WHAT makes a constitution constitutional? Nothing, we most plausibly answer, nor does or can anything make a constitution unconstitutional. Constitutions establish the grounds for constitutionality and unconstitutionality, and in so doing they simply cannot themselves be either constitutional or unconstitutional.
Yet although it thus appears illogical to inquire into the constitutionality of a constitution, there may be a point in framing a question in this manner, especially in the context of thinking about the process of constitutional amendment. For although constitutionalists often assume that something in a constitution provides the grounds for distinguishing between a valid and an invalid amendment, scholars who focus on large-scale political and constitutional transformation just as often maintain that the process of amending an entire constitution, of discarding it and replacing it with another one, takes place outside of both the new and the old constitutions. But if displacing a constitution takes place outside of that constitution, then is it possible that displacing part of it may take place outside the constitution as well, and, if so, that amending it may also occur outside the constitution?
These questions are not just logicians’ puzzles, but logical questions that have considerable import for how we view just what it is that constitutes a constitution.1 I mean to explore these questions here, guided substantially by my readings of the implications of Hans Kelsen’s concept of the grundnorm and H.L.A. Hart’s of the ultimate rule of recognition. I will not, however, engage in exegesis of Kelsen or Hart, and thus it is no part of this project to try to remain faithful to Kelsen’s or Hart’s intentions, especially since those intentions remain obscure and the results of those intentions appear at times confused. Nor, therefore, will I take Kelsen’s or Hart’s views as authoritative, in the sense of using a citation as a substitute for an argument. Still, their views on the extralegal foundations of a legal system are suggestive of an approach to the problem of amendment, an approach I propose to develop in the pages to come.
It is a characteristic of most existing treatments of the process of constitutional amendment, American and non-American, that those treatments are internal to the constitution itself. They take a constitution’s own provisions, however defined,2 as the sole source of legitimate amendment.3 Commonly the starting point for such an analysis is a specific amending clause. As with almost all other constitutions, the Constitution of the United States specifies the conditions for its own amendment, and the Constitution is most easily read as implying that its own specified conditions for valid amendment are to be treated as exclusive. Although Article V does not specify in so many words that the procedures specified therein shall be the sole method of amending the Constitution, nor does it contain the words “inter alia,” any fair literal reading of the text of Article V produces the conclusion that nothing in the Constitution textually authorizes methods of amendment other than the two alternative procedures established in Article V itself.4
Because Article V appears on its face to exhaust the possibilities for amending the Constitution consistent with the Constitution itself, existing American attempts to explain how the 1993 Constitution might legitimately be different from the 1787 Constitution, as amended, other than by amendment in conformity with the procedures specified in Article V, have involved heroic efforts to explain either how other provisions of the Constitution might also allow amendment in different ways, or how different readings of Article V itself might suggest a broader conception of what it takes to amend the Constitution. Bruce Ackerman, for example, castigates an attempt to read Article V according to the historical meaning of “Convention” as “formalistic,” arguing that an updated notion of “Convention” allows the Constitution to be structurally amended by a range of engaged public deliberative actions—constitutional “moments”—that are to be treated as “Conventions” in a less formal sense of that word.5 And Akhil Amar reaches similar conclusions about the nonexclusivity of Article V by relying both on the history of the Constitution’s creation and on his reading of the Preamble, Article VII, and the First, Ninth, and Tenth Amendments.6
My concern here is not with the soundness of efforts such as Ackerman’s and Amar’s, but rather with their impetus. More specifically I am concerned with why it is that “exclusivists” such as Dow, Dellinger, Gunther, Tribe, and Vile, and “nonexclusivists” such as Ackerman, Amar, and Levinson,7 all suppose that the internal resources of the Constitution, however those internal resources are defined, provide the only or most appropriate way of thinking about the process of constitutional change. By contrast, I argue that constitutions rest on logically antecedent presuppositions that give them their constitutional status. As a result, constitutions can and do change not only when they are amended according to their own provisions or their own history, however broadly those provisions or that history may be understood, but whenever there is a change in these underlying presuppositions—political and social, but decidedly not constitutional or legal.8 Constitutions are thus necessarily always subject to amendment as their supporting presuppositions are amended, even though it cannot be the case that the amendment of those supporting presuppositions can be thought of in anything other than factual or other prelegal terms.
Although the issues I address pertain equally to all constitutions, and thus to all national constitutions, to all subnational constitutions such as those of the states in the United States, and to the constitutions of private associations, I will use the Constitution of the national government of the United States as my primary continuing example. Consider, then, the question of the validity of a particular action by or decision of an administrative agency. Because agencies typically have their own rules purporting to govern all of their own actions, the normal test for the validity of an administrative action is whether that action is authorized by and consistent with the agency’s own regulations. The first question about the validity of an administrative action, therefore, is whether it is authorized by some higher (but not very much higher) law, here the set of regulations of the agency itself.
