Continental Airlines, Inc. and Continental Express, Inc. (collectively "Continental"), brought this antitrust action in April 2000 to challenge the installation of templates, which limit the size of carryon baggage, at Dulles Airport. Continental alleges that in agreeing to install the templates, United Air Lines, Inc. ("United"), the primary air carrier at Dulles, and the Dulles Airport Airline Management Council ("AMC"), an unincorporated association of all airlines serving Dulles, unreasonably restrained trade in violation of Section 1 of the Sherman Act, 15 U.S.C. S 1 (1997). Applying an abbreviated "quick-look" analysis, the district court granted summary judgment to Continental, awarded Continental $254,426.85 in trebled damages, and permanently enjoined the use of templates at Dulles. See Continental Airlines, Inc. v. United Air Lines, Inc., 126 F. Supp. 2d 962 (E.D. Va. 2001) (granting summary judgment) (hereafter Continental); Continental Airlines, Inc. v. United Air Lines, Inc., 136 F. Supp. 2d 542 (E.D. Va. 2001) (granting treble damages and an injunction) (hereafter Continental II). Because issues of material fact remain disputed and because both the unique architectural configuration of Dulles Airport and the competitive effects of templates require thorough consideration in a less "quick look," we vacate the judgment of the district court and remand for further proceedings.

Before setting forth the factual background giving rise to this case, we take note of the tragic events of September 11, 2001. Those events and their ramifications have not yet mooted any aspect of this case; although the Federal Aviation Administration ("FAA") has now restricted the number of carry-on bags permitted to each passenger, to date it has not limited their size. Yet we recognize that the tragedies of September 11 now affect any discussion of air travel. Their implications for security1 cannot be overlooked and, on remand, the district court and the parties will undoubtedly have to deal with these issues in considering any prospective relief. Having said that, we turn to the world before September 11, and the facts underlying this case.

In the mid-1990s, an increased volume of air travel and a growing desire to avoid checking luggage meant passengers carried more and more baggage onto commercial flights. "[M]any aircraft did not have sufficient under-seat or overhead bin storage space to stow securely the number and size of bags brought onboard by the increasing number of passengers." Continental, 126 F. Supp. 2d at 965. The Association of Flight Attendants, joined by United and some other airlines, lobbied Congress and the FAA to limit the size of carry-on baggage. Those efforts have failed; instead the FAA simply requires each airline to scan carry-on baggage to "control the size and amount carried on board in accordance with an approved carry-on baggage program in [the airline's] operations specifications." 14 C.F.R. S 121.589(a) (2001). In compliance with this regulation, each airline has published an FAA tariff stating numerical and size limitations on carry-on luggage that it may enforce.

The increased volume of carry-on baggage has been linked to several problems: delays, inconvenience to late boarding passengers from full overhead bins, conflict between airline staff and passengers, and reduced on-board safety if baggage cannot be properly stowed. Carry On Baggage Program, 52 Fed. Reg. 21,472, at 21,475 (June 5, 1987) (codified at 14 C.F.R. pt. 121).2 Airlines have explored various ways to respond to the increase in carry-on baggage, including one critical to this lawsuit -the use of baggage templates.

Baggage templates are pieces of plastic or stainless steel, mounted on hinges, that cover the mouths of x-ray baggage screening machines. When the templates are down, they narrow the mouth of the x-ray machine, preventing oversize baggage from passing through. Generally, even when they are down, the templates permit all baggage sized within the published FAA tariff of any airline to pass through. Templates can be flipped upward to allow authorized larger baggage to bypass the templates. Absent this authorization, when baggage does not fit through the template, it must be checked.

In 1997 Continental experimented with templates at its hub in Houston. Frequent fliers in a focus group tested by Continental found the templates annoying, and templates could not be used in at least one of Continental's hubs because another carrier objected. Thus, in 1998 Continental decided not to install templates, opting instead to expand its overhead bins and provide for more equipment and employees to permit passengers to carry on or gate-check as much luggage as they wished.

In May 1999, a Continental customer advisory board urged the airline to "[e]ncourage people to check bags" and complained of early boarding passengers usurping all carry-on baggage space ("People in row 25 are putting bags in overhead bins in row 7."). The majority of the group stated their belief "that a baggage template is a good idea." Similarly, Continental's Senior Vice President for Airport Services agreed in deposition that templates could reduce or eliminate monopolization of space by early-boarding passengers, flight delays caused by the need to find space for carry-on baggage, conflict between airline crew and passengers over carry-ons, and hazards to crew taking excess carry-on baggage down the jetway to the aircraft.

