Brexit: Foundational Constitutional and Interpretive Principles: II

by | Oct 28, 2016

author profile picture

About Paul Craig

Professor Paul Craig was educated at Worcester College, Oxford, where he subsequently became a Fellow and Tutor in law in 1976. He was appointed to a Readership in 1990, and then became an ad hominem Professor in 1996. He was appointed to an established chair in 1998, the Professorship in English law, which is held at St John's College Oxford. He was made an Honorary QC in 2000, and an Honorary Bencher of Gray's Inn in the same year. He has lectured at many other institutions across the world, including in North America, Europe, China and Australia. He is editor of the Clarendon Law series, co-editor of a monograph series on EU law in Context, and is on the editorial board of various law journals. He is also a delegate of Oxford University Press, and was the alternate UK member on the Venice Commission for Law and Democracy. His research interests include Constitutional Law, Administrative Law, Comparative Public Law and EU Law, and he has published widely in these areas.


Paul Craig “Brexit: Foundational Constitutional and Interpretive Principles: I” (OxHRH Blog, 28 October 2016) <> [Date of Access]

This should be read with the previous posting. They are both designed to reveal underlying issues of constitutional and interpretive principle that pertain to Brexit. The previous post considered constitutional principles that pertain to Parliament’s voice in the Brexit negotiations. This post addresses foundational principles that affect the interpretation of Article 50 TEU, the key issue being whether withdrawal can be revoked once Article 50 has been triggered.  

Article 50 TEU: Four Interpretative Fundamentals

First, it is a cardinal legal principle that a party is not bound by a contract or treaty until agreement has been reached. Prior to that point a party can walk away from the negotiation. This legal principle is not dependent on express written mention. To the contrary it is accepted as part of the legal rules that govern the making of such agreements. The UK is a party to the EU Treaty until the withdrawal agreement is struck, or until the two year period has run its course. Prior to that date it can change its mind. This principle is particularly important in relation to Article 50 since there may be great uncertainty as to the content of the withdrawal agreement when notice of withdrawal is given. It is true that civilian legal systems have principles concerning good faith in contract. There is, however, no reason to believe that such principles would be broken where a state concluded bona fide that withdrawal was not in accord with the national interest when the concrete reality of the deal became apparent.

Secondly, the principle set out above is reaffirmed by Article 68 of the Vienna Convention of the Law of Treaties, VCLT, which provides that “a notification or instrument provided for in article 65 or 67 may be revoked at any time before it takes effect”. Article 65 VCLT stipulates that a state that invokes a ground for withdrawing from a treaty must notify the other parties; Article 67 VCLT determines the manner in which such notification is to be given, namely that it should be in writing. The notification under Article 50 TEU only takes effect when the withdrawal agreement is concluded, or two years has elapsed, since before then the notifying state has all EU law rights and obligations. It follows that the state can revoke before that time. It would be possible for the CJEU to interpret Article 50 so as to displace Article 68 VCLT. However the very architecture of Article 50 TEU is informed by the VCLT principles concerning the giving of notification and the like. The foundational starting point of legal principle, which is reaffirmed by the VCLT, is therefore that a state can revoke prior to the two year period running out. In doing so it is not acting illegitimately or seeking some unwarranted advantage.

Third, when we consider possible interpretations of Treaty provisions we take account of their consequences. The consequences of not being able to revoke would be particularly severe: withdrawal would have to proceed even if invocation of Article 50 triggered an economic meltdown in the country; invocation of Article 50 could not be altered within the two-year period, even if there had been a change of government following an election fought on whether the Member State should exit; there would be intractable problems if the state required a referendum prior to exit, since the voters might decide not to leave and reject the withdrawal agreement; and it would undermine the Constitutional Reform and Governance Act 2010, depriving the provisions designed to give Parliament voice concerning the withdrawal agreement of almost all effect.

Fourth, there are consequentialist arguments that are said to point to the contrary interpretation: if a state could revoke it might trigger revocation at the midnight hour, at the 23rd month, and then make a new notice of withdrawal, thus beginning another two year cycle. This argument is however based on the implausible assumption that the state would get a better deal second time round, which does not accord with reality. The other 27 Member States would regard such action as illegitimate, and would very likely refuse to shift from the pre-existing offer. The courts could moreover prevent such abuse. They regularly peer behind deals artificially structured to, for example, avoid tax, and consider their underlying substance. The CJEU could conclude that the revocation and subsequent re-triggering of Article 50 was designed to undermine the two year time period, and to undermine the need for consent of the other 27 states for an extension of that time limit; that therefore the revocation and subsequent re-triggering should be seen as a single event, not as two discrete events; with the consequence that the state would only have one month before the two year period expired. There is undoubtedly a valid concern that the EU should be disrupted as little as possible by the exit of a Member State. I do not however believe that the feared disruptive consequences represent reality. What would be far more disruptive would be to force a state to leave the EU after triggering Article 50 when it had bona fide changed its mind.

