From a Europe of Bits and Pieces to a Union of Variegated Differentiation

Thirty years after the Europe of bits and pieces of the Maastricht Treaty, the EU legal system has evolved beyond fragmentation to accommodate institutionally structured forms of differentiation. This paper explores several types of new differentiation regimes and argues that they can coexist together without necessarily challenging the unity of the EU legal system. It analyses how the legal system has progressively been adapting to new integration pathways by internalising differentiation and reabsorbing the fragmentation of the Maastricht’s construction. Through the analysis of the Court’s jurisprudence and two case-studies in the areas of economic governance and defence it shows how different strands of differentiation can be blended together in ‘coherent’ differentiated regimes. The paper also considers future differentiation pathways after Brexit and emerging concerns as regards legitimacy and democratic accountability of a differentiated Union.


Introduction *
In the aftermath of the signing of the Maastricht Treaty, I qualified the constitutional structure of the European Union as 'a Europe of bits and pieces.' 1 This qualification implies a challenge to a previously existing unity, both legal and institutional. It represented a legal and institutional analysis that offered a possible -and perhaps at that time obvious-reading of the Treaty of Maastricht (1992). A Europe of 'bits and pieces' implies in its most negative and radical form a broken Europe, a Europe disintegrating. 2 At the very least, it points to deep institutional, legal and political fragmentation. The term fragmentation carries two senses, but not both are negative. First, fragmentation may be used to describe how certain elements areor become -disjointed. Fragmentation can be defined as "the process or state of breaking or being broken into fragments", fragments being themselves defined as a detached element of something: "a small part broken off or separated from something". 3 Second, fragmentation has also a more positive and dynamic meaning. Fragmentation creates not only the danger of "conflicting and incompatible rules, principles, rule-systems and institutional practices" 4 but refers to the positive manner in which law expands and diversifies in terms both of its objects and its techniques. 5 In this sense it is very much about processes of seeking and even redefining relationships among the various seemingly disjointed parts.
After the Treaty of Maastricht, the previously rather unitary legal structure appeared under challenge in novel areas. 6 The argument was not that Europe was disintegrating, as clearly it was not. It was rather messily moving forward on selected key areas closely linked to national sovereignty (economic and monetary union, foreign and security policy, justice and home affairs). A European Union did not -and does not-imply a Europe broken or falling apart. What it did imply was a sense of (more) legal and political fragmentation than was previously the case pre-Maastricht within the axis around the European Economic Community. Fragmentation pertained not only to the structurally seemingly separate intergovernmental 'pillars' alongside the previously existing supranational core but also the inclusion at the Treaty level of an endless array of opt outs for individual Member States on matters of smaller and greater significance. 7 * To be published in: Craig and De Burca, Evolution of EU Law, 3rd Edition (OUP, 2021). This piece is dedicated to my co-author in the two previous editions of this book, Ige Dekker, in gratitude for our stimulating collaboration over the years and for helpful exchanges on this Chapter. I would also like to thank Bruno de Witte for commenting on this and an earlier version and Michal Krawewski, Maria Patrin and Tomas Dumbrovsky for excellent research assistance on this and earlier versions.

2
Robert Schuman Centre for Advanced Studies Working Papers As a metaphor for continued constitutional 'chaos' 8 the contemporary fit of 'bits and pieces' is obviously limited. Beyond the level of formal Treaty provisions, legal and institutional practices rather quickly revealed sustained patterns of 'unity' despite formal fragmentation, albeit 'in disguise '. 9 Later the patterns of unity and the connections emerged ever more 'out of the shadows'. 10 As Ige Dekker and I showed in our two contributions to earlier editions of this book on the evolving EU, European law and European institutions developed in practice myriad new connections and new relationships. 11 The cracks were still visible but quite some original bits and pieces joined up together. Our conclusion ten years ago (2010) bears repeating and refers to the evolution since the first edition (1999): " Our conclusion ten years ago was that the European Union, in general terms, has evolved, as an international organisation, into a legal system with a unitary character overarching a lot ofand sometimes very different -'layers' of cooperation and integration. This unitary character was for some time hidden behind a complex pillar structure and could only be brought into the light on the basis of an analysis of the treaty implementation and legal practices. Ever since, the legal unity of the Union has clearly come out of its shadows and has gained a far more solid and durable basis in its founding Treaties.
These Treaty-level developments were, to a large extent, the result of the ever-increasing legal practices in the field of the common foreign and security policy and the cooperation in the field of justice and home affairs, later reduced to police and judicial cooperation. By these legal practices not only new legal regimes were developed but they also largely contributed to the reforming of the European Union of a mainly socio-economic organisation into a general one, more clearly dominated by political overtones.
Although it is clear that the influence of the legal practices in the second and third pillar can hardly be underestimated it is equally clear that the evolution of the Union has not, as was feared at the time it was established, in general undermined the separate, 'supranational' status of the legal system of the entity formerly known as the European Community. As we expected ten years ago, the Union has proven to be capable of integrating in its overarching legal structure sub-systems containing in certain respects more far reaching principles and rules. In the context of that panorama it is not so surprising that Union law also harbors numerous possibilities for institutional variation but that that fact does not deprive the legal instruments of their character of belonging to a Union legal and political system. The European Union can still be categorized as a highly complex entity with diverse fragments but this complex structure of fragmentsmore reminiscent now of the capricious design of marble than of the old-fashioned and neatly delineated Russian doll of a decade ago -exists within an overall institutional structure and legal and political systems." Ten years after these words were written, far from imploding, the system has held tight. This is not to say that it has not required both expansion and retraction as well as innovative marbling that can and does vary across policy areas. Institutionally structured forms of "differentiation" is a concept better suited to reflect the current institutional and legal reality of an evolving Union. Differentiation reflects an attempt to remain together and seek new relationships despite the centrifugal forces of disintegration that threaten the EU. Different from fragmentation, the notion of differentiation points to a unitary legal order, which is however apt to more systematically accommodate diversity. Differentiation thus internalises the concept of fragmentation and makes it part of its own legal order. As a result, the EU legal system had to adapt to the new paces and forms of integration (or dis-integration), while maintaining its unity. This was no easy task, as it meant overcoming the deeply rooted credo of the "ever closer Union" and the CJEU's mantra of the unity, coherence and indivisibility of the Community legal 8 P. Allott, "Written Evidence" to Select Committee in the European Communities, House of Lords, Political Union: Lawmaking Powers and Procedures, Session 1990-91, 17th Report, 35-38. 9 D. Curtin and I.Dekker, 'The EU as a 'Layered' International Organization :Institutional Unity in Disguise' in P. Craig and G.de Burca (eds.), The Evolution of EU Law (Oxford, OUP, 1999), 83-137. 10 D.Curtin and I. Dekker, "The European Union from Maastricht to Lisbon. Institutional and Legal Unity out of the Shadows" in P. Craig and G.deBurca (eds.) The Evolution of EU Law (Oxford, OUP, 2010). 11 Ibid.

From a Europe of Bits and Pieces to a Union of Variegated Differentiation
European University Institute 3 order. Indeed for a very long time, differentiation was considered as a threat to European integration rather than a form of integration in its own right. Only recently, in the aftermath of a turbulent time for Europe (euro crisis, refugee crisis, Brexit), differentiation has become not only an acceptable compromise, but even the preferred solution to potentially disaggregating forces and to the stalling of the EU integration process.
