The panel has voted to deny the petition for rehearing and to reject the suggestion for rehearing en banc.

The full court has been advised of the suggestion for en banc rehearing, and an active judge called for a vote on whether to rehear the case en banc. A majority of the active judges did not vote for en banc consideration. Fed.R.App.P. 35(b).

The petition for rehearing is denied and the suggestion for a rehearing en banc is rejected.

NORRIS, Circuit Judge, dissenting from the court's rejection of the suggestion for rehearing en banc:

I dissent from the court's failure to rehear this case en banc because I believe the panel decision represents a radical departure from the settled jurisprudence of judicial deference to the policy choices of the political branches of government in regulating the economy. Disregarding the painful lessons of constitutional history--most noteworthy the lessons of the discredited Lochner era--the panel holds that a leasing policy regulating the tenant mix in a government-owned and managed shopping mall is so irrational that it violates the Equal Protection Clause of the Fourteenth Amendment. In my view, judges are rarely, if ever, justified in using their Article III power of judicial review to strike down such garden-variety economic regulation for failing to pass equal protection muster under the rational basis test. As the Supreme Court forcefully stated in City of New Orleans v. Dukes, 427 U.S. 297, 306, 96 S. Ct. 2513, 49 L. Ed. 2d 511 (1976) (per curiam), "Morey was the only case in the last half century to invalidate a wholly economic regulation solely on equal protection grounds, and we are now satisfied that the decision was erroneous."

In essence the panel repeats the mistakes of the Supreme Court in Morey and the Fifth Circuit in Dukes by arrogating to itself the power to sit as a "superlegislature" second-guessing the policy choices of the other branches of government. Dukes, 427 U.S. at 303, 96 S.Ct. at 2516. This mistake should not be masked by the fact that the challenged policy excludes religious organizations as a class from leasing commercial space in the Airport's terminals because the panel specifically factors the nature of the excluded class out of its equal protection analysis.1 The panel instead professes to apply the same sort of rationality review one would apply if, say, the Airport had a policy of not renting to candy stores.

The panel employs two techniques to reach the astonishing conclusion that the Airport's leasing policy fails the equal protection rational-basis test. First, in disposing of the argument that the leasing policy was justified by the Airport's expressed concern that leasing space to some but not all religious groups might raise establishment clause problems, the panel simply makes its own determination that no establishment clause problems would be created and that ergo the policy was irrational. Because the Airport's evaluation of the potential establishment clause problems constituted legislative factfinding, it is entitled to great judicial deference. Thus the relevant judicial inquiry is not whether in the court's opinion the old Airport practice of leasing only to the Christian Science Reading Room in fact violated the establishment clause, but whether an airport commissioner could rationally believe that the old practice raised establishment clause problems. Instead the panel's opinion stands as precedent for the remarkable proposition that legislative assessments of potential legal problems are reviewed de novo.

Second, the panel refuses to even consider the rational relationship between the Airport's new leasing policy and certain purposes asserted in the litigation that were not articulated when the policy was adopted. The panel rests this refusal on its conclusion that these purposes "could not" have been entertained by the Airport. See 792 F.2d at 124, amending 784 F.2d at 1013 & n. 2. However, this conclusion is based solely on the fact that these purposes were not mentioned in the letters and discussions that preceded adoption of the policy while the establishment clause concerns were. Thus, the panel effectively holds that if a legislative body clearly articulates one purpose at the time of the adoption of a challenged classification then a court can refuse to consider the rational relationship between the challenged classification and any purpose that had not been articulated at the time of adoption.

Not only do these erroneous holdings enable the panel to strike down a legislative classification that otherwise would have easily survived rationality review, but they also radically change the methodology of rationality review in this circuit. Unless the rational basis test is applied with great care and precision, rationality review will once again become a vehicle for judicial meddling. Because the panel's opinion warps the methodology of garden-variety rationality review, it threatens to undermine the deference to economic regulation that has been the hallmark of the post-Lochner era.

