dissenting, with whom Circuit Judges MacKINNON and BORK concur:
I concur with the principal dissent in this case because I agree that, if traditional First Amendment analysis is applied to this sleeping, on the assumption that it is a fully protected form of expression, the appellants would nonetheless lose. I write separately to express my willingness to grasp the nettle which the principal dissent leaves untouched, and which the opinions supporting the court’s disposition consider untouchable — that is, flatly to deny that sleeping is or can ever be speech for First Amendment purposes. That this should seem a bold assertion is a commentary upon how far judicial and scholarly discussion of this basic constitutional guarantee has strayed from common and common-sense understanding.
I start from the premise that when the Constitution said “speech” it meant speech and not all forms of expression. Otherwise, it would have been unnecessary to address “freedom of the press” separately' — or, for that matter, “freedom of assembly,” which was obviously directed at facilitating expression. The effect of the speech and press guarantees is to provide special protection against all laws that impinge upon spoken or written communication (which I will, for the sake of simplicity, refer to generically as “speech”) even if they do so for purposes that have nothing to do with communication, such as the suppression of noise or the elimination of litter. But to extend equivalent protection against laws that affect actions which happen to be conducted for the purpose of “making a point” is to stretch the Constitution not only beyond its meaning but beyond reason, and beyond the capacity of any legal system to accommodate.
The cases find within the First Amendment some protection for “expressive conduct” apart from spoken and written thought. The nature and effect of that protection, however, is quite different from the guarantee of freedom of speech narrowly speaking. It involves a significantly different balancing of private rights and public interests, and does not always call for the detailed “First Amendment analysis” characteristic of the speech cases and applied by the majority opinions here. Specifically, what might be termed the more generalized guarantee of freedom of expression makes the communicative nature of conduct an inadequate basis for singling out that conduct for proscription. A law directed at the communicative nature of conduct must, like a law directed at speech itself, be justified by the substantial showing of need that the First Amendment requires. But a law proscribing conduct for a reason having nothing to do with its communicative character need only meet the ordinary minimal requirements of the equal protection clause.1 In other words, the only “First Amendment analysis”2 applicable to laws *56that do not directly or indirectly impede speech is the threshold inquiry of whether the purpose of the law is to suppress communication. If not, that is the end of the matter so far as First Amendment guarantees are concerned; if so, the court then proceeds to determine whether there is substantial justification for the proscription, just as it does in free-speech cases.
Thus, the First Amendment’s protection of free speech invalidates laws that happen to inhibit speech even though they are directed at some other activity (sound amplification,3 campaign contributions,4 littering 5). The more limited guarantee of freedom of expression, by contrast, does not apply to accidental intrusion upon expressiveness but only to purposeful restraint of expression. It would not invalidate a law generally prohibiting the extension of limbs from the windows of moving vehicles; it would invalidate a law prohibiting only the extension of clenched fists.
I believe the foregoing analysis is consistent with all of the Supreme Court’s holdings in this field. I would be content to consign marching and picketing, as the principal dissent suggests, to a category of traditionally expressive conduct which itself qualifies as speech, and thus does not require a showing of expression-suppressing intent. I do not think that exception is necessary, however, to explain the cases. The marching and picketing holdings represent not conduct protected because it is in itself expressive, but rather what the cases and commentators call “speech-plus”6 —conduct “intertwined”7 or “intermingled”8 with speech. The union organizer, for example, cannot convey his spoken or written message to the relevant audience if he is not allowed to be present at the entrance to the employer’s place of business. Those cases differ only in degree from the sound-amplification, campaign-contribution and littering cases referred to above; They deal with laws which, by prohibiting an essential concomitant of effective speech, infringe upon speech itself, and thus call forth the full First Amendment standard of justification.9 (It may be difficult to deter*57mine what particular conduct beyond the physical presence involved in the marching and picketing cases, or the distribution of literature involved in the littering case, is constitutionally deemed an essential concomitant of effective speech; but I consider it self-evident that on-site sleeping is not.)