Nevertheless, Continental rejected templates and other ways to reduce carry-on baggage. Instead, it focused on providing more space for carry-on baggage, and on increasing its gate-checking services, so that when space on board had been exhausted, additional carry-on baggage could be checked at the gate (or at the side of smaller planes) without delaying flights. Continental has received press attention as well as a number of awards, which it attributes to its excellent customer service, including its liberal carry-on baggage policy. It also maintains that this policy has attracted many high-revenue customers.

United has also expanded its overhead bins to meet the problems of increased carry-on baggage. (In fact, according to Continental's Manager for Product Marketing, United began a program of bin expansion before "Continental followed as a competitive response.") In addition, however, United has sought to impose a strict limit on carry-on baggage, restricting each passenger to two carry-on bags of limited size. Finding it difficult to enforce the size limitation because of the subjectivity of the process, United looked for a more objective way to measure size and decided to try baggage templates. United expected many benefits from the use of templates, including a smoother boarding process, better on-time performance, increased passenger comfort, and greater on-board safety.

United initially experimented with templates in mid-1998 at its Los Angeles and Chicago hubs, with mixed results. The test reduced delays by up to 65% in Los Angeles and up to 72% in Chicago. However, United received complaints from high-revenue customers, who threatened to move their business, and an internal United study determined that during the 1998 calendar year, while its "template program ha[d] helped improve carry-on perceptions compared to American and US Airways . . . Continental's larger bins ha [d] provided a solution with more customer value." Nevertheless, United management decided to proceed with a plan to address the problems associated with carry-on baggage by consistently enforcing the number and size restrictions of its carry-on policy, educating consumers as to the benefits of the policy combined with larger overhead bins, and deploying templates wherever possible. (Airlines other than United and Continental, including American, Southwest, and Delta, also considered templates, and a number, like United, adopted them.) At present United has installed templates at security checkpoints used by its passengers in at least twenty-five domestic airports.

In late 1998, United sought to install templates at Dulles, a United hub and one of the fastest growing airports in the country. This presented a special problem because the configuration of Dulles is, as Continental's Chief Executive Officer testified, "unique." Most airports have a number of security checkpoints, so that at most only a few airlines share any given security checkpoint. By contrast, at Dulles every non-connecting passenger for every flight must be screened at one of two common security checkpoints located in the main terminal, which in turn lead to a common "sterile" area. Once in the sterile area, passengers take shuttles to the midfield concourse from which their flights depart. The combination of security checkpoints that must be shared by all airlines, a single sterile area also common to all, and the use of mobile shuttles, distinguishes Dulles from every other major airport in the United States.

The twenty-nine airlines that serve Dulles are members of the AMC, an unincorporated association mandated by the FAA.3 The AMC holds monthly meetings to resolve operating issues that necessarily affect more than one carrier at Dulles. For example, federal law charges all airlines with the responsibility for airport security. 14 C.F.R. S 108.5-.9 (2001). The AMC is thus responsible for the installation, maintenance, and management of security checkpoints at Dulles. As a member of the AMC, each airline receives one vote, regardless of size. Thus, United, with 38 of the 120 gates (making it the largest carrier at Dulles), and Continental, with one gate, each have one vote on the AMC.4

After some deliberation, the AMC agreed to allow United to install templates at the east checkpoint, through which most of United's passengers pass, for a one-month trial period beginning December 30, 1998. The west checkpoint became congested because passengers circumvented the templates by avoiding the east checkpoint. Once the trial period ended, the Chairman of the AMC asked United to remove the templates. United was in the middle of changing station managers and increasing service at Dulles and did not remove the templates quickly. Irritated, the AMC hired a contractor to remove the templates from the east checkpoint in June 1999; United paid for the removal.

Later in 1999, a new United station manager formally asked the AMC to consider reinstalling baggage templates, at least at the east checkpoint. After extensive discussion, the AMC agreed that it would survey the twenty-six carriers eligible to vote on the issue to determine whether they favored templates and to identify any specific areas of concern. Of the carriers responding to the survey, fifteen, including United, favored the templates and five, including Continental, opposed the templates. The survey form instructed that failure to return the ballot would be deemed a vote in favor of templates; in accord with this policy, six airlines that did not respond were included with the fifteen voting in favor of templates to make the final vote 21-5.

