity or evidence of similar facts? Must each incident be the subject of a compartmentalised examination designed to determine whether the constituent elements of each have been established beyond reasonable doubt? Or should consideration of these various instances partake of a rounded evaluation to see whether, taken as a whole, the antecedent evidence establishes a propensity or that the alleged similar facts may properly be described as such and that they tend to support the prosecution’s case that the defendant is guilty of the charges which he or she faces? It will be necessary to explore these issues later in this judgment.

27.              Hints of a different approach can be detected in R v Kilbourne [1973] AC 729. In that case the respondent was convicted of sexual offences against two groups of boys. The trial judge directed the jury that they would be entitled to take into account the uncorroborated evidence of the second group as supporting evidence given by the first group. At p 741, Lord Hailsham of St Marylebone LC said:

“A considerable part of the time taken up in argument was devoted to a consideration whether such evidence of similar incidents could be used against the accused to establish his guilt at all, and we examined the authorities in some depth from Makin v Attorney General for New South Wales [1894] AC 57, through Lord Sumner’s observations in Thompson v The King [1918] AC 221, to Harris v Director of Public Prosecutions [1952] AC 694. I do not myself feel that the point really arises in the present case. Counsel for the respondent was in the end constrained to agree that all the evidence in this case was both admissible and relevant, and that the Court of Appeal was right to draw attention [1972] 1 WLR 1365, 1370 to the ‘striking features of the resemblance’ between the acts alleged to have been committed in one count and those alleged to have been committed in the others and to say that this made it ‘more likely that John was telling the truth when he said that the appellant had behaved in the same way to him.’ In my view this was wholly correct. With the exception of one incident.

‘each accusation bears a resemblance to the other and shows not merely that [Kilbourne] was a homosexual, which would not have been enough to make the evidence admissible, but that he was one whose proclivities in that regard took a particular form’ [1972] 1 WLR 1365, 1369.

I also agree with the Court of Appeal in saying that the evidence of each child went to contradict any possibility of innocent association. As such it was admissible as part of the prosecution case, and since, by the time the judge came to sum up, innocent association was the foundation of the defence put forward by the accused, the admissibility, relevance, and, indeed cogency of the evidence was beyond question. The word ‘corroboration’ by itself means no more than evidence tending to confirm other evidence. In my opinion, evidence which is (a) admissible and (b) relevant to the evidence requiring corroboration, and, if believed, confirming it in the required particulars, is capable of being corroboration of that evidence and, when believed, is in fact such corroboration.”

28.              The primary issue in Kilbourne was, of course, whether evidence which required, as a matter of law, corroboration could be verified by other evidence which itself, again as a matter of law, had to be corroborated. The case is interesting in the present context, however, because of the assertion by Lord Hailsham LC that evidence which is admissible and relevant, if believed, could properly be taken into account as corroborative of the case against the accused.

29.              Counsel for the respondent in this appeal accepted that the use of the phrases “if believed” and “when believed” in the passage quoted did not suggest that Lord Hailsham LC was proposing that proof beyond reasonable doubt of each of the other incidents was required. But, nor did it mean, he suggested, that it could be left to the jury to treat the evidence, once admitted, in whatever manner they chose. The jury was required, he submitted, to make some evaluation of the truth of the similar fact evidence and the speech of Lord Hailsham LC was not necessarily inconsistent with that evidence having to meet a requirement of proof beyond reasonable doubt before it could play a part in their deliberations.

30.              Lord Hailsham LC dealt again with the issue of similar fact evidence in R v Boardman [1975] AC 421. He made a passing reference (at p 453) to its being a matter for the jury to decide on the “degree and weight” to attach to such evidence, if admitted. This does not provide much in the way of authoritative guidance as to the standard of proof to be applied to similar fact evidence. Likewise, in R v Scarrott [1978] QB 1016 Scarman LJ referred to the need for similar fact evidence to be “believed” and the need for the jury “to accept the evidence”. These allusions cannot begin to constitute a statement of commanding clarity as to how a jury should assess such evidence.

