England and Wales Court of Appeal (Criminal Division) Decisions

IN THE COURT OF APPEAL (CRIMINAL DIVISION)
ON APPEAL FROM THE CROWN COURT AT SOUTHWARK
The Hon. Sir Vivian Ramsey
T20107189, T20107367, T20107743
T20107769, T20117368

		Royal Courts of Justice Strand, London, WC2A 2LL
		18/12/2015

Sir Brian Leveson P :

1. This is a judgment of the court to which each of the members of the constitution has made a substantial contribution.

Introduction

2. Writ large throughout this case is the ability of the criminal justice system fairly to manage cases (likely, in the main, to encompass allegations of very substantial fraud) which comprise or comprehend a vast electronic database through the techniques of disclosure which have been developed through the Criminal Procedure and Investigations Act 1996 ("the CPIA") and the various protocols and guidelines which have been issued in an attempt to do so. Thus, it is common ground that, in this prosecution, many computers have been seized containing some 7 terabytes of data. The prosecution case has long since been served, as have prosecution case summaries, updated as time has passed. For five years, however, while proceeding in the Crown Court at Southwark, the case has not progressed beyond what has been contended is necessary for primary disclosure. Neither has this state of affairs come about for want of judicial intervention.
3. Ultimately, by ruling dated 1 May 2015, Sir Vivian Ramsey (who had retired from the High Court bench by the time he issued the ruling) stayed the prosecution in respect of all counts of a draft indictment (which had not reached the stage of being preferred) as an abuse of process. He did so on the basis that the prosecution had for so long failed to comply with its duty of disclosure, a fair trial was no longer possible although he also spoke of prosecutorial misbehaviour. This is an application by the prosecution for leave to appeal that decision on the grounds that the judge had adopted an incorrect approach to the issue of initial disclosure and, in any event, having regard to the issues in the case and all the circumstances, had been wrong to stay the entire prosecution.
4. Notice of intention to appeal, provided pursuant to s. 58 of the Criminal Justice Act 2003 ("the 2003 Act"), was given within the period allowed by the judge and it has been acknowledged that the respondents will be entitled to be acquitted should leave not be obtained or the appeal abandoned.  The application has been referred to the full court by the Registrar on the basis that if leave is granted, the appeal will follow.  Although each of the respondents to the appeal has had separate representation, the thrust of the case in response to the appeal has been advanced by Brendan Kelly Q.C. on a common basis. There have been few idiosyncratic arguments in relation to individuals.
5. In the light of the issues as to the extent of the present operation of the law relating to disclosure in cases of this type, we invited the Attorney General to intervene: he did so and we are grateful for the assistance that Mr Richard Whittam Q.C., on behalf of the Attorney General, provided. Furthermore, because of concerns expressed about the impact of decisions of the Legal Aid Agency in relation to the case, we also invited representations and received submissions from that quarter. In the event, it quickly became clear that decisions as to legal aid had not, in fact, had any impact on the conduct of the case and we deal with such issues only very briefly.

The Nature of the Case

6. The prosecution arises out of an investigation, initiated by HM Revenue and Customs ("HMRC"), into a large-scale tax mitigation scheme which the eight respondents are said to have dishonestly created and sold, thereby placing a very large amount of tax at risk. In summary, it is alleged that between 2005 and 2007, the respondents were involved in setting up a number of UK Limited Liability Partnerships ("LLPs") to create and to trade in Carbon Emissions Reduction Certificates ("CER certificates"). These certificates are used to meet emissions targets set in accordance with the Kyoto Protocol and UK legislation. The LLPs were marketed to wealthy investors who would invest in them by providing 20% of the investment from their own funds and borrowing the remaining 80% from the Environmental Guarantee Corporation ("EGC"), an off-shore lender incorporated in the Isle of Man and set up for the purpose of providing these loans.
7. The partnerships were managed by Carbon Capital Limited ("CCL"), a UK registered company. Its directors included two of the respondents, Adam Page and Michael Richards. The prosecution contend that another of the respondents, Robert Gold, also participated in the company at a managerial level.
8. Once an LLP was funded to the extent of £8.5 million, CCL, on behalf of the LLP, would enter into a contract with Carbon Positive Trading Limited ("CPT"), under which £7.1 million was paid to an offshore researcher to undertake scientific research on land over which the LLP had an option to purchase (the "Scientific Research Agreement"). The intention was for that land to become a cheap source of CER certificates if the research confirmed that carbon sequestration could be achieved there.  Because the LLPs therefore had undertaken only limited trading in CER certificates but had invested (including the loan capital) very substantial sums on research and development, the LLPs made a very large loss (just over 80%) such that the high net worth investors could claim loss relief in relation to their investment. On this basis, investors who had invested 20% of their own money could obtain 40% tax relief on 80% of the total investment, which effectively meant that they could obtain tax relief amounting to 160% of their actual cash investment. A loss of £7.1 million gave rise to claims of approximately £2.84 million.
9. In total, 38 LLPs were subscribed. The commercial model changed over time, with the focus moving from carbon sequestration and reforestation in Brazil, to the growth of bio-fuels in China. But the essential framework remained the same. The amount of tax at stake rose to £107 million on expenditure allegedly incurred of £269.8 million. The amount of cash actually contributed by investors amounted to just over £64.6 million.
10. It became clear to HMRC by the summer of 2005 when the first of the claims for loss relief were made that these LLPs were collectively a tax incentivised scheme, although its existence had not been declared under the rules in relation to marketed tax avoidance. The nature of the lending, the lender itself, the research, and the research expenditure all came under scrutiny.
11. The original prosecution case was that the research and development contracts were a sham; there was no research and development taking place or, at least, not to the extent allegedly agreed. Instead, the true and substantial purpose of the scheme was the personal enrichment of the respondents; the investors' money was being transferred between the various companies involved before ending up in trusts owned by the respondents. However, during the course of the proceedings, and when challenged in the hearing before us, Mr Charles Miskin QC, for the appellants explained why the scheme was a sham whether or not research and development had, in fact, taken place. Three principal reasons were set out.
12. First, the contracts for research and development were not negotiated at arm's length by unconnected parties; rather they were made between connected parties in common ownership. The principal transactions were not commercial because the separate companies were not acting in their own interests and the pricing was not independently reached. In particular, CPT, the company with which the research and development agreements were made, was not a separate and genuine third party, but was set up by Michael Richards and Robert Gold with the assistance of two of the other respondents, Rodney Whiston-Dew and Evdoros Demetriou. The director of CPT was Jon Anwyl, another of the respondents. Therefore the respondents had true beneficial ownership and control over the company. CPT was incorporated in the BVI in 2004 at around the same time as the incorporation of another company, Carbon Positive Limited ("CPL"). Their registered addresses were the same. The case for the prosecution is that CPL is the 100% owner of CPT. As for CPL, the majority of its shares were owned by two trusts of which Michael Richards and Robert Gold were the predominant beneficiaries. The research and development contracts between CCL (on behalf of the LLPs) and CPT were therefore not made at arms length. The respondents – particularly Michael Richards and Robert Gold – were at the same time investors in the LLPs, playing important roles in CCL and yet had a degree of control over CPT.
13. Secondly, the 80% loan funds to be provided by EGC to the individual investors did not exist (other than on paper) and were therefore not available to complete the research contracts entered into. Before the scheme commenced, there was no real money in EGC or CPT beyond the cash contributed by the investors and EGC was therefore not in a position to make the capital loans. The prosecution say that the 20% cash contributions made by LLP members were first transferred to CPT. Once CPT had accumulated a sum equivalent to the total value of the purported capital loan required in order fully fund a particular partnership, this money was passed by CPT to EGC. EGC then purported to lend what had, in fact, originated as the investors own money, back to some of them in order to appear to complete the 100% funding of the first partnership. The LLP passed the monies back to CPT which then transferred it again to EGC to fund the loans to the next LLP in line and so forth until the first group of 8 LLPs were "funded" completely. The scheme needed at least 4 LLPs to work (4 x £1.7 million investment equals the £6.8 million needed for the "loan"). The money in the scheme was at all times under the control of the conspirators and circulated between the various entities under their collective control.
14. Thirdly, the invoices and contracts were merely a means of ensuring substantial sums were paid to CPT by the LLPs, it was never intended that £269.8 million would be spent on research and development. The cost of research and development was fixed; it was set at a level and payment demanded up front in order to facilitate the circulation of funds. In terms of actual expenditure, of the £64.6 million obtained from the investors, the prosecution say that £7.8 million went to a bank in Singapore, DBS, where it was probably mostly spent on the project in China and a further £4.4 million went to Sunshine Technology in a similar fashion. Perhaps as much as £400,000 was spent on research and development in Brazil, however £19.5 million went to the respondents. The prosecution contends that the investors' funds were placed in CPT's bank account in the Netherlands where they were transferred to an account in Switzerland in the name of CPT's parent company, CPL. From there, funds were moved into trusts which were set up and beneficially owned by the respondents.
15. The real issues in the case, as far as they can be established at this early stage, are of central importance to the question of what was sufficient for primary disclosure in this case, which is why it has been necessary to set out the allegations in more detail. The respondents argue that these schemes were entirely legitimate, took account of the prevailing legislation properly to claim appropriate tax relief and conducted research and development fully to justify the claims made. They say they need to demonstrate that the research and development took place and to reconstruct the business to show that the scheme was genuine; therefore they need to see all material that would enable them to do so.
16. However, the prosecution submit that it does not matter whether some research and development in fact took place, and they do not dispute that some did occur in Brazil and China, because the crux of the sham was in the setting up of a bogus structure and lying about it. They say the case is about accrued expenditure, not actual expenditure, and that the real question is whether at each tax year end, the costs invoiced by CPT to the LLPs were properly accrued in the accounts of those partnerships. If they were, they attracted sideways loss relief. If they did not, that did not mean that there was no commercial project, it simply meant that there was no legitimate tax relief.
17. The case advanced by HMRC has been reduced into a draft indictment which alleges two offences of conspiracy to cheat the public revenue (in relation to the scheme), two offences of conspiracy to launder money and one offence of money laundering, four individual offences of cheating the public revenue (in relation to personal tax returns) and one offence of conspiracy to pervert the course of public justice. Because of the remaining issues surrounding initial disclosure, applications have not yet been made to dismiss, the indictment has not been preferred, defence case statements have not been lodged: in short, the case has been stuck to the grave disadvantage of the respondents (who have not had the opportunity of moving on with their lives), but also the public interest in the proper resolution of what is an extremely serious allegation said to involve very substantial loss to the revenue.

The Law Relating to Disclosure

18. The prosecution has long been under a duty to disclose to the defence any unused material in its possession, that is to say material that is not part of its formal case against the defendant, which either weakens its case or strengthens that of the defendant. The central importance of proper disclosure of unused material was underlined by Lord Bingham in R v H [2004] UKHL 3, at paragraph 14:

"Bitter experience has shown that miscarriages of justice may occur where such material is withheld from disclosure. The golden rule is that full disclosure of such material should be made."