One concern that surfaced both in the survey results and in subsequent AMC discussions was the desire not to repeat the mistakes of the prior tests by installing templates at one checkpoint but not the other. Deposition testimony revealed that no one participating in the AMC template discussions, including Continental's representative and those of other carriers that did not favor the installation of templates, perceived that any member of the AMC voting to install templates at Dulles had an anticompetitive purpose in doing so.

The airlines met to discuss implementation of the template program. Several airlines expressed concern that high-revenue passengers not be limited by templates. The AMC focused on an alternative system of tokens, called "medallions," that airlines could distribute to their passengers, who would present the medallions to have the templates lifted for their bags. The AMC adopted the medallion option as part of the template implementation. All AMC members were asked to estimate the number of medallions they required and an initial distribution of medallions was made in proportion to each airline's first-class, business-class, and full coach fare passengers. The templates were installed at Dulles on April 15, 2000.

Shortly thereafter, Continental filed this action, asserting that the Dulles template program violated Section 1 of the Sherman Act. See 15 U.S.C.A. S 1. After filing suit, Continental distributed medallions to its customers until it ran out of its initial supply. Thereafter, it made no attempt to obtain more medallions. Instead, Continental had its employees (and later contractual workers hired by Continental at a cost of approximately $10,000 a month) lift the templates for Continental's passengers who wished to carry on larger bags. The employees also distributed literature notifying customers that Continental was not responsible for the "problems caused" by the template restriction, attributing the decision to install templates at Dulles solely to United (not mentioning the majority vote of the AMC), and relating that Continental was "seeking remedies to the problems caused by United[ ]." Continental II, 136 F. Supp. 2d at 548 n.7.

Nonetheless, the record reveals few complaints about the Dulles templates. The Metropolitan Washington Airport Authority received no complaints about the templates and did not believe passengers found them confusing. Similarly, no complaints about the templates were lodged with the AMC. United received one written complaint and Continental fewer than five. Northwest Airlines, which had opposed installation of the templates, received no complaints and found the template program worked so well that it ultimately found distribution of medallions to be unnecessary.

Continental's station manager testified at deposition that Continental was "able to get all customers [with oversized bags] through the security checkpoint [by lifting the templates] with or without" the contract workers. Continental did not reduce flights or reassign planes as a result of the templates. Continental does not claim that the Dulles templates or the costs of Continental's method of dealing with them affected any ticket price.

The district court granted summary judgment to Continental. The court ruled that the template program constituted a horizontal restraint on output with "manifest" anticompetitive impact, and so required no further proof of anticompetitive effects. Continental, 126 F. Supp. 2d at 977. Although antitrust law would ordinarily condemn joint restraints on output as illegal per se, because airlines must cooperate at airports, the district court determined to employ a "quick-look" analysis. Id. at 976, 978. Accordingly, the court examined each asserted procompetitive justification for the template program -safety, timely departures, and passenger comfort. Id. The court rejected all three, finding "no record support" for the proposition that the template program "promote[d] competition among carriers" in any way. Id. at 979. Having found anticompetitive impact with no redeeming procompetitive justifications, and thus having found an antitrust violation, the district court ruled that Continental had suffered antitrust injury. Id. at 982-84. After another hearing, the court awarded Continental $254,426.85 in trebled damages for its costs in hiring employees to lift the templates, and enjoined United and the AMC from using templates at Dulles. Continental II, 136 F. Supp. 2d at 552-53.

We review a district court's award of summary judgment de novo. M & M Med. Supplies & Serv., Inc. v. Pleasant Valley Hosp., 981 F.2d 160, 163 (4th Cir. 1992). "The evidence of the nonmovant[s] is to be believed, and all justifiable inferences are to be drawn in [their] favor." Anderson v. Liberty Lobby, 477 U.S. 242, 255 (1986). If after examining the entire record, in light of the controlling legal principles, a reviewing court concludes that material facts remain genuinely disputed, it must hold the grant of summary judgment improper. See, e.g., Eastman Kodak Co. v. Image Tech. Servs., Inc., 504 U.S. 451, 477-78 (1992). We turn now to those controlling legal principles and then to their application to the above facts.5

Section 1 of the Sherman Act prohibits "[e]very . . . combination in restraint of trade." 15 U.S.C.A. S 1. The Supreme Court, however, long ago established that Section 1 only outlaws restraints that are "unreasonably restrictive of competitive conditions." Standard Oil Co. v. United States, 221 U.S. 1, 58 (1911). Thus, to prevail on a claim that a horizontal restraint, an agreement among competitors, violates Section 1, a plaintiff must prove that the restraint is unreasonable. A plaintiff cannot prove the unreasonableness of a restraint merely by showing that it caused economic injury. Rather, because "[t]he antitrust laws were enacted for the protection of competition, not competitors," a plaintiff must show that the net effect of a challenged restraint is harmful to competition. Atlantic Richfield Co. v. USA Petroleum Co., 495 U.S. 328, 338 (1990) (citation and emphasis omitted). A proven antitrust violation is a necessary predicate to recovery of antitrust damages. Lee-Moore Oil Co. v. Union Oil Co., 599 F.2d 1299, 1306 (4th Cir. 1979); see also Atlantic Richfield, 495 U.S. at 339-41; Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 489 (1977).