It is only such cases as Stromberg v. California, 283 U.S. 359, 51 S.Ct. 532, 75 L.Ed. 1117 (1931) (flying of a red flag), Brown v. Louisiana, 383 U.S. 131, 86 S.Ct. 719, 15 L.Ed.2d 637 (1966) (silent sit-in), United States v. O’Brien, 391 U.S. 367, 88 S.Ct. 1673, 20 L.Ed.2d 672 (1968) (burning of a draft card), Tinker v. Des Moines School District, 393 U.S. 503, 89 S.Ct. 733, 21 L.Ed.2d 731 (1969) (black arm-bands), and Spence v. Washington, 418 U.S. 405, 94 S.Ct. 2727, 41 L.Ed.2d 842 (1974) (defacing the United States flag), that clearly present situations in which speech — that is, the spoken or written word — is not necessarily involved.10 The holdings of all these cases support the analysis set forth above. Every proscription of expressive conduct struck down by the Supreme Court was aimed precisely at the communicative effect of the conduct. The only reason to ban the flying of a red flag (Stromberg) was the revolutionary sentiment that symbol expressed.11 The only reason for applying the “breach of the peace” statute to the silent presence of black protestors in the library in Brown was the effect which the communicative content of that presence had upon onlookers.12 The only reason for singling out black armbands for a dress proscription (Tinker) was precisely their expressive content, allegedly *58causing classroom disruption.13 The only reason to prevent the attachment of symbols to the United States flag (Spence) was related to the communicative content of the flag.14 In O'Brien, on the other hand, where the Supreme Court upheld a ban on the destruction of draft cards, the law was not directed against the communicative nature of that activity.15
I do not suggest that the dicta of all the expressive conduct cases, as opposed to their holdings, support the distinction set forth above. Some of the opinions merely label the conduct “expressive” and proceed at once to application of First Amendment standards. Only O’Brien, however, really raises the question (though leaves it unanswered) 16 of what it is that avoids required application of those standards in every case. It is true that O’Brien appears to prescribe an inquiry, identical to that which I have described, as one of the four tests to be applied after it is determined that full First Amendment protections obtain. That would be inconsistent with my analysis if the O’Brien formulation were directed exclusively at “expressive conduct” cases — for a test triggered by the protection could hardly be the very test applied to determine whether the protection exists in the first place. In fact, however, the O’Brien discussion is directed at the tests to be applied in order to validate a statute impinging upon any activity protected by the First Amendment — not just expressive conduct, but also conduct “intertwined with speech,” and indeed even religiously motivated or associational conduct.17 For most of these categories the test would not be duplicative; it is only the governmental restriction of purely expressive conduct that escapes the necessity of First Amendment analysis if it is not aimed at repressing expression. This explanation is confirmed by the Supreme Court’s later per curiam opinion in Spence, which, in the context of expressive conduct, describes the inquiry into expression-suppressing purpose — as I have — as a test preliminary to the application of O’Brien’s four-step analysis.18
I find O’Brien supportive of my view since it shows the importance of statutory purpose in the Court’s thinking. The distinction here proposed is described explicitly in the following passage:
The ease at bar is therefore unlike one where the alleged governmental interest in regulating conduct arises in some measure because the communication allegedly integral to the conduct is itself thought to be harmful. In Stromberg v. California, 283 U.S. 359 [51 S.Ct. 532, 75 L.Ed. 1117] (1931), for example, this Court struck down a statutory phrase which punished people who expressed their “opposition to organized government” by displaying “any flag, badge, banner, or. device.” Since the statute there was aimed at suppressing communication it could not be sustained as a regulation of noncommunicative conduct.
