Lee Castleton v Post Office Limited & Anor

Neutral Citation Number: [2026] EWCA Civ 577 Case No: CA-2026-000323 IN THE COURT OF APPEAL (CIVIL DIVISION) ON APPEAL FROM THE HIGH COURT OF JUSTICE BUSINESS AND PROPERTY COURTS OF ENGLAND AND WALES BUSINESS LIST (ChD) Mr Justice Trower and Master Kaye BL-2025-000341 Royal Courts of Justice Strand, London, WC2A 2LL Date: 12/05/2026 Before : LORD JUSTICE PETER JACKSON LORD JUSTICE POPPLEWELLand LORD JUSTICE ZACAROLI – – – – – – – – – – – – – – – – – – – – – Between : LEE CASTLETON Appellant – and – (1) POST OFFICE LIMITED (2) FUJITSU SERVICES LIMITED Respondents – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – Paul Marshall and Andrew Young (instructed by Simons Muirhead Burton) for the Appellant James Bailey KC and Daniel Petrides (instructed by Pinsent Masons LLP) for the First Respondent Stephen Midwinter KC, Frederick Wilmot-Smith and Jagoda Klimowicz (instructed by Morrison & Forster (UK) LLP) for the Second Respondent Hearing date: 30 April 2026 – – – – – – – – – – – – – – – – – – – – – Approved Judgment This judgment was handed down remotely at 10.00am on 12 May 2026 by circulation to the parties or their representatives by e-mail and by release to the National Archives. ……………………….. Lord Justice Zacaroli:

1. This is an appeal against one aspect of an order of Trower J and Master Kaye (the “Judges”), made at a directions hearing on 23 January 2026. Background

2. The claimant, Mr Castleton, was the sub-postmaster of the branch post office at Marine Drive, Bridlington under a contract with the first defendant, Post Office Limited (“POL”), between July 2003 and March 2004. He was suspended in March 2004 on the grounds that the branch accounts disclosed cash shortfall discrepancies. Those records were prepared with the use of the Horizon computerised accounting system, designed by the second defendant, Fujitsu Services Limited (“Fujitsu”) and supplied by it to POL under a facilities management contract. POL brought a civil claim against Mr Castleton (the “OriginalAction”) which culminated in a judgment dated 22 January 2007 for a sum in excess of £25,000 (the “2007 Judgment”). Together with costs, this amounted to a debt of £309,807. Mr Castleton was subsequently made bankrupt.

3. The defects in the Horizon system, the resulting flaws in the civil proceedings pursued against Mr Castleton, and in the civil and criminal proceedings pursued against very many other sub-postmasters, and the devastating consequences for them and their families, are widely known.

4. Mr Castleton was one of a number of people (the “GLO Claimants”) who brought proceedings under a Group Litigation Order against POL, known as Bates & Ors v Post Office Limited (the “GLO Action”). The two principal judgments were handed down in the GLO Action by Fraser J, the first (following a trial of common issues) in March 2019 (Bates and ors v Post Office Ltd (No.3 Common Issues) [2019] EWHC 606 (QB)) and the second (following a trial relating to issues concerning the Horizon system) on 16 December 2019 (Bates and ors v Post Office Ltd (No.6 Horizon Issues Rev.1) [2019] EWHC 3408 (QB)).

5. Between the circulation in draft of the Horizon Issues judgment, and the formal hand down of that judgment, the GLO Action was compromised by a settlement agreement dated 10 December 2019 (the “Settlement Deed”). The Settlement Deed contained a widely drafted ‘general release’ of claims in the GLO Action and of other potential claims against POL.

6. On 14 March 2025, Mr Castleton commenced the present action against POL and Fujitsu (the “2025 Action”). The principal causes of action are unlawful means conspiracies (i) to abuse the process of the court for a collateral purpose and (ii) to deceive the court and Mr Castleton by dishonestly withholding evidence at Mr Castleton’s trial in 2006 and that the judgment was obtained in 2007 by fraud and (iii) to deceive the court and Mr Castleton by withholding evidence at the trial of the GLO Action. The relief sought is a declaration that the 2007 Judgment was obtained by fraud, an order setting aside the 2007 Judgment as having been obtained by fraud, damages exceeding £2 million and an order against POL that the bankruptcy order made against Mr Castleton on 23 May 2007 be cancelled and annulled for having been obtained by fraud.

7. POL and Fujitsu, in addition to challenging the substance of the underlying claims, rely on the Settlement Deed as an answer to the 2025 Action.

