Paradox of Self-Amendment by Peter Suber equally refuse its imprimatur to revolution and to ultra vires amendment. In fact, it may term the latter "peaceful revolution", especially if its subsequent validation is deemed to be the result of an amended rule of recognition. Moreover, once the ultra vires amendment or revolution were authorized by acceptance, the rule of recognition would equally grant its imprimatur. Second, an acceptance theorist may reject the normative practice doctrine, as I am tempted to do, and refer ultimate questions of authority to the same congeries of practice without distorting the latter by calling it a "rule". A final, reflexive question should be asked. If the direct acceptance theory requires that there be no categorically immutable rules, then does it follow that there is one categorically immutable rule to the effect that there cannot be others? The answer seems to depend upon whether the direct acceptance theory categorically or merely contingently prevents categorically immutable rules. If the former, then we are left with a principle analogous to Augustine's —if there were no eternal truths, it would be eternally true that there were no eternal truths. The true answer is at once simpler and more complex. Direct acceptance completely precludes categorically immutable rules. If there were one forbidding others, that would be one too many. It does not follow that a categorically immutable rule is contingently possible. The direct acceptance theory is a theory about law, not a set of rules or principles of or within law. If it bars something (does not recognize something), then it does not follow that legal rules bar that something. For an empirical theory of law completely to preclude something is to decide on the evidence that that something does not appear in law, and not necessarily because it is expressly forbidden by a rule of law. Some things are absent without command, descriptively rather than normatively. According to the direct acceptance theory, categorically immutable rules are completely absent from law, not because they are forbidden but because a true description omits them. There is not even one forbidding others. A legal rule that purported to forbid categorically immutable rules would itself be contingently mutable or immutable. It could be repealed, but its repeal would not thereby enact the possibility of enacting categorically immutable rules. No change of law can enact or permit the enactment of categorically immutable rules because the alegal source of authority of law does not permit it and cannot itself be irrevocably changed by changes of law. Acceptance is that alegal source of authority, and it can serve this role as a "permanent" obstacle to categorically immutable rules even though its status as the alegal source of legal authority is only contingently immutable. Changes of law can replace acceptance with the decrees of a priesthood, military council, or economic elite, but only if such changes are accepted. The contingent immutability of acceptance means that acceptance may always awaken from dormancy and take back the power it once was accepted as having. So if no categorically immutable rules are possible in law, in what sense are rules of change contingently omnipotent? There is one class of rules (categorically immutable rules) that cannot be enacted, even through the contingent favorable acceptance of future generations. That is true, and if it is a limit on omnipotence, then rules of change possess only a further qualified contingent omnipotence. However, rules of change can still authorize rules of any content at any time, which is more than categorically self-embracing or categorically continuing omnipotence could do. Moreover, they may make rules that are never repealed —though only by virtue of a perpetual, contingent failure to repeal. Even irrepealable rules could be made, though only if we mean contingent irrepealability. While contingently omnipotent rules of change can authorize rules of any content at any time, they cannot authorize rules of any modality, if this is taken to refer to the categorical and contingent modes of mutability. The incapacity of a rule of change to make categorically immutable rules is identical to its own contingent mutability or immutability. The contingent mutability of a rule of change allows it to be amended or repealed, but only with contingent revocability. Thus its own status as contingently mutable bars it from any irrevocable self-limitation which is not itself subject to transmutation. Hence its status bars it from categorically irrevocable self-limitation. But since all categorically immutable rules (ex hypothesi) are categorically irrevocable limitations on the amending power, they are all barred; they cannot be made by the only powers available for making law. Hence, by defining rules of law —especially rules of change— as contingently mutable, we have already precluded the possibility of categorically immutable rules. But the property of contingent mutability or immutability in all legal rules is inseparable from the property of contingent omnipotence in rules of change. Both define the maximum legal immutability as one subject to contingent future events, and to that extent as mutable in principle. The property of contingent mutability for rules in general defines this condition from the standpoint of the products of change, and the contingent omnipotence of the rules of change define it from the standpoint of the instrumentality of change. Hence this universal property of contingent mutability implies the impossibility of categorically immutable rules and the necessity of contingent omnipotence of the rules of change; and each of the latter implies each of the other two. Under these conditions it is difficult to conceive the impossibility of categorically immutable rules as a limitation on contingent omnipotence, for the concepts define and imply one another. But certainly if we adopt a broader concept of limitation, then we will find contingent omnipotence limited. For in truth, not even contingent omnipotence can make categorically immutable rules, any more than it could categorically cease being contingent omnipotence. D. A word on the merits of the direct acceptance theory My chief task in this work has been to show how law copes with one family of paradoxes. The actual means by which law copes with paradox —for example, permitting self-amendment without more ado— are of course subject to many interpretations and justifications. As I argued in Section 6, Hart's acceptance theory is one persuasive way to interpret the actual legal methods of coping, but I will refrain from affirming it simply on that ground, for other persuasive interpretations also exist. But clearly the capacity to explain the legality of self-amendment, so intractable to the formalist concepts of reason and law, is a strong argument in its favor. So is its consonance with democratic theory as noted in Section 21.A. As 163