Section 9
Entrenchment, Self-Entrenchment,
and Disentrenchment
of the Amendment Clause Itself
Peter Suber, Paradox of Self-Amendment Table of Contents

A. Types and distinctions

Because entrenchment purports to limit an amending power, it may be reflexive in two ways. First, self-entrenchment (as defined in Section 8) is the entrenchment of a unit of language by words within that unit or, if the entrenchment clause is implied, then by a rule that is inferred from language within the unit it entrenches. The unit may be as large as the whole constitution (e.g., "Nothing in the constitution shall be amended except as provided..."), or as small as the self-entrenching language plus an irreflexive reference (e.g., "Neither rule R nor this sentence shall be amended").[Note 1]

Note that a clause that completely entrenched the whole constitution would thereby deny any effect to the amending power, and to that extent would be less a limit on the AC than a virtual repeal of it. On the other hand, the incomplete entrenchment of the whole constitutionconstitutiony what obtains today, since the constitution cannot be amended at all except by certain "special" procedures.

Second, the language describing the amending power (the AC) may be entrenched, or even self-entrenched in the sense just described. Entrenchment of the AC is reflexive in the sense that the amending power is prohibited from applying to itself. Some self-reference, express or implied, is necessary, even though the entrenchment clause forbids the AC's self-application. If entrenchment clauses that are not also self-entrenched are called "irreflexive" entrenchment clauses, then irreflexive entrenchment of the AC is reflexive because self-referential, irreflexive because not self-entrenched.

The two senses in which entrenchment may be reflexive —self-entrenchment and entrenchment of the AC— may be combined to yield self-entrenchment of the AC. This self-entrenchment may in turn entrench the AC in whole or in part. Note that the irreflexive entrenchment of the AC can be defined by a clause outside or inside the AC itself. If outside it might say, "Nothing in the AC shall be amended." If that clause were inside the AC, then it would become a self-entrenchment clause. Nevertheless a clause inside the AC might constitute irreflexive entrenchment of the AC if reworded to say, "Nothing in the AC, except this sentence, shall be amended."

The last section was devoted to the irreflexive entrenchment and self-entrenchment of clauses of a constitution other than the AC. This section is devoted to two reflexive types of entrenchment: the irreflexive entrenchment of the AC and the self-entrenchment of the AC. It also covers a few variations on entrenchment itself that did not fit into Section 8.

Entrenchment of clauses other than the AC was discussed at all because it threatens the omnipotence of the AC, which raises in constitutional law all the issues of the paradox of omnipotence as they arise in theology. A well-entrenched constitutional rule is the legal equivalent of a stone made by God that God cannot lift. Moreover, self-amendment (which may be paradoxical independently of the paradox of omnipotence, or even with an AC of finite power) might permit transmutation of immutable rules, and therefore the achievement of omnipotence by a previously finite rule of change. The entrenchment of the AC itself also raises the question of limited or unlimited power: if entrenchment or self-entrenchment of the AC is immutable, then the AC is not omnipotent, unless it can transmute its immutable limitations. But entrenchment and self-entrenchment of the AC also more directly raise questions of self-amendment. Even if ordinary or self-limiting self-amendment is possible, shall we allow self-disentrenchment or self-apotheosis? Or if transmutation of the immutable entrenchment of clauses other than the AC is possible, then would the transmutation through self-amendment of the immutable entrenchment of the AC itself be possible?

If self-amendment is self-contradictory but permissible, can we deny the mantle of permissibility to self-disentrenchment of immutable entrenchment clauses just because it is self-contradictory? Or, if self-amendment is a permissible form of self-reference and self-application, then may it be made impermissible in certain cases by "aggravating" its self-reference and self-application in the form of an immutable self-entrenchment clause entrenching the AC and itself? If so, can the AC self-impose such an immutable limitation in an act of permissible self-amendment?

In order to avoid becoming dizzy and to simplify references to complex, nested puzzles, I must distinguish and assign conventional names to several different phenomena. Suppose that an AC has two sections, 1 and 2, and that 1 contains the entrenching language. If 1 entrenches only itself, call that "immediate" self-entrenchment. If 1 entrenches both itself and 2, call that "mediate" self-entrenchment. These terms may be justified by noting a parallel distinction among types of self-reference: immediate self-references refer directly to themselves, while mediate self-references refer to classes of which they are members.[Note 2] A sentence or reference which is neither mediately nor immediately self-referential will be called "irreflexive". A sentence which refers to itself and to Napoleon ("This sentence and Napoleon are short") refers to more than merely itself, but does not refer to a class of which it is a member. By contrast a sentence which refers to sentences and to Napoleon ("Sentences, unlike Napoleon, are immortal") refers to a class of which it is a member and to something outside that class, but not directly to itself. I will call any mediate or immediate self-reference that is combined with an irreflexive reference, "eccentric", and any immediate or mediate self-reference without any such irreflexive reference, "concentric".[Note 3]

If AC 1 entrenches AC 2 without entrenching itself, then no "self-entrenchment" in the strict sense has occurred. I will call that simply the entrenchment of 2 by 1, or the "irreflexive" entrenchment of 2 by 1, for emphasis. An AC of only one section which included entrenching language could be called immediate or mediate self-entrenchment, but I will call it immediate.[Note 4] I will carry over the distinctions from Section 8 between complete and incomplete entrenchment, between original and self-imposed entrenchment, and between limitations on amendment by protection (or as contradiction) and limitations by incompetency (or as ultra vires). Each of these distinctions applies within irreflexive entrenchment as well as within self-entrenchment. "Disentrenchment" is the repeal of an entrenchment clause, and "self-disentrenchment" is the repeal of an entrenching clause on a rule of change by that very rule of change. If a clause said that the AC could not be amended at all, then the AC could repeal that clause only by self-disentrenchment.

Complex variations and combinations of entrenching language are not to be expected in actual constitutions, for most drafters are not genuinely worried that their ACs (unlike many substantive provisions) will be amended in haste or carelessness, and most are concerned that the constitution ring with solemnity and elegance, not perplex like a brain-teaser. Another reason is certainly that a sovereign people will not ordinarily want to limit its power to make law, although often the founding generation does want to limit the power of future generations. One may hazard the generalization that entrenchment clauses are less likely to appear in constitutions written and adopted by the people governed by them than in constitutions written and imposed by a foreign or imperial power.[Note 5]

Among the American states the most common type of entrenchment clause ambiguously proscribes "violation" of large sections of the constitution (more on the latter in Section 18). These clauses usually entrench either the whole constitution, the Bill of Rights, or the AC-like right toconstitutionolish government. The Delaware constitution of 1776, for example, contains such a clause in its AC. One can only conjecture that the prohibition of "violatioconstitution to include the prohibition of amendment as well, for it is superfluous for a constitution to prohibit its violation literally,[Note 6] and it is suspect to prohibit violation of only some sections. This is supported by the Oxford English Dictionary, which lists as 18th century meanings for "to violate": to defile, despoil, treat with violence, disturb, and break in upon. The North Carolina constitution of 1776, similarly, quasi-entrenches its Bill of Rights with the phrase, "ought never to be violated on anconstitutionhatsoever." Arkansas in its 1836 constitution, and Tennessee and Pennsylvania in their present constitutions, take a step toward explicitness by entrenching their Bills of Rights with the declaration that those Articles are "excepted out of the General Powers of government, and shall forever remain inviolate."[Note 7] The word "inviolate" in this context suggests "pristine and unamended" more than "not transgressed".

The Arkansas clause was held effective as an incomplete entrenchment clause in 1851. The Declaration of Rights entrenched by the clause included a provision prohibiting criminal penalties except for offenses charged by indictment or presentment. An 1846 amendment that would have given justices of the peace jurisdiction over assault and battery cases was held to violate the indictment rule and its entrenchment clause. The court held that only a constitutional convention could amend the Declaration of Rights. Eason v. State, 11 Ark. 481 (1851).[Note 8] The court found the "inviolate" language not only an effective entrenchment clause, but also an impliedly incomplete entrenchment clause that permitted amendments by convention. However, the neat "entrenchment" reading of Eason is complicated by the fact that the court also relied on the rationale that the Declaration of Rights is inherently beyond the ordinary (non-conventional) amending power.

But Arkansas, Tennessee, and Pennsylvania do not say anything nearly as explicit as the present and only (1911) constitution of New Mexico, in which one section of the AC says of another that it "shall not be changed, altered or abrogated in any manner except throuconstitution convention called to revise this constitution as herein provided." As we will see, this entrenchment clause was violated, if not repealed sub silentio, with the approval of the state Attorney General. (See Section 9.B below.)

Of the American states, only New Mexico has a clearly entrenched AC which has been self-disentrenched or held by courts to be capable of self-disentrenchment. No state has a self-entrenched or completely entrenched AC. In New Mexico there is legal authority for the proposition that self-disentrenchment of an AC is lawful, but the authority consists only of a few Attorney General opinions that are contradicted by other Attorney General opinions. This is tenuous evidence of legality, and it is a small sample. Self-amendment per se is undoubtedly lawful in America because it has occurred in 47 states without challenge e.g. for self-contradiction, paradox, violating its authority or principles of natural law (see Appendix 2). But one might say that New Mexico was "wrong" to uphold the permissibility of self-disentrenching self-amendments, in a way in which 47 states cannot be wrong. After all, there is reason to believe that New Mexico's disentrenching self-amendment did violate the state constitution. But in one sense that is just the point. Despite the paradox and the contradiction, these acts are accepted as legal by the tests of legality —in New Mexico. In this Section I will be more concerned with the logical than the legal objections, although I do note the small sample and the dubieties of the New Mexican case.