19. This duty was put on a statutory footing in section 3(1)(a) Criminal Procedure and Investigations Act 1996 ("CPIA") which provides a single test for disclosure and requires the prosecution to:

"…disclose to the accused any prosecution material which has not previously been disclosed to the accused and which might reasonably be considered capable of undermining the case for the prosecution against the accused or of assisting the case for the accused."

20. "Prosecution material" is defined in section 3(2) as material:

(a) which is in the prosecutor's possession, and came into his possession in connection with the case for the prosecution against the accused, or

(b) which, in pursuance of a code operative under Part II, he has inspected in connection with the case for the prosecution against the accused.

21. The scheme of the statute proceeds in stages; the primary disclosure required by section 3 is intended to be followed by the service of a defence statement setting out the nature of the accused's defence, including any particular defences on which he intends to rely, and indicating the matters of fact on which he takes issue with the prosecution (section 6A CPIA). Where the prosecutor has complied, or purported to comply with section 3, and the defendant has been charged with an indictable offence, the service of a defence statement is compulsory (section 5(1), (5) CPIA).
22. Disclosure of unused material does not end there. Once the defence statement has been served, the defendant may make an application for specific disclosure under section 8 CPIA of any material which he has reasonable cause to believe should have been disclosed pursuant to section 3. Moreover, the prosecution is under a "continuing duty of disclosure", pursuant to section 7A, which requires it to keep under review the question of whether at any given time there is material which satisfies the test in section 3.
23. As noted in R v H [2004] at paragraphs 17 and 35, section 3 does not require the prosecutor to disclose material which is either neutral or adverse to the defendant; self-evidently, a defendant cannot complain of non-disclosure of material which would lessen his chances of acquittal. More than that, prosecutors have been consistently discouraged from disclosing material that they are not obliged to disclose, not least to avoid over-burdening and distracting the trial process with unnecessary materials.
24. The legislation does not prescribe the method of disclosure, or the process to be adopted by the prosecution; rather it is focussed on the end result: disclosure which complies with section 3. There is however an ample framework of law and guidance, now to be found in the Criminal Procedure Rules ("CrimPR"), the CPIA Code of Practice 2015 ("the Code"), the Attorney General's Guidelines on Disclosure 2013 ("the 2013 Guidelines") and the Judicial Protocol on the Disclosure of Unused Material in Criminal Cases, December 2013 ("the Protocol"), the latter incorporating recommendations contained within the Review of Disclosure in Criminal Proceedings, September 2011, conducted by Gross LJ ("the Review"). For present purposes, nothing turns on the amendments to these various sources post-dating the events with which we are concerned.
25. These materials offer amplification of what the CPIA actually requires, and what would go beyond that. They serve as guidance for the operation of the statutory regime, and how the prosecution ought to go about its task. For present purposes, it is crucial to examine the guidance provided in relation to cases such as this where the unused material is made up of vast quantities of electronic files which would, realistically, be impossible to read and assess in the usual way. How is the prosecution to comply with its obligation under section 3 if it has not read – and could not be expected to read – all the material it has seized? In order to deal with this question it is worth setting out the framework in detail.
26. The scene is set by the Criminal Procedure Rules, underlined by Gross LJ at paragraph 31 of his Review:

"The Rules now consolidate the Court's case management powers and furnish a guide to the underlying culture intended to govern the conduct of criminal trials. Accordingly, the Rules are or should be of the first importance in the proper application of the disclosure regime."

27. In particular, Rule 3.2 imposes a duty on the Court to further the overriding objective "by actively managing the case" which includes "the early identification of the real issues". Rule 3.11(a) requires the Court to establish, with the active assistance of the parties, what disputed issues they intend to explore.
28. Into that picture fits the Code of Practice, required by section 23(1) CPIA, the latest iteration of which records in the Preamble its purpose to set out "the manner in which police officers are to record, retain and reveal to the prosecutor material obtained in a criminal investigation and which may be relevant to the investigation, and related matters". Most notably, the Code sets out the procedure for dealing with relevant prosecution material. First, paragraph 2.1 defines material as relevant if:

" …it appears to an investigator, or to the officer in charge of an investigation, or to the disclosure officer, that it has some bearing on any offence under investigation or any person being investigated, or on the surrounding circumstances of the case, unless it is incapable of having any impact on the case."

29. The Code goes on to provide (at paragraph 6.2) that:

"Material which may be relevant to an investigation, which has been retained in accordance with this code, and which the disclosure officer believes will not form part of the prosecution case, must be listed on a schedule."

30. Scheduling is dealt with in detail (at paragraphs 6.9 – 6.11):

"6.9 The disclosure officer should ensure that each item of material is listed separately on the schedule, and is numbered consecutively. The description of each item should make clear the nature of the item and should contain sufficient detail to enable the prosecutor to decide whether he needs to inspect the material before deciding whether or not it should be disclosed.

6.10 In some enquiries it may not be practicable to list each item of material separately. For example, there may be many items of a similar or repetitive nature. These may be listed in a block and described by quantity and generic title.

6.11 Even if some material is listed in a block, the disclosure officer must ensure that any items among that material which might satisfy the test for prosecution disclosure are listed and described individually."

31. Paragraph 7.1 provides for the disclosure officer to give the schedules to the prosecutor, where practicable, at the same time as giving him the file containing the material for the prosecution case. Paragraph 7.2 provides for the disclosure officer to draw the prosecutor's attention to any retained material which may satisfy the test for prosecution disclosure under the CPIA. Neither the statute nor the Code requires the disclosure of schedules to the accused by the prosecutor; however this has become a requirement in practice for the sake of transparency and in order to command the confidence of the court and the defence that the prosecution has taken the correct approach to disclosure.
32. We now turn to the Attorney General's Guidelines on disclosure of unused material in criminal proceedings and the iteration issued in April 2005. As noted in the Foreword, the Guidelines urged that the disclosure process should not be abused:

"Prosecutors must not abrogate their duties under the CPIA by making wholesale disclosure in order to avoid carrying out the disclosure exercise themselves. Likewise, defence practitioners should avoid fishing expeditions and where disclosure is not provided using this as an excuse for an abuse of process application."

33. The following provisions were intended to assist with the operational approach to paragraphs 6.9 – 11 of the Code of Practice:

"26: … Disclosure officers, or their deputies, must inspect, view or listen to all relevant material that has been retained by the investigator, and the disclosure officer must provide a personal declaration to the effect that this task has been undertaken.

27: Generally this will mean that such material must be examined in detail by the disclosure officer or the deputy, but exceptionally the extent and manner of inspecting, viewing or listening will depend on the nature of material and its form. For example, it might be reasonable to examine digital material by using software search tools, or to establish the contents of large volumes of material by dip sampling. If such material is not examined in detail, it must nonetheless be described on the disclosure schedules accurately and as clearly as possible. The extent and manner of its examination must also be described together with justification for such action."

34. In the Review, Gross LJ drew careful attention to the fact that the 2005 Guidelines did not adopt the approach contained in paragraph 9 of the 2000 Guidelines, often termed "the keys to the warehouse"; in those earlier Guidelines, as summarised in Disclosure in Criminal Proceedings by Corker & Parkinson:

"The solution to this problem… was that if the unused material was too large to inspect and schedule as required by paragraph 6 of the Code, but the possibility that it contained disclosable material could not be eliminated, then not to inspect and schedule but instead to permit the defence controlled access to it. Thus responsibility for ascertaining whether it contained anything of relevance was transferred to the defence…"

35. The later Guidelines issued in 2011 were designed to supplement paragraph 27 of the 2005 version and to meet the rise in investigations where very large volumes of electronic material were found. They are expressed in these terms:

"2. ….The objective of these Guidelines is to set out how material satisfying the tests for disclosure can best be identified and disclosed to the defence without imposing unrealistic or disproportionate demands on the investigator and prosecutor.

3. The approach set out in these Guidelines is in line with existing best practice, in that:

(i) Investigating and prosecuting agencies, especially in large and complex cases, will apply their respective case management and disclosure strategies and policies and be transparent with the defence and the courts about how the prosecution has approached complying with its disclosure obligations in the context of the individual case; and,

(ii) The defence will be expected to play their part in defining the real issues in the case. In this context, the defence will be invited to participate in defining the scope of the reasonable searches that may be made of digitally stored material by the investigator to identify material that might reasonably be expected to undermine the prosecution case or assist the defence."

36. Paragraphs 39 to 45 set out the new guidance on the topic of sifting and examination as prescribed by paragraphs 6.9 – 11 of the Code. In particular:

"41. … It is not the duty of the prosecution to comb through all the material in its possession – e.g. every word or byte of computer material – on the look-out for anything which might conceivably or speculatively assist the defence.

42. In some cases the sift may be conducted by an investigator/disclosure officer manually assessing the content of the computer or other digital material from its directory and determining which files are relevant and should be retained for evidence or unused material.

43. In other cases such an approach may not be feasible. Where there is an enormous volume of material it is perfectly proper for the investigator/disclosure officer to search it by sample, key words, or other appropriate search tools or analytical techniques to locate relevant passages, phrases and identifiers.

44: In cases involving very large quantities of data, the person in charge of the investigation will develop a strategy setting out how the material should be analysed or searched to identify categories of data. Where search tools are used to examine digital material it will usually be appropriate to provide the accused and his or her legal representative with a copy of reasonable search terms used, or to be used, and invite them to suggest any further reasonable search terms. If search terms are suggested which the investigator or prosecutor believes will not be productive – for example because of the use of common words that are likely to identify a mass of irrelevant material, the investigator or prosecutor is entitled to open a dialogue with the defence representative with a view to agreeing sensible refinements. The purpose of this dialogue is to ensure that reasonable and proportionate searches can be carried out."

37. The 2013 Guidelines replace both the 2005 and 2011 versions, with the text of the latter being contained in an Annex. The 2013 Guidelines are intended to operate alongside the Judicial Protocol on the Disclosure of Unused Material in Criminal Cases. These emphasise (at paragraph 3):

"Properly applied, the CPIA should ensure that material is not disclosed which overburdens the participants in the trial process, diverts attention from the relevant issues, leads to unjustifiable delay, and is wasteful of resources. Consideration of disclosure issues should be an integral part of a good investigation and not something that exists separately."

38. Thus, the Guidelines again confirm the important role of the defence statement at paragraph 9. While the expression "dip sample" used in the 2005 Guidelines no longer appears, the 2011 Guidelines reproduced in the Annex still provide for the use of sampling where it would be impossible to manually read and assess the material. Given the continuing duty on the prosecutor under s.7A CPIA to keep disclosure under review, it may be necessary to carry out sampling and searches on more than one occasion: 2013 Guidelines, Annex A, at para.A44. Be that as it may, it is plain that the 2013 Guidelines contemplate the prosecutor, at the stage of initial disclosure, making use of appropriate sampling or the use of appropriate search tools.
39. We turn now to the Protocol prepared following the recommendations of Gross LJ in his Review. These take account of the 'Further review of disclosure in criminal proceedings: sanctions for disclosure failure' prepared by Gross and Treacy LJJ which was published in November 2012. At paragraph 3 it makes clear:

"… [It is] essential that the trial process is not overburdened or diverted by erroneous and inappropriate disclosure of unused prosecution material or by misconceived applications. Although the drafters of the Criminal Procedure and Investigations Act 1996 ('CPIA 1996') cannot have anticipated the vast increase in the amount of electronic material that has been generated in recent years, nevertheless the principles of that Act still hold true. Applications by the parties or decisions by judges based on misconceptions of the law or a general laxity of approach (however well-intentioned) which result in an improper application of the disclosure regime have, time and again, proved unnecessarily costly and have obstructed justice. As Lord Justice Gross noted, the burden of disclosure must not be allowed to render the prosecution of cases impracticable."

