Test Claimants in the Franked Investment Income Group Litigation and others v Commissioners for Her Majesty’s Revenue and Customs

This appeal concerns the correctness of two of the most important decisions on the law of limitation of recent times: the decisions of the House of Lords in Kleinwort Benson Ltd v Lincoln City Council [1999] 2 AC 349 (“Kleinwort Benson”) and Deutsche Morgan Grenfell Group plc v Inland Revenue Comrs [2006] UKHL 49; [2007] 1 AC 558 (“Deutsche Morgan Grenfell”). It arises in the course of long-running proceedings known as the Franked Investment Income (“FII”) Group Litigation.

In view of the length of this judgment, it may be helpful at the outset to explain how it is structured. Matters are dealt with in the following order:

General introduction (paras 3-17)

The history of the proceedings (paras 18-56)

Res judicata, estoppel and abuse of process (paras 57-101)

The background to section 32(1)(c) of the Limitation Act 1980 (paras 102-140)

The Limitation Act 1980 (paras 141-142)

Kleinwort Benson (paras 143-164)

Deutsche Morgan Grenfell (paras 165-171)

Discussion of Deutsche Morgan Grenfell (paras 172-212)

Deutsche Morgan Grenfell: Summary (paras 213-214)

Discussion of Kleinwort Benson (paras 215-241)

Kleinwort Benson: Summary (para 242-243)

The Practice Statement of 26 July 1966 (paras 244-253)

Application to the present proceedings (paras 254-256)

Conclusion (para 257)

The FII Group Litigation was established by a Group Litigation Order (“GLO”) made on 8 October 2003 (“the FII GLO”). The claimants within the FII GLO are companies which belong to groups which include UK-resident companies and non-resident subsidiaries. The defendants are Her Majesty’s Commissioners for Revenue and Customs (“the Revenue”). The purpose of the FII GLO is to determine a number of common or related questions of law arising out of the tax treatment of dividends received by UK-resident companies from non-resident subsidiaries, as compared with the treatment of dividends paid and received within wholly UK-resident groups of companies. The provisions giving rise to those questions concern, first, the system of advance corporation tax (“ACT”) and, secondly, the taxation of dividend income from non-resident sources under section 18 (Schedule D, Case V) of the Income and Corporation Taxes Act 1988 (“ICTA”) (“the DV provisions”). The relevant provisions of ICTA have since been amended. ACT was abolished for distributions made on or after 5 April 1999, and the DV provisions were repealed for dividend income received on or after 1 April 2009. But the problems created by their existence in the past have not gone away.

Under the FII GLO, certain claims were selected as test claims, and the remaining claims were stayed. The test claimants’ case is that the differences between their tax treatment and that of wholly UK-resident groups of companies breached the provisions of article 43 (freedom of establishment) and article 56 (free movement of capital) of the EC Treaty and their predecessor articles. They seek the repayment of the tax so far as it was unlawful under EU law, dating back in some cases to the accession of the UK to the EU in January 1973 and the introduction of ACT in April of that year (expressions such as “the EU” and “EU law” will be used in this judgment, anachronistically but conveniently, to include earlier incarnations of what is now known as the EU). In the alternative, they seek an award of damages under the principles of EU law established in Francovich v Italy (Case C-479/93) [1995] ECR I-3843, given effect in our domestic law in R v Secretary of State for Transport, Ex p Factortame (No 5) [2000] 1 AC 524.

The system of corporate taxation relating to dividends which underlies the FII Group Litigation has also given rise to litigation managed under a number of other GLOs, including the ACT GLO. Whereas the focus of the ACT Group Litigation is on the UK legislation which prevented UK-resident subsidiaries of foreign parents from making group income elections, thereby obliging them to pay ACT when paying dividends to their foreign parents, the focus of the FII Group Litigation is on UK-parented groups with foreign subsidiaries, and on the tax treatment of dividends coming into the UK from abroad. Although the present litigation is therefore concerned with factual situations which are different from those which have given rise to the ACT Group Litigation, some of the most important legal questions are common to both sets of proceedings. The ACT Group Litigation followed the decision of the Court of Justice of the European Union, as the court is now known (“the Court of Justice”), in the Hoechst case (Metallgesellschaft Ltd v Inland Revenue Comrs, Hoechst AG v Inland Revenue Comrs (Joined Cases C-397/98 and C-410/98) [2001] ECR I-1727; [2001] Ch 620). The ACT Group Litigation includes the decision in Deutsche Morgan Grenfell.

A number of other sets of proceedings have also raised issues which arise in the FII Group Litigation. One is the Controlled Foreign Companies (“CFC”) and Dividend Group Litigation, which also concerns claims that the tax treatment of dividends paid by foreign subsidiaries to UK-resident companies was incompatible with EU law. The principal difference from the FII Group Litigation is that the CFC and Dividend Group Litigation includes claims concerned with “portfolio” holdings of less than 10% of the shares of the relevant companies. Another is the Foreign Income Dividends (“FID”) Group Litigation, which concerns claims by pension funds or life companies that the absence of a tax credit in respect of foreign income dividends, in contrast to domestic dividends, was contrary to EU law. Another is the Stamp Taxes Group Litigation, which concerns claims that stamp taxes on issues or transfers of chargeable securities to clearance or depositary services are contrary to EU law. Other relevant proceedings include the Littlewoods proceedings, which concern claims to restitution based on the payment of VAT which was paid under a mistaken understanding of the relevant EU law.

Since the payments with which these various proceedings are concerned go back, in most if not all cases, to the UK’s entry into the EU in 1973, a central issue in the proceedings has been the limitation of actions. Restitutionary claims for the recovery of money are normally subject under English law to a limitation period of six years from the date when the cause of action accrued, on the basis that they are founded on simple contract within the meaning of section 5 of the Limitation Act 1980 (“the 1980 Act”). Francovich claims to damages are subject to the same time limit, on the basis that they are founded on tort within the meaning of section 2 of that Act. Far more than six years had passed between the date when much of the tax was paid, and the right to its recovery therefore accrued, and the date when the claims were brought. As a result, a large element of the claims, together with interest on it over a period of decades, was potentially time-barred.

The only way around that problem was to rely on section 32(1)(c) of the 1980 Act, which applies to an “action for relief from the consequences of a mistake”, and postpones the commencement of the limitation period “until the plaintiff has discovered the … mistake … or could with reasonable diligence have discovered it.” Section 32(1)(c) has therefore been central to all these proceedings. They have all been based on the propositions that (a) a restitutionary claim lies for the recovery of money, including tax, paid under a mistake of law, (b) such a claim falls within the ambit of section 32(1)(c), and (c) the effect of that provision is to postpone the commencement of the limitation period in respect of such a claim until the true state of the law is established by a judicial decision from which there lies no right of appeal.

Each of these propositions was novel to English law. However, the colossal amounts of money at stake in these proceedings have made it worthwhile for every arguable point to be taken, not least points which might affect the applicable limitation period. The result has been a very protracted series of related proceedings. During the many years since these various proceedings were begun, the relevant principles of English law have been undergoing development, largely driven by those proceedings themselves. It may be helpful to note at this stage the principal milestones along the road, beginning with two decisions of the House of Lords which preceded the bringing of these claims, but set the scene for what followed.

In 1992 the House of Lords held that a taxpayer was entitled to recover tax which was paid in response to an unlawful demand: Woolwich Equitable Building Society v Inland Revenue Comrs [1993] AC 70 (“Woolwich”). In 1998 the House of Lords held that a claim for restitution lay in respect of money paid under a mistake of law, and that such a claim fell within the scope of section 32(1)(c) of the 1980 Act: Kleinwort Benson [1999] 2 AC 349. On 8 March 2001, in the ACT Group Litigation, the Court of Justice issued its judgment in Hoechst [2001] Ch 620, establishing the incompatibility with EU law of the UK tax treatment of dividends paid by UK-resident subsidiaries to foreign parents. In July 2003, at a later stage in the ACT Group Litigation, Park J gave judgment in Deutsche Morgan Grenfell, holding that the principles established in Kleinwort Benson applied to tax paid under a mistake of law, including tax paid in ignorance of the fact that the legislation under which it was charged was incompatible with EU law: [2003] EWHC 1779 (Ch); [2003] 4 All ER 645. In accordance with Kleinwort Benson, he also held that the limitation period applicable to such claims was that laid down by section 32(1) of the 1980 Act, namely six years from the date on which the mistake was or could with reasonable diligence have been discovered. That date, he held, was the date on which the Court of Justice gave judgment in Hoechst, establishing the incompatibility of the legislation in question with EU law. On 8 September 2003 the Government announced proposed legislation to exclude the application of section 32(1)(c) in respect of all mistake claims made on or after that date which related to an Inland Revenue taxation matter. Legislation to that effect was enacted in July 2004, in the form of section 320 of the Finance Act 2004 (“FA 2004”). In February 2005 the Court of Appeal reversed Park J’s decision in Deutsche Morgan Grenfell: [2005] EWCA Civ 78; [2006] Ch 243.

In October 2006 the House of Lords gave judgment in Deutsche Morgan Grenfell [2007] 1 AC 558, reversing the judgment of the Court of Appeal and restoring the decision of Park J. It also decided that the fact that the taxpayer might have a concurrent ground of action under the Woolwich principle, which was subject to a limitation period running from the date of the payment, did not prevent it from pursuing its claim on the ground of mistake. The consequence was that claims in the ACT Group Litigation could be brought for the restitution of tax paid as far back as 1973, provided that the claim had been issued prior to the deadline of 8 September 2003 imposed by section 320 of the FA 2004.

Following the decision of the House of Lords in Deutsche Morgan Grenfell, the Government applied to the Court of Justice for the reopening of the hearing of the first reference in the FII Group Litigation so that it could argue for a temporal restriction on the effect of the Court of Justice’s judgment, which had not yet been handed down. On 6 December 2006 the Court of Justice rejected the Government’s application: Order (Case C-446/04) EU:C:2006:761. On the same day, the Government announced proposed legislation excluding the application of section 32(1)(c) of the 1980 Act in respect of mistake claims made before 8 September 2003 and relating to an Inland Revenue matter. A few days later, in the first reference in the FII Group Litigation, the Court of Justice held that the UK tax treatment of dividends paid by foreign subsidiaries to UK resident parents was incompatible with EU law: Test Claimants in the FII Group Litigation v Inland Revenue Comrs (Case C-446/04) [2006] ECR I-11753; [2012] 2 AC 436 (“FII (CJEU) 1”).

In 2007, at a further stage of the ACT Group Litigation, the House of Lords decided that compound interest was payable on the amounts awarded, whether in damages or in restitution: Sempra Metals Ltd (formerly Metallgesellschaft Ltd) v Inland Revenue Comrs [2007] UKHL 34; [2008] 1 AC 561 (“Sempra Metals”). Taken together with Deutsche Morgan Grenfell, this meant that interest could be compounded for a period stretching back to 1973. The day after judgment was delivered in Sempra Metals, the legislation announced in December 2006 was enacted as section 107 of the Finance Act 2007 (“FA 2007”).

In 2012, in the FII Group Litigation, this court held that a Woolwich claim could lie in the absence of a demand (ACT being self-assessed), but that, in order for a claim to fall within the ambit of section 32(1)(c) of the 1980 Act, a mistake must constitute an essential element of the cause of action, and not merely form part of the context: Test Claimants in the FII Group Litigation v Revenue and Customs Comrs (formerly Inland Revenue Comrs) [2012] UKSC 19; [2012] 2 AC 337 (“FII (SC) 1”). The consequence was that section 32(1)(c) did not apply to the Woolwich ground of restitution. The taxpayer could however seek recovery of tax paid in ignorance of the fact that the legislation under which it was charged was incompatible with EU law, on the basis that it had been paid under a mistake. The case was argued and decided on the assumption that the decisions in Kleinwort Benson and Deutsche Morgan Grenfell were correct. The court also held that section 107 of the FA 2007 was incompatible with EU law. The court referred to the Court of Justice the question whether section 320 of the FA 2004 was also incompatible with EU law in so far as it had retrospective effect. In 2013 the Court of Justice held that it was: Test Claimants in the FII Group Litigation v Revenue and Customs Comrs (formerly Inland Revenue Comrs) (Case C-362/12) [2014] AC 1161 (“FII (CJEU) 3”).

These decisions represented a series of defeats for the Revenue. In more recent times, however, they enjoyed greater success. In 2017, in a test case concerned with the restitution of VAT charged incompatibly with EU law, this court reined in the increasingly expansive approach to restitutionary claims which had followed the adoption of the theory of unjust enrichment in Lipkin Gorman v Karpnale Ltd [1991] 2 AC 548 and Banque Financière de la Cité v Parc (Battersea) Ltd [1999] 1 AC 221: see Investment Trust Companies v Revenue and Customs Comrs [2017] UKSC 29; [2018] AC 275. Later that year, in the Littlewoods proceedings, this court held that common law claims to restitution of VAT, together with any right to compound interest based on Sempra Metals, and the limitation regime imposed by the 1980 Act, had been effectively excluded by the statutory provisions governing the recovery of VAT: Littlewoods Ltd v Revenue and Customs Comrs [2017] UKSC 70; [2018] AC 869. In 2018, in the CFC and Dividend Group Litigation, this court held, having regard to Investment Trust Companies, that Sempra Metals had been incorrectly decided in requiring compound interest to be paid on restitutionary awards, and departed from it: Prudential Assurance Co Ltd v Revenue and Customs Comrs [2018] UKSC 39; [2019] AC 929 (“Prudential”).

The principal question raised by the present appeal is whether, as the Revenue contend (drawing to a considerable extent on Dr Samuel Beswick’s articles, “The discoverability of mistakes of law” (2019) Lloyd’s Maritime and Commercial Law Quarterly 112, and “Discoverability Principles and the Law’s Mistakes” (2020) 136 Law Quarterly Review 139), this court should now depart from the decision of the House of Lords in Deutsche Morgan Grenfell in relation to section 32(1)(c) of the 1980 Act. The Revenue were also granted permission to appeal on the question whether the decision in Kleinwort Benson, so far as relating to limitation, was correct. Ultimately, we did not understand the Revenue to press that point, but the court received submissions upon it, partly at its own request, in view of the bearing of the decision on that subsequently taken in Deutsche Morgan Grenfell.

Before considering the question whether the limitation issues in those two cases were correctly decided, however, the court has first to consider whether, as the test claimants contend, the Revenue are barred in the light of the history of these proceedings, including their failure to raise that question in FII (SC) 1, from now raising the question in these proceedings against the test claimants (whatever impact it might have on the claims of other claimants who are party to the FII GLO), because it is res judicata, or because of an estoppel, or because their doing so amounts to an abuse of process.

These proceedings have a long history. That reflects their exceptional complexity and novelty, and the need to make no fewer than three references to the Court of Justice. What follows is not a complete account, but covers the stages in the proceedings which are relevant to the present appeal.

The FII GLO was made on 8 October 2003, and has been repeatedly amended since then. It defined the type of claims falling within the scope of the GLO, identified the initial claimants, and provided a procedure enabling further claimants to be added. It set out the common issues of fact or law which arose for determination, without prejudice to the power of the High Court to add to or vary them. It also laid down a procedure for selecting claims to proceed as test cases and for amending, removing and adding to the common issues. Claims not selected as test claims were stayed.

The claim on behalf of various members of the British American Tobacco (“BAT”) group was selected as a test claim in relation to a number of issues set out in the GLO, including Issue P: “From what date does the limitation period commence?” A claim by members of the Aegis group was chosen as the test claim in relation to Issue Q, which concerned the effect of section 320 of the FA 2004. As explained above, that provision disapplied section 32(1)(c) of the 1980 Act in relation to claims relating to an Inland Revenue taxation matter which were brought on or after 8 September 2003. It did not apply to the BAT claim, which had been issued on 18 June 2003.

The BAT claim sought inter alia the restitution of tax payments made between 1973 and the issue of the claim, with compound interest, on the basis that the tax had been paid pursuant to a mistake of law or unlawful demands. In its defence, the Revenue pleaded inter alia that any right to restitution or damages which accrued more than six years before the claim form was issued (ie prior to 18 June 1997) was barred by the 1980 Act.

On 28 June 2004 the trial of the BAT claim began, but it was immediately apparent that a preliminary reference to the Court of Justice would be needed on the numerous issues of EU law arising. Without delivering a judgment, Park J directed that a reference be made. It included a number of questions concerning the compatibility of domestic tax provisions with EU law, and also questions concerning the classification under EU law of the claims arising in consequence of any incompatibility.

On 12 December 2006 the Court of Justice gave its judgment on the reference (FII (CJEU) 1 [2012] 2 AC 436). It said at para 184 that “[i]t is clear from case law that any less favourable treatment of foreign-sourced dividends in comparison with nationally-sourced dividends must be regarded as a restriction on the free movement of capital in so far as it is liable to make the acquisition of holdings in companies established in other member states less attractive”. The cases cited as establishing that proposition were Staatssecretaris van Financiën v BGM Verkooijen (Case C-35/98) [2000] ECR I-4071 (“Verkooijen”), para 35; Lenz v Finanzlandesdirektion für Tirol (Case C-315/02) [2004] ECR I-7063 (“Lenz”), para 21 and Proceedings brought by Manninen (Case C-319/02) [2004] ECR I-7477; [2005] Ch 236 (“Manninen”), para 23. In the absence of EU legislation, it was for the domestic legal system to lay down the relevant procedural rules governing actions for safeguarding EU rights, including the classification of claims, subject to the obligation of national courts and tribunals to ensure that individuals should have an effective legal remedy enabling them to obtain reimbursement of the tax unlawfully levied on them and the amounts paid to the member state or withheld by it directly against that tax.

In relation to the Francovich claims for compensation, and the requirement that the breach of EU law by the member state must be “sufficiently serious” before such a claim will lie, the Court of Justice stated at [2012] 2 AC 436, paras 215-216:

Following the judgment of the Court of Justice, Rimer J directed that consecutive trials of the BAT and Aegis test claims should proceed. They would try “all GLO issues raised by the test claims, including liability for restitution, save in so far as those issues concern causation or quantification” (para 12 of Rimer J’s order). Directions were also given for the service of amended pleadings and for preparation for trial, including the agreement of a list of questions to be decided by the court.

The BAT claimants amended their particulars of claim on 13 December 2007 so as to aver that they had made the ACT payments “by reason of their mistaken beliefs (i) that the ACT provisions were lawful and enforceable, and/or (ii) that the claimants were lawfully obliged to make those payments”. A similar averment was also made in relation to the DV payments. The BAT claimants also set out detailed averments in support of their reliance on section 32(1)(c) of the 1980 Act.

In relation to the ACT payments, the BAT claimants averred that they discovered their mistakes when the Court of Justice gave its judgment in FII (CJEU) 1 on 12 December 2006, and could not with reasonable diligence have discovered their mistakes any earlier than then, or alternatively any earlier than 8 March 2001, when the Court of Justice gave its decision in Hoechst. In relation to the DV payments, they averred that the fact that those payments were made by mistake depended upon the final determination of the issues in the proceedings, and could not with reasonable diligence be known or discovered at any other time or in any other way. In other words, although they were bringing a claim for the repayment of the DV tax on the basis that it had been paid under a mistake, they submitted that they could not discover the mistake until the question whether the DV provisions were enforceable had been determined by the court in those proceedings. They also added averments explaining why, in their submission, the application of section 107 of the FA 2007 to their claim would be contrary to EU law. As part of their argument that section 107 should not be applied to their claim, they also averred that the Revenue were estopped from denying that section 32 of the 1980 Act applied to their claim, stating that until 6 December 2006 at the earliest (the date when the Revenue announced their proposal that Parliament should enact what became section 107 of the FA 2007), the parties had proceeded on the common understanding that section 32 applied. Alternatively, they averred that, in failing to propose that there be a separate issue within the GLO as to whether section 32 applied to claims commenced before 8 September 2003 (ie claims falling outside the ambit of section 320 of the FA 2004), the Revenue represented that section 32 applied to the BAT claim and others issued before that date.

In response, the Revenue amended their defence on 21 December 2007. In relation to limitation, they denied that the BAT claimants were entitled to rely on section 32(1)(c) of the 1980 Act, and referred to section 107 of the FA 2007. They averred that any right to restitution which accrued more than six years before the date of issue of the claim form was barred by the 1980 Act. They denied that the parties had proceeded on a common understanding that section 32 applied to the BAT claim, averring that the law in that regard was not fully clarified until 25 October 2006 at the earliest (the date of the decision of the House of Lords in Deutsche Morgan Grenfell [2007] 1 AC 558). In fact, they averred, it was their explicit position at all times prior to that date, as advanced in Deutsche Morgan Grenfell, that section 32 did not apply.

In the light of the amended claim and defences, Henderson J amended Issue Q so as to include the effect of section 107 of the FA 2007 as well as section 320 of the FA 2004. Issue P remained unchanged. The BAT claim became an additional test claim in relation to Issue Q so far as relating to section 107 of the FA 2007, as well as remaining a test claim in relation to other issues, including Issue P.

The trial proceeded over 13 days in July 2008, and Henderson J’s judgment was delivered in November of that year: Test Claimants in the FII Group Litigation v Revenue and Customs Comrs (formerly Inland Revenue Comrs) [2008] EWHC 2893 (Ch); [2009] STC 254 (“FII (HC) 1”). The Revenue were recorded as arguing inter alia that the DV claims were excluded by the statutory provisions for recovery of tax overpaid in section 33 of the Taxes Management Act 1970, and that the ACT and DV payments had not in any event been made under any mistake of law.

Those arguments were rejected. Henderson J characterised the mistake of law as “a mistake as to the lawfulness of the ACT regime or the Case V charge” (para 262): a characterisation which was not strictly accurate, since an incompatibility with EU law does not render a United Kingdom statute unlawful under domestic law, but requires the court to disapply the incompatible provision to the extent which is necessary to comply with EU law: R v Secretary of State for Transport, Ex p Factortame Ltd (No 2) [1991] 1 AC 603. The inaccuracy was however immaterial in the present context, since a mistaken belief that the provisions were enforceable, and that the claimants were therefore obliged to make the payments, would equally be a relevant mistake of law for the purposes of a restitutionary claim based on Kleinwort Benson and Deutsche Morgan Grenfell.

On the facts, the judge found that a mistake had been made. In relation to the ACT claims, he was satisfied on the evidence that the mistake “was not obvious to anybody within the BAT group at the time [when the payments were made], since everybody proceeded on the footing that the tax in question was lawfully due and payable” ([2009] STC 254, para 267). The position in relation to the DV claims was said to be similar (para 275).

The evidence bearing on this point was discussed at a later point in the judgment. At para 391, the judge said:

That evidence was consistent with other evidence adduced in relation to the Francovich claim. In that regard, the judge noted the Report of the Committee of Independent Experts on Company Taxation (the Ruding Committee), established by the European Commission in 1990 to evaluate the need for greater harmonisation of tax. In its Report, published in 1992, the Committee noted the adverse impact on overseas investment caused by discriminatory taxation of dividends from profits earned in another member state. There was, however, no suggestion that the discrimination was contrary to EU law. The same was true of the first draft of a paper by the Adam Smith Institute entitled “An Act Against Trade - UK Tax Prejudice Against Trading Abroad: The Problem of Surplus ACT and its Solution”, which was sent to Mr Etherington, the Head of Tax for the BAT Group, in 1989 by the Director of the Institute. Reference was also made to a number of published articles on the subject by tax lawyers. The last of the articles, published in 1998, was the only one to raise the question whether the difference in treatment constituted a violation of EU law (Lodin, “The Imputation Systems and Cross-Border Dividends - the need for new solutions”, EC Tax Review, 1998, p 229). The author concluded that there was very little guidance to be found in earlier decisions of the Court of Justice, and that the outcome of any challenge was difficult to predict. The judge commented that that assessment reflected the uncertainty acknowledged by the Court of Justice in the present proceedings, “which continued at least until the decision in Verkooijen in June 2000” (para 391). He concluded that, prior to that date, “there was admittedly discrimination between the way in which UK tax law treated domestic dividends and foreign dividends, with domestic dividends receiving the more favourable treatment, but whether this form of discrimination involved a breach of articles 43 and 56 remained unclear until the decision in Verkooijen” ([2009] STC 254, para 395).

In relation to limitation, the judge considered the effect of section 320 of the FA 2004 and section 107 of the FA 2007, that is to say, Issue Q in the GLO, and concluded that it was not open to the Revenue to rely on either provision as a defence to the test claims. The judge also identified a number of issues on which a further reference to the Court of Justice was necessary. None of those issues concerned limitation.

Henderson J’s order, dated 12 December 2008, included a declaration (Declaration 17) that “[t]o the extent that claimants paid unlawfully levied ACT and/or corporation tax under Schedule D Case V, such ACT and/or corporation tax was paid under a mistake.” It also ordered (Order 1) that:

Both the test claimants and the Revenue appealed. It was common ground in the appeal that section 32(1)(c) of the 1980 Act applied in principle to the test claims for money paid under a mistake of law, following the decisions of the House of Lords in Kleinwort Benson and Deutsche Morgan Grenfell. The only point arising in relation to limitation was whether the application of section 32(1)(c) was precluded by section 320 of the FA 2004 in relation to the Aegis claim, and by section 107 of the FA 2007 in relation to the BAT claim. The Court of Appeal concluded that EU law did not preclude the application of either provision, since the claimants continued to have Woolwich claims (subject to a six-year limitation period), and those claims were sufficient to meet the requirements of EU law: [2010] EWCA Civ 103; [2010] STC 1251 (“FII (CA) 1”). The court also directed that a further reference should be made to the Court of Justice, in order to seek clarification of its judgment in FII (CJEU) 1 [2012] 2 AC 436.

Accordingly, the order of the court, dated 19 March 2010, varied Henderson J’s Order 1 so as to exclude claims falling within the scope of the issues to be referred to the Court of Justice. Order 4 was also varied so as to state that all claims made outside the applicable limitation periods were unsuccessful.

In November 2010 this court granted both parties permission to appeal on a number of issues, including the question whether the availability of the Woolwich claims sufficed to meet the requirements of EU law. The second reference was then made to the Court of Justice, and it gave its ruling in 2012: Test Claimants in the FII Group Litigation v Revenue and Customs Comrs (formerly Inland Revenue Comrs) (No 3) (Case C-35/11) [2013] Ch 431 (“FII (CJEU) 2”).

In their submissions in the appeal to this court, the Revenue accepted that section 32(1)(c) of the 1980 Act applied to the test claimants’ claims for restitution on the basis of mistake, subject to the effect of section 320 of the FA 2004 and section 107 of the FA 2007. The argument in relation to limitation was therefore concerned with the effect of those provisions, and with the question whether section 32(1)(c) also applied to the Woolwich claims, as the test claimants submitted. The judgments proceeded on the same basis.

