England and Wales Court of Appeal (Criminal Division) Decisions

IN THE SUPREME COURT OF JUDICATURE
COURT OF APPEAL (CRIMINAL DIVISION)
ON APPEAL FROM LIVERPOOL ASSIZES
Mr Justice Cassels

		Royal Courts of Justice Strand, London, WC2A 2LL
		Tuesday 28th October 2003

Lord Justice Rix:

Introduction and synopsis

1. Over 50 years ago, on Saturday 19 March 1949 at about 9.35 pm, a notorious double murder occurred at the Cameo Cinema in Liverpool. Its manager, Leonard Thomas, and assistant manager, John Catterall, were shot dead in the former's office in the course of a robbery of the day's takings. Charles Connolly and George Kelly were subsequently tried in Liverpool for those murders. The Crown alleged that Kelly was the gunman and Connolly his lookout. Kelly and Connolly were arrested on 30 September 1949, on the day following police statements implicating them were made by James Northam and Jacqueline Dickson. Two weeks earlier, on 15 September 1949, a prisoner called Robert Graham had made a police statement that a man known as Donald Johnson had confessed to him, Graham, while in prison, that he, Johnson, had committed the murders and the robbery. Despite that statement, in November 1949 Graham made further statements to the police concerning further alleged prison confessions to the murders, this time by Kelly and Connolly. Graham said that each of them had confessed their roles to him while they were together in Walton Prison between 14 and 16 November 1949. Notice of the additional evidence of what we shall for the sake of convenience describe as Graham's second statement (in fact it would seem that there was no second statement as such but a series of at least three interviews) was served on the defence on the first day of their joint trial, which lasted from 12 to 28 January 1950. Thus it had not been available at the committal proceedings in October 1949. Graham's first statement was never disclosed to the defence, nor even to prosecuting counsel. The discovery of that first statement, still present in police files, in the early 1990s ultimately led to applications to the Criminal Cases Review Commission (CCRC) and, in 2001, to these references by the CCRC to the Court of Appeal.
2. Johnson had already stood trial charged with being an accessory after the fact to the murders and had been acquitted at the direction of the judge, at the time when Graham, in his first statement, said that he (Johnson) had confessed to carrying them out himself. Johnson's trial took place in June 1949, soon after he had made unsigned statements to the police on 2 and 6 May in which he had confessed to helping the gunman, whom he described and disguised as "Charlie Duggan" (or "Dugan"), dispose of the gun. He was acquitted because the judge ruled that the second of his police statements had been obtained by inducements and was inadmissible. Graham said that following that acquittal Johnson, who at this time was in custody on another matter, had returned to prison and told him in glee that he had committed the murders himself and now could no longer be prosecuted for them.
3. That account was given, as we have said, in Graham's first statement dated 15 September 1949. Graham's statement was made to Detective Chief Inspector Balmer who was conducting the police investigation into the murders. CI Balmer reported on this development to his superior, Chief Superintendent Smith, in writing dated 17 September 1949. He said: "I am satisfied GRAHAM is telling the truth. Whether JOHNSON is, is of course, a different matter." CS Smith endorsed the report on 20 September 1949 as follows: "I beg to report that although the statement of GRAHAM is very interesting, it does not seem that we can do anything further regarding JOHNSON at the moment." The report was then passed to the Assistant Chief Constable, who initialled it and added the word "Seen".
4. The subsequent statements of Northam and Dickson were also made to CI Balmer, as was Graham's second statement. In the course of his evidence at the joint trial, CI Balmer said that the first time he had met Graham in connection with the case was on 19 November 1949^[1]. At Kelly's re-trial Graham also said that he had first met CI Balmer on 19 November (both in cross-examination and re-examination), that "the only persons who ever spoke to me about this case was Kelly and Connolly", and that "I am on oath in this box and I can only say I knew nothing about it until I was told by these two people charged with it". A short while later, in direct answer to the judge, Graham again said that he first saw CI Balmer on 19 November. All that evidence was false and probably deliberately so. CI Balmer died on 3 May 1970, and thus no explanation from him regarding the non-disclosure of Graham's first statement or any other matter is available.
5. Despite the evidence of Northam and Dickson, and the evidence of Graham concerning the prison confessions of Kelly and Connolly, their first trial, before Oliver J, ended on 28 January 1950 without any verdicts. Within two days the Liverpool Echo was reporting that the retrial would take place, again in Liverpool, at the next assizes commencing on 31 January 1950. In the event, there was a slight delay, during which the trials of Kelly and Connolly were split. Kelly's retrial began on 2 February and ended on 8 February with his conviction. He was tried only for the manager's murder. He was sentenced to death. The judge, Cassels J, said that "the Jury have rightly found you guilty". He commended Northam and Dickson for their evidence and said he would forward a recommendation regarding Graham. Kelly's appeal was heard and dismissed on 10 March. He was hanged on 28 March. His daughter, Kathleen Hughes, is now the applicant in the reference which has led to this appeal.
6. In a letter dated 28 February 1950 from the Deputy Director of Public Prosecutions, to the Under Secretary of State at the Home Office, the former wrote as follows:

"The all-important evidence for the prosecution consisted of a woman named Dickson and a man named Northam, both persons of bad character, who swore that they were in a Public House on the night of the murder with Kelly and Connolly, that they heard the robbery of the cinema planned by these men, and that after the shooting, Kelly admitted that he had shot the Manager and the Under-Manager and Connolly stated that he had waited outside the cinema but that he ran away as soon as he heard the shots. Graham corroborated the evidence of Dickson and Northam, because he swore that in conversations which he had with both accused in prison, each admitted to him the part which each had played in the commission of the crime, which was substantially the same as the admissions which they had made to Dickson and Northam prior to their arrest.

"I am of the opinion that but for the evidence that Graham gave before Mr Justice Cassels, Kelly would not have been convicted."

7. The Under Secretary replied on 13 March 1950 to confirm the Secretary of State's decision to recommend the remission of the remainder of Graham's sentence and his immediate release from custody.
8. Connolly was due to be tried separately on 13 February 1950, soon after Kelly's conviction. On that day, in circumstances which we will need to consider in detail below, he was given the opportunity to plead to new charges of robbery and conspiracy to rob. He pleaded guilty, and was sentenced to ten years imprisonment on the count of robbery and two years concurrent on the count of conspiracy. The prosecution offered no evidence on the counts of murder, and the jury were directed to acquit.
9. Connolly was released from prison in about 1956. He died on 18 April 1997. His widow, Eileen Connolly, is now the applicant in the reference which has led to this appeal.
10. Before his death Connolly, then a hotel doorman, met a businessman who was a guest in the hotel where he was working. That was in the middle of 1991. The businessman was Mr Luigi Santangeli, who as a teenager had tried to attend the joint trial in Liverpool, but failed because of the great number of people queuing to get in. As a result of what he was told by Connolly, Mr Santangeli resolved to research the case. Connolly had said that neither he nor Kelly had been involved in the murders. Mr Santangeli believed him and felt he had a responsibility to bring the matter into the public domain. Later that year he obtained access from the police to the case papers. Among them he found the original manuscript and a typed version of Graham's first statement. A visit to the public records office in July 1994 produced the notice of additional evidence dated 12 January 1950 relating to Graham's second statement. Also in 1994 he visited the cells in Walton Prison where Kelly and Connolly had been kept on remand awaiting trial. At his request recordings were made of interviews with Connolly conducted in August 1993 and October 1994 by Mr Roger Phillips and Mr Roger Wilkes respectively, both BBC journalists: the transcripts of those interviews are among the material which has come before this court as a result of the CCRC references.
11. Also among that material are the statements and additional statements of Northam and Dickson made on 29 September and 10 October 1949. These were not mentioned in the judge's summing up of the re-trial and do not appear to have been disclosed to the defence: although the defence may have known of their existence there is no sign that they knew of their contents. The significance is that a comparison of the earlier statements with the additional statements, and of the statements as a whole with evidence given at committal and at trial, may have affected those witnesses' credibility.
12. The grounds of appeal in the case of Kelly are that his conviction is unsafe because (1) Graham's first statement was not disclosed; (2) Northam's and Dickson's statements were not disclosed; and (3) Kelly's retrial was severed from that of Connolly's without just or legal cause. The grounds of appeal in the case of Connolly are that his conviction is unsafe because (1) his guilty pleas were induced by duress of circumstances and/or oppression such as effectively denied him a free choice in making his plea; and (2) the first statement of Graham and the statements of Northam and Dickson were not disclosed.
13. On behalf of the Crown, Mr Pownall QC accepted the authenticity of Graham's first statement and that even under the duties of disclosure which applied in 1949 it should have been disclosed and was not. Furthermore he accepted that its disclosure would have been likely to have undermined the integrity of at least Graham and CI Balmer, that the importance of Graham's evidence was accurately reflected in the deputy DPP's letter (see para 6 above: "but for the evidence that Graham gave…Kelly would not have been convicted"), and that the integrity of CI Balmer was also of paramount importance. He said: "If Graham's account had been severely undermined by reason of his earlier account, an already weak case became significantly less compelling." In these circumstances the Crown felt unable to argue that the remaining evidence was of such quality or strength as to permit submissions that the conviction of Kelly was safe. Nevertheless, Mr Pownall submitted that if this had been a modern case, and Kelly were alive, the Crown would have sought a re-trial.
14. In the circumstances, Mr Pownall did not address the two other grounds in Kelly's appeal at length. He submitted, however, that although the defence were aware of the fact that both Northam and Dickson had made statements to the police, they either chose not to see them or asked and were refused. Nevertheless, he accepted that they were clearly disclosable and should in fairness have been disclosed, and that if they had been, the defence would have been able still further to undermine the prosecution case. He submitted that if these further non-disclosures had stood alone, the conviction would have remained safe: as it was, they added significant weight to the principal ground of appeal.
15. As for severance of the re-trial, Mr Pownall accepted that no proper basis existed for it, but again submitted that the fact that Kelly stood trial alone would not of itself have rendered the verdict unsafe.
16. The Crown therefore did not seek to uphold the conviction of Kelly for murder. It did, however, seek to uphold Connolly's convictions, on the ground that his pleas of guilty were unequivocal, freely tendered and not attributable to undue pressure. Nor, it was submitted, were they founded upon any understanding that there was no other material (the undisclosed statements) capable of undermining Graham, Northam or Dickson's evidence. Moreover, the application to sever his trial from Kelly was made by his own counsel, in his presence and thus presumably with his authority.
17. At the hearing of these appeals we decided that they should both be allowed. We gave brief reasons at the time, while reserving our fuller judgment. We said (inter alia):

"Although there was other significant evidence in [Kelly's] trial, in particular the evidence of the witnesses Northam and Dickson, who said that Kelly and Connolly had in their presence planned the robbery of the Cameo Cinema and had later confessed to what had happened there, in the case of Kelly that he had fired the fatal shots and in the case of Connolly that he had stood outside when Kelly went into the manager's office, both of those witnesses, Northam and Dickson, were treated as accomplices and the judge, Cassels J, directed the jury that they should be cautious about convicting Kelly without corroboration. He also directed the jury that they could find corroboration in the evidence of Graham. If Graham's earlier statement implicating Johnson had been disclosed, Kelly's counsel could have made effective use at trial of the oddity that in September 1949 Graham had implicated Johnson and in November 1949 had implicated Kelly (and Connolly) as the Cameo Cinema culprits. The defence might well have been able to discredit Graham's evidence in the eyes of the jury, and in doing so, might also have been able to call into question certain aspects of the police investigations and evidence…

"The Crown's case against Kelly was entirely circumstantial, lacked any forensic support, and rested essentially on the evidence of Northam, Dickson and Graham. The Crown recognised at the time that Graham's evidence had made a significant contribution to Kelly's conviction, and Graham was promptly rewarded by his immediate release from his then current sentence. The jury at the first joint trial of Kelly and Connolly had been unable to agree a verdict. The Crown acknowledges that the failure to disclose Graham's earlier statement was in breach of the Crown's obligations, even as the law recognised them to be at the time.

"We therefore agree that the Crown's failure to disclose Graham's earlier statement renders Kelly's conviction at his retrial unsafe. Unfortunately, Kelly was sentenced to death and, after losing his appeal, hanged.