Share this:

Related Content


  1. Alex Goodman

    Given the issue of whether art 50 can be revoked is integral to the De Santos question of whether the executive can abrogate rights, should the UK courts not be referring this sub-issue to the ECJ? The divisional court seemed in argument to accept that an art 50 notification could not be revoked- so your view would presumably be at least that it is not acte clair?

    • Paul Craig

      Alex, in answer to your comment, my view is that the interpretation of Art 50 as being irrevocable comes nowhere close to being acte clair as that term is interpreted by the CJEU

  2. Sean Feeney

    Like most commentators, Professor Craig addresses a concept, “revocation” of notification, which has literally no support for applicability in the text of the lex-specialist withdrawal procedure of Article 50.

    A decision of intention to withdraw from the European Union by a member state is a challenge to the highest political objective of the European Union – that there be a Union of member states.

    The real question posed by Article 50 is not, I believe, whether notification of intention to leave can be revoked, but whether the “withdrawal agreement” can be given the non-literal interpretation that there can be political agreement that the member state intending to withdraw remains in the EU.

    I think it is crystal clear from a purposive reading of the text of article 50 that the European legislature intended this to be an ad hoc political decision.

    On this analysis, the competence of the CJEU would be limited to confirming this interpretation of Article 50.

    Competence for deciding this political question, whether the UK remains within the EU, would lie where it belongs: with the member states, including the UK, and the European Parliament.

    • Paul Craig

      This comment wholly misinterprets Art 50 as well as my blog, the latter being intended to ensure that Parliament and the people do have a say as to whether we should leave when the terms of the deal are known.

  3. Larry D.

    The reasoning exposed on this blog is absurd. Let’s read article 50 again:
    “3. The Treaties shall cease to apply to the State in question from the date of entry into force of the withdrawal agreement or, failing that, two years after the notification referred to in paragraph 2, unless the European Council, in agreement with the Member State concerned, unanimously decides to extend this period.”
    At the end one can read “unless the European Council […] unanimously decides to extend this period”. It certainly does not say “unless the Member State which decides to withdraw changes its mind”. Besides, why would we need a unanimous decision of the European Council if one single Member State could decide alone? This provision is a guillotine clause which admits no changing of opinion. It aims at making sure that after a certain period (which can be unanimously extended) the question is settled.

    • Paul Craig

      You need to read Article 50 and my blog more carefully for the following reasons.
      1. The unanimous decision of the European Council is required to extend negotiations beyond the two year period. I did not say that a state could revoke its withdrawal after the two year period. To the contrary I stated that it could so before that time elapsed.
      2. I gave two reasons of principle as to why a state should be allowed to revoke prior to that time: one is derived from a fundamental principle of law that is applied to the making of contracts and treaties; the other is affirmation of that principle in Art 68 VCLT, which contains the general principles through which treaties are to be interpreted, which says expressly and unequivocally that a state that tenders a withdrawal notice can revoke prior to the time when it takes effect. I accepted in the blog that the CJEU might displace Art 68, but it nonetheless provides the default position and it is clear moreover that the architecture of Article 50 was framed with the VCLT in mind.

  4. Paolo Sandro

    Dear Prof Craig, thanks a lot for this brief and accessible comment. I also appreciate your restraint in suggesting that the UKSC might have an obligation to refer to the CJEU as to the revocability or not of the notification in art 50 – that conclusion seems to me nonetheless obligatory (at least as a matter of EU law) once your lucid analysis is considered vis-a-vis the majority opinion that sees the notification under art 50 as irrevocable. Whether the UKSC will abide by its obligation under EU law, it’s an entire different matter of course. But the obligation to refer is there. Wondering how the Daily Mail would take it..

    • Paul Craig

      Thanks for the comment: I take your point, which is correct — the CJEU is the ultimate decider of the revocability issue.

  5. Stefan Junkers

    Dear Professor Craig, It has been suggested elsewhere ( that an order for reference by the Supreme Court may be avoidable. Is it any less constitutionally obnoxious that the government’s invocation of Article 50 >couldwould< deprive those subjects of those rights? Is it critical to determining the dispute between the main parties that the revocability of an Article 50 notice be determined, such that the Supreme Court would be obliged to make an order for reference?

    • Paul Craig

      There are ways in which the SC could avoid deciding the revocability issue — it could in effect decide that even if Art 50 might be revocable the government could still not proceed via the royal prerogative because it might not revoke the withdrawal, such that the UK would exit the EU two years down the road if there were no withdrawal agreement.