As explained earlier, the court held that, in order for a claim to fall within the ambit of section 32(1)(c) of the 1980 Act, a mistake must constitute an essential element of the cause of action, and that the provision did not therefore apply to a Woolwich claim: FII (SC) 1 [2012] 2 AC 337. In so holding, the court upheld the earlier decision of Pearson J in Phillips-Higgins v Harper [1954] 1 QB 411 (“Phillips-Higgins”). As Lord Walker of Gestingthorpe pointed out at para 63, if that approach were to be departed from, there would be no principled stopping-place for the expansion of the scope of section 32(1)(c) until it overrode the common law rule that ignorance of the existence of a cause of action does not prevent time from running. The consequence would be that the leading case of Cartledge v E Jopling & Sons Ltd [1963] AC 758 would be seen to have missed the point, and the limits and rationale of sections 11 and 14A of the 1980 Act (which extend the limitation period for actions of damages for personal injuries, and other actions of damages for negligence, respectively, until the facts constituting the cause of action are known) would have to be revisited.

The court also held that section 107 of the FA 2007 was incompatible with EU law, and referred two questions to the Court of Justice, including a question concerning the compatibility with EU law of section 320 of the FA 2004. The Court of Justice delivered its judgment in December 2013: FII (CJEU) 3 [2014] AC 1161. In the light of that judgment, this court held in April 2014 that neither section 320 of the FA 2004 nor section 107 of the FA 2007 could be applied to the test claims.

In the meantime, in May 2013 Henderson J ordered that the trial of the BAT claim be resumed to determine all remaining issues of liability and quantification, apart from a few issues, not relating to limitation, which had been referred to the Court of Justice. Henderson J laid down a timetable for the amendment of the pleadings and the agreement of a list of issues to be decided at the resumed trial.

In their amended particulars of claim, the BAT claimants continued to plead mistakes of law as set out at para 26 above, and those averments were admitted by the Revenue. In relation to limitation, the BAT claimants averred:

Nevertheless, the Revenue informed the BAT claimants that they wished to argue at trial that the relevant date was not 12 December 2006 (the date of the judgment in FII (CJEU) 1) but 8 March 2001 (the date of the judgment in Hoechst). Accordingly, the parties agreed that one of the issues to be decided at the trial was Issue 28:

Following a 16-day trial, Henderson J delivered his judgment in December 2014: Test Claimants in the FII Group Litigation v Revenue and Customs Comrs [2014] EWHC 4302 (Ch); [2015] STC 1471 (“FII (HC) 2”). In relation to Issue 28, he noted that the question as to when the claimants could first have discovered their mistake had been left undecided in FII (HC) 1 [2009] STC 254, and that it was of no practical significance to the BAT claimants, since their claim form was issued on 18 June 2003. That date was within the relevant six-year period, whether that period began on 8 March 2001, as the Revenue argued, on 25 October 2006 (the date of the judgment in Deutsche Morgan Grenfell), as the judge was inclined to think, or on 12 December 2006, as the claimants argued. The issue might, however, be relevant to other claims in the FII GLO.

He observed at para 454 that there was what might at first sight appear to be an insuperable logical difficulty in the claimants’ case on this issue: how could it be said that they neither had discovered, nor with reasonable diligence could have discovered, their mistake until 12 December 2006, when they had already started the present action three and a half years earlier? But, he said, that position necessarily followed from the courts’ jurisprudence. By parity of reasoning with the decision of the House of Lords in Deutsche Morgan Grenfell [2007] 1 AC 558, he considered that it was strongly arguable that it was only when that judgment was delivered, on 25 October 2006, that time began to run against the BAT claimants. That judgment was pertinent, in his view, because it was the first time an appellate court had determined that a restitutionary claim lay for the recovery of tax on the ground that it had been paid under a mistake of law. Although Park J had decided the same point three years earlier, it was only the decision of the House of Lords which achieved finality on the issue. However, in the light of the majority judgments in FII (SC) 1 [2012] 2 AC 33, particularly that of Lord Walker, he concluded that the date when the claimants discovered (or could with reasonable diligence have discovered) their mistake was 8 March 2001, when the Court of Justice delivered its judgment in Hoechst [2001] Ch 620.

In that regard, Henderson J referred to Lord Walker’s discussion of legitimate expectations, in the course of which he had observed at [2012] 2 AC 337, para 103 that, until the Court of Justice issued its judgment in Hoechst, there was “no general appreciation that the UK corporation tax regime was seriously open to challenge as infringing the Treaty”, and had stated at para 104 that, after the date of the judgment in Hoechst, “a well advised multi-national group based in the UK would have had good grounds for supposing that it had a valid claim to recover ACT levied contrary to EU law, with at least a reasonable prospect that the running of time could be postponed until then (but not subsequently)”.

It is relevant to note that, when the parties received the judgment in draft, counsel for the claimants complained to the judge that, if the Revenue wished to argue Issue 28, they must apply to amend their pleadings, and satisfy the court that such an amendment should be permitted. In response, counsel for the Revenue noted that no pleading point had been taken until the draft judgment was released, and stated that the Revenue had not sought to amend their pleadings in the test claim because the issue was of no significance in relation to that claim (ie the BAT claim). Both parties had, however, recognised the significance of the issue for other claims (which had been stayed before being pleaded out), and had agreed that it should be included in the list of issues to be decided at the trial. The judge rejected the complaint, noting that the point was included in the agreed list of issues, and observing that the pleaded position as between the test claimants and the Revenue was not relevant to this issue, since both parties agreed that it made no difference so far as they were concerned.

In his order, dated 30 January 2015, Henderson J granted a declaration (Declaration 24) in the following terms:

In the course of the hearing before the Court of Appeal, in June 2016, counsel for the Revenue observed that the central issue in all of the cases concerned with claims for the restitution of money paid under a mistake was “whether section 32(1)(c) does apply to mistakes of law”. He also observed that this “critical issue” might be a matter for this court in the present proceedings. That appears to have been the first indication, in the papers before this court, that the decisions in Kleinwort Benson and Deutsche Morgan Grenfell might be challenged. The hearing proceeded, however, on the basis that the Court of Appeal was bound to follow the decisions of the House of Lords, and the argument focused on the effect of Deutsche Morgan Grenfell.

In November 2016 the Court of Appeal allowed the test claimants’ appeal on Issue 28: Test Claimants in the FII Group Litigation v Revenue and Customs Comrs [2016] EWCA Civ 1180; [2017] STC 696 (“FII (CA) 2”). Declaration 24A was amended so as to read:

In their judgment, delivered in November 2016, the Court of Appeal noted at paras 348-349 the position on the pleadings. After quoting paras 18 and 18B of the amended particulars of claim (para 44 above), they noted that the Revenue had admitted those paragraphs, and observed that “[t]hat no doubt reflected the fact that even on the basis of the fall-back ‘reasonable discoverability’ date of 8 March 2001 the BAT claimants’ claims were comfortably in time since proceedings had been commenced in 2003”. As they noted, however, other claimants had not commenced proceedings until much later, and the Revenue had made it plain that, although pleadings had not been required in the cases of those claimants, it would be raising a limitation defence in them. So it had been agreed that Issue 28 should be determined.

In relation to that issue, the court noted at para 372 that they were bound by the decision of the House of Lords in Deutsche Morgan Grenfell [2007] 1 AC 558, which in their view established that “in the case of a point of law which is being actively disputed in current litigation the true position is only discoverable, for the purpose of section 32(1)(c) of the 1980 Act, when the point has been authoritatively determined by a final court”. An authoritative determination of a related point by a final court in earlier proceedings would only start time to run, in their view, if it necessarily meant that the same conclusion would follow in the instant proceedings. The provisions in issue in Hoechst were not the same as those in issue in the FII GLO, and it was not contended that the decision in Hoechst necessarily meant that the latter provisions also infringed EU law. On that basis, the court concluded that the limitation period began to run for the test claimants only on the date when judgment was delivered in FII (CJEU) 1: that is to say, 12 December 2006, three and a half years after they had issued their claims.

Thereafter, the Revenue sought permission to appeal to this court on a multiplicity of grounds, including Issue 28, and invited the court to depart from that decision in Deutsche Morgan Grenfell. That ground of appeal was directed at the test claimants (ie the BAT claimants) as well as other claimants. Further submissions were filed following this court’s decision in Prudential [2019] AC 929 (para 15 above), inviting the court also to depart from the decision in Kleinwort Benson as to the scope of section 32(1)(c) of the 1980 Act. Following an oral hearing, permission to appeal on Issue 28 was granted, without prejudice to the test claimants’ entitlement to argue that, even if the court were to hold that those decisions should be departed from, that decision should not affect the outcome of the present case, whether by reason of res judicata, issue estoppel, abuse of process or otherwise. The court also directed that the appeal on Issue 28 should be heard in advance of the appeal and cross-appeal on all remaining grounds. In the event, and partly at the invitation of the court, the arguments at the hearing of the appeal involved a comprehensive consideration of the decisions in Kleinwort Benson and Deutsche Morgan Grenfell, so far as relating to limitation.

In 2015 Parliament again responded to restitution claims relating to taxation in the Finance (No 2) Act 2015 (“F(No 2)A 2015”). In section 38 of that Act Parliament introduced Part 8C of the Corporation Tax Act 2010, which imposed a higher rate of Corporation Tax (45%) on the interest paid on restitution claims for overpaid tax, if the interest was not simple interest at a statutory rate. This measure, which counsel described as a “windfall tax”, was Parliament’s response to the large claims which were being made against the Exchequer. In section 52 of F(No 2)A 2015 Parliament also provided that the rates of interest payable on tax-related judgment debts were those set out in tax legislation.

Until June 2016 the Revenue appear to have given no indication that they might seek to challenge the decisions in Kleinwort Benson and Deutsche Morgan Grenfell and argue that section 32(1)(c) does not apply to mistakes of law: para 51 above. Issue 28 itself (para 45 above) assumes that section 32(1)(c) does so apply, and in the courts below that issue was directed to a debate on whether 8 March 2001 or 12 December 2006 is the relevant date under that subsection for the start of the limitation period.

The emergence of the challenge to the decisions in both Kleinwort Benson and Deutsche Morgan Grenfell in this court after such extensive and costly legal proceedings has, unsurprisingly, caused the claimants to advance a vigorous case in which they argue that the Revenue cannot and, in any event, should not be allowed to make this challenge. The claimants’ primary position is that this court should dismiss the appeal in relation to Issue 28 on the grounds of res judicata, estoppel and abuse of process. Alternatively, they submit that the appeal should be limited to the identification of the relevant date under section 32(1)(c), because the wider challenge would contradict the Revenue’s concessions in the courts below, would amount to an abuse of process and would cause the claimants unfair prejudice. As a fall back, the claimants argue that the court should decline to entertain the appeal on Issue 28 in relation to the test claimants and the other claimants whose claims were issued within six years of 8 March 2001, or order that its determination does not apply to those claimants.

The rules or concepts of res judicata, estoppel, and abuse of process support the same legal policies, namely “that there should be finality in litigation and that a party should not be twice vexed in the same matter”: Johnson v Gore Wood & Co [2002] 2 AC 1, p 31, per Lord Bingham of Cornhill. Lord Bingham went on to state:

The claimants in their pleadings on this appeal use the term “res judicata” not as a portmanteau term to describe the different legal principles of which Lord Sumption spoke in Virgin Atlantic Airways Ltd (above), but equate it with cause of action estoppel. Lord Sumption in that case (para 17) described cause of action estoppel thus:

In his exposition of the law in relation to res judicata, with which the other Justices agreed, Lord Sumption quoted the speech of Lord Keith of Kinkel in Arnold v National Westminster Bank plc [1991] 2 AC 93 (“Arnold”) which described this estoppel in these terms (p 104D-E):

In Virgin Atlantic Airways Ltd (above) Lord Sumption stated ([2014] AC 160, para 22) that Arnold was authority for the following propositions:

From these authorities it is clear that cause of action estoppel operates only to prevent the raising of points which were essential to the existence or non-existence of a cause of action. The claimants’ complaint in short is that the Revenue had conceded both in their pleadings and in counsel’s submissions that section 32(1)(c) applied to mistakes of law and that BAT (and by implication other claimants which had raised proceedings within six years after 8 March 2001) faced no limitation defence. Those concessions relate to the defence of limitation. The effect of limitation is to render an otherwise valid claim unenforceable to the extent that the claim relates to periods beyond the period of limitation. The concessions had and have no bearing on the existence or non-existence of the cause of action which is a claim for restitution based on the payment of tax which was paid under a mistaken understanding of the relevant law. The Revenue therefore are not barred from their challenge by cause of action estoppel.

The second estoppel which we must consider is issue estoppel. This expression, which appears to have been coined by Higgins J in the Australian case of Hoystead v Federal Taxation Comr (1921) 29 CLR 537, 561 and adopted by Diplock LJ in Thoday v Thoday [1964] P 181, 197-198, concerns the principle which Lord Sumption in Virgin Atlantic Airways Ltd (above), para 17 described as:

In Thoday (above), p 198, Diplock LJ observed that issue estoppel was an extension of the public policy underlying cause of action estoppel and described it in these terms:

In Fidelitas Shipping Co Ltd v V/O Exportchleb [1966] 1 QB 630, 642 Diplock LJ expressed the view that in an action in which certain questions of fact or law are tried and determined before others and an interlocutory judgment is given, the parties are bound by the determination of that issue in subsequent proceedings in the same action and their only remedy is to appeal the interlocutory judgment. He saw this as an example of issue estoppel.

In Arnold (above), p 105 Lord Keith said that issue estoppel

Lord Sumption in Virgin Atlantic Airways (above), para 21, explained Lord Keith’s judgment in Arnold (above) in relation to issue estoppel. In the case of that estoppel it was in principle possible to challenge a previous decision on an issue not only by taking a new point which could not reasonably have been taken in the earlier proceedings but also (in contrast to cause of action estoppel) “to reargue in materially altered circumstances an old point which had previously been rejected”. In para 22 he stated that Arnold was authority for the following proposition:

The claimants did not argue in their written case that there is an issue estoppel, but Mr Daniel Margolin QC raised the possibility in his oral submissions and we must address it. The answer to this challenge lies in the terms of the GLO and the way in which the proceedings developed. The question of limitation was raised in Issue P in the GLO (“From what date does the limitation period commence?”) and the BAT claim was the test claim in relation to that issue: para 20 above. Issue P was not argued or determined in Henderson J’s first judgment (FII (HC) 1 [2009] STC 254) or in the appeals which arose out of that judgment. The only question relating to a limitation defence which was decided in the first trial was Issue Q, which concerned the effect of section 320 of the FA 2004 and section 107 of the FA 2007: paras 29 and 35 above. This is unsurprising, as in the first phase of the litigation the Revenue’s only limitation defence to BAT’s mistake of law claims was its reliance on those statutory provisions to exclude the application of section 32(1)(c). In the period leading up to the second trial before Henderson J the BAT claimants asserted in their revised pleadings that the mistake claims were not time barred, and the Revenue admitted those assertions: para 44 above. Notwithstanding that admission in relation to the BAT claimants, the Revenue wished to argue that the relevant date under section 32(1)(c) was 8 March 2001 because that date would support a limitation defence in relation to some of the other claims. As a result, the parties agreed that Issue 28 be decided at the second trial: para 45 above. It would not have been possible for the Revenue to argue at first instance or in the Court of Appeal that either Kleinwort Benson or Deutsche Morgan Grenfell was wrongly decided. But until June 2016 the Revenue gave no indication and made no reservation that they might seek to advance such an argument if the case were to return to the Supreme Court. With the benefit of hindsight, that is unquestionably unfortunate. But it does not give rise to an issue estoppel in circumstances where Issue P had to be determined in the second phase of the proceedings and the argument which the Revenue now wish to advance could be raised only in the Supreme Court.

The claimants advance a closely related argument that this court has no jurisdiction to address the challenge which the Revenue now seek to mount. This is because Henderson J in his second judgment (FII (HC) 2 [2015] STC 1471) made the declaration (Declaration 24) which we have set out in para 50 above. That declaration answered Issue 28 by stating two things. First, in Declaration 24A it stated that the date at which the BAT claimants could have discovered their mistake was 8 March 2001. Secondly, in Declaration 24B it stated:

We reject this argument. The failure to appeal the declaration in question does not exclude the jurisdiction of this court. The declaration is not a judicial determination but records an agreed position at that time. Such an order is not readily the subject of an appeal. The issue to which the declaration of the common position gives rise is whether the Revenue should be allowed to depart from that common position by withdrawing their concession at this late stage in the proceedings. That is a matter which we address in paras 83-100 below.

The claimants’ alternative argument is that the Revenue, by seeking to extend Issue 28 into an argument that Kleinwort Benson and Deutsche Morgan Grenfell were wrongly decided, are guilty of an abuse of process. The principle of abuse of process was first formulated by Wigram V-C in Henderson v Henderson (above) and more recently was analysed by the House of Lords in Johnson v Gore Wood & Co [2002] 2 AC 1. In that case Lord Bingham (at p 31B-E) stated:

The abuse of process doctrine is not confined to the raising of subsequent proceedings after the completion of an action but can apply to separate stages within one litigation. See, for example, Tannu v Moosajee [2003] EWCA Civ 815.

In Virgin Atlantic Airways Ltd (above) Lord Sumption agreed with Lord Millett’s analysis of the relationship between on the one hand the estoppels which come within the law of res judicata and on the other the abuse of process doctrine, stating ([2014] AC 160, para 25):

While the concept of abuse of process informs the exercise of the court’s procedural powers, it is not a question of the exercise by the court of a discretion: Aldi Stores Ltd v WSP Group plc [2017] EWCA Civ 1260; [2008] 1 WLR 748, para 16 per Thomas LJ, para 38 per Longmore LJ. If the court, on making the broad, merits-based judgment of which Lord Bingham spoke, concludes that a claim, a defence, or an amendment of a claim or of a defence involves an abuse of process or oppression of the opposing party, it must exclude that claim, defence or amendment. A finding of abuse of process operates as a bar. Thus, as Lord Wilberforce stated in delivering the judgment of the Judicial Committee of the Privy Council in Brisbane City Council v Attorney General for Queensland [1979] AC 411, 425, the doctrine

From these authorities it is clear that for the court to uphold a plea of abuse of process as a bar to a claim or a defence it must be satisfied that the party in question is misusing or abusing the process of the court by oppressing the other party by repeated challenges relating to the same subject matter. It is not sufficient to establish abuse of process for a party to show that a challenge could have been raised in a prior litigation or at an earlier stage in the same proceedings. It must be shown both that the challenge should have been raised on that earlier occasion and that the later raising of the challenge is abusive.

Applying that test to the circumstances of this appeal, we are not persuaded that it is an abuse of process for the Revenue to challenge the decisions of the House of Lords in Kleinwort Benson and Deutsche Morgan Grenfell at this stage of the GLO proceedings. We have reached this view for the following four reasons.

First, the FII Group Litigation has involved novel and developing legal claims raising legal issues of unparalleled complexity, causing the claimants and the Revenue to amend their pleadings in the light of developments of both EU law and domestic law. Henderson J in FII (HC) 2 [2015] STC 1471, para 468) correctly spoke of “a complex and evolving legal landscape”. The claims were and are located at the interface of two developing systems of law: see paras 9-15 above. In English law the right to claim restitution for money paid under a mistake of law was first recognised only in 1998 and the courts, including this court, have been dealing with the ramifications of that decision since then. This is the second occasion on which the FII claims have reached this court and the claims have been materially affected by the judgment of the House of Lords in Sempra Metals [2008] 1 AC 561 and more recently by the judgments of this court in Littlewoods [2018] AC 869 and Prudential [2019] AC 929. On the European plane, the Court of Justice first recognised the incompatibility of the UK corporation tax legislation with EU law in the ACT Group Litigation in Hoechst in 2001, and the FII claims have since then generated no less than three judgments in references to the Court of Justice in 2006, 2012 and 2013. The claimants in the FII Group Litigation, in the ACT Group litigation, and in similar actions seeking the recovery of tax paid under a mistake of law, have been pursuing their claims at the frontier of legal developments. This in part explains the complexity of the legal proceedings, and why legal questions which are of central importance to those claims have only recently been decided or have not yet been determined. The question whether there has been an abuse of process involves a broad merits-based judgment against this very unusual background.

Secondly, the FII Group Litigation has been the subject of case management by the court, which has determined the order in which the questions of legal principle which the parties had identified have been addressed. In the first phase of the litigation 20 issues were sent to trial for determination by Henderson J. As Mr Margolin QC forcefully submitted, it was intended at that stage of the litigation that the first trial before Henderson J would determine all GLO issues relating to the test claims, including liability for restitution, except in so far as the issues concerned causation or quantification of the claims. It is also clear that at that stage the Revenue did not dispute that section 32(1)(c) would have applied to the mistake of law claims but for Parliament’s intervention by enacting section 320 of the FA 2004 and section 107 of the FA 2007 to exclude the operation of that section in relation to mistake claims relating to Inland Revenue taxation matters. But Issue P (“From what date does the limitation period commence?”) was not determined in the first phase of the litigation, because, as the parties then presented their cases, it made no difference to the outcome of the BAT claims. The question raised by Issue P remained to be addressed in a later phase of the litigation.

Thirdly, it is readily understandable why in the first phase of the litigation the Revenue focused on the statutory provisions which Parliament had enacted, namely section 320 of the FA 2004 and section 107 of the FA 2007. Those provisions would have established in domestic law the Revenue’s limitation defence that all claims accruing more than six years before the date of issue of the relevant claim forms were barred by the 1980 Act, but the provisions were held to be incompatible with EU law in so far as they had retrospective effect. Had the Revenue succeeded in establishing the legal enforceability of those statutory responses to the legal developments, they would not have needed to mount a challenge to Kleinwort Benson and Deutsche Morgan Grenfell. In the context of these actions in a developing area of law, we are satisfied that the Revenue’s failure to raise the wider questions relating to section 32(1)(c), while unfortunate, involved no culpability.

Fourthly, it is not disputed that until the first phase of the FII Group Litigation reached this court in 2012, the Revenue could not have raised a challenge to the decisions in Kleinwort Benson and Deutsche Morgan Grenfell as only this court could review those judgments. The Revenue did not do so. Indeed, in response to a question from this court at that hearing, their counsel disavowed any intention to do so in those proceedings. But, at that time, the Revenue’s defence based on the statutory provisions enacted in 2004 and 2007 was still a live issue and Issue P had not been addressed. With the benefit of hindsight, it would have been better if the issue which the Revenue seek to raise in this hearing had been raised before this court in 2012, not least because the BAT claimants estimate that the limitation defence, if successful, would exclude a very large proportion of the value of their claims. But we do not think that it can be said that in the circumstances which prevailed in 2012 the Revenue should have raised the wider issue then. In the context of a very complex group litigation raising many novel questions of law in which the court had left Issue P for a later phase, the Revenue did not act abusively in not mounting the wider challenge then.

There is therefore no bar arising from an estoppel, lack of jurisdiction or the doctrine of abuse of process which prevents this court from considering the Revenue’s challenge to Kleinwort Benson and Deutsche Morgan Grenfell.

There remains the difficult question of the exercise of this court’s discretion in deciding whether to allow the Revenue to advance the arguments which they now seek to deploy. The claimants argue with no little force that the Revenue in the second phase of the FII Group Litigation never stated that they wished to reserve the right to mount a broader attack in their limitation defence, which included a challenge to the Kleinwort Benson and Deutsche Morgan Grenfell decisions. On the contrary, the Revenue admitted in the pleadings in the BAT test case that BAT’s mistake claims were not time barred: para 44 above. Issue 28 in the second phase, which we have set out in para 45 above, is sufficiently broad to support one of the arguments which the Revenue have advanced in this court, namely that a taxpayer could with reasonable diligence have discovered a mistake of law at the date when the tax was mistakenly paid. But in the context of the Revenue’s admissions, which are reflected in Henderson J’s statement of the common position of the parties in Declaration 24B (para 50 above), the agreed focus of that issue was on the Revenue’s argument that 8 March 2001, which is the date on which the CJEU handed down the Hoechst judgment ([2001] Ch 620), was the relevant date under section 32(1)(c), as Henderson J held in Declaration 24A.

The claimants also argue that they have suffered very serious unfair prejudice by the emergence of the challenge to the Kleinwort Benson and Deutsche Morgan Grenfell decisions so late in these proceedings. We discuss this in paras 91-100 below.

These are matters which the court must consider in the exercise of its discretion, as the Revenue’s broader challenge involves not only the withdrawal of a concession and a pleaded admission as against the BAT claimants, but also the raising of a new point of law on appeal. Several cases illustrate the established approach of the courts to the exercise of this discretion.

In Pittalis v Grant [1989] QB 605 the Court of Appeal addressed an application by the landlord appellants to withdraw a legal concession made at first instance and to amend their grounds of appeal to argue for a different interpretation of a provision in the Rent Act 1977 from that which had been argued at first instance. The Court of Appeal allowed the application. Nourse LJ, who delivered the judgment of the court, stated the rule of procedure which operates as a norm, by quoting from the judgment of Sir George Jessel MR in Ex p Firth, In re Cowburn (1882) 19 Ch D 419, 429:

In Jones v MBNA International Bank (30 June 2000) [2000] EWCA Civ 514; [2000] Lexis citation 3292, Peter Gibson LJ (para 38) summarised the practice of the Court of Appeal in these terms:

In Grobelaar v News Group Newspapers Ltd [2002] UKHL 40; [2002] 1 WLR 3024, the House of Lords had to interpret the verdict of a jury, and addressed an application by the claimant’s counsel to withdraw a concession which he had made in the Court of Appeal as to the inferences of fact to be taken from the jury’s award of damages for libel in favour of his client. He was allowed to do so for reasons which are not material to this appeal, but in a passage on which the test claimants rely, Lord Bingham stated (para 21):

A similar note of appellate caution was sounded in Singh v Dass [2019] EWCA Civ 360 in which a claimant sought to raise a new argument under the 1980 Act which he had not advanced at first instance. Haddon-Cave LJ, who gave the judgment of the court, summarised the relevant principles in these terms:

In Notting Hill Finance Ltd v Sheikh [2019] EWCA Civ 1337; [2019] 4 WLR 146 the Court of Appeal, in a judgment delivered by Snowden J, stated that an appellate court has a general discretion whether to allow a new point to be taken on appeal (para 21) and considered and analysed the practice set out in Pittalis and Singh:

The challenge which the Revenue seek to advance has the potential to affect the quantification of the claims very significantly, and it is raised at a late stage in a complex group litigation. It involves this court making a ruling on a question of law. But the claimants argue that they have acted to their detriment and will suffer serious prejudice if the Revenue were to be allowed to widen Issue 28 into a challenge to the authority of Kleinwort Benson and Deutsche Morgan Grenfell and were to succeed in that challenge. Such an outcome would, as we discuss below, require the parties to amend their pleadings and conduct a further trial on the quantification of the test claimants’ claim.