B. Reflexivity tangles in New Mexico

If the paradox of self-amendment were never raised by Alf Ross, it would be raised by the spectacle of the New Mexican AC. Its legal history is wonderfully complicated. The New Mexican AC contains five sections, the first of which contains a complete description of the non-convention method of amendment. Section 5 explicitly entrenches 1 against all amendment except amendment by convention. This is a clear-cut case of incomplete, irreflexive entrenchment within an AC. But this entrenching language was violated in a fascinating way. The entrenched rule was impliedly self-amended by an explicit amendment to another section of the constitution, with the effect that the amended version of AC 1 is incompletely, mediately self-entrenched. Here's how.

Section 1 of the AC said, inter alia, that "any this constitution may be any regular session" of the legislature. In 1964 AC 1 was used to amend a provision on legislative power (Article 4, 5.B), limiting the types of action which the legislature could consider in even-numbered years to a short list that did not include constitutional amendments. The amendment impliedly amended AC 1 by reducing the years in which constitutional amendments could be considered from all years to odd-numbered years. This effect on AC 1 was confirmed by Attorney General Opinion No. 65-212 (1965). Therefore, AC 1 was amended without a convention despite its incomplete, irreflexive entrenchment by AC 5. Unlike the Australian and South African attempts to circumvent incomplete entrenchment (see Section 8.B, above), this one succeeded, perhaps because the protected rule was only impliedly, not directly, amended. This would not be a tidy legal justification, but it might mean that few people noticed or cared.

Moreover, this implied amendment was an implied self-amendment, since AC 1 was used to amend the legislative article with the side effect of impliedly amending AC 1.

Moreover, because the amended AC 1 now says that constitutional amendments may be considered by the legislature only in odd-numbered years, it has become incompletely, mediately self-entrenched.

Moreover, the entrenchment clause that was initially violated (AC 5) must be considered still valid, not repealed pro tanto, for the people of New Mexico have tried —and failed— to repeal it in 1965, 1970, and 1971. These attempted repeals, furthermore, have not been by convention. If a certain electoral return would signal a successful repeal (which is undetermined), then the irreflexive entrenchment of AC 1 by AC 5 should not be taken to imply the self-entrenchment of AC 5. That an ordinary amendment could repeal the entrenchment clause was held by Attorney General Opinion No. 70-13 (1970).

Moreover, the new rule to consider amendments only in odd-numbered years has frequently been violated, eight times in 1970 alone. It may be considered no rule at all, either because the implied self-amendment is really invalid, or because the implied self-amendment is ineffective and rapidly losing to desuetude.

The Attorney General Opinion that declared that AC 1 had impliedly amended itself ignored the fact that AC 1 was entrenched by AC 5. Attorney General Opinion No. 65-212 (1965). In this sense, then, the Attorney General was answering a self-amendment question, not a self-disentrenchment question. The narrow ruling was that resolutions and constitutional amendments could be considered by the legislature only in odd-numbered years. This Attorney General was Boston E. Witt, and when he was replaced a few years later by James A. Maloney the validity of the implied self-amendment and self-disentrenchment was again brought to the Attorney General's office. Maloney was asked to reconsider Witt's opinion, and was specifically asked to address the self-disentrenchment question.

Maloney upheld Witt's narrow ruling but on new grounds that must be sketched briefly. Prior to 1964 the New Mexican legislature met only in odd-numbered years. When the constitution was amended to provide for annual sessions, the subjects that could be considered in even-numbered years were simultaneously limited, and did not include constitutional amendments. The language of AC 1 that was apparently amended by implication said that amendments could be considered in any "regular" session. Witt took this to designate the new annual sessions, and therefore held that AC 1 had been amended. Maloney read "regular" in AC 1 in its pre-1964 sense as odd-numbered sessions only. He argued that the framers intended to define "regular session" in the AC by reference to the legislative article, which at that time provided only for regular sessions in odd-numbered years. Maloney ignored the amendment of the legislative article on this very point, or more precisely, he held the amendment irrelevant to the meaning of "regular" in the AC. Hence, AC 1 was not amended at all, and had always allowed amendment proposals only in odd-numbered years. Maloney did not indicate whether this ruling implied retroactive invalidity for any ratified amendments. In sum, then, Maloney affirmed Witt's narrow ruling, but denied that self-amendment had occurred, let alone self-disentrenchment. Attorney General Opinion No. 69-105 (1969).

In 1970 Maloney was asked whether the incomplete, irreflexive entrenchment clause in AC 5 could be repealed. He ruled that it could be, and that it could be repealed without a constitutional convention, even under AC 1. Attorney General Opinion No. 70-13 (1970). This is the closest thing in American legal history to authoritative recognition of permissible self-disentrenchment. New Mexico's AC 5 incompletely entrenched AC 1, but 1 was declared capable of repealing 5 without using the special procedures required by 5 (constitutional convention) —a procedure not contained in 1 anyway. The section describing the power to amend was 2. Maloney, however, did not in his opinion visualize the problem as one of paradoxical self-disentrenchment, but as a problem unique to New Mexican history.

If AC 5 was required by the federal government as a precondition of statehood, then would repeal of 5 jeopardize New Mexico's status as a state? Maloney decided that AC 5 was not actually required by the federal government; instead, Congress merely wanted a vote on an amendment to the AC that would make the amending process less difficult. The self-amendment that Congress wanted to be put to a vote would have lowered the majority needed for ratification from two-thirds to a simple majority. The entrenchment clause was already in the pre-statehood constitution, but Congress included it in the AC-package.

Maloney is opposed on this question by Helene Simpson,[Note 9] who has documented her opinion that an amended AC, not merely a vote, was required by Congress as a precondition of admission to the Union.[Note 10] However, even Simpson agrees that Congress did not require AC 5 as a precondition.[Note 11] Maloney and Simpson both note that Arizona was also required to amend its constitution, although not its AC, prior to admission, and that upon admission repealed the required amendment without losing its statehood or running afoul of the courts.[Note 12] Maloney contends that, in any event, no present amendment of the New Mexican constitution could endanger New Mexico's status as a state. The statehood issue clearly overwhelmed the self-disentrenchment issue, as well it might under New Mexico's unique circumstances: no other state was required by Congress to change its AC, or even to vote on changes, as a precondition of admission, even though many other states used the same two-thirds supermajority for ratification of state amendments.[Note 13] In Arizona where the required changes were quickly repealed, the changes had not been self-amendments nor the repeals self-disentrenchments.

Later in the same year Maloney finally reached the self-disentrenchment question when he repeated his earlier ruling that AC 1 had not been amended by implication. This time he based his reasoning on the impermissibility of any amendment of AC 1 except by convention. Attorney General Opinion No. 69-151 (1969).

The underlying issue was still not dead, however, and in 1970 Maloney faced a new dimension of the problem: if the earlier opinions agreed that constitutional amendments could not be proposed in even-numbered years, then what is to be done about the eight that were proposed in 1970? Maloney did not directly say that a court could nullify them if ratified, or enjoin their further consideration, or suggest that he would use his prosecutorial discretion toward these ends. Instead he personally doubted that a court "would" nullify them if ratified, if only because the people would have expressed their will and the courts indulge every presumption in favor of the validity of amendments that the people have ratified. Maloney was evidently trying to say, without abdicating his role as the highest law enforcement officer in the state, that acceptance can cure defects in the most fundamental procedures. In Hart's terms, acceptance can validate laws that, through some defect, are not validated by the rule of recognition. Maloney concludes in a most humble mood, saying that while two different Attorneys General have found that constitutional amendments cannot be proposed in even-numbered years, relying upon different interpretations of the constitution, nevertheless they had not exhausted the reasonable interpretations and a court might disagree with them both. While standing by his opinion, heconstitutionny that a contrary reading of the constitution could be made to appear reasonable, although he can give no hint how. Attorney General Opinion No. 70-81 (1970).

In New Mexico, in sum, a rule that was incompletely entrenched against all amendment except by convention was impliedly amended without a convention, according to one Attorney General. From another angle the same phenomenon may be described as the disregard of a limitation on the amending power in the form of an act of self-amendment. The AC (in 1) did not disentrench itself, or repeal the limitation on its power, except impliedly and pro tanto; it merely "violated" that limitation and got away with it. Moreover, an Attorney General in an advisory opinion held that the limitation may be repealed without any special procedures, even though he held in a prior opinion that it could not be violated by an ordinary amendment. The incomplete, irreflexive entrenchment clause is not impliedly self-entrenched and may be repealed without resort to the special procedure it names. Such repeal would constitute genuine self-disentrenchment, but of a less objectionable sort: repeal of an unentrenched entrenchment clause rather than a self-entrenched entrenchment clause.

This implied self-amendment and self-disentrenchment have not been struck down by the courts and have been upheld by the Attorney General —one ruling that the self-amendment actually occurred, another ruling that the self-disentrenchment actually could.

