Full opinion text
GOLDBERG, Circuit Judge: This case presents us with something mundane, something novel, and something bizarre. The mundane includes commercial law issues now well delimited by precedent. The novel aspects of the case center on intriguing and difficult interrelationships between trademark and antitrust concepts. And the bizarre element is the facially implausible — some might say unappetizing— contention that the man whose chicken is “finger-lickin’ good” has unclean hands. Kentucky Fried Chicken Corporation, a franchisor of fast-food restaurants, brought this action claiming that defendants were infringing its trademarks and engaging in unfair competition by their manner of selling boxes and other supplies to Kentucky Fried franchisees. Defendants placed Kentucky Fried’s trademarks on the supplies without Kentucky Fried’s consent, and they allegedly misled franchisees with respect to the supplies’ source and quality. Defendants counterclaimed, asserting that Kentucky Fried’s franchise agreements, which required franchisees to buy supplies from approved sources, constituted an illegal tying arrangement. The district court, in a penetrating opinion reprinted at 376 F.Supp. 1136 (S.D.Fla.1974), ruled in Kentucky Fried’s favor on every issue and enjoined defendants’ activities. Although some of the issues are not without difficulty, and although we find that franchisors must walk a narrow path when including in their franchise agreements clauses requiring franchisees to buy supplies from approved sources, we affirm. I. Facts Colonel Harland Sanders founded the Kentucky Fried Chicken business in the early 1950s. The Colonel prepared chicken in accordance with his own secret recipe, and among the Colonel’s achievements has been to convince much of the American public that his product bears a close resemblance to the southern fried chicken that preceded peanuts as the south’s most famous cuisine. The Colonel no longer owns the business, having transferred it in five different segments. The plaintiff, Kentucky Fried Chicken Corporation, now conducts the business in 47 states, and four unrelated entities conduct the business in the other three states. Although Kentucky Fried owns some retail stores, its primary manner of conducting business, and the one of importance here, is franchising local outlets for its product. The franchise agreements require franchisees to purchase various supplies and equipment from Kentucky Fried or from sources it approves in writing. The agreements provide that such approval “shall not be unreasonably withheld.” Before purchasing supplies from a source not previously approved, a franchisee must submit a written request for approval, and Kentucky Fried may require that samples from the supplier be submitted for testing. Of crucial importance is the fact that Kentucky Fried has never refused a request to approve a supplier. The supplies that are subject to the approved-source requirement include those around which this litigation revolves: three sizes of carry-out chicken boxes, napkins, towelettes, and plastic eating utensils technically known as “sporks.” Kentucky Fried sells these items to its franchisees, but under the franchise agreement the franchisees may also purchase any or all of these supplies from other approved sources. There are nine independent approved sources of cartons and a tenth that is a subsidiary of Kentucky Fried. The specifications for these supplies require, among other things, that they bear various combinations of Kentucky Fried’s trademarks. The marks, now widely known to the American public, include (1) “it’s finger-lickin’ good,” (2) “Colonel Sanders’ Recipe,” (3) the portrait of Harland Sanders, (4) “Kentucky Fried Chicken,” and (5) “Colonel Sanders’ Recipe, Kentucky Fried Chicken.” Upon its formation in 1972, defendant Diversified Container Corporation (Container) began using Kentucky Fried’s marks without its consent. Container used the marks on chicken cartons, napkins and towelettes that it advertised and sold to Kentucky Fried franchisees. Unlike other suppliers who sought and received approval, Container never requested that Kentucky Fried approve it as a source of these products, and in important respects Container’s products failed to meet Kentucky Fried’s specifications. Container garnered buyers for its low-quality imitations of Kentucky Fried’s supplies by making inaccurate and misleading statements. Container’s advertisements invited franchisees to “buy direct and save” and represented that Container’s products met “all standards.” Container affixed Kentucky Fried’s trademarks to the shipping boxes in which it delivered chicken cartons to franchisees. And when asked by franchisees whether Container was an “approved supplier” of cartons, Container employees evaded the question and said that Container sold “approved boxes.” Kentucky Fried brought this suit to enjoin Container’s activities, relying upon the related theories of unfair competition and trademark infringement. Kentucky Fried did not seek damages. Defendants counterclaimed seeking treble damages for purported antitrust violations. The case was tried to the court, which resolved all claims in Kentucky Fried’s favor. The court’s findings of fact are incorporated in its memorandum opinion. See 376 F.Supp. 1136. The court entered an appropriate injunction. On this