Limitation, Empowerment and the Value of Legal Certainty in the Treaty Incorporation References Case

1. Since the ‘process’ of contemporary devolution first began in 1998 three issues have often given lawyers, academics and critics food for thought. The first issue is external to the settlement: the relationship between devolution and parliamentary sovereignty. The ability of either to survive the existence of the other has been a central question for the discourse, with both sovereignty and devolution potentially posing a deathknell for the other. The second issue is internal to the settlement – the tension between, on the one hand, the immensely technical nature of the devolution statutes themselves and, on the other, their profound constitutional significance. The former has sometimes served to obscure the latter from view. It would be curious, however, given that ‘[t]he carefully chosen language in which these provisions are expressed is not as important as the general message that the words convey’, for the technicalities of the Scotland Act 1998 to stand in the way of its constitutional ‘message’. The third issue is how best to make sense of that message, and to situate it within the constitution more generally. However, it is probably fair to say that the constitutional significance of the settlement itself is, unlike in its earlier years, no longer uncertain. Few would contend now that the devolved legislatures deserve comparisons with an English parish council. Curial attestations of the constitutional value of the devolved legislatures abound, a value which is predicated by their democratic credentials and reinforced in Scotland and Wales by legislative declarations of

permanence, 6 and in Northern Ireland by recourse to popular sovereignty. 7 Yet, this constitutional status is by no means a panacea when a court is confronted with technical points of law, as the Supreme Court was in the Treaty Incorporation References case. 8 Indeed, this case highlights the significance of the three issues outlined above and the Court's conclusions about each of them are telling as to its understanding of both devolution itself, and its relationship with the wider constitution.
2. The Treaty Incorporation References judgment contains a great deal of significant dicta, some of which have been analysed elsewhere. 9 The contribution this article seeks to make is to analyse two key issues in the judgment. The first concerns the s 33 reference procedure itself. This procedure is the mechanism by which this case reached the Supreme Court, but its use has a number of interesting consequences. The second issue explored is how the Court balances the different dimensions of the Scotland Act, namely how it reconciles the Scotland Act's creation and empowerment of a democratic legislature with that Act's simultaneous and explicit provision of limitations to its competences. How the Court balances these competing 'dimensions' of the Scotland Act, and what factors it points to in determining that balance, goes some way to revealing the Court's understanding of the constitutional status of the legislatures in each of the four parts of the UK. Before this discussion, the case in question is briefly outlined.

The Treaty incorporation references
3. In March 2021, the Scottish Parliament passed two pieces of legislation which sought to incorporate into Scottish domestic law two treaties to which the UK is a signatory. The first is the United Nations Convention on the Rights of the Child (Incorporation) (Scotland) Bill ('the UNCRC Bill') and the second is the European Charter of Local Self-Government (Incorporation) (Scotland) Bill ('the ECLSG Bill') which each purported to give domestic effect to their relevant treaties. In order to do so, each Bill contains several important provisions which are themselves mirrored on key provisions of the Human Rights Act 1998 (HRA). First, the UNCRC Bill creates an interpretive obligation that relevant legislation '[s]o far as it is possible to do so … must be read and given effect in a way which is compatible with the UNCRC requirements'. 10  already enacted that is incompatible with those requirements. 11 Thirdly, it gives Scottish courts a power to make an 'incompatibility declarator' against future legislation if it meets that same test. 12 Each of these powers can sound against Acts of the Westminster Parliament which would be within devolved competence.