Having proven an antitrust violation, an antitrust plaintiff seeking damages must show an injury "of the type the antitrust laws were intended to prevent and that flows from that which makes defendants' acts unlawful." Id. Lost profits and the costs of finding alternatives to mitigate the damages caused by an antitrust violation may constitute antitrust injury. Lee-Moore Oil Co., 599 F.2d at 1304-06.

In determining whether a plaintiff has proved that a horizontal agreement violates Section 1, the Supreme Court has authorized three methods of analysis: (1) per se analysis, for obviously anticompetitive restraints, (2) quick-look analysis, for those with some procompetitive justification, and (3) the full "rule of reason," for restraints whose net impact on competition is particularly difficult to determine. The boundaries between these levels of analysis are fluid; "there is generally no categorical line to be drawn between restraints that give rise to an intuitively obvious inference of anticompetitive effect and those that call for more detailed treatment." California Dental Ass'n v. FTC, 526 U.S. 756, 780-81 (1999). Instead, the three methods are best viewed as a continuum, on which the "amount and range of information needed" to evaluate a restraint varies depending on how "highly suspicious" and how "unique" the restraint is. See 11 Herbert Hovenkamp, Antitrust Law P 1911a (1998); see also California Dental, 526 U.S. at 779-81. In all cases, however, "the criterion to be used in judging the validity of a restraint on trade is its impact on competition." NCAA v. Bd. of Regents of Univ. of Okla., 468 U.S. 85, 104 (1984).

The first approach, per se analysis, permits courts to make "categorical judgments" that certain practices, including price fixing, horizontal output restraints, and market-allocation agreements, are illegal per se. Northwest Wholesale Stationers, Inc. v. Pac. Stationery & Printing Co., 472 U.S. 284, 289 (1985); see also NCAA, 468 U.S. at 100. Practices suitable for per se analysis have been found over the years to "be one[s] that would always or almost always tend to restrict competition and decrease output," and that are not "designed to increase economic efficiency and render markets more, rather than less, competitive." Broad. Music, Inc. v. CBS, 441 U.S. 1, 19-20 (1979) (citations omitted) (hereafter BMI). Such restrictions "have such predictable and pernicious anticompetitive effect, and such limited potential for procompetitive benefit, that they are deemed unlawful per se" without any need to conduct a detailed study of the markets on which the restraints operate or the actual effect of those restraints on competition. State Oil Co. v. Khan, 522 U.S. 3, 10 (1997).

At the other end of the spectrum, if the reasonableness of a restraint cannot be determined without a thorough analysis of its net effects on competition in the relevant market, courts must apply a full rule-ofreason analysis. See Oksanen v. Page Mem'l Hosp., 945 F.2d 696, 709 (4th Cir. 1991). In such cases a plaintiff "must prove what market . . . was restrained and that the defendants played a significant role in the relevant market" because "[a]bsent this market power, any restraint on trade created by defendant's action is unlikely to implicate" Section 1. Id. The required analysis varies by case and may extend to a "plenary market examination," California Dental, 526 U.S. at 779, covering "the facts peculiar to the business, the history of the restraint, and the reasons why it was imposed," Nat'l Soc'y of Prof'l Eng'rs v. United States, 435 U.S. 679, 692 (1978), as well as the availability of reasonable, less restrictive alternatives. NCAA, 468 U.S. at 106, 114; BMI, 441 U.S. at 20-23.

Sometimes, the anticompetitive impact of a restraint is clear from a quick look, as in a per se case, but procompetitive justifications for it also exist. Such intermediate cases may "involve[ ] an industry in which horizontal restraints on competition are essential if the product is to be available at all," NCAA, 468 U.S. at 101,or in which a horizontal restraint otherwise plausibly "increase [s] economic efficiency and renders markets more, rather than less, competitive." BMI, 441 U.S. at 20 (citation omitted). For these cases, "abbreviated or `quicklook' analysis" fills in the continuum between per se analysis and the full rule of reason. California Dental, 526 U.S. at 770.