"In the case of Connolly, the matter is complicated by the fact that his trial was severed from that of Kelly. If it had not been, as it should not have been, Connolly may well have been convicted as well, although his case was different since the Crown did not allege that he had been in the manager's office at the time of the murders. Following Kelly's conviction at his retrial, however, Connolly pleaded not guilty to murder, but guilty to new charges of robbery and conspiracy to rob which were added at the last moment to the indictment. The question is whether Connolly's pleas of guilt to the robbery charges are undermined so as to render his convictions on those charges unsafe in the light of the Crown's failure to disclose Graham's earlier statement and also in the light of the circumstances in which those pleas were tendered. The Crown has submitted that those pleas were entirely voluntary and that those convictions remain safe. We disagree. Once Kelly's conviction has been declared unsafe for the reasons stated above, it is in our judgment unrealistic to regard Connolly's convictions, albeit in terms of robbery rather than murder, as safe. If Kelly cannot safely be regarded as the murderer, Connolly cannot safely be regarded as his accomplice in a case where the evidence against the two was essentially the same and came from the same sources. If anything, the evidence against Connolly was weaker. Kelly's conviction, moreover, had been obtained in part by reason of the non-disclosure of Graham's statement. Connolly's convictions, although the product of pleas of guilt, can themselves be said to be founded in part on the Crown's failure to make proper disclosure of the Graham statement as well as on the unenviable position of Connolly who, despite the earlier maintenance of his innocence, had to face up to the fact that Kelly had been convicted and sentenced to death. In those circumstances his own counsel said, in mitigation, that he had advised him to plead guilty on the terms available to him. Had Graham's earlier statement been available, Connolly's counsel's advice may well have been very different. "

18. We turn to a fuller and more detailed explanation of our reasons for allowing the appeals of both Kelly and Connolly and quashing their convictions.

The murders

19. The murders took place at about 9.35 pm on the night of Saturday 19 March 1949. Witnesses who heard the gunshots said variously that they occurred at times between 9.32 and 9.40. Eye witnesses among the staff who saw the gunman leave but could not identify him, described him as wearing a brown overcoat with a belt, a trilby hat pulled down over his head and a black silk scarf masking the lower part of his face. The Home Office pathologist's evidence is significant. Dr Grace said the manager had died from a single bullet wound (which entered the body on the left side of the chest), but that the assistant manager had been shot three times: once through the palm of the right hand (consistent with the hand being raised in an attitude of self-protection), once into the chest by the collar-bone and out again following a superficial course, and thirdly, the fatal shot, in the back. That third shot struck a rib and "went right down through the body and through the liver, causing death, and lodged on the inside of the top of the thigh". This caused Dr Grace to consider that Mr Catterall was probably on his hands and knees at the time.
20. Another witness, Dr Firth, who gave evidence concerning the murder weapon (which was never found) said that the cartridge cases found at the scene revealed that the bullets had undoubtedly been fired from an automatic pistol.
21. The murderer appears to have known the cinema well enough to get to the manager's office, and to cut the wires of a telephone from which the staff sought to raise the alarm.
22. One of the cinema staff who saw the gunman leave was the cinema fireman. He saw the gunman come out of the manager's office with the gun in his hand.

Johnson's role

23. As stated above, Johnson was prosecuted as being an accomplice after the fact, but was acquitted on the direction of the judge. He immediately thereafter was said by Graham to have confessed in prison to being the actual gunman. Johnson's confession to the police was ruled inadmissible because he had been promised bail (in the hope that he would lead the police to the murderer): indeed, a policeman had stood surety for bail.
24. Johnson's (unsigned) statements to the police were to the effect that he had gone with the gunman to the cinema earlier in the week to reconnoitre the place (and that he was in any event familiar with the cinema) and had been shown the gun (which he drew, a picture of an automatic pistol), and had arranged to meet the gunman again outside the Boundary Hotel a short distance from the cinema at 10.45 pm on the Saturday night. While waiting there, a policeman had checked his identity card (a fact later verified by PC 247, PC Thompson): after which the gunman had approached him and given him the gun to conceal, which he had done. In a police report dated 16 May 1949 made by Chief Inspector Morris, reference is made to the statement of PC Thompson to the effect that he had seen a man outside the Boundary Hotel at 10.15 on the night of the murders. The man appeared uneasy and was looking around him. The officer asked what he was doing and was told that he was waiting for someone. The officer then asked him for his identity card, which was produced; but no note was made of it at the time. CI Morris's report comments that, although it would seem that Johnson planned to plead not guilty, as far as could be checked everything said by him had been found to be true. According to a further detailed report concerning the police efforts to solve the crime and Johnson's trial made by CS Smith dated 22 June 1949, PC Thompson testified at Johnson's trial to speaking to a man whom he identified as Johnson. The report also refers to the statement of a prisoner, Bernard McBride, who said that while in custody Johnson continually reiterated his knowledge of the crime, the murderer and where the gun was concealed.
25. Johnson was acquitted on 15 June 1949. Graham's first statement, dated 15 September 1949, said that in the week before his trial, after a visit from Johnson's counsel, Miss Rose Heilbron KC, Johnson told him that she had advised that everything turned on legal argument, and that if she failed on that, he would get life. He said: "Well, if it fails I'm going to tell the truth and go into 'High Wing' [ie the condemned cell], I couldn't do 20 years in here." After his acquittal, however, "he was in great spirits and has been ever since" and spoke often of how he had committed the murders. The statement continued:

"He said, 'I had the place lined up for some time. I often used to go there. Anyway on this night I went up to the manager's office…When I asked them for the cash one of them went for me so I let him have it right away. The other fellow then went for me so I put a bullet in him too but it went into his neck and he dropped to his knees and turned over. The first was killed right away and I thought to myself well, this bugger might recognize me, so I shot him in the back to finish him off. I put three bullets into him before he went out, but the first chap went out when I put the first in him…as I was going out I bumped into the fireman by the top of the stairs. How that fellow didn't recognize me I don't know…

"I ran out but funny enough I didn't like to leave the district. I mooched around and then to finish it off I was stopped by a copper. He asked me for my identity card and I thought he was going to search me but he didn't. If he had I was finished because I had the gun on me then. Anyway I went and got rid of it right away…

"He also told a prisoner named McBRIDE that he's done it.

"I am willing to give evidence any time you want."

26. Following the submission of this statement to the CCRC, forensic investigations were carried out into its authenticity. An original handwritten statement and a typed copy were examined. The conclusions were that they were genuine and they are acknowledged by the Crown to be so. The typed copy states that it was taken by CI Balmer.
27. It is plain from the joint trial and Kelly's second trial, and acknowledged by the Crown, that the statement was not disclosed. The Crown's acceptance that its non-disclosure renders Kelly's conviction unsafe is made irrespective of the motivation of such non-disclosure. Mr Pownall submits that its survival, together with CI Balmer's report on it dated 17 September 1949 which was itself seen and countersigned by CI Balmer's superiors, indicates that there was no conspiracy or bad faith involved. The non-disclosure of the statements of Northam and Dickson as well may simply suggest that there was a practice of non-disclosure of statements: even though it is conceded that even at that time the obligation of disclosure of such statements in the interests of fairness already existed. Nevertheless, when the evidence of Graham at both trials and of CI Balmer at the first trial (see para 4 above) to the effect that the two had not met in the context of the case prior to 19 September is taken into account, and in the absence of any explanation for such testimony, the conclusion that such evidence amounted to deliberate concealment becomes at the least highly likely.
28. The importance of such non-disclosure is not in dispute, and that significance is enhanced if non-disclosure is viewed as concealment. The mere fact that Graham allegedly received confessions of the murders from two different sources would of itself have cast doubt on the totality of his evidence. So would his inaccuracies or lies about his first meeting with CI Balmer, just as the latter's evidence would have been similarly undermined. Of particular importance, moreover, are the details of the shootings described in Graham's first statement. From where did Graham get those details? It is not suggested that they were available in the press. They more or less accurately reflect Dr Grace's testimony. Graham could only have got them from Johnson or the police. Johnson could only have got them because he knew the gunman or was himself the gunman. Some of them emerged again in Northam's evidence (see below). Northam said that he met Kelly on Monday, 21 March (two days after the murders) and that Kelly then told him what had happened in the manager's office: he had shot the manager (once) when another man walked into the room, who, after he had also shot him, "fell to his knees" and was then shot again. Without any knowledge of Graham's first statement, this evidence from Northam was naturally viewed as highly probative. The judge himself understandably stressed this evidence at the very end of his summing-up, linking it with Graham's evidence of Kelly's confession to him, when he said:

" "He fell to his knees" – according to Dr Grace the bullet that killed Catterall, the assistant manager, entered his back, struck his rib, went right down the body and went through the liver on the way down and was finally found on the inside of the thigh. Was Catterall shot while he was on his knees? Who knew that? Only one person. Has Graham imagined his evidence? If you have a reasonable doubt, you will find him not guilty. If, upon the evidence, you come to the conclusion that George Kelly is the man who, on that night of March 19^th of last year, shot that cinema manager, Leonard Thomas, and thus brought his life of 44 years to an end, you will find him guilty. Will you now please consider your verdict."

29. If Graham's first statement had been disclosed, the judge could not have ended his summing-up in this way. The implicit reference to Kelly as the "Only one person" who could have known that Catterall was shot while he was on his knees could not have been made. On this point alone Northam's evidence would have been scrutinised with knowledge of what Graham had said in his first statement as to what Johnson had allegedly told him. Moreover in this respect Graham's evidence as corroboration of Northam's would have suffered a double blow: both because of Graham's own first statement, and because Northam's reference to Catterall being on his knees would have lost its power and become suspect in itself. We will revert to some further difficulties about Northam's and Dickson's evidence below.
30. Graham's first statement was made before the committal proceedings and thus its details could not have been obtained from press reports. In addition to the details of the shootings which correspond to Dr Grace's findings, statements of PC Thompson, of the fireman (Joseph Griffin), and of the prisoner McBride were consistent with or confirmed details of Johnson's alleged account to Graham.

Graham's second statement

31. It appears that there was no such second statement as a document in itself, at any rate none has been found, and that is so despite CI Balmer's evidence that Graham had signed a statement. Instead it appears that Graham's ultimate evidence was based on the notice of additional evidence served on 12 January 1950, which is itself drafted in the form of a statement. Thus it begins: "TAKE NOTICE that upon your Trial an additional witness will be called of whose evidence the following is a copy: ROBERT GRAHAM, 16 Charlotte Street, Preston, says:…" It also appears that this additional evidence came forward as a result of (at least) three interviews between Graham and CI Balmer on 21, 23 and 28 November 1949. There are no immediate records of those interviews, but there is a report by CI Balmer dated 30 November 1949 which purports to set out verbatim what Graham said, and from which it appears that their first meeting was on 21 November and not 19 November as CI Balmer had said at the first trial and Graham had said at both trials. An oddity is that in the joint trial CI Balmer gave evidence that he saw Graham on 19 November and that Graham then began to make a statement, but something happened and he left over the statement until 21 November, and then had seen Graham twice more – so, all in all, four meetings. The report, however, refers only to three meetings.
32. Thus the report states that on 21 November CI Balmer received a telephone call from the Preston police to tell him that Graham, then in Walton Prison, wanted to see him about the Cameo murders; and that the other two meetings were also at Graham's request. The effect of Graham's second statement and evidence was that he, Kelly and Connolly were in separate cells in the prison's hospital. Kelly and Connolly could not communicate with each other from their cells and were not allowed to exercise together, but he, Graham, exercised with each of them on alternate days. They used him to pass messages to each other, and to find out what the other was going to do. He had not known either of them before. He met Connolly first, on 14 November, when Connolly told him what he was in for, but that he had nothing to do with the murders "other than being in the same company with the fellow that did it". He then explained that he was in a pub with Kelly, Northam and Dickson and they decided on a job at the Cameo and he went there with Kelly and Kelly did the shootings, while he stayed outside the door. Next day Graham spoke to Kelly. Kelly said he borrowed a coat from Northam, that he shot the fellows but Connolly would not go in; then he got rid of the hat and coat and "was in my pub having a drink five minutes later. My life hangs on that five minutes". The next day Graham met Connolly again. Connolly complained that Kelly had foolishly pretended on the day of their arrest that they did not know one another. Graham also said that Kelly told him that when he and Connolly met up at the Star pub on the day after the murders (Sunday 20 March) he called Connolly "a yellow bastard" and told Northam that he would give him back his coat the next day; also that on the day of his arrest, he had gone to see CI Balmer about an entirely different matter "and the next thing I was told was that I was going to be charged with murder".
33. CI Balmer commented in his report that Graham had told him things "which were not known to any members of the public, and were not referred to in any press reports. These facts were solely known to Ch. Supt. Smith, myself, and the prisoner who uttered them. It is obvious, therefore, that GRAHAM is telling the truth." CI Balmer then identified these facts, of which there were three: (i) Connolly said that he had sent for CI Balmer soon after his arrest, but changed his mind about saying anything and made an excuse; (ii) Connolly said that on the day before the murders he had just drawn £15 in wages; (iii) Kelly's comment about the circumstances of his arrest (above). CI Balmer's reference to press reports was relevant, since Graham was on remand without loss of privileges and had access to newspaper reports about the committal proceedings.
34. At Kelly's retrial Cassels J summed up Graham's evidence in very much the terms found in CI Balmer's report and in the notice of additional evidence.
35. Kelly at both his trials and Connolly at the joint trial each gave evidence denying any such conversations with Graham. Kelly said that they had no need to use him to pass messages between them, since they could communicate with each other from their respective cells without difficulty. This was a matter investigated by Mr Santangeli on his visit to the hospital wing at Walton Prison, then in the same condition as in November 1949. Kelly's and Connolly's cells had been identified at the trials. In his statement to the CCRC Mr Santangeli said that he, Mr Bill Morrison (playwright) and the assistant governor had conducted an experiment and found that communication between the cells was possible without shouting, albeit at a voice pitch above normal conversational level. The Crown accepted that, contrary to Graham's evidence, it was possible for Kelly and Connolly to speak to one another from their respective cells. Although a subsidiary matter, and evidence which clearly could have been obtained for trial, this is a further pointer to the unsatisfactory nature of Graham's second statement when once that is viewed in the light of the non-disclosure of his first statement and his false evidence which concealed the occasion of its making.