40. Picking up one of the recommendations in the Review, the Protocol provides at paragraph 39:

"The legal representatives need to fulfil their duties in this context with care and efficiency; they should co-operate with the other party (or parties) and the court; and the judge and the other party (or parties) are to be informed of any difficulties, as soon as they arise. The court should be provided with an up-to-date timetable for disclosure whenever there are material changes in this regard. A disclosure-management document, or similar, prepared by the prosecution will be of particular assistance to the court in large and complex cases."

41. Finally, it concludes at paragraph 56:

"Historically, disclosure was viewed essentially as being a matter to be resolved between the parties, and the court only became engaged if a particular issue or complaint was raised. That perception is now wholly out of date. The regime established under the Criminal Justice Act 2003 and the Criminal Procedure Rules gives judges the power – indeed, it imposes a duty on the judiciary – actively to manage disclosure in every case. The efficient, effective and timely resolution of these issues is a critical element in meeting the overriding objective of the Criminal Procedure Rules of dealing with cases justly."

42. It is appropriate to conclude this review of the guidelines and protocol by referring to the Review (at paragraph 156) which is in these terms:

"First, in a good many cases it needs to be recognised that it is likely to be physically impossible or wholly impractical to read every document on every computer seized. It follows that there can be nothing objectionable to search enormous volumes of material by the use of sampling, key words or other appropriate search tools; indeed, there is no other way and full use should be made of such tools. The [2005] Guidelines [at para. 27] and, more especially, the 2011 Guidelines [at paras. 41 et seq] deal in terms with such an approach……

43. Neither is it necessary solely to refer to the Rules or attempts to identify or prescribe practice for the authorities also confirm this approach. In R v Brendan Pearson and Paul Cadman [2006] EWCA Crim 3366, complaint was made that the Crown had failed to comply with its duty of disclosure in relation to records contained on computers which had been seized from the business under investigation. The police had not read every record contained on the computers. The complaint was rejected. Giving the judgment of the Court, Hughes LJ (as he then was) said this (at paragraph 20):

"In the course of evidence given during the trial on a voir dire, a computer expert instructed on behalf of the appellant, when asked how long it would take to read all the computer material that the police had seized, said that it would take a lifetime or more. If the submission is made that it was the duty of the Crown to trawl through every word or byte of this material in order to see whether any of it was capable of undermining the Crown's case or assisting that of the appellant, we do not agree…..Where there is an enormous volume of material, as there was here, it is perfectly proper for the Crown to search it by sample or, as here, by key words…"

Hughes LJ went on to add (at paragraph 22) that where sampling of voluminous material was undertaken "…it is the more important that it is explained exactly how it has been done and what has not been disclosed as a result". These were telling observations.

Summary of the Principles

44. As observed in the Review at para. 8 (i) (and later reaffirmed in the Protocol):

"It is essential that the burden of disclosure should not render the prosecution of economic crime impractical. "

This concern is of the first importance and looms large in our thinking throughout. Whatever its cause, the debacle that has been the present case (with five years of litigation not reaching the stage when the indictment has been put) must not be repeated. As the Review went on to say (at paragraph 8 iii): "The tools are available; they need to be used." The issue confronted in the present proceedings goes to the application of these tools in cases with vast quantities of electronic materials, the scale of which has already been described elsewhere in this judgment.

45. From the provisions and material outlined above, it is possible to draw a number of conclusions about the current law and practice on the disclosure of unused material.

The prosecution is and must be in the driving seat at the stage of initial disclosure

46. The CPIA so provides and considerations of practicality demand it. It must be emphasised that at this stage, the true issues in the case may yet be unclear. It is no accident that the statutory scheme places the responsibility for determining whether material falls to be disclosed under section 3 CPIA on the prosecution.
47. In order to lead (or drive) disclosure, it is essential that the prosecution takes a grip on the case and its disclosure requirements from the outset. To fulfil its duty under section 3, the prosecution must adopt a considered and appropriately resourced approach to giving initial disclosure. Such an approach must extend to and include the overall disclosure strategy, selection of software tools, identifying and isolating material that is subject to legal professional privilege ("LPP") and proposing search terms to be applied. The prosecution must explain what it is doing and what it will not be doing at this stage, ideally in the form of a "Disclosure Management Document". This document, as recommended by the Review and the Protocol, is intended to clarify the prosecution's approach to disclosure (for example, which search terms have been used and why) and to identify and narrow the issues in dispute. By explaining what the prosecution is – and is not – doing, early engagement from the defence would be prompted. Plainly such an approach requires early and careful preparation from the prosecution, tailored to the needs of the individual case. This approach is now embodied in the process for document heavy cases forming part of the Better Case Management ("BCM") initiative. Moreover, it is reflected in the approach to "initial disclosure" (see further below) adopted by the Serious Fraud Office, as helpfully summarised in the respondents' Focused Response (at paragraphs 20 - 22).

The prosecution must then encourage dialogue and prompt engagement with the defence

48. As is clear from the Rules, the duty of the defence is then to engage with the prosecution and thus assist the court in fulfilling its duty of furthering the overriding objective. It is plain that compliance with the test for initial disclosure calls for analysis of the likely cases of prosecution and defence. Absent such analysis, it would not be possible to form a view, even at this stage, of which materials would and which would not undermine the case for the prosecution and/or assist the case for the accused.

The law is prescriptive of the result, not the method.

49. This is particularly relevant in respect of case such as this where the prosecution has recovered vast volumes of electronic material. In our judgment, it has been clear for some time that the prosecution is not required to do the impossible, nor should the duty of giving initial disclosure be rendered incapable of fulfilment through the physical impossibility of reading (and scheduling) each and every item of material seized; common sense must be applied. In such circumstances, the prosecution is entitled to use appropriate sampling and search terms and its record-keeping and scheduling obligations are modified accordingly: we strongly endorse the approach adopted in Pearson (supra) and that contained in the extracts from the 2013 Guidelines to which we have referred.
50. The extent of the duty imposed on the prosecution at this stage, while obviously fact specific, must take account that it is initial disclosure with which the prosecution is then concerned. The right course at the stage of initial disclosure is for the prosecution to formulate a disclosure strategy, canvass that strategy with the Court and the defence and to utilise technology to make an appropriate search or conduct an appropriate sampling exercise of the material seized. That searches and sampling may subsequently need to be repeated (to comply with the prosecutor's continuing duty of disclosure under s. 7A CPIA or to respond to reasoned requests from the defence under s.8) is neither here nor there; the need for repeat searches and sampling does not invalidate the approach to initial disclosure involving such techniques. The problem of vast quantities of electronic documents has, in a sense, been created by technology; in turn, appropriate use must be made of technology to address and solve that problem.
51. The prosecution's duties of record keeping and scheduling must likewise reflect the reality that not every one of perhaps many millions of e-mails is to be individually referenced. Thus, the 2013 Guidelines, at Annex A, paras. 45 – 46, reflecting the Code of Practice, qualify the requirement to keep a "record or log" of all digital material seized and subsequently retained as relevant to the investigation in cases "involving very large quantities of data"; in such cases, the obligation is to make a record of the "strategy and the analytical techniques used to search the data". Similarly, the scheduling duty imposed on the disclosure officer separately to list each item of unused material (as contained in the Code) is modified in favour of "block listing" – albeit that it remains the prosecution's duty to list and describe separately "the search terms used and any items of material which might satisfy the disclosure test": 2013 Guidelines, Annex A, at para. A50.

The process of disclosure should be subject to robust case management by the judge, utilising the full range of case management powers.

52. Though decisions are for the prosecutor, such decisions or prosecution failures are not beyond challenge or somehow immune from the court's case management powers. Richard Whittam Q.C., for the Attorney General, supported by Mr. Miskin, sought to advance the proposition that, at the stage of initial disclosure, judicial powers of case management were (in essence) limited to exhortatory observations or guidance; the danger otherwise was that case management powers, as contained in the Rules, would cut across the CPIA scheme. At the stage of initial disclosure, Mr. Whittam submitted:

"It should not be for the judge to be devising the disclosure scheme ….when the issues haven't been identified."

53. We are wholly unable to accept these submissions. As Mr. Kelly aptly put it, the effect of those submissions would be to marginalise the judge's case management responsibilities as they apply at the stage of initial disclosure and would appear to exclude the power of the judge to enforce case management directions then made (or which the Judge would otherwise wished to have made).
54. In our judgment, the judicial task of active and robust case management is emphatically not confined to the secondary or subsequent stages of disclosure. The tenor of the Rules is quite to the contrary. So too are the various authorities, stretching back to R v Jisl [2004] EWCA Crim 696. Faced, for example, with a manifestly flawed, inadequate or inappropriate prosecution approach to initial disclosure, a judge is not constrained to limit intervention to exhortation and some veiled warning as to later consequences. The court is both entitled and obliged to give orders and directions to address the failing with which it is confronted. Neither is the judge required to watch the case become diverted from its proper course, powerless to stop it doing so until much time and costs have elapsed. The wording of s.3 was not intended to give the prosecution carte blanche to under-perform and, as experience has shown, prosecution failures in this area are of real concern: see, the Further Review conducted by Gross and Treacy LJJ, referred to above.
55. Mr Whittam relied on Crim PR Rule 3.5(1) which sets out the court's case management powers is in these terms:

"In fulfilling its duty under rule 3.2 the court may give any direction and take any step actively to manage a case unless that direction or step would be inconsistent with legislation, including these Rules."

Referring to the obligations of the prosecution in relation to primary disclosure as described in CPIA, he emphasised the final words and argued that the exercise of case management powers at this stage would be "inconsistent with legislation". That approach, however, is to misunderstand the simple principle that the Crim PR cannot override primary legislation. The language of Crim PR Rule 3.5(1) does not begin to restrict the panoply of the case management powers available to the judge at the initial disclosure stage. What the judge needs to do is to have regard to the context of the legislation in exercising those powers.

56. Thus, in the context of initial disclosure, it is incumbent on the judge to consider the obligations of the Crown at that stage, bearing in mind the difficulties (where such exist) of ascertaining the real issues in advance of a defence statement. Moreover, when exercising case management powers at this early stage, it is critical for the court to have regard to the structure of the CPIA scheme – initial disclosure (s.3), followed by a defence statement (s.5), the facility thereafter for a reasoned application by the defence for further disclosure (s.8) and the prosecutor's continuing duty to disclose (s.7A). It should also be plain that, when making case management orders at the stage of initial disclosure, a judge should take care not to subvert the statutory scheme by confusing or conflating the various stages in the process.
57. In seeking to constrain judicial case management at the stage of initial disclosure, Mr. Miskin placed considerable reliance on the decision of this court in R v M (Michael) [2003] EWCA Crim 3764 and, in particular, the following passage in the judgment of Auld LJ:

"49. ….. There is no power in the court to direct primary disclosure, and, even if there were, it is difficult to see how the court could approach its task without knowledge of what, if any, issues were going to be taken with the prosecution case. The scheme of the Act is to rely on the prosecutor at that stage to disclose to the defence any unused material which in his or her opinion might undermine the prosecution case, but not to disclose everything available regardless of any conceivable relevance.

