Valley Liquors, Inc., is a wholesale wine and liquor distributor in northern Illinois, and Renfield Importers, Ltd., is one of its suppliers. Effective November 1, 1981, Renfield terminated Valley as a distributor of Renfield products (which include such popular brands as Gordon's and Martini & Rossi) in two counties, McHenry and Du Page (and part of a third, which we shall ignore to simplify this opinion). Valley sued, charging that Renfield had violated section 1 of the Sherman Act, 15 U.S.C. § 1, which forbids conspiracies or other agreements in restraint of trade. The case is before us on Valley's appeal under 28 U.S.C. § 1292(a)(1) from the denial by the district court of a motion for a preliminary injunction under section 16 of the Clayton Act, 15 U.S.C. § 26.

Until November 1, Renfield generally sold its products to several wholesalers in the same county. But its sales had not been growing as rapidly in Illinois as in the rest of the country, and it decided to adopt a system of restricted distribution whereby it would sell to one, or at most two, wholesalers in each county. (In some instances, however, the plan resulted in an increase in the number of wholesalers from one to two.) Although Valley was Renfield's largest wholesaler in McHenry and Du Page Counties, accounting for some 50 percent of Renfield's total sales there, the new plan terminated Valley and all of Renfield's other distributors in the two counties except Continental and Romano; they were, however, terminated in some other areas. There is unrebutted evidence that Valley had been selling Renfield products at prices five percent below those charged by Renfield's other distributors in McHenry and Du Page Counties and that Valley's termination followed discussions between Renfield and Continental and between Renfield and Romano in which Continental and Romano had expressed unhappiness at Renfield's terminating them in other areas. There is virtually no evidence concerning Renfield's motivation for the adoption of a more restricted distribution system and the concomitant realignment of wholesaler territories, except that it was a reaction to Renfield's disappointing sales in Illinois.

Valley contends that two distinct restraints of trade can be inferred from these facts. The first is a conspiracy among Renfield, Continental, and Romano to increase the wholesale prices of Renfield products in McHenry and Du Page Counties by cutting off Valley-Valley's termination being a concession demanded by Continental and Romano in exchange for consenting to the proposed realignment, under which they lost some of their territories. This is alleged to be a "horizontal" conspiracy, unlawful without more ("per se") under section 1 of the Sherman Act. The second alleged restraint of trade is the exclusion of Valley, pursuant to its distribution agreement with Renfield, from McHenry and Du Page Counties. Valley argues that this "vertical" restriction is unreasonable and hence unlawful under section 1's "Rule of Reason."

The district judge denied a preliminary injunction against Renfield's termination of Valley because he did not think that Valley had demonstrated that it was likely to win the case if tried in full. If the judge was right in his estimation of Valley's chances of success, he was right to deny a preliminary injunction, regardless of other considerations relevant to the exercise of his equitable powers.

If Continental and Romano had agreed to raise the prices of Renfield products in McHenry and Du Page Counties and to that end had persuaded Renfield (perhaps by threatening to discontinue carrying its products if it did not cooperate with them) to terminate Valley, their pesky low-price competitor, then they and their cat's paw Renfield would be guilty of a per se unlawful restraint of trade. Although there was no direct evidence of such a chain of events-in particular no evidence that Continental and Romano ever communicated with each other about Valley-we are asked to infer from the fact that Continental and Romano (separately) expressed unhappiness at being terminated in some of their sales areas that they demanded and received, as a quid pro quo, the termination of their major competitor in the two counties, Valley. However, this hypothesis is too speculative to compel a trier of fact to infer conspiracy, at least if Renfield may have had independent reasons for wanting to terminate Valley. We are asked to exclude that possibility because Valley was Renfield's largest and lowest-priced wholesaler in McHenry and Du Page Counties, and therefore its best. We follow the argument until "therefore." If Renfield had been content with a policy of maximizing wholesaler price competition, it would not have changed to a system of exclusive and dual wholesalers; it would have thought that the more competing wholesalers it had the better off it was. The adoption of a restricted distribution system implies a decision to emphasize nonprice competition over price competition, which such a system tends to suppress. This does not make restricted distribution good, or even lawful; we shall get to that question in a moment. Right now we are just concerned with whether Renfield may have had reasons for terminating Valley that were independent of the desires of Continental and Romano to be rid of the competition of a price cutter. It may have. That possibility is enough to rebut an inference of collusion with those distributors based solely on the termination of Valley.

We are mindful that Cernuto, Inc. v. United Cabinet Corp., 595 F.2d 164 (3d Cir. 1979), held that it is unlawful per se for a manufacturer to terminate a distributor at the behest of a competing distributor who wants to reduce price competition. Valley's allegation of a horizontal conspiracy is an effort to invoke Cernuto. It is a clumsy effort, since Cernuto is not a horizontal case: a horizontal conspiracy is one between two or more competing sellers; a conspiracy between a supplier and a wholesaler is one between firms in different stages of distribution. Perhaps, though, this is a pedantic distinction, and the important point is that Cernuto condemns the vertical expression of a horizontal desire. But it is not enough that the distributor, that is Continental or Romano, have this desire; the supplier must have it too. See Alloy Int'l Co. v. Hoover-NSK Bearing Co., 635 F.2d 1222, 1226 n.6 (7th Cir. 1980); Edward J. Sweeney & Sons, Inc. v. Texaco, Inc., 637 F.2d 105, 111 (3d Cir. 1980). If Renfield wanted to restrict the distribution of its products in order to be a more effective competitor, the antitrust laws would not forbid it to do so merely because its distributors went along so that they would have less price competition. It is therefore not enough to show that Continental and Romano, acting separately (for if they acted together that would be horizontal action), wanted to get rid of a competitor; there must also be evidence that in terminating Valley Renfield was acceding to their desire rather than acting to promote an independent conception of its self-interest. There was no such direct evidence and the circumstantial evidence was as we have pointed out too tenuous to require the trier of fact to draw the inference that Valley asked him to draw.

