Slade & Ors, R. v

As long ago as 20 August 2009, after a lengthy trial at Leeds Crown Court before His Honour Judge Hoffman and a jury, each of the appellants Slade, Pearman and Baxter was convicted of counts of conspiracy to murder, criminal damage, handling and arson.

After that trial they faced a further trial (along with the appellant Hudson) on two counts of conspiracy to rob and two counts of robbery. At various stages Baxter changed his plea to one of guilty on three of the counts which he faced and Hudson pleaded guilty to one count of robbery and one count of conspiracy to rob. After that second trial (also before Judge Hoffman and a jury at Leeds Crown Court) Slade and Pearman were on 1 February 2010 convicted of all four counts and Baxter was convicted of the remaining count of robbery which he faced.

Each of Slade, Pearman and Baxter was sentenced to imprisonment for life on both indictments. In sentencing, the judge made clear that he had considered both indictments together in making his assessment of dangerousness. He described them as callous, dangerous and ruthless and engaged in “Premier League crime”. Slade was assessed as the leader, with Pearman and Baxter described as “loyal henchmen”. They were sentenced to life terms with consecutive notional determinate terms of 23 years (first indictment) and 20 years (second indictment) in the case of Slade; 21 years (first indictment) and 18 years (second indictment) in the case of Pearman; and 21 years (first indictment) and 18 years (second indictment) in the case of Baxter. The resulting specified minimum terms were 21½ years, 19½years and 19½years respectively, with time spent on remand in custody being directed to count towards sentence. A Serious Crime Prevention Order was also made against Slade.

Hudson, who had pleaded guilty at a late stage on the two counts which he faced, was described by the judge as not being a core member of the gang. He did not attract a finding of dangerousness. He was sentenced to a term of 9 years imprisonment, with time spent on remand in custody also being directed to count towards sentence.

Applications of Pearman and Baxter to appeal against their convictions on the second indictment have previously been rejected by this Court. All appellants, however, appeal with leave against the sentences imposed on the second indictment and Slade, Pearman and Baxter also appeal, with leave, against the sentences imposed on the first indictment.

The appeal hearing before us however, was confined to the appeals, with leave of the single judge, of Slade, Pearman and Baxter against their convictions on the first indictment, the arguments focusing on the convictions on the count of conspiracy to murder. It ultimately was common ground before us that the various appeals against sentence would need to be deferred for further consideration depending on the outcome of the appeals against conviction on the first indictment; and the appellant Hudson (who we gather has now been released on licence) made no objection to the hearing of the appeals against sentence being adjourned for this purpose.

A great part of the appeals against conviction has concerned the appellants’ applications to this court to adduce fresh evidence. The fresh evidence is of two kinds.

That, however, is not the only ground for these appeals. Other grounds are advanced. It is said that the Judge should have acceded to an application to discharge the jury. It is said that the Judge erred in failing to accede to submissions of no case to answer made on behalf of all three appellants at the close of the prosecution case. It is further said that thereafter the summing-up to the jury was unfair, unbalanced and heavily slanted in favour of the prosecution and against the defence. Very late in the day, an argument was also sought to be raised on behalf of Baxter to the effect that there was unfair non-disclosure at trial by the Crown of material said to be highly relevant to the defence case.

At trial, all the defendants were represented by experienced leading and junior counsel. Since trial, however, there have, for various reasons, been changes in the legal representation, sometimes more than once. Before us, Slade was represented by Mr Tim Owen QC with Mr Stephen Vullo QC; Pearman was represented by Mr Henry Blaxland QC leading Mr Derek Duffy; and Baxter was represented by Mr Kieran Vaughan QC leading Mr Aaron Watkins. The Crown was represented by Mr Paul Greaney QC leading Mr Nicholas de la Poer. Mr Greaney (then as junior counsel being led by Mr Robert Smith QC) and Mr Duffy had appeared at trial: none of the other counsel appearing before us (or the solicitors instructing them) had been involved in the trial.

We would wish to pay tribute to the meticulously thorough preparation by the legal teams, counsel and solicitors, of all parties for the purposes of this appeal and to the

We should say something about the lapse of time which has occurred before these appeals came on for hearing. It is wholly exceptional that an appeal against conviction comes on for hearing before the Court of Appeal (Criminal Division) some five years after the trial where applications for leave to appeal are lodged shortly after conviction.

To a very great extent the delays have been occasioned by the various changes in the legal representation for the appellants that have occurred from time to time and the sheer complexity of aspects of the fresh evidence sought to be obtained by the appellants since trial, together with associated funding problems. Further, some of the applications for leave to appeal, whether on the first indictment or second indictment, had in fact been brought out of time (the necessary extensions thereafter being granted). In the event, some grounds were then abandoned at various stages, others added. Directions had previously been given by Hughes LJ (then Vice-President of the Court of Appeal, Criminal Division) during 2011 and 2012 which were not, for various reasons, fully complied with. Even by the beginning of 2014 the cases were still a long way away from being ready: and indeed only then, by direction of the Registrar of Criminal Appeals, was the representation order extended to Slade’s new legal team to enable it to instruct voice recognition experts and to prepare the new schedules comprising the CCTV and mobile phone and cell-site evidence. Final perfected Grounds of Appeal, superseding all previous Grounds, were lodged by the appellants during the course of 2014. Further, directions were given at various stages during the year by this Court designed to ensure that the matter came on for hearing in a properly prepared state by the end of the year.

In the result, as we have indicated, the cases as presented were very well prepared and very well focused.

We think it important to provide some context for the first indictment by briefly summarising the facts relevant to the second indictment, which relate to a period of time to some extent spanning or preceding the relevant period for the first indictment. To do so will also help explain why the trial judge described Slade, Pearman and Baxter as Premier League criminals.

Put shortly, the position was this.

Slade, Pearman and Baxter were indisputably close associates. Between January 2007 and March 2008 they conspired to rob drivers and guards of cash in transit vans. A number of premises were targeted. Cars were stolen to assist in their activities and

On 8 March 2006 a stolen low loader blocked the path of a securities cash in transit van just before it reached its depot at Woolston, near Warrington. A stolen agricultural tractor then repeatedly smashed its loading arm into the van’s cash storage compartment. Over £1 million was stolen by men wearing masks and equipped with crowbars. The low loader and tractor were then set on fire with petrol and other vehicles were used to effect a get-away with the cash. That was Count 2.

On 12 December 2007 a telescopic loader was stolen from a building site. On a later date it was used to smash a hole in the wall of a Post Office (at which staff, mainly female, were present). Two masked men ran through the gap created and terrorised the occupants. Four security boxes were taken, although £50,000 was left in an open safe. A stolen car was used to effect the get-away. Some 45 minutes later, Hudson and Pearman were observed by police to hide the car at a garage in Arncliffe Grange in Leeds. That was Count 3.

As for Count 4, that was a count of conspiracy to rob. The conspirators planned between 13 December 2007 and 21 January 2008 to rob a Sainsbury’s store at Colton Retail Park in Leeds, where it was known that some £500,000 was destined for ATM cash machines. They were observed engaging in regular surveillance. By January 2008 they had gained access to the roof and thence to an unused part of the premises and were preparing to smash through it into the ATM room. However, sounds of drilling were heard by staff on 29 January 2008 and the police were summoned. The would-be robbers fled but their activities in the roof area were revealed.

Mobile phone and cell-site evidence featured prominently at the second trial, as well as observations by the police of the accused.

It appeared that at the first trial Slade was, through his counsel, to advance his alleged involvement in such activities, as charged in the second indictment, as a potential explanation for some of the activities which featured in the first trial and in order to rebut the charge of there being a conspiracy to murder.

It was common ground that the matters which ultimately gave rise to the indictment for conspiracy to murder (and related charges) first came to the attention of the police as a result of the observations kept by the police of the movements of Slade, Pearman and Baxter and others associated with them in connection with suspected serious crime, which revealed the robbery matters. Such surveillance included the use of covert audio and tracking devices placed in cars connected with the appellants or their associates.

The prosecution in fact identified the intended victim of the alleged conspiracy to murder as a man called Ralph Roberts. He resided in the East End Park area of Leeds; he was also closely connected to a local pub called The White Horse. Roberts himself had strong connections with criminal activity and had an extensive criminal record, including for violence. (He subsequently refused to co-operate with the police and did not give evidence at the trial.) Roberts resided at an address in York Road, his

On the evening of 4 March 2008, in the hours of darkness, four men drove up outside Roberts’ home in Dawlish Crescent. They were in a black Audi RS6 car. It had been stolen a considerable time previously. Some of the men got out of the car and smashed the windows of a Vauxhall Vectra car, parked in the street, belonging to a man called Maluch. He was a neighbour of Roberts. Mr Maluch looked out of the window and saw the RS6 being driven off at speed. The police had been waiting nearby. They intervened when the sounds of the breaking glass, which were picked up on a covert audio device placed in the RS6, caused them to think that shots had been fired. In the event, the RS6 car and the men in it were not apprehended. But the police at this time also surrounded a black Audi A4 car, which previously had been parked in Ivy Crescent with its engine running, some 320 metres away from Roberts’ house, at around the time the RS6 car had driven up near to Roberts’ house. Inside the A4 car were Slade, Pearman and Baxter. The evidence was that the A4 car had been driven to the area at the same time as the RS6 car and, even if not taking an identical route, had been in “virtual convoy” with it, as the Crown was to put it.

