Robert J. Will, Cross-Appellees v. Comprehensive Accounting Corporation, Cross

*668 EASTERBROOK, Circuit Judge. Franchising spread from hamburgers to the preparation of tax returns and then to the provision of regular accounting services. The Comprehensive Accounting Corporation authorizes accountants to provide service in Comprehensive’s name. The franchisees agree to live up to Comprehensive’s standards and to supply reports to clients in Comprehensive’s style and bearing its trademark. This appears to be useful to clients — perhaps because of the standardized method of doing business, perhaps because of Comprehensive’s policing of its franchisees. Comprehensive is profitable, and so are the franchisees. The business of a franchisee apparently may be sold for more than two times annual gross revenues, while businesses of accountants in solo practice usually fetch much less. Comprehensive’s services to its franchisees include data processing. The accountants send data from clients’ businesses to Comprehensive, which returns reports generated by its large computer. The contract between Comprehensive and its franchisees permits the franchisees to have data processed elsewhere, provided “the Comprehensive E.D.P. [electronic data processing] is not competitive for like services of the same quality, with the same turnabout time.” In recent years small computers have become less expensive, and independent firms have written programs for these small computers that enable them to generate accounting reports and otherwise manage clients’ data. Comprehensive’s franchisees became interested in these smaller computers, which potentially could cut their costs of computation below the prices offered by Comprehensive. Comprehensive did not take this gracefully. It had a large computer ii/place; the cost of this was sunk, so payments from franchisees for computation were mostly profit. Comprehensive’s revenues from data processing reached $3.5 million yearly;- these revenues were the firm’s principal source of profit. The franchisees’ savings from buying their own small computers would translate into losses for Comprehensive. It therefore insisted that franchisees use only small computers that would produce reports that looked exactly like those Comprehensive produced itself. Comprehensive says that it insisted on duplication in order to protect the reputation of its service mark and maintains that it was entitled by contract to be finicky; the franchisees say that Comprehensive was just postponing the inevitable, in breach of contract. Duplication was hard. The commercially available programs were not designed to ape Comprehensive’s reports, and the output of Comprehensive’s printer also looked different from the output of the low-price printers some franchisees wanted to use. Comprehensive promised to find and approve a small computer system that would meet its quality standards. It settled on a system that would be restricted to franchisees with 200 or more clients; most had fewer. Even the approval of this system moved slowly, with several changes of configuration that led some franchisees to conclude that Comprehensive would never be satisfied. In 1982 it announced that it would approve a system with programs Comprehensive had designed itself; it apparently planned to charge enough for these to make up for lost revenues from its larger computer. Several franchisees decided to strike out on their own. Comprehensive threatened them with termination when it found out about this, so some franchisees installed small computers without informing Comprehensive. Comprehensive terminated five franchises in August 1981, sued three of them for substantial sums, and tried to persuade their clients to migrate to other accountants. As other franchisees began to use different systems, Comprehensive terminated them too. Ten of the terminated franchisees and two others brought this suit against Comprehensive, one of its subsidiaries, the chairman of its board, and its president. They maintained that Comprehensive broke its contract by insisting that they purchase computation from Comprehensive even *669 though it was “not competitive for like services of the same quality, with the same turnabout time.” Comprehensive replied that the small computers did not supply the “same quality” service, and that at all events each of the franchisees had violated other provisions of the contract. The franchisees also argued that Comprehensive had violated § 1 of the Sherman Act, 15 U.S.C. § 1, by “tying” data processing (the tied product) to the franchise (the tying product). The district court conducted a jury trial. The jury returned a general verdict for the defendants on the antitrust theory. Six plaintiffs prevailed on the contract theory and six lost. The largest award was $37,-678 to plaintiff Cahill; plaintiffs Rudd and Miller received $1; the others came out in between. Everybody appeals. Comprehensive and the other defendants say that the damages were excessive (they do not challenge the finding of liability); the six franchisees who lost on their contract claims say the verdicts are irrational because there is no significant difference between the winning and losing plaintiffs. All 12 franchisees challenge several of , the instructions to the jury on both antitrust and contract claims; they also dispute some of the district court’s decisions to admit evidence. I Section 1 of the Sherman Act prohibits any “contract, combination ..., or conspiracy, in restraint of trade____” The plaintiffs therefore needed to prove some cooperative