with whom The Chief Justice, Justice White, and Justice Scalia join, concurring in the judgment in part and dissenting in part.
The majority holds that the County of Allegheny violated the Establishment Clause by displaying a creche in the county courthouse, because the “principal or primary effect” of the display is to advance religion within the meaning of Lemon v. Kurtzman, 403 U. S. 602, 612-613 (1971). This view of the Establishment Clause reflects an unjustified hostility toward religion, a hostility inconsistent with our history and our precedents, and I dissent from this holding. The creche display is constitutional, and, for the same reasons, the display of a menorah by the city of Pittsburgh is permissible as well. On this latter point, I concur in the result, but not the reasoning, of Part VI of Justice Blackmun’s opinion.
I
In keeping with the usual fashion of recent years, the majority applies the Lemon test to judge the constitutionality of the holiday displays here in question. I am content for present purposes to remain within the Lemon framework, but do not wish to be seen as advocating, let alone adopting, that test as our primary guide in this difficult area. Persuasive criticism of Lemon has emerged. See Edwards v. Aguillard, 482 U. S. 578, 636-640 (1987) (Scalia, J., dissenting); *656Aguilar v. Felton, 473 U. S. 402, 426-430 (1985) (O’CONNOR, J., dissenting); Wallace v. Jaffree, 472 U. S. 38, 108-113 (1985) (Rehnquist, J., dissenting); Roemer v. Maryland Bd. of Public Works, 426 U. S. 736, 768-769 (1976) (WHITE, J., concurring in judgment). Our cases often question its utility in providing concrete answers to Establishment Clause questions, calling it but a “‘helpful signpos[t]’” or “‘guidelin[e]’” to assist our deliberations rather than a comprehensive test. Mueller v. Allen, 463 U. S. 388, 394 (1983) (quoting Hunt v. McNair, 413 U. S. 734, 741 (1973)); Committee for Public Education & Religious Liberty v. Nyquist, 413 U. S. 756, 773, n. 31 (1973) (quoting Tilton v. Richardson, 403 U. S. 672, 677-678 (1971)); see Lynch v. Donnelly, 465 U. S. 668, 679 (1984) (“[W]e have repeatedly emphasized our unwillingness to be confined to any single test or criterion in this sensitive area”). Substantial revision of our Establishment Clause doctrine may be in order; but it is unnecessary to undertake that task today, for even the Lemon test, when applied with proper sensitivity to our traditions and our case law, supports the conclusion that both the creche and the menorah are permissible displays in the context of the holiday season.
The only Lemon factor implicated in these cases directs us to inquire whether the “principal or primary effect” of the challenged government practice is “one that neither advances nor inhibits religion.” 403 U. S., at 612. The requirement of neutrality inherent in that formulation has sometimes been stated in categorical terms. For example, in Everson v. Board of Education of Ewing, 330 U. S. 1 (1947), the first case in our modern Establishment Clause jurisprudence, Justice Black wrote that the Clause forbids laws “which aid one religion, aid all religions, or prefer one religion over another.” Id., at 15-16. We have stated that government “must be neutral in matters of religious theory, doctrine, and practice” and “may not aid, foster, or promote one religion or religious theory against another or even against the *657militant opposite.” Epperson v. Arkansas, 393 U. S. 97, 103-104 (1968). And we have spoken of a prohibition against conferring an “‘imprimatur of state approval’” on religion, Mueller v. Allen, supra, at 399 (quoting Widmar v. Vincent, 454 U. S. 263, 274 (1981)), or “favor[ing] the adherents of any sect or religious organization,” Gillette v. United States, 401 U. S. 437, 450 (1971).
These statements must not give the impression of a formalism that does not exist. Taken to its logical extreme, some of the language quoted above would require a relentless extirpation of all contact between government and religion. But that is not the history or the purpose of the Establishment Clause. Government policies of accommodation, acknowledgment, and support for religion are an accepted part of our political and cultural heritage. As Chief Justice Burger wrote for the Court in Walz v. Tax Comm’n of New York City, 397 U. S. 664 (1970), we must be careful to avoid “[t]he hazards of placing too much weight on a few words or phrases of the Court,” and so we have “declined to construe the Religion Clauses with a literalness that would undermine the ultimate constitutional objective as illuminated by history.” Id., at 670-671.
Rather than requiring government to avoid any action that acknowledges or aids religion, the Establishment Clause permits government some latitude in recognizing and accommodating the central role religion plays in our society. Lynch v. Donnelly, supra, at 678; Walz v. Tax Comm’n of New York City, supra, at 669. Any approach less sensitive to our heritage would border on latent hostility toward religion, as it would require government in all its multifaceted roles to acknowledge only the secular, to the exclusion and so to the detriment of the religious. A categorical approach would install federal courts as jealous guardians of an absolute “wall of separation,” sending a clear message of disapproval. In this century, as the modern administrative state expands to touch the lives of its citizens in such diverse ways and redi*658rects their financial choices through programs of its own, it is difficult to maintain the fiction that requiring government to avoid all assistance to religion can in fairness be viewed as serving the goal of neutrality.