Another entrenchment clause in the New Mexican AC has not been so fortunate. Within AC 1 itself a clause incompletely entrenches two sections in Article 7 (on voting) and two sections in Article 12 (on education). While ordinary sections may be amended by a simple majority of the electors, these four sections may only be amended by a three-fourths supermajority with the added requirement that there be at least a two-thirds supermajority in each county. The latter requirement makes the New Mexican counties into entities much like the states in the federal Senate; they are represented in these special amendment referenda as units, not by population. For that reason the two-thirds requirement was struck down by the New Mexican Supreme Court for violating the one-person, one-vote principle of the Fourteenth Amendment. State v. State Canvassing Board, 78 N.M. 682, 437 P.2d 143 (1968).[Note 14] It is probably significant that the court struck down the two-thirds requirement after 11 attempts had failed to amend one of the entrenched sections. The three-fourths supermajority requirement was not challenged and still stands. Both the three-fourths and two-thirds requirements were added to AC 1 in the vote demanded by Congress, and were drafted by Congress. The court gave no hint that entrenchment, or at least the incomplete entrenchment before it, was impermissible in itself. One may doubt whether the particular entrenchment clause before the court would have been found to violate the Equal Protection clause if it had not already blocked 11 attempts to amend one section. The normal rule across the country is that the one-person, one-vote principle does not apply to elections that ratify state (or federal) constitutional amendments. See Jackman v. Bodine, 78 N.J. Super. 414, 188 A.2d 642 (1963). To the extent that the history of blocked attempts at amendment influenced the court, the rule of law was covertly bent to accommodate the perceived will of the people, which in Hart's jurisprudence is not only justified, but valid law if subsequently accepted by the people and officials.

The two Attorneys General disagreed in their interpretations of the constitution, but the effect was that one allowed self-amendment in violation of an entrenchment clause, and the other allowed self-disentrenchment of that entrenchment clause. The inference model cannot tolerate either act; but even if it could somehow tolerate self-amendment, it must still balk at self-disentrenchment. If an AC is completely and concentrically self-entrenched, the hardest case (and one that did not arise in New Mexico), and if self-amendment is otherwise permissible, then for self-amendment to be possible in this case the self-entrenching language must be (1) repealed before self-amendment, or (2) repealed simultaneously with the self-amendment of the entrenched AC, or (3) simply violated with impunity. The third is obviously impossible for the inference model. The first is also impossible, for the inference model, because it would require an act of self-amendment to repeal the entrenching language, and therefore could not take place without acting before it acted.

The second is also impossible under the inference model, but for different reasons. If complete entrenchment and self-entrenchment clauses mean anything, they prohibit amendment until they are themselves amended or repealed. If they could be repealed simultaneously with the repeal of the rules they protect, then they would have been no barrier whatsoever to amendment. Even without the exotic logic of self-application, most courts would indulge the presumption that the words in a constitution were not surplusage. The inference model is not committed to preserving the effectiveness of entrenching language as a barrier to amendment, but it is committed to taking rules of change seriously as premises of logical inferences. An explicit entrenchment clause, for the inference model, therefore creates a condition precedent to amendment, namely, the repeal of the entrenchment clause. An AC that could undo its self-entrenching language and amend itself in the same act would be performing the logically prior and the logically posterior simultaneously. This is equivalent to the prisoner opening the outer door simultaneously with the inner door, which is as paradoxical as opening the outer door first. The inference model, then, requires that disentrenchment and self-amendment of the protected rule take place in two steps, not one. The rub is that it cannot allow the step that must come first to occur unless the second has already occurred.

If we put the inference model to one side, then there seems to be no reason why the entrenching language and the rule it protects could not both be repealed in the same act. Douglas Linder rightly notes that "[o]nly a hide-bound formalist would contend that the difference [between one and two amendments] is significant."[Note 15] Linder's view that the right one amendment could do the work of any two is based on the possibility that Congress may intend to repeal both the entrenchment clause and the rule it protects at the same time. That is one non-formalist path out of the problem, but unfortunately Congress is only one of at least 39 legislatures (today) whose intentions would be relevant to one who put stock in the intentions of the law-makers who adopt constitutional amendments. Moreover, intent cannot suffice, for the intent of the framers and adopters of the entrenchment clause was that it should not be repealed, or at least not so easily that it never protected the entrenched rule. Hence, unless the old intent gives way before the new, an irresistible force meets an immovable object —intent v. intent— and the paradox of omnipotence is replicated at the level of the intent of the law-makers.[Note 16]

One type of self-entrenchment of the AC can always be disentrenched by a single amendment; even the inference model would allow it, if it allowed any self-amendment. That is the case of a procedural limitation on the amending process, for example, a limitation on the number of amendments that could be submitted for ratification at the same time. Such limitations are self-entrenchment clauses because (or when) they apply to all amendments and therefore to their own amendment. They are incomplete because they do not forbid all amendment, but only create a hurdle in the path of amendment; yet they specify the same hurdle for "ordinary" amendments and to that extent may not seem "entrenching" at all. Probably the most discussed such limitation, and most discussed self-disentrenching self-amendment that relaxed it, was Illinois' "Gateway Amendment" of 1950.[Note 17] The amendment raised the limit on the number of amendments that could be considered at one time from one to three. It was called the "Gateway Amendment" because it opened up the amending process. Obviously, if self-amendment is permissible, then one amendment suffices to repeal this type of entrenchment clause.

But note that there is no distinction here between the entrenchment clause and the rule that it protects. Only clauses that forbid certain amendments by content create a difference between entrenchment clause and entrenched rule tht might lead a formalist to demand two amendments to repeal both. Only the latter type of clause could construe its own repeal as a violation, and thereby purport to establish its own immutability, or its mutability only on the condition of being repealed before it is repealed. Clauses that forbid certain amendments without reference to content, but only to procedure, do not create the same paradox.

Note that the inference model requires two amendments and prevents them from ever achieving their end. The entrenching language must be repealed before the entrenched rule can be repealed, but the entrenching language can never be repealed under the inference model. All attempts are violations, not repeals, because the self-entrenching language forbids its own amendment as well. The inference model requires two amendments in the sense that a prisoner must open two doors before escaping, and they cannot be opened simultaneously; but if the outer door must be opened first, then neither can be opened —by the prisoner— at all. This dilemma pinches a self-entrenched AC even harder, for it is imprisoned within the double doors. But any self-entrenched entrenchment clause creates the double door prison for some rule, and an AC outside the "prison" cannot free it under the inference model because the repeal of the outer protective clause is always a violation.

Finally, note that it does no good under the inference model to suppose that a self-entrenchment clause could be self-disentrenched because its self-reference has rendered it meaningless and ineffective. If the AC is the self-entrenched rule, then disentrenchment must take the form of self-amendment, which equally involves self-reference. Still, if the barrier is a mirage, then the escape may be a fantasy. If the AC is not the self-entrenched rule, and if the self-entrenchment clause were meaningless, then of course it constitutes no barrier to the amendment of the "entrenched" rules. But the use of self-reference in law cannot be considered to create meaninglessness if self-referential rules and procedures are accepted by officials and citizens (logicians excepted, perhaps) as having meaning. The technical —and contestable— theories of meaning published in philosophy and logic journals have never been, and need not be, the standards used in law. The legal standards are equally technical and contestable, but the point is that they are autonomous and self-sufficient. Just as courts decide insanity on legal grounds, using medical and psychiatric testimony in an advisory capacity only, courts can decide meaningfulness on legal grounds and will take the testimony of logicians "for what it is worth".

When the AC is not entrenched reflexively, completely, or concentrically, then the cases are easier to analyze. Incomplete self-entrenchment, whether mediate or immediate, could always be undone if the proper, difficult procedure were used, and if self-application is allowed. For example, the only limitation on Article V that is still valid is an incomplete, irreflexive entrenchment clause: that "no state, without its consent, shall be deprived of its equal suffrage in the Senate." This may look like complete entrenchment, forbidding a certain type of amendment, but as noted in Section 8 it is actually incomplete. Amendments depriving some states of their equal suffrage in the Senate are not precluded if a particular, difficult procedure is used: getting the consent of each victim state.

It may look self-entrenched as well, but in fact it only (explicitly) entrenches the rule (derived from Article I, 3.1) that the states shall have equal representation in the Senate; it does not entrench itself or forbid its own amendment. However, self-entrenchment is implied for most commentators, and justifiably to the extent that irreflexive entrenchment is pointless without self-entrenchment. The crudest way to infer self-entrenchment is to posit or assume complete self-entrenchment, even though the clause itself is merely an incomplete entrenchment clause. This was the approach of Justice Wayne in Dodge v. Woolsey, 59 U.S. 331, 348 (1855) when he referred to the equal suffrage limitation as one of two "permanent and unalterable exceptions to the power of amendment."[Note 18] It makes more sense to infer incomplete self-entrenchment at the same level of difficulty as the entrenchment of the protected rule. That means that the equal suffrage limitation could be repealed if every potential victim consented, which requires unanimity.[Note 19]

Irreflexive entrenchment can be undone even under the inference model, for the disentrenching act is neither amending itself nor amending an entrenched section nor entrenched against amendment. Under the inference model complete immediate self-entrenchment may be as impossible to undo as complete mediate self-entrenchment. If the section of the AC that describes the power to amend is the power that is immediately self-entrenched, then it must, but cannot, repeal its entrenching language before it can perform any self-amendment. But if the part immediately self-entrenched is not the part describing the power to amend, then the latter part may apply irreflexively to the former and repeal the entrenching language and amend the underlying rule. This of course is not "self-amendment" except that both the amended and amending sections are part of the AC.

If the language that entrenches the AC is outside the AC, then two possibilities arise that require separate treatment. First, if the entrenchment clause is not self-entrenched and not held to be self-entrenched, then it may seem to be repealable by ordinary amendment like any other unentrenched rule. It does not forbid its own amendment (ex hypothesi), but only forbids the amendment of the AC. If one interpreted the repeal of the clause as an amendment to the AC, because the AC will have divested itself of a limitation on its power, then one must consider the clause impliedly self-entrenched; if not, then it may obviously be repealed by ordinary amendment.