31.              Consideration of these authorities (and R v Z [2000] 2 AC 483 to which the respondent also referred) leads inevitably to the conclusion that before the enactment of the 2004 Order (and its England and Wales counterpart, the Criminal Justice Act 2003) there was no clear, definitive statement on the issue now raised as to how juries should treat evidence of similar facts or propensity.

The 2004 Order

32.              Article 4(1) of the 2004 Order (mirroring section 99 of the 2003 Act) abolished the common law rules governing the admissibility in criminal proceedings of evidence of bad character. The Order significantly expanded the circumstances in which bad character evidence could be admitted and the rules which previously restricted admission of such evidence now have no part to play in the decision as to whether it should be received.

33.              Article 6(1) sets out a number of what have come to be known as “gateways” for the admission of evidence of a defendant’s bad character. It provides:

34.              Article 8 deals specifically with propensity. It relates back to gateway (d) of article 6(1). Article 8(1) provides:

35.              In this case the grounds on which the admission of evidence of the respondent’s bad character was sought to be introduced were stated to be that (i) it was relevant to an important matter between the defendant and the prosecution (article 6(1)(d)). The “important matter” was said to ‘include’ the propensity of the respondent “to use knives to wound others”; (ii) it was important explanatory evidence of the character of the respondent (article 6(1)(c)); (iii) it corrected a false impression given by the respondent about herself (article 6(1)(f)); and (iv) it was admissible because the respondent had “attacked the character of another person, namely, the victim, Anthony Robin”.

36.              We have not seen the trial judge’s ruling on the application to admit evidence of bad character. The Court of Appeal dealt with the matter shortly in para 28 where Gillen LJ observed that counsel for the respondent had accepted that the prosecution was entitled to call evidence of her bad character under article 6(1)(d) and 6(1)(g). He also said that Mr O’Donoghue had accepted that “it was at least arguable that the evidence [of bad character] … may have been capable of establishing that the [respondent] had a propensity to arm herself with a knife and to use the knife for the purpose of and with the intention of inflicting serious bodily harm.” If admission had been solely under articles (c), (f) or (g), it would have been necessary to consider at the end of the evidence whether propensity had become a legitimate issue, and how the jury should be directed as to the use which could be made of it. But whether or not it was also admitted under article 6(1)(f), this evidence was plainly admissible under article 6(1)(d), and thus propensity to offend as charged was a relevant matter which the Crown could seek to establish. It was no part of the respondent’s case that the evidence was wrongly admitted.

37.              What the respondent does assert, however, is that the law before the enactment of the 2003 Act and the 2004 Order was that the prosecution was required to prove to the criminal standard the truth and accuracy of evidence said to constitute similar facts or propensity. For the reason given in para 31 above, I do not accept that claim. The respondent is unquestionably right in the submission that neither the 2003 Act nor the 2004 Order stipulates that only the common law rules as to the admissibility of bad character evidence have been abrogated. Common law rules as to how such evidence should be evaluated have not been affected, the respondent says. But, for the reasons earlier given, there are no clear rules on that question. The debate as to how evidence of bad character admitted under the relevant legislation should be regarded by the jury is not assisted by a consideration of the common law position.

38.              It is common case between the parties that the legislation concerning the admission of bad character evidence is silent on the question of whether that evidence must meet the requirement of proof beyond reasonable doubt before it can be taken into account. On one view, this is indicative of a legislative intention that this species of evidence should not be subject to a special regime of independent proof. That it should simply combine with the other evidence in the case for evaluation as to whether the guilt of the accused has been established to the requisite standard. The contrary view is that whether someone has a propensity to engage in activity such as that which constitutes the crime charged or whether they have been involved in acts of a similar nature stands apart from direct evidence of their actual involvement in the crime charged. On that account, so the argument runs, the question whether they have such a propensity or have been involved in events claimed to comprise similar facts, calls for consideration separate from the evidence which directly implicates the accused in the offence for which they are being tried.

