The Judicial Committee of the Privy Council Decisions

John v. The State of Trinidad & Tobago (Trinidad & Tobago) [2009] UKPC 12 (16 March 2009)

Privy Council Appeal No 66 of 2007

Ronald John Appellant

v.

The State of Trinidad and Tobago Respondent

FROM

THE COURT OF APPEAL OF THE REPUBLIC OF

TRINIDAD AND TOBAGO

- - - - - - - - - - - - - - - - -

JUDGMENT OF THE LORDS OF THE JUDICIAL

COMMITTEE OF THE PRIVY COUNCIL

Delivered the 16^th March 2009

- - - - - - - - - - - - - - - - -

Present at the hearing:-

Lord Hoffmann

Lord Rodger of Earlsferry

Baroness Hale of Richmond

Lord Brown of Eaton-Under-Heywood

Sir Jonathan Parker

- - - - - - - - - - - - - - - -

[Majority Judgment Delivered by Lord Brown of Eaton-Under-Heywood]

1. In the early afternoon of Wednesday, 27 November 2002, in the course of a robbery at the Starlite Recreation Club in Palmyra Village, San Fernando, the club's proprietor, Kenneth Boxie, was shot dead. On 8 February 2006, following a three-week trial before Mohammed J and a jury at the San Fernando Assizes, the appellant, Ronald John, was convicted of Boxie's capital murder and, as required by the law of Trinidad and Tobago, sentenced to death. On 28 March 2007, his appeal against conviction was dismissed by the Court of Appeal (Hamel-Smith, Warner and Kangaloo JJA). This further appeal is brought pursuant to special leave granted by the Board on 6 December 2007.
2. The only evidence against the appellant was that given by an accomplice, Jeffrey Lewis, a taxi driver, whose basic account to the jury was that on the occasion of the killing he had driven the appellant and two other (unidentified) men to various locations in the south including the Starlite Club, having understood that the men were just going "to mark a scene" (carry out a reconnaissance), that at the Club the appellant got out of the car with a gun in his hand, that he himself remained in the car under threat during the course of the robbery, that he had then driven the men back to Sea Lots where he had first picked them up, and that he had been in a position to describe the appellant to the police (sufficiently to enable the police to arrest the appellant on 6 December 2002) because for some months previously, once or twice a week whilst on his taxi run, he had used to see the appellant liming (hanging around) on Queen and Nelson Streets, Port of Spain, on one occasion for two or three hours, and so was able to recognise him although he did not know him personally. He had also, of course, seen the appellant on the drive south to the Club on 27 November when, he told the jury, the appellant had hired him and had been the front seat passenger affording him ample opportunity for observation; he said he would have looked at the appellant's face for some 20-25 minutes in all that day.
3. Indisputably Lewis had driven the robbers to the Club: although efforts had been made to obscure the hire car's registration plates, its number had in fact been noted and had led the police on 3 December 2002 to the vehicle's owner who told them that Lewis had been driving it on 27 November. Lewis himself was arrested on 3 December and twice interviewed by the police that day before taking them on the route he had taken with the robbers on 27 November. On 4 December Lewis made the first of two written statements to the police under caution. On 6 December Lewis was again seen by the police and again accompanied them on the route taken on 27 November. On 6 December, as stated, the appellant was arrested, solely on the basis of what Lewis had already by then told the police. On 7 December Lewis made his second (and, so far as is known, final) statement to the police. On 4 December the police had also taken a statement from Lewis's common law wife, Sophia Phillips (who was present when both of Lewis's statements were taken).
4. There can be no doubt that Lewis's story to the police had been in certain respects an evolving one, in particular in that he initially sought to deny completely (to Ms Phillips as well as to the police) and then to minimize his own involvement in the robbery, particularly as to the precise circumstances in which the robbers came to hire his cab and for what purpose. For example, his first recorded—recorded only in the sense of being based upon notes taken at the time—interview (the second interview on 3 December) begins by Lewis saying "sorry for not telling yuh the truth", and even in his statement of 7 December Lewis says that "the statement that I did give [on 4 December] was not totally correct". It was not until 7 December that Lewis admitted knowing that the journey south was to "mark a scene" rather than merely to collect spare parts, as he had earlier said, to repair the appellant's own broken down vehicle. So far as his description of the actual gunman was concerned, however, the only change in Lewis's account was that, whereas initially he had given the impression of not knowing him at all, in his statement on 7 December he said: "I know them fellas by Nelson Street, they do thief and thing, so I ask him if he have anything [i.e. a weapon] on him".
5. By the time of the appellant's arrest on 6 December the recorded information about the gunman given by Lewis consisted of the descriptions "5' 9" tall, brown skin negro, slim build" (interview 3 December) and "I do not know him but he had on a blue and white cap, a blue and white striped jersey, three quarter pants. I can't remember the exact colour. He was a slim, tall, brown-skinned fella with a gold tooth in the top jaw, of African descent with a kinda longish face. He had a small scar on the right side of his face somewhere on his cheek. He had small eyes and was clean shaven" (statement of 4 December, a description not inconsistent with such as were given by those who had witnessed the robbery at the Starlite Club); plus the fact that he was known to the other robbers as "Dollars" (statement of 4 December), and that he lived at Sea Lots (an area near the coast in Port of Spain) where Lewis had picked him up and later returned him on the day of the robbery. The police at trial said that Lewis had mentioned not only Sea Lots but Pioneer Drive in Sea Lots (the address at which the appellant was arrested) and, of course, not everything said by Lewis to the police had been recorded in writing, for example whilst Lewis had twice retraced for them his journey of 27 November. Be that as it may, the police plainly had no doubt that they had correctly identified the person whom Lewis was describing as the gunman (and, on 7 December, was saying that he had in any event been able to recognise from his past knowledge of him on the street). Furthermore, the appellant was already known to the police as "a local villain" (a fair characterisation, Mr Birnbaum QC concedes). On arrest, it is right to say, the appellant denied all involvement in the robbery.
