HKSAR v. MARIA REMEDIOS COADY

CACC 119/1999

IN THE HIGH COURT OF THE

HONG KONG SPECIAL ADMINISTRATIVE REGION

COURT OF APPEAL

CRIMINAL APPEAL NO. 119 OF 1999

(ON APPEAL FROM HCCC NO. 194 OF 1998)

______________

_______________

J U D G M E N T

_______________

Keith JA (giving the judgment of the Court):

Introduction

1. The Applicant was charged with murder. She was tried by Verina Bokhary J and a jury in the Court of First Instance. She pleaded not guilty. However, after a trial lasting a number of days, she was convicted by a majority of 6:1. She was sentenced to imprisonment for life. She now applies for leave to appeal against her conviction.

2. In her summing-up, the judge gave the jury the conventional direction on the mens rea necessary for murder, i.e. a direction in accordance with the "grievous harm" rule. One of the Applicant's grounds of appeal challenged the correctness of that direction. We heard argument on this ground of appeal a few months ago. We held that this ground of appeal failed: [2000] 2 HKLRD 195. We directed that the appeal be re-listed for the other grounds of appeal to be heard, and this is our judgment on those other grounds.

3. Since the ground of appeal relating to the correctness of the "grievous harm" rule involved a question of law alone, leave to appeal on that ground was unnecessary. However, the other grounds of appeal, being questions of mixed law and fact, are grounds for which leave to appeal is required. Since those grounds are arguable ones, we grant the Applicant leave to appeal on those grounds, and from now on we shall refer to her as the Appellant.

The relevant facts

4. Shortly before midnight on 19 December 1997, the emergency services received a 999 call. It was from the Appellant. She was in a distraught and emotional state. She asked for an ambulance to come to the flat in Wan Chai where she was because someone had had a heart attack. She did not say who. When the ambulance men arrived, they found a man lying on the floor of the living room. He had been stabbed several times. The following afternoon, he died as a result of his injuries. He and the Appellant had been lovers, and had spent time together in the flat.

5. At all times, the Appellant denied that she had been responsible for his injuries. She claimed that she and the deceased had been attacked in the street outside the flat by two men who had tried to steal her handbag. In her evidence, the Appellant said that she and the deceased had returned to the flat. He had begun to act strangely, including threatening to kill himself and her. She had told him to calm down, and had gone into the bedroom where she had fallen into a light sleep. After a while, she had heard a loud noise as if someone had fallen over. She had called out a number of times but had got no reply. Eventually she had got up, and had found the deceased sitting in an armchair. After a while, she had realised that the deceased was holding a towel with blood on it over his stomach. He was pale and cold. After failing to get any help from neighbours, she had telephoned the emergency services. As the deceased had been clutching his chest, she had thought that he might have had a heart attack, which was why she had referred to that in the emergency call which she had made.

6. In the light of the Applicant's denial that she had been responsible for the deceased's injuries, there were, on her version of events, three possible ways in which the deceased had sustained his injuries:

(i) they could have been caused during the attack on him and the Appellant in the street;

(ii) they could have been caused by the robbers (or other unknown persons) who had followed the deceased and the Appellant and had come into the flat while the Appellant had been in the bedroom; or

(iii) the deceased could have inflicted the wounds on himself, either in an attempt to commit suicide or to gain sympathy or attention.

7. A bloodstained knife had been found in the flat. There was undisputed evidence that (a) the knife could have caused the deceased's injuries, (b) there was a 1,440:1 probability that the blood on the knife was that of the deceased, and (c) the knife (and a similar one without bloodstains) matched two sheaths found in the kitchen. In the light of this evidence, Mr John Mullick (who represented the Appellant at her trial) told the jury that the defence would not dispute that the knife had been the one which had caused the deceased's injuries, and that it must already have been in the flat. The effect of that concession was to remove the possibility that the deceased's injuries may have been caused during the attack on him and the Appellant in the street.

8. In those circumstances, the forensic evidence as to whether the deceased's injuries were likely to have been self-inflicted became critical. If they had not been self-inflicted, only two possibilities remained: (i) the deceased had been attacked while the Appellant was in the bedroom, or (ii) it was the Appellant who had stabbed him. Since the possibility that the robbers had followed the deceased and the Appellant and had come into the flat and attacked the deceased while the Appellant had been in the bedroom was not a very plausible scenario, and was unlikely to be one which the jury thought was possible, a conclusion that the wounds had not been self-inflicted was likely to lead to the conclusion that, despite her denials, they had been inflicted by the Appellant.

The grounds of appeal

9. Apart from the ground of appeal relating to the correctness of the "grievous harm" rule, two grounds of appeal were advanced by Mr Gerard McCoy SC (who did not appear for the Appellant at her trial). The first was that the judge in her summing-up had not dealt with the expert evidence on how the deceased had died in a balanced way with the result that the Appellant had not received a fair trial on the issue as to whether the wounds from which the deceased had died could have been self-inflicted. The second was that the judge had erred in failing to leave the issue of provocation to the jury.

