Section 2
Preliminaries: Amendment Clauses
Peter Suber, Paradox of Self-Amendment Table of Contents

One of the most important subjects that can engage the attention of the statesmen and people of this country is the extent and the scope of the power to amend the Constitution of the United States.
George Tichnor Curtis[Note 1]

A. Amendment and revolution, lawful and unlawful change of law

Statutes and ordinary enactments cannot change constitutions. Adjudications can settle, even change, the interpretation of constitutions. Some judicial interpretations may take us beyond the limits of the reasonable alternative meanings of the language (see Section 15). But while many "judicial amendments" have vital consequences, the public and the judges themselves would not tolerate implementing certain basic changes by judicial review. Where judges fear to tread, and when change is intended to be fundamental and long-lasting, formal constitutional amendment is necessary.

Constitutions must change, for the societies they govern change constantly and if the constitution lags too far behind, then the rule of law itself is jeopardized. Lester Orfield speculates that the Civil War may have been prevented if the federal constitution had been easier to amend.[Note 2] Adlai Stevenson believed that the difficulty of amending the Illinois constitution before 1950 made evasion of the law a practical necessity in order to avoid "anachronisms" of the Illinois constitution.[Note 3] The constitution must change peacefully and legally, and with roughly the speed and depth of changing circumstances and public dissatisfaction. So long as we are not ready to let courts implement all constitutional change through judicial review and reinterpretation, then only a clause within the constitution itself[Note 4] can provide the authority for most substantive change and the peaceful adaptability within which civilization may flourish. This value was well expressed by John W. Burgess in 1891:[Note 5]

[The amendment clause] is the most important part of a constitution. Upon its existence and truthfulness, i.e., its correspondence with real and natural conditions, depends the question whether the state shall suffer the alternations of sconstitutionetrogression and revolution. A constitution, which may be imperfect and erroneous in its other parts, can be easily supplemented and corrected [through an amendment clause; otherwise] error will accumulate until nothing short of revolution can save the life of the state.

The basic design of government in a constitutional state is embodied in the constitution. An amendment clause (hereafter, "AC") not only aconstitutionlso structures, the change of the constitution under the rule of law.[Note 6]

Basic change brought about under the authority of a constitution, through its AC, preserves the continuity of law in three ways. First, evasion and disregard of anachronistic provisions and outright revolutionary breaks with them are made less urgent, and to that extent less admissible legally and less justifiablconstitutiony. Second, the same authority or constitution is recognized as valid throughout the change, and loses nothing of its validity by reason of the change except the old provison in its particularity that was modified or repealed. Third, the same state that existed before the change exists after the change.

The continuity of law, its identity through change, is a metaphysical idea with crucial legal significance. It is equally important for each of two opposing schools of jurisprudence that we may call the formalist and the non-formalist. These two schools differ primarily on the relations of law to social, political and other "alegal" circumstances, and on the valid sources of authority for legal change. If we conceive law formally, then new law is authorized only by old law; if we conceive it non-formally, then it may be authorized also by present, extra-legal sources of authority such as the consent of the people, the might of the army, or the will of Jehovah. Therefore, for the formalist all legal change is revolutionary and unlawful if it does not proceed from prior law in an unbroken chain of permission and authority, which presupposes and entails a continuing identity for the legal state.

For the non-formalist the same state may exist before and after basic change if the consent of the people uninterruptedly permitted, supervised, and ratified the change; the same may be said of the vigilant eyes of armies and gods. If this continuity is lacking, both the formalist and the non-formalist will hold that the legal change brought by the discontinuity was unlawful or invalid. The difference is that this invalidity is much harder to repair for the formalist than for the non-formalist. If the new law is not authorized by prior law, then the formalist must resort to legal fiction to validate it, for example, an enactment that declares that the gap never occurred or that the new law was properly authorized by antecedent law.[Note 7] Without resorting to fiction, the formalist theory requires that change be continuous or illegal. The non-formalist may make the same appeal to curative fiction, but may also point to extra-legal conditions that remained continuous while the chain of permission and authority was broken, for example, the consent of the people or the support of the army. Such a move implies that otherwise invalid law can be validated directly by alegal sources, not just by prior law.