8. The particulars of claim in the 2025 Action fall into three main parts. Part B sets out the details of Mr Castleton’s claim that POL’s pursuit of the Original Action was an abuse of the process of the court for an improper collateral purpose and involved a conspiracy to injure him by unlawful means. Part C sets out Mr Castleton’s case that the 2007 Judgment was obtained by fraud and by an unlawful means conspiracy between POL and Fujitsu. I will refer to the claims pleaded in Parts B and C of the pleading as the “Historic Claims”.

9. Part A sets out the bases on which Mr Castleton seeks to overcome or avoid the consequences of the Settlement Deed, by one of three routes: (1) As a matter of construction, the Settlement Deed does not apply to the Historic Claims (the “Construction Issue”). (2) It would be unconscionable for POL to rely on the general release under the Settlement Deed (the “Unconscionability Issue”). (3) In 2019, POL dishonestly misrepresented the reason why Gareth Jenkins, a person described by Fraser J at the trial of the GLO Action as central to the operation, efficiency and robustness of Horizon, was not called as a witness for POL at the trial of the GLO Action in 2019. It is said that the real reason Mr Jenkins was not called was that POL knew that he had misled the court by his evidence, and that POL concealed relevant information about Mr Jenkins. It is said that what the court was told about Mr Jenkins during the trial of the GLO Action was false, was known by POL to be false, and that it induced or influenced Mr Castleton to enter into the Settlement Deed (the “Fraud Issue”). The Order under appeal

10. At the hearing on 23 January 2026, the Judges gave consideration, among other matters, to whether it was appropriate to order a separate trial of the Part A claims. They concluded that it was. Their order dated 23 January 2026 (the “Order”) included the following matters: (1) An order for a separate trial to determine preliminary issues, being those set out in Annex 1 to the Order, as follows: “1. Whether, on its true construction, the Settlement Deed released the claims against the First Defendant pleaded in Part B and Part C of the Particulars of Claim.

2. Whether, if on its true construction the Settlement Deed did release the claims against the First Defendant pleaded in Part B and Part C of the Particulars of Claim, it is also effective to release the Claimant’s pleaded claims against the Second Defendant.

3. Whether, if the Settlement Deed did release those claims against the First Defendant, the First Defendant is nonetheless precluded from relying on the effect of that settlement by reason of “unconscionability” (as alleged in paragraph 9 of the Particulars of Claim). 4.Whether: a. The First Defendant fraudulently misrepresented the reasons for not calling Gareth Jenkins as a witness in the Horizon Issues Trial; and b. If so, whether any such fraudulent misrepresentation induced the Group Litigation Claimants in claim numbers HQ16X01238, HQ17X02637 and HQ17X04248 to enter into the Settlement Deed; and

5. If so, to what relief, if any, is the Claimant entitled.” (2) The defendants were required to file defences to the Part A Claims (save for certain passages identified in Annex 2 to the Order, being those that were identified as relating to the Historic Claims, although set out in the Part A section of the pleading); and (3) However, the defendants were not required to file defences to Part B and Part C of the particulars of claim until after judgment was handed down following the trial of the preliminary issues.

11. In the footnote to the Order there is an important proviso to the Unconscionability Issue, requiring some, but only some, of its elements to be assumed, not decided. It is common ground that in order to avoid the Settlement Deed on the grounds of unconscionability, the following four elements must be established: (1) The Historic Claims are viable claims; (2) POL, on 10 December 2019, knew that the Historic Claims were viable claims; (3) Mr Castleton, on 10 December 2019, did not know and could not reasonably have known of the Historic Claims; and (4) POL, on 10 December 2019, knew that Mr Castleton was unaware of the Historic Claims.

12. It is also common ground that establishing unconscionability is not simply a ‘tick-box’ exercise, but that if these four elements are made out the court is required to address the overarching question whether in all the circumstances it would be unconscionable to rely on a settlement agreement in respect of a particular cause of action. I return to this point below.

13. The proviso in the footnote to Annex 1 to the Order is in the following terms: “For the avoidance of doubt, and without the First Defendant or Second Defendant having made any admission, the issue shall be determined on the provisional assumptions, for the purposes of the preliminary issues trial only, that (i) the claims pleaded in Part B and Part C of the Particulars of Claim are viable claims; and (ii) that the First Defendant knew this on 10th December 2019.”

14. Accordingly, the Order envisaged that the preliminary issue would determine the question of unconscionability on the basis that: (1) The court would need to determine, as at the date of the Settlement Deed, whether Mr Castleton was unaware of the Historic Claims and whether POL knew that Mr Castleton was unaware of them; but (2) It was to be “provisionally assumed” that the Historic Claims were viable claims and that as at the date of the Settlement Deed POL knew this.