Our cases reflect this understanding. In Zorach v. Clauson, 343 U. S. 306 (1952), for example, we permitted New York City’s public school system to accommodate the religious preferences of its students by giving them the option of staying in school or leaving to attend religious classes for part of the day. Justice Douglas wrote for the Court:
“When the state encourages religious instruction . . . it follows the best of our traditions. For it then respects the religious nature of our people and accommodates the public service to their spiritual needs. To hold that it may not would be to find in the Constitution a requirement that the government show a callous indifference to religious groups. That would be preferring those who believe in no religion over those who do believe.” Id., at 313-314.
Nothing in the First Amendment compelled New York City to establish the release-time policy in Zorach, but the fact that the policy served to aid religion, and in particular those sects that offer religious education to the young, did not invalidate the accommodation. Likewise, we have upheld government programs supplying textbooks to students in parochial schools, Board of Education of Central School Dist. No. 1 v. Allen, 392 U. S. 236 (1968), providing grants to church-sponsored universities and colleges, Roemer v. Maryland Bd. of Public Works, supra; Tilton v. Richardson, supra, and exempting churches from the obligation to pay taxes, Walz v. Tax Comm’n of New York City, supra. These programs all have the effect of providing substantial benefits to particular religions, see, e. g., Tilton, supra, at 679 (grants to church-sponsored educational institutions “surely aid” those institutions), but they are nonetheless permissible. See Lynch v. Donnelly, supra; McGowan v. *659Maryland, 366 U. S. 420, 445 (1961); Illinois ex rel. McCollum v. Board of Education of School Dist. No. 71, Champaign County, 333 U. S. 203, 211-212 (1948). As Justice Goldberg wrote in Abington School District v. Schempp, 374 U. S. 203 (1963):
“It is said, and I agree, that the attitude of government toward religion must be one of neutrality. But untutored devotion to the concept of neutrality can lead to invocation or approval of results which partake not simply of that noninterference and noninvolvement with the religious which the Constitution commands, but of a brooding and pervasive devotion to the secular and a passive, or even active, hostility to the religious. Such results are not only not compelled by the Constitution, but, it seems to me, are prohibited by it.
Neither government nor this Court can or should ignore the significance of the fact that a vast portion of our people believe in and worship God and that many of our legal, political and personal values derive historically from religious teachings. Government must inevitably take cognizance of the existence of religion . . . .” Id., at 306 (concurring opinion, joined by Harlan, J.).
The ability of the organized community to recognize and accommodate religion in a society with a pervasive public sector requires diligent observance of the border between accommodation and establishment. Our cases disclose two limiting principles: government may not coerce anyone to support or participate in any religion or its exercise; and it may not, in the guise of avoiding hostility or callous indifference, give direct benefits to religion in such a degree that it in fact “establishes a [state] religion or religious faith, or tends to do so.” Lynch v. Donnelly, 465 U. S., at 678. These two principles, while distinct, are not unrelated, for it would be difficult indeed to establish a religion without some measure of more or less subtle coercion, be it in the form of taxation to supply the substantial benefits that would sustain *660a state-established faith, direct compulsion to observance, or governmental exhortation to religiosity that amounts in fact to proselytizing.
It is no surprise that without exception we have invalidated actions that further the interests of religion through the coercive power of government. Forbidden involvements include compelling or coercing participation or attendance at a religious activity, see Engel v. Vitale, 370 U. S. 421 (1962); McGowan v. Maryland, supra, at 452 (discussing McCollum v. Board of Education of School Dist. No. 71, Champaign County, supra), requiring religious oaths to obtain government office or benefits, Torcaso v. Watkins, 367 U. S. 488 (1961), or delegating government power to religious groups, Larkin v. Grendel's Den, Inc., 459 U. S. 116 (1982). The freedom to worship as one pleases without government interference or oppression is the great object of both the Establishment and the Free Exercise Clauses. Barring all attempts to aid religion through government coercion goes far toward attainment of this object. See McGowan v. Maryland, supra, at 441, quoting 1 Annals of Congress 730 (1789) (James Madison, who proposed the First Amendment in Congress, “ ‘apprehended the meaning of the [Religion Clauses] to be, that Congress should not establish a religion, and enforce the legal observation of it by law, nor compel men to worship God in any manner contrary to their conscience’”); Cantwell v. Connecticut, 310 U. S. 296, 303 (1940) (the Religion Clauses “forestall] compulsion by law of the acceptance of any creed or the practice of any form of worship”).
