Blue Cross & Blue Shield United of Wisconsin ("Blue Cross" for short), and its subsidiary, Compcare Health Services Insurance Corporation, a health maintenance organization (HMO), brought suit last year under sections 1 and 2 of the Sherman Act, 15 U.S.C. Secs. 1, 2, against the Marshfield Clinic and its HMO subsidiary, Security Health Plan of Wisconsin, Inc. After a two-week trial, the jury brought in a verdict for both plaintiffs that, after remittitur, trebling, and addition of attorneys' fees, produced a judgment just short of $20 million to which the judge then added a sweeping injunction that we have stayed pending the decision of the defendants' appeal, which we have heard on an expedited schedule. (The plaintiffs filed a cross-appeal, but it was dismissed by agreement of the parties.) The speed with which this complex case was brought to trial is as commendable as it is unusual.

The two plaintiffs have distinct though overlapping claims. Compcare, Blue Cross's HMO, claims that the Marshfield Clinic--a nonprofit corporation owned by the 400 physicians whom it employs--has a monopoly which it acquired and has maintained by improper practices that have excluded Compcare from the HMO "market" in the counties of north central Wisconsin in which the Marshfield Clinic and its HMO subsidiary (Security) operate. This is a section 2 monopolization charge. Blue Cross claims that Marshfield, partly through its own monopoly power and partly by collusion with other providers of medical services, charged supracompetitive prices to patients insured by Blue Cross. This is a section 2 monopolization charge combined with a section 1 price-fixing and division-of-markets charge.

Although Marshfield is a town of only 20,000 people in a largely rural region, the Marshfield Clinic is the fifth largest physician-owned clinic in North America, with annual revenues in excess of $200 million. The Clinic has its main office in Marshfield but it has 21 branch offices scattered throughout the 14 counties of north central Wisconsin. Oddly, we cannot find in the record, nor were counsel able to inform us at argument, what percentage the 400 physicians employed by the Clinic comprise of all the physicians in the 14 counties. We do know that the Clinic employs all the physicians in Marshfield itself and in several other towns--but one of these towns has only one physician--and all the physicians in one entire county--but a county that has only 12 physicians. Security, the Marshfield Clinic's HMO subsidiary, serves its subscribers through the physicians employed by the Clinic plus almost 900 other physicians with whom Security has contracts. These contracts are not exclusive--the physicians are free to work for other HMOs as well as to practice fee-for-service medicine--and in fact work for Security generates only 6 percent of these physicians' total income. (Compcare argues that the 6 percent figure is found only in a study and testimony excluded from evidence, but our reading of the objection to the testimony and the ruling on that objection is that not the figure, but only the witness's effort to characterize it as insignificant, was excluded.) In nine of the 14 counties in the north central region, Security has more than 90 percent of all subscribers to HMOs.

Compcare persuaded the jury that HMOs constitute a separate market, much as banks are a separate market from currency exchanges; and if there is a reasonable basis for this finding in the evidence, we are bound to accept it regardless of what we might think as an original matter. But we have searched the record in vain for evidence that under contemporary principles of antitrust law would justify such a finding. An HMO is not a distinctive organizational form or assemblage of skills, as is plain from the fact that the physicians retained by Security, whether they are employees of the Marshfield Clinic or completely independent, to serve its subscribers serve other patients on a fee-for-service basis. An HMO is basically a method of pricing medical services. Instead of having the patient pay separately for each medical procedure, the patient pays a fixed annual fee for all the services he needs and the HMO undertakes to provide those services with the physicians with whom it has contracts. The different method of pricing used by the HMO has, of course, consequences both for the practice of medicine and for the allocation of the risk of medical expenses. The method of pricing gives the HMO an incentive to minimize the procedures that it performs, since the marginal revenue it derives from each procedure is zero. Hence HMOs are thought to reduce "waste" and to encourage preventive care, although those hostile to the HMO concept believe that the principal effect is merely to reduce the amount of medical care that patients receive. The risk-shifting feature of the concept lies in the fact that if a subscriber incurs above-average medical expenses, the excess cost is borne by the HMO rather than by the subscriber (or by his insurer, or more likely by both because of copayment and deductible provisions in the insurance policy), while if he incurs below-average medical expense the difference enures to the benefit of the HMO rather than to him or his insurer (or, again, both). To control the upside risk that it incurs, the HMO provides medical services through physicians with whom it has contracts specifying their compensation, rather than merely reimbursing some percentage of whatever fee they might happen to charge for their services. This means that the HMO must be able to line up enough physicians with whom to contract to provide its subscribers with a more or less complete menu of medical services. Compcare complains that Security's contracts with physicians require them to refer their patients to the Clinic rather than to "independent" physicians. But that is of the essence of an HMO: the subscriber must take the service offered by the physicians whom the HMO has enlisted.