50. The machinery for testing the objectivity and adequacy of that disclosure, given the prosecution's incomplete knowledge at that stage of what issues lie ahead, is the scheme of secondary disclosure. Once those issues are identified by the defence in a defence statement, if they are so identified, the prosecutor can then revisit his duty of disclosure, better informed than he was at the primary stage, to form a view as to what further disclosure justice requires in the form of material that might reasonably be expected to assist the defence. If, in the light of his then knowledge, he still does not make adequate disclosure, that is when the court can step in, and stay in. It can then consider the material for itself and direct further disclosure if it considers justice requires it.

51. That is the machinery, and that machinery was invoked here over a number of applications in the course of the trial; and the judges, true to the scheme of the Act, ruled as they did……"

58. Properly analysed, we do not think that Michael assists the appellants. The issue in that case concerned a complaint that the prosecution had not gone beyond the requirements of ss. 3 and 8 CPIA and was firmly rejected, a fortiori, absent defence statements. To the extent that Auld LJ referred to there being "no power" in the court to "direct primary disclosure", that observation was obiter and, with respect, in the context of recent developments in the law, must now be read subject to the increased emphasis on case management. Furthermore, although Auld LJ's summary of the manner in which the CPIA scheme generally works is entirely right, his judgment (at paragraph 50) should not be read as confining judicial intervention exclusively to the "secondary" stage of disclosure. In leaving this point, we note too that the Review did not receive any representations from any consultee suggesting limitations at the initial disclosure stage of the judge's case management powers.
59. That said, when exercising those powers, the judge must, of necessity, keep well in mind that he is then concerned with initial disclosure, with the corollary that the true issues in the case may as yet be unclear. The judge's aim, apart from seeking to hold the prosecution to its duty of giving initial disclosure and insisting on defence engagement, must be to drive the case as expeditiously as possible towards the stage where a defence statement is required, the issues can be crystallised and questions of further disclosure dealt with on a reasoned and informed basis pursuant to sections 7A and 8 CPIA.
60. For its part, the respondents underlined that, whereas ss. 5, 6, 7A and 8 CPIA spoke of the situation when the prosecutor has complied or "purported to comply" with his obligations in question, the terms of s.3 simply provided for the prosecution to give initial disclosure – and said nothing about the prosecutor "purporting" to comply with this obligation. Too much should not be made of this point. First, in context, compliance with the prosecutor's duty under s.3 must mean substantial compliance. Realistically, it cannot be supposed that cases will never proceed beyond the stage of initial disclosure merely because some documents have not yet been disclosed. A search for perfection in this area is likely to be illusory. Secondly, both ss. 5 and 6 provide for a defence statement to be given not only when the prosecutor has complied with s.3 but also when he has purported to comply with it. Progress can and should thus be made, even where it is or may be apparent that further prosecution disclosure might be required in the future. It also follows that cases are not doomed to proceed in compartmentalised, consecutive stages; progress can be made in parallel, both completing outstanding initial disclosure and illuminating the true issues in the case pursuant to ss. 5, 6, 7A and 8.

Flexibility is critical

61. Both the review and all other source materials on disclosure emphasise that it is not to be conducted as a "box-ticking" exercise; see too, R v Olu [2010] EWCA 2975; [2011] 1 Cr App R 33, at [42] – [49] and R v Malook [2011] EWCA Crim 254; [2012] 1 WLR 633. In a document heavy case (whether electronic or paper), there can therefore be no objection in principle to the judge, after discussion with the parties, devising a tailored or bespoke approach to disclosure. That must certainly be preferable to dealing with the matter in a mechanistic and unthinking way.
62. There is also no reason in this regard why lessons cannot be learnt from advances in disclosure in civil procedure: see the Review at paras. 79 et seq. However, whatever the approach adopted, there is one overriding proviso: the scheme of the CPIA must be kept firmly in mind and must not be subverted. The constant aim must be to make progress, if need be in parallel, from initial disclosure to defence statement, addressing requests for further disclosure in accordance with s.8. If this proviso is overlooked, the real danger is that an apparently attractive "shortcut" will turn out to be a dead-end, leaving all concerned bogged down in satellite litigation over initial disclosure.
63. Properly applied, the application of these principles will keep the case within the statutory scheme, hold the parties to their duties thereunder and ensure that the proceedings are dealt with fairly, efficiently and expeditiously, in accordance with the overriding objective enshrined in the Rules.
64. Another matter, relevant to the present case ought conveniently to be mentioned here. While it is right that attention must be paid to the format of the material supplied (see the Review, at para. 159), it is no part of the prosecution's duty under s.3 to improve the material seized.
65. Before leaving this part of the case, three other issues must be addressed. The first is to underline one of the "Overarching Principles" set out in the Review of Efficiency in Criminal Proceedings (2015). The principle is "getting it right first time" and its relevance to the present case arises from the fact that the appellant's stance before this court is substantially different from that adopted before Ramsey J. Before the Judge (as discussed in further detail below), the appellant essentially acquiesced in the Judge's proposals as to disclosure. The appellant's case below was that, with more time, they could and would comply with the requirements canvassed with the parties by the judge. On appeal, the case is that those proposals were misconceived with regard to the stage of initial disclosure, imposed upon them under protest and led the parties and the case onto the wrong road.
66. Changes of case of this nature are disconcerting and potentially very wasteful of time and costs. Whether or not in the present proceedings the appellant is permitted to change its case on appeal, it must be emphasised that parties generally can have no expectation that such a course will be open to them. Save very exceptionally, a party is not permitted to acquiesce in an approach to the case before the judge at first instance and then renounce its agreement and advance a fundamentally different approach on appeal. Parties must get it right first time.
67. The second issue concerns the question of a preparatory hearing. As we have noted, this does not appear to be a case where the approach of the judge was imposed upon an unwilling party. The question does, however, arise as to what can be done to challenge an order made in a pre-trial hearing by a judge where, if the parties are left to a post-trial appeal and the judge turns out to have been wrong, the trial will have proceeded on a false footing, delay can be measured in terms of years and the costs in millions of pounds. Such considerations lead naturally to an inquiry as to whether a preparatory hearing could or should have been sought in this case, pursuant to s.29 of the CPIA. Had there been a preparatory hearing and had there been a dispute as to the approach to disclosure favoured by the judge, it could have been challenged by way of appeal from such a hearing rather than many years later by way of an appeal arising from the decision to stay proceedings; it is inherently likely that any such appeal would (or certainly could) have been brought to this court some years earlier.
68. It is plain from their responses to us that the parties did not even consider asking for a preparatory hearing. Additionally, the respondents submitted that a preparatory hearing would have done no good, given the very late change in the appellant's case as to disclosure; on the face of it, there is force in this submission though it is very possible that an earlier appeal by way of that route might have brought about an earlier change in the Crown's approach.
69. The observations which follow therefore look to the future rather than to the present case. In general, parties are discouraged from seeking preparatory hearings. In R v I-I [2009] EWCA Crim 1793; [2010] 1 WLR 1125, Hughes LJ (as he then was), at [21] – [22] observed that given the "co-extensive powers of case management outside the preparatory hearing regime", courts ought to be "very cautious" about directing such a hearing. The mere desire of one party to test a ruling by interlocutory appeal was not a good enough reason for doing so unless the point was one of the few "…genuinely suitable for that procedure and there is a real prospect of such appeal being both capable of resolution in the absence of evidence and avoiding significant wastage of time at the trial". The case then before the court was, however, one of those few; the point was:

"….discrete, novel, certain to arise rather than hypothetical or contingent, involved no factual dispute and needed authoritatively to be determined lest the trial proceed on what might turn out to be a false footing, with consequent risk of the necessity of retrial."

70. We are bound to agree that preparatory hearings should be very few and very far between. Were it otherwise, a glut of interlocutory appeals would overload this court and timetables, both for the proceedings in question and other appeals, would be subject to serious disruption. In an exceptional case, however, where there is essentially a discrete dispute of law (not fact) as to the approach to be followed by way of disclosure, consideration might be given to a preparatory hearing. The advantage of doing so is that it would facilitate an interlocutory appeal to this court, with the attraction of preventing the case from proceeding on a false footing and saving the parties from very substantial losses of time and money.
71. The third issue concerns the position of the Legal Aid Agency ("LAA") and the extent to which its role and responsibilities have impacted on the progress of these proceedings. At this stage, it is only necessary to draw attention to Recommendation 16 of the Review which was in these terms:

"We would welcome more widespread and formalised cooperation between the Court and the LSC [the forerunner of the LAA] – extending to attendance by the LSC at PCMHs where appropriate – to assist the Court with addressing the practicalities in time, approach and costs flowing from an order for disclosure and to assist the LSC with the identification of the real issues in the case. The detail of such cooperation should be considered further in consultation between the professions and the LSC, to be followed by appropriate consultation with the Judiciary."

72. Subsequently, in July 2013, a Protocol ("the LAA Protocol") was finalised and signed on behalf of the LAA and by Gross LJ, as Senior Presiding Judge. Copies of the final version were thereafter circulated, inter alia, to various professional organisations representing both barristers and solicitors. Unfortunately, the LAA Protocol does not appear to have been publicised more widely, though evidence from the LAA placed before us states that the Protocol "codified pre-existing best practice regarding transparency in its communications with the judiciary and the prosecution in [Very High Cost Cases referred to as] VHCCs". The LAA's position is that its staff continue to comply with the LAA Protocol. In our judgment, the LAA Protocol should be revisited by the office of the Senior Presiding Judge with a view to its more widespread dissemination. In relation to the most complex cases, subject to necessary safeguards, there are times when the court can assist the LAA in supplying a clear focus upon the issues (thereby allowing them to focus funding arrangements appropriately) and other times when the expertise of the LAA could prove of assistance to the court.

