dissenting.
The district court held that the New Jersey minute of silence statute, New Jersey P.L. 1982 Ch. 205, was unconstitutional because it violated all of the elements of the three-part disjunctive test set forth in Lemon v. Kurtzman, 403 U.S. 602, 91 S.Ct. 2105, 29 L.Ed.2d 745 (1971): (1) it did not have a secular purpose, (2) its primary effect both advanced and inhibited religion, and (3) it fostered excessive governmental entanglement with religion by promoting divisiveness among religious groups. 572 F.Supp. 1561, 1572-76 (D.N.J.1984). A majority of this panel disagrees with the district court on the second and third points: it finds neither impermissible effect nor excessive entanglement. Majority Op. at 246. With these conclusions I have no quarrel.
Although the majority agrees with the district court that the animating legislative purpose is religious, it reaches this conclusion by an analysis different from that relied upon by the district court. Unlike the district court, the majority interprets the statute to permit, rather than to mandate, student prayer. This reading I believe to be correct. Because of the permissive nature of the statute, the majority concludes that the legislative purpose was not to encourage prayer over other activities, but to accommodate those students who wish to pray in school. Its decision thus turns on the requirement of that the legislature have a secular purpose even when it accommodates religion. The majority holds this accommodation unconstitutional because it accepts the district court’s finding that the statute lacks a secular purpose. I respectfully dissent from this holding.
In my view, a permissive and facially neutral statute such as this is presumptively constitutional. This view is supported not only by common sense, but also by statements in Zorach v. Clausen, 343 U.S. *254306, 72 S.Ct. 679, 96 L.Ed. 954 (1952), and in several of the opinions in Wallace v. Jaffree, — U.S. -, 105 S.Ct. 2479/86 L.Ed.2d 29 (1985). I also believe that the presumption becomes a strong one where, as here, the panel has concluded that the statute has no impermissible effect and fosters no excessive governmental entanglement with religion. As I see it, under these circumstances, only extremely clear proof that the law is devoid of a secular purpose would justify our holding it unconstitutional; such proof was not produced in this case.
Second, I believe that the majority applied an incorrect standard of review when it deferred to the findings of the district court. Bose Corp. v. Consumers Union, 466 U.S. 485, 104 S.Ct. 1949, 80 L.Ed.2d 502 (1984), dictates that our review in this first amendment case be plenary. Under a plenary review standard, after analyzing the evidence of purpose adduced at trial (and bearing in mind the permissive nature of the statute), I conclude that the statute has a legitimate secular purpose, hence is not infirm. Finally, even under a deferential standard of review, I would not accept the district court’s purpose finding because it is unsupported by the admissible evidence. Accordingly I would reverse the judgment of the district court.
In the ordinary course I would commence my discussion with these grounds for reversal. Here, however, in light of what I believe to be the flawed accommodation analysis that informs the majority’s position, I shall discuss the accommodation issue first. Unlike the majority, I do not believe that a statute accommodates religion when it lifts no pre-existing impediment to religious exercise and when, in fact, it creates the alleged religious practice in question. The majority, in the exercise of describing the minute of silence statute as religious, has mistaken the permissive nature of the enactment for an active accommodation of religion. I engage in this discussion of accommodation because this defect is central to the majority’s argument, because it emphasizes the unlikelihood that this statute is unconstitutional, and because it is important to identify the confusion in the jurisprudence about the concept of accommodation. As a preface to this accommodation discussion— and to the opinion itself — I shall explain in a fashion somewhat different from that employed by the majority, why the statute is permissive.
I. NATURE OF THE STATUTE
A. The Permissive Nature of the Statute
The statute provides:
Principals and teachers ... shall permit students to observe a 1 minute period of silence to be used at the discretion of the individual student, ... for quiet and private contemplation or introspection____
It is apparent from the face of the statute that the only mandatory word, “shall,” is directed at principals and teachers. The language applicable to students is, by contrast, permissive: the minute is to be used “at the discretion of the individual student.” Presumably, this language means that students may read, or listen through earphones to radios or tape cassettes, or be excused from class during the minute of silence. As the majority's review demonstrates, see Majority Op. at 246, prior drafts of the minute of silence statute and the legislative history of this version fully support the conclusion that, insofar as student participation is concerned, the law is permissive only.1
*255Of course, students who are so disposed may engage in silent prayer during the minute of silence. But this activity occurs only because the statute allows the students to use the minute of silence at their discretion; the statute neither requires nor encourages prayer. The statute refers to “quiet and private contemplation or introspection.” While these activities may outwardly resemble prayer, they frequently involve quite different attitudes of mind. “Prayer,” and “private contemplation or introspection,” are not synonymous.
Like the majority, therefore, I interpret the statute as permitting, but not coercing, prayer. The majority proceeds to characterize this permissive law as a governmental accommodation of religion. I have strong reservations about the majority’s accommodation analysis.
B. Accommodation of Religion?
The majority reasons that by providing a minute of silence at the beginning of the school day, the government accommodates those students who wish to pray. Because of the neutrality of the statute and the absence of any pre-existing governmental impediment to silent prayer in schools, I disagree with the interpretation of the statute as an accommodation of religion. Moreover, I have doubts as to whether a statute that creates the purported religious observance can be viewed as an accommodation of religion. While I do not accept the majority’s accommodation analysis, the-concept of accommodation is murky enough that I admit that the majority’s approach may be correct. As I have explained, see supra at p. 254, it is important for several reasons to clarify the problems with the majority’s accommodation analysis, not the least of which is that the majority’s resort to the unclear concept of accommodation suggests that this statute is not religious at all, and how unlikely it is that it is unconstitutional.
