Section 3
The Dilemma
Peter Suber, Paradox of Self-Amendment Table of Contents

A. Self-application v. infinite regress

Any legal philosophy that would explain how laws may lawfully be changed faces a dilemma. The authority to change legal rules is provided by a special group of legal rules which we may call (after Hart) rules of change. Rules of change permit and structure the process of enactment, amendment, and repeal. Can the rules of change permit and structure the process of their own change? More pointedly, can they authorize their own change into forms inconsistent with their original forms?[Note 1] Can they make their self-change irrevocable?

The dilemma is sharp. (1) If the rules of change cannot authorize their own change, then they are either immutable, or they may only be changed by a second set of such rules, which in turn must either be immutable or mutable only by a third set, and so on to an arbitrary halt or an infinite regress. (2) But if the rules of change can permit and structure the process of their own change, then they must permit their own limitation and repeal. But at first sight, at least, paradox attaches to any rule that authorizes the limitation or extinguishment of its own power. For it seems that such rules would supply the premises for a transfer of authority in which the conclusion contradicted at least one premise, or an action much like that of an omnipotent being limiting its own power to act. (The paradox of self-amendment is set out more fully in Section 5.)

So it seems that we must choose among immutable rules, infinite regress, and paradox. To this dilemma, or trilemma, we might add revolution. But we are now considering only the lawful change of law. Extra-legal methods of changing the rules are always available, of course, even though they are not lawful and sometimes inefficient. For example, revolution can clearly change the rule of succession for sovereigns, but can only rarely change certain unwritten constitutional norms. In any case, the present inquiry is devoted to the ways in which a rule-governed system may change all its parts, including the rules that govern change. So for present purposes we can omit revolution as a method of legal change.

Of the options remaining —immutable rules, infinite regress, and paradox— immutable rules are the least objectionable logically, although perhaps the most objectionable legally. But in any event that hypothesis that any legal rules are truly immutable faces both logical and legal problems (see Section 8). Legal history also threatens it, for it seems that all kinds of rules of change have been changed; at least this is true of constitutional amending clauses (see Appendix 2).

The problem is to explain how legal change of the rules of change actually occurs, not how a hypothetical set of rules carefully crafted for the purpose could permit its own alteration.[Note 2] Alf Ross, a Danish jurist and logician, has argued that rules of change, or at least constitutional (supreme) rules of change, cannot authorize their own amendment without paradox, and therefore cannot lawfully authorize their own amendment at all.[Note 3] This essay is in effect an extended reply to Ross. I will argue that Ross's paradox is not easily solved or dissolved, but is easily accommodated in law. A powerful explanation of how law can cope with an unflinching contradiction may be found in the legal philosophy of H.L.A. Hart. Hart has laid out a theory of law that provides useful terms and theoretical tools for attacking this problem but that Hart himself did not use for this purpose.[Note 4]

Every legal system contains rules which permit the change of the other rules. Change can occur through the addition of new rules or the modification and repeal of existing rules. (I will use the term "amendment" to include addition and repeal, as well as piecemeal and wholesale alteration.) Sometimes the rules of change are tacit, and cited as needed in a way that resembles invention, but usually they are explicit.[Note 5] There must be many rules of change to deal with the many types of legal rules that must, over time, be changed —constitutional rules, statutes, decisional rules, treaties, executive agreements, administrative regulations, parliamentary rules of order, judicial rules of procedure and evidence, executive orders, canons of professional responsibility, standards of reasonableness, evidentiary presumptions, principles of the priority among rules, canons of interpretation, and even contractual rules. Each of these can be changed by the power that created it, although that power may be difficult to identify and commonly must be more restricted and channeled when used to amend than when originally used to establish law. Most of these types of legal rule can be changed by other powers as well, such as striking through judicial review, reformation in equity, overruling on appeal, judicial reinterpretation, overruling by subsequent enactment, statutory preemption, and constitutional amendment.

In a hierarchy of rules in which some take precedence over others the immutability of the lower level rules is never a real possibility. Most case law is inferior to statutory law, and the latter provides a rule of change for the former. Even if case law provided no rules for its own change (for the overruling and modification of case law by subsequent cases), then some rules of change at higher echelons could always be found, if only in the amending clause (AC). Indeed, one of the defining characteristics of a "higher level" rule, or a rule which takes precedence over other rules, is that it, or a rule of its type, can alter and limit the rules below it in the hierarchy. Hence, we may assume that in every legal system with a hierarchy of rules, the lower level rules, including the lower level rules of change, are always mutable.