As Justice Blackmun observes, ante, at 597-598, n. 47, some of our recent cases reject the view that coercion is the sole touchstone of an Establishment Clause violation. See Engel v. Vitale, supra, at 430 (dictum) (rejecting, without citation of authority, proposition that coercion is required to demonstrate an Establishment Clause violation); Abington School District v. Schempp, supra, at 223; Nyquist, 413 U. S., at 786. That may be true if by “coercion” is meant *661direct coercion in the classic sense of an establishment of religion that the Framers knew. But coercion need not be a direct tax in aid of religion or a test oath. Symbolic recognition or accommodation of religious faith may violate the Clause in an extreme case.1 I doubt not, for example, that the Clause forbids a city to permit the permanent erection of a large Latin cross on the roof of city hall. This is not because government speech about religion is per se suspect, as the majority would have it, but because such an obtrusive year-round religious display would place the government’s weight behind an obvious effort to proselytize on behalf of a particular religion. Cf. Friedman v. Board of County Comm’rs of Bernalillo County, 781 F. 2d 777 (CA10 1985) (en banc) (Latin cross on official county seal); American Civil Liberties Union of Georgia v. Rabun County Chamber of Commerce, Inc., 698 F. 2d 1098 (CA11 1983) (cross erected in public park); Lowe v. Eugene, 254 Ore. 518, 463 P. 2d 360 (1969) (same). Speech may coerce in some circumstances, but this does not justify a ban on all government recognition of religion. As Chief Justice Burger wrote for the Court in Walz:
“The general principle deducible from the First Amendment and all that has been said by the Court is this: that we will not tolerate either governmentally established religion or governmental interference with religion. Short of those expressly proscribed governmental acts there is room for play in the joints productive of a benevolent neutrality which will permit religious exercise to exist *662without sponsorship and without interference.” 397 U. S., at 669.
This is most evident where the government’s act of recognition or accommodation is passive and symbolic, for in that instance any intangible benefit to religion is unlikely to present a realistic risk of establishment. Absent coercion, the risk of infringement of religious liberty by passive or symbolic accommodation is minimal. Our cases reflect this reality by requiring a showing that the symbolic recognition or accommodation advances religion to such a degree that it actually “establishes a religion or religious faith, or tends to do so.” Lynch, 465 U. S., at 678.
In determining whether there exists an establishment, or a tendency toward one, we refer to the other types of church-state contacts that have existed unchallenged throughout our history, or that have been found permissible in our case law. In Lynch, for example, we upheld the city of Pawtucket’s holiday display of a creche, despite the fact that “the display advance[d] religion in a sense.” Id., at 683. We held that the creche conferred no greater benefit on religion than did governmental support for religious education, legislative chaplains, “recognition of the origins of the [Christmas] Holiday itself as ‘Christ’s Mass,’” or many other forms of symbolic or tangible governmental assistance to religious faiths that are ensconced in the safety of national tradition. Id., at 681, 683. And in Marsh v. Chambers, we found that Nebraska’s practice of employing a legislative chaplain did not violate the Establishment Clause, because “legislative prayer presents no more potential for establishment than the provision of school transportation, beneficial grants for higher education, or tax exemptions for religious organizations.” 463 U. S., at 791 (citations omitted). Noncoercive government action within the realm of flexible accommodation or passive acknowledgment of existing symbols does not violate the Establishment Clause unless it benefits religion in a way *663more direct and more substantial than practices that are accepted in our national heritage.
rH h-H
These principles are not difficult to apply to the facts of the cases before us. In permitting the displays on government property of the menorah and the creche, the city and county sought to do no more than “celebrate the season,” Brief for Petitioner County of Allegheny in No. 87-2050, p. 27, and to acknowledge, along with many of their citizens, the historical background and the religious, as well as secular, nature of the Chanukah and Christmas holidays. This interest falls well within the tradition of government accommodation and acknowledgment of religion that has marked our history from the beginning.2 It cannot be disputed that government, if it chooses, may participate in sharing with its citizens the joy of the holiday season, by declaring public holidays, installing or permitting festive displays, sponsoring celebrations and parades, and providing holiday vacations for its employees. All levels of our government do precisely that. As we said in Lynch, “Government has long recognized — indeed it has subsidized — holidays with religious significance.” 465 U. S., at 676.
If government is to participate in its citizens’ celebration of a holiday that contains both a secular and a religious component, enforced recognition of only the secular aspect would *664signify the callous indifference toward religious faith that our cases and traditions do not require; for by commemorating the holiday only as it is celebrated by nonadherents, the government would be refusing to acknowledge the plain fact, and the historical reality, that many of its citizens celebrate its religious aspects as well. Judicial invalidation of government’s attempts to recognize the religious underpinnings of the holiday would signal not neutrality but a pervasive intent to insulate government from all things religious. The Religion Clauses do not require government to acknowledge these holidays or their religious component; but our strong tradition of government accommodation and acknowledgment permits government to do so. See Lynch v. Donnelly, supra; cf. Zorach v. Clauson, 343 U. S., at 314; Abington School District v. Schempp, 374 U. S., at 306 (Goldberg, J., concurring).
There is no suggestion here that the government’s power to coerce has been used to further the interests of Christianity or Judaism in any way. No one was compelled to observe or participate in any religious ceremony or activity. Neither the city nor the county contributed significant amounts of tax money to serve the cause of one religious faith. The creche and the menorah are purely passive symbols of religious holidays. Passersby who disagree with the message conveyed by these displays are free to ignore them, or even to turn their backs, just as they are free to do when they disagree with any other form of government speech.