I. DE NOVO REVIEW OF LEGISLATIVE ASSESSMENTS OF POTENTIAL

The panel decides there was no rational relationship between the Airport's establishment clause concerns and the Airport's leasing policy without deferring in any way to the Airport's assessment that renting to some, but not all, religious groups raised establishment clause problems. Rather, the panel decides that the Airport's concerns bore no rational relationship to the new policy of not leasing commercial space to any religious organization by making its own determination that the Airport's old practice of leasing to the Christian Science Reading Room could not violate either the federal or state establishment clause. See 784 F.2d at 1013-16. Since it can find no establishment clause problems, the panel reasons that it was irrational for the Airport to seek to avoid them.2 Id. at 1016. This approach finds no support in the jurisprudence of rationality review. Rather than deciding de novo whether the old practice could, under any circumstances, violate either the establishment clauses of either the California or Federal constitutions, the panel should have asked whether an airport commissioner could rationally have thought that the old practice might create establishment clause problems under either constitution.

The Airport's determinations that establishment clause problems existed and could be eliminated by its new leasing policy are in essence findings of legislative facts. The panel does not deny that the new policy would be rationally related to the Airport's establishment clause concerns if the old practice in fact raised establishment clause problems. Rather the panel rejects the Airport's premise by deciding de novo that no establishment clause problem existed. Here the panel clearly errs. A wealth of Supreme Court cases establishes that courts should not independently determine legislative facts--including legislative facts such as whether a problem actually exists or whether the challenged classification will in fact have its desired effect of promoting the governmental purpose. Courts should ask only whether a reasonable decisionmaker could have thought a problem existed that would be solved by the classification at issue. The Supreme Court has, for example, explicitly stated that: "states are not required to convince the courts of the correctness of their legislative judgments. Rather, 'those challenging the legislative judgment must convince the court that the legislative facts on which the classification is apparently based could not reasonably be conceived to be true by the governmental decisionmaker.' " Minnesota v. Clover Leaf Creamery Co., 449 U.S. 456, 464, 101 S. Ct. 715, 724, 66 L. Ed. 2d 659 (1981) (quoting Vance v. Bradley, 440 U.S. 93, 111, 99 S. Ct. 939, 949, 59 L. Ed. 2d 171 (1979)); see also Allied Stores of Ohio v. Bowers, 358 U.S. 522, 528, 79 S. Ct. 437, 441, 3 L. Ed. 2d 480 (1959) (classification does not violate equal protection clause "if any state of facts reasonably can be conceived that would sustain it."). More recent Supreme Court opinions continue in the same vein: "An exemption such as that challenged here 'will be sustained if the legislature could have reasonably concluded that the challenged classification would promote a legitimate state purpose.' " Williams v. Vermont, 472 U.S. 14, 105 S. Ct. 2465, 2472, 86 L. Ed. 2d 11 (1985) (quoting Exxon Corp. v. Eagerton, 462 U.S. 176, 196, 103 S. Ct. 2296, 2308, 76 L. Ed. 2d 497 (1983)); see also Western & Southern L.I. Co. v. Board of Equalization, 451 U.S. 648, 668, 672, 101 S. Ct. 2070, 2083, 2085, 68 L. Ed. 2d 514 (1981).3

These Supreme Court cases differ from the present case in one respect--here the legislative factfinding involves an assessment of possible legal pitfalls. But this difference should not alter the level of deference due. Although one reason for judicial deference to legislative factfinding--the limited ability of courts to gather a broad spectrum of information and to make political judgments about how to use that information in rulemaking--is arguably not present, the other reasons for deference continue to hold true. Rulemaking bodies have every right to make their own assessments of what is and is not required of them by law and to make policy choices accordingly. Indeed, this sort of self-policing should be encouraged. Although courts must, of course, continue to decide de novo whether rulemaking bodies have in fact acted within the law, courts should be deferential in deciding whether the attempt to comply with the law was rational. Certainly a court's after-the-fact judgment that a rule change was not required by law does not make it irrational for a rulemaker to have made the change out of concern for violating the law. In short, I see no justification for striking down a legislative attempt to comply with the law solely on the basis of an independent judicial determination that, in hindsight, the attempt was unnecessary. Legislative bodies, after all, must be presumed to have some competence to assess the legal pitfalls of their actions, particularly in a case such as this one where a subdivision of the State of California, relying on the advice of a government lawyer, seeks to comply with state constitutional requirements.