4. The ECLSG Bill contains analogous provisions to those in the UNCRC Bill: an interpretive obligation 13 and a power to make a 'declaration of incompatibility'. 14 The UK Law Officers contended that these provisions in both Bills were outwith competence because they modified Westminster's 'unqualified legislative power' to make laws for Scotland, as protected by s 28(7) of the Scotland Act. 15 The UK Law Officers also contended that a provision of the UNCRC Bill which would make it unlawful for public authorities to act incompatibly with the UNCRC Treaty 16 modified s 28(7), related to reserved matters and modified the law on reserved matters. The UK Law Officers finally contended that these provisions could not be saved by s 101 of the Scotland Act. 17 The UK Law Officers were successful on every point. refer the question of whether a Bill or any provision of a Bill would be within the legislative competence of the Parliament to the Supreme Court for decision'. 23 Section 32 provides that such a Bill cannot be submitted for Royal Assent when such a reference can be made, has been made (but not yet disposed of) or if the Supreme Court has held that, unamended, any provision of the Bill would be outside the legislative competences of the Scottish Parliament. 24 6. An important consequence of the s 33 procedure is that it provides a barrier to the Scottish Parliament's power to make laws which is far more robust than a post-enactment finding to the same effect. If it is the s 33 procedure which finds a Bill to be outside competence, that Bill cannot be put for Royal Assent until it is amended and, when it is amended, it can be referred again. 25 Consider, by way of contrast, the only case where a piece of Scottish legislation has been found to be outside legislative competence after enactment: Salvesen v Riddell. 26 Although that case concerned incompatibility of the Scottish provision in question with the European Convention on Human Rights, Lord Hope (giving the unanimous judgment of the court) was notably careful to limit his conclusions only to the tainted provisions: But the finding of incompatibility ought not to extend any further than is necessary to deal with the facts of this case, and it is important that accrued rights which are not affected by the incompatibility should not be interfered with. As the incompatibility arises from the fact that sections 72(10)(a) and 72(10)(b) are so worded as to exclude landlords of continuing tenancies from the benefit of section 73 if their notices were served or the specified thing occurred before the relevant date, I would limit the decision about the lack of legislative competence to that subsection only. 27 7. Furthermore, Lord Hope went so far as to 'make an order under section 102(2)(b) of the 1998 Act suspending the effect of the finding that [the provision in question] is outside the legislative competence of the Parliament for 12 months or such shorter period as may be required for the defect to be corrected and for that correction to take effect'. 28 This stands in obvious contrast to the effect of a s 33 reference, where the entire Bill must stand still.
8. However, what both Salvesen and s 33 cases do have in common is their prioritisation of legal certainty. Provisions which are impugned because they are outwith competence (and are therefore 'not law') are in both cases excised from (or not permitted to reach) the statute book rather than persisting but being 'disapplied' (or otherwise  28 ibid. In the same paragraph he also explained that he 'would give permission to the Lord Advocate to apply to the Court of Session for any further orders under section 102(2)(b) that may be needed in the meantime to enable the Scottish Ministers to achieve the correction before the suspension comes to an end.' deprived of legal effect) in relevant cases. In other words, a disconnect between the text of a statute and the state of the law is intolerable. 29 The courts, as this case demonstrates, seem to prefer that the picture of the law provided by the Scottish Parliament is as accurate as possible.
9. The s 33 procedure itself also raises the question of how competence limitations apply to devolved legislation. In particular, do these limitations apply automatically, or must they be written into the Bills more explicitly in pursuit of a more accurate account of the legal picture? Although there is at least one arguable exception in the case law, 30 the Supreme Court in Treaty Incorporation References held that competence limitations do not apply automaticallyor implicitlyto the Scottish Parliament's enactments. Rather, Scottish legislation must take account of those limitations and ensure that they are reflected by its text. The court contrasted two types of provision and suggested a preference for those that take express account of the competence limitations, rather than relying on the courts to apply those limitations in future. It said that whereas 'no attempt has been made to confine [s 6 of the UNCRC Bill's] scope to matters falling within the legislative competence of the Scottish Parliament', 31 this section 'might be contrasted with sections 19-21, which expressly confine their scope to legislation which it would be within the legislative competence of the Scottish Parliament to make'. 32 This latter kind of legislation, which provides a clearer, more certain picture of the relationship between competences and the Bills, is clearly, in the court's view, favourable.