39.              A distinction must be recognised between, on the one hand, proof of a propensity and, on the other, the individual underlying facts said to establish that a propensity exists. In a case where there are several incidents which are relied on by the prosecution to show a propensity on the part of the defendant, is it necessary to prove beyond reasonable doubt that each incident happened in precisely the way that it is alleged to have occurred? Must the facts of each individual incident be considered by the jury in isolation from each other? In my view, the answer to both these questions is “No”.

40.              In the case of Ngyuen [2008] 2 Cr App R 9 the appellant had been convicted of murder. On 23 December 2005 in a public house in Woolwich he had struck the victim in the neck with a glass and this caused fatal injuries. The prosecution alleged that this had been a deliberate attack. The appellant claimed that he was acting in self-defence. On 7 December 2005 he had been involved in a similar incident at a different public house which was called “The Great Harry”. On that occasion he had broken a glass and used it to cause injuries to three men. The Crown applied for leave to admit evidence of the incident in The Great Harry. In acceding to that application, the judge said that “the jury will have to be sure of the facts before they can use them, applying the criminal burden and standard of proof”. In directing the jury about the earlier incident, she said that the prosecution relied on three relevant matters - that the appellant had deliberately broken a glass; that he used it with the intention of causing really serious harm; and that he had done so unlawfully. The judge then said, “If you are not sure of any of those facts, the events in The Great Harry are irrelevant to your deliberations on the charge of murder”.

41.              The propriety of the judge’s charge was confirmed by the Court of Appeal. It should be noted, of course, that no challenge to its wording was made by the prosecution or the defence. The appeal, so far as concerns the judge, was based on the claim that she had been wrong to admit the evidence of The Great Harry incident and that the direction on that incident, although fair, was “too much for the jury to apply faithfully and conscientiously”.

42.              It is significant that the only bad character evidence in Ngyuen related to a single previous incident. In order to be convinced that the appellant in that case was possessed of the propensity which the prosecution alleged, it is not surprising that the judge considered that it was necessary for the jury to be persuaded that The Great Harry incident had taken place as alleged. If the jury needed to be sure that the appellant had the alleged propensity (and I am of the view that this was certainly required, if they were to take it into account), how could they, in the circumstances of that particular case, be brought to the necessary point of conviction unless they were convinced that the incident had indeed taken place in the manner that the Crown said that it had? Otherwise, there was simply no factual basis on which to found a conclusion that a propensity existed.

43.              The proper issue for the jury on the question of propensity in a case such as Ngyuen and the present appeal is whether they are sure that the propensity has been proved. In Ngyuen the only way in which they could be sure was by being convinced that the sole incident said to show propensity had been proved to the criminal standard. That does not mean that in cases where there are several instances of misconduct, all tending to show a propensity, the jury has to be convinced of the truth and accuracy of all aspects of each of those. The jury is entitled to - and should - consider the evidence about propensity in the round. There are two interrelated reasons for this. First the improbability of a number of similar incidents alleged against a defendant being false is a consideration which should naturally inform a jury’s deliberations on whether propensity has been proved. Secondly, obvious similarities in various incidents may constitute mutual corroboration of those incidents. Each incident may thus inform another. The question impelled by the Order is whether, overall, propensity has been proved.

44.              As I have said, the existence of a propensity must be proved to the conventional criminal standard. I do not accept the appellant’s argument that it does not call for “special” treatment, if by that it is meant that the existence of a propensity need not be established beyond reasonable doubt. This issue stands apart from the evidence which speaks directly to the defendant’s guilt or innocence of the offences charged. Evidence about a propensity or tendency to commit a specific type of crime or engage in a particular species of misconduct is not in pari materia with testimony that touches on the actual events said to constitute the particular crime involved. It is right, therefore, that the jury should be directed that before they take this into account, they must be convinced that propensity has been proved. That is not to say that the jury must be unanimous on the question of whether it exists. As the judge said in Ngyuen, jurors are at liberty to follow their own evidential track. But the jury should be directed that, if they are to take propensity into account, they should be sure that it has been proved. This does not require that each individual item of evidence said to show propensity must be proved beyond reasonable doubt. It means that all the material touching on the issue should be considered with a view to reaching a conclusion as to whether they are sure that the existence of a propensity has been established.