There is no realistic risk that the creche and the menorah represent an effort to proselytize or are otherwise the first step down the road to an establishment of religion.3 Lynch *665is dispositive of this claim with respect to the creche, and I find no reason for reaching a different result with respect to the menorah. Both are the traditional symbols of religious holidays that over time have acquired a secular component. Ante, at 579, and n. 3, 585, and n. 29. Without ambiguity, Lynch instructs that “the focus of our inquiry must be on the [religious symbol] in the context of the [holiday] season,” 465 U. S., at 679. In that context, religious displays that serve “to celebrate the Holiday and to depict the origins of that Holiday” give rise to no Establishment Clause concern. Id., at 681. If Congress and the state legislatures do not run afoul of the Establishment Clause when they begin each day with a state-sponsored prayer for divine guidance offered by a chaplain whose salary is paid at government expense, I cannot comprehend how a menorah or a creche, displayed in the limited context of the holiday season, can be invalid.4
Respondents say that the religious displays involved here are distinguishable from the creche in Lynch because they are located on government property and are not surrounded *666by the candy canes, reindeer, and other holiday paraphernalia that were a part of the display in Lynch. Nothing in Chief Justice Burger’s opinion for the Court in Lynch provides support for these purported distinctions. After describing the facts, the Lynch opinion makes no mention of either of these factors. It concentrates instead on the significance of the creche as part of the entire holiday season. Indeed, it is clear that the Court did not view the secular aspects of the display as somehow subduing the religious message conveyed by the creche, for the majority expressly rejected the dissenters’ suggestion that it sought “‘to explain away the clear religious import of the creche’” or had “equated the creche with a Santa’s house or reindeer.” Id., at 685, n. 12. Crucial to the Court’s conclusion was not the number, prominence, or type of secular items contained in the holiday display but the simple fact that, when displayed by government during the Christmas season, a creche presents no realistic danger of moving government down the forbidden road toward an establishment of religion. Whether the creche be surrounded by poinsettias, talking wishing wells, or carolers, the conclusion remains the same, for the relevant context is not the items in the display itself but the season as a whole.
The fact that the creche and menorah are both located on government property, even at the very seat of government, is likewise inconsequential. In the first place, the Lynch Court did not rely on the fact that the setting for Pawtucket’s display was a privately owned park, and it is difficult to suggest that anyone could have failed to receive a message oí government sponsorship after observing Santa Claus ride the city fire engine to the park to join with the mayor of Paw-tucket in inaugurating the holiday season by turning on the lights of the city-owned display. See Donnelly v. Lynch, 525 F. Supp. 1150, 1156 (RI 1981). Indeed, the District Court,in Lynch found that “people might reasonably mistake *667the Park for public property,” and rejected as “frivolous” the suggestion that the display was not directly associated with the city. Id., at 1176, and n. 35.
Our cases do not suggest, moreover, that the use of public property necessarily converts otherwise permissible government conduct into an Establishment Clause violation. To the contrary, in some circumstances the First Amendment may require that government property be available for use by religious groups, see Widmar v. Vincent, 454 U. S. 263 (1981); Fowler v. Rhode Island, 345 U. S. 67 (1953); Niemotko v. Maryland, 340 U. S. 268 (1951), and even where not required, such use has long been permitted. The prayer approved in Marsh v. Chambers, for example, was conducted .in the legislative chamber of the State of Nebraska, surely the single place most likely to be thought the center of state authority.
Nor can I comprehend why it should be that placement of a government-owned creche on private land is lawful while placement of a privately owned creche on public land is not.5 If anything, I should have thought government ownership of a religious symbol presented the more difficult question under the Establishment Clause, but as Lynch resolved that question to sustain the government action, the sponsorship here ought to be all the easier to sustain. In short, nothing about the religious displays here distinguishes them in any meaningful way from the creche we permitted in Lynch.
If Lynch is still good law — and until today it was — the judgment below cannot stand. I accept and indeed approve both the holding and the reasoning of Chief Justice Burger’s opinion in Lynch, and so I must dissent from the judgment that the creche display is unconstitutional. On the same reasoning, I agree that the menorah display is constitutional.
*668J-H HH 1 — I
The majority invalidates display of the creche, not because it disagrees with the interpretation of Lynch applied above, but because it chooses to discard the reasoning of the Lynch majority opinion in favor of Justice O’Connor’s concurring opinion in that case. See ante, at 594-597. It has never been my understanding that a concurring opinion “suggest[ing] a clarification of our . . . doctrine,” Lynch, 465 U. S., at 687 (O’Connor, J., concurring), could take precedence over an opinion joined in its entirety by five Members of the Court.6 As a general rule, the principle of stare decisis directs us to adhere not only to the holdings of our prior cases, but also to their explications of the governing rules of law. Since the majority does not state its intent to overrule Lynch, I find its refusal to apply the reasoning of that decision quite confusing.