Compcare's principal argument is that Security has enlisted such a large fraction of the physicians in the 14-county north central region that Compcare cannot find enough "independent" physicians to be able to offer HMO services competitive with Security's--hence Security's huge market shares in 9 of the 14 counties. Supposing this is true--as seems unlikely since Security's almost 900 independent physicians are available to join other HMOs, along with an unknown number of physicians neither employed by the Marshfield Clinic nor retained by Security--it has monopolistic significance only if HMOs constitute a market separate from other contractual forms in which many of the same physicians sell their services. Of this we cannot find any evidence.

In defining a market, one must consider substitution both by buyers and by sellers, IIA Phillip E. Areeda, John L. Solow & Herbert Hovenkamp, Antitrust Law: An Analysis of Antitrust Principles and Their Application p 530a (1995), and let us start with buyers. The record shows, what is anyway well known, that individuals, and their employers, and medical insurers (the real "buyers" of medical services, according to the plaintiffs) regard HMOs as competitive not only with each other but also with the various types of fee-for-service provider, including "preferred provider" plans (generally referred to as "PPOs," for "preferred provider organization") under which the insurer offers more generous reimbursement if the insured patronizes physicians who have contracts with the insurer to provide service at low cost to its insureds. HMOs, though they have made great strides in recent years because of the widespread concern with skyrocketing medical costs, remain relative upstarts in the market for physician services. Many people don't like them because of the restriction on the patient's choice of doctors or because they fear that HMOs skimp on service, since, as we said, the marginal revenue of a medical procedure to an HMO is zero. From a short-term financial standpoint--which we do not suggest is the only standpoint that an HMO is likely to have--the HMO's incentive is to keep you healthy if it can but if you get very sick, and are unlikely to recover to a healthy state involving few medical expenses, to let you die as quickly and cheaply as possible. HMOs compensate for these perceived drawbacks by charging a lower price than fee-for-service plans.

We do not wish to associate ourselves with the critics of HMOs. All that is important to our consideration of this appeal is that many people believe--whether rightly or wrongly is of no moment--that HMOs are not an unalloyed blessing; and this means that the price that an HMO can charge is constrained not only by competition from other HMOs but also by competition from forms of medical-services contracting that are free from the perceived perverse incentive effects of the HMO form. As far as we know or the record shows, even in areas where there is only one HMO most of the people in its service area do not subscribe to it even if its prices are lower than those of fee-for-service providers.

We do not understand Compcare to be contending that insurance companies cannot find enough physicians in the north central region willing to sign contracts with them to staff preferred-provider plans. This has a dual significance. Such plans are particularly close substitutes for HMOs, since they use the insurer's clout with physicians to drive down price. And if, as the record shows, they are feasible even in areas supposedly dominated by the Marshfield Clinic, and even without including any physicians employed by the Clinic, it is impossible to understand what barrier there could be to the formation of HMOs. All that is needed is an array of physicians who among them provide a broad range of medical services, and the same thing is needed for a preferred-provider plan. Ability to create a preferred-provider plan therefore implies ability to create an HMO, and so implies in turn that if Security raised its prices above competitive levels, and if people had a strong preference for HMOs over preferred-provider plans despite the similarities between the two systems for providing medical care, the operators of those plans would convert them to HMOs. If so, those plans are part of the same market with HMOs. So Blue Cross and Blue Shield of Indiana argued in Ball Memorial Hospital, Inc. v. Mutual Hospital Ins., Inc., 784 F.2d 1325 (7th Cir.1986), and we accepted its argument. Id. at 1331-37.