The Supreme Court, in California Dental, recently explained and illustrated at some length the subtlety that the quick-look analysis may require. The Court emphasized that even when a court eschews a full rule-of-reason analysis and so forgoes detailed examination of the relevant market, it must carefully consider a challenged restriction's possible procompetitive justifications. Id. at 773-75.

In California Dental, the Court reviewed a decision of the Ninth Circuit, which had applied quick-look analysis to FTC factual findings based on a full evidentiary record. California Dental Ass'n v. FTC, 128 F.3d 720, 726-30 (9th Cir. 1997). An association of dentists had restricted its members' advertising, banning claims of superior quality and imposing such detailed disclosure requirements for price advertisements that dentists effectively could not advertise "across the board discounts" or even "low" or "affordable" prices. Id. at 723-24. The Ninth Circuit accepted that the advertising restrictions had obvious anticompetitive effects, because they reduced the price information available to consumers. Id. at 727-28. The appeals court limited its analysis to determining whether the proffered justifications for the restrictions had any procompetitive effect, and rejected those justifications because of an asserted lack of record support for them. Id. at 728 (finding "no evidence that the rule has in fact led to increased disclosure and transparency of dental pricing").

The Supreme Court vacated the Ninth Circuit's analysis as too cursory. Ordinarily, advertising restrictions could be expected to harm competition, but the challenged restrictions affected a market for professional services, in which patients typically have much less information about the quality of service than the professionals who serve them. 526 U.S. at 771-73. The high court instructed that although quick-look analysis does "carr[y] the day when the great likelihood of anticompetitive effects can easily be ascertained," a quick look, or at least the kind of quick look employed by the Ninth Circuit, does not suffice if a challenged restraint "might plausibly be thought to have a net procompetitive effect, or possibly no effect at all on competition." Id. at 770, 771. Only if "an observer with even a rudimentary understanding of economics could conclude that the arrangements in question would have an anticompetitive effect on customers and markets" would summary review like that conducted by the Ninth Circuit be proper. Id. at 770. Otherwise, determination of a restraint's impact on competition "call[s] for more than cursory treatment." Id. at 773. "What is required . . . is an enquiry meet for the case, looking to the circumstances, details, and logic of a restraint." Id. at 781. Because the Ninth Circuit failed to provide this, the Supreme Court vacated its judgment and remanded the case for further proceedings. Id.

In quick-look cases involving plausible procompetitive justifications, which require "more than cursory treatment" of "question[s] susceptible to empirical but not a priori analysis," id. at 773, 774, a full record may often be necessary.6 Certainly courts have been wary of summary judgment in the context of quick-look analysis. In fact, the parties have not cited, and we have not found, a single case in which the Supreme Court has approved a quick-look analysis in which the parties received less than a full evidentiary hearing, either before an administrative agency or in court. See FTC v. Indiana Fed'n of Dentists, 476 U.S. 447, 451 (1986); NCAA, 468 U.S. at 88; Nat'l Soc'y of Prof'l Eng'rs, 435 U.S. at 685-86; see also California Dental, 526 U.S. at 762-63.

These principles govern our examination of the case at hand. For the reasons that follow, we believe that application of them to the facts of this case requires that we vacate the district court's judgment. In a nutshell, although the district court demonstrated mastery of many intricacies of antitrust law, it performed too quick an analysis on an insufficiently developed factual record.

In several ways, the district court's approach resembles that of the Ninth Circuit in California Dental.

Like the Ninth Circuit, the district court applied quick-look analysis because it accepted that the challenged restraint had obvious anticompetitive effects. Compare California Dental, 128 F.3d at 727 (finding "a fairly naked restraint" (citation omitted)), with Continental, 126 F. Supp. 2d at 977 (finding "manifest" anticompetitive effects). Then, again like the Ninth Circuit, the district court determined that the proffered procompetitive justifications for the alleged restraint were implausible because of a perceived lack of record support for those justifications. Compare California Dental, 128 F.3d at 728 ("[T]he record provides no evidence that [the restraint] has, in fact, led to increased [procompetitive effects]."), with Continental, 126 F. Supp. 2d at 979 ("There is no record support for the proposition that an agreement to standardize carry-on baggage size . . . promotes competition.").

