(Slip Opinion) OCTOBER TERM, 1999 1 Syllabus NOTE: Where it is feasible, a syllabus (headnote) will be released, as is being done in connection with this case, at the time the opinion is issued. The syllabus constitutes no part of the opinion of the Court but has been prepared by the Reporter of Decisions for the convenience of the reader. See United States v. Detroit Timber & Lumber Co., 200 U. S. 321, 337. SUPREME COURT OF THE UNITED STATES Syllabus MITCHELL ET AL. v. HELMS ET AL. CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT No. 98­1648. Argued December 1, 1999- Decided June 28, 2000 Chapter 2 of the Education Consolidation and Improvement Act of 1981 channels federal funds via state educational agencies (SEA's) to local educational agencies (LEA's), which in turn lend educational materi- als and equipment, such as library and media materials and com- puter software and hardware, to public and private elementary and secondary schools to implement "secular, neutral, and nonideological" programs. The enrollment of each participating school determines the amount of Chapter 2 aid that it receives. In an average year, about 30% of Chapter 2 funds spent in Jefferson Parish, Louisiana, are allocated for private schools, most of which are Catholic or other- wise religiously affiliated. Respondents filed suit alleging, among other things, that Chapter 2, as applied in the parish, violated the First Amendment's Establishment Clause. Agreeing, the Chief Judge of the District Court held, under Lemon v. Kurtzman, 403 U. S. 602, 612­613, that Chapter 2 had the primary effect of advancing re- ligion because the materials and equipment loaned to the Catholic schools were direct aid and the schools were pervasively sectarian. He relied primarily on Meek v. Pittenger, 421 U. S. 349, and Wolman v. Walter, 433 U. S. 229, in which programs providing many of the same sorts of materials and equipment as does Chapter 2 were struck down, even though programs providing for the loan of public school textbooks to religious schools were upheld. After the judge issued an order permanently excluding pervasively sectarian schools in the parish from receiving any Chapter 2 materials or equipment, he re- tired. Another judge then reversed that order, upholding Chapter 2 under, inter alia, Zobrest v. Catalina Foothills School Dist., 509 U. S. 1, in which a public school district was allowed to provide a sign- language interpreter to a deaf student at a Catholic high school as 2 MITCHELL v. HELMS Syllabus part of a federal program for the disabled. While respondents' appeal was pending, this Court decided Agostini v. Felton, 521 U. S. 203, ap- proving a program under Title I of the Elementary and Secondary Education Act of 1965 that provided public employees to teach reme- dial classes at religious and other private schools. Concluding that Agostini had neither directly overruled Meek and Wolman nor re- jected their distinction between textbooks and other in-kind aid, the Fifth Circuit relied on those two cases to invalidate Chapter 2. Held: The judgment is reversed. 151 F. 3d 347, reversed. JUSTICE THOMAS, joined by THE CHIEF JUSTICE, JUSTICE SCALIA, and JUSTICE KENNEDY, concluded that Chapter 2, as applied in Jefferson Parish, is not a law respecting an establishment of religion simply because many of the private schools receiving Chapter 2 aid in the parish are religiously affiliated. Pp. 7­38. (a) In modifying the Lemon test- which asked whether a statute (1) has a secular purpose, (2) has a primary effect of advancing or in- hibiting religion, or (3) creates an excessive entanglement between government and religion, see 403 U. S., at 612­613- Agostini exam- ined only the first and second of those factors, see 521 U. S., at 222­ 223, recasting the entanglement inquiry as simply one criterion rele- vant to determining a statute's effect, id., at 232­233. The Court also acknowledged that its cases had pared somewhat the factors that could justify a finding of excessive entanglement. Id., at 233­234. It then set out three primary criteria for determining a statute's effect: Government aid has the effect of advancing religion if it (1) results in governmental indoctrination, (2) defines its recipients by reference to religion, or (3) creates an excessive entanglement. Id., at 233­234. In this case, the inquiry under Agostini's purpose and effect test is a narrow one. Because the District Court's holding that Chapter 2 has a secular purpose is not challenged, only Chapter 2's effect need be considered. Further, in determining that effect, only the first two Agostini criteria need be considered, because the District Court's holding that Chapter 2 does not create an excessive entanglement is not challenged. Pp. 7­9. (b) Whether governmental aid to religious schools results in relig- ious indoctrination ultimately depends on whether any indoctrination that occurs could reasonably be attributed to governmental action. See, e.g., Agostini, 521 U. S., at 226. Moreover, the answer to the in- doctrination question will resolve the question whether an educational aid program "subsidizes" religion. See id., at 230­231. In distinguish- ing between indoctrination that is attributable to the State and in- doctrination that is not, the Court has consistently turned to the neu- trality principle, upholding aid that is offered to a broad range of Cite as: 530 U. S. ____ (2000) 3 Syllabus groups or persons without regard to their religion. As a way of as- suring neutrality, the Court has repeatedly considered whether any governmental aid to a religious institution results from the genuinely independent and private choices of individual parents, e.g., id., at 226. Agostini's second primary criterion- whether an aid program defines its recipients by reference to religion, 521 U. S., at 234- is closely related to the first. It looks to the same facts as the neutrality inquiry, see id., at 225­226, but uses those facts to answer a some- what different question- whether the criteria for allocating the aid create a financial incentive to undertake religious indoctrination, id., at 231. Such an incentive is not present where the aid is allocated on the basis of neutral, secular criteria that neither favor nor disfavor religion, and is made available to both religious and secular benefici- aries on a nondiscriminatory basis. Ibid. Pp. 9­15. (c) Two rules offered by respondents to govern the determination whether Chapter 2 has the effect of advancing religion are rejected. Pp. 15­27. (i) Respondents' chief argument- that direct, nonincidental aid to religious schools is always impermissible- is inconsistent with this Court's more recent cases. The purpose of the direct/indirect distinc- tion is to present "subsidization" of religion, and the Court's more re- cent cases address this concern through the principle of private choice, as incorporated in the first Agostini criterion (i.e., whether any indoctrination could be attributed to the government). If aid to schools, even "direct aid," is neutrally available and, before reaching or benefiting any religious school, first passes through the hands (lit- erally or figuratively) of numerous private citizens who are free to di- rect the aid elsewhere, the government has not provided any "support of religion." Witters v. Washington Dept. of Servs. for Blind, 474 U. S. 481, 489. Although the presence of private choice is easier to see when aid literally passes through individuals' hands, there is no reason why the Establishment Clause requires such a form. Indeed, Agostini ex- pressly rejected respondents' absolute