But serious doubt may arise as to the mutability of the supreme rule of change. The supreme rule of change, where there is one,[Note 6] may be found by asking which rule of change may change all other rules of the system, directly or indirectly,[Note 7] and may not be changed by any other. In constitutional systems the supreme rule of change is the AC of the constitution, the rule which permits and structures the process of the change of constitutional rules. No lesser power may amend it, and it allows the amendment of all others.[Note 8] This means, among other things, that if an AC cannot authorize its own amendment, then it is legally immutable and may be changed only by illegal or extra-legal means such as revolution.[Note 9]

The self-amendment of a non-supreme rule of change may or may not escape the apparent paradox of self-amendment (see Section 12.B). But to keep the dilemma as sharp as possible, the problem of self-amendment, unless otherwise indicated, will be limited to the problem of the self-amendment of a supreme and omnipotent rule of change (see Sections 8 and 12.B). For the same reason I will consider only changes that make the changed rule inconsistent with its original form (see Section 12.C). Whether it matters that the inconsistent new amendment or self-limitation is revocable or irrevocable will be addressed in Sections 8 and 11.

B. Primary and secondary rules

The term "rule of change" is borrowed from Hart (93-94). Hart conceives law to be the union of primary and secondary rules. Primary rules impose duties on people to behave in certain ways, whether they want to or not (78-79). They create obligations (89, 91) and pertain to action, behavior, and physical movement (79). Secondary rules, by contrast, confer powers and pertain to the primary rules (79). They confer public power, such as the power to legislate, and private power, such as the power to enter contracts and dispose of property through will (79). They create (and are created by) a class of officials who execute public power (113). Hence they also aid the legal system of which they are a part by permitting it to solve problems, without unlawful or alegal manipulations, that arise from the use of primary rules.

For example, primary rules do not themselves determine who, if anybody, is to enact, record, enforce, interpret, adjudicate, apply, or alter them, or by what procedures these tasks are to be performed. Primary rules may exist in written or unwritten form, but in either case they do not themselves declare which versions are authoritative, what is to be done in case two or more primary rules conflict, or exactly which utterances or traditions are to count as primary rules that create obligations. Secondary rules that provide conclusive methods for ascertaining what things are primary rules and which primary rules take precedence over others are called rules of recognition (92-93). Rules that permit and structure the process of enacting, altering, and repealing primary rules are called rules of change (93-94). And rules that empower some people to make authoritative determinations of departures and violations are called rules of adjudication (94-95). These three types exhaust the realm of secondary rules for Hart.

Rules of recognition create a system out of an otherwise disunited congeries of rules (90, 91, 93, 106, 113, 118, 228f) and prevent debilitating uncertainty as to what is law and when we are obligated and at risk of sanctions (92). By providing rules of priority, when needed, the rules of recognition also create a hierarchy of rules. Rules of change allow adaptation to changing conditions and prevent all the ills of a static legal system (93).[Note 10] By centralizing the imposition of sanctions, rules of adjudication prevent the violent and inefficient diffusion of energy in self-help (95, 221). Societies may in fact exist without secondary rules, but they would for that very reason be primitive (89). Indeed, the development of secondary rules "is a step forward as important to society as the invention of the wheel" (41).

A peculiarity of Hart's theory that strikes one immediately is that the secondary rules apply only to primary rules and not also to themselves and one another. Yet rules of recognition must occasionally be changed and ajudicated, rules of change recognized and adjudicated, and rules of adjudication recognized and changed. More to the present point, rules of recognition must be recognized, rules of change changed, and rules of adjudication adjudicated. The self-application of the rules of change will concern us exclusively here, although the possibilities of paradox and contradiction in the self-application of rules of recognition and adjudication should not be overlooked. For none of the three types of secondary rule does Hart explicitly affirm or deny the logical or legal possibility of self-application. Hence, whichever path we take will supplement Hart.[Note 11]

The dilemma may be restated partly in Hart's terms. What rules of change, if any, permit and structure the process of the change of the existing rules of change? Especially, what rule of change, if any, can authorize the change of the supreme rule of change to a form inconsistent with its original form? The existing supreme rule of change is either self-applicable or it is not. (1) If it is self-applicable, then it is apparently self-contradictory in the manner of a self-limiting omnipotent being (see Section 5). (2) If it is not self-applicable, then it must either be immutable, or mutable only by "tertiary" rules or by extra-legal means.