In California Dental, even though the lower court relied on factfinding based on a fully developed record, the Supreme Court vacated the Ninth Circuit's judgment rejecting procompetitive justifications because the appeals court failed to "scrutinize the assumption of relative anticompetitive tendencies." 526 U.S. at 781. Here, given the similar difficulty in the district court's analysis, rendered on summary judgment without any fact-finding, we too must vacate the district court's judgment.

Specifically, in holding the proffered procompetitive justifications for the template program "implausible," the district court underestimated the significance of some undisputed material facts and wrongly ruled that other genuine disputes were immaterial.7 We discuss each problem in turn.

Although the district court properly noted that airlines must cooperate to share airports, rejecting per se analysis on that basis, Continental, 126 F. Supp. 2d at 976, it failed to recognize the extent to which Dulles's indisputably unique architectural configuration requires careful consideration of the template program's competitive effect.

At Dulles, all embarking passengers on all airlines must pass through one of two common security checkpoints, and (because of the congestion problem revealed if templates are installed at just one checkpoint) the two checkpoint locations cannot vary in their treatment of carry-on baggage. Unlike other airports, Dulles cannot use multiple checkpoints to permit variety in its many airlines' treatment of carry-on baggage. In other words, Dulles's checkpoints constitute a single "bottleneck facility that [can]not feasibly be duplicated," and therefore "must be shared among rivals." Paddock Publ'ns, Inc. v. Chicago Tribune Co., 103 F.3d 42, 45 (7th Cir. 1996). Absent an efficient way to differentiate among various airlines' passengers at the checkpoints, but see Part IV, the x-ray machines at the checkpoints must have mouths of some uniform size, and the airlines must cooperate in determining that size: either small, as United and the AMC prefer, or larger, as Continental prefers.

Of course, although operation of the x-ray machines themselves require cooperation, security concerns do not require the airlines to cooperate on the use of templates or justify the airlines' agreement to do so. See 14 C.F.R. SS 108.5-.9 (2001); see generally 11 Herbert Hovenkamp, Antitrust Law S 1906. Templates are not security devices, and no one in this litigation even contends that the need for the airlines to cooperate on security justifies the Dulles template program.8 Cf. NCAA, 468 U.S. at 114 ("[I]t cannot be said that `the agreement on price is necessary to market the product at all.'" (citation omitted)). But unless an efficient way to differentiate among the passengers of different airlines emerges, Dulles's unique architecture does force all of its airlines to cooperate on a single decision as to the use of templates. When the economic implications of physical or geographical limitations require coordination among competitors, the Supreme Court has long applied Section 1 of the Sherman Act with flexibility.

More than ninety years ago, the Court considered a railroad terminal in which the "geographical and topographical situation" required all railroad companies passing through the city to use one terminal, just as all airline passengers may have to pass through one type of security checkpoint at Dulles. United States v. Terminal R.R. Ass'n, 224 U.S. 383, 397 (1912). The Court held that because of the unique configuration of the railroad terminus, competing railway companies could legally combine (indeed, were legally required to do so) to control the facility. Here, if the airlines did not have to share the checkpoints -if Continental had a security checkpoint at Dulles that served only its passengers -Section 1 would not permit the AMC to require it to use templates. However, as in Terminal Railroad Association, the presence of unique architectural constraints requires flexibility in applying Section 1.9

Moreover, beyond the general need for greater cooperation at Dulles than at other airports, United and Continental each make a more specific claim, related to Dulles's unique configuration, as to why their respective preferred outcomes benefit competition. Each argues that only a uniform policy in accordance with its preference will make possible an entire service that would not otherwise be available at Dulles: assertedly, Continental must win to offer flights with carry-on largesse, and United must win to offer flights with carry-on rigor. The district court may ultimately have to choose between two procompetitive claims; either outcome would both help and hurt competition, and which helps competition more than the other may be far from plain. See Part IV.

In such a situation, the Supreme Court has advised that a fuller examination is necessary. In California Dental, too, the context of the restraint plausibly had a similar mixed effect on competition. Ordinarily, the restraint at issue -advertising restrictions -hurts competition, but the restrictions in California Dental occurred in a "professional context," 526 U.S. at 774, a market in which patients had much less information on the quality of service than their dentists had. At bottom, the restrictions' net impact on the information available to patients was unclear. Accordingly, the restraint's net effect could not be determined -it might even have been nil. Id. at 778. The Supreme Court simply could not be sure how such a restraint would affect competition.