10. In Treaty Incorporation References, the Supreme Court gives two reasons for this preference for legal certainty: first, because Westminster, in enacting the Scotland Act, cannot have intended to allow the Scottish Parliament to legislate to create a degree of legal uncertainty that would be intolerable to the rule of law; 33 second, because the existence of the aforementioned pre-enactment safeguards suggest that the relationship between a Bill and the Scottish Parliament's competences should be conclusively discernible at any given moment 29 This raises the distinct but interesting question of whether the Scottish Parliament can enact legislation that is intended to have no legal effect. As is well-known, the majority in the Miller I case held that the Westminster Parliament had not intended s 28 (8)  Although it is perfectly true that there are no express words in section 9 which limit its scope in this way, I am satisfied that it does have such a limited effect. That is because of the simple legal principle, identified by Lord Reed, embodied in the Latin maxim nemo dat quod non habet. Given that the jurisdiction of the Assembly is limited to removing, or delegating the power to remove, functions of Ministers of the Crown when the removal satisfies the requirements of paragraph 6(1)(b) of Part 3 of Schedule 7 to the 2006 Act, the Assembly cannot confer a wider power on Welsh Ministers. Accordingly, the wide words of section 9 must be read as being circumscribed in their scope so as to render the section valid.' 31 In re United Nations Convention on the Rights of the Child (n 8) [59]. 32 ibid [62]. 33 This was particularly important in the Court's analysis of s 101, explored in more detail in Elliott and Kilford (n 9).
in time. 34 In other words, competences should be seen as 'static' and it should always be clear how the Scottish Parliament's legislation fits within them. In the Court's view, the s 33 procedureamong other safeguardswould simply be superfluous if limitations applied automatically. Because of the presence of those safeguards, the Supreme Court considers that its preference for legal certainty is justified. Nonetheless, such a view reveals an internal tension within the Scotland Act: the various pre-enactment safeguards, including s 33, must sit within the context of a statute that provides far reaching constitutional authority to alter much of the statute book in a way that might naturally, by definition, give rise to a natural, and perhaps tolerable, degree of uncertainty. The Court's pursuit of legal clarity and certainty in defence of those safeguards, even if justified, ultimately serves to limit the Scottish Parliament's ability to exercise its constitutional authority as broadly as it might otherwise be able.
Balancing the competing dimensions of the Scotland Act 11. The Scotland Act, as s 33 itself demonstrates, has two distinct dimensions which it expresses simultaneously. On the one hand, it is a statute that establishes a legislature for Scotland and invests it with tremendous powerpower even to repeal Acts of the Westminster Parliament in some circumstances. However, and at the same time, it outlines in immense detail the limitations on the Scottish Parliament's competences and provides several mechanisms for ensuring that those limitations are not overstepped. Indeed, because of the reserved powers model, these limitations need essentially to be exhaustive. 35 The upshot of this is that some of the most significant provisions in the Scotland Act actually explicate what the Scottish Parliament cannot do, rather than what it can. So how, when confronted with such a body of limitations, is the Court to balance them with the more general grant of authority across the Act?
12. The case law seems to provide two (related) answers to this question: (1) by reference to the constitutional significance of devolution; and (2) by reference to the purpose of the devolution statutes. The first of these pulled the strings of the earliest judicial reasoning when the courts initially considered the constitutional place of devolution 36 and took a broad view of where election powers had their source, 37 but has largely fallen out of favour since. 38  guide to their interpretation, reference to the devolved legislatures' institutional characteristics, especially those that might be thought to render them constitutionallike, for example, their democratic credentialshas maintained significance. In particular, it has justified their immunisation from 'judicial review at common law on the grounds of irrationality, unreasonableness or arbitrariness', 40 and in guaranteeing a degree of deference to them as part of a proportionality analysis. 41 13. The interpretive methodology that has found the most consistent favour with the courts is recourse to the purpose of the devolution settlement: to create 'a coherent, stable and workable system within which to exercise … legislative power'. 