Northam's and Dickson's evidence

36. We have already referred to the Crown's acceptance that Northam's and Dickson's statements were not disclosed to the defence, although the existence of at any rate their first statements, those made on 29 September 1949, was known.
37. The full history of Northam's and Dickson's roles in the case is complex, and since the Crown accepts that Kelly's conviction is not saved by their evidence we do not think that it is necessary to set it out in full. Nevertheless, as briefly as we can, the position is as follows.
38. At the trials Northam and Dickson gave evidence directly implicating Kelly and Connolly both in the planning of the Cameo cinema robbery on the evening of 19 March at the Beehive pub and in subsequent confessions, for instance on Sunday 20 and Monday 21 March, as well as about the lending by Northam to Kelly of Northam's coat, and other incriminating matters. If that evidence had come forward from entirely independent witnesses and did not suffer from certain defects which we will seek to set out, it might have been considered to be powerful material. As it was, they were rightly regarded as accomplices on whose evidence Cassels J had to direct the jury that it would be dangerous to convict without corroboration:

"Let me warn you that it is dangerous to convict a prisoner on the uncorroborated testimony of accomplices, although it is within your province to do so if you choose…The evidence of one accomplice cannot be corroborated by another…But a prisoner's conduct in the circumstances of the case may be corroboration. Graham's evidence, if you accept it, may be corroboration, because you may think it strengthens the other evidence. That is a matter for you. Whatever description may be given to Graham – and you may think that he has laid himself open to many – but whatever description is given to that man, he cannot be described as an accomplice. A description of the coat worn by the murderer that night may also be corroboration. That again is a matter for you."

39. Apart from the last throwaway line about the coat, the judge focussed upon Graham's evidence as the possible corroboration; and it was to Graham's evidence that the judge returned, as set out above, in the closing lines of his summing-up.
40. On 4 April 1949 the police received an anonymous letter from someone who wrote that they were "turning informer for gain". The letter said that the writer and others including a girl had heard about the murderer's plan for the robbery: in the end "only two went" and one of them lost his nerve and would not go in and "has not been seen since". The writer had seen the murderer "5 days ago" and was told that he was "in it with him": therefore "what I want to know is how I stand…if I turn King's evidence." As proof of truth the writer said that the gunman had thrown the gun in the pond in the park on Edge Lane. (The pond was subsequently dragged, but no gun was found, although CI Balmer said that he considered that it could still be buried there in the mud.) The letter asked for a reply via the personal column in The Echo, and offered "both there names also some of the bullets he left with me", provided a promise of immunity was given. The police responded as asked, but heard nothing further. The letter is referred to in the report of CS Smith dated 22 June 1949, following the failure of the Johnson trial.
41. The same report states that at some time after the letter was received Dickson "appeared on the scene" and stated that the letter had been written by Connolly. He was therefore interrogated, but he ridiculed Dickson's allegations and "his alibi appeared to be perfect". Dickson then disappeared. Her partner was Northam. The rest of the report is concerned with Johnson and his trial.
42. The anonymous letter was in evidence at the trials. Northam and Dickson said it had been written by themselves and contained both truth and lies.
43. It appears from the evidence of CI Balmer at the retrial that on 13 May 1949 Dickson "sent for" him. On that occasion she said nothing about Kelly, and nothing about the meeting at the Beehive on the evening of 19 March or any other meetings with Kelly or Connolly. She merely told CI Balmer to see Connolly. CI Balmer did not say that she told him that Connolly was the writer of the letter, as CS Smith's report had it. CI Balmer therefore saw Connolly, but got no information from him. He saw Dickson again on 16 May, and she told him "I told you to get Connolly; let him talk. If the gang want to get anybody, they can get him, not me." CI Balmer then said he saw Dickson a third time on 23 May: she again said nothing about the Beehive.
44. In the meantime the police were concentrating on Johnson, from 2 May, when he was seen for the first time, until his acquittal on 15 June.
45. Then on 29 September 1949 both Dickson and Northam made statements. They had been promised immunity. At the retrial CI Balmer said this was the earliest date he learned from those witnesses about Kelly being involved in the murders.
46. Northam's statement of that date began by saying "I want to tell the whole truth about the Cameo Murder…" He related how on the evening of Saturday 19 March he and Dickson had met Kelly and Connolly in the Beehive. They discussed possible robberies. One venue mooted was a funfair, where Kelly said they would have to cut the wires – Northam said he saw the two prongs of a pair of pliers sticking out "when his coat or mac was open". It was Connolly who first mooted the Cameo, adding that a gun (or dummy) would be necessary. Kelly then pulled out a gun and a handful of bullets and loaded the magazine. Northam refused to go along with them. He and Dickson went home. On the next day, Sunday 20 March, the murders were in the papers, but "Jacky [ie Dickson] and me stayed in Birkenhead all day." On Monday, 21 March, Northam met Kelly outside the Palais de Luxe. They discussed the murders, and Kelly accused Connolly of letting him down – "he'd got scared and was yellow". He described the shootings, first the manager ("I shot him") and then the assistant manager who came in at that point ("I saw red then and shot him once. He dropped on his knees and said something to me and so I shot him again…"). He could not open the door and so shot the lock off. He ran off, and Connolly was nowhere to be seen. Later Northam got worried and wrote the anonymous letter and gave it to Dickson to post: but he was too scared of Kelly to answer the police's personal advertisement in The Echo.
47. It may be observed that there is nothing in that statement about lending Kelly his coat; nor about its return; nor about meeting Kelly and Connolly on Sunday 20 March. The reference to the pliers sticking out "when his coat or mac was open" was an oddity. Mention of pliers was relevant, for the cinema's telephone wire had been cut, but nothing was said about this during his evidence at trial, and talk of "his coat or mac" did not suggest the lending of Northam's own coat.
48. The omission about the coat was dealt with in Northam's statement of 10 October 1949. He said Kelly asked him to lend it to him "as he was cold", which he did. Kelly left the pub wearing it "with the belt fastened around him". Kelly showed a woman's small brown apron which he said "would do for a mask". At their meeting on Monday 21 March Kelly said he would fetch the coat and returned with it within 15 minutes. Northam handed to the police an overcoat on 10 October, which he said was the coat in question, "but I cannot now find the belt off it. It has the belt loops on it." (The overcoat was recovered from the home of Kelly's parents.) He also said that on the Monday Kelly told him both that the gun was in the soil in the park and that it was in the lake. There was still nothing about a meeting on Sunday 20 March.
49. In her statement of 29 September Dickson gave similar evidence about the conversation "in a passage" in the Beehive on the Saturday night. She too said nothing about the lending of a coat, and nothing about a further meeting on the Sunday: Northam and she stayed in Birkenhead. On the Monday she went with Northam to Liverpool, left him at Lime Street, and when she met up with him again, he told her he had met Kelly. A few days later she met Connolly, and he had given her a handkerchief filled with bullets: when she showed them to Northam, he kept six and later gave them back to her, and she threw them down a drain. About two weeks after that she saw Connolly again, and he then told her that he and Kelly had gone to the Cameo, and he waited by the door when Kelly went inside. He ran away when he heard a shot. She posted the anonymous letter to the police which Northam had written. In the final paragraph of her statement she said that in the Beehive on the night of the murders Kelly had been wearing "a dark raincoat or overcoat with a belt around it". However, Dickson also made a brief further statement on 10 October in which she said that Northam lent Kelly his overcoat as Kelly was cold, and she identified the overcoat that Northam had provided to the police as the overcoat in question, save that on 19 March it had had a belt which Kelly had fastened. She also said that before Kelly had left the pub that night he took a trilby hat "from his jacket pocket" and also showed them a small brown apron which he said would do for a mask. However, like the evidence about the pliers, the evidence about the trilby hat was not advanced either at the committal or at the trials. Those omissions suggest that not even the prosecution had access to Northam's and Dickson's statements.
50. It was only at the committal proceedings, which took place from 19 to 22 October 1949, that Northam and Dickson said that on Sunday 20 March they had not stayed at home in Birkenhead but had gone, by arrangement made the previous evening, to meet Kelly and Connolly at a pub in Brownlow Hill in Liverpool (the White Star). Connolly was scared and talking of leaving the country. Kelly, however, acted confidently and said CI Balmer had been to see him at home that morning (which had in fact happened) and that he had a good alibi. He called Connolly a yellow bastard, and threatened him (and Northam and Dickson). Northam asked for his coat back, and Kelly said he would fetch it the next day. Dickson also said that Kelly advised Connolly to get himself an alibi, and Connolly said he would figure one out with his wife.
51. These developments in the accounts of Northam and Dickson were obscured in the evidence given at the retrial by reason of the non-disclosure of their statements. Thus at the joint trial CI Balmer had said that he had first taken a statement from Northam and Dickson on 29 September. At the retrial Northam said that he had first told the police about the Beehive meeting (Saturday evening) and the White Star pub meeting (Sunday) and the Lime Street meeting (Monday) in September, but had omitted any mention of lending his overcoat until 9 or 10 October. That last fact appears to have been known to Miss Heilbron, cross-examining on behalf of Kelly, but it is not clear how she came to that knowledge. It may be that it emerged as a result of Northam's production on 10 October of the overcoat to the police. In re-examination by Mr William Gorman KC, who was leading for the Crown, there was this passage:

"Q. On the 29^th September when you gave that statement to the police did you then omit anything at all of what you have given in evidence, either before the Magistrates or here? A. Nothing at all – just the overcoat."

52. That demonstrates that even prosecuting counsel was not aware of the contents of the 29 September statements. Northam's answer, like his answer in cross-examination, was also a lie, so far as concerned any mention of the meeting on the Sunday at the White Star. So it was that the judge summed up this part of his chronology as follows:

"September 29^th: Northam gives all the details to the Police except about the overcoat … October 10^th: the coat handed to the Police."

53. CI Balmer in his evidence also obscured the same point. Miss Heilbron asked:

"Q. Would you tell my Lord and the Jury what was the earliest date you learned from Northam and Dickson about Kelly being involved in this murder? A. The 29^th September.

"Q. What was the date you heard of the Beehive? The same date.

"Q. And the interview on the Sunday? A. Well, I take it about the same date.

"Q. And the Monday in Lime Street? The same date…

"Q. The first you learned of the overcoat was October 10^th? A. Yes."