Government accommodation of religion, as opposed to mere tolerance, has been approved by the Supreme Court, at least where there is some secular purpose of the governmental practice. See Lynch v. Donnelly, 465 U.S. 668, 104 S.Ct. 1355, 1359, 79 L.Ed.2d 604 (1984). While its contours are not entirely clear, an accommodation seems to consist of the lifting of a barrier to religious observance. Thus, without a release time program, public school attendance requirements would interfere with religious instruction classes. See Zorach v. Clauson, 343 U.S. 306, 72 S.Ct. 679, 96 L.Ed. 954 (1952). In his opinion of the Court in Wallace v. Jaffree, — U.S.-, 105 S.Ct. 2479, 86 L.Ed.2d 29 (1985), Justice Stevens recognized that a burden on religious exercise exists prior to a government accommodation of religion. He pointed out that at the time of enactment of the Alabama statute under consideration, “there was no governmental practice impeding students from silently praying for one minute at the beginning of each school day; thus, there was no need to ‘accommodate’ or to exempt individuals from any general governmental requirement because of the dictates of our cases interpreting the Free Exercise Clause.” Id., 105 S.Ct. at 2491 n. 45.
In her opinion in Wallace, supra, Justice O’Connor also described accommodation as a lifting of a governmental impediment to religious practice. 105 S.Ct. at 2505. She concluded that “the Alabama statute at issue today lifts no state-imposed burden on the free exercise of religion, and accordingly cannot properly be viewed as an accommodation statute.” Id. Similarly, prior to the enactment of the New Jersey minute of silence statute, there was “no governmental practice of impeding students from silently praying for one minute at the beginning of each school day.” Under the analysis of two Justices in Wallace, supra, therefore, the statute cannot be considered an accommodation.
*256Moreover, I doubt that accommodation analysis is appropriate when the religious interest in question is created by the very statute that accommodates it. Until the adoption of the minute of silence statute, there was no formally established quiet time during which students in public schools could pray. Thus, the allegedly offending practice — praying during silent time at school — was created by the statute. When it concludes that the statute is a governmental accommodation of religion, the majority engages in bootstrapping. The minute of silence statute does not allow for the observance of a pre-existing religious interest, as did, for example, the release time program for religious training approved in Zorach. And, if the statute provides for any religious activity at all, it provides for a new religious activity; hence, it cannot be said to accommodate an existing one.
There is a potential flaw in the foregoing analysis, one stemming from Lynch, supra. It is possible to read Lynch as approving the use of accommodation analysis where the religious interest is created by the statute or governmental practice in question and where there was no pre-exist-ing burden to religious observance. The Court in Lynch treated Pawtucket’s practice of displaying a nativity scene as an accommodation of religion even though the challenged governmental practice arguably created the religious interest it accommodated, i.e., the interest in viewing the nativity scene. Looking at the problem differently, one might say what was accommodated was not the specific interest created by the government, but rather the more general American religious heritage. Under either interpretation of Lynch, the minute of silence statute can be viewed as an accommodation of religion.
Moreover, it is not clear that the nativity display challenged in Lynch removed a preexisting impediment to religious exercise. In fact, if the religious interest that the Pawtucket display accommodated was created by the display itself, there could have been no pre-existing impediment to religious exercise because there was no exercise to impede. Had the municipality not erected a nativity scene, there would have been no governmental impediment to citizens setting up their own creches, although these nativity scenes would require some expenditure of money by individual citizens. More to the point, the Lynch Court never considered the existence of any impediment to religious observance in the absence of the’ municipal nativity scene.2
Despite Lynch, which is, after all a very different kind of case, I am skeptical about the majority’s characterization of the statute as an accommodation of religion. I believe that my colleagues confuse the permissive nature of the statute with an active accommodation of religious activity. Because the statute is permissive and facially neutral, the majority understandably has difficulty ascribing a religious nature to it. The only way it can do so, it seems, is by considering the law an accommodation of religion (although, ironically, accommodations of religion have been held constitutional by the Supreme Court). The problems with the majority’s accommodation analysis emphasize the need for clarification of this area by the Supreme Court. Moreover, the fact that the majority was forced to turn to accommodation in order to ascribe a religious nature to a permissive, facially neutral statute emphasizes how unlikely it is that such a statute is unconstitutional.
C. The Presumption of Constitutionality
As we have seen, the statute is permissive and facially neutral, i.e., it makes no *257mention of religion. In my view, such a statute is presumptively constitutional.3 When the state has neither given support to a religious group nor enforced any religious observance, it can hardly be said to have established a religion. As I see it, only very strong proof of excessive entanglement, of impermissible effects, or of the absence of any secular purpose will serve to invalidate such a statute.4 Statements in Zorach, supra, and Wallace, supra, underscore the unlikelihood that such a statute will be unconstitutional.
In Zorach, the absence of governmental coercion indicated to the Supreme Court that a school that released students during the day for religous study violated neither the Establishment nor the Free Exercise clauses. The Court’s reasoning was clear, and its holding unequivocal:
No one is forced to go to the religious classroom and no religious exercise or instruction is brought to the classrooms of the public schools. A student need not take religious instruction, he is left to his own desires as to the manner or time of his religious devotions, if any.
There is a suggestion that the system involves the use of coercion to get public school students into religious classrooms. There is no evidence on the record before us that supports that conclusion. The present record indeed tells us that the school authorities are neutral in this regard and do no more than release students whose parents so request. If in fact coercion were used, it it were established that any one or more teachers were using their office to persuade or force students to take the religious instruction, a wholly different case would be presented.
343 U.S. at 311, 72 S.Ct. at 682 (footnotes omitted). The lesson of Zorach, then, is that so long as authorities do not force students to participate, the authorities may allow and provide time and room for voluntary religious participation.
The Zorach reasoning applies directly to the statute at issue here. No student is forced to observe a minute of contemplation, much less prayer. Each student is left to his or her own desires as to how to use that time, and the state is neutral. The statute does no more than provide students who wish to pray silently with an opportunity to do so undisturbed by noise.