Even if Lynch did not control, I would not commit this Court to the test applied by the majority today. The notion that cases arising under the Establishment Clause should be decided by an inquiry into whether a “ ‘reasonable observer’ ” may “‘fairly understand’” government action to “‘sen[d] a message to nonadherents that they are outsiders, not full members of the political community,”’ is a recent, and in my view most unwelcome, addition to our tangled Establishment Clause jurisprudence. Ante, at 595, 620. Although a scattering of our cases have used “endorsement” as another word for “preference” or “imprimatur, ” the endorsement test applied by the majority had its genesis in Justice O’Connor’s concurring opinion in Lynch. See also Corporation of the Presiding Bishop of Church of Jesus Christ of Latterday Saints v. Amos, 483 U. S. 327, 346 (1987) (O’Connor, J., concurring in judgment); Estate of Thornton v. Caldor, Inc., 472 U. S. 703, 711 (1985) (O’Connor, J., concurring); Wal*669lace v. Jaffree, 472 U. S., at 67 (O’Connor, J., concurring in judgment). The endorsement test has been criticized by some scholars in the field, see, e. g., Smith, Symbols, Perceptions, and Doctrinal Illusions: Establishment Neutrality and the “No Endorsement” Test, 86 Mich. L. Rev. 266 (1987); Tushnet, The Constitution of Religion, 18 Conn. Law Rev. 701, 711-712 (1986). Only one opinion for the Court has purported to apply it in full, see School Dist. of Grand Rapids v. Ball, 473 U. S. 373, 389-392 (1985), but the majority’s opinion in these cases suggests that this novel theory is fast becoming a permanent accretion to the law. See also Texas Monthly, Inc. v. Bullock, 489 U. S. 1, 8-9 (1989) (opinion of Brennan, J.). For the reasons expressed below, I submit that the endorsement test is flawed in its fundamentals and unworkable in practice. The uncritical adoption of this standard is every bit as troubling as the bizarre result it produces in the cases before us.
A
I take it as settled law that, whatever standard the Court applies to Establishment Clause claims, it must at least suggest results consistent with our precedents and the historical practices that, by tradition, have informed our First Amendment jurisprudence. See supra, at 655-663; Lynch, supra, at 673-674; Marsh v. Chambers, 463 U. S., at 790-791; Walz v. Tax Comm’n of New York City, 397 U. S., at 671. It is true that, for reasons quite unrelated to the First Amendment, displays commemorating religious holidays were not commonplace in 1791. See generally J. Barnett, The American Christmas: A Study in National Culture 2-11 (1954). But the relevance of history is not confined to the inquiry into whether the challenged practice itself is a part of our accepted traditions dating back to the Founding.
Our decision in Marsh v. Chambers illustrates this proposition. The dissent in that case sought to characterize the decision as “carving out an exception to the Establishment *670Clause rather than reshaping Establishment Clause doctrine to accommodate legislative prayer,” 463 U. S., at 796 (Brennan, J., dissenting), but the majority rejected the suggestion that “historical patterns ca[n] justify contemporary violations of constitutional guarantees,” id., at 790. Marsh stands for the proposition, not that specific practices common in 1791 are an exception to the otherwise broad sweep of the Establishment Clause, but rather that the meaning of the Clause is to be determined by reference to historical practices and understandings.7 Whatever test we choose to apply must permit not only legitimate practices two centuries old but also any other practices with no greater potential for an establishment of religion. See Committee for Public Education and Religious Liberty v. Nyquist, 413 U. S., at 808 (Rehnquist, J., dissenting in part). The First Amendment is a rule, not a digest or compendium. A test for implementing the protections of the Establishment Clause that, if applied with consistency, would invalidate longstanding traditions cannot be a proper reading of the Clause.
If the endorsement test, applied without artificial exceptions for historical practice, reached results consistent with history, my objections to it would have less force. But, as I understand that test, the touchstone of an Establishment Clause violation is whether nonadherents would be made to feel like “outsiders” by government recognition or accommodation of religion. Few of our traditional practices recognizing the part religion plays in our society can withstand scrutiny under a faithful application of this formula.
*671Some examples suffice to make plain my concerns. Since the Founding of our Republic, American Presidents have issued Thanksgiving Proclamations establishing a national day of celebration and prayer. The first such proclamation was issued by President Washington at the request of the First Congress, and “recommend[ed] and assign[ed]” a day “to be devoted by the people of these States to the service of that great and glorious Being who is the beneficent author of all the good that was, that is, or that will be,” so that “we may then unite in most humbly offering our prayers and supplications to the great Lord and Ruler of Nations, and beseech Him to . . . promote the knowledge and practice of true religion and virtue . . . .” 1 J. Richardson, A Compilation of Messages and Papers of the Presidents, 1789-1897, p. 64 (1899). Most of President Washington’s successors have followed suit,8 and the forthrightly religious nature of these proclamations has not waned with the years. President Franklin D. Roosevelt went so far as to “suggest a nationwide reading of the Holy Scriptures during the period from Thanksgiving Day to Christmas” so that “we may bear more earnest witness to our gratitude to Almighty God.” Presidential Proclamation No. 2629, 58 Stat. 1160. It requires little imagination to conclude that these proclamations would cause nonadherents to feel excluded, yet they have been a part of our national heritage from the beginning.9
*672The Executive has not been the only Branch of our Government to recognize the central role of religion in our society. The fact that this Court opens its sessions with the request that “God save the United States and this honorable Court” has been noted elsewhere. See Lynch, 465 U. S., at 677. The Legislature has gone much further, not only employing legislative chaplains, see 2 U. S. C. §61d, but also setting aside a special prayer room in the Capitol for use by Members of the House and Senate. The room is decorated with a large stained glass panel that depicts President Washington kneeling in prayer; around him is etched the first verse of the 16th Psalm: “Preserve me, 0 God, for in Thee do I put my trust.” Beneath the panel is a rostrum on which a Bible is placed; next to the rostrum is an American Flag. See L. Aikman, We the People: The Story of the United States Capitol 122 (1978). Some endorsement is inherent in these reasonable accommodations, yet the Establishment Clause does not forbid them.