appeal the central issues are whether the district court correctly held defendants liable on the unfair competition and trademark infringement theories and whether the court correctly held that Kentucky Fried’s franchise arrangements were not shown to violate the antitrust laws. We must also address defendants’ contentions that the district court should have granted a new trial on the basis of evidence allegedly discovered after trial, that the district court lacked subject matter jurisdiction, and that the district court erred in allowing Kentucky Fried to amend its reply to defendants’ counterclaim. We find a kernel of truth in all Kentucky Fried’s contentions and therefore affirm. II. Antitrust Container’s antitrust counterclaim forces us to confront three contentions: (1) that Kentucky Fried’s conduct constitutes a tie-in and thus a per se antitrust violation, (2) that if Kentucky Fried’s approved-source requirement is not a tie it should nonetheless be held to constitute a new category of per se offense, and (3) that in any event Kentucky Fried’s arrangement contravenes the rule of reason. We reject each contention of the triad. Container’s primary contention is that Kentucky Fried has established a tying arrangement in violation of § 1 of the Sherman Act, 15 U.S.C. § 1. That section prohibits “every contract, combination . or conspiracy in restraint of trade or commerce.” For the most part an arrangement runs afoul of the § 1 mandate only if its restraint on trade is unreasonable. See, e. g., Standard Oil Co. v. United States, 221 U.S. 1, 31 S.Ct. 502, 55 L.Ed. 619 (1911); Chicago Board of Trade v. United States, 246 U.S. 231, 38 S.Ct. 242, 62 L.Ed. 683 (1918). A § 1 plaintiff must therefore generally establish the anticompetitive impact of the. conduct it challenges. Certain categories of business arrangements, however, exhibit a high likelihood of anticompetitive impact and offer virtually no prospect at all of enhancing competition. With respect to such arrangements, antitrust plaintiffs need not demonstrate unreasonableness; the conduct constitutes a per se violation of the Sherman Act. Tying arrangements comprise one such category of behavior that is illegal per se. See, e. g., Fortner Enterprises, Inc. v. United States Steel Corp., 394 U.S. 495, 89 S.Ct. 1252, 22 L.Ed.2d 495 (1969); United States v. Loew’s, Inc., 371 U.S. 38, 83 S.Ct. 97, 9 L.Ed.2d 11 (1962); Northern Pacific Railway Co. v. United States, 356 U.S. 1, 78 S.Ct. 514, 2 L.Ed.2d 545 (1958); Miller v. Granados, 529 F.2d 393 (5th Cir. 1976). The per se label indicates that a plaintiff need not demonstrate that the effects of the tie are unreasonable. Indeed, not only is the plaintiff relieved from establishing that the effects are unreasonable, but in addition the defendant is not free to demonstrate that the effects are reasonable or even affirmatively desirable. The competir tive impact of the arrangement simply is not an issue for trial. Unless a defendant can establish certain narrow affirmative defenses, a finding that the defendant’s conduct falls within the category of per se tying arrangements disposes of the case in the plaintiff’s favor. Here, as elsewhere, however, the per se label can sometimes prove misleading. Per se analysis is susceptible to the unwarranted inference that a plaintiff prevails in a tying case merely by finding some way to characterize the defendant’s conduct as a tie. A tie can be generally defined as an arrangement under which a seller agrees to sell one product (the “tying product”) only on the condition that the buyer also purchase a second product (the “tied product”). See Northern Pacific, ■ supra, 356 U.S. at 5-6, 78 S.Ct. 514. To bring a defendant’s conduct within the category of ties that are per se violations of the Sherman Act, however, a plaintiff must go beyond some color-able characterization of the arrangement as fitting this rough definition. A plaintiff must show that the challenged arrangement is in fact a tie: that two separate products are involved and that, in addition to complying with the literal terms of the imprecise definition, the seller’s behavior follows the general pattern found unacceptable in the earlier tying cases. To measure an arrangement against that general pattern we must take into account the principal evils of tie-ins: they may foreclose the tying party’s competitors from a segment of the tied product market, and they may deprive the tie’s victims of the advantages of shopping around. See Northern Pacific, supra, 356 U.S. at 6, 78 S.Ct. 514. Furthermore, an arrangement falls within the category of per se tying violations only if the seller has sufficient economic power with respect to the tying product appreciably to restrain free competition in the market for the tied product and only if a “not insubstantial” amount of interstate commerce is affected. See, e. g., Fortner Enterprises, Inc. v. United States Steel Corp., 394 U.S. 495, 89 S.Ct. 1252, 22 L.Ed.2d 495 (1969); Northern Pacific, supra; Carpa, Inc. v. Ward Foods, Inc., 536 F.2d 39 (5th Cir. 1976). The problem in the case at bar is to determine whether Kentucky Fried’s arrangement is in fact a tie — i. e., whether its behavior follows the genéral pattern found unacceptable in earlier tying cases. We begin with an analysis of tying in the context of franchise operations. The issue has taken on considerable