On 16 June, by two parallel orders, the EU Court of Justice said the last word on the legality of advocate general Sharpston’s divestment. The Court earmarked Sharpston’s final try for Article 181 fast-track procedure, which is where “manifestly inadmissible” or “manifestly unfounded” appeals fall, without even requiring service on the defendants. The Court could not have been more dismissive. However, while it had hitherto rejected AG Sharpston’s claims exclusively on points of procedure, i.e., as going beyond the bounds of action of annulment, this time it rounded off the argument explaining why such claims would have failed on the merits as well. Not that the Court had second thoughts on admissibility, far from it. It just sought complete closure.
Sharpston, the Court seems to say, has been fighting at windmills, for nobody divested her. Her term elapsed by the operation of law. When the remaining member states, on 29 January 2020, declared that Brexit would have made her position vacant, they “merely took note of the consequences necessarily entailed by the departure of the United Kingdom from the European Union”. By necessary implication because, as the Court explained, under Article 50(3) TEU the Treaties “cease to apply” to the withdrawing state upon the entry into force of the withdrawal agreement (C-684/20 P, paras. 48-49; C-685/20 P, paras. 53-54).
In the end, the Court did little more than reiterate the press statement it made in response to the member states’ declaration: “The Court of Justice takes formal notice of the fact that the withdrawal of the [UK] has the effect of bringing to an end the mandates of the British Members of the Institution with effect from 31 January 2020 at midnight”. When midnight came around everyone – the member states as well as the Court – had just to “take note” and no responsibility. In reality, the member states made a legitimate decision based on an old custom, and the Court could not but oblige.
A bug in the Court’s reasoning
The Court struggled to impart deductive nonchalance to a reasoning that fails to persuade. Commentators have rightly stressed that whilst judges are “from” (treaty language) a member state, advocates-general are not (Article 19(2) TEU). Accordingly, EU law sets the number of advocates-general at eleven and in no way ties it to changes in the EU membership. But the deadly bug is this: there exists no provision in the Treaties whose application would cease, as per Article 50(3) TEU, with the effect of terminating a judicial appointment. Arguing that a break in the application of Article 19(2) TEU as such retroactively affects past appointments is a non sequitur. Regarding advocates-general’s nominations, the provision never applied to states in the first place. It could not possibly cease to apply to the UK. It follows that it was not the Treaty’s operation that caused the British contingent’s early departure from the CJEU. The member states, the UK included, willed it.
The 8th recital of the Withdrawal Agreement’s preamble suggests that the EU and the UK struck the deal on the assumption – possibly the subject-matter of a verbal understanding – that Brexit would put an immediate end to “the mandates of all members of institutions, bodies and agencies of the Union nominated, appointed or elected in relation to the United Kingdom’s membership of the Union”. The notion clearly encompasses the judges “from” the UK as well as Sharpston, a three-time British nominee to the post of advocate general. The UK government’s acquiescent reaction to her replacement tends to confirm that the parties agreed on it. But all this is immaterial, juridically, as no such agreement could supersede security of judicial tenure, a corollary of principles of EU law as essential as judicial independence and the rule of law.
No wonder scholars complained, vehemently sometimes. Halberstam was possibly the first to sound the alarm, claiming that at stake were “the rule of law, and the constitutional structure of the Union”. Kochenov and Butler chided the CJEU for having “ditched one of its members”. AG Sharpston herself told the press that an onslaught on the Court’s independence was underway. The terse statements with which the Court settled the case will likely stoke up suspicions of complicity. And yet the Court made the right call, despite the unpersuasive reasoning.
Interestingly, although the Court and its detractors offer divergent interpretations of what the EU constitution demands, they both rely exclusively on treaty interpretation, never quitting the comforts of written law. To see what they miss, one must descend into a damp, dim-lit recess of the EU’s constitutional edifice.