The United States Code itself contains religious references that would be suspect under the endorsement test. Congress has directed the President to “set aside and proclaim a suitable day each year ... as a National Day of Prayer, on which the people of the United States may turn to God in prayer and meditation at churches, in groups, and as individuals.” 36 U. S. C. § 169h. This statute does not require anyone to pray, of course, but it is a straightforward endorsement of the concept of “turn[ing] to God in prayer.” Also by statute, the Pledge of Allegiance to the Flag describes the United States as “one Nation under God.” 36 U. S. C. § 172. *673To be sure, no one is obligated to recite this phrase, see West Virginia State Board of Education v. Barnette, 319 U. S. 624 (1943), but it borders on sophistry to suggest that the “‘reasonable’ ” atheist would not feel less than a “ ‘full membe[r] of the political community’ ” every time his fellow Americans recited, as part of their expression of patriotism and love for country, a phrase he believed to be false. Likewise, our national motto, “In God we trust,” 36 U. S. C. §186, which is prominently engraved in the wall above the Speaker’s dias in the Chamber of the House of Representatives and is reproduced on every coin minted and every dollar printed by the Federal Government, 31 U. S. C. §§ 5112(d)(1), 5114(b), must have the same effect.
If the intent of the Establishment Clause is to protect individuals from mere feelings of exclusion, then legislative prayer cannot escape invalidation. It has been argued that “[these] government acknowledgments of religion serve, in the only ways reasonably possible in our culture, the legitimate secular purposes of solemnizing public occasions, expressing confidence in the future, and encouraging the recognition of what is worthy of appreciation in society.” Lynch-, supra, at 693 (O’Connor, J., concurring). I fail to see why prayer is the only way to convey these messages; appeals to patriotism, moments of silence, and any number of other approaches would be as effective, were the only purposes at issue the ones described by the Lynch concurrence. Nor is it clear to me why “encouraging the recognition of what is worthy of appreciation in society” can be characterized as a purely secular purpose, if it can be achieved only through religious prayer. No doubt prayer is “worthy of appreciation,” but that is most assuredly not because it is secular. Even accepting the secular-solemnization explanation at face value, moreover, it seems incredible to suggest that the average observer of legislative prayer who either believes in no religion or whose faith rejects the concept of God would not receive the clear message that his faith is out of step with the *674political norm. Either the endorsement test must invalidate scores of traditional practices recognizing the place religion holds in our culture, or it must be twisted and stretched to avoid inconsistency with practices we know to have been permitted in the past, while condemning similar practices with no greater endorsement effect simply by reason of their lack of historical antecedent.10 Neither result is acceptable.
B
In addition to disregarding precedent and historical fact, the majority’s approach to government use of religious symbolism threatens to trivialize constitutional adjudication. By mischaracterizing the Court’s opinion in Lynch as an endorsement-in-context test, ante, at 597, Justice Black-mun embraces a jurisprudence of minutiae. A reviewing court must consider whether the city has included Santas, talking wishing wells, reindeer, or other secular symbols as “a center of attention separate from the creche.” Ante, at 598. After determining whether these centers of attention are sufficiently “separate” that each “had their specific visual story to tell,” the court must then measure their proximity to the créche. Ante, at 598, and n. 48. A community that wishes to construct a constitutional display must also *675take care to avoid floral frames or other devices that might insulate the creche from the sanitizing effect of the secular portions of the display. Ibid. The majority also notes the presence of evergreens near the creche that are identical to two small evergreens placed near official county signs. Ante, at 600, n. 50. After today’s decision, municipal greenery must be used with care.
Another important factor will be the prominence of the setting in which the display is placed. In this case, the Grand Staircase of the county courthouse proved too resplendent. Indeed, the Court finds that this location itself conveyed an “unmistakable message that [the county] supports and promotes the Christian praise to God that is the creche’s religious message.” Ante, at 600.
My description of the majority’s test, though perhaps uncharitable, is intended to illustrate the inevitable difficulties with its application.11 This test could provide workable guidance to the lower courts, if ever, only after this Court has decided a long series of holiday display cases, using little more than intuition and a tape measure. Deciding cases on *676the basis of such an unguided examination of marginalia is irreconcilable with the imperative of applying neutral principles in constitutional adjudication. “It would be appalling to conduct litigation.under the Establishment Clause as if it were a trademark case, with experts testifying about whether one display is really like another, and witnesses testifying they were offended — but would have been less so were the creche five feet closer to the jumbo candy cane.” American Jewish Congress v. Chicago, 827 F. 2d 120, 130 (CA7 1987) (Easterbrook, J., dissenting).