Even if some administrative action were valid according to an agency’s own regulations, however, the question would then arise about the validity of those administrative regulations themselves. Administrative agencies are not of course sovereign bodies, and we know, to oversimplify, that the validity of an administrative regulation is a function of its falling (or not) within the scope of some authorizing act of Congress. An administrative regulation is thus itself valid only insofar as it is authorized by some higher law, here an act of Congress.
Although administrative regulations are valid only insofar as they are authorized by acts of Congress, what is it, however, that makes an act of Congress valid? Here we move one step up the chain, and we know that just as administrative regulations are valid only insofar as they are authorized by acts of Congress, then so too are acts of Congress valid only insofar as they are authorized by and consistent with the Constitution.
But if we are searching for the foundations of law, we might then go one step further, and ask what it is that makes the Constitution valid? It is at this point that some of the most enduring questions of jurisprudence are engaged, and it is here that those questions are directly relevant to the question of constitutional amendment. For when we ask what it is that makes the Constitution valid, the work of Hans Kelsen would indicate that the validity of the Constitution is established by the Grundnorm, whose validity is in turn presupposed or hypothesized as valid. Thus to Kelsen the Grundnorm is the presupposition without which the whole notion of legal validity makes no sense, and which is necessary for statements such as “this is the law” to have any meaning. For Kelsen the Grundnorm functions as sort of a Kantian transcendental understanding, not a fact capable of empirical investigation, but just a way of thinking that is necessary for a range of statements and conclusions about the law to make any sense at all.9
Just what it is that Kelsen meant by the Grundnorm is notoriously obscure.10 Still, the main points are that constitutions need grounding as much as they supply it, and that it is therefore a mistake to think of a constitution as itself being the last or ultimate word. But since Kelsen’s ideas here get a bit mysterious, and since my aim in this essay is not to engage in Kelsenian exegesis, it may be more useful to look instead at H.L.A. Hart’s derivative but clearer idea of the ultimate rule of recognition.11 Among the secondary rules that combine with primary rules to make a legal system, Hart says, are rules of recognition by virtue of which citizens and officials can know what is a law and what is not. Acts of Congress such as the Administrative Procedure Act, therefore, function in part as rules of recognition determing which administrative regulations are to be recognized as valid and which are not. So too with the Constitution, which similarly functions as a rule of recognition determining the validity of state and federal laws. And when we ask what makes the Constitution valid, Hart says, we look to something different, the ultimate rule of recognition, which might specify, for example, that the document commonly known as “the Constitution of the United States of America,” the original version of which was ratified in 1787 and is now to be found in the National Archives in the District of Columbia, is the measure of the validity of all other laws in the United States.
But what if we want to know what makes the ultimate rule of recognition valid? Although Kelsen thinks that this is an answerable question, Hart maintains simply that we have asked the wrong question, because the very notion of validity presupposes some higher legal norm against which a subordinate legal norm might be measured. When we run out of legal norms we have exhausted the concept of legal validity (although we might still use the word valid to refer to some other norm system, such as a moral one),12 and thus the question about the status of the ultimate rule of recognition is a question of fact. The ultimate rule of recognition is a matter of social fact, and so determining it is for empirical investigation rather than legal analysis.13
In referring to the ultimate rule of recognition as a rule, Hart has probably misled us. There is no reason to suppose that the ultimate source of law need be anything that looks at all like a rule, whether simple or complex, or even a collection of rules, and it may be less distracting to think of the ultimate source of recognition, following Brian Simpson, as a practice.14 The ultimate source of law, therefore, is better described as the practice by which it is determined that some things are to count as law and some things are not.
Hart’s analysis is also thin on the question of just whose determination and whose practice makes for the existence of the ultimate rule of recognition. Hart says that the ultimate rule of recognition exists in what officials, particularly judges, recognize as the sources of law,15 but here Hart’s own argument against the Realists may be used against him.16 Just as saying that law is what the judges say it is tells us nothing about why the judges say what they say, or what judges look at in determining what the law is, or what arguments might be usable in legal argument, so too does identifying as the ultimate rule (or practice) of recognition what the judges actually take to be the sources of law tell us little about where judges will look to determine what the legitimate sources of law are. This, it turns out, is a question of considerable importance in cases of revolution or commensurate transformation, for it may then be the task of judges to try to determine just what the ultimate rule of recognition is. Perhaps they just choose sides on political grounds of their own, but perhaps in some cases they also try to determine whether the people in general, or other officials in general, or just law-enforcing officials, have shifted in what they believe the ultimate source of law to be, such that the judges are looking externally as well as internally in determining the content of the ultimate rule of recognition.17 And this is why at the level of determining what the ultimate rule of recognition is, questions of efficacy are central, because only with some degree of efficacy in fact (Kelsen said that the existence of a legal system required that the laws of that system were obeyed and applied “by and large”)18 is it possible to say that some norm or practice exists as all or part of the ultimate rule of recognition. As Ilmar Tammelo put it, “The foundation rules of a legal system can never be derived from within the system itself; they are a political fact.”19
For purposes of thinking about the process of amendment, not all of these complications need be pursued to the same extent that we might pursue them were we concerned with the total displacement of legal systems. Still, the central import of this line of thought is recognition of the necessarily factual basis of the ultimate rule of recognition, and recognition of the associated point that something preconstitutional as well as factual is the logical predicate for the truth of the proposition that the document in the National Archives is the Constitution of the United States.