Moreover, a decision by the Ninth Circuit that the old Airport practice of leasing to the Christian Science Reading Room created no establishment clause problems cannot lay the Airport's concerns to rest. The same issue could be raised again in future litigation brought by different parties, upon whom our court's decision would have no preclusive effect. Nor, in such litigation, would our decision be binding precedent on the United States Supreme Court or, with respect to the state constitution's establishment clause, on any state court. Thus, the Airport could still be held in violation of the state or federal establishment clause for failing to adopt the leasing policy that our court holds they were irrational in adopting. Cf. Edgar v. Mite, 457 U.S. 624, 651, 102 S. Ct. 2629, 2645, 73 L. Ed. 2d 269 (1982) (Stevens, J., concurring in part).

Finally, the panel reads the Airport's establishment clause concerns much too narrowly. The panel declares the establishment clause concerns unfounded because the prior practice of leasing space to the Christian Science Reading Room did not violate the establishment clause. See 784 F.2d at 1014-1016. But it is clear that the Airport was not only concerned that it might be violating the federal and state establishment clauses in leasing to the Reading Room. It was also concerned about "requests from other religious denominations and persons for space at the Airport." See Letter from Richard Heath, Director of Airports, to Donald Mayer (April 13, 1981). These requests, "combined with" the Airport's "overall shortage of rentable space," were part of the Airport's broader establishment clause concerns. Id. Read in this uncramped fashion, the Airport's establishment clause concerns seem much more powerful: if many different religious groups began vying for limited space, then the Airport would be forced to choose among religious groups if it leased to any. Although keeping the Reading Room as a tenant arguably may not itself offend establishment clause principles, a decision to lease space to some religious organizations but not to others would obviously raise much more serious constitutional questions. See Memorandum of Donald Garibaldi, Airports General Counsel, at p. 2 (April 10, 1981) (noting that "governmental action must be 'religiously neutral' " under the California Constitution). At the very least, the panel should have addressed the question of whether the Airport could have legitimately and rationally been motivated by a desire to avoid constitutional challenges by religious organizations whom the Airport might be unable to accommodate as tenants.

The San Francisco Airport Commission, a governmental body charged with the heavy responsibility of managing a major commercial enterprise, serving the traveling public, and conserving public resources, must constantly make a myriad of tough business decisions. Like the management of any such enterprise, the commissioners have limited time and resources available to make an exhaustive record of the concerns that may animate their decisions. If a court is indeed intent on considering only articulated concerns, then it should at least read those concerns in a manner that shows proper appreciation for the practical difficulties of requiring detailed articulation.

II. FAILURE TO CONSIDER UNARTICULATED PURPOSES

The airport's policy would have survived even the panel's de novo review of legislative factfinding involving assessment of establishment clause problems if the panel had not refused to consider whether two other purposes offered by the Airport were rationally related to its policy. The two purposes advanced at trial which the panel refused to consider--(1) maximizing revenue, and (2) serving the desires of the traveling public--were clearly legitimate. Nonetheless, starting from the assumption that its first step should be to "determine the policy's purpose," 784 F.2d at 1013, the panel bases its refusal to consider either of these two purposes on its finding that the "record fully supports" the fact that "the sole purpose" of the policy was to comply with the Establishment Clauses of the United States and California Constitutions. Id. In making this finding, the panel finds "dispositive" certain letters and discussions, id., which refer to the Airport's establishment clause concerns but did not mention the two other purposes advanced at trial, see id. at 1011. In essence, the panel holds that a court can use the mere articulation of one purpose by a governmental decisionmaker as its sole basis for refusing to consider other unarticulated purposes. This holding is not only wrong but also throws the law of this circuit on the methodology of rationality review into a state of confusion.

Because we are not a "superlegislature" that sits in judgment on "the wisdom or desirability of legislative policy," Dukes, 427 U.S. at 303, 96 S.Ct. at 2517, we must start with a strong presumption that governmental classifications do not violate the equal protection clause unless they burden a suspect class or a fundamental interest. Ordinarily this presumption will include a willingness to consider the rational relationship between any asserted purpose and a challenged classification. At most there are two exceptions to this presumptive willingness to consider any asserted purpose. One, a court may not need to consider asserted purposes that "could not" have been a purpose behind a challenged classification. This narrow exception, however, permits courts to refuse to consider unarticulated purposes only when they are inconsistent either with the articulated purposes or with the governmental classification on its face. See infra at 1471. Two, if a classification does not trigger scrutiny that is stricter than rationality review but nonetheless classifies people based upon sensitive (even though not quite suspect) personal characteristics, or approaches burdening fundamental rights, a court may ascertain the actual purposes behind the classification and make that a factor in deciding whether it may refuse to consider asserted purposes. See infra at 1472. My review of the jurisprudence of rationality review methodology demonstrates that the Airport's policy falls within neither of these two exceptions.