undertaking. Establishing the necessary combination in a tying case requires exceeding subtlety, because the substantive theory of tying law depends on coercion to take two products as a package. The joint sale of two products is a “tie” only if the seller exploits its control of the tying product “to force the buyer into the purchase of a tied product that the buyer either did not want at all, or might have preferred to purchase elsewhere on different terms.” Jefferson Parish Hospital District No. 2 v. Hyde, 466 U.S. 2, 104 S.Ct. 1551, 1558, 80 L.Ed.2d 2 (1984). See also Jack Walters & Sons Corp. v. Morton Building, Inc., 737 F.2d 698, 703-05 (7th Cir.), cert. denied, - U.S. -, 105 S.Ct. 432, 83 L.Ed.2d 359 (1984). A tie within the meaning of antitrust depends on showing that the buyer did not want to take both products from the same vendor. “[TJhere is nothing inherently anticompetitive about packaged sales. Only if [buyers] are forced to purchase [the tied] services as a result of the [seller’s] market power would the arrangement have anticompetitive consequences.” Hyde, supra, 104 S.Ct. at 1565. If the buyer wants both products together — as, for example, the buyer of an automobile wants both chassis and engine together, even though they could be sold separately — there is no forcing, and so there is no tie-in. As a linguistic matter, proof that the buyer took both products in a package against his will negates the existence of a “contract, combination, or conspiracy.” The plaintiffs’ position here is complicated by their contract, which they insist establishes that they did not agree to buy their computing services from Comprehensive. Tying is not cooperation among competitors, the focus of § 1, it is aggressive conduct akin to monopolization under § 2 of the Sherman Act. Tying usually is challenged under § 3 of the Clayton Act, 15 U.S.C. § 14, which addresses the practice explicitly. Joint action becomes an issue only when the plaintiff tries to take advantage of per se rules under § 1. Perma Life Mufflers, Inc. v. International Parts Corp., 392 U.S. 134, 88 S.Ct. 1981, 20 L.Ed.2d 982 (1968), overruled in part on other grounds by Copperweld Corp. v. Independence Tube Corp., 467 U.S. 752, 104 S.Ct. 2731, 81 L.Ed.2d 628 (1984), provides plaintiffs with an escape hatch. The Court stated that a franchisee “can clearly charge a combination between [the franchisor] and himself, as of the day he unwillingly complied with the restrictive franchise agreements, ... or between [the franchisor] and other franchise dealers, whose acquiescence in [the] firmly enforced *670 restraints was induced by 'the communicated danger of termination’ ...” 392 U.S. at 142, 88 S.Ct. at 1986. Although the franchise contract in Perma Life established the tie-in of which the franchisees complained, the essential principle — that “unwilling compliance” satisfies the joint action requirement of § 1 — applies to our case too. This portion of Perma Life survived both Copperweld and Monsanto Co. v. Spray-Rite Service Corp., 465 U.S. 752, 104 S.Ct. 1464, 79 L.Ed.2d 775 (1984). Copperweld rejected an alternative holding in Perma Life that an agreement among jointly owned corporations satisfied the combination requirement of § 1, but it expressly refrained from disturbing the holding we quoted above (see 104 S.Ct. at 2739). Monsanto made it harder to show joint action concerning price within a system of distributing products that is concededly full of joint action; Monsanto helps to separate the legitimate and illegitimate components of a generally cooperative system of distribution and has nothing to say about the meaning of the joint action requirement in a tying case. See Black Gold, Ltd. v. Rockwool Industries, Inc., 732 F.2d 779, 780 (10th Cir.), cert. denied, - U.S. -, 105 S.Ct. 178, 83 L.Ed.2d 113 (1984). We therefore must decide whether the district court’s charge correctly stated the Perma Life rule. The court told the jury: A conspiracy or agreement .. .■ may be shown to have existed in either of two ways. First, as between Comprehensive and the individual plaintiffs themselves, if the individual plaintiffs unwillingly accepted and agreed with the alleged tying arrangement. Second, the conspiracy may also be shown to have existed between Comprehensive and franchise dealers other than the plaintiffs whose agreement to Comprehensive’s alleged tying agreement was induced by a communicated danger of termination____ The plaintiffs in this case must present direct or circumstantial evidence that Comprehensive and its franchisees had a conscious commitment to a common scheme designed to achieve an unlawful purpose____ This concept of a conscious commitment or a meeting of the minds means more than a mere showing that any franchisee conformed or acquiesced in the use of Comprehensive’s mainframe or its approved mini computer for its data processing. It means that the franchisee unwillingly agreed to use Comprehensive for its data processing, an agreement induced by a fear of termination. The first paragraph of this instruction is lifted straight from Perma Life. The plaintiffs say that the second and third paragraphs required the jury to find that the franchisees not only knuckled under to Comprehensive but also wanted to produce anticompetitive effects. This additional requirement, they say, was erroneously smuggled into the instructions from Monsanto. Certainly the plaintiff in a tying case need not show that it wanted the anticompetitive scheme to succeed; the buyer in a tying case is a victim, and few victims want