significance in recent years; franchising has increased while tying strictures have grown tighter. In the commonly recurring situation, the tying product is the franchise itself, and the tied products may be such things as the equipment the franchisee will use to conduct the business, the ingredients of the goods the franchisee will ultimately sell to consumers, or the supplies the franchisee will distribute to the public in connection with the main product. As an original matter it is less than self-evident that such arrangements should be treated as garden-variety tie-ins, to be analyzed in accordance with the same tying principles developed in other contexts. Unlike that of the tying party in a prototypal tying case, a franchisor’s own success may depend in large measure on the success of the tie’s “victim”, the franchisee. The franchisor will succeed only by establishing a favorable reputation among the consuming public, and in building that reputation the franchisor must depend largely on the quality of the franchisee’s performance. A franchisor will rarely have an opportunity to explain to a dissatisfied customer that the fault was only that of the particular franchisee. The franchisor may therefore have legitimate as well as illegitimate reasons for restraining the franchisee’s choices in the tied product market. The tie may have a benevolent or malevolent loop. Although in the archetypal case “[tjying arrangements serve hardly any purpose beyond the suppression of competition”, Standard Oil Co. v. United States, 337 U.S. 293, 69 S.Ct. 1051, 93 L.Ed. 1371 (1949), in the franchising context ties may well serve acceptable purposes. Nevertheless, tying principles are fully applicable to franchise sales. See Carpa, Inc. v. Ward Foods, Inc., 536 F.2d 39 (5th Cir. 1976); Warriner Hermetics, Inc. v. Copeland Refrigerator Corp., 463 F.2d 1002 (5th Cir. 1972); cert. denied, 409 U.S. 1086, 93 S.Ct. 688, 34 L.Ed.2d 673 (1972); Siegel v. Chicken Delight, Inc., 448 F.2d 43 (9th Cir. 1971). As these cases make clear, when a franchisor conditions the sale of a franchise on the buyer’s agreement to buy additional products from the franchisor, the law of tying comes into play. In order to establish a per se violation, the tying claimant need only demonstrate the requisite economic power and “not insubstantial” effect on commerce. Moreover, a franchisor who requires franchisees to use trademarked supplies does not escape the impact of tying principles to any extent. The franchisor’s right to prevent others from selling supplies bearing its trademarks must yield to the antitrust laws’ command to open the tied market to competitors. See Chicken Delight, supra, 448 F.2d at 52; cf. Timken Roller Bearing Co. v. United States, 341 U.S. 593, 599, 71 S.Ct. 971, 95 L.Ed. 1199 (1951). Despite this relatively low threshold for invoking the per se doctrine, however, the franchisor retains a potentially significant defense — one designed to accommodate the franchisor’s interests in the franchisee’s performance. The franchisor is free to demonstrate that the tie constitutes a necessary device for controlling the quality of the end product sold to the consuming public. See Carpa, supra, 536 F.2d at 46-47; Warriner Hermetics, supra, 463 F.2d at 1016; cf. Dehydrating Process Co. v. A.O. Smith Corp., 292 F.2d 653 (1st Cir.), cert. denied, 368 U.S. 931, 82 S.Ct. 368, 7 L.Ed.2d 194 (1961) (not in franchising context); United States v. Jerrold Electronics Corp., 187 F.Supp. 545 (E.D.Pa.1960), aff’d per curiam, 365 U.S. 567, 81 S.Ct. 755, 5 L.Ed.2d 806 (1961) (same). Product protection through tying can have a legal legitimacy. As part of this defense, however, the franchisor must establish that the tie constitutes the method of maintaining quality that imposes the least burden on commerce. If there are less burdensome alternatives, a franchisor is, obligated to employ them rather than the tie. See Carpa, supra, 536 F.2d at 47; Warriner Hermetics, supra, 463 F.2d at 1016; cf. Copper Liquor, Inc. v. Adolph Coors Co., 506 F.2d 934 (5th Cir. 1975) (not in tying context). The burden of proof on this issue rests with the franchisor seeking to justify the tie, and the burden is a heavy one. The defense fails in the usual situation because specification of the type and quality- of the product to be used in connection with the tying device is protection enough. The only situation, indeed, in which the protection of good will- may necessitate the use of tying clauses is where specifications for a substitute would be so detailed that they could not practicably be supplied. Standard Oil Co. v. United States, 337 U.S. 293, 306, 69 S.Ct. 1051, 1058, 93 L.Ed. 1371 (1949). With this background we turn to Container’s claim that Kentucky Fried’s arrangement constitutes a tie. The legal tightrope upon which we walk is very taut. Kentucky Fried does not expressly require franchisees to purchase from it the allegedly tied products. Instead, the franchise agreements permit franchisees to purchase the supplies from any source Kentucky Fried approves in writing. At the time of trial there were ten approved sources for cartons, only one of which was an affiliate of Kentucky Fried. Franchisees were free to recommend additional suppliers for approval, and the franchise agreement mandated that Kentucky Fried’s approval “not be unreasonably withheld.” The