Jump to ContentJump to Main Navigation
Oxford Studies in Philosophy of LawVolume 1$

Leslie Green and Brian Leiter

Print publication date: 2011

Print ISBN-13: 9780199606443

Published to Oxford Scholarship Online: September 2011

DOI: 10.1093/acprof:oso/9780199606443.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 24 February 2017

Can There Be a Written Constitution?

Can There Be a Written Constitution?

(p.162) 5 Can There Be a Written Constitution?
Oxford Studies in Philosophy of Law

John Gardner

Oxford University Press

Abstract and Keywords

The existence of unwritten constitutions, such as that of the UK, strikes some as puzzling. However the existence of unwritten constitutions turns out to be easier to explain than the existence of written constitutions, such as that of the US. This chapter explores some tricky conceptual questions thrown up by written constitutions. The biggest of these is: What is a constitution? This chapter suggests (a) that a constitution is not only an important part of the law but also a set of extra‐legal political constraints; (b) that the fact that a constitution is written allows, and indeed requires, that the constitution also has a life in the case law and customary law of the system; (c) that what H.L.A. Hart called the rules of recognition of a legal system are not best regarded as part of its constitution, at any rate when the constitution is written. As the chapter goes on the criteria for regarding certain propositions of law as belonging to constitutional law (as opposed to administrative law or other parts of public law) are progressively refined. There is also discussion of some particular rules of constitutional amendment that apply in the UK and the US, which show the alleged differences between the two constitutions to be less radical than they are sometimes imagined to be.

Keywords:   constitution, comparative, rule of recognition, H.L.A. Hart, amendment, delegation, interpretation


Does the United Kingdom have a constitution? Some people doubt it. But there is no room for doubt. A constitution is a conceptual necessity of every legal system. In every legal system there are rules that specify the major institutions and officials of government, and determine which of them is to do what, and how they are to interact, and how their membership or succession is to be determined, and so forth. Without some such rules, as H. L. A. Hart explained in The Concept of Law, there is no legal system.1 These rules without which there would be no legal system (or some of them, invariably in combination with some other rules) make up the constitution of that system. Since the United Kingdom undeniably has some law, and since all law is necessarily the law of one or another legal system, the United Kingdom necessarily has at least one constitution.

I say ‘at least’ because the United Kingdom has three distinct municipal legal systems (the law of England and Wales, the law of Scotland, and the law of Northern Ireland). Does it correspondingly have three constitutions? And if it has three doesn't it inevitably have four, for surely there must also be an overarching constitution of the whole union that determines the relations among the other three? Or do the three (p.163) distinct legal systems share just one constitution between them? Is that a conceptual possibility? Can it be squared with the proper criteria for individuating legal systems? These questions are vigorously debated, and not only by theorists.2 Some of them are political hot potatoes. For present purposes, however, I will ignore them. At the risk of being hijacked for the unionist cause, I will speak brazenly of ‘the United Kingdom constitution’ (or ‘the UK constitution’ for short). For the questions I want to tackle here are unaffected by the possibility that the UK has a cluster of interrelated constitutions. Whether there is one constitution or several, each alike exhibits the feature that causes people to doubt whether the UK has any constitution at all.

The feature that provokes the doubts, notoriously, is that the UK constitution is an unwritten one. When they ask what the constitution of some country or state has to say on some subject, people nowadays typically expect to be directed to a canonical constitutional master‐text on the model of the Constitution of the United States of America (complete with its amendments). Thus the political furore about the proposed ‘Constitution for Europe’ was not about whether the European Union should have a constitution—as it already has a legal system, necessarily it already has a constitution—but whether it should acquire a canonical constitutional master‐text to match.3

In the UK, there is no canonical constitutional master‐text, and no live proposal to have one. This does not mean, I hasten to add, that there are no rules of the UK constitution that have canonical texts. On the contrary, there are many. To take a few famous examples, there are constitutional provisions in Magna Carta 1297, the Laws in Wales Acts 1535–42, the Habeas Corpus Act 1679, the Bill of Rights 1689, the Act of Union (with Scotland) 1707, the Act of Union (with Ireland) 1800, the Judicature Acts 1873 and 1875, and the Statute of Westminster 1931. (p.164) More debatable examples, among many, include the Great Reform Act of 1832 and its successor Representation of the People Acts, the Parliament Acts 1911 and 1949, and the European Communities Act 1972. But none of these, nor all of them stapled together, is ‘the UK constitution’, in the sense that many inquirers have in mind. They are only some written fragments of the otherwise unwritten (and so maybe we should say ‘uncodified’ rather than ‘unwritten’) UK constitution. This is not altered by the fact that two of them—the Acts of Union—are founding documents, in the sense that they establish the UK as a union. For what establishes a union is not necessarily its constitution. A union may be established (as in the case of the UK) by uniting its parts complete with elements of their existing constitutions, and without, in the process, reducing those elements to writing. In that case the founding document of the union only adds another fragment to its constitution.4

Notice that all of the constitutional fragments listed above are Acts of Parliament that were passed and promulgated in the ordinary way. Their status as part of the constitution is not determined by some special origin or process of enactment. They do not have ‘constitutional’ printed or stamped on them at point of issue. Even if they did, it is not clear that this by itself would have any legal effect.5 The UK constitution does not provide for Parliament to act as a primary legislator other than by passing an Act of Parliament in the ordinary way. Nor is there any other primary legislator constituted by the UK constitution6 whose enactments bind Parliament.7 That is one reason why there is no live (p.165) proposal to have a canonical constitutional master‐text in the UK. Even if one were thought desirable, it is far from clear how anyone would set about creating one. The UK constitution includes no procedure for its own deliberate amendment, never mind for its own encapsulation in a canonical master‐text.

If the UK constitution contains no procedure for its own deliberate amendment, how do certain Acts of Parliament come to have constitutional effect? Their constitutional effect is not determined by how they are created but rather by how they are received, by their treatment in either the customs or the decisions of certain law‐applying officials, principally the courts. The Acts in question are regarded, usually only some time after enactment, as placing constitutional limits on what various major political institutions, including the courts themselves, can do (i.e. are empowered or permitted to do) by law. In that sense, even though written, they remain part of an unwritten constitution, for their constitutional status—their entry into the constitution—comes of the unwritten law of the law‐applying officials who subsequently treat them as having that status. The debatable cases, meanwhile, are debatable precisely because of a continuing indeterminacy in the way they are treated by the relevant law‐applying officials. The indeterminacy may come of official dissensus or official circumspection. Or the issue may never have come up for official determination. This is not to deny, of course, that the Acts in question are determinately recognised as valid Acts of Parliament. It is only their constitutional status that has yet to be put to a decisive test.

When commentators are thinking about what might count as a decisive test, they often focus on whether the Act in question could validly be repealed by a future Parliament. Would a purported repeal of the European Communities Act 1972 be legally valid, even if it were in breach of the Treaty of Rome? How about a purported repeal of the Statute of Westminster 1931, aimed at recolonizing various former British colonies over the heads of their own constitutions? These are intriguing, although sadly imponderable, questions of law.8 As tests for (p.166) the constitutional status of an Act of Parliament, however, they are too strict. Even if a certain Act A could validly be repealed by Parliament in the ordinary way, i.e. by subsequently passing Act B to repeal it, Act A may meanwhile be recognized as having an effect on the operation of other parts of the UK constitution. Most importantly, without yet being recognized by the UK courts as either repealable or not repealable, Act A may be recognized by the UK courts as immune from the normal doctrine of implied repeal. By the normal doctrine of implied repeal, in the event of a conflict between Act A and Act B, the later Act B prevails even if Parliament did not notice, and hence in Act B did not make any provision for resolving, the conflict. This is a rule of the UK constitution. So if Act A is treated by the courts as an exception to it (on the basis that Act A provides for itself to be an exception or for any other reason), Act A is thereby elevated by the courts to a special constitutional status. The European Communities Act 1972, for example, has whatever constitutional status it has in the UK because and to the extent that the courts have held it immune from implied repeal.9 That it could still be expressly repealed by Parliament, if it could, does not strip it of that constitutional status. It only goes to show that at least some parts of the UK constitution are less deeply entrenched than are at least some parts of, say, the US constitution. They are not immune from the normal process of deliberate legislative change even though they are immune from some forms of accidental legislative change.10

I have emphasized the courts as principal arbiters of the constitutional status of legislation in the UK. To do so is to invite further political controversy. Richard Bellamy cautions against a ‘legal constitutionalism’ that presents the courts as the true guardians of the UK constitution. He advocates instead a ‘political constitutionalism’ according to which ‘the constitution is identified with the political rather (p.167) than the legal system, and in particular with the ways political power is organised and divided’.11 This, however, is a false contrast. My emphasis on the courts does not prevent me from endorsing Bellamy's view of the constitution. Let me mention just a few reasons why.

  1. (i) I already agreed that constitutions, not just in the UK but everywhere, are concerned with ‘the ways political power is organised and divided’. To be exact I said that a constitution, even when thought of as part of a legal system, is the part that specifies the ‘major institutions and officials of government’ and how they are to function, interact, be appointed, etc.