to pay monopoly overcharges. But we do not read the instructions as plaintiffs do. The court had somehow to get the jury to distinguish between the voluntary purchase of two products together, which is not a tie at all, and submission to the seller’s forcing, which is both the definition of a tie and (under Perma Life) the definition of the agreement. The last sentence makes the point in a way lay jurors can understand. Plaintiffs received the Perma Life instruction to which they were entitled. II The court told the jury that in order to find a violation of the antitrust laws, it had to find six elements: (1) that data processing and the franchise are separate products; 1 (2) that the products were tied; (3) *671 that there was an agreement; (4) that Comprehensive had market power; (5) that the plaintiffs suffered direct injury; and (6) that the damages were reasonably ascertainable. The court also instructed the jury that Comprehensive was entitled to use a tie-in if that was the “least restrictive means” of upholding its standards under its trademark. See Benjamin Klein & Lester F. Saft, The Law and Economics of Franchise Tying Contracts, 28 J.L. & Econ. 345 (1985) (discussing the rationale of this defense). Plaintiffs challenge several of these instructions, most vigorously those concerning the justification of protecting the trademark. These instructions may have a lot more to do with plaintiffs’ defeat than the conspiracy instruction. After all, six plaintiffs prevailed on their contract claims, and several recovered damages based on the overcharge they paid for Comprehensive’s service. The jury must have found that they were forced to (and did) use Comprehensive’s computing services on pain of termination. Under the district court’s instructions, they therefore also would have found a combination satisfying § 1. The plaintiffs must have lost on their antitrust claims for some other reason. We need not sift through the other instructions, however, because plaintiffs’ case has a fatal weakness. They did not establish market power. They failed as a matter of law. This failure makes every other element of the anti-trust case irrelevant. The purpose of the rule against certain tying arrangements is to stop the extension of market power from one product to another. Hyde, supra, 104 S.Ct. at 1558-60 & n. 20; Northern Pacific Ry. v. United States, 356 U.S. 1, 78 S.Ct. 514, 2 L.Ed.2d 545 (1958); International Salt Co. v. United States, 332 U.S. 392, 68 S.Ct. 12, 92 L.Ed. 20 (1947); Carl Sandburg Village Condominium Ass’n No. 1 v. First Condominium Development Co., 758 F.2d 203 (7th Cir.1985). The idea is that a firm with a monopoly of one product may refuse to deal except on terms that will lead to a monopoly of another. It also may be possible to use tying arrangements to extract a higher profit through price discrimination. 2 Both the extension of power and the practice of price discrimination are impossible unless the seller has substantial market power. This means power over price, see Hyde, supra, 104 S.Ct. at 1566 n. 46, the ability to induce buyers to pay more money by cutting back the supply of goods available for purchase. See Broadcast Music, Inc. v. Columbia Broadcasting System, Inc., 441 U.S. 1, 9-10, 19-20, 22 n. 40, 99 S.Ct. 1551, 1556-57, 1562-63, 1564 n. 40, 60 L.Ed.2d 1 (1979); National Collegiate Athletic Ass’n v. Board of Regents, - U.S. -, 104 S.Ct. 2948, 2961-63, 2967-68, 82 L.Ed.2d 70 (1984). The best way to show power over price is to establish directly that the price of the tied package is higher *672 than the price of components sold in competitive markets. The early tying cases involved patented products, and the Supreme Court assumed that the patents conferred market power. Other eases have required the plaintiffs to prove the defendant’s power, and the Supreme Court’s two most recent tying decisions have gone for defendants in light of the plaintiffs’ failures. In Hyde the Court assumed that the defendant had a market share of 30% of surgical procedures, which it exploited to force patients to take anesthesia from an unwanted provider; still, it concluded, this was not the sort of power that a plaintiff must show, and it cited earlier monopolization cases to suggest that the plaintiff must establish more than just a substantial share of all sales. Hyde, supra, 104 S.Ct. at 1566-68 & n. 43. The Vertical Restraint Guidelines of the Department of Justice accurately reflect this authority in treating as lawful any tying arrangement affecting less than 30% of a relevant market. Vertical Restraint Guidelines § 5.3 (1985). United States Steel Corp. v. Fortner Enterprises, Inc., 429 U.S. 610, 97 S.Ct. 861, 51 L.Ed.2d 80 (1977) (Fortner II), considered “uniqueness,” the other traditional way to establish market power in tying cases. U.S. Steel offered nonrecourse, low-interest, 100% credit to those who bought its prefabricated houses; you could not get the loan without taking the house. The Court treated this as unique. No one else offered a similar package. Yet it held that as a matter of law factual uniqueness was not enough; in order to establish market power by showing a unique package, the plaintiff must prove that “the seller has some advantage not shared by his competitors in the market for the tying product.” 429 U.S. at 620, 97 S.Ct. at 868. In other words, the plaintiff must show a barrier to entry that prevents competition. 