There is, we admit, a certain unreality in the careful parsing of motives that Cernuto seems to require. If a supplier wants his distributors to emphasize nonprice rather than price competition, which as we said is the usual reason why he would restrict his distribution, he will be hostile to price cutters because they will make it harder for his other distributors to recoup the expenditures that he wants them to make on presale services to consumers and on other forms of nonprice competition, and of course the undersold distributors will be equally or more hostile. The motive of supplier and distributors alike could thus be described as wanting to eliminate price cutters yet there would be no per se illegality so long as the supplier was not just knuckling under to the distributors' desire for less competition. It is difficult to see how a court could distinguish empirically between such a case and the pure antipathy to price competition envisaged by Cernuto. But the unraveling of this skein can be left for another occasion. It is enough that in this case the plaintiff did not prove an improper motive by its supplier.

We turn to the vertical aspect of the case. If we accept, as on the state of the record we must, that Valley sold at lower prices than the other distributors in McHenry and Du Page Counties, then the territorial restriction pursuant to which it was terminated in those two counties has reduced price competition among wholesalers of the brands supplied by Renfield ("intrabrand price competition"). Valley contends that this reduction establishes a prima facie case of unreasonable restraint of trade which shifts to Renfield the burden of showing an offsetting increase in competition between brands supplied by Renfield on the one hand and brands supplied by other importers or national distributors of alcoholic beverages on the other hand ("interbrand competition").

We reject the casual equation of intrabrand price competition with interbrand competition. The elimination of a price cutter who is taking a free ride on the promotional efforts of competing distributors will tend to stimulate nonprice competition among the distributors at the same time that it dampens price competition among them, so that the net effect on intrabrand competition need not be negative. In any event, the suggestion that proof of a reduction in intrabrand competition creates a presumption of illegality is inconsistent with the test that the courts apply in restricted distribution cases. Building from a suggestive footnote in Continental T.V., Inc. v. GTE Sylvania Inc., 433 U.S. 36, 57 n.27, 97 S. Ct. 2549, 2561 n.27, 53 L. Ed. 2d 568 (1976), the courts have held that the effects on intrabrand and on interbrand competition must be balanced in deciding whether a challenged restriction on distribution is unreasonable. See, e.g., Muenster Butane, Inc. v. Stewart Co., 651 F.2d 292, 296 (5th Cir. 1981); Eiberger v. Sony Corp. of America, 622 F.2d 1068, 1076 (2d Cir. 1980). And it is not generally true of balancing tests that the plaintiff, in order to make out a prima facie case, has only to show that if you put something on his side of the empty balance the balance will tilt his way. The plaintiff in a restricted distribution case must show that the restriction he is complaining of was unreasonable because, weighing effects on both intrabrand and interbrand competition, it made consumers worse off.

Admittedly, this test of illegality is easier to state than to apply, the effects to be weighed being so difficult to measure or even estimate by the methods of litigation. The courts have therefore looked for shortcuts. A popular one is to say that the balance tips in the defendant's favor if the plaintiff fails to show that the defendant has significant market power (that is, power to raise prices significantly above the competitive level without losing all of one's business). That is the approach of the Fifth and Ninth Circuits. See, e.g., Muenster Butane, supra, 651 F.2d at 298; Cowley v. Braden Indus., Inc., 613 F.2d 751, 755 (9th Cir. 1980). The Second Circuit seems divided on the question. Compare Oreck Corp. v. Whirlpool Corp., 579 F.2d 126, 130 n.5 (2d Cir. 1978) (en banc), with Eiberger, supra, 622 F.2d at 1081. We agree with the Fifth and Ninth Circuits. A firm that has no market power is unlikely to adopt policies that disserve its consumers; it cannot afford to. And if it blunders and does adopt such a policy, market retribution will be swift. Thus its mistakes do not seriously threaten consumer welfare, which is the objective that we are told should guide us in interpreting the Sherman Act. See, e.g., Reiter v. Sonotone Corp., 442 U.S. 330, 343, 99 S. Ct. 2326, 2333, 60 L. Ed. 2d 931 (1979). Even if there is some possibility that the distribution practices of a powerless firm will have a substantial anticompetitive effect, it is too small a possibility to warrant trundling out the great machinery of antitrust enforcement.

Since market power can rarely be measured directly by the methods of litigation, it is normally inferred from possession of a substantial percentage of the sales in a market carefully defined in terms of both product and geography. See, e.g., Red Diamond Supply, Inc. v. Liquid Carbonic Corp., 637 F.2d 1001, 1005 (5th Cir. 1981). In this case no evidence of market share was presented. In fact, no market was defined, either in product or in geographical terms, so that we do not have even a rough idea whether Renfield was a big firm in its market or a small firm. Nor did Valley seek to establish Renfield's market power by some alternative route, not involving proof of relevant market and market share.

On this as on all the other issues in this case our comments on the weight of the evidence have reference only to the evidence introduced in support of and in opposition to Valley's motion for a preliminary injunction. We are not prejudging Valley's right to a permanent injunction at the end of the trial. That right depends on the evidence introduced at trial, which for all we know may cure the deficiencies in Valley's proof that require us to order that the denial of its motion for a preliminary injunction be, and it hereby is,

AFFIRMED.