line. 521 U. S., at 225. To the extent respondents intend their direct/indirect distinction to require that any aid be literally placed in schoolchildren's hands rather than given directly to their schools, Meek and Wolman, the cases on which they rely, demonstrate the irrelevance of such formalism. Further, respondents' formalistic line breaks down in the application to real- world programs. Whether a program is labeled "direct" or "indirect" is a rather arbitrary choice that does not further the constitutional analysis. See Allen, supra, at 243­245. Although "special Establish- ment Clause dangers" may exist when money is given directly to re- ligious schools, see, e.g., Rosenberger v. Rector and Visitors of Univ. of Va., 515 U. S. 819, 842, such direct payments are not at issue here. Pp. 4 MITCHELL v. HELMS Syllabus 17­21. (ii) Respondents' second argument- that provision to religious schools of aid that is divertible to religious use is always impermissi- ble- is also inconsistent with the Court's more recent cases, particu- larly Zobrest, supra, at 18­23, and Witters and is also unworkable. Meek and Wolman, on which respondents appear to rely for their di- vertibility rule, offer little, if any, support for their rule. The issue is not divertibility but whether the aid itself has an impermissible con- tent. Where the aid would be suitable for use in a public school, it is also suitable for use in any private school. Similarly, the prohibition against the government providing impermissible content resolves the Establishment Clause concerns that exist if aid is actually diverted to religious uses. See, e.g., Agostini, supra, at 224­226. A concern for divertibility, as opposed to improper content, is also misplaced be- cause it is boundless- enveloping all aid, no matter how trivial- and thus has only the most attenuated (if any) link to any realistic con- cern for preventing an establishment of religion. Finally, any aid, with or without content, is "divertible" in the sense that it allows schools to "divert" resources. Yet the Court has not accepted the re- current argument that all aid is forbidden because aid to one aspect of an institution frees it to spend its other resources on religious ends. E.g., Committee for Public Ed. and Religious Liberty v. Regan, 444 U. S. 646, 658. Pp. 21­27. (d) Additional factors cited by the dissent- including the concern for political divisiveness that post-Aguilar cases have disregarded, see, e.g., Agostini, supra, at 233­234, are rejected. In particular, whether a recipient school is pervasively sectarian, a factor that has been disregarded in recent cases, e.g., Witters, supra, is not relevant to the constitutionality of a school-aid program. Pp. 27­31. (e) Applying the two relevant Agostini criteria reveals that there is no basis for concluding that Jefferson Parish's Chapter 2 program has the effect of advancing religion. First, Chapter 2 does not define its recipients by reference to religion, since aid is allocated on the ba- sis of neutral, secular criteria that neither favor nor disfavor religion, and is made available to both religious and secular beneficiaries on a nondiscriminatory basis. 521 U. S., at 231. There is no improper in- centive because, under the statute, aid is allocated based on school enrollment. Second, Chapter 2 does not result in governmental in- doctrination of religion. It determines eligibility for aid neutrally, making a broad array of schools eligible without regard to their re- ligious affiliations or lack thereof. See id., at 225­226. It also allo- cates aid based on the private choices of students and their parents as to which schools to attend. See id., at 222. Thus, it is not prob- lematic that Chapter 2 could fairly be described as providing "direct" Cite as: 530 U. S. ____ (2000) 5 Syllabus aid. Finally, the Chapter 2 aid provided to religious schools does not have an impermissible content. The statute explicitly requires that such aid be "secular, neutral, and nonideological," and the record in- dicates that the Louisiana SEA and the Jefferson Parish LEA have faithfully enforced this requirement insofar as relevant to this case. Although there is evidence that equipment has been, or at least eas- ily could be, diverted for use in relgious classes, that evidence is not relevant to the constitutional analysis. Scattered de minimis statu- tory violations of the restrictions on content, discovered and remedied by the relevant authorities themselves before this litigation began almost 15 years ago, should not be elevated to such a level as to con- vert an otherwise unobjectionable parishwide program into a law that has the effect of advancing religion. Pp. 31­37. (f) To the extent that Meek and Wolman conflict with the foregoing analysis, they are overruled. Pp. 37­38. JUSTICE O'CONNOR, joined by JUSTICE BREYER, concluded that Agos- tini v. Felton, 521 U. S. 203, controls the constitutional inquiry pre- sented here, and requires reversal of the Fifth Circuit's judgment that the Chapter 2 program is unconstitutional as applied in Jeffer- son Parish. To the extent Meek v. Pittenger, 421 U. S. 349, and Wolman v. Walter, 433 U. S. 229, are inconsistent with the Court's judgment to- day, they should be overruled. Pp. 1­33. (a) The plurality announces a rule of unprecedented breadth for the evaluation of Establishment Clause challenges to government school-aid programs. That rule is particularly troubling because, first, its treatment of neutrality comes close to assigning that factor singular importance in the future adjudication of Establishment Clause challenges to school-aid programs. Although neutrality is im- portant, see, e.g., Agostini, 521 U. S., at 228, 231­232, the Court has never held that a government-aid program passes constitutional muster solely because of the neutral criteria it employs as a basis for distributing aid. Rather, neutrality has heretofore been only one of several factors the Court considers. See, e.g., id., at 226­228. Sec- ond, the plurality's approval of actual diversion of government aid to religious indoctrination is in tension with this Court's precedents. See, e.g., id., at 226­227. Actual diversion is constitutionally imper- missible. E.g., Bowen v. Kendrick, 487 U. S. 589, 621­622, 624. The Court should not treat a per-capita-aid program like Chapter 2 the same as the true private choice programs approved in Witters v. Washington Dept. of Servs. for Blind, 474 U. S. 481, and Zobrest v. Catalina Foothills School Dist., 509 U. S. 1. Because Agostini repre- sents the Court's most recent attempt to devise a general framework for approaching questions concerning neutral school-aid programs, and involved an Establishment Clause challenge to a school-aid pro- 6 MITCHELL v. HELMS Syllabus gram closely related to the instant program, the Agostini criteria should control here. Pp. 2­9. (b) Under Agostini, the Court asks whether the government acted with the purpose of advancing or inhibiting religion and whether the aid has the "effect" of doing so. 521 U. S., at 222­223. The specific criteria used to determine an impermissible effect have changed in recent cases, see id., at 223, which disclose three primary criteria to guide the determination: (1) whether the aid results in governmental indoctrination, (2) whether the program defines its recipients by ref- erence to religion, and (3) whether the aid creates an excessive en- tanglement between government and religion, id., at 234. Finally, the same criteria can be reviewed to determine whether a program constitutes endorsement of religion. Id., at 235. Respondents