*59391 U.S. at 382, 88 S.Ct. at 1681. To the same effect is the following statement in Buckley v. Valeo, 424 U.S. 1, 17, 96 S.Ct. 612, 633, 46 L.Ed.2d 659 (1976):
Even if the categorization of the expenditure of money as conduct were accepted, the limitations challenged here would not meet the O’Brien test because the governmental interests advanced in support of the Act involve “suppressing communication.” The interests served by the Act include restricting the voices of people and interest groups who have money to spend and reducing the overall scope of federal election campaigns. Although the Act does not focus on the ideas expressed by persons or groups subject to its regulations, it is aimed in part at equalizing the relative ability of all voters to affect electoral outcomes by placing a ceiling on expenditures for political expression by citizens and groups. Unlike O’Brien, where the Selective Service System’s administrative interest in the preservation of draft cards was wholly unrelated to their use as a means of communication, it is beyond dispute that the interest in regulating the alleged “conduct” of giving or spending money “arises in some measure because the communication allegedly integral to conduct is itself thought to be harmful.” 391 U.S. at 382, 88 S.Ct. at 1681.
The effect of the rule I think to be the law may be to permit the prohibition of some expressive conduct that might be desirable. Perhaps symbolic campsites19 or symbolic fire bases20 are a good idea. But it is not the function of the Constitution to make such fine judgments; nor is it within the practical power of the courts to apply them. There is a gap between what the Constitution requires and what perfect governance might sometimes suggest, in the area of expression as in other fields. So long as the Park Service is held to evenhanded application of its rules, I doubt that the political pressures generated in a representative democracy will tolerate the proscription of all expressive conduct, in Lafayette Park or anywhere else. The Park Service’s judgment will not be distorted, however — nor its time and ours consumed— in the mistaken pursuit of a supposed constitutional answer.
Where expressive conduct unrelated to speech is at issue, I think it worthwhile to engage in the preliminary step of analysis that separates conduct-prohibiting from expression-prohibiting laws and exempts the former from rigorous First Amendment scrutiny. The government argued in the present case, with some justification, that the posture in which it now finds itself— prohibiting sleep, but permitting all of the external manifestations of sleeping, including tents — is attributable to its efforts to comply with the directives of this court relating to the special justification needed to prohibit expressive conduct. See Women Strike for Peace v. Morton, 472 F.2d 1273 (D.C.Cir.1972). The Park Service has in effect been required to split each of its regulations into two: one that applies to people who are not engaging in the prohibited conduct for an expressive purpose, which can be enforced as written; the other that applies to demonstrators, which can be enforced only if supported by the substantial governmental interest that the First Amendment requires. That necessity may be unavoidable with regard to the relatively narrow range of conduct essential to effective speech. But to expand it to all conduct, even including sleep, seems to me unreasonable and unlikely to work. Park Service officers who have even less assurance of the proper application of the O’Brien four-part test than the various opinions of this court display will (against their sound administrative judgment) permit “symbolic” intrusions that need not be *60allowed; and the rule for demonstrators will inevitably (and perhaps rightly) tend to become the rule for the public at large — all with needlessly harmful effect upon the agreeability of our parks and public places. The unfortunate result is described by Justice Jackson’s statement in Saia v. New York, supra, 334 U.S. at 566, 68 S.Ct. at 1152, which I take the liberty of adapting to the facts of this case: “I dissent from this decision, which seems to me to endanger the great right of free speech by making it ridiculous and obnoxious, more than the Park Service regulation in question menaces free speech by proscribing sleep.”
. For a description of those requirements, see, e.g., City of New Orleans v. Dukes, 427 U.S. 297, 303, 96 S.Ct. 2513, 2516, 49 L.Ed.2d 511 (1976).
. I refer here only to the First Amendment’s guarantees of freedom of speech and press— not to other guarantees, such as freedom of religion or the right of personal autonomy or privacy which some cases have rested in part upon the First Amendment. See, e.g., Stanley *56v. Georgia, 394 U.S. 557, 564, 89 S.Ct. 1243, 1247, 22 L.Ed.2d 542 (1969); Griswold v. Connecticut, 381 U.S. 479, 482-83, 85 S.Ct. 1678, 1680-1681, 14 L.Ed.2d 510 (1965).