Arguably, the acceptance of a differentiated future for the EU has implied a change in the concept of integration itself, towards a more composite and fluid understanding of European integration as opposed to a traditional unitary model. Political scientists have pointed in this direction. For instance Frank Shimmelfenning has extensively argued that European integration can only be understood through the prism of differentiation. Richard Bellamy and Sandra Kroeger have investigated under which circumstances institutional designs of differentiated integration can be considered as fair and therefore acceptable 12 . In addition, concrete proposals on how to implement differentiation in policy-making have blossomed recently. The most discussed is certainly the European Commission's White Paper on the Future of Europe, featuring, as arguably one of the preferred scenario for moving forward, a "model of differentiation allowing willing member states to do more together in specific areas" (Scenario 3: Those who want more do more) 13 . Even more radical proposals have also come from think-tanks and academia. A 2018 Bruegel paper written by Demertzis, Pisany-Ferry, Sapir, Wieser and Wolff proposed a hybrid model of differentiated integration combining a bare-bones EU (mainly based on economic integration) in which all EU members participate with a set of differently integrated clubs in specific policy areas (e.g.EMU, Migration and Schengen,security and foreign policy etc.) 14 . Although the feasibility of these projects is controversial, they show that the scientific and political debate on European integration has by now accepted differentiation as an unavoidableand sometimes welcomed-destiny.
As the integration dynamics have changed, so too has the EU legal system. Differentiation has become a stable element of the EU legal system 15 . It has progressed partly as a feature of the EU legal order, by incorporating differentiation mechanisms and opt-outs in the Treaties. Sectoral opt-outs and enhanced cooperation clauses are typical example of this Treaty-based differentiation. But it has also featured outside the EU legal order in the form of international law instruments (inter-se treaties) 16 . This typicallybut not exclusivelyhappens in crisis situations. Organizations facing crisis may spawn new institutional arrangements including in the case of the EU the pursuit of EU law objectives outside the framework of EU law. 17 The response to the financial crisis for example includes new EU international agreements and the structuring of the new banking union partially outise of the EU legal framework. As I will be argue in this paper, these different forms of legal differentiation, however, are not contradictory nor opposed to each other but they are closely connected. Often, they constitute two components of the same differentiation objectives.
Overall, this chapter offers an initial account of different strands of new differentiations that are emerging and though still, in a certain sense, forming 'bits and pieces' they can be understood as part of a larger and cohering framework. which incorporates varying levels and forms of internal differentiation, and, potentially more problematically, also includes provisions that, although they are technically not EU law, are so connected and interlinked that they almost form an annexed component of that same system. The structure of this paper is as follows. I first discuss the evolution which has led from a fragmented to the compositions of a differentiated European Union, thereby discussing the different types and forms of differentiation. I then turn to analyse how the EU legal system has evolved to adapt to the increasing differentiation through the jurisprudence of the CJEU. As a third step I propose to look at two exemplary case-studies, economic governance and defence, to show how the different forms of differentiation are blended together to achieve the overarching objective to provide for sustainable and 'coherent' differentiated regimes. I conclude with some overarching considerations as regards the future differentiation pathways after Brexit and on emerging concerns as regards legitimacy and democratic accountability of an increasingly differentiated Union.

Hybrid diversity through the decades
The ongoing debate about 'differentiated integration" is not a new debate. Under the heading of "multispeed", "variable geometries" or "core Europe", it raises its head periodically when a group of Member States wish to proceed further and deeper in a given policy area. In a nutshell, differentiated integration indicates a model of integration whereby some Member States proceed towards closer integration without the participation of others 18 . Among legal scholars differentiation was not originally perceived as a threat to European integration but rather as a tool to promote further integration. The idea was that a group of Member States would move ahead with enhanced cooperation while leaving the door open for the non-participating Member States to join later (for example the Schengen and Prüm Conventions).
The first explicit suggestion for differentiated integration was put forward by Willy Brandt in November 1974 in the aftermath of the oil crisis and when the Luxembourg compromise were taking its due on the decision-making process. 19 He called for a graduated integration. 20 His argument was that given the levels of economic diversity among the Member States at that time, the objectively stronger countries could lead the way in certain areas and would have a centripetal effect in pulling the weaker countries along into the core group. In the early seventies discussion of a 'two-speed' Community was prompted by the simultaneous discussion of Economic and Monetary Union (EMU) and the effort of digesting enlargement which included the UK, alongside Ireland and Denmark. The first major differentiation came in 1979 with the establishment of the European Monetary System (EMS) setting limits for fluctuation of exchange rates. The EMS was not entirely a Community initiative, nor entirely outside it. 21 Only EC Member States were allowed to participate but they were not obliged to do so. The UK in fact did not join the system.  Paris, Nov. 19, 1974, 30 EA D33 (1975.

From a Europe of Bits and Pieces to a Union of Variegated Differentiation
European University Institute 5 The Dooge Committee, preparing the Single European Act, estimated in 1985 that differentiation would enhance both the decision-making and negotiation of the Single Act. 23 At this time, major early studies on differentiated integration were published (Langeheine 1983, 24 Ehlermann 1984, 25 Grabitz 1984, 26 and H. Wallace 1985 ). Already in 1984, Ehlermann cited in his article all major differentiation concepts: two speed or two-tier Community, Europe à géométrie variable, Europe à la carte, abgestufte Integration and admitted that no generally accepted definition existed. 28 Neither the policy-makers nor the scholarship seems to move significantly further on this matter. 29 The competition was effectively left open between advocates of closer integration and those who wanted looser intergovernmental cooperation. This pattern of living apart together is embedded in the fabric of the EU as we know it today.
Various forms of differentiation have in fact become a stable feature of the EU legal system and have resulted in a "regime of legally constrained differentiation" enshrined in the Treaties. 30 The very notion that European integration progresses at varying speeds is based on the assumption that a group of Member States advances towards a common policy objective, and is later joined by those States which are unwilling or unable to do so from the start. The non-participation of some Member States in the integration of a given policy area is regarded as inherently temporary with catch-up built in. This is visible in the use of transitional periods and safeguard clauses granted to those states in order to prepare for the full implementation of the instruments for the integration of that sector. 31 'Variable geometry', by contrast, suggests that the differing degrees of integration of the Member States may be formalised in permanent terms by using the instrument of the opt-out. Modest, more ad hoc or bespoke differentiation has been facilitated almost from the very beginning of the Community as a way of taking account of immediate stresses and strains. The Treaty of Maastricht in the early 1990's made opt-outs more structural and embedded. This was built on further in later revision Treaties. Some domains of EU law know numerous opt-outs. Some of the more structural cases have been the UK's opt-out from the Charter of Fundamental Rights, or Denmark's opt-out from the Area of Freedom, Security and Justice. 32 Langeheine, "Legal Problems Related to a Proposed 'Two-Tier System' of Integration within the European Community", Opt-outs refer to the unwillingness or inability of a Member State to participate in an EU policy where it is assumed that all Member States will join. Yet differentiation can also be seen where the opposite occurswhere the Member States do not advance together to integrate a new policy, with only a few choosing to do so. Enhanced cooperation is a means to overcome disagreement of the few when a clear majority have joined the cooperation. Both the last resort principle (Art. 20 (2) TEU) and the principle of openness to other Member States (Art. 328 (1) and 330 TFEU) reveal the departure point of ideally as many as possible to participate. As of 2020, there are four authorised enhanced cooperations with each Member State taking part in at least one them. This shows a relatively high acceptance of the differentiation mechanism as such and puts the idea of a homogenous core group or "first class" of Member States into question. 34 Finally, the Lisbon Treaties assign a specific meaning to differentiation in targeted policy areas, detailing out the legal forms that enhanced cooperation in these fields should take. They namely provide for two special regimes of differentiation for economic governance and for defence. Art. 136(1) TEU allows for the adoption of euro-area specific legislation on economic governance. The provisions recall the rules on enhanced cooperation, except that no authorization is required by the Council acting in its normal composition. This is because euro differentiation is open only to the eurozone members and requires the participation of all of them. Under this perspective, the eurozone is the closest form of a 'core Europe', albeit open to all Member States who might in the future adopt the euro currency and join EMU 35 . Similarly, Art. 42 (6) TEU and the Protocol 10 to the Lisbon Treaties allows Member States to enter in permanent structured cooperations (PESCO) in the field of defence. This regime is again similar to enhanced cooperation, yet it is based on codified and objective criteria linked to the military capabilities of the Member States wishing to participate 36 . Art. 42(6) TEU states that: "Those Member States whose military capabilities fulfil higher criteria and which have made more binding commitments to one another in this area with a view to the most demanding missions [to] establish permanent structured cooperation within the Union framework". 37 In addition, the decision-making process of PESCO remains fully intergovernmental and reflects the special status enjoyed by Common Security and Defence Policy (CSDP) in the Treaties as a special area of competence of a distinct legal nature 38 .