neither question the Chapter 2 program's secular purpose nor contend that it creates an excessive entanglement. Accordingly, the Court need ask only whether Chapter 2, as applied in Jefferson Parish, results in governmental indoctrination or defines its recipients by reference to religion. It is clear that Chapter 2 does not so define aid recipients. Rather, it uses wholly neutral and secular criteria to allocate aid to students enrolled in religious and secular schools alike. As to the in- doctrination inquiry, the Chapter 2 program bears the same hall- marks of the program upheld in Agostini: Aid is allocated on the basis of neutral, secular criteria; it is supplementary to, and does not sup- plant, non-federal funds; no Chapter 2 funds reach the coffers of re- ligious schools; the aid is secular; evidence of actual diversion is de minimis; and the program includes adequate safeguards. Regardless of whether these factors are constitutional requirements, they are sufficient to find that the program at issue does not have the imper- missible effect of advancing religion. For the same reasons, the Chapter 2 program cannot reasonably be viewed as an endorsement of religion. Pp. 9­14. (c) Respondents' contentions that Agostini is distinguishable and that Meek and Wolman are controlling here, must be rejected. Meek and Wolman created an inexplicable rift within the Court's Estab- lishment Clause jurisprudence. Those decisions adhered to the prior holding in Board of Ed. of Central School Dist. No. 1 v. Allen, 392 U. S. 236, that statutes authorizing the lending of textbooks to religious school students did not violate the Establishment Clause, see, e.g., Meek, 421 U. S., at 359­362 (plurality opinion), but invalidated the lending of instructional materials and equipment to religious schools, e.g., id., at 362­366, on the ground that any assistance in support of the pervasively sectarian schools' educational missions would inevi- tably have the impermissible effect of advancing religion, see, e.g., id., at 365­366. The irrationality of this distinction is patent. See Cite as: 530 U. S. ____ (2000) 7 Syllabus Wallace v. Jaffree, 472 U. S. 38, 110. Respondents' assertion that materials and equipment, unlike textbooks, are reasonably divertible to religious uses is rejected because it does not provide a logical dis- tinction: An educator can use virtually any instructional tool, even a textbook, to teach a religious message. Pp. 14­22. (d) The Court should follow the rule applied in the context of text- book lending programs: To establish a First Amendment violation, plaintiffs must prove that the aid actually is, or has been, used for religious purposes. See, e.g., Allen, supra, at 248. Agostini and the cases on which it relied have undermined the assumptions underly- ing Meek and Wolman. Agostini's definitive rejection of the presump- tion that public-school employees teaching in religious schools would inevitably inculcate religion also stood for- or at least strongly pointed to- the broader proposition that such presumptions of relig- ious indoctrination are normally inappropriate when evaluating neu- tral school-aid programs under the Establishment Clause. Respon- dents' contentions that Agostini should be limited to its facts, and that a presumption of religious inculcation for instructional materials and equipment should be retained, must be rejected. The assumption that religious-school instructors can abide by restrictions on the use of government-provided textbooks, see Meek, supra, at 384, should extend to instructional materials and equipment. School Dist. of Grand Rapids v. Ball, 473 U. S. 373, 399­400 (O'CONNOR, J., concur- ring in judgment in part and dissenting in part), distinguished. Pp. 22­25. (e) Respondents' contention that the actual administration of Chapter 2 in Jefferson Parish violated the Establishment Clause is rejected. The limited evidence amassed by respondents during 4 years of discovery (which began approximately 15 years ago) is at best de minimis and therefore insufficient to affect the constitutional inquiry. Their assertion that the government must have a failsafe mechanism capable of detecting any instance of diversion was re- jected in Agostini, supra, at 234. Because the presumption adopted in Meek and Wolman respecting the use of instructional materials and equipment by religious-school teachers should be abandoned, there is no constitutional need for pervasive monitoring under the Chapter 2 program. Moreover, a review of the specific safeguards employed under Chapter 2 at the federal, state, and local levels dem- onstrates that they are constitutionally sufficient. Respondents' evi- dence does not demonstrate any actual diversion, but, at most, proves the possibility of diversion in two isolated instances. The evidence of violations of Chapter 2's supplantation and secular-content restric- tions is equally insignificant and, therefore, should be treated the same. This Court has never declared an entire aid program unconsti- 8 MITCHELL v. HELMS Syllabus tutional on Establishment Clause grounds solely because of viola- tions on the miniscule scale of those at issue here. The presence of so few examples tends to show not that the "no-diversion" rules have failed, but that they have worked. Pp. 26­33. THOMAS, J., announced the judgment of the Court and delivered an opinion, in which REHNQUIST, C. J., and SCALIA and KENNEDY, JJ., joined. O'CONNOR, J., filed an opinion concurring in the judgment, in which BREYER, J., joined. SOUTER, J., filed a dissenting opinion, in which STEVENS and GINSBURG, JJ., joined. Cite as: 530 U. S. ____ (2000) 1 Opinion of THOMAS, J. NOTICE: This opinion is subject to formal revision before publication in the preliminary print of the United States Reports. Readers are requested to notify the Reporter of Decisions, Supreme Court of the United States, Wash- ington, D. C. 20543, of any typographical or other formal errors, in order that corrections may be made before the preliminary print goes to press. SUPREME COURT OF THE UNITED STATES _________________ No. 98­1648 _________________ GUY MITCHELL, ET AL., PETITIONERS v. MARY L. HELMS ET AL. ON WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT [June 28, 2000] JUSTICE THOMAS announced the judgment of the Court and delivered an opinion, in which THE CHIEF JUSTICE, JUSTICE SCALIA, and JUSTICE KENNEDY join. As part of a longstanding school aid program known as Chapter 2, the Federal Government distributes funds to state and local governmental agencies, which in turn lend educational materials and equipment to public and private schools, with the enrollment of each participating school determining the amount of aid that it receives. The que s- tion is whether Chapter 2, as applied in Jefferson Parish, Louisiana, is a law respecting an establishment of religion, because many of the private schools receiving Chapter 2 aid in that parish are religiously affiliated. We hold that Chapter 2 is not such a law. IA Chapter 2 of the Education Consolidation and Improv e- ment Act of 1981, Pub. L. 97­35, 95 Stat. 469, as 2 MITCHELL v. HELMS Opinion of THOMAS, J. amended, 20 U. S. C. §§7301­7373,1 has its origins in the Elementary and Secondary Education Act of 1965 (ESEA), Pub. L. 89­10, 79 Stat. 27, 55, and is a close cousin of the provision of the ESEA that we recently considered in Agostini v. Felton, 521 U. S. 203 (1997). Like the provision at issue in Agostini, Chapter 2 channels federal funds to local educational agencies (LEA's), which are usually public school districts, via state educational agencies (SEA's), to implement programs to assist children in elementary and secondary schools. Among other things, Chapter 2 provides aid "for the acquisition and use of instructional and ed u- cational materials, including library services and m a- terials (including media materials), assessments, re f- erence materials, computer software and hardware for instructional use, and other curricular materials." 