The police were armed. When arrested, none of the three appellants resisted. No weapons of any kind were found on them or in the A4 car itself. However, subsequent examination showed that within the car were, among other things, a black bin liner bag containing clothing and a pair of binoculars; a rucksack of a colour and kind Slade had been observed often to carry, containing dog and cat repellent and a frequency recorder; a Tupperware box containing four car keys and a electronic fob and a yellow card (Slade’s fingerprints were on the box); gloves; a magnet; a plastic bag containing adhesive straps, with two blades; and a tracker (MAC4) with batteries attached and which had Slade’s fingerprints and DNA on it. The tracker was found on the rear nearside floor (where Pearman had been sitting): when arrested, Slade, who had been sitting in the front passenger seat, had been observed turning to place something at the feet of where Pearman had been sitting. The appellants were taken to separate police stations. A mobile phone attributed to Baxter was found in the car and mobile phones were also seized from Slade and Pearman. All had had their batteries removed – as the Crown was to say, with the perceived view of frustrating any subsequent cell-site analysis. In this regard, there was evidence that Pearman, from previous criminal proceedings, had had experience of the use of such cell-site evidence. Further, a mobile phone jammer and other anti-avoidance materials were subsequently found at Slade’s house and there was evidence of accessing by computers belonging to Slade and Baxter of anti-avoidance and jamming techniques. Literature relating to mobile phone jamming was also found in Baxter’s Mercedes ML car.

At 19.23, before the windows of Mr Maluch’s car were smashed, the police had alerted Roberts. He had then left the area.

At 19.42, after the smashing of the car windows of Mr Maluch’s car, the RS6 had been observed to stop at Rushwood Gardens nearby. A young witness saw four men get out of the car. A man on a motorcycle then drove up. The driver of the RS6 gave him something bulky, which had been hidden under his jacket, got on to the motorcycle and the motorcycle was then driven away. The remaining three men were seen and heard by the witness to be arguing. One was heard to say words to the

Subsequent examination of the burned out RS6 showed it to be a stolen car. Its number plates had been removed. In their place were double-sided adhesive strips, similar (although not identical) to those found in the bag in the A4. A magnetic disc was found in the debris, of a like kind to that fixed to the tracker MAC4 found in the rear nearside footwell of the A4.

The Crown’s case was that the men in the RS6 were operating in conjunction with the appellants in the A4 (which had been borrowed from a car leasing firm that very day). The actual period of the alleged conspiracy to murder was stated to be from 21 January to 4 March 2008. The Crown’s case was that the plan was to lure Roberts into the street when the car windows were smashed, when he would then be killed. An alternative suggestion by the Crown was that, at the least, Roberts was to be lured into the street, or otherwise distracted, with a view to a tracker being affixed to Roberts’ car in order to assist in his subsequent killing.

The Crown’s case had a number of strands.

Apart from clear evidence of association between the three appellants, a considerable amount of evidence was adduced which was designed to show that they were associated with the RS6. That car had been stolen on 27 September 2005 from an address in Leeds. Its identity thereafter was variously changed by the application of two sets of “cloned” number plates, lawfully registered to other Audi RS6 cars.

There was an abundance of evidence to link the appellants to the RS6 in 2007. Police observation also showed that the RS6 had been kept at various locations, including two lock-up garages at Arncliffe Grange and Unit 9, Riffa Business Park; at 6 Primrose Court on Primley Park View; and High Thorn Court on Shadwell Lane. Recording devices had been installed by the police at these locations, for the purposes of the investigations connected with the suspected serious crime, as well as at other locations. Further, tracking and recording devices were covertly affixed by the police to the RS6 and also to a Peugeot and a BMW connected to the appellants. Although there had been observations on a number of occasions of Pearman and Baxter in the RS6 on false plates up to the end of 2007, there were no such observations maintained by the police thereafter. However, Baxter was on an occasion on 26 February 2008 observed in his Mercedes ML car parked a few feet away from the RS6, itself parked at Primrose Court. Further, when the Mercedes ML was subsequently searched following the arrest of the appellants a key to the lock of the lock-up garage in Arncliffe Grange was found in it.

All three appellants were linked to the lock-up garage at Riffa Business Park, where the RS6 had from time to time been kept. Pearman’s fingerprints were on the tenancy agreement for the unit. A key to the garage padlock was found in Baxter’s Mercedes ML along with other keys. In the box found in the A4 car (on which were Slade’s fingerprints) was a fob for the roller-door of the garage at Riffa Business Park.

Another vehicle connected both to the RS6 and to the appellants was a Toyota Hiace van. That vehicle had been bought for cash on 9 August 2007 by two men who wore motorcycle helmets throughout the acquisition transaction. A false name and address

The Toyota Hiace was operated under false plates, cloned to authentic plates of another Toyota Hiace lawfully kept by its owner in Merseyside. A false plate fixed by magnet to the vehicle was in due course found to have Pearman’s prints on it. In addition, a key to the Toyota Hiace was later found in Baxter’s Mercedes ML. Further, there was evidence that the vehicle had been observed on a significant number of occasions to have been used by all three appellants. On occasions in 2007, moreover, the Toyota Hiace was observed in convoy with the RS6, with Pearman and Baxter, if not Slade also, inside.

Latterly, on 1 March 2008 the Toyota Hiace, with three men inside it, drove repeatedly and slowly around a number of streets in the East End Park area, including Dawlish Crescent and the area close to the White Horse pub. It was then driven to Swarcliffe (where Pearman lived). On 3 March 2008, in the evening three men (again the appellants, on the Crown’s case) were in the Toyota Hiace, in the Dawlish Crescent area. They were under police observation. Two of the men in the vehicle got out of the Toyota Hiace and walked around, wearing hats and with their faces concealed by scarves.

A very considerable amount of time at the trial was devoted to evidence of mobile phone use and cell-site analysis. Detailed and complex schedules were prepared. A number of mobile phones variously attributed to Slade, Pearman and Baxter were the subject of such analysis. It is not necessary for present purposes to go into details: although we will have to come back to the point when considering Mr Vullo’s submissions. Suffice it here to say that the Crown’s case at trial was that such evidence was consistent with the alleged conspiracy and that the cell-site analysis and tracker evidence accorded with the movements of the appellants in furtherance of that alleged conspiracy. It was necessarily accepted that the cell-site evidence could only show the (approximate) location of the phone in question: it could not necessarily identify the individual with such phone.

The Crown also adduced evidence of the recordings of the CCTV cameras at Slade’s house (of which there were four, with a centralised control). These showed, among other things, Slade during this period being variously picked up and dropped off at 2 Sandmoor Drive by the Mercedes ML associated with Baxter. In this regard, the evidence gave rise to an important issue. Police evidence was adduced relating to their attendance at 2 Sandmoor Drive later on 4 March 2008 after the appellants had been arrested. That evidence, if not entirely precise, established that on that date the clock on Slade’s CCTV showed a time some 22 minutes slower than real time. That matter played an important part at trial and has in turn played an important part in the arguments before us. The CCTV evidence taken from cameras at Slade’s house in Sandmoor Drive was to the effect that Baxter in the Mercedes ML car would regularly pick Slade up (Slade often taking with him a rucksack and bag) and then would return him in the same car some time later. It was common ground at trial, as we were told, that the Mercedes ML was driven always by Baxter.

A very important part of the Crown’s case at trial also related to covert audio recordings (by probe) of conversations between the male occupants of the RS6 car on various occasions in the conspiracy period. The Crown’s case was that such conversations were variously to be attributed to Pearman and Baxter and also (on two occasions) to Slade.

In brief summary, during the days of the alleged conspiracy the position was this.

When interviewed over a number of occasions, Slade and Baxter made no comment. Pearman gave answers denying any involvement in any conspiracy to murder. In the course of his answers he was to say – what the prosecution stated was a clear lie – that he did not even know any Damon.

As Mr Greaney observed, the appellants conducted a highly tactical defence.

Not only did Slade and Baxter make no comment in interview but none of the appellants, as we were told, served any defence statement. After they had failed in their submissions of no case to answer, none of them gave evidence.

The position of the appellants at trial was, among other things, that neither Pearman nor Baxter was shown to be in the RS6 at any stage during the indictment period. Slade likewise denied being in the RS6 on the two occasions (24 and 26 February) when he was said to have been in it. The prosecution necessarily had to concede that none of the appellants were in the RS6 on the evening of 4 March 2008 when the windows of Mr Maluch’s car were smashed and the RS6 thereafter burned.

It seems not to have been seriously disputed at trial, at all events on behalf of some of the appellants, that there was ample room for drawing an inference that the three appellants were engaged in some kind of criminal joint enterprise. The same approach has been pragmatically taken on behalf of all appellants for the purposes of this appeal. What is stressed by the appellants, however, is that that is not enough. What was alleged by the prosecution was a conspiracy to murder (with related counts) to which the appellants were alleged to be party: and it is that which the prosecution had to prove to the criminal standard.

The trial commenced on 19 May 2009. Much of the prosecution evidence was directed at mobile phone and cell-site analysis and the tracker and probe evidence; and much evidence also was directed at proving a connection between the three appellants and the RS6 and the Toyota Hiace and associated premises.

The only positive case advanced at trial by any of the three appellants by way of explaining their presence in the East End Park area (in the A4 if not also previously in the Toyota Hiace) was advanced on behalf of Slade. It was said – although not, of course, said by him in interview – that he was there to collect a debt from a man called Forrest. Pearman and Slade were said to be there supporting him (although it seems that they themselves at trial, through counsel, advanced no such positive case). Forrest and his partner were called at trial on behalf of Slade. It would seem that their evidence, for whatever the jury made of it, gave rise to more questions than answers. It was also, for example, left unexplained why, if that were the real explanation, the appellants had the various items in the A4 when arrested.

There had been no direct evidence of observation of any of the appellants being in the RS6 at the time the various conversations were recorded making reference to a gun, cartridges and killing. In this regard, however, the Crown adduced the evidence of a voice recognition expert, Mrs Elizabeth McClelland. Slade had obtained the report of a voice recognition expert, Mr Martin Duckworth. Baxter had obtained the report of Dr Frederika Holmes. Pearman had obtained the report of Professor Peter French (assisted by Mr Harrison).

In the result, neither Mr Duckworth nor Dr Holmes was called. As the jury were informed, they had reached an agreed position with Mrs McClelland.