6. So sure were the police that they had arrested the right man that he was never put up on an identification parade. There was no reason to suppose that any of the few other witnesses of this robbery would have been able to identify him and, certainly by 7 December, the police regarded Lewis's identification essentially as recognition rather than observation evidence. On 9 December the appellant was charged with Boxie's murder. The same day, Lewis was released from police custody.
7. In the event, it was not until 19 February 2004 (over a year after the appellant's arrest and during the preliminary enquiry into the murder charge against him, indeed some four months after that had begun) that Lewis for the first time since the robbery came face to face with him. Without objection from the defence, Lewis's evidence on that occasion was as follows: "I knew the person who flagged me down a few months before by seeing him. I know him by seeing him on Nelson Street. I used to see this person probably once or twice a week. I see this person I speak of in court today. If I see him again I would be able to point him out (witness points to accused)."
8. Earlier that very day (19 February) Lewis had been served with a written undertaking signed the previous day by the DPP, in effect undertaking not to prosecute him providing that he made full and truthful witness statements and gave evidence in accordance with them when required to do so in relation to the matters giving rise to the appellant's indictment for murder.
9. As stated, the appellant's trial before the jury lasted three weeks. He was most ably defended. Lewis, the prosecution's all-important witness, was strenuously cross-examined. Essentially it was put to him that he was lying, rather than mistaken, in stating that the accused was the gunman, lying to save his own skin. His evidence was that he was well able to recognise the appellant. He said that the immunity "gave me a chance to speak the truth." When it was put to him that he identified the appellant in the dock because: "No matter who you saw as the accused person in the Magistrates Court or this Court – because the State wanted you to involve Ronald John. If not, you would have [been] charged with murder [under the felony murder rule]", Lewis replied: "All the State wanted me to do is to speak the truth and I did so." The judge having rejected a submission of no case to answer at the close of the prosecution case, the appellant chose not to give evidence. Ms Phillips was called on his behalf but really could say nothing of help. The judge's summing up to the jury took about a day and a half. Although criticising it as "overlong", Mr Birnbaum acknowledges that the judge "clearly strove to be fair" and took meticulous care in dealing with the evidence.
10. For their part their Lordships regard the summing up as a model of fairness and clarity. More than once the judge emphasised that "the State's case stands or falls on the evidence of Jeffrey Lewis. Everything therefore depends on what you make of him as a witness. If at the end of the day you say that Jeffrey Lewis is not speaking the truth then you go no further; you will have to find the accused not guilty . . . if you say that he may be speaking the truth but you are not sure, then you will still have to find the accused not guilty. You are required to look at his evidence in the most rigorous fashion, putting it through a fine tooth comb. You must approach his evidence, because of his status as an accomplice and because he has been granted an immunity from prosecution, with a great amount of special care and special caution. There are weaknesses and potential weaknesses in his evidence, depending on how you view them; they may have an effect on his credibility and reliability" (excerpts variously from pp127, 153, 164-165 of the record). The judge rehearsed all the evidence that could be said to cast doubt on various aspects of Lewis's evidence, including, insofar as these were inconsistent, his earlier statements. He catalogued "all the factors that have the potential, depending entirely on how you view them, to undermine the credibility of Jeffrey Lewis. All of these factors are equally and just as importantly relevant to your assessment of whether Jeffrey Lewis is a reliable witness in terms of the accuracy and correctness of his identification of the person he says is the accused" (record p162). He emphasised, again more than once, the relevance and importance of Lewis not only being an accomplice, but also having received an immunity from prosecution. For example he said:

"An accomplice, by his very status, is a suspect witness and may have a motivation to give false evidence, and that is the general reason why I have warned you that the evidence of Jeffrey Lewis as an accomplice, which is the way the State has approached him, must be treated and approached with a great deal of special care and special caution. But in addition to that, an accomplice witness who receives an immunity from prosecution avoids prosecution altogether once he abides by the terms of the agreement. . . . If an accomplice starts off with a false account or a partly false account, all that an immunity may do is to have the effect in the accomplice's mind of tying him down to those initial accounts so that you must bear this possible danger in mind." (record pp 149-150)

He fully rehearsed all the evidence going to the question of Lewis's prior knowledge of the appellant and told the jury in terms that a dock identification (if that is how Lewis's evidence implicating the appellant was to be regarded) "would be worthless".