By analogy to Hart's distinction between primary and secondary rules, one might postulate rules which apply to the secondaries, called tertiaries, and rules which apply to the tertiaries, called quaternaries, and so on.[Note 12] Immutability and revolution are not the only alternatives to one who denies self-applicability: one may postulate a tertiary rule of change which authorizes the amendment of the secondary rules of change. On this branch of the forking path one must ask whether a tertiary rule of change can exist if the secondary rule is, ex hypothesi, actually supreme. In any case, the tertiary rule of change would replicate the dilemma perfectly, for it itself would either be immutable, or mutable only by self-application, by revolution, or by a quaternary rule, and the quaternary rule likewise, and so on to an arbitrary halt or an infinite regress.

Several horns of the dilemma may be removed for various reasons. But before pruning, the dilemma (tetralemma) breaks into four paths: Supreme secondary rules of change are either (1) immutable or mutable. If they are mutable, then it is by means of (2) higher rules of change, (3) self-application, or (4) extra-legal means such as revolution. If they are mutable by higher rules of change, then those higher rules replicate this tetralemma at their own level, and so on ad infinitum.

The dilemma may be simplified if we eliminate two possibilities. First, we should eliminate the possibility of extra-legal or revolutionary change of the rules of change, if only because we want to explain the phenomenon of their legal change. For the legal change of supreme and inferior rules of change does occur (see Section 5.C and Appendix 2). Even if their extra-legal change occurs as well, the explanation for the latter cannot, and need not, be stretched to cover the former.

Second, we should eliminate the possibility of the immutability of the secondary rules of change as empirically false and question begging. Even supreme rules of change are legally mutable, as evidenced by the history of the amendments to the ACs of the various state constitutions in the United States (see Appendix 2). Lesser rules of change are mutable under the higher rules of change and the ACs. One might argue that the existing (and mutable) ACs are only secondary rules of change, and that there exist rules of change at the tertiary level or above that are genuinely immutable. Alf Ross takes this position (see Section 5). But we are investigating actual ACs, which have historically often been amended. They at least are mutable, and we are only eliminating the assertion of their immutability from the dilemma. The option of affirming the immutability of higher level, but tacit, rules of change is preserved by the denial of the immutability of secondary rules, and will be examined on its merits (see Sections 8 and 9).

It is important to note that to postulate immutable legal rules of any kind is to beg the question of the mutability of the rules of change. For if the rules of change may be changed, then new rules of change might reach all the rules of the system and "transmute" the "immutable" rules. Therefore, Hart's unsupported statements that some rules can lie beyond the amending power of the system, and that the amending power can be limited (71, 72, 76, 103) must be taken as unfortunate obiter dicta that beg the question under investigation. Even if they are taken as indirect denials of the self-applicability of rules of change, which they are, they are not supported by evidence or argument. The proposition that an AC may be limited beyond undoing will be examined in Section 9.

We may eliminate from the dilemma the possibility that the rules of change are immutable, simply on the ground that some supreme rules of change, or actual ACs, are legally mutable. We need not reach the question whether all are mutable. However, we must eventually face the question whether actual ACs have been mutable precisely because they were not actually supreme.

After eliminating the two possibilities of revolution and immutability, the dilemma may be stated very simply. Secondary rules that permit the change of primaries either permit their own change or they do not. If they do permit their own change, then they are self-applicable and prima facie paradoxical. If not, then they are changeable only by tertiary rules, which in turn could be either immutable, or mutable by self-application or by quaternary rules, and so on to an arbitrary halt or an infinite regress.