3 If rivals may design and offer a similar package for a similar cost, there is no barrier, and without a barrier there is no market power. Fortner II, supra, 429 U.S. at 622, 97 S.Ct. at 868; “Without any evidence that the [seller] had some cost advantage over its competitors— or could offer a form of financing that was significantly differentiated from that which other lenders could offer if they so elected — the unique character of its financing does not support the conclusion [of] market power”. See Spartan Grain & Mill Co. v. Ayers, 735 F.2d 1284 (11th Cir.1984), cert. denied, - U.S. -, 105 S.Ct. 785, 83 L.Ed.2d 779 (1985), which overturned a jury verdict and held as a matter of law that “uniqueness” means the inability of a seller’s rivals to offer a similar package, not simply the fact that no rival has chosen to do so. See also W. Kenneth Jones, The Two Faces of Fortner: Comment on a Recent Antitrust Opinion, 78 Colum.L. Rev. 39, 40-43 (1978); Kenneth W. Dam, Fortner Enterprises v. United States Steel: “Neither a Borrower, Nor a Lender Be”, 1969 Sup.Ct.Rev. 1, 23-28 (an analysis of “uniqueness” repeatedly cited with approval in both Fortner II and Hyde). Plaintiffs did not claim, let alone offer evidence to show, that the price they paid for the franchise-computation package is higher than the price for those two products purchased in separate markets. The evidence — which shows that franchisees’ businesses were worth more as a result of *673 being members of the Comprehensive system — tends to show that the package Comprehensive furnished on the whole was beneficial, not priced at a monopoly level. They therefore failed to show market power directly. Kypta v. McDonald’s Corp., 671 F.2d 1282 (11th Cir.), cert. denied, 459 U.S. 857, 103 S.Ct. 127, 74 L.Ed.2d 109 (1982), holds that unless the plaintiff shows that the package price was elevated the suit must be dismissed without further ado. That is a sound position, for unless the package price exceeds the competitive price of its components, we have a replay of Fortner II. There U.S. Steel used “cheap financing in order to sell expensive houses” (429 U.S. at 622, 97 S.Ct. at 868); here, for all the evidence shows, Comprehensive licensed its trademark cheaply in order to sell expensive computation. . To the extent plaintiffs may use market share or uniqueness as proxies for power over price, they failed on both counts. There is no evidence about Comprehensive’s market share. The whole Comprehensive system is a pygmy among the large, national firms. There are also thousands of smaller firms. Many businesses produce accounting services internally, and these in-house services also are part of the market because they are substitutes for what the plaintiffs do. Plaintiffs’ principal argument was uniqueness. All they proved, however, is that Comprehensive’s franchising system is unusual; there are few similar systems, and Comprehensive may be the most successful franchisor of accountants. Spartan holds that such a showing is inadequate. Plaintiffs did not show or try to show that Comprehensive has a cost advantage over rivals and potential rivals, that there is a barrier to entry into the business of franchising. They did not show or try to show that rivals could not produce a similar package for a similar cost; without such a showing, they must lose. 4 The Supreme Court emphasized in Hyde, 104 S.Ct. at 1558-59,1565, and again in National Collegiate Athletic Ass’n, supra, 104 S.Ct. at 2962 n. 26, that tying may have competitive benefits. Sometimes the sale of the package is just a way to compete. Some accounting firms have offices throughout the nation and offer comprehensive services; a package franchising system may be a way to compete with the larger firms while retaining the advantages of independent ownership. Just about every conceivable method of organizing service is used in this business — national partnerships with blanket coverage, independent one-man offices, smaller partnerships, franchised systems, and in-house production. Each, including methods that revolve around tied packages, may be beneficial to some customers. In a competitive market the customers will pick the arrangements that work best for them. Antitrust law is based on the premise that when markets are competitive, the process of sellers’ rivalry and buyers’ choice produces the best results. Unless courts insist on a showing of market power, they run the risk of deleting one of the existing options and so re- *674 ducing rather than enhancing the vigor of competition and the welfare of consumers. Comprehensive had no market power. For that matter, it is not to be believed that Comprehensive wanted to monopolize (or posed a threat of monopolizing) computation services, the tied product. Competition in the computation business is exceptionally vigorous; plaintiffs’ contract case depended on establishing that this vigorous competition made it cheaper to secure computation from vendors other than Comprehensive. Even if Comprehensive had some power, then, plaintiffs still must lose. “One of the threshold criteria the plaintiff must satisfy ... is that there is a substantial danger that the tying seller will acquire market power in the tied product market.” Carl Sandburg Village Condominium Ass’n, supra, 758 F.2d at 210. This case therefore did not need to go to the jury. Ill We have bypassed the challenges to the jury instructions on the antitrust issues because, in light of their failure to show market power, the plaintiffs had no case. The contract issues are altogether different. Six plaintiffs prevailed on the contract claims, and the defendants have not challenged the sufficiency of the evidence to support these