Second, if it is self-entrenched or impliedly self-entrenched, then it is more difficult to dispose of. Even if self-entrenchment clauses are repealable by extraordinary procedures like transmutation (see Section 8.A), then a clause entrenching the AC and itself is still in a different class. In a case of "ordinary" transmutation, the rule to be repealed purports to be immutable and to prohibit the amendment that would repeal it. The AC might change itself prior to transmutation to give itself explicit power to repeal protective language as well as to give itself the benefit of recency (a tie-breaker under some canons of interpretation). But if the AC is prohibited to amend itself and if the prohibition also prohibits its own amendment, then the AC cannot acquire new, explicit language by self-amendment prior to self-amendment. Although the logic of this situation is more complex than ordinary entrenchment, and the barriers to repeal are more formidable, the overkill of insurmountable hurdles, the answer is the same as for ordinary disentrenchment: under the inference model the multiplication of impregnable walls simply reinforces the conclusion of irrepealability, while under the acceptance model repeal is permissible if accepted. The theory cited by a judge to uphold such disentrenchment might be more tortured and elliptical, and its implausibility might contribute to a climate in which acceptance fails to grow. In fact, and this is the sharp edge of the theory, the legal rationale might be self-contradictory, or illogical for some other reason, without thereby ceasing to articulate the acceptance of the people that authorizes the amendment.

Moreover, the judge is not wholly without theories to cite, for she may appeal to the continuing omnipotence of the people, the undemocratic consequence of the rule of the living by the dead, the anti-democratic intention to bind future generations irrevocably, the perilous enhancement of revolution as a political option, and so on. She might also cite the precedents of complete entrenchment repealed or declared repealable, such as Opinion of the Justices, 263 Ala. 158, 81 So.2d 881 (1955), and McCullen v. Williamson, 221 Ga. 358, 144 S.E.2d 911 (1965).[Note 20]

C. Self-disentrenchment of the AC

What if a self-entrenched AC forbids not only the amendment of the AC but also the repeal of the entrenchment clause? We should assume that a clause that says anything like "this AC shall not be amended", when the clause is part of the AC, already forbids both. That condition is built into the definition of self-entrenchment. Logically, it may seem that an amendment that would repeal the entrenchment clause violates, not repeals, the language of the constitution. Indeed, that is how the formalist or inference model sees it, and this reading would be shared by even more legally sophisticateconstitutionlaw that interpreted words of the constitution by their "plain meaning". But disentrenchment is permissible in New Mexico despite plain meaning and the logic of violation. In New Mexico the rule protected by an entrenchment clause was allowed to amend itself in violation of the clause, and the Attorney General gave an advisory opinion that the entrenchment clause could itself be repealed without any special procedure. Both these acts would seem to violate, not repeal, the clause, but they were declared permissible.

A valuable perspective on the slippery and fundamendal distinction between violation and repeal exists in the common law. Most cases sit atop a history of precedent that contains inconsistent rulings. No matter how the next court decides a case, it will be "violating" valid law in the form of some living precedent. Even if one denies that this is inevitable or frequent (see Section 21.B), one must acknowledge that it is permissible. Yet the decision that "violates" valid rules of case law is not a simple violation; it makes new law itself. That is part of the reason why it escapes nullification for its violation. (Another part of the reason is that the decision itself argues that the "violated" precedents were distinguishable, or inapplicable to the case, in light of stronger precedents, and these arguments were persuasive to a good number of people either in fact or pro forma.) The decision might even be taken to repeal the old rule pro tanto.

Similarly, in international customary law there may be a dearth of custom on a certain question, but the little that exists might all point in the same direction. An act that departs from that custom may for that reason be a violation or the beginning of a new custom. When an act with law-making effect in the present violates laws made by similar acts in the past, then it is not a crime, a delict, or nullity; it is an amendment or repeal. It is a repeal because the act that made it is a rule of change for laws of its type, just as common law decisions may repeal rules of common law, custom-making acts and trends may repeal customary law, statutes may repeal statutes, and constitutional amendments may repeal constitutional rules. This is contingent, however, not necessary. It just happens that in our legal system every law-making power is also an amending power or rule of change for the type of law it makes. Conceivably a judge could create common law rules but not amend them; indeed, many who insist upon unwavering adherence to precedent forget that stare decisis permits occasional change and repeal of rules. This rule, then, that all law-making powers are amending powers for the type of law they make, should only be applied where there is a history of application. It should not be used, except by analogy, to conclude that an AC cannot make an immutable self-entrenchment clause, because we just do not have enough case law on the question to know. The principle is not an a priori truth.

Even if a constitutional rule is completely self-entrenched, and even if that rule is the AC, the entrenchment clause may be repealed under the acceptance model. The desire for repeal will itself contribute to the climate of acceptance, even if it is not officially incorporated by rules to indulge all presumptions favoring amendments ratified by the people. If the people want to repeal the language entrenching their AC so that they may amend their AC, they will do so. The only criterion of the legal validity of their act is the procedure outlined in the AC to be disentrenched, although a judge could validly appeal to the language of the entrenchment clause. The entrenchment clause forbidding certain amendments, including its own amendment, can prevent only inference-like legal change, not other types. The verdict of the principle of non-contradiction, without more, is even less relevant than the judgment of a law professor in a journal article. Legislators, judges, and voters may take it into account, but need not, and probably will not. This is especially likely if the judgment of some formal logic is not articulated by a participant but left to frown on history from ideality.

Linder believes that entrenchment clauses that limit the AC, not by entrenching the AC itself but by purporting to make other constitutional rules immutable, may both be repealed and, if not repealed, ignored and left unenforced by courts.[Note 21] He focuses on the strong, traditional policy in the United States against allowing one generation to bind its successors irrevocably, and believes that a court could justifiably cite this policy (or cases that cited it) in upholding amendments that violated or repealed limitations that purported to make themselves immutable. This policy is sufficiently close to "traditional judicial considerations" that it would not undermine judicial legitimacy to the same extent as a direct appeal to acceptance or "purely political considerations."[Note 22] This part of Linder's conclusion is attractive, but is weakened by the fact that he assumes without question that entrenchment implies self-entrenchment, that self-entrenchment clauses must be immutable absent judicial nullification or nonenforcement, and that all self-limiting self-amendments must be prohibited as immutable. Moreover, he is moved to find an implied limitation against immutable amendments on grounds of democratic policy and values, but must make his limitation immutable. He does not consider the question whether democracies are best served by the people's self-embracing or continuing omnipotence. He is right that immutable rules are fundamentally undemocratic, but that proposition must be qualified by how we answer the question whether (one generation of) the people's will should prevail even when it wants an immutable rule.

What if the entrenching language is more explicit and prohibits not only amendment of the entrenchment clause itself and its underlying rule, but specifically prohibits the attempted repeal of the entrenchment clause? Again, under the inference model this would more than suffice to prevent a valid attempt at disentrenchment, but under the acceptance model it matters not. A judge could certainly enjoin the attempted amendment or its vote of ratification, and in that case we would not face the problem of an amended prohibition of an attempted amendment. But if things got as far as a completed amendment, then a court could nullify the amendment or let it stand, under the acceptance model, with equal validity. She could appeal to democratic values as a justification for disentrenching the immutable limitation on the amending power. The same values would justify her in heeding the manifest will of the people who ratified the amendment more recently than they ratified the immutable limitation. Or she could appeal to the same democratic values for the opposite result, and choose to enforce the people's self-paternalism by deferring to their earlier judgment that subsequent generations might adopt bad amendments in a wave of hysteria.

An AC that prohibited even the attempt of amendment may be likened to a will with an in terrorem clause. An in terrorem or forfeiture clause in a will says that any beneficiary who contests the will's validity (say, to cut out another or to invalidate the will and take a bigger cut under the intestacy laws) will get nothing under the will. When enforced, in terrorem clauses only deter those who think they might lose the contest, for if the will is invalidated, the in terrorem clause is invalidated too. When one aims at a king, one must not miss.

In terrorem clauses are incomplete, mediate self-entrenchment devices. They do not totally prohibit contests that might change the will so much as raise the price or risk of trying. They are mediate by applying to the whole will, not merely to themselves. While they purport to penalize all contests, they are almost never read to penalize successful ones. Occasionally they do not even penalize unsuccessful ones; courts in some circumstances allow beneficiaries to challenge wills with in terrorem clauses, lose, and still take their designated share of the testator's estate. The rationale is that the testator's wish in her will to avoid the expense or trauma of a contest, or to avoid posthumous loss of reputation, should not prevent good faith contests designed to reveal fraud, coercion, or undue influence. A person who schemed to make the will in the first place, and to include the in terrorem clause, would otherwise be free to make her fraud self-insulating. Hence, beneficiaries may circumvent the clause if they proceed with good faith and probable cause to believe the will is invalid.[Note 23]

This rule is a triumph of sound policy over abstract logic. The in terrorem clause would punish all failing attempts to challenge the will of which it is a part, and thereby deter many contests that should succeed; yet not all such contests are punished. The rule violates the testator's apparent intent in order to preserve the possibility of ascertaining her actual intent. It is designed to avoid a vicious circle, or fraudulent self-concealment. Mechanical compliance with rules normally binding is suspended in the name of a sound policy to avoid an unjust result, which is how equity usually works and how the formalist inference model could never work.