difference between this arrangement and a traditional tie is readily apparent. Here the franchise agreement does not require franchisees to take the “tied” product (supplies) from Kentucky Fried in order to obtain the “tying” product (the franchise). Franchisees need not purchase a single unit of the supplies in question from Kentucky Fried; they can take their entire requirements from other sources. Container, however, argues that the effect of the agreement is the same as a traditional tie because Kentucky Fried coerces franchisees into purchasing from it the supplies in question. We agree with Container that if such coercion were proved, the per se doctrine would apply. A tie need not be reduced to writing to come within the per se proscription. A tie claimant establishes a tie when it proves that a franchisor makes a practice of coercing franchisees into purchasing supplies or other products from the franchisor. See Response of Carolina, Inc. v. Leasco Response, Inc., 537 F.2d 1307, 1326-31 (5th Cir. 1976). In such cases sale of the franchise is, as a practical matter, conditioned upon sale of the tied supplies. The claimant then need only establish the tying prerequisites — sufficient economic power with respect to the tying product appreciably to restrain free competition in' the market for the tied product and a “not insubstantial” amount of affected commerce in the tied product market. Kentucky Fried concedes that these prerequisites are satisfied here. Kentucky Fried denies, however, that Container succeeded in proving that franchisees were coerced into taking their supplies from Kentucky Fried. The burden of proof on this issue rests on Container; demonstrating the existence of a tie is part of the claimant’s case in chief. See Response of Carolina, Inc. v. Leasco, Inc., supra, 537 F.2d at 1328. The district court resolved the question against Container, concluding that Kentucky Fried had not coerced its franchisees in this regard. Coercion is a question of fact, and we therefore review the district court’s conclusion only to determine whether it is clearly erroneous. See Fed.R.Civ.P. 52. Our review of the record convinces us that Container has not only failed to demonstrate clear .error but has also failed to adduce any support at all for its allegation of coercion. Coercion can be circumstantially established; it need not be an instrument under seal or proven with sound and music. Coercion cannot, however, be merely conjured. It is neither a fantasy nor a figment. It is a fact that must be established, whether inferentially or deductively. A fundamental distinction must be drawn between coercing franchisees to purchase from Kentucky Fried and coercing franchisees to purchase from approved sources. The record is barren of any suggestion that Kentucky Fried engaged in the former type of coercion to any extent at all. Thus Kentucky Fried left franchisees free to purchase the purportedly “tied” products from sources other than Kentucky Fried, sources in which Kentucky Fried had no interest and on whose sales Kentucky Fried earned no commission. We conclude that this arrangement simply does not constitute a tie. A monolithic tie may bring down the wrath of per se guilt, but not every use of string tangles with the antitrust laws. The principal evils of tie-ins are the foreclosure of competitors in the tied market and the denial to buyers of the advantages of shopping around. See Northern Pacific Railway Co. v. United States, 356 U.S. 1, 6, 78 S.Ct. 514, 2 L.Ed.2d 545 (1958). Kentucky Fried’s system, at least on its face, presents neither of these evils in anything like the degree associated with tie-ins. Competitors in the supplies market are not foreclosed from reaching a single potential buyer. Such competitors are subject to the approval requirement, but the record contains no showing that the approval requirement has had the effect in practice of foreclosing competitors in the supplies market. Kentucky Fried’s uncontradicted assertion is that it has never withheld approval from a single supplier who requested it. We might speculate that a new entrant’s arrival might be delayed, but nothing in this record demonstrates that the delay would be substantial, and Container does not aver that the perceived onerousness of this delay prevented it from seeking approval. This record shows no closure of competition; it was an invitational affair. The rope hung loosely. It was not a noose. Turning to the second principal evil presented by tie-ins, we find that it, too, is lacking. The franchisees retain the option to shop around. The option is limited to approved suppliers, but there are ten such suppliers of cartons, and franchisees are free to nominate additional suppliers whenever they can be found. There is no allegation that Kentucky Fried exerts any influence over the terms at which its competitors sell to franchisees, and there has been no showing that the competing suppliers have combined to reduce the benefits of competition to the franchisees. Franchisees might fare better if fifty or a hundred suppliers competed for their business, but a market with ten suppliers and unrestrained entry surely poses a far different problem than the market available to the victim of a traditional tie: one supplier and no entry. We conclude that the tie-in’s second principal evil, like the first, is not present