42 Lord Hope explained the position in Imperial Tobacco in a passage worth repeating in full: the description of the [Scotland] Act as a constitutional statute cannot be taken, in itself, to be a guide to its interpretation. The statute must be interpreted like any other statute. But the purpose of the Act has informed the statutory language. Its concern must be taken to have been that the Scottish Parliament should be able to legislate effectively about matters that were intended to be devolved to it, while ensuring that there were adequate safeguards for those matters that were intended to be reserved. That purpose provides the context for any discussion about legislative competence. So it is proper to have regard to the purpose if help is needed as to what the words actually mean. The fact that section 29 provides a mechanism for determining whether a provision of an Act of the Scottish Parliament is outside, rather than inside, competence does not create a presumption in favour of competence. But it helps to show that one of the purposes of the 1998 Act was to enable the Parliament to make such laws within the powers given to it by section 28 as it thought fit. It was intended, within carefully defined limits, to be a generous settlement of legislative authority. 43 14. The emphasis this reasoning places on workability reflects ideas expressed in prior cases, 44 and has been influential since. 45 'Workability', as this dictum rightly explains, points in two directions: it suggests both that the 'generous' legislative power of the Scottish Parliament should be freely exercisable, and that the limitations on its competence should be clear and effective. In other words, both competing dimensions of the settlement should be enforced to their proper extent. If the legislative competences themselves are too broadly construed, then the limits on competence might become meaningless, but the opposite is also a possibility. paramount importance, the courts showing care to ensure that the empowering provisions of the legislation are not deprived of effect. 46 However, other cases have shown that stabilityand its concomitants of certainty and clarityare at the core of the purpose of the devolution statutes. 47 15. Where the Court comes down on this issue in the Treaty Incorporation References judgment is fascinating. In short, the Court 'weighs' its reasoning heavily in favour of a restrictive approach: it reads the empowering provisions in the Scotland Act narrowly, and it takes a very broad view of the provisions that provide competence limitations, something it justifies primarily by reference to legal stability and certainty. In order to demonstrate this, the Court's contrasting treatment of ss 28(7) and 101 is analysed.
The Court's view of section 28 (7) 16. The Court's expansive view of the limitations in the Scotland Act is arguably most apparent in its treatment of s 28(7). The nature of the s 28(7) limitation is important to keep in mind. It is an enactment under Sch 4 which is protected from 'modification' by the Scottish Parliament. This process, as the Supreme Court noted in Continuity Bill, should be carefully distinguished from the reserved matters mechanism under Sch 5: When the UK Parliament decides to reserve an entire area of the law to itself, it does so by listing the relevant subject-matter in Schedule 5. When it has not taken that step, but has protected a particular enactment from modification by including it in Schedule 4, it is not to be treated as if it had listed the subject-matter of the enactment in Schedule 5. 48 17. Section 28(7) does not provide a reserved matter, and nor is it listed in s 29 as a discrete limitation to competences. It also stipulates that it is 'this section', rather than the Scottish Parliament's legislation (although such legislation does arguably give life to s 28), which 'does not affect the power of the Parliament of the United Kingdom to make laws for Scotland'. The Court, however, endorses a broad view of both the meaning of this guarantee for Westminster (consisting, in part at least, of its immunity from 'political opprobrium' 49 ) and couples this with an equally broad test for modification from Continuity Bill. 50 The cumulative effect is a potent cocktail for limiting the Scottish Parliament's legislative freedom. 51 In fact, that this provision's equivalent might provide an additional limitation to the competence of the Northern Ireland Assembly has since been rejected in that context. 52 The Court's view of section 101 18. By contrast to its broad interpretation of s 28(7) and the severe imposition of limitations on the Scottish Parliament that results, the Court takes a distinctly narrow reading of s 101 of the Scotland Act. Section 101 essentially nudges the courts to prefer interpretations of devolved Bills and Acts that are favourable for the devolved institutions. It provides that, where 'a provision' 53 'could be read in such a way as to be outside competence', it 'is to be read as narrowly as is required for it to be within competence, if such a reading is possible'. As the explanatory notes make clear, this provision 'is intended to ensure that the courts will not invalidate such legislation merely because it could be read in such as a way as to make it outside competence' and, even though '[a]rguably, it does no more than replicate' the principle of efficacy, 54 it nonetheless gives further credence to the respect that is owed to the devolved legislatures which the courts have demonstrated in other contexts.