20 U. S. C. §7351(b)(2). LEA's and SEA's must offer assistance to both public and private schools (although any private school must be nonprofit). §§7312(a), 7372(a)(1). Participating private schools receive Chapter 2 aid based on the number of children enrolled in each school, see §7372(a)(1), and allocations of Chapter 2 funds for those schools must generally be "equal (consistent with the number of chi l- dren to be served) to expenditures for programs . . . for children enrolled in the public schools of the [LEA]," §7372(b). LEA's must in all cases "assure equitable pa r- ticipation" of the children of private schools "in the pu r- poses and benefits" of Chapter 2. §7372(a)(1); see - - - - - - 1 Chapter 2 is now technically Subchapter VI of Chapter 70 of 20 U. S. C., where it was codified by the Improving America's Schools Act of 1994, Pub. L. 103­382, 108 Stat. 3707. For convenience, we will use the term "Chapter 2," as the lower courts did. Prior to 1994, Chapter 2 was codified at 20 U. S. C. §§2911­2976 (1988 ed.). Cite as: 530 U. S. ____ (2000) 3 Opinion of THOMAS, J. §7372(b). Further, Chapter 2 funds may only "supplement and, to the extent practical, increase the level of funds that would . . . be made available from non-Federal sources." §7371(b). LEA's and SEA's may not operate their programs "so as to supplant funds from non-Federal sources." Ibid. Several restrictions apply to aid to private schools. Most significantly, the "services, materials, and equipment" provided to private schools must be "secular, neutral, and nonideological." §7372(a)(1). In addition, private schools may not acquire control of Chapter 2 funds or title to Chapter 2 materials, equipment, or property. §7372(c)(1). A private school receives the materials and equipment listed in §7351(b)(2) by submitting to the LEA an applic a- tion detailing which items the school seeks and how it will use them; the LEA, if it approves the application, pu r- chases those items from the school's allocation of funds, and then lends them to that school. In Jefferson Parish (the Louisiana governmental unit at issue in this case), as in Louisiana as a whole, private schools have primarily used their allocations for nonrecu r- ring expenses, usually materials and equipment. In the 1986­1987 fiscal year, for example, 44% of the money budgeted for private schools in Jefferson Parish was spent by LEA's for acquiring library and media materials, and 48% for instructional equipment. Among the materials and equipment provided have been library books, compu t- ers, and computer software, and also slide and movie projectors, overhead projectors, television sets, tape re- corders, VCR's, projection screens, laboratory equipment, maps, globes, filmstrips, slides, and cassette recordings. 2 - - - - - - 2 Congress in 1988 amended the section governing the sorts of mat e- rials and equipment available under Chapter 2. Compare 20 U. S. C. §3832(1)(B) (1982 ed.) with §7351(b)(2) (1994 ed.). The record in this case closed in 1989, and the effect of the amendment is not at issue. 4 MITCHELL v. HELMS Opinion of THOMAS, J. It appears that, in an average year, about 30% of Cha p- ter 2 funds spent in Jefferson Parish are allocated for private schools. For the 1985­1986 fiscal year, 41 private schools participated in Chapter 2. For the following year, 46 participated, and the participation level has remained relatively constant since then. See App. 132a. Of these 46, 34 were Roman Catholic; 7 were otherwise religiously affiliated; and 5 were not religiously affil iated. B Respondents filed suit in December 1985, alleging, among other things, that Chapter 2, as applied in Jeffe r- son Parish, violated the Establishment Clause of the First Amendment of the Federal Constitution. The case's tortu- ous history over the next 15 years indicates well the d e- gree to which our Establishment Clause jurisprudence has shifted in recent times, while nevertheless retaining anomalies with which the lower courts have had to stru g- gle.In 1990, after extended discovery, Chief Judge Heebe of the District Court for the Eastern District of Louisiana granted summary judgment in favor of respondents. Helms v. Cody, Civ. A. No. 85­5533, 1990 WL 36124 (Mar. 27), App. to Pet. for Cert. 137a. He held that Chapter 2 violated the Establishment Clause because, under the second part of our three-part test in Lemon v. Kurtzman, 403 U. S. 602, 612­613 (1971), the program had the primary effect of advancing religion. Chapter 2 had such effect, in his view, because the materials and equipment loaned to the Catholic schools were direct aid to those schools and because the Catholic schools were, he concluded after detailed i n- quiry into their doctrine and curriculum, "pervasively se c- tarian." App. to Pet. for Cert. 151a. Chief Judge Heebe relied primarily on Meek v. Pittenger, 421 U. S. 349 (1975), and Wolman v. Walter, 433 U. S. 229 (1977), in which we held unconstitutional programs that provided many of the Cite as: 530 U. S. ____ (2000) 5 Opinion of THOMAS, J. same sorts of materials and equipment as does Chapter 2. In 1994, after having resolved the numerous other issues in the case, he issued an order permanently excluding perv a- sively sectarian schools in Jefferson Parish from receiving any Chapter 2 materials or equipment. Two years later, Chief Judge Heebe having retired, Judge Livaudais received the case. Ruling in early 1997 on postjudgment motions, he reversed the decision of former Chief Judge Heebe and upheld Chapter 2, pointing to several significant changes in the legal landscape over the previous seven years. Helms v. Cody, 1997 WL 35283 (Jan. 28), App. to Pet. for Cert. 79a. In particular, Judge Livaudais cited our 1993 decision in Zobrest v. Catalina Foothills School Dist., 509 U. S. 1, in which we held that a State could, as part of a federal program for the disabled, provide a sign-language interpreter to a deaf student at a Catholic high school. Judge Livaudais also relied heavily on a 1995 decision of the Court of Appeals for the Ninth Circuit, Walker v. San Francisco Unified School Dist., 46 F. 3d 1449, upholding Chapter 2 on facts that he found "virtually indistinguis h- able." The Ninth Circuit acknowledged in Walker, as Judge Heebe had in his 1990 summary judgment ruling, that Meek and Wolman appeared to erect a constitutional distinction between providing textbooks (permissible) and providing any other in-kind aid (impermissible). 46 F. 3d, at 1464­1465; see Board of Ed. of Central School Dist. No. 1 v. Allen, 392 U. S. 236 (1968) (upholding textbook pr o- gram). The Court of Appeals viewed this distinction, however, as "thin" and "unmoored from any Establishment Clause principles," and, more importantly, as "rendered untenable" by subsequent cases, particularly Zobrest. 46 F. 3d, at 1465­1466. These cases, in the Ninth Circuit's view, revived the principle of Allen and of Everson v. 6 MITCHELL v. HELMS Opinion of THOMAS, J. Board of Ed. of Ewing,3 that "state benefits provided to all citizens without regard to religion are constitutional." 