15. The parties liaised over the form of order. There was no debate about the passages from Part A of the pleading to include in Annex 2 (save for a minor disagreement over three words, which is not relevant to the issues raised on this appeal). The scope of permission to appeal

16. Permission to appeal was sought on seven grounds. Permission was granted by Lewison LJ on 4 March 2026 on one ground only, being ground 6: “The Court wrongly relieved the Respondents of the mandatory requirement to comply with CPR 16.5(1) by directing that they were not required to plead defences to the five paragraphs/phrases of Part A listed in Annex 2 to the Order.”

17. In giving permission to appeal on ground 6, however, Lewison LJ said this: “There is a real prospect of successfully arguing that if the court directs the trial of preliminary issues it should either (a) provide for all the facts to be found or (b) direct a trial on the basis that the facts pleaded by the claimant are true. The hybrid approach of the court in this case may well be wrong.”

18. Mr Bailey KC, who appeared with Mr Petrides for POL, submitted that the wording of ground 6, particularly when read in the light of the matters on which permission was refused, meant that the scope of the appeal is limited to the narrow question whether the respondents should be relieved of the requirement to plead to the five identified passages set out in Part A of the pleading pending determination of the preliminary issue. In the grounds of appeal for which permission was refused Mr Castleton sought to challenge the judges’ decision to order a trial of preliminary issues, their decision as to which issues to be determined as preliminary issues, and their decision to relieve the defendants of the obligation to plead to the Part B and Part C Claims until after determination of the preliminary issues. The defendants further contend that an appeal on this ground is not open to Mr Castleton, because his lawyers specifically agreed the material parts of Annex 2 (though Mr Marshall submitted that these had been directed by the court, rather than agreed).

19. In my judgment, however, it is necessary to have due regard to the reason given by Lewison LJ in granting permission, which specifically identified the argument that could be pursued on appeal, namely that it was wrong to order that, in the circumstances of this case, the trial of preliminary issues should be conducted partly on the basis of assumed facts and partly on the basis of findings made by the court. There is no conflict between this interpretation and the fact that permission was refused on the remaining grounds, because none of the other grounds of appeal specifically addressed this hybrid approach to the trial of the Unconscionability Issue. Moreover where, as here, for the reasons developed below, the issue affects the efficient use of court resources as well as the private interests of the parties, the court has its own interest in it being addressed.

20. Accordingly, I address the substance of the appeal on the basis that permission has been granted on the wider basis implied by the reason Lewison LJ gave for granting permission. The assumption as to the existence of viable claims

21. Although it was suggested in the appeal skeleton argument filed by Mr Castleton’s counsel that the hybrid approach adopted by the Judges was wrong in principle and unprecedented, no objection was pursued, at the hearing, to the preliminary issues trial proceeding on the assumption that Part B and Part C of the pleading set out viable claims. Such an assumption is likely to be essential in any case where a preliminary issue is ordered as to whether a settlement agreement is a complete answer to the pursuit of claims purportedly settled by that agreement. Mr Castleton himself accepts that the issue of construction of the Settlement Deed could appropriately be determined on a preliminary basis, on the basis that the facts pleaded in the particulars of claim are true.

22. In any event, I would not accept that a hybrid approach is, as a matter of principle, unacceptable. The real question, and the focus of the submissions by Mr Marshall, who appeared with Mr Young, at the hearing, is whether it is appropriate to proceed on the assumption that POL knew there were viable claims, but to require a determination on the facts of the other elements of the claim in unconscionability. The assumption as to POL’s knowledge

23. There appears to have been little debate before the Judges as to the nature and consequences of the assumption as to POL’s knowledge. The objections to the hybrid approach taken by Mr Marshall on this appeal were not taken below.

24. That may have been because of the way in which the point arose. The first time that the suggestion was made that POL’s knowledge of the Historic Claims might be assumed, for the purposes of the preliminary issues, was when it appeared in a footnote to POL’s skeleton for the hearing below (and in the footnote to Annex 1 of the draft order which was appended to POL’s skeleton). As to the effect of the assumption, all that appears to have been said was that (per Mr Bailey), its purpose was to excise an issue of law as to the corporate knowledge of POL in 2019 of events which happened in 2007 where, for example, “no one in the building overlapped”.

25. In stating the Judges’ decision at the hearing (with reasons to follow), Trower J said that they had decided to order a split trial “which takes the form of the split that is suggested by the Post Office in their skeleton argument”. The terms of the footnote in the skeleton were carried through without change into the footnote to Annex 1 to the Order.