Chronological Analysis

73. Turning to a brief chronology, as set out above, HMRC became aware of the scheme in 2005 when various individuals made claims for loss relief against their tax. On 19 July 2007, search warrants were executed at 6 business premises and 16 residential addresses; four of the respondents were arrested (with the other four respondents being arrested in August 2007, September 2007, March 2009 and June 2011 respectively). A substantial amount of hard-copy material and electronic material held on computers and other digital media was seized during the searches: in total, 85 computers and other similar electronic devices were seized, along with 152,865 pages of hard-copy documentation. The computers contained some 7 terabytes of information.
74. In addition to the material held on the 85 digital devices, the prosecution served disclosure schedules which addressed all the other relevant items gathered or created by the investigators. This included material generated by HMRC's civil investigation and any internal HMRC electronic documentation. We are informed that initial disclosure in this context (viz. the non-digital material) was complete by July 2012, although disclosure remained continuously under review. It is of note that by 24 August 2014, what has been called the Non-Sensitive Disclosure Schedule ('NSDS') contained 25,892 items. 
75. In relation to the 7 terabytes of seized electronic material, the Digital Forensic Group of HMRC ("DFG") placed images of all the material onto their system before the computers were returned to those from whom they were seized. As a result, the prosecution was dealing only with 'imaged copies' of the computer drives. Furthermore, the respondents had full access to everything on their own computers and were therefore in a position to recreate the nature of their involvement by reference to contemporaneous material as well as to demonstrate that in which they were involved or of which they had knowledge.
76. On 5 November 2007, HMRC sent letters to all those from whom digital or other material had been seized, inviting indications as to whether, inter alia, the computers contained documents that might attract legal professional privilege ("LPP"). In the early part of 2008 the independent counsel and those representing the then suspects accessed some of the material held by DFG. Relevant items were 'bookmarked' for LPP purposes, and those files were removed before the HMRC investigation team began the process of identifying potential evidence in the remainder of the files. Hard copy packs of documents were produced and distributed to each of the suspects in advance of their interviews in April 2008.
77. On 7 April 2008 a solicitor acting for one of those due to be interviewed informed HMRC that the interview packs appeared to reveal that the investigators might have been looking at documents from sources containing LPP material which had yet to be reviewed. The HMRC investigating team thereon ceased work and independent counsel carried out an inspection. On 23 November 2009 the investigating team was once again granted access to the computer material that had been seized.
78. In addition to retaining independent counsel for this purpose, the prosecution used forensic software called FTK v1.7. This was a reputable digital forensic application, and it had an acknowledged facility for presenting emails as well as an efficient search facility. It would appear that, at the time, it was an appropriate tool to use for this purpose, particularly given the need to apply many thousands of legal professional privilege 'bookmarks'. It is of note, however, that problems in this context had not disappeared because on 13 November 2010 it became apparent that some items bookmarked as LPP had been included in documentation provided to HMRC.
79. We observe in passing that the respondents have argued that the evidence revealed a considerable number of problems with the accuracy of the metadata fields in the dataset, following the use of FTK, by reason of human error or because insufficient data was transferred, or because of differences that resulted from the application of different forensic tools, corruption or the presence of foreign characters. As Ms Malcolm succinctly submitted (day 3, page 38, 41):

"Where the matter went awry … is that … FTK, which was the investigative tool, was never an e-disclosure tool. …. That is the point where this case departed from the proper procedure that is exercised across the other authorities and in particular the SFO. …

"[H]ad the underlying data … had integrity, this case would have been, it may sound ridiculous now, it would have been lauded as the way to deal with these matters, properly, efficiently, pragmatically and using case management skills."

80. Meanwhile, on 4 March 2010, three of the respondents (Michael Richards, Robert Gold and Rodney Whiston-Dew) were charged. This led to a preliminary hearing before the Recorder of Westminster on 12 July 2010 at which the prosecution were ordered to serve their primary evidence by 10 December 2010, together with a full case summary; a direction was given that the case was to be tried in October 2011. At a further hearing on 24 November 2010, before Bean J, the respondents were ordered to serve statements of issues in response to the case summary served by the prosecution. The prosecution contend that this was never done in practice.
81. The next directions hearing was fixed for 20 April 2011 before Ramsey J (who had been appointed to manage the case and then conduct the trial) which was six days after the case statement drafted by prosecution counsel had been served. At this directions hearing, disclosure of the electronic material was discussed. At this stage, the disclosure officers were in the process of scheduling all of the seized material, including the electronic material. Mr Miskin underlined that the principle of initial disclosure was whether unused material undermined the prosecution case or assisted the defence and that the law did not allow the defence to be provided with "the keys to the warehouse door". Lord Carlile, then acting for one of the respondents identified what was required in these terms:

"[O]ne of the key issues from [my client's] point of view: was there a trade, and if there was a trade was it a trade which probably fell within tax considerations. ... That means that if there is evidence of trade, that materially assists the defence case or materially undermines the prosecution case potentially."

82. In other words, it was suggested that in order for the defence to challenge the evidence that the scheme was a sham from the outset, it was necessary for them to be in a position to recreate the entirety of the business to show that it was operating as a legitimate enterprise. This proposition ignores the fact that the concept of a sham ("made to appear what it is not") not only encompasses an operation in which there is no trade (which would be undermined by proof of business) but also an operation which falls foul of the characteristics identified by Mr Miskin to which we refer above.
83. At this stage, Ramsey J noted that scheduling volumes of e-material had been abandoned in civil proceedings several years ago. He was concerned that detailed scheduling would deprive the respondents of timely access to the material and gave an early indication that consideration should be given to e-disclosure (providing respondents with digital material wholesale on a searchable sub-database).
84. It had originally been envisaged by the prosecution that disclosure would be effected by disclosing documents on a computer-by-computer basis, in light of the prosecution search terms that had been provided by the individual defendants. The approach of the Disclosure Officers was to dip sample the devices. They considered 10% of the files that had been 'hit' by the search terms and which they had reviewed, and 1% of the files 'hit' by the search terms but not reviewed by them. They disclosed the results of that dip sampling to all the defendants, and by the hearing on 7 June 2011 55 of the digital devices had been dealt with in this way. The disclosure officers worked towards providing a schedule of the unused material in accordance with a Draft Disclosure Protocol that had been provided to the court in advance of the hearing listed on 7 June 2011.
85. In any event, at the April hearing, the judge directed the prosecution to set out a statement as to its position on disclosure. The prosecution served a Disclosure Update and a Draft Disclosure Protocol, stating that initial disclosure would be completed by 31 August 2011. Thereafter, although the case proceeded throughout on the premise identified by Lord Carlile, one of the important issues in this application is whether that is a correct analysis of the position. Mr Miskin contends that it is not, but that the Crown were driven down the path that represented that approach and that to such extent as thereafter they acquiesced, it was because (a) there was no mechanism to challenge it; and (b) in any event, it was felt that the disclosure exercise being undertaken would satisfy the requirements of that approach.
86. At a further hearing before Ramsey J on 7 June 2011, the method of providing disclosure of the electronic material was further discussed. The prosecution proposed to apply search terms chosen by them and provide the respondents with the results of the so called "dip sampling process". Both the judge and the respondents expressed concern at this proposal. Ramsey J made the point:

"If you apply the undermine and assist analysis to that, I don't see how you can do it without going through each of those documents on the sub database. … Therefore and I put this forward in the civil context there is now a reverse burden of disclosure which is that a claimant doesn't have to go through to check that there is standard civil disclosure within the sub database, it produces it to the other side … and because it's not a copy, the other parties can then interrogate it with their search terms … I know warehouse keys and so on can be mentioned in connection with that, but that is the practical way in which [it] has to happen because, as you say, for the prosecution to go through each, this matter won't come on for five years. "

87. Thus, the judge suggested an alternative method of disclosure whereby for each computer or digital media item, the prosecution would serve an expanded description of the item, a printout of its directory setting out folder names, and a list of the search terms that had been applied to that item. The respondents could then propose any additional search terms.
88. Effectively, therefore, the proposal (influenced by discovery in civil law) was that the prosecution would provide all of the material within a database so that it could be interrogated by the defence teams. Subject to any response from the Crown by 21 June 2011, he directed that for each computer or other digital device, the prosecution was to serve an expanded description of the device, and identify from where it had been seized. A printout of its directory structure was to be provided, setting out of the folder names and a list the search terms that the investigators applied. The defendants were then to be asked to propose any additional search terms for any particular folders. Finally, the Crown was to apply all of the search terms to the individual devices and disclose a dataset of all of the documents or e-mails identified.  Thus, the consequence of the judge's proposal was that the prosecution was to effect disclosure of all the materials identified by the defence search terms. The prosecution agreed to consider the proposal (although there was little, if any, alternative) and a trial date was fixed for 10 September 2012. 
89. The prosecution subsequently communicated its agreement to the alternative method proposed on 20 June 2011, subject to being allowed time to comply with it. If search terms were provided by the respondents by the end of July 2011, the process ought to be completed by October 2011. It proposed using Access Data Forensic Toolkit software ("FTK software") to create a report and database to be provided to the respondents. We note in passing that email threading had not emerged as a requirement at this stage, and underline that the obligation on the prosecution was to provide separate datasets for each of the electronic devices as opposed to a single unique global dataset for each defendant, which included the CPIA and the PACE disclosed material.
90. On 6 July 2011 the complete folder structures of the computer drives of the computers were served on the defence and the respondents each served a list of search terms. The consolidated list contained a total of some 3,791 search terms which was clearly unworkable. Thus, on 29 September 2011, Ramsey J directed the respondents to provide a reduced list of search terms. The prosecution were also required to disclose the documents identified and reviewed during the initial sift of the material by the investigation team as soon as possible. This was the classic tick list/unused material.
91. At a directions hearing on 24 October 2011, it became apparent that the FTK software proposed was not be capable of carrying out the necessary optical character recognition search of PDF documents ("OCR"). DFG proposed updating the software used from FTK v1.7 to FTK3 which did have an OCR facility and which would address certain outstanding LPP issues. By this time, the respondents had served a reduced list of just over 200 search terms.
92. All the documents and files which had been 'hit' by the prosecution search terms and which had been viewed by the investigators - some 166,000 items - were provided to the defence by 16 December 2011 on a hard drive. This has been referred to as the 'classic unused' material or the 'tick list' because - subject to the Iron Mountain issue described below - each item viewed by the investigators was electronically ticked when it was viewed.
93. There was a difficulty in relation to 312,500 files containing graphics because, as we have indicated, FTK.v1.7 did not permit OCR of files of this type. This had the consequence, at least potentially, that files in this category that may be relevant could not be identified by the search terms that were used. The solution to this problem adopted by the prosecution was to review the graphics files 'in bulk'. This involved a number of stages. First, they were reviewed by scrolling through screens of thumbnail images, 30 to a screen, and those of interest were opened and 'ticked'. Second, the unopened items over a certain size (312,500 in total) were sent in January 2012 to a company called Iron Mountain for OCR and thereafter the same 232 defence search terms were applied. Any disclosure arising out of this exercise was complete by 21 September 2011. OCR did not work for some 21,000 items and they were individually reviewed, resulting in disclosure of 189 items on 21 September 2011.
94. Pausing in the narrative, and putting the matter shortly, it is our view that, at the latest by 21 September 2011, the prosecution had sufficiently discharged its primary disclosure obligations. Not only was it wholly unfeasible for the Crown to read more than a minimal part of the overall material located on the electronic devices, for the purposes of initial disclosure a review by way of proportionate dip sampling of the databases met the statutory disclosure requirement: indeed, this reflected the approach taken by the Crown prior to the hearing on 7 June 2011. As set out above, Hughes LJ in R v Brendan Pearson and Paul Cadman put beyond doubt that it was not the duty of the Crown 'to trawl through every word or byte' of the seized material in order to comply with its disclosure obligation and that when there is an enormous volume of material the Crown can 'perfectly properly' search by sample or by key words. The prosecution, therefore, was not obliged to consider every document identified by the defence search terms (the validity of which, we note, had not been investigated by the court in this case), most particularly at this early stage of the case which was prior to service of the defence statement. The search terms produced a virtual mountain of documents, and it would only be after service of the defence statements that it would be possible for parties, and the court if called on to review the matter, to identify whether a more detailed or focussed approach to disclosure needed to take place.
95. In November 2011 the suggestion emerged (in part from the Legal Aid Agency) that instead of serving the datasets of each individual computer, a global dataset covering all the computers was to be provided. This presented the prosecution with a very difficult technical task, and given this involved over 70 computers and 13 other devices the Crown stressed the potential impracticability of this step. This emerged as the course to be followed by the prosecution.
96. The combination of the orders made at the 17 February 2012 and 16 March 2012 hearings meant that the judge had ordered the prosecution to provide the emails in a format (on a database or by way of a search system) that treated them as emails (paragraph 32 of the note of 17 February 2012). The prosecution indicated that it would be necessary to use software tools such as Nuix or Intella for this purpose (the latter was eventually selected). In essence, the judge directed that the emails from all the computers should be provided en masse (from which any LPP material had been removed), together with the software to search them. Ramsey J observed (at paragraph 42 (ii) of the note of 16 March 2012):