. See Saia v. New York, 334 U.S. 558, 561, 68 S.Ct. 1148, 1150, 92 L.Ed. 1574 (1948) (“Loudspeakers are today indispensable instruments of effective public speech”).
. See Buckley v. Valeo, 424 U.S. 1, 16, 96 S.Ct. 612, 633, 46 L.Ed.2d 659 (1976):
We cannot share the view that the present Act’s contribution and expenditure limitations are comparable to the restrictions on conduct upheld in O’Brien. The expenditure of money simply cannot be equated with such conduct as destruction of a draft card. Some forms of communication made possible by the giving and spending of money involve speech alone, some involve conduct primarily, and some involve a combination of the two. Yet this Court has never suggested that the dependence of a communication on the expenditure of money operates itself to introduce a nonspeech element or to reduce the exacting- scrutiny required by the First Amendment.
. See Schneider v. State, 308 U.S. 147, 163, 60 S.Ct. 146, 151, 84 L.Ed. 155 (1939):
It is argued that the circumstance that in the actual enforcement of the Milwaukee ordinance the distributor is arrested only if those who receive the literature throw it in the streets, renders it valid. But, even as thus construed, the ordinance cannot be enforced without unconstitutionally abridging the liberty of free speech. As we have pointed out, the public convenience in respect of cleanliness of the streets does not justify an exertion of the police power which invades the free communication of information and opinion secured by the Constitution.
. See American Radio Ass’n v. Mobile Steamship Ass’n, 419 U.S. 215, 231, 95 S.Ct. 409, 418, 42 L.Ed.2d 911 (1974); W. Lockhart, Y. Kamisar, & J. Choper, Constitutional Law 1136 (1980).
. See, e.g., Cameron v. Johnson, 390 U.S. 611, 617, 88 S.Ct. 1335, 1338, 20 L.Ed.2d 182 (1968); Cox v. Louisiana, 379 U.S. 559, 563, 85 S.Ct. 476, 480, 13 L.Ed.2d 487 (1965).
. See, e.g., Amalgamated Food Employees Union Local 590 v. Logan Valley Plaza, Inc., 391 U.S. 308, 313, 88 S.Ct. 1601, 1605, 20 L.Ed.2d 603 (1968).
. These cases would be compatible with the analysis I have set forth, even if they were to be regarded as involving not “speech-plus” but purely nonspeech expressive conduct. The picketing cases, for example, do not invalidate *57general prohibitions against walking back and forth, or against obstructing entrances, but rather banning such activities when engaged in for the (expressive) purpose of inducing people to refrain from trading or working. See, e.g., Thornhill v. Alabama, 310 U.S. 88, 91-92 [60 S.Ct. 736, 738-739, 84 L.Ed. 1093] (1940), where the statute forbade “[a]ny person ... [to] go near to or loiter about the premises or place of business of any other person ... for the purpose, or with the intent of influencing, or inducing other persons not to trade with, buy from, sell to, have business dealings with, or be employed by such persons .... ” See also Carlson v. California, 310 U.S. 106 [60 S.Ct. 746, 84 L.Ed. 1104] (1940). The marching cases typically turn upon the use of a vague ordinance for the very purpose of suppressing only expressive activity. See, e.g., Shuttlesworth v. City of Birmingham, 394 U.S. 147, 153, 89 S.Ct. 935, 940, 22 L.Ed.2d 162 (1969); Edwards v. South Carolina, 372 U.S. 229, 236, 83 S.Ct. 680, 683, 9 L.Ed.2d 697 (1963).