C. Is law finite?

Perhaps the possibility of infinite regress should also be removed. An infinite regress is objectionable, not because legal systems contain only a finite number of rules, but because obviously no legislator or judge appeals to an infinity of rules to justify any change of law, including self-amendment. Whatever is justified in law, as elsewhere, is justified by a finite number of rules or premises. Conceivably this need not be so, but in fact judges would rather invent rules to justify their holdings, or double the number of cited precedents to hide the fact that their holdings were unjustified, than to appeal to an infinity of rules.[Note 13]

Ronald Dworkin, for one, believes that the number of legal rules and principles is infinite.[Note 14] So does Robert Birmingham.[Note 15] Even if they are not considered to have proved their conclusions, they make it difficult to assume without argument, as Kelsen does, that "the legal order can be composed only by a definite number of rules."[Note 16] Dworkin's thesis is based on the difficulty of individuating legal rules or principles. If it leads us to postulate an infinity of rules, it is because of the innumerable interstices and shades of meaning in any legal system. Birmingham's thesis is based on a concept of rule by which the solution to each distinct legal problem becomes a new rule. Not even Dworkin and Birmingham, however, would recognize an infinity of duplicate rules, each identical to the last except for its place in an infinite regress.

The Dworkin-Birmingham thesis is important to eliminate one possible solution to the paradox of self-amendment. If we could count all the possible laws by some definite method for identifying what was and what was not a possible law, then we could say whether self-amendment was included in the series. Or if we could list all the laws that our theories would like to count as possible laws, ensuring to exclude (or include) self-amendment, then we could define the universe of possible laws as the enumerated series. This looks more absurd for law than for logic, where such methods are sometimes useful.[Note 17] We need not believe the number of laws is actually infinite to reject the utility of this method even in principle; we need only believe that the number of laws cannot be counted or iterated, either because they are always changing, or because they cannot be individuated without distortion and controversy.[Note 18]

D. The paradox of omnipotence, the barber, and the liar

The paradox of omnipotence is widely thought to be of the "Barber-type" (see Section 1), for it seems to prove that omnipotence cannot exist as naively conceived. If we naively assume that a deity can do any act at any time, then can she make a stone so heavy that she cannot lift it? If she can, then she is not omnipotent in the naive sense (because there is something she cannot lift); and if she cannot, then she is not omnipotent in the naive sense (because there is something she cannot create). The natural solution is to deny that there is a deity omnipotent in the naive sense, with the power to do any act at any time.

Similarly, can a constitutional amending clause amend itself —especially, can it do so when it is the only authority for the amendment, when it is the supreme rule of change in that legal system, when the new version of the clause would be inconsistent with the original, when the amendment would diminish the amending power, and when the amendment purports to be irrevocable? If we regard constitutional amending clauses as legally omnipotent, on the evidence that they are the supreme rules of change in their respective systems, then we have replicated the theological version of the paradox of omnipotence. If we do not regard amending clauses as legally omnipotent, then some irrevocable limitation on their power must exist; for if all limitations on it were revocable, some power would be legally omnipotent.

There is no contradiction in assuming that such irrevocable limitations exist. They would imply the existence of immutable rules, which are equally difficult to posit in theory and to find in history. If we insist they are there even though we cannot find them, then no legal power is omnipotent. For even one immutable rule forecloses the possibility of omnipotence, unless we redefine omnipotence. That is a logically permissible solution. If we think immutable rules are not to be found, or if we suppose for the purposes of inquiry that they are not to be found, then we thereby posit at least one omnipotent legal power. For the absence of immutable rules implies that any legal changes can be made, perhaps by a composite of powers working together. If we deny immutability and acknowledge omnipotence, then we have clearly lost the right to reject the assumption of omnipotence —as we reject the assumption of the Barber's existence— when the paradox of omnipotence shows a contradiction in it. So we must apparently choose between the universal absence of legal omnipotence and the permissible presence of legal contradiction. In general I will take the latter path, even though it implies that the paradox of self-amendment is more "Liar-like" than "Barber-like".


1. On the difficulty of deciding when the new and old rules of change are "inconsistent" see Section 12.C. [Resume]

2. See Appendix 3 in which a game is presented in which changing the rules is a move. Also see Sections 4 and 13 for some of the logical problems involved in developing such rules. [Resume]