In sum, Treaty revisions have progressively refined and fleshed out the different options for differentiated integration in the EU. In doing so, they have attempted to control and legalise the "disintegration potential" of fragmentation. By allowing differentiation within the EU framework they have also admittedly sought to limit differentiation outside of the Treaties, albeit with disputable results. The use of international legal instruments for purposes of differentiation has always been part of the DNA of the EU and has often led to important steps forward in the integration process. One clear example is the Schengen regime, which, born as an international convention between some EU and non-EU countries, was later incorporated into the Union's legal order (Amsterdam Treaty). Strengthened Treaty provisions on enhanced cooperation have in fact not prevented differentiation to take place outside of the Union's legal framework in recent times too. The legal regime adopted to face the financial and economic crisis is made of a mixture of internal and external legal tools. So called inter-se 33 Ibid 185. agreements form a big chunk of it. In addition, not only crisis-related measures were taken outside of the EU legal framework, as showed by the Agreement on a Unified Patent Court which was eventually adopted under international law 39 . Do these develoments signify a return to fragmentation, to a 'Europe of bits and pieces', at least as far as its legal order is concerned? If one looks at the messy mushrooming of legal instruments, regulations and agreements there indeed seems to reign considerable confusion on where the system is going and how it is developing. Looking more carefully, however, it is striking that all these instruments, internal and external, are compatible and are kept together by the reference to EU law. The EU legal system acts as a type of magnetic sphere attracting all the satellite agreements under a unitary legal regime 40 . As Dekker and I put it in our first contribution: "the European Union can still be categorized as a highly complex entity with diverse (eclectic) fragments but these fragments exist within an overall institutional structure and legal system 41 . This is clearly the case for EMU, as it will be showed in more detail later, where the "external side" of EMU is in fact embedded in the EU institutional setting and is inextricably connected to internal EU law. Also attempts to "repatriate" external agreements into the EU legal system are frequent, admittedly with mixed results. Surely it can be argued that the system as it has been developed is somewhat dysfunctional. However it is typical of the fluid character of the integration process, that experiments different legal instruments and means to find pragmatic solutions 42 . Unsurprisingly, these solutions might appear as unorthodox and not always consistent, especially when adopted under the impetus of crisis situations. Ultimately, this is the way in which the legal order has developed to redefine and reaccomodate relationships between member states and institutions in new, and often challenging, environments.

Judicial uniformity
The Court of Justice of the European Union (CJEU), often assisted by the legal services of the Commission, has traditionally insisted on maintaining that the Treaties and the legal and institutional order created by them forms an indivisible whole. The European Court of Justice historically acts as a custodian of the uniformity of EU law. The purpose of uniformity was indeed the reason why the Court a single judicial body having a monopoly over the supply of authoritative interpretations of EU law 43 and the preliminary reference mechanism were created to begin with. Besides Article 19 (1)  to be closely connected with the equality of EU citizens. The preliminary reference mechanism guarantees that citizens across the EU enjoy equal protection under EU law. 47 However, what is striking if one considers the early cases, some of which are landmark cases of EU constitutional law, is that the disunity of the EU (EEC) legal order is perceived as an existential threat for the EU(EEC) itself. The Court pledged early on to defend the "unity" of the common market and the "uniform application" of Community law against efforts to achieve community policies outside the treaty frameworks. 48 As Barents explains, the very Community character of Community law determined the indivisibility and therefore invariability of the EEC's power. 49 In the Court's vision, if it would "vary from one Member State to the other", then the very "authority of Community law" would be compromised. 50 In its famous ruling in Hauer, which concerned the relation between EU law and national constitutional provisions on fundamental rights, the Court's language is permeated by a sense that fragmentation leads to inescapable catastrophe: 51 "the introduction of special criteria for assessment stemming from the legislation or constitutional law of a particular member state would, by damaging the substantive unity and efficacy of community law, lead inevitably to the destruction of the unity of the common market and the jeopardizing of the cohesion of the community". 52 This same concern with preserving legal unity as a matter of constitutional law is visible to this day. In Opinion 1/91, on the creation of the European Economic Area, the Court used the plural in referring to the various Treaties instituting the different Communities, while speaking of the 'Community' legal order in the singular. 53 Even in relatively recent case law, the Court restates that the protection of fundamental rights enshrined in the Charter may not compromise the "primacy, unity and effectiveness of EU law". 54 In Opinion 2/13, the same argument reappeared as one of the reasons why the Court turned down the EU's accession to the ECHRthat loss of unity was (allegedly) foreshadowed by the risk of fragmentation of the ECJ's exclusive competences. 55 Beyond the area of human rights, the same concern about the unity of the EU legal order is visible in Opinion 1/09 on a proposed agreement on a unified patent litigation system, including an international court. 56 Faced with a relatively new threat of the evergrowing range of international tribunals interpreting EU law, the Court has visibly sharpened its approach. 57 We have seen most recently in Achmea 58 that the Court will not hesitate from making bold decisionssuch as to dismantle the State-investor arbitration system across the EU Member Statesdespite considerable consequences for national policies -to preserve the "autonomy" of Union law, understood in terms of the fullness of its own jurisdiction over EU law. 59 47 K. Lenearts, "New Horizons for the Rule of Law Within the EU", (2020)  Moreover, in Pupino, 60 the Court held that the principle of loyal cooperation applied transversally across the various pillars. This contributed to depillarize and somewhat mend the fragmentation emerging from Maastricht's intergovernmental elements.
Despite the fierce protection of the unity and autonomy of EU law, the recent case-law of the CJEU also points to a slightly shifting stance as regards the uniformity of the content of the EU legal order. The concerns for coherence remain at the core of the CJEU jurisprudence but leave space open for the integration of differentiation. Arguably the Court safeguards the unity of the legal order by incorporating differentiated integration and submitting it under its control and jurisdiction rather than simply defying it. This evolution clearly emerges from recent jurisprudence on internal and international differentiation.