Counsel argues that if the BAT claimants had known that the Revenue might seek to withdraw their admission that the claims which pre-dated 8 March 2007 were not time-barred, they would not have appealed Henderson J’s Declaration 24A (para 50 above) on behalf of the eight claimants who were adversely affected by the decision that the relevant date for the calculation of the limitation period was 8 March 2001. Secondly, they submit that there was a clear demarcation in the phases between liability and quantification and the question of limitation properly belonged to the first phase. Thirdly, the claimants would suffer enormous prejudice if the Revenue’s new case on limitation were to succeed, because the test claimants had expended very substantial resources in the past six years in litigating legal issues relating to the quantification of their claims in the second phase of the Group Litigation and also in challenging the “windfall tax” imposed by the F(No 2)A 2015. Counsel estimated that the claimants had incurred costs of about £9.8m, net of recovery through awards of costs, on the FII Group Litigation and the “windfall tax” challenge. Fourthly, if the Revenue were to succeed, this might necessitate a retrial of questions of quantification. The test claimants also assert that they have been prejudiced because this court in its judgments in Littlewoods [2018] AC 869 and Prudential [2019] AC 929, which have a materially adverse effect on the quantification of their claims by excluding compound interest on those claims, was influenced by the disruption to public finances which the application of section 32(1)(c) to claims for the repayment of tax would entail.

We consider this challenge to be the most difficult to determine of the claimants’ preliminary challenges to the scope of this appeal. With hindsight, there is no doubt that it would have been better if the Revenue at the start of the second phase of the FII Group Litigation had reserved their right to mount the challenge which they seek to make in this court. It is important that there be discipline in the conduct of actions which are the subject of Group Litigation Orders and it is important that there be finality in the determination of issues raised in such actions. An appellate court, in the interests of justice, will normally seek strenuously to avoid an outcome which results in the parties, who have already gone to trial on the quantification of a claim, having to amend their pleadings and to adduce further evidence to apply its ruling on a new issue of law to the facts of their case. In a normal litigation, the need for a re-trial would be a strong and normally determinative pointer against allowing a party to withdraw a concession which had influenced the way in which a litigation had been conducted.

There are nonetheless several factors which point in the other direction which make it appropriate not to apply the normal rule. The court is being asked to exercise a discretion not in an individual case but in the context of a group litigation order, a procedural phenomenon which did not exist when Lord Herschell wrote his speech in The Tasmania (1890) 15 App Cas 223. One must also have regard to the nature and subject matter of this group litigation and the manner in which it has been conducted. It is not suggested that the BAT claimants have not had time to deal with the legal challenge. We do not accept that, as the FII Group Litigation progressed, there was a complete demarcation between liability and quantum in the first and second phases: the BAT claimants accept that in the second phase, 19 of the 29 issues related to quantification. The others did not. Issue P, which became Issue 28, remained to be resolved and Issue 17 (namely whether the tax credits given to shareholders for ACT prevented the Revenue from being enriched) raised an issue of principle which could have had a material effect on the quantification of the claims.

Because Kleinwort Benson and Deutsche Morgan Grenfell were rulings by the House of Lords, the Revenue could not have mounted the challenge in the courts below in the second phase; the Revenue could only have given notice that such a challenge might be made. If such notice had been given, how far would the BAT claimants have acted differently?

We are persuaded that Henderson J’s Declaration 24A would have been appealed. In the context of the FII Group Litigation, the starting date of the six-year limitation period was of material importance to the claimants who were prejudiced by Henderson J’s determination. Any one of those claimants could have applied for permission to appeal that declaration under CPR Part 19, rule 19.12(2). It is important to bear in mind the context of this litigation in which this court is asked to make rulings on issues of legal principle which will affect directly or indirectly other claimants besides the BAT claimants, both within and outside the particular GLO. In that context, the loss of the opportunity for the BAT claimants to secure a procedural advantage to close off the issue so far as it related to their claims and those of the other 18 claimants who were not prejudiced by Henderson J’s determination by not appealing against Declaration 24A is a consideration which carries only limited weight.

It is possible that the BAT claimants’ approach to the sequencing of the issues in Phase 2 of the litigation, and in particular the quantification of their claims, would have been different. They might have wished the challenge to Kleinwort Benson and Deutsche Morgan Grenfell to be resolved before they expended time and money on quantification. But the claimants in the FII GLO would still have substantial claims, which the Revenue estimate to be between £80m and £130m, if the limitation challenge which the Revenue now seek to pursue were to succeed; and they needed to complete the litigation to establish those claims. Further, the BAT claimants and the other claimants were prepared to incur the costs in relation to quantification when there was no final determination of issues such as Issue 17 and when the question whether there was an entitlement to compound interest, which was determined adversely to their interest by this court in Littlewoods and Prudential, had yet to be conclusively resolved. It is therefore mere speculation on the information before this court for us to say what the claimants might have done if the Revenue had reserved their position on Kleinwort Benson and Deutsche Morgan Grenfell. Insofar as the BAT claimants are able to persuade the court that they have suffered prejudice by incurring costs which they would not have incurred but for the admission that there was no time bar defence in relation to the BAT claimants and (by implication) the 18 other claimants who commenced proceedings before 8 March 2007, it may be possible to provide a remedy by revising the orders for costs which have been made in the proceedings or by making a further order for costs.

We do not consider that the costs which the claimants have incurred in their challenge to the “windfall tax” in the F(No 2)A 2015 are a relevant consideration as that is a separate litigation relating to different statutory provisions. That legislation was enacted before several decisions which have materially affected the value of the claimants’ claims had been determined. It predated this court’s judgments in Littlewoods and Prudential, which excluded claims for compound interest as a component of a claim for restitution. We cannot know whether Parliament would have acted differently in 2015 if the Revenue had reserved a right to challenge the Kleinwort Benson and Deutsche Morgan Grenfell decisions before this court at a future date.

We also consider that the points which we have made in para 78 above in relation to the abuse of process claim are both relevant and of great weight when considering the exercise of this discretion. The nature of the claims, depending as they do on a developing area of law, means that it is important that this court address the legal questions which the Revenue wish to raise. The size of the claims and their impact on the public purse are also relevant considerations, as it would be wrong to uphold such claims if they are based on an incorrect understanding of the law. As we have said, even if the Revenue’s challenge to the application of section 32(1)(c) succeeds, the claimants will have claims of substantial value. The legal question is also of great importance to other claimants outside the FII Group Litigation, including claimants in the litigations to which we have referred in paras 5-6 above, who also have claims of high value.

In the end, the task for the court is to make an evaluation of what justice requires in the circumstances of this group litigation. We are persuaded for the reasons set out above that we should allow the Revenue to withdraw their concession and to amend their pleadings to remove the admission on which the test claimants found.

The final preliminary matter which we must consider is the test claimants’ application that, in the event that the court allows the Revenue to withdraw their concession and mount the challenge, the court should decline to entertain the appeal in relation to the 19 claimants whose claims were issued within six years after 8 March 2001 or, by analogy with CPR rule 19.12, order that any judgment or order which it makes shall not be binding on those claims. For the reasons which we have set out in paras 94-100 above (other than the effect of the determination on claimants outside the FII Group Litigation) and in particular that, if we were to hold that either Kleinwort Benson or Deutsche Morgan Grenfell was wrongly decided in relation to the interpretation of section 32(1)(c) of the 1980 Act, those claims would to that extent be based on an incorrect understanding of the law, we are not persuaded that the interests of justice require this court to make such orders.

The 1980 Act is a consolidation statute, designed to consolidate the 1939 Act and a number of subsequent enactments. Section 32(1), in particular, is a re-enactment of section 26(b) of the Limitation Act 1939 (“the 1939 Act”), subject to a minor amendment which appears in section 32(1)(b). Nevertheless, as its interpretation raises questions of substantial difficulty, it is both permissible (Farrell v Alexander [1977] AC 59, 72-73) and necessary to consider the previous law in some detail, as the House of Lords did in Kleinwort Benson and as this court did in FII (SC) 1 [2012] 2 AC 33.

When considering the state of the law prior to the 1939 Act, in so far as it related to “action[s] for relief from the consequences of a mistake”, and the limitation period applicable to such actions, it is necessary to distinguish between actions at law and claims for equitable relief. So far as common law actions are concerned, there were a number of types of action which might be described as “action[s] for relief from the consequences of a mistake”. But the mistake was invariably one of fact, rather than law. In particular, it had been established for almost 200 years that no claim lay at common law for the recovery of money paid under a mistake of law: see, for example, Bilbie v Lumley (1802) 2 East 469. That was settled law in 1939, and continued to be so until the decision in Kleinwort Benson.

As Atkinson J pointed out in Anglo-Scottish Beet Sugar Corpn Ltd v Spalding Urban District Council [1937] 2 KB 607, 615-616, in most cases of payment by mistake the person paying has paid because of a mistake as to his legal right or obligation, and “whether the payment can be recovered or not depends upon whether that mistake as to legal right is due to a mistake of fact or a mistake of law”. The distinction between these alternatives gave rise to disputes in borderline cases, and was considered in a multitude of authorities, in which fine distinctions were sometimes drawn.

There were a number of statutes concerned with limitation in relation to common law actions. The most important for present purposes was the Limitation Act 1623 (21 Jac 1, c 16), as amended by the Administration of Justice Act 1705 (4 & 5 Anne c 16) and the Mercantile Law Amendment Act 1856 (19 & 20 Vict, c 97) (“the 1623 Act”). It imposed time limits of 20 years on the bringing of real actions and six years, running from the accrual of the cause of action, on the bringing of certain personal actions, including trespass, trover, replevin, actions of account, action on the case and actions of debt. It is apparent from the names of the forms of action to which the statute applied, and from the fact that they were referred to as “actions”, that the only proceedings barred were actions at law.

Actions on the case included actions of indebitatus assumpsit on a count for money had and received, which was the relevant form of action for restitution of money paid under a mistake. In such cases, the cause of action accrued on the date of the payment: Baker v Courage & Co Ltd [1910] 1 KB 56. The limitation period therefore began to run on that date.

The position in equity is more complex. As Lord Walker observed in FII (SC) 1 [2012] 2 AC 337, para 62, “the authorities are rather short on clear exposition of the relevant principles of equity”. It is also necessary to bear in mind that cases which involved a mistake also often involved other factors which formed the justification for equitable relief, such as fraud, misrepresentation or abuse of a fiduciary position. For present purposes, in the light of the decision in FII (SC) 1, it is also necessary to distinguish between cases where mistake was an essential element of the claim for relief, and cases where it was not.

The law as it was understood in the 1930s is broadly summarised in Snell’s Equity, 21st ed (1934), p 428:

Snell listed four kinds of case in which equitable relief could be given from the consequences of a mistake. First, mistake was accepted as being a ground in some circumstances for refusing specific performance of a contract. Secondly, mistake could in some circumstances justify the exercise of an equitable jurisdiction to grant rescission of a contract (it is unnecessary to consider in this appeal whether such a jurisdiction survived the decision in Bell v Lever Bros Ltd [1932] AC 161: a question considered in Great Peace Shipping Ltd v Tsavliris Salvage (International) Ltd [2002] EWCA Civ 1407; [2003] QB 679). It is relevant to note that in a leading case of common mistake where equity intervened, a distinction was drawn between ignorance of the general law, which could not justify rescission, and a mistake as to private rights of ownership, which could, but was categorised as a mistake of fact: Cooper v Phibbs (1867) LR 2 HL 149, 170 (where the plaintiff contracted to purchase property from the defendant which, unknown to either of them, the plaintiff already owned in equity). Thirdly, equity could provide relief where a written contract failed to express correctly the parties’ antecedent agreement, by providing the remedy of rectification. Fourthly, although it was a general rule in equity, as at common law, that money paid under a mistake of law could not be recovered, there were said to be certain exceptions.

The general rule was stated in Snell’s discussion of mistake at pp 439-440:

Snell mentioned a number of supposed exceptions to the general rule. As Snell noted at p 440, the first supposed exception, where the mistake was as to foreign law, was merely apparent, since foreign law was treated as a matter of fact. The second supposed exception was where money was paid to an officer of the court, such as a trustee in bankruptcy, under a mistake of law. It was held that the court could prevent its officer from taking advantage of the mistake: see, for example, Ex p James, In re Condon (1874) LR 9 Ch App 609 and Ex p Simmonds, In re Carnac (1885) 16 QBD 308. In these cases, however, the grant of relief was not based on mistake, but on the court’s jurisdiction to enforce high ethical standards on the part of its officers.

Vaughan Williams LJ explained this in In re Tyler, Ex p The Official Receiver [1907] 1 KB 865. Referring to Ex p James, he said at p 869:

The third supposed exception was where the mistake was induced by fraud or by the breach of a fiduciary duty. The authorities cited by Snell (British Workman’s and General Insurance Co v Cunliffe (1902) 18 TLR 425, Harse v Pearl Life Assurance Co [1904] 1 KB 558 and Phillips v Royal London Mutual Assurance Co (1911) 105 LT 136) were concerned with claims for the return of premiums, brought by persons who had entered into contracts of insurance which were illegal and void (for want of an insurable interest) as a result of misrepresentations made by or on behalf of the insurance company. Where the misrepresentation was innocent, the money was irrecoverable. Where the misrepresentations were fraudulent, relief was granted, but on the basis of fraud, not mistake: see Harse v Pearl Life Assurance Co at p 563, where Sir Richard Collins MR indicated that relief might also be granted in cases of duress or oppression, or where the defendant stood in a fiduciary relationship towards the plaintiff.

Accordingly, the authorities provide examples of equitable relief being given where there had been mistakes of law as well as mistakes of fact. However, Snell provides no example of a money claim for relief from the consequences of a mistake of law, where the occurrence of the mistake was an essential element of the claim. The judgments in cases such as Rogers v Ingham and In re Tyler, Ex p The Official Receiver indicate that a money claim could not be brought on that basis.

As was mentioned earlier, it is necessary in the light of FII (SC) 1 to distinguish between cases where mistake is an essential ingredient of the cause of action, and cases where there may have been a mistake but the claim has another legal basis. There were by the 1930s a number of established types of claim in equity which fell into the latter category, in addition to those already mentioned. One was a claim for an account, based on a duty to account arising from the relationship between the parties, but where the claim might have been prompted by the discovery of a mistake. Another example, although not a claim at all, was the correction of errors of account between trustees and beneficiaries: the courts would allow a trustee or personal representative to deduct sums overpaid under a mistake of law from future instalments due to the overpaid beneficiary. On the other hand, there does not appear to be any reported case where a trustee or personal representative recovered money paid under a mistake of law from the recipient, and there are dicta to the effect that such a claim must fail because of the general rule barring such recovery.

Another example of a claim which might be brought where a mistake had occurred, but where the mistake was not the justification for the grant of relief, was a claim brought where an executor administering the estate of a deceased person paid out funds to someone other than the person to whom they were properly due, and that person then sought to recover them from the recipient. The remedy available to the person to whom the money was legally due lay in the first instance against the executor, but he could also recover from the recipient any amount which he was unable to recover from the executor. Such a claim was not, however based on mistake: it was, as the Court of Appeal said in In re Diplock [1948] Ch 465 (“In re Diplock”), p 502, “an equitable claim independent of a mistake of fact or of any mistake”. It was based, rather, on the fact that the payment had been made by the executor to a person who was not entitled to it, in breach of the rights of the person to whom it was legally due, as Lord Davey explained in Harrison v Kirk [1904] AC 1, 7.

So far as limitation is concerned, there was not before 1833 any statute which explicitly barred any suit in equity. In so far as the Court of Chancery applied statutes of limitation, it did so by analogy, as explained below. From 1833 onwards, however, a number of statutes were enacted which imposed limitation periods on the bringing of particular types of suit in equity. For example, the Real Property Limitation Acts of 1833 and 1874 introduced limitation periods in respect of equitable proceedings to recover interests in land, and the Trustee Act 1888 established a limitation period for certain claims against trustees. Many types of equitable proceedings remained subject to no limitation period: for example, there was no provision imposing a time limit on proceedings to rescind transactions induced by undue influence or innocent misrepresentation, and no time limit within which proceedings for rectification must be brought. The statutes did not modify the equitable doctrines of laches and acquiescence.

Where equity provided a remedy corresponding to a remedy at law, and the latter was subject to a limitation period, the courts of equity (or after the Judicature Acts, courts asked to give equitable relief) applied the statutes of limitation by analogy, as Lord Westbury explained in Knox v Gye (1872) LR 5 HL 656, 674-675:

The common law courts were bound to apply the statutes according to their terms, but the Court of Chancery, when it applied them by analogy, developed a principle that a defendant whose unconscionable conduct had denied the plaintiff the opportunity to sue in time should not in conscience be permitted to plead the statute to defeat the plaintiff’s claim, provided the claim was brought timeously once the plaintiff discovered or should have discovered the basis of his claim. Accordingly, where the plaintiff’s claim in equity was founded on the fraud of the defendant, time did not begin to run against the plaintiff until he discovered the fraud or had a reasonable opportunity of discovering it.

This equitable rule received partial recognition in section 26 of the Real Property Limitation Act 1833 (the lineal ancestor of section 26 of the 1939 Act and section 32 of the 1980 Act), under which the right to bring a suit in equity for the recovery of land or rent of which the claimant or his predecessors were deprived by concealed fraud was deemed to have accrued “at and not before the time at which such fraud shall or with reasonable diligence might have been first known or discovered”.

In cases where the claim for equitable relief arose in circumstances where the claimant had been unaware of the matter in question as the result of a mistake, and where equity applied the statutes of limitation by analogy, allowance was similarly made for the period before the mistake was or could with reasonable diligence have been discovered. The point is illustrated by the judgment of Alderson B in Denys v Shuckburgh (1840) 4 Y & C Ex 42, where the profits of a mine had for many years been distributed between the parties under a mistake as to their respective shares. When the mistake was discovered, the plaintiff filed a bill for an account, and the question arose whether more than six years’ profits could be recovered in equity. The plaintiff relied on Alderson B’s earlier judgment in Brooksbank v Smith (1836) 2 Y & C Ex 58, where the court applied the 1623 Act by analogy but held that time did not run until the mistake was discovered, since it was only then that laches commenced. In Denys v Shuckburgh, on the other hand, Alderson B explained at p 53 that the position was different where the mistake could reasonably have been discovered earlier than it was:

In this context, a distinction was drawn between a mistake as to the facts supporting a claim for equitable relief, and ignorance that known facts gave rise to a claim. Knight Bruce LJ observed in Stafford v Stafford (1857) 1 De G & J 193, 202 that “[g]enerally, when the facts are known from which a right arises, the right is presumed to be known”. Similar observations were made by Sir Richard Collins MR in Molloy v Mutual Reserve Life Insurance Co (1906) 94 LT 756, 761, in a judgment which is discussed at paras 204-208 below.

In 1934 the Law Revision Committee was invited to consider various aspects of the law of limitation, including the scope of the rules on concealed fraud. The Committee reported in 1936 (Fifth Interim Report, on Statutes of Limitation, Cmd 5334), and its Report formed the background to the 1939 Act. The passages in the Report which are relevant for present purposes begin with the Committee’s explanation of the limitation of claims for equitable remedies, at paragraph 13:

The concluding words in that passage described what the Committee later referred to as “the equitable rule”. As will be explained, it was the Committee’s recommendation to extend that rule to common law claims which resulted in the enactment of section 26 of the 1939 Act, effectively re-enacted as section 32 of the 1980 Act.

In relation to cases of fraud, the Committee noted at paragraph 22 the problem that “[a]s a general rule it is no answer to a plea of the Statutes of Limitation to say that the plaintiff was unaware of the existence of his cause of action until after the expiry of the statutory period”. Exceptions to that general rule included section 26 of the Real Property Limitation Act 1833, and “the equitable doctrine that a plaintiff is not to be affected by the lapse of time where his ignorance is due to the fraud of the defendant, and he has had no reasonable opportunity of discovering such fraud before bringing his action”. It also noted that, following the Judicature Act 1873, there were inconsistent decisions as to whether the equitable doctrine applied to actions in which a court of law would previously have had exclusive jurisdiction. The Committee considered that the position should be clarified so as to prevent defendants from relying on a lapse of time which was due to their fraudulent conduct. It also considered that exception created by the equitable rule should be extended so as to apply not only where a cause of action was founded on a concealed fraud, but also where a cause of action unconnected with fraud was fraudulently concealed from the plaintiff or someone through whom he claimed.

The Committee then turned to cases of mistake, and stated at paragraph 23:

When, in that passage, the Committee stated that “the mere fact that a plaintiff is ignorant of his rights is not to be a ground for the extension of time”, it did not have in mind a situation in which a mistake of law gave rise to a cause of action falling within the scope of statutory limitation, directly or by analogy: as we have explained, no such cause of action existed at that time, and therefore the possibility of such a situation did not arise. It was, as we understand it, reaffirming the principle stated in Stafford v Stafford and Molloy v Mutual Reserve Life Insurance Co (para 122 above) that, whereas allowance could be made for a mistake where it formed one of the ingredients of a cause of action, allowance could not be made, where the ingredients of a cause of action were known, for ignorance that those circumstances gave rise to a cause of action.

Accordingly, in relation to cases involving fraud or mistake, the Committee recommended at paragraph 37:

The 1939 Act gave effect to those recommendations, and also made other changes to the law. Part I laid down periods of limitation for different classes of action, subject under section 1 to the provisions of Part II, “which provide for the extension of the periods of limitation in the case of disability, acknowledgment, part payment, fraud and mistake”. Section 2(1) laid down a six-year limitation period, running from the date on which the cause of action accrued, for a number of categories of action, including “(a) actions founded on simple contract or on tort”. Section 2(7) provided:

In Part II, section 26 provided, so far as material:

Part III contained general provisions. Section 29 preserved the equitable jurisdiction to refuse relief on the ground of acquiescence or otherwise. Section 34 repealed all relevant subsisting statutory provisions for limitation.

It is apparent from the opening words of section 26 of the 1939 Act that it was concerned only with actions for which a period of limitation was prescribed by the Act. Section 26(c), which applied where “the action is for relief from the consequences of a mistake”, was therefore confined to actions meeting that description for which a period of limitation was prescribed by the Act. It had to be construed in the light of section 2(7), and therefore extended to claims for equitable relief for which a period prescribed in section 2 applied by analogy, in the same way as the earlier statutes of limitation (repealed by section 34) were previously applied.

It follows that section 26(c) applied to claims for the recovery of money paid under a mistake of fact. Actions at law of that kind had previously fallen within the ambit of section 3 of the 1623 Act, and were intended by Parliament to fall within the scope of section 2(1) of the 1939 Act, as the Court of Appeal held in In re Diplock at p 514. Equivalent claims in equity (eg where the plaintiff was not the person who made the payment under a mistake) fell within the ambit of section 2(7). However, section 26(c) was not understood to apply to actions for the recovery of money on the ground that it had been paid under a mistake of law, since no action of that description, whether in law or in equity, was recognised until long after the 1939 Act had been repealed.

The 1939 Act was considered in two cases which are relevant in the present context. The first was In re Diplock. The proceedings were brought after executors distributed the residue of an estate in accordance with a provision in the will directing them to hold it in trust and divide it between such charitable or benevolent objects as they might think fit, without further specification. The next of kin challenged the validity of the trust, and it was held by the House of Lords to be void for uncertainty. More than six years (but less than 12 years) after the distributions had been made, the next of kin sought a declaration that the recipients of the money were liable to refund it to them. The claims were made on two bases. The first was a claim in personam based on the right of an unpaid beneficiary to recover money wrongly paid to a stranger to the estate. The second was a claim in rem, based on tracing the trust assets into the hands of the defendants. It is unnecessary to consider the latter aspect.

At first instance, the judge failed to recognise that the personal claims fell within the ambit of the principle relating to the wrongful distribution of estates, and instead treated them as claims for money had and received. On that basis, he held that no claim was available, either at law or in equity, since the mistake was one of law: In re Diplock [1947] Ch 716. The judge’s decision on that point was reversed on appeal: [1948] Ch 465. The Court of Appeal correctly held that an equitable claim lay against a recipient who was paid more than he was entitled to receive under a will, regardless of whether the overpayment was made under a mistake, either of fact or of law. As discussed in para 116 above, the court explained that the basis of equitable relief was not mistake, but the receipt of a share or interest in the estate to which the recipient was not entitled, at the expense of the person entitled to it. The primary claim lay however against the executors, and the equitable cause of action was therefore for recoupment of such amounts as were irrecoverable from them.

The Court of Appeal further held that limitation was governed in such a case by section 20 of the 1939 Act, and not by section 2(1) or (7). Since section 20 laid down a 12-year period, it followed that the claims were not time-barred. However, the court went on to consider, obiter, the position if, contrary to their view, the claims fell within the scope of section 2(7). On that hypothesis, the court considered that section 26(c) would be relevant, on the basis that the claims sought relief from the consequences of a mistake. This obiter dictum preceded the line of authority, culminating in the decision in FII (SC) 1, which entailed that a claim such as that in In re Diplock, for which a mistake was not an essential ingredient of the cause of action, did not fall within the scope of section 26(c) of the 1939 Act or section 32(1)(c) of the 1980 Act.

The Court of Appeal’s decision was affirmed by the House of Lords: Ministry of Health v Simpson [1951] AC 251. In a speech with which the other members of the Appellate Committee expressed agreement, Lord Simonds emphasised at p 265 “that the particular branch of the jurisdiction of the Court of Chancery with which we are concerned relates to the administration of assets of a deceased person”. Lord Simonds next cited the dictum of Lord Davey in Harrison v Kirk which was mentioned in para 116 above, and stated at p 266:

The argument that this jurisdiction was limited to payments made under a mistake of fact, rather than law, was rejected by Lord Simonds at pp 269-270, on the basis that the equitable doctrine was not based on the existence of a mistake at all, but on the making of a wrongful payment. As he said at p 270, “a legatee does not plead his own mistake or his own ignorance but, having exhausted his remedy against the executor who has made the wrongful payment, seeks to recover money from him who has been wrongfully paid”. In relation to limitation, Lord Simonds agreed with the Court of Appeal that the claims were governed by section 20 of the 1939 Act. He added at p 277 that it was unnecessary to say anything about section 26 by way of approval or disapproval of what fell from the Court of Appeal. He observed that it was a section which presented many problems.

The other case from this period which should be noted is Phillips-Higgins v Harper [1954] 1 QB 411, a decision of Pearson J. The plaintiff brought a claim for an account and payment of money due under a contract over a period of 13 years. The defendant argued that as more than six years had passed since the initial payments were due, it followed that the claim was to that extent time-barred, under section 2(2)(a) and (7) of the 1939 Act. In response, the plaintiff relied on section 26(c), arguing that she had not known that the money that had been paid to her was less than was due under the contract, and was therefore seeking relief from the consequences of a mistake.