We have a similar difficulty. Here also, the context in which the restraint operates -an airport with unique features -makes it possible that the restraint yields both positive and negative effects on competition. Because the airlines must share the checkpoints, a decision that aids one airline's service may necessarily hamper another's. Therefore, without additional analysis, it is as difficult for us as it was for the Supreme Court in California Dental to determine the nature of the challenged restraint's net effect on competition. The district court's failure to give careful consideration to Dulles' unique architecture when it examined the templates' impact on competition thus requires us to vacate its judgment.

The district court also erred in ruling that factual disputes over the proffered procompetitive justifications for the templates -improved on-board safety, more timely departures, and a better on-board experience for passengers -were immaterial. The court recognized the existence of genuine factual disputes as to the justifications, but miscalculated in ruling that they were not material. The district court concluded that "the flying public wants . . . sufficient onboard carry-on storage capacity and flexible carry-on policies," and that United's "attendant incidental" problems with safety, timeliness, and passenger comfort only occurred because United had failed to meet that desire. Continental, 126 F. Supp. 2d at 980. Even if the templates did improve safety, timeliness, and comfort, they were not procompetitive because, according to the district court, customers preferred that such problems be solved through generous carry-on policies like that of Continental. Thus in the district court's view, the proffered justifications for the templates were, as a matter of law, not "plausible or material." Id. at 981.

Again, the district court's rationale resembles the Ninth Circuit's in California Dental. The Ninth Circuit found no record evidence that the challenged restrictions had the asserted positive effects, and further suggested that they were unlikely to have such effects. California Dental, 128 F.3d at 728. Here, the district court recognized a dispute as to whether the templates had the asserted positive effects, but ruled that even if so, there was "no record support for the proposition" that achieving safety, timeliness, and comfort through templates "promote[d] competition." Continental, 126 F. Supp. 2d at 979. In essence, each court dismissed the proffered procompetitive justifications because it found them implausible. Compare California Dental, 128 F.3d at 728, with Continental, 126 F. Supp. 2d at 979-81. The Supreme Court held that the "plausibility of competing claims" as to the effect of the challenged restriction -that is, the possibility that the restriction might have procompetitive effects or no effect at all -"rule[d] out the indulgently abbreviated review" of the Ninth Circuit. 526 U.S. at 778. So it is here.

Review of the record in this case simply does not support the conclusion that the proffered justifications for the template program are, as a matter of law, implausible. Rather, the record reveals factual disputes as to whether the proffered justifications plausibly enhance competition. Continental has, to be sure, offered evidence indicating that these justifications are illusory. But United has submitted evidence that a size restriction on carry-on luggage can improve onboard safety, on-time takeoffs, and passengers' flying experience, and that templates increase its efficiency in offering a new output, a service with these improved features. Moreover, evidence in the record, including studies by both airlines, also supports the view that some customers value and have sought such benefits, achieved by restricting carry-on baggage, including by the use of templates. See also Carry-On Baggage Program, 52 Fed. Reg. at 21,472, 21,473.

Given this record, we cannot conclude that these justifications are indisputably implausible. Like the advertising restrictions at issue in California Dental, installation of templates at Dulles might "plausibly be thought to have a net procompetitive effect, or possibly no effect at all." 526 U.S. at 771.

In addition to its dismissal of the templates' possible procompetitive justifications, which mirrors the Ninth Circuit action that the Supreme Court disapproved in California Dental, the district court erred in a more fundamental respect. It failed to recognize that the parties genuinely disputed whether, given the lifting and medallion alternatives, the template program actually restrained trade, i.e., whether the program actually restricted Continental's carry-on service. Continental offered evidence that even with medallions and lifting, some of its passengers were unable to avoid the templates. But United offered evidence to the contrary including the fact that no other airline found the medallion solution inadequate and the testimony of Continental's general manager at Dulles that, after installation of the templates, Continental was still able "to get all customers" with oversized bags to its gate. On this record, there is thus a genuine dispute as to whether the template program affected the ability of any Continental passenger to proceed directly to the gate with all desired carry-on baggage. The district court overlooked not just the existence but the central importance of this dispute: If all of Continental's passengers reached the gate with all desired luggage, Continental has not shown a restraint of trade.

The district court only addressed the lifting and medallion options in finding that Continental had demonstrated antitrust injury. We agree with the district court that Continental unquestionably incurred costs in supplying labor to lift the templates for its passengers.10 We further agree that if related to a violation of the antitrust laws, those costs would "flow[ ] from that which makes defendants' acts unlawful" under the Sherman Act, Brunswick Corp., 429 U.S. at 489, and would be recoverable as mitigation damages. Continental, 126 F. Supp. at 983, citing Lee-Moore, 599 F.2d at 1306. But whether there was a violation of the antitrust laws is very much in dispute. The parties disagree as to whether the template program prevented Continental passengers from bringing their desired luggage to the gate. If it is proved that the program did not do this, then the template program did not affect trade by restraining output.