As we said earlier, the definition of a market depends on substitutability on the supply side as well as on the demand side. Even if two products are completely different from the consumer's standpoint, if they are made by the same producers an increase in the price of one that is not cost-justified will induce producers to shift production from the other product to this one in order to increase their profits by selling at a supracompetitive price. SBC Communications Inc. v. FCC, 56 F.3d 1484, 1493-94 (D.C.Cir.1995); cf. Ball Memorial Hospital, Inc. v. Mutual Hospital Ins., Inc., supra, 784 F.2d at 1335. The services offered by HMOs and by various fee-for-service plans are both provided by the same physicians, who can easily shift from one type of service to another if a change in relative prices makes one type more lucrative than others.

We thus do not believe that a reasonable jury, confronting the record compiled in the district court, could find that HMOs constitute a separate market. (We emphasize "record compiled in the district court." Terms such as "HMO" and "PPO" are used to refer to a variety of different types of plan, which may vary in respects crucial to antitrust liability.) Nor, if we examine a more plausible market such as all physician services in north central Wisconsin and thus shift our focus from Security, the HMO, to the Marshfield Clinic itself, is there any basis in the record for treating the Clinic not as the alliance of 400 physicians that is the Marshfield Clinic but instead as the almost 1300 physicians who are either employed by the Clinic or have contracts with Security. The 900 independent contractors derive only a small fraction of their income from those contracts, as we have seen, and the contracts do not forbid them to join other HMOs or otherwise compete with Security or with the Marshfield Clinic.

It is true that because of the sparsity of the physician population in north central Wisconsin the physicians employed by the Clinic have a large share of the market for physician services, since, for primary care anyway (an important qualification--people will go a long way for a liver transplant), that market is a local one. Remember, there is one entire county in which all the physicians are employed by the Clinic. We do not understand Compcare to be arguing that this is the market that the Clinic monopolized; the damages to Compcare from being excluded (if it was excluded--we do not know) from that tiny market would be tiny. Compcare paints with a broader brush, drawing a dizzying series of concentric circles around the Clinic's offices and counting the physicians who can serve people living in the circles. It would have been much simpler and we suppose just as good--having in mind the desirability of avoiding a hunt for the snark of delusive exactness--to have treated the counties as the markets. But whatever the precise method chosen, Compcare is unable to show that in what it chooses to regard as the relevant geographic market or series of linked geographic markets for physician services the Marshfield Clinic employed 50 percent or more of the physicians serving the market. Fifty percent is below any accepted benchmark for inferring monopoly power from market share, United States v. Rockford Memorial Corp., 898 F.2d 1278, 1285 (7th Cir.1990); Fineman v. Armstrong World Industries, Inc., 980 F.2d 171, 201-02 (3d Cir.1992); Colorado Interstate Gas Co. v. Natural Gas Pipeline Co., 885 F.2d 683, 694 n. 18 (10th Cir.1989); United States v. Aluminum Co. of America, 148 F.2d 416, 424 (2d Cir.1945) (L. Hand, J.), and the Clinic has given us no reason for lowering the mark. Compcare is able to derive larger shares only by defining submarkets of specific, very narrowly defined medical procedures, called "Diagnostic Related Groups," such as circumcision of a male 17 years old or older, circumcision of a male under 17, hysterectomy, and reconstructive surgery for the uterine system, in a few of which (14 out of 494, according to Compcare's expert witness) the physicians employed by or under contract with the Marshfield Clinic performed most of the procedures in north central Wisconsin. As the examples we have given show, classification in a DRG is unrelated to the conditions of supply. Many, no doubt most, physicians perform or are capable of performing more than one procedure, and are therefore part of the market even if at present not active in it. If the Clinic overprices a particular procedure, other physicians capable of performing that procedure will have an added incentive to do so, knocking down the excessive price.