In Spain and Italy v. Council, the Court reviewed the legality of the authorization to start enhanced cooperation in the field of unitary patent protection. It provided a rather flexible interpretation of enhanced cooperation, as an instrument to overcome vetos in the Council 61 . According to the Court, when unanimity applies enhanced cooperation can namely allow to proceed further on a piece of legislation even in the presence of some Member Sates' disagreement on the substance of legislation. The judgment opens the way to an easier recourse to enhanced cooperation. More strikingly, the Court has also given its (conditional) approval to forms of differentiated cooperation outside of the EU Treaty framework, even when this happens in policy areas that are closely connected to EU law. In the Pringle case the Court was called to rule on the legality of the Treaty establishing the European Stability Mechanism (ESM Treaty), adopted by the countries of the eurozone as a crisis prevention-mechanism in 2012. The Treaty took the form of an inter-sé agreement external to the EU legal system. The Court ruled in favour of the compatibility of the ESM with EU law, however it established some important limitations to the externalisation of differentiated cooperation among Union members, including the respect of exclusive Union's competences (or shared competences where the Union has already acted); the duty to comply with EU law; and the respect of the principle of loyal cooperation 62 . The Pringle judgment thus provides some legal clarity and guidance in a much discussed and controversial field, namely the relation between international legal instruments and the EU legal order. It does so by accepting the fact that Member States may resort to international law to regulate policy areas which, although not part of the EU competences, are strictly related to them. In particular the Court defends the legality of the ESM Treaty arguing that it pertains to the area of economic policy and not of monetary policy, which would have been an exclusive EU competence 63 . In sum, the Court accepts the compatibility of the ESM with EU law, but at the same time keeps tight control on the conditions under which recourse to international law should take place. It is only allowed in an area close to EU policies when it is solidly anchored to the EU legal system and it does not threaten its unity. Inter-se agreements are satellite agreements which are de-facto embedded in EU law and form a "semi-connected part" of the Union's legal system 64 . This emerges for instance if one looks at the use of institutions made in the framework of the ESM, where the European Commission and the European Central Bank are effectively "borrowed" to perform some tasks under an international Treaty signed by only 17 Member States. This use too was endorsed by the CJEU in Pringle.
In both cases the Court's stance towards instruments of differentiation points to some flexibility in accommodating partial deviation from uniformity, which are ultimately necessary to safeguard the overarching unity of the legal system. As pointed out by Bruno De Witte, the increase in differentiated integration did not formally challenge the existence of one single EU legal order but rather resulted in the fact that "that EU legal order is no longer uniform… The Court's old ideal of EU legal rules being 'fully applicable at the same time and with identical effects over the whole territory of the Community' has become unattainable" 65 . In the absence of this uniformity, the Union's legal order has itself changed to accommodate differentiation.
In sum, the case law of the Court of Justice suggests a tension between the very "Community logic" of the EU and the practice of differentiated integration. The fact is that many of the EU's policy areas are fractured along national borders. The Court's jurisprudence has thus evolved to reflect this development, integrating several elements of differentiation in its system or even allowing for a connected use of differentiation through international legal agreements, which however remains solidly embedded in and "controlled" by EU law. In addition, the jurisdiction of the Courta single apex judicial body with the competence to interpret and apply EU law in all its segmentshas hitherto functioned as a constitutional umbrella that mitigates the risks of differentiated integration. Thanks to its transversal jurisdiction, the Court is able to secure the uniform application of EU constitutional concepts such as effective judicial protection and other individual fundamental rights, including more general principles such as the principle of institutional balance and of proportionality. As follows from the Opinions 1/91 and 2/13, the indivisibility of the Court's jurisdiction over the interpretation and application of EU law in its various segments may appear as the main constraint upon differentiated integration. 66

Introduction
As pointed out above, the future of the EU is being shaped around an increased use of opt-outs, enhanced cooperation and inter-se agreements. An argument can be made that what is distinctive about differentiation in recent years is how the 'old' forms of fragmentation are combined and overlap. Differentiation has resulted in an increasing complex and multilayerd system, where several forms of fragmentation are brought together to allow for diversified integration paces. In most policy areas, differentiation has thus advanced in parallel pathways, internal and external, blending opt-outs, regimes of enhanced cooperation and, sometimes, international law instruments. The pace and interconnectness of this progress is determined between the pragmatic search for a solution that it typical of differentiated integration and the flexibility of the EU legal system to accommodate heterogeneity and differentiation. How can the single EU institutional framework cope with this ever-growing and multi-layered differentiation trend? The contemporary debates on further integration in two fields in particular -EMU and defence-throw up old and new questions and seem to blend existing types of differentiation in varied and at times structural compositions.
In several regards EMU and Defence represent two privileged areas for testing differentiation, and arguably for opposing reasons. In the former case, the distinction between eurozone and non eurozone countries is so embedded in the European Union's structure that differentiation in this field has almost become an irreversible fact. Furthermore supranational integration has advanced quicker than in other areas. If there is one lesson to be learnt from the economic and financial crisis it is that euro-area members share a common destiny. Indeed the only viable option out of the crisis was to further strengthen integration and extend it to economic and financial policy coordination. On the other hand, defence and security represents one of the least integrated areas of EU cooperation, and maintains a solid intergovernmental character. It is also an "exceptional" area of core state power, which is subject to sovereignty and national identity concerns. According to Rittberger, Leuffen and Schimmelfenning, deeper integration in such areas of sensitive core state powers is particularly likely to develop forms of differentiation 67 .
Interestingly, defence and EMU represent the only two special regimes of enhanced cooperation which are enshrined in the Treaties. Also, enhanced cooperation is based primarily on objective criteria (the convergence criteria on the one hand and criteria on military capability on the other), while it also factors in the will of the Member States to participate, openly in the case of defence, but de-facto also in EMU, where several Member States have opted out or have deliberately decided not to respect the euro criteria. Finally, both areas blend different forms of internal differentiation (multi-speed, variable geometries) with forms of external cooperation and make innovative use of the EU institutional framewok. They will now be explored further in two consecutive paragraphs.

Differentiated integration in EMU
Within the EMU differentiation primarily occurrs in terms of membership. In the early days the opt outs negotiated by the UK and by Denmark in the Treaty of Maastricht were voluntary and represented a political choice in the perceived national interest. They had nothing to do with social and economic factors for non-participation. The individual opt-outs negotiated were permanent in nature. It was not multiple speed which implies a catch up at some stage. But rather a type of what has been called 'variable geometry'. The transition to the third stage of the EMU is on the other hand an example of multiple speed integration. This does not impair the functioning of the internal market as such. Rather it enables a peleton group of Member States to proceed with further integration according to objectively defined criteria. The intention is that the other Member States will, with the ripeness of time, join the leading Member States in the path ahead they forged. The possibility of this leading group was explicitly accepted by all the Member States in the TEU, including the UK and Denmark, even if they obtained an opt-out for themselves. For this reason a single institutional framework was maintained.
Overall, there seems to be a mismatch between the idea of multi-speed differentiation enshrined in the Treaties, whereby adopting the euro is described as a common objective that must be shared and achieved by all Union's members -only at different paces, and the de-facto development of EMU in practice. It seems indeed that the EU may rather move towards a form of semi-permanent multi speed or even 'core' Eurozone Europe. This has been and continues to be an incremental process. A new conceptualization of multi-speed might indeed rest in its "semi-permanent" character, that is understood as a dynamic relationship. Some Member States would move to, say level 4, others would move to level 3, meanwhile some of the former group would move to level 5, etc.