In the case of Slade, it was stated that with regard to one recording (Holdall 55) no identification was possible. In other respects it was agreed that the reports did not conflict. This agreement was to the effect that sections of the recorded passages for 26 February 2008 were consistent with being spoken by Slade: with, as Mrs McClelland said, moderately distinctive features. “Consistency” was, however, such as potentially to extend to thousands of other men’s voices. It was stressed at all events by the experts that the opinions had to be used in conjunction with all the other evidence. In the case of Baxter, the joint finding, by reference to the selected recordings, was one of consistency. However, Dr Holmes found no levels of distinctiveness whereas Mrs McClelland found that some features might be regarded as moderately distinctive.

As to Pearman, by the time of his final report Professor French, in contrast to Mrs McClelland, found no consistency; and he found it very unlikely that the speaker identified by the Crown as Pearman was Pearman: although it was a remote possibility. He gave evidence at trial to that effect. Dr Foulkes, also in Professor French’s consultancy, agreed that it was very unlikely that Pearman was the speaker in the questioned recordings.

In the course of cross-examination, Mrs McClelland in fact had seemed to withdraw any attribution of at any rate one of the questioned recordings to Slade. However she then withdrew this apparent concession in re-examination and maintained there was (low-level) consistency. She accepted, however, that consistency could mean consistency with thousands of voices. She also accepted that “moderate distinctiveness” was of limited assistance in voice identification. She further accepted that if independent evidence showed that one of the defendants was not in the car when said to have been speaking on the recording then that would indicate that such defendant was not speaking then or on any other occasion attributed to that same defendant by the voice recognition evidence.

Overall, the outcome of her oral evidence was such that Mrs McClelland’s evidence was strongly criticised by the defence. At all events the prosecution were in no position to say, and did not say, that the voice recognition evidence of Mrs McClelland was in itself, if accepted by the jury, capable of proving the presence of the appellants in the RS6. They relied on her evidence, along with all the other evidence, as supporting an inference of such presence on these occasions. We discuss Mrs McClelland’s evidence at trial in more detail later in this judgment.

On 22 June 2009 the judge rejected an application to discharge the jury. On 2 August 2009 the judge handed down a written ruling rejecting the appellants’ respective submissions, themselves made primarily in writing, of no case to answer. The judge commenced his detailed summing-up on 12 August 2009. By that time the jury were reduced to 10 in number. As we have said, the jury convicted on 20 August 2009.

We turn to the Grounds of Appeal. We will take them not entirely in the order in which they were presented to us. First Ground: application to discharge the jury

We are not impressed by this ground.

Some six weeks after the trial had started the court was informed, through an usher, that one of the jurors – Mrs Jones – had attended a social function the previous weekend. She there had met Leah Gatt, who she learned was a cousin of Slade. Mrs Gatt, when she was told that Mrs Jones was doing jury service, made reference to the fact that “our Dennis is up”. Mrs Jones then removed herself from the conversation.

The judge caused Mrs Jones to be brought into court. She said that she did not know

The judge then, after discussion with counsel, had all the other jurors brought into court individually (not as a group) and questioned each of them about what they had heard from Mrs Jones. One said that Mrs Jones had just mentioned that her friend was related to a defendant, presumably Slade. Another said Mrs Jones had mentioned a friend whose cousin was on trial. That juror had only overheard part of the conversation and confirmed that he had heard nothing to compromise his position as a juror. Other jurors said that they had heard nothing at all from Mrs Jones. One had heard her say that a friend of hers was related to Slade, but nothing else. Another again had overheard part of a conversation in which Mrs Jones had mentioned “something about a friend and possibly a cousin”.

The judge was then invited to discharge the entire jury, he having already agreed to discharge Mrs Jones. He declined to do so. He found that none of the remaining jurors knew anything to the disadvantage of the defendants. The judge said that he was entirely satisfied that the remaining jurors had “in no way been tainted, either directly or indirectly, by anything that they should not know”.

Mr Vaughan, who led the appellants’ arguments on this point, said that there was an appearance of bias, given the circumstances and applying the test set out in Khan & Hanif [2008] 2 Cr. App. R 13. He further submitted that the account that Mrs Jones had given to the judge was not reconcilable with the account given by the other jurors and that, on Mrs Jones’ account, at least three other jurors knew what she knew (in particular about the Birley family, its criminal involvement and its connection with Slade).

We reject this argument. The judge had questioned all the jurors in the light of what Mrs Jones had said. He was entirely satisfied, having regard to their answers, that they in fact knew nothing potentially adverse to the defendants. In our judgment, he was entitled to draw that conclusion on the evidence which he had heard. In consequence he was entitled in his discretion to decline to discharge the entire jury. This ground accordingly fails.

In the course of directions given earlier in 2014, this court had, in order to achieve finality and clarity, directed the appellants to lodge final perfected grounds, superseding all previous versions of the grounds. This was done. However shortly before the appeal hearing Mr Vaughan indicated that he wished to raise a further ground on behalf of Baxter.

This was to the effect that, shortly before the conclusion of the trial on the first indictment, the prosecution had disclosed material in the second indictment proceedings. This related to mobile phone usage and cell-site analysis on the part of Baxter which the Crown was potentially going to use to demonstrate Baxter’s involvement in the robberies and conspiracies to rob. It was set out in a mobile phone and cell-site report served by the prosecution. It is now said that this might tend to show that Baxter was not the kind of person not to use his telephone at the relevant times if engaged in a conspiracy to murder; and also might tend to show that he was not the kind of man to create false alibis by giving his phone on occasion to others.

This is wholly speculative. Baxter had not at that stage even pleaded guilty to any count on the second indictment: his eventual pleas were, in fact, very late in the day. Moreover, as Mr Greaney pointed out, Baxter would have known what use he had made of his mobile phones: he did not need the prosecution to make his case for him. Further, whilst it is asserted that such non-disclosure deprived the defence of an opportunity to decide whether to rely upon the disclosed report at the first trial, it is entirely speculative as to just what reliance could meaningfully have been so placed. Yet further, how such material could meaningfully be deployed at the first trial was problematic on any view.

This proposed new ground is perhaps revealing of the acutely tactical and opportunistic way in which these appellants ran their cases. At all events, this proposed further ground is entirely lacking in the substance that would have been needed to justify this court even entertaining it at so late a stage. Since it seems to us to be devoid of merit, and since it is raised so unacceptably late in the day without any proper explanation, we reject it without more ado. We refuse leave to amend the grounds to add it.

As we have said, a submission of no case to answer was made on behalf of all three appellants at the close of the prosecution case. It seems that the respective submissions were made primarily in writing (and this court has seen a number, although not all, of such written submissions as presented to the trial judge). The judge in due course rejected the submissions of no case to answer by full written ruling dated 10 August 2009.

We were addressed at some length on the applicable principles: but there was no real dispute before us about them and we thus need not enter into a detailed exegesis here. The bedrock authority remains Galbraith (1981) 73 Cr App R 124. In the context in particular of a case mainly founded on circumstantial evidence, we were also referred to the decision of a constitution of this court in G & F [2012] EWCA Crim 1756. In the course of giving the judgment of the court, Aikens LJ reviewed a number of relevant authorities. Having done so, he summarised the position at paragraph 36 of the judgment in this way:

“36.

We think that the legal position can be summarised as follows: (1) in all cases where a judge is asked to consider a submission of no case to answer, the judge should apply the ‘classic’ or ‘traditional’ test set out by Lord Lane CJ in Galbraith. (2) Where a key issue in the submission of no case is whether there is sufficient evidence on which a reasonable jury could be entitled to draw an adverse inference against the defendant from a combination of factual circumstances based upon evidence adduced by the prosecution, the exercise of deciding that there is a case to answer does involve the rejection of all realistic possibilities consistent with innocence. (3) However, most importantly the question is whether a reasonable jury, not all reasonable juries, could, on one possible view of the evidence, be entitled to reach that adverse inference. If a judge concludes that a reasonable jury could be entitled to do so (properly directed) on the evidence, putting the prosecution case at its highest, then the case must continue; if not it must be withdrawn from the jury.”

This was accepted before us, for the purposes of this appeal, as a convenient summary of the correct legal approach.

That the Crown’s case was essentially a circumstantial one is plain. It had, as we have already said, a number of different, albeit linked, strands. The conversations recorded in the RS6 provided evidence to the jury of a conspiracy to kill. As to the question of whether the appellants were party to such a conspiracy there were, among other points, these factors. First there was the clear evidence of regular association between Slade, Pearman and Baxter. Second, there was the abundant evidence of association between them with both the RS6 (a stolen car frequently used with false number plates) and the Toyota Hiace (a vehicle converted into a surveillance vehicle and frequently used with false number plates); as well as their association with the premises where they were stored (as evidenced by the finding of the fob and keys and so on). Third, there was the evidence of their use of the Toyota Hiace, adapted for surveillance purposes, sometimes in conjunction with the RS6. Fourth, there was the evidence of the RS6 being regularly in the East End Park area during the period of the alleged conspiracy and also, on occasion, being involved in what could be concluded to be the tailing of Roberts’ Toyota Avensis. Fifth, there was the evidence of the Toyota Hiace, with three men in it, being in the East End Park area when the RS6 was not being used on 1 and 3 March 2008. Fifth, there was the evidence of the A4 in effect going to the East End Park area in what could be said to be virtual convoy with the RS6 on 4 March 2008 itself. Sixth, there was the indisputable evidence that the three were arrested in the A4 on that date when parked just 320 metres or so from and at the same time as the incident involving the men in the RS6 car damaging Mr Maluch’s car in the street outside Roberts’ house. Seventh, there was the evidence of the trackers, magnets and other potentially incriminating items found in the A4 car at the time, along with the mobile phones with the batteries removed. Eighth, there was the conduct of the four men in the RS6 burning the car and then departing in the circumstances witnessed by the boy. Ninth, Slade and Baxter gave no comment interviews. There were various other matters as well, which we need not rehearse here.