A. The Presumption Against Refusing to Consider Asserted Purposes.

A strong presumption against refusing to consider asserted purposes is fully supported by the case law as well as by the policies that animate deferential rationality review. A discussion of the case law must start with Fritz, in which the Supreme Court, after acknowledging that previous cases had been inconsistent, purported to "state the proper application of the [equal protection rational-basis] test." United States R.R. Retirement Bd. v. Fritz, 449 U.S. 166, 177 n. 10, 101 S. Ct. 453, 460 n. 10, 66 L. Ed. 2d 368 (1980). The Fritz Court, having decided that an asserted reason was rationally related to a challenged classification, emphatically stated: "Where, as here, there are plausible reasons for Congress' action, our inquiry is at an end. It is, of course, 'constitutionally irrelevant whether this reasoning in fact underlay the legislative decision,' Fleming v. Nestor, 363 U.S. , at 612 [80 S. Ct. 1367, at 1373, 4 L. Ed. 2d 1435 (1960) ], because this Court has never insisted that a legislative body articulate its reasons for enacting a statute." Fritz, 449 U.S. at 179, 101 S.Ct. at 461.

At a minimum Fritz definitively rejects two possible rationales for refusing to consider asserted purposes under garden-variety rationality review: (1) it rejects the idea that a court can refuse to consider a purpose simply because the purpose was not articulated by the governmental decisionmaker; and (2) it rejects the idea that a court can refuse to consider a purpose simply because the court makes a finding that it did not "actually" motivate the challenged classification. The rejection of both these rationales comports with our deferential role in reviewing social and economic legislation that does not burden a suspect class or fundamental interest. The nature of the legislative process involves compromise, trading votes, and voting with several purposes in mind--some conflicting and some of which a legislator may rightfully prefer not to articulate. We cannot fairly sit in judgment on the rationality of such a process by considering only those purposes that are reflected in the public record or by engaging in an artificial factfinding process of determining the "actual" purposes.

Moreover, the Fritz Court's rejection of both these rationales is firmly established in both previous and subsequent case law of the Supreme Court on the methodology of rationality review. The Court has several times rejected the notion that asserted purposes are ineligible for basic rationality review simply because the legislative history never articulates them. See, e.g., McDonald v. Board of Election Commissioners, 394 U.S. 802, 809, 89 S. Ct. 1404, 1408, 22 L. Ed. 2d 739 (1969); Allied Stores, 358 U.S. at 528, 79 S.Ct. at 441; cf. L. Tribe, American Constitutional Law, Sec. 16-3, at 996 (1978).4 Furthermore, the practice of the Court continues to be to consider asserted post hoc purposes even if they were not previously stated by the governmental decisionmaker. See, e.g., Bowen v. Owens, --- U.S. ----, 106 S. Ct. 1881, 1887, 90 L. Ed. 2d 316 (1986) (inferring a possible purpose), and id. 106 S.Ct. at 1888-89 (Marshall, J., dissenting) (pointing out that the legislative history never hints at the purpose considered by the majority); Williams, 105 S. Ct. at 2473-74 (considering, although rejecting, post hoc justifications that were not the stated purpose of the statute).5

Nor has the Supreme Court shown any willingness to ascertain actual purposes and reject all other purposes under garden-variety rationality review. Indeed, in countless cases the Supreme Court has endorsed the proposition that any "conceivable" purpose should be considered. See, e.g., Regan v. Taxation With Representation, 461 U.S. 540, 547-48, 103 S. Ct. 1997, 2001-02, 76 L. Ed. 2d 129 (1983); Clements v. Fashing, 457 U.S. 957, 963, 102 S. Ct. 2836, 2843, 73 L. Ed. 2d 508 (1982) (plurality opinion); Hodel v. Indiana, 452 U.S. 314, 332, 101 S. Ct. 2376, 2387, 69 L. Ed. 2d 40 (1981); McDonald, 394 U.S. at 809, 89 S.Ct. at 1408; see also L. Tribe, American Constitutional Law, Sec. 16.3, at 996 (1978).6 And in other cases the Court has pointedly ignored dissenting arguments that the purposes it is considering under rationality review are not actual purposes. See, e.g., Bowen, 106 S. Ct. at 1887 (ignoring dissent's argument that the considered purposes were not the actual ones); Kassel v. Consolidated Freightways Corp., 450 U.S. 662, 669-679, 101 S. Ct. 1309, 1315-1321, 67 L. Ed. 2d 580 (1981) (plurality) (ignoring the same sort of argument from a concurring opinion); id. at 702-705, 101 S.Ct. at 1332-1334 (dissent) (rejecting the same).7

