Wisconsin severely restricts the wearing of jewelry by jail and prison inmates. A regulation forbids the possession of "items which because of shape or configuration are apt to cause a laceration if applied to the skin with force," and the state refuses to make an exception for religious jewelry, such as crucifixes, which (unless made of cloth) fall within the ban. Inmates brought this suit against the relevant officials to enjoin, as a violation of the Religious Freedom Restoration Act of 1993, 42 U.S.C. §§ 2000bb to 2000bb-4, the defendants' refusal to make such an exception. The officials defend their refusal primarily on the ground that the Act is unconstitutional, exceeding the power that section 5 of the Fourteenth Amendment gives Congress to enforce the amendment, and secondarily on the ground that the ban of religious jewelry satisfies the Act's stringent test for permissible burdening of religion. The district judge rejected the constitutional challenge (as two other circuits in similar cases have done, Flores v. City of Boerne, 73 F.3d 1352, 1359-60 (5th Cir.1996); EEOC v. Catholic University, 83 F.3d 455, 469-70 (D.C.Cir.1996)), Sasnett v. Department of Corrections, 891 F. Supp. 1305, 1315-21 (W.D.Wis.1995), and went on to grant summary judgment for the plaintiffs.

The background and content of the Religious Freedom Restoration Act are discussed in our recent opinion in Mack v. O'Leary, 80 F.3d 1175 (7th Cir.1996), enabling us to be brief. The Act forbids government, federal or state, to "substantially burden" a person's exercise of his or her religion unless the government shows that the burden is the "least restrictive means of furthering [a] compelling governmental interest." 42 U.S.C. § 2000bb-1(b). The Act was motivated by a desire to supersede Employment Division v. Smith, 494 U.S. 872, 110 S. Ct. 1595, 108 L. Ed. 2d 876 (1990), which held that the free-exercise clause of the First Amendment allows the government to do just what the Act forbids, provided that the government does so by means of a law of general applicability not motivated by hostility toward religion or toward a particular sect. In Smith the law was a general prohibition of controlled substances applied to the use of peyote in a religious ceremony of an Indian tribe. Before Smith the Supreme Court had used something much like, perhaps identical to, the test adopted in the Act in interpreting the free-exercise clause. The Act thus seeks to return the courts, when a law burdening religious observance is challenged, to the approach they had taken before Smith. The term "restoration" in the Act's title supports this interpretation of what Congress was about, as do the Act's text and legislative history.

The state argues that section 5 of the Fourteenth Amendment does not authorize Congress to create new rights, Civil Rights Cases, 109 U.S. 3, 11, 3 S. Ct. 18, 21, 27 L. Ed. 835 (1883); City of Rome v. United States, 446 U.S. 156, 210-12, 220, 100 S. Ct. 1548, 1578-80, 1583, 64 L. Ed. 2d 119 (1980) (dissenting opinion), and that this is just what the Religious Freedom Restoration Act did. Smith had held that a person has no right to demand special treatment by government just because he will find it difficult to practice his religion otherwise. The Act creates such a right. This is one way to look at what Congress did, but it cannot be the only way. Whenever Congress passes a law under the authority of section 5, it creates a right. The question is whether it has exceeded its authority in creating the particular right at issue. It has not if the right is reasonably designed to secure a right created by the Fourteenth Amendment itself. The clearest examples come from the Voting Rights Act of 1965, 42 U.S.C. §§ 1973 et seq. The Supreme Court held in Lassiter v. Northampton County Bd. of Elections, 360 U.S. 45, 79 S. Ct. 985, 3 L. Ed. 2d 1072 (1959), that literacy tests for voters do not violate the Fourteenth Amendment--whereupon Congress prohibited literacy tests for voters, 42 U.S.C. § 1973b(e), and the Court upheld the prohibition. Katzenbach v. Morgan, 384 U.S. 641, 86 S. Ct. 1717, 16 L. Ed. 2d 828 (1966). And on the same day that the Court held in Mobile v. Bolden, 446 U.S. 55, 100 S. Ct. 1490, 64 L. Ed. 2d 47 (1980), that at-large electoral systems which have the practical effect of preventing the election of any blacks do not violate the Fourteenth or Fifteenth Amendments, it held in City of Rome v. United States, supra, that the provision of the Voting Rights Act that authorizes the Department of Justice to veto such systems if they have an exclusionary effect is valid under the Fifteenth Amendment's counterpart to section 5.