This trend towards the creation of a stable eurozone avantgarde was strengthened by the EU response to the euro-crisis, which took an array of different forms, further complicating the relationship between eurozone and non-eurozone members. Indeed the economic and financial crisis urgently demanded solutions and powers extension targeted to only some Member States 68 . As a first step, ample use was made of the special cooperation regime under Art. 136 (1) TFEU, with the adoption of the two-pack and six-pack legislation, strengthening budgetary and macro-economic surveillance (under the excessive deficit procedure and a new macro-economic imbalance procedure). Interestingly, some six-Pack operate when new powers are necessary for somein this case for the Eurozonebut when not all member statesnotably some outside the Eurozoneare willing to make this possible?" Robert Schuman Centre for Advanced Studies Working Papers provisions apply to the eurozone only (under Art. 136(1) TFEU), while some others to the EU as a whole (under Art. 121 TFEU on the multilateral surveillance procedure of the Union) 69 . In addition, however, because of the opposition of the UK in particular, Eurogroup members had to move outside the Treaties to negotiate international agreements external to the EU. In 2012 the Fiscal Compact (Treaty on Stability, Coordination and Governance in the Economic and Monetary Union -TSCG) and the ESM Traty were adopted under international law. The Fiscal Compact aimed at increasing budgetary discipline, among else requiring Member States to enshrine into national law the "golden rule" of a structural deficit not exceeding 0.5% of the GDP 70 . The ESM Treaty created a permanent emergency fund to support eurozone members in distress 71 . Whereas the Fiscal Compact was signed by 25 Member States (with the exclusion of the UK and the Czech Republic), the ESM only applied to euro-area Member States.
Finally, another conundrum relates to the EU institutions. Until now the single institutional framework has been maintained, even when Eurogroup members negotiated international agreements external to the EU. This happened before with the Schengen Convention but with the variation that new institutions were set up and the acquis was later incorporated back fully into the EU (an example of legal fragmentation being repatriated). In the more recent ESM Treaties the link with the EU single institutional framework was explicitly maintained and, as pointed out above, it was endorsed by the CJEU in the Pringle judgment. 72 Since the Fiscal Compact incorporated some of the key points that were supposed to amend the EU treaties, it effectively amended EU primary law without the approval of all the Member States. In addition, they did so by using the same Union's institutional framework. It has been argued that Treaties inter se can thus constitute a new form of EU law. 73 This is considered highly problematic from an EU constitutional point of view. However it also shows that the EU legal system is resistant to disaggregation forces. Indeed, the intention was not to unbundle the new legal instruments from EU law. On the contrary, provisions were explicitly made for repatriation within the EU frameworksomething that was not the case in earlier variants of 'repatriation' such as with the Schengen Convention and with the pillars of the Maastricht Treaty. The Fiscal Compact in Article 16 explicitly provides that: "Within five years, at most, of the date of entry into force of this Treaty, on the basis of an assessment of the experience with its implementation, the necessary steps shall be taken, in accordance with the Treaty on the European Union and the Treaty on the Functioning of the European Union, with the aim of incorporating the substance of this Treaty into the legal framework of the European Union." This five-year deadline for the Fiscal Compact ended on January 1 st , 2018. Accordingly, the EU Commission published a proposal for its incorporation into EU law. 74 It can be argued that Art. 16 is not an obligation under EU law and does not fall within the jurisdiction of the CJEU, which according to  The two principal options are incorporation into EU primary law through a treaty amendment and incorporation via a (normal) secondary legal act. Only 25 of the 28 Member States ratified the Fiscal Compact, not all of whom signed all parts of it. 81 The United Kingdom and the Czech Republic were in particular opposed to the adoption of the Fiscal Compact in the first place, and particularly in crisis-hit Member States such as Greece, implementing once more a legal instrument that is mainly incorporating austerity measures may not be supported.
One specific way of 'repatriating' the Fiscal Compact despite the resistance of certain member states is pursuant to the enhanced cooperation procedure foreseen in Art. 20 TEU and Art. 326-334 TFEU. Enhanced cooperation allows groups of member states to create secondary EU law not binding all member states. This way, member states interested in a legal measure can proceed with further integration steps, while unwilling states can stay out. This also means that vetoes of single member states or groups can be circumvented. However, the enhanced cooperation procedure is subject to a number of limiting conditions that have to be met in advance and has not been initiated even if some maintain that this procedure could be a "perfect fit" 82 in the future. This, however, would be hardly sustainable if non-members freely choose parts of EMU to join, yet not join other connected parts.
The Commission's approach is less thorough and more ad hoc. It published a proposal for a Council directive incorporating the Fiscal Compact. 83 The idea is that this directive would be passed by all Member States together, although its application would be limited to Eurozone Member States and all Member States who want to participate (i.e. the non-Euro TSCG parties), Art. 1(2) and 4 Draft Directive. In its roadmap paper, the Commission explicitly points out that "[t]he proposed Directive integrates the essence of Article 3 of the Treaty on Stability, Coordination and Governance, which forms part of the so-called Fiscal Compact." 84 Some dispute this on the basis that there are in fact substantial divergences from the TSCG in the new draft Directive and that the Commission's method of repatriation 75 D. Fromage, B.  85 Most of the Fiscal Compact obligations, particularly the structural deficit limit of 0,5% of the GDP is for example nowhere to be found. Instead, the draft provision refers to the provisions in Art. 1 of Protocol (No 12) on the Excessive Deficit Procedure, which contain the much less restrictive limits of 3 % of the GDP for the structural deficit and 60 % of the GPD for the overall government debt. Given the decreased popularity of austerity measures among the member states, the public pressure and the generally critical attitude of some member states towards the fiscal compact, in terms of legal unity and apparent coherence this would be a striking development if in fact adopted and implemented. To date there has been no follow up to the Commission's initiative and in the meantime a new legislature has begun its new term, leaving the fate of repatriation very uncertain. Moreover, a reform of the ESM is currently being undertaken by Member States under international law and is getting a new role with the Covid19 crisis. 86 With the Fiscal Compact effectively frozen, the EU legal order will continue to resist the centrifugal forces of disintegration, by closely connecting the international side of EMU to the core of EMU law. In this respect, indeed, inter-se agreements in EMU have in fact becomeas paradoxically as it may sound -a heterogenous extension of the EU legal order outside of the Treaty framework.
This blended mixture between EU internal and external law instruments seems to fit the differentiations needs of current EU policy-making. In the field of EMU, where it is widely used, it points to the "growing acceptance of such decisions being taken only among the Eurozone Member States, signalling the emerging acceptance of differentiation of integration within the EU as well as for breaking with "the principle of institutional unity"' 87 . As such it may well provide a new model for advancing differentiation in other policy areas. For instance, the package on Banking Union, adopted in 2014, reproduces similar patterns of interconnections between internal-EU and external-international law. A key pillar of the Banking Union is the Single Resolution Mechanism (SRM), which decides on the resolution of failing banks, making sure this happens in an orderly manner 88 . The Single Resolution Board (SRB)an EU Agencyis the central decision-making authority in the SRM and it manages the Single Resolution Fund (SRF), which was however created via an international law agreement 89 . Participation in the SRM is mandatory for euro-zone members but it is also open to non eurozone members 90 . As for the SRF, the Intergovernmental Agreement (IGA) was signed by 26 Member States, with the exception of Sweden and the UK. 91 . The system thus shows variable patterns of internal and international differentiation, deliberately composed.