I want to illustrate this point with a bizarre, but I hope still useful, example. Let me, right here on these pages, offer a constitution for the United States:
The Constitution of the United States of America
We the People of the United States, in order to place our welfare and our future in the soundest possible hands, grant to Frederick Schauer, of the City of Cambridge, County of Middlesex, Commonwealth of Massachusetts, the plenary and sovereign powers of governance, without limitation, over the territory known in 1994 as the United States of America.
Article I. Frederick Schauer or those he may designate shall possess all of the legislative powers of the United States.
Article II. Frederick Schauer or those he may at any time designate shall hold all powers of enforcement and application of the laws.
Article III. All judicial powers shall be exercised by judges appointed by whoever at the time holds power under Article II of this Constitution.
Article IV. The powers under Articles I, II, and III shall descend according to the laws of descent and distribution in force in the Commonwealth of Massachusetts on December 31, 1992.
Article V. This Constitution may be amended only with the joint consent of those then exercising power under Articles I and II.
Article VI. This Constitution shall be established and in force upon signing by the individual named in Article I.
/s/
Frederick Schauer
There you have it. Not only have I just presented a document entitled “The Constitution of the United States of America” purporting to be the constitution for the United States of America, but I have also ensured that it is valid and in force according to its own terms, just as the document to be found in the National Archives is also valid and in force according to its own terms. And as a result of my having drafted this constitution and satisfied its own internal conditions for validity, I can state with confidence that there now exist within the territory known as the United States (at least) two internally valid documents, each purporting to be the Constitution of and for the United States of America.
We know, of course, that the document in the National Archives is, more or less (which I will get to presently), the Constitution of the United States, and we know just as well that the silly collection of words just above is not. But how do we know this? We know this not because of anything internal to one document or the other, because internally they are equally valid. Rather, we know that one is the Constitution and the other is not because of what we know empirically and factually about the world, because we know that one is efficacious and the other not, because we know that one document has been accepted by the American people, by American officials, and by American judges, while the other has been accepted by no one, not even its author. Yet had the facts, and only the facts, been reversed, the conclusion would be just the opposite. Had the American people and the American officials accepted my constitution as governing law rather than the thing in the National Archives, then the document above would be the Constitution of the United States. Thus a change in the raw fact of efficacy or acceptance could make my document the Constitution while the thing in the National Archives would be but a legally irrelevant historical relic.
Although my example is both simple and silly, it nevertheless resembles more closely than might at first appear a number of genuine issues that arise whenever there is a case of dramatic constitutional transformation. Whenever there is a revolution, or whenever there is some other form of dramatic change in government, there is usually an attempt by the successor government to substitute a new ultimate rule of recognition for the one that had been in force prior to the transformation. As with the choice between my constitution and the existing U.S. Constitution, however, the question of whether the ultimate rule of recognition has indeed been changed by revolution is a question that the internal resources of neither the old nor the new constitutions can answer.20 The fact of constitutional displacement is just that—a fact—and the social choice between the new and the old is just like the social choice between the existing U.S. Constitution and some other, such as mine.21
Most of the existing literature about the transformation of legal systems has focused on full-scale displacements of just this type. As with the cases in Pakistan, Uganda, and Southern Rhodesia, courts and various international bodies have been required to determine whether the ultimate rule of recognition had changed, in order to make the logically subsequent determination of which particular constitution was then in force. And in making this determination, it is clear that courts could not rely on anything in either of the contesting legal systems, for the status of the very documents was exactly what was at issue.
Yet although the history and the literature typically focus on cases of total constitutional displacement, the essential point about the externality (to a constitution) of the determination of constitutional change applies equally to partial displacement. We understand now that the determination that Constitution B has replaced Constitution A cannot be made on the basis of anything in Constitutions A or B, but the same lesson holds in the case of partial rather than total replacement. Suppose that Constitution A contained two parts, Part A.1 outlining the structure of government and Part A.2 being a bill of rights limiting the powers of government. And suppose as well that Constitution B also contained two parts, again one establishing the structure of government and the other constraining that government with a bill of rights, these two parts being designated B.1 and B.2. Now suppose that at some time a population has decided that its existing constitution, A, is unsatisfactory, because it contains a defective bill of rights. Revolutionary forces have proposed B as a substitute, there is some fighting, and then there is a cease-fire. The result is that the competing forces agree, with widespread public support, that the country should henceforth be governed according to a combination of the structure of government provisions of the earlier constitution and the bill of rights of the proposed replacement, that is, A.1 and B.2.