Compcare also asked the jury to infer monopoly power directly from the Clinic's high prices, and high rate of return, relative to the prices and rates of return of its competitors. But when dealing with a heterogeneous product or service, such as the full range of medical care, a reasonable finder of fact cannot infer monopoly power just from higher prices--the difference may reflect a higher quality more costly to provide--and it is always treacherous to try to infer monopoly power from a high rate of return. Taking the second point first, not only do measured rates of return reflect accounting conventions more than they do real profits (or losses), as an economist would understand these terms, Susan Rose-Ackerman, "Unfair Competition and Corporate Income Taxation, 34 Stan.L.Rev. 1017, 1025 n. 28, 1031 n. 43 (1982); Michael Boudin, "Forensic Economics," 97 Harv.L.Rev. 835, 837 (1984), but there is not even a good economic theory that associates monopoly power with a high rate of return. Firms compete to become and to remain monopolists, and the process of competition erodes their profits. Conversely, competitive firms may be highly profitable merely by virtue of having low costs as a result of superior efficiency, yet not sufficiently lower costs than all other competitors to enable the firm to take over its market and become a monopolist. As for high prices, one of the complaints against HMOs, remember, is that they skimp on service. One HMO may charge higher prices than other HMOs (and Security does charge higher prices) not because it has a monopoly but because it is offering better service than the other HMOs in its market. Compcare itself stresses the quality of the Marshfield Clinic's doctors, as part of its argument that it cannot succeed unless the Clinic is forced to join it. Generally you must pay more for higher quality.

So Security is not a monopolist of HMO services, because, subject to our earlier qualification, HMOs are not a market. The Marshfield Clinic, its parent, is not a monopolist either. It does not control its independent contractors; and once they are excluded, the Clinic (which is to say the physicians who own and are employed by it--who are the firm, functionally speaking) does not have a monopoly of physician services in the north central region or in any parts of the region other than parts too small to support more than a handful of physicians. If the Marshfield Clinic is a monopolist in any of these areas it is what is called a "natural monopolist," which is to say a firm that has no competitors simply because the market is too small to support more than a single firm. If an entire county has only 12 physicians, one can hardly expect or want them to set up in competition with each other. We live in the age of technology and specialization in medical services. Physicians practice in groups, in alliances, in networks, utilizing expensive equipment and support. Twelve physicians competing in a county would be competing to provide horse-and-buggy medicine. Only as part of a large and sophisticated medical enterprise such as the Marshfield Clinic can they practice modern medicine in rural Wisconsin.

We are mindful that a concept of essential or bottleneck facilities has been used from time to time to require a natural monopolist to cooperate with would-be competitors. E.g., Otter Tail Power Co. v. United States, 410 U.S. 366 (1973); MCI Communications Corp. v. AT & T Corp., 708 F.2d 1081, 1132-33 (7th Cir.1983). The principal case remains the old St. Louis terminal case. United States v. Terminal Railroad Ass'n, 224 U.S. 383, 32 S. Ct. 507, 56 L. Ed. 810 (1912). A consortium of 14 of the 24 railroads that shipped freight across the Mississippi River at St. Louis got control of the terminal facilities at each side of the river. The Supreme Court, while assuming that the operation of these facilities as a single entity was the most efficient way to operate them (that is, they comprised a natural monopoly), held that the Sherman Act required the consortium to provide access to the terminal facilities to the 10 other railroads on nondiscriminatory terms. The decision, along with the rest of the decisions in the essential-facilities line, has been criticized as having nothing to do with the purposes of antitrust law. E.g., Herbert Hovenkamp, Federal Antitrust Policy: The Law of Competition and Its Practice Sec. 7.7 (1994); Phillip Areeda, "Essential Facilities: An Epithet in Need of Limiting Principles," 58 Antitrust L.J. 841 (1990); Scott D. Makar, "The Essential Facilities Doctrine and the Health Care Industry," 21 Fla.St.U.L.Rev. 913 (1994). Had the terminal facilities been owned by a firm unaffiliated with any railroad, the firm could have charged whatever prices it wanted, including prices that discriminated against some of the users (monopolists frequently price discriminate), because the antitrust laws do not regulate the prices of natural monopolists. A natural monopolist that acquired and maintained its monopoly without excluding competitors by improper means is not guilty of "monopolizing" in violation of the Sherman Act, National Reporting Co. v. Alderson Reporting Co., 763 F.2d 1020, 1023-24 (8th Cir.1985); United States v. Aluminum Co. of America, supra, 148 F.2d at 430, and can therefore charge any price that it wants, Ball Memorial Hospital, Inc. v. Mutual Hospital Ins., Inc., supra, 784 F.2d at 1339; Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263, 296-98 (2d Cir.1979), for the antitrust laws are not a price-control statute or a public-utility or common-carrier rate-regulation statute.