in the approved-source system disclosed by this record. When the victim of an alleged tie-in is not required to buy a single unit of the tied product from the tying party or from any source in which the tying party has an interest or on whose sales the tying party earns a commission, the arrangement simply does not constitute a tie. Kentucky Fried has not imposed a tie-in. That the arrangement is not a tie does not, of course, prevent per se treatment; tying is not the only per se antitrust violation. We deem it inappropriate, however, to add approved-source requirements to the list of per se violations. When business arrangements exhibit consistently adverse competitive effects or are totally without redeeming virtue, per se treatment is desirable. Proving the adverse consequences in particular circumstances may prove difficult and, at any rate, will consume valuable court time. The chance that anticompetitive effects will go undetected and the cost in judicial resources make the prudent course to condemn all arrangements in a given category rather than to attempt to sort the harmful from the harmless. See, e. g., Northern Pacific, supra, 356 U.S. at 5, 78 S.Ct. 514. We are not prepared to say, however, that approved-source requirements are so universally devoid of redeeming virtue that they warrant per se treatment. As we noted in developing the background law of franchise tying, ties themselves are not as completely objectionable in the franchise context as in the contexts in which tying law originally developed. Moreover, franchise arrangements may sometimes create better competitive markets than would otherwise exist. A system under which an independent franchisee’s choices are somewhat restricted may nevertheless prove superior to a system in which retail outlets are owned by the national firm. If, for example, Kentucky Fried had chosen not to franchise local outlets but rather to own them outright, the antitrust laws would leave it relatively free to supply the individual stores solely through the national office. Competition at the national level for Kentucky Fried’s supplies business would continue, just as competition to sell Kentucky Fried the supplies it will in turn sell to franchisees is currently unencumbered. But competition at thn local level would be as nonexistent under a system of national ownership of local stores as it would be under a franchise system utilizing explicit ties. These principles are insufficient to take franchise tying out of the per se arena. The existence of the quality defense assures that if a tie is ever truly essential to maintenance of the franchise method of conducting business, the tie will be permissible. To be sure, cases will undoubtedly occur in which ties, while in no sense essential to retention of a franchise arrangement, will prove beneficial to or convenient for the franchisor, and in such cases we could speculate that the somewhat harsh treatment of ties might induce a company to opt for national ownership rather than franchising. We can safely assume, however, that for the most part the application of tying principles to franchise operations will not affect a business’s decision whether to engage in franchising. When we turn from tying to approved-source requirements, however, the situation is somewhat different. The threat that franchisors will abandon franchising does not affect us, but the potential pro-competitive effects of franchising lead us to proceed cautiously lest we unduly shackle franchisors without achieving discernible competitive benefits. We must encourage business ingenuity so long as it is not competitively stifling. We deal here not with tie-ins, whose adverse effects and lack of redeeming virtue are by now quite familiar, but instead with approved-source requirements. When we become more familiar with large-scale franchising and with approved-source requirements, we may discern that the latter are wholly unnecessary to the former. Indeed, we may one day learn that approved-source requirements are consistently hurtful of competition or that sorting the anti-competitive provisions from the innocuous ones is a task too elusive or time consuming to warrant the effort. It will be time enough, however, to declare such requirements to be per se violations when that day arrives. It is enough to decide today’s cases today and leave future cases to the wisdom and experience of future years. Economic decisions derive from contemporary economic analysis. The Supreme Court adopted this cautious approach when first confronted with vertical territorial restrictions, saying We do not know enough of the economic and business stuff out of which these arrangements emerge to be certain. . We need to know more than we do about the actual impact of these arrangements on competition to decide whether they have such a “pernicious effect on competition and lack any redeeming virtue” [Northern Pac. R. Co. v. United States, 356 U.S. 1, 5, 78 S. Ct. 514, 2 L.Ed.2d 545 (1958)] and therefore should be classified as per se violations of the Sherman Act. White Motor Co. v. United States, 372 U.S. 253, 263, 83 S.Ct. 696, 702, 9 L.Ed.2d 738 (1963). We think we should likewise adopt that prudent course here. We are unable at this time to declare approved-source provisions per se violations. Our conclusion is that Kentucky Fried has not committed a per se