Michaelmas Term [2020] UKSC 38 On appeal from: [2020] EWCA Civ 574 JUDGMENT Enka Insaat Ve Sanayi AS (Respondent) v OOO Insurance Company Chubb (Appellant) before Lord Kerr Lord Sales Lord Hamblen Lord Leggatt Lord Burrows JUDGMENT GIVEN ON 9 October 2020 Heard on 27 and 28 July 2020 Appellant Respondent David Bailey QC Robin Dicker QC Toby Landau QC David Joseph QC Marcus Mander Niranjan Venkatesan Clara Benn (Instructed by Kennedys Law LLP (London) ) (Instructed by Shearman & Sterling LLP (London) ) Page 2 LORD HAMBLEN AND LORD LEGGATT: ( with whom Lord Kerr agrees) I. Introduction 1. Where an international commercial contract contains an agreement to resolve disputes by arbitration, at least three systems of national law are engaged when a dispute occurs. They are: the law governing the substance of the dispute; the law governing the a greement to arbitrate; and the law governing the arbitration process. The law governing the substance of the dispute is generally the law applicable to the contract from which the dispute has arisen. The law governing the arbitration process (sometimes referred to as the “curial law”) is generally the law of the “seat” of the arbitration, which is usually the place chosen for the arbitration in the arbitration agreement . These two systems of law may differ from each other. Each may also differ from the law which governs the validity and scope of the arbitration agreement. 2. The central issue on this appeal concerns which system of national law governs the validity and scope of the arbitration agreement when the law applicable to the contract containing it differs from the law of the seat of the arbitration. 3. This is an issue which has long divided courts and commentators, both in this country and internationally. On one side there are those who say that the law that governs a contract should generally also govern an arbitration agreement which, though separable, forms part of that contract. On the other side there are those who say that the law of the chosen seat of the arbitration should also generally govern the arbitration agreement. There have been Court of A ppeal decisions falling on either side of this divide: Sulamérica Cia Nacional de Seguros SA v Enesa Enge nharia SA [2012] EWCA Civ 638; [2013] 1 WLR 102 and C v D [2007] EWCA Civ 1282; [2008] Bus LR 843. 4. In its judgment in the present case [2020] EWCA Civ 574, the Court of Appeal considered that “the time has come to seek to impose some order and clarity on this area of the law” (para 89) and held that , unless there has been an express choice of the law that is to govern the arbitration agreement, the general rule should be that the arbitration agreement is governed by the law of the seat , as a matter of implied choice, subject only to any particular fe atures of the case demonstrating powerful reasons to the contrary (para 91). Page 3 5. On this appeal the appellant argues that this conclusion is heterodox and wrong and that the correct approach is that , in the absence of strong indications to the contrary, a choice of law for the contract is a choice of th at law to govern the arbitration agreement. The appellant contends that i n the present case the parties have chosen Russian law to govern the construction contract between them and that the implication that they intended the arbitration agreement included in that contract to be governed by Russian law is not displaced by the ir choice of London as the seat of arbitration. 6. If that issue is decided in its favour, the appellant goes on to argue that the Court of Appea l was wrong to grant an injunction to restrain it from pursuing proceedings in Russia in alleged breach of the arbitration agreement. The appellant’s case is that, because the arbitration agreement is governed by Russian law, the Russian courts are best pl aced to decide whether or not the arbitration agreement applies to the claim which the appellant has brought against the respondent in Russia and that, as a matter of comity or discretion, the English courts ought not to interfere with those proceedings by granting an anti -suit injunction. II. Factual background (i) The construction contract 7. On 1 February 2016 a power plant situated at Berezovskaya in Russia was severely damaged by fire. The appellant (“ Chubb Russia”) is a Russian insurance company which ha d insured the owner of the p ower plant, a company now named PJSC Unipro (“ Unipro”), against such damage. Chubb Russia is part of the Chubb Group, which is the world’s largest publicly traded property and casualty insurer. 8. The company responsible for the design and construction of the power plant under a contract made with Unipro in May 2011 was a Russian company called CJSC Energoproekt. The respondent (“ Enka”) was engaged by Energoproekt as one of many sub- contractors involved in the construction project. Enka is a global engineering and construction company incorporated and based in Turkey with a substantial presence and history of operations in Russia, amongst other countries. 9. The contract between Energoproekt and Enka dated 27 June 2012 (“the c onstruction contract ”) is a substantial document running to 97 pages, with around 400 pages of attachments. It was executed in parallel Russian and English versions (though it provides that the Russian language version has precedence). Page 4 10. The construction contract contains , in article 50 , a dispute resolution clause in these terms : “ Resolution of disputes 50.1. The Parties undertake to make in good faith every reasonable effort to resolve any dispute or disagreement arising from or in connection with this Agreement (including disputes regarding validity of this agreement and the fact of its conclusion (hereinafte r - ‘ Dispute ’) by means of negotiations between themselves. In the event of the failure to resolve any Dispute pursuant to this a rticle within 10 (ten) days from the date that either Party sends a Notification to the opposite Party containing an indication of the given Dispute (the given period may be extended by mutual consent of the Parties) any Party may, by giving written notice, cause the matter to be referred to a meeting between the senior managements of the Contractor and Customer (in the case of th e Contractor senior management sh all be understood as a member of the executive board or above, in the case of Customer, senior management shall be understood as general directors of their respective companies). The parties may invite the End Customer to s uch Senior Management Meeting. Such meeting sh all be held within fourteen (14) calendar days following the giving of a notice. If the matter is not resolved within twenty (20) calendar days after the date of the notice referring the matter to appropriate h igher management or such later date as may be unanimously agreed upon, the Dispute shall be referred to international arbitration as follows: • the Dispute shall be finally settled under the Rules of Arbitration of the International Chamber of Commerce , • the Dispute shall be settled by three arbitrators appointed in accordance with these Rules, • the arbitration shall be conducted in the English language, and • the place of arbitration shall be London, England. Page 5 50.2. Unless otherwise explicitly stipulated in t his Agreement , the existence of any Dispute shall not give the Contractor the right to suspend Work. 50.3. Not used. 50.4. Not used. 50.5. All other documentation such as financial documentation and cover documents for it must be presented in Russian.” 11. On 21 May 2014 Energoproekt transferred its rights and obligations under the construction contract to Unipro pursuant to an assignment agreement made between Energoproekt, Unipro and Enka. By clause 7.5 of that agreement, the parties agreed that disputes between Unipro and Enka were to be finally and exclusively resolved by arbitration in accordance with the provisions of article 50.1 of the construction c ontract. 12. After the fire in February 2016 Chubb Russia paid 26.1 billion roubles ( approximately US$400m ) to Unipro under its property insurance policy and thereby became subrogated to any rights of Unipro to claim compensation from third parties for the damage caused by the fire . (ii) The Russian proceedings 13. On 25 May 2019 Chubb Russia filed a claim in the Moscow Arbitrazh (ie commercial) Court against Enka and ten other defendants whom it claimed were jointly liable for the damage caused by the fire. Chubb Russia was required by the Moscow court to provide further details of its claims, following which the claims were accepted by the court on 3 September 2019. 14. On 17 September 2019 Enka filed a motion in the Russian p roceedings to have Chubb Russia’s claim against it dismissed (or “left without consideration”) pursuant to art icle 148(5) of the Arbitrazh Procedure Code , which is intended to give effect to Russia’s obligations under a rticle II(3) of the Convention on the Recognition and Enforcement of Foreign Arbitral Awards 1958 (“ the New York Convention”) to refer to arbitration parties wh o have agreed to submit to arbitration a dispute of which a court of a contracting state is seized . Enka argued that the claim against it fell within the scope of the arbitration agreement contained in article 50.1 Page 6 of the construction c ontract and ought therefore to be resolved, not by the Russian courts, but by an arbitration conducted in accordance with that provision in London. The Moscow court decided to deal with Enka’s motion at the same time as the merits of Chubb Russia’s claim s at a hearing fixed for 2 2 January 2020. 15. Following th at hearing , which continued on two later dates, on 18 March 2020 the judge in the Russian proceedings announced her decisions (a) not to grant Enka’s motion to refer the claim against it to arbitration and (b) to dismiss Chubb Russia’s claims against all the defendants on the merits. The reasons for those decisions were given in a written judgment handed down on 6 May 2020. 16. Chubb Russia and Enka have both filed appeals in the Russian proceedings ( in relati on to the decision on the merits and the decision to refuse Enka’s a pplication, respectively). (iii) The English proceedings 17. Meanwhile, Enka had on 16 September 2019 brought an arbitration claim in the Commercial Court in London seeking an anti- suit injunction to restrain Chubb Russia from further pursui ng the Russian proceedings against Enka on the ground that this was a breach of the arbitration agreement in article 50.1 of the construction c ontract . Enka also sought injunctions against other members of t he Chubb Group said to be “caught up” in Chubb Russia’s breach of the arbitration agreement, namely Chubb UK Lt d, Chubb European Group SE (“ Chubb Europe”) and the ultimate parent company of the Chubb Group which is incorporated in Switzerland . 18. On 15 October 2019 Carr J declined to grant an interim anti -suit injunction but gave directions for an expedited trial. The trial took place on 11 and 12 December 2019 before Andrew Baker J. He gave judgment on 20 December 2019, dismissing Enka’s claims against all the defendants . His primary reason for doing so was that he considered the appropriate forum to decide whether Chubb Russia’s claim against Enka falls within the arbitration agreement to be the Moscow Arbitrazh Court and not the English Commercial Court . 19. Enka applied to the Court of Appeal for permission to appeal from this decision as it applied to Chubb Russia (alone). The application was granted on 6 February 2020 and the appeal was heard on 7 and 8 April 2020. On 29 April 2020 the Court of Appeal (Flaux, Males and Popplewell LJJ) allowed Enka’s appeal and issued an anti-suit injunction restraining Chubb Russia from continuing the Russian proceedings . Page 7 (iv) The arbitration proceedings 20. On 10 January 2020 Enka gave notice to Chubb Russia and Chubb Europe of a “Dispute” under article 50 of the construction contract. This was followed on 11 March 2020 by a request for arbitration filed with the International Chamber of Commerce (“ICC”) in which Enka sought a declaration that Chubb Russia ’s claims in the Russian court fall within the scope of the arbitration agreement and damages. 21. On 22 May 2020 Chubb Russia and Chubb Europe filed their answer to the r equest for a rbitration in wh ich they challenged the jurisdiction of the arbitrators and denied that Enka is entitled to any of the relief claimed. 22. On 12 June 2020 the ICC notified the parties of the appointment of Mr Michael Brindle QC as president of the arbitral tribunal. The other members of the tribunal are Lord Hoffmann, nominated by Enka, and Lord Mance, nominated by Chubb Russia and Chubb Europe (without prejudice to the ir objection s to the jurisdiction of the tribun al). (v) This appeal 23. On 26 May 2020 Chubb Russia applied to the Supreme C ourt for permission to appeal from the decision of the Court of Appeal. On 5 June 2020 this court granted permission to appeal and also stayed the anti -suit injunction upon Chubb Russia giving suitable undertakings to protect Enka’s position pending the outcome of the appeal. The appeal was expedited and heard over two days on 27 and 28 July 2020. 24. It is a striking feature of the English proceedings that the trial, the appeal to the Court of Appeal and the appeal to the Supreme Court have all been heard in just over seven months. This is a vivid demonstration of the speed with which the English courts can act when the urgency of a matter requires it . III. The English conflict of laws rules (i) The Rome I Regulation 25. Where a court of England and Wale s has to decide which system of national law governs a contract, the court must normally apply the provisions of the “Rome I Regulation” (a shorthand for Regulation (EC) No 593/2008 of the European Parliament and of the Council of 17 June 2008 on the law a pplicable to contractual Page 8 obligations ). By a rticle 1(1), t he Rome I Regulation appl ies, in situations involving a conflict of laws, to contractual obligations in civil and commercial matters. Article 1(2)(e), however, excludes from its scope “arbitration ag reements and agreements on the choice of court”. 26. Pursuant to article 3, a contract to which the Rome I Regulation applies is governed by the law chosen by the parties, where the choice is made expressly or clearly demonstrated by the terms of the contract or the circumstances of the case. In determining whether the parties have made a choice of law, the court should adopt a b road Regulation -based approach, not constrained by national rules of contractual interpretation: see eg Dicey, Morris & Collins on The Conflict of Laws , 15th ed (2012), para 32 -048. Article 4 contains rules for determining the law applicable to the contrac t to the extent that no such choice has been made. Article 4(1) sets out presumptions or prima facie rules that apply in relation to particular types of contract. However, where it is clear from the circumstances of the case that the contract is manifestly more closely connected with another country, or where none of the prima facie rules applies, articles 4(3) and 4(4) respectively provide for the contract to be governed by the law of the country with which it is most closely connected. (ii) The common law rules 27. Because t he Rome I Regulation does not apply to a rbitration agreements , an English court which has to decide which system of law governs the validity, scope or interpretation of an arbitration agreement must apply the rules developed by the common law for determining the law governing contractual obligations. Those rules are that a contract (or relevant part of it) is governed by: (i) the law expressly or impliedly chosen by the parties; or (ii) in the absence of such choice, the law with which it is most closely connected: see eg Dicey, Morris & Collins on The Conflict of La ws, 15th ed (2012), rule 64(1). 28. In view of the similarity between the common law rules and the rules provided by the Rome I Regulation , cases in which the two regimes would yield different results are likely to be rare. But in principle, where an English c ourt has to determine which law governs an arbitration agreement incorporated in a contract, it is the common law rules alone which - because of the exclusion of arbitration agreements from the scope of the Rome I Regulation by article 1(2)(e) - the court must apply. Page 9 (iii) Party choice 29. The starting point at common law (as under the Rome I Regulation) is that contracting parties are free to choose the system of law which is to govern their contract, provided only that their choice is not contrary to public p olicy. The court must therefore construe the contract t o see whether the parties have agree d on a choice of law to govern it . As Lord Diplock explained in Cie Tunisienne de Navigation SA v Cie d ’Armement Maritime SA [1971] AC 572 , 603 : “ The first stage, therefore, when any question arises between parties to a contract as to the proper law applicable to it, is to determine whether the parties intended by their contract to exercise any choice at all and, if they did, to determine what was the system of law which they selected. In determining this the English court applies the ordinary rules of English law relating to the construction of contracts.” 30. The exclusion of arbitration agreements from the scope of the Rome I Regulation by article 1(2)(e) does not prevent an arbitration clause from being taken into consider ation for the purposes of a rticle 3 in determining whether there has been a choice of the law applicable to other parts of the contract, as noted in Giuliano and Lagarde, Council Report on the Convention on the law applicable to contractual obligations (OJ EU No C 282- 1) at p 12 . By the same token, the fact that other parts of the contract are within the scope of the Rome I Regulation does not prevent them from being taken into consideration in determining in accordance with the English common law rules of construction whether the parties have agreed on a choice of law to govern the arbitration clause. Like any question of contractual interpretation, this is a unitary exercise which requires the court to construe the contract, including the arbitration clause, as a whole. (iv) Law of the forum 31. Where an English court has to decide whether a contract which is said to be governed by a foreign system of law is valid, the court applies the “putativ e applicable law”, in other words the law which would govern the contract if it were validly concluded. At the prior stage, however, of determining what is the applicable law or putative applicable law of the contract, all the leading authorities proceed o n the basis that it is English rules of law which apply, as stated by Lord Diplock in the passage quoted above. In the Tunisienne case, for example, a contract for the transport of oil in several shipments contained a provision (clause 13) that the contract “shall be governed by the laws of the flag of the vessels carrying the goods …”. The first question which the House of Lords ha d to decide was whether, in the Page 10 circumstances of the case which included the fact that vessels flying different flags were used to ship the oil, this clause conveyed a choice of French law to govern the contract, as the shipowners argued. To answer that qu estion the House did not apply the rules of French law governing the interpretation of contracts, but (only) those of English law. 32. The same approach was adopted in Whitworth Street Estates (Manchester) Ltd v James Miller & Partners Ltd [1970] AC 583, where the House of Lords held that subsequent conduct of the parties could not be looked at to construe a contract in order to decide whether it was intended to be governed by English (rather than Scottish) law. The exclusion of subsequent conduct as an aid to interpretation is a consequence of the objective principle of interpretation in English law, which searches not for what the parties subjectively thought or intended the effect of their contract to be but for what reasonable people in their position would be understood to have meant by the language used. Although in the Whitworth Street Estates case English law was one putative applicable law of the contract, there is no suggestion in the speeches that this was the basis for applying English principles of c ontractual interpretation. 33. In our view, it is both consistent with authority and sound in principle to apply English law as the law of the forum to ascertain whether the parties have agreed on the law which is to govern their contract (and, if not, what la w governs it in the absence of agreement). To apply any other law for this purpose would introduce an additional layer of complexity into the conflict of laws analysis without any clear justification and could produce odd or inconsistent results. As the au thors of Dicey, Morris & Collins on The Conflict of Laws, 15th ed (2012) observe, at para 32-036 , by reference to a case in which subsequent conduct was taken into account to construe a contract found to be governed by Chilean law because it was admissible under that law: “But it would be very odd if when a question arose as to whether a contract was governed by English law or Chilean law, subsequent conduct would not be taken into account in determining whether a choice of English law could be inferred, but it could be taken in to account in determining whether Chilean law applied.” 34. The Court of Appeal in the present case asserted (although without explanation) that, in construing the contract to determine whether a choice of governing law applies to an arbitration agreement with in it, the court should apply the principles of construction of the main contract law if different from English law (see paras 90 and 105(2) of the judgment). We do not consider this to be correct. As we have indicated, the proper approach in determining w hether there has been a Page 11 choice of law is to apply English law as the law of the forum. Where the question is whether there has been a choice of the law applicable to an arbitration clause, the relevant English law rules are the common law rules which requi re the court to interpret the contract as a whole applying the ordinary English rules of contractual interpretation. The main contract law, if different, has no part to play in the analysis. (v) Express or implied choice 35. Many of the cases applying the comm on law rules distinguish between a choice of law which is “express” or “implied”. Article 3 of the Rome I Regulation draws a similar distinction in referring to a choice which is “ made expressly or clearly demonstrated ”. The terminology is useful in reflec ting the fact that an agreement on a choice of law to govern a contract, like any contractual term, may be explicitly articulated or may be a matter of necessary implication or inference from other terms of the contract and the surrounding circumstances. T he distinction, however, is not a sharp one: language may be more or less explicit and the extent to which a contractual term is spelt out in so many words or requires a process of inference to identify it is a matter of degree. Determining whether the parties have agreed on a choice of law to govern their contract is in every case a question of interpretation. It is also important to keep in mind that whether a choice is described as express or implied is not a distinction on which any legal consequence tu rns. An implied choice is still a choice which is just as effective as a choice made expressly. (vi) The default rule 36. Where a choice of law cannot be identified by interpreting the contract, the approach of the common law was at one time to presume that th e parties must nevertheless have intended their contract to be governed by some particular system of national law and to impute a relevant intention to them. This is reflected, for example, in the first edition of Dicey’s treatise on the conflict of laws, which defined the law governing a contract as “the law or laws to which the parties intended, or may fairly be presumed to have intended, to submit themselves”: Dicey, A Digest on the Law of England with reference to the Conflict of Laws , 1st ed (1896), ru le 143. In the second half of the 20th century, however, the test of presumed intention came gradually to be superseded by an acknowledgement that at this stage of the analysis the court is no longer concerned with intention at all and is applying a positi ve rule of law, with the rule being that the contract is governed by the system of law with which it has its “closest and most real connection”: see Dicey, Morris & Collins on The Conflict of Laws , 15th ed (2012), paras 32 -006 - 32- 007; Hellenic Steel Co v Svolamar Shipping Co Ltd (The Komninos S) [ 1991] 1 Lloyd’s Rep 370, 37 4 (Bingham LJ) . Lord Diplock stated the modern position clearly in the Tunisienne case , at p p 603 -604 : Page 12 “ If, applying these rules [sc the ordinary rules of English law relating to the construction of contracts ], the court reaches the conclusion that the parties did not intend to exercise any choice of proper law, or is unable to identify what their choice was, it becomes necessary for the court to proceed to the second stage, of determi ning itself what is the proper law applicable. In doing so, the court applies the English rule of the conflict of laws … that the proper law is that system of law with which the transaction has its closest and most real connection: Bonython v Commonwealth of Australia [1951] AC 201, 219. My Lords, this is applied as a positive rule of English law. It is applied not because it is the choice of the parties themselves but because they never intended to exercise their liberty to make a choice or, if they did, t hey have failed to make their choice clear.” 37. Whether the parties have agreed on a choice of law is a matter which inevitably may sometimes give rise to differences of opinion. In the Tunisienne case three members of the House of Lords appellate committee (Lord Morris of Borth -y - Gest, Viscount Dilhorne and Lord Diplock) held that clause 13 (quoted earlier) was in its context to be construed as an agreement that French law was to govern the contract. The other two members of the committee (Lord Reid and Lord Wilberforce) did not consider that the clause could be so construed but still concluded at the second stage of the analysis that French law was the governing law. In Amin Rasheed Shipping Corp n v Kuwait Insurance Co (The Al Wahab) [1984] AC 50, Lord Diplock (with whose speech three of the other law lords agreed) applied the principles he had identified in the Tunisienne case to determine whether an insurance contract was governed by English law or the law of Kuwait . He concluded (at p 62) that on thei r proper construction the provisions of the contract, taken as a whole, “by necessary implication point ineluctably to the conclusion that the intention of the parties was that their mutual rights and obligations under it should be determined in accordance with the English law of marine insurance ”. Lord Wilberforce reached the same result on the basis that English law was the system of law with which the contract had the closest and most real connection. (vii) Splitting the contract 38. English common law (alon g with other legal systems) recognises the possibility that different parts of a contract may be governed by different laws - a concept known in conflict of laws theory as dépeçage. This is also expressly provided for in the Rome I Regulation . Article 3(1) includes the statement: Page 13 “ By their choice the parties can select the law applicable to the whole or to part only of the contract.” 39. There are many English cases in which courts have contemplated that different obligations in the same contract ma y be governed by different laws . The earliest such case to which we were referred was the decision of the Court of Appeal in Jacobs , Marcus & Co v Cr édit Lyonnais (1884 ) 12 QBD 589 . There appear to be few cases, however, in which such a situation has been found to exist (although one such case is Libyan Arab Foreign Bank v Bankers Trust [1989] QB 728, 746-747). No doubt this is because , as Lord MacDermott said in Kahler v Midland Bank Ltd [1950] AC 24 at 42, “the courts of this country will not split the co ntract in this sense readily or without good reason.” It is generally reasonable to assume that parties would intend or expect their contract to be governed by a single system of law. To apply different systems of law to different parts of a contract has t he potential to give rise to inconsistency and uncertainty. This is particularly so where questions about the validity or enforceability of contractual obligations arise. As observed in Dicey, Morris & Collins on The Conflict of Laws , 15th ed (2012) at para 32 -026 : “ Even if different parts of a contract are said to be governed by different laws, it would be highly inconvenient and contrary to principle for such issues as whether the contract is discharged by frustration, or whether the innocent party may terminate or withhold performance on account of the other party’s breach, not to be governed by a single law. ” 40. The assumption that, unless there is good reason to conclude otherwise, all the terms of a contract are governed by the same law applie s to an arbitration clause, as it does to any other clause of a contract. As Mustill J said in Black Clawson International Ltd v Papierwerke Waldhof -Aschaffenburg AG [1981] 2 Lloyd ’s Rep 446 , 456: “In the ordinary way, this [sc the law of the arbitration a greement] would be likely to follow the law of the substantive contract .” An arbitration clause may, however, more readily than other clauses be governed by a different law. One reason for this is that an arbitration clause has a different subject matter a nd purpose from the rest of the contract. It is concerned not with establishing substantive rights and obligations of the parties but with providing a mechanism by which a dispute about such rights and obligations will be resolved. A second reason flows from the principle of separability of the arbitration agreement. This is a cardinal principle of arbitration law, codified in section 7 of the Arbitration Act 1996 . Section 7 provides that, unless otherwise agreed by the parties, “an arbitration agreement wh ich forms or was intended to form part of another agreement … shall not be regarded as invalid, non -existent or ineffective because that other agreement is invalid, or did not come into existence or has become ineffective, and it shall for that purpose be treated as a distinct agreement.” Page 14 41. As counsel for Chubb Russia emphasised, the principle of separability is not a principle that an arbitration agreement is to be treated as a distinct agreement for all purposes but only that it is to be so treated for the purpose of determining its validity or enforceability. That is clear from the words “for that purpose” in section 7 of the 1996 Act. Thus, the separability principle does not require that an arbitration agreement should be treated as a separate agreement for the purpose of determining its governing law. Nevertheless, the principle is relevant to the conflict of laws analysis because it alleviates the difficulty identified by Dicey, Morris & Collins in the passage quoted at para 39 above in treating different parts of a contract as governed by different laws . Where the separability principle is recognised by the putative applicable law of the arbitration agreement , no inconsistency will arise from treating issues such as whether the contract is discharged by frustration, or whether the innocent party may terminate or withhold performance on account of the other party’s breach , or whether the contract has been rescinded for misrepresentation, as gov erned by a different law from the law of the arbitration agreement, as the resolution of those issues will not affect the validity or enforceability of the arbitration agreement. 42. The possibility that an arbitration agreement may be governed by a different system of law from the contract of which it forms part is also implicitly recognised by the exclusion of arbitration agreements from the scope of the Rome I Regulation , with the consequence that the law applicable to an arbitration agreement and the law ap plicable to the rest of the contract must be determined independently by different conflict of laws regimes. IV. Choice of law for the whole contract (i) Significance of a governing law clause 43. It is rare for the law governing an arbitration clause to be sp ecifically identified (either in the arbitration clause itself or elsewhere in the contract). It is common, however, in a contract which has connections with more than one country (or territory with its own legal system) to find a clause specifying the law which is to govern the contract. A typical clause of this kind states: “This Agreement shall be governed by and construed in accordance with the laws of [name of legal system].” Where the contract also contains an arbitration clause, it is natural to inte rpret such a governing law clause, in the absence of good reason to the contrary, as applying to the arbitration clause for the simple reason that the arbitration clause is part of the contract which the parties have agreed is to be governed by the specifi ed system of law. As stated in Redfern and Hunter : Law and Practice of International Commercial Arbitration, 6 th ed (2015) at para 3.12: Page 15 “Since the arbitration clause is only one of many clauses in a contract, it might seem reasonable to assume that the law chosen by the parties to govern the contract will also govern the arbitration clause. If the parties expressly choose a particular law to govern their agreement, why should some other law - which the parties have not chosen - be applied to only one of the clauses in the agreement, simply because it happens to be the arbitration clause?” 44. This approach is supported by other leading commentaries. For example, Merkin on Arbitration Law , Issue 84 (2020), para 7.12, states that: “… e ven if there is no express contractual statement to that effect, a choice of law clause for the entire agreement is likely to be construed as extending to the arbitration clause. There are numerous decisions to this effect … However, that presumption may be ousted in appropriate ci rcumstances …” See also Dicey, Morris & Collins on The Conflicts of Laws , 15th ed (2012) at para 16- 017: “If there is an express choice of law to govern the contract as a whole, the arbitration agreement may also be governed by that law.” (ii) Domestic cas e law 45. There is a considerable body of English case law which proceeds on the assumption that a choice of law for the contract will normally apply to an arbitration clause in the contract . The approach was summarised by Colman J in Sonatrach Petroleum Corpn (BVI) v Ferrell International Ltd [2002] 1 All ER (Comm) 627 at para 32: “Where the substantive contract contains an express choice of law, but the agreement to arbitrate contains no separate express choice of law, the latter agreement will normally be governed by the body of law expressly chosen to govern the substantive contract.” Page 16 46. It has not generally been considered to make any difference in this regard that the arbitration clause provides for arbitration to take place in a different count ry from the country whose law has been chosen to govern the contract. Examples of decisions in which a choice of law clause in the contract has been treated as applying to the arbitration agreement despite the seat of arbitration being in a different juris diction include: Cia Maritima Zorroza SA v Sesostris SAE ( The Marques De Bolarque) [1984] 1 Lloyd’s Rep 652 , 653; Union of India v McDonnell Douglas Corp n [1993] 2 Lloyd’s Rep 48, 49 -50; Sumitomo Heavy Industries Ltd v Oil and Natural Gas Commission [1994] 1 Lloyd’s Rep 45 , 57; Deutz AG v General Electric Co (Thomas J, 14 April 2000) at p 17; Peterson Farms Inc v C&M Farming Ltd [2004] EWHC 121 (Comm); [2004] 1 Lloyd’s Rep 603 , paras 43 -46; Leibinger v Stryker Trauma GmbH [2005] EWHC 690 (Comm) , para 38; and Svenska Petroleum Exploration AB v Government of the Republic of Lithuania [2005] EWHC 2437 (Comm) ; [2006] 1 All ER (Comm) 731 , paras 76 -77 . 47. A different view was expressed in XL Insurance Ltd v Owens Corning [2001] 1 All ER (Comm) 530, a case concern ing a policy of insurance on “Bermuda form” terms which provid e for New York law to govern the policy but for disputes to be determined by arbitration in London. The English court granted an injunction to restrain the insured from pursuing a claim against the insurers in the courts of Delaware. The insured argued that the choice of New York law to govern the policy included the arbitration agreement and that this agreement was invalid under the F ederal Arbitration Act which formed part of New York l aw. Toulson J rejected that argument and concluded that, by stipulating for arbitration in London under the provisions of the 1996 Act, the parties had impliedly chose n English law to govern the arbitration agreement (see p 543b). We will consider his reason ing later in this judgment . 48. In C v D [2007] EWCA Civ 1282; [200 8] Bus LR 843, another case concerning a Bermuda form insurance policy , the Court of Appeal likewise expressed the view (obiter) that the arbitration agreement was governed by English law . In C v D , however, Longmore LJ (with whom the other members of the court agreed) reached this conclusion, not on the basis of implied choice, but on the basis that there was no choice of law for the arbitration agreement so that it w as necessary to identify the law with which it was most closely connected. He considered this to be the law of the place where the parties ha d chosen to arbitrate rather than the law of the insurance contract (paras 25- 26). 