An important feature of the Crown’s case, of course, necessarily was the content of the conversations, recorded by the probe, of the men inside the RS6 on the various occasions in this period, as we have summarised above. This was of particular importance: because the Crown’s primary case was not, of course, simply that the

The appellants at trial were, it is true, in a position to say that there was no direct observation of the appellants in the RS6 after the end of 2007 (and no doubt also were in a position to mount an argument that the RS6 was to be regarded as, as it were, a “pool car” for general use by the criminal fraternity). But even there the prosecution were able to point to the observation evidence showing Pearman at Primley Park View very close to the RS6 on the evening of 26 February 2008 in circumstances where, it could be inferred, he was checking on it.

Given all this, one can certainly see a case for saying that the appellants were parties to a conspiracy to murder whether or not they themselves variously were actually in the RS6 and actually party to the conversations recorded by the probe as summarised above.

But that was not the position advanced by the prosecution at trial. Its position throughout was that the two men in the RS6 recorded by the probe in this period were Pearman and Baxter: joined, on the two occasions, by Slade. That was a position founded essentially on inference. It is true that some support for such a case could be found in the evidence of Mrs McClelland, if accepted by the jury. But, as the prosecution throughout fairly and rightly accepted, such expert evidence could not, even taken at its highest, identify the men in the RS6 on those occasions as the appellants. Her evidence was, in effect limited to saying that the voices were (in her opinion) consistent with those of Slade, Pearman and Baxter and with varying degrees of distinctiveness. On her own evidence, the voices could also be consistent with voices of men other than the appellants.

The judge recorded in his written ruling that it was a “central plank” of the prosecution case that the recorded voices in the RS6 were those of the appellants and “crucial to proof of their case” against the appellants. He said of the prosecution: “They have nailed their flag to that particular mast. Nor do they seek to retreat from that position…”

When this court studied the written submissions of the Crown advanced at trial in response to the written submissions of no case to answer, it was not entirely obvious to us from those written submissions that the Crown indeed were “nailing their flag to that particular mast”: even if it certainly was a part of the case then being presented and had been the basis on which the case was opened. However Mr Greaney told us, with commendable candour and fairness, that that correctly described the position. He confirmed that it was indeed a “central plank” of the case then being advanced at the trial: and no alternative position was being advanced. The judge’s description of the position thus was accurate.

Accordingly, as it seems to us, as a matter of fairness the question of whether or not the submission of no case to answer was rightly rejected by the judge has to be assessed by reference to the case then being advanced: which case included the proposition that the three voices in the RS6 were those of the appellants. That it may not, in truth, have been crucial to establish a case against the appellants on that particular basis is thus, for present purposes, neither here nor there. If what was being presented to the jury by the prosecution as “crucial” to the proof of its case did not, on analysis, represent a conclusion properly open to the jury on the evidence, then as we see it the submission of no case to answer had to be decided accordingly. It thus follows that it is not for this court, in the circumstances of these particular appeals and in assessing whether or not these particular convictions were safe, to approach the matter on an alternative basis (viz that actual presence of the appellants in the RS6 on the disputed occasions was not, in truth, necessarily crucial to the prosecution case) which was not advanced to the trial judge by the prosecution and which was not in accordance with the way the case was in fact conducted at trial. Nor did Mr Greaney, for the purposes of the present argument, seek to contend otherwise before us. He maintained that – taking all the evidence as a whole – there was indeed evidence from which a reasonable jury could properly and safely infer that the men recorded on these occasions in the RS6 by the probe were Pearman and Baxter and also (on two of the occasions) Slade.

But we also add this. The judge’s task was to assess the submission of no case to answer by reference to the arguments addressed to him and to the evidence thus far deployed at the trial. There was, in our view, no obligation on the judge to review for himself all the evidence in order to see if there were yet other possible arguments or evidential points available to the appellants, even if they were not then actually advancing them in support of the submissions of no case to answer. We say this because, as we will come on to identify further, aspects of the submissions made to us – notably those of Mr Vullo – undoubtedly involved an analysis of the materials and a much wider ranging factual appraisal (and in some respects deployment of factual matters not raised at trial) which went well beyond the analysis and argument undertaken on behalf of the appellants before the trial judge on the submissions of no case to answer. We should, however, make clear that no ground of appeal is advanced before us based on inadequate representation by trial counsel: indeed any such suggestion was in terms disclaimed.

That said, therefore, we turn to assess whether the judge could properly conclude that there was a basis whereby a reasonable jury could properly accept what was then being put forward as a central plank of the prosecution case and could properly reject any other contrary conclusion.

It is a reasonable starting point to note that the evidence was on any view that others also had been associated with the RS6 at this time – for example, in various of its movements and parking activities. Moreover, self-evidently there were four others in the RS6 engaged in the events outside Roberts’ house on 4 March 2008 itself: because the appellants were themselves at that time parked nearby in the A4. No doubt it was for this reason that the conspiracy particularised on the indictment alleged a conspiracy between the appellants and “persons unknown”.

On this “central plank” of the prosecution case, the arguments advanced before the trial judge on behalf of the appellants – as reflected in the written submissions made in particular on behalf of Baxter, which we have seen – came to this. It was said that no

The essential points open to be made were as follows:

There was a further point available to the defence on the CCTV timings. Even if the inference was not that the CCTV timings were consistently 22 minutes slow (as they argued), then they could alternatively say that the timings, if reflecting real time, still,

Furthermore, if all these points were valid they would operate to show that no further support could be derived from Mrs McClelland’s voice recognition evidence. As we have said, it was accepted that if the voice recognition expert evidence was controverted by other evidence it would yield to that evidence; and further Mrs McClelland had herself stated the view that the voices were, in her view, the same throughout; and if she was wrong in one instance with regard to a particular male then she could not be taken as right in the other instances with regard to that male.

The judge rejected the arguments advanced to him, preferring the arguments of the prosecution:

The appellants submit that the judge was not justified in rejecting the submissions of no case to answer as he did; and submitted that the “central plank” of the prosecution case as showing the appellants being party to a conspiracy to murder was indeed not capable of being sufficiently proved. No inference, to the exclusion of other inferences, could properly be drawn as argued for by the prosecution. In particular, they criticise as wholly inadequate the prosecution statement to the judge that the CCTV times “may or may not” correspond with the time shown on 4 March; and the judge’s approach to the CCTV timing, they submitted, was tantamount to reversing the burden of proof. Overall, the case should have been withdrawn from the jury given the way the prosecution had sought to advance it at trial.

We have considered these various points with care. There is no doubt that they are powerful points and they merited close consideration by the judge himself.

In this regard, however, it is necessary to bear in mind that the judge was plainly wholly on top of the evidence as it had emerged during the prosecution case (which perhaps in part explains the relatively short-hand way in which the appellants’ written submissions were presented at trial). It is at all events evident from his full ruling, which ran to 34 paragraphs, that he had careful regard to all the submissions made to him. We also bear in mind that Galbraith itself makes clear that matters close to the line on a submission of no case to answer “can safely be left to the discretion of the judge”.

We have concluded, assessing matters as best we can on the basis of the arguments and materials we understand to have been put before the judge, that we should not interfere with his ruling. As the judge said, on the evidence adduced at trial the only sure fact with regard to the CCTV timing was that on 4 March 2008 it was some 22 minutes slow compared to real time. It did not follow, as an inference, that it must likewise have been 22 minutes slow on each day in the intervening period from 21 February 2008. Nor did it follow that it showed the real time on those days either. As the jury had heard and could conclude, the appellants were surveillance “savvy”: and one question the jury could properly ask themselves, in the absence of explanation, was as to how and why the CCTV timing at Slade’s house came to be out by 22

Similar considerations – in particular by reference to the appellants’ surveillance and mobile phone “savviness” – could be applied to the apparent “alibis” arising from relevant mobile phones being on occasion elsewhere from the RS6 at the time of some of the critical instances relied on by the prosecution.

When these points are set in the context of all the many other factors cumulatively lending powerful support to the Crown’s case that the appellants were party to a conspiracy to murder Roberts as set out by the judge, our view, on balance, is that the judge was entitled to rule as he did.

This ground, as formulated in this way, thus fails.

But we have also concluded that that cannot, given the circumstances of this particular case, be the end of the matter. What this court ultimately has to consider, after all, is whether these convictions are safe. And we are simply not able to put out of our consideration the impressively powerful presentation by Mr Vullo, based on the meticulously prepared schedules put before us collating the relevant CCTV sightings at 2 Sandmoor Drive with the mobile phone/cell-site evidence and the tracker observation evidence: which we will come on to summarise.

We have nevertheless deliberated as to whether it is right for us to entertain these points as advanced by Mr Vullo at all. To a significant extent the points now advanced to us do make, albeit with much more elaboration, points sought to be advanced to the trial judge on the submissions of no case to answer. But they also go very much further. For one thing, they focus on the CCTV timings (both on the basis that they were 22 minutes slow throughout and on the alternative basis that they show the real time throughout) with a view to positively demonstrating that the CCTV clock was indeed consistently 22 minutes out on each day in the alleged conspiracy period. This was done in much more detail than was advanced to the trial judge – where the defence stance had been confined in effect to arguing for an inference as to CCTV timings. Further, this exercise was also conducted before us in relation to days (not simply 22, 24 and 26 February 2008) other than those on which the judge had been addressed. This was with a view to, as Mr Vullo explained, showing a consistent pattern throughout this period as to the CCTV timings: a potentially very important point. A yet further development was deployment by Mr Vullo (and, in one instance, Mr Vaughan) of telling materials which, although based on evidence available at trial, do not appear to have been advanced in evidence or argument before the judge.

Does, then, the importance of the one-trial principle – that the defence should present the best case available to it at trial – preclude our having regard to these matters, be they styled fresh arguments or fresh evidence (albeit the evidence on which the arguments to us were based was not “fresh”: in the sense that it was evidence available at trial even if not deployed by the defence for these purposes)? We have come to the conclusion, however, that, as a matter of justice and in the circumstances of this particular case, we should not regard ourselves as so precluded and that we should have regard to these matters as advanced by Mr Vullo.