In this way both formalist and non-formalist theories of law distinguish between amendment and revolution, although to the formalist the line is rigid and to the non-formalist it is tentative and not a final difference of kind. Both allow revolution to be cured by subsequent ratification, and in that sense converted to lawful amendment, although for the formalist this must be through the sheerest fiction. The non-formalist may allow cure not only by ratification, but also by mere acquiescence or direct consent. For both, discontinuity invalidates, but can be cured retroactively. When cured, a history of legal change is a cumulation of lawful amendments, not a series of new beginnings.[Note 8]

How can an entity (system, rule, power, authority, agent) authorize its own change or destruction? Let us be precise. We are speaking of the modification of a constitution. Such change is authorized, when authorized, first by the AC, andconstitutionhatever it is that authorizes the constitution itself. But much basic change can occur without any strict self-amendment taking place. The legislative branch may be abolished, the executive absolutized, and cruel punishments permitted by ordinary amendments, without any legal entity authorizing its own change except in the broad sense in which any amendment comprises the use of the constitution's supremacy against itself (see Section 1.C). The articles pertaining to the legislature, executive, and punishments would not have authorized their own change or repeal; the AC would have authorized these changes. The broad view that any amendment is self-amendment blurs the importance of the inner articulation of parts in a constitution. Genuinely reflexive, or circular, self-amendment and self-repeal would not occur unless the AC were used to amend or repeal the AC itself.[Note 9] It is the latter that I wish to explore.

Is self-amendment an inherently discontinuous act, and hence inherently unlawful? When we think about a rule being displaced by a successor, we may imagine a temporal seam or adjacency. If the predecessor is the sole authority for the successor, then there is (or may as well be) no temporal overlap to guarantee continuity. But is overlap necessary? Do the bygone rules have any present power to authorize their successors? (See Section 10.)

We should resist the temptation to cast the problem wholly as a logical problem. The legal version of the paradox of omnipotence is first and foremost a legal problem, just as the theological version is primarily a theological problem. Hasty or exclusive treatment of them as logical problems may overlook the theories, very significant if true, that law-makers and deities may perform contradictions. In this sense, logic is just another metaphor for law. Like the metaphor of seams and overlapping, it should be used to reach clarity and discarded when it prevents us from seeing the reality as it is.

If the rule or power embodied by the words of an AC is abstracted and formalized, and treated solely as a proposition in a logical system rather than a rule in a legal system, then we may judge questions of self-application, limitations on power, and so on, by the false precision of formal logic. That is, we may let formal consistency be the supreme criterion of permissibility, as it is in logic, regardless of its place in the hierarchy of criteria in law. If the resulting view denies legality to acts which are legally commonplace and never challenged, then the model must be wrong, not as logic, of course, but as law.

An AC, then, allows the basic document to change in ways that would be revolutionary but for their permissibility. To speak of an "amendment clause" and of the constitution as a "document" is clearly to engage more than just words. The words of an AC create a power, and simultaneously channel and regulate that power. The reflexivity of an AC amending itself is more than the self-reference of words; it is also the self-application of a power. If legal rules and powers are more than the words defining them, then we are dealing as much with causation as with inference in discussing how they may be affected.

I will limit my discussion to the problem of the authority of an AC to amend itself (Part One) and some variations of this problem, chiefly concerning alternative methods of amendment and self-amendment (Part Two). If an AC can validly amend itself, then the new AC would be authorized, like any other amendment, by the old AC and by the constitution itself from which the authority of the old AC derconstitutionrns out that the authority of the constitution itself must be addressed in order to explain the legality of self-amendment (see Section 7). Sometimes a constitution is authorized by the AC of its predecessor, but that merely postpones the question of validity to the predecessor. I will generally defend H.L.A. Hart's theory that basic rules of law are made valid by the acceptance and usage of the people and officials, which I take to be a variant of Locke's consent theory. More precisely, I will argue that Hart's acceptance theory can explain the legality of self-amendment, and, while it has theoretical difficulties of its own to overcome, it is more plausible on its merits than the formalist model of law as a quasi-logical system (see Sections 6 and 21.D).