Were all of this to happen, and were this resolution of the conflict thus accepted by the people and by officials, the content of the ultimate rule of recognition would be such that it then recognized as the supreme law this combination of A.1. and B.2. If, therefore, the ultimate rule of recognition were to recognize as supreme law this A.1/B.2 constitution, then the correct answer to the question “What is the constitution?” would be “the combination of A.1 and B.2.”
Once we accept that “the constitution” according to the ultimate rule of recognition need not be something written at one time,22 we see Kelsen’s mistake in assuming that the only two possible Grundnormen are a changeable one in countries such as Great Britain, New Zealand, and Israel, which have no single document titled “the Constitution,” or what he referred to as the “historically first constitution.”23 Where there is a written constitution, Kelsen asserted, “coercive acts ought to be carried out only under the conditions and in the way determined by the ‘fathers’ of the constitution or the organs delegated by them.”24 As the father of the Austrian Constitution of 1930, it is understandable why Kelsen would promote this view, but it is nevertheless a poor interpretation of Kelsen’s own idea of the Grundnorm, and independently mistaken in its own right. Although Kelsen’s quoted view could be the Grundnorm or ultimate rule of recognition in a country with a written constitution, and although we well understand why the drafter of a constitution would wish for this to be the Grundnorm, there is no reason why the Grundnorm must be one that treats the law-recognizing functions of a particular written constitution as exclusive. If the Grundnorm were “coercive acts ought to be carried out only under the conditions and in the way determined by the ‘fathers’ of the constitution or the organs delegated by them, or whenever such acts are authorized by a referendum of 90 percent of the population,” this would still qualify as a Grundnorm or ultimate rule of recognition, and were this the Grundnorm actually in force in some society, then coercive acts authorized by 90 percent of the population in a referendum would qualify as “valid law” even though those acts would not comport with the requirements of the written constitution taken in isolation.25
What all of this shows, of course, is that once we recognize that constitutional displacement takes place outside of the constitution, then so can partial displacement take place outside of the constitution. And the next step is no step at all, or at least a very small one. If partial displacement can take place outside of the “primary” written constitution, then so too can partial “supplementation” take place outside of the written constitution. Thus it is quite likely, as Kent Greenawalt has developed at length, that the ultimate rule of recognition in the United States may refuse to recognize parts of the written Constitution of 1787 as valid law, and most certainly recognizes as valid law sources of law not traceable to or through the Constitution of 1787.26 And thus if law can exist outside of the Constitution, then so too can it be amended outside of that Constitution. Moreover, since the Constitution thus rests on the extraconstitutional foundations that make up the ultimate rule of recognition, then the Constitution necessarily can be amended by the extraconstitutional process of amending, socially and politically, those extraconstitutional foundations.27 To take an example, if the American people came to the realization that the Second Amendment’s seeming protection of the right to keep and bear arms was simply obsolete and unwise in light of the realities of 1994, and if that view were shared by legislative, executive, and judicial officials, and if all proceeded to treat the Second Amendment as a nullity despite the fact that it had not been repealed according to the provisions of Article V, then it would be accurate to say that the Constitution of the United States did not contain the provision designated as “Amendment II” in most versions of the document titled “the Constitution of the United States.” The small c constitution would thus have been amended by virtue of this amendment to the ultimate rule of recognition, even though it could also be accurately said that the large C Constitution had not been validly amended according to its own terms.
Nothing I have said in the foregoing analysis is intended to take a position on whether it would be a good idea, in a country having a canonical written constitution, to adopt this type of somewhat more open-ended ultimate rule of recognition instead of the more bounded one that Kelsen preferred. Although Kelsen was mistaken in supposing that the existence of a historically first written constitution mandated a Grundnorm treating that historically first constitution (and therefore its amending clauses) as exclusive, it might still be a wise course of action, as a matter of social/constitutional policy, for a nation having a written constitution with internally comprehensive pretensions to treat that constitution as the exclusive “supreme criterion”28 for the recognition of all other laws. This is almost certainly not the American approach, but that is not to say that it could not be a quite plausible one.
My primary goal here is to explain why the choice between a written constitution as an exclusive or a nonexclusive supreme criterion of law is antecedent and external to the constitutional text (and therefore also to the history of that text) rather than to explore the reasons why a polity should choose one or the other. Still, it is worth noting that at this antecedent level the issue is merely an instantiation of the familiar questions about the desirability of rule-based constraint, albeit at one remove and in a slightly different setting. For just as the amendment power itself grows out of a rule-based view of legal or constitutional constraint (were rules merely rules of thumb then there would be no need to amend those rules to take care of the instances in which the rules produced a suboptimal result),29 a Grundnorm treating the amendment power as exclusive would be a rule-based view that it is better to have some ideally desirable amendments not take place because of the constraints of the amendment procedure than to have too many unwise amendments take place because of an excessively permissive environment that treated the amendment power as merely establishing a rule of thumb.