At least one other case supports the proposition that moment of silence statutes that are permissive and facially neutral presumptively have a secular purpose. In Abington School District v. Schempp, 374 U.S. 203, 281, 83 S.Ct. 1560, 1602, 10 L.Ed.2d 844 (1963) (Brennan, J., concurring), Justice Brennan said that “the observance of a moment of reverent silence at the opening of the class [may serve] the solely secular purposes of devotional activities without jeopardizing either the religious liberties of any members of the community or the proper degree of separation between the spheres of religion and government.” (quoted in Wallace, supra, 105 S.Ct. at 2499 (O’Connor, J., concurring)).
The vitality of such reasoning has recently been reaffirmed in Wallace, which, like the instant case, involved a moment of silence statute. The statute at issue in Wallace gave each teacher the discretion to announce a period of silence in her or his classroom; however, if the teacher did impose the minute of silence, the students were required to pray. The Supreme Court found the statute unconstitutional. In doing so, the Court gave clear indication that a statute that did not require, but merely permitted, a student to pray, would be constitutional.
*258In his opinion for the Court, in which Justices Brennan, Marshall, and Blackmun joined, Justice Stevens explicitly distinguished the statute before the Court from a statute, like the one before us, that allows students opportunity to pray but does not mandate prayer:
The legislative intent to return prayer to the public schools is, of course, quite different from merely protecting every student’s right to engage in voluntary prayer during an appropriate moment of silence during the school day.
Id. at 2491. I believe it to be implicit in this discussion that statutes that “merely protect[] every student’s right to engage in voluntary prayer” carry a strong presumption of constitutionality.
The concurring and dissenting opinions in Wallace, supra, lend further support to this view. In her concurring opinion, Justice O’Connor stated that “[b]y mandating a moment of silence, a State does not necessarily endorse any activity that might occur during that period.” Id. at 2499 (O’Connor, J., concurring). This statement is a reaffirmation of the Zorach principle as applied to moments of silence: if the legislation allowing for such moments does not endorse religious activity, it is not unconstitutional. And Justice Powell observed that he would “vote to uphold the Alabama statute if it also had a clear secular purpose” in addition to the religious purpose for which it was enacted. 105 S.Ct. at 2495. The citation he used to support this statement implies thal. he would discern secular purpose from the face of some statutes. He said, “See Mueller v. Allen, 463 U.S. 388, 394-95, 103 S.Ct. 3062, 3066, 77 L.Ed.2d 721 (1983) (the Court is ‘reluctan[t] to attribute unconstitutional motives to the state, particularly when a plausible secular purpose may be discerned from the face of the statute’).” Id. I believe that the text of the statute, reflects a clear secular purpose, i.e., the providing of a moment of calm before the beginning of classes. See also infra p. 253.
In their dissenting opinions in Wallace, Chief Justice Burger and Justice White emphasized the absence of coercion as a reason for holding the Alabama statute unconstitutional. The Chief Justice described the statute in Wallace thus: “Without pressuring those who do not wish to pray, the statute simply creates an opportunity to think, to plan, or to pray if one wishes — as Congress does by providing chaplains and chapels.” 105 S.Ct. at 2507-08 (Burger, C.J., dissenting). He went on to state, “[i]f the government may not accommodate religious needs when it does so in a wholly neutral and noncoercive manner, the ‘benevolent neutrality’ that we have long considered the correct constitutional standard will quickly translate into the ‘callous indifference’ that the Court has consistently held the Establishment Clause does not require.” Id. at 2508.
According to Justice White, “if a student asked whether he could pray during [the moment of silence], it is difficult to believe that the teacher could not answer in the affirmative.” Id. (White, J., dissenting). This suggests, I think, that Justice White would consider a statute constitutional if it permitted but did not require, prayer.
The Wallace opinions also emphasize how improbable it is that a permissive, facially neutral statute will fail the effects test of Lemon. In his concurring opinion, Justice Powell stated that “the ‘effect’ of a straightforward moment-of-silence statute is unlikely to ‘advanc[e] or inhibit] religion’.” 105 S.Ct. at 2495. Justice O’Con-nor provided a lengthier explanation of why a mere minute of silence, without any encouragement to pray, passes the effect test of Lemon. She stated:
First, a moment of silence is not inherently religious. Silence, unlike prayer or Bible reading, need not be associated with a religious exercise. Second, a pupil who participates in a moment of silence need not compromise his or her beliefs. During a moment of silence, a student who objects to prayer is left to his or her own thoughts, and is not compelled to listen to the prayers or *259thoughts of others____ It is difficult to discern a serious threat to religious liberty from a room of silent, thoughtful schoolchildren.
105 S.Ct. at 2498-99. Under this reasoning, it is hardly possible to hold that the statute challenged here has the effect of advancing or inhibiting religion. And, indeed, the majority has conceded that the New Jersey minute of silence has no unconstitutional effect.
Logic, precedent, and the writings of a majority of the Justices suggest that a permissive, facially neutral statute such as the New Jersey moment of silence law, which has no impermissible effects and fosters an impermissible entanglement, carries with it a strong presumption of constitutionality which only the clearest proof of an absence of secular purpose can vitiate. With the nature of the New Jersey statute in mind, I turn to an analysis of its purpose, the ground upon which the majority finds the law unconstitutional. As I demonstrate in part II, clear proof of the absence of a secular purpose is surely lacking here.