B. Amendment and democracy

The AC of a constitution, and even the technical problem of self-amendment, assume special importance on the classical American or Lockean consent theory that law and government derive their legitimacy from the consent of the governed. If the people's consent wanes and revolution is to be avoided, then a lawful method of basic change must be provided. For supreme, constitutional rules, the constitution change must be on a paconstitutiononstitution itself. The AC provides such a method. The pressure to change the constitution, and hence to use the AC, is admittedly irreflexive most of the time. Slavery must be abolished or alcohol prohibited. Rarely do citizens clamor for a change in the rule that authorizes change of the rules. But it has happened, and most of the clamor has focused on making the amendment process either much more difficult or much less difficult.[Note 10] The difficulty of the amendment process is closely related to the ability of citizens to manifest their consent to the constitution and hence (under a consent theory) to authorize it to govern them.

One might well argue that the citizens who vote to ratify a constitution are bound by it under a consent or contract theory. Those who consented to constitutionon election but voted against the constitution, and those who consented to the product of all lawful procedures but voted against every initiative, could be bound under a simple extension of the theory. But what about the future generations that never voted and nevconstitution to put ratification to a vote? Consent theorists must say either (1) that the constitution remains authoritative despite the lack of manifestations of consent by future generations, or (2) that it has continuing authority if and only if the people manifest continuing consent.[Note 11] The former alternative makes the first generation the proxy of all subsequent generations, and justifies the imposition of a system of law on a population whether or not it would have consented and whether or not its members would have appointed their ancestors as proxies. If that is plausible, at least it departs widely from classical consent theory and any authentic sense of consent. It also raises the omnipotence paradox of a sovereign binding its successors or its own future conduct, perhaps irrevocably. The second alternative requires some continuing show of authorizing consent which might well be inferred from the citizens' use and contentment not to use their AC.[Note 12]

One might say that citizens' failure to overthrow their constitution and government in revolution is a suconstitutionifestation of consent to give the constitution and constitutional form of government legitimacy. Continuing failure to revolt would be continuing consent. But this view amounts to the proposition that a regime has legitimate authority if, and so long as, it successfully puts down and postpones revolution. The theory is strengthened if it makes legitimacy conditional on the postponement even of attempted revolution. But this does not strengthen it enough —and in other respects over-strengthens it. For of course the prevention of revolution may be accomplished by soul-satisfying justice, or it may be accomplished by midnight break-ins, kidnaping, constant surveillance, indoctrination, intimidation, terror, and torture. The point of a consent theory of legitimacy is gutted if the requisite consent may be coerced.

But if the use and contentment not to use the AC is a measure of popular consent to the constitution, then the fairness and difficulty of the amending process must be brought into the calculation. The weight of the use and non-use of the ACconstitution of consent to be governconstitutionnstitution is (roughly) in inverse proportion to the difficulty of amending the consticonstitutionhe general public. Dissatisfaction could rise to a level needed for successful revolution and still not suffice to amend the constitution if the amending process required the unanimous votes of every state and county legislature, or referred the question to an aristocratic elite.