10. Although Mr McCoy and Mr Andrew Bruce SC for the prosecution (who likewise did not appear at the trial) addressed us initially on the issue of provocation, we think that analytically the question whether the summing-up on the issue whether the deceased's wounds had been self-inflicted was a balanced one should be decided first. After all, whether the wounds had been self-inflicted or whether it had been the Appellant who had stabbed the deceased had to be determined before the question as to whether she had been provoked to do so arose.

The expert evidence

11. The expert evidence relating to whether the wounds from which the deceased had died had been self-inflicted concentrated on (a) the deceased's injuries and (b) the damage which had been done to the shirt and jacket which he had been wearing at the time. As for (a), the unchallenged evidence of Dr Beh Swan Lip, the forensic pathologist who examined the deceased two days after he died, was that the deceased had been wounded in three places: a superficial cut on his right arm, a cut on his right thumb, and a deep wound in his abdomen which injured his liver and from which he died. As for (b), the unchallenged evidence of Bobby Cheung, the forensic scientist who examined the deceased's clothing, was that the deceased's jacket had been cut in seven places on the right side (four in the front, two on the sleeve, and one in the back). All but one of these cuts corresponded to cuts in the deceased's shirt. The ambulance men had cut the deceased's shirt in such a way that Mr Cheung could not tell whether there was a cut in the shirt which corresponded with the remaining cut in the jacket.

12. The thrust of Mr Cheung's evidence was that the nature, trajectory and position of the cuts in the deceased's clothing made it very unlikely that the wounds had been self-inflicted. That particularly applied to the cut on the back of the jacket and the shirt. For his part, Dr Beh was unable to proffer an opinion, from his examination of the deceased's wounds as opposed to the deceased's clothing, as to whether the wounds had been self-inflicted, though when asked in examination-in-chief whether the possibility that the wounds had been self-inflicted had occurred to him when he had first seen the wounds, he said that that possibility had not.

13. Mr Cheung's opinion was questioned by Prof. Anthony Ansford, a forensic pathologist called on behalf of the Appellant. He disputed Mr Cheung's expertise, and he expressed the opinion that the fact that (a) the cuts were only on the right side of the deceased's clothing and (b) there were more cuts on the deceased's clothing than on his body suggested that the wounds had been self-inflicted. He regarded it as more likely that the wounds had been self-inflicted.

The judge's treatment of the expert evidence

14. The criticism of the judge relates to the way in which she summarised Prof. Ansford's evidence on this topic when compared with the way in which she summarised that of Mr Cheung and Dr Beh. Almost 12 pages of the transcript of her summing-up (pp. 28R-40E) are devoted to her summary of the evidence of Mr Cheung and Dr Beh - in particular, Mr Cheung's opinion as to whether each of the cuts in the deceased's jacket and shirt were consistent with the deceased having wielded the knife himself - whereas she dealt with Prof. Ansford's evidence in just over one page (pp. 40J-41L). What she said about Prof. Ansford's evidence was this:

"I think that the essence of Professor Ansford's evidence, relied upon by the defence, is that the more likely scenario is self- infliction whether by a deliberate suicidal act, attention seeking or threatening, and that the idea that the marks on the deceased's clothings and his wounds were caused by someone attacking him is unusual or unlikely. Professor Ansford does not agree with Mr Cheung, and disputes Mr Cheung's expertise to give the evidence which he gave.

Members of the jury, you will recall the passage in Professor Ansford's statement, commented upon by Dr Beh, which contains the sentence:

'Such a linear pattern of wounds is extremely unusual in a homicidal stabbing, particularly if there is a struggle and is highly suggestive of self-infliction.'

[Counsel] for the prosecution, when cross-examining Professor Ansford, put that sentence to Professor Ansford and asked him if he withdrew it now that he had modified his assertion that all the deceased's wounds and the damage to his clothing were virtually in a straight line down the right side. Professor Ansford answered 'I withdraw "the highly suggestive" '.

Professor Ansford adhered to his disagreement with Mr Cheung and Mr Cheung's qualification to say how the deceased's clothing was damaged. But Professor Ansford did agree with this statement put to him by [counsel for the prosecution] from a book:

'Damage to clothing is commonly encountered in serious crimes of violence such as homicide and rape. Examination of damaged clothing and other textiles may provide valuable information as to the possible implement that caused the damage and the manner in which it was caused. Damage analysis may corroborate or refute a particular crime scenario.' "

Having summarised Prof. Ansford's evidence in that way, the judge returned to Mr Cheung's evidence. She repeated his conclusion that self-infliction was "extremely remote" (pp. 53S-54B), and summarised again the comments which Mr Cheung had made as to whether the cuts in the deceased's clothing were consistent with the deceased having caused them himself (pp. 54K-55N).