II. THE STATUTE’S SECULAR PURPOSE
The majority affirms the holding of unconstitutionality, because it believes that it cannot overturn the district court’s finding that the statute lacks a secular purpose. I believe that the majority proceeds under an erroneous standard of review. I shall first explain why I believe that Bose Corp. v. Consumers’ Union, 466 U.S. 485, 104 S.Ct. 1949, 80 L.Ed.2d 502 (1984), requires that our standard of review in this case should be plenary. Next, I shall engage in plenary review of the evidence bearing on legislative purpose, and explain why I believe the evidence shows that the statute had a secular purpose. Finally, I shall demonstrate that even under a deferential standard of review, the district court’s purpose finding is unsupportable.
A. The Relevance of Bose
In Bose Corp. v. Consumers Union, 466 U.S. 485, 104 S.Ct. 1949, 80 L.Ed.2d 502 (1984) the Supreme Court held that an appellate court is required to exercise plenary review of a lower court’s finding of actual malice in a libel case. The majority would apparently limit Bose to its facts, exercising plenary review only in libel or defamation cases. I believe that this limitation represents a misreading of Bose. Bose’s language and reasoning were broad; it drew from precedents in a number of first amendment cases. Because I read Bose as a statement about the role of appellate courts in first amendment cases generally, I believe that it dictates our scope of review here.
The Bose determination of whether a speaker or writer had “actual malice,” like the determination of whether a legislature had a secular purpose, is a factual question concerning intent. Because it is a factual question, Fed.R.Civ.P. 52(a) would appear to require federal appellate courts to review district court’s findings on the matter under a deferential standard. In Bose, however, the Court held that as a matter of constitutional law, plenary review was required:
The requirement of independent appellate review ... is a rule of federal constitutional law. It emerged from the exigency of deciding concrete cases; it is law in its purest form in our common law heritage. It reflects a deeply held conviction that judges — and particularly members of this Court — must exercise such review in order to preserve the precious liberties established and ordained by the Constitution.
104 S.Ct. at 1965. In my view, this language is applicable to any findings of a district court that implicate a party’s first amendment interests. Indeed, the Court arrived at its position in Bose after a lengthy review of first amendment cases. It pointed out that it had exercised plenary review in cases involving libel, fighting words, incitement to riot, obscenity and child pornography. Id. at 1961-63.
Although the majority in this case apparently would have it differently, the Court *260regarded Bose not as an aberration but as an imperative of first amendment principles and jurisprudence. Plenary review is necessary to protect the delicate balance that the first amendment entails. As the Court explained, “[providing triers of fact with a general description of the type of communication whose content is unworthy of protection has not, in and of itself, served sufficiently to narrow the category, nor served to eliminate the danger that decisions by triers of fact may inhibit the expression of protected ideas.” Id. at 1962 (footnote omitted). Under Bose, plenary review is appropriate in a first amendment case whenever important “constitutional facts” are at issue, that is, whenever the district court’s findings of fact presuppose a legal standard and have potentially grave effects on first amendment rights. The Supreme Court has recently stated this general principle: “Where, for example, as with proof of actual malice in First-Amendment libel cases, the relevant legal principle can be given meaning only through its application to the particular circumstances of a case, the Court has been reluctant to give the trier of fact’s conclusions presumptive force and, in so doing, strip a federal appellate court of its primary function as an expositor of law.” Miller v. Fenton, — U.S.-,-, 106 S.Ct. 445, 452, 88 L.Ed.2d 405 (1985). See also Bender v. Williamsport Township, 741 F.2d 538, 542 n. 3 (3d Cir.1984) (“in the area of ‘constitutional fact,’ an appellate court is free to draw its own inferences from the record”), cert. granted, — U.S.-, 105 S.Ct. 1167, 84 L.Ed.2d 319 (1985).5
It should be clear that plenary review is appropriate in this case. Whether the statute had a permissible purpose is an important constitutional fact, for permissibility is determined by a legal standard and potentially grave effects on first amendment rights are at stake. Indeed, as noted the one at issue in Bose in that both the determination of actual malice and that of secular purpose involve inquiries into states of mind, guided by complicated legal standards. Distinguishing what is religious from what is secular is a process as overlaid with constitutional considerations as deciding the difference between protected and unprotected speech. The principles that underpin Bose are thus equally forceful here.
To complete my argument in favor of plenary review, I will consider here two distinctions that one might arguably draw between the instant case and Bose and explain why I believe that neither is material. First, one might argue that Bose should be limited to cases where an appellate court reviews decisions denying assertions of first amendment rights, whereas the district court’s decision in this case arguably upholds the expressive rights of citizens. I do not believe, however, that the scope of review in first amendment cases is determined by which side is favored by the district court. The appellate court’s concern should be to draw first amendment boundaries correctly, not to favor one side over the other. Moreover, in its extensive summary of cases, see 104 S.Ct at 1961-63; supra at 259-260, the Bose Court cited cases in which the lower court had both upheld and denied first amendment claims; the Court did not distinguish between the two sets of cases. Thus, the first arguable distinction is simply not tenable.
Second, one might argue that Bose should be limited to cases involving speech, and perhaps to cases involving free exercise claims, but should not be extended to establishment cases. This position requires the implicit assumption that plenary review is more important when freedom of speech (or free exercise of religion) is at *261issue because an erroneous determination that speech (or religious exercise) is unprotected inhibits the expression of individuals’ ideas and thus impairs personal freedom. To make explicit this assumption is to reveal its falsity, for one of the most fundamental of the Founding Fathers’ tenets was that individuals must be free from governmentally imposed religion. The constitutional prohibition of established religion is surely as important as the guarantee of freedom of speech.6
In the next section, I shall examine under a plenary standard of review the evidence relating to the purpose of the statute.
B. Evidence of Legislative Intent
The evidence presented to the district court may be classified in five categories: testimony of three witnesses who were present at legislative hearings; newspaper accounts of the hearings; evidence of the perceptions of various citizens and citizens’ groups of New Jersey about the statute’s purpose; expert testimony about the pedagogical value of the statute; and seventeen prior bills proposing moments of prayer or moments of silence. I shall consider these in turn, and also consider one final piece of evidence of legislative purpose, the text of the statute itself.