antitrust violation: it has not established a de facto tie through coercive tactics, and its approved-source provision is not a per se violation. Container’s attack on Kentucky Fried’s arrangement is not yet exhausted, however, for the rule of reason remains. An antitrust claimant who unsuccessfully seeks to establish a per se violation may nonetheless prevail by showing that its adversary’s conduct unreasonably restrains competition. See Fortner Enterprises, Inc. v. United States Steel Corp., 394 U.S. 495, 499-500, 89 S.Ct. 1252, 22 L.Ed.2d 495 (1969). The burden of proving unreasonable effects rests with the antitrust plaintiff. In the case at bar Container has failed to carry this burden. First, Container has not demonstrated that Kentucky Fried’s arrangement adversely affects competition. Indeed, Container has presented no evidence at all of the actual competitive effect of Kentucky Fried’s system. Antitrust claims need not be established by euclidean proof, but they cannot be merely fantasized. For all that appears in this record, competition among suppliers of the franchisees is as open and vigorous as it would be under a system in which Kentucky Fried exerted no control at all over franchisees. Kentucky Fried has excluded not a single supplier from the market; it has narrowed the negotiations between franchisees and their suppliers in not a single respect. The approved-source provision is hardly a boon to competition, but on this record we can only conclude that this approved-source requirement is as innocuous as any could be. Unless we were willing to condemn all approved-source requirements, we could not condemn this one. We have refused, however, to make such provisions per se violations, and Container’s failure to adduce evidence of this provision’s adverse impact therefore defeats its claim. Moreover, Container’s proof is deficient in another respect. Even if a franchisor’s conduct adversely affects competition, the conduct does not contravene the rule of reason if it is designed to control the quality of the franchisee’s product and if the gain in quality is more beneficial than the attendant detriment to competition. Here, Kentucky Fried seeks to justify its approved-source requirement as a device for controlling quality. Kentucky Fried’s argument possesses a substantial measure of intuitive appeal. A customer dissatisfied with one Kentucky Fried outlet is unlikely to limit his or her adverse reaction to the particular outlet; instead, the adverse reaction will likely be directed to all Kentucky Fried stores. The quality of a franchisee’s product thus undoubtedly affects Kentucky Fried’s reputation and its future success. Moreover, this phenomenon is not limited to the quality of the chicken itself. Finger-lickin’ good chicken alone does not a satisfied customer make. Kentucky Fried has a legitimate interest in whether cartons are so thin that the grease leaks through or heat readily escapes, Jm whether the packet of utensils given a carry-out customer contains everything it should and in whether the towelette contains a liquid that will adequately perform the Herculean task of removing Augean refuse from the customer’s face and hands. Kentucky Fried contends that by approving sources only if they comply with minimum standards, it ensures that the various supplies will not be of such poor quality that customers will be alienated. Container strongly counters that the quality control program is a sham, but aside from its own vociferous lamentations Container marshals no support for its contention. In this respect it is important to note that Container bears the burden of proof; we are now analyzing quality control as an element of Container’s claim that the approved-source provision is an unreasonable restraint of trade. We must emphasize the distinction between quality control as an affirmative defense to a per se tying violation — with the defendant bearing the burden of proof and having to establish that the tie-in is the least'’burdensome method for effectively controlling quality — and quality control as a factor in determining whether the defendant’s conduct accords with the rule of reason. We deal here with the latter situation. Kentucky Fried’s reliance on the quality control rationale therefore is not necessarily misplaced solely because less burdensome alternatives for controlling quality are available. The presence of such alternatives is a factor to be considered in the reasonableness analysis, but it is not necessarily the decisive factor. Container has failed to establish that Kentucky Fried’s system is not a reasonable means of controlling quality. Kentucky Fried must not be compelled to subject its chicken to the vagaries of unsuitable packaging. We therefore conclude that Container has not prevailed on its rule-of-reason contention, both because it has failed to demonstrate adverse competitive impacts and because it has failed to show that Kentucky Fried’s system is not a reasonable method for achieving quality control. The district court correctly held for Kentucky Fried with respect to Container’s antitrust counterclaim. This knot was not conceived as a loophole in our antitrust statutes. III. Unfair Competition We turn now to Kentucky Fried’s claims for affirmative relief against Container’s activities. The district court held for Kentucky Fried on the basis of both unfair