46 F. 3d, at 1465. The Ninth Circuit also relied, id., at 1467, on our observation in Board of Ed. of Kiryas Joel Village School Dist. v. Grumet, 512 U. S. 687 (1994), that "we have frequently relied explicitly on the general availability of any benefit provided religious groups or individuals in turning aside Establishment Clause challenges," id., at 704. The Ninth Circuit purported to distinguish Meek and Wolman based on the percentage of schools receiving aid that were parochial (a large percentage in those cases and a moderate percentage in Walker), 46 F. 3d, at 1468, but that court undermined this distinction when it observed that Meek also upheld "the massive provision of textbooks to parochial schools." 46 F. 3d, at 1468, n. 16. Thus, although the Ninth Circuit did not explicitly hold that Meek and Wolman were no longer good law, its reasoning seemed to require that conclusion. Finally, in addition to relying on our decision in Zobrest and the Ninth Circuit's decision in Walker, Judge Livau- dais invoked Rosenberger v. Rector and Visitors of Univ. of Va., 515 U. S. 819 (1995), in which, a f ew months after Walker, we held that the Establishment Clause does not require a public university to exclude a student-run reli g- ious publication from assistance available to numerous other student-run publications. Following Judge Livaudais's ruling, re spondents ap- pealed to the Court of Appeals for the Fifth Circuit. While that appeal was pending, we decided Agostini, in which we approved a program that, under Title I of the ESEA, provided public employees to teach remedial classes at - - - - - - 3 Everson v. Board of Ed. of Ewing, 330 U. S. 1 (1947) (upholding reimbursement to parents for costs of busing their children to public or private school). Cite as: 530 U. S. ____ (2000) 7 Opinion of THOMAS, J. private schools, including religious schools. In so holding, we overruled Aguilar v. Felton, 473 U. S. 402 (1985), and partially overruled School Dist. of Grand Rapids v. Ball, 473 U. S. 373 (1985), both of which had involved such a program. The Fifth Circuit thus faced a dilemma between, on the one hand, the Ninth Circuit's holding and analysis in Walker and our subsequent decisions in Rosenberger and Agostini, and, on the other hand, our holdings in Meek and Wolman. To resolve the dilemma, the Fifth Circuit aba n- doned any effort to find coherence in our case law or to divine the future course of our decisions and instead f o- cused on our particular holdings. Helms v. Picard, 151 F. 3d 347, 371 (1998). It thought such an approach r e- quired not only by the lack of coherence but also by Agos- tini's admonition to lower courts to abide by any applicable holding of this Court even though that holding might seem inconsistent with our subsequent decisions, see Agostini, 521 U. S., at 237. The Fifth Circuit acknowledged that Agostini, by recognizing our rejection of the rule that "all government aid that directly assists the educational fun c- tion of religious schools is invalid," id., at 225, had rejected a premise of Meek, but that court nevertheless concluded that Agostini had neither directly overruled Meek and Wolman nor rejected their distinction between textbooks and other in-kind aid. The Fifth Circuit therefore co n- cluded that Meek and Wolman controlled, and thus it held Chapter 2 unconstitutional. We granted certiorari. 527 U. S. 1002 (1999). II The Establishment Clause of the First Amendment dictates that "Congress shall make no law respecting an establishment of religion." In the over 50 years since Everson, we have consistently struggled to apply these 8 MITCHELL v. HELMS Opinion of THOMAS, J. simple words in the context of governmental aid to religious schools.4 As we admitted in Tilton v. Richardson, 403 U. S. 672 (1971), "candor compels the acknowledgment that we can only dimly perceive the boundaries of permissible go v- ernment activity in this sensitive area." Id., at 678 (plur- ality opinion); see id., at 671 (White, J., concurring in judgment). In Agostini, however, we brought some clarity to our case law, by overruling two anomalous precedents (one in whole, the other in part) and by consolidating some of our previously disparate considerations under a revised test. Whereas in Lemon we had considered whether a statute (1) has a secular purpose, (2) has a primary effect of a d- vancing or inhibiting religion, or (3) creates an excessive entanglement between government and religion, see 403 U. S., at 612­613, in Agostini we modified Lemon for purposes of evaluating aid to schools and examined only the first and second factors, see 521 U. S., at 222­223. We acknowledged that our cases discussing excessive enta n- glement had applied many of the same considerations as had our cases discussing primary effect, and we therefore recast Lemon's entanglement inquiry as simply one crit e- rion relevant to determining a statute's effect. Agostini, supra, at 232­233. We also acknowledged that our cases had pared somewhat the factors that could justify a fin d- ing of excessive entanglement. 521 U. S., at 233­234. We then set out revised criteria for determining the effect of a statute: "To summarize, New York City's Title I program does not run afoul of any of three primary criteria we - - - - - - 4 Cases prior to Everson discussed the issue only indirectly, see e.g., Vidal v. Philadelphia, 2 How. 127, 198­200 (1844); Quick Bear v. Leupp, 210 U. S. 50, 81 (1908), or evaluated aid to schools under other prov i- sions of the Constitution, see Cochran v. Louisiana Bd. of Ed., 281 U. S. 370, 374­375 (1930). Cite as: 530 U. S. ____ (2000) 9 Opinion of THOMAS, J. currently use to evaluate whether government aid has the effect of advancing religion: It does not result in governmental indoctrination; define its recipients by reference to religion; or create an excessive entangl e- ment." Id., at 234. In this case, our inquiry under Agostini's purpose and effect test is a narrow one. Because respondents do not challenge the District Court's holding that Chapter 2 has a secular purpose, and because the Fifth Circuit also did not question that holding, cf. 151 F. 3d, at 369, n. 17, we will consider only Chapter 2's effect. Further, in determining that effect, we will consider only the first two Agostini criteria, since neither respondents nor the Fifth Circuit has questioned the District Court's holding, App. to Pet. for Cert. 108a, that Chapter 2 does not create an excessive entanglement. Considering Chapter 2 in light of our more recent case law, we conclude that it neither results in religious indoctrination by the government nor defines its recipients by reference to religion. We therefore hold that Chapter 2 is not a "law respecting an establishment of religion." In so holding, we acknowledge what both the Ninth and Fifth Circuits saw was inescapable- Meek and Wolman are anomalies in our case law. We therefore conclude that they are no longer good law. A As we indicated in Agostini, and have indicated else- where, the question whether governmental aid to religious schools results in governmental indoctrination is ult i- mately a question whether any religious indoctrination that occurs in those schools could reasonably be attributed to governmental action. See Agostini, supra, at 226 (quoting Zobrest, 509 U. S., at 10 (presence of sign- language interpreter in Catholic school " `cannot be attrib- uted to state decisionmaking' ") (emphasis added in Agos- tini)); 521 U. S., at 230 (question is whether "any use of 10 MITCHELL v. HELMS Opinion of THOMAS, J. [governmental] aid to indoctrinate religion could be a t- tributed to the State"); see also Rosenberger, 515 U. S., at 841­842; Witters v. Washington Dept. of Servs. for Blind, 474 U. S. 