Something very basic about legal systems, as distinct from logical systems, is manifest here. Though both are superficially systems of rules, every practicing lawyer can think of dozens of cases in which policy was —wisely or unwisely— put ahead of literalistic compliance with clear rules. Even if they cannot envision the strange circumstances that would make formalist compliance with a given rule unjust or perverse, working lawyers acknowledge from long experience with strangeness that every rule has its nether world of application where exceptions will be recognized, or where finesse, art, judgement, and unabashed consideration of policy are needed. In logic it is difficult to make all rules explicit and to state all the principles that are taken for granted, but it is possible. In law it is impossible (see Section 5.C), and even when principles are articulated, they may require setting a few normally binding rules aside for the given case and looking at the outcome as a value-laden phenomenon, not merely as a product of rules. In law, policies and principles regularly suspend rules, and not because law is hopelessly unsystematic, undecided, or illogical, and not because there is always a rule commanding the suspension, but because it is law, not formal logic.

If a state put an in terrorem clause into its AC, self-entrenching the AC and penalizing any legislator who would propose to amend it, regardless whether the proposal passed, then courts would almost certainly put policy ahead of mechanical compliance (at least in the "right case"), and certainly could permissibly do so.[Note 24] The rule might evolve in much the same way as it did for wills, until an implied exception was recognized for good faith proposals of amendment that arose from a reasonable belief in the deficiency of the present AC, perhaps as evidenced by the widespread public support that resulted in ratification. In the right cases, the exception could also rest on a theory that the framers' intent could be better served by allowing an adjustment in the letter of their law in order to reach the spirit. If the in terrorem clause applied to the whole constitution, and not just to the AC, then the docile compliance of courts is even less likely. As Lester Orfield said in another context, "[t]he courts should be slow to constitutiontruction which would permit a new constitution [and a fortiori a mere amendment] only by revolution."[Note 25]

Because an in terrorem clause is just a self-entrenchment clause that forbids both attempted and consummated amendment, it may be repealed on the same terms as other self-entrenchment clauses. A judge would have even firmer grounds to enjoin the first stirrings of a repeal amendment, but need not do so and may be authorized de facto in restraining herself by the acceptance of the people and officials.

In 1939, when Henry Rottschaefer's Handbook of American Constitutional Law appeared, New Mexico had not amended its constitution in violation of an entrenchment clause, nor had it yet discussed the repeal of the entrenchment clause. Hence, Rottschaefer was not at all sure that either could be done, but in this passage he shows the sort of policy considerations that would come into play.[Note 26]

There has been found no case in which the power to amend has been employed to directly or indirectly modify a constitutional provision expressly excepted from that power. The issues that such an attempt could raise could not be settled by any reasoning derived from logical processes from prevailing conceptions of sovereignty, and those based on considerations of convenience and expediency point to the solution that such attempts to limit the power of amendment should be held futile. The necessities of orderly government do not require that one generation should be permitted to permanently fetter all future generations.

Rottschaefer was also unaware of the Eason case (1851) discussed above (Section 9.A), which fits the description.[Note 27]

Note that if we simply "held futile" all attempts to limit the power of amendment, we would not actually solve the problem. We would make the AC omnipotent, but in which sense? Would it be powerful enough to limit itself irrevocably? If so, we are back to a limited amending power. But if not, then it is not powerful enough to limit itself, and lives from birth under an irrevocable limitation. The paradox of self-amendment is a paradox because it identifies a contradiction whose obvious alternatives are also contradictions.

It is important that most entrenchment and self-entrenchment clauses limit the content of amendments. Either they protect actual rules, and thereby forbid amendments with a content that would amend or repeal them, or they describe the prohibited content, such as any amendment that would deprive a state of its equal suffrage in the Senate without its consent (although the latter could also be said to protect the rule in Article I, 3.1). Rules within an AC itself are mostly rules of procedure. If one rule says that all amendments must receive affirmative votes from three-fourths of those voting in a popular referendum, then in a sense it is incompletely self-entrenched, for it can only be amended by using the procedure it specifies. However, because that procedure is not more difficult than the procedure for ordinary amendments, it probably should not be called "entrenchment" at all, even if the proposed amendment would lower the supermajority needed for ratification.

But note that if such rules could be called self-entrenched in a broader sense, then all law is self-entrenched in that sense. All law is a hurdle to its own amendment or repeal insofar as it must be heeded until changed. The people do not rule themselves directly, but only by means of rules that are somewhat difficult to change. Once legislative will reaches a certain level of intensity or extent, it can make a law, and if that will later subsides, the law stays behind as its relic. Strict positivists and formalists, like Alf Ross, would insist that statutes and constitutional rules persist in validity forever until amended or repealed by a properly channeled act of legislative will. The acceptance theory can more comfortably accommodate a rule of desuetude (see Section 19), and allow laws to fade away with their waning acceptance and usage, slowly disentrenched by the elanguescence of memory, relevancy, and desire.

D. Entrenchment and time

Entrenchment has an ineliminable temporal element. Entrenchment protects rules against future amendment. Therefore the paradox of omnipotence, as it arises in theology, has been said to fail to "get a grip" if the deity is conceived as eternal, or literally outside of time.[Note 28] Even if this works for deities, it cannot work for ACs or sovereigns.

The temporal element of entrenchment suggests another way to conceive entrenchment itself. "Protecting" and "insulating" a rule from amendment are metaphorical. The reality underlying the metaphor is that entrenchment makes the protected rule logically prior to certain temporally posterior rules of change. When an amendment is adopted that conflicts with an entrenched rule, other than an amendment designed to repeal the entrenching language, then the conflict is resolved in favor of the entrenchment clause, which takes the field to itself or amends the newer rule pro tanto. The ordinary rule of priority favors the recent rule in cases of conflicts, unless the older rule is of a legally or logically prior type. In these cases the rule favoring recency is reversed by the explicit terms of the entrenching language. This makes entrenchment clauses into rules of recognition that determine the priority of rules in conflict. They may be thought to attain that status by virtue of their own self-referential assertions or by virtue of a more traditional rule of priority, the lex specialis principle, which favors specific over general language.

Entrenched clauses are reflexive or tacitly self-referential in their attempt to insure their own priority in future conflicts between themselves and attempts to amend or repeal them. They strive to become what is ordinarily forbidden —judges in their own case. They are not alone in this adjudicative hubris. The supremacy clause of the federal constitution, for exconstitutionnly declares the supremacy of the constitution of which it is a part, thus declaring its own supremacconstitutiones further and specifically adds that all state judges are bound by the federal constitution and this clause, "anything in the Constitution or laws of any State to the contrary notwithstanding" (Article VI, 2). We are not suspicious of the supremacy clause, perhaps because it restates the obvious, or more subtly, the obviously necessary, in a federal system; in short, because it is good policy. But we are suspicious of entrenchment and self-entrenchment clauses that share the logic of self-elevation. We might suspect entrenchment clauses because they presumptuously attempt to settle in their own favor all disputes in which they figure as parties. We recall that John Locke located the tyranny of absolute monarchs precisely in their capacity to be judges in their own cases,[Note 29] and that Madison advocated a republic with a system of checks and balances over a pure democracy to avoid the same evil.[Note 30] But the fact is that the supremacy clause is equally presumptuous in settling its conflicts with state law, in advance, by its own terms. I submit that the reason we suspect one and not the other is entirely a difference of the wisdom of the policy that is attempting to prevail by its own terms.

Bound up with this reflexive view of entrenchment is a temporal view, for entrenchment and self-entrenchment clauses, at their humblest and at their most presumptuous, only feign to prevail in future disputes in which they might be implicated. This allows us to see that ordinary implied repeal and implied amendment (see Section 16) are the temporally symmetrical correlates of entrenchment. In conflicts they tell us to prefer the recent rule and to amend all earlier rules in conflict with it pro tanto. Entrenchment and implied repeal are symmetrical temporally, but are not equally weighted legally. Implied repeals or preference for the recent rule is the norm, and entrenchment or preference for the earlier rule is the exception (when the rules are on the same level). Indeed, preference for the new rule is presumed, and is only rebutted (when rebutted) by the affirmative specific act of adopting the entrenching language. In the language of presumptions we may say that, historically, presumptions against entrenchment are rarely rebutted by implication and require explicit language; and on the other side, when there is explicit entrenching language, the presumption in favor of its plain meaning has always been rebuttable, sometimes by explicit language in a repeal amendment, sometimes by inattention or winks, but most often by appeal to public policy.

The temporal perspective on entrenchment allows us to define a new type. Ordinary entrenching language protects a rule from amendment for an indefinite period of time. Most entrenchment clauses do not have "sunset clauses" in them that provide for automatic expiration after a certain date; but they could (see Section 14). Still, an entrenchment clause silent on the duration of its own effectiveness does not impliedly make itself valid forever. Nor on the other hand does it impliedly defer to a rule of desuetude. If the entrenched rule is a constitutional rule, then the entrenching language expires with the constitution or underlying rule. That may wisely be presumed for ordinary entrenchment clauses, just as in terrorem clauses lose their validity if the will in which they appear is successfully overturned. (In part, that is what "successfully overturning" the will means.) But suppose an entrenchment clause said, "this section shall not ever be amended or repealed, and shall not ever lose its validity, not even by or after the replacement of this constitution, by revolution, by complete change of the form of government, or by the extinction of the human race".