15. It is not suggested that there was anything inaccurate in the judge's summary of the evidence. The criticism relates to the disproportionate time which the judge devoted to the evidence of Mr Cheung and Dr Beh on the one hand and that of Prof. Ansford on the other. Two consequences were said by Mr McCoy to flow from that. First, it undermined the significance of Prof. Ansford's opinion in the eyes of the jury. The judge's treatment of Prof. Ansford's evidence might have led the jury to think that she was sceptical about the correctness of his opinion that the wounds were more likely to have been self-inflicted. If the judge was sceptical about the correctness of his opinion, the jury would have been likely to treat his opinion with scepticism as well. It is true that, on a number of occasions in the course of her summing-up, the judge gave the jury the standard direction that if she happened to express her personal view about any aspect of the evidence, the jury should disregard that view unless they happened to agree with it. However, Mr McCoy said that that did not overcome the problem:

"A judge .... is not entitled to comment in such a way as to make the summing up as a whole unbalanced .... It cannot be said too often or too strongly that a summing up which is fundamentally unbalanced is not saved by the continued repetition of the phrase that it is a matter for the jury": Lloyd LJ (as he then was) in R. v. Gilbey (unreported), 26 January 1990.

This dictum was cited with approval by the Privy Council in Mears v. R. [1993] 1 WLR 818 at p. 822F. In other words, the recitation of the conventional mantra will not neutralise the effect of an unbalanced summing-up.

16. Secondly, the effect of the judge's summing-up was to remind the jury of the reasons for Mr Cheung's conclusion that the wounds were unlikely to have been self-inflicted, but not to remind the jury of the reasons for Prof. Ansford's conclusion that the wounds were likely to have been self-inflicted. That meant that the jury had had to evaluate Prof. Ansford's evidence without having been reminded of the details of it, even though (for reasons which it is unnecessary to detail in this judgment) his evidence had been given 18 days before the judge commenced her summing-up to the jury.

17. These are powerful arguments, but we have not been persuaded by them. Two important points must be made. First, Mr Cheung's opinion that the wounds were unlikely to have been self-inflicted was based on the nature, trajectory and position of each of the cuts in the deceased's clothing. It was therefore necessary for him to set the scene by giving evidence about where each of the cuts were and what they were like, and then to explain why each of them supported his conclusion. Prof. Ansford did not have to set the scene as that had been done by Mr Cheung. Indeed, Prof. Ansford complimented Mr Cheung on the photographs and diagrams produced by him, and on the thorough and impressive job he had done. Moreover, Prof. Ansford's opinion that the wounds were likely to have been self-inflicted was not based on each of the cuts when looked at individually, but on their pattern as a whole. The transcript of his evidence records him as saying:

"I would not seek to make comments on individual wounds necessarily, either or individual cuts, but the pattern, to me, of the cuts and the wounds is what is important rather than the .... individual cuts" (p. 269M-N).

And a little later he said:

"The pattern of the cut holes in the jacket is to me what is significant not the individual cuts taken in isolation. And the fact that they're all on one side" (p. 274L-M).

Thus, apart from the need for Mr Cheung to set the scene, on the critical question as to whether the cuts were more or less consistent with self-infliction, the basis on which Mr Cheung formed his opinion by definition required greater explanation by him (and therefore a longer summary by the judge) than that on which Prof. Ansford's opinion was formed.

18. Secondly, we reject the suggestion that the jury was only reminded of Prof. Ansford's views in the passage at pp. 40J-41L of the transcript to which we have referred. Prof. Ansford had prepared a report containing his conclusions, and although that report had not been an exhibit in the case and had not therefore been placed before the jury, significant passages in the report had been put to Mr Cheung and Dr Beh for their comments when they were giving evidence. In her summing-up, the judge reminded the jury of the salient passages in the report on which the comments of Mr Cheung and Dr Beh had been sought. In one of those passages, Prof. Ansford had referred to the deceased's wounds and the cuts in his clothing as all being on the right side (though in his evidence he retracted his assertion that they were virtually in a straight line). In another passage, he expressed the opinion that "such a linear pattern of wounds [was] extremely unusual in a homicidal stabbing, particularly if there is a struggle and is highly suggestive of self-infliction" (though in his evidence he retracted the assertion that it was "highly" suggestive). And in yet another passage, he said:

"It is highly unlikely that the deceased would have stood calmly while the [accused person] stabbed him seven times in a line with at least two penetrating the body. If he was asleep and she stabbed him, in my view, she would have been more likely to stab him in a potentially fatal area, namely, where she thought the heart to be."

All these passages were read to the jury by the judge in her summing-up. Having compared these passages with the evidence which Prof. Ansford gave, Mr Bruce was, we believe, correct in his submission that those passages broadly speaking represented the reasons which Prof. Ansford gave in his evidence for his opinion. We note that Mr McCoy did not dispute that contention.

19. In short, although Mr McCoy's argument had a superficial attractiveness when first encountered, we do not think that, when the argument is analysed, the judge has been shown to have devoted a disproportionate amount of time to the reasons for Mr Cheung's opinion in comparison to that of Prof. Ansford. The jury could have been left in no doubt as to the grounds on which Prof. Ansford's opinion was based. And to the extent that the jury might have thought that the judge's personal view was to prefer Mr Cheung's opinion to that of Prof. Ansford, the summing-up had not been so unbalanced as to neutralise the effect of the conventional direction which the judge had given to the jury.