54. Thus CI Balmer deliberately, as it would appear, obscured the point that not only had the evidence about the Sunday meeting at the White Star not emerged at the time of the first statements, but that those statements positively asserted that the witnesses had stayed at home in Birkenhead all day. CI Balmer was, however, well aware of the late appearance of the meeting at the White Star on Sunday, for on 26 October 1949, a few days after the committal proceedings, he wrote a report about it to CS Smith. In it he referred to his earlier report of 12 October and stated that since that time Northam and Dickson had "recalled a further meeting", and he set out their further evidence. He also stated: "The additional information was given to us by the witnesses NORTHAM and DICKSON prior to the Court proceedings and was embodied in their statements." However, no supplementary or refashioned statements have been found.
55. Two further points about the late appearance of evidence about the Sunday meeting may be mentioned. One is that, in the absence of any mention of it, Northam has his next meeting with Kelly, that in Lime Street on the Monday, as being the occasion when Kelly told him about the murders. If, however, they had met on the Sunday, it would have been more likely for that conversation to have occurred on that day rather than on the Monday. Secondly, part of Graham's second statement has Kelly make reference to the Sunday meeting: how he had met Connolly and Northam at the Star on that day, and arranged to return the coat, and called Connolly a yellow bastard. The link between a vulnerable part of Northam's and Dickson's evidence and material contained in Graham's second statement (made of course in November 1949 and thus after the committal proceedings) magnifies the potential consequences of the non-disclosures on the fairness of the trial and the safety of Kelly's conviction.

Kelly's evidence

56. Kelly gave evidence in his defence at both trials. He said that on the evening of 19 March he was in the Coach and Horses (some distance from the Beehive) with a man called Skelly, and later had been in the Leigh Arms, then at the Spofforth Hotel, and then back to the Leigh Arms and was there at the time of the murders and until closing time. The licensee of the Leigh Arms, Frederick Thomalla, who was called by the defence, said he saw Kelly, who was a regular, enter the premises at 9 pm, without a coat or a hat. The licensee of the Spofforth Hotel, a Mr Ellis, another prosecution witness, said he saw Kelly in his pub between 9.15 and 9.25 pm, again without coat or hat. Another prosecution witness, the barmaid Edna Bore, saw Kelly at the bar of the Leigh Arms at 9.45 pm. The clock at the Leigh Arms was set ten minutes fast. Kelly called two brothers, Reginald and Walter Brampton, to say that he was back at the Leigh Arms by about 9.30 pm.
57. The prosecution case was that Kelly had sufficient time to obtain the overcoat and hat from wherever he had stowed them, commit the murders, discard the coat (and put it or hide it somewhere safely, for on the prosecution case it was returned to Northam), and dispose of the gun in the park, some distance away, and arrive back at the Leigh Arms by 9.45 pm.
58. In this connection the evidence of Mr Thomalla, the licensee, was important, for he said he saw Kelly at the Leigh Arms both before and after the murders. Mr Thomalla was a former police officer, who had given a statement to CI Balmer. At the retrial he said that the second time he saw Kelly was at 9.50 pm in the buffet, again in a suit, his usual clothes, not a coat. He was cross-examined by Mr Gorman for the Crown. Mr Gorman put to him his signed statement, which read: "I am quite definite that George Kelly was not in my house at 9.30 p.m. on the night of the Cameo murder, but he was in the buffet at ten to 10." Mr Thomalla told Mr Gorman that he "could not say it is true" and denied making that statement to CI Balmer. He said that what he had told CI Balmer, when he was asked by him if Kelly was in his pub at 9.30, was "Not to my knowledge". Mr Gorman pursued the point:

"Q. Why did you sign it? A. At the moment I signed this I was in a motor car, and I did not notice the word "definitely".

Q. What did you, a police officer of 23 years' service, think you were signing? A. I signed that statement going along in a motor car.

Q. Do you say you did not read it? A. I did not read it.

Q. When did you decide that he may have been in at half past 9? A. When I told Mr. Balmer.

Q. Then why did you not say to Mr Balmer: "I am not sure about it; he may have been in or he may not have been "? A. When I signed that I did not read it. I thought Mr. Balmer had put in "to my knowledge"."

59. The strange discrepancy that has now emerged by reason of the material which has come forward as a result of the CCRC references, to which Mr Pownall himself has drawn attention in his skeleton, is that in a draft unsigned statement for Mr Thomalla dated 4 October 1949 which CI Balmer has referred to in and attached to his report dated 12 October 1949, Mr Thomalla is represented as saying not what was in his signed statement but the following:

"Although I saw Kelly about 9.0 p.m. and again just before closing time, I cannot say whether or not he was in the house between those times. It would have been easy for him to leave between those times and I certainly did not serve him with any other drink" (emphasis added).

60. We are not sure whether Mr Thomalla made two statements and stuck to his first rather than his second (as Mr Pownall submitted), or whether Mr Thomalla's remarks to the police were first drafted in one form and then drafted in another, in the form which he ultimately signed – in a motor-car. We would rather infer from the evidence which he gave at trial and the material we have seen (and we have not seen any signed statement) that the latter is what happened.

CI Balmer's role

61. The death of CI Balmer and thus the unavailability of any new evidence from him, or indeed from any other member of the police team investigating the Cameo murders, concerning the matters on which the appellants rely render the task of this court in assessing his role in the investigations and at trial difficult and troubling. The Crown accepts that on any view of the reason for the statements' non-disclosure, even mere inadvertence, Kelly's conviction is unsafe. It does appear, however, that, in addition to the non-disclosures, CI Balmer's evidence was tailored to conceal both the fact that he had in September 1949 taken Graham's first statement and the fact that Northam and Dickson had changed their account in relation to Sunday 20 March. Moreover, both Graham and Northam similarly concealed the same respective vulnerabilities concerning their evidence, apparently confident that the truth would not emerge.
62. As the Crown acknowledged, the integrity of CI Balmer was of paramount importance in the case. His evidence challenged Kelly's alibi at several points, such as his account of what Kelly had said to him, when he visited him at his home on 20 March, about his whereabouts the previous evening. CI Balmer's evidence was in turn challenged in cross-examination by Miss Heilbron. In this connection the discrepancy about Mr Thomalla's statement is again relevant. CI Balmer was concerned with the close timings of the evidence about Kelly's alibi and referred to them in his report of 26 October. He there pointed out that the Leigh Arms was an eight minute walk from the Cameo, and he discussed the timings of various sightings of Kelly at that pub.
63. In the circumstances we agree with Mr Pownall's acceptance on behalf of the Crown that the matters identified before us must raise at the very least the strong possibility that proper disclosure might have undermined not merely Graham's evidence, but also the integrity of CI Balmer which was itself central to the prosecution's case.

The splitting of the retrials

64. The joint trial ended on 28 January 1950 without any verdicts. On 30 January The Echo published a report that the retrial would take place at the next assizes commencing on 31 January. On that day Miss Heilbron, for Kelly, appeared before Oliver J, who had conducted the joint trial, to make an application and was referred to Cassels J. It is clear from the transcript of the hearing that day before Cassels J that an administrative decision had already been made that the retrial would be split for the convenience of counsel. It would seem that this decision was occasioned by the Parliamentary duties of Mr Basil Nield KC, as he then was, who had defended Connolly. Thus Miss Heilbron had to apply for a joint trial, objecting to the decision that had already been taken. She began:

"I understand that this case is to be re-tried separately, that is to say, Kelly is to be tried first. My first point, on behalf of Kelly, is that I object to the re-trial separately."

65. Among the reasons she cited was reliance on the authority of R v. Grondkowski and Malinowski [1946] KB 369. The judge asked Connolly's junior counsel then in court, Mr Clover, what his attitude to Miss Heilbron's application was. Mr Clover said he had had no notice of it and no specific instructions. The Crown's counsel, Mr Blackledge, said he had nothing to say on the subject. The judge said he would hear the matter debated on the first day of trial, which he was prepared to postpone until 2 February.
66. On that day, in the presence of both accused, Mr Edmund Rowson KC, who in the meantime had replaced Mr Nield as Connolly's leading counsel, asked for a separate trial. Miss Heilbron, for Kelly, asked that the retrial should be a joint one, like the first trial, and developed her reasons. Mr Gorman, for the Crown, merely said that it was a matter for the judge's discretion "and I do not think it proper to say one word as to the exercise of that discretion", but, in answer to the judge, he said he offered no objection to a separate trial. Mr Rowson did not volunteer, and was not asked to advance, any reasons for his request for a separate trial.
67. The judge said:

"The Court has considered the possibility of there being a separate trial in this case. Objection has been taken to that by learned Counsel appearing for the defence of the prisoner George Kelly. Learned Counsel appearing for the defence of the defendant Charles Connolly has formally applied for a separate trial, and learned Counsel for the prosecution says that he takes no part in the matter other than to say that the matter is one for the exercise of judicial discretion.

"In the exercise of that judicial discretion, I direct that there shall be in this case a separate trial, and that the first of the defendants to be tried shall be George Kelly. A separate trial provides a plain issue for the jury concerning the one person to be tried. So far as I read the depositions, they seem to disclose that the two cases for the Prosecution differ in fact and in law in some important respects. A separate trial eliminates any evidence which is not directly against the person tried. I have come to the conclusion that it is in the interests of justice that the jury, in a long case like this, particularly on a re-trial and on a capital charge, should not have to dissect the evidence of individual witnesses and relate it to more than one person under trial. I do not think that the Defence is prejudiced, but, on the contrary, it may well be favoured by a separate trial, in that there will be before the jury no evidence other than that relevant to the issue being tried concerning one prisoner. Witnesses, whether for the prosecution or Defence, will be cross-examined once only, and the jury's task and recollection will be simplified. I therefore direct, as I stated at the beginning, that there shall be a separate trial, and that that of George Kelly shall be taken first."

68. On these appeals, Mr Pownall for the Crown does not seek to support that decision as being a proper exercise of the judge's discretion but acknowledges that no proper basis existed for the trials of Kelly and Connolly being severed within five days of a jury failing to reach verdicts against either. We think that concession to be rightly made. In Grondkowski and Malinowski, which, although cited by Miss Heilbron, the judge did not address, Lord Goddard LCJ said at 371:

"Prima facie it appears to the court that where the essence of the case is that the prisoners were engaged on a common enterprise, it is obviously right and proper that they should be jointly indicted and jointly tried, and in some cases it would be as much in the interest of the accused as of the prosecution that they should be."

69. That case involved cut-throat defences, but Lord Goddard went on to emphasise that even in such cases the interests of justice would usually favour a joint trial. The cases of Kelly and Connolly, however, did not involve cut-throat defences, but complete denials: and although there were some differences in the Crown's case against each, in particular that Connolly was only a secondary party and of course the Crown had to deal separately with the alibis of each, nevertheless in essence the Crown relied on the same evidence and principally on Northam, Dickson and Graham against both. The judge considered that Kelly would not be prejudiced and could even be favoured by separate trials, but it is hard to see the basis for either proposition. The jury would have heard Connolly's evidence as well as Kelly's, ie that not only was Connolly not a party to the murders, but he did not plan it with Kelly or go with him to the Cameo; and in general his evidence would have added his contradictions to the evidence of Northam, Dickson and Graham.
70. What is also disturbing is that it is clear that the original decision in favour of a split trial was ascribed to the convenience of counsel, and yet nothing was said about that by Cassels J in giving his reasons for the exercise of his discretion – even though when the issue came to be debated no reasons had been advanced on behalf of either the Crown or Connolly as to why there should be split trials. In his skeleton argument Mr Pownall accepted that "The reason for ordering separate trials though largely left unsaid in the course of argument appears to have been the fact that Connolly's Counsel Mr Nield was involved in Parliamentary business".
71. Mr Pownall nevertheless submitted that this error would not by itself have rendered the conviction of Kelly unsafe, and that it was, in itself, of no real weight. The test stated by Lord Goddard in Grondowski and Malinowski at 374 is:

"The real test, after all, which must be applied by a court of criminal appeal on a matter which is essentially one of discretion is, has the exercise of discretion resulted in a miscarriage of justice? If improper prejudice has been created whether by a separate or by a joint trial – for as we showed at an earlier stage of this judgment prejudice might be caused to one prisoner by ordering a separate trial on the application of the other – this court will interfere but not otherwise."