  2. (ii) As Bellamy seems to agree,12 the courts themselves are political players. In the UK, as elsewhere, they jockey for power and position no less than the other main political institutions.

  3. (iii) At the heart of the UK constitution is the doctrine of Parliamentary Sovereignty. As a legal doctrine, this has been developed principally by the courts. Some of the rules that go to make it up (those concerning express and implied repeal) have been sketched above. The doctrine, as we saw, severely restricts the ways in which and the extent to which Parliament can bind its successors, for Act A cannot anticipatorily immunize itself against its repeal by a future Act B (even if sometimes the courts may later grant Act A a limited immunity).13 But Parliamentary Sovereignty also limits the extent to which Parliament is bound by the courts. As the doctrine is applied today, the courts determine the legal effect of Acts of Parliament but (p.168) Parliament may always relegislate to overrule the courts, subject again to the courts’ determining the legal effects of the relegislation. Parliament can always get its way in the end by progressively more definite reiteration. Unless Parliament tires of the process, the courts only get to postpone their own defeat.

  4. (iv) The courts are the principal law‐applying institutions that determine the legal effect of Acts of Parliament and other legislation, but they are not the only ones. There are also various tribunals, commissions, regulatory agencies, and quangos that share at least some of the work of the courts. In small pockets, work of this type is also undertaken by the police, immigration staff, planning officers, etc. What distinguishes this work is that it involves authoritative applications of the law, i.e. determinations of legal effect that others are legally bound to follow.14

  5. (v) The rule of the courts, such as it is, is dependent on the submission to that rule of numerous others, including other officials, such as police and military commanders, and the submission to those commanders of ordinary soldiers and police officers, and ultimately the submission to these of the wider population.15 It also depends on détente with other institutions to whom there may be greater official or popular loyalty. The $64,000 question about every constitution is this: In a severe constitutional crisis, whose loyalties will lie where (and who has the weapons and who has the numbers on their side)?16

  6. (vi) Finally, the courts and other law‐applying institutions are only the guardians of constitutional law. No constitution is exhausted by its law. In the UK, for instance, there are also what Dicey dubbed the (p.169) ‘conventions of the constitution’: customary constitutional rules which the courts may note, rely upon, and accommodate in applying the law but of which the courts’ applications are not authoritative, even for the immediate purposes of the case before them.17 These extra‐legal customs, like any other rules, could be transformed into law by the courts themselves treating them as law. But the courts have many reasons to be self‐denying in respect of such a change, including but not limited to the risk of breaking the détente mentioned in (v), and thereby occasioning a coup against themselves.18

When thinking about the constitution of any country, theorists of law are naturally most interested in its constitutional law, which leads them in turn to emphasise the courts as arbiters of constitutionality. It does not follow, as Bellamy seems to assume, that they regard constitutional law as the most socially important, let alone the most socially desirable, part of the constitution.19 They may even think, as I tend to think, that something is amiss in the public life of a society when constitutional questions often have to be settled in the courtroom. Indeed, one might add, there is something amiss in the public life of a society when questions of any type often have to be settled by the courts. A theorist of law need not be an enthusiast for law, let alone for more of it. With that caveat entered, I will continue here to pursue my interest as a theorist of law. So I will persist in focusing on the legal aspect of constitutions except where otherwise indicated. My talk of ‘written constitutions’ should be understood accordingly.


In the UK, to recap, what determines the status of certain law as constitutional law is its reception into constitutional law by certain law‐applying officials, principally the courts. Couldn't this claim be (p.170) extended to constitutions generally? Couldn't it be argued that, in every jurisdiction and at every time, what really determines the status of something as part of the constitution is how it is received by its official users, principally the courts, never mind whether it has ‘constitutional’ stamped on it at point of issue? If so the tables are turned. For now the doubts do not hang over the possibility of an unwritten constitution like the UK's. Instead they hang over the possibility of a written constitution. On closer inspection, it may seem, it is part of the nature of a constitution that it is unwritten, and that its so‐called written parts are only parts of it because of their reception into the unwritten law that is made by the customs and decisions of the courts and other law‐applying officials. If that much is true, then ‘The Constitution of the United States of America’ is a serious misnomer, for inasmuch as it is a name given to a document containing canonical formulations of law, it involves a category mistake. Constitutions cannot be, or be contained in, documents. That, at any rate, is the heretical view that I will be exploring here. Although I will reject it in due course, some aspects of it strike me as salutary. They help us to see the exaggerations of the opposite view according to which a written constitution is somehow a more normal case.

Let me begin the argument in earnest by putting the challenge to the possibility of written constitutions in a slightly more elaborate form. As a student of UK constitutional law I was given to understand—or maybe simply came to understand by myself—that the distinction between questions of constitutional law and questions of ordinary public law (also known as administrative law) lies in the type of institutions that these respective parts of the law regulate. Administrative law regulates institutions whose powers are delegated. Constitutional law regulates those that do the delegating, i.e. institutions whose powers are not delegated but are, as it is sometimes put, inherent or original.20 Thus, in the UK, constitutional law regulates Parliament—or the Queen in Parliament as the institution is more acurately known— (p.171) together with the high offices of the Crown (mainly government ministers), the Privy Council,21 and the Queen herself in her official capacity. It also regulates the High Court and the judicial bodies known as the Appellate Committee of the House of Lords and the Judicial Committee of the Privy Council. Constitutional law does not, however, regulate the Court of Appeal, which is a creature of statute.22 Its judges are constitutional officials only in virtue of the fact that they remain, constitutionally speaking, judges of the High Court even after they have been elevated to the Court of Appeal. Nor does constitutional law regulate the police, the army, local councils, magistrates and their courts, tribunals and public inquiries, quangos or arms‐length government agencies, industry‐wide regulators, tax inspectors, or other similar institutions and officials.23 These institutions and officials hold their legal powers by delegation, usually from Parliament although sometimes by Royal Preogative (i.e. from the inherent constitutional powers of the Crown, exercised on behalf of the Queen by government ministers or the Privy Council).

It does not follow, of course, that these delegate institutions and officials are untouched by the law of the constitution. Most obviously, they can only have delegated powers in law if the original powers are there to delegate, which means that delegate institutions are subject to whatever restrictions the constitution places on the exercise of the same powers by those who delegated them. Moreover, the constitution may regulate which delegations are to take place, by what means, to whom, under what conditions, and so forth. That a delegate institution is picked out for mention in the constitution in this way does not (p.172) transform it into a constitutional institution. Consider the armed forces in the UK. Thanks to a provision of the Bill of Rights 1689, the Crown lost its constitutional power to enlist and maintain a standing army in the absence of Parliamentary consent. This is an aspect of the Royal Prerogative of national defence which came under early Parliamentary control24 and which already puts the army at the centre of an important constitutional relationship. But it does not turn the army into a constitutional institution. The provision regulates the Crown and Parliament, not the army itself, which gets its legal powers, if any, only by Parliamentary or Prerogative delegation.

This 1689 provision also helps to forestall other common misunderstandings. An institution of inherent or original power is sometimes understood to mean an institution that determines its own powers,25 or (more radically still) an institution with unlimited powers.26 But these definitions, we can now see, are too demanding. The Crown did not cease to be an institution of inherent power merely because it lost its power unilaterally to maintain a standing army. Nor did the Crown cease to be an institution of inherent power merely because it was Parliament, through the Bill of Rights 1689, that restricted the power of the Crown to maintain a standing army. Nor did Parliament itself cease to be an institution of inherent power merely because it was the courts that gave to the Bill of Rights its constitutional status, and so made it the case that Parliament, by enacting the Bill of Rights, had altered the constitutional powers of the Crown in relation to the maintenance of a standing army. So the point cannot be that institutions of inherent power are not subject to legal limitations imposed by (p.173) others. The point is only that their powers are not delegated to them by others.

Whether there is a delegation is not to be decided by looking at the history of the institutions in question. Historically, in the UK, both Parliament and the High Court (or at least part of it) had their constitutional powers carved out of the constitutional powers of the monarch. The King or Queen from time to time delegated some of his or her personal powers to them or to their predecessors. But these powers, although they were delegated at the time, are not now to be regarded as delegated powers. The legal position—the position under the UK constitution today—is that both Parliament and the High Court have inherent or original powers. They are not delegates and the Queen cannot, constitutionally, revoke their powers. If the Queen purported to revoke their powers they could, constitutionally, ignore the revocation and continue to sit. They might lose some or all of their powers in some other way—as the 1689 example shows, constitutional institutions are not immune from having their powers restricted by others—but they cannot lose their powers by revocation. This is a hallmark of institutions that fall under constitutional, as opposed to administrative, law.