And the charging of a high price is, so far as potential competitors are concerned, an attracting rather than an excluding practice. Consumers are not better off if the natural monopolist is forced to share some of his profits with potential competitors, as required by Terminal Railroad Ass'n. Similarly, if the practice of medicine in some sparsely populated county of north central Wisconsin is a natural monopoly, consumers will not be helped by our forcing the handful of physicians there to affiliate with multiple HMOs. Those physicians will still charge fees reflecting their monopoly.

We are not authorized to abrogate doctrines that have been endorsed and not yet rejected by the Supreme Court--as we acknowledged, indeed emphasized, with specific reference to the "essential facilities" doctrine, in Olympia Equipment Leasing Co. v. Western Union Telegraph Co., 797 F.2d 370, 376 (7th Cir.1986)--but we do not understand Compcare to be trying to pin the tail on the natural-monopoly donkey. Its argument is not that the Marshfield Clinic "owns" some tiny natural monopolies scattered throughout its market area, to which it must admit Compcare, but that the Clinic itself is an essential facility that any HMO may demand access to on equal terms, like the excluded railroads in the terminal case. To be an essential facility, however, a facility must be essential. The terminal association controlled 100 percent of a market, assumed to be properly defined, consisting of terminal services at St. Louis. The Clinic does not control 100 percent--or even 50 percent--of any properly defined market. It did have a monopoly market share of the HMO "market," but we have held that that is not a proper market; and likewise the DRGs. Nor can the Clinic's market share be levered up to 100 percent by the argument that the Clinic's reputation is so superb that no one will sign up with an HMO or preferred-provider plan that does not have the Clinic on its roster. The suggestion that the price of being "best" is to be brought under the regulatory aegis of antitrust law and stripped of your power to decide whom to do business with does not identify an interest that the antitrust laws protect. "The successful competitor, having been urged to compete, must not be turned upon when he wins." United States v. Aluminum Co. of America, supra, 148 F.2d at 430.

Since monopoly power was not proved, we need not evaluate the practices by which the Clinic acquired or maintained it. But we will not conceal our skepticism that they are exclusionary in an invidious sense. Compcare argues that the Clinic refused to allow its employees to "cross-cover" with "independent" physicians, in the sense of physicians not employed by or affiliated by contract with the Clinic (that is, agree to care for another physician's patients when that physician is on vacation or otherwise unavailable in exchange for a reciprocal agreement by the other physician), discouraged hospitals that its staff controlled from joining HMOs that compete with Security, and restricted staff privileges at those hospitals of "independent" physicians at those hospitals. All these practices are ambiguous from the standpoint of competition and efficiency. Hospitals are not public utilities, required to grant staff privileges to anyone with a medical license. The Marshfield Clinic's reputation for high quality implies selectivity in the granting of staff privileges at hospitals affiliated with the Clinic. Physicians employed by the Clinic, which has its own HMO, are hardly to be expected to steer their patients to another HMO, as they would be doing if they used their control of hospital staffs to induce the hospital to join another HMO. And given the extensive network constituted by the physicians either employed by or contracting with the Clinic, they would have little occasion to "cross-cover" with other physicians and would be reluctant to do so if, as is completely consistent with Compcare's version of the "essential facilities" doctrine, the Clinic maintains a reputation for high quality by being selective about the physicians to whom it entrusts its customers. But the important point is that, even if these practices are as we doubt tortious interferences with Compcare's business, they do not constitute monopolizing in violation of section 2 of the Sherman Act in the absence of acceptable proof, here lacking, of monopoly power.