19. When construing 'possible', the Court was directed to an analogous provision in the HRA. 55 However, despite the similarities between these provisions, and despite them even arguably forming part of the same reform programme, 56 the Supreme Court was eager to distinguish the two. The Court held that s 101 is far weaker than s 3 HRA: 'Section 101 of the Scotland Act', said the court, is 'fundamentally different from section 3 of the Human Rights Act'. 57  Ireland, even in relation to devolved matters, is therefore preserved (although that Parliament will normally only do so after a legislative consent motion has been sought from, and passed by, the Assembly). However, within its sphere of competence, the Assembly is entitled to pass laws modifying any provision made by an Act of Parliament in so far as it is part of the law of Northern Ireland. By section 98(1), 'modifying' is defined, in relation to an enactment, to include amendment or repeal. Thus, provided the Assembly is not acting beyond its competence as defined by sections 6-8 of the NIA, it may repeal any provision made by an Act of the Parliament of the United Kingdom as a matter of the law of Northern Ireland.' He also held that '[t]here is no limitation on the Assembly's power to legislate for transferred matters, other than those relating to legislative competence more generally': ibid [48]. 53 The meaning of which is given by s 101(1). 54 Explanatory Notes to the Scotland Act 1998, s 101. 55 Section 3(1) of Human Rights Act 1998 provides that '[s]o far as it is possible to do so, primary legislation and subordinate legislation must be read and given effect in a way which is compatible with the Convention rights'. 56 That the courts should take a different interpretation of the same words in different statutory contexts is justifiable, but they have given the same effect to, for example, the words 'relates to' across the devolution settlement because of their systemic similarities, despite that actually giving rise to arguably problematic consequences: see Tomkins (n 41). 57 In re United Nations Convention on the Rights of the Child (n 8) [ awkward in a case where the question is whether a provision … is incompatible with Convention rights'. 59 Despite the DS dictum therefore suggesting that interpretive tools should be selected because they are least likely to result in a finding that a provision is outwith competence, the Court concludes that this authority weakens the s 101 obligation.
20. The Court also draws on authority that in practice s 101, and its analogues in the other devolution legislation, 'have not been given as far-reaching an effect as section 3 of the Human Rights Act'. 60 The Court uses as its contrast 'Lord Neuberger's statement in the Welsh Byelaws case that "[i]t would not be permissible to invoke [s 154 of the Government of Wales Act] if it was inconsistent with the plain words of [the provision in question]" gives section 154 a more restricted scope than section 3 of the Human Rights Act, as interpreted in Ghaidan v Godin-Mendoza'. 61 However, that s 3 HRA is capable of more in extreme cases does not mean that s 101 is not capable of matching its potency in ordinary, less demanding ones. The test, if that is right, is whether the Court is being asked to construe provisions inconsistently with their plain words.
21. Perhaps most telling, however, is the Court's view of the relationship between s 101 and the pre-enactment safeguards, as outlined above. In its analysis the Court reveals clearly how it is balancing the competing interests in the Scotland Act. Shortly, the Court said that to allow the Scottish Parliament to intentionally place heavy reliance on s 101 would have 'the effect of rendering nugatory the pre-enactment safeguards provided by the Scotland Act'. 62 This is an interesting dictum for two reasons. First, it represents the first time in the judgment that the wider statutory context informs the Court's analysis of a particular provision in the Scotland Act. This stands in contrast, for example, to the Court's treatment of s 28(7). In interpreting that provision broadly, the Court overlooks the risks that such an interpretation might similarly emasculate the remainder of s 28. Selectively restraining provisions by varying reference to their context is one of the ways that this reasoning demonstrates its weight towards constraint over empowerment.
22. The second reason the prioritisation of the pre-enactment safeguards is intriguing is because this conception of the relationship between s 101 and those safeguards is by no means obvious. If either s 101 or the pre-enactment safeguards must give way, it is not clear why it is s 101 which should yield, as it does in the Court's judgment. In fact, ascribing such significance to these safeguards, many of which are political, stands in contrast to the courts' treatment of similar pre-enactment safeguards in the HRA whichwhen combined with the courts' powers under that regimeare plural precisely because they are ripe for disagreement and are in any case of little legal significance. Indeed, the final determinative statement onin the HRA case compatibility, and in this case competenceis the Supreme Court's judgment on the issue. It is not extraordinary to suggest that it should be completely free to express its view, rather than being bound by political declarations from other forums. The Court's prioritisation of the safeguards over s 101 again demonstrates its preference for limitationjustified by legal certaintyover legislative freedom.
case got that balance right remains to be seen, but its decision seems likely to cast a long shadow indeed.