72. This issue was itself raised on Kelly's 1950 appeal, but rejected. Mr Pownall submits that nothing new has arisen since then to make it proper for this court to take a different view.
73. We see the force of that submission (even though a transcript of the judgment on appeal is not available); and, since we agree with the Crown's concession that Kelly's conviction is rendered unsafe by the non-disclosure of Graham's first statement (and, in conjunction with that, by the further non-disclosures of Northam's and Dickson's statements), it is not necessary for us to form a judgment about the independent effect of the erroneous splitting of the retrials. That, however, is subject to two further considerations. The first, is that we consider that the matter of the splitting of the retrials cannot be divorced from the new material which has emerged about the non-disclosure of the statements: and that viewed in that light, of which this court in 1950 did not have the benefit, the splitting of the trials becomes a material and substantial ingredient in our overall conclusion that Kelly did not have a fair trial and that his conviction is unsafe.
74. The second, is that on behalf of Connolly Mr William Waldron QC submitted to us that the decision to split the retrials was in truth conditioned not only on the convenience of counsel but on something further, namely an at least incipient discussion of a deal between the Crown and Connolly's counsel that, if he would plead guilty to robbery, he would be acquitted of murder and receive a sentence of ten years. Indeed, as part of his submissions in relation to severance, Mr Newman QC, on behalf of Kelly, had submitted that there was an irresistible inference that the essence of such an accommodation had already been reached between the Crown and Connolly's lawyers. Although the severance of the retrials was not a separate ground of appeal in the case of Connolly's appeal, as it was in Kelly's case, nevertheless as oral argument progressed it became clear that on analysis Connolly's ground of appeal to the effect that his plea of guilty was obtained by undue pressure incorporated a complaint that part at least of that pressure was brought to bear on him by reason of the improper splitting off of his retrial. Mr Pownall, on the other hand, submitted that the splitting of the retrials and Connolly's guilty plea were separate matters, and that just as the latter was unequivocal and voluntary, so the former must have been fully discussed with and authorised by Connolly: and in this connection Mr Pownall referred to the facts that on 31 January Mr Clover said that he had no specific instructions from Connolly, while on 2 February Mr Rowson applied for a split trial in the presence of Connolly. The question, he argued, must have been discussed in the interim.
75. We therefore turn to the circumstances of Connolly's plea at his retrial.

Connolly's guilty plea

76. Connolly had pleaded not guilty at the joint trial and given evidence in his defence. How did he come to plead guilty to new charges of robbery and conspiracy to rob? Inevitably, the material on this issue now before this court is new. In principle all of it could of course have been before the trial judge. In practice, however, such material inevitably emerges only after the event. In the present case, that material largely consists of the transcripts made from the taped interviews given by Connolly in 1993 and 1994. There is also some correspondence with his trial solicitor, Mr Maxwell-Brown, and an unsigned record of an interview with Mr Maxwell-Brown conducted by Detective Superintendent Semple at the former's home on 17 January 2003. DS Semple was in charge of the police enquiries which were necessitated by the CCRC references. Mr Maxwell-Brown is now very old and the Crown did not require his attendance at court. Everyone else who might have been able to assist is now dead. Mr Pownall did not formally object to admitting the transcripts, but he submitted, we think rightly, that they should be viewed with the greatest circumspection. He does, however, positively rely on the Maxwell-Brown material.
77. On 18 December 1991, which is after Mr Santangeli had taken an interest in his case, Connolly wrote to Mr Maxwell-Brown asking him to confirm that "I was advised by Mr Rowson KC to plead guilty". On 30 December 1991 Mr Maxwell-Brown replied that he was unable to do so and continued:

"After Kelly had been found guilty by the jury on his re-trial for murder, Counsel for the Crown informed Mr Rowson that the Prosecution would be prepared to seek the approval of the trial Judge, to the addition to the indictment against you of a charge of robbery. Also, if you were prepared to plead guilty to the lesser charge, the Court would be asked to consent to the murder charge being withdrawn. We were informed that this proposal had been mentioned to the Judge who had indicated his approval of the matter proceedings in that way.

"Mr Rowson, Junior Counsel and myself immediately arranged to see you at Walton and what had been proposed was explained to you. No advice as to how you should plead to either charge was sought by you or offered by Counsel nor, I have no doubt whatsoever, would experienced Counsel, such as Mr Rowson, even consider doing so."

78. There the matter rested, for Mr Maxwell-Brown ended his letter by saying that he regarded the correspondence as closed, until Connolly's taped interviews. These were extensive and dealt with many matters, but we refer to them only for what Connolly there said about the circumstances of his plea. In the first of them, held on 26 August 1993, Connolly said that after the end of the first trial he did not see Mr Nield again; and that he did not see his new silk, Mr Rowson, until two days before his retrial (which was due to begin on 13 February 1950). In the meantime, although he had been present in court when it was decided to split off his trial, he had not been consulted upon that: "I didn't have any say in the matter. I was just sitting there." Then, at the conference two days before his retrial Mr Rowson, Mr Clover and Mr Maxwell-Brown came to see him. Mr Rowson said he had no chance of an acquittal ("The evidence against him [ie Kelly] is against you…") and sought to persuade him to accept a deal, if one could be made, in order to save his life from the gallows. Otherwise his conviction would be just a formality. At first he refused, asserting his innocence, and then he said he would consult his parents (he was 26), but Mr Rowson said he could not do that, it was between them; and "Good God man…can't you see I'm trying to save your life?" He asked what he would get, and was told ten years. "I said how do you know I'll get ten years? He said I know you'll get ten years and that was it." Connolly said he'd think about it. In the end, he only finally made up his mind in the taxi going down to court. In court, he was surprised by having put to him an additional charge of conspiracy which Mr Rowson had not mentioned, and he said "Not guilty": but Mr Rowson turned to him in the dock and said it did not matter, it was a formality, and so he pleaded guilty to that as well. But he was, he said, completely innocent.
79. Over a year later, on 25 October 1994, Connolly was interviewed again, this time by a different journalist. Connolly's account of his sole consultation with Mr Rowson was the same; as was his description of his last minute decision and the mix-up over his plea to the charge of conspiracy (Mr Rowson: "it's immaterial, the sentence is still the same"). But he also went into further detail about his motivation. He was concerned to be reassured that a plea of guilty would not affect Kelly's appeal. He also said – "I think it's only fair to say that the strongest instinct in man prevailed and that's to save my own life." He told his mother about it and she said "do what you want but I would sooner have you alive than dead". But he had spent the rest of his life telling people of his innocence.
80. He was also taken again to the argument about the split trial. He confirmed that he had no say in it all. In the cells, he asked Mr Clover why, if there were to be separate trials, he could not go first, and was told the argument was over.
81. He was asked about Mr Rowson's plea in mitigation. The transcript of 12 February 1950 shows the following:

"The difficulty of Connolly became obvious from the moment that a jury had accepted, as one must assume they had accepted, the evidence of Northam, Dickson and Graham. In those circumstances, my Lord, it was obvious to anyone that there would be some difficulty in persuading a jury, as was attempted at the first trial of Connolly, to accept his evidence of an alibi. In those circumstances, my Lord, he has, on my advice and those interested in the Defence in this case, pleaded guilty to the charge of robbery."

(Mr Pownall submitted that that passage did not amount to a statement by Mr Rowson that he had advised Connolly to plead guilty and so did not confirm Connolly's evidence that he had so advised him, and in any event did not confirm any degree of pressure.) Mr Rowson continued his mitigation by reference to those passages in the evidence of Northam, Dickson and Connolly which would permit the submission that Connolly's role was a subsidiary one and that

"he had not the slightest intention or knowledge that violence was going to be used. No doubt the fact that all that evidence was given by those three has induced the Prosecution in this case to accept the plea of not guilty".

82. So far there was nothing in Mr Rowson's mitigation, over and above the plea of guilty itself, to compromise Connolly's case that he only accepted a plea of guilty to robbery under undue pressure. Towards the end of the mitigation, however, there was this passage:

"To a very great extent I venture to think, my Lord, that the events with which he was mixed up were not really the outcome of his own desire in any way, and he does, through me, for what it is worth, express his very deep regret for having been mixed up in this matter. It is a regret that is not expressed here for the first time, because he expressed it to Graham in Walton Gaol when he told him that he had been a fool to be mixed up with the job."

83. It was this passage which Mr Pownall relied on to submit that Connolly's plea was an entirely voluntary one and that at that time he fully accepted his guilt as a participant to robbery. In his second interview Connolly was asked about this passage. Connolly answered the question rather obliquely by reference to an earlier incident in his life when he struck a doorman at a dance because he would not let him in. He pleaded guilty, and received a fine and a suspended sentence. Mr Rowson referred to this previous conviction in the course of cross-examination of the police officer who was dealing with Connolly's record. He suggested to the officer that the assault on the doorman had happened because Connolly was drunk: to which the officer merely replied, "Possibly". In his interview Connolly said:

"…I was dance crazy you know – he said it was a bit too late you can't come. I said I've paid to come in. One word led to another and I hit him…Mr Rowson brought this up, said the case of Connolly's past of an assault to this doorman was when he was drunk and had no control over…I didn't say…that's what I'm trying to get over to you, the same thing as he's saying, what you just repeated now."

84. This is a little obscure, but we interpret it as meaning that Connolly was denying that he had instructed Mr Rowson to apologise on his behalf, just as he was denying that on the earlier occasion he had struck the doorman in drink, as distinct from frustration at being denied access to a dance he had paid to go to.
85. In the light of Connolly's appeal Mr Maxwell-Brown was interviewed on 17 January 2003. DS Semple then wrote up a record of the interview and submitted it to Mr Maxwell-Brown for his comments. It was returned with manuscript amendments, but unsigned. In relevant part it reads as follows (the manuscript additions are italicised):

"Mr Maxwell-Brown remembers being approached by the prosecution the day before the trial date regarding Connolly's plea. He stated that he would be pleading not guilty. He told counsel about the approach that had been made but was then not involved in the subsequent talks between prosecution and defence counsel and the presiding judge.

"He was aware that there was talk of a deal but that the judge was not happy initially and counsel made a second approach to the judge…Finally an agreement was reached and Mr Maxwell-Brown was informed that an alternative charge of robbery would be put on the indictment. He and defence counsel, Mr Rowson and Mr Gordon Clover, went to see Connolly and explained the situation. There was very little time, he thinks it may have been that same afternoon.

"Connolly had been aware that Kelly had been found guilty of murder and was reminded of this fact by Mr Rowson and Mr Maxwell-Brown, who wished to ascertain what Connolly would plead. The situation in relation to the prosecution approach and a possible charge regarding robbery were explained to him. Connolly asked what he would get for the robbery charge and Mr Rowson informed that he would probably get ten years imprisonment. Connolly wanted to see his wife or family and Mr Rowson pointed out that it was he, not his wife, who was going to hang. He was repeatedly told that it was a decision, which he himself would have to make.

"So far as I can remember, the meeting lasted about half an hour and Connolly decided that he would plead guilty to the alternative charge…Mr Maxwell-Brown does not now remember anything specifically about a conspiracy charge, but has no doubt that it would have been mentioned at this meeting. Although Connolly was given 'advice', it was in the form of answers to his questions not that he was advised to plead guilty to robbery. Connolly never protested about pleading guilty to robbery. He was not shocked by the suggestion; he was paying attention and made his own decision. Having made his decision he appeared to be relieved."