competition and trademark infringement, and parts of the injunction respond to each theory. We therefore find it necessary to address each contention. The first theory upon which Kentucky Fried relies is unfair competition. Unfair competition is a common law tort that occurs when one business entity “palms off” its products as those of another. See, e. g., Boston Professional Hockey Association, Inc. v. Dallas Cap & Emblem Manufacturing, Inc., 510 F.2d 1004 (5th Cir. 1975), cert. denied, 423 U.S. 868, 96 S.Ct. 132, 46 L.Ed.2d 98 (1975); B.H. Bunn Co. v. AAA Replacement Parts Co., 451 F.2d 1254 (5th Cir. 1971). The determinative question is whether the tortfeasor’s practices are likely to mislead customers into believing that the product emanates from or has been endorsed by the claimant. A claimant need not demonstrate that any customers have suffered actual confusion; the test is likelihood of confusion. See, e. g., World Carpets, Inc. v. Dick Littrell’s New World Carpets, 438 F.2d 482, 489 (5th Cir. 1971). Determining likelihood of confusion entails findings of fact at two distinct levels. The first level consists of the primary activities of the alleged tortfeasor. Did the alleged tortfeasor use the plaintiff’s name? Did it use pictures of the plaintiff’s products in its advertisements? Did it evade potential customers’ inquiries with respect to its connection with the plaintiff? Such questions require findings of fact; to clarify the analytical process we have referred to these primary findings as “digits”. See B. H. Bunn Co. v. AAA Replacement Parts, 451 F.2d 1254 (5th Cir. 1971). Our review of these findings is of course circumscribed by the “clearly erroneous” standard. See Fed.R.Civ.P. 52(a). To be clearly erroneous a finding of fact must leave us with the “definite and firm conviction that a mistake has been committed.” B. H. Bunn Co. v. AAA Replacement Parts Co., 451 F.2d 1254, 1260 (5th Cir. 1971), quoting Chaney v. City of Galveston, 368 F.2d 774, 776 (5th Cir. 1966). In the case at bar the district court compiled an impressive array of fourteen separate findings of fact, or digits, that contributed to its unfair competition holding. These digits provide the proper background for addressing Container’s arguments, and we therefore set them forth here: (1) The Defendants’ cartons bear displays of the Plaintiff’s trademarks, trade name and trade dress colors, but the Defendants’ cartons do not have the Defendants’ own name printed on them. (2) The Defendants’ advertisements are printed in Plaintiff’s red and white striped trade dress and colors. (3) The Defendants’ envelopes for its advertisements and brochures are printed in Plaintiff’s red and white striped trade dress and colors. (4) The Defendants’ envelopes and advertisements contain elaborate displays of the Plaintiff’s trademarks on them. (5) The Defendants’ envelopes, advertisements and products do not contain any notice that the Defendants are not authorized by or a part of KFC Corp. (6) . . . contains the statements “take advantage of your option to buy direct and save”, “Save 50$ per case on KFC boxes that meet exact specifications and quality of the boxes you are now buying”, and “BUY DIRECT AND SAVE.” (7) . . . of Defendants’ advertisements, states: “Meets all standards.” (8) . . . of Defendants’ advertisements, states: “Buy KFC boxes direct and save $1.00 more.” (9) . . one of the Defendants’ envelopes, states: “Now buy direct and save $2.00/M on KFC boxes.” (10) The Defendant’s invoices . describe the Defendants’ products for which invoices are sent to purchasers as “KFC dinner boxes”, and “KFC snack boxes.”. (11) The Defendants’ food cartons are packed and shipped to purchasers in shipping cases that also bear the Plaintiff’s trademarks and service marks. (12) The shipping cases in which the Defendants’ cartons are packed and shipped bear the part numbers, No. “6015002” and “6014002” on the outside. These part numbers are not exactly the same as Plaintiff’s standard part numbers for cartons (including No. 6020041 and 6020051), but Defendants’ part numbers are obviously very similar to those used by the Plaintiff. And the evidence has established that the Defendants do not use these part numbers in invoicing goods. Since the Defendants only sell a total of four products, their use of these seven-digit part numbers appears calculated to mislead. (13) Defendants’ employees, when asked during their telephone solicitations whether Container is one of Plaintiff’s “approved suppliers” of cartons, avoid answering and represent to such potential customers that Container sells “approved boxes.” (14) As an additional “digit” in this analysis, the Court finds that Samuel Alpert, president of Folding Cartons, Inc., E. John Tamblyn, Jr., sales manager for Folding Cartons, Inc. and one-half owner of Container, and Sanford Gubernik, president and one-half owner of Container combined and contrived to deceive the Plaintiff as to their association and purposes. The purposes included the omission of Folding Cartons, Inc.’s name or mark from the cartons it manufactured and sold to Defendant Container in order to make it difficult or impossible for the Plaintiff to learn the identity of the manufacturer of the cartons that were being sold by Container. This association resulted in profit to Folding Cartons, Inc. and Container by virtue of their