49. Many commentaries and authoritie s, including XL Insurance and C v D , were considered by the Court of Appeal in Sulamérica Cia Nacional de Seguros SA v Enesa Enge nharia SA [2012] EWCA Civ 638; [2013] 1 WLR 102. In a judgment with which the other members of the court agreed, Moore -Bick LJ said (at para 11): Page 17 “It is common for parties to make an express choice of law to govern their contract, but unusual for them to make an express choice of the law to govern any arbitration agreement contained within it; and where they have not done so, the natural inference is that they intended the proper law chosen to govern the substantive contract also to govern the agreement to arbitrate.” 50. Moore- Bick LJ expressed reservations about the dicta of Longmore LJ in C v D , noting that the court i n that case did not have the benefit of full citation of authority and that a rule that an arbitration agreement is governed by the law of the seat even where there is a choice of law clause in the contract cannot “ easily be reconciled with the earlier aut horities or with the established principles for determining the proper law” (para 24). His conclusion (at para 26) was in the following terms: “ In the absence of any indication to the contrary, an express choice of law governing the substantive contract is a strong indication of the parties ’ intention in relation to the agreement to arbitrate. A search for an implied choice of proper law to govern the arbitration agreement is therefore likely (as the dicta in the earlier cases indicate) to lead to the concl usion that the parties intended the arbitration agreement to be governed by the same system of law as the substantive contract, unless there are other factors present which point to a different conclusion. These may include the terms of the arbitration agreement itself or the consequences for its effectiveness of choosing the proper law of the substantive contract …” 51. This approach was followed in Arsanovia Ltd v Cruz City 1 Mauritius Holdings [2012] EWHC 3702 (Comm) ; [2013] 2 All ER (Comm) 1. In that case a contract contained clauses providing that it was to be governed by the laws of India and that disputes were to be settled by arbitration in London. It was held that, as a matter of construction, the parties had chosen Indian law to govern the arbitration agreement. 52. Recently, in Kabab -Ji SAL (Lebanon) v Kout Food Group (Kuwait) [2020] EWCA Civ 6; [2020] 1 Lloyd’s Rep 269 the Court of Appeal similarly construed a clause in a contract which stated “ This Agreement shall be governed by and construed in accordance with the laws of England” as m eaning that all the term s of the contract were governed by English law including an arbitration clause which provided for arbitration in France . This conclusion was reinforced by the fact that Page 18 the contract included a clause which stated that “ This Agreement consists of … the terms of agreement set forth herein below …” . (iii) Considerations of principle 53. A number of further considerations confirm the reasonableness of, as a general rule, construing a choice of law to govern the contract as applying to an arbitration agreement set out in a clause of the contract, even where the law chosen to govern the contract differs from that of the place chosen as the seat of the arbitration: i) This approach provides a degree of ce rtainty. The parties can be assured that an agreement as to the governing law will generally be an effective choice in relation to all of their contractual rights and obligations and to all of their disputes. ii) It achieves consistency. The same system of law governs all the parties’ rights and obligations. It can be unsatisfactory for potentially closely related issues such as the identity of the contracting parties or the proper approach to the interpretation of their bargain to be governed by different systems of law, depending on whether it relates to the main contract or the arbitration agreement. iii) It avoids complexities and uncertainties . As soon as the relationship between the parties is subject to two systems of law, problems can arise as to where and ho w to draw the boundaries between them . This is exemplified by the increasing prevalence of multi -tier dispute resolution clauses . If the arbitration agreement is governed by a different system of law from the main body of the contract, provisions that requ ire negotiation and/or mediation and/or expert determination in advance of arbitration raise potentially difficult questions as to whether they are governed by the law applicable to the arbitration agreement or by the law generally applicable to the contra ct, and indeed as to whether those questions should be answered by applying the common law rules or the Rome I Regulation. Article 50.1 of the construction contract is an example of such a clause. Although we explain later how these difficulties may be add ressed, if there is only one system of law then no such difficulties arise. iv) It avoids artificiality. The principle that an arbitration agreement is separable from the contract containing it is an important part of arbitration law but it is a legal doctrine and one which is likely to be much better known Page 19 to arbitration lawyers than to commercial parties. For them a contract is a contract; not a contract with an ancillary or collateral or interior arbitration agreement. They would therefore reasonably expect a choice of law to apply to the whole of that contract. v) It ensures coherence. It is consistent with the treatment of other types of clauses whose validity is also insulated from challenges to the contract, such as choice of law or choice of court clauses. Such clauses are generally presumed to be governed by the law of the contract of which they form part: see Dicey, Morris & Collins on The Conflict of Laws , 15th ed (2012) at paras 12- 103 and 12- 109. 54. As a matter of principle and authority there are therefore strong reasons why an agreement on a choice of law to govern a contract should generally be construed as applying to an arbitration agreement set out or otherwise incorporated in the contract. (iv) The international perspective 55. As to the international pe rspective, although there is no uniformity, there are many commentators on international arbitration who support such an approach, at least where there is an express choice of governing law for the contract. Examples to which we were referred include: Bant ekas, “The Proper Law of the Arbitration Clause : A Challenge to the Prevailing Orthodoxy” (2010) 27 Journal of International Arbitration 1 , 1 -2; Born, International Commercial Arbitration , 2 nd ed (2014) , p 592 ; Grover, “Dilemma of the Proper Law of the Arbitration Agreement: An Approach Towards Unification of Applicable Laws ” (2014) 32 Sing L Rev 227, 255; Choi, “Choice of Law Rules Applicable for International Arbitration Agreements ” (2015) 11 Asian International Arbitration Journal 105, 108 -1 09 ; Khatch adourian, “ Fortifying the Arbitration Clause ” in Ziadé (ed), Festschrift Ahmed Sadek El -Kosheri (2015), pp 53- 56; and Miles and Goh, “A Principled Approach Towards the Law Governing Arbitration Agreements ” in Kaplan and Moser (eds) , Jurisdiction, Admissibility and Choice of Law in International Arbitration: Liber Amicorum Michael Pryles ( 2018) Chapter 24, p 393. 56. This is also said to be the approach generally adopted by ICC arbitrators (see Lew, “The Law Applicable to the Form and Sub stance of the Arbitration Clause: 40 Years of Application of the New York Convention ” in van den Berg (ed), Improving the Efficiency of Arbitration Agreements and Awards , (1998) ICCA Congress Series Vol 9 , pp 143 -14 4). It would appear that the same approac h has been adopted in a number of common law and civil law jurisdictions. These include Singapore, India, Page 20 Pakistan, Germany and Austria. According to Chubb Russia they also include Hong Kong, Australia and Switzerland, although this was questioned by Enka. 57. Singapore provides an instructive example. In FirstLink Investments Corpn Ltd v GT Payment Pte Ltd [2014] SGHCR 12 it was held that the law of the seat should generally apply to the arbitration agreement. In BCY v BCZ [2016] SGHC 249; [2016] 2 Lloyd’s Rep 583 Steven Chong J disagreed and held that the approach in Sula mérica should be followed as it “is supported by the weight of authority and is, in any event, preferable as a matter of principle” (para 49). Having set out detailed reasons why that was so, he concluded that, as the arbitration agreement in that case was contained in a contract expressly g overned by New York law , the presumption was that New York law governed the arbitration agreement and th is presumption was not displaced by the choice of Singapore as the seat of arbitration . 58. BCY v BCZ has been approved by the Singapore Court of Appeal - s ee BNA v BNB [2020] 1 Lloyd’s Rep 55, para 44, where it was accepted by both parties as a correct statement of the law. V. The approach of the Court of Appeal (i) The Court of Appeal’s judgment 59. The Court of Appeal reached a contrary conclusion in the present case. Leaving aside cases in which, exceptionally, a choice of the law governing the arbitration agreement is specified in the arbitration agreement itself, Popplewell LJ (with whom Flaux and Males LJJ agreed) was prepared to accept that an express choice of the law applicable to the contract containing the arbitration agreement may sometimes, as a matter of construction, amount to an express choice of the law applicable to the arbitration agr eement (para 90). But he considered that this conclusion would follow only in a minority of cases and that in all other cases there is a strong presumption that the parties have impliedly chosen the law of the seat of the arbitration to govern the arbitrat ion agreement. This was said to be the general rule, “subject only to any particular features of the case demonstrating powerful reasons to the contrary” (para 91). (ii) Separability 60. Our first difficulty with this proposed general rule is that we do not ag ree that it is only in a minority of cases that an express choice of law to govern the contract should properly be construed as being a choice of law to govern an arbitration agreement included in the contract. As we have discussed, a clause such as “ This Page 21 Agreement is to be governed by and construed in accordance with the laws of [a named country]” is naturally and sensibly understood to mean that the law of that country should govern and determine the meaning and effect of all the clauses in the contract which the parties signed including the arbitration clause . It is unclear to us why more should be needed - or what more on the Court of Appeal’s approach is required - to make it clear that a phrase such as “This Agreement” means the whole agreement and not just part of it. 61. The Court of Appeal justified its approach on the ground that a choice of law to govern the contract “has little if anything to say about the [arbitration agreement] law choice because it is directed to a different and separate agreement” (para 92). This was said to follow from the doctrine that an arbitration agreement is separable from the rest of the contract. In our view, this puts the principle of separability of the arbitration agreement too high. For reasons given earlier, the requi rement that a n arbitration clause is to be treated a s a distinct agreement for the purpose of determining its validity, existence and effectiveness makes it more amenable than other parts of a contract to the application of a different law. The rationale u nderlying the separability principle is also relevant, as we will mention later, in cases where applying the governing law of the contract to the arbitration clause would render the arbitration agreement invalid or ineffective. But it does not follow from the separability principle that an arbitration agreement is generally to be regarded as “ a different and separate agreement ” from the rest of the contract or that a choice of governing law for the contract should not generally be interpreted as applying to an arbitration clause. 62. Descriptions of an arbitration clause as, for example, “collateral to the main contract in which it is incorporated” (Paal Wilson & Co A/S v Partenreederei Hannah Blumenthal ( The Hannah Blumenthal) [1983] 1 AC 854, 917, per Lord Dip lock) or “a separate contract, ancillary to the main contract ” (Bremer Vulkan Schiffbau und Maschinenfabrik v South India Shipping Corpn Ltd [1981] AC 909, 9 98 , per Lord Scarman) need to be seen in their context as ways of expressing the doctrine that the discharge by frustration (or for other reasons) of the substantive obligations created by the contract will not discharge the parties’ agreement to arbitrat e. The arbitration clause is nonetheless part of the bundle of rights and obligations recorded i n the contractual document. So, for example, an assignment of the contract will include an arbitration clause without the need for any separate or additional assignment: see Schiffahrtsgesellschaft De tlev von Appen GmbH v Voest Alpine Intertrading GmbH (The Jay Bola) [1997] 2 Lloyd ’s Rep 279 , 285; Shayler v Woolf [ 1 946] Ch 320; and Cockett Marine Oil DMCC v ING Bank NV (The M/V Ziemia Ciesznska) [2019] EWHC 1533 (Comm); [2019] 2 Lloyd ’s Rep 541. As Colman J put it in construing the words “ any clause of this Agreement ” as including an arbitration clause in JSC Zestafoni G Nikoladze Ferroalloy Plant v Ronly Holdings Ltd [2004] EWHC 245 (Comm); [2004] 2 Lloyd ’s Rep 335, para 31: Page 22 “There is nothing in the intrinsic character of an arbitration agreement as having an attribute of separability which prevents it from being included in that phrase.” 63. Moore- Bick LJ summed up the position clearly when he said in the Sul américa case at para 26: “ The concept of se parability itself, however, simply reflects the partie s’ presumed intention that their agreed procedure for resolving disputes should remain effective in circumstances that would render the substantive contract ineffective. Its purpose is to give legal effect to that intention, not to insulate the arbitration agreement from the substantive contract for all purposes. ” 64. In his lead judgment in t he Court of Appeal Popplewell LJ quoted this passage (at para 9 3) and appeared there to recognise that it is wrong to characterise an arbitration clause generally as a separate agreement. He went on, however, to make a more specific point that one of the purposes for which an arbitration agreement is treated as separat e and severable is that of applying the curial law which , w here the parties have chosen a different arbitration seat - and hence curial law - from the law applicable to their contract , is distinct from the latter system of law . The rhetorical question was posed, at para 94 : “Why then should [the law applicable to the contract] have anything to say about the closely related aspect of the very same arbitration agreement, namely the [ law which governs it] (absent express language to that effect so as to give rise to an express choice of [the arbitration agreement ] law)?” Leaving aside what should count as “express language” in this regard, this argument rests on the premise that the curial law which governs the arbitration process is so closely related to the law governing the arbitration agreement that a choice of law to govern the contract should generally be presumed not to apply to an arbitration clause when the parties have chosen a different curial law. It is to this argument, which was central to the Court of Appeal’s reasoning, that we therefore turn. (iii) The overlap argument 65. This argument, which we will call the “overlap argument ”, seems to have made its first appearance in XL Insurance Ltd v Owens Corning [2001] 1 All ER (Comm) 530, mentioned earlier, where Toulson J considered that, by stipulating for arbitration in London under the provisions of the Arbitration Act 1996 , the parties had impliedly chose n English law to govern the validity of the arbitrati on agreement despite the choice of New York law as the governing law of the policy (see p 543b). Page 23 His essential reasoning (at p 541e) was that the substance and process of arbitration “ are closely intertwined ” and that the 1996 Act “ contains various provisi ons which could not readily be separated into boxes labelled ‘ substantive arbitration law’ or ‘ procedural law ’, because that would be an artificial division ”. 66. The Court of Appeal in the present case endorsed and elaborated on this reasoning, concluding tha t “the overlap between the scope of the curial law and that of the [arbitration agreement] law strongly suggests that they should be the same” (para 96). They further considered that, given this overlap and the fact that the curial law which regulates the arbitration process is a matter of choice which comes with an express choice of seat, it seems “natural to regard” a choice of seat as an implied choice of the law applicable to the arbitration agreement (para 101). On this basis they held that there is a “strong presumption” that a choice of seat is an implied choice of the law which is to govern the arbitration agreement (para 105(3)). (iv) Choice of curial law 67. On this appeal Chubb Russia disputed the initial premise that a choice of seat for an arbitration involves any choice of law at all, procedural or substantive . C ounsel for Chubb Russia submitted that the application of the curial law of the seat is something that follows automatically from a choice of place of arbitration rather than bei ng itself a matter of choice . They cited as an analogy a hypothetical case postulated by Redfern and Hunter : Law and Practice of International Commercial Arbitration, 6 th ed ( 2015 ), para 3.63, of an English motorist who takes her car to France. Redfern and Hunter comment that: “… it would be an odd use of language to say that this notional motorist had opted for ‘French traffic law’; rather, she has chosen to go to France - and the applicability of French law then follows automatically. It is not a matter of choice.” 68. We agree that it would be inapt to describe the tourist in this example as having made a choice to be regulated by French traffic law. But as Mr Dicker QC for Enka submitted, it is difficult to conceive that a person’s decision to visit France might be informed by a desire to be governed by French traffic law. By contrast, the nature and sco pe of the jurisdiction exercised by the courts of a country over an arbitration which has its seat there is a highly material consideration in choosing a seat for the arbitration. That is reinforced by the fact that the seat of an arbitration is a legal co ncept rather than a physical one. A choice of place as the seat does not dictate that hearings must be held, or that any award must actually be issued, in that place. As the Court of Appeal observed (at para 46), it is perfectly possible to conduct an arbitration with an English seat at any convenient location, anywhere in Page 24 the world. Furthermore, under section 53 of the Arbitration Act 1996, unless otherwise agreed by the parties, where the seat of an arbitration is in England and Wales, any award in the proceedings shall be treated as made there, regardless of where it was signed, despatched or delivered to any of the parties (see also article 31(3) of the UNCITRAL Model Law on International Commercial Arbitration adopted by the United Nations Commission on International Trade Law on 21 June 1985). The point of agreeing a seat is to agree that the law and courts of a particular country will exercise control over an arbitration which has its seat in that country to the extent provided for by that country’s la w. A choice of seat can in these circumstances aptly be regarded as a choice of the curial law. 69. As noted at the beginning of this judgment, however, the curial law which applies to the arbitration process is conceptually distinct from the law which governs the validity and scope of the arbitration agreement. Whether a choice of the curial law carries any implication that the parties intended the same system of law to govern the arbitration agreement - and, if so, the strength of any such implication - must depend on the content of the relevant curial law. (v) Relationship between curial law and arbitration agreement law 70. In Carpatsky Petroleum Corp n v PJSC Ukrnafta [2020] EWHC 769 (Comm); [2020] Bus LR 1284 , the claimant applied to enforce in England and Wales an arbitration award made in Sweden. Enforcement was resisted on the ground (among others) that there was no valid arbitration agreement in the contract between the parties. This argument depended on the ass umption that the validity of the arbitration agreement was governed by the law of Ukraine. The contract provided for the “ law of substance of Ukrain e” to apply “on examination of disputes”. Butcher J held (at paras 67 -71) that this was not a choice of Ukra inian law to govern the arbitration agreement and that, in the circumstances, the choice of Stockholm as the seat for any arbitration demonstrated an implied choice that the validity and interpretation of the arbitration agreement should be governed by Swe dish law. His reasons were that: (1) it was reasonable to infer that the parties had deliberately chosen a neutral forum to resolve their disputes and hence “intended the law of that jurisdiction to determine issues as to the validity and ambit of that cho ice”; and (2) by choosing Sweden as the seat for the arbitration, the parties agreed to the application of the Swedish Arbitration Act, including section 48 which provides that, in the absence of agreement on a choice of law to govern an arbitration agreem ent with an international connection , the arbitration agreement shall be governed by the law of the country in which, by virtue of that agreement, the arbitration proceedings have taken place or will take place. It follows that, by providing for a Swedish seat , the parties were impliedly agreeing that Swedish law should govern the arbitration agreement . Page 25 71. A similar inference could also be drawn where a contract contains an agreement for arbitration in Scotland. Section 6 of the Arbitration (Scotland) Act 2010 provides : “ Where - (a) the parties to an arbitration agreement agree that an arbitration under that agreement is to be seated in Scotland, but (b) the arbitration agreement does not specify the law which is to govern it, then, unless the parties otherwise agree, the arbitration agreement is to be governed by Scots law .” 72. There is, however, no similar provision in the Arbitration Act 1996. The argument made by Enka, and accepted by the Court of Appeal, is that the 1996 Act contains provisions which are subst antive as well as provisions which are procedural in nature, and that there is no clear division between the two. In these circumstances it is argued that, by choosing an English seat in the knowledge that the Arbitration Act 1996 will apply where the seat of the arbitration is in England , the parties are by implication choosing English law to govern at least some aspects of their substantive rights under the arbitration agreement . Furthermore, as suggested by Toulson J in the XL Insurance case, the provisions which affect substantive rights are intertwined with, and cannot readily be separated from, procedural provisions of the Act. The natural inference is said to be that the parties intended all their rights under the arbitration agreement to be governed by English law. (vi) Section 4(5) of the 1996 Act 73. We agree that there is a close relationship between provisions of the Arbitration Act concerned with the arbitration agreement and provisions of the Act concerned with the arbitration process and that the d istinction between them is not always clear or easy to draw. But we do not accept that this justifies the conclusion that a choice of an English seat of arbitration is an implied choice that the arbitration agreement will be governed by English law. In our view, a conclusive answer to that argument lies in a point raised by Chubb Russia on this appeal which was not fully developed in the Court of Appeal . The point in short is that almost all the provisions of the 1996 Act relied on to support the overlap argument are non- mandatory and, where the arbitration agreement is governed a foreign law, by reason of s ection 4(5) Page 26 the non -mandatory provisions of the Act which concern arbitration agreements do not apply to it. As the legislation contemplates and specific ally provides for a situation in which the arbitration agreement will be governed by a foreign law even though English law governs the arbitration process, no necessary inference can be drawn that, by choosing an English seat and with it English law as the curial law, parties are also impliedly choosing English law to govern their arbitration agreement. 74. Section 4(5) of the 1996 Act states: “The choice of a law other than the law of England and Wales or Northern Ireland as the applicable law in respect of a matter provided for by a non- mandatory provision of this Part is equivalent to an agreement making provision about that matter. For this purpose an applicable law determined in accordance with the parties’ agreement, or which is objectively determined in the absence of any express or implied choice, shall be treated as chosen by the parties.” 75. The clear meaning and effect of this provision is that, where a foreign law is applicable to an arbitration agreement (whether by choice or as determined in the absence of choice by the closest connection test), that fact alone is enough to disapply any non- mandatory provision of the Act in so far as it would otherwise affect a matter governed by the law applicable to the arbitration agreement. This is beca use the applicability of a foreign law is treated as equivalent to an agreement to make contrary provision about a matter. It is not necessary to inquire whether or not the foreign law does in fact make such contrary provision. 76. Even if there were otherwise considered to be any ambiguity in the meaning of section 4(5), it is dispelled by the Supplementary Report on the Arbitration Act 1996, dated January 1997, produced by the Departmental Advisory Committee on Arbitration (the “ DAC” ), which explains the gene sis of the provision. As originally drafted, clause 2 of the Bill provided: “(1) The provisions of this Part apply where the law of England and Wales or Northern Ireland is applicable, or the powers of the court are exercisable, in accordance with the rules of the conflict of laws. (2) They apply, in particular - Page 27 (a) to matters relating to or governed by the arbitration agreement, where the applicable law is the law of England and Wales or Northern Ireland; and (b) to matters governed by the law applica ble to the arbitral proceedings, where the seat of the arbitration is in England and Wales or Northern Ireland.” 77. The DAC Supplementary Report, at para 7(ii), observed that the purpose of clause 2(2) was to avoid the danger that all the provisions of Part I of the Act would be imported if English law was found to govern one particular aspect of an arbitration. For example: “… an arbitration may have a French seat, with French law governing the procedure, but English law governing the arbitration agreement. In such a situation, only those provisions of the Act which concern arbitration agreements should apply. It would be quite wrong to apply provisions of the Act which concern arbitral procedure, as this would be governed by French law. ” Plainly, this reasoni ng applies equally in reverse to an arbitration with an English seat and English law governing the procedure, but French law governing the arbitration agreement. In such a situation, only those provisions of the Act which concern arbitral procedure should apply and not those which concern the arbitration agreement, as this would be governed by French law. 78. The clause as drafted, however, was considered unworkable in practice (although sound in principle) - one reason being that, to apply clause 2(2), it woul d have been necessary individually to characterise and separate all those provisions of the Act which concerned the arbitration agreement, as distinct from all those that concerned the arbitral procedure (see para 9(ii) of the DAC Supplementary Report). It was noted that the attempt to do this “had proved an extremely difficult and complex exercise”. Furthermore: “Many provisions concern both arbitration agreements and arbitral procedure, and there appeared to be a divergence of view with respect to many ot hers.” 79. In the light of these difficulties, the DAC decided to recommend recasting the whole provision so as to establish in section 2(1) the basic rule that Part I of the Page 28 Act applies to arbitrations which have their seat in England and Wales or Northern Ireland (see paras 10 -11 of the DAC Supplementary Report). In such a case, however, as explained in para 12: “If … a foreign law has been chosen to govern any particular aspect of the arbitration, such as the arbitral procedure or the arbitration agreement, or is o therwise applicable to any such aspect, this is catered for by section 4(5). Therefore, reference may be made to this Act in the first instance, and then back to another law with respect to a specific issue. Whilst a process of characterisation may still have to be done, the combination of section 2 and section 4(5) avoids the dangers that: – a choice of English law with respect to one part of an arbitration will import other parts of the Act that concern other aspects of the arbitration; – a choice of England as the seat of the arbitration will necessarily entail the imposition of every provision of the Act.” 80. We observe that the “recasting” carried out on the recommendation of the DAC did not remove the need individually to characterise the provisions of the Ac t as substantive or procedural (or partly substantive and partly procedural) whenever the applicable law is in issue - an exercise described by the DAC as “extremely difficult and complex”. Nevertheless, the legislative history confirms that sections 2 and 4(5) of the 1996 Act as enacted were intended to have the effect that, where England is chosen as the seat of an arbitration but the arbitration agreement is governed by a foreign law, the non -mandatory provisions of the Act do not apply to any matter con cerning the parties’ substantive rights and obligations under the arbitration agreement. The fact that the Act contains some provisions which are substantive, or partly substantive, cannot therefore - where those provisions are non - mandatory - support an i nference that, by choosing an English seat of arbitration, parties must be taken to have contemplated and intended that the validity and scope of their arbitration agreement should be governed by English law. 81. The only mandatory provisions of the 1996 Act a re sections 12, 13 and 66 to 68. Section 12 gives the court power to extend time for beginning an arbitration where there is a contractual time limit. This could only have any bearing on the law applicable to the arbitration agreement if the arbitration ag reement includes a contractual time limit (which the relevant clause in this case does not). Section 13 applies the Limitation Acts to arbitrations. As these Acts include the Foreign Page 29 Limitation Periods Act 1984, which applies foreign limitation law to any substantive obligation governed by foreign law, this cannot support an inference that the arbitration agreement is governed by English law. Sections 66 to 68 are concerned with enforcement of the award and applications to the court to challenge an award. T hey are procedural in nature and cannot be said to determine the law applicable to the arbitration agreement. 82. The provisions of the Arbitration Act 1996 therefore do not justify any general inference that parties who choose an English seat of arbitration t hereby intend their arbitration agreement to be governed by English law. (vii) Enka’s case on section 4(5) 83. Enka put forward three responses to this reasoning, none of which we have found persuasive. 84. First, counsel for Enka submitted that section 4(5) is co ncerned only with a choice of foreign law as the curial law for the arbitration process, and not with a choice of foreign law to govern the arbitration agreement. This, however, is not a tenable reading of section 4(5), which is manifestly not limited in t his way and expressly applies whenever a foreign law is applicable in respect of a matter provided for by a non -mandatory provision of the Act. As emphasised on Enka’s own case, the matters provided for by non -mandatory provisions of the Act include some m atters which concern the substance of the arbitration agreement as well as matters of procedure. Nor does section 4(4) support a different interpretation, as suggested in Enka’s written case. Section 4(4) provides that it “is immaterial whether or not the law applicable to the parties’ agreement is the law of England and Wales …”. This makes it clear that, if the parties have made arrangements by agreement in place of any non -mandatory provision of the Act, it is irrelevant whether or not that agreement is governed by English law. There is no inconsistency between that provision and the rule established by section 4(5) that a choice of foreign law in respect of a matter is equivalent to an agreement making provision about that matter. 85. The second argument adv anced by Enka is that, if - as we think clear - section 4(5) is not confined to a choice of curial law and also covers cases where a foreign law is applicable to the arbitration agreement, section 4(5) nevertheless applies only where the arbitration agreem ent makes specific reference to the matter provided for by a non- mandatory provision of the Act. Page 30 86. As authority for this restrictive interpretation, Enka relied on a dictum of Lord Steyn in Lesotho Highlands Development Authority v Impregilo S pA [2005] UKHL 43; [2006] 1 AC 221. That case involved an attempted challenge under section 68 of the 1996 Act to a decision by an arbitral tribunal to award interest under section 49(3) on principal sums awarded. The challenge failed because the House of Lords held that substantial injustice had not been established, as required to invoke section 68. However, Lord Steyn, who gave the leading speech, went on to point out that the challenge had also faced other formidable difficulties. In particular, the power under sectio n 49(3) to award interest was prima facie available: the only question was whether there had been an agreement to the contrary for the purpose of section 49(2). In that context Lord Steyn noted (at para 37) that the judge at first instance had appeared to take the view that the law of Lesotho, as the law applicable to the construction contract under which the claim arose, might be relevant - presumably on the basis that it constituted an agreement to the contrary. In relation to this, Lord Steyn remarked: “ Ignoring for the moment the fact that one does not know what the law of Lesotho is, this view comes up against the difficulty that only an agreement in writing as defined in the Act can qualify as an agreement to the contrary under section 49: section 5(1). … The law of Lesotho is not an agreement to the contrary in writing.” 87. Lord Steyn made no mention of section 4(5) of the Act: the point that he made was based on section 5(1), which states that an “agreement between the parties as to any matter is effective for the purposes of this Part only if in writing”. Nevertheless, in C v D , at para 19, Longmore LJ treated Lord Steyn’s dictum as supporting the view that section 4(5) requires a choice of law “with regard to the specific provision of the [ 1996] Act which the parties agree is not to apply”. This statement was in turn relied on by Burton J in National Iranian Oil Co v Crescent Petroleum Co International Ltd [2016] EWHC 510 (Comm); [2016] 2 Lloyd’s Rep 146, paras 12 -17, to conclude that a choice of Ira nian law to govern an arbitration agreement was not sufficient to disapply section 7 of the 1996 Act, which codifies the principle of separability of the arbitration agreement, and that nothing less than an agreement expressly disapplying section 7 or the English law governing separability would have sufficed for that purpose. 88. The notion that section 4(5) applies only where parties have specifically excluded a non- mandatory provision of the Act by the terms of their arbitration agreement cannot, in our view , be accepted. It is not consistent with the language of section 4(5). The words “in respect of a matter provided for by a non -mandatory provision” require only that the matter governed by the foreign law should be a matter provided for by a non -mandatory provision of the Act. They cannot reasonably be read as requiring the parties’ specific agreement that the foreign law Page 31 and not the non- mandatory provision will govern the matter. Apart from anything else, the second paragraph of section 4(5) makes it expli citly clear that no choice or agreement of the parties at all is required for section 4(5) to apply. The interpretation contended for by Enka is also inconsistent with the legislative intent, as explained in the DAC Supplementary Report. Furthermore, as th e late Mr VV Veeder QC observed, if correct, it would make a practical nonsense of the 1996 Act by requiring parties choosing a foreign law to govern an agreement for arbitration in England to analyse and identify individually in their agreement each of th e 35 or so non- mandatory provisions of the 1996 Act which they wish to disapply. We agree with Mr Veeder’s comment that the absurd consequences of such an interpretation speak for themselves: see Kaplan and Moser (eds), Jurisdiction, Admissibility and Choi ce of Law in International Arbitration: Liber Amicorum Michael Pryles (2018) , Chapter 23, p 382. 