That argument was rejected by the judge. As he noted, section 26 dealt differently with fraud and mistake. In relation to fraud, provision was made for two situations: first, where “(a) the action is based upon … fraud”, and secondly, where “(b) the right of action is concealed by … fraud”. It followed that, in cases falling within (b), the action need not be based upon fraud. In relation to mistake, on the other hand, provision was made for only one situation: where “(c) the action is for relief from the consequences of a mistake”. In the judge’s view, that wording was “carefully chosen to indicate a class of action where a mistake has been made which has had certain consequences and the plaintiff seeks to be relieved from those consequences” (p 418). No provision was made for the situation where the right of action was concealed by a mistake. In the instant case, the plaintiff’s claim was to recover money due to her under a contract. The fact that she had been unaware of the right of action by reason of a mistake was insufficient to bring her within the ambit of section 26(c). The judge expressed the opinion at p 419 that “[p]robably provision (c) applies only where the mistake is an essential ingredient of the cause of action”. He added (ibid) that it was no doubt “intended to be a narrow provision, because any wider provision would have opened too wide a door of escape from the general principle of limitation”. That reasoning, subsequently approved in FII (SC) 1, entailed that section 26(c) could not apply to a claim of the kind considered in In re Diplock, since such a claim was not based on mistake, as explained in paras 116 and 137-138 above.

As previously mentioned, the 1980 Act is a consolidation statute, designed to consolidate the 1939 Act and a number of subsequent enactments. Section 5 lays down a six-year limitation period for actions founded on simple contract. Like section 2 of the 1939 Act, it has been held to apply to claims for the recovery of money on the ground that it was paid under a mistake: Westdeutsche Landesbank Girozentrale v Islington London Borough Council [1994] 4 All ER 890, 942-943; Aspect Contracts (Asbestos) Ltd v Higgins Construction plc [2015] UKSC 38; [2015] 1 WLR 2961, para 25.

Section 32(1) of the 1980 Act corresponds to section 26 of the 1939 Act, subject to the deletion (originally effected by the section 7 of the Limitation Amendment Act 1980) of the reference to concealment by fraud and the substitution in section 32(1)(b) of the concept of deliberate concealment of relevant facts. It provides (so far as material):

The case of Kleinwort Benson [1999] 2 AC 349 concerned claims by a bank for the recovery of sums which it had paid to local authorities under interest rate swap agreements which it had believed to be valid, but which were subsequently held, initially by the Divisional Court and subsequently by the House of Lords, to be ultra vires and therefore void: Hazell v Hammersmith and Fulham London Borough Council [1992] 2 AC 1. The bank then recovered, in a first set of proceedings, such sums as had been paid within the six years preceding the issue of its writ, on the ground that there had been a failure of consideration. In a second set of proceedings, the bank sought to recover the sums which had been paid more than six years previously, by relying on section 32(1)(c) of the 1980 Act to postpone the commencement of the limitation period. A preliminary issue arose as to (1) whether the bank had a cause of action based on mistake, and (2) if so, whether the bank could rely on section 32(1)(c).

There was no reasoned judgment at first instance, the judge concluding on issue (1) that he was compelled by authority to deny liability. For the same reason, the leapfrog procedure was employed to bypass the Court of Appeal. Before the House of Lords, the bank sought first to establish that there was a cause of action to recover money paid under a mistake of law, and that its claim could be brought on that basis, on the footing that the payments had been made under the mistaken belief that the contracts were legally valid. In response, the local authorities did not attempt to defend the rule that money paid under a mistake of law was irrecoverable, but argued that recovery should not lie where a payment was made in accordance with a settled understanding of the law which was later changed by a judicial decision. The better course, it was argued, was to leave the law to be altered by Parliament, particularly in view of the problems arising in relation to the law of limitation. Since the Government had accepted the Law Commission’s recommendations on the issue, it would be wrong for the courts to pre-empt legislative reform. In reply, the bank accepted (as was noted at pp 362 and 391) that a payment made on the basis of a settled understanding of the law would not be made under a mistake, even if the law was later changed by a judicial decision, but argued that the law on the issue in question had not been settled prior to the House of Lords’ decision in Hazell.

By a majority of three to two, the Appellate Committee accepted the bank’s argument, and went beyond it by holding that the right to recover payments made under a mistake of law applied whether or not the basis on which the payment was made was in accordance with settled law. Lord Goff, in his final speech before his retirement, focused on the retrospective effect of judicial decisions. He accepted that the question whether a payment was made under a mistake was determined as at the time when the payment was made (Baker v Courage at p 66), and observed that “when the judges state what the law is, their decisions … have a retrospective effect” (p 378). It was because of that retrospective effect, he asserted, that it was “plain” (p 379) that a previous understanding of the law which was overturned by a judicial decision was mistaken as at the time when the payment was made. The cause of action for the recovery of money paid under such a mistake of law therefore accrued on the date when the payment was made (p 386). That was so even though Lord Goff disavowed the declaratory theory of judicial decision-making, with the consequence that the previous understanding might be regarded as having been correct as the law stood at the time of the payment: a situation which Lord Hoffmann described as a “deemed” mistake. It is unnecessary for present purposes to consider the merits of that reasoning. It was disputed by Lord Browne-Wilkinson and Lord Lloyd of Berwick, and has been criticised by a number of academic commentators (and approved by others), but is not challenged in these proceedings.

On that basis, Lord Goff and the other members of the majority rejected the argument that cases where the court departed from a previous decision, or from a settled practice, should be distinguished from cases where the court determined the law for the first time. In each of those events, the court’s decision had a retrospective effect: that was “an inevitable attribute of judicial decision-making” (p 379). In each event, the effect of the court’s decision was to falsify the belief or assumption which had caused the claimant to make the payment, and that was sufficient to create a restitutionary claim based on mistake.

The next step in the bank’s argument was to establish that the cause of action for the recovery of money paid under a mistake of law fell within the scope of section 32(1)(c) of the 1980 Act. In that regard, counsel for the bank relied on the obiter dicta of the Court of Appeal in In re Diplock, discussed at paras 136-138 above, and argued that, even if such a claim would not have been recognised at the time when the provision was enacted, it should be construed in accordance with the “always speaking” principle of statutory interpretation (referring to R v Ireland [1998] AC 147). On that basis, counsel argued that section 32(1)(c) extended to mistakes of law once the law recognised such mistakes as giving rise to a right of action. In response, counsel for the local authorities argued that section 32(1) should be construed as at the moment of its enactment, when it could only have applied to mistakes of fact. Furthermore, the language of section 32(1) was not apt to apply to mistakes of law, since the law could rarely be said to be objectively ascertainable, so as to be capable of being “discovered” with reasonable diligence.

The majority of the Appellate Committee decided this issue in favour of the bank, and the minority concurred, on the hypothesis (contrary to their opinion) that there was an actionable mistake. The reasons the majority gave for reaching that conclusion were brief and rested principally on what appears to us, with respect, to have been an inaccurate understanding of the pre-1939 law. The statutory concept of “discoverability” was not discussed. The proceedings had not reached the stage at which it was necessary to determine when the mistake of law was discovered, or could with reasonable diligence have been discovered, and only the two judges in the minority considered the question.

Lord Goff did not refer to the bank’s argument based on the “always speaking” principle, but briefly addressed the local authorities’ argument concerning the language of section 32(1), stating at pp 388-389:

The first of the authorities which Lord Goff cited, Earl Beauchamp v Winn (1873) LR 6 HL 223, was a similar case to Cooper v Phibbs, mentioned in para 109 above. It concerned a bill seeking the equitable rescission of a contract for the exchange of property, on the ground of common mistake as to the parties’ respective rights to the properties in question. The principal issues were whether there had been a common mistake, and if so, whether relief was barred either by the impossibility of restitutio in integrum or on the ground that the appellant could readily have discovered the true position before entering into the agreement, having the relevant title deeds in his possession but having failed to read them.

The passage in the speech of Lord Chelmsford which Lord Goff cited was concerned with three matters. The first, as Lord Chelmsford put it at p 233, was the principle “that where a party is put upon inquiry, and by reasonable diligence he might have obtained knowledge of a fact of which he remained in ignorance, Equity would not relieve him”. The second, as it was put at p 234, was “the objection, that the mistake (if any) was one of law, and that the rule ‘Ignorantia juris neminem excusat’ applies”. In that regard, Lord Chelmsford followed Cooper v Phibbs in distinguishing between “ignorance of a well-known rule of law” and ignorance of the true construction of a deed. The third issue was the equitable doctrine of acquiescence. In the event, after these objections had been considered and rejected, there was held to have been no mistake. There was no discussion of the statutes of limitation, or of the equitable rule mentioned by the Law Reform Committee, or of the question whether it might have any application to mistakes of law.

Lord Goff also cited the obiter dictum of the Court of Appeal in In re Diplock. As was explained in para 136 above, that dictum proceeded on the hypothesis that personal claims against the wrongful recipient of property during the administration of an estate fell within the scope of section 2(7) of the 1939 Act rather than section 20. The Court of Appeal had already rejected that hypothesis, and the House of Lords also rejected it, on appeal, in Ministry of Health v Simpson, as explained at para 138 above. Furthermore, since the right of action with which In re Diplock was concerned was not based on a mistake, as explained in paras 116 and 137-138 above, it followed from the decision in Phillips-Higgins v Harper [1954] 1 QB 411, later endorsed in FII (SC) 1 [2012] 2 AC 33, that it could not fall within the ambit of section 26(c) of the 1939 Act, or section 32(1)(c) of the 1980 Act: see paras 41 and 140 above.

Lord Goff did not discuss the local authorities’ argument that the law could rarely be said to be objectively ascertainable, so as to be capable of being discovered with reasonable diligence. As the decision in Kleinwort v Benson itself illustrates, points of law present a problem for a test of “discoverability”, if discovery requires the ascertainment of the truth. On the assumption that it did, the local authorities argued in Kleinwort Benson that the test of discoverability could not be applied to mistakes of law, and that they therefore fell outside the scope of section 32(1). As will appear, the House of Lords, proceeding on the same assumption, decided in Deutsche Morgan Grenfell that the truth could not be discovered until it had been established by an authoritative judicial decision, and that time could not therefore begin to run under section 32(1) until such a decision had been taken. It will be necessary to consider at a later point whether the underlying assumption, that the test of discoverability requires the ascertainment of the truth, is well-founded.

Before summarising his conclusions, Lord Goff stated at p 389:

With great respect to an eminent judge, that statement suggests that some important matters were insufficiently considered. The fundamental purpose of limitation statutes is to set a time limit for the bringing of claims. As the Law Reform Committee stated at paragraph 7 of its Report, “the purpose of the statutes [of limitation] goes further than the prevention of dilatoriness; they aim at putting a certain end to litigation and at preventing the resurrection of old claims, whether there has been delay or not”. Lord Goff’s statement accepts that the result of the majority’s decision as to the effect of section 32(1)(c) is “that the cause of action in a case such as the present may be extended for an indefinite period of time”. That is also a possibility in the case of mistakes of fact, but it may be argued that the risk is potentially higher, and the consequences potentially more serious, in the case of a mistake of law arising retrospectively as a result of a judicial decision. Lord Goff’s statement that “this consequence may not have been fully appreciated at the time when this provision was enacted” lays the responsibility at Parliament’s door. But the question which the Appellate Committee should itself have considered was whether the result of its decision would be consistent with Parliament’s intention in enacting the 1980 Act. It is the duty of the court, in accordance with ordinary principles of statutory construction, to favour an interpretation of legislation which gives effect to its purpose rather than defeating it. Lord Goff did not, however, undertake any analysis of section 32(1), and made no attempt to give it a purposive interpretation. It will be necessary to return to this issue, after section 32(1) has been examined in the light of the decision in Deutsche Morgan Grenfell [2007] 1 AC 558.

Turning to the other majority judgments, Lord Hoffmann, like Lord Goff, rejected the possibility of distinguishing in the law of restitution between cases where a judicial decision changed a settled view of the law, or settled what was previously an unsettled view, on the one hand, and cases where the mistake of law lacked any retrospective element, on the other hand. In Lord Hoffmann’s view, there was no basis in principle for drawing such a distinction.

In relation to limitation, Lord Hoffmann stated at p 401:

Lord Hope of Craighead also rejected the possibility of distinguishing between different kinds of mistake of law for the purposes of the law of restitution, because of the difficulty of establishing a clear and principled approach. He identified a number of situations in which there might be said to be a mistake of law. The mistake might be caused by a failure to take advice, by omitting to examine the available information, or by misunderstanding the information which had been obtained. Or it might be due to a failure to predict correctly how the court would determine issues which were unresolved at the time of the payment, or to foresee that there was an issue which would have to be resolved by the court. Within the latter categories, there might be cases where the court overturned an established line of authority, and cases where there was no previous decision on the point. He concluded, at p 411, that it was preferable to avoid being drawn into a discussion as to whether a particular decision changed the law or was merely declaratory, since “[i]t would not be possible to lay down any hard and fast rules on this point”.

In relation to limitation, Lord Hope observed at p 417 that the word “mistake” appeared in section 32(1) without qualification, and that there was nothing in the words used which restricted the application of the subsection to mistakes of fact. More questionably, he added that the origin of the section, in paragraph 23 of the Report of the Law Revision Committee, suggested that the absence of restriction was intentional. No other member of the Appellate Committee supported that reading of the Report, and we can find no indication of such an intention in paragraph 23 or elsewhere: see in particular para 126 above.

Lord Hope also noted that in In re Diplock the Court of Appeal had said that section 26(c) of the 1939 Act would operate to postpone the running of time in the case of an action to recover money paid under a mistake of fact. He continued, at p 417:

The authorities cited in that passage might be regarded as illustrating the fine distinctions sometimes drawn between mistakes of fact and of law, but they did not dissolve the distinction. They were not, in particular, concerned with claims for the recovery of money on the basis that it had been paid under a mistake of law. Nor were they concerned with limitation. For the reasons explained at paras 150-151 above, Earl Beauchamp v Winn does not in our opinion offer any guidance in relation to the application of section 32(1)(c) of the 1980 Act to claims of the kind with which Kleinwort Benson was concerned. Cases concerned with the recovery of payments made under an error of law to a trustee in bankruptcy as an officer of the court, such as Ex p James, also appear to us to have no bearing on the point. As was explained in paras 111-112 above, claims to recovery in cases of that kind were not based on mistake, and did not question that both the legal and the equitable title had passed. The case of R v Tower Hamlets London Borough Council, Ex p Chetnik Developments Ltd [1988] AC 858 also appears to us to offer no assistance, except in explaining the principle underlying the line of authority including Ex p James. It was a case in public law, concerned with the exercise of a statutory discretion to repay rates which had been paid in the absence of any liability to pay.

In relation to the risk that the decision of the majority would result in serious problems, Lord Hope stated at p 417:

By contrast, Lord Browne-Wilkinson considered that, if the law recognised claims for the recovery of money paid under a mistake of law, including claims arising retrospectively as the result of a judicial decision, then the disruption of legal certainty resulting from the application of section 32(1)(c) would be so great that the Appellate Committee ought not to develop the law so as to recognise such claims. He observed at p 364 that “[o]n every occasion in which a higher court changed the law by judicial decision, all those who had made payments on the basis that the old law was correct (however long ago such payments were made) would have six years in which to bring a claim to recover money paid under a mistake of law”. Since all the members of the Appellate Committee accepted that this position could not be cured save by primary legislation altering the relevant limitation period, he concluded that “the correct course would be for the House to indicate that an alteration in the law is desirable but leave it to the Law Commission and Parliament to produce a satisfactory statutory change in the law which, at one and the same time, both introduces the new cause of action and also properly regulates the limitation period applicable to it”. Similar views were expressed by Lord Lloyd of Berwick (p 398).

The decision in Kleinwort Benson in relation to section 32(1) does not stand or fall on the persuasiveness of the speeches. It will be necessary to return at a later point in this judgment to the question as to whether, on a proper understanding of section 32(1), the decision was correct. First, however, it is necessary to consider the construction of section 32(1), which was one of the matters examined in Deutsche Morgan Grenfell.

The case of Deutsche Morgan Grenfell concerned legislation under which, where a company paid a dividend, it was liable to pay ACT, calculated as a proportion of the dividend, which could later be set off against its liability to pay mainstream corporation tax on its profits. The Revenue thereby obtained early payment of the tax and, in cases where the ACT exceeded the mainstream corporation tax, the payment of tax which would not otherwise have been due. Where, however, the dividend was paid to a parent company, and both the company paying the dividend and its parent were resident in the UK, a group income election could be made. The result of such an election was that the subsidiary did not pay ACT, but instead paid the appropriate amount of mainstream corporation tax when it became due. Deutsche Morgan Grenfell (“DMG”) was a UK subsidiary of a German parent and was therefore unable to make an election. As a result, it paid tax, in the form of ACT, earlier than it would have done if an election had been possible. In Hoechst [2001] Ch 620, the Court of Justice held that the legislation was incompatible with EU law in so far as it denied to the subsidiaries of non-UK resident parents the ability to make a group income election. That decision endorsed the opinion of the Advocate General, promulgated six months earlier.

A month after the Advocate General’s opinion was promulgated, and five months before the decision of the Court of Justice, DMG began proceedings to recover “compensation” for its early payment of the tax. Its claim was based on the proposition that it paid the tax when it did under a mistake of law, and was therefore entitled to restitution in accordance with the principle established in Kleinwort Benson: a principle which, it argued, applied to payments of tax as it did to other payments, notwithstanding the availability of a right to recover undue tax under the Woolwich principle. On the other hand, the Revenue argued that the reasoning in Kleinwort Benson did not apply to payments of tax, and that the only common law cause of action to recover tax was that based on the decision in Woolwich [1993] AC 70. It is unnecessary for us to consider the Appellate Committee’s decision on those questions, which is not in issue in the present appeal. DMG also argued that the mistake was not discoverable until the decision in Hoechst (although it had begun its action before then), and that section 32(1) of the 1980 Act postponed the commencement of the limitation period until then. In reply, the Revenue argued that the mistake was discovered when DMG learned in 1995, six years before the decision in Hoechst, that the relevant provisions were the subject of serious legal challenge in the Hoechst proceedings and might not be lawful.

DMG’s arguments on that question were accepted by the House of Lords ([2007] 1 AC 558), by a majority of three to two. In considering the application of section 32(1)(c), Lord Hoffmann stated at para 31 that the “reasonable diligence” proviso depended upon the true state of affairs being there to be discovered:

This statement is based on a number of premises. One is that a mistake of law is a “mistake” within the meaning of section 32(1)(c), as had been held in Kleinwort Benson, and therefore falls within the ambit of the discoverability test. It will be necessary to return to that point. Lord Hoffmann’s statement also assumes that “discovery”, within the meaning of section 32(1), means the ascertainment of the truth, and that, as a consequence of the abandonment of the declaratory theory, judicial decisions which establish a point of law thereby bring the truth into existence for the first time. It will be necessary to examine those assumptions in the context of the dissenting speech of Lord Brown of Eaton-under-Heywood.

Lord Hope emphasised at para 71 that DMG’s claim was disputed by the Revenue until the matter was finally decided in DMG’s favour by the Court of Justice:

Lord Walker concurred, stating at para 144 that it was the judgment of the Court of Justice in Hoechst “that first turned recognition of the possibility of a mistake into knowledge that there had indeed been a mistake”. Like Lord Hope, he emphasised that, until that judgment, the Revenue denied that DMG had a cause of action:

Lord Brown dissented, on the view that DMG discovered the mistake, within the meaning of section 32, when it first became aware of the Hoechst proceedings. It will be necessary to return to Lord Brown’s speech. Lord Scott of Foscote also dissented, on the view that DMG’s cause of action properly lay in tort, and therefore fell outside the ambit of section 32(1)(c) of the 1980 Act.

We shall begin our discussion of the two decisions placed in question in the present appeal by considering Deutsche Morgan Grenfell, on the hypothesis that the decision in Kleinwort Benson, that mistakes of law fall within the ambit of section 32(1)(c), was correct. We shall then consider Kleinwort Benson [1999] 2 AC 349. We approach the decisions in that order because it was only in Deutsche Morgan Grenfell [2007] 1 AC 558 that the Appellate Committee considered how section 32(1) operated in practice, in relation to “discoverability”, if mistakes of law fell within its scope. It is best to consider that issue, in the light of the contrasting views of the majority and of Lord Brown, before attempting to answer the question whether such mistakes do fall within the scope of the provision, purposively construed.

A paradox results from the approach adopted in Deutsche Morgan Grenfell, most clearly articulated by Lord Hoffmann: a claimant can be unable to discover the existence of his cause of action even after he has brought his claim: he cannot discover it until his claim succeeds. The paradox is well illustrated by the Court of Appeal’s decision in FII (CA) 2 [2017] STC 696, based on the application of Deutsche Morgan Grenfell. As was explained in para 54 above, the court held that the decision in Deutsche Morgan Grenfell established that “in the case of a point of law which is being actively disputed in current litigation the true position is only discoverable, for the purpose of section 32(1)(c) of the 1980 Act, when the point has been authoritatively determined by a final court”. On that basis, the court concluded that time began to run for the test claimants only on the date when judgment was delivered in FII (CJEU) 1, three and a half years after they had issued their claims. The paradox is particularly striking because the test claimants were successful before the Court of Justice. Its decision confirmed that they had been correct when they issued their claim form in 2003, asserting that they had paid tax under a mistake of law. It was the Revenue who were mistaken. That result illustrates the illogicality inherent in the reasoning in Deutsche Morgan Grenfell: the test claimants were able to identify correctly a mistake of law for the purpose of pleading a cause of action, while supposedly being unable to discover it for the purpose of the limitation period applicable to that cause of action.

That illogicality results from a specific difference between Lord Hoffmann’s approach to the accrual of a cause of action based on mistake, on the one hand, and his approach to the limitation period applicable to that cause of action, on the other hand. Where a payment has been made at time T1 on the basis of the law as it stood at that time, and the law is subsequently changed (as Lord Hoffmann would describe it) by a judicial decision taken at time T2, Lord Hoffmann says that the effect of the decision at T2 is that the law at T1 retrospectively becomes what it was decided to be at T2. The consequence is that the payment at T1 is retrospectively deemed to have been made under a mistake. A cause of action is therefore retrospectively deemed to have accrued at T1. However, when it comes to limitation, a different approach is adopted. The change in the law which is said to have been brought about by the decision at T2 is treated as occurring at T2, and therefore as being discoverable only at that time. Thus the mistake of law which, for the purpose of the accrual of a cause of action, is deemed to have occurred at T1, is simultaneously deemed not to have occurred at TI, but at T2, for the purpose of the law governing the discoverability of the mistake. It is because T2 occurs after the claim has been brought, and at the point when it is finally decided, that the paradox arises, that the mistake which forms the basis of the claim is not discoverable unless and until the claim succeeds. It is for the same reason that there arises the equally paradoxical result, that a limitation period applicable to the commencement of proceedings cannot begin to run until the proceedings have been completed. “Paradoxical” is indeed a generous term. One might say more candidly that this approach has consequences which are illogical and which frustrate the purpose of the legislation.

One possible response, arguably consistent with the abandonment of the declaratory theory, would be to argue that a “deemed” mistake is in reality no mistake at all. That is not, however, being argued in the present case. In any event, any attempt to draw a clear and principled distinction between “deemed” and actual mistakes faces real difficulties. As Lord Hope, in particular, indicated in his speech in Kleinwort Benson [1999] 2 AC 349, determining whether a particular decision changed the law or was merely declaratory would be a difficult exercise, not merely evidentially, but at a much deeper level. For example, when the House of Lords held in Murphy v Brentwood District Council [1991] 1 AC 398 that the case of Anns v Merton London Borough Council [1978] AC 728 had been “wrongly decided” (per Lord Keith of Kinkel at p 472), was the law changed, or was there a non-fictional sense in which the law at the time of Anns was other than the House of Lords had then declared it to be? Ultimately, the drawing of a line between “deemed” and actual mistakes, and even the question whether such a distinction can be drawn, depends on a theory of the nature of judicial decision-making, and indeed of the nature of law. The resultant scope for argument as to where the line should be drawn in any particular case would undermine one of the basic objectives of limitation statutes, namely to produce certainty as to the time limit for the bringing of a claim.

In any event, the issue raised by Lord Hoffmann’s reasoning is not confined to “deemed” mistakes, or conditional on his rejection of the declaratory theory. Judges cannot avoid having to decide at T2 what the law was at T1, and if their decision does not reflect how the law was understood by the claimant at T1, then it will ordinarily be uncontroversial to say that the claimant was mistaken at T1. The consequence, following the decision on the law of restitution in Kleinwort Benson, is that a cause of action accrued at T1 if a payment was made then on the basis of the mistaken understanding, regardless of the date of T2. On the limitation side of the analysis, on the other hand, the concept of discoverability is designed to protect claimants who could not reasonably be expected to know of the existence of the circumstances giving rise to their cause of action until sometime after it accrued. It must therefore be concerned with discoverability in reality, at a date which may be later than T1. It does not, however, follow that the “discoverability” of a mistake of law, within the meaning of section 32(1), must necessarily be tied to the date of a judicial decision, ie T2. The problems identified in para 174 above suggest that tying “discoverability” to the date of a judicial decision is a mistake. It will be necessary to return to that point in the context of Lord Brown’s dissenting speech in Deutsche Morgan Grenfell [2007] 1 AC 558.

That thought is reinforced by other considerations. Section 32(1) applies where the claimant does not know and cannot reasonably be expected to discover a mistake which forms an essential ingredient of his cause of action. Its effect is that the limitation period commences not on the date when the cause of action accrues, but on the date when the claimant discovers, or could with reasonable diligence discover, the mistake in question. The result of that postponement of the commencement date of the limitation period is to postpone the deadline for the bringing of a claim, so that the time during which the claimant was disadvantaged by the mistake does not count against him. Lord Hoffmann’s approach, whereby the limitation period does not begin until the truth has been established by a final judicial decision, does not merely extend the limitation period to the extent necessary to overcome the disadvantage arising from the mistake, but has the remarkable consequence of excusing the claimant from the necessity of bringing a claim until he can be certain that it will succeed: indeed, until it has in fact succeeded. This places the claimant in a case based on a mistake of law in a uniquely privileged position, since other claimants are required to bring their claims at a time when they have no such guarantee: the limitation period runs alike for claims which fail as for claims which succeed.

If the limitation period can begin to run at a time when a claim is uncertain of success, then, in addition to the logical problem discussed earlier, there is also a lack of realism in treating the date of a judicial decision authoritatively establishing the true state of the law as the earliest date when the claimant discovers, or could with reasonable diligence discover, the mistake in question. In the first place, the courts do not act on their own initiative, but only when their jurisdiction is invoked: normally, by the issuing of a claim. A point of law could often have been decided earlier, if a claim had been brought at an earlier time. Secondly, thinking about the law evolves over time. Developments in judicial thinking, in particular, do not take place in a vacuum. Judgments are the culmination of an evolution of opinion within a wider legal community, to which practitioners, universities, legal journals and the judiciary all contribute. And it is not only judges who are influenced by that evolving body of opinion. Claimants and their advisers respond to the same developments in their understanding of the state of the law, and their decisions as to whether or not worthwhile claims may exist. It is therefore possible to investigate how legal thinking on a particular question (for example, in the present case, whether the UK tax treatment of dividends received by UK-resident companies from non-resident subsidiaries was compatible with EU law) developed over time, and to ascertain, by means of evidence, the time by which a reasonably diligent person in the position of the claimant (such as, in the present case, a UK-based multi-national company) could have known of a previous mistake of law, to the extent of knowing that there was a real possibility that such a mistake had been made, and that a worthwhile claim could therefore be made on that basis. This line of thought suggests that the focus of attention under section 32(1) of the 1980 Act should not be on judicial decisions, but on the claimant’s ability to discover that he had a worthwhile claim.