The constitution, as I said, is what regulates institutions of inherent power. The emerging problem is this. The institutions of inherent power in any legal system are also those that are identified by what H. L. A. Hart called ‘rules of recognition’. To be exact, they are identified by the ultimate rules of recognition of that legal system.27 And the ultimate (p.174) rules of recognition of a legal system, as Hart explained, cannot be enacted, or otherwise canonically formulated. They cannot exist in legislation. They can exist only in the practices of officials. Putting it more technically, they can only be customary laws in foro.28 Why? To greatly simplify: Any attempt to create an ultimate rule of recognition by legislation requires that there be a superior legislator with the power to confer an original or inherent power. But that is a contradiction. If there is a superior legislator conferring the power, it follows that the power conferred is not original or inherent but delegated. The rule of recognition created is not, in other words, an ultimate rule of recognition. It follows that an ultimate rule of recognition cannot be a legislated rule.29 And one may conclude from this that a constitution, as the part of the law that regulates institutions of original or inherent power, can't be legislated either. It can't take the form of written law, existing in canonical formulation. All constitutions must therefore be unwritten like the UK one.

Many legal theorists embrace something like this argument. Most take it to be a reductio.30 They regard it as axiomatic that there are written constitutions, such as the US constitution, and therefore treat the argument as casting doubt on what Hart has to say about rules of recognition. Either, pace Hart, some legal systems do not have ultimate rules of recognition, or else an ultimate rule of recognition is capable, pace Hart, of being a legislated rule. Both of these conclusions are, however, inadequately supported by the argument just sketched. For the argument fails to establish any incompatibility between a Hartian (customary) rule of recognition and a written constitution. It is a faulty argument. Let me focus on three of its faults.

(p.175) III.

(1) The rules identifying institutions of inherent power do not exhaust the rules that regulate those institutions. Indeed some rules that identify institutions of inherent power do not regulate those same institutions at all. Hart occasionally encouraged the view that the ultimate rules of recognition of each legal system are those that confer the system's inherent powers.31 If this was ever Hart's view it should not have been, for it is incompatible with Hart's own careful enumeration and differentiation of the types of rules on the possession of which the existence of any legal system depends. Hart distinguished rules of recognition from rules of adjudication and rules of change, and argued that every legal system has distinct rules of all three types.32 Rules of adjudication and of change confer powers to apply the law and to change the law respectively.33 In the constitution—where we find the ultimate rules of adjudication and change—they confer inherent powers to apply the law and to change the law. But if the legal system's ultimate rules of recognition have already conferred these powers on the same institutions, why are rules of adjudication and change needed? Aren't they just duplicative?

The answer is that they are needed because a rule of recognition does not confer these or any other legal powers. A rule of recognition is a duty‐imposing rule.34 It imposes a legal duty on law‐applying officials. One of the UK's ultimate rules of recognition—Hart's stock example of a rule of recognition—is the rule by which what the Queen in Parliament enacts is law.35 This rule gives law‐applying officials the (p.176) duty, in law, to apply whatever rules Parliament enacts. Does it impose that duty on Parliament itself? No. Except on a narrow range of matters concerning the privileges of its own members, and leaving aside the anomalous position of the Appellate Committee of the House of Lords as a judicial body sitting within Parliament, Parliament does not make (and has no legal power to make) binding rulings on the legal effect of its own laws. It is not a law‐applying institution but only a law‐making one, and cannot be bound by the rule of recognition.36 So just as the army is identified by the 1689 Bill of Rights but not regulated by it, so Parliament is identified by this famous rule of recognition but not regulated by it. It is regulated instead by a counterpart rule of change which confers on it the legal power to pass enactments, in the process imposing, by the rule of recognition, a legal duty on law‐applying officials to apply the rules contained in those enactments. So Hart's argument to the effect that the ultimate rules of recognition of a legal system can only be customary rules does not entail that only customary rules regulate institutions of inherent power. It leaves open the possibility that in some legal systems the constitutional rules of adjudication and change—the rules that endow the institutions of inherent power with their inherent powers—could be non‐customary. Thus there remains a possible subject‐matter for a written constitution, namely the allocation of inherent legal powers to match the duties imposed by the system's ultimate rules of recognition.

Jeremy Waldron resists the idea that the (duty‐imposing) rule of recognition here is genuinely distinct from the (power‐conferring) rule of change. ‘The idea of a power,’ he writes,

is the idea of a capacity to change people's duties. So if the rule of change empowers Congress [or, we should add, Parliament] to legislate, it necessarily (p.177) enables it to do something that will change the duties of other actors in the system.37

Waldron is not quite right about the idea of a power. One may have a power, yet no ability to change any duties. One's ability may be limited to changing other powers.38 But it is true that the powers of Parliament and Congress do include the ability to change legal duties, and that someone who doesn't know this by that token doesn't understand the powers of Parliament or Congress (as the case may be). This does not show, however, that the power‐conferring rule is also the duty‐imposing one. It shows only that the power‐conferring rule presupposes the existence of the duty‐imposing one. Actually, we could go further. The duty‐imposing rule also presupposes the existence of the power‐conferring one. Yet the point remains. Two rules that presuppose each other's existence are not the same rule.

This was one of his several battles with Hans Kelsen that Hart won decisively. The power‐conferring rule enabling a court to pass sentence for careless driving is incomplete to the point of unintelligibility unless there is also a duty‐imposing rule making careless driving a criminal offence. That is because there being a criminal offence of φing is built into the very idea that φing is something that attracts a sentence.39 Conversely, a duty‐imposing rule making it a criminal offence to drive carelessly is incomplete to the point of unintelligibility unless it is joined by a power‐conferring rule enabling a court to sentence the offender for committing the offence.40 That is because there being a power to sentence for φing is built into the very idea that φing is a criminal offence. Finding just one of these rules in the law, a competent law‐applier will hold the other rule to exist by necessary implication (so that, if it does not exist already, he will have to bring it into existence himself, (p.178) or else put out to pasture the bereft rule that he started with).41 The fact that there is a necessary implication here does not entail, you can now see, that there is only one rule. In fact it entails the opposite: a relationship of necessary implication between rules can hold only if there are (at least) two rules for it to hold between.

As with the rules in a defectively drafted criminal statute, so too with a legal system's ultimate rules of recognition, change, and adjudication. They cannot but cross‐refer, and hence depend on each other for their intelligibility, yet each has its own normative force.42 Each regulates different actions, or different agents, or the same actions of the same agent in a different way. Each is therefore a distinct rule.43 So the fact that a legal system's rules of recognition must be customary rules does not show that the system's rules of change and adjudication cannot be legislated rules, laid down, perhaps, in a written constitution.

(2) A connected point. Hart is uncertain whether to classify the ultimate rules of recognition of a legal system as themselves legal rules.44 He is right to be uncertain. As he says, for some purposes and on some occasions it is harmless and natural to classify them as legal. They are, after all, rules specific to that legal system. They pertain exclusively to it. But do they quite belong to it? For present purposes it is perhaps better to think that they do not. Why? Because in a way they lie beyond the constitution. One needs rules of recognition even in order to identify the rules of the constitution. One needs to know, even of these rules, that they satisfy the ultimate criteria of legal validity for the legal system one is looking at, before one can identify them as the constitutional rules of that system. Putting the point rather paradoxically, one might (p.179) say that even the constitution needs to be constituted somehow. Is it constituted by law? Kelsen thought that it must be, and ended up facing a new version of the old problem of infinite regress, ended only by what he latterly came to call the ‘fiction’ of the validity of some historically first legal act.45 Hart avoided the same problem by presenting the ultimate rules of recognition of legal systems as borderline cases of legal rules.46 They are rules providing what Hart calls the ‘criteria’ by which law (the law of a particular system) can be recognized as law (the law of that system). But by their nature they need not themselves meet those criteria. They are found in the custom of law‐applying officials but it does not follow that they must (although they may) identify the custom of law‐applying officials as a source of law. In that sense they are above the law rather than part of it. This allows us to recognize that there are ultimate rules of recognition that are, so to speak, above the constitution while at the same time agreeing that there is no law that is above the constitution. Constitutional law is as high as the law goes. Correspondingly, a written constitution may exist even though there must be a customary rule of recognition above it, one that identifies it as the constitution, and binds the legal system's law‐appliers, qua law‐appliers of that legal system, to follow it.