In The Constitution as an Institution, Llewellyn chastised doctrines obsessed with textual interpretation, doctrines which, in his view, covered “perhaps half” of what should count as constitutional materials, throwing “the other half in Stygian black”. Critical to the understanding of Sharpston is what Llewellyn would have called the EU’s constitutional “penumbra”, a kind of customary law, a law that, long predating the European integration process, has always governed the apportionment of judicial seats among the member states. The EFTA Court’s early history offers an arresting snapshot of that law in action.
EFTA’s vicissitudes include a little known parallel to the Sharpston affair, and to Brexit (no, not this one). On 1 January 1995, when the EFTA Court was hardly through its inaugural year, EFTA lost to the EU three of its five members – Austria, Finland, and Sweden – a massive exit. As its longest-serving ex-President recalls, the Court survived by dint of an arrangement between the member states, remaining and leaving alike. In disregard of treaty law, the arrangement determined that the judges nominated by the leaving states had to help with the pending cases until 30 June 1995, and then resign. The Court found nothing objectionable in the circumstance that states forced three of its members to leave.
The wording of Article 28 of the agreement establishing the EFTA Court makes the parallel even more striking. In all its successive versions, Article 28 provides that the EFTA Court “shall consist of [n] judges”, with n matching the number of member states. Unlike Article 19(2) TEU, which features the more flexible formula “one judge from each Member State”, Article 28 does not specify that each state is entitled to a judge. On a literal reading of it, each of the three EFTA Court judges prematurely dismissed could have asked, “Why me?”, a bit like AG Sharpston did.
To our pragmatic minds, Article 28 reads like Article 19(2) TEU: the fact that the number of judges coincides with the number of states can only mean that each state “owns” a judge. In realty, meaning comes not from the wording – Article 28 is silent about the judges’ provenance – but from an unwritten constitutional rule. The converse is true as well: Article 19(2) reads like Article 28, in the sense that both are taken to mean that there have to be as many judges as states at each point in time. On this reading, a post-Brexit Court of Justice temporarily consisting of 28 judges would be inconceivable. But, again, this is not what the Treaty says. Unlike its EFTA counterpart, the TEU is not specific about the number of judges. An equally plausible reading of it – and one more in tune with security of tenure – would have allowed the British judges to stay beyond Brexit.
For Pech, “should [a] Member State leave the EU, the post of judge associated with that Member State must logically disappear on the day that State’s withdrawal is effective”. This is correct, but it is not logic that makes it so. It is, again, a customary rule lurking in the background. This rule, which applies to judges and advocates-general alike, made AG Sharpston’s seat vacant at midnight before Brexit.
An ancient custom
The rule stems from an ultra-secular tradition of governance of international judicial institutions. Since the early 20th century, states have been making those institutions in two distinct senses: they set them up, after which they concur in the appointment or election of judges and other officials, following pre-established apportionment conventions. Such arrangements arguably make the very existence of said institutions possible.
In 1907, during the Second Hague Peace Conference, lack of agreement among states on apportionment of judicial seats doomed the blueprint for a permanent international court to failure, shining a light on the main stumbling block to the establishment of an international judiciary. Post-World War I diplomacy broke the impasse. For Manley Hudson, the provisions of the Statute of the Permanent Court of International Justice dealing with the election of judges marked “the greatest triumph achieved in 1920”. Critical to it was the tacit understanding that the bench had to include nationals of the great powers. One century later, the procedure for the appointment of judges to the defunct EU Civil Service Tribunal failed to run smoothly, and eventually spawned intractable disputes, because no apportionment convention for filling the Tribunal’s seven seats ever crystallized on top of the formal requirement to ensure a geographically “balanced composition”.
Since apportionment conventions coexist in tension with the principle of judicial independence, they are often tacit or deceptively consigned to forms of sub-legal existence, making occasional appearances in non-binding declarations. In legal texts, the same tension reverberates in elliptical allusions to underlying power-sharing arrangements (as in provisions that spell out just the number of judges) or in delightfully euphemistic phraseology (“from each Member State”). Equally important in explaining insincerity is the conventions’ frequent departure from the principle of equality of states. Apportionment conventions are more likely to make it to the surface of a legal text when they do not contradict that principle. Article 19(2) TEU instantiates this model insofar as it applies to judges. The smaller pool of advocates-general complicates the matter and makes the relevant rule slide into a juridical penumbra.