3. See Section 1, note 17. Ross's argument is summarized in Section 5. [Resume]

4. H.L.A. Hart, The Concept of Law, Oxford University Press, 1961. Subsequent references to this work in this section will be kept in the text in parentheses. In a later essay Hart explicitly faced the problem of self-amendment and replied to Ross. See Hart, "Self-Referring Laws," Festskrift Tillägnad Karl Olivecrona, Stockholm: Kungl. Boktryckeriet, P.A. Norstedt & Söner, 1964, pp. 307-16. I will argue that Hart's explicit response to Ross in this essay is less adequate than one that can be wrought from the doctrines addressing other questions in The Concept of Law. The virtues of the approach of his book are set out in Section 7, and the defects of the approach of his essay are set out in Section 10. In his essay Hart knew Ross's paradox only through Ross's less complete exposition in On Law and Justice, London, 1958, pp. 80-84. In Ross's full exposition, "On Self-Reference and a Puzzle in Constitutional Law," Mind, 78 (1969) 1-24, he shows familiarity with Hart's later essay (at p. 10.n.1). Hence, Ross's best exposition knew both Hart's book and essay, while neither of Hart's works knew Ross's best exposition. Hart reprinted his essay, "Self-Referring Laws," in his collection, Essays in Jurisprudence and Philosophy, Oxford University Press, 1983, pp. 170-178, but does not revise it in light of Ross's fuller essay except to cite it as critical of his own. [Resume]

5. The constitution of Italy lacks an AC but Italian jurists conclude that it is

tacitly understood that amendment may be made either by the ordinary legislative method, or by the same power by which the constitution was originally adopted.

Lester Bernhardt Orfield, The Amending of the Federal Constitution, University of Michigan Press, 1942, p. 39. For the resolution of similar difficulties in the early American state constitutions that lacked ACs see In Re Opinion to the Governor, 55 R.I. 56, 178 A. 433, 449-51 (1935), and Walter Farleigh Dodd, The Revision and Amendment of State Constitutions, Johns Hopkins University Press, 1910, pp. 44-45. [Resume]

6. This qualification is intended to avoid (1) an implied assertion that all legal systems have hierarchies of rules, and (2) an Austin-like inference of the necessary existence of an unlimited and illimitable power or sovereign in every legal system. I do not mean to deny Austin's inference, but I do not, without inquiry, want to endorse his position by my language. For my purposes there may well be in every legal system a rule or power which is not limited by any other power and which limits all others; but I need not take the matter out of the sphere of empirical investigation. See Sections 8 and 9. [Resume]

7. For example, the AC may change constitutional rules directly, and statutory rules indirectly by changing the standards by which we define permissible statutes. [Resume]

8. The definition of supremacy used here may imply that a supreme or constitutional rule of change must also be exclusive or the only rule that can change the constitution. The complexity simplified by this usage is discussed in Section 12.A. [Resume]

9. In Part Two I will examine methods of constitutional amendment that do not use the AC. Each may conceivably amend the AC and thereby prevent the absolute immutability of the AC even if strict self-amendment is not allowed. Because all these methods are either surreptitious, unofficial, de facto rather than legal de jure, or ill-defined in scope or operation, they show how the informality of actual legal systems can save us from logical monsters and political follies. However, in Part One I will assume that (supreme) ACs are lawfully mutable only by self-amendment, if only to tighten the dilemma. [Resume]

10. The permanent need for change and openness to change was well expressed by Robert W. Goedecke in his Change and the Law, Florida State University Press, 1969, at p. 212:

[T]he dominance of reason in the rule of law does not mean the sterile maintenance of a mechanical jurisprudence...but rather the ability to change the bases of the law to allow for continued practical determination in new circumstances and, at the same time, allow for continued argument about these bases... [Moreover, we must allow dissenters to persuade us, and permit] the previous dissenters [to] become the new majority....Change is necessary, therefore, to have the rule of law.


11. Hart ignores the question of the self-application of secondary rules in both The Concept of Law and "Self-Referring Laws" (see note 4 above), although the latter does address some of the problems of the self-application of a rule of change. In The Concept of Law he does not ignore all kinds of self-application, however. For example, he addresses the problem of sovereign self-limitation at 117, 145f, 215f, and 219; the self-binding force of law (which, unlike Austinian commands, bind their makers) at 42, 48, 57, and 77; self-reference at 116, 145-146, and 146-147; self-embracing omnipotence at 146; judges ruling on, and occasionally creating, their own competency at 148f; and the impossibility of self-interpretation at 123 and 139.