Quite simply, the detailed analysis presented to us, coupled with the deployment of the new evidential points not advanced at trial, has persuaded us that the CCTV timing has positively been demonstrated to be some 22 minutes out on each day in the period. The judge’s “Who knows?” position with regard to the CCTV times – taken at the stage of the submissions of no case to answer and then as put to the jury in the summing-up – has now been shown to have been wrong. When one links that – now demonstrated – fact with the other arguments that were and remain available (the “alibi” evidence of the phones, the “Div” texts, the accessing of the computer at Baxter’s house on 25 February 2008), and as further confirmed by other points advanced to us, in our view it is established that the central plank of the Crown’s case, as advanced at trial, was one that was not available. On all the evidence now before this court it could not and cannot be proved by the prosecution that Baxter, if not also the other, appellants variously were in the RS6 discussing the killing of Roberts as

If one then takes into account the fact that in due course the jury were, albeit understandably at the time, directed in the summing-up (consistently with the prosecution case as advanced at trial) that it was open to them to conclude the appellants had been the men recorded on the various occasions in the RS6 and to reject the defence “alibi” arguments, then the conclusion simply has to be that the convictions are not safe: for that central plank, as advanced by the Crown to the jury, has been removed. Indeed, that aspect of the Crown’s case may – adversely to the appellants – have potentially achieved still greater prominence in the minds of the jury by that stage, given the prosecution’s and judge’s understandable repeated emphasis on the fact that none of the appellants had given evidence, or (for the most part) answers in interview, to provide any explanations.

We turn to summarise the points advanced before us by Mr Vullo, by reference to his schedules, to explain our conclusion in a little more detail.

The main focus of the schedules and arguments before us was directed at the movements of Baxter. But this in turn potentially may impact on the position of Slade, outside whose house Baxter’s Mercedes ML is on occasion recorded as arriving or leaving, as the case may be. The position of Pearman, in turn again, may be rather different. But Mr Greaney pragmatically did not seek to argue before us that there could be one outcome on this appeal for one or more of the appellants but a different outcome for the other appellants.

We would pay tribute to the meticulous attention to detail which underpinned Mr

The schedules deployed before us are essentially, as we gather, all founded on materials available for use at trial. They relate to each day from 22 February to 4 March 2008. They are compiled in the following way:

By this process, and by cross-referencing events recorded on the CCTV at 2 Sandmoor Drive at the unadjusted times with the cell-site/mobile phone material identifying contact by or to the phones attributed to Baxter, there is scarcely any reconciliation on any of the days between the CCTV images showing Baxter’s Mercedes ML outside (or near to) 2 Sandmoor Drive and the cell-sites accessed by the phones attributable to him.

By way of contrast, if one then undertakes the same exercise adjusting the CCTV timings by 22 minutes then the telephone usage relating to Baxter’s phones on each day closely ties in with the events recorded in the CCTV footage. Those phones are, at such adjusted times, then accessing cell-sites which are either best server cell-sites for 2 Sandmoor Drive or very close to it. Further, and importantly, the pattern is repeated on each day throughout the period (including 4 March 2008, when the Crown itself accepted the CCTV clock timings were 22 minutes out). This had particular importance, as Mr Vullo observed, because it went beyond the point relating to the phones allegedly providing an “alibi” on 24 and 26 February 2008.

In such circumstances, the appellants contend that there is clear evidence compelling a conclusion that the CCTV timings were indeed 22 minutes out on each day during the entire period of the alleged conspiracy. The CCTV timings therefore were neither operating at the times they showed as real time nor were they operating on a persistently erratic basis.

A further schedule provided by Mr Vullo seeks to confirm the point by different means. Two regular events in the daily routine at 2 Sandmoor Drive, as caught on the CCTV, are taken. These are the morning school run and the arrival of the cleaner on weekdays. Throughout, there is broad consistency in these timings: this is accordingly both indicative of there being no deliberate change in the timings of the CCTV recordings and also indicative of the timing mechanism operating consistently in this period.

Yet further, a series of stills taken from the CCTV on 22 and 27 February and 1 March 2008 were provided to us. These appear to show Slade (or, once, Baxter) using a

We thus have concluded that these schedules demonstrated, to our way of thinking, that the internal clock on the CCTV was indeed running consistently some 22 minutes slow on each day throughout the period of the alleged conspiracy.

The resulting position can be assessed by taking as examples the days 22, 24 and 26 February, as detailed on the schedules. On the footing that the CCTV clock was indeed 22 minutes slow, then on 22 February 2008 the position becomes as follows:

On 24 February 2008 the position becomes as follows:

On 26 February 2008 the position becomes as follows:

We repeat that there is no difference in pattern between the various days in the entirety of this period (including 4 March 2008). Yet further, this material was, as Mr Vullo submitted, to be put in the context both of the “alibi” phone occasions and also of the texts from Baxter’s relevant mobile phone to that of Div. On this last point he relied – as had been done on behalf of Baxter at trial – in particular on the entries relating to 26 February 2008, which showed a series of exchanges between 18.04 and 19.38 (when the RS6 was in use, but when the relevant phone was in a location elsewhere than at the location of the RS6, being at or near 2 Sandmoor Drive) as well as such Div texts on other dates. In this regard he was also entitled to re-iterate the point which had been made at trial: that the highly personal nature of such texts is not readily consistent with anyone other than Baxter being the user.

Moreover, Mr Vullo’s argument derived support from yet another point. There were available at trial records of Automated Number Plate Recognition sightings. Mr Vaughan, on behalf of Baxter, submitted that, from those, it was demonstrated that on 24 February 2008 at 20.41.30, at a time when the RS6 was tracked to the East End Park area, Baxter’s Mercedes ML was in fact identified at a location in Leeds over five miles away. This too is evidence of a kind which was available at trial but was not used (perhaps because it was overlooked). But it lends further support to Mr Vullo’s points, if further support is needed.

Overall, Mr Greaney had the greatest difficulty in formulating submissions to controvert these points made by Mr Vullo and as adopted by all counsel for the appellants. In our view, as will be gathered, they were not controverted.

We think it appropriate to deal, nevertheless, with the other grounds advanced.

The summing-up of the judge was lengthy and detailed. It was clearly structured; it dealt fully with the evidence and outlined the issues. It was a very long way indeed from being a summing-up of the kind involving reading out verbatim the notes from the judge’s notebook or anything like that.

All trial counsel for the appellants nevertheless made objection on the record, at the conclusion of the summing-up, as to its fairness and balance. That objection is maintained before us.

We were taken in considerable detail through a number of the passages in the summing-up. It was complained that in many places it read like a speech for the prosecution. It was said that the judge persistently in the course of the summing-up adopted a technique not only of emphasising and commending prosecution points but also, when dealing with defence points designed to rebut prosecution points, then himself making comments designed to depreciate those defence points; and further did so when summarising the defence case at the end of the summing-up. Moreover he did so, it is said, in trenchant and dismissive terms. It is further complained that the judge dealt with the points strongly relied on by the defence – the CCTV timings, the “alibi” phone calls and so on – in a cursory and dismissive way.

It was further observed that, towards the end of the summing-up, the judge twice said that he was not seeking to undermine the defence case: a coded acknowledgement, it was suggested, of an appreciation that that is precisely how it may otherwise have been viewed.

A particular illustration, it was said, of the unfair comments made (among others) can be taken from a passage of the summing-up relating to the fact that there had been no actual observations made by the police of any of the three appellants being in the RS6 during the days of the alleged conspiracy between 21 February and 4 March 2008. The judge as to that commented:

“Obviously it is an omission in the investigations and observations that the police were doing and no doubt somebody will be wishing that a camera had been set up long before it was to monitor comings and goings…”

It is complained that that conveys the inference that, had only such observations been made, further incriminating evidence would have emerged. Numerous other passages were also relied on to similar purport.

Given our overall conclusion, we can take these submissions shortly. The judge was here dealing with a highly tactical defence case, where all the indications are that leading counsel for the appellants at trial had perhaps taken to the limit what might properly be advanced in closing speeches in the absence of any answers in interviews, any defence case statements or any evidence at trial from the appellants themselves; and when the prosecution had by then, of course, made its own closing speech and had no further right of response. The judge had to deal with that situation arising. We can accept that some passages of the summing-up might in places have been better framed or balanced or have been put in somewhat different language. But overall we are not persuaded that this summing-up was unfair or unbalanced to such an extent as to render the convictions unsafe.

This ground fails accordingly.

Our ultimate conclusion also means that it is strictly unnecessary to deal with the applications formally to adduce fresh evidence. But we think it appropriate to do so. This is not only out of deference to the detail of the evidence (in the first instance, received by us de bene esse) and to the careful arguments presented to us on it; it is also because our views may perhaps in some respects be of relevance for wider purposes.

The proposed fresh evidence was in each instance expert evidence. The evidence was, as we have indicated, directed at two distinct areas: first, relating to the timings of the internal clock relating to the CCTV system at Slade’s residence at 2 Sandmoor Drive; second, relating to voice recognition.

The statutory criteria for admitting fresh evidence are, of course, those set out in s.23 of the Criminal Appeal Act 1968. Ultimately the court has to consider whether or not it is necessary or expedient in the interests of justice for such evidence to be received. The

For the purpose of exercising its statutory jurisdiction, the court has to balance two competing principles. The first – to which we have previously alluded – is that it is incumbent on a defendant to present his whole case at trial (the one trial principle). It is, in general terms, contrary to the proper administration of justice for a defendant to advance on appeal a case different from or other than the case he is able to present at trial. The second, however, is that it is generally abhorrent to the proper administration of justice that a defendant may have been wrongfully convicted. The court has to weigh these matters and all other relevant matters (including those specified in s.23(2)(a) to (d)) in deciding whether or not to receive the proposed fresh evidence. At all events, the impression sometimes is given by appellants – and, on occasion, perhaps was sometimes given in the arguments advanced before us on behalf of the appellants in this case – that if only the fresh evidence may afford a ground for allowing an appeal then that of itself justifies its reception into evidence. But demonstrably the consideration has to be wider than that: the ultimate question being whether it is necessary or expedient in the interest of justice to receive the evidence.