If Continental cannot show any effect on price or output, then it has shown only that it incurred costs in hiring people to lift the templates. Those costs may establish a presumption that Continental lost profits when its costs rose with no corresponding increase in revenue. Continental II, 136 F. Supp. 2d at 545-46. But to be recovered as antitrust damages, a competitor's loss of profits must "stem[ ] from a competition-reducing aspect or effect of the defendant's behavior," Atlantic Richfield, 495 U.S. at 344 (original emphasis deleted), that is, from "acts that reduce output or raise prices to consumers." Chicago Prof'l Sports Ltd. P'ship v. NBA, 961 F.2d 667, 670 (7th Cir. 1992); see also Bd. of Trade of Chicago v. United States, 246 U.S. 231, 240 (1918) (noting that the challenged restraint had not affected price or reduced output, and that it might have increased output); cf. Brunswick Corp., 429 U.S. at 489 (noting that antitrust injury "should reflect the anticompetitive effect either of the violation or of anticompetitive acts made possible by the violation"). Thus, Continental cannot recover its costs of hiring template lifters unless it can demonstrate that the template program had any anticompetitive effect, something it has not yet done.

Before concluding, we note that we recognize the seeming unfairness in "punishing" Continental for perhaps finding a way in which its passengers could avoid the templates. Of course,"[a]n antitrust plaintiff is not obliged to pursue any imaginable alternative, regardless of cost or efficiency, before it can complain that a practice has restrained competition." CBS, Inc. v. ASCAP, 620 F.2d 930, 936 (2d Cir. 1980) (on remand from BMI, 441 U.S. 1). However, a market innovation "does not restrain trade if an alternative opportunity . . . is realistically available." Buffalo Broad. Co. v. ASCAP, 744 F.2d 917, 925 (2d Cir. 1984) (citation omitted). Whether an alternative is "realistic[ ]" depends on its cost. Id. at 926. Here the antitrust plaintiff -Continental -may have come up with a fully effective alternative, the cost of which apparently did not affect ticket prices. The templates are not "a restraint unless it were proven that there are no realistically available alternatives." Id. at 933.

The Supreme Court's consideration of alternatives in BMI guides the proper analysis of alternatives here. In that case, the challenged restraint added an entirely new good, a blanket license to perform multiple pieces of music, to the market, without eliminating the goods that had previously been available, individual licenses to perform single pieces of music. BMI, 441 U.S. at 21-24. The Court accepted the restraint's potential legitimacy, because permitting the challenged restraint offered "real choice." Id. at 24; cf. NCAA, 468 U.S. at 106 (considering whether a "real choice" was available). Here, United argues, and the facts may show, that only templates permit United to add an entire service to the market, as in BMI .

Unlike the situation in BMI, however, here the same argument weighs on the other side as well, complicating the problem. Continental argues, and the facts may show, that the addition of United's service would simultaneously eliminate Continental's own entire service. If neither side is able to offer its service through an efficient alternative, the district court may thus have to balance the relative contributions to competition of the two services. See Continental, 126 F. Supp. 2d at 981 n.52 (discussing United's alternatives); 13 Herbert Hovenkamp, Antitrust Law S 2132d."[E]ven efficient ventures may . . . be organized in alternative ways that will reduce dangers to competition." 13 Herbert Hovenkamp, Antitrust Law S 2100c; see also 11 Herbert Hovenkamp, Antitrust Law, S 1912i.

The district court will have an opportunity to consider the issues we have discussed on a full record. Of course, the resolution of some disputes could obviate the need to reach others. Given the multiple factual disputes at present, we express no view as to which issues the district court may need to reach, let alone what their outcomes might be. We merely rule that "we cannot reach these conclusions as a matter of law on a record this sparse." Eastman Kodak, 504 U.S. at 486.

For all of these reasons, we vacate the judgment of the district court granting summary judgment, damages, and an injunction in favor of Continental, and remand the case for further proceedings consistent with this opinion. We leave to the district court the "question whether on remand it can effectively assess" the alleged restraint by a modified quick-look analysis, or whether it must undertake "a more extensive rule-of-reason analysis." California Dental, 526 U.S. at 768 n.8; see also California Dental, 224 F.3d 942, 947 & n.2 (9th Cir. 2000) (on remand from California Dental, 526 U.S. 756).