B. The "Could Not" Exception.

The panel, however, attempts to escape from this general presumption by holding that the purposes of maximizing revenue and serving the traveling public need not be considered because the Airport "could not" reasonably be understood to have entertained them. See 792 F.2d at 124, amending 784 F.2d at 1013. This reliance on the exception for purposes that "could not" have been legislative purposes is misplaced. Worse, it threatens to open up enormous possibilities for judicial intrusion.

The "could not" exception emanates from Weinberger v. Wiensenfeld, 420 U.S. 636, 95 S. Ct. 1225, 43 L. Ed. 2d 514 (1975), in which the Court stated that: "the mere recitation [in court] of a benign, compensatory purpose is not an automatic shield which protects against any inquiry into the actual purposes underlying a statutory scheme.... This Court need not in equal protection cases accept at face value assertions of legislative purposes, when an examination of the legislative scheme and its history demonstrates that the asserted purpose could not have been a goal of the legislation." 420 U.S. at 648 & n. 16, 95 S. Ct. at 1233 & n. 16 (the second quoted sentence footnoted the first).8 On its face, this language creates a very narrow exception to the presumption in favor of considering all asserted purposes--excluding from consideration only those purposes that are at odds with the statute or legislative history. Yet the panel seizes upon this language to create a new and expansive exception. The panel holds that the asserted purposes "could not" have been purposes behind the Airport's new leasing policy on the sole basis that that the Airport articulated another purpose without mentioning the asserted purposes. In effect, the panel rules that whenever the legislative history articulates purposes, a court should consider only those articulated purposes under the rational-basis test. Such an expansive reading of the "could not" exception creates a new opening for judicial second-guessing of legislative judgments.9

A much more reasonable interpretation of the "could not" exception is that a court can infer that an asserted purpose "could not" have been a legislative purpose only if the asserted purpose is inconsistent either with the purposes articulated in the legislative history or with the legislation on its face. Here the panel makes no claim that there is any inconsistency between the articulated purpose and the asserted purposes. And indeed there is no inconsistency between the Airport's articulated interest in avoiding potential establishment problems and its asserted interests in maximizing revenue and serving the traveling public. Nor, obviously, is it otherwise inconceivable that an airport would be interested in maximizing revenue and serving the traveling public.

Admittedly, the Court has yet to expressly state that inconsistency is the core of the "could not" test. Nonetheless, so far the Wiesenfeld line of cases has in fact used the articulation of purposes as a basis for refusing to consider other unarticulated purposes only where the articulated purpose (or statutory scheme) is inconsistent with the unarticulated purpose. See Lucas, 427 U.S. at 508 n. 14, 96 S. Ct. at 2763 n. 14 (rejecting an argument that it should not consider an asserted purpose where the legislative history did not "belie" the purpose); Wiesenfeld, 420 U.S. at 648-53, 95 S.Ct. at 1233-36; id. at 655, 95 S.Ct. at 1236 (Rehnquist, J., concurring in the result) (rejecting the asserted purpose because it is "totally at odds with the context and history of" the challenged statute); Fritz, 449 U.S. at 186, 101 S.Ct. at 465 (Brennan, J., dissenting) (arguing that asserted purpose should be rejected because it "directly conflicts with Congress' stated purpose."). The Court regularly has upheld governmental action based upon unarticulated purposes even when other purposes are articulated. Moreover, in several cases, the Court has declined to follow the suggestions of dissenting or concurring opinions that it should make adverse inferences simply from the fact that a legislative history exists but fails to articulate the asserted purpose. See Bowen v. Owens, 106 S. Ct. at 1889 (Marshall, J., dissenting); Kassel, 450 U.S. at 681-83 & n. 3, 101 S. Ct. at 1321-23 & n. 3 (Brennan, J., dissenting); Michael M., 450 U.S. at 494-96, 101 S.Ct. at 1216-18 (Brennan, J., dissenting); Wilson, 450 U.S. at 244-45, 101 S.Ct. at 1087-89 (Powell, J., dissenting).