In the context of the proposed reception of fresh expert evidence we were referred to a number of authorities. These included the helpful summary given by Aikens LJ in Chattoo [2012] EWCA Crim 190 at paragraph 70 of the judgment; see also Workman [2014] EWCA Crim 575 at paragraphs 59 to 63 of the judgment. The observations of the court in those cases, as was not disputed before us, provide a convenient summary of the required approach.

Mr Owen sought to place particular emphasis on the decision of the Privy Council in the case of Lundy [2013] UKPC 28. That was a New Zealand case, whereby the underlying principles were broadly comparable to (even if not identical to) those set out in s.23 of the 1968 Act. It was a striking case, in that expert evidence was, in the circumstances of that case, permitted to be received in evidence by the Privy Council even though its availability had for the most part been known to counsel at trial. Mr Owen referred in particular to the observations of Lord Kerr at paragraph 120 of his judgment and to his observation in paragraph 128 that “where the new evidence presents a direct and plausible challenge to one of the central elements of the prosecution case, this factor [the one trial principle] ceases to be of such importance”. Ultimately, however, as we see it, just where the balance lies in assessing what is necessary or expedient in the interests of justice has to be decided by reference to the particular circumstances of each case; and in this regard (as emphasised in Chattoo at paragraph 70), the appellate court must, among other things, be careful not to allow the trial process unjustifiably to be subverted.

The appellants sought to advance as fresh evidence an expert report dated 28 July 2014 from Mr Sudeep Joseph Abraham, a computer crime consultant working with Systems Technology Consultants Ltd (Sytech). Mr Abraham’s expertise was not in dispute. He was demonstrably a credible and reliable witness.

Mr Abraham had in 2014 been asked to perform tests on the CCTV system which had been seized from 2 Sandmoor Drive on 4 March 2008. It will be recalled that the evidence given at trial was that, and the entire trial was conducted on the basis that, as at that date (4 March 2008) the CCTV time clock was some 22 minutes slow compared to real time. The tests Mr Abraham was asked to perform were undertaken with a view to showing that the clock had been similarly slow – that is, 22 minutes slow – in the days and weeks before 4 March 2008. If that was so, then that, as will be gathered, undermined the prosecution case of Baxter’s and/or Slade’s presence in the RS6 (as recorded) and corroborated the “alibi” arguments advanced by the defence. Thus this expert evidence was designed in this respect to show what Mr Vullo had set out to show by reference to his schedules.

Mr Abraham received the CCTV system from the West Yorkshire Police on 23 June 2014. The system was then checked to see that it was working properly, when powered by mains electricity; and was synchronised. Thereafter it was run for the period between 16 July and 28 July 2014. It was then assessed for time differentiation: and was found to have lost two seconds in that 12 day period. Mr Abraham’s calculation was, on that basis, that it would take about a year for the CCTV clock to lose one minute.

He further reported on the pronounced time variations identified when the CCTV system was examined by the police in May 2008 as compared to 4 March 2008 – to the effect that on 28 May 2008 the date on the CCTV system when then examined was some seven weeks out and the time 13 minutes out, as was explained in evidence given at trial by DC Lyon. Mr Abraham said that was to be explained by the CCTV system having been entirely switched off in the interim and the battery’s power in consequence having become degraded. He did, however, agree in cross-examination before us that something must have happened to cause the clock to be 22 minutes late on 4 March 2008 – for example, as one possibility, interference by the user. 132.The appellants therefore say that Mr Abraham’s evidence is very important evidence. If, as was the unchallenged evidence, the CCTV clock was indeed “out” by some 22 minutes on 4 March 2008 then the evidence of Mr Abraham goes positively to confirm the defence case at trial: that it would indeed likewise have been “out” by some 22 minutes in the preceding weeks and days. And, if that were so, then such timings – by reference in particular to the CCTV showing Baxter in his Mercedes ML variously picking up or dropping off Slade at 2 Sandmoor Drive – demonstrated the defence point that Baxter and/or Slade simply could not have been in the RS6 on various of the occasions relied on by the prosecution at trial. That, of course, was precisely the point Mr Vullo was seeking to make.

It is, as we have said, the case that at trial the defence had understandably been seeking to make much of the argument that the CCTV time was, as accepted, some 22

Mr Greener in his trial report had, among other things, considered the date and time accuracy of the CCTV system retrieved from 2 Sandmoor Drive. He said they were

Mr Greener had been asked to attend at trial. He told us that there was a conference at court with the defence legal team (which included discussion of the 22 minute point); and in the event he was not required to give evidence.

In such circumstances, we are not prepared formally to admit the evidence of Mr Abraham. The defence were alive at trial to the potential importance of the 22 minute point and the CCTV footage; and Slade had commissioned a report from Mr Greener at the time for that purpose. Mr Greener had not been instructed at the time to make the further investigations to which he had alluded nor had he been asked to give evidence at trial. The evidence thus was, or reasonably could have been, available. It was submitted to us that the defence could properly have taken it – and it was said that Slade’s trial counsel did take it – that the Crown was not disputing that the CCTV time was around 22 minutes out throughout the relevant period. But the Crown clearly had not accepted that as a fact. Further, one can envisage that the defence tactically might prefer to make their points – as they did – by reference to inferences which they asserted might be drawn rather than, for example, exposing Mr Greener to the possibility of cross-examination – for example as to how the CCTV time came to be out by 22 minutes on 4 March 2008, with a potential suggestion that Slade (as enduser), and against the background evidence that the appellants were surveillance “savvy”, could have re-programmed the clock each day accordingly.

In our view, therefore, it is not necessary or expedient in the interests of justice that Mr Abraham’s evidence be received. This was evidence of a kind available to the defence at trial which it elected not to explore further or to deploy. In any event, the point sought to be made is now covered by Mr Vullo’s analysis. Indeed Mr Owen himself accepted in his written argument that the new Sytech evidence “merely confirms” the point now being made as to the CCTV timing.

We accordingly formally refuse the application to receive this evidence of Mr Abraham.

This ground raises issues altogether more complex than the application to adduce the Sytech evidence.

It was, as will be gathered, the prosecution case that the men recorded speaking in the car on the various occasions were the appellants. In support of that allegation, which

Expert evidence of this nature commonly involves a comparison of a questioned sample of recorded speech (the recording of a person alleged to be involved in crime) with a reference sample comprising a recording of what is known to be the voice of the suspect. From that comparison the expert witness forms an opinion as to how far the two recordings are consistent with having originated from the same speaker. The usual methods of comparison are auditory analysis and acoustic analysis. The former usually involves the expert witness repeatedly listening to the recording through highpowered headphones, and making a judgment. The latter involves the application of computer software to make a computerised analysis which may include spectograms of specific sounds and formant measurements (that is, computerised measurements of resonance or areas of high energy in the recorded speech).

Mrs McClelland’s methodology used only auditory analysis. She was sceptical about acoustic analysis generally, one of her reasons being that she did not believe there to be a sufficient database to enable the expert to know what could be expected of the population as a whole, and therefore there was nothing with which reliably to compare the findings in a particular case. She also took the view that formant measurements cannot reliably be used, and could be misleading when the recording concerned had not been made in clear conditions free of interference. She acknowledged that in this respect she is in a minority amongst practitioners in the field: the majority use a combination of auditory and acoustic analysis.

In Robb (1991) 93 Cr App R 161 this court held that expert evidence of voice identification, based on auditory analysis alone, is admissible. In 2002 the Court of Appeal in Northern Ireland took the view that time had moved on since the decision in Robb, and concluded that in that jurisdiction no prosecution should be brought in which one of the planks was voice identification based on auditory analysis alone: see O’Doherty [2003] 1 Cr App R 5. In England and Wales, however, there has been no similar development in the law; and in Flynn and St John [2008] 2 Cr App R 20 a constitution of this court declined to follow the decision in O’Doherty: see, in particular, paragraph 62 of the judgment of the court given by Gage LJ.

Mrs McClelland had been provided with copies of some of the recordings made in the RS6, in which the male voices were designated M1, M2 etc. These recordings carried engine, traffic and other noises as well as the voices of the speakers. She had also been provided with reference samples of the speech of each of the appellants. Some of these came from police interviews, but reference samples of the voices of Baxter and Slade also included a recording (exhibit “Holdall 12”) which was admitted to be of them speaking in a different vehicle, the Peugeot car. The reference samples of Pearman included a recording (“Holdall 42”) of him speaking in a third vehicle, the BMW car. The recordings in the Peugeot and BMW had been made covertly, and they too carried engine and traffic noise.

In relation to the voices of the men speaking in the RS6, Mrs McClelland was asked to consider whether each of the voices designated as M1, M2 and M3 was the same voice in each of the recordings in which it was thought to be heard. She was also asked to consider whether the reference samples of the appellants were phonetically and linguistically consistent with the questioned voices.

Mrs McClelland explained to the jury that forensic voice analysis first entailed forming an opinion as to whether the questioned and reference samples were consistent with the speech having originated from the same person. If so, it then entailed going on to express how closely matched the samples were, using a five-point scale of distinctiveness ranging from “not distinctive” to “exceptionally distinctive”. The distinctiveness scale enabled the expert to consider detailed phonetic and acoustic features over and above the broad consistency which had first been noted. This involved considering – by intensive listening over many hours – the fine detail of articulation of vowel and consonant sounds, rhythm, pitch, intonation and voice quality (“timbre”), and marking on the scale how many of those features, and to what extent, were judged to be distinctively similar across the reference and questioned samples.

As we have said, it was never the prosecution case that voice identification evidence alone could prove that a particular appellant was a particular speaker in the RS6. The evidence was put forward only as one aspect of the prosecution case. Mrs McClelland also made clear, at an early stage of her evidence-in-chief, that forensic voice analysis could not operate as a freestanding identification of an individual. She said –

“It’s important to recognise that forensic voice analysis isn’t equivalent in any sense to identification evidence from, for instance, DNA or fingerprinting. The conclusions we can reach in terms of voice identification should only be used in conjunction with other evidence as part of a picture. You can’t identify an individual solely using an opinion from a forensic voice analyst.”