20. We note that the judge did not remind the jury why Prof. Ansford thought that Mr Cheung lacked the professional expertise to express an opinion on whether the cuts in the clothing had been inflicted by the deceased. In short, Prof. Ansford's point was that Mr Cheung's experience as a biochemist qualified him to comment on the nature of the cuts in the clothing and on whether they had been caused by a knife, but not on what he described as "the various dynamics of the situation", by which we take him to have meant how and in what circumstances the knife had been wielded. We think that it might have been more appropriate for the judge to have reminded the jury of that part of Prof. Ansford's evidence, but a failure to do so does not, in our opinion, go anywhere near undermining the summing-up as a whole. The fact remains that the judge did remind the jury that Prof. Ansford had questioned Mr Cheung's qualification to express the opinion which he did.

21. Finally, we have not overlooked a point which Mr McCoy made on a particular comment which the judge made to the jury. When she had almost completed her summing-up, the judge adjourned the hearing until the following day. In the jury's absence, Mr Mullick made the point which has become this ground of appeal, namely that the judge had not summarised Prof. Ansford's evidence in sufficient detail. The judge said that she would consider the matter overnight. She decided not to revert to Prof. Ansford's evidence. The following morning, having once again given the jury the conventional direction about rejecting any view which she may appear to have expressed if it did not appeal to them, she said:

"I have dealt with the evidence of some witnesses at greater length or in more detail than the evidence of other witnesses. This is based simply on what reminders I have thought would be helpful to you. It is not based on any view of my own as to the weight of the evidence. And in any event, any view which I may hold of the facts do not matter. What matters is your views of the facts." (Emphasis supplied)

This passage was heavily criticised by Mr McCoy. The judge was in effect saying, said Mr McCoy, that it would not be helpful for her to remind the jury about the detail of Prof. Ansford's evidence. The inference which the jury might have drawn from that was that she thought that Prof. Ansford's evidence was itself unhelpful. We think that this is reading far too much into what the judge said. We think that in this passage the judge was simply telling the jury not to form their own view of the significance of the evidence of a particular witness by reference to the length of time which the judge had taken to summarise it. A direction of that kind is unimpeachable. We do not think that the jury would have got any more out of this passage than that.

22. For these reasons, we do not think that the judge's treatment of the evidence of Prof. Ansford can be seriously criticised, and this ground of appeal must be dismissed. We wish to add two things, though. First, Prof. Ansford had commented on some of the individual cuts in the deceased's clothing, even though it was their overall pattern which he regarded as important. We do not criticise the judge for not reminding the jury of that part of his evidence, but with the advantage of hindsight it may have been helpful to the jury to have been reminded of Prof. Ansford's evidence on those of the individual cuts on which he had commented. Secondly, having summarised the case for the prosecution and the defence in terms which we regard as sufficient, the judge returned to the prosecution case and summarised that again (pp. 50R-55N of the transcript). We do not think that that was necessary, and in an extreme case that could result in an otherwise unimpeachable summing-up losing the balance which it has to maintain. It is not a practice which we would recommend, but looking at the summing-up as a whole, we cannot say that it so failed to hold the balance between prosecution and defence as to undermine its fairness.

Provocation

23. In R. v. Cox [1995] 2 Cr. App. R. 513, the Court of Appeal commented on counsel's duties in a case in which the issue of provocation might be said to arise. At p. 518D-E, Glidewell LJ said:

".... if it appears to counsel, whether for the Crown or the defendant, that there is evidence on which the jury could find provocation, they should regard it as their duty to point it out to the judge, before the judge sums up, and should invite him to consider whether he agrees that the evidence would justify a finding of provocation, and if so, remind him that he in those circumstances is required by the statute to leave the remaining issues to the jury."

In the present case, the issue of provocation was discussed between the judge and counsel. The judge was referred to various passages in what was then the latest edition of Archbold, but neither counsel pointed to any evidence on which the jury could find provocation, and neither counsel suggested to her that the issue of provocation should be left to the jury. In the event, the judge ruled that the issue of provocation should not be left to the jury. Mr McCoy contended that that ruling was wrong, and that the jury should have been directed that, if they were sure that it had been the Appellant who had stabbed the deceased, it was open to them to convict her of manslaughter on the basis that she had been provoked.

24. The role of the court. At first blush, it is not readily apparent why a judge should be criticised for not doing something which she was not urged at the trial to do. As was said in Cox at pp. 516G-517A:

"It may seem to many who regard our system of criminal justice as purely adversarial, strange that when counsel for a defendant has not at trial sought to rely on provocation, the judge's failure to direct the jury to consider that possible defence can sometimes be a valid ground for alleging misdirection."