For my purposes here I need not decide the maximal degree of difficulty for an amendment process before it fails to capture the actual consent of the governed. However, I am intrigued by the idea that the manifestation of consent sufficient to adopt a constitution in the first place is the most that can be expected in constitutionvide continuing authority to that constitutioconstitutionns that if amendment is constitutionlt than the original ratification, then the constitution has lost some degree of authority or legitimacy. The authority of a constitution over generations of citizens who did not ratify it would diminish roughly to the extent that the difficulty they face in amendmconstitutionthe difficulty of the original ratification. For such citizens, legitimate amendment would be more difficult than a revolution or discontinuity that would establish a new constitution with equal or greater legitimacy than their ancestors had in adopting the old one.[Note 13] If amendment becomes impossible after the first generation, because there is no AC or because courts or tyrants invalidate all attempts under it, then authority under the consent theory would drop to about zero plus any surcharge which citizens accord to rules of law, qua rules of law, before deciding to disobey.[Note 14]

The consent theory of authority is widespread and long-lived. Without endorsing it here I may observe that it is expressly endorsed by the Declaration of Independence[Note 15] and many state constitutions, and that it puts great weight on the AC of a constitution and the difficulty it creates for the amending process. Because most attempts to change ACs are attempts to raise or lower that level of difficulty, the problem of the applicability of an AC to itself is of urgent practical importance. Under the consent theory the self-amendment of the AC may increase or decrease the opportunities for the people to constitutionir consent to be governed by that constitution.[Note 16] Self-limiting self-amendments, which most strongly raise the specter of paradox, may enhance or diminish the legitimacy of the constitution.

The consent of the governed as it appears in the amendment process not only requires that difficulty of the process be within certain bounds, but also that the consent meet some minimal standards of cognitive and volitional competency, like any other consent. Detailing the distortions to consent, hence to legitimacy, introduced by ideology and violence is generally beyond the topic here. But we may say that the validity of consent to an amendment may, for evidentiary purposes, be presumed conclusively from the requisite vote, but that does not preclude inquiry into the "actual" consent of the people. In Bedner v. King, 110 N.H. 475, 272 A.2d 616 (1970) the New Hampshire Supreme Court refused to find that the voters' consent to an amendment was invalid merely because publicity about its content and effects was inaccurate. Cognitive incompetency is apparently irrelevant in New Hampshire, or at least not to be inferred from false advertising alone.

Serious questions of volitional competency have been raised about the three Civil War amendments. The validity of the Thirteenth, Fourteenth, and Fifteenth Amendments has long troubled those who perceive them as exactions of the victor not voluntarily endorsed by the southern states whose votes were needed for ratification. Compare the case of Japan, whose constitution of 1946 was imposed on the nation by the Supreme Command for the Allied Powers (SCAP).[Note 17] The constitution was ratified by the Japanese Diet but under pressure from SCAP while Japan was under military occupation. For these reasoconstitutionnese scholars argue that the 1946 constitution is invalid.[Note 18] John Maki reports that among the most urgently demanded reforms is the self-amendment of the Japanese AC, which Commissioner Kanikawa objected is one of the most difficult to use in the world and "possesses the character of a law of occupation control enacted by the Allied nations."[Note 19]

I am not concerned to justify the consent theory per se or to suggest the optimal level of difficulty which the people should impose on themselves through their AC.[Note 20] I will concentrate hereafter on the theoretical problems of self-amendment, hoping that the reader will see behind the logical problem a very fundamental problem of legitimacy which implicates many of the basic values of our system of law and popular sovereignty.[Note 21]

The self-amendment of ACs has frequently occurred (see Appendix 2). Hence the question of the logical coherence of self-amendment is not merely academic. By exploring its logical and legal characteristics I hope to shed light on our hidden models of legal change and concepts of law itself, rather than to set logic over law as sovereign and retroactively declare the impossibility, or even question the possibility, of the commonplace.