11. Before closing his summing up, the judge invited the jury to withdraw and asked defence counsel whether there was anything he would like added. He duly then included express reference to the various points counsel made. Shortly afterwards the jury retired and, two and a quarter hours later, returned a unanimous verdict of guilty.
12. Different counsel was instructed on the appeal when a single ground of appeal was relied upon: the judge, it was said, had wrongly permitted prejudicial evidence of bad character to go before the jury, that evidence being what Lewis had said in his statement of 7 December as to "them fellas by Nelson Street . . . do thief and thing", the reason why he had asked if the appellant was armed. The point was a hopeless one and was rightly rejected. It is not renewed before the Board and no more need be said about it.
13. Now advanced before the Board, however, is a series of completely fresh points, several of which are no less hopeless and to most of which their Lordships need not refer. The one point which does, however, deserve the fullest consideration is Mr Birnbaum's submission that an identification parade should have been held in this case, that the absence of one caused the appellant an injustice, and, a linked argument, that the judge erred in not having warned the jury that the failure to hold a parade constituted a substantial weakness in the prosecution's case (although this was not something which, at the end of the summing up, counsel had asked the judge to deal with).
14. As a basic rule, an identification parade should be held whenever it would serve a useful purpose. This principle was initially stated by Hobhouse LJ in R v Popat [1998] 2 Cr App R208, BAILII: [1998] EWCA Crim 1035, 215 and endorsed by Lord Hoffmann giving the judgment of the Board in Goldson & McGlashan v R (2000) 56 WIR 444, [2000] UKPC 9 Plainly an identification parade serves a useful purpose whenever the police have a suspect in custody and a witness who, with no previous knowledge of the suspect, saw him commit the crime (or saw him in circumstances relevant to the likelihood of his having done so, for example en route to a robbery). Often, indeed usually, that is the position and, when it is, an identification parade is not merely useful but, assuming it is practicable to hold one, well-nigh imperative before the witness could properly give identifying evidence. In such a case, Lord Hoffmann said in Goldson, "a dock identification is unsatisfactory and ought not to be allowed," although he added: "Unless the witness had provided the police with a complete identification by name or description, so as to enable the police to take the accused into custody, the previous identification should take the form of an identification parade."
15. At the opposite extreme lies a case where the suspect and the witness are well known to each other and neither of them disputes this. It may be, of course, that on the critical occasion when the witness saw the crime being committed (or, for example, the person concerned en route), he thought it was the person he knew but was mistaken as to this. An identification parade obviously cannot help in this situation. Indeed, as Lord Hoffmann pointed out in Goldson, a parade then would be not merely unnecessary but could be "positively misleading":

"The witness will naturally pick out the person whom he knows and whom he believes that he saw commit the crime. In fact, the evidence of the parade might mislead the jury into thinking that it somehow confirmed the identification, whereas all that it would confirm was the undisputed fact that the witness knew the accused. It would not in any way lessen the danger that the witness might have been mistaken in thinking that the accused was the person who committed the crime.

16. A third situation arises when the witness claims to know the suspect but the suspect denies this. This indeed was the situation in Goldson itself, certainly so far as one of the two accused was concerned. The witness, Claudette Bernard, herself shot in the face by one of the gunmen (who then shot dead her boyfriend lying next to her), subsequently identified them simply as men known to her by their street names. One of the two accepted that she knew him and the question in his case was simply whether she had recognised him on the occasion of the shooting (essentially, therefore, the second of the situations considered above); the other, however, whom she said she had seen two or three times a week on the street for three years but had spoken to only once and who had a girlfriend called Ginger, disputed that she knew him at all, said that he had no such girlfriend, and gave evidence to that effect.
17. The advantage of holding an identification parade in such circumstances was, as counsel pointed out:

"If Claudette had failed to pick out the accused on the parade, her assertion that the accused were known to her would have been shown to be false. By not holding identification parades, the police had denied the accused an opportunity to demonstrate conclusively that she was not telling the truth."

18. The Board referred to two English cases where there was "a dispute over whether the accused was in fact a person known, or sufficiently known, to the witness" and where the convictions had been set aside as unsafe because, in the absence of a parade, the evidence of identification (in each case by way of dock identification) was regarded as too weak to support the conviction. In R v Conway (1990) 91 Cr App R 143 the witness said that she knew the accused, had seen him in a public house and entertained him to dinner, but did not know his name, where he lived, or anything of importance about him. No identification parade had been held despite the accused having denied that the witness knew him and having expressly requested a parade. In R v Fergus [1992] Crim. L.R. 363, where the witness had claimed only to have seen the accused once and to have heard his name from someone else, the Court observed:

"The case where the complainant had seen the assailant only once or on a few occasions before might well be treated as that of identification rather than recognition".