In both sequences Congress was held to be empowered by the enforcement clauses of the Reconstruction amendments to outlaw a practice that while not unconstitutional deprived a constitutional right of practical efficacy. The legislation bore a necessary and proper relation to the underlying right. Katzenbach v. Morgan, supra, 384 U.S. at 650, 86 S. Ct. at 1723. The plaintiffs argue that RFRA bears a similar relation to the constitutional right to exercise one's religion without interference by government. After Smith the only way to prove a violation of the free-exercise clause is by showing that government discriminated against religion, or a particular religion, by actually targeting a religious practice, rather than hit it by accident while aiming at something else. A tax on churches, a prohibition (shades of Elizabethan England) against conducting a mass, a ban on wearing yarmulkes--only intentional discrimination, as illustrated by these hypothetical cases, is actionable under Smith. Even under this demanding test a law neutral on its face but intended to discourage a particular religious practice or belief infringes the free-exercise clause, Church of the Lukumi Babalu Aye, Inc. v. City of Hialeah, 508 U.S. 520, 534, 113 S. Ct. 2217, 2227, 124 L. Ed. 2d 472 (1993), but the intent behind such a law may be difficult to prove. Its apparent neutrality would be sand in the eyes of the finder of fact. Congress could, the plaintiffs argue, have established the stiffer test of RFRA to prevent government from resorting to such difficult-to-detect methods of discrimination, in just the same way that the Voting Rights Act forbade literacy tests for voters because it was so difficult to prove (what everyone suspected) that the hidden purpose of the tests was to disfranchise blacks. On this view, instead of having to prove that the Wisconsin prison system was motivated by hostility to religion in refusing to permit the wearing of religious jewelry, all that the plaintiffs would have to prove was that the refusal imposed a substantial burden on their religious practice. By proving this they would shift to the state the burden of proving if it could that it had no other way of achieving a legitimate and important governmental objective.

Granted, the legislative history of the Act contains only hints of this theory of what the Act was intended to do. See, e.g., H.R.Rep. No. 88, 103d Cong., 1st Sess. 6 (1993) U.S.Code Cong. & Admin.News 1993, 1895; "The Religious Freedom Restoration Act," Hearing before the S. Comm. on the Judiciary, 102d Cong., 2d Sess., ser. no. J-102-82, at 95-96 (Sept. 18, 1992) (statement of Prof. Douglas Laycock). But a statute's constitutionality should not depend on the ability of congressional staff to write a plausible brief for it into the committee reports. Katzenbach v. Morgan, supra, 384 U.S. at 653, 86 S. Ct. at 1724. Legislative history is too easy to manufacture. And anyway it is not the motive of the legislators that is important (at least when a statute is not attacked on the basis of an invidious purpose, as in Edwards v. Aguillard, 482 U.S. 578, 591-92, 107 S. Ct. 2573, 2581-82, 96 L. Ed. 2d 510 (1987), and Wallace v. Jaffree, 472 U.S. 38, 56-61, 105 S. Ct. 2479, 2489-92, 86 L. Ed. 2d 29 (1985)), but whether the statute they pass is within the scope of their constitutional authority. The motivation behind the Religious Freedom Restoration Act was in fact disagreement with the Supreme Court's interpretation of the Constitution. The Act could still be within the scope of section 5 if its effect is, without infringing any constitutional rights, to make the legal remedies against violations of the free-exercise clause more effective because requiring less proof.