Hart himself unfortunately casts doubt on this healthy way of thinking about written constitutions when he writes:

If a constitution specifying the various sources of law is a living reality in the sense that the courts and officials of the system actually identify the law in accordance with the criteria it provides, then the constitution is accepted and actually exists. It seems a needless reduplication to suggest that there is a further rule to the effect that the constitution (or those who laid it down) are to be obeyed.47

It is certainly true of legal systems with unwritten constitutions that they lack a rule of recognition with anything like this content. But Hart seems to think that the same could also be true of legal systems with (p.180) written constitutions, constitutions which are ‘laid down’.48 As Raz says, ‘the constitution, in such cases, should presumably be regarded as created both by legislation [qua written] and by custom [qua rule of recognition], a position which…needs some explaining’.49 Is it explicable? I doubt it. Of course there are examples, as we saw, of constitutions made up of some written and some unwritten law. Probably most constitutions are like this, albeit in various configurations that allow us to think of some as basically unwritten and others as basically written. But that is beside the point. The picture that Hart seems to be trying to conjure up is of a constitution containing some law that is both written and unwritten—legislated and customary—at the same time. The only picture he succeeded in conjuring up for me, however, is of written law which was displaced, perhaps one step at a time, by unwritten law, so that the formerly legislated constitution lost its force in favour of customary rules with similar content. Hart may have thought that this is what happens when the law of a written constitution is developed over time by the courts. The written constitution eventually becomes a dead letter, referred to only honorifically. As we will see towards the end of Section IV, however, the ‘living reality’ of a written constitution calls for a different analysis, one consistent with and indeed conducive to the view that the rule of recognition is indeed a customary rule lying beyond the constitution itself, a rule ‘to the effect that the constitution (or those who laid it down) are to be obeyed’. As I just pointed out on Hart's behalf in reply to Waldron, this is not ‘needless reduplication’ of the rules in the written constitution but a separate rule of recognition without which there is no written constitution to contain those rules.50

(p.181) (3) Finally, my argument for regarding the UK Parliament and High Court as institutions of inherent rather than delegated power rested in part on the proposition that their powers are not, as the constitutional law of the UK now stands, revocable. The monarch cannot lawfully step back in and reclaim the powers that, in the middle ages, were delegated by her predecessors to Parliament and to (what is now) the High Court. This suggests a possible way of thinking about constitutional powers which is consistent with their having been endowed by another through legislation, and consistent with constitutional law recognizing that endowment as the source of the powers. One may say that it is possible for constitutional institutions to have been endowed with those powers by a higher legislative institution so long as that institution cannot revoke them. The easiest way for that condition to be met is for the institution in question no longer to exist. This is indeed standard practice in constitution‐building today. A temporary constitutional caucus or assembly is conjured up which then endows constitutional powers upon other institutions designed to be permanent. The law then recognizes that the power was endowed but adds that the method of endowment is not reusable, within the constitution, as a method of revocation, for the delegating body has wound itself up or has been wound up. The delegation is rendered irrevocable. This suggests a possible revision of my original proposal for determining the scope of constitutional law, as distinct from administrative law. Constitutional law regulates those institutions that, according to the law, have either inherent or irrevocably delegated powers. Over time institutions with irrevocably delegated powers may come to be regarded, in law, as having inherent powers. But strictly speaking it need not be so.

The US constitution illustrates that it need not be so. It is a written constitution. It is a piece of legislation. The legislator was a temporary institution, a Convention of delegates (yes, delegates) from twelve states meeting at Philadelphia in 1787. With only the ratification of subsidiary ad hoc Conventions in the States, this Convention created the standing governmental institutions of today's federal United States, including notably (but not only) the Presidency, Congress with its twin houses, and the Supreme Court. The powers with which these institutions were (p.182) endowed were delegated to them by the 1787 Convention, which held itself to hold all the powers of the new Federal legal system that it was creating, including the power to delegate those powers.51 However the Convention did not make provision for its own future existence. Instead it made alternative provision for later amendments to the constitution, in which the amenders would be institutions created by the constitution itself, operating under special procedures designed for the purpose. Thus all the inherent powers that the Convention took itself to have, and is taken by current federal US law to have had, were delegated away to the new institutions that it created. The Convention itself was wound up, putting an end even to its own legal power to reconvene and hence to revoke the delegation.

Or so one version of the story goes. We are straying into another political minefield with this vignette of US constitutional history. Twin amendment procedures are set out in Article V of (the 1787 enactment now known as) the United States Constitution. The first is the well‐known and much‐used power of Congress to propose amendments (by a two‐thirds majority) which take effect upon ratification by three‐quarters of the States. The second is this less well‐known one:

The Congress…on the Application of the Legislatures of two thirds of the several States, shall call a Convention for proposing Amendments, which…shall be valid to all Intents and Purposes, as Part of this Constitution, when ratified by the Legislatures of three fourths of the several States or by Conventions in three fourths thereof.

According to some, this is a power to reconvene the original 1787 Convention.52 It is irrelevant to this claim that an Article V Convention (p.183) has to be called by Congress on the application of the States, and that its proposals then have to be ratified by the States. The 1787 Convention was itself called under the old Articles of Confederation by Congress on the application of the States, and its product—the constitutional enactment in which Article V appears—was also ratified by the States before coming into effect. It is of equally scant assistance to note that an Article V Convention is limited to proposing amendments to take effect ‘as Part of this Constitution’. These words obviously throw up the classic problem of the identity of wholes with fungible parts, which threatens to render them vacuous.53 But even if they add something, there is another problem. Does ‘this Constitution’ refer to the institutional arrangements of the US constitution or does it refer to their canonical formulation in the 1787 enactment as amended? This is important because the 1787 Convention preserved parts of the institutional set‐up from the Articles of Confederation under which it was summoned (e.g. the existence of Congress, the constitutional recognition of the States as ratifiers). So the 1787 Convention too can be argued to have enacted the new constitution ‘as Part of’ the old (never mind that the new then took on a life of its own and was held, legally speaking, not to owe its validity to the old). In which case there is nothing in these words to distinguish an Article V Convention from the 1787 Convention.

It is true, of course, that an Article V Convention can only amend—taking ‘amendment’ to designate a lawful mode of constitutional (p.184) change—if it acts within its constitutional powers. But that only brings us back to our original question. What are those constitutional powers? Are they delegated powers or are they a continuation of the inherent powers by which the 1787 Convention itself is now regarded as having acted? Does the US constitution recognize and require the continuing existence of its own super‐legislator, a legislator which, although dormant, could be stirred back to consciousness by Congress on the application of two‐thirds of the States and could validly revoke—and revoke, perhaps, to tabula rasa except for its own continuing powers—the various legal powers conferred upon the institutions created at its own previous meeting in 1787?54

The answer is nowhere close at hand. If the triggering condition is met—if a valid application is received from the States—it is probably mandatory under Article V for Congress to call a Convention.55 But whether the triggering condition has ever been met is a matter of some dispute. It depends on how State applications are to be individuated and counted. When States apply severally rather than jointly, perhaps with differently worded and differently scoped applications many years apart, are their various applications to be aggregated until the two‐thirds line is reached? Or, as most constitutional lawyers assume, is more coordination or convergence among States required before their petitions come together to qualify as an application under Article V? Whatever the answer, Congress has never called an Article V Convention, and the Supreme Court has never ruled on the legality of Congress's not having done so. Nor is it clear that the Supreme Court would rule on this matter if it were petitioned. The Supreme Court has ruled in the past that other Congressional decisions concerning the amendment process under Article V are non‐justiciable, i.e. not subject to the court's (p.185) authoritative rulings on their constitutionality.56 Other federal courts have extended the same rule to the calling of an Article V Convention.57 This may mean that constitutional questions about the calling of an Article V Convention are non‐legal questions. They are to be settled elsewhere in the political process, regulated not by law but at most by what Dicey might have called the conventions of the US constitution.58 This leaves us several frustrating steps away from achieving any legal determinacy on the vexed question of whether an Article V Convention, were one to be called, would have the original powers of the 1787 Convention or merely delegated powers conferred by or under the 1787 Convention.

Can't we find the answer to this vexed question in another way? Surely US constitutional law is by now amply determinate in classifying the Presidency, Congress, and the Supreme Court as constitutional institutions. If the 1787 Convention were merely sleeping, wouldn't the powers of these institutions have to be reclassified as revocably delegated powers, and hence as non‐constitutional, powers? In which case—shock, horror!—these famous institutions would not be constitutional institutions after all. Indeed the 1787 constitution (the document that calls itself ‘The Constitution of the United States of America’) would no longer qualify as the US constitution, for now it would be merely repealable delegating legislation made under the pre‐1787 constitution. And most of what is known in US law schools as ‘constitutional law’ would be nothing of the kind.

Things are getting out of hand. A power should not be classified as revocable, for our purposes, merely because its revocation is imaginable. Even if Congress and the Supreme Court together control and always conspire to deny access to the only process by which their own powers could imaginably be revoked, then those powers, albeit delegated, are irrevocable enough to qualify as constitutional powers. At this point (p.186) attention returns to the questions of loyalty—the loyalty of petty officials and the loyalty of the wider population—which we touched upon in Section I. Unflinching refusal of Congress and the Supreme Court to grant an application for an Article V Convention could imaginably be overcome, but only by mass defection to rival institutions. If we regard the 1787 constitution—the one headed ‘The Constitution of the United States of America’—as the true US constitution, then that mass defection would qualify as a new American revolution. If, on the other hand, we deny that the 1787 document is the true US constitution, treating it as a mere delegating act, then such mass defection would arguably be no more than a coup.59 Either way, however, it would be a usurpation of legal powers under the constitution, for on either view it is only by usurpation that the power to call an Article V Convention can be wrestled from the hands of a peristently refusnik Congress and Court. And once the possibility of usurpation has been opened up, we are no longer talking about amendment, understood as a lawful mode of constitutional change. We are changing the subject.60

We may conclude, with a sigh of relief, that the Presidency, Congress, and the Supreme Court are not mere administrative bodies regulated by a jumped up kind of administrative law. They are constitutional bodies and they are constitutional bodies because their powers, although delegated to them and hence not strictly speaking inherent or original, are delegated to them irrevocably by the 1787 constitution, as amended. This turns out to be true irrespective of whether an Article V Convention, if called, would be a reawakening of the 1787 Convention.