Recall that the Arkansas, Pennsylvania, and Tennessee constitutions said that their Bills of Rights shall "forever remain inviolate". That is close, but Rhode Island came closer in its constitution of 1842. After announcing the right to alter or abolish government, the Bill of Rights added that "the Constitution which at any time exists, till changed by an explicit and authentic act of the whole people, is sacredly obligatory upon all". This sentence has three remarkable features. Fconstitutionthe only language in any American constitution that says of the whole instrument, in effect, "valid and supreme until amended", which suggests self-embracing, not continuing omnipotence. Second, it provides something of a test for the legality of rebellion —"an explicit and authentic act of the whole people" (see Section 18). Third, it imposes a duty, or sacred obligation, upon "all" (which for the present may be restricted to Rhode Islanders) to obey "the Constitution which at any time exists". It not only says that we should obey the rules (see the first rule of Nomic in Appendix 3), but that we should obey whatever rules will be in effect after any number of changes. It anticipates its own eventual obliteration, but tries to imposes a duty that will outlast its own validity. The duty to obey is entrenched beyond the life of the language of the duty, and even beyond the life of the language of the entrenchment clause. If future Rhode Islanders, living under a new constitution, were moved to another insurrection, then the 1842 rule would (if it could) rise up and remconstitutiontheir continuing duty to obey the constitution.

Is this super-entrenchment effective? The Rhode Island clause was probably not meant to protect the duty to obey in anything so fragile as constitutional language. Its framers probably believed that the duty to obey the law was eternally inherent in law. They tried to say so in language that inadvertently but unavoidably left openings for someone like me to take it literally. But let us take it literally here.

Under the inference model, it might not be effective. The logic of parts and wholes may require the expiration of all the parts with the expiration of the whole. But if super-entrenchment is effective, followers of the inference model must sigh in dismay to think of all the ancient obligations imposed on them of which they are unaware, and which cannot be repealed even by revolution. Under the acceptance model, the answer could be that the language is ineffective, if forgotten or regarded as superseded, but this would depend on the empirical facts of what is accepted. Some rules can be accepted as authoritative after their language has expired (see Section 14).

(In Section 10.C I argue that the "binding until changed" language of the clause negates the super-entrenching effect.)

J.M. Finnis has argued that the validity of such super-entrenchment follows from the jurisprudence of Hans Kelsen.[Note 31] Kelsen argued that a constitutionconstitutionconstitution/a> -->rule purporting to repeal a later constitution, whether or not the later constitution authorizes such a repeal.

Finnis rightly notes that this problem is not resolved by stating, or assuring, that the two constitutions were never in effect at the same time, for the same reason that this objection does not dissolve the paradox of self-amendment: the earlier authorizes the later regardless whether they overlapped in time or not (see Section 10).

This type of super-entrenchment is disarmed by any theory, such as the acceptance theory, which denies that a constitution can be validated only by its predecessor. For the acceptance theory constitutions might be validated instead by present and continuing acceptance or consent. But Kelsen and Finnis do not deny this. Finnis eventually denies Kelsen's position in favor of a variant of Hart's acceptancconstitutionut even Kelsen believes that if a constitution is not authorizeconstitutiondecessor, because there wconstitutionecause it was overthrown, then the present constitution is valid by something very close to acceptance, and the AC of the old constitution cannot be super-entrenched or presently valid. On this view an AC need not amend itself in any special way in order to become super-entrenched against its successors: it is automatically cantilevered over its successors until one of them is superseded by revolution.

In the cognate situation of a sovereign releasing a subject (or former colony), therefore, Finnis notes that[Note 35]

[g]reat care has, on occasion, been taken to perform some unauthorized act in the course of transfer of authority, so that it may be claimed that, because there has been a revolution, therefore the validity of the new constitution[al] legal order cannot be traced back to the British Imperial Constitution.

In this sense Americans should be relieved that the present American constitution was invalidly established under the AC of the Articles of Confederation (see Appendix 1.D). Thomas Reed Powell once bemoaned the fact, which is a fact only to a formalist, that the American legal system was illegal on account of the defective self-amendment of the Articles of Confederation.[Note 36] But if initial illegality is cured by acceptance or acquiescence, or by the self-justifying terms of the ratification clause of the new constitution -->constitution> --> (Article VII in the present constitution) (see Section 7.B), then initial illegality is preferable to perpetual subjection to the prior AC, pending a revolution.

Another form of super-entrenchment may be said to occur when a clause attempts to forbid its amendment by appeal to a moral duty which is presumed to be immutable, perhaps thereby acknowledging that a legal duty will not survive a change of regime. One example is the old Virginia statute quoted in Section 8.C which defined a right of free exercise of religion and asserted that repeal, while technically lawful, would infringe the right defined in the statute, which ultimately derived from natural law. Another example is K.W.B. Middleton's contribution to the long-running English debate on the question whether parliament can commit suicide. Dicey said it could, and Hart thinks that "might be conceded."[Note 37] On the other side are those, perhaps a majority, who believe the British Parliament has continuing omnipotence. Middleton takes a different approach, and decides that the "purpose" of parliament is to convert the "inchoate will of the nation into a set of concrete rules," and therefore that it could not commit suicide,[Note 38]

legally....For the effect would be to deprive the nation, which can only speak through Parliament, of all legal means of expressing itself, and that would be clearly contrary to the purpose for which it exists.

Middleton suggests that insofar as power is restrained by the purposes of its existence, parliament cannot commit suicide; but he leaves the door wide open for the interpretation that parliamentary suicide would be improper, not illegal, a betrayal of trust more than a violation of law. Using similar reasoning, William Marbury has argued that the "power to 'amend' the Constitution was not intended to include the power to destroy it."[Note 39]

In the theological version of the paradox of omnipotence the parallel argument is that the deity has self-embracing omnipotence, and logically could limit or destroy itself, but will not because some auxiliary blessing, such as infinite goodness, argues in favor of the greater utility of remaining unlimited in power.[Note 40] Like the moral injunction instead of the legal one in the old Virginia statute, this argument is something of a cheat. The deity's goodness is usually described in terms that make its verdicts necessary, not contingent. Hence, the deity whose goodness prevents self-limitation should actually be said to possess continuing, not self-embracing, omnipotence, unless the omnipotence can repeal the goodness: but presumably the goodness itself will prevent that.

One may wonder why the self-destruction of parliament or Jehovah, or the self-repeal (without replacement) of a constitution, might not be even more permissible than the repeal or destruction of lesser entities. Wittgenstein noted, in a different context, that one who doubts that Napoleon existed is not of sound mind; but one who doubts that the world exists is on firmer ground.[Note 41] If one accepts the world but doubts Napoleon, then one is accepting a system of evidence that strongly supports the existence of Napoleon. But if one rejects the whole system of evidence, nothing remains to confute one's doubt. Similarly, if an AC repeals itself, rules are left behind that might suffice to declare the act invalid, such as (for Alf Ross) legal rules incorporating the principle of non-contradiction. But if the constitution is repealed, or better, if the British parliament commits suicide, then at the very least the validity of the act will be injusticiable, if not actually beyond the strictures of the rules and criteria it took with it.[Note 42] Moral and logical rules will remain after every such act, except possibly after nuclear annihilation, and while both may find self-destruction objectionable, they cannot find it unlawful.

E. Self-repeal

Suppose someone were to object that a particular self-entrenched AC could not disentrench itself because it had continuing, not self-embracing, omnipotence. This would be a plausible objection for any AC, for self-disentrenchment presupposes some self-embracing power, although not necessarily self-embracing omnipotence. However, the objection is ill-founded. If the AC is truly characterized by continuing omnipotence, then it must have the capacity to authorize all amendments that are not inconsistent with its continuity or omnipotence. That means it must have the capacity to undo by amendment all limitations on its continuity or on its remaining omnipotence, and that implies the power of self-disentrenchment. All limitations on its power (except the limitation that makes it continuing) are revocable by it.

An AC of continuing omnipotence could not entrench itself in the first place, except revocably. If the self-entrenchment clause were original, not self-imposed, then the AC could still undo it through a self-application that merely revoked the revocable limitation. If the original limitation is irrevocable, then the AC does not have continuing omnipotence. Therefore, the inability of continuing omnipotence to amend itself in significant ways is not a barrier to self-disentrenchment. If the limitations to be disentrenched were self-imposed, then they are necessarily revocable, because continuing omnipotence cannot limit itself irrevocably. If they are original, then either they are revocable, or the AC does not actually have continuing omnipotence.

A subtler problem lurks here. How do we know whether an AC has self-embracing or continuing omnipotence? Hart says the question is empirical. He does not say that the decisive empirical question is whether self-limiting self-amendment has occurred. He says it is "an empirical question concerning the form of rule which is accepted as the ultimate criterion in identifying the law."[Note 43] This suggests that the empirical inquiry seeks to ascertain whether the rule of recognition sees the AC as self-embracing or continuing. This just postpones the question of what evidence is to count either way.

One empirical test that might be the language of the AC. However, no AC that I know of explicitly says whether it is self-applicable. The closest that I have seen is New Mexico's, which implies self-applicability by denying that one section of the AC can be amended except in convention. Many ACs say they apply to "this constitution", which implies self-applicability, but not enough to satisfy Alf Ross, for example. But self-applicability is not really decisive on this question, even if it is suggestive. Self-embracing omnipotence requires self-applicability. But so does continuing omnipotence, in order to repeal the revocable limitations on its continuing power or to add others.

Note that if an AC were reworded so that it explicitly permitted all kinds of self-amendment, including self-repeal, self-entrenchment, and self-disentrenchment, then the paradox of self-amendment would not be any closer to solution or dissolution. The reason is that a self-amendment under the old AC inconsistent with its original form would still make an invalid inference. And of course an amended AC could still become inconsistent with its original form by, say, prohibiting what was once permitted or permitting what was once prohibited (see Section 12.C).