verdicts. We deal here with the defendants’ argument that the damages are excessive. Then we take up in Part IV the losing plaintiffs’ contentions. The plaintiffs who recovered the largest verdicts are Cahill ($37,678) and Van Winkle ($15,000). They are the only two among the 12 plaintiffs who are still franchisees. Comprehensive maintains that the jury must have awarded damages to Cahill and Van Winkle for the difference between the price they paid to Comprehensive for computation and the cost they would have incurred had they been able to secure computation elsewhere. This award is impermissible, Comprehensive says, because its only breach of contract was the termination (or threatened termination) of those who took their computing business elsewhere. The provision of computation for a price was not itself a breach. (This argument applies to plaintiffs Partridge and Leebelt, too.) Comprehensive also asserts that the verdict for Cahill must have included the costs Cahill incurred in defending an arbitration proceeding Comprehensive commenced to obtain a declaration that Cahill was in breach of contract for not using Comprehensive’s computation. The arbitrator concluded that Cahill’s small computer complied fully with the contract and ordered Comprehensive to retain Cahill as a franchisee in good standing. Under Illinois law, Comprehensive tells us, a party out of pocket the legal costs in one proceeding may not commence another to obtain reimbursement. Ritter v. Ritter, 381 Ill. 549, 46 N.E.2d 41 (1943). The plaintiffs argue that the awards of damages for computing overcharges is appropriate because Comprehensive’s tenacious defense of its computation revenues violated the clause in the contract allowing franchisees to go elsewhere. As for the costs of defending the arbitration, Cahill insists, Illinois permits recoveries to be based on the costs of litigation, when the very institution of the prior suit was a part of the defendant’s wrongful conduct. Sorenson v. Fio Rito, 90 Ill.App.3d 368, 45 Ill.Dec. 714, 413 N.E.2d 47 (1980). Cahill believes that the jury was entitled to find that Comprehensive instituted the arbitration as part of a campaign to make it too costly for the franchisees to stand on their rights. The plaintiffs introduced a. memo written by the president of Comprehensive advising other managerial personnel that “if we do not take action of a punitive nature .the rest of the [franchisees] will feel that it is okay to put in a mini-computer because there was no penalty in so doing.” Although this memo did not discuss the Cahill arbitration, we have been invited to infer that Comprehensive was bent on punishing all franchisees who defected to small computers, no matter the contract. If the district court had granted summary judgment or prevented the matters from *675 going to the jury, we would have two very difficult issues of law. Was the sale of computation under threat of termination a breach in addition to the terminations themselves? Did Comprehensive act in wilful disregard of the contract in commencing the costly arbitration against Cahill? The difficulty is that the judge sent the case to the jury, and the instructions did not ask the jury to consider either question. It was simply told to find the plaintiffs’ damages. Comprehensive objected to the introduction of the evidence of the costs of arbitration, but it did not propose an instruction to the jury describing what it would have to find in order to award these costs as elements of damages. Although part of its argument now may be construed as a claim that the jury could not award these costs even if the arbitration had been instituted in bad faith, it did not propose an instruction excluding these costs from consideration. Fed.R.Civ.P. 51 requires us to assume that the instructions given were correct; in a civil case each party must live with the legal theory reflected in instructions to which it does not object. United States v. Atkinson, 297 U.S. 157, 56 S.Ct. 391, 80 L.Ed. 555 (1936); Tweed’s Case, 83 U.S. (16 Wall.) 504, 515-16, 21 L.Ed. 389 (1872). We have general verdicts. When a party does not ask for an instruction limiting the kinds of damages the jury may award (or informing the jury of the things it must find before it may award an item of damages), we may reverse the district court’s denial of a judgment n.o.v. only if no reasonable juror could have found the evidence sufficient under the instructions it heard. If a party could argue that the evidence was insufficient under the correct legal standard, this would be an indirect method of attacking the instructions, which we have just held defendants may not do. Once the law of this case is settled by failure to object to the instructions, the parties may argue only that the jury did not play its proper part. Because the jury’s part is to apply the instructions to the facts, we may scrutinize its conclusions only to find out whether it did so. It did. Cahill incurred substantial expenses defending the arbitration, and the jury could have found these attributable to Comprehensive’s effort to make an example of Cahill for the edification of the others. One of Comprehensive’s managers told Cahill that the arbitration would serve as an example, and even after Comprehensive lost the arbitration it informed other franchisees it had won, that the arbitrator had required Cahill to conform to all of Comprehensive’s demands. If Comprehensive was using Cahill as an example, and this was part of its breach of contract, the jury was entitled to compensate Cahill. Comprehensive must administer such lessons at its own costs. Similarly, under the instructions actually given, there was evidence from which the jury could have concluded that the excessive charges for computing — backed up by threats of termination that Comprehensive was not entitled to issue — were a breach of contract. IV We come now to the plaintiffs’ efforts to upset the verdicts on the contract claims. There are four: two based on evidentiary rulings, one based on an instruction, and one based on inconsistency among the verdicts. 