89. We do not think it credible that Lord Steyn in the Lesotho case intended to endorse such an interpretation of section 4(5), and to do so without giving any rea sons or even mentioning that provision of the Act at all. The likely reason why no reference was made to section 4(5) is that it was not relevant to the power to award interest. The Court of Appeal in the Lesotho case characterised the power to award inter est under section 49(3) of the 1996 Act as discretionary and procedural - a characterisation which Lord Steyn seems to have endorsed when referring to the reasoning of the Court of Appeal in para 38 of his speech. The fact that section 49(3) was treated by both the Court of Appeal and the House of Lords in the Lesotho case as procedural in nature was later relied on by the Court of Appeal in Maher v Groupama Grand Est [2009] EWCA Civ 1191; [2010] 1 WLR 1564, para 38, to support a similar characterisation of the power of a court to award interest under section 35A of the Senior Courts Act 1981. Because section 49(3) is procedural, the choice of the law of Lesotho to govern substantive contractual rights was not in respect of a matter provided for by section 4 9(3) and therefore did not engage section 4(5). As it was not in doubt that the curial law governing the arbitration process was English law, to disapply section 49(3) would accordingly have required a specific agreement (in writing), as Lord Steyn observe d. Whether or not Lesotho law contained any equivalent procedural power was in these circumstances not relevant. Even if it did, the law of Lesotho concerning that matter could not amount to an agreement to the contrary. 90. This is, we think, how Lord Steyn’s dictum should be understood. But whether this was what was meant or not, we are satisfied that section 4(5) does not require a specific agreement to disapply a non -mandatory provision of the Act. It follows that Longmore LJ’s statement to that effect in C v D was erroneous and that the National Iranian Oil Co case was wrongly decided on this point. 91. The third response of Enka was to contend that the consequences of giving section 4(5) what we consider to be its unambiguous meaning would be “as far- Page 32 reaching as they are surprising” because it would cause numerous non -mandatory provisions, which parties to a London arbitration are unlikely to have intended to exclude, nonetheless to be excluded. To support this contention, Enka relied as examples on sections 5, 7, 30 and 58 of the 1996 Act. 92. Of these provisions, only section 7 which codifies the principle of separability concerns the validity or scope of the arbitration agreement. Section 5, which states that Part I of the Act applies only where the arbitration a greement is in writing, is not concerned with the validity or scope of the arbitration agreement but with the circumstances in which the provisions of the Act will apply. If the requirement of writing is not met, Part I of the Act will not apply to the arb itration agreement but it will be regulated by, and will still be valid at, common law (see section 81). Section 30, which empowers the arbitral tribunal to rule on its own jurisdiction, is procedural. It does not deal with the parties’ substantive rights under the arbitration agreement but with the competence of the tribunal to determine the validity and scope of those rights. Section 58, which provides for the finality of an arbitral award, is also procedural in nature. (For that reason, the insurer’s arg ument in C v D that, as a result of section 4(5), section 58 was disapplied by a choice of New York law to govern the arbitration agreement was misconceived.) These and other procedural non -mandatory provisions will only be excluded in the unusual event th at the parties have chosen a foreign procedural law for an English -seated arbitration: see Dubai Islamic Bank PJSC v Paymentech Merchant Services Inc [2001] 1 All ER (Comm) 514 , para 31; C v D [2007] EWHC 1541 (Comm); [2007] 2 All ER (Comm) 557, paras 25 -2 6 (Cooke J); Sterling v Rand [2019] EWHC 2560 (Ch); [2019] 2 Lloyd’s Rep 577, para 58. As observed in the DAC Supplementary Report, para 7(ii) (quoted at para 77 above), in such a case it would be wrong to apply non- mandatory provisions of the Act which co ncern arbitral procedure, as this would be governed by foreign law. 93. We accept that characterising individual provisions of that Act as procedural or substantive can, as recognised by the DAC, be a difficult and complex exercise . But we are satisfied that giving section 4(5) its plain meaning does not lead to surprising or untoward consequences and is inconsistent with the contention that choosing English law as the curial law of an arbitration involves an implied choice of English law as the law applicable to the arbitration agreement. 94. For these reasons, we do not consider the overlap argument as accepted by the Court of Appeal to be well founded. While a choice of seat and curial law is capabl e in some cases (based on the content of the relevant curial law) of supporting an inference that the parties were choosing the law of that place to govern the arbitration agreement, the content of the Arbitration Act 1996 does not support such a general i nference where the arbitration has its seat in England and Wales. Page 33 VI. Avoiding invalidity (i) The validation principle 95. It is a well -established principle of contractual interpretation in English law, which dates back at least to the time of Sir Edward Coke (see Coke upon Littleton (1628) 42a), that an interpretation which upholds the validity of a transaction is to be preferred to one which would render it invalid or ineffective. In the days when Latin was commonly used in the courts, it was expressed by th e maxim “verba ita sunt intelligenda ut res magis valeat quam pereat” - translated by Staughton LJ in Lancashire County Council v Municipal Mutual Insurance Ltd [1997] QB 897, 910, as “the contract should be interpreted so that it is valid rather than ineffective”. 96. This principle may apply if, in determining whether the parties have agreed on a choice of governing law, a putative governing law would render all or a part of the contract ineffective. For example, in In re Missouri Steamship Co (1889) 42 Ch D 321 a contract for the carriage of cattle by sea from Boston to England contained a clause that the carrier should not be liable for the negligence of the master or crew of the ship. The clause was valid under English law but void under the law of Massachu setts as being against public policy. The cattle were lost by the negligence of the master and crew, and the shipper claimed against the carrier for the loss. In concluding that the parties intended the contract to be governed by English law, the judge and the Court of Appeal placed reliance on the presumption that, in the words of Fry LJ at p 341, “the law which would make the contract valid in all particulars was the law [intended] to regulate the conduct of the parties.” 97. In that case the potential invali dity of a significant clause in a contract was relied on as indicating the law intended to govern the entire contract. Where the clause in question is an arbitration clause, because of its severable character its putative invalidity may support an inferenc e that it was intended to be governed by a different law from the other provisions of the contract - or may at least negate an inference that the law generally applicable to the contract was intended to apply to the arbitration clause. (ii) Hamlyn v Talisk er 98. An early but authoritative instance of such reasoning is the decision of the House of Lords in Hamlyn & Co v Talisker Distillery [1894] AC 202. A contract between an English company and a Scottish company, to be performed in Scotland, contained the foll owing provision: Page 34 “Should any dispute arise out of this contract, the same to be settled by arbitration by two members of the London Corn Exchange, or their umpire, in the usual way.” It was common ground that th is arbitration clause was valid according to English law but in valid according to the law of Scotland because the arbitrators we re not named. The Court of Session held that the contract was governed by Scottish law as the law of the place of performance of the contract and that, in consequence, the a rbitration clause was invalid. The House of Lords unanimously reversed that decision. As Lord Wilberforce subsequently noted in the Tunisienne case (at p 596), the only question decided by the House of Lords was whether the arbitration clause was governed by Scottish law or by English law. The members of the appellate committee were careful to limit their opinions to that question and to express no view on which law governed the other provisions of the contract. 99. Two reasons were given for concluding that th e arbitration clause was governed by English law. One reason, most fully expressed by Lord Watson (at pp 212 -213), was that the language of the arbitration clause showed that the parties were contracting with reference to English law, as the clause require d the arbitrators to be members of a commercial body in London and to decide disputes “in the usual way” - in other words, in the manner customary in London. This reasoning did not, however, as it seems to us, justify treating the arbitration clause itself as governed by English law irrespective of which law governed the rest of the contract. It was a reason for inferring that the parties intended the arbitrators to apply English law in deciding any dispute under the contract and therefore for regarding the parties’ substantive contractual obligations as governed by English law. The question whether the arbitration clause was valid determined whether the arbitrators had jurisdiction, which was not at that time a matter that the arbitrators themselves were seen as competent to decide. This reasoning is therefore an early example of an approach we will consider shortly which treats a choice of seat of arbitration as an implied choice of law to govern the contract as a whole. 100. The principal enduring significance of Hamlyn v Talisker lies in the second reason given for the decision, which was clearly articulated by Lord Herschell LC and Lord Ashbourne. It was this reason which justified treating the arbitration clause as potentially governed by a different law from rest of the contract. In Lord Herschell’s words (at p 208): “… the contract with reference to arbitration would have been absolutely null and void if it were to be governed by the law of Scotland. That cannot have been the intention of the parties; it is not reasonable to attribute that intention to them if the contract may be otherwise construed; …” Page 35 Lord Ashbourne made the same point, stating graphically (at p 215) that “ the arbitration clause becomes mere waste paper if it is held that the parties were contracting on the basis of the application of the law of Scotland, which would at once refuse to acknowledge the full efficacy of a clause so framed.” He continued: “It is more reasonable to hold that the parties contracted with the common intention of giving entire effect to every clause, rather than of mutilating or destroying one of the most important provisions.” (iii) The decision in Sulamérica 101. It was this reasoning which led the Court of Appeal in the Sulamérica case to conclude that the arbitration clause in that case was governed by English law despite, as discussed earlier, starting from the position that an express choice of law to govern the contract is normally intended to apply to the arbitration clause. 102. In the Sulamérica case claims were made by Brazilian companies involved in a construction project in Brazil under two insurance policies. Each policy contained an express choice of Brazili an law to govern the policy and a clause conferring exclusive jurisdiction on the courts of Brazil, but also mediation and arbitration clauses. These provided that any dispute should be referred to mediation and that, if the parties failed to agree the amo unt to be paid under the policy through mediation, the dispute should then be referred to arbitration in London. The insurers commenced arbitration proceedings in London and applied successfully to the English court for an interim injunction to restrain th e insured from pursuing proceedings in the courts of Brazil. An appeal by the insured was dismissed by the Court of Appeal. 103. The insured’s case was that the contract, including the arbitration agreement, was governed by Brazilian law and that under Brazilia n law the arbitration agreement was not enforceable against them without their consent. As noted earlier, Moore- Bick LJ (with whom Hallett LJ and Lord Neuberger MR agreed) accepted that the choice of Brazilian law to govern the contract was a strong indica tion that the parties intended that system of law to govern the arbitration agreement. However, Moore -Bick LJ identified two factors pointing the other way. The first was the overlap argument which we have just discussed: that by choosing London as the sea t of arbitration, the parties must have foreseen and intended that the provisions of the Arbitration Act 1996 should apply to any arbitration, including those provisions which are more substantive than procedural in nature (para 29). For the reasons alread y given, we do not think that this argument is sound, as it overlooks the fact that, if the arbitration agreement was governed by Brazilian law, the non - Page 36 mandatory substantive provisions of the Act would be excluded by section 4(5). It was the second factor, however, which the Court of Appeal regarded as decisive. This was the possible existence of a rule of Brazilian law which would render the arbitration agreement enforceable only with the insured’s consent (para 30). Moore - Bick LJ reasoned that, given the terms of the mediation and arbitration clauses, the parties could not have intended to choose a system of law that “either would, or might well, have that effect” (para 31). As he also put it, Brazilian law could not have been intended to govern the arbit ration agreement when “there is at least a serious risk that a choice of Brazilian law would significantly undermine that agreement”. 104. In these circumstances it was necessary to identify the system of law with which the arbitration agreement was most closel y connected. On this point Moore- Bick LJ said (at para 32) that: “an agreement to resolve disputes by arbitration in London, and therefore in accordance with English arbitral law, does not have a close juridical connection with the system of law governing the policy of insurance, whose purpose is unrelated to that of dispute resolution; rather, it has its closest and most real connection with the law of the place where the arbitration is to be held and which will exercise the supporting and supervisory juri sdiction necessary to ensure that the procedure is effective.” On this basis he concluded that the arbitration agreement was governed by English law. 105. Although reasoning of this kind was not relied on in the XL Insurance case - where, as discussed earlier, Toulson J relied on the overlap argument - it provides in our view a better justification for the result reached in that case. The fact that the arbitration clause would arguably have been invalid under New York law was itself a strong reason for interpret ing the choice of New York law to govern the insurance policy as not extending to the arbitration agreement. (iv) Commercial purpose of an arbitration clause 106. The principle that contracting parties could not reasonably have intended a significant clause in their contract, such as an arbitration clause, to be invalid is a form of purposive interpretation, which seeks to interpret the language of the contract, so far as possible, in a way which will give effect to - rather than defeat - an aim or purpose which the parties can be taken to have had in view. The strength Page 37 of the inference that an interpretation of the contract would defeat an aim of the parties is, however, a matter of degree. An interpretation which would without doubt mean that an arbitration cla use is void and of no legal effect at all gives rise to a very powerful inference that such a meaning could not rationally have been intended. That was the position in Hamlyn v Talisker , where it was common ground that, if the arbitration clause were governed by Scottish law, it would have been (in Lord Herschell’s words [1894] AC 202, 208 ) “absolutely null and void”. In the Sulamérica case the inference was weaker. There was a serious risk - but not a certainty - that, if Brazilian law applied to the arbit ration clause, it would render the agreement to arbitrate enforceable only with the insured’s consent. That would not have meant that the arbitration clause was of no effect at all. As Moore -Bick LJ acknowledged, although most arbitration agreements permit either party to refer disputes to arbitration, some provide for arbitration only at the option of one or other party. He did not think it reasonable, however, to attribute to the parties in that case an intention to enter into “a one -sided arrangement of that kind” (para 30). 107. In Fiona Trust & Holding Corp n v Privalov [2007] UKHL 40; [2007] Bus LR 1719, the House of Lords affirmed the principle that “the construction of an arbitration clause should start from the assumption that the parties, as rational bus inessmen, are likely to have intended any dispute arising out of the relationship into which they have entered or purported to enter to be decided by the same tribunal” (see para 13, per Lord Hoffmann). Contrary to a submission made on behalf of Chubb Russ ia, this is not a parochial approach but one which, as the House of Lords noted in the Fiona Trust case, has been recognised by (amongst other foreign courts) the German Federal Supreme Court (Bundesgerichtshof), the Federal Court of Australia and the Unit ed States Supreme Court and, as stated by Lord Hope at para 31, “is now firmly embedded as part of the law of international commerce.” In his monumental work on International Commercial Arbitration , 2nd ed (2014), p 1403 Gary Born summarises the position a s follows: “ In a substantial majority of all jurisdictions, national law provides that international arbitration agreements should be interpreted in light of a ‘ pro-arbitration ’ presumption. Derived from the policies of leading international arbitration conventions and national arbitration legislation, and from the parties’ likely objectives, this type of presumption provides that a valid arbitration clause should generally be interpreted expansively and, in cases of doubt, extended to encompa ss disputed claims. That is particularly true where an arbitration clause encompasses some of the parties’ disputes and the question is whether it also a pplies to related disputes, so that all such controversies can be resolved in a single proceeding (rather than in multiple proceedings in different forums).” Page 38 108. To the extent that a putative applicable law fails to recognise this presumption that arbitration has been chosen as a one stop method of dispute resolution, it is inherently less likel y that reasonable commercial parties would have intended that law to determine the validity and scope of their agreement to arbitrate (rather than litigate) disputes. 109. What degree of impairment to the commercial purpose of an arbitration agreement will be e nough to negate the assumption that a choice of law to govern the contract is intended to apply to the arbitration agreement is not a question which can be answered in the abstract. As with any question of construction, it will be necessary to have regard to the particular words used in the contract and the surrounding circumstances, as well as the nature and extent of the risk that the purpose of the arbitration agreement would be undermined if its validity and scope were governed by the relevant system of law. We cannot improve on the formulation of Moore -Bick LJ in the Sulamérica case, para 31, that commercial parties are generally unlikely to have intended a choice of governing law for the contract to apply to an arbitration agreement if there is “at lea st a serious risk” that a choice of that law would “significantly undermine” that agreement. VII. Relevance of the arbitration seat to the main contract law 110. During the 20th century a line of authority developed which treated a choice of place of arbitratio n, where there was no express choice of governing law clause in the contract, as a strong indication that the parties intended the contract to be governed by the law of that place. This inference hardened into a rule of law and reached its high -water mark in Tzortzis v Monark Line A/B [1968] 1 WLR 406, where the Court of Appeal held that a London arbitration clause gave rise to an implication that the parties intended English law to govern their contract which could only be rebutted by an express provision to the contrary. 111. In the Tunisienne case the House of Lords held that this put the strength of the implication too high and that the implication stemming from a choice of arbitral forum could be overridden by contrary indications derived from the express pr ovisions of the contract or relevant surrounding circumstances. Nevertheless, Lord Wilberforce (at p 596B) described the inference that the parties intended the law of the place of arbitration to govern their contract as “ a sound general rule”. Lord Diploc k went further and said (at p 609E) that he did not “ wish to throw any doubt upon the proposition that an arbitration clause is generally intended by the parties to operate as a choice of the proper law of the contract as well as the curial law and should be so construed unless there are compelling indications to the contrary ...”. Page 39 112. As is apparent from, for example, the submissions of Robert Goff QC in defence of this approach in the Tunisienne case (at p 579D), its rationale was that contracting parties, by agreeing to arbitration in a particular place, must normally be taken to have expected the arbitrators to be resident in that place and to apply the law with which they are familiar. Lord Wilberforce expressed some reservation about this reasoning, observ ing (at p 596C): “I venture to think that in commercial matters, at the present time, this may give insufficient recognition to the international character of the City of London as a commercial centre - the reason, rather than any preference for English rules, for which arbitration in London is selected.” 113. In the half century since the Tunisienne case was decided international arbitration has undergone major evolution and exponential growth. This has been accompanied by the development of internati onal arbitral institutions such as the ICC’s International Court of Arbitration, the International Centre for Dispute Resolution established by the American Arbitration Association and the London Court of International Arbitration. The primary reason for s electing London as a place of arbitration is no longer the international character of London as a commercial centre but its attractiveness specifically as a forum in which to arbitrate international disputes. In some cases where the parties have chosen Eng lish law as the governing law of their contract, the ready availability of expert English lawyers may be a relevant factor in choosing London as the arbitration venue. But even in the kinds of arbitration where the members of the arbitral tribunal are chos en for their legal expertise (rather than solely or mainly for their commercial experience), there is nothing to prevent the appointment of lawyers qualified in other jurisdictions to act as arbitrators in a London -seated arbitration, or English lawyers to act as arbitrators in a foreign -seated arbitration, and such appointments are frequently made. Furthermore, experienced international arbitrators qualified as lawyers in England and Wales or in other jurisdictions are perfectly familiar with applying systems of law other than their own. There can in these circumstances be no general implication that a choice of London (or any other major arbitration centre) as the seat of arbitration demonstrates an intention that the parties’ contractual obligations will be governed by the law of that place. This is equally so whether the question of implied choice is governed by article 3 of the Rome I Regulation (in relation to the main body of the contract) or the common law conflict rules (in relation to the arbitratio n agreement). 114. There are still cases in which an arbitration clause providing for arbitration in London by, for example, English maritime arbitrators, or by London brokers, or by a local association or exchange, may in combination with other factors be rega rded as conveying an implied choice of law. An example is Egon Oldendorff v Libera Corpn (No 2) [1996] 1 Lloyd’s Rep 380, where an arbitration clause in a Page 40 charterparty made between Japanese owners and German charterers provided for arbitration in London by arbitrators appointed by the London Maritime Arbitrators’ Association. Also relevant to Clarke J’s decision that the parties intended English law to govern the charterparty were: (1) the fact that it was made on a well -known standard form containing claus es with well-known meanings in English law; and (2) that having agreed a “neutral” forum, the parties intended that forum to apply a “neutral” law, namely English law and not German or Japanese law. In such cases that implied choice of law will equally app ly to the arbitration agreement: see Habas Sinai Ve Tibbi Gazlar Istihsal v VSC Steel Co Ltd [2013] EWHC 4071 (Comm); [2014] 1 Lloyd’s Rep 479, para 102. 115. Such a situation may be contrasted with one in which the arbitration clause, although it specifies a p lace of arbitration, does not provide for a method of identifying the arbitrators except through appointment by an international arbitral body such as the ICC. As Andrew Baker J observed in his judgment in this case (at para 62), the ICC is “a quintessenti ally and deliberately supranational institution”, with its own internal, and so again supranational, supervisory apparatus of the International Court of Arbitration and its Secretary General and Secretariat. In a case of this kind the parties could not rea sonably assume that the selection of London as the seat of arbitration, even where it is a neutral forum, points ineluctably by necessary implication to a choice of English law to govern the contract so as to make the express designation of a governing law unnecessary. 116. Enka did not seek to argue on this appeal that the choice of London as the seat of arbitration in this case implies that the parties intended the construction contract as a whole to be governed by English law. But counsel for Enka submitted t hat, even though such an inference cannot be drawn in relation to the law intended to govern the parties’ substantive contractual obligations, it can nevertheless be drawn in relation to the arbitration agreement itself. 117. We do not accept this. Where there is insufficient reason to infer that the parties chose London as the seat of arbitration because they wanted the arbitrators to be versed in English law, that applies as much to any issues concerning the validity or scope of the arbitration agreement which the arbitrators might be asked to decide as it does to the substance of any dispute. Nor can any necessary implication be drawn from the possibility that issues concerning the validity or scope of the arbitration agreement might have to be decided by the English courts in the exercise of their supervisory jurisdiction. Questions of foreign law are dealt with in the English Commercial Court on a daily basis - the trial of the present case being an example - and, as Steyn LJ said in Star Shipping AS v China Shipping Foreign Trade Transportation Corp n (The Star Texas) [1993] 2 Lloyd’s Rep 445, 451- 452, even an express choice of jurisdiction does not by itself give rise to an implied choice of law. We therefore do not consider that a choice of the seat of arbitration can by Page 41 itself be construed as an implied choice of the law applicable to the arbitration agreement. VIII Applying the closest connection test 118. So far we have been considering the question whether the parties to a contract have chosen the law applicable to the arbitration agreement, either specifically or by choosing a system of law to govern the contract as a whole including the arbitration agreement. We now turn to the situation in which no such choice has been made. As discussed earlier (see para 36 above), the court must in these circumstances determine, objectively and irrespective of the parties’ intention, with which system of law the arbitration agreement has its closest connection . This exercise is different in nature from the attempt to identify a choice (whether express or implied), as it involves the application of a rule of law and not a process of contractual interpretation. 119. Even where the parties have not agreed what law is to govern their contract, it is reasonable to start from an assumption - for reasons given earlier - that all the terms of the contract, including an arbitration clause, are governed by the same system of law. Where, however, the parties have selected a place for the arbitration of disputes, t here is authority for, as a general rule, regarding the law with which the arbitration agreement is most closely connected as the law of the seat of arbitration . As we have seen, this was the approach adopted by the Court of Appeal in the Sulamérica case (see para 104 above). It was also endorsed by the Court of Appeal in C v D (see para 48 above), albeit that in that case insufficient reason was given, in our opinion, for rejecti ng the inference that the law chosen to govern the insurance contract was intended to apply to the arbitration clause. Among commentators, this rule notably has the support of Dicey, Morris & Collins on The Conflict of Laws , 15th ed (2012), rule 64(1)(b) a nd para 16-016; see also Russell on Arbitration , 24th ed, (2015) at para 2 -121 . 120. There are a number of reasons of principle and policy which in our opinion justify as a general rule regarding the law of the place chosen as the seat of arbitration as the law most closely connect ed with the arbitration agreement which in the absence of choice will apply by default. (i) The place of performance 121. The starting point is that the seat of arbitration is the place where (legally, even if not physically) the arbitratio n agreement is to be performed. In identifying the system of law with which a contract (or relevant part of it) has its closest and Page 42 most real connection, the place where the transaction is to be performed is the connecting factor to which the common law ha s long attached the greatest weight (since the place where the contract was concluded ceased to be seen as significant): see eg Dicey, Morris & Collins on The Conflict of Laws , 15th ed (2012), para 32 - 073. This is justified by the fact that states have an interest in regulating transactions taking place within their territory and by the consequent natural assumption that the law of the territory in which a transaction is taking place will govern it in the absence of a contrary indication. By agreeing to a s eat of arbitration the parties submit themselves to the jurisdiction of the courts of that place and to its law and coercive powers for the purposes of deciding any issue relating to the validity or enforceability of their arbitration agreement. Thus, as w e discuss later in this judgment (see Part XI below), the courts of the seat have jurisdiction to grant an injunction to restrain proceedings brought in breach of the agreement to arbitrate. The parties also by their choice of seat impliedly agree to bring any claim for a remedy relating to the existence or scope of the arbitrators’ jurisdiction (including any issue as to the validity or scope or the arbitration agreement), and any challenge to an arbitral award, in the courts of that place: see C v D [2007] EWHC 1541 (Comm); [2007] 2 All ER (Comm) 557, paras 29 -34 (Cooke J); C v D [2007] EWCA Civ 1282; [2008] Bus LR 843, para 17 (CA); Minister of Finance (Inc) v International Petroleum Investment Co [2019] EWCA Civ 2080; [2020] Bus LR 45, paras 36 -49; Dicey , Morris & Collins on The Conflict of Laws , 15th ed (2012), para 16- 036. The seat of arbitration is in these circumstances the place to whose system of law the arbitration agreement is most closely attached. 122. By contrast, there is no reason to regard the pl ace of performance of the substantive obligations created by the contract as a significant connection for the purpose of determining the law applicable to the arbitration agreement (as opposed to for the purpose of determining what law the arbitrators shou ld apply in deciding a dispute). This is because (as noted at para 40 above) the subject matter and purpose of an arbitration agreement are different from those of the contract in which it is incorporated. The irrelevance of the place of performance of the main contract is illustrated by the fact that seats of arbitration are frequently chosen which have no connection with where the parties’ substantive obligations are to be performed (or otherwise with the contract) and sometimes precisely because they hav e no such connection. Other factors connecting the main contract to a country or its laws are equally irrelevant in regard to the arbitration agreement. For example, article 4 of the Rome I Regulation adopts a presumption that the contract is most closely connected with the country where the party required to effect the characteristic performance of the contract has his habitual residence. There is no reason to regard this as a factor which should have any bearing on the law applicable to the arbitration ag reement. Page 43 123. We therefore agree with the view of Moore -Bick LJ in the Sulamérica case quoted at para 104 above and also with statement of Longmore LJ in C v D [2007] EWCA Civ 1282; [2008 ] Bus LR 843, para 26, that: “ an agreement to arbitrate will normally have a closer and more real connection with the place where the parties have chosen to arbitrate than with the place of the law of the underlying contract in cases where the parties have deliberately chosen to arbitrate in one place disputes which have arisen under a contract governed by the law of another place.” 124. We do not consider that the importance of the connection between the law governing the arbitration agreement and the law of the seat is undermined by the fact that some national laws, suc h as the Arbitration Act 1996 in England and Wales, allow the parties a wide degree of freedom to make their own arrangements, either by choosing another system of law to govern their arbitration agreement or arbitral procedure (see section 4(5) of the 199 6 Act, discussed earlier) or by agreeing to the application of institutional rules made by an arbitral body such as the ICC (see section 4(3) of the 1996 Act). The extent to which the parties are free to make such arrangements is itself a matter for the la w of the seat. Furthermore, any national law is likely to include mandatory provisions, described in section 1(b) of the 1996 Act as “such safeguards as are necessary in the public interest,” which have effect notwithstanding any agreement to the contrary. As noted earlier, in the 1996 Act these include sections 66 to 68, which govern any challenge to an award made in England including any challenge to the substantive jurisdiction of the arbitrators on grounds that the arbitration agreement is invalid or un enforceable or does not cover the dispute referred to arbitration. Such provisions of themselves establish a close nexus between the law determining the validity and scope of the arbitration agreement and the law of the seat of arbitration. (ii) Consistency with international law and legislative policy 125. A second, and in our view compelling, reason for treating an arbitration agreement as governed by the law of the seat of arbitration in the absence of choice is that such a rule accords with international law as embodied in the 1958 New York Convention and other international instruments, as well as with the national law which gives effect to the New York Convention in England and Wales. 126. The New York Convention , to which the United Kingdom became a party in 19 75 and which more than 160 states have now signed, has been described as “the single most important pillar on which the edifice of international arbitration rests,” and as “perhaps … the most effective instance of international legislation in the Page 44 entire history of commercial law”: see Redfern and Hunter : Law and Practice of International Commercial Arbitration, 6th ed (2015), para 2.11, quoting Wetter, “The present status of the International Court of Arbitration of the ICC: An appraisal” (1990) 1 Am Rev In tl Arb 91, p 93, and Mustill, “Arbitration: History and background” (1989) 6 J Intl Arb 43 , p 49. The essential aim of the Convention was to establish a single uniform set of international legal standards for the recognition and enforcement of arbitration agreements and awards. Its success is reflected in the fact that, according to Born, International Commercial Arbitration , 2nd ed (2014) , p 113, the New York Convention has been implemented through national legislation in virtually all contracting states. 