.In my view, the nude entertainment holdings do not deal with mere expressive conduct. Schad v. Borough of Mount Ephraim, 452 U.S. 61, 101 S.Ct. 2176, 68 L.Ed.2d 671 (1981), struck down the challenged ordinance on over-breadth grounds, since it included all live entertainment — including spoken entertainment. Southeastern Promotions, Ltd. v. Conrad, 420 U.S. 546, 95 S.Ct. 1239, 43 L.Ed.2d 448 (1975), involved a prohibition not of nudity alone, but of the entire stage production “Hair” because it included nudity. It stands for the well-established principle that a spoken or written work which has “serious artistic value” cannot be banned simply because it includes matter which, in isolation, might be proscribable. In California v. LaRue, 409 U.S. 109, 118, 93 S.Ct. 390, 397, 34 L.Ed.2d 342 (1972), the Court said that “at least some of the performances” covered by the regulation banning nudity and sexual acts “are within the limits of the constitutional protection of freedom of expression” (the case in any event upheld the regulation); and in Doran v. Salem Inn, Inc., 422 U.S. 922, 932, 95 S.Ct. 2561, 2568, 45 L.Ed.2d 648 (1975), it said that the nude barroom dancing might be protected “under some circumstances.” Both these cases may have had in mind only nudity in connection with a spoken or sung performance. In any case, I find it difficult to believe that exhibitory nudity will, on the ground that it is independently “communicative,” be accorded greater constitutional protection than the nondemonstrative sort, such as nude bathing, see, e.g., Chapin v. Town of Southampton, 457 F.Supp. 1170 (E.D.N.Y.1978). In other words, to the extent the nude entertainment cases speak to nudity apart from spoken or sung performances they seem to me based upon the “personal autonomy” rather than the “free speech” line of cases. See note 2, supra.
. The statute in Stromberg forbade the flying of “a red flag, banner or badge ... as a sign, symbol or emblem of opposition to organized government ....” 283 U.S. at 361, 51 S.Ct. at 532.
. “The statute was deliberately and purposefully applied solely to terminate the reasonable, orderly, and limited exercise of the right to protest the unconstitutional segregation of a public facility.” 383 U.S. at 142, 86 S.Ct. at 724.
. “The school officials banned and sought to punish petitioners for a silent, passive expression of opinion unaccompanied by any disorder or disturbance on the part of petitioners.” 393 U.S. at 508, 89 S.Ct. at 737.
. “If [Washington’s interest in preserving the national flag as an unalloyed symbol of our country] is valid, we note that it is directly related to expression in the context of activity like that undertaken by appellant. For that reason and because no other governmental interest unrelated to expression has been advanced or can be supported on this record, the four-step analysis of United States v. O’Brien ... is inapplicable.” 418 U.S. at 414 n. 8, 94 S.Ct. at 2732 n. 8 (citation omitted).
. “[B]oth the governmental interest and the operation of the 1965 Amendment [banning draft card burning] are limited to the noncommunicative aspect of O’Brien’s conduct.” 391 U.S. at 381-82, 88 S.Ct. at 1681-1682.
. “We cannot accept the view that an apparently limitless variety of conduct can be labeled ‘speech’ whenever the person engaging in the conduct intends thereby to express an idea. However, even on the assumption that the alleged communicative element in O’Brien’s conduct is sufficient to bring into play the First Amendment, it does not necessarily follow that the destruction of a registration certificate is constitutionally protected activity.” 391 U.S. at 376, 88 S.Ct. at 1678.
. See the cases cited at 391 U.S. at 376-77 nn. 22-27, 88 S.Ct. at 1678-1679 nn. 22-27.
. See note 14, supra.
. See Vietnam Veterans Against the War v. Morton, 506 F.2d 53 (D.C.Cir.1974) (per curiam).
. See Reply to Appellees’ Opposition to Appellants’ Emergency Motion for Injunction Pending Appeal and Opposition to Appellants’ Motion for Summary Affirmance at 46 (Dec. 14, 1982) (description of Vietnam Veterans’ May 1982 demonstration).