Alternatively, some society might recognize that rules, including the rules of amendment, are necessarily either actually or potentially both under- and overinclusive vis-à-vis their background justifications, or visà-vis the best all-things-considered decision for the case at hand, such that there might be circumstances in which the rules of amendment, if followed faithfully, would turn out either to be insufficiently or excessively stringent. Recognizing this, and not wishing to be bound in the way that rules bind, this society might choose instead an ultimate rule of recognition that treated the amendment provision as merely hortatory, or simply as a rule of thumb, or presumptive but not conclusive, or perhaps even to be dispensed with entirely.30
Insofar as a society feared its own excess willingness to amend its constitution outside of the ways designated by that constitution, it might prefer the former approach to the latter, seeking to bind itself, at least presumptively but perhaps even conclusively, to an amendment process premised on textual exclusivity. As should now be clear, however, this state of affairs cannot be accomplished simply by writing that exclusivity into a constitutional text, for that writing would itself be subject to the very kind of antecedent amendment of presuppositions that I have been exploring here.31 Nevertheless, entrenchment can be accomplished psychologically even if not logically, and so a society with this rule-based approach to the amendment process might choose to adopt and publicly articulate as its Grundnorm that the textually designated amendment process was to be taken as exclusive.32 Were that to happen, it is likely that the very act of public articulation and public written canonization would serve to make it substantially more psychologically difficult to amend that presupposition than would have been the case without that public commitment. Just as making even an internal New Year’s resolution may produce a resolve to keep that resolution stronger than would have existed for the same conduct without the crystallized internal articulation of the resolution, and just as publicly announcing that resolution may create an informal but still effective interpersonal enforcement mechanism that is even more constraining than an internal resolution,33 so too may publicly announcing and discussing the importance of retaining textual exclusivity in the amendment process serve to entrench psychologically an ultimate rule of recognition that, as we have seen, cannot be entrenched by legal or constitutional means.
The case of constitutional amendment, therefore, turns out to exemplify the familiar (although often abused) point that rules cannot determine their own application. Rules exist against a background of presuppositions—such as the presupposition that when confronting an arrow on a highway one should drive in the direction of the point rather than the tail,34 or the presupposition that the language in which a rule is written should, or should not, be interpreted as ordinary nontechnical English—that are themselves subject to change. Amendment provisions are of course written in a language, and it is a mistake to assume that the contingency of linguistic meaning entails a world in which there is a possibility of frictionless linguistic change. Moreover, it is also a mistake to assume that the difficulties in explaining the grounding of language entail either the indeterminacy of language, or the indeterminacy of rules that may be written in it.
As an empirical matter, the amendment provisions of most constitutions do have identifiable semantic content sufficient to indicate an outcome in a large number of potential instances. Where an amendment provision is this concrete, as Article V of the Constitution of the United States has proved to be, where the public commitment to recognition of the constitution containing those provisions is strong, and where the public commitment to recognition of that amendment provision in particular is equally strong, then an amendment provision could reflect public recognition of the desirability of an exclusive approach to that amendment provision, especially if its words are also taken to be understood literally. In other words, a polity could make a New Year’s resolution about the exclusivity of its amending clause, and that resolution could make a difference in the degree of entrenchment of that exclusivity. But all of this is contingent, and all of this is therefore subject to change. A society could, for example, shift from a literal to a nonliteral understanding of its amendment provisions, and given the prevalence of nonformal and nonliteral modes of legal and constitutional interpretation in the United States,35 this may in fact be what has happened. Or it could shift from an exclusive to a nonexclusive approach when confronted with a situation in which its amending clause was widely understood to have proved unsatisfactory. But whether and when these eventualities have occurred are not issues that can be determined by looking at the Constitution, however the Constitution is defined. And if these presuppositions are to change, the resources for that change are again not something that any amount of massaging of the Constitution itself can produce.
The process of constitutional amendment, therefore, can take place on one of two levels. On the constitutional level, it can take place within the contours of the constitution itself, either according to a literal reading of an amending clause, or according to an interpretive understanding consistent with the understanding of just what the constitution encompasses. But because constitutions owe their “constitutionality” to logically and politically antecedent conditions, the process of constitutional amendment may also take place at another level, when these logically and politically antecedent conditions are themselves amended. Because these antecedent conditions are not themselves legal or constitutional in any important sense of those terms, however, it remains necessarily the case that constitutions are always subject to amendment by changes—amendments—in the practices of a citizenry, in the practices of its officials, and in the practices of its judges. Whether these changes have occurred will be a question of social and political fact and not a question of law, constitutional or otherwise. And whether these changes should occur will be the necessarily political and moral question of what status a constitution should have, and what status its particular provisions should have. There may not be any harm in thinking of these questions as constitutional ones, but neither is there any harm—and there may indeed be some good—in recognizing the way in which thinking about the presuppositions of a constitution is quite different from thinking about constitutionalism against the background of assumed presuppositions.