26. I recognise that it may be said that Mr Castleton ought not to be allowed to challenge the hybrid approach on appeal, given the lack of objection taken below on his behalf. In fact, much of the detail of Mr Marshall’s arguments on this point was set out in amendments made to his skeleton argument for this appeal. This was the subject of an opposed amendment application.

27. Whether or not parties identify particular problems with a proposed trial of preliminary issues, if it is likely to create difficulties for the trial of the action then that is a matter of concern to the court itself. No prejudice has been caused to the defendants by Mr Castleton being allowed to advance these arguments on appeal, or by the amendments to the skeleton. POL has taken the opportunity to respond fully and has filed its own short responsive skeleton.

28. The matter is complicated further in this case because – as I describe in more detail below – it became apparent for the first time at the hearing of the appeal that there is significant uncertainty over the meaning and effect of the “provisional assumption” set out in the footnote to Annex 1 to the Order. In all the circumstances, Mr Castleton should be permitted to advance the case as developed in his counsels’ amended skeleton argument on this appeal. For completeness, I would allow the application to amend the appellant’s skeleton, on the basis that POL is permitted to rely on its responsive skeleton.

29. On the closer analysis which has been undertaken on this appeal of the consequences of the hybrid approach, I consider that it is indeed problematic.

30. The difficulty lies principally in the fact that a determination of POL’s knowledge would have involved findings on various factual matters (as to what was known by particular individuals and when) as well as on the legal consequences as to whether that knowledge could be attributed to POL as at the date of the Settlement Deed. The provisional assumption required to be made on the Judges’ approach is, however, as Peter Jackson LJ put it in argument, a legal assumption unpopulated by any facts. That might not be a problem if other elements of the Unconscionability Issue did not – at least potentially – involve an investigation into the same kind of facts that would have populated the issue of POL’s knowledge. For the reasons developed below I consider that there is scope for significant overlap between the investigation into those other issues and the facts which would be likely – if it were to be determined – to populate the sub-issue as to POL’s knowledge of the Historic Claims.

31. To explain this, it is necessary to say something more of the nature of a claim to set aside a settlement agreement on the basis of unconscionability. This was canvassed in Bank of Credit and Commerce International SA v Ali [2001] UKHL 8; [2002] 1 AC 251, a case concerning the construction of a settlement agreement and general release signed by an employee of the bank “in full and final settlement of all or any claims … of whatsoever nature that exist or may exist”. Although not applicable on the facts of that case, their Lordships were satisfied that the law could provide a remedy in cases of “sharp practice” (which subsequent cases have equated with the concept of unconscionability). For present purposes, it is sufficient to refer to the following passage in the speech of Lord Nicholls, at §32: “Thus far I have been considering the case where both parties were unaware of a claim which subsequently came to light. Materially different is the case where the party to whom the release was given knew that the other party had or might have a claim and knew also that the other party was ignorant of this. In some circumstances seeking and taking a general release in such a case, without disclosing the existence of the claim or possible claim, could be unacceptable sharp practice. When this is so, the law would be defective if it did not provide a remedy.”

32. As is evident from that passage, establishing that the party seeking to rely on a settlement agreement knew of claims and knew that the other party did not know of them is not in itself enough. That may give rise to a remedy, but the overarching question is whether reliance on the settlement agreement would in all the circumstances be unconscionable.

33. The parties are agreed that this is a developing area of the law, that questions remain as to what is meant by “knowledge” in relation to each of the three elements to which it is relevant, and that it may be that it means different things as between some or all of the elements.

34. The first area of overlap between the issues is between the assumption that POL knew of the Historic Claims and the issue as to its knowledge that Mr Castleton was unaware of them. It may be, by way of example, that resolution of that issue will turn in part on how and when individuals within POL became aware of the matters giving rise to the Historic Claims, and the extent to which the matters of which they became aware were available beyond POL or, conversely, were suppressed within POL, and so would or would not have been likely to come to Mr Castleton’s attention.

35. More broadly, in considering the overarching question of unconscionability, the precise extent and nature of POL’s knowledge may well be important. Relevant to that inquiry might be matters such as the seniority of those within POL who knew of the possibility of Historic Claims, precisely what they knew of them, what they understood as to their viability, and to what extent information relating to them had been actively suppressed.

36. Relatedly, there is scope for uncertainty in the management of the preliminary issues trial going forward. For example, does the assumption as to POL’s knowledge of the Historic Claims preclude there being disclosure, witness evidence or cross-examination on matters going to that knowledge, even though that may be relevant in determining either POL’s knowledge of Mr Castleton’s lack of awareness of the claims or the overarching issue of unconscionability? Or does it preclude cross-examination of a particular witness on aspects of POL’s knowledge of the Historic Claims which may be relevant in undermining the credibility of their evidence as to their understanding of Mr Castleton’s awareness or otherwise of them? It may not be until the trial itself that – by reason of an answer by one of POL’s witnesses – it becomes apparent that it is necessary to explore in cross-examination further aspects of POL’s knowledge as to the Historic Claims. That is likely to create difficulties if relevant disclosure has not been given.