1. George Tichnor Curtis, Constitutional History of the United States From Their Declaration of Independence to the Close of the Civil War, Harper and Bros., 2 vols., 1889-96, 2:152. [Resume]

2. Lester Bernhardt Orfield, The Amending of the Federal Constitution, University of Michigan Press, 1942, p. 210. [Resume]

3. Stevenson's remarks are quoted in Robert L. Farwell, "Gateway to What?" DePaul Law Review, 10 (1961) 274-85 at p. 278. See also Charles L. Laughlin, "A Study in Constitutional Rigidity," University of Chicago Law Review, Part I, 10 (1943) 142-76, Part II, co-authored with Kenneth G. Sears, 11 (1944) 374-442. [Resume]

4. Constitutional amendment through means other than amendment clauses is surveyed in Part Two. [Resume]

5. John William Burgess, Political Science and Comparative Constitutional Law, London: Ginn & Co., 2 vols., 1891, 1:137. [Resume]

6. As I will use the term, an "amendment clause" or "AC" is any rule authorizing the change of the rules of the -->constitutionion. When the constitution is written, then the AC is usually explicit and easy to find, but not always. I intend this broad definition to capture tacit amendment rules of other constitutional systems.

When it is explicit, of course, it is not just a rule but a provision or specimen of language. The difference between an AC qua language and an AC qua rule is considerable, and will be observed whenever it is relevant. "Amendment clause" ("AC") covers both senses.

One reason that the paradox of self-amendment cannot arise in deontic logic (the logic of obligation and permission, often used in analyzing law) is that no equivalent of an AC so defined exists in any such system. Whether an equivalent may be constructed —a fascinating challenge— is beyond my topic here. See Section 12.C and Appendix 3. [Resume]

7. See H.L.A. Hart, The Concept of Law, Oxford University Press, 1961, pp. 115-16. [Resume]

8. In subsequent sections I will call the formalist theory the "inference model" and two non-formalist theories the "acceptance model" and "procedural model". See Sections 6, 7, and 10. [Resume]

9. In Part One I will discuss the general application of the AC of a system to itself. In Sections 4 and 13 I will discuss the application of one section of a complex AC to another section as a method of amending the AC without strict self-application. [Resume]

10. Orfield, op. cit., Chapter VI. [Resume]

11. Our Lockean ears resonate with the proposition that the people are sovereign and that they are bound to obey their laws by contract principles. Yet the paradox of omnipotence arises in another form if the first generation of sovereign people can bind its successors (see Section 8). The adoption of a constitution with an AC is a revocable act, because the AC permits piecemeal change and wholesale replconstitution long as the establishment of the constitution is revocable by later generations, and the method of amendment is fair, then the first generation is not oppressively bindingconstitutionors. But if the method of amendment is not fair, or is too difficult, then the constitution inherited by future generations does oppress and is partially illegitimate. The Lockean consent theory is strengthened as a normative theory of justice, and protected from the paradox of omnipotence, if we insist that the legitimacy of law requires the continuing consent of the governed, not just the consent of the founding generation. [Resume]

12. Paul Brest makes the point that failure to amend does not show consent to the constitution in its current form, but only constitutionnt judicial interpretation of the constitution. Paul Brest, "The Misconceived Quest for the Original Understanding," Boston University Law Review, 60 (1980) 204, 236. This point is well taken. To the extent that the constitution is its current judicial interpretation for purposes of application, the point I wish to make survives Brest's qualification. Under the consent theory the changes in the constitution that are necessary to sustain the approval of the people through changing circumstances may arise in any way consistent with that continuing approval, e.g., by judicial review (see Section 15), by international agreements (see Section 17), by desuetude (see Section 19), or by use of the methods of the AC. In fairness to Brest I should note that, while he believes the consent theory must be extended to require continuing consent, he ultimately rejects even the extended theory. Id. at 225ff.

An excellent general discussion of the politics of self-binding may be found in Jon Elster, Ulysses and the Sirens: Studies in Rationality and Irrationality, Cambridge University Press, rev. ed. 1984, esp. pp. 87-103. [Resume]

13. The reasonableness of this idea is suggested by the simplified situation in which all the parameters of difficulty in the amendment process are focused on the supermajority needed to adopt. If the first generation (G1) needed 70% approval to adopt the constitution, and the second generation (G2) needed 75% to amend, then failure by G2 to aconstitution