1. Three people who attended the legislative hearings testified to legislators’ statements concerning the religious purpose of the statute. Although the statute was debated by the full assembly and the full senate, the witnesses mentioned only two assembly members who ever referred to religion: Ms. Rhodes and Mr. Chuman repeated statements reflecting a religious purpose made by Assemblyman Zangari, and Reverend Sarfarty described a similar remark made by Assemblyman Adubato.
This evidence is simply insufficient to support a finding of a religious intent on the part of the legislature. In the first place, given its paucity, the evidence is at best underwhelming. Two assemblymen cannot be said to speak for the intent of the entire assembly and senate. See United States v. O’Brien, 391 U.S. 367, 384, 88 S.Ct. 1673, 1683, 20 L.Ed.2d 672 (1968) (“[w]hat motivates one legislator to make a speech about a statute is not necessarily what motivates scores of others to enact it, and the stakes are sufficiently high for us to eschew guesswork.”). Yet there is no evidence on the record that any other as-semblymember or senator had a religious purpose in mind. More importantly, even this testimony is equivocal because each of the witnesses also testified to legislators’ statements that the statute had a secular purpose.7 Because we may not invalidate a statute that has a secular purpose, Lemon, supra, 403 U.S. at 611, 91 S.Ct. at 2110; *262see also Wallace, 105 S.Ct. at 2490, testimony that reveals both a religious and a secular purpose cannot support a finding of unconstitutionality.8
2. The district court also relied on newspaper articles describing the legislative hearings on the statute. Because the New Jersey legislature does not keep a record of its hearings and debates and because the district court excused the legislators from testifying about their motives in enacting the statute,9 the district court found that there was sufficient necessity to admit the articles under the residual hearsay exception, Fed.R.Evid. 803(24).10 The newspaper reports admitted suggested that the purpose behind the statute was to return prayer back into the public schools, surreptitiously.
There are two problems with the newspaper evidence. First, I do not believe that the newspaper accounts satisfy the requirements for admission of Fed.R.Evid. 803(24).11 That section reads, in relevant part, as follows:
The following are not excluded by the hearsay rule, even though the declarant is available as a witness:
* * * * * *
(24) other exceptions. A statement not specifically covered by any of the foregoing exceptions but having equivalent circumstantial guarantees of trustworthiness, if the court determines that (A) the statement is offered as evidence of a material fact; (B) the statement is more probative on the point for which it is offered than any other evidence which the proponent can procure through reasonable efforts; and (C) the general purposes of these rules and the interests of *263justice will be served by admission of the statement into evidence.
The legislative history of the Rule indicates that it was to be narrowly construed and invoked only sparingly. See S.Rep. No. 1277, 93d Cong. 2d Sess. (1974), reprinted in [1974] U.S.Code Cong. & Admin.News, pp. 7051, 7065; H.Conf.Rep. No. 1597, 93d Cong., 2d Sess. (1974), reprinted in [1974] U.S.Code Cong. & Admin.News, pp. 7051, 7105. See also In re Japanese Electronics Products, 723 F.2d 238, 302 (3rd Cir.1983) (“We note, moreover, that those exceptions are intended to have a narrow focus.”); United States v. Bailey, 581 F.2d 341, 346-47 (3d Cir.1978); Robinson v. Shapiro 646 F.2d 734, 742 (2d Cir.1981); Huff v. White Motor Corp., 609 F.2d 286, 291 (7th Cir.1979).
It is clear that two of the explicit prerequisites of Rule 803(24) were not met. As the majority has observed, Maj.Op. at 252, n. 9, the plaintiffs made no showing of inability to locate observers who attended the legislative debates about the statute and three of them in fact testified. The live testimony of these witnesses was far preferrable to newspaper accounts of the debate, and, while the plaintiffs might have been able to get the names of additional observers through reasonable efforts, they made no effort to do so. Cf. Debra P. by Irene P. v. Turlington, 730 F.2d 1405 (11th Cir.1984) (consulting firm’s report admitted pursuant to 803(24) only after a showing that there was no better available evidence). Thus, the newspaper clippings did not satisfy the 803(24)(B) requirement that the evidence be more probative than any other evidence that the plaintiffs could have procured through reasonable efforts.
Nor did the clippings have the requisite “circumstantial guarantees of trustworthiness.” Courts admitting evidence under 803(24) require some showing that “the declarant’s perception, memory, narration, or sincerity,” are reliable. United States v. Friedman, 593 F.2d 109, 119 (9th Cir.1979).12 Under this standard, the newspaper reports are of dubious validity. On the basis of the record, we do not know anything about the reliability of the articles; perception, memory, narration, and misrepresentation all present potential problems. The reporters may have been present for only part of the hearings they reported, and may not correctly have remembered or interpreted the speeches they heard. They may have simply recorded their impressions of the hearings, in which case the clippings would provide only very slender evidence of legislative intent. We have no idea of the amount of editorializing that went into the articles, which could have been written from a biased point of view. It is not unknown for reporters to stretch some facts or omit others in order to arouse public indignation.13
I thus conclude that the newspaper accounts were inadmissible.14
*264The second problem with the newspaper evidence is that, even if the accounts were admissible, they are ambiguous at best in establishing that the legislature had forbidden motivations in passing the statute. The newspaper accounts contain descriptions of legislators’ secular purposes, as well as statements about returning prayer to the schools. For example, one article on which plaintiffs rely quoted Assemblyman Zangari as saying, inter alia, that “regardless of what some people say, this is not a prayer bill. Rather it allows each student to participate in a one-minute period of voluntary silence to be used at his or her own discretion.” The Star Ledger, P.-, col.-, Oct. 25, 1982. It appears to me that this is clear evidence of a secular purpose that the majority and the district court have ignored.