86. Mr Pownall submitted that Mr Maxwell-Brown's recollections were inconsistent with and did not support Connolly's account. There are some differences, in particular in respect of a concluded decision, reached by the end of the meeting, with which Connolly was comfortable, but in the main we think that it confirms rather than contradicts the earlier account and, if it is inconsistent with anything, it is with Mr Maxwell-Brown's own letter of December 1991. Thus the meeting took place only a day or so before the day fixed for trial; a deal had already been reached with the Crown which Mr Rowson presented to Connolly at their conference; there was very little time (for a life and death decision); Mr Rowson had given firm advice, reminding Connolly of Kelly's conviction; Connolly did want to consult his family but was reminded in graphic terms by Mr Rowson of the consequences of a wrong decision; ten years were spoken of as the sentence he would receive; and nothing was said (or could be remembered as being said) about a charge of conspiracy.
87. Mr Pownall points out that there is nothing on the transcript of Connolly's plea of guilty which reflects his account of what happened when the conspiracy charge was put to him. But it is possible that the stenographer would not have heard Mr Rowson's comment to Connolly; and may well merely have recorded the final answer of guilty. The point is in any event not determinative.
88. In our judgment, for all that we have considered those transcripts with circumspection, the essential lines of Connolly's account of the meeting seem to be genuine. There was some debate before us about the extent to which it was proper for counsel to advise a defendant as to his plea: but in our view it is unrealistic to suppose that against the background of Kelly's conviction experienced counsel would have been doing his practical duty if he did not impress his client with his view of the likely consequences of a decision in one direction or another: and especially in a case where, without previously taking instructions from the client, a deal had already been reached with the prosecution. No doubt Mr Maxwell-Brown is right to say that Connolly was also told that ultimately the decision was his: but it is difficult to suppose that he was not urged to think of his own safety.
89. We are sceptical that the rather carefully phrased passage cited above from Mr Rowson's mitigation concerning Connolly's regret can be relied on as confirmation of Connolly's genuine acceptance of guilt. However, whether in the overall context of the case it can be said that Connolly suffered from undue pressure is a question we will come to below. But we are satisfied that he received advice in the strongest terms that he should accept the deal that was on the table. Mr Rowson himself told the judge that Connolly pleaded guilty "on my advice".
90. Mr Maxwell-Brown said nothing to suggest that the availability of the deal had been discussed at some earlier time than a day or so before Connolly's retrial. Nevertheless, Mr Waldron submitted that the circumstances in which Connolly's retrial was split off points towards the probability that there had already been some discussion between lawyers at some level as to a possible deal; and Mr Newman had put a similar submission even more strongly. Mr Newman pointed out, correctly, that the application was for a split trial, not for an adjournment to accommodate Mr Rowson's late entrance. The decision to split the trials is sufficiently strange to give some force to those submissions; but it still remains possible that that decision may well be sufficiently explained by the need to replace Mr Nield as Connolly's leading counsel: there appears to have been an unspoken understanding, which emerged explicitly in the judge's decision, that if the trials were to be split Kelly's would go first. We are not satisfied that any discussion of a deal with respect to Connolly's role did enter into the judge's decision to split off and put back his retrial. Whether it entered into the Crown's or Mr Rowson's calculations is impossible to say.
91. In the circumstances we do not think it matters very much whether instructions were received from Connolly authorising the splitting of the retrials. He had lost Mr Nield, in whom he had, as he said in his taped interviews, reposed trust. We accept that he had yet to meet Mr Nield's replacement, Mr Rowson. The initial decision to split the trials had been taken over his head and without his instructions, as Mr Clover's comment at the hearing of 31 January 1950 makes clear. The reasons for Mr Rowson's formal application on 2 February 1950 for a split trial were never articulated. In the circumstances they remain speculative. Was it to enable a new leader to master the brief? If so, then Connolly must have felt that he was left in the lurch. Was it because there was a calculation that if Kelly were acquitted, the Crown would drop its case against Connolly, whereas if Kelly were convicted, a deal could be made, as in the end it was? If so, we are satisfied on both Connolly's and Mr Maxwell-Brown's accounts that that was not explained to him. We are therefore inclined to accept that Connolly's account that he was never consulted is at the very least a realistic possibility. Mr Maxwell-Brown did not say – of course he may not have been asked – that Connolly was consulted. Connolly's statement that, after the judge had made his decision to split the trials, he asked Mr Clover if his trial could go first, is inconsistent with him being told that the need to defer his trial was due to the change in counsel. He was not told of the possibility of a deal. But even if instructions to split the trial were sought, that must have happened in a most perfunctory way, for what could he be told? In the circumstances he must simply have been prepared to follow advice, the reasoning of which remains obscure. There is something deeply unsatisfactory about the difficulty of resolving such issues after all this time, when all the leading actors are dead or unavailable for examination. What remains, however, is this: that, as a result of a ruling to split the trials which in our judgment, as well as on the Crown's concession, should not have been made, Connolly was left shortly before his retrial with a life and death decision whether to accept a deal, negotiated without his involvement, which involved his pleading guilty to robbery.

Kelly's 1950 appeal

92. Kelly's appeal was heard on 6 March 1950 before Lord Goddard CJ. There is no transcript available in respect of the proceedings on that day, but, in addition to grounds already lodged Miss Heilbron raised a new point, of which she had only just learned, regarding the possible disqualified status of a member of the jury. The appeal was therefore adjourned to 10 March for enquiries to be made. On that day the appeal in relation to the lodged grounds of appeal was dismissed: as stated above, one of those grounds related to the splitting of the retrials, but the others related to the summing-up. No transcript exists of the judgment, but a press report in The Echo says that Lord Goddard described the appeal as hopeless. Then, on 15 March a separate judgment was delivered, which does exist, dealing with the point about the status of the jury. The submission that the trial was a nullity was rejected.

The law in relation to appeals from old convictions

93. It has been established in R v. Bentley [2001] 1 Cr App R 21 at para 4 (per Lord Bingham of Cornhill CJ) that in reviewing the safety of old convictions pursuant to references from the CCRC –

"The conduct of the trial and the direction of the jury must be judged according to the standards which we would now apply in any other appeal under section 1 of the 1968 Act."

See also R v. King [2000] 2 Cr App R 391 at 402, where Lord Bingham again stressed that this court is concerned only with the safety of the conviction.

94. In this case, however, it is not controversial that even in 1950, let alone today, the requirements of a fair trial required the disclosure of Graham's first statement and the Northam and Dickson statements: see R v. Ward (1993) 96 Cr App R 1 at 23/25, which discusses the position at any rate 30 years ago in 1974, and particularly this passage at 25:

"To return, however, to the position in 1974, Mr Mansfield submits, rightly, that paragraphs 443 and 443a of Archbold (38^th ed.) were by no means exhaustive. They were merely aspects of the defendant's elementary common law right to a fair trial which depends on the observance by the prosecution, no less than the court, of the rules of natural justice. No authority is needed for this proposition but it is illustrated by the decision of the Divisional Court in Leyland Justices, ex p. Hawthorn [1979] Q.B. 283. On the broad basis of this right, the defendant is plainly entitled (subject to statutory limitations on disclosure, and the possibility of public interest immunity, which we discuss below) to be supplied with police evidence of all relevant interviews with him. We would adopt the words of Lawton L.J. in Hennessey (1979) 68 Cr. App. R. 419, 426, where he said that the courts must,

"keep in mind that those who prepare and conduct prosecutions owe a duty to the Courts to ensure that all relevant evidence of help to an accused is either led by them or made available to the defence …" "

The new evidence

95. We have received and already discussed in our judgment above a body of new evidence, which includes the undisclosed statements themselves, the police reports of the 1949 investigations, the statement of Mr Santangeli, the transcripts of Connolly's taped interviews (together with statements from the journalists who conducted them), the Maxwell-Brown material, and statements from the appellants' families. The Crown does not oppose the reception of any of this new evidence under the provisions of section 23(1) of the Criminal Appeal Act 1968, subject to its caution about the taped transcripts. Mr Pownall accepts Mr Santangeli as a witness of the highest integrity.

Kelly's appeal

96. We are now in a position to revert to our reasons for concluding that Kelly's conviction is unsafe and thus for allowing his appeal.
97. The non-disclosure of Graham's first statement, and to a lesser but cumulative extent the non-disclosure of Northam's and Dickson's statements, deprived Kelly and his lawyers of highly relevant material. Subject only to the fact that Miss Heilbron was plainly aware of the existence of at least some of the latter statements (and of the fact that the lending of the overcoat was not mentioned until 10 October), and thus subject to the possibility that the non-disclosures were simply part of a local culture, there is every sign that those non-disclosures, and particularly that of Graham's statement, were due to deliberate concealment. Indeed, the lies of Graham and CI Balmer in relation to when they first saw one another in relation to the Cameo murders are a strong indication of the need for concealment. The similar lies in relation to the time when the Sunday meeting at the White Star was first revealed to CI Balmer suggest the same conclusion in respect to the statements of Northam and Dickson. Disclosure of the statements would have enabled Miss Heilbron to go further than she was able in attacking the credibility of all three principal witnesses for the Crown.
98. Did the non-disclosures make the conviction of Kelly unsafe? In our judgment, they did. The evidence against Kelly was, in Mr Pownall's words, far from overwhelming. There was no forensic or scientific evidence to link Kelly with the murders. He was not identified at the scene. Northam and Dickson were rightly treated as accomplices. The jury were therefore cautioned against convicting without corroboration. The only corroboration they were offered apart from Graham's evidence was the coat: it is true that the gunman was seen wearing a belted overcoat, but the only thing to connect Kelly with such a coat was Northam's and Dickson's evidence. Outside that evidence he was never seen in hat or coat. The essential corroboration presented to the jury was Graham's evidence. It was with that that the judge concluded his summing-up – as well as with the point from Northam's evidence that Kelly had said that the assistant manager was "on his knees" when he shot him again. Unwittingly, however: for that point, which must have seemed so powerful at trial, would have been revealed by disclosure of Graham's first statement to have been more than double-edged.
99. On the prosecution's own case, Kelly's opportunity to commit the murders was very narrow. He was at the Spofforth Hotel as late as 9.25 pm. He was in the Leigh Arms at 9.45 pm. The shooting was at around 9.35 pm. Kelly had to obtain the hat and coat and go to the Cameo. There was a side door to the cinema at the bottom of a spiral staircase from which the gunman escaped, but that could not be opened from outside. The cut telephone wires were at the bottom of the spiral staircase. They must have been cut in advance of the shooting. The gunman had difficulty getting out of the manager's office because the lock was shot off. He had to get rid of the gun (on the prosecution's case, in the lake in the park), stow away the hat and coat, and return to the Leigh Arms.
100. The prosecution itself did not regard its case as strong. A minute in the DPP file dated 13 October 1949, just before the committal proceedings, says "It is not a strong case." A letter dated 14 October to the DPP from the prosecuting solicitor says: "You will probably agree that the evidence is not very strong." After the committal a report dated 22 October says: "A strong prima facie case has been established against both accused, but it will require to be well supported at trial". It was supported by Graham's second statement, but to the exclusion of his first. After Kelly's conviction, the deputy DPP wrote (see above) that in his opinion that conviction would not have happened but for Graham's evidence. The Crown accepts that opinion as correct. Northam's and Dickson's evidence suffered from the deficiencies which Miss Heilbron was able to elucidate at trial (the ambivalent letter, the failure to mention anything prior to 29 September, the failure to mention the lending of the overcoat until 10 October) but in addition the concealed failure to mention the meeting on Sunday until the committal proceedings. Thus Graham's evidence was at the forefront of Mr Gorman's powerful speech for the prosecution.
101. However, if Graham's first statement had been disclosed, it is difficult to think that his evidence would have had the impact which Mr Gorman obviously believed it forensically deserved. It is not impossible that Graham would never have been called. If the statement had been disclosed between the first trial and the retrial as it should have been, all the more so in that CI Balmer in his evidence at the first trial had lied about meeting Graham for the first time in the case on 19 September (Graham had similarly lied about the date of his first meeting with CI Balmer), it is not impossible that the retrial would not have taken place.
102. In all these circumstances we consider that Kelly did not have a fair trial, his conviction is unsafe and must be quashed, and the appeal constituted by this reference must be allowed. We would merely add as a postscript, because no point was taken on it at these appeals, that the judge gave no direction to the jury as to how they should view Graham as a witness, other than that he may be regarded "as of importance" and that his evidence could corroborate that of Northam and Dickson. There was no warning of caution in that he may have been seeking an advantage for himself. In this connection, see now Benedetto v The Queen [2003] UKPC 27, [2003] 1 WLR 1545, in particular at paras 32/33.

The law in relation to appeals from pleas of guilty

103. It follows that if the retrials had not been severed, and if both Kelly and Connolly had been convicted at their retrial, Connolly's conviction would have been quashed for the same reasons. Does it matter that Connolly pleaded guilty to the new charges of robbery and conspiracy to rob? Mr Pownall submits that it does and that Connolly's conviction is safely based upon his own unequivocal plea.
104. This is not because a plea of guilty in law prevents a finding on appeal that the conviction is unsafe. Mr Pownall acknowledges that that is so, and that "conviction" in section 2(1) of the Criminal Appeal Act 1968 includes a conviction upon a plea of guilty. Thus a plea of guilty is not necessarily a bar to an appeal against conviction: DPP v. Shannon [1975] AC 717, 59 Cr App R 250. But he submits that there was nothing in the facts to remove Connolly's case from that of a voluntary plea. Similarly, Mr Waldron acknowledges that the existence of the death penalty at that time, being sanctioned by law, could not by itself render Connolly's plea one obtained by duress. But he submits that it is relevant as a factor to be taken into account when attention is focussed on those matters which affected his decision and were not in accordance with law.
105. In R v. Forde [1923] KB 400 Avory J said at 403 that –

"A plea of Guilty having been recorded, this Court can only entertain an appeal against conviction if it appears (1.) that the appellant did not appreciate the nature of the charge or did not intend to admit he was guilty of it, or (2.) that upon the admitted facts he could not in law have been convicted of the offence charged."