127. Article V(1)(a) of the Convention specifies, among the limited circumstances in which r ecognition or enforcement by the courts of a Convention state of an award made in another Convention state may be refused, proof that the arbitration agreement “is not v alid under the law to which the parties have subjected it or, failing any indication thereon, under the law of the country where the award was made”. As stated in Dicey, Morris & Collins on The Conflict of Laws , 15th ed (2012), para 16- 014: “In the light of the pervasive reach of the New York Convention in modern times, this rule, although not itself prescribing a choice of law rule of general application, nevertheless provides a strong indication of one …” 128. Article V(1)(a) - enacted into English law by section 103(2)(b) of the Arbitration Act 1996 - has two limbs, which are intended to be treated as uniform international conflict of laws rules: see Dallah Real Estate and Tourism Holding Co v Ministry of Religious Affairs of the Government of Pakistan [2008] EWHC 1901 (Comm); [2009] 1 All ER (Comm) 505 , para 78 (Aikens J); and [2010] UKSC 46; [2011] 1 AC 763 , para 123 (Lord Collins). The first, and primary, rule is that the validity of the arbitration agreement is governed by “the law to which the parti es [ have ] subjected it” - in other words the law chosen by the parties. The second, default rule, which applies where no choice has been indicated is that the applicable law is that of “the country where the award was made”. Where the parties have chosen t he seat of arbitration, this will be (or be deemed to be) the law of the seat. In English law this is expressly provided by section 100(2)(b) of the 1996 Act. 129. There is a division of opinion among commentators over whether the first limb of article V(1)(a) applies only where there is an express choice of law to govern the arbitration agreement or whether it also encompasses a choice that is implied - for example from a choice of law to govern the contract in general: compare van de n Berg, The New York Arbitr ation Convention of 1958 (1981) , p 293 and Born, International Commercial Arbitration , 2nd ed (2014), pp 564- 565. We think the Page 45 latter is the better view. As discussed earlier, a choice of law for the arbitration agreement may be clearly indicated by a choice of law for the contract of which it forms part and a choice conveyed impliedly is just as much a choice entitled to respect in accordance with the principle of party autonomy as a choice stated expressly. Furthermore, the broader interpretation is supported by the language of article V(1)(a), which applies the default rule only failing “any indication” of the law to which the p arties have subjected the arbitration agreement . 130. Where proceedings are brought in a court of a contracting state in respect of a matter covered by an arbitration agreement to which the New York Convention applies, article II(3) of the Convention requires t he court, at the request of one of the parties, to refer the parties to arbitration, unless the agreement “is null and void, inoperative or incapable of being performed.” Article II does not itself specify rules for identifying the law by which the validit y of the arbitration agreement is to be determined. There is, however, a strong and widely accepted argument that the Convention is to be interpreted as requiring the same conflict rules to be applied in relation to article II(3) as are specifically required at the stage of enforcement by article V(1)(a ). Thus, Professor van de n Berg, a leading authority on the New York Convention, has written: “A systematic interpretation of the Convention, in principle, permits the application by analogy of the conflict rules of a rticle V(1 )(a) to the enforcement of the agreement. It would appear inconsistent at the time of the enforcement of the award to apply the Convention ’s uniform conflict rules and at the time of the enforcement of the agreement to apply possibly different conflict rules of the forum. It could lead to the undesirable situation of the same arbitration agreement being held to be governed by two different laws: one law determined according to the conflict rules of the forum at the time of the enforcement of the agreement, and the other determined according to article V( 1)(a) at the time of enforcement of the award.” van de n Berg, The New York Arbitration Convention of 1958 (1981) , p 126- 7; and see Born, International Commercial Arbitration , 2nd ed (2014), pp 494, 495- 499; Lew & Mistelis, Comparative International Commercial Arbitration ( 2003) , para 6- 55; Schramm, Geisinger & Pinsolle, “Article II” in Kronke , Nacimiento et al (eds), Recognition and Enforcement of Foreign Arbitral Awards : A Global Commentary on the New York Convention (2010), p 55. 131. This approach is also supported by other international instruments. The 1961 European Convention on International Commercial Arbitration adopts the conflict rules set out in article V(1)(a) of the New Yor k Convention and, by a rticle Page 46 V I(2), provides for those rules to be applied at any stage when a court of a c ontracting state is required to rule on the existence or validity of an arbitration agreement - in other words, whether the question arises pre - or p ost-award. 132. Article 36 of the Model Law adopted by the United Nations Commission on International Trade Law (UNCITRAL ) on 21 June 1985 parallels article V of the New York Convention in its list of grounds (set out in article 36) on which recognition or enfo rcement of an arbitral award may be refused. The Model Law takes this a step further in article 34 by restricting any challenge to an arbitral award to an application brought in the state in which the award was made and by limiting the grounds on which an award may be set aside to those on which recognition or enforcement of a foreign award may be refused. 133. The primary reason for the exclusion of arbitration agreements from the Rome I Regulation was that such agreements were already adequately regulated by international conventions: see McParland, The Rome I Regulation on the Law Applicable to Contractual Obl igations (2015) , paras 7 -126 - 7 -127. The exclusion can accordingly be seen as a recognition of the fact that arbitration agreements are already subject to international uniform conflict rules derived, in particular, from the 1958 New York Convention and the 1961 European Convention. 134. Although the United Kingdom has not signed the 19 61 European Convention and has not in all respects adopted the UNCITRAL Model Law, the rules laid down in a rticle V of the New York Convention (and article 36 of the Model Law) relating to the recognition or enforcement of awards have been directly incorporated into English law by section 103 of the 1996 Act. Thus, under section 103(2)(b) the grounds on which recognition or enforcement of an award made in another Convention state may be refused include proof that: “the arbitration agreement was not valid un der the law to which the parties subjected it or, failing any indication thereon, under the law of the country where the award was made;” 135. While this provision only applies directly in proceedings brought to enforce an award made in another Convention state , it would be illogical to apply different conflict rules to determine which law governs the validity of the arbitration agreement where the arbitration is seated (and the award therefore treated as made) in England . Thus, in cases where the parties have n ot chosen the law of the arbitration agreement but have chosen the seat of arbitration, it would be illogical if the English courts were to treat the validity of the arbitration agreement as governed by the law of the seat if the parties have chosen a fore ign seat but by the law of the main contract if they have been chosen an English seat of arbitration. Such an Page 47 approach would be all the more incoherent given that, if proceedings were brought in another Convention state to enforce an award made in England, the foreign court would apply the law of the seat (and not the law of the main contract, if different) to determine the validity of the award as required by article V(1)(a) of the Convention. 136. As pointed out by Professor van de n Berg in the passage quoted at para 130 above, it would be equally illogical if the law governing the validity of the arbitration agreement were to differ depending on whether the question of validity is raised before or after an award has been made. To ensure consistency and coheren ce in the law, the same law should be applied to answer the question in either case. Again, the incoherence that would result if English common law were to adopt a different conflict rule from the New York Convention’s uniform rule would be compounded when the international perspective is considered. As one commentator has observed: “ It is fair to say that today, the conflict rule contained in article V(1)(a) New York Convention … has developed into a truly transnational conflict rule for the determination of the law governing the substantive validity of the arbitration agreement. This rule has been applied in numerous international arbitral awards, is favo ured by international arbitral doctrine and has been accepted by domestic courts .” See Berger, “Re Page 48 Exactly the same points apply to the ap proach taken by national courts to the scope of an international arbitration agreement. 137. As with questions of validity, issues about whether a dispute falls within the scope of the arbitration agreement may arise at any stage from when a party wishes to refer a dispute to arbitration to the stage of seeking to enforce an award. Article V(1)(c) of New York Convention provides that recognition and enforcement may be refused if “[t]he award deals with a difference not contemplated by or not falling within the t erms of the submission to arbitration, or it contains decisions on matters beyond the scope of the submission to arbitration … ”. Section 103(2)(d) of the 1996 Act contains an almost identical provision , as does article 36(i)(a)(iii) of the UNCITRAL Model Law and article IX(1)(c) of the European Convention . 138. The general approach in the conflict of laws, adopted by both the common law and the Rome I Regulation, is to treat the validity and scope of a contract (as well as other issues such as the consequences o f breach and ways of extinguishing obligations) as governed by the same applicable law. This makes good sense, not least because the boundary between issues of validity and scope is not always clear. Thus, it is logical to apply the law identified by the c onflict rules prescribed by article V(1)(a) of the New York Convention and section 103(2)(b) of the 1996 Act to questions about the scope or interpretation of the arbitration agreement as well as disputes about its validity. 139. This also accords with the appr oach taken by the American Law Institute in the final draft of the Restatement (Third) of the US Law of International Commercial and Investor- State Arbitration (24 April 2019). Section 2.14 of the draft Restatement recommends a rule that a court should determine whether an international arbitration agreement is null and void in accordance with: (1) the law to which the parties have subjected the arbitration agreement; or (2) in the absence of such a choice of law, t he law of the seat of arbitration. This approach is consistent with article V(1)(a) of the New York Convention. The comment on the applicable law explains: “ On balance, the present s ection favors ensuring symmetry between pre -arbitration and post -award sta ndards for determining the validity of an arbitration agreement. There is no reason in principle why a court should answer that question differently depending on the stage of the proceedings, and doing so would inject unnecessary uncertainty and complexity into the analysis.” Page 49 140. Section 2.15 of the draft Restatement adopts the same rule for the purpose of determining whether a matter falls within the scope of an arbitration agreement, taking the position that the law applicable to determining the scope of an a greement to arbitrate should parallel the law applicable to determining whether the agreement is valid. 141. Accordingly, whatever merit there might be, if one were designing a system of law from scratch, in a conflicts rule which treated the law of the main co ntract as applicable to the arbitration agreement in the absence of choice, it would in our view be wrong for the English common law to adopt a rule out of step with both the legislative policy of the 1996 Act and the underlying uniform rule established by the New York Convention. The court should apply the same conflict rules to identify the governing law irrespective of whether the arbitration has a domestic or foreign seat and irrespective of the stage at which an issue about the validity or scope of the arbitration agreement is raised. Internal coherence of English law, as well as harmony with international law and practice, is achieved by treating the applicable law in all cases, in the absence of a choice by the parties, as the law of the seat of arbit ration. (iii) Giving effect to commercial purpose 142. A third reason for applying the law of the seat as a default rule is that it is likely to uphold the reasonable expectations of contracting parties who have chosen to settle their disputes by arbitration in a specified place but made no choice of law for their contract. This is particularly so where, as is often the case in contracts made between parties of different nationalities, a popular seat of international arbitration has been chosen as a neutral foru m with which neither party is connected. In such circumstances, if the parties had been required to make a common choice of law to govern their arbitration agreement at the time of contracting, it is inherently unlikely that they would have agreed on eithe r of their national systems of law and much more likely that they would have settled on the law of the place which they had chosen as the seat of arbitration. Not only does this provide a neutral choice of law but it is already the law of that place which - in countries which have implemented the Model Law or are parties to the New York Convention - will determine the validity of an award if an application is made to set it aside or if its enforcement in the other party’s home state is resisted. 143. Countries frequently chosen as neutral seats of arbitration can also be expected to have legal regimes which are supportive of arbitration and which seek to give effect to the parties’ intention that they do not wish to have their disputes decided by a court. That is the case for all the most preferred seats of international arbitration - which, according to the most recent 2018 international arbitration survey conducted by the School of International Arbitration at Queen Mary Page 50 University of London, are London, Paris, Singapore, Hong Kong and Geneva. As discussed earlier, it is reasonable to assume that parties who have chosen to settle their disputes by international arbitration want an arbitration that resolves all (and not only some) disputes through an award that is binding and enforceable and which is immune from collateral attacks, particularly in the home country of one of the parties. As a general rule, applying the law of the chosen seat of arbitration is calculated to achieve that purpose. (iv) Legal certainty 144. Finally, there is merit is recognising a clear default rule in the interests of legal certainty. Applying a general rule that, in the absence of choice, an arbitration agreement is governed by the law of the seat of arbitration (where a seat has been desig nated) enables the parties to predict easily and with little room for argument which law the court will apply by default. The benefits of certainty are further enhanced if the same law is applied irrespective of the country in which the proceedings are bro ught and whether the question of the validity or scope of the arbitration agreement is raised before or after an award has been made. Certainty might not be a sufficient reason to recognise a clear and uniform rule if the rule interfered with party choice. But here there is no risk of such interference because we are concerned with the situation in which the parties have not exercised their freedom to choose the law to be applied so that the court must make the selection for them. It is desirable that parti es should be able to know with certainty what law a court will apply in this situation. If they do not like the default option, they can always choose a system of law that they prefer. (v) Conclusion on the default rule 145. Chubb Russia did not argue against t he contention that the law most closely connected with the arbitration agreement, which in the absence of choice will apply by default, will in general be the law of the seat of the arbitration. Indeed, leading c ounsel for Chubb Russia at one point in oral argument volunteered the suggestion that an appropriate default rule would be that the arbitration agreement is governed by the law of the seat. He noted that such a rule would have the advantages of certainty and consistency with article V(1)(a) of the N ew York Convention. Counsel later withdrew that suggestion and in reaching our conclusion on this issue we have placed no reliance on the fact it was made. But it was in our view no more than a realistic acknowledgement of the overwhelming case for recogni sing such a general rule. 146. A case can be made for recognising an exception to the ordinary default rule where the arbitration agreement would be invalid under the law of the seat but not Page 51 under the law governing the rest of the contract: see eg Merkin & Flannery on The Arbitration Act 1996, 6th ed (2019) , para 46.10.5 and Born, International Commercial Arbitration , 2nd ed (2014), pp 54 2-54 9; for a contrary view, see Glick and Venkatesan, “Choosing the Law Governing the Arbitration” Agreement’ in Kaplan and Moser (eds) , Jurisdiction, A dmissibility and Choice of Law in International Arbitration: Liber Amicorum Michael Pryles (2018) , Chapter 9, pp 148 -149 . Since the issue does not arise in the present case, it is not necessary to decide whether such an exception should be recognised. Even if there be no such exception, where the law of the seat is English law an arbitration agreement will only be invalid in limited circumstances and for good reason. Where the law of the seat is not English law, an award made under an arbitration agreement invalid under that law is liable in any event to be set aside by the courts of the seat, whose decision would normally be followed by the English courts: see the discussion by Lord Mance in “Arbitration - a law unto itself?” (2016) 32 Arbitration International 223. There can also be cases where no seat has been designated, where it may be appropriate to apply the law applicable to the rest of the contract. But such exceptional cases apart, we consider that the law of the seat will apply by default. IX The law applicable to the arbitration agreement in article 50.1 147. Applying the principles discussed above to the present case, it is common ground that the parties have not chosen a system of law specifically to govern the arbitration agreement contained in article 50.1 of the construction contract . Chubb Russia, however, contends that the parties have chosen Russian law to govern the contract as a whole including the arbitration agreement. Enka disputes this. Enka accepts that the main body of t he construction contract is governed by Russian law but maintains that this is so only because of the connections between the c onstruction contract and the law of Russia and not as a matter of choice. (i) No choice of law 148. The first thing to note is that th e construction contract does not contain a choice of governing law clause. Amongst almost 100 pages of primary text and another 400 pages of appendices, there is no provision which says that the contract shall be governed by or interpreted in accordance wi th a specified system of law. In a detailed and professionally drafted commercial contract made between substantial organisations based in different countries, such a clause is an entirely standard clause, almost invariably included along with a clause spe cifying the forum in which any dispute is to be resolved. It is difficult to conceive that the omission of such a clause in this case - despite the inclusion of a detailed provision dealing with the resolution of disputes - was accidental. We agree with co unsel for Enka that an obvious explanation for its absence is that the parties were not able to agree on a choice of the governing law. Page 52 149. Chubb Russia contends that a choice of Russian law can nonetheless be discerned from the use in the construction contrac t of the term “Applicable Law” , taken together with the definition of that term in A ttachment 17 as : “Law of the Russian Federation, including legislation of the Russian Federation, all regulatory legal acts of the State Authority Federal Bodies, State Aut horities of the constituent entities of the Russian Federation, legislation of the constituent entities of the Russian Federation, regulatory legal acts by Local Authorities and any other applicable regulatory legal acts .” There are numerous references throughout the body of the contract to the “ Applicable Law ”, as well as other references to the law or laws of the Russian Federation . Counsel for Chubb Russia submitted that, read as a whole, the language of the c onstruction contract makes it clear that the parties were contracting by reference to Russian law and chose Russian law as the law applicable to their agreement. 150. Had it been the parties’ choice, however, that the construction contract should be governed by the “Applicable Law” as defined in Attachmen t 17, it would have been simple to say so. Yet, as noted, there is no clause which s tates this. Rather, the term “Applicable Law” is used in specific provisions of the contract which impose obligations on the contractor to comply with laws and regulations applicable in the country where the construction work was to take place . As the Court of Appeal observed ( at para 107 ), it is a common technique in international construction co ntracts to define such an applicable law or laws and to impose an obligation to comply with them separately from any choice of the law that is to govern the validity and interpretation of the parties’ contractual rights and obligations. 151. As evidence of this practice, the Court of Appeal cited a leading text on the widely used standard forms of international construction contract issued by the International Federation of Consulting Engineers (“FIDIC”): see Baker Mellors Chalmers and Lavers on FIDIC Contracts : Law and Practice, 5th ed (2009) , paras 2.126, 2.140 and 2.145. Counsel for Chubb Russia pointed out that the contract in this case was not made on a FIDIC standard form and, unlike contracts made on FIDIC forms, does not contain a governing law clause. Th ey observed that the technique employed in drafting FIDIC contracts is to select a governing law and then to apply a different law (usually the local law) expressly to certain provisions in such a way that the contractor will be obliged to comply with that law. That was not done here, where the only law specified was the “Applicable Law”. Page 53 152. The drafting technique to which the Court of Appeal referred is not , however, peculiar to FIDIC standard forms. Authoritative texts cited by counsel for Enka confirm that other standard forms of international construction contract also typically include provisions which require the contractor to comply with applicable laws or with laws of the country where the works are carried out: see Huse, Understanding and Negotiating T urnkey and EPC Contracts, 4th ed (2020) , paras 4 -110 - 4 -112; Bailey, Construction Law , 2nd ed (2016), para 18.11. The clear purpose of such provisions is to protect the employer against the risk of incurring liability through failure by the contractor to comply with local laws such as building regulations, health and safety and environmental laws, tax laws and other applicable regulatory requirements. The rationale for including such provisions is not affected by the presence or absence of a governing law clause in the contract. 153. There is no necessary inference that the validity and interpretation of a contractual obligation requiring compliance with a law or laws of a particular country is itself to be determined by applying the contract law of that country . This is underlined by the point which Chubb Russia itself makes that the law chosen to govern a contract made on a FIDIC standard form (or, we would add, other forms of international construction contract) may and often does differ from the “applicable” law with which the contractor is required to comply in performing the contract. In any case the contractual obligations of Enka were not limited to compliance with the “Applicable Law”. Article 4.1 of the construction contract provides: “The Contractor sha ll ensure performance of the Work in accordance with: a) The requirements of this Agreement (including references to the non-mandatory rules of Applicable Law but to the extent the provisions of the Agreement are not at variance with mandatory rules of App licable Law); b) Applicable Law (including the Mandatory Technical Rules constituting a part of such Applicable Law); c) An Implied Covenant of Good Faith and Fair Dealing.” Page 54 The definition in Attachment 17 of the phrase “Implied Covenant of Good Faith and Fair Dealing” imports standards applied by “ experienced international contractor organisations” engaged in similar projects . As well as such standards, the construction contract and its attachments set out many specific requirements for the work which do not form part of the “ Applicable Law”. 154. Quite apart from this, there are numerous rights and obligations established by the con struction contract which make no reference to the “Applicable Law” (or to laws of the Russian Federation). Examples are clauses dealing with the consequences of delay (article 26), force majeure (article 31), payment of the price (article 33) and terminati on (article 43). 155. In these circumstances , it cannot be said that the parties have in the construction contract expressly selected a system of law to govern the validity and interpretation of their contractual obligations nor that the terms of the contract c onstrued in their context point ineluctably to the conclusion that the parties intended Russian law to apply. To the contrary, t he obvious inference from the fact that the parties have not anywhere in the contract stated what system of law is to govern any of their contractual obligations - as opposed to creating obligations to comply with applicable laws - is that they have not agreed (for whatever reason) on a choice of governing law. This inference applies to the arbitration agreement as much as to the rest of the contract. (ii) Closest connection 156. In the absence of any choice of the law that is to govern the arbitration agreement, it is necessary to fall back on the default rule and identify the system of law with which the arbitration agreement is most c losely connected. In accordance with our earlier analysis, this will generally be the law of the seat chosen by the parties, which in this case is London. 157. As already mentioned, Chubb Russia did not actively oppose this conclusion if it is necessary to iden tify the law with which the arbitration agreement is most closely connected . Chubb Russia’s case has been put solely on the basis that the parties chose Russian law as the law governing the contract including the arbitration agreement. No alternative argum ent has been advanced that, if this is wrong, Russian law nevertheless applies as the law most closely connected with the arbitration agreement. 158. Chubb Russia has put forward an argument, however, about the proper interpretation of particular terms of the construction contract which it remains Page 55 relevant to consider. This argument is that the agreement to arbitrate disputes is embedded in a clause of the contract (article 50) dealing with dispute resolution which contains other obligations in addition to the obligation to arbitrate and which itself is , as Mr Bailey QC put it, “buried deep inside” the contract and inextricably connected to other provisi ons of it. It is said that in these circumstances the parties must have intended all the obligations in article 50, including the arbitration agreement , to be governed by the same system of law as each other and as the rest of the contract. For the purpose of this argument, it is necessary to determine the law applicable to the main body of the construction contract. As discussed earlier, for that purpose the court must apply the Rome I Regulation. (iii) The law applicable to the main contract 159. Although it w ould be a mistake to interpret the Rome I Regulation through the prism of the common law, there does not appear to be any substantial difference (save possibly in relation to the admissibility of subsequent conduct) between the approach of the common law t o determining whether there has been an express or implied choice of law and the approach to be followed in deciding whether a choice has been made expressly or “clearly demonstrated” for the purpose of article 3 of the Rome I Regulation. Thus, in Lawlor v Sandvik Mining and Construction Mobile Crushers and Screens Ltd [2013] EWCA Civ 365; [2013] 2 Lloyd’s Rep 98, para 33, the Court of Appeal held that the test of whether a choice has been “clearly demonstrated” is objective and is equivalent to Lord Diplock’s formulation of the common law test, requiring the court to be satisfied that the only reasonable conclusion to be drawn from the circumstances is that the parties should be taken to have intended the putative law to apply. 160. For the reasons already given when considering the position at common law, the parties have not in this case expressly made or clearly demonstrated a choice of law to govern the construction contract but are, as it seems to us, reasonably to be understood as having not agreed on a choice of law. The governing law is therefore to be determined by applying article 4 of the Rome I Regulation. 161. Under the construction contract Enka was engaged to install a boiler and auxiliary equipment, with the equipment and materials (except for consumable materials) to be supplied by Energoproekt as customer. The contract was therefore, at least predominantly, a contract for the provision of services by Enka. Article 4(1)(b) of the Rome I Regulation establishes a prima facie rule that, to the extent that the law applicable to it has not been chosen in accordance with article 3, a con tract for the provision of services shall be governed by the law of the country where service provider “has his habitual residence”. This rule points towards the law of Turkey as the country where the contractor, Enka, had its place of central administrati on and therefore habitual residence (see article 19(1)). However, the Page 56 other party to the contract, Energoproekt, was a Russian company, as was the “End Customer”, Unipro. The contract was for the performance of construction work in Russia and required comp liance with Russian laws and regulations. It is written in the Russian language (as the authoritative version); notifications under it were likewise required to be written in Russian and English but with the Russian version taking precedence and, when sent to the contractor, were to be sent to its Moscow office. The price for the work, although calculated in US dollars, was to be paid in roubles to a Russian bank account. The fact that the dispute resolution clause provides for arbitration in London is not a sufficient connection to indicate that English law should govern the contractual obligations of the parties. It is clear from all the circumstances of the case that the main body of the construction contract is manifestly more closely connected with Russ ia than with any other country. Pursuant to article 4(3) of the Rome I Regulation, it is therefore governed by Russian law. (iv) The dispute resolution clause 162. Chubb Russia’s argument that the arbitration agreement cannot reasonably be detached from the res t of the contract in terms of its governing law has two aspects. The first is that article 50.1, which contains the arbitration agreement, must be governed by a single law. The second is that it makes no sense for that law to differ from the law applicable to the rest of the construction contract. 163. Article 50.1 sets outs a series of procedures of increasing formality which the parties have agreed to follow for resolving any dispute, with arbitration being the last resort. Thus, where a “Dispute” as defined i n the first sentence of article 50.1 arises, the parties are first of all obliged to “make in good faith every reasonable effort” to resolve it by negotiations. If the Dispute is not resolved within ten days of either party sending a “Notification” (a term defined in article 51.2 of the contract) to the o pposite party containing an indication of the Dispute, either party may then give a written notice causing it to be referred to a meeting between the parties’ senior managements. It is only if the matter is not resolved within a further 20 calendar days that the obligation arises to refer the Dispute to international arbitration. 164. Enka accepts that article 50.1 can only reasonably be interpreted as governed by a single system of law, as it is clearly intended to establish a single, staged dispute resolution process and it would make no sense for the meaning or scope of a “Dispute” as defined in the earlier part the clause to be determined by applying a different system of law from the law governing the validit y and scope of the obligation to arbitrate. But it is Enka’s case that the implication in terms of governing law flows in the opposite direction from that contended for by Chubb Russia, and that it follows from the identification of English law as the law which (on Enka’s case) governs the arbitration agreement that English law applies to the Page 57 whole of article 50.1. That conclusion should be reached, Enka contends, either by applying the common law rules to the whole of article 50.1 on the basis that the who le of that clause constitutes an “arbitration agreement” within the meaning of a rticle 1(2)(e) of the Rome I Regulation or by applying the principle of dépeçage and treating article 50.1 as a severable part of the contract for the purpose of the Rome I Reg ulation. 165. Mr Bailey QC for Chubb Russia drew attention to connections between article 50.1 and other parts of the contract: in particular the use of capitalised terms such as “Notification” which are defined elsewhere. He also pointed out that article 42.2 of the contract includes provision for referring disputes arising out of the operation of the change control procedure to arbitration pursuant to article 50. He submitted that the dispute resolution clause is not hermetically sealed from the rest of the co ntract but is inextricably bound up with it, and that this points strongly to the conclusion that the arbitration agreement and the other obligations contained within the contract must all be governed by the same system of law. 166. This contention could be formulated on the basis of implied choice or by reference to the closest connection test. As to the former, no doubt parties could in principle agree that the whole of their contract, including an arbitration agreement within it, should be governed by a singl e system of law even though they have not agreed on what that law should be. But this does not seem to us an inherently likely agreement for contracting parties to make. To establish such an agreement a clearer demonstration of intent would be necessary th an the mere fact that the arbitration agreement forms part of a wider dispute resolution clause which is referred to elsewhere and uses terms defined elsewhere in the contract. 167. In terms of connections, we agree with both parties that article 50.1 makes sen se only as an integrated whole governed by one system of law. But we do not regard the connections to which Chubb Russia drew attention between article 50.1 and the rest of the contract as particularly strong or sufficient to require the application of the same law in circumstances where no choice of law has been made by the parties. There is no difficulty in principle in using within a contract or clause of a contract governed by a particular system of law a term defined in another part of the contract or in a separate instrument governed by a different system of law. In such a case the term will carry its defined meaning by agreement. The reference in article 50.1 to a “Notification” can readily operate in this way. Likewise, the cross - reference in article 42.2 of the construction contract to the dispute resolution clause does not require both clauses to be governed by the same system of law. 168. It has become increasingly common for commercial parties to include in their contracts provisions which require other forms of dispute resolution, such as good faith negotiation or mediation, to be undertaken without success before a dispute is Page 58 referred to arbitration. We find it difficult to see how, as a matter of principle or policy, the fact that such an approach is adopted can justify the application of a different law to determine the validity or scope of the arbitration agreement. All the reasons tha t we have identified for, as a general rule, regarding the law governing the arbitration agreement in the absence of choice as the law of the seat of arbitration apply equally and with equal force where the arbitration agreement is contained in a wider dis pute resolution clause (or integrated set of clauses) as where it is self - contained. We do not think that reasonable commercial parties would expect the law applied to determine the validity and scope of their arbitration agreement to depend on which form of dispute resolution procedure is chosen. Rather, it is reasonable to expect that, where a multi -tiered procedure is chosen, the law which determines the validity and scope of the arbitration agreement will determine the validity and scope of the whole di spute resolution agreement. 