45.              In the case of R v Lafayette [2008] EWCA Crim 3238; [2009] Crim LR 809 the defendant was convicted of murder, having stabbed his victim with a knife. He claimed that the victim had produced the knife and that, in self-defence, he had grabbed the victim’s hand and accidentally caused the knife to enter his body. Cross examination of some of the eye witnesses called by the prosecution about their criminal records led to an application by the prosecution for permission to cross examine the defendant about his own previous convictions including one in 2003 for criminal damage to a window in a flat occupied by his partner. The application was granted. The partner had made a statement to the police that during this incident he had threatened to slit her throat. He had not been charged with making that threat. He denied having made it. As it happened, his partner gave evidence on the defendant’s behalf on his trial for murder. She was asked about the threat alleged to have been made in 2003. She claimed that she was not able to remember it. On the appeal against his conviction for murder, counsel for Lafayette accepted that the jury would have been entitled to conclude that the partner had adopted the written statement that she had made in 2003 in which reference to the threat to kill had been made.

46.              The Court of Appeal said (in para 36 of its judgment) that the judge should have directed the jury that they should not rely on the allegation that he had threatened to kill his partner “unless they were sure that [he] had made the threat.” One can understand why this conclusion was reached. A very specific threat had been imputed to the defendant and the evidence about it was, at best, equivocal. The incident to which the evidence related was not similar to other instances of criminal conduct which were referred to by the Crown, in support of its claim that the defendant had a general propensity to crimes of violence. If the judgment of the Court of Appeal can be interpreted as suggesting that the jury should consider this item of evidence completely separately from other alleged incidents of violence in the defendant’s past and should leave it entirely out of account unless satisfied that all aspects of the incident were proved to the criminal standard, I would not agree with it. The evidence relating to the threat required to be considered by the jury along with other evidence which was called to establish propensity and a determination ought to have been made on whether all that testimony, taken in combination, proved the claimed propensity. Each item of evidence in relation to individual instances of alleged propensity must be examined and conclusions on the primary facts should be reached but, in its deliberations as to whether propensity has been proved, the jury should consider the evidence on the subject as a whole rather than in individual compartments.

47.              The practical difficulties in dealing with each avowed instance of bad character tending to show propensity or similar facts are well demonstrated in the case of O’Dowd [2009] 2 Cr App R 16. In that case the trial of a single defendant on charges relating to one victim lasted six and a half months. He had been convicted of falsely imprisoning, raping, sexually assaulting and poisoning a woman. The Court of Appeal stated that a major reason for the length of the trial was the introduction of bad character evidence. This concerned three allegations of rape, two of which related to events that had occurred 22 and 17 years before the indicted charges. The first of the allegations resulted in an acquittal, the second in a conviction and the third was stayed on the ground of abuse of process.

48.              Interestingly, it had originally been argued on behalf of the defendant that the trial judge was wrong to have directed the jury that the bad character allegations were capable of mutually supporting the truth of other allegations. Beatson J, who delivered the judgment of the court, dealt with that argument in para 6 of his judgment:

“The second ground upon which leave to appeal was granted concerned the judge’s directions as to the use the jury could make of the bad character evidence. This ground has two limbs. The first concerns the direction that the bad character allegations were capable of mutually supporting the truth of the other allegations. … [Counsel] did not pursue the first limb. He was right not to do so. It was unarguable in the light of the decisions of this court in [R v Wallace [2007] 2 Cr App R 30; R v DM [2009] 1 Cr App R 10, and R v Freeman [2009] 1 WLR 2723].”

49.              In light of what I have said at para 43 above, I obviously agree with this. It would be misleading and confusing for a jury to be instructed that they should ignore the significance of one incident tending to show propensity when they come to form their views about another. Indeed, it would be unrealistic to expect that they perform the counter intuitive intellectual exercise of segregating various incidents for separate consideration without considering the possible impact of one on the other. Decisions about propensity should not be the product of a review of facts about separate episodes in hermetically sealed compartments.