1. The plaintiffs presented to the jury all of Comprehensive’s machinations to induce them and franchisees in general to keep using Comprehensive’s large computer. They also introduced evidence about the termination of and suits against three other franchisees in August 1981. The district court balked, however, when they tried to show how Comprehensive had reacted to 22 additional franchisees who installed small computers. The court thought this evidence cumulative and excluded it under Fed.R.Evid. 403. The court also kept out the brief Comprehensive filed in the Cahill arbitration, the arbitrator’s “restatement” of the reasons for his award, and portions of the testimony of one witness to a particularly offensive retaliatory act by Comprehensive against a per *676 son not a party here (carting away the files of a franchisee who used a small computer). All of this was cumulative, some unduly prejudicial. The trial spans more than 3,000 pages of transcript, and there was apparently no end to the number of incidents plaintiffs could have produced to show that Comprehensive used every device it could imagine to quash the threat of small computers. The district court was entitled to call an end, once the jury got the general picture. The district judge knows better than we how much is too much; he did not abuse his discretion. See United States v. Watson, 623 F.2d 1198, 1203 (7th Cir.1980). 2. The district court let in one bit of evidence that plaintiffs wanted kept out: the testimony of Thomas A. Mass, Comprehensive’s outside counsel during the time Comprehensive was trying to round up its renegade franchisees. He had given advice to Comprehensive about the meaning of its contract and had served as a lawyer in the Cahill arbitration and several suits against terminated franchisees. The court allowed Mass to testify that he had advised Comprehensive that its acts (and the procedures limiting small computers to those that produced identical reports) were necessary to protect Comprehensive’s trademarks and trade secrets. Plaintiffs say that Mass, having served as counsel in this litigation until he was listed as a witness, should have been precluded from testifying. But an attorney is a competent witness, see Fed.R.Evid. 601, 611; 3 Weinstein’s Evidence 601-33 to -37. When an attorney withdraws from active representation before testifying, the jury is not likely to be confused about his role. The rules of ethics (see Disciplinary Rule 5-102) say that a lawyer should not be both witness and counsel in the same case (although there are exceptions for hardship, see DR 5-101(B)(4)). Mass withdrew as counsel well before he testified, and the district court did not abuse its discretion in permitting Mass to do this. Cf. United States v. Morris, 714 F.2d 669, 672 (7th Cir.1983). 3. Comprehensive filed counterclaims against the plaintiffs, contending that they used its name and trademarks in an unauthorized way. See the Lanham Act, 15 U.S.C. §§ 1051-1127. Although the district court held that Comprehensive’s claim for statutory damages was “equitable” and therefore would not be submitted to the jury, it told the jury that Comprehensive was entitled to terminate franchisees who palmed off their reports as Comprehensive’s or wrongfully used its trade secrets. The plaintiffs concede that the instruction was accurate, but they say it was confusing because their use (or misuse) of the marks and secrets depended on the franchise agreement. If the termination was wrongful, they point out, they were entitled to go on using the marks and methods. The instructions do not establish a vicious circle. Another instruction told the jury that “[i]n order to deal with these claims you must determine first if there was a breach of contract and, second, which party was responsible for breaching or breaking the contract first.” The instructions read together told the jury that if the franchisees breached by palming off their reports as Comprehensive’s (or in some other way), they could be let go; if Comprehensive breached first, they could continue to act as franchisees. Although the franchisees say they were put in a no-win position by Comprehensive’s demand to make all reports “identical” to those produced by its computer — compliance with which exposed them to a claim of “palming off” when Comprehensive refused, as they say, to provide trademarked paper on which to print the reports — the resolution of disputes of this sort is exactly the function of juries. 