481, 488­489 (1986); Mueller v. Allen, 463 U. S. 388, 397 (1983); cf. Corporation of Presiding Bishop of Church of Jesus Christ of Latter-day Saints v. Amos, 483 U. S. 327, 337 (1987) ("For a law to have forbidden `effects' under Lemon, it must be fair to say that the government itself has advanced religion through its own activities and influence"). We have also indicated that the answer to the question of indoctrination will resolve the question whether a program of educational aid "subsidizes" religion, as our religion cases use that term. See Agostini, 521 U. S., at 230­231; see also id., at 230. In distinguishing between indoctrination that is attri b- utable to the State and indoctrination that is not, we have consistently turned to the principle of neutrality, uphol d- ing aid that is offered to a broad range of groups or pe r- sons without regard to their religion. If the religious, irreligious, and areligious are all alike eligible for gover n- mental aid, no one would conclude that any indoctrination that any particular recipient conducts has been done at the behest of the government. For attribution of indoctr i- nation is a relative question. If the government is offering assistance to recipients who provide, so to speak, a broad range of indoctrination, the government itself is not thought responsible for any particular indoctrination. To put the point differently, if the government, seeking to further some legitimate secular purpose, offers aid on the same terms, without regard to religion, to all who ad e- quately further that purpose, see Allen, 392 U. S., at 245­ 247 (discussing dual secular and religious purposes of religious schools), then it is fair to say that any aid going to a religious recipient only has the effect of furthering that secular purpose. The government, in crafting such an aid program, has had to conclude that a given level of aid Cite as: 530 U. S. ____ (2000) 11 Opinion of THOMAS, J. is necessary to further that purpose among secular recip i- ents and has provided no more than that same level to religious recipients. As a way of assuring neutrality, we have repeatedly considered whether any governmental aid that goes to a religious institution does so "only as a result of the gen u- inely independent and private choices of individuals." Agostini, supra, at 226 (internal quotation marks omitted). We have viewed as significant whether the "private choices of individual parents," as opposed to the "unmed i- ated" will of government, Ball, 473 U. S., at 395, n. 13 (internal quotation marks omitted), determine what schools ultimately benefit from the governmental aid, and how much. For if numerous private choices, rather than the single choice of a government, determine the distrib u- tion of aid pursuant to neutral eligibility criteria, then a government cannot, or at least cannot easily, grant special favors that might lead to a religious establishment. Pr i- vate choice also helps guarantee neutrality by mitigating the preference for pre-existing recipients that is arguably inherent in any governmental aid program, see, e.g., Gilder, The Revitalization of Everything: The Law of the Microcosm, Harv. Bus. Rev. 49 (Mar./Apr. 1988), and that could lead to a program inadvertently favoring one reli g- ion or favoring religious private schools in general over nonreligious ones. The principles of neutrality and private choice, and their relationship to each other, were prominent not only in Agostini, supra, at 225­226, 228, 230­232, but also in Zobrest, Witters, and Mueller.5 The heart of our reasoning in Zobrest, upholding governmental provision of a sign- - - - - - - 5 JUSTICE O'CONNOR acknowledges that "neutrality is an important reason for upholding government-aid programs," one that our recent cases have "emphasized . . . repeatedly." Post, at 3 (opinion concurring in judgment). 12 MITCHELL v. HELMS Opinion of THOMAS, J. language interpreter to a deaf student at his Catholic high school, was as follows: "The service at issue in this case is part of a general government program that distributes benefits neu- trally to any child qualifying as `disabled' under the [statute], without regard to the `sectarian- nonsectarian, or public-nonpublic nature' of the school the child attends. By according parents freedom to select a school of their choice, the statute ensures that a government-paid interpreter will be present in a sectarian school only as a result of the private dec i- sion of individual parents. In other words, because the [statute] creates no financial incentive for parents to choose a sectarian school, an interpreter's presence there cannot be attributed to state decisionmaking." 509 U. S., at 10. As this passage indicates, the private choices helped to ensure neutrality, and neutrality and private choices together eliminated any possible attribution to the go v- ernment even when the interpreter translated classes on Catholic doctrine. Witters and Mueller employed similar reasoning. In Witters, we held that the Establishment Clause did not bar a State from including within a neutral program providing tuition payments for vocational rehabilitation a blind person studying at a Christian college to become a pastor, missionary, or youth director. We explained: "Any aid . . . that ultimately flows to religious instit u- tions does so only as a result of the genuinely ind e- pendent and private choices of aid recipients. Was h- ington's program is made available generally without regard to the sectarian-nonsectarian, or public- nonpublic nature of the institution benefited and . . . creates no financial incentive for students to unde r- take sectarian education. . . . [T]he fact that aid goes Cite as: 530 U. S. ____ (2000) 13 Opinion of THOMAS, J. to individuals means that the decision to support r e- ligious education is made by the individual, not by the State. . . . . . "[I]t does not seem appropriate to view any aid ult i- mately flowing to the Inland Empire School of the B i- ble as resulting from a state action sponsoring or sub- sidizing religion." 474 U. S., at 487­488 (footnote, citations, and internal quotation marks omi tted).6 Further, five Members of this Court, in separate opinions, emphasized both the importance of neutrality and of private choices, and the relationship between the two. See id., at 490­491 (Powell, J., joined by Burger, C. J., and REHNQUIST, J., concurring); id., at 493 (O'CONNOR, J., - - - - - - 6 The majority opinion also noted that only a small portion of the overall aid under the State's program would go to religious education, see Witters, 474 U. S., at 488, but it appears that five Members of the Court thought this point irrelevant. See id., at 491, n. 3 (Powell, J., joined by Burger, C. J., and REHNQUIST, J., concurring) (citing Mueller v. Allen, 463 U. S. 388, 401 (1983), to assert that validity of program "does not depend on the fact that petitioner appears to be the only handicapped student who has sought to use his assistance to pursue religious trai n- ing"); 474 U. S., at 490 (White, J., concurring) (agreeing with "most of JUSTICE POWELL's concurring opinion with respect to the relevance of Mueller," but not specifying further); id., at 493 (O'CONNOR, J., concurring in part and concurring in judgment) (agreeing with Justice Powell's reliance on Mueller and explaining that the program did not have an impermissible effect, because it was neutral and involved private choice, and thus "[n]o reasonable observer is likely to draw from the facts before us an inference that the State itself is endorsing a religious practice or belief"). More recently, in Agostini v. Felton, 521 U. S. 203 (1997), we held that the proportion of aid benefiting students at religious schools pursuant to a neutral program involving private choices was irrelevant to the constitutional inquiry. Id., at 229 (refusing "to conclude that the constitu- tionality of an aid program depends on the number of sectarian school students who happen to receive the otherwise neutral aid"); see also post, at 13 (O'CONNOR, J., concurring in judgment) (quoting this passage). 