Further, a consideration of the policy interests at stake demonstrates that inconsistency with articulated purposes should be the core element of the "could not" test. The issue of which grounds a court can permissibly rely upon to infer that an asserted purpose "could not" have been a legislative purpose must take into account the strong policies against requiring legislative articulation, see Fritz, 449 U.S. at 179, 101 S.Ct. at 461; Bowers, 358 U.S. at 528-29, 79 S.Ct. at 441-42; ante at 1473 & n. 9, and against delving into legislative motivation, see Michael M., 450 U.S. at 469-70, 101 S.Ct. at 1204-05 (plurality) (citing cases); ante at 1473 & n. 8. Moreover, in every rationality review case, a court is required to indulge the government with a presumption that it had a constitutional motivation. See McDonald, 394 U.S. at 809, 89 S.Ct. at 1408. These countervailing policies necessarily limit the inferences courts can draw from unclear legislative histories and must be borne in mind when evaluating the permissible grounds for determining what "could not" have been a governmental motivation.

On the one extreme, if the preamble to a statute specifically stated that X was not the purpose of the statute, there seems little reason for a court to consider purpose X under its rationality review, even if some other legislature could have rationally been motivated by that purpose. The countervailing policies of the Fritz line of cases would not be adversely affected. The refusal to consider purposes eschewed by the legislature would not force articulation since it in no way penalizes nonarticulation. The examination of intent in such a case seems unproblematic since the legislature has itself expressed its intent. Finally, any presumption in favor of the legislature has clearly been rebutted by its own statements. At the other extreme, a court cannot infer that a purpose could not have motivated the classification simply because it had never been articulated. See ante at 1473. In between these two extremes is the question of whether a court can infer that an asserted purpose could not have been an actual purpose when the legislative history articulates purposes but not the purpose asserted.

It is this intermediate case that we face today. For the following reasons, the caselaw limiting the "could not" exception to cases where the the articulated purposes are inconsistent with the articulated purposes minimizes the encroachment of the "could not" exception on the countervailing policies against forcing articulation, against delving into legislative motive, and for presuming that legislative action is constitutionally rational. First, the refusal to consider purposes inconsistent with articulated purposes will rarely penalize nonarticulation. Second, the examination of intent in such a case seems relatively unproblematic. Finally, the presumption in favor of the legislature will have been largely rebutted by its own statements.

This methodology for screening asserted purposes under rationality review also comports best with the various policies behind rationality review deference and with an understanding of governmental decisionmaking. The panel's methodology in essence takes at face value those purposes specifically mentioned on the public record at the time a policy is adopted or legislation is passed. But to my mind, attributing to the decisionmaker only those purposes that are memorialized in the public record reflects a simplistic view of the legislative process. Policymakers are not judges, expected to set out the reasons for their decisions by identifying the applicable legal principles that dictate the result. Individual legislators--or in this case commissioners--may choose, for whatever reason, not to make a public record of their actual purposes in voting for a particular proposal. Moreover, even if an individual commissioner or legislator is willing to make a public record of one of her purposes, often a commissioner or legislator must act for a variety of reasons, some of which she would rather not express. It may, for example, be poor politics to inform the Reading Room that its lease is terminated because the traveling public would prefer to play video games. Surely it is not for us to require that every public official state all her reasons for supporting legislative action. The equal protection clause is not a warrant for forcing legislators or airport commissioners to live in a fishbowl, with every motive for every action out in the open.

Nor is the equal protection clause a warrant for eliminating the legislative practice of compromise and votetrading. Whereas compromise and vote trading have no place in the judicial process, they are an expected and necessary aspect of the legislative process. We therefore cannot always expect to find a detailed public record setting forth all the reasons behind legislative action. We should not draw any inference that an asserted purpose "could not" have been a purpose behind a governmental classification simply because other purposes for the classification were set out in the public record. As long as the legislative history does not either specifically deny that the legislative body was motivated by an asserted purpose or offer other purposes that are so inconsistent with the asserted purpose that we may safely conclude that the presumption is rebutted that some individual decisionmaker had the purpose in mind, then the asserted purpose must be considered eligible for rationality review.