Justice Blackmun employs in many respects a similar analysis with respect to the menorah, principally discussing its proximity to the Christmas tree and whether “it is . . . more sensible to interpret the menorah in light of the tree, rather than vice versa.” Ante, at 617; see also ante, at 635 (O’Connor, J., concurring in part and concurring in judgment) (concluding that combination of tree, menorah, and salute to liberty conveys no message of endorsement to reasonable observers). Justice Blackmun goes further, however, and in upholding the menorah as an acknowledgment of a holiday with secular aspects emphasizes the city’s lack of “reasonable alternatives that are less religious in nature.” Ante, at 618; see ibid, (noting absence of a “more secular alternative symbol”). This least-religious-means test presents several difficulties.12 First, it creates an internal inconsistency in J ustice Blackmun’s opinion. J ustice Blackmun earlier suggests that the display of a creche is sometimes constitutional. Ante, at 598. But it is obvious that there are innumerable secular symbols of Christmas, and that there will always be a more secular alternative available in place of a creche. Second, the test as applied by Justice Blackmun is unworkable, for it requires not only that the Court engage in the unfamiliar task of deciding whether a particular alterna*677tive symbol is more or less religious, but also whether the alternative would “look out of place.” Ante, at 618. Third, although Justice Blackmun purports not to be overruling Lynch, the more-secular-alternative test contradicts that decision, as it comes not from the Court’s opinion, nor even from the concurrence, but from the dissent. See 465 U. S., at 699 (Brennan, J., dissenting). The Court in Lynch noted that the dissent “argues that the city’s objectives could have been achieved without including the creche in the display.” Id., at 681, n. 7. “True or false,” we said, “that is irrelevant.”
The result the Court reaches in these cases is perhaps the clearest illustration of the unwisdom of the endorsement test. Although Justice O’Connor disavows Justice Black-mun’s suggestion that the minority or majority status of a religion is relevant to the question whether government recognition constitutes a forbidden endorsement, ante, at 634 (O’Connor, J., concurring in part and concurring in judgment), the very nature of the endorsement test, with its emphasis on the feelings of the objective observer, easily lends itself to this type of inquiry. If there be such a person as the “reasonable observer,” I am quite certain that he or she will take away a salient message from our holding in these cases: the Supreme Court of the United States has concluded that the First Amendment creates classes of religions based on the relative numbers of their adherents. Those religions enjoying the largest following must be consigned to the status of least favored faiths so as to avoid any possible risk of offending members of minority religions. I would be the first to admit that many questions arising under the Establishment Clause do not admit of easy answers, but whatever the Clause requires, it is not the result reached by the Court today.
IV
The approach adopted by the majority contradicts important values embodied in the Clause. Obsessive, implacable resistance to all but the most carefully scripted and secu*678larized forms of accommodation requires this Court to act as a censor, issuing national decrees as to what is orthodox and what is not. What is orthodox, in this context, means what is secular; the only Christmas the State can acknowledge is one in which references to religion have been held to a minimum. The Court thus lends its assistance to an Orwellian rewriting of history as many understand it. I can conceive of no judicial function more antithetical to the First Amendment.
A further contradiction arises from the majority’s approach, for thd Court also assumes the difficult and inappropriate task of saying what every religious symbol means. Before studying these cases, I had not known the full history of the menorah, and I suspect the same was true of my colleagues. More important, this history was, and is, likely unknown to the vast majority of people of all faiths who saw the symbol displayed in Pittsburgh. Even if the majority is quite right about the history of the menorah, it hardly follows that this same history informed the observers’ view of the symbol and the reason for its presence. This Court is ill equipped to sit as a national theology board, and I question both the wisdom and the constitutionality of its doing so. Indeed, were I required to choose between the approach taken by the majority and a strict separationist view, I would have to respect the consistency of the latter.
The suit before us is admittedly a troubling one. It must be conceded that, however neutral the purpose of the city and county, the eager proselytizer may seek to use these symbols for his own ends. The urge to use them to teach or to taunt is always present. It is also true that some devout adherents of Judaism or Christianity may be as offended by the holiday display as are nonbelievers, if not more so. To place these religious symbols in a common hallway or sidewalk, where they may be ignored or even insulted, must be distasteful to many who cherish their meaning.
*679For these reasons, I might have voted against installation of these particular displays were I a local legislative official. But we have no jurisdiction over matters of taste within the realm of constitutionally permissible discretion. Our role is enforcement of a written Constitution. In my view, the principles of the Establishment Clause and our Nation’s historic traditions of diversity and pluralism allow communities to make reasonable judgments respecting the accommodation or acknowledgment of holidays with both cultural and religious aspects. No constitutional violation occurs when they do so by displaying a symbol of the holiday’s religious origins.
Justice Stevens is incorrect when he asserts that requiring a showing of direct or indirect coercion in Establishment Clause cases is “out of step with our precedent.” Ante, at 650, n. 6. As is demonstrated by the language Justice Stevens quotes from Engel v. Vitale, 370 U. S. 421, 430 (1962), our eases have held only that direct coercion need not always be shown to establish an Establishment Clause violation. The prayer invalidated in Engel was unquestionably coercive in an indirect manner, as the Engel Court itself recognized in the sentences immediately following the passage Justice Stevens chooses to quote. Id., at 430-431.
The majority rejects the suggestion that the display of the creche can “be justified as an ‘accommodation’ of religion,” because it “does not remove any burden on the free exercise of Christianity.” Ante, at 601, n. 51. Contrary to the assumption implicit in this analysis, however, we have never held that government’s power to accommodate and recognize religion extends no further than the requirements of the Free Exercise Clause. To the contrary, “[t]he limits of permissible state accommodation to religion are by no means coextensive with the non-interference mandated by the Free Exercise Clause.” Walz v. Tax Comm’n of New York City, 397 U. S. 664, 673 (1970). Cf. Texas Monthly, Inc. v. Bullock, 489 U. S. 1, 38 (1989) (Scalia, J., dissenting).