At least a history of self-limiting self-amendment would be a strong empirical indication of self-embracing power. It would not be decisive unless we knew that the limitations were immutable, which is unknowable so long as future generations may find a legal theory, such as acceptance, sufficient to amend them. A history of self-disentrenchment or transmutation would be a strong sign of continuing omnipotence. It would not be decisive unless we knew that all future limitations will be revocable, which is also unknowable. Because self-disentrenchment proves that the repealed limitations were repealable, the former (and therefore, present) existence of the limitation is consistent with both continuing and self-embracing omnipotence.

What if an AC has both histories? What if it has validly limited itself through self-amendment and validly disentrenched its limitations? The present New Mexican AC has amended itself to a form inconsistent with its original form, and by reducing the years in which amendments may be considered by half, has limited itself. One section subject to a valid entrenchment clause has impliedly amended itself before or without expressly disentrenching itself. And the entrenched amending rule has been said to be capable of repealing its entrenchment clause without using the latter's special procedure, which is to disentrench itself. How do we interpret these signs?

Self-embracing and continuing omnipotence are mutually exclusive because the former could limit itself in violation of the latter, and the latter could resist limitation in violation of the former (see Section 10). But self-embracing omnipotence is consistent with a limited power of disentrenchment, and continuing omnipotence is consistent with a limited power of self-amendment. The New Mexican AC may be one or the other of these logically coherent species, or it may actually possess both self-embracing and continuing omnipotence simultaneously and incoherently. Under the acceptance model, that is at least possible. That is, the people and officials could accept an immutable self-limitation, a sign of self-embracing omnipotence, and then accept its repeal, perhaps after transmutation, a sign of the revocability of the limitation and the continuing omnipotence of the AC.

An AC may have continuing omnipotence as evidenced by a history of self-disentrenchment, and also have self-embracing omnipotence as evidenced by a history of revocable, temporary self-entrenchment and self-limitation. Stated this way, there is not even a logical problem. But self-embracing omnipotence could result in total self-repeal, which is clearly inconsistent with continuing omnipotence —if not self-contradictory in itself. And of course it could result in irrevocable, permanent self-entrenchment and self-limitation, which are also inconsistent with continuing omnipotence. These cannot coexist logically, but may they coexist legally?

Self-repeal is cleaner and simpler than irrevocable self-entrenchment or self-limitation. Whether it has ever occurred in the United States depends upon one's definition. The mere replacement of an old constitution by a new one, unconstitutionority of the old AC, when the new constitution has a completely different AC, may or may not be taken to comprise self-repeal. In any case, it is commonplace in American history and clearly lawful (see Appendix 2). If an AC authorizes the making of a new constitution that lacks an AC, then that is a more troublesome kind of self-repeal. It has apparently occurred once in America. The AC of the 1776 Pennsylvania constitution was used to nullify the instrument of which it was a partconstitutionntly used to make the next (17constitution. That constitution contained no constitutionbut announced in its final paragraph the conditional nullification of the whole constitution "if reconciliation between Great Britain and these Colonies should take placconstitutionises the question whether a reconciliation would have resulted in total self-repeal, or only in the irreflexive repeal of the constitution minus the repealing authority in the last sentence. Would the last sentence remain valid forever to tell us that the repeal was effective? (For problems of this kind, see Section 14.) For one like Alf Ross who believes that self-amendment and self-repeal are illegal, the reconciliation between Britain and the United States would have caused the anomaly in New Jersey of peace triggering revolution.


1. I defer to Ross's argument, outlined in Section 5.D, that a sentence that refers only to itself, refers to a hall of mirrors rather than to a determinate subject. Even if Ross is wrong on this point, I need not show it for my present purposes. I also defer to Hart's argument that an "asymptotic stutter" can be avoided (at least) if the self-referential sentence refers to something other than itself as well. Hence, "this sentence shall not be amended" is (or may well be) subject to Ross's criticism. But no actual rules of law are so whimsical. A sentence that announced an irreflexive rule and added, "this section, including this sentence, shall not be amended", or more minimally, "neither rule R nor this sentence shall be amended", is perfectly meaningful. [Resume]

2. These terms are ambiguous on the question whether a self-referential sentence refers to its words or to its meaning. While this ambiguity should be clarified through appropriate terminology, it is unnecessary to clarify it for my purposes here. Hence I will not confuse things by pursuing clarity through another distinction. [Resume]

3. In these terms, Ross objected to statutes that were self-referential per se, but eventually narrowed his criticism to those that were immediately self-referential. Hart countered by showing that mediate and eccentric self-reference can be meaningful by avoiding the "asymptotic stutter" or incompleteable series of references comprising the subject of the sentence. [Resume]

4. This choice may be defended by noting that a reference that referred directly and solely to itself is immediately self-referential, even though it may be said to refer to the class of which it is the only member. [Resume]

5. In support I can cite the entrenched sections in the Australian and South African constitutions, discussed in Section 8, both of which were written by the British parliament. The British North America Act, which served as Canada's constitution until 1982, and which was written mostly in Britain, had no amending power that constitutionemselves could use. The new 1982 constitution, however, was estaconstitutionarily to transfer the amending power from Britain to Canada, "repatriating" the consticonstitutionompleting the sovereignty of Canada. The entrenchment of the equal suffrage of the states in the Senate in the United States constitution was included at the insistence of the states who feared encroachments on their sovereignty. The entrenchment of 1 of the New Mexican AC by 5 was originally imposed by the New Mexicans on themselves, but Congress left it unchanged and even demanded that it remain as a precondition of admission to the Union. The clause within New Mexico's AC 1 that entrenched sections of other articles on voting and education were demanded by Congress. The clearest exception to my generalization is the entrenchment clause in the present (1901) Alabama AC, that explicitly and completely entrenches the rule that representation in the state legislature shall be based on population (see entry on Alabama in Appendix 2). [Resume]

6. In real legal systems a tacit rule to obey the other rules is either unnecessary or may always be inferred. Hart argues that they are unnecessary; Hart, The Concept of Law, Oxford University Press, 1961, p. 230. The same is probably true of games, but in the spirit of making everything explicit in a game in which the rules are constantly changing, and to offer a tantalizing rule to change, I have made the rule to obey the rules explicit in the game Nomic, in Appendix 3. [Resume]

7. Each state uses the same language, except that Arkansas inserts "the" before "government." [Resume]

8. This case was first brought into this context by Douglas Linder, "What In the Constitution Cannot Be Amended?" Arizona Law Review, 23 (1981) 717-33, at 724.n.46. He deserves great credit for discovering it, for Henry Rottschaefer, writing in 1939, believed that no case of an attempted amendment of an entrenchment clause had occurred in America. See text accompanying note 27, below. [Resume]

9. Helene Simpson, "Procedural Problems in Amending New Mexico's Constitution," Natural Resources Journal, 4 (1964-65) 151-59, at 156ff. [Resume]

10. I have not been able to discover whether the vote on the AC that Congress demanded was conducted under the terms of the old AC that would be replaced. Certainly it was not proposed under the terms of the old AC. [Resume]

11. Simpson, op. cit. at 157. [Resume]

12. Simpson, op. cit. at 158-59. Arizona had originally subjected all public officials to recall by popular vote. Congress demanded an exception for judges, an exception that the voters adopted before submission and then repealed after admission. The Arizona amendment was demanded by a Joint Resolution of Congress, like the New Mexican amendment, not by the federal Enabling Act. The Arizona Supreme Court has ruled, by contrast, that the provisions in the Arizona constitution required by the Enabling Act "cannot be altered, changed, amended, or disregarded without an act of Congress." Murphy v. State, 65 Ariz. 338, 181 P.2d 336, 340 (1947). Wyoming made a similar ruling in Merrill v. Bishop, 74 Wyo. 298, 287 P.2d 620 (1955). Three New Mexican attempts to circumvent the terms of its Enabling Act were struck down as unconstitutional. Regents of the University of New Mexico v. Graham, 33 N.M. 214, 264 P. 953 (1928), State v. Llewellyn, 23 N.M. 43, 167 P. 414 (1917), cert. den. 245 U.S. 666 (1918), and State v. Marron, 18 N.M. 426, 137 P. 845 (1913). Cf. Ervien v. United States, 251 U.S. 41 (1919). See Simpson, op. cit. at 157. In United States v. Sandoval, 231 U.S. 28 (1913) the Supreme Court rejected a challenge by New Mexico to a provision of its Enabling Act that prohibited liquor in Pueblo Indian land. Congress can make a state assent to such a provision as a precondition to admission so long as Congress has jurisdiction to regulate the subject matter of the provision. New Mexico's Enabling Act is 36 Stat. 557 (1910), which is included in the New Mexico constitution at Article 21, 10. Article 21, by the way, is incompletely entrenched by AC 4. The Enabling Act did not require the amendment of the original AC; that was demanded by a Joint Resolution of Congress, 37 Stat. 39 (1911). [Resume]

13. One wonders whether New Mexico was alone in this respect because of its large Hispanic —and Catholic— population. [Resume]

14. See Arthur O'Neal Beach, "Constitutional Revision — Constitutional Amendment Process [in New Mexico]," Natural Resources Journal, 9 (1969) 422-29 at 425f. [Resume]