However, the authorities on the point were all considered in R. v. Cambridge [1994] 1 WLR 971. The court concluded that the issue of provocation should be left to the jury if there was evidence to support it, even if the defence had not sought to rely on it. In reaching that conclusion, the court cited with approval a passage from the opinion of the Privy Council in Bullard v. R. [1957] AC 635, in which Lord Tucker had said at p. 642:

"It has long been settled law that if on the evidence, whether of the prosecution or of the defence, there is any evidence of provocation fit to be left to a jury, and whether or not this issue has been specifically raised at the trial by counsel for the defence and whether or not the accused has said in terms that he was provoked, it is the duty of the judge, after a proper direction, to leave it open to the jury to return a verdict of manslaughter if they are not satisfied beyond reasonable doubt that the killing was unprovoked." (Emphasis supplied)

25. That principle applies not merely to cases in which the defence has not sought to rely on provocation, but also to cases in which the defence expressly disavows provocation and positively does not want the issue of provocation to be left to the jury. In Cambridge, the court had to decide whether the issue of provocation should have been left to the jury when the defendant's case had been that he had not been the assailant. At p. 975C-D, Lord Taylor CJ said:

".... even though the defence may prefer provocation not to be raised, in the fear that it may be a distraction offering the jury a possible compromise verdict, the judge must leave it to the jury if there is evidence."

Similar observations were made in R. v. Dhillon [1997] 2 Cr. App. R. 104. The relevant part of the headnote reads:

"... the judge had realised that provocation could have been pleaded and had sought counsel's help on the matter. Defence counsel's dilemma was that provocation did not sit comfortably alongside the defence of accident. The judge was anxious to be fair, and had done what both experienced counsel had agreed that he should do. Nevertheless, the law was clear and compelled the conclusion that the judge had been wrong not to follow his first inclination that there was sufficient evidence to impose on him the duty of leaving provocation to the jury. His failure to do so was a misdirection ....."

At p. 114B-C, Ward LJ observed:

"The result, making some mockery of our hallowed adversarial procedure which strives to do justice to both sides, is that the appellant is able both to have his cake at trial and also to eat it on appeal."

Finally, the same point was made by the Privy Council, though not in the context of provocation, in Von Starck v. R. [2000] 1 WLR 1270, in which Lord Clyde said at p. 1275G-H:

"For tactical reasons counsel for a defendant may not wish to enlarge upon, or even to mention, a possible conclusion which the jury would be entitled on the evidence to reach, in the fear that what he might see as a compromise conclusion would detract from a more stark choice between a conviction on a serious charge and an acquittal. But if there is evidence to support such a compromise verdict it is the duty of the judge to explain it to the jury and leave the choice to them."

26. So far as we can tell, this line of authority has not been considered in Hong Kong before. We take this opportunity to say that we agree with all these observations. We therefore conclude that we are entitled to consider the correctness of the judge's decision not to give the jury a direction on the issue of provocation notwithstanding the course which the trial took. In defence of the judge, however, we should add that the arguments before us have been very different from those which were advanced to her. In particular, it has been submitted to us that there was evidence on which the jury could have found provocation, whereas neither counsel then appearing before her pointed to any such evidence.

27. The law of provocation. Section 4 of the Homicide Ordinance (Cap. 339) provides:

"Where on a charge of murder there is evidence on which the jury can find that the person charged was provoked (whether by things done or by things said or by both together) to lose his self-control, the question whether the provocation was enough to make a reasonable man do as he did shall be left to be determined by the jury; and in determining that question the jury shall take into account everything both done and said according to the effect which, in their opinion, it would have on a reasonable man."

Section 4 is in identical terms to section 3 of the English Homicide Act 1957. There is no difference between Mr Bruce and Mr McCoy as to the applicable law:

"Section 3 involves two questions: (a) is there any evidence of specific provoking conduct of the accused, and (b) is there any evidence that the provocation caused him to lose his self- control? If both questions are answered in the affirmative, the issue of provocation should be left to the jury notwithstanding the fact that in the opinion of the judge no reasonable jury could conclude on the evidence that a reasonable person would have been provoked to lose his self-control: R. v. Gilbert (1978) 66 Cr. App. R. 237": Archbold 2000, para. 19-53.

Although the judge in her ruling rolled up these two questions, and asked herself whether there was "evidence of a specific act or words of provocation resulting in a loss of self-control", we think that the better course is to address each of the two questions separately.

28. The quality of the evidence for the purpose of answering these two questions was considered in Cambridge. At p. 975D-F, Lord Taylor CJ said:

"But what sort of evidence gives rise to the duty? Clearly, it is not for the judge to conjure up a speculative possibility of a defence which is not relied on and is unrealistic: see Fazal Mohammed v. The State [1990] 2 AC 320, 332. There must be some evidence, but of what strength? In Bullard v. R. [1957] AC 635, the phrase used was: 'any evidence .... fit to be left to a jury.' It is true that in R. v. Camplin [1978] AC 705, 716D, Lord Diplock used the phrase 'however slight', but he used it to describe the measure of the provocative acts or words, not the strength of the evidence that such acts or words in fact occurred and caused the defendant to lose his self-control. Likewise in R. v. Rossiter (1992) 95 Cr. App. R. 326, 332 when Russell LJ referred to 'material capable of amounting to provocation, however tenuous it may be,' the word 'tenuous' described the provocative acts and words not the evidence of their existence."