We have taken a long detour into the dimmer recesses of the US constitution. The point was mainly to illustrate a realistic possibility. The possibility is that the ultimate rules of recognition of a legal system may identify different institutions from those mentioned in its ultimate rules of adjudication and change. This seems to be the case in the US. The ultimate rules of recognition of the federal US, like all ultimate (p.187) rules of recognition around the globe and throughout history, are indeterminate in numerous respects.61 But federal US law is by now entirely determinate in treating the 1787 constitution, as amended, as its constitution. This is the custom of the Supreme Court and of all the other federal courts and of other authoritative law‐applying agencies. There is little doubt, then, that one of the ultimate rules of recognition of the US says something like: ‘Whatever is laid down in the constitution produced by the 1787 Philadelphia Convention, as amended, is law.’ Notice that this rule does not mention Congress, the President, or the Supreme Court. They are mentioned for the first time only in the rules of change and adjudication that are in the 1787 constitution.

So constitutional authorities—to go back to our original problem—need not be identified in the system's ultimate rules of recognition, even in that rule of recognition by virtue of which they are constitutional authorities. If the constitution is a written one, only (the author of)62 the written constitution need be identified in the relevant rule of recognition, while the main constitutional institutions are those identified in the written constitution. By this route too we find that a customary rule of recognition can be squared with a written constitution.


What I have said so far was designed to fend off a certain set of objections to the idea that constitutions can be written. It did so by tweaking the scope of constitutional law to cover irrevocably delegated powers as well as inherent or original powers, by distinguishing the rules that identify a constitutional institution from those that regulate it, and by placing the rule of recognition of a legal system outside (above) the constitution itself. Actually, to be more exact, we only placed the rule of recognition above the law of the constitution. We should always keep in mind the important warning at the end of Section I and repeated more (p.188) than once already. Constitutions are not exhausted by their law. In every country with a constitution, and hence in every country with a legal system, there are also constitutional rules which are distinguished from the rest of the constitution precisely in being rules of which the authoritative law‐applying institutions do not get to make authoritative applications. This makes space for there to be actions which are unconstitutional but which are neither illegal nor legally invalid. The role of this fact may vary from legal system to legal system. In the UK it is a fairly prominent aspect of the constitution. In the US it is less prominent, but—as we just saw in subsection II.3—it is certainly not absent.

Is an ultimate rule of recognition one of those rules of which the law‐applying institutions do not get to make authoritative applications? Is it in that way (although plainly not in other ways) akin to a Diceyan ‘convention of the constitution’? Hart seems to have thought as much.63 Of course, he insisted that every ultimate rule of recognition is a rule made by the authoritative law‐applying institutions of that legal system of which it is a rule of recognition. But he thought that it was made by the customs of the officials of those institutions in the course of exercising their authority over other things, not in exercising authority over the ultimate rules of recognition themselves.64 Authoritative law‐applying institutions, typically courts, get to decide cases authoritatively. In all but the most rudimentary legal systems, when they decide cases authoritatively the higher courts (not only and not always courts of inherent jurisdiction) also get to make a kind of law known as case law. This is a kind of unwritten law that is found in the premisses of the arguments that higher courts use to arrive at their authoritative decisions. It differs in several ways from their customary law in foro.65 For a start, case‐law can be made by one law‐applying official (or one committee of such officials) in one case—by a single exercise of judicial authority—whereas customary law in foro requires for its existence a temporally extended pattern of relevantly convergent behaviour by multiple law‐applying officials. Why couldn't an ultimate rule of recognition be made by the case‐law method instead of by the customary‐law method? Hart did not explain. Maybe he did not appreciate that case (p.189) law differs from both legislated law and customary law. Certainly he tended to speak of judge‐made law, when not customary, as legislated.66 Subsequent discussion of the issue has been hampered not only by repetition of this error, but also by the tendency (to which, as we saw, Hart also gave some succour) to confuse rules of recognition on the one hand with rules of adjudication and change on the other.

It would take us too far afield to pursue these problems about the rule of recognition here. But our remarks about judge‐made law, including judicial contributions to the constitutional rules of adjudication and change, provoke a new question about written constitutions. Perhaps there can be written constitutions—I have just argued that there can— but can there be entirely written constitutions? To make the issue less complex I will set aside those parts of the constitution that are not constitutional law (the Diceyan ‘conventions of the constitution’) as well as the ultimate rules of recognition of the system. So my question is this: Can (the rest of) constitutional law be entirely written? Or must it always also include some customary law or case law?

The answer seems plain enough. On the day it is enacted a new constitution is wholly written law. But that day does not last. As soon as ripe disputes begin to arise that concern the meaning of constitutional provisions, written constitutional law inevitably needs to be filled out with case law and/or customary law. What is written in the constitution needs to be invested with more determinate meaning, and by and large this has to be done at the point of its authoritative application, principally by judges. With the passage of time, such judge‐made law tends to predominate over the parts of constitutional law that exist apart from it. With the passage of time, one knows an ever‐smaller proportion of the law of the constitution—or at any rate an ever‐smaller proportion of the material that goes to make up the law of the constitution—simply by reading the constitutional text.

Like other things we have said, this plain answer may strike some as political explosive. It may seem to lead us straight into the big controversies of contemporary US constitutional law. In the red corner, (p.190) ladies and gentlemen, those who insist that the constitution is found in what was written and nothing but what was written by the 1787 founders and their authorized legislative amenders. And in the blue corner, those who say that the constitution is a living body of law and should not be regarded as frozen in time at the moment of enactment. Is this the fight we are getting into? No. Nothing I have just said takes sides in this or any other debate about how the US constitution should be interpreted. I do not doubt, of course, the profound political significance of such debates and—especially when the disputants are Supreme Court judges—their huge potential consequences for the future direction of America. My only point is this. Inasmuch as these debates have huge potential consequences for the future direction of America, that is because they have huge potential consequences for the future development of American constitutional law. And they have huge potential consequences for the future development of American constitutional law precisely because, in both corners, we have people with proposals for how American constitutional law should be developed. Both sides are assuming that it will be developed, and that it will be developed by judges. The only question is, how will it be developed? Which way are the judges to take it? Both sides—or since it is not really a two‐corner fight, I should really say all sides—must be in agreement that US constitutional law is not just what is contained in the text of the written constitution. For all of them it must also include judge‐made law. For if it did not include judge‐made law there would be no point in fighting over how judges should make constitutional decisions. The only possible reason for choosing a textualist Supreme Court nominee over a purposivist, or an originalist‐textualist over a strict‐constructionist‐textualist, or a original‐intent‐originalist over an original‐meaning‐originalist, or indeed a baggist over a raggist, is that each of them, or at any rate each of them in combination with some like‐minded judges, will have the power to change the law of the constitution by giving the constitution a meaning different from the one that it would have under the authority of a judge or a combination of judges from some rival camp.

Aren't there a few elaborate theoretical positions, in these debates over constitutional interpretation, that genuinely include the proposition that judges don't change the law? Doesn't Ronald Dworkin famously say exactly that? At any rate, doesn't he say that when judges (p.191) get their decisions on points of constitutional law right—when they give ‘right answers’—all they are doing is applying the law that is already in the constitution, albeit maybe implicitly rather than explicitly?67 I doubt whether Dworkin still holds this view.68 Possibly he never held it.69 But consider its implications. Its implications include that judges only change the law of the constitution when they get their decisions wrong. It follows that, according to this view, constitutional law may only ever be changed for the worse by judges, for each change necessarily introduces yet another error. Better, then, if the law doesn't change at all. But then if it should never change at all, judges should never have added all the case law that they added during the couple of centuries that they have done so. Better if they had answered each question of law by saying: ‘Just use what you already have, it's all there. Don't come bothering us.’ On this reading of his ‘right answer’ thesis, Dworkin does not become a mere closet textualist. He departs even from the mainstream textualist view in holding that judges should not develop US constitutional law in either a textualist or an opposing, more innovative, direction. They should not develop US constitutional law at all. At each moment they give the ‘right answer’ only if they leave the law exactly as it is, complete with all the previous accumulated errors.

This is a crazy view, which explains why I am reluctant to ascribe it to Dworkin. Even if he once held this crazy view, however, it would have been hard for him to deny that (worse luck!) there has been a huge judicial contribution to US constitutional law. Even if the judges should not have added to it, they have added to it. Thanks to them there is a lot more of it now than there was in 1790, or in 1832, or in 1896, or in 1926, or even in 1964. And since judges are fallible human beings like the rest of us it had to be that way. So even if one says, crazily, that judicial law‐making is always erroneous, one cannot avoid reaching the same result: Any constitution that provides for authoritative adjudications regarding (p.192) its own application cannot but be to some extent a living constitution, i.e. cannot but contain less law at its inception than it comes to contain later.

It is tempting to conclude from all this that a constitution cannot be entirely written. From soon after its birth, it seems, the written constitution must constitute only part of the constitution, the rest being made up of judge‐made law.70 But does that follow? Let me suggest two ways in which it is apt to mislead.