We turn to Blue Cross's case. Blue Cross claims that it was overcharged by the Clinic. Many of the individuals whom Blue Cross insures in north central Wisconsin are customers of the Marshfield Clinic, paying for the services they receive from the Clinic on a fee-for-service basis. (The Clinic's doctors devote only a small part of their time to Security's HMO clientele.) Blue Cross pays the Clinic directly the portion of the fee that Blue Cross has agreed with its insureds to cover. The Clinic seeks to head off Blue Cross's claim at the pass, arguing that since the Clinic's fee-for-service contracts are with the patients themselves--it has no contract with Blue Cross--only the patients have "standing" (a term here not used in the jurisdictional sense, Associated General Contractors of California, Inc. v. California State Council of Carpenters, 459 U.S. 519, 535 n. 31, 103 S. Ct. 897, 907 n. 31, 74 L. Ed. 2d 723 (1983)) to complain about the alleged overcharging.

If the patients paid the entire fees to the Clinic and then were reimbursed in whole or part by Blue Cross, the Clinic would be right: only the patients could sue. The Supreme Court has been emphatic that only the direct purchaser from an allegedly overcharging defendant has standing to maintain an antitrust suit. Kansas v. Utilicorp United Inc., 497 U.S. 199, 110 S. Ct. 2807, 111 L. Ed. 2d 169 (1990); Illinois Brick Co. v. Illinois, 431 U.S. 720, 97 S. Ct. 2061, 52 L. Ed. 2d 707 (1977). But here the money went directly from Blue Cross to the Clinic, and although the two entities were not linked by any overarching contract, each payment and acceptance was a separate completed contract. We do not think more is required to establish Blue Cross's right to sue to collect these overcharges. Pennsylvania Dental Ass'n v. Medical Service Ass'n, 815 F.2d 270, 276-77 (3d Cir.1987); Todorov v. DCH Healthcare Authority, 921 F.2d 1438, 1455 (11th Cir.1991); cf. Kartell v. Blue Shield of Mass., Inc., 749 F.2d 922, 926 (1st Cir.1984). It would be cumbersome, to say the least, for patients of the Marshfield Clinic to organize into a class action to recover money that the patients never paid and that if they received in a judgment or settlement they would have to share with Blue Cross; for Blue Cross could seek and probably obtain restitution of moneys paid by mistake to the insureds because paid for expenses that the insureds had not in the end incurred. 2 George E. Palmer, The Law of Restitution Sec. 11.2(d), p. 491 (1978); Home Ins. Co. v. Honaker, 480 A.2d 652 (Del.1984); Ticor Title Ins. Co. v. Graham, 576 N.E.2d 1332, 1336-37 (Ind.App.1991).

Against this it can be argued that if Blue Cross reimburses an insured for say 80 percent of a $100 charge that should have been only $50, Blue Cross has lost $40 but the insured has lost $10 and so there is still a basis for a class action. Still, in this example anyway it is Blue Cross that bears the lion's share of the loss, as well as being able to proceed on its own without the aggregation of separate plaintiffs required for a class action; and the class action would enable a full recovery of the damages inflicted by the defendant only if the class members received a tremendous windfall--only, that is, if the class is allowed to recover $40 rather than $10 (and the entire $50 will of course be trebled). It can further be argued that the insured might pay the whole $50, because Blue Cross would reduce the percentage that it reimburses in order to shift the cost of the monopolist's overcharge to the hapless insureds. But the insureds may not be hapless--may in fact be members of the employees' group insurance plan of a large employer with plenty of insurance options. We need not pursue these issues. Blue Cross paid Marshfield Clinic directly, in accordance with Blue Cross's contractual obligations to its insureds, and if it paid too much because the Clinic violated the antitrust laws then it ought to be allowed to sue to recover these damages.