This essay was completed while I was a visiting fellow of the Research School of Social Sciences, Australian National University. I am grateful for both financial and intangible support to the Australian National University, and also to the Joan Shorenstein Barone Center on the Press, Politics and Public Policy. An earlier version was presented at the Academica Sinica, Taipei, Taiwan, and I am grateful for the audience comments on that occasion.
1The relationship between the problems of amendment and the logical paradoxes of self-reference are well known. See, for example, Peter Suber, The Paradox of Self-Amendment: A Study of Logic, Law, Omnipotence and Change (London: Peter Lang, 1990); Patrick Fitzgerald, “The ‘Paradox’ of Parliamentary Sovereignty,” Irish Jurist, n.s., 7 (1972): 28–48; J. C. Hicks, “The Liar Paradox in Legal Reasoning,” Cambridge Law Journal 29 (1971): 275–91; N. Hoerster, “On Alf Ross’s Alleged Puzzle in Constitutional Law,” Mind 81 (1972): 422–27; Alf Ross, “On Self-Reference and a Puzzle in Constitutional Law,” Mind 78 (1969): 1–14; Ilmar Tammelo, “The Antinomy of Parliamentary Sovereignty,” Archiv für Rechts-und-Sozialphilosophie (1958): 495–513.
2I mean to encompass a great deal by “however defined.” That is, I do not mean to limit my notion of “internal” to a text-based approach to constitutional interpretation. Rather, my notion of “internal” encompasses any interpretive methodology that its propounder would claim to be an interpretation of this constitution, rather than an interpretation of some other constitution, and rather than a political or moral proposal its proponent admitted could not be connected to the constitution.
3Good examples would include Walter Dellinger, “The Legitimacy of Constitutional Change: Rethinking the Amendment Process,” Harvard Law Review 97 (1983): 386–432; Walter Dellinger, “Constitutional Politics: A Rejoinder,” Harvard Law Review 97 (1983): 446–50; Gerald Gunther, “The Convention Method of Amending the Constitution,” Georgia Law Review 14 (1979): 1–25; Laurence H. Tribe, “A Constitution We Are Amending: In Defense of a Restrained Judicial Role,” Harvard Law Review 97 (1983): 433–45; Katherine Swinton, “Amending the Canadian Constitution: Lessons from Meech Lake,” University of Toronto Law Journal 42 (1992): 139–69; George Winterton, “An Australian Republic,” University of Melbourne Law Review 16 (1988): 475–78.
4 See David R. Dow, “When Words Mean What We Believe They Say: The Case of Article V,” Iowa Law Review 76 (1990): 4, 39–44; John R. Vile, “Legally Amending the United States Constitution: The Exclusivity of Article V’s Mechanics,” Cumberland Law Review 21 (1991): 271–307; John R. Vile, “Judicial Review of the Amending Process: The Dellinger-Tribe Debate,” Journal of Law and Politics 3 (1986): 21–50; John R. Vile, “The Amending Process: Alternative to Revolution,” Southeastern Political Review 11 (1983): 49–96; John R. Vile, “American Views of the Constitutional Amending Process: An Intellectual History of Article V,” American Journal of Legal History 35 (1991): 44–69.
5Bruce Ackerman, “The Storrs Lectures: Discovering the Constitution,” Yale Law Journal 93 (1984): 1013–72; see also Ackerman’s “Constitutional Politics/Constitutional Law,” Yale Law Journal 99 (1989): 453–96, and We the People: Foundations (Cambridge, Mass.: Harvard University Press, 1991).
6Akhil Reed Amar, “Philadelphia Revisited: Amending the Constitution Outside Article V,” University of Chicago Law Review 55 (1988): 1043–1104.
7Sanford Levinson, “Accounting for Constitutional Change (Or, How Many Times Has the United States Constitution Been Amended? (A) < 26; (B) 26; (C) > 26; (D) All of the Above),” Constitutional Commentary 8 (1991): 409–31, reprinted, in expanded form, as Chapter 2, above.
8By drawing this distinction I plead guilty to legal positivist predispositions. See Frederick Schauer, “Constitutional Positivism,” Connecticut Law Review 25 (1993): 797–828. Nevertheless, nothing in this article assumes any stronger a positivist perspective than even, say, Ackerman and Amar presuppose. As long as they and others seek to situate their conclusions in a reading of the constitutional text, or in a reading of history that focuses more on events related to that text than to other historically contemporaneous events, then they have accepted the broad extensional disequivalence between the realms of the constitutional/legal on the one hand and the social/political on the other that is presupposed in my argument.
9On the questions I discuss here, the primary Kelsenian text is Hans Kelsen, General Theory of Law and State, trans. Anders Wedberg (New York: Russell & Russell, 1961), especially pp. 115–36. See also Hans Kelsen, Pure Theory of Law, trans. Max Knight (Berkeley: University of California Press, 1967), pp. 46–50, 202–11; Hans Kelsen, “On the Pure Theory of Law,” Israel Law Review 1 (1966): 1–7; Hans Kelsen, “Professor Stone and the Pure Theory of Law,” Stanford Law Review 17 (1964): 1128–48.