19. The Board in Goldson, having concluded that it would have been good practice for the police there to have held an identification parade, had then to address the question whether the failure to do so had in fact caused a serious miscarriage of justice as in Conway and Fergus. Concluding not, the Board contrasted Fergus, where "the claimed previous knowledge was very slight indeed", and Conway, where the accused had expressly requested an identification parade and been refused one, with Goldson where neither accused had requested a parade and where there had been no objection to the dock identification at the preliminary inquiry. The Board then continued:

"The position is therefore that although one may speculate about the possibility that a parade would have destroyed the prosecution's case . . . it is not possible to say that the absence of a parade made the trial unfair. The judge was entitled to leave the question of credibility to the jury on the evidence before them. And once she was accepted as a credible witness, no criticism was or could be made of the judge's directions that the jury were to be careful about accepting her evidence that they were the gunmen.

[Counsel] submitted that the judge should have given the jury a specific direction about the absence of an identification parade and the dangers of a dock identification. But their Lordships consider that in the present case such directions were unnecessary. The judge told the jury that they should first consider whether Claudette Bernard was a credible witness. If they thought she was lying, the accused had to be acquitted. This appears to their Lordships to be sufficient, because if she was not lying, it would follow that there had been no need for an identification parade and the dock identification would have been the purely formal confirmation that the men she knew were the men in the dock."

20. The Board has had occasion to deal with failures to hold identification parades in a number of subsequent cases. Amongst them are Aurelio Pop v The Queen [2003] UKPC 40 and Pipersburgh and Robateau v The Queen [2008] UKPC 11, each an appeal from the Court of Appeal of Belize, both resulting in the quashing of the appellant's convictions, and in both of which Lord Rodger of Earlsferry delivered the judgment of the Board. It is unnecessary to rehearse here the detailed facts of either case. Both, however, in their different ways involved unsatisfactory recognition evidence and dock identifications only. In Pop, the witness Adolphus who identified the accused as the gunman, only made the link between the man he knew simply as R and the accused as the result of an improper leading question by prosecuting counsel (see paras 7 and 10 of the judgment). That, coupled with the failure to hold an identification parade which should have been held under Belize law (see para 9 of the judgment) required that the judge should have "warn[ed] the jury of the dangers of identification without a parade and should have explained to them the potential advantage of an inconclusive parade to a defendant such as the appellant. For these reasons, he should have explained, this kind of evidence was undesirable in principle and the jury would require to approach it with great care" (para 9) and he should have "pointed out to the jury that [because of counsel's leading question] they required to take even greater care in assessing Adolphus's evidence that it was the appellant who had shot the deceased" (para 10).
21. In Pipersburg (an appeal heard, the Board regretfully recorded, with the DPP unrepresented) no identification parade had been held because the suspects' pictures had been published in the press and it was feared that they would be identified from these—an inadequate justification for dock identifications over 18 months later. It is sufficient for present purposes to cite paragraph 17 of the Board's judgment:

"In the present case, it may well be that the judge bemoaned the fact that no identification parade had been held and pointed out the advantages of such a parade. But, despite what the Board had said in Pop, he did not point out that Mr Robateau had thereby lost the potential advantage of an inconclusive parade. Moreover, while giving directions on the care that needs to be taken with identification evidence in general, the judge did not warn the jury of the distinct and positive dangers of a dock identification without a previous identification parade. In particular, he did not draw their attention to the risk that the witnesses might have been influenced to make their identifications by seeing the appellants in the dock. And, perhaps most importantly, even if the judge's directions would have ensured that the jury appreciated that this type of identification evidence was undesirable in principle, he did not explain that they would require to approach that evidence with great care. On the contrary, the closing words of the direction really left the whole matter to the jury on the basis that the witnesses said that they knew the men and it was simply up to the jury to accept or reject their evidence."