3. Another source of evidence of religious purpose is citizens’ perceptions of the statute. Proceeding under the theory that “[i]t is significant what those who opposed the Bill conceived its purposes and effects to be, because in many instances they were among those most directly affected by it,” May v. Cooperman, 572 F.Supp. at 1565, the district court considered reaction to the statute by the Ethical Culture Society of Bergen County, the New Jersey School Board Association, the American Baptist Churches of New Jersey, the New Jersey Education Association, and several parents of public school students.
I believe that this evidence is very weak. Community perception is simply too amorphous and unreliable to provide the ground for constitutional decisions. The opinions and perceptions of the community are shaped by many factors — editorials, personal biases, gossip, news broadcasts, and peer pressure, for example. Such perceptions are thus unreliable indicators of what the legislative purpose of the statute in fact was.15 These problems were exacerbated by the district court’s notion that the perceptions of those who opposed the statute were particularly valuable. The perceptions of those who oppose the statute were no truer or more reliable than the perceptions of the statute’s supporters. Thus, in addition to being inherently unreliable, the evidence that the district court heard was likely skewed.16
4. The district court heard three experts testify about the effectiveness of the min*265ute of silence as a teaching device. Dr. Scrupski, an associate professor at Rutgers Graduate School of Education, testified that the minute of silence served a useful pedagogical function by creating a boundary between school and non-school time. The boundary was important, Dr. Scrupski explained, because it would emphasize to the students the importance of the school hours, and the fact that school hours were a time for serious work rather than undisciplined play. Dr. Kuriloff, an associate professor of education at the University of Pennsylvania, and Dr. Rosner, a professor at the Temple College of Education, testified that they believed that the moment of silence before class would not be a valuable educational tool.
Once again, I find the evidence inadequate to support a finding of no secular purpose. Not only was the testimony that there is no pedagogical merit to a minute of silence contested, but even if it had stood alone and uncontradicted it would not establish lack of secular purpose. The testimony of the plaintiffs’ experts may cast doubt upon the effectiveness of the statute as a transitional, or boundary, device. But proof that the statute was an unwise or ineffective measure is not proof that it lacked a secular purpose. It is significant only if the statute seems so patently ineffective or irrelevant with respect to the alleged secular purpose that it is unreasonable to believe that the legislature enacted the statute to serve that purpose. The expert testimony in this case could not give rise to such an inference; indeed, as noted, there was disagreement among the experts as to the pedagogical merit of the statute. Under the circumstances, the expert testimony was of little value in establishing the lack of a secular purpose.
5. The district court described the attempts by prior New Jersey legislatures to pass a minute of silence or moment of prayer statute. This evidence has a more prominent role in the majority’s opinion than it did in the district court’s. It is clear, however, that both the district court and the majority consider the perceived purposes of the previous bills to be indicative of the purpose underlying the statute.
I believe that the previous bills are irrelevant. The bills were submitted to different legislatures with significantly different memberships. The overlap between the sponsors and supporters of the statute and the sponsors and supporters of the statute and the sponsors and supporters of the previous bills is minimal. Thus the circumstances surrounding the prior bills supply little evidence of the legislative purpose behind the present enactment. Earlier drafts of the same bill can shed light on the intent behind the final formulation, but the histories of completely separate bills have no such relevance.
6. The final evidence of legislative purpose may not be properly classifiable as evidence at all: the language of the statute itself. But the Supreme Court has stated that “a plausible secular purpose for [a] state’s program may be discerned from the face of the statute.” Mueller v. Allen, 463 U.S. 388, 394-395, 103 S.Ct. 3062, 3066, 77 L.Ed.2d 721 (1983). Here the neutral language of the statute is evidence in favor of a secular legislative purpose, i.e., the providing of a moment of calm before the beginning of classes.
In sum, reviewed under a plenary standard none of the evidence on which the plaintiffs rely is sufficient, separately or cumulatively, to prove an absence of any secular purpose behind the statute. The evidence is equivocal at best, and much of it may be irrelevant. The single unambiguous piece of evidence — the face of the statute itself — attests to the secular motives of the legislators. Parts of the testimony of the witnesses who attended the hearings also support this conclusion. I therefore believe that the statute has a secular purpose, that it is therefore constitutional, and that the majority is in error.
In the next section, I shall explain that I would reach the same conclusion even if we were obliged to defer to the findings of the district court.
*266C. Result Under a Deferential Standard of Review
My conclusion that a statute is constitutional does not depend on either the standard of review I have employed or the admissibility of the newspaper clippings. Even if we were required to defer to the district court’s findings, and even if the newspaper articles were admissible under Fed.R.Evid. 803(24), I would vote to reverse the judgment of the district court. All that is required by Lemon is a secular purpose. Wallace, supra, 105 S.Ct. at 2494 (Powell, J., concurring) (“We have not interpreted the [purpose] prong of Lemon, supra, however, as requiring that a statute have ‘exclusively secular’ objectives.”); Lynch, supra, 104 S.Ct. at 1363 n. 6 (“We hold only that Pawtucket has a secular purpose for its display, which is all that Lemon requires.”). Indeed, the Supreme Court has refused to hold unconstitutional, on the ground that there was some secular reason for them, practices that would seem to be purely religious. In Lynch, for example, the municipality’s celebration of Christmas and its depiction of the origins of that holiday were considered legitimate and sufficient secular purposes.
The evidence simply does not demonstrate that the statute lacks any secular purpose. Although the newspaper articles and the testimony of persons who attended the legislative hearings reveal that some legislators had prayer in mind, these very sources of evidence also contain descriptions of a secular purpose, the purpose of providing a moment of calm before the beginning of classes. As the Supreme Court has stated, “a statute that is motivated in part by a religious purpose may satisfy” the purpose prong of Lemon, as long as it has a secular aim, also. Wallace, supra, 105 S.Ct. at 2490.