First, possibly the law of the constitution (or constitutional law) should not be thought of as identical with the constitution itself, even if we restrict our attention to the parts of the constitution that form part of the law. We can bear this thought out somewhat by considering two ways of responding to the question: What does the constitution have to say about this? One response would be to hand over the document, the canonical text, assuming one is in a country that has one. A rival response would be to mention a case, a judicial decision in constitutional law, which is relevant to the issue. Both are intelligible responses but their rivalry shows that there are, where constitutions are concerned, at least two rival objects of interpretation. There is the constitutional document, the text, as one possible object of interpretation. And then there is constitutional law, the rules, as another possible object of interpretation. They are not the same but they often bear the same name. Many confusions in constitutional theory come of a failure to clarify which object of interpretation is at stake in which debates. I hazard a guess that much of the debate between the different ‘‐isms’ of US constitutional interpretation is crippled by such confusions, and would be better abandoned and restarted in entirely different (and I dare to hope, less philosophically pretentious) terms.

Second, one possible link between the two possible objects of interpretation just mentioned—constitutional law and the canonical constitutional text—is that the former may present itself as an interpretation of the latter. In other words interpreting constitutional law is partly a second‐order activity in which one interprets the attempts and claims of others, mainly judges, to be interpreting a canonical text. (p.193) Interpreting a text is explaining (or exhibiting) some meaning that it has.71 This truth about interpretation has led some to think that, inasmuch as constitutional case law has interpreted the constitutional text to include norms that are not already part of its meaning, that cannot count as interpretation. But it is no part of the concept of interpretation that the meaning one explains by interpreting must already be part of the text before one so interprets it. One may also interpret a text by giving it some meaning, such that it has that meaning from now on in virtue of one's having given it.72 There are undoubtedly moral and political questions about how innovative, or how retrievalist, one should be on a certain interpretative occasion, but none of these is settled by the concept of interpretation. This suggests that we can afford to take a more expansive view of what a written constitution, and written law more generally, is. We can include under the heading of ‘written law’ both the text and its meaning (for without meaning there is no law in the text) and we should include in its meaning whatever meaning it has, legally speaking. This includes meaning that was authoritatively extracted from it by interpretative retrievers as well as meaning that was authoritatively attached to it by interpretative innovators.

Of course there may be some meanings that were authoritatively attached to it and then authoritatively removed (by the judicial overruling or distinguishing of an earlier case, or by a change in judicial custom). But until such a removal takes place, all the meaning is there. To the extent that it conflicts, and those conflicts are not resolved by legal rules for ranking the conflicting interpretations (such as rules of stare decisis), it leaves the law of the constitution correpondingly indeterminate. The indeterminacy comes not of there being too little constitutional law, but of there being too much. Not only are there inconsistent but coexisting legal rules about judicial review of legislation, constitutional amendment, the domestic reccognition of international law, and so on. There are also inconsistent but coexisting legal (p.194) rules for determining what the constitution says about such things, i.e. for interpreting the constitution.73

Thinking this way puts controversies about interpretative technique in their place. They are straightforward moral and political debates about what judges should do when they interpret, debates which concede on all sides that whichever way judges lean—more retrievalist or more innovative—they change the meaning of the text accordingly. For our purposes the most startling implication of this is also a simple one. Where there is a written constitution, there is no logical obstacle to the whole law of the constitution being written law. For there is no logical obstacle to its all being contained in the text, either because it was found there by subsequent judgments of courts or because it was put there by subsequent judgments of courts. Written constitutions, in short, may be entirely written constitutions, for their developments in case law, by way of interpretation, cannot but become part of their meaning qua written.74


This chapter is dedicated to the memory of Neil MacCormick 1941–2009. A very early draft was presented as a public lecture to celebrate the 25th Anniversary of Southern Methodist University's Law Program in Oxford. Later drafts of the paper were presented at the Universities of Palermo and Girona. Many thanks to my hosts and audiences on these occasions for inspiring discussion. Thanks especially to John Attanasio of SMU for valuable points that are taken up in subsection III.3 below. In finalizing the text I also benefited from the detailed and expert comments of Nick Barber, Les Green, and an anonymous reviewer.

(1) The Concept of Law (Oxford, 1961), 95–6. Hereafter abbreviated as CL.

(2) For a predictably excellent theoretical treatment, see Neil MacCormick, Questioning Sovereignty (Oxford, 1999), ch 4. For a similarly excellent discussion of the like problem of the individuation of legal systems in the European Community, see Julie Dickson, ‘How Many Legal Systems? Some Puzzles Regarding the Identity Conditions of, and Relations Between, Legal Systems in the European Union’, Problema 2 (2008), 9.

(3) For the now‐abandoned constitutional text see Official Journal of the European Communities C310/01 (2004). Mads Andenas and I tried to introduce some of the theoretical issues underlying the political debate in ‘Can Europe Have a Constitution?’, King's College Law Journal 12 (2001), 1—in particular the idea of a ‘capital‐C Constitution’, which I will not make use of here.

(4) Contrast Neil MacCormick, ‘Does the United Kingdom Have a Constitution? Reflections on MacCormick v Lord Advocate’, Northern Ireland Legal Quarterly 29 (1978), 1.

(5) Thus the grandly‐titled Constitutional Reform Act 2005 has yet to establish its constitutional credentials. Likewise the Constitution Act 1982, which did not have constitutional effect at the time of its enactment but only later: see Manuel v Attorney General [1983] Ch 77, discussed below n. 8.

(6) I say ‘constituted’ rather than ‘recognized’ because arguably the European Commission, which is constituted by the European Community constitution not the UK constitution, has nevertheless come to be recognized by the UK constitution as a primary legislator whose enactments are capable of binding Parliament. See Factortame v Secretary of State for Transport [1991] AC 603 at 658–9 per Lord Bridge for a strong judicial statement to that effect.

(7) Which is not to say that no other primary legislator is constituted by the UK constitution. There is also the Privy Council, which makes Orders in Council as a primary legislator under (what is left of) the Royal Prerogative: Council of Civil Service Unions v. Minister for the Civil Service [1985] AC 374 at 399, per Lord Fraser of Tullybelton. I say ‘what is left of’ because these Orders do not bind Parliament. They may be and often have been countermanded and superseded by Acts of Parliament, and with them those parts of the Royal Prerogative to which they pertained. There are also, a separate matter, Orders in Council that serve as delegated legislation under Acts of Parliament.

(8) See Manuel v Attorney General [1983] Ch 77 at 87–9 per Megarry VC for interesting obiter discussion of the second question. At 89, Megarry VC distinguishes what is it is possible for Parliament to do ‘as a matter of abstract law’ from what it is possible to do ‘effectively’. One reason why the question is imponderable, however, is that this distinction collapses too quickly. There must be a supporting practice (a measure of effectiveness) among law‐applying officials before a legislature can do anything, even as a matter of abstract law. Here, as Hart rather cryptically put it, ‘all that succeeds is success’: CL, 149.

(9) Thoburn v Sunderland City Council [2002] QB 151 at 186–7 per Laws LJ; Jackson v Attorney General [2006] 1 AC 262.

(10) How does the express/implied distinction relate to the deliberate/accidental one? I am assuming that Parliament always has the ability to say what it means, so that if it intends a certain change not to take place, it may always say so. For further discussion of these distinctions as applies to legislated law, as well as to case law, see my ‘Some Types of Law’ in Douglas Edlin (ed.), Common Law Theory (Cambridge, 2007).

(11) Political Constitutionalism: A Republican Defence of the Constitutionality of Democracy (Cambridge, 2007), 5. Bellamy's main aim in this book is to defend the UK doctrine of Parliamentary Sovereignty against those who would like there to be a judicial power to invalidate held‐to‐be‐odious provisions of Acts of Parliament. This debate about ‘constitutional judicial review’ will be of no concern to us here. It is only very indirectly related to the question of how much of the constitution should be legalized. And each of these questions is only very indirectly related to the question of whether the UK could or should have a written constitution. A written constitution is not the same as, and need not include, a bill of rights, let alone one that allows for judicial invalidation of primary legislation. Bellamy seems to mix these questions up.

(12) Ibid. 5, citing Martin Shapiro's ‘Political Jurisprudence’, Kentucky Law Journal 52 (1964), 294.

(13) A revisionist view of the doctrine, defended by Ivor Jennings in The Law and the Constitution (London, 1959) and by R. V. Heuston in his Essays in Constitutional Law (London, 1964), and recently revived by Jeffrey Goldsworthy in The Sovereignty of Parliament (Oxford, 1999) would have it that sovereignty is partly ‘self‐embracing’: Parliament can bind its successors as to the ‘manner and form’ of future legislation but not as to its content. I doubt whether UK law draws or ever drew such a distinction, but my formulation of the doctrine here is intended to remain agnostic on the point.

(14) In Practical Reason and Norms (London, 1975), at 134–7, Joseph Raz calls the relevant law‐appliers the ‘primary law‐applying organs’ of the system.

(15) In an unfortunately much‐quoted passage, Hart boiled all this down to ‘a complex…practice of the courts, officials, and private persons’ (CL, 107). But as he promptly made clear in less‐quoted passages (110–11, 113), he was referring to several related practices. We should distinguish (a) the courts’ practice of applying certain rules as rules of law; (b) a practice among petty officials of submitting to the rulings of the courts; and (c) a practice of the wider population of submitting to, or at least not defying, those petty officials (whether because they are officials of law or otherwise). If (b) and (c) go, so too does (a): without widespread efficacy, these are no longer the courts, and this is no longer the legal system.