No mere practice
The rule which controls the apportionment of the CJEU’s seats started life in the early 1950s, as an informal arrangement. Its kernel – great-power privilege limited to the selection of the advocates-general – has remained unchanged since then. In the 1950s and 1960s it was understood, yet never put down in writing, that France and Germany “owned” the only two posts of advocate general, a clear sign of the role’s significance. In the 1970s, the number of posts doubled, making room for two other permanent holders, Great Britain and Italy. The current regime, combing permanent seats with a rotation system, began to take shape in the 1980s. In 1995, a declaration reported that Spain had joined the states that “have a permanent Advocate-General”. A later declaration, annexed to the Lisbon Conference Final Act, chronicled the co-option of Poland among those “haves”. Marking an unfortunate return to obfuscation, the 29 January 2020 declaration portrays as an automatic consequence of Brexit the political decision to “smelt” in the rotating pool the British seat, instead of handing it over to Romania.
For Halberstam, “the colloquial use of the word ‘have’” in those documents signals that the apportionment of judicial seats is “the stuff of non-binding understandings”, material of scant legal value. However, the member states take that “have” very seriously, indeed literally. Solecisms often tell the truth. Indeed, the gist of the customary rule in question is that states have advocates-general for as long as they remain members of the Union. The same (less controversially) applies to judges.
The CJEU could hardly see in it a “mere practice” that “cannot override the provisions of the Treaty”. In 2007, AG Poiares Maduro proposed to apply customary international law’s two-element test to distinguish transient political practices from intra-EU unwritten rules of law. However, already in 1983 the Court itself suggested that unlike public international law, which ordinarily requires proof of a “widespread and representative” practice, EU law demands consensus. Back then, the Court denied that the European Parliament’s practice of holding some of its plenary meetings in Luxembourg “had created a custom”, because a single member state, France, had consistently opposed that practice. By contrast, the practice of holding meetings of parliamentary committees in Brussel had, in the Court’s view, ripened into a custom as no member state ever questioned it. The same goes for the practice here under scrutiny.
One would be hard-pressed to deny that member states accept such practice as law (opinion juris). Being aware of the CJEU’s consistent rejection of practices at variance with the Treaties, the member states must believe that the old apportionment convention they unfailingly follow is lawful and in harmony with EU law’s supreme principles. Unlike “mere practices”, which supervene on the Treaty and depart from it, the customary rule in question is foundational in character, a point of departure.
A British-style constitutional convention
The rule that made AG Sharpston’s vacant upon Brexit is not a public international law remnant as marginal to the EU’s constitutional makeup as Lutetia ruins in Paris. It is part of the EU constitution’s bedrock and is best thought of as a constitutional convention, British-style (yes, there’s a little irony in this), a customary rule that modifies the scope of application of colliding supreme principles. That rule enabled AG Sharpston’s divestment, by way of exception to the principle of security of judicial tenure. This is a limited exception that only circumstances as radical as changes in the EU membership trigger.
The member states could have agreed to let AG Sharpston complete her term (as much as they could have consigned the UK’s permanent seat to Romania instead of suppressing it). Had they done so, they would have set a precedent, changing in part the relevant custom. As the UK Supreme Court held in Miller, judges “are neither the parents nor the guardians of political conventions”, whose “operation or scope” are matters to be “determined within the political world”. The limited subject-matter of such conventions – in the case at hand, the adjustment of a power-sharing arrangement accompanying a change in the EU membership – stands as a guarantee against their abuse.
To sum up: AG Sharpston’s term came to an end on 31 January 2020 “at midnight” not because some treaty provision ceased to apply, as the Court eventually ruled, but because an old custom continues to apply. No constitutional crisis has taken place. The Sharpston affair just holds up a mirror to an ancient trait of the EU constitution, deeply rooted in the sometime unsavoury history of international adjudication. But such excavations are better left to scholars. The Court was wise to cut it short and blame it all on Article 50(3) TUE, no matter what happened that night.