169. The fact that two conflict of laws regimes are potentially in play complicates the analysis but provides no reason to alter the result. Where, as in this case, an obligation to arbitrate disputes is embedded in a single dispute resolution agreement which provides for other steps to be undertaken before the obligation to arbitrate arises, we do not think it unreasonable to regard the whole dispute resolution agreement as an “arbitration agreement” for the purpose of article 1(2)(e ) of the Rome I Regulation. On this basis, applying the common law conflict of laws rules, article 50.1 of the construction contract is governed by English law. X Conclusions on applicable law 170. It may be useful to summarise the principles which in our judgm ent govern the determination of the law applicable to the arbitration agreement in cases of this kind: i) Where a contract contains an agreement to resolve disputes arising from it by arbitration, the law applicable to the arbitration agreement may not be the same as the law applicable to the other parts of the contract and is to be determined by applying English common law rules for resolving conflicts of laws rather than the provisions of the Rome I Regulation. ii) According to these rules, the law applicable to the arbitration agreement will be (a) the law chosen by the parties to govern it or (b) in the absence of such a choice, the system of law with which the arbitration agreement is most closely connected. Page 59 iii) Whether the parties have agreed on a choice of law t o govern the arbitration agreement is ascertained by construing the arbitration agreement and the contract containing it, as a whole, applying the rules of contractual interpretation of English law as the law of the forum. iv) Where the law applicable to the a rbitration agreement is not specified, a choice of governing law for the contract will generally apply to an arbitration agreement which forms part of the contract . v) The choice of a different country as the seat of the arbitration is not, without more, suff icient to negate an inference that a choice of law to govern the contract was intended to apply to the arbitration agreement. vi) Additional factors which may, however, negate such an inference and may in some cases imply that the arbitration agreement was int ended to be governed by the law of the seat are: (a) any provision of the law of the seat which indicates that , where an arbitration is subject to that law, the arbitration agreement will also be treated as governed by that country’s law ; or (b) the existence of a s erious risk that, if governed by the same law as the main contract, the arbitration agreement would be in effective. Either factor may be reinforced by circumstances indicating that the seat was deliberately chosen as a neutral forum for the arbitration. vii) Where there is no express choice of law to govern the contract, a clause providing for arbitration in a particular place will not by itself justify an inference that the contract (or the arbitration agreement) is intended to be governed by the law of that place. viii) In the absence of any choice of law to govern the arbitration agreement, the arbitration agreement is governed by the law with which it is most closely connected. Where the parties have chosen a seat of arbitration, this will generally be the law of the seat, even if this differs from the law applicable to the parties’ substantive contractual obligations. ix) The fact that the contract requires the parties to attempt to resolve a dispute through good faith negotiation, mediati on or any other procedure before referring it to arbitration will not generally provide a reason to displace the law of the seat of arbitration as the law applicable to the arbitration agreement by default in the absence of a choice of law to govern it. Page 60 171. Ap plying these principles, we have concluded that the contract from which a dispute has arisen in this case contains no choice of the law that is intended to govern the contract or the arbitration agreement within it. In these circumstances the validity and scope of the arbitration agreement (and in our opinion the rest of the dispute resolution clause containing that agreement) is governed by the law of the chosen seat of arbitration, as the law with which the dispute resolution clause is most closely connec ted. We would therefore affirm - albeit for different reasons - the Court of Appeal’s conclusion that the law applicable to the arbitration agreement is English law. 172. We have not found it necessary to consider arguments made by Enka that, if the arbitration agreement were governed by the law of Russia as the place of performance of the construction project and country with which the parties’ substantive contractual obligations have their closest connection, there would be a serious risk that the parties’ int ention of having their disputes finally settled by arbitration in a neutral forum would be defeated. This was disputed by Chubb Russia, but in the light of the conclusion we have reached there is no need to resolve this further issue. XI The anti -suit inju nction 173. If, as we have held, the arbitration agreement is governed by English law, Chubb Russia does not dispute that it was legitimate for the Court of Appeal to exercise its discretion whether to grant an anti -suit injunction afresh and does not contend that it erred in so doing. Its challenge to the order made by the Court of Appeal rests on the assumption that the arbitration agreement is governed by Russian law. Chubb Russia contends that the English courts ought in these circumstances to defer to the d ecision of the Russian courts on whether their dispute must be referred to arbitration or may be resolved by litigation in the Russian courts. On Chubb Russia’s case the English court’s approach to the grant of anti -suit injunctions should differ according to whether the arbitration agreement is governed by English law or a foreign law. As we have held that the arbitration agreement is governed by English and not Russian law, it is not necessary to address this further ground of appeal. Nevertheless, given that it has been fully argued and the importance of the issues raised, we shall briefly address it. 174. As already noted, by choosing a seat of arbitration the parties are choosing to submit themselves to the supervisory and supporting jurisdiction of the cour ts of that seat over the arbitration. A well established and well recognised feature of the supervisory and supporting jurisdiction of the English courts is the grant of injunctive relief to restrain a party from breaching its obligations under the arbitra tion agreement by bringing claims which fall within that agreement in court Page 61 proceedings rather than, as agreed, in arbitration. A promise to arbitrate is also a promise not to litigate. 175. As explained by Lord Hoffmann in West Tankers Inc v RAS Riunione Adriatic a di Sicurtà SpA (The Front Comor) [2007] UKHL 4; [2007] 1 Lloyd’s Rep 391 , at paras 20 -22: “ 20. Of course arbitration cannot be self- sustaining. It needs the support of the courts … Different national systems give support in different ways and an important aspect of the autonomy of the parties is the right to choose the governing law and seat of the arbitration according to what they consider will best serve their interests. 21. The Courts of the United Kingdom have for many years exercised the jurisdiction to restrain foreign court proceedings as Colman J did in this case: see Pena Copper Mines Ltd v Rio Tinto Co Ltd (1911) 105 LT 846. It is generally regarded as an important and valuable weapon in the hands of a court exercising supervisory jurisdiction over the arbitration. It promotes legal certainty and reduces the possibility of conflict between the arbitration award and the judgment of a national court. … it saves a party to an arbitration agreement from having to keep a watchful eye upon parallel court proceedings in another jurisdiction, trying to steer a course between so much involvement as will amount to a submission to the jurisdiction … and so little as to lead to a de fault judgment. That is just the kind of thing that the parties meant to avoid by having an arbitration agreement. 2 2. Whether the parties should submit themselves to such a jurisdiction by choosing this country as the seat of their arbitration is, in my o pinion, entirely a matter for them. The courts are there to serve the business community rather than the other way round. No one is obliged to choose London. The existence of the jurisdiction to restrain proceedings in breach of an arbitration agreement cl early does not deter parties to commercial agreements. On the contrary, it may be regarded as one of the advantages which the chosen seat of arbitration has to offer. ... ” 176. In the same case Lord Mance stated at paras 31 -32: Page 62 “ 31. The purpose of arbitration (enshrined in most modern arbitration legislation) is that disputes should be resolved by a consensual mechanism outside any court structure, subject to no more than limited supervision by the courts of the place of arbitration. Experience as a commercial j udge shows that, once a dispute has arisen within the scope of an arbitration clause, it is not uncommon for persons bound by the clause to seek to avoid its application. Anti -suit injunctions issued by the courts of the place of arbitration represent a ca refully developed - and, I would emphasise, carefully applied - tool which has proved a highly efficient means to give speedy effect to clearly applicable arbitration agreements. 3 2. It is in practice no or little comfort or use for a person entitled to th e benefit of a London arbitration clause to be told that (where a binding arbitration clause is being - however clearly - disregarded) the only remedy is to become engaged in the foreign litigation pursued in disregard of the clause. Engagement in the fore ign litigation is precisely what the person pursuing such litigation wishes to draw the other party into , but is precisely what the latter party aimed and bargained to avoid.” 177. In granting an anti -suit injunction the English courts are seeking to uphold and enforce the parties’ contractual bargain as set out in the arbitration agreement. In principle it should make no difference whether that agreement is governed by English law or by a foreign law. In both cases the e nquiry is whether there has been a breach of the arbitration agreement and whether it is just and convenient to restrain that breach by the grant of an anti -suit injunction. The detail of the e nquiry may differ, but its nature is the same. 178. Chubb Russia contends that as a matter of discretion the considerations to be taken into account are different where the arbitration agreement is governed by foreign law. It submits that issues of scope and breach of the arbitration agreement are generally best left to th e foreign court which has the requisite expertise in the applicable foreign law. 179. The judge’s view was that different considerations arise where the arbitration agreement is governed by foreign law by reason of the doctrine of forum conveniens. We agree wit h the Court of Appeal that forum conveniens, which is a matter that goes to the court’s jurisdiction, is not relevant. By agreeing to arbitrate in London the parties were agreeing to submit to the supervisory and supporting Page 63 jurisdiction of the English courts, including its jurisdiction to grant anti -suit injunctions. 180. Chubb Russia’s principal argument is that considerations of comity nevertheless make it appropriate to defer to the foreign court as a matter of discretion. Comity, however, has little if any role to play where anti -suit injunctive relief is sought on the grounds of breach of contract. As Millett LJ stated in Aggeliki Charis Cia Maritima SA v Pagnan SpA (The Angelic Grace) [1995] 1 Lloyd’s Rep 87, 96: “… in my judgment there is no good reason fo r diffidence in granting an injunction to restrain foreign proceedings on the clear and simple ground that the defendant has promised not to bring them. … The courts in countries … party to … the New York Convention, are accustomed to the concept that they may be under a duty to decline jurisdiction in a particular case because of the existence of an … arbitration clause. I cannot accept the proposition that any court would be offended by the grant of an injunction to restrain a party from invoking a jurisd iction which he had promised not to invoke and which it was its own duty to decline.” 181. Although The Angelic Grace concerned an arbitration agreement governed by English law, that was not material to the reasoning of the Court of Appeal. The rationale for the court’s approach was the fact of the promise made, not the law by which it was governed. That accords with principle. 182. Nor does article II(3) of the New York Convention make any difference. As noted earlier, under this article a court of a Convention state is required to refer the parties to arbitration when it is seized of a matter which the parties have agreed to arbitrate (unless the arbitration agreement is null and void, inoperative or incapable of being performed). The New York Conven tion is concerned with recognition and enforcement of arbitration agreements and awards , not jurisdiction - see, for example, Shashoua v Sharma [2009] EWHC 957 (Comm); [2009] 2 All ER (Comm) 477, paras 36 -38. If a court is seized of jurisdiction under its own law or rules, article II(3) obliges it to exercise that jurisdiction to enforce arbitration Page 64 agreements. It does not purport to nor does it confer any primacy over the jurisdiction of the courts of the seat. 183. The grant of an anti -suit injunction is alway s a matter of discretion. There may be circumstances in which it would be appropriate to await a decision of a foreign court. If, for example, the scope of the arbitration agreement was about to be determined by the highest court in the country of the gove rning law in unrelated proceedings, then it might be sensible for the English court to await that decision. Where, however, the issue arises in proceedings brought in alleged breach of the arbitration agreement, deference to the foreign court should genera lly give way to the importance of upholding the parties’ bargain and restraining a party to an arbitration agreement from doing something it has promised not to do. 184. We therefore agree with the Court of Appeal that the principles governing the grant of an a nti-suit injunction in support of an arbitration agreement with an English seat do not differ according to whether the arbitration agreement is governed by English law or foreign law. Forum conveniens considerations are irrelevant and comity has little if any role to play. The court’s concern will be to uphold the parties’ bargain, absent strong reason to the contrary, and the court’s readiness to do so is itself an important reason for choosing an English seat of arbitration. 185. It follows that if the agreeme nt to arbitrate disputes contained in article 50.1 of the construction contract had been governed by Russian law, it would have been necessary for the English court to determine whether under the law of Russia the agreement is valid and the claim which Chu bb Russia is seeking to pursue in Russia falls within its scope. If those questions were answered in the affirmative, it would in any event have been appropriate to grant an anti -suit injunction. XII Overall conclusion 186. Although our approach to the determin ation of the law applicable to the arbitration agreement differs from that taken by the Court of Appeal, we have similarly concluded that the arbitration agreement in this case is governed by English law. It is common ground that in these circumstances the arbitration agreement is valid, the dispute between the parties falls within it and that the injunction granted by the Court of Appeal to restrain Chubb Russia from proceeding against Enka in Russia was properly granted. It follows that we would dismiss t he appeal. Page 65 LORD BURROWS : (dissenting) (with whom Lord Sales agrees) 1. Introduction 187. In this case, we are presented with an intriguing question of law which courts and commentators have been grappling with for many years. What is the proper law (in the English common law conflicts of law) of an arbitration agreement where there is no expre ss choice of law clause in the arbitration agreement? In particular, should the proper law of the arbitration agreement be the law of the main contract in which the arbitration agreement is contained or should it be the law of the seat of arbitration? In s horthand, should one determine the proper law of the arbitration agreement by the “main contract” approach or the “seat” approach? In this case, the seat of the arbitration is England but the proper law of the main construction contract, in which the arbit ration agreement is contained, is Russian law (although there is a dispute as to the precise reason for that). Although the ultimate question for this court is whether to issue an anti -suit injunction to stop proceedings in Russia it is first helpful, and arguably essential, to determine the proper law of the arbitration agreement . That proper law issue is of wide public importance and this (dissenting) judgment is almost entirely devoted to it. 188. A bare outline of the facts will here be sufficient. The claim ant and respondent to this appeal (“Enka”) is a Turkish engineering company that had been engaged as a subcontractor in construction work at a power plant in Russia. The head -contractor (CJSC “Energoproekt”) assigned its rights against Enka to the owner an d developer (PJSC “Unipro”). There was an arbitration agreement (in article 50.1) in the construction contract (“the main contract”) between Enka and the head -contractor that disputes would be determined by way of International Chamber of Commerce (“ICC”) arbitration with London seat . Following a massive fire at the power plant, the Russian first defendant insurer and the appellant in this appeal, OOO “Insurance Company Chubb” (which I shall refer to throughout as “Chubb Russia”), paid an in surance claim ma de by the owner and was subrogated to any rights the owner had against Enka. Chubb Russia brought a claim against Enka (and others) in Russia. Enka con tended that those proceedings were in breach of the arbitration agreement and applied to the Russian court to dismiss Chubb Russia’s claim . It also brought a claim in England for an anti -suit injunction against the defendants, all members of the Chubb group of companies . 189. At first instance, Andrew Baker J declined to reach a decision on the proper law of the a rbitration agreement but dismissed Enka’s claim for an anti -suit injunction on the ground of forum non conveniens: [2019] EWHC 3568 (Comm). Subsequently Enka’s claim in Russia to dismiss the Russian proceedings, as being in breach of the arbitration agreement , failed although Chubb Russia’s claim on the merits against Enka also failed. Both Enka and Chubb Russia are appealing that Page 66 decision to the Russian appeal court (and the appeal is set for late October 2020). Meanwhile the Court of Appeal here (Flaux, Males and Popplewell LJJ) ([2020] EWCA Civ 574) allowed Enka’s appeal against Andrew Baker J’s decision. It held that the proper law of the arbitration agreement was English and granted Enka an anti -suit injunct ion to stop any Russian appeal going ahead as being in breach of the arbitration agreement . 190. Chubb Russia applied to the Supreme Court for permission to appeal from the decision of the Court of Appeal. This court granted permission to appeal and also stayed the anti -suit injunction upon Chubb Russia giving suitable undertakings to protect Enka’s position pending the outcome of this expedited appeal. 191. It will be helpful to set out immediately t he arbitration agreement . This appears within article 50.1 of the m ain construction contract in the following terms: “ Resolution of disputes 50.1. The Parties undertake to make in good faith every reasonable effort to resolve any dispute or disagreement arising from or in connection with this Agreement (including disputes regarding validity of this agreement and the fact of its conclusion (hereinafter - ‘ Dispute ’) by means of negotiations between themselves. In the event of the failure to resolve any Dispute pursuant to this a rticle within 10 (ten) days from the date that either Party sends a Notification to the opposite Party containing an indication of the given Dispute (the given period may be extended by mutual consent of the Parties) any Party may, by giving written notice, cause the matter to be referred to a meeting between the senior managements of the Contractor and Customer (in the case of the Contractor senior management sh all be understood as a member of the executive board or above, in the case of Customer, senior management shall be understood as general direct ors of their respective companies). The parties may invite the End Customer to such Senior Management Meeting. Such meeting sh all be held within fourteen (14) calendar days following the giving of a notice. If the matter is not resolved within twenty (20) calendar days after the date of the notice referring the matter to appropriate higher management or such later date as may be unanimously agreed upon, the Dispute shall be referred to international arbitration as follows: Page 67 • the Dispute shall be finally settl ed under the Rules of Arbitration of the International Chamber of Commerce, • the Dispute shall be settled by three arbitrators appointed in accordance with these Rules, • the arbitration shall be conducted in the English language, and • the place of arbitration shall be London, England. 50.2. Unless otherwise explicitly stipulated in this Agreement , the existence of any Dispute shall not give the Contractor the right to suspend Work. 50.3. Not used. 50.4. Not used. 50.5. All other documentation such as financial documentation and cover documents for it must be presented in Russian. ” 192. This judgment builds up to answering the question as to the proper law of the arbitration agreement by initially clearing the ground in three sections. The first sets out som e clear or undisputed points of law, the second explains that the issue in this case concerns interpretation not invalidity, and the third clarifies why the proper law of the main contract is Russian. There is then an overview of the case law on the proper law of the arbitration agreement before I come to the central sections of the judgment on determining the proper law of the arbitration agreement in this case and generally. The analysis enables me to provide a statement of the common law on the proper la w of an arbitration agreement that is principled, straightforward , clear and easy to apply. Page 68 2. Clear or undisputed points of law 193. A number of important matters of law relevant to deciding the proper law of the arbitration agreement are not in dispute (or are clear) and are worth setting out immediately. They are: (i) The seat of the arbitration is England as set out in article 50.1. (ii) The pr oper (or “applicable”) law of the main construction contract, which is determined by applying the Rome I Regulation (EC) No 593/2008 (laying down the EU and therefore English conflict of law rules to determine the proper law for contractual obligations), i s Russian law. But there is a dispute as to how that conclusion is reached. The relevant provisions of the Rome I Regulation are as follows: “ Article 3 Freedom of choice 1. A contract shall be governed by the law chosen by the parties. The choice shall be made expressly or clearly demonstrated by the terms of the contract or the circumstances of the case. By their choice the parties can select the law applicable to the whole or to part only of the contract. … Article 4 Applicable law in the absence of choic e 1. To the extent that the law applicable to the contract has not been chosen in accordance with article 3 …, the law governing the contract shall be determined as follows: … (b) a contract for the provision of services shall be governed by the law of the country where the service provider has his habitual residence; … Page 69 2. Where the contract is not covered by paragraph 1 or where the elements of the contract would be covered by more than one of points (a) to (h) of paragraph 1, the contract shall be governe d by the law of the country where the party required to effect the characteristic performance of the contract has his habitual residence. 3. Where it is clear from all the circumstances of the case that the contract is manifestly more closely connected wit h a country other than that indicated in paragraphs 1 or 2, the law of that other country shall apply. 4. Where the law applicable cannot be determined pursuant to paragraphs 1 or 2, the contract shall be governed by the law of the country with which it is most closely connected.” David Bailey QC, for Chubb Russia, submitted that the proper law of the main contract is Russian because, applying a rticle 3(1) of Rome I, the choice of Russian law has been made expressly or clearly demonstrated. Robin Dicker QC, for Enka, denied that there has been an express or implied (ie clearly demonstrated) choice of Russian law. Mr Dicker accepted that Russian law is the proper law by reason of a rticle 4 of Rome I but he did not pinpoint why that was so (but because Enka, a s the service-provider, is Turkish this must presumably be because Russia is the country with which the contract is manifestly more closely connected than Turkey). (i ii) Although there is no bar to having different proper laws applying to different clauses of the same contract (the so -called concept of d épe çage) , the general position taken at common law (not least on grounds of practical convenience) is that a contract has a single proper law. See, for example, Kahler v Midland Bank [1950] AC 24 , 42 (per Lord MacDermott); Libyan Arab Foreign Bank v Bankers Trust Co [1989] QB 728 , 747 (per Staughton J); Dicey , Morris & Collins , T he Conflict of Laws , 15th ed (2012) , para 32 - 026. It is worth stressing that the arbitration agreement here is contained in the main contract. We are not concerned with a free -standing arbitration agreement (see para 230 below). (iv) The Rome I Regulation does not apply (directly) to an arbitration agreement because of an exclusion from the Regulation of “arbitration agreements and agreements on the choice of court” in article 1(2)(e) of the Regulation. The proper law of the arbitration agreement must therefore (in an English court) be determined by applying English common law conflict Page 70 of laws rules. They require a court to look for (applying English law) an express choice, an implied choice or, if neither of those applies, the system of law with which the arbitration agreement has its closest and most real connection: Bonython v Commonwealth of Australia [1951] AC 201, 219; Sulamérica Cia Nacional de Seguros SA v Enesa Enge nharia SA [2012] EWCA Civ 638 ; [2013] 1 WLR 102, paras 9 and 25. The first two of those stages are both concerned with ascertaining the parties’ objective intentions. One can regard the exercise as being one of interpretation of the main contract and the arbitration agreement. There is no express choice of law clause in the arbitration agreement in this case , ie there is no mention of choice of law in article 50.1 of the contract. (v) Mr Bailey at one stage in oral argument appeared to concede that, if the proper law of the arbitration agreement was not Russian by reason of an express or implied choice, it must be English because, as the seat of the arbitration was England, one could not decide that the arbitration agreement had its closest and most real connection to Russia. But he later withdrew that concession. I consider that he was correct to do so (I return to this in para 25 6 below). (vi) What is commonly referred to as the curial law is, according to Mustill and Boyd, Commercial Arbitration , 2 nd ed (1989) , pp 60 -62, 64 -68, the law dealing with “the manner in which the parties and the arbitrator are required to conduct the reference of a particular dispute” (p 60) and includes “the procedural powers and duties of the arbitrator” (p 62). The curial law is (almost) invariably the law of the seat of the arbitration. As the law of the seat is England, the curial law here is English. Inextricably linked to this is what may be referred to as the curial or supervisory jurisdiction of the courts. This is concerned with the courts’ jurisdiction to support and enforce the arbitration. It includes, for example, the power to remove or replace an arbitrator, to enforce or set aside an arbitral award, and to grant injunctions to support the arbitration including anti -suit injunct ions. Like the curial law, the curial or supervisory jurisdiction of the courts is (almost) invariably determined by the seat of the arbitration. Here, therefore, it is not in doubt that the English court s ha ve curial or supervisory jurisdiction in relation to the arbitration and this includes the jurisdiction to grant an anti- suit injunction in this case to restrain the Russian proceedings . In summary, as Popplewell LJ expressed it in the Court of Appeal at para 46, “The significance of the choice of a seat is … a legal one as to the curial law and the curial court.” (vii) If the proper law of the arbitration agreement is determined to be English, the anti -suit injunction ordered by the Court of Appeal is appropriate. This was conceded by Mr Bailey. The dispute as to whether an Page 71 anti -suit injunction should be ordered therefore arises only if the proper law of the arbitration agreement is determined to be Russian. 3. A preliminary important point: the dispute concerns the interpretation (or scope) of the arbitration agreement not its validity 194. The reason why the parties respectively favour Russian or English law as the proper law of the arbitration agreement is because English law may take a wider interpretation of the arbitration agreement in this case than Russian law. The precise basis for this is not entirely clear. The most obvious basis is that English law regard s tort as well as contractual claims between the part ies to be included within the scope of the disputes covered by the arbitration agreement (see Aggeliki Charis Cia Maritima SA v Pagnan SpA ( The Angelic Grace ) [1995] 1 Lloyd’s Rep 87), whereas Russian law may interpret disputes as applying only to contract ual disputes between the parties. However, it may be that the true basis is slightly more complex than that and involves Russian law tending to interpret the arbitration agreement as not covering joint tortious liability whereas English law appears to include that. Whatever the precise basis for respectively favouring Russian or English law, t he important point is that the issue between the parties is as to the scope or interpretation of the arbitration agreement. It is not about the validity of the arbitration agreement . 195. Andrew Baker J recognised this in his judgment at paras 11 -12 (and also at para 88). He said: “11 . … [I]t is common ground that there exists between Enka and Chubb Russia a valid and binding arbitration agreement. That is so even though Chubb Russia is suing in Moscow, and is therefore sued here, as subrogated insurer of Enka ’s original contractual counterparty. Whether Russian law or English law governs that question, it is common ground that such an insurer is bound by its insured ’s applicable arbitration agreement. The dispute between the parties, then, again as it was in The Angelic Grace, is whether the claim being pursued in the target proceedings is a claim in tort that falls outside the scope of the agreement to arbitra te. 12. The detail is more complex than it was in The Angelic Grace, however, because in that case there was no dispute but that the claim as brought in Italy was a claim in tort, and it was common ground that the question whether it fell within the scope of the arbitration agreement was governed by English Page 72 law. Here … the law applicable to the question of the scope of the arbitration agreement is disputed; and it is also contentious between the parties whether the claim as brought under Russian law in the Moscow Claim is a claim in tort, or, more strictly, whether it is viable as such. Furthermore, it is effectively common ground that if the question of the scope of the arbitration agreement is governed by English law, then that claim, however it is to be c haracterised under Russian law, is within that scope. The defendants ’ argument that the claim, if rightly characterised as a claim in tort, falls outwith the scope of the arbitration agreement, only arises at all if they are right that scope is a matter of Russian law.” 196. That interpretation or scope, not validity, is in issue is b orne out by the decision of the Russian court on 6 May 2020 which decided a preliminary question as to whether, applying Russian law, the court proceedings should go ahead despite t he arbitration clause. The Russian court made clear that the issue was as to the interpretation or scope of the arbitration agreement and not the validity of the arbitration agreement . The Court’s short judgment on this preliminary question was as follows: “So it is article 965 of the R ussian Federation Civil Code that establishes the right of the claimant to file against the persons liable for the losses, regardless of what served as the grounds for their occurrence. Therefore, the arbitration clause to wh ich Enka refers does not encompass this dispute and does not extend to it , as the participants are not Enka alone, but also the other ten co -defendants who did not enter into an arbitration clause, and the subject of the dispute is the general obligation o f all 11 co -defendants to indemnify the losses caused. On the basis of the above, the arbitral clause set out in point 50.1 of the contract is not applicable and because of this the motion declared by defendant 11 that the claim should be left on file shou ld not be granted.” ( E mphasis added) 197. However, Mr Dicker has now submitted that there is also an issue about the validity of the arbitration agreement under Russian law that does not arise under English law. He referred to a Russian decision on 8 February 2018 (in an unrelated matter) on enforcement of an arbitral award under this type of arbitration agreement. The decision was that the arbitration agre ement was too uncertain to be enforceable under Russian law apparently because of uncertainty about whether there should have been a reference in the arbitration agreement to the Int ernational Court of Arbitration . It was submitted by Mr Bailey in Chubb Ru ssia’s written case (at para 22) that “there is no question of the arbitration agreement being invalid under Page 73 Russian law”; and, as we have seen in the last paragraph, such an argument about invalidity played no part in the reasoning of the Russian court in the 6 May 2020 decision . In any event, our attention was drawn to a note on the website of Debevoise & Plimpton LLP, dated 7 January 2019, indicat ing that the February 2018 decision in Russia is inconsistent with the usual approach of the Russian S upreme Court and is not a binding authority. Although Mr Dicker submitted that, in the light of th at case, there is “ a serious risk ” (to use the language in Sulamérica Cia Nacional de Seguros SA v Enesa Enge nharia SA [2012] EWCA Civ 638 ; [2013] 1 WLR 102 , para 31 : see below para 217 ) that the arbitration agreement would be struck down as invalid under Russian law, that is not a submission that I can accept without having been provided with proper evidence as to the Russian law on the point. One can accept that there may be a triable issue as to whether there is a serious risk of invalidity in this case by reason of that 2018 case. However, we must decide the issue before us as to the proper law of the arbitration agreement on the evidence presented and on the matters pleaded (which do not include this invalidity point). In any event, the arbitration agreement in ques tion in this case was entered into in 2012 and it would seem that, for the purpose of determining the proper law of the arbitration agreement , we must assess the parties’ intentions and all other relevant factors as at that point in time unaffected by subs equent legal developments in 2018. 