C. The Exception for Sensitive Classifications.

The Supreme Court, without ever spelling it out explicitly, seems to also recognize an exception to the general presumption when reviewing the rationality of sensitive classifications. Specifically, it has shown a greater willingness to indulge in the ascertainment of actual purposes under rationality review when the challenged classification is--instead of a garden-variety economic or social regulatory classification--a classification based upon sensitive personal characteristics or a classification that approaches burdening a fundamental right. By definition, because rationality review applies, these are cases that do not sufficiently burden a suspect class or a fundamental interest to trigger stricter scrutiny. Nonetheless, the willingness to ascertain actual purposes in these cases seems sensible since such classifications would regularly raise concerns that the actual purposes behind them might be illegitimate or irrational, and yet the potential discriminatees would not be protected as suspect classes or under the rubric of fundamental rights. By screening out post hoc rationalizations in these special cases, the Court is better able to isolate and evaluate the legitimacy and rationality of the motivations actually behind the more sensitive classifications.

Thus, it seems perfectly appropriate for courts to vigorously screen asserted purposes behind sensitive classifications such as those based on gender, legitimacy of birth, sexual orientation, alienage, mental retardation or state of residency even though the classifications are reviewed under the rational basis test.10 In fact, in many cases involving challenges to such sensitive classifications the Supreme Court has rejected asserted purposes based on its finding that the asserted purposes did not actually motivate the relevant decisionmaker. See, e.g., Hogan, 458 U.S. at 728-730, 102 S.Ct. at 3338-3339 (gender); Wiesenfeld, 420 U.S. at 648-653, 655, 95 S. Ct. at 1233-1236, 1237 (gender); Trimble v. Gordon, 430 U.S. 762, 775-76, 97 S. Ct. 1459, 1467-68, 52 L. Ed. 2d 31 (1977) (illegitimacy); cf. City of Cleburne v. Cleburne Living Center, 473 U.S. 432, 105 S. Ct. 3249, 3258-3260, 87 L. Ed. 2d 313 (1985) (reviewing the rationality of a classification based on mental retardation in a manner that seemed designed to weed out post hoc purposes in part because the Court suspected that the actual purpose was "irrational prejudice against the mentally retarded"); see also id., 105 S.Ct. at 3262-63 (Stevens, J., concurring) (same). In other such cases, the Court seems careful to establish that an asserted purpose was an actual purpose of a classification before considering it. See, e.g., Michael M., 450 U.S. at 470, 101 S.Ct. at 1204 (plurality opinion of Rehnquist, J., joined by Burger, C.J., and Stewart & Powell, JJ.) (gender); Hughes v. Alexandria Scrap Corp., 426 U.S. 794, 813 n. 23, 96 S. Ct. 2488, 2499 n. 23, 49 L. Ed. 2d 220 (1976) (state residency); cf. Hampton v. Mow Sun Wong, 426 U.S. 88, 103, 96 S. Ct. 1895, 1905, 48 L. Ed. 2d 495 (1976) (alienage).

Similarly, it would seem perfectly appropriate for courts to vigorously screen asserted purposes behind a classification that comes close to burdening fundamental rights. And, although the case law is rather sketchy, it appears that the Supreme Court has been willing to ascertain actual purposes under rationality review where fundamental interests are in the background. For example, in Eisenstadt v. Baird, 405 U.S. 438, 92 S. Ct. 1029, 31 L. Ed. 2d 349 (1972), the Court rejected asserted reasons for a statute that allowed only married people to obtain contraceptives with an analysis that, although couched in terms of rationality, seemed to be based on the Court's conclusion that the purposes did not actually motivate the state. Id. at 443, 447-50, 92 S. Ct. at 1033, 1035-37. And in McGinnis v. Royster, 410 U.S. 263, 93 S. Ct. 1055, 35 L. Ed. 2d 282 (1973), the Court, in upholding a parole classification that discriminated between prisoners jailed before conviction and those that had been free on bail, stated that the asserted purpose had to be "a purpose," "nonillusory," and could not be an "imaginary basis or purpose." Id. at 274-77, 93 S.Ct. at 1061-63.

The panel, however, makes no attempt to rely on an exception for sensitive classifications. Indeed, the panel expressly declines to discuss whether all religious organizations as a group constitute a suspect or quasi-suspect class or whether the Airport's policy infringes rights under the free exercise clause. See 784 F.2d at 1012-13 & n. 1. Instead, the panel professes to be employing garden-variety rationality review. Its opinion thus sets a bad precedent that could generally affect rationality review of nonsensitive classifications.