50.              In para 32 of his judgment Beatson J set out the case made by the defendant that the evidence of bad character should not be admitted. He said this:

51.              What Beatson J characterised as the most serious difficulty in acceding to the Crown’s application to have the bad character evidence admitted was described in para 55 of his judgment. He said that proof of the alleged misconduct “would require the trial of three collateral or satellite issues as part of the trial …”. This perceived difficulty led to the Court of Appeal’s conclusion (at para 65) that a trial of those collateral issues was required. It was in this context that the statement of Moses LJ in DM (referred to in para 16 above) was quoted with approval. Beatson J said that the jury would “have to be sure those allegations were true before relying on them in relation to the index offence”. Because, in effect, the trial was lengthened by a considerable period because of the perceived need to conduct what were in effect three mini-trials, the Court of Appeal considered that the fairness of the proceedings was irredeemably compromised and the convictions were quashed.

52.              I can understand why the Court of Appeal reached the decision that it did. Three trials of the earlier incidents were in fact conducted and the appellate court had to deal with that unalterable fact. But I am of the clear view that it was inappropriate for the jury to be directed that it had to examine in insulated compartments the evidence in relation to each previous incident and that it had to be sure that each incident had been proved before it could take any account of it. It was, of course, necessary to lead evidence of the three incidents. The jury should have been directed to consider whether the sum of that evidence established to the criminal standard that the defendant was possessed of the propensity which was alleged. The evidence in relation to those incidents should have been considered cumulatively, not as separate aspects of the case for a propensity, isolated one from the other.

53.              O’Dowd nevertheless illustrates an important consideration which must be borne in mind by trial judges when determining applications to adduce evidence of propensity under articles 6(1)(d) and 8(1)(a). The jury is not asked to return a verdict on any previous allegations relied upon, and indeed should be reminded that the defendant is not on trial for them. It should be told to focus on the indicted offence(s). Reliance on cumulative past incidents in support of a case of propensity may indeed illuminate the truth of the currently indicted allegations, but excessive recourse to such history may skew the trial in a way which distracts attention from the central issue. Article 6(3) requires the judge to consider actively whether the effect of admitting the bad character evidence will have such an adverse effect on the fairness of the trial that it ought to be excluded. That species of adverse effect can arise through the sheer weight of disputed evidence on other uncharged allegations. And that can happen even though the jury will in due course be directed to consider propensity cumulatively, if the volume of evidence received is sufficiently strong to support a conviction. It is a truism that satellite litigation is often inimical to efficient trial.

54.              In so far as the Court of Appeal in the present case has suggested that each incident claimed by the prosecution to show a propensity on the part of the defendant required to be proved to the criminal standard. I would not agree with it. For the reasons that I have given, the proper question to be posed is whether the jury is satisfied that a propensity has been established. That assessment depends on an overall consideration of the evidence available, not upon a segregated examination of each item of evidence in order to decide whether it has been proved beyond reasonable doubt.

55.              It is necessary to emphasise, however, that propensity is, at most, an incidental issue. It should be made clear to the jury that the most important evidence is that which bears directly on the guilt or innocence of the accused person. Propensity cannot alone establish guilt and it must not be regarded as a satisfactory substitute for direct evidence of the accused’s involvement in the crime charged.

56.              It is clear in the present case, however, that the trial judge failed to give adequate directions as to how the question of propensity should be approached by the jury. On that account the conviction was unsafe and it was properly quashed.

57.              The current Bench Books for Northern Ireland and England and Wales contain specimen directions which might be considered to suggest that juries require to be directed that they need to be satisfied of the truth of every allegation of propensity before they may take it into account. For the reasons that I have given, I consider that such a suggestion is misconceived. It will be for the authors of those works to reflect on whether an amendment to the relevant sections of the Bench Books is required.