Although extended recourse to this hybrid legal model can create confusion and is unlikely to become a general rule, it does provide a practical way to advance differentiation in selected policy-areas. It confirms the trend towards incorporating differentiation as a constitutiveand accepted-feature of the EU legal order, even when this partly happens outside of that same legal order. 85 This, of course, was also said about the TSCG itself: S. Peers, "The Stability Treaty: Permanent Austerity or Gesture

Institutional (dis)unity on the horizon
The idea of a more semi-permanent multi-speed EU with at its very core the Eurozone members takes some further shape precisely in this Brexit withdrawal phase. Macron's declared policy for the EMU calls for a "multi-speed Europe" and an "executive" for the Eurozone in the form of a prime minister or finance minister, who would be answerable to Eurozone subsection of the European Parliament. A shortterm idea is that the Eurogroup would get a full-time chairperson, who would present the strategy and evaluate the situation in the Euro area before the national parliaments. Regular practice of the Euro area summits at the level of Heads of State and Government would be restored, instead of the current tendency to bring them together only in the event of a crisis.
These changes do not require modifying the Treaties and indeed already are largely the practice. In the longer term, and in particular if a regional Euro budget is created, 92 more demanding and quite farreaching changes are envisaged in terms of differentiated EU executive power. The Juncker Commission has launched the reflection by proposing the creation of a European Minister of Economy and Finance in the framework of its reform package on EMU 93 . In the Commission's proposal the Minister would combine the positions of EMU Commissioner and permanent Chair of the Eurogroup. Interestingly, a new central executive power for the eurozone would be created at the core of the EU institutional architecture. The Finance Minister would indeed be the Minister for the Union, while at the same time effectively acting as the main executive of the Eurozone. How this would play out in terms of relationship and balance between eurozone and non-eurozone members is only one of the several questions left open by the Commission's proposal 94 . In the meantime a new Commission has taken office and no follow up has been given to the initiative. Yet, the debate on how to strengthen executive power in the field of economic governance and in a differentiated Union very much remains on the table even as Europe grapples with the consequences of the Corona virus public health and economic emergency.
It seems clear that a strong (differentiated) Eurozone executive power in any event needs to be balanced by a stronger democratic institution for the Eurozone. Until now the Commission's and even the European Parliament's proposals are fairly weak. The Commission proposes an 'agreement' on the democratic accountability for the euro area in time for the European Parliament elections June 2019. 95 The European Parliament itself, in its Verhofstadt report 96 , proposes distinct voting rights for Eurozone MEPs as well non-binding advisory votes for non-Eurozone MEP's. This is linked to the adoption of a budgetary capacity for the Eurozone which in the future could require taxation.
This concept of 'differentiated representation' within the European Parliament itself is based on the idea that Eurozone measures only produce effects within the countries of the Euro area. In legal terms it is true that only the States that are part of the EMU are bound to comply with tighter rules with specific rights and duties for their citizens. Doubts can be expressed however about the economic and the political effects of those measures in non-Eurozone countries. Eurocrisis measures after all produce 92 See German Grand Coalition Agreement between CDU, CSU, and SPD, p. 9. Available at https://www.spd.de/fileadmin/Dokumente/Koalitionsvertrag/Koalitionsvertrag_2018.pdf. 'spillover effects' also on countries outside the EMU. 97 The living conditions of European citizens outside the Euro area are to some extent affected by those measures and their institutions are used to take them into consideration even if there are no legal obligations in that regard. Moreover, the EMU itself has become quite differentiated with the introduction of the practice that non-Eurozone MSs may join some or even most of Eurocrisis measures. The Treaty on Stability, Coordination and Governance (TSCG) was ratified by all Member States except for the UK, the Czech Republic and Croatia. 98 In the area of banking supervision, the Single Supervisory Mechanism (SSM) allows "close cooperation" by non -Eurozone MS with the ECB as the new supervisor of the commercial banks' financial health. 99 So far, Bulgaria and Croatia are officially negotiating a enhanced cooperation with the SSM and the ECB. Moreover, as was noted just above, in the resolution pillar of the banking union there is both jurisdictionalas not all Member States participate -and legal fragmentation -some Member States join it outside the framework of EU law, via international agreements.
In implementing a system of differentiated representation within the EP, the level at which the rights of MEPs elected outside the Eurozone will be limited is not obvious. Are only voting rights to be constrained? Or should the right to take part in parliamentary debates about the Eurozone measures also be restricted to Eurozone MEPs'? This would limit freedom of expression of the other MEPs. The idea to limit the rights of members of parliament risks creating second-rate MEPs and implicitly recognizes second order European citizens represented in the European Parliament. This result challenges the very nature of a Parliament as a democratically elected Assembly of representatives protected by a free mandate. The 'trustee model' of representation, on which the concept of the free parliamentary mandate is based, runs in favour of a truly European mandate of MEPs, at least as an aspiration. Finally, it is unclear how differentiated rights for MEPs would be compatible with the introduction of transnational lists, whereby a share of MEPs would not be elected on a national contingent but on Europe-wide lists. Transnational lists, which are deemed to enhance the democratic legitimacy of European elections and therefore of the MEPs themsleves, would provide for a genuine European mandate, independent from the distinction between Eurozone and non Eurozone countries 100 .
A bolder version of differentiation would break radically with the existing single institutional framework for further Eurozone integration. A number of prominent French academics, including Thomas Piketty, have proposed the establishment of a new Eurozone institution. 101 This is to be called the Parliamentary Assembly of the Euro area. This would be brought about in a far-reaching, parallel, reform Treaty on the democratization of the governance of the euro area (T-Dem). The idea is that the new Assembly would have much more far-reaching powers than currently enjoyed by the European Parliament in this area. Thus, it would take part to the preparation of the Euro Group and of the Euro summit meetings; it would have legislative capacity; and it would have the final say on the vote of the Euro area budget. The new Assembly would be composed for 4/5ths by the representatives of national parliaments and for the remaining 1/5th by the representatives of the European Parliament. The free mandate is certainly inadequately addressed. Given the free mandate principle, who do the 4/5th and 1/5 th represent? A Member State, the European Parliament, the Eurozone citizens? Changes in the composition, voting or participation rights of MEPs requires an amendment of the existing European Treaties. All 27 Member States would, under the existing Treaties, have to agree that the Eurozone group would establish their own institutions alongside or even within the full EU institutions. This would constitute a form of permanent multi-speed policy at the core of the EU. Moreover, it would give rise to a form of EU law and of new EU institutions that only applies to less than the full cohort of Member States. Despite the democratic ambitions of its promoters 102 , some criticisms justifiably claim T-DEM "highjacks" existing bodies of the EU and transports them outside in a modified fashion 103 . Arguably, moreover, the reform would weaken the legitimacy basis of the European Parliament and enhance institutional confusion 104 .

Defence battalions gear up
The second area in which differentiation is taking placeand may expand in the futureunder varying forms and features is defence. Defence represents the field par excellence of intergovernmental cooperation, often happening aside of the Union's framework and in ad hoc compositions. The first reason why cooperation does not often use the Union's framework is obviousand is also enshrined in the Treaties: most of military cooperation goes through the NATO. However, also when it comes to actions which are exclusively European in character, Member States tend to prefer to establish bilater and multilateral cooperations outside of the EU institutional and legal framework. This was for instance the case of the military operations in Lybia in 2011 which was carried out by individual European countries led by the UK and France, fully independently from the EU 105 . Under these circumstances tracking international differentiation in the field of defence is a difficult task, and it is arguably not very useful, since in most of the cases one cannot truly speak of integration, but rather of ad hoc military operations.