This theme was picked up by Lord Steyn in R. v. Acott [1997] 2 Cr. App. R. 94. At p. 102D-E, he said:

".... there can only be an issue of provocation to be considered by the jury if the judge considers that there is some evidence of a specific act or words of provocation resulting in a loss of self-control. It does not matter from what source that evidence emerges or whether it is relied on at trial by the defendant or not. If there is such evidence, the judge must leave the issue to the jury. If there is no such evidence, but merely the speculative possibility that there had been an act of provocation, it is wrong for the judge to direct the jury to consider provocation. In such a case there is simply no triable issue of provocation."

29. The first question. We agree with these observations and we therefore turn to consider the two questions which on the authorities the statutory provision poses. We address the first question first: was there any evidence of conduct, whether of words or acts or a combination of both, of a provoking nature, i.e. conduct which was capable of provoking the Appellant to lose her self-control? We have not been provided with a transcript of the Appellant's evidence, and we are left with how the judge summarised the Appellant's evidence in her summing-up. She summarised the Appellant's evidence about her relationship with the deceased as follows:

".... the relationship between her and the deceased would seem to have been a turbulent one at times. And he, again going by her account, was a rather unusual person at times: prone, in addition to other objectionable behaviour, to violence and displays and threats of violence, directed at her, her son and himself too."

30. The Appellant's evidence was that on the evening in question the deceased had been drinking heavily. Having been to a restaurant in Lan Kwai Fong and a bar in Robinson Road, they caught a taxi with a friend of hers. The friend was dropped off, and an argument ensued between the Appellant and the deceased. She wanted to go to Chung Hom Kok (which was one of the places at which she was staying at the time) but he wanted her to go to the flat in Wan Chai. They got out of the taxi and the argument continued in the street. She turned down the deceased's suggestions that they should go to a hotel or to a topless bar. It was then that the attempted theft of her handbag took place.

31. The deceased managed to ward off the robbers with his walking stick, and told the Appellant to run. After a while he caught up with her. The judge's summary of her evidence continued:

"Seeing that he was in fact able to run although he had been using a walking stick, she accused him of lying and said that she would get a taxi to Chung Hom Kok. The deceased then threatened her and also pleaded with her. The threat was to the effect that if she left him, something would happen to her son. As far as pleading was concerned, he was talking about being sick and having problems in Hong Kong. For the sake of her son, she said, she went back to the flat with the deceased. They walked to the flat.

When they arrived at the flat, he pushed her in.... He laughed saying she was not leaving. He hit her ankle with his stick. There was an argument. He threatened to kill her and himself if she was going to leave, saying that he had a knife. He waved his arms about beating himself. She did not see any knife. He said something about her leaving him again; about him dying; about her following him; about her dying in gaol and about her dying in hospital.

He looked strange. She suspected that he might be hiding something. She feared that he would stab her. She suspected that he might have a knife. She tried to see but could not. When he lifted his arms, his arms struck her face and she felt dizzy. She told him that she needed a rest; that he should calm down and that he should get changed. She then went into the bedroom and fell into a light sleep."

What she had said then happened has been summarised earlier in this judgment.

32. There was one other piece of evidence which was relevant to the issue of provocation. A friend of the Appellant and the deceased gave evidence that as the deceased was leaving the restaurant in Lan Kwai Fong he had said that he was going to beat the Appellant once more. The Appellant was following the deceased at the time, but there was no evidence as to how far behind him she was, or whether she had heard what he had said. The witness said that the deceased had said this in such a way that he had been unable to tell whether the deceased was serious or whether he had been joking.

33. All this evidence was direct evidence of the deceased's conduct on the evening in question. The question is whether this conduct on the part of the deceased (assuming that he had behaved in the way alleged) was capable of provoking the Appellant to lose her self-control. In our view, it was. His threats to beat and kill her, against the background of the demands he was making on her, for example, to visit a topless bar, coupled with the blow on her ankle with his walking stick, and his disturbing behaviour overall, amounted in our view to conduct which was capable of provoking the Appellant to lose her self-control.

34. But even if the deceased's conduct on the evening in question was not, by itself, capable of provoking the Appellant to lose her self-control, it became capable of provoking the Appellant to lose her self-control when seen against the turbulent nature of her relationship with the deceased which she had described in her evidence. In this connection, we have in mind the following passage from Archbold 2000, para. 19-58, which was referred to by the judge in her ruling:

"A general approach can .... be discerned from the authorities, namely that evidence of previous provocative acts or past conduct, particularly in cases of domestic violence, is admissible in order to place in its appropriate context the reaction of the accused to the alleged provocation on the occasion of the killing: see R. v. Thornton (No. 2) [1996] 2 Cr. App. R. 108 at p. 118. For the proper approach in relation to a history of provocation leading to a post-traumatic stress syndrome or 'Battered Woman Syndrome', see R. v. Ahluwalia (1993) 96 Cr. App. R. 133 and R. v. Thornton (No. 2) (a jury might more easily find that there was a sudden loss of self-control triggered even by a minor incident if the defendant had endured abuse over a period, on a last straw basis."