37. Mr Midwinter KC, who appeared with Ms Klimowicz for Fujitsu, made the point that questions as to the ambit of disclosure are for the Judges who are case managing the preliminary issues, and it is not for us to second guess what orders may be made. That, however, misses the point that in determining whether the hybrid approach is appropriate it is necessary to explore its practical consequences for the trial of the preliminary issues.

38. During the course of the hearing of the appeal, it became apparent that there was a serious misunderstanding, or at least difference of opinion, between the parties as to what is meant by a “provisional” assumption “for the purposes of the preliminary issues trial only”, as specified in the footnote to Annex 1 of the Order. Specifically, does it prevent POL, if it lost the preliminary issue trial, from having a second bite of the cherry via a further trial of the question whether it actually knew of the Historic Claims?

39. Mr Bailey made it clear at the hearing of the appeal that it was not intended to preclude POL doing so. That came as a surprise to Mr Marshall (who appeared both before us and below) and Mr Midwinter. Mr Midwinter described, in his skeleton argument, the assumption as having the same effect, in substance, as an admission. That is not how POL intended it to be understood.

40. This was not an issue that was addressed before the Judges. It seems most likely to me, however, that the Judges shared Mr Marshall’s and Mr Midwinter’s view of the effect of the proviso. That is the more obvious reading of paragraph 2 of the Order, read together with Annex 1, because it ordered a trial “to determine” the issues in Annex 1, including the Unconscionability Issue. On Mr Bailey’s interpretation the preliminary issues trial would not determine the Unconscionability Issue, but only aspects of it. Moreover, on POL’s understanding of the assumption, there would be a material risk of overlap in the evidence required for the two trials that may now be required, and the associated risk of inconsistent findings. The fact that these matters, which would often count against directing a trial of a preliminary issue, were not explored at the hearing below supports the view that this interpretation of the assumption to be made was POL’s alone.

41. As to the practical difficulties identified above, if the effect of the assumption as to POL’s knowledge of the Historic Claims is to preclude investigation of the facts going to POL’s knowledge for all purposes, it is difficult to see how there could be a fair determination of the other issues to which those facts are, or are potentially, relevant. If it was sought to avoid that unfairness by excising determination of those other issues from the preliminary issues trial, that would further undermine the utility of trying the Unconscionability Issue on a preliminary basis. It would almost certainly preclude the court determining the overarching question of unconscionability for the reasons set out above. That would be yet further removed from the objective contemplated by the Order – that the issue would be determined. Conclusion

42. For the above reasons, I consider that on the closer analysis that has been possible in this Court than was undertaken before the Judges, the proposal that the Unconscionability Issue be determined on the basis of the provisional assumption as to POL’s knowledge of the Historic Claims creates too many practical difficulties to be workable.

43. To that extent, therefore, if the other members of this Court agree, the Order cannot stand in its present form. It does not necessarily follow that the assumption as to POL’s knowledge should simply be removed from the footnote to Annex 1 to the Order. The extent to which, if POL’s knowledge of Historic Claims is not to be assumed, the preliminary issues trial in its (otherwise) present form remains appropriate, and whether that is a matter best determined by us or on remittal to the Judges, is something on which the parties should be invited to file short submissions.

44. That leaves the appellant’s arguments on the pleading point raised by ground 6 itself. The short answer to this is that, in light of the conclusion I have reached above, the identification of those parts of the pleading in Part A to which a defence can be postponed will need to be revisited. I would only make the further observation that, despite the terms of Annex 2 to the Order being largely agreed, certain of the passages referred to (for example paragraph 12 of the particulars of claim) would on any view appear to relate to matters that fall within the issues that are to be determined at the preliminary issues trial.

45. Finally, Mr Castleton requested that the matter be remitted, not to the Judges (to whom this case has been assigned) but to a differently constituted court. The fact that an appeal court disagrees with a case management decision of a first instance judge is no reason, in itself, to require the case to be remitted to a different judge. Judges are routinely required to reconsider a case following a successful appeal. In any event, given that the key objections were not taken below, the mere fact that I have reached a different view to that of the judges does not in this case imply any criticism of them. There is in my view no basis which requires this case to be remitted to different judges. Accordingly, I would decline that request. Lord Justice Popplewell

46. I agree. Lord Justice Peter Jackson

47. I also agree.