We note a tension between Blue Cross's claim and Compcare's. Blue Cross claims that the Clinic overcharged. The higher its prices, the greater the opening for competitors, such as Compcare. Even if Compcare could not get a foothold in the HMO "market," it could, of course, provide medical services through other means, such as a preferred-provider plan; and the higher the prices charged by the Marshfield Clinic, the more attractive such a plan offered by Compcare would be to employers and other purchasers of medical plans. To put the point differently, Blue Cross has a dual role in this case, as a buyer of medical services from Marshfield Clinic and, through its Compcare subsidiary, as a competitor of the Clinic. As a competitor its interest is in the Clinic's charging high prices, but as a buyer its interest is in the Clinic's charging low prices. Damages to Compcare are windfalls to Blue Cross, and vice versa. The damages awarded by the jury to the two plaintiffs (since they are economically one entity) should have been netted against each other rather than aggregated.

But that is a detail, since Compcare, for the reasons discussed earlier, was not entitled to recover any damages at all. Forget Compcare, then, and think of Blue Cross solely as a purchaser of medical services. Insofar as it paid high prices because the Marshfield Clinic was a monopolist, its claim must fail because the Clinic was not shown to be an illegal monopolist; a lawful monopolist can charge what it wants. But Blue Cross also claims that the Clinic colluded with competitors to raise prices above competitive levels. That claim is independent of the Clinic's being a monopolist.

Two forms of collusion are charged, with various furbelows and arabesques that can be ignored. The first is collusion between the Clinic and its affiliated (as distinct from employed) physicians. Here the tension between the two plaintiffs' cases again comes to the surface, since it will be recalled that Compcare in its case treats the affiliated physicians as a part of the Clinic, and a firm cannot collude with itself. Forget all that; the only evidence of collusion is that the Clinic, when buying services from the affiliated physicians either directly or through Security, would not pay them more than what these physicians charge their other patients. This is said to put a floor underneath these physicians' prices, since if they cut prices to their other patients their reimbursement from the Clinic will decline automatically. This is an ingenious but perverse argument. "Most favored nations" clauses are standard devices by which buyers try to bargain for low prices, by getting the seller to agree to treat them as favorably as any of their other customers. The Clinic did this to minimize the cost of these physicians to it, and that is the sort of conduct that the antitrust laws seek to encourage. It is not price-fixing. Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., --- U.S. ----, ---- - ----, 113 S. Ct. 2578, 2589-90, 125 L. Ed. 2d 168 (1993); Hospital Corp. of America v. FTC, 807 F.2d 1381, 1392 (7th Cir.1986); Ball Memorial Hospital, Inc. v. Mutual Hospital Ins., Inc., supra, 784 F.2d at 1338. Perhaps, as the Department of Justice believes, these clauses are misused to anticompetitive ends in some cases; but there is no evidence of that in this case.

The other form of collusion alleged is more troublesome. The Clinic is said to have agreed with some of its competitors, in particular the North Central Health Protection Plan, an HMO that had 37,000 subscribers, to divide markets, a practice that violates section 1 of the Sherman Act. Palmer v. BRG of Georgia, Inc., 498 U.S. 46, 111 S. Ct. 401, 112 L. Ed. 2d 349 (1990) (per curiam). The analogy between price-fixing and division of markets is compelling. It would be a strange interpretation of antitrust law that forbade competitors to agree on what price to charge, thus eliminating price competition among them, but allowed them to divide markets, thus eliminating all competition among them.

The charge of a division of markets in this case is backed up by some pretty strong documents. In one of them an executive of the Marshfield Clinic, after noting that "at the time in which management of Marshfield Clinic and North Central Health Protection Plan established the Free Flow arrangements, the parties involved purposely chose not to place in writing clear descriptions of their respective Plan service areas so as to minimize any risk in terms of violation of antitrust laws," remarks that some years ago physicians affiliated with the NCHPP "wished to establish a practice in Marshfield and we took the position that the Free Flow agreement did not support that activity"--and they backed off. The implications of this only slightly Aesopian wording are backed up by statistics showing that Security had fewer than 30 percent of the HMO subscribers in the two counties in which NCHPP was active, while in the counties surrounding those two counties on three sides it had more than 90 percent of the HMO subscribers.