In the area of defence a lot hinges on practical considerations: the defence capabilities of individual Member States, their willingness to deploy them and their ability to pay for them. The conundrum for the EU is complicated by Brexit. UK military strength lies in its ability to provide troops, hardware, and expertise. It cannot be matched by any other EU state. Leadership of Germany in military matters is politically problematic and France's resources are more limited. A more informal, flexible approach outside the EU treaty frameworks is therefore often the preferred way forward although it gives rise to legal fragmentation. Given longstanding lack of consensus on the creation of shared defence structures legal fragmentation may be the necessary first step -similar to what happened originally with regard to Schengen. Practice indicates that Member States maintain a political appetite to act in parallel to or even independently from EU initiatives. An approach dubbed "Schengen for Defence" is that like-minded countries would begin sharing military capabilities and resources on the basis of an ad hoc agreement. This would then be open for others to join and could later become a formalised part of the Treaties at a more advanced stage.
Be it as it may, what is interesting is the fact that this intergovernmental cooperation has run in paralleland has probably also acceleratedthe establishment of a cooperation framework for defence inside the EU system. This framework allows a maximum level of differentiation, to reflect the 102 Ibid 3. 103 S. Platon, Democratizing the Euro Area without the European Parliament: Benoît Hamon's "T-Dem", in Verfassungsblog, 13 March 2017, available at: https://verfassungsblog.de/democratizing-the-euro-area-without-the-european-parliamentbenoit-hamons-t-dem/ (last consulted 24/04/20). 104 De Witte (n 16) 246. Robert Schuman Centre for Advanced Studies Working Papers importance of national sovereignty in this area. It also provides for a highly proceduralised legal structure, that once again aims at internalising differentiation at the price of uniformity. It brings together opt-outs, special enhanced cooperation regimes and, lately, research and industrial development through the Union's budget.
As pointed out in section 2, the Treaty of Lisbon provides a legal basis for what is termed a 'Permanent Structured Cooperation' in Defense. 106 This anticipates differentiated integration from the startand in fact, Art. 1 (b) Protocol (No 10) on PESCO refers to the year 2010, showing that the Treaty envisaged the formation of a defence avant-garde almost from the beginning of its entry into force.
In institutional terms, PESCO provides a number of notable provisions compared to the general enhanced cooperation procedure. Most noteworthy is the virtual absence of all supranational institutions when compared to the normal enhanced cooperation procedure. The Treaty tries to keep Member States from cooperating outside the Treaties by imposing light conditions on them: PESCO is established by (only) qualified majority voting in the Council and no European Parliament involvement. The requirements are much stricter for establishing general "enhanced cooperation", which needs either authorization by the qualified majority in the Council and consent of the EP or even by unanimity in the Council if the cooperation falls within the common foreign and security policy. In addition, in PESCO it is explicitly provided that participating Member States can withdraw by means of unilateral notification (Article 46(5) TEU) and participation at a later stage is subject to a Council decision adopted by qualified majority voting. Moreover, the Council may adopt a decision by qualified majority voting suspending the participation of a Member State when it does not/no longer fulfil the criteria or meet the commitments. This decision is taken by the Council, with only participating Member States, by qualified majority (Article 46 (4) TEU). Enhanced cooperation, on the other hand, is governed by the principle of openness enshrined in Art. 328 (1) and 330 TFEU. According to this, the Member States must promote the participation of as many Member States as possible, and cannot hinder a Member State from participating, let alone suspend participation. A withdrawal, although being discussed, 107 is not foreseen explicitly.
All other PESCO decisions are taken by the participating Council members in unanimity with no passerrelle-clause foreseen to apply qualified majority voting. Unlike in enhanced cooperation, the presence of non-participating Member States in the deliberations is not foreseen explicitlynevertheless, the drafting history points toward them being allowed to participate. 108 The High Representative for Foreign Affairs and Security Policy is to be "fully involved" according to the preamble of Protocol (No 10), without this involvement being subject to a formal procedure. 109 No specific role is foreseen for the European Parliament-only regular consultations with the High Representative for Foreign Affairs and Security Policy according to Art. 36 TEU. No specific role is foreseen for the Commission either. The Court of Justice in principle does not have jurisdiction over PESCO (Art. 275 TFEU).
In light of the strong intergovernmental character of the PESCO one may query the use of such detailed and complex provisions (a protocol full of criteria and rules is devoted to the matter). After all, Member States wishing to proceed further in defence matters could do so quite easily by reverting to ad hoc agreements outside of the EU legal framework, as it has often been the case in the past. Yet anchoring enhanced cooperation in defence into the EU legal order internalises differentiation, thus providing avenues for experimenting new cooperation regimes within the institutional framework of the Union. This arguably serves the purpose of preventing cooperation taking place outside of the EU legal framework and allows the EU to keep control of differentiation through well defined procedures and rules 110 . These provisions ultimately provide for legal solutions which would enable groups of Member States to do together within a set EU framework what they may otherwise choose to do beyond EU structures altogether. As a matter of fact, PESCO has not prevented military cooperation from also happening outside of the EU but it has given a chance to differentiated integration in this area to also develop within an innovative EU intergovernmental framework.
In fact, there has been no obstacle to the use of PESCO in the past and the UK could not have prevented it. But for a long time the desire to use it was lacking. 111 The EU leaders eventually agreed in 2017 to set up a 'Military Planning and Conduct Capability' facility. This potentially constitutes a first step towards a EU military headquarter. 112 The June 2017 European Council conclusions on security and defence presented initiatives that have not been seen since the failure of the European Defence Community in the late 1950's. A formal decision was taken to launch Permanent Structured Cooperation (PESCO) provided for by the Lisbon Treaty. 113 Only the UK and Denmark do not participate and Malta still considers joining depending on the actual implementation. Already seventeen initial collaborative projects have been chosen. 114 Besides PESCO, EU leaders have further committed to establish a multi-billion-euro defense fund to finance joint military projects. This European Defence Fund is without peer in terms of the financial resources committed and the type of activities envisaged. It enjoys the strong support of both Macron and Merkel. The European Defence Fund, administered by the Commission, is intertwined with PESCO: Projects conceived through the Permanent Structured Cooperation receive higher co-funding from the fund. 115 Hence, despite its absence in the Treaty provisions, the Commission is trying to introduce some supranational involvement in PESCO.
These new initiatives can also be understood as a response to Brexit in two ways. First, UK military capabilities may be replaced by the other EU states or at least maintained in a different way. Second, for some in the EU, Brexit creates an opportunity for much deeper integration in military matters. The UK has consistently argued the primacy of NATO and avoided any real progress in EU defense cooperation. How far like-minded Member States can really go remains not only to be seen but also to be considered in relation to constitutional obligations to the non-participating Member States, and (some of) those who maintain a policy of neutrality. In addition, Brexit may be seen also as an incentive to go "outside of the EU". In 2018 under French leadership a joint military project between 14 European countries was established, including the UK and Denmark. The so called European Intervention Initiative has the objective to enhance cooperation on security matters, "thereby creating a framework for selective cooperation outside the structures of the EU" 116

The art of marbling
Differentiationcruciallydoes not necessarily mean either fragmentation or disintegration. At least on paper, the Union has long followed an ideal of diversity in unity -indeed, "united in diversity" has been its official motto since the year 2000. Diversity may exist without unity being compromised. One particular challenge is the preservation of unity in spite of diversity. In the light of President Juncker's 117 and President Macron's 118 proposals to reshape Europe after Brexit, which have emphasized the virtues of 'variable geometry' or 'multispeed Europe' as an organising principle for reform, some Member States have expressed fears that such fragmentation could lead to the collapse of the EU. 119 In this way the debates on further integration and threats of disintegration remain superficially inter-twined.