10For a sampling of the commentary, see J. W. Harris, Law and Legal Science: An Inquiry into the Concepts Legal Rule and Legal System (Oxford: Clarendon Press, 1979); Joseph Raz, The Concept of a Legal System: An Introduction to the Theory of Legal System, 2d ed. (Oxford: Clarendon Press, 1980); Julius Stone, Legal System and Lawyers’ Reasonings (Sydney: Sweet and Maxwell, 1964), pp. 132–34, 202–5; George C. Christie, “The Notion of Validity in Modern Jurisprudence,” Minnesota Law Review 48 (1964): 1049–79; J. W. Harris, “When and Why Does the Grundnorm Change?” Cambridge Law Journal 29 (1971): 103–33.
11H.L.A. Hart, The Concept of Law (Oxford: Clarendon Press, 1961), especially pp. 97–114, 245–47. Hart’s ideas about the ultimate rule of recognition are usefully explained and at times criticized in Neil MacCormick, H.L.A. Hart (Stanford, Calif.: Stanford University Press, 1981), and in P.M.S. Hacker, “Hart’s Philosophy of Law,” in P.M.S. Hacker and J. Raz, eds., Law, Morality and Society: Essays in Honour of H.L.A. Hart (Oxford: Clarendon Press, 1977), pp. 1–25.
12Thus Stone, Legal System and Lawyers’ Reasonings, pp. 202–5, maintained that what Stone called the “apex norm” could still be measured by “socioethical” standards.
13The precursor to both Kelsen and Hart was Salmond, who noted that the ultimate rule of law was “historical only, not legal.” J. Salmond, Jurisprudence, 12th ed. (London: Sweet and Maxwell), p. 111.
14A.W.B. Simpson, “The Common Law and Legal Theory,” in A.W.B. Simpson, ed., Oxford Essays in Jurisprudence, 2d ser. (Oxford: Clarendon Press, 1973), pp. 77–99. And at one point Hart himself says that the ultimate rule of recognition “exists only as a complex, but normally concordant, practice of the courts, officials and private persons in identifying the law by reference to certain criteria.” Hart, The Concept of Law, p. 107.
15The act of recognition on the part of judges and other officials may not be explicit, but may just consist in what officials use and apply in making their decisions. See Roger Shiner, Norm and Nature: The Movements of Legal Thought (Oxford: Clarendon Press, 1992), pp. 170–74.
16Hart, The Concept of Law, pp. 121–50.
17There is a remarkable series of cases in which postrevolutionary judges, required just after a period of major political upheaval to decide what body of law, what legal system, they were supposed to enforce, relied quite explicitly on Kelsenian ideas in framing their inquiries. The most famous of these cases arose out of the Unilateral Declaration of Independence (UDI) of what was then Southern Rhodesia, Madzimbamuto v. Lardner-Burke N.O., JD/CIV/23/66, 9 Sept. 1966 (1968) 2 S.A. 284 (App. Div.) [1969] A.C. 645 (P.C.), and earlier and similar issues arose in Pakistan, State v. Dosso (1958) 2 Pakistan S.C.R. 180, and in Uganda, Uganda v. Commissioner of Prisons, ex parte Matovu (1966) E.A. 514. There has been an extensive and jurisprudentially sophisticated commentary on these cases, and much of it is germane to the issue of amendment. Good examples are J. M. Eekelaar, “Splitting the Grundnorm,” Modern Law Review 30 (1967): 156–75; J. M. Eekelaar, “Rhodesia: The Abdication of Constitutionalism,” Modern Law Review 32 (1969): 115–18; Harris, “When and Why Does the Grundnorm Change?” A. M. Honore, “Reflections on Revolution,” Irish Jurist, n.s., 2 (1967): 268–78; S. A. de Smith, “Constitutional Lawyers in Revolutionary Situations,” Western Ontario Law Review 7 (1968): 93–110.
18Kelsen, General Theory of Law and State, pp. 118, 437.
19Tammelo, “The Antinomy of Parliamentary Sovereignty,” p. 504.
20There is a curious literature addressing the question of whether the existing U.S. Constitution is illegal according to the Articles of Confederation. Those arguing that the 1787 Constitution was illegal according to the Articles of Confederation include Bruce Ackerman, “Discovering the Constitution,” pp. 1017, n.6, 1058; Richard S. Kay, “The Illegality of the Constitution,” Constitutional Commentary 4 (1987): 57–80; John Leubsdorf, “Deconstructing the Constitution,” Stanford Law Review 40 (1987): 181, 186–88; and those maintaining that the 1787 Constitution can be taken to have been ratified consistent with the authorization of the Articles include Amar, “Philadelphia Revisited,” and Julius Goebel, History of the Supreme Court (New York: Columbia University Press, 1971), pp. 198–204. The debate is curious, however, because both the debate and the reason for having it are again external to the resources used by the combatants. No amount of illegality according to the Articles would render the 1787 Constitution any less the law now, and conversely no amount of legality according to the Articles would make the 1787 Constitution the law absent the necessary factual preconditions of acceptance and efficacy. It is likely, therefore, that the debate is premised on the mistaken supposition that the only options are violent and armed revolution, on the one hand, and legal continuity, on the other. Yet if we accept the fact that there can be peaceful, orderly, and deliberative revolutions (consider the current legal orders in some number of countries that were formerly members of the Soviet Union), then one can recognize the illegality of the 1787 Constitution under the Articles while still not concluding that the Framers of the 1787 Constitution were “lawless” in the sense of that word that implies violence or tyrannical usurpation of power rightfully held by others.