One can imagine a case in which the use of passive symbols to acknowledge religious holidays could present this danger. For example, if a city chose to recognize, through religious displays, every significant Christian holiday while ignoring the holidays of all other faiths, the argument that the city was simply recognizing certain holidays celebrated by its citizens without establishing an official faith or applying pressure to obtain adher*665ents would be much more difficult to maintain. On the facts of these cases, no such unmistakable and continual preference for one faith has been demonstrated or alleged.
The majority suggests that our approval of legislative prayer in Marsh v. Chambers is to be distinguished from these cases on the ground that legislative prayer is nonseetarian, while creches and menorahs are not. Ante, at 603. In the first place, of course, this purported distinction is utterly inconsistent with the majority’s belief that the Establishment Clause “mean[s] no official preference even for religion over nonreligion.” Ante, at 605. If year-round legislative prayer does not express “official preference for religion over nonreligion,” a créche or menorah display in the context of the holiday season certainly does not “demonstrate a preference for one particular sect or creed.” Ibid. Moreover, the majority chooses to ignore the Court’s opinion in Lynch v. Donnelly, 465 U. S. 668 (1984), which applied precisely the same analysis as that I apply today: “[T]o conclude that the primary effect of including the creche is to advance religion in violation of the Establishment Clause would require that we view it as more beneficial to and more an endorsement of religion . . . than . . . the legislative prayers upheld in Marsh v. Chambers . . . .” Id., at 681-682.
The creche in Lynch was owned by Pawtucket. Neither the creche nor the menorah at issue in this case is owned by a governmental entity.
The majority illustrates the depth of its error in this regard by going so far as to refer to the concmrence and dissent in Lynch as “[o]ur previous opinions. ...” Ante, at 602.
Contrary to the majority’s discussion, ante, at 604-605, and nn. 53-54, the relevant historical practices are those conducted by governmental units which were subject to the constraints of the Establishment Clause. Acts of “official discrimination against non-Christians” perpetrated in the 18th and 19th centuries by States and municipalities are of course irrelevant to this inquiry, but the practices of past Congresses and Presidents are highly informative.
In keeping with his strict views of the degree of separation mandated by the Establishment Clause, Thomas Jefferson declined to follow this tradition. See 11 Writings of Thomas Jefferson 429 (A. Lipscomb ed. 1904).
Similarly, our Presidential inaugurations have traditionally opened with a request for divine blessing. At our most recent such occasion, on January 20,1989, thousands bowed their heads in prayer to this invocation:
“Our Father and our God, Thou hast said blessed is the nation whose God is the Lord.
“We recognize on this historic occasion that we are a nation under God. This faith in God is our foundation and our heritage. . . .
*672“As George Washington reminded us in his Farewell Address, morality and faith are the pillars of our society. May we never forget that.
“We acknowledge Thy divine help in the selection of our leadership each 4 years.
“All this we pray in the name of the Father, the Son, and the Holy Spirit. Amen.” 136 Cong. Rec. 303 (1989) (Rev. Billy Graham).
If the majority’s test were to be applied logically, it would lead to the elimination of all nonsecular Christmas caroling in public buildings or, presumably, anywhere on public property. It is difficult to argue that lyrics like “Good Christian men, rejoice,” “Joy to the world! the Savior reigns,” “This, this is Christ the King,” “Christ, by highest heav’n adored,” and “Come and behold Him, Born the King of angels” have acquired such a secular nature that nonadherents would not feel “left out” by a government-sponsored or approved program that included these carols. See W. Ehret & G. Evans, The International Book of Christmas Carols 12, 28, 30, 46, 318 (1963). We do not think for a moment that the Court will ban such carol programs, however. Like Thanksgiving Proclamations, the reference to God in the Pledge of Allegiance, and invocations to God in sessions of Congress and of this Court, they constitute practices that the Court will not proscribe, but that the Court’s reasoning today does not explain.
Justice Blackmun and Justice O’ConnoR defend the majority’s test by suggesting that the approach followed in Lynch would require equally difficult line drawing. Ante, at 606; ante, at 629-630 (O’ConnoR, J., concurring in part and concurring in judgment). It is true that the Lynch test may involve courts in difficult line-drawing in the unusual case where a municipality insists on such extreme use of religious speech that an establishment of religion is threatened. See supra, at 661. Only adoption of the absolutist views that either all government involvement with religion is permissible, or that none is, can provide a bright line in all cases. That price for clarity is neither exacted nor permitted by the Constitution. But for the most part, Justice Blackmun’s and Justice O’ConnoR’s objections are not well taken. As a practical matter, the only cases of symbolic recognition likely to arise with much frequency are those involving simple holiday displays, and in that context Lynch provides unambiguous guidance. I would follow it. The majority’s test, on the other hand, demands the Court to draw exquisite distinctions from fine detail in a wide range of cases. The anomalous result the test has produced here speaks for itself.
Of course, a majority of the Court today rejects Justice Blackmun’s approach in this regard. See ante, at 636-637 (O’ConnoR, J., concurring in part and concurring in judgment).