  6. Greg Callus

    Dear Professor Craig – many thanks for writing this. It is with great respect and trepidation that I dare to disagree.

    (1) It appears you see withdrawal and negotiation of the aftermath as though it were the starting point. Surely the better analog is an existing contract (TEU and TFEU), whereby any party (here the UK) has an option to terminate and thereby be outside the contract entirely (WTO terms), subject to any other agreement made between the parties. The consequence is not that there must be a new contract, but that there may be, yet absent its agreement, notice of termination has effect. Whether or not such a notice could be withdrawn would be a matter of the proper construction of the first contract (TEU) not first principles as to whether one had been formed.

    (2) I’m afraid I think it almost certain the CJEU would interpret Article 50 TEU in terms, rather than by the VCLT.

    (3) You treat a feature as though it were a bug: the whole point of A50 is to have a compressed two-years-max chance at an alternative deal (subject to unanimous extension) otherwise the economic apocalypse hits the leaving state. It it designed to deter Member States from leaving, and then to give the remaining EU members the maximum leverage in any negotiations that do happen. Prejudice to the leaving state, and making sure it is quite certain it wants to leave before it triggers Article 50, is the entire point of the stringent provisions: to construe A50 as though the EU was neutral on members leaving it, and wanting to make it relatively painless, is to miss some rather important context.

    (4) You correctly identify it would be absurd and abusive if a state could withdraw and then re-issue just to get an extension beyond two years, without meeting the requirement of unanimity. It would be, and having put such an onerous burden on even a first, modest extension, it seems unlikely that there would be no fetter whatsoever on unilateral withdrawal (letting the member back into all Council meetings etc etc). Your elaborate suggestions for how this could be solved (letting the CJEU decide whether or not the UK was still in the EU or not, on the basis of abuse of process arguments!) would mean utter chaos, and cause such profound uncertainty as to be insensible. The ‘irrevocable’ argument benefits from absolute certainty at all stages. Such gymnastics as must be performed by the argument for revocability are only necessary in case of the almost absurd outcome whereby a country decides that it wants to leave the EU, notifies under Article 50, then changes its mind within 2 years! I accept, of course, that this might well happen in the UK, but the idea that it was impliedly catered-for in A50 itself strikes me as far-fetched.

    Ultimately, the answer to the point is short: an A50 notification is irrevocable, otherwise the extension of time stringency makes no sense, and so as per A50(3) itself: the Treaties cease to apply within 2 years of a notice (unless extended) come Hell or high water, and a state that changes its mind needs to re-apply (or persuade all 27 members to strike a deal otherwise) under A49, as set out under A50(5).

    • Paul Craig

      Thanks for the comment. People clearly disagree on this. A quick response.

      1. Clearly notice of termination has legal effect. It triggers the beginning of the Art 50 process, and the start therefore of the two year deadline (subject to agreement on extension thereof). This does not tell us anything as such as to whether the process is revocable. Indeed the most detailed full length study in a recent paper by Aurel Sari from the University of Exeter concluded that read in the light of the evidence overall Art 50 should be considered to be revocable.

      2. I never claimed that Art 68 VCLT provided the answer in itself. It is however a powerful piece of evidence as to what the international community, which includes the MS of the EU, think should apply in circumstances such as these. Art 68 is moreover intended to apply in precisely the circumstances where the Treaty provision provides no particular answer one way or another. I think therefore that your brushing aside of the VCLT is neither legally nor normatively warranted.

      3. Your third point just presumes the answer to the question in issue, it provides no independent basis for it. Insofar as there is an independent basis it is highly questionable. Thus the idea that it would be best for the EU, or that it is in the EU’s interest, for Art 50 to be completely irrevocable is not self-evident at all. Indeed to the contrary, I could think of few things that could be worse for the EU, whether from a political, economic or symbolic perspective, to force out a state that had bona fide changed its mind in the 2 year period.

      4. I do not think that the suggestion of the legal technique for preventing abuse of Art 50 is elaborate, or that it would lead to chaos. Courts interpret legal rules to prevent their abuse all the time. Nor do I think that the scenario painted of a state that seeks to abuse Art 50 comports with political or economic reality: the assumption that it would do so in order to secure some better deal in a ‘new’ two year period is predicated on the idea that the other MS would be willing to revise whatever the previous deal was in order to give an improved version thereof to the departing state. There are however plenty of reasons why the EU would not do this, and no reason why it would.

      5. My guess is that the CJEU would interpret Art 50 as allowing revocability, but probably not unilateral, and would require a qualified majority from the other MS.

Submit a Comment