(16) As Dwight Eisenhower famously showed on 24 Sept. 1957 in Little Rock, Arkansas, invoking powers under Article I of the US constitution, together with the might of the 101st Airborne Division, to compel the fidelity of the Arkansas National Guard to the order of a federal judge. A perhaps less heroic example, from UK colonial law and policy, is Madzimbamuto v Lardner‐Burke [1969] 1 AC 645, some of the background of which is discussed in H. H. Marshall, ‘The Legal Effects of UDI (based on Madzimbamuto v Lardner‐Burke)’, International and Comparative Law Quarterly 17 (1968), 1022.

(17) A. V. Dicey, Introduction to the Study of the Law of the Constitution (London, 1885), 23–4.

(18) R v Secretary of State for the Environment ex parte Nottinghamshire County Council [1986] AC 240 at 250–1 per Lord Scarman, where this self‐denial is itself elevated to the status of convention of the constitution.

(19) Or that they regard the constitution as the most socially important, let alone socially desirable, part of public culture. Arguably Bellamy and others with similar views decry excessive legalism about the constitution only to fall into a similar trap at the next level by being excessively constitutionalist about politics, or at the next level still by being excessively political about life.

(20) Both these terms also have other meanings. In some contexts, e.g. in Article III of the Constitution of the United States, ‘original jurisdiction’ is contrasted with ‘appellate jurisdiction’. In the sense intended here, however, an appellate jurisdiction may also be an original one. It is original, or inherent, in being a jurisdiction ‘which, as the name indicates, requires no authorizing provision’: R v Forbes ex parte Bevan (1972) 127 CLR 1 at 7 per Menzies J.

(21) The example of the Privy Council shows that some institutions fall under both headings. Depending on the subject‐matter, the Privy Council acts either as a constitutional institution, making primary legislation under the Royal Prerogative, or as a delegate institution, making secondary legislation on the Parliamentary authority of the Statutory Instruments Act 1946 (see n. 7 above).

(22) The Court was created from scratch by the Judicature Act 1873. See Lord Justice Cohen, ‘Jurisdiction, Practice and Procedure in the Court of Appeal’, Cambridge Law Journal 11 (1951), 3. A criminal jurisdiction was added by the Criminal Appeal Act 1966, transferring the powers of the Court of Criminal Appeal, itself created by the Criminal Appeals Act 1907: R v Collins [1970] 1 QB 710 at 713–14; R v Shannon [1975] AC 717 at 745–8 and 756–7.

(23) Here I once again stress that we are talking about constitutional law. The humble legal position of the police (which is below the constitutional radar) is counteracted by powerful constitutional conventions (which lend the police a vast constitutional importance). For brilliant analysis, see Geoffrey Marshall, Constitutional Conventions (Oxford, 1984), ch. 8.

(24) These days the armed forces are predominantly regulated by Act of Parliament. However the power to deploy troops, and to declare war or peace, still belongs to the Royal Prerogative, albeit with an emerging constitutional convention that Parliament's consent is required for deployment: House of Lords Select Committee on the Constitution, Waging War: Parliament's Role and Responsibility, vol i (HL 236.1, London 2006), 34–5.

(25) Canada Trust Co v Stolzenberg [1997] 4 All ER 983 per Millett LJ at 989.

(26) John Laws, ‘Illegality: The Problem of Jurisdiction’, in M. Supperstone and J. Goudie (eds.), Judicial Review (1st edn., London 1991), 69–70. As Hart shows, using the UK doctrine of Parliamentary Sovereignty to illustrate, there is no such thing as an institution with unlimited powers: CL, 145–6. (But note that ‘unlimited jurisdiction’ is used by many lawyers as a misleading but otherwise innocent synonym for what I am calling inherent or original jurisdiction. See, e.g. R (on the application of Cart et al) v The Upper Tribunal et al [2009] EWHC 3052, per Laws LJ.)

(27) In this formulation I am guarding against two errors. (a) As in Hart's most careful formulations (e.g. CL, 112) I speak of ultimate rules of recognition because there are plenty of (here irrelevant) lower‐level rules of recognition in every legal system, e.g. those identifying a police officer as an issuer of legally binding traffic signals, those identifying a local council as the source of legally binding planning decisions, etc. (b) I also speak of ultimate rules of recognition (plural). Hart sometimes suggested that each legal system has only one ultimate rule of recognition. Not so. All but the most rudimentary legal systems have several ultimate rules of recognition, the inevitable conflicts between which may well come to a head only on rare occasions when they have implications lower down the system. Hart sometimes suggested that there could be no such conflicts. There must be a transitive ranking of the various validity‐criteria of each legal system, and hence a single rule providing the ranking: CL, 103. However Hart is not entirely consistent about this. See CL, 92 for talk of a system's ‘rules of recognition’ in the plural and the concessionary remark that ‘provision may be made for their possible conflict’ (emphasis added). For the purposes of this paper I will overlook Hart's apparent indecision on this point and talk as if he held the correct pluralist view. However most of what I say could be adapted to sit no less comfortably with the rival monist view.

(28) Jeremy Bentham, A Comment on the Commentaries and A Fragment on Government, ed. Burns and Hart (London, 1977), 182–4.

(29) CL, 103–7.

(30) See e.g. Laurence H. Tribe, ‘Taking Text and Structure Seriously: Reflections on Free‐Form Method in Constitutional Interpretation’, Harvard Law Review 108 (1995), 1221 at 1246–7; Charles Fried, ‘Foreward: Revolutions?’, Harvard Law Review 109 (1996), 13 at 26 n. 66; Frank I. Michelman, ‘Constitutional Authorship’, in Larry Alexander (ed.), Constitutionalism: Philosophical Foundations (Cambridge, 1998) at 70–2. For a non‐reductio use, see Michael Dorf, ‘Con Law in 12,008’, 〈http://www.michaeldorf.org/2008/06/con‐law‐in‐12008.html〉.

(31) See e.g. CL, 95, where Hart spoke of a rule of recognition ‘confer[ring] jurisdiction’. For more analysis of this passage, see n. 43 below.

(32) CL, 95: the ‘heart of a legal system’ is ‘the combination of primary rules of obligation with the secondary rules of recognition, change, and adjudication’.

(33) CL, 93 and 94.

(34) Joseph Raz, The Concept of a Legal System (Oxford, 1970), 198–9; Neil MacCormick, H. L. A. Hart (London, 1981), 103–6.

(35) Kent Greenawalt reports that, in correspondence, Hart confessed to ‘a slip’ in presenting this as a complete statement of the rule, omitting the further criteria concerning case law, customary law, and other types of primary legislation: Greenawalt, ‘The Rule of Recognition and the Constitution’, Michigan Law Review 85 (1987), 621 at 631 n. 30. But this was not Hart's slip. He was right to regard this as a complete statement of the rule. He was wrong, however, to regard this rule as the UK's only ultimate rule of recognition, for there are several other rules relating to case law, etc. See n. 27 above.

(36) It does not follow that Parliament cannot be bound in its law‐making by other duty‐imposing laws, including duty‐imposing laws of the constitution. However, the only examples I know of are found in European Community Law that is incorporated into UK law, e.g. the compensation duty in Joined Cases 46/93 & 48/93, Brasserie du Pècheur SA v Germany and R v Secretary of State for Transport ex parte Factortame Ltd (1996) 1 CMLR 889. The situation where Parliament ‘binds’ its successor Parliaments, if there can be such a situation, is not a case of duties imposed but of powers removed.

(37) Waldron, ‘Who Needs Rules of Recognition?’, in M. Adler and K. Himma (eds.), The Rule of Recognition and the US Constitution (New York, 2009), 343.

(38) Or to granting permissions. Not every grant of permission affects the incidence or force of a duty. Instead it may conflict with a duty. For more discussion of powers to empower and permit, see my ‘Justification under Authority’, Canadian Journal of Law and Jurisprudence 23 (2010), 71.

(39) In Hart's terms, the one idea ‘involves’ the other: CL, 39. At this point Hart is dismantling Kelsen's view, set out in General Theory of Law and State, trans. Wedberg (Cambridge, Mass., 1945), 53–4, that there is only one legal norm here, and that what looks like a second is merely a fragment of the first.

(40) CL, 35–41.

(41) A common mistake is to think that every rule, the existence of which is entailed by a legal rule is also a legal rule. Why is this a mistake? See J. Raz, ‘Legal Rights’, Oxford Journal of Legal Studies 4 (1984), 1 at 9–12.

(42) See MacCormick, H. L. A. Hart, 108–11, showing that a search for logical priority between the three types of rules is ill‐fated. This led MacCormick to suspect a vicious circularity. But he later came to see that the suspicion was unfounded: see his H. L. A. Hart (2nd edn., Stanford, 2008), 151.