Mrs McClelland’s evidence was that the designation of M1, M2 and M3 was consistent across all the questioned samples recorded in the RS6. In some of the recordings the level of background noise was such that she could not form any view about the degree of distinctiveness. Other sections of the recordings were clearer and enabled her to express an opinion on the distinctiveness scale. Her findings, in summary, were as follows.

On behalf of Pearman, Professor French gave evidence at trial. He has worked in this field for some 25 years and has given expert evidence in many cases. He is an Honorary Professor in the Department of Language and Linguistic Science in the University of York and President of the International Association for Forensic Phonetics and Acoustics. His work in this case had been peer-reviewed by his colleague Dr Foulkes, who also gave evidence at the trial.

Professor French had used a combination of auditory and acoustic analysis. He regarded Mrs McClelland’s approach, of using auditory analysis alone, as no longer acceptable practice. He disagreed with Mrs McClelland’s evidence relating to Pearman. When he first prepared a report in this case, he found consistency between Pearman’s voice and the voice of M1. However, he subsequently analysed the Holdall 42 recording from the BMW, and concluded – on the basis of voice quality and phonetic differences, which he explained to the jury - that there was no consistency and that it was very unlikely that Pearman was the speaker in the RS6. He did not feel able to eliminate him completely; he said there was a remote possibility that M1 was Pearman.

Neither of the other appellants called any expert evidence on this topic, as we have explained: although the jury were made aware that Mr Duckworth had been instructed on behalf of Slade, and Dr Holmes on behalf of Baxter. The jury were also aware that Mr Duckworth agreed with Mrs McClelland’s attribution of M3 to Mr Slade in a recording on 26^th February as “moderately distinctive”. Dr Holmes accepted that Baxter’s voice was consistent, but no more than consistent, with M2.

The judge in his summing-up gave an appropriate direction to the jury about their approach to this evidence, explaining to them that although they had heard the recordings for themselves, the expert witnesses had the advantage, which the jury could not have, of repeated listening to the recordings on specialised audio equipment.

After the trial, Professor French, and his colleague Mr Harrison, were instructed to conduct further voice comparisons using Automatic Speaker Recognition (“ASR”) technology in the form of a system known as Batvox. Initially they did their work solely in relation to the appellant Pearman. But they were later instructed also to consider the recorded voices said to be those of the other appellants. In their first report on 1^st June 2012, dealing with Pearman alone, they concluded that the results of the Batvox tests were 37 or 38 times more likely if M1 was not Pearman than if he was. A new version of the Batvox system then became available. This was used in carrying out tests in relation to all three appellants, in the course of which the Holdall 12 recordings from the Peugeot car were used as an additional reference sample. The overall result of their work in relation to Pearman and Slade was summarised as follows in their report of 17^th July 2014:

“In our view, the voice evidence in respect of these two appellants provides exceptionally strong support for the defence claim. We consider that Richard Pearman and Dennis

As to Baxter, Professor French and Mr Harrison concluded that he could be eliminated with “a fairly high degree of confidence”.

All three appellants have sought to adduce this evidence as fresh evidence. This court heard, de bene esse, evidence from Professor French and Mr Harrison, as well as evidence from Mr Allen Hirson who was called by the respondent. In their submissions as to the considerations which are relevant for the purposes of this section 23 application, the parties focused on two issues. Was Professor French able at trial to give evidence based on ASR technology? And was it necessary or expedient for this court to receive the evidence?

In their report of 1^st June 2012, Professor French and Mr Harrison explained that ASR systems work on the principle that individual voices may be distinguished from one another by virtue of the vocal tracts from which they emanate having different anatomical dimensions and proportions. These differences give rise to acoustic differences: namely, differences in the structure of the resonance frequencies found across the speech of individuals. ASR systems take the recorded voices of individuals, perform complex mathematical operations on them and reduce them to statistical models. A statistical model is made of the known sample and the questioned sample, and the two are compared to produce a measure of the similarity or difference between them. In order to determine whether the measure is indicative of their having come from the same or different speakers, the extracted features are also compared with a set of statistical models from a reference population of other speakers held within the system. The characteristics of the reference population, including gender, language and recording conditions, should ideally be the same as those of the suspect’s recording.

The measure of the difference or similarity between the compared recordings is expressed by the ASR system as a likelihood ratio: that is, an expression of the ratio of how likely it is to have found the voice evidence if the samples were to have come from the same speaker against the likelihood of having found that evidence if they had come from a different person. A likelihood ratio of more than 1 means that the evidence would be more likely to occur if the speaker were the same; a likelihood ratio of less than 1 means that the evidence would be more likely to occur if they had come from different speakers.

Professor French said in evidence that he first acquired Batvox in March/April 2009, and trialled it in July/August of that year. He therefore had access to the Batvox system at the time of the trial of these appellants. However, he took the view that at that stage it had not had sufficient testing with samples of English voices for him to use it in his forensic work. He said that his subsequent testing included tests which showed the Batvox system to be sensitive to regional accents, but without large differences, and further testing in circumstances in which the quality of the sample recording was less clear. The results of those tests, and testing of the system by others, led him to regard it as a system which could be used in casework by the end of 2010/start of 2011. He has not, however, used the system in any of his forensic work other than this case.

In their evidence to this court, Professor French and Mr Harrison explained that the Batvox software which they had used had a database or reference population of 100 male speakers of standard southern British English, aged 18-25, who had been recorded in the form of simulated police interviews. From that population the software automatically chose a smaller subset on the basis of their statistical models matching most closely that of the suspect. That subset could be as low as 20; but they had set the system so that a subset of 35 was used. It was not altogether clear from their evidence which 35 sample voices the system chose for comparison with each of the appellants nor why a figure of 35 was selected. Insofar as the subsets may have differed from one another, at all events, the court has no information as to what those differences were.

At trial, it was common ground between Mrs McClelland and Professor French that voice quality/timbre is an important marker of a voice. Professor French told this court that he and his colleague had initially considered that ideally the reference sample should comprise speakers who came (like the appellants) from West Yorkshire; but ultimately they decided that was not necessary because “the system considered the reference population appropriate for the tasks”. Mr Harrison explained that the software’s reference population of 100 is not to be thought of in the same way as the database used for DNA analysis. He said that the purpose of the reference sample is to assess how usual or unusual a feature of speech is, and so to show whether the score is what one would expect as between the questioned and the reference samples. He said that there is no regional variation in vocal tracts, and therefore it was not particularly important to know where the persons in the reference sample came from. He also said that their work had been peer-reviewed; and there had been no criticism of the size of the reference population.

Professor French and his colleague had reported that when 27 different voices were compared with the known voices in the BMW and Peugeot recordings, the system showed a likelihood ratio of less than one in 63% of cases but (incorrectly) identified the speaker with a likelihood ratio of more than one in the other 37%. The report said of this –

“In summary, the system obtains the correct result for all same speaker comparisons and for the majority of different speaker comparisons. When it does make an error it is biased towards making false identifications rather than false rejections. In other words the bias would be towards acceptance of the prosecution rather than the defence claims”.

In their second report Professor French and Mr Harrison had said that they

“agree with the view expressed in a research paper by Becker et al (2012) that owing to the possibility of errors made by ASR systems in specific cases, it is necessary to accompany an ASR analysis with auditory and acoustic-phonetic analysis … We would not use an ASR system as a stand-alone method for comparing speakers in a forensic case. In our view, it should be used in conjunction with a human-based auditory and acousticphonetic analysis, which is what we have done in this case.” 163.In the light of their testing of the system and their use of it in this case, Professor French and Mr Harrison gave evidence supporting the use of Batvox (and by extension, of ASR systems generally), in conjunction with other methods of analysis, as a reliable means of excluding a suspect’s voice from any similarity with a questioned recording. They did not put it forward as capable of making a reliable positive identification of a suspect as the speaker whose voice could be heard in a questioned recording.

The suggested advantages of ASR are that it is largely independent of the analyst who operates it, so that two persons processing the same recordings through the system should get the same result; that it produces a numerical estimation of likelihood; and that it compares the voice of the suspect not only with the questioned recording but also with the reference population of the system. Neither Professor French nor Mr Harrison was aware of any research or any academic articles which had concluded that ASR systems were not reliable for court purposes; and it is said that such systems are used in court proceedings in many other countries.

Professor French was asked why he would only use the Batvox system in conjunction with other forms of analysis, and not as a freestanding method. He said that Batvox reflects, mathematically, the resonance characteristics of the vocal tracts. That is the only aspect of the voice which is measured. It is possible that two men may have similar vocal tracts, but differ for example in accent or pitch; therefore use of ASR alone “could give a false hit”. He accordingly felt it better to use Batvox in conjunction with testing of other characteristics which Batvox was not designed to test. He accepted in cross-examination that whereas in relation to his conventional testing he could explain to a jury what distinctive features he had found, in relation to Batvox he could only explain that the system identified differences in the geometry of vocal tracts: but he would not be able to say precisely what those differences were.

He also said that some of the different ASR systems are “more conservative” than others; but he did not know which nor where Batvox was positioned on that spectrum. It was put to him that this would seem to raise the possibility of different systems producing different results. He accepted that; but he said that he did not expect that one would have one system producing a likelihood ratio below 1 and a different system producing a likelihood ratio above 1 in respect of the same sample. However, he conceded that in one of his tests, which compared the voice of Mr Harrison with the voice of Pearman, the result had been a likelihood ratio of 0.8, and he further conceded that he could not say whether a different system might have given, for example, a likelihood ratio of 1.1.

Professor French was asked why, in view of the claimed advantages of ASR, he had not used the system in any of his forensic casework apart from this case. He said that in this case, he had reference recordings of the known voices of the appellants speaking in a car: without those, he would have felt it necessary to do wider testing on generic voices in cars before using ASR as part of his analysis.