In other words, the deceased's conduct on the evening in question, even if unprovocative in itself, was capable, in our view, of constituting the proverbial straw which broke the camel's back and which finally provoked the Appellant into losing her self-control.

35. The second question. We turn, then, to the second question which on the authorities the statutory provision poses: was there any evidence on which the jury could reasonably have concluded that the deceased's conduct on the evening in question in fact caused the Appellant to lose her self-control? This was the question on which Mr Bruce concentrated his submissions, because although he did not accept that the first question should have been answered in the Appellant's favour, he acknowledged that there were grounds for concluding that the deceased's conduct was at least capable of provoking the Appellant to lose her self-control. Mr Bruce strongly argued that there was no evidential basis on which the jury could reasonably have concluded that the deceased's conduct had in fact caused the Appellant to lose her self-control. Any suggestion that the Appellant had lost her self-control had been expressly disavowed by her. As Mr Bruce put it in his outline submissions:

".... if it was true that she was provoked by the deceased to the extent that she was no longer the master of her mind [the phrase used by Devlin J (as he then was) when summing up to the jury in R. v. Duffy and referred to by the Court of Criminal Appeal in [1949] 1 All ER 932] .... as a result of that provocation then surely she would have said so. She said, in effect, the opposite."

36. We cannot go along with this argument. If the Appellant had been relying on the defence of provocation, no doubt she would have said that she had lost her self-control, and it would then have been for the jury to evaluate the truth of that assertion. The only reason why the Appellant did not say that she had lost her self-control was because she was saying that she had not been the assailant. She could not have said that she had lost her self-control without at the same time abandoning her main defence to the charge. The fact that she did not suggest to the jury that she had lost her self-control does not, of itself, mean that the second question could not have been answered in the Appellant's favour. As was said in Bullard in the passage to which we have already referred, if there is evidence to support provocation, the issue of provocation must be left to the jury "whether or not the accused has said in terms that he was provoked".

37. The fact is that evidence that the deceased's conduct caused the Appellant to lose her self-control did not have to come from the Appellant. It could have come from other sources. And the evidence did not have to be direct. It would have been sufficient if there had been evidence of primary facts from which the loss of self-control on the part of the Appellant as a result of the deceased's conduct could have been inferred. In our view, although there was no direct evidence that the Appellant had lost her self-control, we think that this was one of those cases from which a jury could reasonably infer such loss of self-control from the nature of the deceased's injuries. We do not think that they can be described as frenzied, but the seven attempts which had been made to stab the deceased were such as to render a loss of self-control on the part of the Appellant at least a possibility, given the nature of their relationship and the domestic setting in which the attack on the deceased occurred. Moreover, the deceased's conduct immediately before the attack on him was such that a jury might reasonably have felt able to infer that if the Appellant had lost her self-control, the loss of self-control was caused by that conduct.

38. In reaching this conclusion, we have not overlooked the decision of the Court of Final Appeal in Zeng Liang Xin v. HKSAR (1997-98) 1 HKCFAR 12. In that case, the court held that the issue of provocation should not have been left to the jury. However, that was a conclusion on the second question reached on the particular facts of the case. The court concluded that there had been no evidence on which the jury could reasonably have concluded that the deceased's conduct had caused the defendant to lose his self-control.

39. Conclusion. It follows that, in our view, the issue of provocation should have been left to the jury, and the failure to do so constituted a material irregularity in the trial. That implies no criticism of the trial judge. After all, she had to address the question (a) without the benefit of any argument that the issue had been raised on the evidence, and (b) without the benefit of those cases having been cited to her which showed that she had to consider the issue of provocation even if counsel did not wish to rely upon it. We refrain from commenting on whether this was a strong or weak case of provocation, because for the reasons to which we shall come it may be that the issue will have to be considered by another jury in due course.

40. Finally, we were at one time concerned that if we ruled that the issue of provocation should have been left to the jury in this case, the issue would have to be left to the jury in many cases in which such a course would not usually be regarded as appropriate. However, that conclusion is not one which troubles us. This case was extremely unusual in that, although provocation was not being relied upon by the defence, the evidence which would have had to be adduced if it had been relied on was in fact adduced - presumably for the purpose of explaining why it would not have been so unlikely for the deceased to have inflicted the injuries upon himself.

The proviso

41. Mr Bruce submitted that, in the event of our concluding (as we have done) that the issue of provocation should have been left to the jury, we should nevertheless apply the proviso to section 83(1) of the Criminal Procedure Ordinance (Cap. 221), which provides that:

"..... the Court of Appeal may, notwithstanding that it is of opinion that the point raised in the appeal might be decided in favour of the appellant, dismiss the appeal if it considers that no miscarriage of justice has actually occurred."