We think the evidence of a division of markets, though a little scanty, was sufficient to sustain the jury's verdict on this (and only this) aspect of the case--provided this is not one of those cases in which a division of markets or other cartel-like activity is actually essential to the provision of a lawful service. Broadcast Music, Inc. v. Columbia Broadcasting System, Inc., 441 U.S. 1, 99 S. Ct. 1551, 60 L. Ed. 2d 1 (1979); National Collegiate Athletic Ass'n v. Board of Regents, 468 U.S. 85, 104 S. Ct. 2948, 82 L. Ed. 2d 70 (1984); General Leaseways, Inc. v. National Truck Leasing Ass'n, 744 F.2d 588 (7th Cir.1984). The "Free Flow" agreement to which the document from which we quoted refers was an agreement whereby Security's physicians and NCHPP's physicians could refer patients to each other without getting the permission of their superiors. The plan of the physician who rendered the service would bill the other plan for its cost. This was no doubt a valuable service, expanding the range of physicians to whom people enrolled in these two HMOs could turn. But the Clinic has failed to explain why, to provide this service, it was necessary for the two plans to agree not to open offices in each other's territories. We can imagine an argument that would run as follows: if NCHPP, say, opened an office in Marshfield, the doctors staffing that office would constantly be referring patients to the Clinic because it has such a plethora of doctors in Marshfield (remember that it employs all the doctors working in that town). The office would be running a referral business, "free riding" on the reputation and quality of Marshfield's doctors, and perhaps it would be difficult to devise a reimbursement plan under which these doctors would be compensated for these services. It is not a bad argument, but it is not made by the Clinic and it has no support in the record.

We conclude that the jury verdict on liability must stand insofar as the charge of a division of markets is concerned, and therefore the provisions of the injunction (IV(A)(3), (B)(3), (C)(3), (D)(3), and (E)(2), and V(D)(1)--and possibly others; but for the sake of clarity and completeness the injunction ought to be rewritten from scratch in light of our invalidating most of it) that forbid the Clinic to divide markets with competing plans or groups. The rest of the injunction falls with the charges that we have held to be without legal or factual basis. No part of the award of damages or of costs and attorneys' fees can stand, because these items were not segregated by offense. There must be a new trial limited to damages for dividing markets. The burden will be on Blue Cross to show how much less it would have paid the Clinic had the Clinic refrained from that illegal practice.

We have not discussed every contention of the parties, but have said enough to indicate the basis for our decision. As a detail, we urge the bench and bar of this circuit not to imitate the special verdict on liability that went to the jury in this case. The verdict form contained 18 questions. If the jury wanted to find for the plaintiff on every contested point, all it had to do (and all it did do) was write "yes" 18 times. A verdict so configured, like a plea colloquy in which the answer that the defendant is expected to give to every one of the judge's questions is "yes," Castillo v. United States, 34 F.3d 443, 445 (7th Cir.1994), invites rote answers rather than the careful consideration of the structure of the plaintiff's case that the use of a special verdict is intended to foster. Cf. W.T. Rogers Co. v. Keene, 778 F.2d 334, 346 (7th Cir.1985). None of the purposes ascribed to the giving of a special verdict, Stewart & Stevenson Services, Inc. v. Pickard, 49 F.2d 635, 644 (11th Cir.1984); 9A Charles Alan Wright & Arthur R. Miller, Federal Practice and Procedure Sec. 2508 (1995), of which probably the most important is to minimize the likelihood or scope of a retrial in a case in which there is more than one independent ground for recovery, are served by a verdict form that does not invite the jurors to use their heads. Given the broad discretion of the trial judge in the formulation of special verdicts, e.g., Hibma v. Odegaard, 769 F.2d 1147, 1157 (7th Cir.1985), and the fact that no objection was made to the form employed here, we do not suggest that this form would have constituted reversible error if the point had been argued; we merely offer for what it is worth our belief that it is not the best possible form, because it does not force the jurors to think before filling it out.

The judgment is affirmed in part, reversed in part, and remanded for further proceedings consistent with this opinion. Costs on appeal to the appellants.

AFFIRMED IN PART, REVERSED IN PART, AND REMANDED.