Finally, in relation to Lord Hoffmann’s reasoning, it is also, with great respect, susceptible to the criticism that it pays insufficient regard to the principle of statutory construction that legislation should be given a purposive interpretation. If section 32(1) is interpreted in accordance with Kleinwort Benson as applying to mistakes of law, and if those mistakes of law are not considered to be “discoverable” within the meaning of the provision until after a final judgment has been delivered, as was held in Deutsche Morgan Grenfell [2007] 1 AC 558, then the object of the limitation statute is defeated. That object is to set a time limit for the bringing of claims. That object is frustrated if the limitation period does not begin to run until the proceedings have been completed. It is true that the limitation period so set will not be completely pointless in a situation where other people have identical claims which are not being pursued in the same proceedings, since time will begin to run for the bringing of those other claims. But in more usual situations, where an individual claim is brought, or where multiple claims are brought together in a group litigation (as in Deutsche Morgan Grenfell itself, which was a test case in the ACT Group Litigation), this approach to limitation defeats Parliament’s purpose in enacting limitation periods. It is therefore a result which Parliament cannot have intended when it enacted the 1980 Act.

As we have explained, Lord Brown dissented in Deutsche Morgan Grenfell [2007] 1 AC 558 on the view that DMG discovered the mistake, within the meaning of section 32, when it first became aware of the Hoechst proceedings and recognised that there was a serious challenge to the legality of the ACT regime under EU law. He stated at para 165:

Lord Brown noted that DMG had continued to make payments of ACT after July 1995, when they learned that Hoechst had issued proceedings, and that they had issued their own claim five months prior to the decision of the Court of Justice in Hoechst [2001] Ch 620. Referring to Lord Hope’s statement that, when DMG paid the ACT, “it was not then obvious that the payments might not be due”, Lord Brown commented at para 172 that he had some difficulty with that conclusion:

In support of his views, Lord Brown pointed first, at para 167, to the parallel treatment in section 32 of fraud, deliberate concealment and mistake:

The view expressed in that passage is supported by a number of authorities concerned with the application of section 32(1) in cases of fraud. The first which might be mentioned is the judgment of Arden LJ, with which Aldous and Robert Walker LJJ agreed, in Biggs v Sottnicks [2002] EWCA Civ 272. In deciding when the appellants “could with reasonable diligence have discovered” a fraud, for the purposes of section 32(1) of the 1980 Act, her Ladyship treated the relevant date as “the correct date when the appellants’ solicitors had sufficient information in their hands for the purposes of this deceit claim” (para 62), that is to say, the date when “the appellants were in a position to plead their own case” (para 64). A similar approach was adopted in Law Society v Sephton & Co [2004] EWHC 544 (Ch); [2004] PNLR 27, para 44, where the court proceeded on the basis of the parties’ agreement that a claimant did not discover a fraud until he had material sufficient to enable him properly to plead it.

Reference should also be made to the judgment of Lord Hoffmann NPJ, with which the other members of the Hong Kong Court of Final Appeal agreed, in Peconic Industrial Development Ltd v Lau Kwok Fai [2009] HKCFA 16; [2009] WTLR 999. The case raised the question, under a legislative provision in the same terms as section 32(1) of the 1980 Act, whether the claimants could with reasonable diligence have discovered a fraud committed more than six years before proceedings were issued. Lord Hoffmann stated at para 56:

The approach adopted in those cases differs from that proposed by Lord Brown only in its focus on the date when the claimant (or his lawyers) had sufficient material properly to plead a claim in fraud. Lord Brown put the matter differently in paras 165 and 167 of his judgment in Deutsche Morgan Grenfell [2007] 1 AC 558, when he treated the mistake as being discovered as soon as the claimant recognises that “a worthwhile claim arises”, or that “he has a worthwhile case … to pursue a claim”. It will be necessary to return to this point. As will be explained, Lord Brown’s approach is consistent with that adopted authoritatively in analogous contexts where fraud was not in issue, and is also in accordance with principle.

What is more important for present purposes, however, is that the approach adopted in these cases of fraud, like that proposed by Lord Brown for cases of mistake, treats the relevant date, for the purposes of the commencement of the limitation period, not as the date when the claimant knows or can establish the truth, but as the date when he can recognise that a worthwhile claim arises, in Lord Brown’s formulation, or can plead a statement of claim, in the formulation preferred in the fraud cases.

Lord Brown also found support for his position in Deutsche Morgan Grenfell in authorities concerned with the interpretation of other provisions of the 1980 Act which postpone the commencement of the limitation period until the claimant knows or could reasonably have known the facts forming the basis of his cause of action. That approach is applied, for example, to actions for damages in respect of torts causing personal injuries, by section 11 of the 1980 Act. Under section 11(4), read together with section 14(1), the limitation period generally runs from the date on which the cause of action accrued, or, if later, the date on which the person injured had knowledge that the injury was significant and was attributable to the act or omission relied on, and knowledge of the identity of the defendant. For these purposes, knowledge is defined as including “knowledge which he might reasonably have been expected to acquire” (section 14(3)). The language of these provisions differs from section 32(1) in that they refer to having knowledge, rather than discovering. But that is on its face an insubstantial difference, since discovery ordinarily refers to the acquisition of knowledge. And sections 11, 14 and 32 have the same rationale, namely that the limitation period should only run from the time when the claimant knows or could reasonably have known of the existence of his cause of action. Sections 11 and 14 are explicitly concerned with knowledge of the facts forming the cause of action, and not with their legal consequences. But the same is true of section 32(1), even in its application to mistakes of law. As is explained below, the relevant fact that has to be discovered, in that context, is the fact that the claimant made a mistake, that being an essential ingredient of his cause of action. A claimant’s ignorance of the legal consequences of the facts forming his cause of action is not something with which section 32(1) is concerned, as Lord Walker made clear in FII (SC) 1 [2012] 2 AC 33, para 63 (para 41 above). That is consistent with the intention of the Law Revision Committee, as was explained at para 127 above.

Sections 11 and 14 were considered by this court in AB v Ministry of Defence [2012] UKSC 9; [2013] 1 AC 78, where proceedings were begun by the claimants at a time when they believed that their injuries had been caused by their exposure to radiation by the defendant, but had no objective basis for their belief. Their contention that they did not then have knowledge of the facts forming the basis of their cause of action was rejected. The court held, by a majority, that knowledge did not mean knowing for certain and beyond possibility of contradiction, but that mere suspicion was not enough; that in order to amount to knowledge a belief had to be held with sufficient confidence to justify embarking on the preliminaries to issuing proceedings; and that it was, therefore, a legal impossibility for a claimant to lack knowledge for the purposes of section 14(1) at a time after he had issued his claim.

In relation to the last of those points, Lord Wilson, Lord Walker, Lord Brown and Lord Mance all made it clear that, in deciding whether a claim was statute-barred, the court had to assume that, when the claimant issued his claim, he had knowledge of the facts necessary to support his pleaded cause of action. Lord Wilson stated at para 6 that it was “heretical” that a claimant could escape the requirement to assert his cause of action for personal injuries within three years of its accrual “by establishing that, even after his claim was brought, he remained in a state of ignorance entirely inconsistent with it”. Lord Walker said at para 67 that he did not see how a claimant who had issued a claim form could be heard to suggest that he did not, when it was issued, have the requisite knowledge for the purposes of the 1980 Act. Lord Brown said at para 71 that “once a claimant issues his claim, it is no longer open to him to say that he still lacks the knowledge necessary … to set time running”. Lord Mance agreed, observing at para 84 that a claimant bringing proceedings necessarily asserts that he or she has a properly arguable claim.

Considering more precisely the point in time at which a claimant acquires “knowledge” for the purposes of sections 11(4) and 14(1) of the 1980 Act, the majority of the court in AB v Ministry of Defence endorsed the test earlier approved by the House of Lords in relation to claims falling under section 14A (inserted by the Latent Damage Act 1977), which applies to actions for damages for negligence, other than those involving personal injuries. In Haward v Fawcetts [2006] UKHL 9; [2006] 1 WLR 682, para 9, Lord Nicholls of Birkenhead stated:

The formulation adopted in Halford v Brookes [1991] 1 WLR 428, Haward v Fawcetts and AB v Ministry of Defence places the commencement of the limitation period slightly earlier than the fraud cases discussed earlier. The relevant time is when the claimant knows “with sufficient confidence to justify embarking on the preliminaries to the issue of a writ”, rather than the point in time when he could plead a statement of claim. This is not the occasion on which to review the formulation used in the fraud cases, which reflects the special standards applicable to the pleading of fraud. The formulation used in Halford v Brookes, Haward v Fawcetts and AB v Ministry of Defence is, however, consistent with the way in which the point was expressed by Lord Brown in Deutsche Morgan Grenfell (para 180 above) and by Lord Walker in FII (SC) 1 [2012] 2 AC 33 (para 48 above).

It is also consistent with principle. The limitation period normally begins to run on the date when the cause of action accrues. It is not postponed until the claimant has consulted a solicitor, carried out investigations, and is in a position to plead a statement of claim. For example, a pedestrian who is knocked down and injured by a car while using a zebra crossing has a cause of action against the driver, which accrues on the date of the accident. It will take time before he can issue a claim: he will need to consult solicitors, and counsel may have to be instructed to draft the claim. There may be many matters which have to be investigated, and that may take time. And it may be that his claim will fail in the end, if, for example, it is found that he suddenly ran into the path of the car, or that the driver had a heart attack and lost control of the vehicle. Nevertheless, the limitation period begins to run on the date of the accident. It is not postponed until he has completed his investigations, or until he knows that his claim is guaranteed to succeed.

The purpose of the postponement effected by section 32(1) is to ensure that a claimant is not disadvantaged, so far as limitation is concerned, by reason of being unaware of the circumstances giving rise to his cause of action as a result of fraud, concealment or mistake. That purpose is achieved, where the ingredients of the cause of action include his having made a mistake of law, if time runs from the point in time when he knows, or could with reasonable diligence know, that he made such a mistake “with sufficient confidence to justify embarking on the preliminaries to the issue of a writ, such as submitting a claim to the proposed defendant, taking advice and collecting evidence”; or, as Lord Brown put it in Deutsche Morgan Grenfell [2007] 1 AC 558, he discovers or could with reasonable diligence discover his mistake in the sense of recognising that a worthwhile claim arises. We do not believe that there is any difference of substance between these formulations, each of which is helpful and casts light on the other.

It is true that Haward v Fawcetts [2006] 1 WLR 682 and AB v Ministry of Defence [2013] 1 AC 78 were not concerned with section 32, but with other provisions of the 1980 Act, expressed in different language: sections 14(3) and 14A(10) are concerned with “knowledge which [the claimant] might reasonably have been expected to acquire”, whereas section 32(1) is concerned with what he “could with reasonable diligence have discovered”. It is also true that sections 14 and 14A explicitly provide that knowledge “that any acts or omissions did or did not, as a matter of law, involve negligence”, is irrelevant. They are, however, concerned with the same problem as section 32(1), namely that a cause of action can accrue before the claimant comes to know of it, and they address that problem in a similar way, by postponing the commencement of the limitation period until the claimant knew, actually or constructively, the facts on which the cause of action is based. The close connection between sections 11, 14, 14A and 32 of the 1980 Act was made clear by Lord Walker’s reasoning in FII (SC) 1, para 63 (para 41 above).

In those circumstances, it appears to us to be impossible to reconcile the reasoning in Haward v Fawcetts and AB v Ministry of Defence with that in Deutsche Morgan Grenfell and the cases which have followed it. The former line of authority proceeds on the basis that the commencement of the limitation period is postponed until the claimant knows, actually or constructively, the essential facts on which the cause of action is based, “with sufficient confidence to justify embarking on the preliminaries to the issue of a writ, such as submitting a claim to the proposed defendant, taking advice and collecting evidence”. The dissenting judgment of Lord Brown in Deutsche Morgan Grenfell is consistent with that approach: time does not begin to run until the claimant knows, actually or constructively, that he made a mistake (that being an essential ingredient of the cause of action), to the standard that “a worthwhile claim arises”. The latter line of authority, on the other hand, proceeds on the basis that the limitation period does not run until a court has authoritatively established that the claimant’s assertion that he made a mistake of law is true. Mistakes of law are thus treated differently from mistakes of fact, and the difficult and much criticised distinction between the two remains of crucial importance.

Furthermore, only the former line of authority is consistent with the rationale of limitation periods. It is in the nature of litigation that facts and law are commonly disputed. It is the function of courts to resolve those disputes. Until the court has done so, the parties can, at best, have only a reasonable belief that their assertions are correct. If a limitation period is to serve its purpose, in fixing a time within which claims must be brought, it can therefore only be concerned with beliefs, and not with the truth established by judicial decisions, whether in the proceedings in question, or in other proceedings. That is reflected in Lord Donaldson’s statement in Halford v Brookes [1991] 1 WLR 428, endorsed by Lord Nicholls in Haward v Fawcetts [2006] 1 WLR 682 (para 190 above) and by Lord Wilson in AB v Ministry of Defence [2013] 1 AC 78, para 11, that “reasonable belief will normally suffice”.

Returning to Deutsche Morgan Grenfell [2007] 1 AC 558, Lord Brown found further support for his argument in an authority concerned with the meaning of “discover” in the context of tax legislation. The Income Tax Act 1918 (and later tax statutes) contained a provision enabling additional assessments to be issued where it was “discovered” that profits chargeable to tax had been omitted from an initial assessment. In Earl Beatty v Inland Revenue Comrs [1953] 1 WLR 1090, the assessments under appeal were made under that provision, at a time when the Commissioners had a strong suspicion that there had been an undeclared transfer of assets by the appellant or his wife. It subsequently transpired that there had indeed been undeclared transfers, not by the appellant or his wife, but by his brother acting on his behalf. The assessments were challenged on the ground that they were not based on a “discovery” within the meaning of the legislation, since a suspicion, especially if inaccurate, did not amount to a discovery. The argument was rejected, the judge observing at p 1095:

Like a claim form, an assessment is not a statement of established verities. It is a formal statement of a claim made by the Commissioners and forms the basis of an inquiry into the facts in the event that it is challenged. In those circumstances, the test of “discovery” could not sensibly require that the truth had already been established. The same is true in the present context.

The approach adopted in the fraud cases discussed in paras 180-186 above, and in the cases concerned with analogous provisions of the 1980 Act, discussed in paras 187-196 above, is consistent with the nature of a plea of limitation: it is legally distinct from the merits of the claim in question, and is often conveniently dealt with as a preliminary issue. The 1980 Act proceeds on the basis that a cause of action has accrued, without concerning itself with the question whether or not the action is well-founded. Section 32(1)(a) applies where “the action is based upon the fraud of the defendant”, and section 32(1)(c) applies where “the action is for relief from the consequences of a mistake”. If the action runs its full course, it may transpire that there was no fraud or mistake, indeed no cause of action at all. But where, at the stage of an inquiry into the defendant’s plea that the action is time-barred, the claimant relies on section 32(1)(a) or (c), the question is not whether there was in reality any fraud or mistake: that will not be established unless and until the court issues a judgment on the merits of the case. The question under section 32(1)(a) and (c) of the 1980 Act is whether, upon the assumption that there was fraud or mistake, as identified by the claimant in the way in which he pleads his case, it was discovered or could with reasonable diligence have been discovered at such a time as would render the claim time-barred.

One might compare the approach adopted to the issue of laches in Earl Beauchamp v Winn (1873) LR 6 HL 223, where Lord Chelmsford stated at p 233 that “in considering this part of the case it has been assumed, for the purpose of the argument, that the late Earl was under a mistake as to his interest … Mr Winn, upon this assumption, was also under a mistake … The case must be dealt with, therefore, as one of mutual mistake.” Once the issue of laches had been disposed of on that basis, the House of Lords went on to hold that there had in fact been no mistake.

Hence the situation which may seem paradoxical, but sometimes arises in practice (as, for example, in Law Society v Sephton & Co [2004] EWHC 544 (Ch); [2004] PNLR 27), where in a trial on limitation the defendant disputes the claimant’s assertion that he could not have known or discovered a fact which, in relation to the merits of the claim, the defendant denies is a fact at all. There is in reality no paradox, because at the stage of an inquiry into limitation the existence of the cause of action, and therefore the truth of the facts relied on by the claimant to establish it, is not the relevant issue. Put in general terms, the question is not whether the claimant could have established his cause of action more than six years (or whatever other limitation period might be relevant) before he issued his claim, but whether he could have commenced proceedings more than six years before he issued his claim. The existence of the constituents of the cause of action - such as fraud or mistake - as verified facts is not the issue.

That point emerges clearly from the majority judgments in AB v Ministry of Defence [2013] 1 AC 78. Lord Wilson, for example, stated at para 2, in relation to section 11(4) of the 1980 Act:

That approach is also consistent with the well-established test for determining whether, for the purposes of section 32(1), the claimant “could with reasonable diligence” have discovered a fraud. Authoritative guidance on that topic was given by Millett LJ in Paragon Finance plc v DB Thakerar & Co [1999] 1 All ER 400, 418:

The foregoing approach is also supported by the pre-1939 principle of equity on which section 26 of the 1939 Act and section 32(1) of the 1980 Act were modelled. In that regard, the decision of the Court of Appeal in Molloy v Mutual Reserve Life Insurance Co (1906) 94 LT 756 is particularly helpful. The plaintiff took out a life assurance policy after being told by the insurer’s agent that, under the policy, the premiums would remain at a fixed rate. When the insurer later increased the premiums, the plaintiff brought proceedings in the County Court to recover the overpayments. The County Court held, however, that the insurer was entitled under the policy to charge the increased premiums. Several years later, another policy holder brought similar proceedings in the High Court, in which he succeeded. That decision was overturned on appeal, but the Court of Appeal, and ultimately the House of Lords, held that the contract should be rescinded, and the premiums returned, on the ground of fraudulent misrepresentation. The plaintiff (in the Molloy case) was by then out of time to bring a common law claim for the return of his premiums, but instead brought proceedings in equity for rescission, an account of the premiums paid (as a consequence of the setting aside of the contract), and payment of the amount found due on the account. Since the claim to an account was subject by analogy to the statutory limitation period, the plaintiff sought to rely on the equitable principle allowing for its extension in a case of fraud, and argued that he had been unable to discover that he had a cause of action prior to the decision of the House of Lords.

That argument was accepted by Swinfen Eady J, who considered that time did not begin to run while the plaintiff “waited … to be fully informed as to what his legal rights were, and [until] the position was definitely and finally ascertained”: (1906) 94 LT 756, 759. The Court of Appeal (Sir Richard Collins MR, Romer and Cozens-Hardy LJJ) disagreed.

The Master of the Rolls gave several reasons at p 761 for rejecting the argument. First, he pointed out that the plaintiff had known the facts which were essential to his cause of action long before the House of Lords gave its decision. The limitation period ran from the time when the plaintiff discovered the facts essential to his cause of action. It was immaterial that he did not understand their legal significance, or that it was only the decision of the House of Lords as to the construction of the policy that put that element of the cause of action beyond dispute:

That is equally true in a situation where one of the facts essential to the cause of action is that the claimant has made a mistake, whether of fact or of law. The fact that he has made a mistake needs to be discoverable (in the relevant sense) with reasonable diligence, but he does not need to know that he is consequently entitled to bring a claim. As the Law Revision Committee stated, “the mere fact that a plaintiff is ignorant of his rights is not to be a ground for the extension of time” (para 126 above). That is why, on the facts of Kleinwort Benson [1999] 2 AC 349, the relevant matter which needed to be discoverable was that the swaps contracts were ultra vires, as had been established in Hazell, and not that a cause of action lay for payments made under a mistake of law, as was established in Kleinwort Benson itself. For the same reason, Henderson J was in error in FII (HC) 2 [2015] STC 1471, in favouring the view (para 47 above) that it was only when the House of Lords gave judgment in Deutsche Morgan Grenfell [2007] 1 AC 558 that time began to run against the BAT claimants, since that was the first time an appellate court had held that a restitutionary claim lay for the recovery of tax on the ground that it had been paid under a mistake of law. The relevant fact was that the belief that the tax was payable had been mistaken; not that there was a right to restitution.

The second reason given by the Master of the Rolls for rejecting the plaintiff’s argument is also relevant to these proceedings:

It remains to consider whether the test of discoverability suggested at para 193 above, taken together with the standard of “reasonable diligence” discussed at para 203, provides an approach to the application of section 32(1) to mistakes of law which is likely to be reasonably practical and certain in its operation. To recap:

As was explained, the suggested test of discoverability is that a mistake of law is discoverable when the claimant knows, or could with reasonable diligence know, that he made such a mistake “with sufficient confidence to justify embarking on the preliminaries to the issue of a writ, such as submitting a claim to the proposed defendant, taking advice and collecting evidence”; or, as Lord Brown put it in Deutsche Morgan Grenfell, he discovers or could with reasonable diligence discover his mistake in the sense of recognising that a worthwhile claim arises. We do not believe that there is any difference of substance between these formulations, each of which is helpful and casts light on the other.

The standard of reasonable diligence is how a person carrying on a business of the relevant kind would act, on the assumption that he desired to know whether or not he had made a mistake, if he had adequate but not unlimited staff and resources and was motivated by a reasonable but not excessive sense of urgency. The question is not whether the claimant should have discovered the mistake sooner, but whether he could with reasonable diligence have done so. The burden of proof is on the claimant. He must establish on the balance of probabilities that he could not have discovered the mistake without exceptional measures which he could not reasonably have been expected to take.

In practice, the application of that approach will depend on the circumstances of the case. For example, in cases where the claimant has made a payment on the basis of a mistaken understanding of the law which has resulted from ignorance, the mistake will normally have been discoverable immediately, by seeking legal advice. Section 32(1) only has effect where a mistake could not have been discovered at the time of the payment with the exercise of reasonable diligence. On the other hand, where the payment was made in reliance on a precedent that was subsequently overruled, or an understanding of the law that was later altered by a judicial decision, the question will be whether the claim was brought within the prescribed period beginning on the date when it was discoverable by the exercise of reasonable diligence that the basis of the payment was legally questionable, so as to give rise to a worthwhile claim to restitution. Depending on the circumstances, it may be difficult to identify a specific date, but doubtful cases can be resolved by bearing in mind that the burden of proof lies on the claimant to prove that his claim was brought within the prescribed limitation period.

Clearly, where a payment was made in accordance with the law as it was then understood to be, the point in time at which the claimant could, with reasonable diligence, have discovered that the basis of the payment was legally questionable, so as to give rise to a worthwhile claim to restitution, will have to be established by evidence. The focus of that evidence is likely to be upon developments in legal understanding within the relevant category of claimants and their advisers, as explained in para 178 above. Thus, in the circumstances of the present case, Lord Walker referred in FII (SC) 1 [2012] 2 AC 33 (para 48 above) to there being a reasonable prospect that the limitation period could be deferred until the time when “a well advised multi-national group based in the UK would have had good grounds for supposing that it had a valid claim to recover ACT levied contrary to EU law”. This point is considered in greater detail in para 255 below. Evidence in relation to matters of this kind may well include expert evidence concerning the state of understanding of the law within the relevant categories of professional advisers during the relevant period.

It is true that this approach involves a more nuanced inquiry than a mechanical test based on the date on which an authoritative appellate judgment determined the point in issue. But it would be unduly pessimistic to conclude at this stage that it will prove to be unworkable in practice, or too uncertain in its operation to be acceptable.

Taking stock of the discussion so far, the position can be summarised as follows:

Limitation periods set a time limit for issuing a claim, which normally begins to run when the cause of action accrues. They apply whether the substance of the claim is disputed or not. They apply to claims regardless of whether there is in truth a well-founded cause of action.

Section 32(1) of the 1980 Act postpones the running of time beyond the date when the cause of action accrues, in cases where the claimant cannot reasonably be expected to know at that time the circumstances giving rise to the cause of action, by reason of fraud, concealment or mistake. Its effect is that the limitation period commences not on the date when the cause of action accrues, but on the date when the claimant discovers, or could with reasonable diligence discover, the fraud, concealment or mistake.

Consistently with (1) above, section 32(1) cannot be intended to postpone the commencement of the limitation period until the claimant discovers, or could discover, that his claim is certain to succeed.

Consistently with (1) above, section 32(1) cannot be intended to postpone the commencement of the limitation period until the proceedings have been completed.

In tying the date of “discoverability” of a mistake of law in section 32(1) to the date when “the truth” as to whether the claimant has a well-founded cause of action is established by a judicial decision, the decision in Deutsche Morgan Grenfell [2007] 1 AC 558 contravenes (3) above, and is therefore inconsistent also with (1) above.

In tying the date of discoverability to the date of a judicial decision, with the consequence that the limitation period for issuing a claim may not begin to run until the proceedings have been completed, the decision in Deutsche Morgan Grenfell also contravenes (4) above, and is for that reason also inconsistent with (1) above.

Tying the date of discoverability to the date of a decision by a court of final jurisdiction, as the House of Lords appear to have done in Deutsche Morgan Grenfell, and as the Court of Appeal held in FII (CA) 2, compounds the mistake (para 54 above).

In tying the date of the discoverability of a mistake of law to the date of a judicial decision which establishes that a mistake was made, the decision in Deutsche Morgan Grenfell also has the illogical consequence that mistakes are not discoverable by a claimant until after he has issued a claim on the basis of the mistake: (paras 173-174 above).

The decision in Deutsche Morgan Grenfell therefore frustrates Parliament’s intention in enacting section 32(1) (para 179 above).

The decision in Deutsche Morgan Grenfell is also inconsistent with authorities concerned with section 32(1) in relation to fraud (paras 180-186 above).

The decision in Deutsche Morgan Grenfell is also inconsistent with authorities at the highest level concerned with analogous provisions of the 1980 Act (paras 187-196 above).

The decision in Deutsche Morgan Grenfell is also inconsistent with the meaning given by the courts to “discovery” in another statutory context (paras 197-198 above).

The purpose of the postponement effected by section 32(1) is to ensure that the claimant is not disadvantaged, so far as limitation is concerned, by reason of being unaware of the circumstances giving rise to his cause of action as a result of fraud, concealment or mistake. That purpose is achieved, where the ingredients of the cause of action include his having made a mistake of law, if time runs from the point in time when he knows, or could with reasonable diligence know, that he made such a mistake “with sufficient confidence to justify embarking on the preliminaries to the issue of a writ, such as submitting a claim to the proposed defendant, taking advice and collecting evidence”; or, as Lord Brown put it in Deutsche Morgan Grenfell, he discovers or could with reasonable diligence discover his mistake in the sense of recognising that a worthwhile claim arises (paras 193 and 209).