22. Pop and Pipersburg are really the high watermark of the appellant's case. Mr Birnbaum submits that through the failure to hold an identification parade here, this appellant too "lost the potential advantage of an inconclusive parade". In the context of the present case, however, there could only have been "an inconclusive parade" if Lewis was actually lying—as, of course, the appellant said he was—in claiming to know him. If he did know him in the sense of recognising him from the streets or even, indeed, merely from having driven him down south and back again on the occasion of the robbery, he could hardly have failed to pick him out on the parade.
23. Only if Lewis had not in fact clearly known what the appellant looked like could he have failed to pick him out on a parade. Lewis cannot simply have invented a description of the gunman (and given him the street name Dollars and Sea Lots as his whereabouts) with no particular person he knew in mind, a description which then led the police by sheer chance to identify and arrest the appellant. Plainly, therefore, Lewis was attempting to describe someone he knew. To protect the actual gunman, he might, of course, have described to the police someone else he knew, perhaps, indeed, the appellant from seeing him around Nelson Street as he described. But how, in that case, would he have known his street name Dollars? And more to the point in the present context, how would that bald lie have been exposed by an identification parade? How, then, had an identification parade been held, could Lewis possibly have failed to pick out the appellant as the man he had described to the police? Only, surely, if the police, acting upon Lewis's genuine description of the gunman, had then in fact arrested the wrong man—wholly coincidentally a man who broadly fitted Lewis's physical description, a man known as Dollars (not, of course, a unique street name but hardly a common one), and a man who happened to live at Sea Lots. This possibility seems to their Lordships decidedly far-fetched. But assume that it occurred. It would necessarily then follow that the person whom Lewis saw in the dock at the preliminary enquiry over a year later, he would have been seeing for the very first time. He would have been expecting to see the actual gunman in the dock but would in fact be seeing someone different and someone, therefore, he knew perfectly well to be innocent. Is it really to be supposed that in those circumstances Lewis, just because that morning he had been given immunity from prosecution conditional upon his cooperating with the police in the matter of the appellant's murder trial, would blithely then knowingly identify and give lengthy evidence calculated to convict an innocent man? Again this appears to their Lordships a somewhat fanciful possibility.
24. Certainly the DPP's undertaking could not be construed as requiring him to give false evidence. As he himself said (see para 9 above): "All the State wanted me to do is to speak the truth" and the immunity "gave me a chance to [do so]". It may well be, indeed, that, without the indemnity in place, Lewis would have declined to identify the appellant on a parade even had one been held.
25. It therefore seems to their Lordships that, realistically, on a true analysis of the evidence, an identification parade in this case would have served less purpose not only than in either Pop or Pipersburg but also than in Goldson itself. Mr Birnbaum seeks to distinguish Goldson on the basis that whereas the identifying witness there was a victim, here it was an accomplice, and he submits that accordingly here, not only should there have been a parade (as the Board thought in Goldson) but (in contrast to the holding in Goldson) that the lack of one here caused justice to miscarry. The argument is a difficult one. In the first place, unlike the position in Goldson, this was a case where the witness provided the police with sufficient particulars of identification "to enable the police to take the accused into custody". But, perhaps more importantly, the very fact that Lewis was an accomplice meant that, assuming always he was telling the truth, he was altogether better placed to know who the killer was than Claudette (the first to be shot and seeing the gunmen for moments only) had been in Goldson. As the State suggested to the jury, "there must have been some element of familiarity between Jeffrey Lewis and the person said to be the accused, because a person would not reasonably go from North to South to mark a scene with a complete and total stranger," and as the judge observed: it was open to the jury to find Lewis's evidence "the best evidence available since it comes from a person who might be best positioned to know what allegedly transpired as the alleged driver of the car." In short, the only purpose of an identification parade here would have been to guard against the possibility (a) that the police might have arrested the wrong man (someone completely unknown to Lewis) but that nevertheless (b) Lewis might falsely identify him in the dock when seeing him for the very first time in the belief that the indemnity required him to do so—both possibilities which their Lordships have already indicated appear somewhat fanciful.
26. All that said, the Board nevertheless concludes that the police here should have held an identification parade, this being a case of capital murder. There was on the face of it nothing to lose by holding a parade (although, as already suggested, it might have required there to be an indemnity already in place). Just conceivably it might have produced another identifying witness (from the Club), not that that would have advantaged the appellant. And it would have eliminated the risk (however small) of a lying dock identification on the unlikely hypotheses indicated above. Finally, the Board would note, Mr Guthrie QC for the State, whilst pointing out that no identification parade was requested by the accused (as it had been in Conway), was disposed to "accept that it might have been desirable".
27. It by no means follows, however, that the failure to hold a parade here can be regarded as having caused a miscarriage of justice. Rather, for all the reasons already rehearsed, their Lordships find themselves quite unable to reach that conclusion and on the contrary regard the position here as a fortiori to that arrived at in Goldson. Similarly their Lordships do not see this case as comparable to Pop or Pipersburg. Realistically, there was not the same possibility of mistaken recognition in this case as in each of those and the summing up in this case was altogether fuller, fairer and more favourable to the accused than in either of those.
28. That, in their Lordships' view, effectively concludes this appeal. The judge in summing up noted the defence submission that the police "could have placed the accused on an identification parade and called Jeffrey Lewis, or . . . the witnesses from the bar". True, he did not himself adopt that implicit criticism of the police's failure to hold a parade or warn the jury that it weakened the prosecution's case. Since, however, for the reasons already given, realistically it did no such thing (as perhaps is reflected in counsel's decision not to include this amongst the points he asked the judge to add at the conclusion of his summing up), their Lordships cannot regard this as a substantial defect in the summing up, still less as one causing any miscarriage of justice.
29. Nor is the appellant's criticism of the dock identification sustainable on the facts here. As both Pop and Pipersburg make plain, dock identifications as such are not inadmissible and, again for the reasons already given, the dock identifications here, not objected to either at the preliminary inquiry or at the trial, occasioned no miscarriage of justice.
30. Realistically, the criticism of the DPP for granting Lewis an indemnity here is not a separate point either. Although the practice of allowing an accomplice immunity in return for giving evidence for the prosecution has long been recognised both as "distasteful" and as creating an obvious risk of the accomplice giving false evidence, plainly there are occasions when nevertheless an indemnity can properly be given—see, for example, Turner (Bryan) (1975) 61 Cr APP R 67, Chan Wai-Keung v Regina [1995] 2 Cr App R 194 and R v Smith [2003] EWCA Crim 3847. (In England and Wales, such arrangements have now been placed on a statutory footing under Chapter 2 of the Serious Organised Crime and Police Act 2005.)
31. Obviously great care must be taken in these cases and judges must give the jury appropriately strong directions about the risks. As already indicated, however, the judge's directions in the present case were both full and impeccable. It may, of course, be that Lewis continued throughout to understate his own part in this robbery and possibly, indeed, he could have identified the other two robbers as well as the appellant (who, as the actual killer, the police were above all concerned to arrest). It would not follow, however, that his evidence identifying the appellant as the gunman—ultimately, the one critical question for the jury—was in any way suspect. On the contrary, the more involved Lewis personally was and the better he knew the robbers, the plainer it was that he could recognise the appellant.
32. Mr Birnbaum's contention that the judge should have acceded to the defence submission of no case to answer is an impossible one. It was open to the jury to accept Lewis as an honest and reliable witness in so far as he identified the appellant as the gunman and, provided only that they did so, they had no alternative but to return a verdict of guilty. The judge could not properly have withdrawn the case from them.
33. In the result, the Board has reached the clear conclusion that none of the appellant's newly advanced grounds of appeal are sustainable. Their Lordships' attention was drawn to the United Nations' Secretary General's note to the General Assembly dated 7 October 1996, annexing the Report of the Special Rapporteur of the Commission on Human Rights on extrajudicial, summary or arbitrary executions which, at para 110, includes this:

"The Special Rapporteur wishes to reiterate that proceedings leading to the imposition of capital punishment must confirm to the highest standards of independence, competence, objectivity and impartiality of judges and juries, as found in the pertinent international legal instruments. All defendants facing the imposition of capital punishment must benefit from the services of a competent defence counsel at every stage of the proceedings. Defendants must be presumed innocent until their guilt has been proved beyond a reasonable doubt, in strict application of the highest standards for the gathering and assessment of evidence."

Mr Birnbaum submits that the failure to hold an identification parade here meant that the highest standards for the gathering of evidence were not met. The Board has already indicated at length why it concludes that the appellant was not in fact disadvantaged by the lack of a parade in this case. Only too well aware as they are that this is a capital case, their Lordships cannot in all conscience allow this appeal. Rather they must dismiss it.

Concurring Judgment by Lord Hoffmann

34. As the advice of the Board is not unanimous, I propose (unusually) to say briefly why I join in the opinion of Lord Brown of Eaton-under-Heywood.
35. There are three points upon which the Board is agreed. The first is that it would have been better if there had been an identification parade. The second is that the absence of a parade was not a ground for withdrawing the case from the jury. That is important, because if it was fatal to the State's case that the appellant was not given the opportunity to demonstrate, by a failed identity parade, that Lewis was lying or mistaken, then logically the case should have been stopped. The third is that the judge gave the jury careful directions about assessing the accomplice's credibility and the possibility that even a credible witness might make a mistaken identification.
36. The issue, then, is whether the judge should have invited the jury to cry over spilt milk and told them what they would have been able to infer, on various hypotheses, if there had been evidence of an identification parade. That would not have been without complication, because on the most probable assumption, namely, that the jury believed that Lewis knew the appellant, perhaps better than he was willing to admit, the answer would have been: nothing at all. In my opinion it was not necessary for the judge to tell the jury how it might have assisted their task if the evidence had been different. His duty was to instruct them how to approach the evidence they had actually heard.

Dissenting Judgment by Baroness Hale of Richmond

37. It is accepted that the death penalty may be imposed and executed for the crime of murder consistently with the Constitution of Trinidad and Tobago: see Matthew v State of Trinidad and Tobago [2005] 1 AC 433, where the debate concerned the mandatory imposition of the death penalty in all cases of murder irrespective of the circumstances. It is accepted that our task as the supreme court for Trinidad and Tobago is to uphold the laws of Trinidad and Tobago whether we like them or not. But it is also accepted that the pre-condition for imposing the ultimate penalty for the ultimate crime is that the investigation, prosecution and trial of an alleged murderer have been conducted with such fairness and propriety that there can be no real possibility that a mistake has been made and the wrong person convicted.
38. The jury, as all judges remind them when summing up the case, are the sole judges of fact, of whom to believe and whom not to believe. It is not our job to substitute our view of the evidence for theirs. But experience has shown that deciding whether a witness is telling the truth, or telling a lie, or simply mistaken in his observation or recollection, is not an easy task. Even twelve careful and conscientious members of a jury can get it wrong. That is why all sorts of rules and practices have been developed in an effort to avoid some of the most obvious pitfalls for those who have to find the facts.
39. Amongst these are the rules and practices relating to identification. These are particularly important in a case such as this, where there is no evidence of any sort to connect the appellant with the crime apart from the evidence of Jeffrey Lewis. There is no confession, no forensic evidence, no weapon or other suspicious article which can be traced to him. It all depends upon Jeffrey Lewis.
40. The safeguards against wrongful identification are of two sorts. First, of course, the jury must be properly and carefully directed about the difficulties and dangers of identification evidence. This the trial judge, Mohammed J, undoubtedly did, in the course of a conspicuously clear, careful and impartial summing up. Secondly, however, the investigation must be conducted in such a way as to minimise the chances of the wrong person being identified as the offender. For once a person has been identified and put on trial, the case gains a momentum and persuasiveness which it may be difficult for even the most competent defence to counter. All are agreed that defence counsel in this case was indeed very competent.
41. The principal safeguard is an "identification procedure", an objective test of whether the witness is indeed able to identify the accused as the person who did the deed. The most familiar of these procedures is, of course, an identification parade. In England and Wales, such a procedure must be held in the circumstances laid down in para 3.12 of Code D of the Codes of Practice issued under the Police and Criminal Evidence Act 1984:

"Whenever:

(i) a witness has identified a suspect or purported to have identified a suspect prior to any identification procedure . . . having been held; or

(ii) there is a witness available, who expresses an ability to identify the suspect, or where there is a reasonable chance of the witness being able to do so, and they have not been given an opportunity to identify the suspect in any of [those] procedures . . .

and the suspect disputes being the person the witness claims to have seen, an identification procedure shall be held unless it is not practicable or it would serve no useful purpose in proving or disproving whether the suspect was involved in committing the offence. For example, when it is not disputed that the suspect is already well known to the witness who claims to have seen them commit the crime."

42. The Board has held more than once in Caribbean cases that a parade should be held unless it would serve no useful purpose: see Goldson and McGlashan v The Queen [2000] UKPC 9, Aurelio Pop v The Queen [2003] UKPC 40, Ebanks v The Queen [2006] UKPC 6. Most recently, in Pipersburgh, Robateau v The Queen [2008] UKPC 11, the Board has said that "In their Lordships' view, in a serious case such as the present, where the identification of the perpetrators is plainly going to be a critical issue at any trial, the balance of advantage will almost always lie with holding an identification parade". Mr Guthrie QC, for the State, agrees that, with hindsight, an identification parade would have been desirable.
43. That was a sensible concession to make. Para 3.12 of Code D would clearly have applied. Mr Lewis had given the police two pieces of information before the appellant was arrested: that the other men in the car had called him "Dollars" and that he had asked to be taken to Sea Lots. That could perhaps be said to be an identification within para 3.12 (i). More pertinently, Mr Lewis had also had a better opportunity than anyone else to identify the person who had flagged him down, so para 3.12 (ii) applied. The appellant was denying any involvement, so his identity was in dispute. Nor, in my view, could it be said that a parade would serve no useful purpose. Mr Lewis never said that he knew the appellant by name, or that he had ever talked to him, before the day of the murder. There is nothing in the police notes of their interviews and investigations, or in the statements which Mr Lewis had given to the police at that stage, to indicate that the appellant was well known to him. The highest he had put it in any of the recorded material before the committal proceedings was his statement on 7 December that "I know them fellas by Nelson Street, they do thief and thing . . .". Whatever he may have meant by that, it certainly did not suggest a long or close acquaintance with this particular man. And in any event the appellant did not accept that Mr Lewis knew him. It is not enough to say that the jury later believed Mr Lewis' evidence that he had frequently seen the appellant on the streets. The object of a parade is to test the reliability of the identification at a much earlier stage, before either a witness or a jury has had a chance to go wrong. It is pulling oneself up with one's own bootstraps to say that a parade may not be necessary in a recognition case and then to decide whether or not it was a recognition case by reference to the findings of the jury on the evidence of the same witness who supplied which believed the identification.
44. The failure to hold an identification parade therefore deprived the accused of the possibility (whether fanciful or not is a matter to which I will return) that Mr Lewis would not have picked him out. (I appreciate that, had there been a parade, and Mr Lewis had picked him out, the jury would have had to be given some careful directions about that. But we are concerned with a different situation.) What then is the consequence? In England and Wales, there would have been a vigorous debate about whether the judge should have allowed the accused to be identified in the dock or admitted evidence of the dock identification in the committal proceedings. The test which appears in some of the cases is whether the judge is sure that the witness knew the accused so well that he would inevitably have picked him out: see R v Trevor Elton Gardner [2004] EWCA Crim 1639. In the circumstances of this case, that would have been difficult for him to conclude without knowing what the jury made of Mr Lewis' evidence. But no-one questions that evidence of a dock identification is admissible.
45. Once a dock identification was permitted, the jury should have been directed about the circumstances in which an identification parade should have been held and warned that the failure to hold one deprived the accused of the possibility of an inconclusive parade. This case is if anything stronger than the case of Pipersburgh, Robateau v The Queen [2008] UKPC 11. There, as was pointed out at para 17, "it may well be that the judge bemoaned the fact that no identification parade had been held and pointed out the advantages of such a parade". Here the judge did not even do that. He merely referred, among other points in a sentence setting out some of the defence arguments, to the argument of defence counsel that the police could have held a parade. In Pipersburgh, the Board went on to say that the judge "did not point out that Mr Robateau had thereby lost the potential advantage of an inconclusive parade", as the Board had said should be done in Aurelio Pop v The Queen [2003] UKPC 40. Here the judge did not do that either.
46. In Pipersburgh, the Board continued:


". . . the judge did not warn the jury of the distinct and positive dangers of a dock identification without a previous identification parade. . . . And, perhaps most importantly, even if the judge's directions would have ensured that the jury appreciated that this type of identification evidence was undesirable in principle, he did not explain that they would require to approach that evidence with great care. On the contrary, the closing words of the direction really left the whole matter to the jury on the basis that the witnesses said that they knew the men and it was simply up to the jury to accept or reject their evidence".