We are given some pause, however, by the lack of a recent history of governmental discrimination against religious observance. That history is not a complete blank. See Douglas Laycock, "Continuity and Change in the Threat to Religious Liberty: The Reformation Era and the Late Twentieth Century," 80 Minn. L.Rev. 1047, 1096 (1996). But it is not easy to take entirely seriously the proposition that the enactment of RFRA was necessary in order to prevent the states from engaging in forms of intentional discrimination that, unlike the ordinance invalidated in Church of the Lukumi Babalu Aye, could not readily be shown to be intentional. The factual basis for the law, when interpreted as a law designed simply to make it easier to prove a violation of an existing constitutional right declared by the Supreme Court, is feebler than the factual basis for the Voting Rights Act. And there is nothing here that corresponds to one of the grounds of City of Rome v. United States, supra, 446 U.S. at 176-77, 100 S. Ct. at 1561-62--that it was necessary to legislate against practices not unconstitutional in themselves in order to extirpate the effects of past discrimination. Both grounds emphasized by the Supreme Court in its decisions upholding provisions of the Voting Rights Act--the desirability of simplifying the proof of discrimination and the desirability of eliminating the present effects of past discrimination--are simply applications of the familiar principle of equitable remedies that an injunction or other equitable decree may fence the defendant in, forbidding lawful as well as unlawful conduct in order to prevent the evasion of the core prohibition in the decree and to extirpate any lingering effects of the violation sought to be remedied. FTC v. National Lead Co., 352 U.S. 419, 429-30, 77 S. Ct. 502, 509-10, 1 L. Ed. 2d 438 (1957); Szabo v. U.S. Marine Corp., 819 F.2d 714, 721 (7th Cir.1987); Larsen v. Sielaff, 702 F.2d 116, 118 (7th Cir.1983).

But we do not think that the scope of section 5 of the Fourteenth Amendment is limited to the strictly remedial aim evident, for example, in the ban on literacy tests. Rights are instruments for achieving concrete goals. The right to exercise one's religion free of government interference is an instrument for achieving the goal of religious freedom. That freedom is impaired by practices that do not, as well as by practices that do, infringe the constitutional right itself as it was authoritatively construed in the Smith case. Religions that have strong support among influential people have enough influence in the political process to be able to make sure that legislation is not passed that will inadvertently burden the observances required or encouraged by the religion. It is not an accident that we have Sunday closing laws rather than Thursday closing laws. Religions that have fewer members, especially if those members are drawn from the margins of society, do not have sufficient influence over the legislative process to avoid being flailed by the dinosaur's tail of legislation of general applicability, legislation not motivated by any animus toward minor sects but merely insensitive to their interests--possibly even oblivious to their existence. The formal right of the members of these sects to the free exercise of their religion may have little practical value if observance is made onerous by general-purpose legislation. It seems to us that Congress can properly legislate under section 5 to make the formal right of religious freedom of persons who lack political power in individual states (yet are somehow able to form or benefit from the formation of an effective coalition to obtain legislative assistance at the national level) a meaningful right. The analogy to the empowerment theory of the Voting Rights Act (see Katzenbach v. Morgan, supra, 384 U.S. at 652-53, 86 S. Ct. at 1724-25) is close. Prisoners, Christian or otherwise, are a good example of a group whose claims to be able to exercise a meaningful freedom of religion are unlikely to be heard in state legislatures.

We are mindful that by progressively incorporating almost the entire Bill of Rights, expansively interpreted, into the due process clause of the Fourteenth Amendment, the Supreme Court has greatly enlarged the reach of section 5, Hutto v. Finney, 437 U.S. 678, 98 S. Ct. 2565, 57 L. Ed. 2d 522 (1978); id. at 717-18, 98 S.Ct. at 2587-88 (dissenting opinion); United States v. Price, 383 U.S. 787, 789, 86 S. Ct. 1152, 1154, 16 L. Ed. 2d 267 (1966), and we share the state's anxiety lest the clause swallow up the remaining powers of state government. Suppose that Congress, concerned that many policemen flout Fourth Amendment rights, which are among those incorporated in the Fourteenth Amendment, passed a law requiring every state and local government to provide two weeks each year of instruction to all its law enforcement officers in the principles of the Fourth Amendment. Or forbade these governments to hire police officers who had ever been found to have violated a suspect's Fourth Amendment rights. Or required high schools to offer instruction in the law of search and seizure. These hypothetical cases and others that could be put suggest the necessity of imposing limits on the scope of the enforcement clause beyond the obvious one that it may not be used to infringe other constitutional rights. We think that in addition it may not be employed beyond reasonable limits fixed with due regard for the autonomy and responsibility of state and local government. Exactly what those limits are is unclear, but we do not think that the Religious Freedom Restoration Act reaches or exceeds them. It does not impose substantial direct costs on the states, as in the examples we have given, because it makes an exception for regulations that are necessary to the furtherance of compelling state interests. It does not enlist state officials as federal functionaries. It does not make the states dance to a federal tune, beyond requiring them to accord a limited measure of protection to religious interests that may not be represented in the politics of the state.