"There has to be a discussion about e-mail production, and what is needed for there to be a sensible search by the defence of e-mails. E-mail threads are important and so is the e-mail header information. Equally, because they will form a good amount of the documentation, there needs to be effective de-duplication. Consider whether there is some approach that could produce filtered databases limited to the e-mail files earlier so that there could be a combined PST (or other e-mail database file), which could be searched with easily available commercially available software at reasonable cost."

97. The migration of data from FTK v1.7 to FTK v3.4 was taking place, albeit by January 2012 it was clear that this was not a straightforward exercise and would be time consuming. Consideration was given to changing instead to FTK v1.8, but ultimately it was decided to pursue the option of making the transition to FTK 3.4 for some of the drives.
98. The Intella dataset was disclosed on 13 July 2012. This dataset contained approximately 5.5 million files that had been identified by the 232 defence search terms. We note that the prosecution also supplied the hardware and software required to navigate the dataset. In due course, because of concerns that emerged over possible breaches of legal professional privilege, access to the Intella data set was restricted, in the sense that only those individuals who had originally had access to a particular computer were provided with the items emanating from that computer. Material, access to which was restricted in this way, has been referred to as the 'PACE material': it was provided to the individual accused under section 21(4) Police and Criminal Evidence Act 1984.
99. Meanwhile, a hearing date for the trial of 10 September 2012 was increasingly unlikely. Both the respondents and Ramsey J raised concerns that the e-disclosure had become protracted, partly due to lack of sufficient skilled resources, partly due to problems with the operation of the FTK software. On 20 April 2012, the judge expressed dissatisfaction with the progress made by the prosecution and the level of resources which it was dedicating to complete the task. 
100. As a result of the difficulties encountered, the prosecution was forced to apply to adjourn the trial date of 10 September 2012. This application was heard by Ramsey J on 21 June 2012. Following the direction that he had set, the judge found that the prosecution was responsible for 11 months' delay from the seizure of documents in July 2007 until November 2009, with a further delay up until April 2011 without much progress being made. He took the view that, by June 2012, the prosecution had still not provided disclosure. Having said that, he also concluded that there could still be a fair trial, the judge granted an adjournment and the trial was re-fixed for 16 September 2013.
101. On 13 July 2012, the prosecution made available to the respondents a dataset containing 276Gb, or 5.5m files, of compressed data hit by the defence search terms. However in September 2012, the defence expert instructed by the respondents identified various problems with the dataset provided; namely the presence of LPP material and missing or corrupted files. At a hearing on 19 November 2012, directions were given for the exchange of expert reports to address the issues identified. DFG advised in December 2012 that the removal of LPP duplicates meant that e-disclosure would take at least a further nine months.
102. On 21 December 2012, the respondents invited the court the reopen the adjournment application heard in June 2012 and a further hearing was fixed to consider disclosure issues. On 8 February 2013, Ramsey J vacated the trial date of 6 September 2013 and directed that e-disclosure be completed by 28 March 2013. After that hearing, HMRC instructed Deloitte to assist in making initial disclosure. On 28 March, the prosecution informed the court that it could not comply with the deadline of that day, citing the technical difficulties in relation to 'embedded LPP files' such that none of the product produced for disclosure thus far could be guaranteed as free of privileged material.
103. Until shortly before 28 March 2013 it had been hoped that bespoke software would enable the prosecution to remedy the LLP issues. However, the software failed to work. The judge identified various deficiencies (paragraphs 89 - 94 judgment 13 May 2015) with the dataset that DFG had produced using FTK software (loaded onto the Relativity system to which we refer below). The most significant of these was that there was no tool to search for and place bookmarks on 'LPP Duplicates' (viz. copies of LPP documents which had been found on drives other than those which a particular defendant had been permitted to search). In addition, a large number of files on the original hard drive were not present ('the COM error') and emails and their attachments were not appropriately linked (the 'parent and child' deficiency). There were also difficulties using FTK with whole word searches and the failure to display some email addresses in full. The judge expressed the view that this meant that the prosecution had failed to provide disclosure in an acceptable form.
104. It was because of the difficulties that the platform called Relativity had been identified by the prosecution as the means by which the outstanding LLP issues could be resolved. Relativity is a web-hosted and platform-managed database or 'relational information system' provided by Deloitte. It has very considerable computational power. Given the disclosure orders made by the judge, it had become necessary to transfer the dataset to a web-based platform of this kind. This contained all of the files identified by the defence search terms, and they had been reviewed for material covered by legal professional privilege. This has been referred to as the 'CPIA' dataset (and it contains 3,288,563 files).
105. The respondents made further application to stay the indictment as an abuse of process: this was heard between 1 and 5 July 2013; Ramsey J handed down judgment on 15 October 2013. It was argued that an abuse arose from the prosecution's disclosure of documents which attract LPP and also on the basis of the delay by the prosecution in giving electronic disclosure.  The first limb of the application was rejected and, furthermore, because the predicted delay in the trial was from September 2013 to 2015, the point had not been reached where the prosecution should be stayed because of delay; there was no proper risk of there not being a fair trial.
106. In respect of the state of electronic disclosure at the time, Ramsey J was satisfied that, with the necessary training and experience, the respondents would be able to access and work on the dataset loaded onto the Relativity platform which, he noted, was an industry standard product. However he also noted that the Relativity system was only as good as the dataset which had been loaded onto it. He identified four areas in which the underlying data provided through FTK needed to be identified and removed: (i) all duplicates and near duplicates of LPP material (wherever and on which ever computer they were found); (ii) attachments to e-mails bookmarked as privileged; (iii) items attracting LPP which had not been identified because of the 'whole word search' issue; and (iv) items attracting LPP which had not been identified because of the 'friendly name'. Additionally, some files from particular computers were missing and were to be added to the dataset.
107. At a case management hearing on 22 November 2013, the prosecution outlined a three-phase process for removing LPP material thereby remedying the deficiencies identified by Ramsey J. Independent counsel was to be instructed to review the whole dataset for LPP. Additionally, independent counsel, together with the individual accused, were to review their drives using their original LPP search terms, adapted for Relativity. Thereafter, Deloitte would use the Relativity Analytics toolset to find documents that were similar to the LPP documents. It was proposed that initial disclosure by the prosecution would be completed by 5 May 2014, allowing for a trial date of 6 April 2015.  
108. May 2014 passed with the analysis proving more complicated and time consuming than expected. At a hearing on 27 June 2014, Ramsey J agreed to the prosecution's revised timetable and ordered that initial disclosure be given by 8 September 2014, leading to a trial date of 11 January 2016, but warned that the consequences of not complying with the 8 September 2014 date would be serious. At the hearing, concerns were raised in respect of the proposed method of e-disclosure; in particular whether the prosecution would be providing an unfiltered set of emails, or whether the search terms would be applied to the emails to reduce the number. The respondents were in favour of unfiltered emails because they understood Relativity to be incapable of "re-threading" emails, that is to say, joining up connected e mails; however the prosecution confirmed that Relativity had techniques available to re-thread.
109. Delays occurred for a variety of reasons, but, on 8 September 2014, access was provided to the first of the Relativity data sets. This comprised a CPIA dataset for all the accused (the 3,288,563 files) and a separate PACE dataset for each individual defendant relating to their own devices, which included their LPP material. The CPIA data set, therefore, was available to all the defendants whereas the PACE material was restricted to individual defendants. There had been particular difficulties with 58,057 documents, because they were in a foreign language, they were unreadable or they consisted of duplicates or family member documents.
110. On 10 September 2014 the prosecution wrote to the defendants indicating that initial disclosure was complete. Shortly thereafter, (starting on 21 September 2014) a very large proportion of the emails contained in the image of each individual defendant's data set were 'threaded'. Furthermore, we were given a clear indication from Mr Miskin during the hearing of the appeal that the LPP issues have been resolved, in that the prosecution is able to provide a complete set of the documents with all the material covered by LPP removed (day 1 pp. 125 - 127).
111. There has been considerable criticism of the Relativity dataset, particularly by Bernhard Sebesta of LDM Global who was jointly instructed by the respondents. The judge summarised the criticism as being that the CPIA dataset was 'wholly unusable' - that it lacked 'forensic integrity' because the functionality of the search process was compromised, which resulted in wholly inaccurate and perverse results.
112. The expert evidence the judge heard focussed on four main topics: i) 'parent/child determination', ii) the inadequacy of the metadata and missing metadata; iii) email threading and iv) duplicate documents. The 'parent/child determination' concerned the separation of (child) attachments from the (parent) emails to which they had been attached when the data source was FTK (they were not exported as MSG files). On occasion emails and attachments have been incorrectly paired. As to the inadequacy of the metadata and missing metadata, it was infeasible to 'reunite' satisfactorily all the separated emails and attachments. There were difficulties, for which at least a partial remedy had been found, as regards missing information, or wrong additional information, in the "Email from" field. There were some examples of erroneous or corrupt metadata fields that made it impossible to carry out effective searches for documents, albeit the evidence from Deloitte tends to indicate this problem is limited in scope. Mr Sebasta identified some missing, erroneous and inconsistent date fields, but when the 'Document Date Field' was used this problem to a significant extent disappeared, and it only related to a very small quantity of documents. The email threading was only successful for between 60% to 80% of emails. In some instances exact duplicate items were stored in modified formats, thereby hiding the duplicate nature of the documentation; this was a result of the manner in which the items were exported out of FTK.
113. Mr Sebasta suggested that, viewed overall, the product of the search process was inherently unreliable and incomplete. Deloitte responded to the concerns in October 2014. At a hearing on 30 October 2014, the respondents developed their criticism of the CPIA dataset and Ramsey J was called upon to decide whether the criticisms were justified and, if so, what effect that had on the prosecution's compliance with the order to provide e-disclosure by 8 September 2014. Directions were given for the respondents to provide their remaining concerns about the dataset and for the prosecution to serve a report answering those concerns. A hearing was set for 16 December 2014 to resolve the issue. The timetable was both tight and critical: Ramsey J was due to retire from the bench thereafter. In the event, there was no cross examination of the experts notwithstanding requests to do so; further expert reports had to be prepared subsequent to the hearing which were then submitted in writing.
114. On 1 May 2015, Sir Vivian Ramsey (as the judge had become) handed down a lengthy and comprehensive judgment. He concluded, having heard from experts for the prosecution and the defence, that the dataset provided on 8 September 2014 had defects in it which were substantially those relied on by the respondents: see, in particular, paras. 132 to 252. His overarching conclusion in this context was as follows:

"286 […] The dataset is […] not 'fit for purpose' in the respects in which Mr. Sebesta explained his view. That is, the functionality of the search process is compromised resulting in wholly inaccurate and in many instances perverse results and also that the product of the search process is inherently unreliable and incomplete. Whilst I would not go as far as to say that the dataset is 'wholly' unusable, it does not represent a dataset which, in my judgement, would allow proper and efficient searching and analysis of the dataset to take place. Equally of importance is Mr Sebesta's conclusion that the dataset lacks forensic integrity. The difficulties explained above, in particular with missing or corrupt metadata, mean that the export of data from FTK to Relativity deprived the dataset of that essential forensic integrity. "

115. He considered that in a fraud case such as this, a properly usable and searchable dataset was required which depended upon the forensic integrity of the underlying data. That integrity had been lost during the original exercise using FTK software and the export of that data to the Relativity platform which meant that the functionality of the search process was compromised resulting in inaccurate and inherently unreliable results. Ramsey J concluded that the prosecution had failed on 8 September 2014, and have continued to fail, to make available a dataset which can properly be regarded as being CPIA compliant. 
116. There can be no doubt that the judge's objectives, reflected in the orders that he made, were entirely laudable and accorded with the principles underpinning good case management, to say nothing of common sense. He sought to ensure that as early as possible the defence was in possession of what he described as a 'properly compiled dataset'. As he explained in the course of the judgment of 13 May 2015, he went on to say at [287] that he:

"accepted the submissions made on behalf of the Defendants that they should not commence work on the dataset before it is a properly CPIA compliant dataset. To do otherwise would mean that time and cost would be wasted and processes would have to be repeated at substantial cost, in this case to public funds. In this context I note the position of the LAA that they were not willing to authorise work on searches or a review of materials if that work might then have to be repeated at a later date."

117. The underlying reasoning summarised by the judge was as follows (at [283]):

"In a fraud case such as this, evidence of the involvement of each defendant will be important. In that context each defendant will want to see what else was happening at a particular time in the form of communications between others. Equally […] in order to establish that the business was genuine and not fraudulent they will want to reconstruct the business by gathering together relevant documents from various sources. These aspects require a properly usable and searchable dataset which has forensic integrity in the documents contained in it. Metadata is also of great importance in establishing provenance and involvement in documents. "

118. It was clearly desirable that the defence should be able to access with ease the documents which may undermine the prosecution case or assist the defence case. Furthermore, it is undisputed on this appeal that a global, fully searchable dataset of the material seized in July 2007 had become feasible by 2014. For the future, therefore, the problems that have beset this case ought not to be repeated. The prosecution will need a 'properly compiled dataset' for its own purposes, in order to conduct its own investigations. This will then be available to ensure that the disclosure exercise is conducted properly, no doubt based - in the main - on approved search terms.
119. Having said that, however, a global, fully searchable dataset that encompassed the 85 digital devices and which threaded the email exchanges was unavailable in 2007 and instead the technology has improved during the life of this case, up until the judge's decision on 13 May 2015. The manner in which the materials were exported out of FTK into Relativity deprived the dataset of the forensic integrity that parties today when working on major cases of this kind are likely to expect. It is to be regretted that the prosecution experienced the series of difficulties summarised above in relation to the seized digital material, and the ability to search and handle it with confidence in the results. The errors as regards LPP are particularly of concern.
120. The question, however, is not whether the materials were made available in a manner which failed to provide an optimal search or review facility, but instead whether the prosecution had 'failed to give CPIA compliant disclosure by 8 September 2014' which was the principal reason that the judge ordered the stay of proceedings (paragraph 318 of the judgment of 13 May 2015). The judge had disapproved the prosecution's proposal that was based on dip sampling the materials in their 'raw or native format' (to adopt the prosecution's terminology), and the orders he made in this regard are conveniently summarised in his judgment of 19 November 2012 (at [103]):

"At the hearing on 7 June 2011 objections were raised to the method of disclosure of electronic documents by using search terms and then dip sampling on a list of files which had been identified by the use of the relevant search terms which appeared to be what was proposed. After a general discussion on the way in which the Prosecution was to give disclosure of electronic documents, the Prosecution was ordered to respond to a proposal which involved the Defendants proposing additional search terms for particular electronic material and then being provided with the results of those searches."

121. For the reasons identified in our analysis of the requirements of primary disclosure, we find ourselves wholly unable to agree with the approach which was adopted in this case. With the best of intentions, the learned Judge was, effectively, prepared to grant the respondents the keys to the warehouse and was diverted from a clear analysis of what could truly undermine the prosecution case or assist the defence. What had been the original and sustainable course the prosecution adopted to initial disclosure (based on the use of search terms and sampling) – an entirely proper approach, as already underlined – became far more wide ranging and went significantly beyond the statutory requirements, as interpreted by the case law and explained in the various Disclosure Reviews, Guidelines and Protocols. As we have indicated, we have no doubt that, at the latest by 21 September 2011, the prosecution had sufficiently discharged its primary disclosure obligations. In those circumstances, it is unnecessary for us to analyse or resolve the issues between the appellants and the respondents about Mr Sebasta's evidence and conclusions or, indeed, the extent to which the judge should have allowed cross examination to test the issues between the parties.
122. In its anxiety to follow the judge's lead and its belief (however well or ill founded that was) that it could, in fact and within the dates set, achieve that which the judge directed, the Crown did not do all that it could or should have done to bring the case back on course and it is not surprising that the respondents were keen to maintain the pressure on the prosecution doubtless in the belief that it could only enure to their benefit.

Abuse of process

123. A court has the power to stay criminal proceedings for an abuse of process in two categories of case, namely (i) where it will be impossible to give the accused a fair trial and (ii) where it offends the court's sense of justice and propriety to be asked to try the accused in the particular circumstances of the case, in short, where it would not be fair to try the defendant: see, for example, R v Horseferry Road Magistrates Court, ex p Bennett [1994] 1 AC 42, HL and, in particular, per Lord Lowry who observed (at page 74F) :

"… [P]rima facie it is the duty of a court to try a person who is charged before it with an offence which the court has the power to try and therefore the jurisdiction to stay must be exercised carefully and sparingly and only for compelling reasons. The discretion to stay is not a disciplinary jurisdiction and ought not to be exercised in order to express the court's disapproval of official conduct… 'pour encourager les autres'".

124. More recently, the test has been elucidated in R v Maxwell [2011] 1 WLR 1837 by Lord Dyson SJC in these terms (at paragraph 13):

"It is well established that the court has the power to stay proceedings in two categories of case, namely (i) where it will be impossible to give the accused a fair trial, and (ii) where it offends the court's sense of justice and propriety to be asked to try the accused in the particular circumstances of the case. In the first category of case, if the court concludes that an accused cannot receive a fair trial, it will stay the proceedings without more. No question of the balancing of competing interests arises. In the second category of case, the court is concerned to protect the integrity of the criminal justice system. Here a stay will be granted where the court concludes that in all the circumstances a trial will offend the court's sense of justice and propriety (per Lord Lowry in R v Horseferry Road Magistrates' Court, Ex p Bennett [1994] 1 AC 42, 74g ) or will undermine public confidence in the criminal justice system and bring it into disrepute (per Lord Steyn in R v Latif [1996] 1 WLR 104 , 112 f)."

125. In reaching his conclusions, Sir Vivian decided that the trial date of January 2016 could no longer be maintained and, in circumstances where (as he found albeit contrary to our conclusion) a fundamental reprocessing of the documentation was required before a CPIA compliant database could be made available to the respondents, he considered whether it was now appropriate to stay the proceedings on the basis of delay.  He concluded that the point had been reached, on this third occasion where he had been asked to consider the application, where the delay had deprived the case of the fairness which the respondents and the public were entitled to expect.  He considered the only "appropriate sanction" was to stay the prosecution.
126. This conclusion was on the basis that the defendants would not get a fair trial due to the delay which had resulted from the prosecution failure to provide CPIA compliant primary disclosure. On the face of it, therefore, the decision fell squarely within first limb abuse of process (impossible now to have a fair trial). However, Sir Vivian also referred to public interest considerations and at times appeared to state his concern about the integrity of the criminal justice system. As outlined above, he referred to the stay as a "sanction" and repeatedly referred to the prosecutorial failings. Such considerations are not relevant to a consideration of first limb abuse which should only be concerned with whether it is possible for the defendant to have a fair trial. They only come to the fore during the balancing exercise required in a consideration of where there is second limb abuse; namely whether it is fair to try the defendant.
127. Having set out the relevant authorities on delay and abuse of process in all three of his judgments, the judge also dealt with the risk to a fair trial as a consequence of delay. In October 2013, he said "the delay in this case can be explained but is clearly unjustified" ([397]) and "this is a case where there is clearly fault on the part of the prosecution" (398]), concluding (at [399]) that there had been prejudice to the respondents in not having their case tried and determined in September 2012 or September 2013". However, he noted (at [400]) that:

"this was a case which was likely to depend to a large extent on evidence from documents rather than from recollections and that, to the extent that recollections were necessary, contemporary documentation would allow recollections to be refreshed".

128. The authorities also make it clear that where delay is said to be the basis for a stay, serious prejudice must be shown: unjustified delay by itself is not a sufficient reason. In R v S (P) [2006] 2 Cr App R 23 (at [21]):

"In the light of the authorities, the correct approach for a judge to whom an application for a stay for abuse of process on the ground of delay is made, is to bear in mind the following principles:

(i) Even where delay is unjustifiable, a permanent stay should be the exception rather than the rule;

(ii) where there is no fault on the part of the complainant or the prosecution, it will be very rare for a stay to be granted;

(iii) no stay should be granted in the absence of serious prejudice to the defence so that no fair trial can be held;

(iv) when assessing possible serious prejudice, the judge should bear in mind his or her power to regulate the admissibility of evidence and that the trial process itself should ensure that all relevant factual issues arising from delay will be placed before the jury for their consideration in accordance with appropriate direction from the judge;

(v) if, having considered all these factors, a judge's assessment is that a fair trial will be possible, a stay should not be granted."

129. The court went further in R v F (S) [2011] EWCA Crim 1844 expressing the position in these terms:

"40 The explanations for delay are relevant to an application to stay only if they bear on how readily the fact of prejudice may be shown. Unjustified delay in the making of the complaint, and even more so institutional prosecutor misconduct leading to delay (which is what the court was considering in Attorney General's Reference (No 1 of 1990) ) may make the judge more certain of prejudice, which may even have been the aim of the delay. That is the import of the references in the cases to the reasons for the delay. That is, however, a long way from the proposition that unjustified delay is by itself a sufficient reason for a stay. It is not.