4. Six franchisees won and six lost. The losers say that there is no rational foundation for the difference. Six franchisees used identical small computer systems; four won and two lost. Two used a different system; one won and one lost. Plaintiffs say that the nature and quality of reports produced by these small systems therefore cannot account for the different *677 verdicts. Comprehensive replies that five of the six losers printed their reports on paper lacking Comprehensive’s trademark, which the contract required them to use; the plaintiffs reply that no one used this paper, because Comprehensive would not sell it to them. Comprehensive tells us that the losing plaintiffs incurred debt to acquire their systems without giving Comprehensive advance notice, a notice required by contract to protect Comprehensive’s ability to collect existing debts. (Comprehensive apparently lends substantial sums to its franchisees for a variety of purposes.) The plaintiffs reply that some of the winners incurred unauthorized debt and anyway that they were justified in not telegraphing their punches, given Comprehensive’s hostility to small computers. We cannot resolve this dispute short of becoming the triers of fact. All parties broke their word. The jury was required to find out which breaches came first and were most important. Some plaintiffs may have breached their obligations earlier and more substantially than others, and different cumulations of breaches may explain the divergent results. The outcome may be explicable or it may not. We need not decide. As a rule civil juries must return consistent verdicts. Gallick v. Baltimore & Ohio R.R., 372 U.S. 108, 119, 83 S.Ct. 659, 666, 9 L.Ed.2d 618 (1963); Bates v. Jean, 745 F.2d 1146, 1149-52 (7th Cir.1984). Criminal juries need not be consistent, see United States v. Powell, - U.S. -, 105 S.Ct. 471, 83 L.Ed.2d 461 (1984), but there is a special reason for this: the double jeopardy clause blocks any review of a verdict in favor of the defendant, and a demand for consistency therefore levels everything down to an acquittal; yet because the jury may have meant to do no more than dispense clemency on one count, leveling down is inappropriate. Id. 105 S.Ct. at 477. Cf. Standefer v. United States, 447 U.S. 10, 100 S.Ct. 1999, 64 L.Ed.2d 689 (1980). The civil jury has no power to dispense clemency, and verdicts in the teeth of the evidence may be set right. Inconsistency may reflect nothing but compromise, and compromise dilutes the requirement that verdicts be unanimous. True, compromise may arise from the fact that the evidence is inconclusive. The evidence may leave a jury on the line between the permissible outcomes, and a compromise may accurately reflect the uncertainties in human affairs. In a case with many plaintiffs, the compromise may work out just fine for the defendant (whose total judgment comes out right), but the cost is defeat for some plaintiffs. Each plaintiff is entitled to have his case decided separately, on a preponderance-of-the-evidence standard that itself recognizes the inevitable uncertainties in evidence. So although “[ajppellate courts should be slow to impute to juries a disregard of their duties, and to trial courts want of diligence or perspicacity in appraising the jury’s conduct,” Fairmont Glass Works Co. v. Cub Fork Coal Co., 287 U.S. 474, 485, 53 S.Ct. 252, 255, 77 L.Ed. 439 (1933), they may not abide inexplicable inconsistency when different parties are involved. 5 This conclusion does not get plaintiffs where they want to go. A jury that returns inconsistent verdicts may be *678 ordered to resume its deliberations. See Bates, supra; Barnes v. Brown, 430 F.2d 578 (7th Cir.1970); Cundiff v. Washburn, 393 F.2d 505 (7th Cir.1968). Fed.R.Civ.P. 49 explicitly authorizes this in case of special verdicts. A district court also may find inconsistent verdicts an adequate ground for a new trial. Either procedure removes the inconsistency. The plaintiffs did not ask the jury to continue its deliberations, however; they asked only for a poll of the jury. 6 Although they asked for a new trial, they restricted this request to the six losing plaintiffs; the six losers also seek judgment notwithstanding the verdict. The six prevailing plaintiffs have not offered to abide the outcome of another trial. Plaintiffs may not simply assume, however, that the “right” verdicts are the ones in plaintiffs’ favor. The problem here is inconsistency, not a judgment in the teeth of the evidence. The rule that each plaintiff is entitled to have his case determined independently is one source of our conclusion that juries may not compromise across plaintiffs to do right by defendants. It also means, however, that plaintiffs fare no better than if each case had been tried separately. In such a series of 12 trials, each plaintiff could have lost. The plaintiffs are not entitled to judgment notwithstanding the verdict or a new trial unless no rational jury could have brought back a verdict for the defendants. Brady v. Southern Ry., 320 U.S. 476, 479, 64 S.Ct. 232, 234, 88 L.Ed. 239 (1943); Freeman v. Franzen, 695 F.2d 485 (7th Cir.1982), cert. denied, 463 U.S. 1214, 103 S.Ct. 3553, 77 L.Ed.2d 1400 (1983); cf. Allied Chemical Corp. v. Daiflon, Inc., 449 U.S. 33, 36, 101 S.Ct. 188, 190, 66 L.Ed.2d 193 (1980). The evidence here was conflicting; a verdict against all 12 plaintiffs would have been sustained, as a verdict for all 12 would have been. The six losing plaintiffs want another bite at the apple without exposing the six prevailing plaintiffs to jeopardy. The rule that verdicts should be consistent does not tilt the scales in favor of plaintiffs, and they are not entitled to a remedy that locks in their gains while they try for more. Affirmed. Footnotes [1] No mean feat because "franchises" (the tying product here) are just names and methods of doing business, not "products" and some courts have held that as a matter of law there cannot be a tie-in between a name and a product, see Krehl v. Baskin-Robbins Ice Cream Co., 664 F.2d *671 1348 (9th Cir.1982); Principe v. McDonald's Corp., 631 F.2d 303 (4th