198. Why is it an important point that the dispute concerns the interpretation or scope of the arbitration agreement not its validity? There are two linked reasons. First, it is a general principle within the English conflict of laws that , as between two possible proper laws, the courts should favour the proper law that would uphold the validity of an agreement rather than one that would invalidate it (see, for example, In re Missouri Steamship Co (1889) 42 Ch D 321, 341; South African Breweries Ltd v King [1899] 2 Ch 173, 18 1; Coast Lines Ltd v Hudig and Veder Chartering NV [1972] 2 QB 34, 44 (per Lord Denning MR), 48 (per Megaw LJ); Chitty on Contracts , 33rd ed (2018) , para 30-12). Mr Bailey referred to this (in reliance on th e work of Gary Born, International Commercial Arbitration , 2nd ed ( 2014) , pp 542 -549, and Robert Merkin and Louis Flannery, The Arbitration Act 1996 , 6 th ed (2019) , para 46.10.5) as “the validation principle”. It rests on the rational assumption that p arties would prefer to have an agreement upheld than not. But if it is correct that there is no dispute about the validity of the arbitration agreement in this case, the validation principle is not a reason here for favouring English law over Russian law as th e proper law of the arbitration agreement . 199. Secondly, Mr Dicker submitted that, even if the dispute goes to the interpretation of the arbitration agreement and not its validity, the rational assumption is that parties would prefer to have all their disputes referred to arbitration rather than just some ie that “rational businessmen are likely to have intended” (using Lord Hoffmann’s words in Fiona Trust & Holding Corpn v Privalov [2007] UKHL 40; [2007 ] Bus LR 1719, para 13) that a wider rather than a narrower interpretation of disputes which should be arbitrated was intended. Page 74 However, there is a n important difference between, on the one hand, upholding as valid an undisputed agreement which the parties have reached and, on the other hand, determining the correct interpretation or scope of the agreement where the very question at issue is what is it that the parties have agreed. Without empirical evidence about what rational businessmen, one Russian and one Turkish, concludin g a contract for work to be carried out in Russia, would be likely to have intended, I am reluctant to place weight on the idea that these parties would have intended a wider rather than a narrower interpretation of their arbitration agreement . The rationa l assumption is that the parties intended their agreement to be interpreted in such a way that matches what they agreed . Rationally they do not want to be held to have agreed something wh ich is outside their agreement. And one cannot say that, just because English law may adopt a wider rather than a narrower approach to interpretation of an arbitration agreement than Russian law, that will ensure the correct interpretation of the arbitration agreement . I therefore agree with Mr Bailey’s written submission o n this point where he said: “[T]here is no suggestion of invalidity in this case, so as to engage the ‘validation principle’. The argument is simply that English law should be taken to apply because it construes AAs [ie arbitration agreements] more liberally. That point only has to be articulated to reveal its parochialism. It is impossible to say that just because Russian law takes a narrower view of AAs than English law does … that the parties must have intended English law to apply. That is results -based reasoning that ignores the fact that there are legitimate reasons for adopting a narrower approach (such as, in this very case, that a broad interpretation of AAs can lead to an undesirable fragmentation of disputes and proceedings where many different pa rties are involved).” 4. Why is the proper law of the main contract Russian? 200. As I have explained in para 193(ii), while it is not in dispute that the proper law of the main construction contract is Russian, the route to that conclusion through the Rome I R egulation is disputed. This matter is of central importance because it has a significant impact on determining the proper law of the arbitration agreement . 201. As we have seen in para 193(ii), the Rome I Regulation provides in a rticle 3.1 that the governing la w is that chosen by the parties where a choice is “ made expressly or is clearly demonstrated by the terms of the contract or the circumstances of the case ”. In the absence of such choice a rticle 4 provides that in a contract for the provision of services t he governing law is prima facie that of the habitual residence of the service provider but that the law of another country applies “ where Page 75 it is clear from all the circumstances of the case that the contract is manifestly more closely connected with [ that] country ”. 202. Mr Bailey submitted that Russian law had been expressly chosen as the proper law. He relied on the definition of “Applicable Law ” in Attachment 17 to the contract which reads : “ Law of the Russian Federation, including legislation of the Russian Federation, all regulatory legal acts of the State Authority Federal Bodies, State Authorities of the constituent entities of the Russian Federation, legislation of the constituent entitie s of the Russian Federation, regulatory legal acts by Local Authorities and any other applicable regulatory legal acts.” Although this was not a classic choice of law clause of the type “This Agreement is governed by Russian law” Mr Bailey submitted that i t had the same effect. I am not persuaded by that. The applicable law article (A ttachment 17) does not say “This Agreement is governed by the Applicable law ”. Rather article 1 of the contract provides that “The terms used in this Agreement shall have the d efinitions set forth in Attachment No 17 to this Agreement ”. Admittedly, the term “Applicable Law ” is used in a l arge number of specific provisions . But Mr Dicker submitted that one is here talking about an incorporation by reference of relevant legislativ e provisions and that that is how the phrase “Applicable law” is used in international construction contracts (and he here referred us to a major practitioner work on standard contracts issued by the International Federation of Consulting Engineers (“FIDIC ”): Baker Mellors Chalmers and Lavers on FIDIC Contracts, Law and Practice at paras 2.126, 2.140, 2.145 ). Mr Dicker took as a t ypical article in the main contract, article 4.1(b) which provides that Enka shall ensure performance of the work in accordance w ith the Applicable Law. This ensures that, incorporated into the contract, are local laws and regulations, such as those governing planning, health and safety, labour laws, taxes and customs . Admittedly the main contract was not a FIDIC contract. And it ma y be thought odd to incorporate, where specified, all the relevant law of the Russian Federation (as the first phrase of Attachment 17 requires) including presumably the Russian law of contract in the Russian Civil Code, if all one is concerned with are pa rticular mandatory regulations. My view is that, although there is some ambiguity about the role of the “Applicable Law” definition, Mr Dicker is correct that Attachment 17 does not constitute an express choice of law clause. 203. However, Attachment 17 is not alone. There are many other additional references to Russian law in the contract. So, for example, at article 24.2 there is reference to the provisions of the Russian Civil Code, there is reference to “RF law” in article 4.15, and there are numerous refere nces (eg at articles 4.5, 4.26, 19.2 and Page 76 36.1) to “law” which, in the context, are clearly references to Russian law. It is helpful here to refer to Title II, article 3, para 3 of the Giuliano -Lagarde Report on the Convention on the law applicable to contr actual obligations which was the report that lay behind the Rome Convention which was the predecessor of the Rome I Regulation (and had the same wording as article 3.1 except that the formulation was “ The choice must be expressed or demonstrated with reaso nable certainty” rather than “ The choice shall be made expressly or clearly demonstrated ”): “ The choice of law by the parties will often be express but the Convention recognizes the possibility that the court may, in the light of all the facts, find that t he parties have made a real choice of law although this is not expressly state d in the contract. For example ... references in a contract to specific a rticles of the French Civil Code may leave the court in no doubt that the parties have deliberately chose n French law, although there is no expressly stated choice of law.” (OJ C282/17) 204. One can add to those express words in the contract, several other circumstances. The head -contractor in the contract with Enka was Energoproekt, a Russian company and the owner and end customer, Unipro, was also Russian. The place of performance was Russia n. The effects of any breach would be suffered in Russia. The primary language of the contract was Russian. And the price for the work was to be paid in Russian currency to a Russian bank account. Indeed, the only non -Russian elements of the contract are that Enka is a Turkish company and that the seat of the arbitration is England. 205. My conclusion, therefore, is that, applying article 3.1 of the Rome I Regulation, Russian law is the proper law of the main contract chosen by the parties because, even though n ot expressly chosen, that choice has been “clearly demonstrated by the terms of the contract or the circumstances of the case”. 206. The most powerful argument to the contrary is that the parties could easily have inserted a choice of law clause into the contra ct and yet failed to do so. Mr Dicker submitted that, in the context of a professionally drafted, detailed, and long contract, the most obvious explanation for that was that the parties could not agree on which law should be the governing law. But we have seen no evidence as to the circumstances in which this contract was drawn up and it seems to me more plausible as an objective interpretation of the parties’ intentions that, given that there was some ambiguity over the role of the “Applicable Law” definit ion, the parties thought it was clear, and did not need to be further stated, that Russian law was the proper law. Page 77 207. Although there may be marginal differences as between article 3.1 of the Rome I Regulation and the first two stages (express or implied choic e) of the common law test for the proper law, they are very closely aligned: see Lawlor v Sandvik Mining and Construction Mobile Crushers and Screens Ltd [2013] EWCA Civ 365; [2013] 2 Lloyd’s Rep 98. In my view, English common law, which I here refer to by analogy , would in this case regard there as having been an implied choice of Russian law. Even though there was no express term to that effect, the correct objective interpretation of the contract is that Russian law has been chosen by the parties. 208. I should stress that the lower courts did not decide this question as to why Russian law was the proper law of the main contract. Andrew Baker J, at paras 91 - 93, simply said that whether there was a choice of Russian law as the proper law is “far from clear in E nka’s favour” (ie it was not clear that no choice had been made). The Court of Appeal decided that there was no express choice of proper law but appeared to leave open whether there had nevertheless been a clearly demonstrated choice under art icle 3(1) of the Rome I Regulation. 5. The case law on the proper law of the arbitration agreement 209. In the Court of Appeal in this case, Popplewell LJ said, at para 89, “In my view the time has come to seek to impose some order and clarity on this area of the law, in particular as to the relative significance to be attached to the main contract law on the one hand, and the curial law of the arbitration agreement on the ot her, in seeking to determine the AA law. The current state of the authorities does no credit to English commercial law which seeks to serve the business community by providing certainty.” As this passage suggests, the English cases on this question, which appear to have been proliferating in recent years, do not speak with one voice. Certainly in seeking to provide the clarity which Popplewell LJ was rightly seeking, one cannot simply examine the relevant cases and hope to find in them a definitive answer to our question. With reasoning and decisions going both ways, the major purpose of looking at past cases is rather to put the task facing us in context and to ensure that all relevant considerations have been borne in mind. But ultimately, and without any authority binding this court, the way forward rests on a re -examination of principle. It also follows that no attempt is here being made to cover all relevant cases. Rather I shall focus on the most important cases to which we were referred by counsel . Page 78 210. The earliest case we were referred to was the House of Lords decision in Hamlyn & Co v Talisker Distillery [1894] AC 202. This concerned a contract between an English and Scots firm, made in London but to be performed in Scotland, with an arbitration clause for arbitration by “two members of the London Corn Exchange, or their umpire, in the usual way”. It was held that the interpretation of the arbitration clause was governed by English law (ie in modern terminology, the proper law of the arbitration agreement was English). But in determining the respective weights of the proper law of the main contract and the proper law of the seat of the arbitration, the case does not take one very far for tw o reasons. First, the proper law of the main contract was not clarified and indeed it seemed to be assumed that the proper law of the arbitration agreement would also be the proper law of the main contract. In the words of Lord Herschell LC, at p 209: “I s ee no difficulty whatever in construing the language used as an indication that the contract, or that term of it [ie the arbitration agreement ], was to be governed and regulated by the law of England.” Secondly, it was regarded as an important consideratio n that the arb itration clause was invalid in Scotland - because the arbitrators were not named - but valid in England. It was for this reason that Mr Bailey submitted that this case was an example of the application of the “validation principle”. 211. In Cie Tu nisienne de Navigation SA v Cie d ’Armement Maritime SA [1971] AC 572, the House of Lords was deciding on the proper law of the main contract (a contract for the carriage of goods by sea) in a context where that proper law was specified as being governed “b y the laws of the flag of the vessel carrying the goods”. There was an arbitration clause with London as the seat. It was held that the proper law of the main contract was French. The majority (Lords Morris, Dilhorne and Diplock) reasoned that this was bec ause there was a choice of French law as the proper law (because, on the true construction of the choice of law clause, the relevant flag was French). Lords Reid and Wilberforce reasoned that, although there was no operative choice of law clause (because t he dispute could not be related to a specific vessel or shipment), the rest of the contract and the relevant surrounding facts meant that the contract had the closest connection with France (the majority preferred to treat this as an alternative reason for their decision). Their Lordships placed considerable weight on the seat of the arbitration as a strong indication of the proper law of the main contract (and implicitly the proper law of the arbitration agreement ) b ut held that that strong indication was here negatived by the choice of law clause (per the majority) or by the other factors linking the contract most closely to French law (per Lords Reid and Wilberforce). Lord Diplock’s analysis of the “curial law” is particularly helpful. He said the followi ng at p 604: “ My Lords, it is possible for parties to a contract to choose one system of law as the proper law of their contract and a different system of law as the curial law. Although they may want their mutual rights and obligations under the contract to be ascertained by reference to the system of law of a country with Page 79 which the transaction has some close and real connection, they may nonetheless consider that the arbitral procedure adopted in some other country, or the high reputation and commercial e xpertise of arbitrators available there, make the curial law of that country preferable to the curial law of the country whose system of law they have chosen as the proper law. It is not now open to question that if parties to a commercial contract have ag reed expressly upon the system of law of one country as the proper law of their contract and have selected a different curial law by providing expressly that disputes under the contract shall be submitted to arbitration in another country, the arbitrators must apply as the proper law of the contract that system of law upon which the parties have expressly agreed. But the cases which have given rise to difficulty are those where the parties have made a choice of curial law by a clause of their contract expre ssly agreeing to arbitration in a particular country but have made no express provision as to the proper law applicable to the contract .” 212. We were then referred to two judgments of Lord Mustill, who was the co - author, with Stewart Boyd QC, of Commercial Arbitration (the first edition of which appeared in 1982 with a second edition in 1989). In Black Clawson International Ltd v Papierwerke Waldhof -Aschaffenburg AG [1981] 2 Lloyds Rep 446 Mustill J as he then was said, at p 455: “Where the laws diverge at all, one will find in most instances that the law governing the continuous agreement [sc the arbitration agreement] is the same as the substantive law of the contract in which it is embodied …” And at p 456: “In the ordinary way, this [sc the proper law of the arbitration agreement] would be likely to follow the law of the substantive contract.” These statements offer support to the proper law of the arbitration agreement being the same law as the main contract rather than being the law of the seat. This is Page 80 consistent with the approach favoured in Mustill and Boyd , Commercial Arbitration , 2 nd ed (1989) , at p 63: “The starting point is to determine the proper law of the contract in which the arbitration is embodied. As a general rule the arbitration agreement will be governed by the same law, since it is part of the substance of the underlying contract .” However, in the Black Clawson case itself, the force of Mustill J’s support for the “main contract” approach is somewhat diminished because he went on to trea t the parties ’ choice of Zurich as the place of arbitration as indicating an intention that the law governing the arbitration agreement should be the law of Zurich. 213. Subsequently, we see Lord Mustill favouring the “main contract” approach in the House of Lo rds in Channel Tunnel Group Ltd v Balfour Beatty Construction Ltd [1993] AC 334 . At p 357, Lord Mustill said : “It is by now firmly established that more than one national system of law may bear upon an international arbitration. Thus, there is the proper l aw which regulates the substantive rights and duties of the parties to the contract from which the dispute has arisen. Exceptionally, this may differ from the national law governing the interpretation of the agreement to submit the dispute to arbitration. Less exceptionally it may also differ from the national law which the parties have expressly or by implication selected to govern the relationship between themselves and the arbitrator in the conduct of the arbitration: the ‘curial law ’ of the arbitration, as it is often called.” ( E mphasis added) 214. In XL Insurance Ltd v Owens Corning [2001] 1 All ER (Comm) 530 Toulson J was concerned with an insurance policy which (to simplify slightly) had a New York governing law clause and an arbitration clause with a Lond on seat which included reference to the Arbitration Act 1996. It was alleged that the arbitration agreement was unenforceable because it was not in the correct written form under New York law. It was held, inter alia, that the enforceability of the arbitration agreement should be governed by English law as the law of the seat. Although Toulson J’s reasoning is open to various possible interpretations - and certainly his reasoning lends support to hiving off arbitration from the rest of the main contract as dealing with “a particular method of resolving disputes” (at 54 1e) - one interpretation is that, as he was satisfied that the parties had made an arbitration Page 81 agreement , the validation principle was being applied so as to ensure that that arbitration agreement was upheld. 215. The primary importance of C v D [2007] EWCA Civ 1282 ; [2008] Bus LR 843 is obiter dicta of Longmore LJ supporting the “seat” approach. The case dealt with an insurance contract governed by New York law with an English arbitration cla use (ie an English seat). The question was which law, New York or English, governed challenges to the arbitral award. It was held that English law applied to determine that question. That seems straightforward because that question was one of curial law an d curial jurisdiction and the seat of arbitration (here England) almost invariably determines that law. The proper law of the arbitration agreement and the proper law of main contract were irrelevant in this case. However, Longmore LJ went on , in obiter di cta, to look at the proper law of the arbitration agreement and said this, at para 2 2: “The question then arises whether, if there is no express law of the arbitration agreement, the law with which that agreement has its closest and most real connection is the law of the underlying contract or the law of the seat of arbitration. It seems to me that if (contrary to what I have said above) this is a relevant question, the answer is more likely to be the law of the seat of arbitration than the law of the underlying contract.” 216. It is worth interjecting here that, in line with Longmore LJ’s obiter dicta, the 15th edition of Dicey , Morris & Collins , The Conflict of Laws , published in 2012 has the following main rule (r ule 64(1)): “ The material validity, scope and i nterpretation of an arbitration agreement are governed by its applicable law, namely: (a) the law expressly or impliedly chosen by the parties; or, (b) in the absence of such choice, the law which is most closely connected with the arbitration agreement, which will in general be the law of the seat of the arbitration.” 217. We then come to what can probably be regarded as the leading case: Sulamérica Cia Nacional de Seguros SA v Enesa Enge nharia SA [2012] EWCA Civ Page 82 638 ; [2013] 1 WLR 102. Moore -Bick LJ’s leading judgment (with which Hallett LJ and Lord Neuberger MR agreed) was cited by both Mr Bailey and Mr Dicker in support of their submissions. Claims were brought by Brazilian companies under two i nsurance policies covering construction work in Brazil. The insurers denied liability on the basis of an exclusion clause and material non -disclosure. There was an express choice of Brazilian law as the governing law in the insurance contracts and an exclu sive jurisdiction clause in favour of Brazilian courts. However, the arbitration clause specified England as the seat. In the insurers’ application for an anti -suit injunction , the central question was what was the proper law of the arbitration agreement . Under Brazilian law, there was a “serious risk ” (per Moore - Bick LJ at para 31) that the insured was not bound by the arbitration clause as the insured may not have specifically consented to it s enforcement. The Court of Appeal held that English law was the proper law of the arbitration agreement . But it is not easy to determine whether Moore -Bick LJ’s judgment supports the “main contract” or “seat” approach. 218. The following passage, at para 26 , supports the “main contract” approach provided there is an express choice of law clause in the main contract: “… where the arbitration agreement forms part of a substantive contract an express choice of proper law to govern that contract is an important factor to be tak en into account. … In the absence of any indication to the contrary, an express choice of law governing the substantive contract is a strong indication of the parties ’ intention in relation to the agreement to arbitrate. A search for an implied choice of p roper law to govern the arbitration agreement is therefore likely … to lead to the conclusion that the parties intended the arbitration agreement to be governed by the same system of law as the substantive contract, unless there are other factors present w hich point to a different conclusion. These may include the terms of the arbitration agreement itself or the consequences for its effectiveness of choosing the proper law of the substantive contract.” 219. Moore- Bick LJ went on to decide that there were two “co nflicting indications” (para 31) that meant that the parties had not impliedly chosen Brazilian law as the proper law of the arbitration agreement . The first was that England was the seat, which inevitably imported English law, and hence the provisions of the Arbitration Act 1996, relating to the conduct and supervision of the arbitration (ie the curial law was English and the English courts had supervisory jurisdiction). The second was the “ serious risk” that the arbitration agreement might not be binding, as against the insured, under Brazilian law. He then turned to the third stage of the Page 83 common law approach and, in a passage which supports the “seat” approach he said this at para 32 : “One then has to consider with what system of law the agreement has the closest and most real connection. Although [counsel for the appellant] submitted that the agreement has a close and real connection with the law of Brazil, being the law governing the substantive contract in which the arbitration agreement itself is embed ded, I think his argument fails adequately to distinguish between the substantive contract and the system of law by which it is governed. No doubt the arbitration agreement has a close and real connection with the contract of which it forms part, but its n ature and purpose are very different. In my view an agreement to resolve disputes by arbitration in London, and therefore in accordance with English arbitral law, does not have a close juridical connection with the system of law governing the policy of ins urance, whose purpose is unrelated to that of dispute resolution; rather, it has its closest and most real connection with the law of the place where the arbitration is to be held and which will exercise the supporting and supervisory jurisdiction necessary to ensure that the procedure is effective. Its closest and most real connection is with English law. I therefore agree with the judge that the arbitration agreement is governed by English law.” 220. Subsequent to Sulam érica , there have been two significant fi rst instance decisions . In Arsanovia Ltd v Cruz City 1 Mauritius Holdings [2012] EWHC 3702 (Comm) ; [2013] 2 All ER (Comm) 1 , Andrew Smith J was faced with an express choice of Indian law in the main contract and an arbitration agreement with a London seat. Distinguishing Sulamérica , because there were no indications conflicting with the express choice of law , he held that the proper law of the arbitration agreement was Indian law. 221. Then we come to the valiant attempt by Hamblen J (as he then was) in Habas Si nai Ve Tibbi Gazlar Istihsal Endustrisi AS v VSC Steel Co Ltd [2013] EWHC 4071 (Comm) ; [2014] 1 Lloyd’s Rep 479, to set out, as clearly as possible, the relevant principles to be derived from the cases in this tangled area. In relation to the question of the proper law of the arbitration agreement it was assumed that there was no choice of law in the main contract but that it was governed by Turkish law as the law with which it was most closely connected. The parties had agre ed (as found by Hamblen J) a London arbitration clause. It was held that the proper law of the arbitration agreement was English. At para 101, Hamblen J said: Page 84 “101 . The leading authority is the recent Court of Appeal decision in Sul América Cia Nacional d e Seguros SA v Enesa Engenharia SA [2012] 1 Lloyd ’s Rep 671 . Moore-Bick LJ (with whom Hallett LJ and Lord Neuberger MR agreed), summarised the test for determining the law applicable to arbitration agreements at paras 26-32. The Court of Appeal ’s decision w as considered but distinguished by Andrew Smith J in Arsanovia Ltd v Cruz City 1 Mauritius Holdings [2013] 1 Lloyd’s Rep 235. The guidance provided by these authorities may be summarised as follows: (1) Even if an arbitration agreement forms part of a matrix contract (as is commonly the case), its proper law may not be the same as that of the matrix contract. (2) The proper law is to be determined by undertaking a three -stage enquiry into : (i) express choice ; (ii) implied choice ; and (iii) the system of law with which the arbitration agreement has the closest and most real connection. (3) Where the matrix contract does not contain an express governing law clause, the significance of the choice of seat of the arbitration is likely to be ‘ overwhelming ’. That i s because the system of law of the country seat will usually be that with which the arbitration agreement has its closest and most real connection. (4) Where the matrix contract contains an express choice of law, this is a strong indication or pointer in r elation to the parties ’ intention as to the governing law of the agreement to arbitrate, in the absence of any indication to the contrary. (5) The choice of a different country for the seat of the arbitration is a factor pointing the other way. However, it may not in itself be sufficient to displace the indication of choice implicit in the express choice of law to govern the matrix contract. Page 85 (6) Where there are sufficient factors pointing the other way to negate the implied choice derived from the express c hoice of law in the matrix contract the arbitration agreement will be governed by the law with which it has the closest and most real connection. That is likely to be the law of the country of seat, being the place where the arbitration is to be held and w hich will exercise the supporting and supervisory jurisdiction necessary to ensure that the procedure is effective. 102. In relation to point (3), I would add that the terms of the arbitration clause may themselves connote an implied choice of law. It is recognised that they may operate as an implied choice of law for the matrix contract itself - see, for example, Cie Tunisienne de Navigation SA v Cie d’Armement Maritime SA [1971] AC 572 , Lord Wilberforce at p 596 and Lord Diplock at pp 604-605; … In such cases they must surely equally operate as an implied choice of law for the arbitration agreement. 103. The present case is one where there is no express choice of law in the matrix contract. In such a case the Sul Am érica decision is clear authority t hat the applicable law will be that of the country of seat. This was acknowledged by Habas who reserved the right to challenge the decision should this case go further.” 222. The reference to “overwhelming” in point (3) appears to refer to the words of Moore- Bick LJ in the Sulam érica case, at para 26, but it should be noted that Moore- Bick LJ was using that description in the context of a free -standing agreement to arbitrate not an arbitration agreement contained in a main contract. 223. Ham blen J’s summary represent s clear support for the “seat” approach: unless there is an express choice of law clause in the main contract, the seat will very likely determine the proper law of the arbitration agreement ; and even where there is such an express choice of law clause, th ere may be sufficient factors pointing towards the “seat” determining the arbitration agreement ’s proper law. 224. I interject at this point that there was a careful analysis of these issues by Steven Chong J (as he then was) in BCY v BCZ [2016] SGHC 249; [2016 ] 2 Lloyd’ s Rep 583 in the High Court of Singapore . In a judgment which favoured the “main contract” approach, he said at para 65: Page 86 “ w here the arbitration agreement is part of the main contract, I would hold, adopting Sul América , that the governing law of the main contract is a strong indicator of the governing law of the arbitration agreement unless there are indications to the contrary. The choice of a seat different from the law of the governing contract would not in itself be sufficient to displace that starting point. ” 225. The approach in the BCY case was subsequently assumed to be the correct law in Singapore by the Singaporean Court of Appeal (Sundaresh Menon CJ, Judith Prakash JA, and Steven Chong JA), and by the parties, in BNA v BN B [2019] SGCA 84; [20 20] 1 Lloyd’s Rep 55 , paras 44 -95. 226. Popplewell LJ’s approach in the Court of Appeal in the present case may be regarded as somewhat similar to that of Hamblen J’s in the Habas case. At para 91, Popplewell LJ said that, subject to an express choice of law in the main contract, “the general rule should be that the arbitration agreement law is the curial law, as a matter of implied choice, subject only to any particular features of the case demonstrating powerful reasons to the contrary”. And at para 105, he sa id the following: “I would therefore summarise the principles applicable to determining the proper law of an arbitration agreement, what I have called the AA law, when found in an agreement governed by a different system of law, as follows: (1) The AA law is to be determined by applying the three stage test required by English common law conflict of laws rules, namely (i) is there an express choice of law? (ii) if not, is there an implied choice of law? (iii) if not, with what system of law does the arbitration agreement have its closest and most real connection? (2) Where there is an express choice of law in the main contract it may amount to an express choice of the AA law. Whether it does so will be a matter of construction of the whole contrac t, including the arbitration agreement … Page 87 (3) In all other cases there is a strong presumption that the parties have impliedly chosen the curial law as the AA law. This is the general rule , but may yield to another system of law governing the arbitration ag reement where there are powerful countervailing factors in the relationship between the parties or the circumstances of the case.” 6. What is the proper law of the arbitration agreement ? (1) The proper law of the arbitration agreement is Russian law by rea son of an implied choice 227. We are now in a position to decide what is the proper law of the arbitration agreement . As I have said at para 193(iv) above, this is to be resolved by the common law choice of law rules ie one is looking for an express choice, an implied choice or, if neither of those applies, the system of law with which the arbitration agreement has its closest and most real connection. In this case, the three most important factors in deciding this issue are: (i) There is no express choice of la w clause in the arbitration agreement here ie there is no mention of choice of law in article 50.1 of the main construction contract. (ii) The seat of the arbitration, as laid down in the arbitration agreement, is England. (iii) The proper law of the main construction contract, as we have established at paras 200-208 above, is Russian law by reason of the implied choice of the parties. 228. It is my view that that combination of factors leads to the conclusion that, under English commo n law, the proper law of the arbitration agreement is, by reason of an implied choice, Russian law. A s the parties have impliedly chosen Russian law for the main contract it is natural , rational and realistic to regard that choice for the main contract as encompassing, or carrying across to, the arbitration agreement . That implied choice is simply the correct objective interpretation of the parties’ main contract and arbitration agreement . Page 88 229. Although the decision as to the proper law of the arbitration agreem ent turns on the interpretation of the main contract and the arbitration agreement, there are a number of general reasons (ie reasons that do not turn on the interpretation of these particular contracts) which support the view that, absent an express choic e of law in the arbitration agreement , there is a presumption (or general rule) that the proper law of the main contract is also the proper law of the arbitration agreement; and there is no such presumption (or general rule) that the law of the seat is the proper law of the arbitration agreement . In short, these are reasons for favouring the “main contract” rather than the “seat” approach. 