Subsequent authority has shown, however, that that is too narrow a formulation.

106. In R v. Peace [1976] Crim LR 119 it was held that an accused who pleaded guilty following and adopting the advice of his counsel, albeit unhappily and with reluctance, could not be said to have lost his power to make a voluntary and deliberate choice and thus say that his plea amounted to a nullity. An appeal could only lie where it had been entered under pressure or threats or the like in circumstances in which the defendant had no free choice but was driven to adopt a certain course whether he liked it or not. It would be a serious matter if it was accepted as a principle that where counsel gave strong advice to a client indicating the prospect of his being found guilty as an alternative to pleading guilty, it was then to be said that the plea was not really his own but something which had been forced on him. It was a question of fact in every case.
107. R v. Turner [1970] 2 QB 321, 54 Cr App R 352 is the leading authority on plea bargaining. It held that counsel must be free to do his duty of giving the accused the best advice he can and if need be in strong terms, but will emphasise that he should not plead guilty unless he has committed the acts constituting the offence charged.
108. R v. Inns (1975) 60 Cr App R 231 is an example of a defendant adjudged not to have a free choice. There the pressure was put on him by his counsel's report of conversations with the judge. Lawton LJ said (at 233):

"When the accused is making a plea of guilty under pressure and threats, he does not make a free plea and the trial starts without there being a proper plea at all."

109. R v. Hunt [1986] QB 125, 82 Cr App R 244 concerned the case of a plea of guilty following an erroneous ruling by a judge on a point of law, there the meaning of "controlled drug". This court held that the judge had given a correct ruling, but on erroneous grounds. At 132, 249 Robert Goff LJ said –

"Since the applicant's plea of guilty was founded upon that ruling, we shall grant the applicant leave to appeal from his conviction."

110. However, the appeal was dismissed because, although had the judge's grounds been correctly formulated the accused could in theory have adduced relevant evidence on the issue of his guilt, in practice he would not have done. Therefore his conviction would have been inevitable.
111. In R v. Preston (1992) 95 Cr App R 355 this court adopted Robert Goff LJ's test that a plea of guilty following an erroneous ruling of law must be founded on that ruling. Woolf LF (at 381) referred to Viscount Dilhorne's dictum in Shannon at 757 and 264 that –

"the respondent having pleaded guilty, [the court of appeal] had only power to quash the conviction if either there had been a wrong decision on a question of law or a material irregularity in the course of trial"

and continued:

"It is not any error of law or any irregularity which is sufficient for this purpose…In our judgment before an appellant who has pleaded guilty can rely upon an erroneous ruling on a point of law or a material irregularity, he must show that his plea "was founded" upon the erroneous ruling of law or material irregularity."

112. Woolf LJ then went on to distinguish between a situation where a plea was influenced by the judge's rulings and a situation where the defects in the rulings were critical. In that court's opinion what was critical was a separate decision by the judge to admit certain evidence, as a result of which "Preston had no prospect of successfully contesting his guilt" (ibid).
113. We now come to the first of a number of recent authorities which are of particular relevance to Connolly's appeal because they concern the effect of non-disclosure on pleas of guilty. R v. Schlesinger [1995] CLR 137 arose out of the background facts to the Scott Inquiry about arms to Iraq. In 1985 the defendants had been convicted of illegally exporting arms to Iraq. In the preparations for trial the defendants had been told that the arms they had supplied, although paid for by Iraq, were really presents for Sudan and Jordan. They were assured that they would get witnesses from the embassies of those countries in Britain to support that account, but then learned that authority for such witnesses to give that evidence had been withdrawn. What they did not know, and was only revealed as a result of the Inquiry, was that the embassies had been approached on the initiative of the Customs and the Foreign Office (in the belief that the account of the arms really being presents to Sudan and Jordan was false) to scotch any possibility of such statements. Ignorant of this non-disclosure, the defendants pleaded guilty, assisted in that decision by being told that they did not face a prison sentence in any event. It was conceded for the Crown that what had occurred was an indefensible abuse of process, but nevertheless the appeals were resisted. It was submitted for the appellants that the prosecution would have been stayed if the court had known the truth of these matters. This court agreed that was so, and that the convictions should be set aside irrespective of prejudice; but in any event on the ground of prejudice. The digest of the report states (at 139):

"Whilst the Court was usually slow to set aside pleas of guilty which had been unequivocally made, where they had been made in ignorance of malpractice having operated to a defendant, different considerations might apply."

114. The brief comment by the late Professor Sir John Smith QC reads:

"The appellants had pleaded guilty but they would, apparently, not have done so if the witnesses whom they hoped to call had been available. Guilty or not, their right to have the charge against them proved beyond reasonable doubt had been, at least, impaired. It seems there was really no need to rely on the extension of the concept of abuse of process made in the Horseferry Road case. A defendant does not get a fair trial if he is precluded from calling witnesses whom he believes to be necessary to his defence."

115. R v. Blackledge (1996) 1 Cr App R 326 was another case arising out of the arms to Iraq affair. Licence to export the arms to Jordan had been obtained, and the defence was that the authorities knew that Jordan was being used as a conduit for exports to Iraq and were turning a blind eye. The defendants sought disclosure of policy and guideline documents governing the grant of export licences, but were refused by the trial judge, whereupon, after assurances from the prosecution that they would not be presenting the case as one of the utmost gravity and indications from the judge that suspended sentences could be expected, they changed their pleas to guilty. Subsequently, ministerial documents came to light substantiating the defendants' allegations. Their appeals were allowed. It was held that the failure to disclose was a material irregularity and that the guilty pleas had been founded on that irregularity. Lord Taylor CJ said this (at 338/339):

"It is submitted that the material irregularity created by the non-disclosure deprived the appellants of material which would have enabled them to make properly informed decisions as to what course they should take…Even without that documentation, the appellants could have run their defence and given evidence in support of it. It is stressed, however, that in addition to being deprived, so to speak, of ammunition, the appellants were put into a difficult dilemma. For the offences charged, they could have been sentenced to a substantial period of imprisonment. At the time of the trial, after the second Gulf War, they had an understandable fear as to what sentence a court might consider properly reflected public disapproval of assistance to Saddam Hussein. In those circumstances, the prospect of a muted presentation of the facts by the prosecution, followed by a suspended sentence, put pressure on the defendants to go quietly…We have considered the aggregate of all the unusual circumstances of this case – the material irregularity, the judge's ruling based on an unawareness by him and by prosecuting counsel of the undisclosed documents, and finally the pressure added to those factors by the discussions leading to the changes of plea. We consider the pleas of guilty to be "founded on" the material irregularity and the judge's ruling coupled with the pressure to which we have referred. In the result, we cannot regard the convictions as safe and satisfactory."

116. In R v. Togher [2001] 1 Cr App R 33 the appellants were indicted for importing drugs on two separate indictments (the "Frugal" and "Madrid" indictments). They were convicted on the first (Frugal) and pleaded guilty on the second (Madrid). Their appeal against their conviction on the first indictment was allowed and a retrial ultimately stayed on the ground that the prosecution had at the first trial by reason of non-disclosure deprived the defence of its ability to mount a challenge to the integrity of the prosecution. They then appealed against their conviction on the second indictment, but failed. The case is principally authority for the proposition that a broad approach (as per R v. Mullen [1999] 2 Cr App R 143, [2000] QB 520) should be adopted to the statutory test of safety so as to embrace not only circumstances where the conviction might be wrong but also where the appellant had been deprived of a fair trial. In applying that broad test, this court nevertheless concluded that the retrial judge had been wrong to find that there had been any abuse of process requiring the staying of the retrial. Lord Woolf CJ concluded as follows (at paras 58/59):

"58…The defendants had not lost the ability "properly to defend themselves" at a retrial when they would be well aware of the failures on the part of the prosecution. The defendants could, if they were prepared to take the consequences of doing so, exploit the earlier non-disclosure to challenge the bona fides of the officers in charge of the prosecution "viewed as a single entity".

"59. The shortcomings on the part of the prosecution are not of the category of misconduct which would justify interfering with the defendants' freely entered pleas of guilty. We see this case as being in a wholly different category from the exceptional case Lord Lowry was considering in his speech in Bennett. When the appellants pleaded guilty they were not aware of the matters relied upon before Turner J. for obtaining a stay of the retrial, but they were aware that they were appealing against their conviction. They therefore should have appreciated that the appeal against their conviction might succeed. If this had happened they would still be bound by their pleas of guilty. They were never ignorant of any evidence which went directly to their innocence of guilt. They were only unaware of material which could, but for their pleas, have been used to attack the credibility of the prosecution witnesses. Ignorance of this kind does not justify reopening their pleas of guilty. While there was an irregularity in their trial on the Frugal indictment, the appellants' pleas to the Madrid indictment were not "founded on" and were independent of that irregularity."

117. Both parties relied on that authority. Mr Pownall submitted that the non-disclosures in this case were similar to those in Togher and that the last four sentences just cited are directly applicable. Mr Waldron, however, submitted that the ultimate test proposed is whether the misconduct complained of "would justify interfering with the defendants' freely entered pleas of guilty" and that in this case it clearly would. We would observe that it is clear from the facts discussed elsewhere in Lord Woolf's judgment that the non-disclosure in question related to the authorisation of covert surveillance work which had produced 57 tape-recordings which were disclosed in the course of the Frugal retrial. The argument was that, although the new material "did not relate to any issue of fact in the trial which had taken place" (at para 35), the possibility that the surveillance was not properly authorised could lead to an attack on the integrity of the investigation in a case where the defendants alleged that they had been framed. That was the context in which Lord Woolf said that the material of which the defendants were unaware did not go directly to their guilt.
118. R v. Bhatti (CACD, 19 December 2000, unreported) contains an extensive summary and analysis of the relevant authorities. The appellant pleaded guilty to causing death by dangerous driving on the basis of a police report that the cause of the accident was the combination of the driver's poor steering and harsh acceleration. A defence report said a component failure could not be ruled out. The car had been scrapped. The appellant was told that the police expert could not accept the opinions of the defence expert. The appellant pleaded guilty. Later the CPS informed the appellant that independent experts had reviewed the police expert's report and found it "totally unreliable and riddled with error". The appeal was allowed. Potter LJ said (at paras 30/33):

"30. However, when the appeal is in respect of a conviction following a plea of guilty the considerations which apply are very different and the circumstances in which it may be appropriate or proper to allow an appeal are of necessity very limited. That is because the safety of the conviction depends not on some legal error or procedural irregularity which has arisen in the course of the adversarial process of the trial, thereby leading to a verdict of guilty which might otherwise have been not guilty; it rests upon the question whether, and in what circumstances the Court should look behind the plea of guilty (which represents a voluntary recognition of guilt) and enter on an examination of the reasons or motives of the defendant in deciding so to plead. That in turn requires the Court to reach a decision based not upon objective matters of record, namely the procedures adopted and decisions reached openly in the course of the trial, but on the subjective recollection and subsequent account of the appellant and/or his advisers as to the reasons for his plea. Such a procedure is an inherently unsatisfactory basis for interference with an unequivocal plea, voluntarily made at the proper time, for reasons or motives which may be infinitely various. Of course, if the defendant can establish that he pleaded guilty without understanding the nature of the charge or otherwise without intending to admit his guilt of what was alleged, the conviction may be quashed; as may also be the case if the appellant can show circumstances which effectively deprived him of a free choice as to plea: see Archbold (2001) at paragraph 7-103 and cases there cited [see now Archbold (2003) at para 7-291]. In such cases the voluntary nature of the plea is vitiated.

"31. It will also sometimes be the case that an appellant can establish that, as a result of trial processes prior to his plea, such as an erroneous ruling on a point of law (see R v. Preston 95 Cr App R 355 at 381) the basis of the plea is fundamentally undermined. Similarly, where by reason of some act of deception or non-disclosure on the part of the prosecution the appellant has effectively no alternative but to plead guilty, when he would not otherwise have done so, his conviction may thereafter be set aside (see for instance R v. Schlesinger and others [1995] Crim LR 137 and R v. Blackledge and others [1998] 2 Cr App R 100). However, that will not be the case where, in the face of evidence bona fide presented by the prosecution, the accused has pleaded guilty on the basis of advice from counsel, however reluctantly accepted: see R v. Peace [1976] Crim LR 119. As made clear in R v. Boal (1992) 95 Cr App R 272, although a plea of guilty does not deprive the court of jurisdiction to hear an appeal against conviction, it is highly relevant to the issue whether the conviction is unsafe that the defendant knew what he is doing, intended to plead guilty, did so without equivocation and after receipt of expert advice…

"33…Thus once the defendant has pleaded guilty and been sentenced on the basis of his plea, it will only be in the rarest of cases that circumstances should be regarded as vitiating or undermining the voluntary nature of the plea to such an extent that the conviction should be regarded as unsafe. Certainly it will not in the ordinary way be sufficient to demonstrate that the evidence of a particular witness, which at the time appeared reliable, has been shown subsequently to be unreliable or untrue."