Although under a deferential standard of review, I would accept the district court’s decision to believe the plaintiff’s expert witnesses and to disbelieve the conflicting expert testimony, I still find that this evidence about the effectiveness of the legislation has little bearing on its purpose. Nor do evidence of community perception of the legislative intent or the history of the seventeen prior minute of silence bills have much probative value in a purpose inquiry. Against this weak and possibly irrelevant evidence stands the permissive and neutral wording of the statute, with its strong indi-cia of constitutionality, see Part I, supra, bolstered by statements in newspaper articles and observer’s testimony that at least some legislators endorsed the statute for a secular reason. I therefore conclude that the district court clearly erred in finding that the statute had no secular purpose.
III. CONCLUSION
In sum, I agree with the majority that the statute is permissive and that it neither produces impermissible effects nor fosters excessive governmental entanglement with religion. Although the characterization of the statute as an accommodation of religion is not determinative of its constitutionality, I nonetheless have serious reservations about the majority’s accommodation analysis and believe that its flaws emphasize how unlikely it is that the statute is unconstitutional. In my view, a permissive and facially neutral statute such as the New Jersey moment of silence is presumptively constitutional; that the presumption is all the stronger when, as here, the enactment passes the entanglement and effects test of Lemon; and that such a presumption cannot be overcome in the absence of clear evidence of a secular purpose.
I find that the evidence concerning lack of a secular purpose in the case does not overcome this presumption of constitutionality. I reach the same conclusion both under the plenary standard of review that I believe must employ and under the deferential standard that the majority employs. I therefore would reverse the judgment of the district court. Accordingly, I respectfully dissent.
. The tape of a segment of a CBS news program that was placed into evidence provides further proof, if any were needed, that the statute mandates no particular behavior on the part of the students. The tape depicted the experience of the public schools in Sayreville, New Jersey, after the enactment of the challenged statute. An actual minute of silence was shown during which students’ attitudes and behavior were far from uniform. Their posture can only be described as that of studied irreverence. Cf. Wallace v. Jaffree, — U.S.-, 105 S.Ct. 2479, 2495 n. 9, 86 L.Ed.2d 29 (Powell, J., concurring) (“Given the types of subjects youthful minds, are primarily concerned with, it is unlikely that many children would use a simple 'moment of *255silence' as a time for religious prayer. There are too many other subjects on the mind of the typical child.")
. I note that Chief Justice Burger characterized the moment of silence statute at issue in Wallace as an accommodation of religion. Wallace, supra, 105 S.Ct. at 2507-08 (Burger, C.J., dissenting). It is significance that Justice Stevens’ discussion in Jaffree of a pre-existing impediment requirement mentions only "our cases interpreting the Free Exercise Clause.” Governmental impediments to religious observance are at the core of free exercise claims. However, Justice Stevens ignores Lynch, in which accommodation analysis was used to dispose of a claim that the government had violated the Establishment Clause.
. I speak of presumptions not in a technical sense, relating to the parties’ burden of proof, but in a broader sense in which it means that such a statute is probably constitutional.
. In this case, the majority has found that the statute has no impermissible effects and does not foster governmental entanglement with religion. Thus, as noted supra, the presumption of constitutionality is particularly strong, for it is most unlikely that such an enactment has a purely religious purpose.
. That Bender involved appellate review of a summary judgment record whereas we have a full record in this case replete with factual disputes is not material. There was a disputed factual record in Bose. The appellate court’s responsibility to exercise plenary review is constant no matter how the case before it arose; the inferences may be drawn from disputed as well as from undisputed records.
. Indeed in Bob Jones University v. United States, 461 U.S. 574, 103 S.Ct. 2017, 76 L.Ed.2d 157 (1983), decided a year before Bose, the Supreme Court reviewed the district court’s determination whether an institution’s purpose was primarily educational or religious under a plenary standard of review. See also Presbyterian & Reformed Publishing Co. v. Commissioner of Internal Revenue, 743 F.2d 148, 158 n. 9 (3d Cir.1984) (Tax Court's determination that a publishing house was not sufficiently connected with a church to qualify for tax-exempt status was not entitled to deferential standard of review); Cf. Bender v. Williamsport Township, 741 F.2d 538 (3d Cir.1984) (plenary review of whether speech was secular or religious).
. Mr. Chuman’s testimony on the point was as follows;
QUESTION: You said you heard Senator — or
Assemblyman — James Zangari testify?
ANSWER: Yes.
QUESTION: Do you recall what he stated— was he in favor of the adoption of this law?
ANSWER: Oh, very much so.
QUESTION: Did he state his reasons?
ANSWER: Well, yes he did, he thought that the moment of silence would be useful in the schools and would serve an educational purpose.
QUESTION: What purpose?
ANSWER: Educational purpose.
QUESTION: A teaching purpose?
ANSWER: Yes.
QUESTION: Did he expand on that and say in what way?
ANSWER: he may have, but I don’t recall. Ms. Rhodes and Reverend Sarfarty recounted statements by other legislators that described the minute of silence as an opportunity for the students to collect their thoughts and prepare for the day’s learning and as an aid in the creation of discipline and order.
. By contrast, in Wallace, the testimony of Senator Holmes, the sponsor of the bill concerning the religious purpose of the moment of silence, was unrelieved by statements concerning a secular purpose.