Moreover, the policies behind a sensitive classifications exception to the general presumption against refusing to consider asserted purposes do not support extending the exception to this case. The more probing look into actual purposes permitted under a sensitive classifications exception is appropriate only in reviewing classifications that regularly raise concerns that the actual purposes are illegitimate or irrational and for which there is no independent doctrinal basis permitting closer judicial review. It is inappropriate here where the case law on the free exercise and establishment clauses already stands as a fully elaborated doctrinal restraint on discriminatory governmental action. Indeed, allowing a court to resort to such probing inquiry in these circumstances means allowing a court to evade the careful balance the Supreme Court has tried to strike in its cases between free exercise and establishment clause concerns. With the case stripped of the question of whether the Board's classification burdens the right to free exercise of religion, the classification should be evaluated as a purely business judgment about leasing commercial space in a government-owned airport. Cf. Johnson v. Robison, 415 U.S. 361, 375 n. 14, 94 S. Ct. 1160, 1169 n. 14, 39 L. Ed. 2d 389 (1974) ("since we hold ... that the Act does not violate appellee's right of free exercise of religion, we have no occasion to apply to the challenged justification a standard of scrutiny stricter than the traditional rational-basis test.").

One would have thought that by now the law concerning rationality review methodology would be clear. Instead, the case law reveals surprising intricacy and confusion. This state of affairs, however, is all the more reason why our court should have taken this case en banc. Because every federal and state law that classifies persons is potentially subject to equal protection rationality review, confusion about the appropriate methodology for such review poses serious problems. For an issue this important, it behooves us to at least try to straighten out the law of our own circuit.

That this case demonstrates an intracircuit confusion over the appropriate methodology for rationality review seems indisputable. Our previous cases have manifested a broad willingness to consider any asserted governmental purpose. See Bunyan v. Camacho, 770 F.2d 773, 774 (9th Cir.1985) ("We may hypothesize the possible legislative purposes for enacting the [challenged] statute, and must uphold it if any legitimate purpose is served."), cert. denied --- U.S. ----, 106 S. Ct. 3271, 91 L. Ed. 2d 562 (1986); Hoffman, 767 F.2d at 1437; In re Lara, 731 F.2d 1455, 1460 n. 7 (9th Cir.1984) (not limited to articulated purposes); Brandwein, 708 F.2d at 1470-71; Lamb v. Scripps College, 627 F.2d 1015, 1021 n. 9 (9th Cir.1980). In fact, even cases decided after this one seem to conflict with it. See Jackson Water Works v. Public Util. Comm'n, 793 F.2d 1090, 1096-97 (9th Cir.1986) (stating that a court can look at any conceivable reason regardless of whether it actually underlay the legislative action; "[c]lassifications are set aside ... only if no grounds can be conceived to justify them.").

Moreover, the panel's decision seems to place our circuit in conflict with other circuits as well. The Fifth Circuit reads the Supreme Court's case law as implicitly accepting the principles that any "conceivable natural basis" can be looked at and that "the legislature or other governing body need not express all its purposes when it acts and that the defendant in a constitutional case need not prove the legislative purpose as a historical fact." Shelton v. City of College Station, 780 F.2d 475, 484 (5th Cir.), cert. denied --- U.S. ----, 106 S. Ct. 3276, 91 L. Ed. 2d 566 and --- U.S. ----, 107 S. Ct. 89, 93 L. Ed. 2d 41 (1986). Arguably the panel's decision also places our circuit in conflict with the Third Circuit because, although the panel adopted the language of the screening test enunciated in a Third Circuit case, in that case the Third Circuit allowed itself to consider post hoc rationales because it was "reluctant to hold that the [classification] challenged here can survive equal protection scrutiny only if shown to further some legitimate goal actually set forth in the legislative history." Delaware River Basin Commission v. Bucks County, 641 F.2d 1087, 1096 (3d Cir.1981).

In short, our court allows to stand an opinion that further muddles the already confused case law on the methodology of rationality review. Because I believe that the panel's methodology is fundamentally mistaken, and because considerable confusion exists both within and without this circuit over the proper methodology for rationality review, I dissent from this court's refusal to rehear this case en banc.