By tying the concept of discovery to the ascertainment of the truth, the decision in Deutsche Morgan Grenfell contradicts the principle that limitation periods apply to claims regardless of whether they are ill- or well-founded. The claimant cannot be required to have ascertained the truth, in order for a limitation period to apply. Consistently with authorities concerned with analogous provisions of the 1980 Act, a reasonable belief will normally suffice (para 196).

Tying the concept of discovery to the ascertainment of the truth is also inconsistent with the nature of a plea of limitation. The question under section 32(1) is not whether there was in reality any fraud, concealment or mistake as the claimant has pleaded, but whether, upon the assumption that there was, it was discovered, or could with reasonable diligence have been discovered, at such a time as would render the proceedings time-barred. The existence of a mistake as a verified fact is not the issue (paras 199-202).

Authorities concerned with the meaning of “reasonable diligence” in section 32(1) also indicate that it is concerned with the steps which a person could reasonably be expected to take before issuing a claim (para 203 above). The standard of reasonable diligence is how a person carrying on a business of the relevant kind would act, on the assumption that he desired to know whether or not he had made a mistake, if he had adequate but not unlimited staff and resources and was motivated by a reasonable but not excessive sense of urgency. The question is not whether the claimant should have discovered the mistake sooner, but whether he could with reasonable diligence have done so. The burden of proof is on the claimant. He must establish on the balance of probabilities that he could not have discovered the mistake without exceptional measures which he could not reasonably have been expected to take (para 209).

Authorities concerned with the pre-1939 equitable rule on which section 32(1) is based also support the view that the limitation period runs from the time when the claimant discovers the facts essential to his cause of action, and not from the date of a judicial decision supportive of his claim (paras 204-208 above).

In adopting a different approach to the discoverability of mistakes of law from that which applies to mistakes of fact, the decision in Deutsche Morgan Grenfell perpetuates the problem of distinguishing between the two, contrary to the intended effect of the decision in Kleinwort Benson (para 195 above).

It follows, for all these reasons, that even if it is accepted that Kleinwort Benson was correctly decided, Deutsche Morgan Grenfell [2007] 1 AC 558, so far as it concerned limitation, was not.

We have not yet considered a more fundamental issue: the argument that an action for the recovery of money paid under a mistake of law, unlike an action for the recovery of money paid under a mistake of fact, is not an “action … for relief from the consequences of a mistake” within the meaning of section 32(1)(c), and therefore falls outside the ambit of the discoverability test. This argument challenges the correctness of the decision in Kleinwort Benson [1999] 2 AC 349, so far as it related to limitation.

As we have explained, at the time when section 26(c) of the 1939 Act was enacted, and equally at the time when section 32(1)(c) of the 1980 Act was enacted, the only recognised actions for which a period of limitation was prescribed, and which fitted the description of an “action … for relief from the consequences of a mistake”, were common law actions based on mistakes of fact, such as actions for the recovery of money paid under a mistake of fact, and analogous equitable claims also based on mistakes of fact. In our opinion, that is the effect of the pre-1939 authorities, notwithstanding the contrary views expressed in Kleinwort v Benson and discussed at paras 149-152 and 159-161 above. Although there were some recognised forms of equitable relief from the consequences of mistakes of law, such as rectification, they were not subject to statutory limitation either directly or by analogy prior to 1939; and that position was preserved by the 1939 and 1980 Acts: see paras 117-118, 123, 129, 131 and 142 above.

When the House of Lords recognised in Kleinwort Benson [1999] 2 AC 349 the existence of an action for the recovery of money paid under a mistake of law, it recognised another action which fitted the description of an “action … for relief from the consequences of a mistake”, if those words are construed according to their ordinary meaning. The question nevertheless arises whether that construction is in accordance with the purpose of the provision.

It is debatable, but ultimately does not matter, whether this question should be approached by focusing specifically on the “always speaking” principle, as counsel for the bank did in Kleinwort Benson. That somewhat vague expression is commonly used in connection with statutory terms which change in their connotations over time, such as “family” (Fitzpatrick v Sterling Housing Association Ltd [2001] 1 AC 27). The case of R v Ireland [1998] AC 147, cited by counsel in Kleinwort Benson, was of a similar kind. The question was whether the words “bodily harm”, in the Offences Against the Person Act 1861, should be interpreted in the light of contemporary knowledge as applying to psychiatric injury. The “always speaking” principle is also invoked where the question arises whether a statutory expression should be interpreted as including a novel invention or activity which does not naturally fall within its meaning, and was not envisaged at the time of its enactment, but which may nevertheless fall within the scope of its original intention. Examples of the latter kind of case include Victor Chandler International Ltd v Customs and Excise Comrs [2000] 1 WLR 1296, which concerned the question whether a teletext fell within the scope of the statutory term “document”, and R (Quintavalle) v Secretary of State for Health [2003] UKHL 13; [2003] 2 AC 687, which concerned the question whether an embryo created by the novel technique of cloning, rather than by the traditional method of fertilisation, fell within the scope of the statutory expression “embryo where fertilisation is complete”. Another well-known example is the case of Royal College of Nursing of the United Kingdom v Department of Health and Social Security [1981] AC 800, where the question was whether the statutory expression “a pregnancy … terminated by a registered medical practitioner” should be interpreted as including a novel technique of termination which was carried out by a nurse acting on the instructions of a medical practitioner.

The question in the present case is not of precisely the same kind. The cause of action recognised in Kleinwort Benson undoubtedly falls within the scope of the language used in section 32(1)(c), if that language is given its ordinary meaning. A mistake of law was understood to be a “mistake” in 1939, and in 1980, just as much as it is today. Nevertheless, the decision taken in Kleinwort Benson to recognise a cause of action for the recovery of money paid under a mistake of law could not have been foreseen in 1939 or 1980. The question therefore arises whether section 32(1)(c) applies to those unforeseen circumstances: a question which ultimately boils down to the same issue as arises when considering the “always speaking” principle, and indeed in all cases concerned with statutory interpretation: what is the construction of the provision which best gives effect to the policy of the statute as enacted?

A number of points can be made in support of a construction which would accommodate mistakes of law. First, and most importantly, the purpose of section 32(1)(c) is to postpone the commencement of the limitation period in respect of a claim for relief from the consequences of a mistake where, as a result of the mistake, the claimant could not reasonably have known of the circumstances giving rise to his cause of action at the time when it accrued. The effect of section 32(1)(c) is that the time when the claimant could not reasonably have known about those circumstances does not count towards the limitation period. Those were also the rationale and effect of the equitable rule applicable prior to 1939, and of the recommendation made in the Report of the Law Reform Committee. The equitable rule did not apply where the claimant had been aware of all the relevant circumstances at the time when his cause of action accrued and had merely been ignorant that those circumstances gave rise to a cause of action: see para 122 above. That aspect was also reflected in the Committee’s Report: see paras 126 and 127 above.

As we have explained, when section 32(1) was enacted, it could only have applied to claims in respect of mistakes of fact, since those were the only mistakes which gave rise to an “action … for relief from the consequences of a mistake”. However, the law subsequently developed in Kleinwort Benson so as to allow claims to be brought for relief from the consequences of mistakes of law. That development has to be addressed in the law of limitation in a way which is consistent with the legislative policy of the 1980 Act and avoids discord in the law, as Lord Hoffmann explained in Johnson v Unisys [2003] AC 518 (para 157 above). In principle, it is consistent with the purpose of section 32(1)(c) for it to apply to claims brought on that basis. The rationale of section 32(1) - to postpone the commencement of the limitation period in respect of a claim for relief from the consequences of a mistake where, as a result of the mistake, the claimant could not reasonably have known of the circumstances giving rise to his cause of action at the time when it accrued - applies with equal force to a mistake of law as to a mistake of fact. To construe the provision in a sense which excluded such claims would not be consistent with Parliament’s intention to relieve claimants from the necessity of complying with the time bar which would apply in the absence of section 32(1), at a time when they could not reasonably be expected to do so. Nor would such a construction reflect the ordinary meaning of the language which Parliament used: a mistake of law is, and always was, a “mistake” in the ordinary sense of the word.

For similar reasons, it would not be consistent with the intention of Parliament to exclude “deemed” mistakes from the ambit of section 32(1)(c), even assuming (contrary to the conclusion reached at paras 175-176 above) that a principled and workable distinction could be drawn between “deemed” and “actual” mistakes. There would, in the first place, be no warrant in the language of the provision for drawing such a distinction; and the court cannot effectively amend the legislation under the guise of interpretation. Furthermore, to draw such a distinction would undermine the purpose of the provision: a provision which, as explained earlier, has its origins in equity. The person who has made a “deemed” mistake is no less deserving of an extension of the time permitted for bringing a claim, until he could have discovered his mistake, than a person who has made a mistake in circumstances where, on any view, the law has remained unchanged. In the latter situation, the person could at least have discovered his mistake at the time if he had consulted a lawyer.

It is also relevant to note that there is some authority in other jurisdictions accepting that provisions equivalent to section 32(1)(c) apply to restitutionary claims based on mistakes of law. The question arose in an Australian case in relation to section 27(c) of the Limitation of Actions Act (Vic), which is materially identical to section 26(c) of the 1939 Act and section 32(1)(c) of the 1980 Act. In the case of Paciocco v Australia and New Zealand Banking Group Ltd [2014] FCA 35; (2014) 309 ALR 249, Gordon J held at para 365 that the concept of a mistake, within the meaning of section 27(c), included a mistake of law. On appeal, the Full Court expressed their agreement with that conclusion, obiter: [2015] FCAFC 50; (2015) 236 FCR 199, paras 192, 396 and 398. In particular, Besanko J considered the question whether the provision should be construed as what is sometimes termed a “fixed time” provision, which must be construed in the sense in which it would have been applied at the time of its enactment, or as a provision which is “always speaking” and can be given an ambulatory construction. He concluded that the latter view was to be preferred. On a further appeal to the High Court of Australia, only Nettle J considered the point, again obiter, and agreed with the views expressed in the courts below: [2016] HCA 28; (2016) 90 ALJR 835, para 374.

Although the point does not appear to have been specifically considered elsewhere, that conclusion is consistent with the application, in a number of other jurisdictions, of provisions materially identical to section 32(1)(c) of the 1980 Act to claims based on a mistake of law. That can be seen, for example, in the Hong Kong case of Ho Kin Man v Comr of Police [2012] HKCFI 1064; [2013] 1 HKC 13, and in a number of decisions of the Supreme Court of India, including Assistant Engineer (D1) Ajmer Vidyut Vitran Nigam Ltd v Rahamatullah Khan Alias Rahamjulla [2020] INSC 188.

Nevertheless, it is necessary to consider whether construing section 32(1)(c) in that way would have other consequences which would be contrary to Parliament’s intention. As we have explained, the reasoning in Deutsche Morgan Grenfell would indeed have that effect, since the mistake of law was, according to that reasoning, undiscoverable until it had been established by an authoritative judicial decision, which might not occur until the proceedings in question had been completed: a result which defeats the object of limitation. That is not, however, the effect of section 32(1)(c) if it is construed in accordance with the test proposed in Deutsche Morgan Grenfell [2007] 1 AC 558 by Lord Brown, and consistently with the approach established in Haward v Fawcetts [2006] 1 WLR 682, AB v Ministry of Defence [2013] 1 AC 78 and Paragon Finance plc v DB Thakerar & Co.

Even so, there are a number of arguments which need to be considered: notably, that to construe section 32(1)(c) as applying to mistakes of law would be destructive of legal certainty, and therefore contrary to the policy of Parliament; that previous decisions have indicated that section 32(1) should be restrictively construed; and that to treat mistakes of law as falling within the scope of section 32(1)(c) would undermine this court’s ability to reverse decisions of the Court of Appeal and to depart from its own decisions in accordance with the 1966 Practice Statement (Practice Statement (Judicial Precedent) [1966] 1 WLR 1234).

In relation to legal certainty, Lord Goff was correct in stating in Kleinwort Benson [1999] 2 AC 349 at p 389 that “the cause of action in a case such as the present may be extended for an indefinite period of time”; and Lord Hope was right to acknowledge at p 417 that “[t]he objection may be made that time may run on for a very long time before a mistake of law could have been discovered”, and that “[i]t may also be said that in some cases a mistake of law may have affected a very large number of transactions, and that the potential for uncertainty is very great”. In that regard, it is relevant to note that mistakes of law differ from mistakes of fact in that the facts are fixed at the relevant point in time, even if they may remain undiscovered until much later, whereas the law can be altered from time to time by judicial decisions. For these reasons, it can be argued with force that to apply section 32(1)(c) to mistakes of law undermines the basic purpose of limitation statutes, namely “putting a certain end to litigation and … preventing the resurrection of old claims”, as the Law Reform Committee stated at paragraph 7 of its Report.

A number of points can be made in response. First, section 32(1)(c), like the equitable rule which preceded it, necessarily qualifies the certainty otherwise provided by limitation periods. It means that the 1980 Act does not pursue an unqualified goal of barring stale claims: its pursuit of that objective is tempered by an acceptance that it would be unfair for time to run against a claimant before he could reasonably be aware of the circumstances giving rise to his right of action. Even as it applies to mistakes of fact, section 32(1)(c) (like sections 14 and 14A) has the consequence that the cause of action may be extended for an indefinite period of time. The point can be illustrated by the facts of In re Baronetcy of Pringle of Stichill [2016] UKPC 16; [2016] 1 WLR 2870, where DNA evidence established in 2016 that a person born in 1903 had wrongly succeeded to a title in 1919, with the effect of impugning the title inherited by successive generations of his descendants. The position would have been the same if he had been born centuries earlier.

Secondly, the uncertainty which is liable to result from the application of section 32(1)(c) to mistakes of law should not be exaggerated. In most cases where a mistake of law is made, the application of section 32(1)(c) will not produce disruptive consequences. That is because the mistake will normally have been discoverable at the time of the transaction in question by the exercise of reasonable diligence, by obtaining legal advice. The commencement of the limitation period will not, therefore, be postponed. Cases where advice as to the correct state of the law was not reasonably available at the time of the transaction, and where a right to restitution might in principle be available, are likely to be unusual. One example is the class of cases where the mistake of law is the retrospective result of a judicial decision which upsets an established rule of law on the basis of which payments have been made: what Lord Hoffmann described as a “deemed” mistake. Cases of that kind should, however, be highly unusual, as courts do not often overturn established rules of law, and in considering whether to do so they attach particular importance to the security of settled transactions. The present proceedings, and the other proceedings mentioned in paras 5 and 6 above, are not however concerned with deemed mistakes but with actual mistakes arising from a unique set of circumstances: the UK’s entry into the EU, with supranational laws which had to be given priority over domestic statutes, resulting in the gradual application of the EU principles of freedom of establishment and free movement of capital in the field of taxation, and the eventual realisation that UK tax legislation might be incompatible with those principles. Such circumstances are of a wholly exceptional nature.

Thirdly, to the extent that this objection to the result of Kleinwort Benson is based on policies attributed to Parliament, we would refer to the discussion of the intention of Parliament at paras 219-222 above. Furthermore, it is reasonable, 22 years after the decision in Kleinwort Benson [1999] 2 AC 349, to note that Parliament has evinced no concern about its consequences, except in relation to tax. Recommendations for reform were made by the Law Commission in a Report published almost 20 years ago (Limitation of Actions (2001) (Law Com No 270), HC 23), and were accepted in principle by the Government in 2002 (Hansard, HL Deb, 16 July 2002, col 127 WA), but in 2009 the Government announced that reform of the law of limitation would not after all be taken forward (Hansard, HC Deb, 19 November 2009, col 13 WS).

In relation to tax, the legislative measures introduced in section 320 of the FA 2004 have succeeded in protecting public revenues prospectively with effect from 8 September 2003. In relation to tax levied before that date, Parliament acted to mitigate the impact on public revenues in the F(No 2)A 2015 (para 56 above). That impact was further reduced, dramatically, by the decision in Prudential [2019] AC 929, and will be reduced further if this court departs from Deutsche Morgan Grenfell and adopts instead the interpretation of section 32(1) which was explained in para 209 above.

With the exception of claims in relation to tax that was unlawful under EU law, there has been no noticeable surge of claims for restitution of money paid under mistakes of law. Were such claims to be made after a long lapse of time, the defence of change of position might well be available, as it has been held to be in the present proceedings (FII (CA) 1 [2010] STC 1251), albeit not made out on the facts because of the absence of a clear relationship between tax receipts and public expenditure (FII (CA) 2 [2017] STC 696).

Another argument for holding that section 32(1) should not be interpreted as applying to mistakes of law is that the courts have made clear the risks involved in giving the provision a broad interpretation, in Phillips-Higgins [1954] 1 QB 411 and FII (SC) 1 [2012] 2 AC 33: see paras 41 and 139 above. In those cases, however, what was being rejected was an attempt to extend section 32(1)(c) to cases where mistake was not an essential ingredient of the cause of action, but where the claimant had merely been ignorant that he had a cause of action: an extension which is not being suggested in this judgment, and which would be inconsistent with the Law Reform Committee’s intention that “the mere fact that a plaintiff is ignorant of his rights is not to be a ground for the extension of time” (para 126 above).

That is not, however, a reason for excluding from the scope of section 32(1) cases where a mistake of law is an essential ingredient of the cause of action. That is because, in such cases, ignorance that he made a mistake renders the claimant unaware of one of the facts giving rise to his cause of action, just as a claimant who is ignorant that he made a mistake of fact is unaware of one of the facts giving rise to a cause of action based on a mistake of fact. In neither case is the limitation period postponed merely because the claimant is ignorant of his rights. Were matters otherwise, FII (SC) 1 could hardly have been decided as it was, since the claim based on Kleinwort Benson would then have been struck at just as much as the claim based on Woolwich [1993] AC 70.

A closely related argument is that discoverability is concerned with the facts which are essential to a cause of action, and not with their legal consequences. The principle is well illustrated by Knight Bruce LJ’s statement in Stafford v Stafford (1857) 1 De G & J 193, 202 that “[g]enerally, when the facts are known from which a right arises, the right is presumed to be known”, and by Sir Richard Collins MR’s judgment in Molloy 94 LT 756 (para 206 above). As we have explained, the reforms recommended by the Law Revision Committee were not intended to impinge upon that principle. The principle is reflected in the terms of sections 14(1)(d) and 14A(9) of the 1980 Act, which specifically provide that knowledge that the relevant acts or omissions involved negligence or other breaches of duty is irrelevant. Although Parliament did not set out a corresponding provision in section 32(1), the same principle nevertheless permeates section 32(1) just as much as it does the remainder of the 1980 Act. It might be argued, on that basis, that mistakes of law fall outside the ambit of section 32(1)(c).

The cause of action created by Kleinwort Benson depends on the claimant having had a mistaken understanding of the law at the time when the payment was made, and on a causal relationship between that mistaken understanding and the making of the payment. Those are the relevant facts, as discussed in para 207 above. Once those facts are or could with reasonable diligence be discovered, the limitation period begins to run. It is not postponed because the claimant, with actual or constructive knowledge of those facts, is ignorant that they give rise to an entitlement to restitution.

In those circumstances, to treat the cause of action recognised in Kleinwort Benson as falling within the scope of section 32(1) involves no breach of the general principle that “when the facts are known from which a right arises, the right is presumed to be known”. Nor, recalling Sir Richard Collins MR’s judgment in Molloy, is there any inconsistency with his statement that:

A further argument is that to treat mistakes of law as falling within the scope of section 32(1)(c) would undermine this court’s ability to reverse decisions of the Court of Appeal or to depart from its own decisions in accordance with the 1966 Practice Statement, since to do so might trigger widespread liabilities under the law of restitution. The first point to be made in response is that Parliament cannot have had the Practice Statement in mind in 1939. Nor can it bear on the interpretation of a 1939 provision which is re-enacted in a consolidation statute in 1980, since no change in meaning is taken to have been intended. The Practice Statement has to be operated within a framework established by statute, including the 1939 and 1980 Acts, rather than the Practice Statement affording guidance as to how those statutes should be interpreted.

Secondly, as we have indicated, courts, including this court, do not often overturn settled rules of law, and in considering whether to do so they attach particular importance to the security of settled transactions. The decisions in Kleinwort Benson and Sempra Metals [2008] 1 AC 561 were exceptional in their readiness to overturn centuries of authority, as the House of Lords enthusiastically adopted the theory of unjust enrichment. Those decisions were criticised by this court in Prudential [2019] AC 929 at para 63 because of their disregard of the need for judicial development of the law to be justifiable by reference to existing legal principles. Normally, as was stated in a recent judgment of this court, “[i]n order to preserve legal certainty, judicial developments of the common law must … be based on established principles, building on them incrementally rather than making the more dramatic changes which are the prerogative of the legislature”: R (Elgizouli) v Secretary of State for the Home Department [2020] UKSC 10; [2020] 2 WLR 857, para 170. Considering the Practice Statement in particular, it states specifically that the court “will bear in mind the danger of disturbing retrospectively the basis on which contracts, settlements of property, and fiscal arrangements have been entered into”. This court, like the House of Lords before it, has followed that practice.

Thirdly, the potential problem which concerned the minority in Kleinwort Benson is significantly alleviated if the approach to discoverability which was adopted in Deutsche Morgan Grenfell is departed from, as suggested above. For example, Lord Browne-Wilkinson posited at p 358 a case where the law was established by a decision of the Court of Appeal in 1930. In 1990 the claimant made a payment which was only due if the Court of Appeal decision was good law. In 1997 the House of Lords overruled the decision of the Court of Appeal. Lord Browne-Wilkinson commented at p 359:

Furthermore, as was explained at para 232 above, in considering whether the overturning of a precedent might result in restitutionary claims going back for long periods of time, it is necessary to bear in mind the defence of change of position. The circumstances which led to the rejection of that defence in the present case - the absence of any demonstrable connection between the tax paid and public expenditure - were unusual. In the event, such claims have not been a notable feature of the period since Kleinwort Benson was decided.

Taking stock of the discussion of Kleinwort Benson [1999] 2 AC 349, the position can be summarised as follows:

The decision in Kleinwort Benson, that claims for the restitution of money paid under a mistake of law fall within the ambit of section 32(1)(c) of the 1980 Act, was not supported by convincing reasoning (paras 148-161 above).

When section 32(1)(c) was enacted, it was not contemplated that it might extend to actions for the restitution of money paid under a mistake of law: no such action was recognised at that time.

Nevertheless, giving the words used in section 32(1)(c) their ordinary meaning, they include such actions. That is not, however, conclusive. The provision has to be construed consistently with its purpose.

The purpose of section 32(1)(c) is to postpone the commencement of the limitation period in respect of a claim for relief from the consequences of a mistake where, as a result of the mistake, the claimant could not reasonably have known of the circumstances giving rise to his cause of action at the time when it accrued (para 220 above).

If, after the enactment of section 32(1)(c), the law developed so as to allow actions to be brought for relief from the consequences of a mistake of law, then in principle it would be consistent with that purpose for section 32(1)(c) to apply to such claims. To construe the provision in a sense which excluded such claims would not be consistent with Parliament’s intention to relieve claimants from the necessity of complying with the time bar which would apply in the absence of section 32(1), at a time when they could not reasonably be expected to do so (para 221 above).

That argument applies equally to “deemed” mistakes of law as to “actual” mistakes, even assuming that a principled and workable distinction can be drawn between the two (para 222 above).

The construction of section 32(1)(c) as applying to mistakes of law as well as of fact also gains some support from the case law of other jurisdictions (paras 223-224 above).

On the other hand, it can be argued that such a construction of section 32(1)(c) undermines the primary policy of the 1980 Act as a whole, namely to put a certain end to litigation (para 227 above).

A number of points can be made in answer to that argument:

Section 32(1)(c) necessarily qualifies the certainty otherwise provided by limitation periods, in recognition of the unfairness of allowing time to run against a claimant before he could reasonably be aware of the circumstances giving rise to his right of action (para 228 above).

Nevertheless, in most cases where a mistake of law is made, the application of section 32(1)(c) will not produce disruptive consequences. The correct state of the law is normally ascertainable at the time of a transaction. Cases where it is not, and where a right to restitution might in principle be available, are likely to be unusual. In particular, cases where the courts upset an established rule of law with retrospective effect, so as to affect settled transactions, should be highly unusual. The present proceedings arise from a unique set of circumstances (para 229 above).

The policy consequences have not prompted legislation, except in relation to tax. On the contrary, the Government has declined to implement reforms recommended by the Law Commission (para 230 above).

In relation to tax, the consequences of Kleinwort Benson have been addressed by Parliament, and have also been mitigated by subsequent judicial decisions. They will be mitigated further if this court departs from Deutsche Morgan Grenfell (para 231 above).

Other than in relation to tax, the decision in Kleinwort Benson has not resulted in a surge of stale claims. Were such claims to be made, a defence of change of position might well be available (para 232 above).

It can also be argued that section 32(1)(c) should be restrictively construed, consistently with dicta in Phillips-Higgins and FII (SC) 1. However, those cases were concerned with attempts to extend section 32(1)(c) to cases where mistake was not an essential ingredient of the cause of action, but where the claimant had merely been ignorant that he had a cause of action. Claims for the restitution of money paid under a mistake of law do not fall into that category. On the contrary, such a claim was regarded as unobjectionable in FII (SC) 1 (para 233-234 above).

It can also be argued that to apply section 32(1)(c) to claims for restitution of money paid under a mistake of law contravenes the principle that ignorance of the law is not a ground for the extension of the limitation period. However, that is a mistaken view. The commencement of the limitation period is postponed while the claimant is unaware of the fact that he had a defective understanding of the law at the time when he made the payment. It is not postponed because he is ignorant that, in those circumstances, he has a right to restitution (paras 234-237 above).

It can also be argued that to treat mistakes of law as falling within the scope of section 32(1)(c) would undermine this court’s ability to reverse decisions of the Court of Appeal or to depart from its own decisions. The force of that argument appears to us to be diminished, however, when regard is had (a) to the importance which this court attaches in any event to legal certainty and to the security of settled transactions, (b) to the significance of adopting the approach to “discoverability” discussed above, and (c) to the importance of the defence of change of position (paras 239-241 above).

The claimant seeking restitution of money paid under a mistake of law does not, therefore, come within the scope of section 32(1) because he is unaware of his rights. He comes within it where, and during the period that, he is unaware that his understanding of the law at some point in the past was defective (the mistake in question being one which forms an essential element of a cause of action). He ceases to come within it at the point when he knows, or could with reasonable diligence know, that he made such a mistake “with sufficient confidence to justify embarking on the preliminaries to the issue of a writ, such as submitting a claim to the proposed defendant, taking advice and collecting evidence”; or, as Lord Brown put it in Deutsche Morgan Grenfell, he discovers or could with reasonable diligence discover his mistake in the sense of recognising that a worthwhile claim arises.