14 MITCHELL v. HELMS Opinion of THOMAS, J. concurring in part and concurring in judgment); see also id., at 490 (White, J., concurring). The tax deduction for educational expenses that we upheld in Mueller was, in these respects, the same as the tuition grant in Witters. We upheld it chiefly because it "neutrally provides state assistance to a broad spectrum of citizens," 463 U. S., at 398­399, and because "numerous, private choices of individual parents of school-age chi l- dren," id., at 399, determined which schools would benefit from the deductions. We explained that "[w]here, as here, aid to parochial schools is available only as a result of decisions of individual parents no `imprimatur of state approval' can be deemed to have been conferred on any particular religion, or on religion generally." Ibid. (cita- tion omitted); see id., at 397 (neutrality indicates lack of state imprimatur). Agostini's second primary criterion for determining the effect of governmental aid is closely related to the first. The second criterion requires a court to consider whether an aid program "define[s] its recipients by reference to religion." 521 U. S., at 234. As we briefly explained in Agostini, id., at 230­231, this second criterion looks to the same set of facts as does our focus, under the first crit e- rion, on neutrality, see id., at 225­226, but the second criterion uses those facts to answer a somewhat different question- whether the criteria for allocating the aid "creat[e] a financial incentive to undertake religious indo c- trination." Id., at 231. In Agostini we set out the follow- ing rule for answering this question: "This incentive is not present, however, where the aid is allocated on the basis of neutral, secular criteria that neither favor nor disfavor religion, and is made available to both religious and secular beneficiaries on a nondiscriminatory basis. Under such circu m- stances, the aid is less likely to have the effect of a d- Cite as: 530 U. S. ____ (2000) 15 Opinion of THOMAS, J. vancing religion." Ibid. The cases on which Agostini relied for this rule, and Agos- tini itself, make clear the close relationship between this rule, incentives, and private choice. For to say that a program does not create an incentive to choose religious schools is to say that the private choice is truly "indepen d- ent," Witters, 474 U. S., at 487. See Agostini, supra, at 232 (holding that Title I did not create any impermissible incentive, because its services were "available to all chi l- dren who meet the Act's eligibility requirements, no ma t- ter what their religious beliefs or where they go to school"); Zobrest, 509 U. S., at 10 (discussing, in successive sentences, neutrality, private choice, and financial ince n- tives, respectively); Witters, supra, at 488 (similar). When such an incentive does exist, there is a greater risk that one could attribute to the government any indoctrination by the religious schools. See Zobrest, supra, at 10. We hasten to add, what should be obvious from the rule itself, that simply because an aid program offers private schools, and thus religious schools, a benefit that they did not previously receive does not mean that the program, by reducing the cost of securing a religious education, creates, under Agostini's second criterion, an "incentive" for par- ents to choose such an education for their children. For any aid will have some such effect. S ee Allen, 392 U. S., at 244; Everson, 330 U. S., at 17; see also Mueller, 463 U. S., at 399. B Respondents inexplicably make no effort to address Chapter 2 under the Agostini test. Instead, dismissing Agostini as factually distinguishable, they offer two rules that they contend should govern our determination of whether Chapter 2 has the effect of advancing religion. They argue first, and chiefly, that "direct, nonincidental" aid to the primary educational mission of religious schools 16 MITCHELL v. HELMS Opinion of THOMAS, J. is always impermissible. Second, they argue that prov i- sion to religious schools of aid that is divertible to religious use is similarly impermissible.7 Respondents' arguments are inconsistent with our more recent case law, in pa r- ticular Agostini and Zobrest, and we therefore reject them. - - - - - - 7 Respondents also contend that Chapter 2 aid supplants, rather than supplements, the core educational function of parochial schools and therefore has the effect of furthering religion. Our case law does provide some indication that this distinction may be relevant to dete r- mining whether aid results in governmental indoctrination, see Agos- tini, 521 U. S., at 228­229; Zobrest v. Catalina Foothills School Dist., 509 U. S. 1, 12 (1993); but see School Dist. of Grand Rapids v. Ball, 473 U. S. 373, 396 (1985), but we have never delineated the distinction's contours or held that it is constitutionally required. Nor, to the extent that the supplement/supplant line is separable from respondents' direct/indirect and "no divertibility" arguments, do we need to resolve the distinction's constitutional status today, for, as we have already noted, Chapter 2 itself requires that aid may only be supplemental. 20 U. S. C. §7371(b). See also post, at 33 (O'CONNOR, J., concurring in judgment) (declining to decide whether suppl e- ment/supplant distinction is a constitutional requirement); but see post, at 17 (explaining that computers are "necessary" to "the educational process"). We presume that whether a parish has complied with that statutory requirement would be, at the very least, relevant to whether a violation of any constitutional supplement/supplant requirement has occurred, yet we have no reason to believe that there has been any material statutory violation. A statewide review by the Louisiana SEA indicated that §7371(b) receives nearly universal compliance. App. 112a. More importantly, neither the District Court nor the Fifth Circuit even hinted that Jefferson Parish had violated §7371(b), and respondents barely mention the statute in their brief to this Court, offering only the slimmest evidence of any possible violation, see id., at 63a. Respondents argue that any Chapter 2 aid that a school uses to comply with state requirements (such as those relating to computers and libraries) necessarily violates whatever supplement/supplant line may exist in the Constitution, but our decision in Committee for Public Ed. and Religious Liberty v. Regan, 444 U. S. 646 (1980), upholding reimbursement to parochial schools of costs relating to state-mandated testing, rejects any such blanket rule. Cite as: 530 U. S. ____ (2000) 17 Opinion of THOMAS, J. 1 Although some of our earlier cases, particularly Ball, 473 U. S., at 393­394, did emphasize the distinction b e- tween direct and indirect aid, the purpose of this distin c- tion was merely to prevent "subsidization" of religion, see id., at 394. As even the dissent all but admits, see post, at 22 (opinion of SOUTER, J.), our more recent cases address this purpose not through the direct/indirect distinction but rather through the principle of private choice, as incorp o- rated in the first Agostini criterion (i.e., whether any indoctrination could be attributed to the government). If aid to schools, even "direct aid," is neutrally available and, before reaching or benefiting any religious school, first passes through the hands (literally or figuratively) of numerous private citizens who are free to direct the aid elsewhere, the government has not provided any "support of religion," Witters, supra, at 489. See supra, at 10­11. Although the presence of private choice is easier to see when aid literally passes through the hands of individuals- which is why we have mentioned directness in the same breath with private choice, see, e.g., Agostini, supra, at 226; Witters, supra, at 487; Mueller, supra, at 399- there is no reason why the Establishment Clause requires such a form. Indeed, Agostini expressly rejected the absolute line that respondents would have us draw. We there explained that "we have departed from the rule relied on in Ball that all government aid that directly assists the educational function of religious schools is invalid." 