230. I should make clear at the outset that, everything that is here said, relates to an arbitration agreement that is conta ined in a main contract. While a free -standing arbitration agreement entered into at the same time would not be treated differently, a free -standing arbitration agreement entered into at a different time and under different circumstances would require a di fferent analysis. (2) Reasons why, absent an express choice of law in the arbitration agreement, there is a presumption (or general rule) that the proper law of the main contract is also the proper law of the arbitration agreement (i) Dépe çage is the excep tion not the rule 231. If one were to treat the arbitration agreement in the same way as all the other clauses in the main contract, the general rule would be that the same proper law would apply throughout . D épe çage is the exception not the rule. See para 193(iii) above . (ii) The rationale of the separability doctrine 232. Under the separability doctrine, an arbitration agreement is viewed for certain purposes, both at common law and under s ection 7 of the Arbitration Act 1996, as a separate contract from the main contract. The reason for that is in order to ensure that the arbitration agreement is effective despite the non -existence, invalidity, termination or rescission of the main contract. In other words, it stops the argument that the parties have not agreed to arbitration to deal with disputes about the non - existence, invalidity or initial ineffectiveness of the main contract; and it also stops the argument that the arbitration agreement cannot deal with disputes once the main contract has been terminated or rescinded. This explains the wording of s ection 7 of the Arbitration Act 1996: Page 89 “ Separability of arbitration agreement Unless otherwise agreed by the parties, an arbitration agreement which forms or was intended to form part of another agreement (whether or not in writing) shall not be regarded as invalid, non -existent or ineffective because that other agreement is invalid, or did not come into existence or has become ineffective, and it shall for that purpose be treated as a distinct agreement.” ( E mphasis added) 233. This statutory wording makes clear that the separability doctrine has been devised for a particular purpose. For that purpose, it treats (one might say somewhat fictionally) the arbitration agreement as a separate agreement when, in reality, it is not a free -standing agreement but is merely part of the main contract. However, that purpose does not extend to working out the conflict of laws rules applicable to an arbitration agreement . It follows that in decid ing on the proper law of the arbitration agreement , the arbitration agreement should be regarded as part of the main contract. 234. I therefore agree with the characteristically clear and helpful exposition by Adrian Briggs, Private International Law in English Courts (2014) , paras 14.37 - 14.38: “If the agreement to arbitrate is a term of a larger contract, the law which governs the con tract as a whole will generally determine the scope of the terms of that contract. For even though the arbitration agreement is for some important purposes notionally severable from the substantive contract, those purposes do not include the need for its g overning law to be separate or different from that of the substantive contract in which the arbitration agreement is contained. It would be perverse to deduce from the principle of severability a rule that the law governing the agreement to arbitrate shoul d be identified without reference to the substantive contract in which the parties included it as a term. The autonomy of the arbitration agreement is one thing; its hermetic isolation would be quite another. To put the point yet another way: the agreement to arbitrate is severable , but that does not mean it is separate. Prior to any severance it will have been governed by the law which governs the contract; after severance, it must remain governed by the same law, for otherwise ‘ it’ is not being severed; something else is instead being created. Page 90 The result is that if the law which governs the substantive contract is identified by the Rome I Regulation, that law is very likely to govern the agreement to arbitrate, and will therefore al so be used by the court to determine the validity, meaning and scope of the arbitration agreement. The fact that the Rome I Regulation makes no claim to identify the applicable law for arbitration agreements does not prevent the common law rules of private international law applying their own solution to the question, which is that the agreement to arbitrate is generally governed by the law of the contract of which it is a term if it is a term of a substantive contract.” (footnotes omitted) (iii) Dividing t he arbitration agreement from the rest of the contract may be problematic 235. There may sometimes be practical problems in drawing the line for proper law purposes between the arbitration agreement and the rest of the main contract. This case provides an excel lent example. This is because the arbitration agreement is itself part of a wider dispute resolution clause , ie article 50.1 (set out at para 191 above) includes an obligation to resolve the dispute in good faith and for there to be a meeting of senior man agement and only after that should the dispute, if still unresolved, be referred to international arbitration. It would be very odd and inconvenient to apply one proper law to interpret the earlier sentences in article 50.1 and a different proper law to in terpret the later sentences. Moreover, the terms “notification” and “written notice” are used in article 50.1 - and therefore impact on the time when the matter can be referred to arbitration - and the meaning of those terms is set out in article 51.4 of t he main contract. It might be said that the whole of article 50.1 should be separated off from the main contract for the purposes of deciding the proper law. But while that would avoid the difficulty of different proper laws applying within the same disput e resolution clause, it creates the problem of how to ensure consistency with other terms of the main contract, such as article 51.4 (or another example, article 51.2 which is an entire agreement clause). To have a different proper law applying to the defi nitional article 51.4 than applies to article 50.1 would be problematic. All these difficulties would be avoided if the proper law of the arbitration agreement were the same as the proper law of the main contract. 236. Let us further assume that, instead of put ting the arbitration agreement in a dispute resolution clause, the contract, as is often the case, had two separate clauses: a dispute resolution clause operative prior to arbitration and an arbitration agreement . Surely using two clauses instead of one ca nnot make all the difference to the proper law issue. Yet on the face of it that is what the seat approach would require. Page 91 237. One can envisage other examples of the difficulties that this division of the proper law would cause. Take, for example, the English l aw rule of interpretation that pre -contractual negotiations are not to be taken into account. Let us assume (as appears to be the case) that that is different from the law on interpretation in New York. Then, let us assume, that there is a main contract go verned by New York law which includes an arbitration agreement with London as the seat. There may be pre - contractual negotiations that are relevant to understanding the contract including the arbitration agreement . It would be most odd to take those negoti ations into account in interpreting the main contract (governed by New York law) but to exclude them when interpreting the arbitration agreement (governed by English law). Again that problem is avoided if the same proper law applies across the board. 238. Anoth er problematic example arises because of different possible approaches to a “no oral modification” clause. Such a clause is effective to prevent subsequent oral variations of a contract in English law (as laid down in MWB Business Exchange Centres Ltd v Ro ck Advertising Ltd [2018] UKSC 24 ; [2019 ] AC 119 ). Let us assume, as appears to be the case, that the contrary position is taken under New York law. Let us then assume that there is a contract containing a no oral modification clause and an arbitration agreement . The main contract is governed by New York law but London is the seat of the arbitration. If one applies different proper laws to the main contract and to the arbitration agreement , that would appear to produce the odd result that a subsequent oral variation, which might affect the arbitration agreement , would be effective in relation to the main contract but would be ineffective in relation to the arbitration agreement . Again there would be no such problem if the proper law that applied to the main contract applied also to the arbitration agreement . 239. No doubt one can envisage many other such practical problems arising from the division required by the seat approach. They indicate the underlying truth that, in contrast to the main contract approach, th e seat approach cuts across a principled way forward. (iv) In past cases excessive weight has been given to the seat of arbitration 240. It is not easy to pinpoint why, in several past cases (as we have seen in paras 209 -226 above) the seat of arbitration has b een thought to be of such major importance in determining the proper law of the arbitration agreement . True it is that the seat of arbitration (almost) invariably carries with it the curial law and the courts’ curial or supervisory jurisdiction (see para 193(vi) above). So in this case it is not in dispute that the curial law of the arbitration agreement here is England and that the English courts have curial or supervisory jurisdiction. It may be, therefore, that in the past there has sometimes been a failure to distinguish between, on the one hand, Page 92 the curial law and the curial/supervisory jurisdiction of the courts - which are (almost) invariably determined by the law of the seat - and, on the other hand, the proper law of the arbitration agreement . 241. As Adrian Briggs has written in Private International Law in English Courts (2014) , para 14.41: “ [T]he id entification of the seat is a reliable indicator of the law which was intended or expected by the parties to apply to the proceedings before the arbitral tribunal, to their support, supervision, and control, but it is not a statement of the law which will govern the initial validity and scope of the agreement to arbitrate. The parties may say that they wish to have arbitration in London, and it may well be true that they expect the Arbitration Act 1996 to provide the template for the procedure which will be followed once the arbitration is underway. But it does not follow, or does not need to follow, that the validity of the contract by means of which that agreement was or [was ] not made must also be understood to be governed by English law, for that is anot her question entirely .” 242. Another possible explanation for the weight given to the seat in older cases is that this has rested on the now outdated assumption (given the way modern international arbitration works) that arbitrators at the seat would only be co mfortable applying their own law. In Cie Tunisienne de Navigation SA v Cie d’Armement Maritime S A [1971] AC 572, the House of Lords reasoned that the choice of seat in an arbitration clause was an indicator as to the proper law of the main contract. A subm ission put forward in support of that was that a reason for choosing an English seat was because English arbitrators would be most familiar with English law. Lord Wilberforce rejected that submission. He said, at 596, “I venture to think that in commercial matters, at the present time, this may give insufficient recognition to the international character of the City of London as a commercial centre - the reason, rather than any preference for English rules, for which arbitration in London is selected. In th is case the arbitrators had no difficulty in finding for French law and I do not suppose they would find ascertainment of the French law as to damages any more difficult than the English law of anticipatory breach.” 243. And as Popplewell LJ said in the Court o f Appeal in this case, at para 72: “I doubt that [that submission] would now be accorded significant weight in the context of most international Page 93 arbitration in England, in which English arbitrators are often asked to decide questions under a foreign governing law and are regarded as equipped to do so. A fortiori it is inapplicable to a case such as the present involving arbitration under the ICC Rules which commonly involves appointment of foreign arbitrators from different legal traditions and discip lines notwithstanding that the seat of the arbitration is in London.” 244. Mr Dicker submitted that the seat might often be chosen to ensure neutrality . However, the desire for neutrality is surely normally concerned with the quality and integrity of the decisi on-makers and rarely has anything to do with the proper law to be applied (ie the relevant neutrality is referring to the decision -maker not the proper law to be applied by that decision -maker). T here may have been an implication in Mr Dicker’s submission that the parties in this case precisely chose England as the seat because they did not trust the Russian courts. Certainly one can readily accept that neutrality away from home courts may be a reason why parties choose international arbitration, and that t he curial or supervisory jurisdiction of the courts at the seat may be significant. But the desire for neutrality does not explain why the parties would choose the law of the seat rather than the law of the main contract as the proper law of the arbitratio n agreement. Moreover, in this case if the parties really did not trust the Russian courts, one would have expected there to have been an exclusive jurisdiction clause (requiring any litigation to come before the English courts) in the main contract. In an y event, we were supplied with no evidence to support any suggestion that the parties in the present case did not trust the Russian courts. Clearly they preferred to resolve the matter by arbitration rather than litigation but that is a different point. (v ) In past cases insufficient weight has traditionally been given to the implied choice of the parties 245. Although it is very difficult to rationalise all past cases, the apparent rationalisation given by the Court of Appeal in this case (mirroring other judic ial attempts), in seeking to put the law on a sound footing, with respect places insufficient weight on the implied choice of the parties. That approach was to say that, in general, the proper law of the arbitration agreement was dictated by the seat chose n for the arbitration unless there was an express choice of proper law in the main contract (see Popplewell LJ’s judgment at paras 90 -91 and 105 and above para 226 ). But why should only an express choice of proper law in the main contract have this effect? As Mr Bailey persuasively submitted , in his written case, “it is the fact that the parties have made a choice which matters, not the way in which that choice was manifested.” In other words, it makes no rational sense to place heavy weight on an express c hoice in the main contract while placing little weight on an implied choice in the main contract. Page 94 (vi) The curial law and curial jurisdiction can be separated out from the proper law of the arbitration agreement 246. A central submission of Mr Dicker, in line w ith the views of Popplewell LJ in the Court of Appeal at paras 96 to 99, is that one cannot properly separate out the curial law of the arbitration from the proper law of the arbitration agreement. They are intertwined. It follows, so the submission goes, that the parties are unlikely to have intended the proper law of the arbitration agreement to be different from the curial law (and we know that the latter is English by reason of the choice of seat). While in general terms, the curial law may be said to b e dealing with arbitral procedure, and the proper law of the arbitration agreement with the substance of the parties’ arbitration agreement (its existence, validity and scope), one cannot in this context neatly divide procedure and substance. This is illus trated, so the submission goes, by the provisions of the Arbitration Act 1996. If the seat is England, the provisions of that Act apply whatever the proper law of the arbitration agreement ; and many of these provisions ( for example, sections 5, 7, 12, 28(1), 58, 60, 71(4), 79 and 82(2)) are substantive not procedural. 247. Looked at in the overall context of the English rules on the conflict of laws, this may be thought a surprising submission. This is because it has long been recognised that, while there may be issues at the margins in drawing the distinction, there is an important difference between matters of procedure that are governed by the law of the forum and matters of substance that are governed by the particular proper law; and in modern times it would not be suggested that the forum chosen, governing procedure, would be a decisive , or even an important , factor in deciding on the proper law determining the substantive rights of the parties. 248. It should also be noted that one would face the same issue of s eparating out the curial law from the proper law of the arbitration agreement if there were an express choice of law clause in the main contract specifying a different proper law than the curial law. Yet there is wide acceptance that an express choice of law clause in the main contract would override the choice of seat in determining the proper law of the arbitration agreement . 249. Moreover, as regards the Arbitration Act 1996, I accept the submissions of Chubb Russia, put forward so persuasively on this matter by Toby Landau QC, that Mr Dicker’s submissions (and the reasoning of Popplewell LJ on this) are incorrect for the following two reasons : (i) Almost all the provisions of the Arbitration Act 1996 being referred to as substantive not procedural are non -mandatory. And in relation to such non -mandatory provisions, section 4(5) of the 1996 Act lays down (as one Page 95 would expect in any event) that a fore ign proper law for the arbitration agreement means that the non -mandatory provisions of the 1996 Act do not apply. This provision was not relied on by Chubb Russia in the Court of Appeal and was not mentioned in the Court of Appeal’s judgment. Section 4 re ads as follows: “ Mandatory and non -mandatory provisions (1) The mandatory provisions of this Part are listed in Schedule 1 and have effect notwithstanding any agreement to the contrary. (2) The other provisions of this Part (the ‘ non-mandatory provisions’ ) allow the parties to make their own arrangements by agreement but provide rules which apply in the absence of such agreement. (3) The parties may make such arrangements by agreeing to the application of institutional rules or providing any other means by which a matter may be decided. … (5) The choice of a law other than the law of England and Wales or Northern Ireland as the applicable law in respect of a matter provided for by a non -mandatory provision of this Part is equivalent to an agreement making pr ovision about that matter. For this purpose an applicable law determined in accordance with the parties ’ agreement, or which is objectively determined in the absence of any express or implied choice, shall be treated as chosen by the parties.” As the Suppl emental Report of the Department’s Advisory Committee on Arbitration Law (DAC) said, at para 12, section 4(5) “avoids the dangers that … a choice of England as the seat of the arbitration will necessarily entail the imposition of every provision of the Act ”. (ii) The remaining provisions of the 1996 Act relied on by Mr Dicker ( sections 12- 13 and 66 -68) appear to be procedural not substantive (they are concerned with extending time limits for beginning arbitration proceedings, Page 96 limitation periods, and the enforcement and setting aside of an award). But even if one regards them as substantive (see Popplewell LJ at para 96) it is clear that, in themselves, they cannot be regarded as having any bearing on the proper law of the arbitration agreement . (vii) Section 103(2)(b) of the Arbitration Act 1996 (codifying article V(1)(a) of the 1958 New York Convention) is neutral 250. Mr Dicker sought to pray in aid a rticle V(1)(a) of the 1958 New York Convention on the Recognition and Enforcement of Foreign Arbitral Awards, whi ch has been codified in what is now section 103(2)(b) of the Arbitration Act 1996. This statutory provision (which is materially identical to a rticle V(1)(a) of the 1958 New York Convention) reads as follows: “ 103 . Refusal of recognition or enforcement. (1 ) Recognition or enforcement of a New York Convention award shall not be refused except in the following cases. (2) Recognition or enforcement of the award may be refused if the person against whom it is invoked proves - (a) that a party to the arbitration agreement was (under the law applicable to him) under some incapacity; (b) that the arbitration agreement was not valid under the law to which the parties subjected it or, failing any indication thereon, under the law of the country where the a ward was made; …” This statutory provision therefore deals with the refusal of recognition or enforcement of a non- domestic arbitral award (ie an award made in a territory outside the UK in a state which is a party to the New York Convention: Arbitration A ct 1996 s ection 100(1)) where an arbitration agreement is invalid under the law to which the parties subjected it or, failing any indication thereon, under the law of the seat. True it is that that constitutes legislative acceptance of the relevance of the law of the seat . But this provision is only directly concerned with the enforcement or recognition of arbitral awards . It is not directly concerned with the validity of an arbitration agreement prior to any question as to its enforcement or recognition Page 97 al though Mr Dicker submitted (relying on Albert van den Berg, The New York Convention of 1958 (1981) , pp 126 -128) that what is relevant at the end should also be relevant at the start. Mr Bailey ’s response was that, even if one were to regard this provision as having relevance at the pre -enforcement stage, the provision tend ed to support his case because the relevance of the law of the seat is only at the default level : where the parties have chosen the proper law of the arbitration agreement , including impliedly, the law of the seat does not apply. In other words, his submission was that this statutory provision was simply irrelevant where there has been an implied choice (as on the facts of this case). I agree with that. However, it is important to add that the statutory provision is irrelevant to this case for a wider reason: as I have made clear at paras 194- 199 above, this case is concerned with the interpretation of an arbitration agreement and not with its validity. It should also be stressed that the aw ard in this case, because the seat is England, would be a domestic award to which section 103(2)(b) does not apply. 251. Nevertheless, I am here concerned to articulate reasons that apply generally to favour the “main contract” as opposed to “seat” approach. On the face of it, the statutory provision (and article V(1)(a) of the New York Convention) does offer support in relation to the validity of the arbitration agreement and, at least at the enforcement and recognition stage, for applying the law of the seat w here there has been no choice of law, express or implied, made by the parties. One may say that it represents a legislative policy, and a policy of international arbitration, which the common law should respect. However, in so far as one might apply this p rovision so as to make a practical difference to the determination of the proper law of the arbitration agreement (ie where one would be applying, as the proper law of the arbitration agreement , at the pre-enforcement stage, the law of the seat rather than the law of the main contract) there is a difficulty with reconciling that provision with the “validation principle”. We have explained in para 1 98 above that that principle is the general principle whereby the courts favour the proper law that would uphol d an arbitration agreement rather than one that would invalidate it; and this can be seen to rest on the assumption that rational parties would prefer to have an agreement upheld than not. It follows that, unless one is to accept the unfortunate conclusion that the legislative provision may (sometimes) override the validation principle (of course sometimes it will be consistent with it), one will need to interpret the provision in such a way that, where the arbitration agreement would be invalid under the l aw of the seat but valid under the law of the main contract, the law of the seat will give way to the law of the main contract. The most obvious way of achieving this is to recognise that the provision confers a discretion. The relevant statutory words are that recognition or enforcement of the award “may be refused”. Assuming there is such a discretion, it should be exercised to accommodate the “validation principle”. The consequence would be that any practical difference, as to validity, between the prope r law of the seat and the proper law of the main contract would be nullified. Page 98 252. Ardavan Arzandeh and Jonathan Hill, “Ascertaining the Proper Law of an Arbitration Clause under English Law” (2009) Journal of Private International Law 425 , 442, stress, correct ly in my view, that, while superficially attractive, it is problematic to decide the proper law of the arbitration agreement by reading across from article V(1)(a) of the New York Convention (and hence from section 103(2)(b) of the 1996 Act): “Although international harmonisation of choice -of -law rules on the basis of the rules enshrined in a rticle V(1)(a) of the New York Convention is superficially attractive, it is not wholly unproblematical. If a national court may, in the exercise of discre tion, order enforcement of an award notwithstanding the fact that the underlying arbitration clause is invalid according to the law specified by a rticle V(l)(a), it is legitimate to question whether it would be logical or sensible to treat the choice -of -la w rules endorsed by article V(l)(a), as interpreted by van den Berg, as being automatically applicable in contexts other than the enforcement of arbitral awards, contexts in which the element o f discretion is absent.” 253. The overall position, therefore, is th at not only does section 103(2)(b) have no direct relevance to the facts of this case (because we are concerned with interpretation not validity and the award would be a domestic award), it also has no direct relevance to our general enquiry because we are not concerned with the enforcement or recognition of an award. This is in line with the view of Robert Merkin, Arbitration Law (Issue 84, 2020) para 7.15 that the provision “has a more limited effect” than may at first sight appear. In any event, it would appear that the provision’s support for the “seat approach” can, and should, be limited so as to adhere to the “validation principle” (thereby nullifying any practical difference, as to validity, between the proper law of the seat and the proper law of th e main contract). For all these reasons, it seems reasonable to regard section 103(2)(b) of the Arbitration Act 1996 as an essentially neutral consideration that should not be regarded as inconsistent with, or as standing in the way of, a principled soluti on. (viii) The analogy to an exclusive jurisdiction clause 254. In deciding on a principled approach to the proper law of an arbitration agreement , it is helpful to think of the analogy between an arbitration agreement and an exclusive jurisdiction clause. Say one has a contract governed by Russian law but with a jurisdiction clause giving the English courts exclusive jurisdiction. What is the proper law of the exclusive jurisdiction agreement? Although Mr Dicker submitted that that clause would be governed (pre sumptively) by English law - as the courts (and p lace) chosen by the parties - he was not able to support that Page 99 submission with any convincing references. It would be surprising if, at least normally, the proper law of the jurisdiction clause is anything ot her than the same as the proper law of the main contract. Certainly that is the position favoured by Dicey , Morris & Collins , The Conflict of Laws , 15th ed (2012) , para 12.103: “[A]s a matter of common law, normally a jurisdiction agreement (like arbitration agreements, which are also excluded by article 1(2)(e) from the application of the Rome I Regulation) is governed by the law applicable to the contract of which it forms a pa rt. Accordingly, and as a matter of the common law principles of the conflict of laws, the law which governs the contract will also generally govern the jurisdiction agreement. This means … that this law governs the construction and interpretation of the a greement …” ( ix) Conclusion 255. Taken together, these reasons provide a convincing case for favouring the “main contract” as opposed to “seat” approach to determining the proper law of the arbitration agreement . They should be viewed as supporting a presumption (or general rule) that the proper law of the main contract is also the proper law of the arbitration agreement . In this case, they support the conclusion that the proper law of the arbitration agreement is Russian law by reason of an implied choice. (3) The proper law of the arbitration agreement is Russian law even if there has been no implied choice 256. I would arrive at the same conclusion - that the proper law of the arbitration agreement is Russian law - for the reasons that have been set out in paras 231-255 above , even if the pro per law of the main contract was Russian under article 4, rather than under article 3 (1), of Rome I Regulation at least if the reason for that was that Russia is the country with which the contract is most closely connected . That would then carry across to the third stage of the common law approach and would mean that, despite the seat for the arbitration being England, the arbitration agreement also has the closest and most real connection with Russia. That one arrives at the same result at common law whet her applying the implied choice or the default rule is unsurprising. It has long been recognised that there is a thin distinction between those two stages: they represent the distinction between implied and imputed intention. In Amin Rasheed Shipping Corp n v Kuwait Insurance Co [1984] AC 50 the majority, led by Lord Diplock, decided that English law was the proper law by necessary implication whereas Lord Wilberforce came to the same conclusion applying the closest and most real connection test while recogn ising, at p 69, that Page 100 the two “merge into each other”. But although, in general terms, it is important to recognise that one would arrive at the same conclusion if one applied the third stage of the common law approach, this case can be decided without goin g beyond the choice of the parties. The proper law of the arbitration agreement is Russian because that is the law which they have impliedly chosen . (4) Stating the common law on the proper law of an arbitration agreement 257. The reasoning above enables me to state the common law on the proper law of an arbitration agreement (contained in a main contract) in the following straightforward and principled way which (had this view found favour) would have been easy to apply and would have been one way of providing the clarity that Popplewell LJ was rightly seeking : (i) The proper law of the arbitration agreement is to be determined by applying the three stage common law test . Is there an express choice of law? I f not, is there an implied ch oice of law? If not, with what system of law does the arbitration agreement have its closest and most real connection? (ii) Where there is an express proper law clause in the arbitration agreement (which is rare) that will be determinative. (iii) Where there is no su ch clause, there is a presumption or general rule that the proper law of the main contract is also the proper law of the arbitration agreement . That presumption or general rule can assist the enquiry at any of the three stages of the common law approach. ( It is most appropriate to use the language of a presumption where one is considering the parties’ choice at the first two stages of the enquiry - ie it is a presumption of the parties’ intentions - and to use the language of a general rule where one is con sidering the third stage of the closest and most real connection.) (iv) That presumption may most obviously be rebutted, or there is an exception to that general rule, where the standard “validation principle” (of the English conflict of laws) applies ie where the law of the seat (or another relevant jurisdiction) would treat the arbitration agreement as valid whereas the proper law of the main contract would treat the arbitration agreement as invalid ( or, as in the Sulam érica case, not binding on one of the par ties). In very rare cases that presumption would also be rebutted where it is clear that the parties have chosen the law of the seat as the proper law of the arbitration agreement even though there is no express proper law clause in the arbitration agreement . Page 101 258. The above statement of the common law on the proper law of an arbitration agreement does not undermine the well -established and uncontroversial position that the curial law and curial jurisdiction are (almost) invariably determined by the seat c hosen for the arbitration. 7. Concluding remarks on the proper law of an arbitration agreement 259. We were referred to the writings of many commentators on this issue. Several (for example, Gary Born, International Commercial Arbitration , 2nd ed ( 2014), Chapte r 4; and Julian Lew, The Law Applicable to the Form and Substance of the Arbitration Clause in Albert van de n Berg (ed) Improving the Efficiency of Arbitration Agreements and Awards: 40 Years of Application of the New York Convention , (1998) ICCA Congress Series Vo l 9, 114, 114 -145) refer to the international context and I have been very conscious throughout that it would be inappropriate to lay down an approach for the English common law that would be inconsistent with accepted principles of international arbitration law. Although the commentators, as with the judges, do not speak with one voice on the issue facing us, I have found illuminating most of the writings to which we were referred. I have derived particular help from the work I have earlier mentio ned of Lord Mustill and Stewart Boyd, Gary Born, Robert Merkin and Louis Flannery , Albert van den Berg, Adrian Briggs, and Ardavan Arzandeh and Jonathan Hill . In addition, I have been helped by an excellent case -note on the Court of Appeal decision in this case by Edwin Peel, “The Proper Law of an Arbitration Agreement” (2020) 136 LQR 534. 260. It will be clear from all that I have said above that, while there are large measures of agreement between us (for example, that (at least in general) an express or impli ed choice of the proper law for the main contract carries across to be the proper law of the arbitration agreement , irrespective of the specified seat of arbitration) I cannot agree, with great respect, with the overall approach or conclusion in this case of my colleagues, Lords Hamblen and Leggatt (with whom Lord Kerr agrees). In their view, the proper law of the arbitration agreement is here English law because there has been no choice of law for the arbitration agreement, express or implied, and the arbitration agreement has the closest and most real connection to England as the seat of the arbitration. Their decision would have been different had the proper law of the main contract been Russian law by reason of an express or implied choice. But beca use the proper law of the main contract is, in their view, Russian law, only because it has the closest and most real connection to Russia, that means that the proper law of the arbitration agreement is English law. That is to rest crucially different cons equences on a divide between the choice and default stages of the Rome I Regulation and between the second and third stages of the common law approach in a way that, with respect, I do not believe to be justified in principle. I also consider that that app roach produces undesirable practical and unprincipled consequences (especially by forcing a division of the proper laws) such as those set out in paras 235- 239 above. I also have misgivings about the idea that Page 102 the English common law should depart from a principled solution on the basis of a supposed - but in my view unproven - consensus as to international arbitration policy favouring the seat approach (in the absence of choice). My view is that the proper law of the arbitration agreement is Russian. That i s because the proper law of the main contract is Russian by implied choice and that implied choice encompasses, or carries across to constitute, an implied choice of Russian law for the arbitration agreement . Even if my reasoning on the proper law of the m ain contract is wrong - and the proper law of the main contract is Russian by reason of Russia having the closest and most real connection rather than by implied choice - I would still regard the proper law of the arbitration agreement as being Russian law by reason of the arbitration agreement having the closest and most real connection with Russian law. This is to apply the general rule, to which there is here no exception, that the proper law of the main contract is also the proper law of the arbitration agreement . 