119. In the light of these principles Potter LJ turned to the facts of the case:

"35…As a result, we are satisfied that:

1. Had the Crown prior to the appellant's plea of guilty been aware of the deficiencies of PC Desmond's report the prosecution would not have proceeded.

2. Equally, and in any event, had the defence been aware that the report was not reliable, the appellant would not have entered a plea of guilty.

3. Had the deficiencies of the report emerged and had the Crown disclosed them at any stage prior to sentence, as we are satisfied would have been done had the truth been known, the defendant would have sought to vacate his plea of guilty.

4. The Crown would have supported the appellant in that application and, had it been successful, would have elected not to proceed.

5. If, contrary to what we are told would have happened, the Crown had proceeded, the defendant would have called his expert witness to attest to the possibility of mechanical failure in which event it is possible, and indeed probable, that the jury would have acquitted the defendant.

"36. In those circumstances, we are satisfied that the assumption of both the prosecution and the defence as to the reliability of PC Desmond's report as to the reasons for the accident and his dismissal of the suggestion of mechanical failure, and the subseq uent revelation of the worthlessness of that report and opinion fatally undermine the plea."

120. Mr Pownall submitted that items 1-4 in the citation of para 35 above did not apply in the present case.
121. Finally, in R v. Montague-Darlington [2003] EWCA Crim 1542 (unreported, CACD 23 May 2003) the appellant had pleaded guilty to importation of 90 packages of cocaine which she had swallowed. She was advised that she had a defence of duress, but chose to plead guilty. A year later the solicitor to the Customs wrote to her solicitors to inform them of material which had only recently come to the former's attention and to say that if it had been to hand in time he would have regarded it as disclosable but likely to attract public interest immunity. This court was satisfied that the material was disclosable but that rather than disclose it the prosecution would not have been commenced or proceeded with. Kennedy LJ said:

"7. This court will only rarely entertain an appeal against conviction where there has been a plea of guilty, but the circumstances in which an appeal may be successful are not confined to those identified by Avory J in Forde [1923] 2 KB 400, as is clear from the recent judgment in Togher [2001] 1 Cr App R 457. The court has to consider whether the appellant had a fair trial. It is difficult to see how the appellant can be said to have had a fair trial when it is now the case for the prosecution that she should not have been tried at all. Furthermore if, when advising in relation to plea, her legal advisers had access to all of the relevant material it is clear that she would have been strongly advised not to plead guilty, and there is no reason to think that she would have refused to accept that advice."

Connolly's appeal: discussion and conclusion

122. In the light of these authorities Mr Pownall submitted that Connolly's plea was voluntary, not born of undue pressure, and not founded on any material irregularity or error of law, and that his convictions therefore remain safe. In particular, he submitted that the severance of Connolly's trial was authorised by him; that Mr Maxwell-Brown's evidence demonstrates that the plea was voluntary; that Graham was not the only witness in the case; that even if Graham's statement had been disclosed, there would probably have been a trial in any event, just as the committal was achieved without Graham's evidence; that there was independent support for Northam's and Dickson's accounts in the facts that Connolly handed in his notice on 23 March 1949, applied for a free passage to Australia on 31 March 1949, and had lied to the police by saying that he had been at work on the day of the murders; whereas his alibi that he had been at a dance with his wife was capable of being subject to detailed attack.
123. However, we have already rejected a finding that Connolly was consulted about the severance of his trial: we consider that, to put it at its lowest, there was a realistic possibility that he was not. As for Mr Rowson's consultation with Connolly shortly before his trial, we have already found that Connolly's account is to be broadly accepted, and that in its essence it is supported by Mr Maxwell-Brown's evidence: indeed, given Mr Maxwell-Brown's earlier letter of December 1991, we consider that if there is any difference between the accounts we would prefer Connolly's. We regard his account of the consultation as having the ring of truth. We think that Mr Rowson went to the very limit of what his duty allowed in giving strong, realistic and practical, indeed life and death, advice: we doubt that any leader placed as he was would have acted differently, but it must have amounted to enormous, indeed irresistible, pressure to take the course which would preserve life rather than gamble it. We agree that there was nothing illegitimate in such advice in itself, and that the mere existence of capital punishment was part of the fabric of the then law and could not in itself constitute unlawful or undue pressure. However, that penalty, the availability of the deal, and the advice tendered by Connolly's lawyers (for we feel sure that no one at that consultation disagreed with Mr Rowson's advice) did not stand alone. The police had failed to provide even for prosecuting counsel, let alone for the defence, to have the statements of the three leading witnesses in the case. Even if the non-disclosure of Northam's and Dickson's statements may not have been enough by themselves to have rendered Kelly's conviction unsafe, a question we have not had to determine, it added materially to the vice of the non-disclosure of Graham's first statement, a matter which the Crown acknowledges by itself requires the quashing of Kelly's conviction. We consider that those non-disclosures meant that Kelly did not have a fair trial, and the severance of the retrials in the event compounded that unfairness. It was Kelly's conviction on a capital charge in an unfair trial that was the critical background to the advice tendered by Mr Rowson – who was of course ignorant of the non-disclosures – and to the deal negotiated with the prosecution which he held out to Connolly.
124. We do not think it wrong in these circumstances to say that Connolly's decision to plead guilty was founded on an unsafe conviction in an unfair trial. Both parties to the plea bargain negotiated on behalf of Connolly, and indeed the judge who had to approve that bargain, were acting on the basis that Kelly had been properly convicted in a fair trial. Although Kelly had his appeal, it is plain that, as matters were then known, that appeal was given no weight in the scale. It seems to us that the cases which are closest to the present are Schlesinger and Blackledge and, in a slightly different way, Bhatti. Those cases indicate that the concept of a plea of guilty being founded on an irregularity goes wider than the paradigm case of a plea which turns entirely on, say, a decision of law as to the width of a statute imposing criminal liability. The passage cited above from Blackledge in particular illustrates how, in a far less serious context, the dilemma faced by a defendant may be part of a combination of circumstances which effectively undermines or vitiates the essential condition of a plea being voluntary.
125. As for Togher, we consider that the circumstances there were very different. The Madrid indictment (concerning 33 kilos of cocaine seized by the Spanish at a hotel in Spain) related to a different consignment of drugs from that involved in the Frugal indictment (which concerned 309 kilos of cocaine seized in Sussex where the drugs had been brought on a boat called the Frugal). The non-disclosure did not even relate to any issue of fact in the Frugal indictment, let alone the Madrid indictment. The submission was merely that the new material, if known about in time, might have enabled the defendants to develop a collateral attack on the credibility of the investigation and, so it was merely assumed (at para 7), would have led the defendants to defend the Madrid indictment. There was no submission, as far as we can see, that the pleas were other than freely entered. That is why Lord Woolf began his concluding reasoning at para 59 with the premise that the defendants had freely chosen to enter pleas of guilty, when he said: "The shortcomings on the part of the prosecution are not of the category of misconduct which would justify interfering with the defendants' freely entered pleas of guilty". He had previously described those shortcomings as "regrettable muddle and confusion and incompetence" (at para 58). In those circumstances it is not surprising that the defendants' submission that the non-disclosure by itself somehow unlocked their voluntary acceptance of guilt failed. In the present case, however, the non-disclosure, in the case of Graham, went to the credibility of critical corroborative testimony which spoke directly of the defendants' own confessions, and, in the cases of Northam and Dickson, went to the credibility of the main prosecution witnesses whose testimony again spoke directly of the defendants' prior planning and subsequent confessions; and all in circumstances, unlikely to be repeated, where one defendant was forced by an unsupportable decision to sever the retrials to choose, following the conviction of the man he denied being his accomplice in guilt, between his desire to plead his innocence and his desire to save his life.
126. The test of a plea of guilty being "founded on" some material irregularity or error of law expresses a strong and determinative causal connection. The critical and fundamental causative part which the trial of Kelly played in the decision of Connolly can be illustrated by the consideration that if the retrial had not been wrongly severed, Connolly would have been tried with Kelly, and either suffered his fate, or acquitted on the basis on which the Crown were ultimately prepared to accept a plea of not guilty to murder. The critical part which Graham's evidence played in the trial of Kelly is demonstrated by the fact that if his first statement had been disclosed, it is entirely possible that he would never have been called, or if called discredited: in circumstances where the view expressed after the event was that his was the evidence without which Kelly's conviction would not have been secured. And if the disclosure had taken place between the first joint trial and Kelly's retrial, then in the light of the lies told at the first trial, it is possible that there may have been no second trial at all. Moreover, if immediately after Kelly's retrial the non-disclosures (and lies) had come to light and been understood as they have now been understood, then the Crown would then have been acknowledging that Kelly's conviction could not survive and it is highly unlikely that Connolly would have been advised in the way in which he was advised, or that he would have changed his plea to one of guilty (see Bhatti).
127. Ultimately, however, the test is of the safety of the conviction. For the reasons expressed in Bhatti the scope for finding that an unequivocal and intentional plea of guilty can lead to an unsafe conviction must be exceptional and rare. However, undue pressure or errors of law or unfairness in the trial process may all be of such an important causative impact on the decision to plead guilty that the conviction which follows on such a plea can, in an appropriate case, be described as unsafe. In our judgment such is this case. Ultimately, as the authorities emphasise, it is a question of fact in each case.
128. In the circumstances the factors raised by Mr Pownall in his final submission as supporting the safety of Connolly's conviction are really neither here nor there. They did not lead to a conviction at his first trial and were not tested in a retrial. If Kelly, the alleged gunman and thus primary party cannot be regarded as safely convicted, it is unrealistic to regard Connolly, an allegedly merely secondary party, as safely convicted once the basis for upholding his conviction on his own plea is fatally undermined. In certain respects, for instance the irrelevance in the case of Connolly of Northam's and Dickson's evidence about the coat and the absence of any evidence against Connolly of knowledge that the assistant manager had been shot while on his knees, the evidence against Connolly was weaker than that against Kelly. In any event, the matters raised by Mr Pownall are controversial. Thus, as to working on the Saturday: he had not worked on the Saturday, but he had on the Friday, and at the first trial Oliver J had said (The Echo report of the summing-up):

"That was on May 14 and the murder was 8 weeks before. The jury might think quite possibly that the man be honestly mistaken, when he said he was at work at Bibbys because he was in fact at work there up to March the 18^th…"

129. Similarly, as for planning to go to Australia, at the first trial evidence was given by Revd Thomas Frayne that Connolly had consulted him about leaving for Australia already in about May 1948. As for his alibi, Connolly's wife said that they had been out dancing together on Saturday 19 March: she was attacked in cross-examination on the basis that she had muddled the date and that she was thinking of some other Saturday: but she was able to prove that she had the date right by reference to a letter post-marked 18 March. In the end, the dispute about the alibi was as to whether Connolly was seen at the dance before or only after 10 pm: witnesses were called by prosecution and defence respectively on either sides of that issue. When it was tried, at the first trial, Connolly was not convicted.

Conclusion

130. For these reasons we considered that both convictions were unsafe, and we therefore allowed the appeals in the references concerning both Kelly and Connelly and quashed their convictions. As we intimated at the time of the hearing of these appeals, in the light of our decisions the Cameo murders remain an unsolved crime. If these appeals had happened timeously, fresh trials might have been held, if the Crown would have proceeded with them in the light of the new material, and these matters could have been subjected to the scrutiny of a new jury. As it is, that cannot happen. However, having considered that there was in these cases a breakdown in the due administration of justice and a failure to ensure a fair trial, we consider that the consequence was a miscarriage of justice, which must be deeply regretted.

Note 1   CI Balmer said, in answer to a question from the judge, that “The first time I saw him in connection with this case was on this date – 19th March”. The reference to “March” was clearly a slip for “November”, since 19 November not March was the subject matter of the questioning at this point and “this date” had already been identified as 19 November. 19 March was the date of the murders.     [Back]