In this chapter I have attempted to show that forms of differentiation can coexist together without necessarily challenging the unity of the EU legal system. In fact the EU legal system has been progressively accommodating and internalising differentiation, adapting to the new integration needs. Also when taking place outside of the EU legal framework, differentiated integration has maintained a tight connection with EU law and could arguably not exist in isolation. The CJEU has validated current use of differentiation instruments, both as regards enhanced cooperation and the international ESM Treaty. I have argued that the EU legal system has adapted to new integration pathways, by internalising differentiation and somehow reabsorbing the fragmentation of the Maastricht's construction. Until this process is controlled and steered by the EU institutions and the CJEU, and until the respect of the Union's foundational principles is ensured, a certain level of differentiation can be accommodated within a system which is unitary yet in a variety of layered ways.
What is key is that differing positions and variable participation in integration goals do not endanger the coherence of the integration project as a whole. Such is only possible if, whenever differentiation elements emerge, accommodation is made to ensure loyal cooperation between participating and nonparticipating Member States. Some common tools that are used to address that challenge are the status of associated membership (e.g. Denmark in Europol after the referendum which rejected its participation in the AFSJ), or the right of Member States to participatewithout voting rightsin deliberations concerning policies subject to enhanced cooperation in which they do not take part. Above all, differences in integration must respect a certain core understanding of the Union, which can only happen if its values and constitutional principles are followed as elements that foster coherence of the overall political and legal system. The challenge, more than ever, is to ensure that the European Union's foundational principles still inform and permeate the variations in integration created by its Member States.

From a Europe of Bits and Pieces to a Union of Variegated Differentiation
European University Institute 21 relflects the application of an aqueous surface design onto a pre-existing system, which can produce patterns similar to smooth marble or other kinds of stone.
The broader reality is that there is now a more open discussion after Brexit of the need for certain policies to deepen for the EU 27. This is explicitly not just an institutional matter but a more explicit focus on substantive policy areas such as defence, security cooperation and EMU requiring a sharper focus on the competition between integration and cooperation. Paradoxically, the departure of the UK, the most vocal advocate of differentiation within the Union, has relaunched the debate on differentiated integration among the remaining members 120 . For sure Brexit will not stop differentiation, it may however modify the forms that it takes. Brexit will not stop it because it is arguably born out of the same concerns and dynamics that are nurturing the differentiation appetite. It happened at a time of general crisis of the Union, that faced (or is facing) challenges ranging from the euro-crisis to the management of refugees flows -and now with the coronavirus even a global health crisis of unprecedent dimension that will have perduring effects on the economic performance of the Member States. Under these circumstances what drives differentiation forward are the diverging views of Member States and of European people at large about the potential solutions to these crisis and, as a result, about the objectives and use of European integration. We may certainly expect even more differentiation in the future. Yet the UK leaving might for instance reduce the importance of the opt-out instruments, while boosting other more structured forms of differentiation. After all, the UK is the country which enjoyed the most Treaty opt-outs and that has been constantly pushing to enlarge the range of its exceptional treatment. Without the UK, the practice of opt-outs might lose its appeal, leaving isolated those Member States who tend to rely on this instrument, such as Denmark and the Czech Republic. Conversely, forms of institutionally structured cooperation, such as enhanced cooperation and other special regimes, might gain new impetus, in order to advance integration in selected areas. A recent resolution of the European Parliament points in this direction, acknowledging that Brexit may offer the opportunity "to move away from models of 'opting out' towards non-discriminatory and supportive models of 'opting in '". 121 The fact that Member States hold divergent views on the finalité de l'Europe does not per se mean that further integration is not possible. In fact only rarely has EU integration advanced on the basis of fully shared objectives. Even at the times of the Paris Treaties and the creation of the European Coal and Steel Community, the interests of France and Germany pointed in the same direction but were not identical. It may be argued that in fact this is what integration is all about: bringing together diversity in a common project. And often this common project is based on compromises rather than on full alignment. This is meant to be only a cautious warning not to confuse the heterongeneity of Member States' interests with the necessary need for differentiation. Of course, there must be a sense of direction and purpose, perhaps that ideal of "an ever closer Union" that is missing today. One cannot force Member States to remain together, as clearly showed by the UK experience. Are we heading too far apart? Only time will tell. Differentiated integration is not the answer to every situation in which Member States do not agree. However, given the gloomy perspectives of further traditional integration, as for now it provides a viable plan B for those countries who still wish to hang out together despite all the recent upheavals.
Arguing that differentiated integration is possible, or actually that it is already an undeniable reality, however, does not mean to say that this is unproblematic. In particular the effects of differentiation on the democratic legitimacy of the Union are questionable 122 . First, differentiation makes the already complex EU system even more complicated, with negative drawbacks on transparency and communications. Second, it can hamper the creation of a common "European sentiment", as citizens do not feel part of the same project. Daniel Thym in particular stresses the risk that differentiation, especially in the case of opt-outs, may strengthen the self-perception of the out-members as being Robert Schuman Centre for Advanced Studies Working Papers different, thus paving the way to further distancing, as happened with the UK 123 . Finally, it can negatively affect the connections between citizens and the European democratic institutions. As shown in the case of EMU, one of the main challenges is to know how the European Parliament can at the same time guarantee the democratic oversight of the eurozone and the represenativeness of all EU citizens. Overall differentiated integration blurs the lines between supranational and intergovernmental, between ins-and outs-members, between EU and international law. The clarity of the European project is affected, and so is the democratic accountability line between the Union's institutions and its citizens. The EU's track-record in ensuring democratic legitimacy has never been brilliant, as witnessed by the ever-lasting debate on the Union's democratic deficit. Guaranteeing fair and democratic representation as well as enhancing the legitimacy of the Union's action possibly remain the main challenge in times of differentiation.
To conclude, the choice for deeper and differentiated integration in distinct policy areas is logical for the EU 27 and very much part of the existing trajectory of European integration over time and space. This is (far) less than what Valery Giscard d'Estaing has argued fora re-booting of the European project with a "strong and federated" entity of 12 Member States who go it alone with specific institutions, a single budget and tax system, without any treaty change. 124 An original preference for 'outsiderness' over full membership pursued by certain Member States (eg the UK and Denmark) especially in areas that were previously exclusive to the nation state, are transformed by specific authorities (eg law enforcement communities and policy-makers) into a desire for greater functional integration (retaining access to relevant EU databases and agencies). Political sociologists have long challenged the claim that opt outs lead to the marginalisation of certain Member States and contribute to European disintegration. 125 On the contray the types of coping strategies used by for example Danish or British officials reveal that even where national referenda take place (on membership of Europol or of the EU in its entirety) all kinds of convoluted legal and practical arrangements may be made to ensure that, below the political radar, officials, national and European, can continue sharing information and cooperating even 'inter-operably' despite negative referêndum results. 126 In a significant sense, fragmentation and coherence are indeed not immutable and static aspects of a legal and political system but lie in the eye of the beholder. 127 The European Commission supports the EUI through the European Union budget. This publication reflects the views only of the author(s), and the Commission cannot be held responsible for any use which may be made of the information contained therein.