8. The anti- suit injunction 261. Had my conclusion on the proper law of the arbitration agreement prevailed - that the proper law of the arbitration agreement is Russian - the following question would have arisen . Should this matter be remitted to t he English Commercial Court to decide if an anti -suit injunction should be granted or, as Mr Bailey submitted, should the matter be left to the Russian courts by refusing an anti -suit injunction (overturning the Court of Appeal)? It is not in dispute that the English courts, because England is the seat of the arbitration, have curial or supervisory jurisdiction to support and enforce the arbitration agreement (see para 193(vi) above ). It is also clear that the English Commercial Court has the means and expe rience, relying on expert evidence on Russian law, to decide on the correct interpretation of the arbitration agreement applying Russian law. I consider that, in these circumstances, had my view on the proper law of the arbitration agreement been the majority view, the appropriate course would have been for the question as to whether an anti-suit injunction should be ordered to be remitted to the English Commercial Court which would have been required to determin e whether, applying Russian law to interpret the arbitration agreement , the proceedings in Russia constitute d a breach of the arbitration agreement . That court would also have been required to determine , if Enka had been given permission to plead the point, whether, applying Russian law, there was a serious risk of the arbitration agreement being held invalid under Russian law as at the time this arbitration agreement was entered into (see para 1 97 above). Had my view on the proper law prevailed, t he stay of execution of the anti -suit injunction w ould not therefore have been lifted and the undertakings given by the parties, pending the outcome of this appeal, w ould have had to be extended to protect Enka’s position. Page 103 9. Conclusion 262. Contrary to the joint judgmen t of Lord Hamblen and Lord Leggatt (with whom Lord Kerr agrees), it is therefore my view that, on the main issue in the case, Chubb Russia is correct that the proper law of the arbitration agreement is Russian, not English, law; and, on that basis, I would have remitted the question, whether an anti -suit injunction should be ordered, to the English Commercial Court. LORD SALES : 263. I agree with the judgment of Lord Burrows. In relation to determining the proper law of an arbitration agreement contained in a mai n contract my view is that the “main contract” approach should be preferred to the “seat” approach. I add a short judgment of my own to explain my position in relation to the points on which there is a difference of view within the court and to indicate th e areas where I am in agreement with the judgment of Lord Hamblen and Lord Leggatt. 264. The court is taking this opportunity to clarify the position regarding the approach to determining the proper law of an arbitration agreement which is a provision within a main contract. The main contract may or may not contain a provision stating the proper law of the contract. Where the main contract contains such a provision, it is not usual for the parties also to include a distinct term to state the proper law of the arbitration agreement embedded in the main contract. 265. According to English conflict of laws rules, the proper law of the main contract will usually be determined by application of the Rome I Regulation, but that does not apply in relation to the arbitration agreement. In relation to the arbitration agreement, the proper law is determined by reference to the conflict of laws rules of the common law: the proper law is that chosen by the parties (i) expressly or (ii) by implication, according to the terms of any agreement between them, and (iii) in the absence of such choice is the law of the jurisdiction with which the arbitration agreement has the closest and most real connection. Choice of the parties 266. Where the main contract includes a provision st ating the proper law of that contract, I agree with Lord Hamblen and Lord Leggatt that the ordinary effect of the provision is that this indicates that the parties have chosen the same proper law for the arbitration agreement. Page 104 267. I further agree with Lord Ham blen and Lord Leggatt that for these purposes there is not necessarily a sharp division between an express choice of law and an implied choice of law. The point can be illustrated by the decision in Sulam érica Cia Nacional de Seguros SA v Enesa Engenharia SA [2012] EWCA Civ 638; [2013] 1 WLR 102 (“Sulam érica ”). That case concerned an arbitration agreement contained in a main contract which included a term stating that the proper law of the contract was Brazilian. In his judgment, Moore -Bick LJ assumed that what was in issue was whether the parties had thereby made an implied choice of law in relation to the arbitration agreement, and held that by virtue of the application of the validation principle the choice of law term could not be interpreted as having t hat effect: paras 25- 26 and 31. However, one might analyse the effect of the proper law provision in the main contract by asking whether on the true construction of its express terms the statement that the proper law of the contract was Brazilian law exten ded to cover the arbitration agreement which was part of that contract. Again, application of the validation principle would indicate that in the particular circumstances of the case the parties did not intend that statement to extend so far. 268. Stages (i) an d (ii) of the common law rule are aligned with the test in article 3(1) of the Rome I Regulation. The first main point of difference between the judgment of Lord Burrows and the judgment of Lord Hamblen and Lord Leggatt is whether in the circumstances of t he present case the parties impliedly chose Russian law as the law governing the main contract, including the arbitration agreement. On that question, I agree with Lord Burrows that they did. Although the parties did not include an express choice of law st atement in the main contract, they included many references in the main contract to make it clear that they intended that Russian law should govern their relationship. In the circumstances of the case, and given the nature of the task to be performed by En ka, it would have been bizarre for them to assume that any other law was to apply. The guidance in the report by Giuliano and Lagarde on the Rome Convention which later became the Rome I Regulation (para 203 above) is strong support for this view. Unlike i n Sulam érica , there was no good countervailing reason to indicate that the parties intended that the choice of law they had made for their contract should not extend to the arbitration agreement which was part of it. 269. Where the parties to a main contract in clude an arbitration agreement as part of that contract, then in general terms there are strong grounds to infer that they intend their choice of the law to govern that contract to cover the arbitration agreement as well, as Lord Hamblen and Lord Leggatt p oint out : para 53 above. T here is a presumption that in ordinary circumstances a contract has a single proper law since otherwise “a serious element of uncertainty” would be introduced into mercantile agreements : Jacobs , Marcus & Co v Credit Lyonnais ( 1884) 12 QBD 589, 602- 603 per Bowen LJ ; see also Kahler v Midland Bank [1950] AC 24, 42 (Lord MacDermott ). A contract contains a unified package of rights and obligations, created in the same set of circumstances, so the usual and natural inference is that Page 105 the parties intend, on an objective basis, that the same proper law should apply in relation to it. An arbitration agreement contained in the main contract imposes an obligation to take disputes to arbitration in certain circumstances , as part of the package of rights and obligations created by and set out in the main contract . In usual circumstances, I can see no good reason to infer that the parties to the main contract intended the interpretation of the obligation to arbitrate to be governed by any different system of law than the system of law which governs the interpretation of all the other obligations in their contract. 270. Applying the same system of law to govern the construction of the whole of the contract the parties have made ensures simplicity and coherence in its interpretation . It avoids the uncertainty associated with subjecting different parts of the contract to interpretation according to different systems of law. Any national system of law may be expected to have internal coherence, whic h will not be the case when two national systems of law are set side by side or are overlaid. Each will have an internal logic and in dealing with particular matters which is at variance from the internal logic of the other. Each may have different solutio ns to practical problems which are coherent within that system, but are opposed to the solutions given by the other system according to what is coherent within that other system. The presumption that a contract has a single proper law thus reflects the usu al expectations of the parties to a contract, since it is a reasonable inference that they prefer certainty, coherence and simplicity in working out the practical implications of their agreement. 271. In my view, these points underlie the observation by Lord Mu still in Channel Tunnel Group Ltd v Balfour Beatty Construction Ltd [1993] AC 334, 357- 358, that it would be exceptional for the proper law of the arbitration agreement to be different from the proper law of the main contract. Lord Mustill’s opinion in thi s area carries great weight. He also pointed out that it is less unusual for the curial law in relation to an arbitration to be different from the proper law of the main contract (and the proper law of an arbitration agreement contained in the main contrac t). The explanation for this is that the curial law follows the choice of seat. When the parties choose a particular seat, their reasons for doing so include the relationship stipulated by the law of the jurisdiction of the seat as to the grounds on which the courts of that jurisdiction may interfere with the arbitral process or its outcome and the extent to which those courts may take action positively to support the arbitral process and uphold the agreement to arbitrate, including by the grant of injuncti ve relief. These reasons apply whatever the proper law of the main contract or the arbitration agreement may be. Hence I do not consider that ordinarily the choice of the seat provides any sound basis to infer what the parties intended or might have expect ed the proper law of the main contract or the arbitration agreement to be. 272. Two comments may be made about this. First, many decades ago it was understood that when the parties stipulated that the seat for their arbitration would Page 106 be in a particular jurisdic tion their intention was that the arbitrators would be local lawyers chosen for their expertise in the law of that jurisdiction, so that the inference could be drawn that the parties intended that they would apply that law in determining issues in dispute, including as to the proper interpretation of the arbitration agreement and the main contract. But changes in the way international arbitration was conducted meant that such an inference was already suspect by 1970, as Lord Wilberforce explained in Cie Tunisienne de Navigation SA v Cie d’Armement Maritime SA [1971] AC 572, 596. Under modern conditions of international arbitration, in which arbitrators may be drawn from different jurisdictions and are regularly expected to receive evidence about and to apply foreign law, it is now no longer a plausible inference. 273. Secondly, parties may sometimes choose arbitration for resolution of their disputes with a seat in a neutral jurisdiction because one or other of them does not have complete trust in the impartiality of the courts of the state of the other. But a preference for a neutral seat does not support any inference as to the parties’ intentions as to the law which the arbitrators should apply when interpreting the main contract or the arbitration agreement. Arbitrators can be expected to apply any relevant law, of whichever legal system is appropriate, in an impartial way and the courts of the neutral jurisdiction will be impartial in applying the curial law. If an inference is sought to be drawn as to the prop er law of the main contract or the arbitration agreement, something more is required: an indication that the parties wished the law to be applied to govern the interpretation of their contract to be neutral in the sense that it is not aligned with the home jurisdiction of either of them. Exceptionally, the circumstances may support such an inference: see eg Egon Oldendorff v Libera Corpn [1995] 2 Lloyd’s Law Rep 64, 69 and see para 114 above. However, the circumstances of the present case show that no such inference can be drawn here. The parties have stipulated that Enka’s obligations under the main contract should incorporate norms of Russian law. Accordingly, it is my view that Lord Hamblen and Lord Leggatt overstate the significance of the choice of the seat in this case. The choice of curial law associated with the choice of the seat is directed to a different subject -matter (regulation of the relationship between the courts of place of the seat and the arbitral process) than the rules directed to determ ining the proper law of a contract for the purpose of interpreting it, so it is not appropriate to use the former as a basis for establishing what the latter should be. 274. The inference that the parties who made the contract in the present case intended that the interpretation of the whole of it should be governed by Russian law is especially strong , since the arbitration agreement is contained in a complex main contract with many interacting parts which have to live together in a coherent relationship. In particular, the parties’ intention, judged objectively, is that the obligation to arbitrate set out in the arbitration agreement contained in article 50.1 of the main contract should be interpreted in a way which makes it coherent with the other obligations i n the same provision to seek to negotiate in good faith to find Page 107 a resolution for disputes. I can see no reason why the interpretation of the latter set of obligations is not governed by Russian law, like all the other obligations in the main contract. The obligation to arbitrate in article 50.1 is likewise just another obligation set out in the main contract and it is so closely related to the other dispute resolution obligations in the main contract that the obvious inference is that the parties intended t he interpretation of the whole of the provision to be governed by the same law, ie Russian law. 275. The separability principle which exists in relation to an arbitration agreement contained within a main contract does not alter this analysis. That principle ha s limited significance. As reflected in section 7 of the Arbitration Act 1996, it allows for the survival of an arbitration agreement contained in a main contract if the validity, existence or effectiveness of the main contract is called in question, so th at the arbitrators can rule on such matters. This tells one nothing about the legal system which the parties intended or might reasonably have expected to govern the interpretation of the arbitration agreement as part of the main contract. 276. By contrast, the validation principle, as illustrated by Hamlyn & Co v Talisker Distillery [1894] AC 202 and Sulam érica , does allow one to draw an inference as to the system of law which the parties intended should govern the interpretation of the arbitration agreement. T he principle can provide a basis for distinguishing the proper law of the arbitration agreement from that of the main contract or, where the proper law of the main contract is uncertain, it may provide a basis for an inference also to be drawn that the pro per law of the main contract is intended to follow the choice of proper law for the arbitration agreement (in Hamlyn v Talisker Lord Herschell LC referred to this possibility at p 209). 277. In my view, the validation principle is an aspect of the general objective approach to determining the intention of the parties to a contract ut res magis valeat quam pereat (so that the main object of the agreement is upheld and not destroyed). Where the main contract contains an arbitration agreement, it will be clear that the parties intend that the obligation to arbitrate as set out in the arbitration agreement should be valid and effective. T he parties are presumed to know the state of the law at the t ime they contract . If it appears that according to the law which governs the main contract the arbitration agreement would be invalid, then it can be inferred that the parties intended that a different law should govern the arbitration agreement in order t o uphold its validity and effect. The same is true if it appears that according to the law which governs the main contract the arbitration agreement would be subject to a serious risk of being found to be invalid or that its binding force would be destroye d (as in Sulam érica ), since the inference is that the parties would choose certainty rather than uncertainty in upholding the effect iveness of this part of their contract. Usually, since the legal system which governs the main contract is ruled out by this reasoning, the obvious conclusion is that the parties intended the law of the jurisdiction of the seat which they have stipulated to apply instead. The terms of Page 108 the arbitration agreement, set against the background of the state of the law in the two candi date jurisdictions, show that the parties intended the law of the jurisdiction of the seat to apply in this sort of case. This reasoning does not apply where what is in issue is the choice of the proper law to determine the scope of the arbitration agreeme nt rather than whether it would be invalid or would not impose a binding obligation to go to arbitration if one system of law were applied rather than another. 278. In Sulam érica , Moore -Bick LJ rightly held that the validation principle applied so as to negativ e any choice of Brazilian law as the proper law of the arbitration agreement. He seems to have drawn the conclusion that this meant that the parties had formed no intention regarding what was to be the proper law of the arbitration agreement (see para 31) and so proceeded to analyse the position by reference to the common law default rule at stage (iii), in order to conclude that English rather than Brazilian law governed the arbitration agreement contained in the main contract. However, in my opinion, foll owing the reasoning above, the better view is that the validation principle showed that the parties intended that English law should govern the arbitration agreement. This conclusion should have been reached at stage (i)/stage (ii) of the common law analys is. 279. In the present case, subject to one argument introduced by Enka for the first time on the appeal to this court (see para 197 above), the validation principle has no application. Up to the hearing in this court, it has been common ground that under Russian law the arbitration agreement in article 50.1 is valid and binding in its effect; the issue that has divided the parties is the effect that application of Russian law would have regarding the interpretation of its scope. As to Enka’s new argument that the validation principle does in fact apply, I agree with Lord Burrows that if our view regarding the proper law of the arbitration agreement had prevailed the case should have been remitted to the Commercial Court and that it would have been for that court to consider whether the new argument could be introduced and, if it were, then to rule upon it alongside the other issue of Russian law which is in dispute between the parties, namely whether the interpretation of article 50.1 according to Russian law wo uld be narrower or the same as that given by English law. 280. The second main area of disagreement appearing from the judgment of Lord Burrows and the judgment of Lord Hamblen and Lord Leggatt relates to the operation of the common law default rule at stage (i ii), if the parties have made no choice at stage (i) or stage (ii). On the analysis of Lord Burrows, with which I agree, the parties to the main contract impliedly intended that the interpretation of the AA in that contract should be governed by Russian la w, at stage (ii). If that were right, stage (iii) would not be reached. However, the majority do not agree about this. On their analysis it is necessary to consider the position on the footing that the parties have made no choice at stage (i) or stage (ii). Page 109 The default rule 281. In the early formulation of the common law rule by Dicey in 1896 (para 36 above), the difference between stage (i)/stage (ii) and stage (iii) was described as one between what the parties (actually) intended and what they “may fairly be presumed to have intended”. Obviously, imputed choice is something different from actual choice. Later, the common law default rule at stage (iii) was formulated in terms of the system of law with which the contract has its closest and most real connection . But this does not mark a radical change. Rather, focusing on the closest and most real connection serves the same underlying policy, which is to seek to reflect the likely expectations of the parties as businesspeople, by producing an outcome which is re asonable and coherent in its own terms and does not place excessive emphasis on the boundary between stage (ii) and stage (iii). If, on analysis, the parties have not made a choice of proper law themselves - perhaps because they did not think about it or t hey chose to leave matters unclear in the interests of arriving at an agreement without having to argue about it and in the hope that a dispute might never arise which required a determination of the issue - the policy of the common law, as expressed in th e default rule at stage (iii), is to produce the answer which it is plausible to think businesspeople in the position of the parties, acting reasonably, would have been likely to have chosen for themselves if they had to confront the issue. 282. Many of the fac tors relevant to an argument that an implied choice of proper law can be identified at stage (ii) will also be relevant to the alternative argument based on the default rule at stage (iii). In broad terms, businesspeople would expect them to be likely to p roduce similar outcomes. That has certainly been the judicial approach until fairly recently, as illustrated by the decision of the House of Lords in Amin Rasheed Shipping Corpn v Kuwait Insurance Co [1984] AC 50. In that case, the majority of the Appellat e Committee determined the proper law of the contract by reference to stage (ii), while Lord Wilberforce reached the same conclusion by reference to the test at stage (iii), for closely similar reasons. Similarly, in the Cie Tunisienne case all members of the Appellate Committee arrived at the same conclusion regarding the proper law of the contract, but they did so by different routes; some found that the parties had made a choice, others that the default rule in stage (iii) applied. In th e leading authorities referred to in the Cie Tunisienne case, Bonython v Commonwealth of Australia [1951] AC 201, and In re United Railways of Havana and Regla Warehouses Ltd [1961] AC 1007 , the test applied to determine the proper law of the contract was that stated by Lord Simonds in Bonython, at p 219: “the system of law by reference to which the contract was made or that with which the transaction has its closest and most real connexion” , which elides the question of party choice and the default rule , and deliberately so. The close alignment of the approach under stage (ii) and that under stage (iii) was traced by Toulson LJ in Lawlor v Sandvik Mining and Construction Mobile Crushers and Screens Ltd [2013] 2 Lloyd’s Law Rep 98, paras 20- 27. Page 110 283. Since the boundary between stage (ii) and stage (iii) is by no means crystal clear and there is scope for eminent judges to reach different views about which stage of the common law analysis supplies th e answer in any given case, it would risk the appearance of arbitrariness to adopt a default rule at stage (iii) which was radically at variance in the results it produced by comparison with stage (i) and stage (ii). Further, if the common law adopted a ra dically divergent default rule, so that significant differences in outcome turned on this, that would be an incentive for parties to litigate the question of whether a case was to be analysed as falling within stage (i)/ stage (ii) or within stage (iii). T his would be contrary to the interest of businesspeople to avoid expensive litigation to resolve disputes, so far as possible. If the parties appreciate that all roads lead to Rome, so to speak, the need for litigation to decide which road should be taken is avoided. 284. This analysis prompts a further comment on Sulamérica . Having held (contrary to my view at para 278 above) that the application of the validation principle meant that the parties had made no choice as to the proper law of the arbitration agreem ent within the main contract, Moore -Bick LJ proceeded to apply the default rule at stage (iii) (para 32). However, in doing so he took the view that the arbitration agreement had its closest and most real connection with the law of the place of the seat (E ngland); and this despite the fact that, subject to the application of the validation principle, he thought that at stage (ii) the parties impliedly intended that the proper law of the main contract (Brazilian law) would also apply to the arbitration agree ment (paras 26-27). I think it is evident that Moore - Bick LJ’s analysis at both stage (ii) and stage (iii) was rightly designed to give effect to the validation principle and to uphold the effective binding force of the arbitration agreement in that case i n line with the parties’ intention. But unfortunately in doing so he proposed a solution which, if taken at face value and generalised, would give rise to the kind of radical divergence of outcome between stage (i)/stage (ii) and stage (iii) which the comm on law default rule in fact seeks to avoid, and which does not reflect the previous authorities referred to above. 285. The court in Sulam érica did not need to take the step of saying that the arbitration agreement had its closest and most real connection with the law of the place of the seat in order to produce the appropriate result, which was to uphold the binding effect of the arbitration agreement in line with the parties’ intention by application of the validation principle: see para 278 above. One might a lso say that the valid ation principle is capable of operating at stage (iii) as well as at stage (i)/stage (ii), as an aspect of the common law default rule, as an expression of the policy of the common law to uphold the validity and binding effect of an a rbitration agreement which the parties have chosen to enter into. But again, that would mean that the law of the place of the seat (England) was applicable as the proper law of the arbitration agreement as the only remaining candidate once Brazilian law ha d been eliminated as a candidate by application of the validation principle. At the end of this process of analysis, it could be said that the arbitration agreement had its Page 111 closest and most real connection with the law of the place of the seat; but that is only in the very limited sense that this was the only system of law with which the arbitration agreement could be said to have any connection, if the validation principle was to be given effect. However, the way in which Moore -Bick LJ explains his reasoni ng at para 32 makes it sound as though the general starting point, if the analysis at stage (i)/stage (ii) does not give a result, is always that the arbitration agreement contained in the main contract has as its proper law the law of the place of the sea t rather than generally following the proper law of the main contract. In my respectful opinion, that approach is erroneous and contrary to principle and authority. 286. In my view, the powerful points which Lord Hamblen and Lord Leggatt make at para 53 of thei r judgment regarding the expectations of businesspeople to the effect that their contractual arrangements should have internal coherence (so that if the parties have chosen the proper law of the main contract they would ordinarily expect the same proper la w to apply in relation to an arbitration agreement contained within it) also apply in relation to the operation of the default rule at stage (iii) where the circumstances mean that it is clear what the proper law of the main contract is, even when that is not as a result of the exercise of choice within the meaning of article 3(1) of the Rome I Regulation. The main contract carries with it the legal system which governs its interpretation and application. Accordingly, the need for and expectation that there will be coherence between the main contract and the arbitration agreement contained within it means that the arbitration agreement has its closest and most real connection with the legal system which constitutes the proper law of the main contract in whic h it is contained. 287. By contrast, it is my opinion that the argument for a connection between the arbitration agreement and the law of the place of the seat is much weaker. The parties obtain the benefits of the curial law of the place of the seat in any eve nt, whatever the proper law of the arbitration agreement: see para 271 above . Therefore the choice of seat does not point to any particular connection with the arbitration agreement in terms of providing guidance as to its proper law. To the extent that th e courts of the place of the seat exercise a supervisory function in relation to the arbitration, for example to ensure that the arbitrators act within the scope of the arbitration agreement according to its true construction, they can readily do that by reference to evidence about any foreign law which is identified as the proper law of the arbitration agreement. 288. In the present case, Enka disputes that there has been a choice of proper law within article 3(1) of the Rome I Regulation for the main contract but accepts that article 4, as the default rule set out in the Regulation, has the effect that the proper law of the main contract is Russian. This concession must be based on an acceptance that it is clear from all the circumstances that the main contract is manifestly more connected with Russia than with any other country (including the country where Page 112 Enka is habitually resident, Turkey): see article 4(3). The assessment under article 4(3) involves inquiring into the country with which the contract taken a s a whole has its closest connection. Where, in this case, following this path of analysis, the main contract taken as a whole manifestly has its closest connection with Russia so that Russian law is taken to be its proper law, it seems to me that the reas oning above indicates that the arbitration agreement contained in the main contract similarly has its closest and most real connection with Russian law. There is no good reason to conclude that the law of the seat is more closely connected or provides a be tter guide for the purposes of application of a rule designed to identify the law which is to govern the interpretation of the arbitration agreement. 289. In their judgment, in relation to stage (iii) of the common law rule Lord Hamblen and Lord Leggatt rely on article V(1)(a) of the New York Convention and section 103(2)(b) of the Arbitration Act 1996 in support of their view that at that stage the arbitrat ion agreement in the main contract has its closest and most real connection with the law of the seat (England) rather than with the law which governs the main contract. In my opinion, this is to give those provisions excessive weight in analysing the appli cation of the common law rule. As I have sought to show, the policy of the common law as reflected in the default rule at stage (iii) is to align that rule with the likely result the parties would have wished to achieve to produce reasonable coherence acro ss their whole contractual relationship. Application of article V(1)(a) would defeat that policy, because it would produce a radical divergence between the effect of stage (i)/stage (ii) and stage (iii) of the common law rule. 290. Another way of putting this i s to say that the points made by Lord Hamblen and Lord Leggatt at para 53 of their judgment do not drop out of the analysis for the purposes of the common law at stage (iii), but continue to have validity and force at that stage as well. By contrast, when one is applying article V(1)(a) those points do drop out of the picture and have no force, precisely because the New York Convention legislates for a rule which excludes them from being relevant. 291. Moreover, article V(1)(a) does not provide a good guide as to the application of the common law rule. Article V(1)(a) sets out a default rule within the scheme of the Convention which is different from the default rule under the common law and which, if ap plied, would undermine the validation principle when it is applied by the common law as an aspect of stage (iii) (see para 285 above). The provision states that, in the absence of a choice by the parties, recognition of an arbitral award may be refused if the arbitration agreement “is not valid … under the law of the country where the award was made”. That seems to say that recognition may be refused if the arbitration agreement is invalid according to the law of the place of the seat; but under the common law in such a case the validation principle would apply and the court would identify another system of law as the proper law of the arbitration agreement in order to uphold and give effect to the arbitration Page 113 agreement. Article V(1)(a) thus sets out what ca n fairly be described as a very simple and inflexible default rule for the purposes of the Convention regime which is different from the more flexible and nuanced common law default rule of “closest and most real connection” and should not be taken to displace that rule. Within the Convention regime, the rationale for the choice of a simple test is not difficult to understand. It is a clear rule by reference to which it is reasonably easy to judge whether the actions of states party to the Convention comply with it or not. By contrast, the common law default rule has been established for a very long period of time, well before international policy arguably came to crystallise in line with article V(1)(a) of the New York Convention, and it reflects different policy objectives, as set out above. So far as choice of proper law for an arbitration agreement is concerned (as distinct from regulation of the recognition of foreign arbitral awards, which is governed by section 103(2)(b) of the 1996 Act), article V(1)(a) of the New York Convention is part of an unincorporated treaty and it is unclear by what process of legal reasoning it could be taken to have displaced the well established common law default rule. None of the leading common law authorities give any wei ght to article V(1)(a) in the formulation or application of the common law rule. 292. For present purposes, it is not necessary to determine the position where it is not article 4(3) but one of the other more mechanical rules in article 4 which determines the p roper law of the main contract. It suffices to say that I think there is force in the argument that the analysis above tends to indicate that also in that sort of case the proper law of the main contract will usually provide the best indication of the prop er law of an arbitration agreement contained within it, at stage (iii) of the common law rule. Again, the points made by Lord Hamblen and Lord Leggatt at para 53 of their judgment should not drop out of the picture here. This approach would reflect how the parties are likely to have approached matters themselves, by starting with their agreement on the substantive aspects of the main contract and then adding the arbitration agreement into that framework, with the general intention and expectation that the m ain contract and the arbitration agreement would form a coherent whole. It would also have the merit of making the analysis in any case as simple and clear as possible. One would start by identifying the proper law of the main contract according to the cho ice of the parties pursuant to article 3 of the Rome I Regulation and, in default of any choice, by reference to the rules in article 4 of the Regulation, and then the presumption would be that the proper law of the arbitration agreement is the same. The anti -suit injunction 293. Finally, if the interpretation of article 50.1 were governed by Russian law, as Lord Burrows and I think it is, and a Russian court is about to pronounce on the interpretation of that provision according to Russian law in the parallel p roceedings between the parties in Russia, the question arises whether this makes it inappropriate for the English court to issue an anti -suit injunction in favour of Enka, whether on Page 114 grounds of forum non conveniens, comity or otherwise. On that issue, I ag ree with section IX of the judgment of Lord Hamblen and Lord Leggatt, which is in line with Lord Burrows’ judgment. The English court, as the court of the place of the seat of the arbitration chosen by the parties, has a particular responsibility to ensure that the arbitration agreement is upheld and applied in accordance with its terms. On the basis of expert evidence of foreign law adduced in the usual way, the English court could determine the meaning of article 50.1 according to Russian law. If article 50.1, so construed, imposes an obligation on Chubb Russia to proceed by way of arbitration rather than by litigation, the English court could and should enforce that obligation by way of an anti -suit injunction.