We defer to the Fifth Circuit's analysis of why the Act also does not violate the separation of powers or the establishment clause of the First Amendment. Flores v. City of Boerne, supra, 73 F.3d at 1361-64. We add only that Congress has not "overruled" Smith or assumed a paramount authority of constitutional interpretation in derogation of the Supreme Court's authority under Article III. Smith remains undisturbed within its domain, that of the meaning of the First Amendment. The Religious Freedom Restoration Act does not alter that meaning. It creates a new statutory right designed to buttress the constitutional right.

That the Act, assuming its constitutionality, was soundly applied in this case is not in doubt, given the test we adopted in Mack v. O'Leary, supra, 80 F.3d at 1178-80. The test, so far as applicable to the challenged regulation, is whether the adherents of a religion are being prevented, without justification based on a compelling interest, from engaging in religiously motivated conduct or expression, whether or not the burdened practice is mandatory for adherents. The crucifix is to many Christians the central symbol of their faith. Wearing it on a chain around one's neck, while certainly not required by the laws of the Roman Catholic Church or, so far as we know, any other Christian sect, is religiously motivated. Prohibiting the wearing of it thus places a substantial burden, within the meaning of the Religious Freedom Restoration Act, on the plaintiffs' observance of their religion.

So the burden of justification is on the state; and it has not been carried. The regulation to which the defendants refuse to make an exception forbids the wearing of a crucifix even if it is too small or light to be a weapon (at least more of a weapon than a fist, a tooth, or a fingernail), too inexpensive to barter for a weapon, invisible because worn under clothing, and not a gang symbol or easily confused with one. The state allows prisoners to have rosaries, which could be used to strangle a fellow prisoner or a guard, and bans crucifixes even in correctional facilities wholly occupied by white-collar prisoners who do not belong to gangs or get into fights with each other or the guards. These features of the state's practice blast the case for regarding a ban on crucifixes and other religious jewelry as a serious and measured response to a concern with violence or a concern with gangs, legitimate and important as these concerns are. If particular types of religious jewelry (or religious jewelry of any type in the hands of prisoners reasonably believed prone to use it for purposes of weaponry, barter, or gang insignia) pose a genuine threat to prison security, the state can ban them; prison security is a compelling state interest. Id. at 1180.

We asked the parties to file supplemental briefs concerning the possible bearing on this case of the Prison Litigation Reform Act of 1995, Pub.L. 104-134, Title VIII (to be codified at 18 U.S.C. § 3626). The Act restricts prisoner litigation in a variety of ways, and one of the restrictions--that any injunctive relief be as narrowly drawn as possible, § 802(a) (to be codified at 18 U.S.C. § 3626(a)(1))--is expressly applicable to pending litigation. § 802(b)(1). The parties agree, however, that we should not attempt ourselves in the first instance to determine whether the injunction that the district court issued is consistent with this new provision. It is always open to a defendant to ask the district court to modify an injunction in light of changed circumstances, including a change in the applicable law. Rufo v. Inmates of Suffolk County Jail, 502 U.S. 367, 388-90, 112 S. Ct. 748, 762-63, 116 L. Ed. 2d 867 (1992); Balark v. City of Chicago, 81 F.3d 658, 661-65 (7th Cir.1996); In re Hendrix, 986 F.2d 195, 198 (7th Cir.1993). If the defendants can show that the injunction does not conform to the new law, the district judge will modify the injunction.

AFFIRMED.