For these reasons, although there is undeniable force in the argument that section 32(1)(c) should be construed as being confined to mistakes of fact, the balance of the arguments in our view favours giving the language of section 32(1)(c) its ordinary meaning, so that it is applicable also to actions for relief from the consequences of a mistake of law. That approach to the construction of the provision best gives effect to Parliament’s intention to relieve claimants from the necessity of complying with a time limit at a time when they cannot reasonably be expected to do so, and does not have unacceptable consequences for the legal certainty which the 1980 Act is primarily designed to protect. That is only so, however, if the court departs from the reasoning in Deutsche Morgan Grenfell [2007] 1 AC 558, since that reasoning would defeat Parliament’s intention. On that basis, we consider that this court should adhere to the decision in Kleinwort Benson, so far as relating to limitation.

We must also give due weight to the importance of maintaining legal certainty by the preservation of precedent. The use of precedent, as the 1966 Practice Statement acknowledges, is an indispensable foundation upon which judges decide what the law is and apply the law in individual cases. It is, in Lord Goff’s words in Kleinwort Benson (p 379), “the cement of legal principle” providing stability to the common law. As is well known, this court has held that the Practice Statement has effect as much as it did before the Appellate Committee in the House of Lords: Austin v Southwark London Borough Council [2010] UKSC 28; [2011] 1 AC 355, para 25 per Lord Hope. It is necessary therefore to consider with care whether it is appropriate for this court to depart from prior decisions of the House of Lords.

It is unquestionable that there is a general public interest in certainty in the law. It is not a sufficient basis for this court to reverse a previous decision which it or the House of Lords has made that this court considers that a previous decision was wrong. As Lord Reid stated in R v Knuller (Publishing, Printing and Promotions) Ltd [1973] AC 435, 455,

Lord Reid explained his understanding of the rationale of the Practice Statement in R v National Insurance Comr, Ex p Hudson [1972] AC 944, 966 when he stated that there were a number of reported decisions which were impeding the proper development of the law or which led to results which were unjust or contrary to public policy. Some situations, such as a fundamental change in circumstances, or where a decision has resulted in unforeseen serious injustice, have been recognised as permitting a departure from precedent: Rees v Darlington Memorial NHS Trust [2003] UKHL 52; [2004] 1 AC 309, para 31 per Lord Steyn. In Horton v Sadler [2006] UKHL 27; [2007] 1 AC 307, para 29, Lord Bingham considered that too rigid adherence to precedent might cause injustice in a particular case and unduly restrict the development of the law. But, in the same paragraph, he acknowledged that the power had been exercised rarely and sparingly.

In view of this well-established approach to precedent, would it be right for this court to depart from Kleinwort Benson or Deutsche Morgan Grenfell in relation to the law of limitation on this appeal?

Kleinwort Benson effected a radical change in the law of restitution by opening up claims for the recovery of money paid under a mistake of law. By applying section 32(1)(c) of the 1980 Act to such claims it created the potential that a cause of action may be extended for an indefinite period of time and thereby undermine security of transactions, as each of the majority, Lord Goff, Lord Hoffmann and Lord Hope, expressly recognised (paras 154, 157 and 162 above). The minority, Lord Browne-Wilkinson and Lord Lloyd, saw this potential as a basis for leaving the proposed change of law to Parliament (para 163 above). But there has been little evidence of any surge of claims for restitution of money paid under mistakes of law. The most significant claims have been in the field of taxation, such as the various group litigations which we have mentioned in paras 5 and 6 above. Those challenges have exposed the Exchequer to claims which go back many years and involve very large sums of money. But Parliament has intervened, as we have explained, by enacting section 320 of the FA 2004 which, while ineffective to undermine claims under EU law retrospectively, has protected public revenues prospectively with effect from 8 September 2003 (paras 10 and 14 above). There is therefore no apparent danger of similarly large claims arising in future in the field of taxation which have the potential to disrupt the fiscal planning of the executive.

In paras 242 and 243 above, we have summarised our position in relation to Kleinwort Benson. The considerations stated there and those in para 248 above suggest to us that preserving the authority of Kleinwort Benson would not be contrary to principle or give rise to serious uncertainty in the law. Upholding Kleinwort Benson would be unlikely to give rise to serious injustice in individual cases in the future, and it would not impede the proper development of the law.

On the other hand, from our discussion which we have summarised in para 213 above, it is clear that the decision in Deutsche Morgan Grenfell [2007] 1 AC 558 on the question of discoverability in section 32(1)(c) has very unfortunate consequences. Several matters are of particular relevance to the application of the Practice Statement. The decision defeats the purpose of limitation, and in so doing appears to be contrary to the intention of Parliament in enacting the 1939 and 1980 Acts. It creates incoherence in interpretation both within section 32(1) and between that section and analogous provisions of the 1980 Act. It creates the legal paradox to which we have referred (paras 173-174 above). It also perpetuates the problem of distinguishing between matters of fact and matters of law, a result which, as we have discussed, is contrary to the intended effect of Kleinwort Benson [1999] 2 AC 349. In so doing, it impedes the coherent development of the law.

It is necessary to balance against those considerations the possibility that a departure from Deutsche Morgan Grenfell, in relation to discoverability, will itself result in some claims to restitution. Such claims may be made on the basis that payments have been made under a mistake of law, because the claims for restitution, which that decision supported and which led to those payments, were, on a proper understanding of the law, already subject to a limitation defence on the interpretation of section 32(1)(c) which we favour. That would be unfortunate. But we would not expect the number of claims to be significant for two reasons. First, there has not been a surge of claims for restitution of money paid under mistakes of law, following the Kleinwort Benson decision, outside the tax litigation to which we have referred. Secondly, the recipients of such payments made in restitution may have a defence of change of position if the payer, such as the Revenue, were to seek to recover them. We are not persuaded that the possibility of such claims should deter us from departing from Deutsche Morgan Grenfell in relation to discoverability if that is the only way in which to promote coherence in the law of limitation.

When the Appellate Committee determined the appeal in Deutsche Morgan Grenfell [2007] 1 AC 558 in 2006, Lord Hope (para 68) suggested that the legislature could intervene to stop time running indefinitely in all mistake cases, if there was a problem. There was then some prospect that Parliament would consider a reform of the law of limitation of actions. As we have explained (para 230 above), the Government initially accepted the Law Commission’s recommendations to reform the law of limitation and proposed to legislate, but by 2009 it had announced that it would not take forward those reforms. There is therefore now no prospect that Parliament will enact a legislative solution to remove the anomalies which the Deutsche Morgan Grenfell judgment has created.

In these circumstances, we are persuaded that this is an appropriate case in which to depart from the decision in Deutsche Morgan Grenfell in relation to discoverability in section 32(1)(c) of the 1980 Act.

This appeal is brought against the decision of the Court of Appeal in FII (CA) 2 [2017] STC 696, and this judgment is concerned with the appeal only in so far as it relates to Issue 28: the issue of limitation. As was explained at para 54 above, the decision of the Court of Appeal on Issue 28 was based on the application of the approach established in Deutsche Morgan Grenfell. For the reasons we have explained, that approach cannot be upheld, and the appeal on Issue 28 must therefore be allowed.

This court cannot, however, determine in the abstract the point in time when the test claimants could with reasonable diligence have discovered, to the standard of knowing that they had a worthwhile claim, that they had paid tax under a mistaken understanding that they were liable to do so. That depends on an examination of the evidence. As we have explained, EU law, in relation to tax regimes which discriminated between companies based in one member state and companies based in another, developed through a series of judgments of the Court of Justice, including Verkooijen [2000] ECR I-4071 (2000), Lenz [2004] ECR I-7063 (2004) and Manninen [2005] Ch 236 (2004), discussed at paras 24 and 33-34 above, Hoechst [2001] Ch 620 (2006) and FII (CJEU) 1 [2012] 2 AC 436 (2006). Each of those judgments was itself the result of a claim made some years earlier. In Hoechst, for example, the claim was filed in 1995, 11 years before the judgment of the Court of Justice. DMG was aware of the claim almost immediately, and it was for that reason that, in Deutsche Morgan Grenfell, Lord Brown considered that time began running for DMG in July 1995. But the date when the claimant became aware of another claim, and appreciated its potential implications for its own situation, is not conclusive, if a claimant acting with reasonable diligence could have discovered that it had a worthwhile claim at an earlier time. Equally, the answer to the question arising under section 32(1) does not depend upon the characteristics of the particular claimant: whether, for example, it was inclined to await further developments, and to allow other taxpayers to make the running. The standard is “could”, as Millett LJ emphasised in Paragon Finance (para 203 above). And the test is objective, as Millett LJ explained in the same passage of his judgment, and as Lord Walker made clear in FII (SC) 1 [2012] 2 AC 33, when he referred (para 48 above) to the time when “a well advised multi-national group based in the UK would have had good grounds for supposing that it had a valid claim to recover ACT levied contrary to EU law”.

In the circumstances of the present proceedings, if the date of commencement of the limitation period requires to be judicially determined, that matter will have to be decided by the High Court, after the parties have had an opportunity to amend their pleadings.

We would allow the appeal on Issue 28 and would make an order remitting that issue to the High Court to allow the parties to amend their pleadings on discoverability of the mistake and to determine the date of commencement of the limitation period.

In large measure we agree with the judgment of Lord Reed and Lord Hodge, which sets out the issues and explains this litigation and the course of the previous litigation in this area with such admirable clarity. The issue on which we find ourselves in respectful disagreement is whether this court should overrule Kleinwort Benson Ltd v Lincoln City Council [1999] 2 AC 349 as regards the interpretation of section 32(1)(c) of the Limitation Act 1980 and hold that it does not apply in relation to payments made on the basis of a mistake of law. In our view, we should do so.

In outline, we have reached that view for three main reasons. First, we are convinced that the House of Lords was plainly wrong in Kleinwort Benson to interpret section 32(1)(c) as extending to mistakes of law. Secondly, we do not consider that the large inroads upon the overall purpose of the Limitation Act in undermining legal certainty in relation to settled transactions, recognised by all their lordships in that case, are sufficiently addressed by the limited departure from Deutsche Morgan Grenfell Group plc v Inland Revenue Comrs [2006] UKHL 49; [2007] 1 AC 558 which the majority propose. The outcome will, we fear, place a serious brake upon judicial modernisation of the common law which we are sure Parliament cannot have intended. This issue has to be confronted in this court, because the hopes of their lordships in Kleinwort Benson that Parliament would legislate to deal with the problem have not been fulfilled. Thirdly, we have serious reservations about whether the test proposed by the majority, based upon Lord Brown’s dissenting speech in Deutsche Morgan Grenfell, will prove to be workable in practice. It is not in our view plausible to infer that Parliament intended that section 32(1)(c) should be read as being subject to such a test.

Although the 1980 Act is a consolidation statute, in construing section 32(1)(c) the House of Lords in Kleinwort Benson and this court in Test Claimants in the FII Group Litigation v Revenue and Customs Comrs (formerly Inland Revenue Comrs) [2012] UKSC 19; [2012] 2 AC 337 (“FII (SC) 1”) found it necessary to look at the background and intended effect of the predecessor provision in the Limitation Act 1939, section 26(c). As was made clear in Farrell v Alexander [1977] AC 59, where there is significant doubt about the meaning and effect of a provision in a consolidation statute it is appropriate to investigate the meaning and effect of the earlier legislation from which it is derived.

Lord Reed and Lord Hodge set out the law as regards limitation of actions as it stood prior to 1939 at paras 103 to 122 above. In summary, statute set out periods of limitation for claims arising at common law while generally equity applied the doctrines of laches (which included reference to statutory limitation periods in relation to equitable claims which were analogous to claims at law) and acquiescence. For the purposes of those doctrines, in certain circumstances time did not run where a claimant was labouring under a mistake until the mistake was discovered, or could with reasonable diligence have been discovered. A claim would arise at common law where money was paid to another by reason of a mistake of fact by way of an action for money had and received, which had historically been vindicated using the old form of action known as indebitatus assumpsit. The time limit for bringing such a claim was six years from the date of the payment: Baker v Courage & Co Ltd [1910] 1 KB 56. It had been established by the case of Bilbie v Lumley (1802) 2 East 469 that this form of action was not available, and hence this type of claim did not arise, to claim back money paid under a mistake of law. By 1939 this was a well-established rule of law. On the other hand, equity never provided relief in relation to money paid away by reason of a simple mistake, whether of law or fact, without more. Equity granted relief to vindicate certain underlying property rights, or rights arising under a trust or in relation to the execution of a will. Mistake, including in some cases a mistake of law, was just a relevant factor to be taken into account in deciding whether equity would intervene to vindicate those rights in a particular case.

As Lord Reed and Lord Hodge observe (para 119), in cases of fraud the equitable rule was that time would not run by analogy with statute until the claimant could with reasonable diligence have discovered the fraud, since it would be unconscionable for a defendant in such a case to rely on the statute to defeat the claim. Clearly, that reasoning does not apply in a case where a claimant labours under a mistake which the defendant has done nothing to induce. But in Brooksbank v Smith (1836) 2 Y & C Ex 58 Alderson B expressed the view that the rule in cases of fraud should apply in cases of mistake as well, without explaining why.

In 1936 the Law Revision Committee (“the LR Committee”) produced its Fifth Interim Report on the law of limitation: see paras 123-128 above. Its recommendations were enacted in the 1939 Act. The LR Committee rejected the idea of a general power of extension of limitation periods, on the grounds that it might be impossible to predict how such a power would be exercised, in which case “the fundamental benefit conferred by statutes of limitation, namely the elimination of uncertainty, would be prejudiced” (para 7). At paragraph 13 the LR Committee recommended leaving the equitable doctrines of laches and acquiescence in place. At paragraphs 22 and 23 the LR Committee examined the merits of applying equitable principles to common law claims; in doing so, it discussed fraud claims and mistake claims separately. The inference from the way in which the LR Committee separated its discussion of fraud and mistake for common law claims is that it recognised that the equities between the parties and the policy issues arising in the two cases are very different. In FII (SC) 1 Lord Walker (para 63) and Lord Sumption (paras 183-185) explain the contrasting policy issues and the risks of uncertainty attendant on an over-broad extension of limitation periods in cases of mistake as distinct from fraud.

At paragraph 22 the LR Committee recommended adopting the equitable rule regarding extension of time for the purposes of common law claims based on fraud. It identified two ways in which fraud might have an impact (“Either the cause of action may spring from the fraud of the defendant or else the existence of a cause of action untainted in its origin by fraud may have been concealed from the plaintiff by the fraudulent conduct of the defendant”) and observed, “[i]t is obviously unjust that a defendant should be permitted to rely upon a lapse of time created by his own misconduct.” Its recommendation was that time should not start to run in either case until the fraud was or could with reasonable diligence be discovered. This recommendation was followed in section 26(a) and (b) of the 1939 Act (re-enacted as section 32(1)(a) and (b) of the 1980 Act), reflecting the two ways in which fraud could operate, respectively: see FII (SC) 1, paras 179-180 (Lord Sumption). The equities are, of course, entirely different in cases of ordinary mistake where the defendant has done nothing unconscionable to create the delay before the claimant seeks to litigate.

At para 23 the LR Committee recommended adopting the equitable rule regarding extension of time in relation to the common law action for relief from the consequences of a mistake. This recommendation was carried into the 1939 Act at section 26(c). As explained in FII (SC) 1, at paras 42-63 (Lord Walker) and paras 177-185 (Lord Sumption), the LR Committee’s recommendation was limited to cases where the mistake itself gave rise to a cause of action. Given the established state of the law in 1936, this meant that the recommendation was confined to cases where a payment was made by reason of a mistake of fact. As Pearson J said in the leading case on the ambit of section 26(c), “No doubt it was intended to be a narrow provision, because any wider provision would have opened too wide a door of escape from the general principle of limitation” (Phillips-Higgins v Harper [1954] 1 QB 411, 419, cited with approval by Lord Sumption in FII (SC) 1, para 183). The LR Committee did not recommend any change in the substantive law regarding claims at common law based on mistake and did not make any recommendation which addressed the very different policy issues which would arise in respect of a claim to recover a payment based on a mistake of law. That such a claim might be recognised was something entirely outside its contemplation. Further, the LR Committee was at pains to state that “the mere fact that a plaintiff is ignorant of his rights is not to be a ground for the extension of time. Our recommendation only extends to cases when there is a right to relief from the consequences of a mistake.” A mistake of law occurs where a claimant is ignorant of his rights. The only right to relief from the consequences of a mistake which was in the contemplation of the LR Committee was where there was a mistake of fact. It was fundamental to the approach of the LR Committee that it regarded the need to protect past payments from claims for repayment many years later by persons alleging ignorance of their rights as being satisfied by the absence of any cause of action, either in law or in equity, for repayment on the ground of mistake of law. In terms of the policy issues which arise, we consider that no sensible distinction can realistically be drawn between ignorance of the right to restitution on the ground of mistake of law and ignorance of the underlying rights which constitutes the mistake of law on which that right depends: cf para 220 above. Accordingly, with respect, we do not agree that cases of mistake of fact and cases of mistake of law can be equiparated (see para 236 above) so far as concerns the policy and effect of either the equitable rule or the recommendation of the LR Committee.

It follows that, where, implementing the recommendation of the LR Committee, section 26(c) was enacted referring to an action “for relief from the consequences of a mistake”, Parliament meant by that phrase an action for relief from the consequences of a mistake of fact: see para 133 above. In neither the 1939 Act nor the 1980 Act, when section 26(c) was re-enacted as section 32(1)(c), did Parliament attempt to address the distinct policy issues regarding limitation which arise when a claim is recognised for recovery of money paid under a mistake of law, as happened in Kleinwort Benson [1999] 2 AC 349.

With that change in the law, the question arose for the first time whether the phrase “action for relief from the consequences of a mistake” in section 26(c) of the 1939 Act and section 32(1)(c) of the 1980 Act covered not only claims for recovery based on mistake of fact but also claims based on mistake of law. The Appellate Committee in Kleinwort Benson held that it did, but proceeded on a mistaken understanding as to the state of the law prior to the enactment of the 1939 Act: paras 148-163 above. The only substantive reasoning in support of construing section 32(1)(c) as extending to claims for recovery of money paid under a mistake of law was by Lord Goff and Lord Hope. It is a remarkable feature of the case that the reasoning of all members of the Appellate Committee implicitly recognised that the effect of reading section 32(1)(c) as including claims for recovery based on mistake of law as well as mistake of fact would dramatically undermine the intention of Parliament in the 1939 Act and the 1980 Act to set out clear and readily applicable periods of limitation.

We consider that the House of Lords erred in Kleinwort Benson in giving section 32(1)(c) this interpretation. Lord Reed and Lord Hodge question whether it is appropriate to consider this issue through the prism of the doctrine that statutes are to be taken to be “always speaking”. We think that it is helpful and appropriate to do so but, as they observe, nothing really turns on this.

The guidance regarding the ambit of the “always speaking” doctrine is in fact concerned with the fundamental underlying issue of whether Parliament can be taken to have intended by a statutory provision passed at one point in time, using language directed to the circumstances at that time, to cover a new set of circumstances which has come into existence since then. As we understand it, Lord Reed and Lord Hodge agree that this is the fundamental issue raised by the decision in Kleinwort Benson regarding the application of section 32(1)(c): see paras 155 and 157 above. The issue of how broadly one should construe the language of the statutory provision to cover new matters arising after its enactment necessarily involves consideration of what inferences can be drawn from the language used and the circumstances of the enactment as to Parliament’s policy intention in promulgating the provision. If the inference can be drawn that Parliament’s policy intention was broad and the new matters are aligned with that broad intention and are covered by it, a court will be justified in concluding that the provision applies; conversely, if there is not sufficient congruence between the policy issues raised by the new matters and Parliament’s intention as expressed when it enacted the provision, the provision does not apply. Since the case law on the “always speaking” doctrine addresses this question, we will make reference to it. In our view, the question to be posed is whether the phrase using the term “mistake” in section 26(c) of the 1939 Act (and re-enacted in section 32(1)(c) of the 1980 Act), where in the legal context in 1939 and 1980 the word could only refer to a mistake of fact, should in the light of the change in legal doctrine made in Kleinwort Benson now be taken to include also a mistake of law.

The ambit of the “always speaking” doctrine was explained by Lord Wilberforce in Royal College of Nursing of the United Kingdom v Department of Health and Social Security [1981] AC 800, 822:

As read in light of the LR Committee’s Report on limitation periods on which the 1939 Act was based, the Act had two features which are relevant for present purposes. First, as a matter of general policy, in the interests of predictability, certainty and security of transactions, it re-enacted the previous six-year time limit for actions at common law for the recovery of money paid under a mistake. This was a continuation of the established policy of the Statute of Limitations of 1623 to promote finality, certainty and security of receipt of money, as emphasised by Sir Richard Collins MR in Molloy v Mutual Reserve Life Insurance Co (1906) 94 LT 756, 761 (see para 206 above). It also represents the principal policy to which effect was given in the 1939 Act, in light of which any derogation falls to be interpreted on a restrictive basis: see paras 263-266 above. It was in line with the general policy of the 1939 Act to enact and regulate limitation periods on a comprehensive and coherent basis. This policy objective was recognised in In re Diplock [1948] Ch 465, 514, where the Court of Appeal noted that the wording of section 2(1)(a) of the 1939 Act, which enacts a six year limitation period for claims in contract, was not entirely apt to cover claims in quasi-contract to recover money paid under a mistake (or in unjust enrichment, as it would be categorised today), but nonetheless concluded that it should be so interpreted. In other words, the court considered that the policy of the 1939 Act to introduce certainty in relation to limitation was so strong that such claims were to be treated as falling within the scope of this provision.

Secondly, section 26(c) of the 1939 Act was directed to addressing a very specific issue, ie modifying the ruling in Baker v Courage & Co Ltd regarding the time limit for an action at law to claim recovery of money paid under a mistake of fact, but on a narrow basis. The restriction of that common law action to recovery of money paid under a mistake of fact was well established in 1939; there was no equivalent claim in equity; and there was no call at the time for the ambit of the common law action to be expanded to cover recovery of money paid under a mistake of law. Even in equity, the courts were at pains to emphasise the difference between the sort of error of law which might be relevant to a claim for equitable relief (ie error of law as to private rights, where the analogy with mistake of fact was very close: see Cooper v Phibbs (1867) LR 2 HL 149, 170 per Lord Westbury; Earl Beauchamp v Winn (1873) LR 6 HL 223, 234 per Lord Chelmsford; and Ministry of Health v Simpson [1951] AC 251, 268-270), and error regarding general law. At paragraph 23 of its Report, the LR Committee made it clear that it was not recommending that limitation should be extended where a party had made a mistake about his rights. Therefore, it was not in Parliament’s contemplation that the common law could be changed in the direction taken in Kleinwort Benson [1999] 2 AC 349.

Moreover, the policy issues which would arise in relation to limitation if section 26(c) applied in respect of recovery of money paid under a mistake of law are of a wholly different scale and character from those which were confronted and debated by the LR Committee in its Report, focused as it was on the existing common law claim for recovery of money paid under a mistake of fact. The speeches in Kleinwort Benson itself make the difference plain. It flows from the process by which the common law develops and changes over time while at the same time adhering to a declaratory theory of the law according to which decisions have retrospective effect (see in particular [1999] 2 AC 349, 377-379 and 381-382 per Lord Goff).

In our view, the House of Lords in Kleinwort Benson [1999] 2 AC 349, by changing the law to bring a new type of legal claim into existence, created a new state of affairs which did not fall within the intention or purpose of Parliament in enacting section 26(c) of the 1939 Act:

The new state of affairs did not fall within the same genus of facts as those by reference to which the expressed policy had been formulated. Mistake of law is something very different from mistake of fact. Mistake of law is a concept liable to change over time as the common law develops and changes, and to do so with retrospective effect, thereby wholly undermining the central policy of the 1939 Act and other Limitation Acts of achieving certainty after a fixed period of time. By contrast, mistake of fact is something fixed in time by reference to the facts which really were in existence at the time when the cause of action arose. As Lord Lloyd put it in Kleinwort Benson [1999] 2 AC 349 (p 393), “[f]acts are immutable, law is not”. The scale of disruption to the central policy of the Limitation Acts is completely different in the two cases;

It is not possible to detect a clear purpose in the legislation which can only be fulfilled if the extension is made. On the contrary, interpreting “mistake” in the phrase “the action is for relief from the consequences of a mistake” as it applies to the common law action for recovery of money paid under a mistake to cover a mistake of law as well as a mistake of fact would defeat the clear primary purpose of the legislation, to produce certain time limits within which claims may be brought. It would also undermine the policy intention expressed in paragraph 23 of the LR Committee’s Report that time should not be extended in cases of ignorance of rights;

The nature of the 1939 Act, to produce a comprehensive and effective limitation regime, as its principal policy, and the narrow and precise phraseology employed in section 26(c) (see paras 265-266 above), are both strong indications that the word “mistake” cannot, on a purposive construction, be construed to apply to a common law claim for recovery of money paid under a mistake of law. It is clear that this particular provision was designed to be restrictive and circumscribed in its operation rather than liberal or permissive, and much more circumscribed than the equitable doctrine of laches, which did not depend upon the claim in equity being founded upon mistake, in the sense of it being an integral part of the cause of action. Further, the language in section 26(c) of the 1939 Act and section 32(1)(c) of the 1980 Act of a mistake being “discovered”, or “discovered with reasonable diligence”, in the context of a common law claim, is not apt to cover a mistake of law of a general kind, to which the common law claim now extends, pursuant to Kleinwort Benson. That is also true in relation to mistake of law in equity, where the emphasis was always on the analogy between mistake as to private rights and mistake of fact. Contrary to the observation of Lord Goff in Kleinwort Benson [1999] 2 AC 349, 388H-389A, the pre-existing equitable rule did not apply to all mistakes, whether of fact or law. Equity was more nuanced than that, and it did not include a claim for simple recovery of money paid under a mistake of law: see Rogers v Ingham (1876) 3 Ch D 351, 355 per James LJ;

The subject matter, an action at common law for money paid under a mistake of law, is different in kind and in the dimension of its implications from that for which the legislative provision was passed, to cover an action at common law for money paid under a mistake of fact. The debate regarding the merits of a change in the substantive law to allow recovery for mistake of law, reviewed in Kleinwort Benson, itself reveals the different issues of principle which arise in the two cases: see [1999] 2 AC 349, 371E-372A per Lord Goff (“… as the majority judgments in Brisbane v Dacres [5 Taunt 143] show, the rule [in Bilbie v Lumley] was perceived, after due deliberation, to rest on sound legal policy … the difficulties now faced in formulating satisfactory limits to a right to recover money paid under a mistake of law reveal that there was more sense in the rule than its more strident critics have been prepared to admit”).