Mr McCoy contended that we had no power to do so. He pointed to what had been said in Bullard at p. 644:

"Every man on trial for murder has the right to have the issue of manslaughter left to the jury if there is any evidence upon which such a verdict can be given. To deprive him of this right must of necessity constitute a great miscarriage of justice and it is idle to speculate what verdict the jury would have reached."

Mr McCoy argued that since the question whether the proviso should be applied involves a consideration of what verdict the jury would have reached if the issue of provocation had been left to them, the fact that it is idle to speculate about that means that the proviso can never be applied in such a case.

42. The question as to whether the proviso to section 2(1) of the Criminal Appeal Act 1968, which is the equivalent in England of section 83(1), can be applied where there had been a failure to leave the issue of provocation to the jury was considered in Cox. Glidewell LJ referred to what had been said in R. v. Whitfield (1976) 63 Cr. App. R. 39, in which the Court of Appeal had held that the issue of provocation should have been left to the jury. At p. 43, Lord Widgery CJ had said:

"We are impressed by the fact that Parliament in the [Homicide Act 1957] has taken an unusual step of deliberately insisting that a particular issue shall be tried by the jury and no one else. In other words the reaction of the reasonable man must be assessed by the jury because the section says so. In this case no jury had an opportunity of expressing a view on the point at the Court of trial. There is no jury in this Court. If we apply the proviso we are in fact determining that issue otherwise than by the verdict of the jury. Although we are not prepared to go to the length of saying that the proviso is never appropriate in this kind of case, we do not feel that it would be right to apply it in this case."

In Cox, Glidewell LJ noted that in that last sentence it had not been said that as a matter of law the proviso could not apply where there had been a failure to leave the issue of provocation to the jury. Indeed, the court dealt with the point about section 3 of the Homicide Act referring expressly to the jury by saying that the section was dealing with the trial, not with the appeal. We agree. Having read section 3 of the Homicide Act with care, we do not think that the reference to the jury is anything more than a recognition of the fact that the tribunal of fact in cases of murder is the jury. Where on appeal, for the purpose of the application of the proviso, the appellate court has to decide whether the jury would inevitably have convicted the defendant if the material irregularity in the trial had not occurred, the appellate court is always putting itself in the position of the tribunal of fact. The mere fact that section 3 of the Homicide Act (and section 4 of the Homicide Ordinance) refers to the jury expressly does not affect that.

43. We therefore conclude that it is open to the Court of Appeal, if the evidence justifies it, to apply the proviso even where the issue of provocation should have been left to the jury but was not. We think, though, that the circumstances in which the proviso can be applied will be rare indeed. That is because it will only in a rare case be possible for the court to conclude that the jury would inevitably have convicted the defendant of murder even if the issue of provocation had been properly left to them. The considerations which caused the court to conclude that the issue of provocation should have been left to the jury will usually mean that the court cannot be confident about what the ultimate outcome would have been. We note that in Cox the court did apply the proviso, and although we understand perfectly the reasons for that in that case (see Glidewell LJ's observations at pp. 521E-522A), we regard that case as exceptional. For our part, we decline to apply the proviso in the present case, because we cannot say that the jury would inevitably have convicted the Appellant of murder if the issue of provocation had been left to them.

Conclusion

44. For these reasons, this appeal must be allowed, and the Appellant's conviction for murder must be quashed. All that remains is to decide what should happen now. Should we order a re-trial, or should we substitute for the Appellant's conviction for murder a conviction for manslaughter on the ground of provocation on the basis that the Appellant is entitled to the benefit of a verdict for the lesser offence? The position of both Mr Bruce and Mr McCoy was that, if we allowed the appeal only on the basis that the issue of provocation should have been left to the jury, we should defer further consideration of the issue as to what should then happen to a further hearing at which any communications between prosecution and defence during the trial which might be relevant to the exercise of our discretion would be brought to our attention and at which submissions specifically directed to this issue would be addressed to us. Accordingly, we direct that this appeal be re-listed for hearing on a date to be fixed after consultation with counsel's diaries for that issue to be addressed. The Appellant will remain in custody in the meantime.

45. However, it may be helpful if we express our provisional thoughts on the matter. The usual order would be for a re-trial. After all, substituting a verdict of manslaughter would deny the prosecution the opportunity of obtaining a finding from the jury that such provocation as there might have been was not sufficient to justify a verdict of manslaughter. We recognise that ordering a re-trial would enable the Appellant to have a second opportunity to run the defence that she had not been the assailant at all, when no valid criticism of the trial on that issue has been established. But that is the price which would have to be paid for both sides not pointing out to the judge the evidence which raised the issue of provocation.

46. We cannot stress too much, though, that these observations are tentative, and apart from anything else they are subject to whatever may have passed between the prosecution and the defence at the trial relating, for example, to the possible acceptance by the prosecution of any plea of guilty to manslaughter. In any event, it should be noted that, even if a re-trial is ordered, it would still be open to the prosecution to accept a plea of guilty to manslaughter if one was tendered and if the prosecution thought that such a course was appropriate.