521 U. S., at 225. Agostini relied primarily on Witters for this conclusion and made clear that private choice and neutrality would r e- solve the concerns formerly addressed by the rule in Ball. It was undeniable in Witters that the aid (tuition) would ultimately go to the Inland Empire School of the Bible and would support religious education. We viewed this a r- rangement, however, as no different from a government issuing a paycheck to one of its employees knowing that 18 MITCHELL v. HELMS Opinion of THOMAS, J. the employee would direct the funds to a religious instit u- tion. Both arrangements would be valid, for the same reason: "[A]ny money that ultimately went to religious institutions did so `only as a result of the genuinely ind e- pendent and private choices of' individuals." Agostini, supra, at 226 (quoting Witters, 474 U. S., at 487). In addi- tion, the program in Witters was neutral. 521 U. S., at 225 (quoting Witters, supra, at 487). As Agostini explained, the same reasoning was at work in Zobrest, where we allowed the government-funded interpreter to provide assistance at a Catholic school, "even though she would be a mouthpiece for religious instruction," because the interpreter was provided a c- cording to neutral eligibility criteria and private choice. 521 U. S., at 226. Therefore, the religious messages inte r- preted by the interpreter could not be attributed to the government, see ibid. (We saw no difference in Zobrest between the government hiring the interpreter directly and the government providing funds to the parents who then would hire the interpreter. 509 U. S., at 13, n. 11.) We rejected the dissent's objection that we had never before allowed "a public employee to participate directly in religious indoctrination." See id., at 18 (Blackmun, J., dissenting). Finally, in Agostini itself, we used the rea- soning of Witters and Zobrest to conclude that remedial classes provided under Title I of the ESEA by public e m- ployees did not impermissibly finance religious indoctrin a- tion. 521 U. S., at 228; see id., at 230­232. We found it insignificant that students did not have to directly apply for Title I services, that Title I instruction was provided to students in groups rather than individually, and that instruction was provided in the facilities of the private schools. Id., at 226­229. To the extent that respondents intend their di- rect/indirect distinction to require that any aid be literally placed in the hands of schoolchildren rather than given Cite as: 530 U. S. ____ (2000) 19 Opinion of THOMAS, J. directly to the school for teaching those same children, the very cases on which respondents most rely, Meek and Wolman, demonstrate the irrelevance of such formalism. In Meek, we justified our rejection of a program that loaned instructional materials and equipment by, among other things, pointing out that the aid was loaned to the schools, and thus was "direct aid." 421 U. S., at 362­363. The materials-and-equipment program in Wolman was essentially identical, except that the State, in an effort to comply with Meek, see Wolman, 433 U. S., at 233, 250, loaned the aid to the students. (The revised program operated much like the one we upheld in Allen. Compare Wolman, supra, at 248, with Allen, 392 U. S., at 243­245.) Yet we dismissed as "technical" the difference between the two programs: "[I]t would exalt form over substance if this distinction were found to justify a result different from that in Meek." 433 U. S., at 250. Wolman thus, although purporting to reaffirm Meek, actually undermined that decision, as is evident from the similarity between the reasoning of Wolman and that of the Meek dissent. Com- pare Wolman, supra, at 250 (The "technical change in legal bailee" was irrelevant), with Meek, supra, at 391 (REHNQUIST, J., concurring in judgment in part and di s- senting in part) ("Nor can the fact that the school is the bailee be regarded as constitutionally determinative"). That Meek and Wolman reached the same result, on pro- grams that were indistinguishable but for the d i- rect/indirect distinction, shows that that distinction played no part in Meek. Further, respondents' formalistic line breaks down in the application to real-world programs. In Allen, for ex- ample, although we did recognize that students the m- selves received and owned the textbooks, we also noted that the books provided were those that the private schools required for courses, that the schools could collect students' requests for books and submit them to the board 20 MITCHELL v. HELMS Opinion of THOMAS, J. of education, that the schools could store the textbooks, and that the textbooks were essential to the schools' teaching of secular subjects. See 392 U. S., at 243­245. Whether one chooses to label this program "direct" or "indirect" is a rather arbitrary choice, one that does not further the constitutional analysis. Of course, we have seen "special Establishment Clause dangers," Rosenberger, 515 U. S., at 842, when money is given to religious schools or entities directly rather than, as in Witters and Mueller, indirectly. See 515 U. S., at 842 (collecting cases); id., at 846­847 (O'CONNOR, J., concur- ring); see also Bowen v. Kendrick, 487 U. S. 589, 608­609 (1988); compare Committee for Public Ed. and Religious Liberty v. Regan, 444 U. S. 646 (1980), with Levitt v. Com- mittee for Public Ed. & Religious Liberty, 413 U. S. 472 (1973).8 But direct payments of money are not at issue in - - - - - - 8 The reason for such concern is not that the form per se is bad, but that such a form creates special risks that governmental aid will have the effect of advancing religion (or, even more, a purpose of doing so). An indirect form of payment reduces these risks. See Mueller, 463 U. S., at 399 (neutral tax deduction, because of its indirect form, allowed economic benefit to religious schools only as result of private choice and thus did not suggest state sanction of schools' religious messages). It is arguable, however, at least after Witters, that the principles of neutrality and private choice would be adequate to address those special risks, for it is hard to see the basis for deciding Witters differently simply if the State had sent the tuition check directly to whichever school Witters chose to attend. See Rosenberger v. Rector and Visitors of Univ. of Va., 515 U. S. 819, 848 (1995) (O'CONNOR, J., concurring) (explaining Witters as recon- ciling principle of neutrality with principle against public funding of religious messages by relying