Lord Reed and Lord Hodge have explained how Lord Goff and Lord Hope misunderstood the legal position as it existed when Parliament legislated in 1939. This had the effect that their reasoning in Kleinwort Benson regarding the interpretation of section 26(c) in relation to the new claim to recover money paid under a mistake of law was flawed, because they did not properly understand the limited object which Parliament sought to achieve in 1939 in enacting that provision: see para 272 above. Kleinwort Benson provides no other basis for applying section 26(c) of the 1939 Act and then section 32(1)(c) of the 1980 Act to mistakes of law. This flaw was compounded by their failure to appreciate that the major degree of uncertainty in the law which would be introduced by interpreting section 32(1)(c) of the 1980 Act and its predecessor section 26(c) of the 1939 Act as covering the new type of claim, which all members of the Appellate Committee identified would be the consequence, showed that such an interpretation was completely at odds with the policy and intent of both statutes. This latter point deserves emphasis.

In Kleinwort Benson [1999] 2 AC 349, Lord Browne-Wilkinson considered (p 364) that, on the footing that Lord Goff was correct in holding that section 32(1)(c) of the 1980 Act applies to actions for recovery of money paid under a mistake of law, the disruption to settled entitlements every time the law was changed or developed by judicial decision would be so great that the House of Lords ought not to make the change to the substantive law which the majority decided upon, to allow recovery of money paid under a mistake of law. He took that view even though he thought that would be a desirable reform of substantive law. As he said, the consequence would be that “[o]n every occasion in which a higher court changed the law by judicial decision, all those who had made payments on the basis that the old law was correct (however long ago such payments were made) would have six years in which to bring a claim to recover money paid under a mistake of law”; as a result, in his judgment “the correct course would be for the House to indicate that an alteration in the law is desirable but leave it to the Law Commission and Parliament to produce a satisfactory statutory change in the law which, at one and the same time, both introduces the new cause of action and also properly regulates the limitation period applicable to it”.

In other words, Lord Browne-Wilkinson recognised that the change in the substantive law, if the limitation position was as stated by Lord Goff, would be massively disruptive of settled transactions and would unduly undermine security of receipt of money on a very wide scale. Lord Lloyd of Berwick agreed with him. He emphasised the “intense uncertainty” which would follow from the conclusion of the majority in the case to change the substantive law, with transactions unsettled and liable to be reopened, a consequence which he viewed “with alarm”: [1999] 2 AC 349, 397-398. In our view, however, the logical conclusion should have been that the change in the substantive law was of such a character as fell outside the policy and intent of the 1939 Act and the 1980 Act and outside the meaning of section 26(c) and section 32(1)(c) respectively, on a purposive, always speaking, construction.

Lord Reed and Lord Hodge challenge Lord Browne-Wilkinson’s reasoning on this point, on the footing that if their view that section 32(1)(c) is subject to a test of the discoverability of a mistake of law is accepted, the extent of disruption contemplated by him is reduced: para 240 above. But if, as we think, that test cannot plausibly be said to be part of the meaning which Parliament intended section 32(1)(c) to have, we fear that their challenge is misplaced. Indeed, it is in our view revealing that such an interpretation of section 32(1)(c) did not occur to any member of the Appellate Committee, who were addressing the meaning of the provision for the first time and without any preconceptions. In any event, it does not seem to us that their proposed reading of section 32(1)(c) does adequately deal with the points made by Lord Browne-Wilkinson and Lord Lloyd. Clearly, there may be many cases where there is a long period of time, far exceeding the usual six year limitation period, between a payment being made on the basis of some settled common law rule and some later development in legal opinion which calls that rule into question to the threshold standard of discoverability which Lord Reed and Lord Hodge endorse. We consider that Lord Browne-Wilkinson’s point remains a good one. In our view, to apply section 32(1)(c) to payments made under mistake of law would give rise to levels of uncertainty which conflict with the policy objective stated by the LR Committee (see paragraph 7 of its Report) and the underlying policy of the 1939 Act and the 1980 Act as limitation statutes, and could not have been regarded by Parliament as acceptable.

Lord Goff made the statement set out at para 154 above in which he recognised that great uncertainty in the law would arise from the application of section 32(1)(c) to claims for recovery of money paid under a mistake of law. We agree with the criticism of this passage by Lord Reed and Lord Hodge at para 155. With respect to Lord Goff, he omitted to consider the question of the application of section 32(1)(c) in terms of the object of the 1980 Act and to adopt a purposive construction in the light of that. In our view, if that had been done, he would have been constrained to accept that the points he himself made showed that to treat that provision as applicable would clearly undermine the policy of the 1980 Act, with the result that section 32(1)(c) could not bear the interpretation he sought to place on it. As he said, the dramatic consequences produced by a combination of the recognition of the new cause of action in Kleinwort Benson and an extended interpretation of section 32(1)(c) had not been appreciated at the time of the enactment (indeed, they were completely outside what was in Parliament’s contemplation when it passed both the 1939 Act and the 1980 Act), and were of such a profound character as to “call for legislative reform to provide for some time limit”, as opposed to (in practice) a wholly indefinite limit. But this serves only to emphasise that his proposed reading of that provision was contrary to the policy of the enactments.

Lord Hoffmann made similar points at p 401, in the passage set out at para 157 above. He noted that the combination of the change in substantive law to allow claims for recovery of payments made under a mistake of law and the application of section 32(1)(c) might be said to go too far in undermining security of transactions, and observed in that regard that “[t]he most obvious problem is the Limitation Act, which as presently drafted is inadequate to deal with the problem of retrospective changes in law by judicial decision”. We agree with the comment about this by Lord Reed and Lord Hodge at para 157.

Thus, faced with the same dilemma as Lord Browne-Wilkinson, Lord Hoffmann favoured changing the law on recovery of payments made under a mistake of law, notwithstanding that he recognised that the Limitation Act was inadequate to deal with retrospective changes of the law by judicial decision. But in our view this was a false dilemma. The proper conclusion to be drawn from this assessment was that section 32(1)(c) should not be construed to cover the new form of claim. It clearly fell outside the policy of the Act in relation to that provision, which was addressed specifically to claims for recovery of payments made under mistake of fact. Construing the provision as referring only to such claims, and not claims for recovery of money paid under mistake of law, would serve to maintain a proper balance of the public interest in the security of transactions, which would be assured after a limitation period of six years from the date of payment.

Lord Hope indicated (p 417) that he thought the LR Committee intended the word “mistake” to extend to all mistakes of law, but this is not correct: see para 159 above. A proper reading of the Report leads to the opposite conclusion. Later in his speech (pp 417-418) he made the statement set out at para 162 above. Although he accepted that “time may run on for a very long time before a mistake of law could have been discovered with reasonable diligence” and there was potential for uncertainty, in his view this was a problem for the legislature to resolve. He observed that the problem did not arise under the statutory limitation regime for Scotland, since the relevant prescriptive period of five years could be extended only where the creditor was induced to refrain from making a claim by fraud or error induced by the debtor’s words or conduct or was under a legal disability.

Similar points may be made about this part of Lord Hope’s reasoning as in relation to Lord Goff’s speech. In our view, Lord Hope’s own account indicates why his interpretation of section 32(1)(c) is contrary to the policy of the 1980 Act, read as a whole and also specifically in relation to the provision itself. As he acknowledged, his interpretation of the provision creates very long periods before limitation could apply (and, of course, since there will be new judicial decisions in future, any of which might effect a relevant change in the law, any limitation period which appears to be closed could always be reopened to run again). The potential for uncertainty thereby created was indeed “very great”. The conclusion from this ought to be that mistake of law as a ground of recovery of money paid, in an action at common law, was never contemplated by Parliament to be capable of falling within section 32(1)(c) and that a purposive interpretation of that provision, in its statutory context, means that it cannot be construed in that way. The comparison with the position in Scotland underlines this point, for it is difficult to see why Parliament would have wished to produce such a radical difference of limitation outcome in the two jurisdictions in relation to a cause of action of a character which is equally viable and capable of vindication on both sides of the border (unlike purely equitable claims in English law).

Moreover, in our view, when the context of the 1939 Act and the 1980 Act as limitation statutes designed to produce reasonably determinate limits for the bringing of claims and the specific purpose of section 26(c) of the 1939 Act and section 32(1)(c) of the 1980 Act are brought into account, the argument based on the language of those provisions and the use of the word “discovered”, although dismissed by Lord Goff, acquires particular force as indicating that the provisions were not intended to apply to cases of payment under mistake of law. It can readily be seen that the language of discovery is apt in the context of a limitation statute when speaking of discovery of a mistake of fact. It is far more difficult to square it with a deemed mistake of law produced by the retrospective effect of a later judicial decision. This is indeed what led to the conundrums debated in Deutsche Morgan Grenfell [2007] 1 AC 558 and then again in the judgment of Lord Reed and Lord Hodge.

In our judgment, therefore, there was a clear misstep by the House of Lords in Kleinwort Benson [1999] 2 AC 349 when it construed section 32(1)(c) as it did. In our view, the decision that section 32(1)(c) applies to common law claims based upon mistake of law was wrong, as a matter of construction of the provision. This is where, with respect, we part company with Lord Reed and Lord Hodge.

In the next part of their judgment (paras 165 and following) they consider the decision of the House of Lords in Deutsche Morgan Grenfell [2007] 1 AC 558. Since no one in that case raised the issue of whether the House of Lords in Kleinwort Benson was right to construe section 32(1)(c) as applying to payments under a mistake of law, the members of the Appellate Committee all proceeded on the footing that it did so apply. The question therefore was when such a mistake, as produced by the retrospective effect of a court decision delivered after the payment was made, could be regarded as being capable of discovery for the purposes of the section. The majority view in Deutsche Morgan Grenfell was that the mistake could only be discovered when the later court decision was made. Lord Brown dissented, saying that the possibility of a mistake (ie the possibility of the reversal of the rule of law on the basis of which the claimant made a payment) would be capable of being identified before the reversal by the later court decision actually occurred and it was from when it could be discovered that the prospect of this occurring was sufficiently developed that the limitation period would run. Upon reconsideration of this point, Lord Reed and Lord Hodge prefer the solution proposed by Lord Brown. They conclude that this reflects the proper interpretation of section 32(1)(c) on a purposive approach in line with Parliament’s intention in enacting section 26(c) of the 1939 Act. That is to say, section 32(1)(c) does apply to claims for recovery of payments made under a mistake of law, but on the basis that where the mistake arises from the retrospective effect of a later court decision the mistake is to be taken to have been capable of discovery when the prospect that the law would be changed was sufficiently well developed. In this way, Lord Reed and Lord Hodge seek to develop a new argument, not set out in Kleinwort Benson, why the word “mistake” in the critical phrase in section 26(c) of the 1939 Act and section 32(1)(c) of the 1980 Act should be interpreted to cover claims based on mistake of law as well those based on mistake of fact, albeit that was the only type of claim to which these provisions were directed when enacted.

We do not agree that this is the correct interpretation of section 26(c) of the 1939 Act and section 32(1)(c) of the 1980 Act. Ingenious though the reasoning is to square the concept of discoverability of a mistake with the effect produced by the retrospective effect of a change in the law, in our opinion it still produces a result which is seriously at odds with the policy and intent of those provisions. Further, it seems to us, with respect, that the argument presented in support of this interpretation (see para 236(3) above) is excessively linguistic. The ordinary meaning of words, to which Lord Reed and Lord Hodge make appeal, is an inadequate tool for this process of construction, when the words in question cannot possibly have had the meaning now contended for when enacted. Instead, as set out above, the focus should be on purposive construction of the provision, arrived at in light of consideration of the policy of the limitation statutes in which it appeared and the object Parliament sought to pursue in enacting the particular provision in that context.

Although in Kleinwort Benson [1999] 2 AC 349 the House of Lords decided that for the purposes of the law of unjust enrichment there was no sufficient difference between mistake of fact and mistake of law to justify distinguishing them as the basis for recovery of money paid, that was a matter of judicial policy in the development of the common law. It did not reflect any legislative policy adopted by or attributable to Parliament relating to the Limitation Acts. When Parliament enacted section 26(c) of the 1939 Act and section 32(1)(c) of the 1980 Act it addressed the law as it stood at the time, in which the only cases in which recovery was possible was where the payment had been made on the basis of a mistake of fact. Parliament has never addressed the distinct and difficult policy issues which arise in the context of these provisions when one moves from recovery in the case of mistake of fact to recovery in the case of mistake of law, and in our view it is not possible to assume that its policy in enacting those provisions covered the latter type of case.

There are three striking features of the latter class of case to which, in our view, Lord Reed and Lord Hodge do not give sufficient weight. First, any application of section 32(1)(c) to mistakes of law which include judicial rewriting of the law is bound to risk opening up very old claims indeed. This was not possible prior to Kleinwort Benson, because the claim would have had to have been based on mistake of fact.

Although section 32(1)(c) involves some departure from a clear and certain limitation cut-off of six years in that sort of case, this is a very modest extension the potential for application of which is likely to narrow considerably as time goes by and the underlying true facts come to light. The opposite is true in the case of mistakes of law identified by retrospective application of later judicial decisions which change the law. Particularly in the field of the common law, the scope for the law to be changed by judicial decision increases as time goes by and the law is perceived as no longer reflecting social values or legal policy, a gradual head of steam builds up among judges and commentators calling for it to be changed and then the courts eventually respond. No purposive interpretation of section 26(c) of the 1939 Act and section 32(1)(c) of the 1980 Act or application of the “always speaking” doctrine could lead to the conclusion that Parliament intended, by a new provision in a Limitation Act, to open up such stale claims.

One example serves to illustrate the point. In 2018, in Rock Advertising Ltd v MWB Business Exchange Centres Ltd [2018] UKSC 24; [2019] AC 119, this court was asked to depart from the much-criticised decision of the House of Lords in Foakes v Beer (1884) 9 App Cas 605. The doctrine from which the court was asked to depart had only reluctantly been affirmed out of loyalty to the Court of Appeal in Pinnel’s case (1602) 5 Co Rep 117a. If the court had departed from Foakes v Beer, this would have undermined settled payments made for over 130 years, or 416 years if Pinnel’s case had also been overruled. But claimants for repayment (or their estates if individuals) would have had until 2024 to bring their claims, with no limitation defence to impede them. In Kleinwort Benson [1999] 2 AC 349 Lord Lloyd gave a number of other examples (p 393). The decision in that case to depart from the rule of law laid down in 1802 in Bilbie v Lumley is itself a further example.

Lord Reed and Lord Hodge seek to meet this point by saying that mistakes of fact might emerge after a long period of time, and give the example (at para 228) of In re Baronetcy of Pringle of Stichill [2016] UKPC 16; [2016] 1 WLR 2870. It was not a case about a common law claim in mistake nor about the interpretation of section 32(1)(c). We would make three points about this example. (i) The case arose in unusual circumstances and is one of the most extreme forms of mistake of fact case one can imagine. The more usual type of mistake of fact case is one where the mistake is liable to emerge after a much shorter period, by contrast with what happens in relation to mistake of law: para 290 above. (ii) It seems to us that the reasoning of Lord Hodge for the Board of the Privy Council in this case tends to demonstrate that Parliament cannot have intended section 32(1)(c) to apply in the case of mistakes of law. It involved a very late challenge to entitlement to the honour of a baronetcy in which the modern discovery of DNA and the use of DNA testing to determine parentage had the effect of unsettling the operation of various rules of law which previously would have made such a challenge very difficult indeed after the baronetcy had been held by an individual for a very long period. The particular form of claim was not one to which any limitation period had been enacted by Parliament, either in English law (para 39) or Scots law (paras 50-61). The policy concern at potential disruption of property transactions in other cases was so obvious that the Board felt that it should call attention to the lacuna in the limitation statutes (para 85). Yet the reasoning of Lord Reed and Lord Hodge in the present case would have the effect of exacerbating this problem by extending the application of section 32(1)(c) to cover mistakes of law. (iii) Most importantly, if one imagined a relevant common law claim arising from facts similar to those in the Stichill Baronetcy case, although it would be an unusual case it would fall squarely within the meaning which Parliament intended section 32(1)(c) to have, as involving a mistake of fact. But the question in the current case is different. It is whether a mistake of law, which has arisen only because of a change in the law long after a relevant payment was made, falls within the intention of Parliament in the legislation it enacted, even though Parliament could never have contemplated that it did. In our view, for the reasons we have given, it is not possible to draw such an inference. The unusual circumstances of the Stichill Baronetcy case were such as to unsettle only one transaction, the inheritance of the baronetcy, and one small set of people were interested in that question. But the extension of section 32(1)(c) to cover payments made under mistake of law will tend to unsettle whole classes of transactions, such as were governed by rules of law of general application.

Secondly, the phenomenon of judicial decisions changing the law occurs across a wide range of cases. As was pointed out by Lord Browne-Wilkinson and Lord Lloyd in Kleinwort Benson [1999] 2 AC 349 (at pp 363-364 and 393-394, respectively), it extends from situations in which rules of the common law are derived from practice and the understanding of lawyers skilled in the field, through decisions of lower courts being overturned by superior courts (a very common feature of the legal system), to this court deciding in comparatively rare cases to re-open and overturn previous decisions of itself or the House of Lords. The law is often settled by a decision of the Court of Appeal, or even at first instance, as was thought to have happened in relation to floating charges in Siebe Gorman & Co Ltd v Barclays Bank Ltd [1979] 2 Lloyd’s Rep 142: see In re Spectrum Plus Ltd (in liquidation) [2005] 2 AC 680, paras 1-17 per Lord Nicholls. If this court (or the Court of Appeal, in the case of a first instance decision) concludes that the earlier decision is wrong, then it will overrule it, and with retrospective effect, with little scope for considering the risks to the security of settled transactions. In Kleinwort Benson the House of Lords departed from law which had been settled by a lower court in Bilbie v Lumley. Again, it seems to us that neither a purposive interpretation of the relevant provisions nor the application of the “always speaking” doctrine could lead to the conclusion that Parliament intended that such uncertainty and potential for undermining the security of transactions should be introduced into the law across such a wide range of cases, least of all in a Limitation Act, the general object of which is to achieve the opposite effect: see para 271 above. The extent of the contradiction between the uncertainty created by alterations in the law made by the courts and the policy of the Limitation Acts was already great in 1939, since common law rules established by professional practice or decisions of courts up to and including the Court of Appeal could always be changed. The extent of the contradiction has been greatly increased with the 1966 Practice Statement (Practice Statement (Judicial Precedent) [1966] 1 WLR 1234), which has the effect that even rules established by the House of Lords or the Supreme Court can now be changed with retrospective effect. Therefore, in our view, the reasons why section 26(c) of the 1939 Act and section 32(1)(c) of the 1980 Act cannot be construed as applying to mistakes of law have become even stronger than they were in 1939.

Lord Reed and Lord Hodge say that the 1966 Practice Statement was not in the mind of Parliament in 1939 and suggest that reference to it is therefore inapposite: para 238. We respectfully doubt that. The court has to infer what should be regarded as the true intention of Parliament in enacting section 26(c) of the 1939 Act (and then re-enacting that provision in section 32(1)(c) of the 1980 Act) as to whether it should apply in new circumstances which Parliament did not have in its contemplation in 1939. To do that, the court has to take account of the entire impact of the new circumstances on the policy underlying Parliament’s choice to enact section 26(c) in the terms it did. It seems to us that the major transformation of the legal landscape produced by the 1966 Practice Statement and the major change in doctrine in Kleinwort Benson both have to be brought into consideration to address that question.

We note that Lord Reed and Lord Hodge at para 241 refer in similar fashion (correctly in our view, as a matter of principle) to the change of position defence, which also developed after 1939. So far as that defence is concerned, we do not consider that it provides an adequate answer to the policy objections to treating section 32(1)(c) as covering mistake of law. We cannot see how the merits of an alleged change of position could be examined over intervening decades or centuries. Moreover, many public authority defendants, including the Revenue, may be unlikely in practice to be able to rely on it. The majority in Kleinwort Benson likewise referred to the new defence of change of position as a possible answer to, or at least amelioration of, the problems of injustice and uncertainty to which their interpretation of section 32(1)(c) gave rise. However, experience, including claims in the field of tax as affected by EU law for recovery of payments made under mistakes of law dating back to 1973, has shown that this hope has not been realised.

Thirdly, as is clear from the LR Committee’s Report, section 26(c) of the 1939 Act (and now section 32(1)(c) of the 1980 Act) enacts what was previously largely an equitable principle, namely that relief should be available in the occasional case where the particular circumstances of the claimant would otherwise render the rigid application of the law unconscionable. Thus it would be unconscionable for claimants to have time running against them when, either because they were labouring under a fraud or, because of a mistake as to the facts, they were unaware of their cause of action. The occasional claimant thus disadvantaged could rely upon the exceptional extension of the running of time. But if section 32 is applied to extend time where there has been a retrospective judicial change in the law, then every potential claimant is benefitted by the exception. By definition every potential claimant was suffering from the same deemed mistake when making the relevant payment. In the make-believe world view necessitated by the need to give retrospective effect to judicial law-making, no one knew what the law then really was on the point in issue. It seems to us that to extend the application of these provisions to this class of case goes well beyond the narrow equitable principle which was intended to apply. As we have noted, the equitable principle grew from the idea of the unconscionability of a defendant relying on his own fraud and was given a modest extension to cover individual cases of mistake of fact. By contrast, a claim based on a deemed mistake which has arisen only because of a retrospective change in the law and which affects all cases within the purview of the rule of law which is overruled lies very far indeed from any concept which could be grounded in the equitable principle which the LR Committee identified and which Parliament intended to apply to common law claims.

In our view, the approach of Lord Brown in Deutsche Morgan Grenfell [2007] 1 AC 558, which Lord Reed and Lord Hodge endorse, does not provide an answer to these objections. Plainly, there may be a very long period when a rule of law is taken to be established by professional practice or judicial decisions before the threshold of discoverability proposed by Lord Brown to suggest it might be wrong is crossed. Accordingly, even though Lord Brown’s approach ameliorates to some degree the conflict between section 32(1)(c) and the basic object of the Limitation Acts which arises when that provision is taken to apply to mistakes of law, it does so only to a very limited and inadequate extent.

There are several additional reasons which reinforce our view that it is not plausible to identify Lord Brown’s interpretation of section 32(1)(c) as representing the intention of Parliament in any genuine sense, including the extended sense to which the “always speaking” doctrine refers:

Our conclusion regarding the proper interpretation of section 32(1)(c) would open the door to a departure from Kleinwort Benson [1999] 2 AC 349 on that issue under the 1966 Practice Statement. In our view, it would be appropriate for the Practice Statement to be applied to restore the proper interpretation of section 32(1)(c) which we consider Parliament intended, as set out above. We express our views shortly, as we are in a minority so far as concerns reversing Kleinwort Benson by construing “mistake” in section 32(1)(c) to mean only a mistake of fact, in accordance with the law as it stood in 1939 and 1980.

In our judgment, for the reasons we have set out, the decision in Kleinwort Benson that section 32(1)(c) applies to mistakes of law was wrong for reasons of much greater solidity and significance than a mere intellectual difference of opinion. The reasoning in Kleinwort Benson is in our respectful view gravely undermined by an underlying view of the equitable antecedents to what is now section 32(1) which cannot be squared with previous authority and by an apparent failure to weigh in the balance how serious a departure from the overall policy of the 1980 Act is involved in a conclusion that section 32(1)(c) does apply to claims based upon a mistake of law. Furthermore the readiness of the House of Lords in both Kleinwort Benson and Deutsche Morgan Grenfell to acknowledge common law claims based on mistake of law where settled law had been changed with retrospective effect, and that section 32(1)(c) applied to such claims, was heavily based upon a hope that Parliament would remedy the unsatisfactory consequences, a hope which has now clearly been shown to have been misplaced, save to a limited degree in relation to tax.

An important consideration underlying the Practice Statement is that where possible past transactions should not be rendered uncertain or insecure. As the Practice Statement says, in deciding whether it is right to depart from a previous decision of the House of Lords (or, now, this court) the court “will bear in mind the danger of disturbing retrospectively the basis on which contracts, settlements of property and fiscal arrangements have been entered into …”. The greater the degree of such disruption, the less will it be regarded as acceptable for the court to change the law for the future. It is possible that some past transactions might be unsettled by changing the interpretation of section 32(1)(c) adopted in Kleinwort Benson; however, for the reasons we have explained, the application of section 32(1)(c) to cases of mistake of law will unsettle past transactions and will generate such uncertainty to a very much greater degree. Changes in the law produced by higher courts reversing decisions of lower courts or correcting professional practice are commonplace. The period of time before such a decision is produced, and in which parties will have entered transactions on the basis of the previous understanding of the law, may be very long.

Further, this open-ended prospect of unravelling past transactions without limit of time is likely to act as a very serious and chilling constraint upon any departure from an earlier decision by the House of Lords or this court under the 1966 Practice Statement. The older the decision from which departure is being considered, the greater the peril to settled transactions, and the greater the difficulty which this court will face in assessing whether that peril is sufficient to prohibit an otherwise worthwhile change. The speeches of Lord Browne-Wilkinson and Lord Lloyd in Kleinwort Benson [1999] 2 AC 349 illustrate this point. Absent their concerns about the absence of an effective limitation cut-off, they would both have wished to support the substantive change in the law of unjust enrichment produced by the majority. The new circumstances in which the proper interpretation of section 32(1)(c) falls to be assessed as a statutory provision which is “always speaking” include not just the change in the law in Kleinwort Benson but the change in the practice of the House of Lords effected by the 1966 Practice Statement. Parliament cannot have intended that section 32(1)(c) should have the practical effect of acting as a serious impediment to desirable judicial modernisation of the common law pursuant to the 1966 Practice Statement.

Accordingly, it is our view that correction of the wrong turn taken in Kleinwort Benson regarding the true interpretation of section 32(1)(c) is justified pursuant to the Practice Statement. It would tend to reduce, rather than promote, insecurity of transactions across time. It would also secure the ability of this court to review and amend substantive legal doctrine in the interests of promoting doctrinal coherence and keeping the law broadly in line with changing social expectations and values.

Finally, since our view regarding the proper interpretation of section 32(1)(c) is not accepted by Lord Reed and Lord Hodge and the majority in the court, we address the position which arises under the Practice Statement if their interpretation of section 32(1)(c) prevails, as it does. On their interpretation, there is still considerable scope for uncertainty in the law to arise and unsettle transactions dating far back in time. Their interpretation, following Lord Brown’s approach in Deutsche Morgan Grenfell [2007] 1 AC 558, is also productive of a degree of uncertainty because of the test of discoverability of a mistake which they say should apply. To that extent, therefore, it seems to us that the argument for applying the Practice Statement in relation to both Kleinwort Benson and Deutsche Morgan Grenfell is weakened. Nonetheless, we consider that their interpretative approach better reflects the legislative purpose in the context of the Limitation Act of securing a degree of certainty in relation to past transactions than does that of the majority in Deutsche Morgan Grenfell [2007] 1 AC 558 and the approach which the Appellate Committee in Kleinwort Benson assumed would apply. Therefore, on the footing that the interpretation of section 32(1)(c) preferred by Lord Reed and Lord Hodge must be accepted, we agree that it is appropriate to apply the Practice Statement in relation to those decisions and in favour of now adopting their interpretation.