15. Linder, op. cit. at 729. [Resume]

16. Linder gets out of this problem, partially, by denying entrenchment clauses any effect at all, apparently because democratic values take precedence over the intent of the makers: "no principled decision could depend upon whether the amendment [to be repealed] did or did not include a clause expressly declaring the amendment not to be subject to repeal." Linder, op. cit. at 730. Therefore, Linder's single amendment need not and would not repeal both the entrenchment clause and the protected rule. Because the former is a nullity, only the latter need be repealed, and no obstacle stands in the way. Linder gives no reason to think that democratic values should sometimes supersede the intent of the law-makers and sometimes be superseded. Nor does he reconcile his appeal to democratic values with his per se rule invalidating entrenchment clauses, except to note that it is undemocratic to allow one generation to bind its successors irrevocably. I agree with that. But it is arguably as undemocratic to ignore the manifest will of the people in ratifying an entrenchment clause. In other terms, he does not explain why democracy requires continuing, not self-embracing, omnipotence, nor why democracy can abide the immutable limitation that makes continuing omnipotence continuing. I believe he is right in this normative preference for the continuing omnipotence of the people, but if he had articulated his reasons, then he probably could not, so easily, have justified the total nullification of entrenchment clauses. See Section 21. [Resume]

17. In chronological order the chief articles are Clarence Norton Goodwin, "How Should the Illinois Constitution be Amended?" Illinois Bar Review, 9 (1915) 601-11; A.L. Powell, "A Plan for Facilitating Constitutional Amendment in Illinois," Illinois Law Review, 30 (1935) 59-67; Special Committee on Constitutional Convention, "The Gateway Amendment," Illinois Bar Journal, 38 (1950) 308; L.A. Kohn, H.I. Green, and T.J. Welds, "Debating the Gateway Amendment," Illinois Bar Journal, 39 (1950) 127-33; Robert L. Farwell, "Gateway to What?" DePaul Law Review, 10 (1961) 274-85; and Samuel W. Witwer, "Workable Amending Clause or Strait-Jacket?" Chicago Bar Record, 50 (1968) 10. [Resume]

18. The second such limitation is hard to find. Justice Wayne had just summarized the original three limitations on the federal amending power. Two of them expired in 1808: the irreflexive entrenchment of Article I, 9.1 and 9.4. Grammatically he must have meant that the second of these, and the equal suffrage limitation, were immutable. But if so, he is clearly wrong on the entrenchment of I.9.4, since it expired (see Section 14). I believe he was wrong about the equal suffrage limitation as well, for even if it is impliedly self-entrenched, it is incompletely self-entrenched, and therefore mutable by a special procedure. [Resume]

19. See Lester Bernhardt Orfield, The Amending of the Federal Constitution, University of Michigan Press, 1942, at 84-85. See my Section 8, note 47, above. [Resume]

20. In the former case the most complete entrenchment clause in any American constitution was declared repealable, although perhaps because it was not self-entrenched; see entry under Alabama in Appendix 2. The latter case was discussed in Section 8.C; an amendment that delegated the decision on its effective date to its beneficiaries was upheld on the assumption that delegation was prohibited and that the prohibition was completely entrenched. [Resume]

21. Linder, op. cit. passim. [Resume]

22. Linder, op. cit. at 723-24. [Resume]

23. See the learned opinion of Hartz' Estate v. Cade, 247 Minn. 362, 77 N.W.2d 169 (1956). See also 26 ALR 755, 52 ALR 83, 125 ALR 1135, and 157 ALR 584; Harold Kertz, "Contesting a Will in the Face of a Forfeiture Clause," Georgia Law Journal, 45 (1956) 200-213; Jack Leavitt, "Scope and Effectiveness of No-Contest Clauses in Last Wills and Testaments," Hastings Law Journal, 15 (1963) 45-91; and W. Harvey Jack, "No-Contest or In Terrorem Clauses in Wills— Construction and Enforcement," Southwestern Law Journal, 19 (1965) 722-39. Leavitt says the following (at 91): "The value of laws cannot be determined by their internal consistency or chaos, for legal symmetry is admirable only when its subject matter is equally symmetrical....What if...deserving parties would have been harmed by [a truly logical result]?" [Resume]

24. But see Demosthenes on the Locrians,

In that country the people are so strongly of the opinion that it is right to observe old-established laws, to preserve the institutions of their forefathers, and never to legislate for the gratification of whims, or for a compromise with transgression, that if a man wishes to propose a new law, he legislates with a halter round his neck. If the law is accepted as good and beneficial, the proposer departs with his life, but, if not, the halter is drawn tight, and he is a dead man.

Demosthenes, "Against Timocrates," @139, p. 463 of vol. III of the Loeb Classical Library edition of his works, trans. J.H. Vince, Harvard University Press, 1951. See also Jon Elster, Ulysses and the Sirens: Studies in Rationality and Irrationality, Cambridge University Press, rev. ed., 1984, pp. 88f. [Resume]

25. Orfield, op. cit. at 44. [Resume]

26. Henry Rottschaefer, Handbook of American Constitutional Law, West Pub. Co., 1939, at 9-10. [Resume]

27. See note 8, above. [Resume]

28. Richard Swineburne, "Omnipotence," American Philosophical Quarterly, 10 (1973) 231-37, at 237. This view has been criticized by Edward J. Khamara, "In Defense of Omnipotence," Philosophical Quarterly, 28 (1978) 215-28 at 215-16. [Resume]

29. John Locke, The Second Treatise of Government, Section 13, in e.g., John Locke, Two Treatises of Government, ed. Peter Laslett, Cambridge University Press, 1963. [Resume]

30. James Madison, The Federalist #10. For more discussion of the reflexivity of becoming a judge in one's own case, see Section 20. [Resume]

31. J.M. Finnis, "Revolutions and the Continuity of Law," in A.W.B. Simpson (ed.), Oxford Essays in Jurisprudence, Second Series, Oxford University Press, 1973, pp. 44-76, at 52f. [Resume]

32. Hans Kelsen, The Pure Theory of Law, University of California Press, 1967, at 200. [Resume]

33. Alf Ross, "On Self-Reference and a Puzzle in Constitutional Law," Mind, 78 (1969) 1-24, at 22-23. [Resume]

34. Finnis, op. cit. at 53. [Resume]

35. Finnis, ibid. at 52 (emphases in original). [Resume]

36. Thomas Reed Powell, "Changing Constitutional Phases," Boston University Law Review, 19 (1939) 509-32, at 511-12. More precisely, Powell argued that the new constitution was not established until it had been ratified unanimously, as required by the Articles of Confederation. Ratificaconstitutionunanimous two years after the new constitution's own terms for its establishment (ratification by nine states) had been satisfied. For more discussion of this problem, see Appendix 1.D. [Resume]

37. H.L.A. Hart, The Concept of Law, Oxford University Press, 1961, at 148. [Resume]

38. K.W.B. Middleton, "New Thoughts on the Union of England and Scotland," Juridical Review, 66 (1954) 37-60 at 48; the quotation to follow appears at 52. [Resume]

39. William Marbury, "The Limitations Upon the Amending Power," Harvard Law Review, 33 (1919) 223-35, at 225. This idea has some support from the more radical statement by Roger Sherman, delegate to the Constitutional Convention, made in the first session of Congress:

All that is granted us by the Fifth Article is, that whenever we shall think it necessary we may propose amendments to the Constitution, not that we may propose to repeal the old, and substitute a new one.

Annals of Congress, vol. 1, at 742. Note that Sherman's principle forbids a type of self-amendment that is common in the states, and that arguably (but if so, defectively) established the present federal constitution under the AC of the Articles of Confederation (see Appendix 1.D). For other doctrines of Roger Sherman that, by contrast, may have led to the self-amendment of the federal AC, see Section 16. [Resume]

40. Swineburne, op. cit. at 236. [Resume]

41. Ludwig Wittgenstein, On Certainty, Harper and Row, 1969, Section 185, p. 26e. [Resume]

42. J.M. Eekelaar, "Principles of Revolutionary Legality," in A.W.B. Simpson (ed.), Oxford Essays in Jurisprudence, Second Series, Oxford University Press, 1973, 22-43. Eekelaar argues that the legality of revolutionary acts is justiciable by the next or new regime, according to principles that survive revolution. But see Wayne B. Wheeler, "The Constitutionality of the Constitution is Not A Justiciable Question," Century Law Review, 90 (1920) 152. The best recent summary of the law of the justiciability of amendments is Judith Elder, "Article V, Justiciability, and the Equal Rights Amendment," Oklahoma Law Review, 31 (1978) 63-109.

Howard Newcomb Morse, "May an Amendment to the U.S. Constitution Be Unconstitutional?" Alabama Lawyer, 10 (1949) 199-200, offers the unusual view that amendments may conflict with the spirit of the constitution and, if so, the conflict should be settled according to the rules of priority used in conveyances of real property. Morse's rationale is that the tests of sovereignty are all property law tests, such as the power to take property by eminent domain, the quality of having property escheat to it, and immunity to claims of adverse possession. [Resume]

43. Hart, op. cit. at 146. [Resume]

44. Jacob Tanger, "Amending Procedure of the Federal Constitution," American Political Science Review, 10 (1916) 689-99. For the general proposition that express amendment clauses in the supreme law were the invention of Americans, see Ralph R. Martig, "Amending the Constitution; Article V; The Keystone of the Arch," Michigan Law Review, 35 (1937) 1253-85, and Carl Joachim Friedrich, Constitutional Government and Democracy: Theory and Practice in Europe and America, Blaisdell Pub. Co., 4th ed., 1968, at 138. [Resume]

This file is one section of the book, The Paradox of Self-Amendment. Return to the Table of Contents.

Ribbon] Peter Suber , Department of Philosophy , Earlham College , Richmond, Indiana, 47374, U.S.A. . Copyright 1990, Peter Suber.