(43) This returns us to the passage at CL, 95 where Hart says that there may be a rule of recognition that is ‘also’ a rule of adjudication conferring jurisdiction on some law‐applying body. This suggests identity, not entailment. However Hart quickly goes on to speak, more carefully, of two ‘inseparable’ rules, one a rule of recognition, the other a rule of adjudication.

(44) CL, 108.

(45) Kelsen, General Theory of Norms (Oxford, 1991), 256. I am not denying that Kelsen's fictitious basic norm might supply a good answer to some other philosophical question. Possibly it helps us to understand the normativity of law even though it fails as an attempt to explain the possibility of legal validity.

(46) CL, 108.

(47) CL, 246.

(48) He says that it is ‘particularly clear’ of unwritten constitutions, not that it is particular to them: CL, 246.

(49) Raz, The Concept of a Legal System, 198.

(50) It is worth remembering that when he writes these words Hart is bending over backwards to distinguish his ultimate rules of recognition from Kelsen's basic norm. He may well be bending too far. He rightly points out that Kelsen's basic norm always has the same content, roughly: ‘one should always and only obey the historically first constitution’. Hart is right that, unlike this basic norm, ultimate rules of recognition have diverse content, varying from time to time and from legal system to legal system: CL 245–6. It does not follow, however, that there are no ultimate rules of recognition anywhere with content akin to that of the Kelsenian basic norm. Nor would the redundancy of this content qua content of a basic norm entail its redundancy qua content of an ultimate rule of recognition, since a Kelsenian basic norm and an ultimate rule of recognition do different jobs and are not rivals for the same explanatory space. See Stanley Paulson, ‘Christian Dahlman's Reflections on the Basic Norm’, Archiv für Rechts und Sozialphilosophie 91 (2005), 96 at 105.

(51) Although not of course all the powers of the various state legal systems, all of which have their own ultimate rules of recognition and change and adjudication. See Greenawalt, ‘The Rule of Recognition and the Constitution’, at 645–7. Curiously, having explained that state institutions are not delegates of federal institutions, and so enjoy recognitional independence of federal law, Greenawalt goes on to formulate a single list of criteria of recognition for ‘the American legal system’ as seen from ‘someplace within the United States’, consolidating federal and state criteria: ibid. 659. He also presents these criteria as adding up to a single rule of recognition, so perhaps (like Hart, see above n. 27) he resists the idea that there could be conflicts between ultimate rules of recognition; perhaps he also extends that resistance so that it applies not only to the rules of recognition of a single legal system but also to the rules of recognition of multiple legal systems applicable in the same territory, such as state and federal US law.

(52) The discussion is generally cast as one about the terms of reference of an Article V Convention. Can it be limited as to subject‐matter of amendment or must it have a roving brief? If the latter, what stands (constitutionally) in the way of a full Constitutional Convention like that of 1787? For defence of the ‘roving brief’ view see Charles Black, ‘Amending the Constitution: A Letter to a Congressman’, Yale Law Journal 82 (1972), 189; Walter Dellinger, ‘The Recurring Question of the “Limited” Constitutional Convention’, Yale Law Journal 88 (1979), 1623. As Black says at 199, his view ‘does not imply that a “runaway” [Article V] convention is possible, for…no convention can be called that has anything to run away from’. The 1787 Convention lives on!

(53) Plutarch, Lives, vol i, trans. Perrin (Cambridge, Mass., 1914), 49: ‘The ship on which Theseus sailed with the youths and returned in safety…was preserved by the Athenians down to the time of Demetrius Phalereus. They took away the old timbers from time to time, and put new and sound ones in their places, so that the vessel became standing illustration for the philosophers in the mooted question of growth, some declaring that it remained the same, others that it was not the same vessel.’ I will not attempt to sample the vast modern literature on this ‘mooted question’, except to note that it is sometimes recast as the ‘problem of constitution’: see e.g. Michael Rea, ‘The Problem of Material Constitution’, Philosophical Review 104 (1995), 525.

(54) Could it even amend the voting powers of the states in the Senate, a matter explicitly excluded from its amending power by Article V? Why not begin by amending Article V to remove the exclusion? For valuable reflections, see Akhil Reed Amar, ‘Popular Sovereignty and Constitutional Amendment’, in Sanford Levinson (ed.), Responding to Imperfection: The Theory and Practice of Constitutional Amendment (Princeton, 1995), 90–2. Notice that we are raising here the question of whether the power of a revived 1787 Convention would be ‘self‐embracing’ (n. 13 above). The issue of self‐embracingness is not unique to the doctrine of UK Parliamentary Sovereignty but, as Hart said, ‘can arise in relation to ultimate criteria of legal validity in any system’ (CL, 148).

(55) United States v Sprague 282 US 716 (1931).

(56) Coleman v Miller 307 US 433 (1939).

(57) Walker v United States, unreported C00–2125C, US District Court Western District (2001); Walker v Members of Congress, unreported, US District Court Western District C04‐1977 RSM (2004); US Court of Appeals 9th Circuit 05–35023 (2005); certiorari denied Supreme Court 06–244 (2006).

(58) On the role of Diceyean conventions in the US constitution, see H. W. Horwill, The Usages of the American Constitution (New York 1925) which in some ways parallels Marshall's work on the UK, Constitutional Convertions.

(59) On the problematic distinction between revolutions and coups, see J. M. Finnis, ‘Revolutions and the Continuity of Law’, in A. W. B. Simpson (ed.), Oxford Essays in Jurisprudence: Second Series (Oxford, 1973).

(60) But cf Bruce Ackerman, ‘Discovering the Constitution’, Yale Law Journal 93 (1984), 1013, for doubts about the amendment/usurpation distinction as it applies to the US constitution. Ackerman tends to think that the US constitution somehow invites or compasses its own overthrow.

(61) As Hart explained at length: CL, 144–50.

(62) On the parenthetical words, see Greenawalt, ‘The Rule of Recognition and the Constitution’, 640. Greenawalt leans towards the view that ‘the legal authority of…the original Constitution is established by its continued acceptance’, not by its having been validly enacted in 1787. On a rival view, it is the validity of its enactment in 1787 that is now accepted, and that gives the Constitution its legal authority. For the purposes of the argument here the point is not crucial. Hence the parentheses.

(63) CL, 107–8.

(64) CL, 99.

(65) I have explored some of the main differences in ‘Some Types of Law’.

(66) See e.g. CL, 131–2. At 149–50, Hart seems to suggest, collapsing two distinctions that are only very indirectly related, that the ability of the courts to make law is either customary‐and‐inherent or legislative‐and‐delegated. But cf. 93 and 98 where he distinguishes both custom and precedent from legislation.

(67) The least equivocal rendition is Dworkin, ‘No Right Answer?’, in P. M. S. Hacker and J. Raz (eds.), Law, Morality, and Society (Oxford, 1977). My formulation is most influenced by Hart's famous summary in ‘Legal Duty and Obligation’, in H. L. A. Hart, Essays on Bentham (Oxford, 1982), 147–8.

(68) See Dworkin, Law's Empire (Cambridge, Mass., 1986), 255–63, where a ‘right answer’ seems to be relativized to the convictions of each judge.

(69) See Dworkin, Justice in Robes (Cambridge, Mass., 2006), 266 n. 3, where he denies that he has changed his mind about the thesis.

(70) Thus—to take one example from a huge selection—Thomas C. Grey's ‘The Uses of an Unwritten Constitution’, Chicago‐Kent Law Review 64 (1988), 211 has as its main topic creative judicial interpretation of the written constitution.

(71) J. Raz, ‘Interpretation Without Retrieval’, in Andrei Marmor (ed.), Interpretation in Law (Oxford, 1995), 155.

(72) Ibid. 169–72, explaining how such meaning‐giving is constrained by properties of its object, and hence still qualifies as interpretation.

(73) In general these rules are permissive, not mandatory. American theorists tend to assume, mistakenly, that the main US rules of legislative and constitutional interpretation must be mandatory rules, such that in cases of dissensus, at least one of the dissentients must be in breach of legal duty (if only we could work out which). British theorists, by contrast, almost universally acknowledge the permissive character of the main UK rules of legislative interpretation: see e.g. the classic formulations by Rupert Cross in his Statutory Interpretation (London, 1976), 43. Such rules conflict only in that, as a user of them, one must sometimes choose between legal permissions that would give rival meanings to the object of interpretation. One makes one's choice on other grounds, usually moral grounds that are local to a particular object of interpretation, i.e. a particular provision or Act. Thus one need not be predisposed to use the same method in successive cases involving different provisions or Acts. Still less need one have a ‘theory’ of interpretation, with inevitably monopolistic aspirations.

(74) I am not suggesting that the US constitution falls into this category. Doubtless some parts of US constitutional law are judge‐made but not by way of interpretation of the canonical text or of previous interpretations of it. They are independent judicial or legislative contributions to constitutional law of the kind that are characteristic of UK constitutional law. For a very illuminating if overstated account, see David Strauss, ‘The Irrelevance of Constitutional Amendments’, Harvard Law Review 114 (2001), 1457. I say ‘overstated’ because, as his title hints, Strauss sometimes slips into the rival analysis evoked by Hart at CL, 246 (see the remarks at the end of III.2 above).