Cir.1980), cert. denied, 451 U.S. 970, 101 S.Ct. 2047, 68 L.Ed.2d 349 (1981); Jack Walters, supra, 737 F.2d at 704-05 (reserving the question). A tie involves products that may be sold in separate markets. But a method of doing business (the franchise) is not sold separately from the ingredients that go into the method of business. A franchiser and its franchisees are part of a business organization not altogether different from vertical integration. See G. Frank Mathewson & Ralph A. Winter, The Economics of Franchise Contracts, 28 J.L. & Econ. 503 (1985). We need not consider the implications of this approach, however. [2] There has been a debate among both judges and scholars concerning the effects of tying arrangements. The close division on the rationale in Hyde was attributable in part to this. On the scholarly debate, compare Phillip Areeda & Donald F. Turner, V Antitrust Law ff 1129c, 1134b (1980), Robert H. Bork, The Antitrust Paradox 140-44, 372-75 (1978), and Roger D. Blair & David L. Kaserman, Antitrust Economics 381-405 (1985), with Louis Kaplow, Extension of Monopoly Power Through Leverage, 85 Colum.L.Rev. 515 (1985). See also S.J. Liebowitz, Tie-in Sales and Price Discrimination, 21 Economic Inquiry 387 (1983). Many scholars, and at least four Justices, would analyze tying arrangements under the Rule of Reason or permit them outright even when the seller has substantial market power. We need not join the debate, given the conclusions discussed below. [3] The use of barriers to entry as a proxy for market power is familiar in the law of mergers. Unless barriers to entry prevent rivals from entering the market at the same cost of production, even a very large market share does not establish market power. See e.g., United States v. Waste Management, Inc., 743 F.2d 976 (2d Cir.1984) (market share exceeding 50% does not establish power over price when entry is easy); Echliri Manufacturing Co., - F.T.C. - (June 28, 1985). Cf. William M. Landes & Richard A. Posner, Market Power in Antitrust Cases, 94 Harv.L.Rev. 937, 947-51, 962-68 (1981); Areeda & Turner, supra, at f 504 (1978). True, even in the absence of barriers the inevitable delay in - entry by rivals may enable those with a large share of a market to raise the price and make supracompetitive profits. But the possibility that short-run market power exceeds long-run power does not apply to a case, such as this, in which the restraints in question have lasted for many years. Rivals to Comprehensive have long since been able to enter, so that barriers (and not lags) are what matter. [4] To the extent Digidyne Corp. v. Data General Corp., 734 F.2d 1336 (9th Cir.1984), cert. denied, - U.S. -, 105 S.Ct. 3534, 87 L.Ed.2d 657 (1985), holds that a plaintiff may prevail, without establishing power over price in the relevant market, by showing that rivals cannot produce exactly the same package, it conflicts with Spartan, Kypta, and several other cases, including Jack Walters in our circuit. E.g., Systemized of New England, Inc. v. SCM, Inc., 732 F.2d 1030 (1st Cir.1984) (no market power as a matter of law in a unique system); Domed Stadium Hotel, Inc., v. Holiday Inns, Inc., 732 F.2d 480 (5th Cir.1984) (no market power as a matter of law in a factually unique reservation system that occupied more than 20% of its market). Sellers may strive to differentiate their products from others in order to compete for the custom of patrons with slightly different needs or tastes. It would be perverse to turn this ordinary attribute of the competitive struggle into a source of illegality. No one may copy Comprehensive’s trademark or its copyrighted materials precisely, but rivals may create similar items for similar costs. Only when there is a barrier to entry — when rivals’ cost of creating similar items is higher than the full costs of the original creator — may differences in the design of the package be treated as proof of market power. Spartan, Kypta, and the other cases we have cited are in line with Fortner II and Hyde; we follow them rather than Data General. [5] Merchant v. Ruhle, 740 F.2d 86 (1st Cir.1984), holds that general civil verdicts may be inconsistent. See also Alexander M. Bickel, Judge and Jury — Inconsistent Verdicts in the Federal Courts, 63 Harv.L.Rev. 649 (1950). Merchant assumes that the rule in criminal cases should apply in civil cases too; we think not, for reasons we have explained. It is also established that special verdicts of civil juries must be consistent. Inconsistent general verdicts may be much rarer, but there is no good reason to treat them differently from inconsistent special verdicts as a rule. There may be something to the First Circuit’s argument when the inconsistency involves but one party, for this may arise in a borderline case. There is no special virtue to a rule of all-or-nothing when several different claims may cluster right around the boundary between judgment for plaintiff and judgment for defendant; a very small difference may put one claim over while the other fails. Different outcomes on very similar facts therefore may not be inconsistent at all. Whatever the appropriate treatment of one-party cases, inconsistency among parties is impermissible. [6] Barnes and Cundiff hold that a request to have the jury resume its deliberations is the only appropriate response to special verdicts that are inconsistent with general verdicts, see Rule 49(b), and that if a party does not act in time he waives any later challenge. We need not decide whether this is also the rule for inconsistent general verdicts.