21This is decidedly not to say, however, that the task of a judge in such circumstances is limited to determining the social or political fact of the efficacy of a new legal system. The judge is herself part of the very efficacy that determines whether the ultimate rule of recognition has been transformed, and thus there is an unavoidable normative component in a judge’s determination that a new ultimate rule of recognition should replace and has replaced the old one.
22On the way in which the constitution of a country is more than just the document it labels as “the Constitution,” see W. R. Lederman, “The Supreme Court of Canada and Basic Constitutional Amendment: An Assessment of Reference Re Amendment of the Constitution of Canada (Nos. 1, 2 and 3),” McGill Law Review 27 (1982): 527–40.
23Kelsen, Pure Theory of Law, p. 50.
24Kelsen, General Theory of Law and State, pp. 115–16.
25For a similar critique of Kelsen, see Raz, The Concept of a Legal System, p. 138.
26Kent Greenawalt, “The Rule of Recognition and the Constitution,” Michigan Law Review 85 (1987): 621–71.
27Thus it may be important to distinguish what appears to be Ackerman’s and Amar’s question—“Can the Constitution be amended outside of a literal reading of Article V consistent with a certain internal and accepting attitude toward the Constitution as a whole?”—from the quite different question of whether the Constitution can be amended outside of a literal reading of Article V. Putting the question in the former way, and thus trying to get out of a literal reading of Article V only by use of the Constitution’s own internal resources, may signal a certain kind of “pro” attitude toward the existing Constitution not necessarily shared by the explanatory theorist, and not necessarily shared by a population inclined toward large-scale change in their constitution.
28Hart, The Concept of Law, p. 103.
29 I have developed elsewhere the picture of rules that undergirds the brief statement in the text. Frederick Schauer, Playing By the Rules: A Philosophical Examination of Rule-Based Decision-Making in Law and in Life (Oxford: Clarendon Press, 1991).
30Descriptively, my own view is that the American legal and constitutional culture treats the procedures of Article V as presumptively but not conclusively constraining, just as that same culture treats most written legal norms in just this presumptive way. See Frederick Schauer, “Rules and the Rule of Law,” Harvard Journal of Law and Public Policy 14 (1991): 645–94.
31Thus, the supposed paradox of self-amendment is either no paradox at all, or is only a paradox within a closed domain whose presuppositions are stipulated not to be subject to alteration. But the fact that a change at one level might be inconsistent with something said at another level is simply not paradoxical. See Hoerster, “On Alf Ross’s Alleged Puzzle in Constitutional Law,” Tammelo, “The Antimony of Parliamentary Sovereignty” Thus there is nothing paradoxical about a change in constitutional presuppositions having the effect of repealing some provision in a text, or giving it a meaning different from its ordinary language meaning.
32As should be clear by now, an ultimate rule of recognition could treat the entire constitution as exclusive, and thus necessarily treat its amendment clause as exclusive. But a different ultimate rule of recognition, indeed the one advocated by Walter Dellinger, “The Legitimacy of Constitutional Change,” pp. 389, 418–19, and seemingly presupposed by many other commentators, could treat all of a constitution as nonexclusive except for its amendment provision, which it would treat as exclusive. Such an approach, however, puts the greatest strain on any distinction between interpretation and amendment, because at some point the interpretive freedom allowed by such an approach through all of the rest of the constitution will turn into the very kind of nonexclusive approach to the amendment process that this combination of views seeks to avoid. See Chapter 2 in this volume, by Sanford Levinson.
33See Thomas C. Schelling, “The Intimate Contest for Self-Command” and “Ethics, Law, and the Exercise of Self-Command,” in Choice and Consequence (Cambridge, Mass.: Harvard University Press, 1984), pp. 57–112.
34Ludwig Wittgenstein, Philosophical Investigations, trans. G.E.M. Anscombe (Oxford: Basil Blackwell, 1958).
35The point in the text is exemplified by works such as Guido Calabresi, A Common Law for the Age of Statutes (Cambridge, Mass.: Harvard University Press, 1984), and contrasted with other approaches in Robert Summers and Patrick Atiyah, Form and Substance in Anglo-American Adjudication (Oxford: Clarendon Press, 1986).