. The majority states that "the district court's recognition of a legislative privilege not to testify in a case in which legislative purpose is an issue may itself have been an error.” See supra at 252 n. 9. For support of this view, it relies upon United States v. O’Brien, 391 U.S. 367, 88 S.Ct. 1673, 20 L.Ed.2d 672 (1968), in which the Supreme Court stated that it would look to statements by legislators for guidance "in a very limited and well-defined class of cases where the very nature of the constitutional question requires an inquiry into legislative purpose.” Id. at 383 n. 30, 88 S.Ct. at 1682 n. 30. However, the O’Brien Court was not referring to depositions or other testimony taken from legislators, but to statements contained in the official record of the legislative proceedings. In my view, the district court acted fully within its discretion in refusing to allow the plaintiffs to depose New Jersey legislators. Although the testimony in question does not fall within the provisions of either the federal or the state constitution, there may well be a common-law speech and debate privilege that protects the legislators in this situation. At all events, it would be unseemly to parade legislators before the court to testify about their intentions in enacting a statute, and, more important, the reliability of such testimony would be highly suspect. Legislators, concerned that the statute might be held unconstitutional, could view their intentions differently in hindsight. As Justice O’Connor stated in Wallace v. Jaffree, 105 S.Ct. at 2500 (”[i]t is particularly troublesome to denigrate an expressed secular purpose due to post-enactment testimony by particular legislators----”). Thus, even if there is no common-law privilege, such testimony would be subject to a ruling that it is inadmissible under Rule 403, for its probative value is substantially outweighed by the danger that it will be inaccurate and therefore confuse the issues.
. Ordinarily, when offered to prove the truth of the matters stated therein, newspaper articles are held inadmissible as hearsay. See, e.g., Pallotta v. United States, 404 F.2d 1035 (1st Cir.1968); Poretto v. United States, 196 F.2d 392 (5th Cir.1952); De La Cruz v. Dufresne, 533 F.Supp. 145 (D.Nev.1982). Robert Stigwood Group v. O'Reilly, 346 F.Supp. 376 (D.Conn.1972), rev’d on other grds., 530 F.2d 1096, cert. denied, 429 U.S. 848, 97 S.Ct. 135, 50 L.Ed.2d 121 (1976); United States v. Crocker-Anglo National Bank, 277 F.Supp. 133, 183 n. 36 (1967). See also Annot., Admissibility of Newspaper Articles as Evidence of the Truth of the Facts States Therein, 55 ALR3d 663. These authorities do not discuss the applicability of Rule 803(24), however.
. The majority, too, had some doubts about the admissibility of these documents. See Maj.Op. at 252. The majority concluded, however, that if the evidentiary ruling was indeed error, it was harmless because there was ample other evidence supporting the trial court’s conclusion that the legislative purpose was religious.
. See also Weinstein & Berger, Evidence ¶ 803(24)(1): "Such factors as the nature — written or oral — and character of the statement, the relationship of the parties, the probable motivation of the declarant in making the statement, and the circumstances under which it was made must be assessed. Also significant are the knowledge and qualifications of the declarant." (Footnotes omitted.)
. Cf. Pittsburgh Press Club v. United States, 579 F.2d 751, 759-60 (3d Cir.1978) (ruling a survey inadmissible under Rule 803(24) because it “lacked the necessary safeguards of accuracy and flexibility") The court stated that ”[i]n the context of polls and surveys, the circumstantial guarantees of trustworthiness are for the most part satisfied if the poll is conducted in accordance with generally accepted survey principles, and if the results are used in a statistically correct way, since proper survey and statistical methods are intended to assure a poll’s reliability.” Id. at 758. Not only are there no comparable generally accepted principles of reporting, so that the entire process is less reliable than poll-taking, but we also know nothing of the methods or motivations of the reporters who wrote the articles that were admitted into evidence.
. This conclusion is unaffected by Dallas County v. Commercial Union Assurance Co., 286 F.2d 388 (5th Cir.1961), relied upon by appellees. In Dallas County, which was decided before the adoption of the Federal Rules of Evidence, the court held a newspaper article admissible under Fed.R.Civ.P. 43(a), using the kind of approach later codified in Rule 803(24). The reasons the court gave for finding that the article met the hearsay requirements of necessity and trustwor*264thiness are not present in this case. The article in Dallas County was more probative than any other evidence that could be obtained through reasonable efforts because the fire it described had occurred fifty-eight years before the trial. 286 F.2d at 396. Indeed, the Dallas County court went on to state that the rationale behind the "ancient documents” exception to the hearsay rule was applicable to the newspaper article in question. Id. Unlike the fifty-eight-year-old article at issue in Dallas County, the newspaper clippings admitted in this case cannot even be considered ancient documents, for under Fed.R.Evid. 803(16), an ancient document is twenty years old.
Similarly, the Dallas County court’s analysis of trustworthiness does not necessarily apply the articles admitted in this case. The article admitted in that case described a fire in a public building in a small town, an event that a reporter could not fabricate without embarrassment. Unlike a fire in a public building, the statements of New Jersey legislators at hearings concerning the bill are not facts of such a public nature that they would be generally known throughout the community. While it might be inconceivable that a reporter would state that legislative hearings occurred when in fact they did not, it would not be inconceivable that he or she might, either consciously or unconsciously, give a skewed or one-sided account of what was said at the hearing. Fear of embarrassment is not such a powerful guarantee of truthfulness in this case, both because the nature of the material reported is more subjective and because the reporters were not writing for small, close-knit communities.
. I have found only one case, Duffy v. Las Cruces Public Schools, 557 F.Supp. 1013 (D.N.M.1983) in which a court found public perception probative of legislative intent. Duffy is distinguishable because in that case there was significant direct evidence of legislative intent and the evidence of public perceptions was primarily cumulative: here, however, the other evidence of legislative intent is scant. If Duffy is not distinguishable, then I think that it was incorrectly decided, for the reasons stated in the text.
. It is interesting to note that in her opinion in Wallace, Justice O'Connor described community perception of a minute of silence as evidence of impermissible effect, rather than of impermissible purpose. 105 S.Ct. at 2500-01.