That is in effect what the judge did in this case. He pointed out that if they agreed with the defence on the issue of whether or not Mr Lewis knew the accused beforehand, then

"this would be a core or fundamental weakness because a witness could then easily point out to, and implicate anyone who is sitting in the dock at the first opportunity, which is at the Magistrates' Court, and such a dock identification would not be a reliable and good one in the eyes of the law. Such an identification would be worthless. So this is a very important issue in this case for your determination'.

47. The judge was thus leaving the jury to assess the reliability of the dock identification on the basis of whether they accepted Mr Lewis's evidence that he had known the accused by sight before the murder. This is what the Board had said in Pipersburgh was not sufficient. Moreover, Pipersburgh was in many ways a stronger case than this. It was not disputed that the appellants were employed as truck drivers to make regular deliveries to the premises where the guards were shot, so that there was good objective evidence to support the witness's claims to know them. There was no such evidence to support Mr Lewis's claim to have known this accused.
48. In my view, therefore, thorough and careful though this summing up undoubtedly was, it did not deal satisfactorily with the lack of an identification parade and the potential advantage that this might have brought, whether or not they believed that Mr Lewis did know the accused by sight beforehand.
49. The question remains, however, whether the Board can be satisfied that there was no miscarriage of justice as a result. It is not possible to answer this question simply by saying that the jury believed Mr Lewis, and that they did so despite having been directed that they must approach his evidence with very great care, not only because he was an accomplice in the robbery but also because he had been given immunity from prosecution in return for his testimony. The jury not having been given the help which they should have been given in evaluating his evidence, it is necessary to review the various possibilities on the basis, either that Mr Lewis was not telling the truth, or that he was mistaken.
50. The first possibility is that Mr Lewis had never seen the accused before, was lying when he eventually said that he had, and was lying when he purported to identify the accused in the dock. I do not regard this as fanciful. According to the police notes, Mr Lewis had given them only two clues, "Dollars" and a request to go to Sea Lots, before they arrested the accused. The police knew a "Dollars" who lived in Sea Lots. So they arrested him. They then got a statement from Mr Lewis in which he admitted that his earlier story about giving the man a tow and then driving off to look for parts had been an invention. Mr Lewis was in the frame. Then he was released. No further statement was taken from him. Fourteen months later, just as he was due to give evidence in the committal proceedings, he was given a written undertaking not to prosecute him for any offence he had disclosed, provided that he identified all the people whom he knew or believed to be involved in those offences, made full witness statements and gave evidence in accordance with those statements. It is to say the least possible that a person who might otherwise face prosecution for the capital offence of murder will tell lies in the witness box in order to avoid that risk, even to the extent of identifying someone he does not know. An identification parade would in all probability have prevented this.
51. Another possibility is that Mr Lewis had seen the accused before, though perhaps not as often as he made out in the witness box, but was lying when he said that the accused was the man who had flagged him down. In that case, an identification parade would not have been so useful, as Mr Lewis might well have identified the person he had seen around. He made it clear in his evidence how hard he was finding it being in police custody and how anxious he was to get away. But it does not follow that because he was prepared to lie in the witness box months later he would have been prepared to make a false identification so soon after the event.
52. A further possibility is that, whether or not he had seen the accused around, Mr Lewis was mistaken in his eventual identification. He did spend a considerable amount of time in the company of the person who he said had flagged him down and had good opportunities to see his face. So mistake may be less likely. But Mr Lewis had every incentive to convince himself that the police had got the right man. He will have wanted so much to believe that they had. He had fourteen months of knowing that they had a man in custody for the offence and that he had been set free as a result. If he had not seen the man before, an identification parade would have prevented this. Even if he had, it would have made the risk less likely, because on this hypothesis Mr Lewis is an honest witness who is trying to help. He has had less time to convince himself that the police have got the right man.

The reality is that Mr Lewis provided the police with two clues which enabled them to pick up the accused and after that no further steps were taken to confirm that they were right. This was a serious failure in a case which depended entirely upon the evidence of an accomplice. The majority may believe that the possibility that the police had leapt to the wrong conclusion is so slim that there is no risk of a miscarriage of justice. But this would not be the first time that the police had, quite understandably, leapt to a conclusion which turned out to be wrong. I may be more cynical than the majority, but I could not in all conscience send a man to his death on that basis.