In his evidence for the respondent, Mr Hirson made clear that although he has many years experience as a senior lecturer in phonetics at the City University, London, he has very little, or no, experience of actually using an ASR system. His objections are based on principle, not (as he openly accepted) on technical knowledge of the machine. His objections therefore relate to all ASR systems. The nub of his objection is that it is asserted that Batvox relates to the geometry of the vocal cord; but he regards that as speculation. If it does so relate, then the properties of the vocal cords do not correspond directly to the sounds of speech. He regards the presentation in mathematical form as spurious science, a misleading portrayal of quantitative data. He was not satisfied there has been a sufficient testing of the Batvox system; because a proper evaluation would in his view require a comparison of hundreds of speech samples under conditions similar to those in this case. Although he was much criticised by the appellants’ counsel on the grounds of lack of objectivity and of lack of expertise in relation to ASR systems, he was, in our view, entitled on the basis of his expertise in the field generally to set out those objections of principle for the court’s consideration.

On behalf of the appellants, counsel submitted that the court should receive the evidence of Professor French and Mr Harrison as fresh evidence pursuant to section 23 of the 1968 Act, in accordance with the principles conveniently summarised in Chattoo (cited above). Counsel submitted that the Batvox system should be admissible in evidence as an aid to existing methods of voice identification, even if not as a freestanding diagnostic method. They emphasised that the Crown – though well aware of the issues – had not sought to adduce any evidence to contradict the evidence of Professor French and Mr Harrison as to the reliability of the operation of the system. They submitted that, consistently with Professor French’s evidence, it may be that the system cannot be used as a positive identification of a speaker, because it cannot express probability by reference to a large database; but it is, they submitted, capable of categoric exclusion. They argued that the evidence shows that, for that purpose, the system picks an appropriate sample; and no evidence had been adduced to the contrary.

For the Crown, it was submitted by Mr Greaney that the Batvox system was available to Professor French at the time of the trial, that accordingly this is not fresh evidence and that the court should uphold the important one trial principle. Further, it was the appellants who sought to have this evidence admitted and it was therefore for them to establish its admissibility, whether or not specific rival evidence was called. Mr Greaney in addition put forward a number of reasons why the court should not regard the Batvox system as reliable, and should hold the evidence of the results of using that system to be inadmissible. He argued, in particular, that the Batvox results are expressed in the form of likelihood ratios, with no clear basis for asserting that such is appropriate; and he invited the court to conclude there are far too many variables and uncertainties in the data relied on by Professor French and Mr Harrison for them to be permitted to express an opinion based on the use of mathematical formulae.

In support of the last point, Mr Greaney also relied on the decision of a constitution of this court in T [2011] 1 Cr App R 9, in which the issue was raised as to whether it permissible to use mathematical formulae and likelihood ratios based on statistics to arrive at an evaluative opinion in footwear mark cases.

The court referred in that case to the permissible use, in DNA cases, of match probabilities “not directed to whether DNA came from the suspect but to the probability of obtaining a match that came from an unknown person who is unrelated to the suspect but has the same profile”. The court continued at paragraph 78:

“However, no case was drawn to our attention which suggests that a mathematical formula is appropriate where it has no proper statistical basis. ... If there are reliable statistics and data, it would then be necessary to consider how likelihood ratios should be used and how their use should be explained to a jury.”

The court then considered the reliability of the statistics and data in relation to footwear. It concluded that there were far too many uncertainties and variables in the data to enable an expert to express an opinion based on a mathematical formula, saying at paragraph 86:

“There are no sufficiently reliable data on which an assessment based on data can properly be made for the reasons we have given. An attempt to assess the degrees of probability where footwear could have made a mark based on figures relating to distribution is inherently unreliable and gives rise to a verisimilitude of mathematical probability based on data where it is not possible to build that data in a way that enables this to be done; none in truth exists for the reasons we have explained. We are satisfied that in the area of footwear evidence, no attempt can realistically be made in the generality of cases to use a formula to calculate the probabilities. The practice has no sound basis.”

That conclusion made it unnecessary for the court to consider, how, in the context of footwear impressions, the use of likelihood ratios should be explained to a jury. The court nonetheless referred briefly to Adams [1996] 2 Cr App R 467, Doheny [1997] 1 Cr App R 369 and Adams (no 2) [1998] 1 Cr App R 377. A review of those cases led to the conclusion at paragraph 90 that:

“It is quite clear therefore that outside the field of DNA (and possibly other areas where there is a firm statistical base), this court has made it clear that Bayestheorem and likelihood ratios should not be used.”

We turn to our conclusions about this aspect of the appeal.

We would not think it right to refuse to receive the evidence of Professor French and Mr Harrison solely on the ground that it was available at the time of trial. We accept the submissions of the appellants’ counsel to the effect that it would be unfair to hold it against Professor French that he had taken time to test and research the system before deciding that he could properly use it in his forensic work.

For other compelling reasons, however, we do not think it necessary or expedient in the interests of justice to admit the evidence of Professor French and Mr Harrison. Even accepting that their evidence is fresh evidence for the purposes of section 23 of the 1968 Act, we find a number of areas of concern which lead us to conclude that the appellants have failed in this case to justify the use of a likelihood ratio, generated by ASR software, as a means of determining voice identification or exclusion. In view of our overall decision on other grounds of appeal, however, it is neither necessary

First, we are far from persuaded that the very small reference population selected by the software provides a sufficient basis for a reliable conclusion. T emphasised the importance of a proper statistical basis if the use of a likelihood ratio is to be justified. We accept the submission of the Crown that it has not been sufficiently convincingly demonstrated in this appeal that a group of 20 or 30 speakers adequately covers all the possible features of the geometry of human vocal cords. The outcome of the testing, in which a likelihood ratio of more than 1 was incorrectly found in over a third of the tests, does not inspire confidence. Professor French gave evidence that he was not troubled by those results, because he was focusing on the use of the system to exclude, rather than include, a suspect. We do not share that response. Whichever way the error goes, it is an indication of fallibility which has not been satisfactorily explained. In the context of evidence which is said to enable the categorical exclusion of a suspect, on the basis of a comparison with the voices of 20 or 30 speakers whose ages and accents may differ substantially from those of the suspect, such fallibility is troubling. The fact that ASR systems are, as we were told, used for forensic purposes in other jurisdictions does not resolve those concerns; because we heard no evidence as the purpose for which the technology is used in those jurisdictions, or as to whether its use there is subject to any qualifications or limitations. Further, such jurisdictions might have quite different rules of evidence compared to those pertaining in England and Wales.

Secondly, in a number of respects it seems to us that the evidence ultimately amounts to little more than a bare assertion that the software is so designed as to ensure the right results: with no explanation of how the court can be confident that is so. For example, the selection by the software of the subset of voices from the reference population has not been explained; and no clear reason has been shown why the court should simply accept the assertion that the system has made the best choices. It does not seem to us to be a sufficient answer to this concern to say that it is only proposed that ASR should be used in conjunction with other forms of analysis. In this case, indeed, it is apparent that Professor French was caused to adjust his opinion by the results of the Batvox testing, which – in his words – took him beyond his original conclusion that the voice of M1 was very unlikely to be that of Mr Pearman. It is also apparent that the Batvox results, amounting effectively to categorical exclusion or elimination with a fairly high degree of confidence, were inconsistent with the evidence of Mrs McClelland and with the views (so far as made known to this court) of Dr Holmes and Mr Duckworth. The court therefore has to consider that it is being asked to admit what in actuality is advanced as decisive evidence, not simply supportive evidence.

Thirdly, we are concerned that counsel for the appellants were unable to give any satisfactory answer to the question of how evidence of this nature should or could appropriately be presented to a jury. It seems that even a witness as distinguished as Professor French would in reality be telling a jury that the system had produced a certain result, with no real explanation of what features had contributed to that result and therefore no real scope for cross-examination. How are the jury to evaluate such evidence? What, for example, are a jury to make of the proposition that the results of the Batvox tests are 38 times more likely if the suspect was not the man whose voice can be heard in the questioned recording than if he was? Are they to regard “38 times more likely” as significantly high or insignificantly low? The absence of any satisfactory answer to such questions reflects the fact that the system simply produces a result, expressed in a mathematical formula (with the attendant danger of a potentially misleading appearance of certainty), but without any explanation of which features of similarity or dissimilarity have contributed to that result.

Fourthly, it seems clear from the evidence before this court that different ASR systems may produce different results from testing of the same samples, if only because one system is “more conservative” than another. The submissions on behalf of the appellants were unable to satisfy us as to how a jury should approach such a conflict.

Lastly, making every allowance for a proper wish to test the system and for the fact that the outcome of this appeal was awaited, it was not clear why Professor French and Mr Harrison had not thus far used the Batvox system in any other trial or in other casework, if only for their own purposes and by way of a cross-check on their other analyses. Professor French’s answers on this, when raised with him, were, with respect, hesitant. Nor was it clear to us why the fortuitous availability of the recordings in the Peugeot car was regarded as a reason to use the system in this case even though it was not being used in other cases. These troubling features of the evidence suggested a certain lack of confidence in the system, and in particular in the adequacy of the system’s own reference population. This is only reinforced by the insistence that evidence based on such a system should only be used in conjunction with other auditory recognition/acoustic evidence.

For those reasons we decline to admit this evidence, which in our view left many important questions unanswered.

Although we have rejected many of the points advanced by the appellants we have, for the reasons given above, overall come to the conclusion that these convictions are unsafe. They are therefore quashed.

We have borne in mind that the jury were given a separate treatment direction at trial:

We also have borne in mind that most of the new points and arguments addressed to us by reference to the schedules which we have been prepared to receive and accept related primarily to Baxter (albeit also to some extent, by extension, to Slade). At all events, not all of them, by any means, were directed to Pearman’s position. But understandably in the circumstances, Mr Greaney, as we have already indicated, did

We will receive the submissions of counsel on any consequential applications or other matters arising from this judgment when it has been handed down. Such submissions will need to extend to dealing with the outstanding appeals against sentence (which also include the appeal of Hudson).