When C.F.H. Tapper speaks of the "self-application" of Hartian rules he means only that the people bound by a (primary) rule apply it to themselves, as opposed having it applied to them by a separate class of officials who are not bound by it. C.F.H. Tapper, "Powers and Secondary Rules of Change," in A.W.B. Simpson (ed.), Oxford Essays in Jurisprudence, Second Series, Oxford University Press, 1973, pp. 242-77 at pp. 249, 266. See Hart, op. cit. at 38. [Resume]

12. The need for tertiary rules in Hart's scheme has been argued by D. Gerber, "Levels of Rules and Hart's Concept of Law," Mind, 81 (1972) 102-05, and rejected by Theodore M. Benditt, "On 'Levels of Rules and Hart's Concept of Law,'" (a reply to Gerber), Mind, 83 (1974) 422-23. Gerber points out that secondary rules of adjudication are changed and, because Hart makes no provision for the application of secondary rules to one another, tertiary rules must be used to change secondary rules. Benditt suggests that legislative rules of adjudication (statutes serving the purposes which Hart attributes to rules of adjudication) may be changed by legislatures by means of new rules on the same level as those being changed (i.e., by new statutes); hence, resort to tertiary rules is unnecessary.

Benditt confuses the logical level of rules postulated by Hart with the legal levels of rules illustrated by the hierarchy in which constitutional rules supersede legislative, legislative rules supersede adjudicative, and so on. For Hart, primary and secondary rules may exist at the same legal level (say, at the statutory level); he does not mean that primary rules that are statutes can only be changed by secondary rules that are legally superior to statutes. A new statute may change an old statute in accordance with a rule of recognition which gives new statutes priority over any conflicting pre-existent statutes (see Section 16). The power of one statute to amend another indicates a logical superiority that Hart captures in the term "secondary rule". A secondary rule of adjudication, following Gerber's example, may be changed only by rules of a higher logical order; rules of a higher legal order are not necessary. Neither Gerber nor Benditt examined the problems of permitting self-application, and neither passed beyond exegesis of Hart to face the dilemma of circularity and infinite regress. Hart says his primary and secondary rules exist on different "levels" at 92, 94, and 108. [Resume]

13. The position that legal rules are authorized or validated only by other legal rules, which is central to a formalist model of law, implies either an infinite regress, a self-validating rule, or an exceptional rule that is authorized, say, by consent or acceptance. No formalist has chosen the infinite regress, and most have ignored the dilemma.

Hans Kelsen asserts that all legal norms are coercive in that they provide sanctions for departures and violations. A judge applies the coercive rule against burglary to a convicted burglar; the judge does so because she is constrained by a coercive rule to decide burglary cases under burglary rules. The coercive rule against burglary is applied by a judge under a coercive rule about adjudication. Do we need a higher coercive rule to ensure that the coervice rule about adjudication is applied? It might appear that we need coercive rules in an infinite regress in order to make each lower rule fully coercive and enforceable. But for Kelsen this series is not infinite because a rule is authoritative merely by providing a sanction and need not be authorized by a higher coercive rule. Hans Kelsen, General Theory of Law and the State, Russell and Russell, 1961, "The Never-Ending Series of Sanctions," pp. 28-29. [Resume]

14. Ronald Dworkin, Taking Rights Seriously, Harvard University Press, 1977, at p. 44: If "we tried actually to list all the principles in force we would fail. They are controversial...they are numberless, and they shift and change so fast that the start of our list would be obsolete before we reached the middle." See also pp. 25, 66, 75, 76. [Resume]

15. Robert Birmingham, "'Into the Swamp': More on Rules," Archiv für Rechts- und Sozialphilosophie, 64 (1978) 49-61. Birmingham does not rely on Dworkin, but uses an original argument that legal rules are paths from a set of facts to a holding, and therefore are as numberless as distinct cases. [Resume]

16. Kelsen, op. cit. at 28-29. [Resume]

17. Jaakko Hintikka believes this method is theoretically possible for the paradox of legal omnipotence, but "inadvisable" for "practical reasons". Hintikka, "Remarks on a Paradox," Archiv für Rechts- und Sozialphilosophie, 44 (1958) 514-16, at 515. [Resume]

18. For an excellent discussion of the problem of the individuation of laws, with less emphasis on the question of the cardinality of the set of laws, see Joseph Raz, The Concept of a Legal System, second edition, Oxford University Press, 1980, Chapter IV, "On the Individuation of Laws," pp. 70-92. [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.