…

47 When abuse of process submissions on the grounds of delay are advanced, provided the principles articulated in R v Galbraith [1981] 1 WLR 1039 and Attorney General's Reference (No 1 of 1990) [1992] QB 630 are clearly understood, it will no longer be necessary or appropriate for reference to be made to any of the decisions of this court except R v S (P) [2006] 2 Cr App R 341 and the present decision. These four authorities contain all the necessary discussion about the applicable principles. Their application, whether in the Crown Court or in this court, is fact-specific, and is to be regarded, unless this court in any subsequent judgment expressly indicates the contrary, as a fact-specific decision rather than an elaboration of or amendment to the governing principles. In this court, but not the Crown Court, the separate question of the safety of the conviction, if there is one, may also arise for decision. Again, however, the issues which may arise are illustrated by R v B [2003] 2 Cr App R 197 and R v Smolinksi [2004] 2 Cr App R 661. No further citation of authority is needed."

130. Having stated that there had been prejudice to the respondents, Ramsey J went on to note that "many of the Defendants have now been involved for some eight years since the searches and arrests in 2007 and have been charged since 2010. They are all on bail and some are subject to stringent bail conditions and restraint orders and the further period of delay in giving disclosure and delaying the trial to an uncertain date is additional prejudice". He was thus rightly concerned, as we are, about the effect that the delay has had on the respondents personally; the significant inconvenience and distress it will have caused to have these proceedings hanging over them. However, the judge appears to have placed greater weight on this personal prejudice rather than considering whether there is serious prejudice in the sense that they will be deprived of a fair trial.
131. As the judge accepted, the case turned in large part on documentary evidence and to the extent that recollections were necessary, documents would allow memories to be refreshed. While he was concerned 10 years on from the date that the alleged conduct occurred, memories would have faded, this is arguably true of many prosecutions. Furthermore, in this case, the respondents have had the prosecution Case Summary and statements of the prosecution witnesses for many years and have thus have had and will have ample opportunity (even if they needed it) to re-acquaint themselves with the detail of what is an almost exclusively paper based prosecution case.
132. Annette Henry Q.C. (for Mr Whiston-Dew) pointed to the absence of contemporaneous or computer based material available to him: that feature of the case will doubtless be deployed to demonstrate the very limited extent of the evidence of his knowledge of the circumstances and it will be for the Crown to prove to the contrary. Although he may not be able to deploy the arguments of those more intimately involved in the day to day business of the companies, he will be able to adopt whatever arguments they advance: that is no different from the position that arises in many cases where the role of those on the periphery (without deciding whether or not this respondent is in that position), depends in part on what is proved against the principals.
133. Suffice to say, in our judgment, the delay, of itself, was not sufficient to warrant a conclusion that the respondents could not now receive a fair trial and Sir Vivian was wrong so to find. They clearly then could and we have no doubt that they still can. In the event of an adverse verdict, to such extent as the trial judge takes the view that the delay amounts to a breach of the reasonable time requirements of Articles 5(3) or 6(1) of the ECHR, there is clear authority for the proposition that such delay may be cured by a reduction in sentence. Thus, in Spiers v Ruddy [2008] 1 AC 873, having reviewed the authorities (including those emanating from Strasbourg), Lord Bingham observed (at para 16B):

"The authorities relied on and considered above make clear, in my opinion, that such delay does not give rise to a continuing breach which cannot be cured save by a discontinuation of proceedings. It gives rise to a breach which can be cured, even where it cannot be prevented, by expedition, reduction of sentence or compensation, provided always that the breach, where it occurs, is publicly acknowledged and addressed."

134. Having concluded there was no basis on which to stay the prosecution under first limb abuse of process, would it nevertheless be unfair to try the respondents now? As noted above, Sir Vivian focussed on the prosecutorial failings in this case. That brings into play the balancing exercise identified in R v Latif and Shahzad [1996] 1 WLR 104, by Lord Steyn (at page 113A-B):

"[I]n a case such as the present the judge must weigh in the balance the public interest in ensuring that those charged with the gravest crimes should be tried and the competing public interest in not conveying the impression that the court should adopt the approach that the end justifies any means".

135. The problem arises because maintaining confidence in the criminal justice system (or, as it has been put, avoiding "an affront to the public conscience") is an aim or aspiration which has to be perceived from different directions. On the one hand, there is gross misconduct which the criminal justice system cannot approbate (as in cases such as Bennett and R v Mullen [2000] QB 520). On the other hand, however, it is important that conduct or results that may merely be the result of state incompetence or negligence should not necessarily justify the abandonment of a trial of serious allegations. As has been observed, there is no bright line and a broad brush approach is likely to be necessary.
136. In this case, it is beyond argument that there has been no deliberate misconduct or bad faith on the part of the prosecution. Every effort has been made to comply with the disclosure strategy to which it had, rightly or wrongly, agreed even if some of the steps taken have been insufficiently thought through or have proved to be ineffective. There was no deliberate disregard for a clear direction of the court, as there was in R v Boardman [2015] EWCA Crim 175. In any event, that case did not involve an allegation of abuse of process but concerned a refusal to adjourn a fixed trial with consequential orders relating to the admissibility of evidence made pursuant to s. 78 of the Police and Criminal Evidence Act 1984.
137. In submissions made on behalf of one of the respondents, Rodney Whiston-Dew, Ms Henry QC clearly recognised the difficulty that this analysis presented for the respondents. As a result, she went on to submit that the second limb abuse of process could, and should, be extended to include egregious breaches of case management orders in cases such as this. One of the obvious problems with that submission, which we raised at the hearing, was how such a system would work the other way around. There is no ultimate sanction for a defendant who fails to comply with case management orders: unlike in a civil case (in which the claim or the defence can be struck out for non-compliance), a defendant cannot lose the right to defend himself.
138. The availability of other sanctions with teeth was a source of concern for the court in R v S(D) and S(T) [2015] EWCA Crim 662 (see [71]) and exercised the attention of the recent Review of Efficiency in Criminal Proceedings. To allow successful abuse of process applications where neither prosecutorial misconduct of the type identified in the authorities nor delay such as would prejudice a fair trial can be established would, however, provide a perverse incentive for those charged with criminal offences to procrastinate and seek to undermine the prosecution by creating hurdles to overcome all in the hope that, at some stage, a particular hurdle will cause it to fail. We emphasise that we are not suggesting that the respondents to this appeal deliberately set about to undermine the prosecution and, indeed, Sir Vivian was emphatic that they had done all they could to assist the speedier resolution of the claim albeit that they did so in the context of what we consider to have been a flawed submission in relation to primary disclosure which the Judge then adopted. Suffice to say that there will be cases (such as Boardman) where prosecutorial failures can bring a prosecution summarily to an end but these can only be decided on a case by case basis and it is difficult to generalise as to the circumstances in which they arise. The search for an effective sanction will continue but improvements are likely to be based in the adoption of other aspects of the Review of Efficiency (not least the requirement to "get things right first time").
139. In conclusion, bearing in mind our conclusion that primary disclosure had been completed as long ago as September 2011, the jurisdiction to stay the trial simply does not arise either on the grounds of delay such that a fair trial is no longer possible or because a trial would offend the court's sense of justice and propriety or undermine public confidence in the criminal justice system. Even if we had reached a different view in relation to the disclosure exercise, in a case of this nature, it is by no means clear that Sir Vivian's conclusion in relation to abuse of process was justified: in reality, there is no basis in law for extending the abuse jurisdiction and, in the circumstances of this case, it is difficult to see that the constituent elements of either of the limbs of the abuse jurisdiction are established.
140. There are a number of arguments that we have not felt it necessary to deal with either because it is unnecessary to do so in the light of our general conclusions concerning primary disclosure (e.g. as to the detailed findings regarding the IT problems of addressing listing, disclosure or presentation) or because they fall away (e.g. the separate arguments about the personal tax charges). In relation to the former, over-focus on the detail to the extent of the skeleton arguments of both sides (necessary although it was, given the approach of the judge) will not assist the ultimate resolution of this prosecution.

Conclusion

141. The appeal having been brought pursuant to s. 58 of the 2003 Act, s. 61(1) provides that the court may confirm, reverse or vary any ruling to which it relates but (as prescribed by s. 67):

"The Court of Appeal may not reverse a ruling on an appeal under this Part unless it is satisfied –

(a) that the ruling was wrong in law.

(b) that the ruling involved an error of law or principle.

(c) that the ruling was a ruling that it was not reasonable for the judge to have made."

142. In our judgment, staying the case as an abuse of process was a ruling which was not reasonable for the judge to have made, having regard to his failure to appreciate that, on the principles of law which we have sought to expound, primary disclosure had long been addressed sufficiently to comply with s. 3 CPIA. In any event, there was an error of principle in that, given the nature of the case, it was wrong to conclude (if, in reality, he did so conclude) that the respondents could not receive a fair trial or (further or in the alternative) that the misconduct of the prosecution was such that it offends the court's sense of justice and propriety to be asked to try the respondents in the particular circumstances of the case.
143. In the circumstances, leave is granted pursuant to s. 57(4) of 2003 Act and this appeal is allowed: the stay is lifted and in respect of each offence which is the subject of the appeal the proceedings must be resumed in the Crown Court. Sir Vivian having retired, a Queen's Bench judge will be nominated to conduct the trial. He will be able to give consideration to the question whether or not it is necessary to order a preparatory hearing and must ensure that the trial can now be progressed, the prosecution papers having been served many years ago. Given the schedules prepared by or on behalf of the respondents for the purposes of the hearing before Ramsey J and the knowledge that each will have from their own records (or provable absence of knowledge from the fact that there were no relevant contemporaneous documents seized from them), we trust that all parties will co-operate to ensure progress with the service of defence case statements, secondary disclosure (if any) and directions for trial.
144. Although we have reached a different conclusion to Sir Vivian Ramsey to that which he formed about the nature and extent of what was required for primary disclosure, we cannot leave the case without expressing our deep admiration for his mastery of the detail and the comprehensive way that he dealt with many of the arguments advanced. It may be that an over focus on that detail has meant that he did not stand back to look at the overall allegation and the need to approach the case on the basis of the thrust of the case fully set out in the prosecution summary. That summary fully recognised that research and development had been undertaken, but argues that its nature and extent did not require analysis at the level of individual e mails but, rather, at contractual relationships and payments in the context of the overall financing of the schemes.
145. We deal finally with publication of this judgment. There are restrictions on reporting proceedings of this type contained within s. 71 of the 2003 Act and, save for those particulars permitted by statute to be reported, we underline those restrictions and order that they apply until the conclusion of the trial. The only exception to this order is that we permit (and, indeed, encourage) the publication of a redacted and anonymised extract dealing with the analysis of the law and practice both of disclosure and abuse of process. These parts of the judgment are of general application and may well be relevant and affect the approach to other prosecutions now proceeding through the courts.