utilization of thinner and cheaper board which did not meet Plaintiff’s specifications in their cartons, a fact concealed from Plaintiff. The Court, finds that Samuel Alpert, acting for Folding Cartons, Inc., deliberately sold cartons manufactured from thinner board to Container knowing that those cartons did not meet the Plaintiff’s thickness specifications. The Court further finds that cartons manufactured from this thinner board were provided by Folding Cartons, Inc. to Container and that Container put these goods into commerce bearing the plaintiff’s trademarks both before .and after Plaintiff learned of the thinner board’s utilization and rejected its use. 376 F.Supp. at 1143-44. Container does not challenge a single finding as clearly erroneous. Instead, it advances the seemingly remarkable contention that- despite these facts it has not been guilty of unfair competition. To assess Container’s argument we must advance to the next level of unfair competition review. As we indicated, the digits comprise only one of the two levels of findings of fact involved in an unfair competition case. The second level is the determination whether the defendant’s activities, considered as a whole, evince a likelihood of confusion and therefore constitute unfair competition. Likelihood of confusion is a finding of fact, and once again our review is circumscribed by the “clearly erroneous” standard. See, e. g. Volkswagenwerk Aktiengesellschaft v. Rickard, 492 F.2d 474, 478 (5th Cir. 1974). The district court’s conclusion that the digits it listed totaled to a finding of likely confusion cannot be said to be clearly erroneous, and we do not understand Container to assert the contrary. One cannot read the digits set forth above and reach any conclusion other than that Container intended to mislead potential customers and was likely to succeed. Container, however, invokes the established principle that when a district court labors under a misapprehension concerning the governing legal norms, the “clearly erroneous” standard no longer circumscribes appellate review. - See, e. g., Continental Motors Corp. v. Continental Aviation Corp., 375 F.2d 857 (5th Cir. 1967). An appellant who relies on this principle must first pinpoint a controlling legal standard that the district court misapplied. Container has arT ticulated its attack somewhat imprecisely; we are not entirely certain which legal standard Container believes the district court misconstrued. The court clearly, understood that the determinative issue in an unfair competition case is likelihood of confusion, and it also properly utilized our Bunn analysis to specify the digits on which it relied in assessing likelihood of confusion. The court’s statement of the law and its analytical technique were impeccable.- Container’s attack is apparently directed to the propriety of utilizing several of the individual digits as factors in the analysis. As we said in Bunn, the question whether a particular digit can properly be taken into account is a question of law. See 451 F.2d at 1263. When a district court considers a digit'that, as a matter of law, should not be counted against a defendant, the court’s findings lose the shield of the “clearly erroneous” standard. Container contends that, for several reasons, it was entitled to use Kentucky Fried’s trademarks on the supplies it sold to franchisees. The district court, however, listed Container’s use of the marks as part of its first digit. Container’s claim of a right to use the Kentucky Fried marks is not frivolous, as our later discussion of trademark infringement will show. The issue need not be resolved in this litigation, however, and we therefore decline to resolve it. We assume for purposes of analyzing the unfair competition claim that Container did have the right to use the Kentucky Fried marks .on the products it sold to franchisees. The upshot of this assumption-— and we reiterate that it is only an assumption — is not necessarily that the district court erred when it considered as an unfair competition digit the fact that Container’s cartons bore Kentucky Fried’s marks and characteristic red and white trade dress. In assessing unfair competition, a court can and often should look to behavior that would be perfectly legitimate standing alone. A court should consider all the circumstances. The argument that particular conduct was legal and can be explained in a completely innocent manner ordinarily affects only the weight to be assigned the conduct in reaching the unfair competition balance; the propriety of including the conduct as a digit is not affected. Here, however, the district court listed Container’s use of Kentucky Fried’s marks as the first digit, and the court apparently viewed the use as illegitimate conduct to be weighed heavily -rather than as otherwise legal behavior to be counted as a small part of the total circumstances relevant to the unfair competition claim. On the basis of our assumption that the use was legal standing alone, therefore, we conclude that the district court erred as a matter of law in its treatment of this digit. In sum, then, we are willing to accept solely for the purposes of argument Container’s contention that the district court’s analysis failed to follow the governing legal standards. The result, however, is not an automatic rejection of the district court’s holding. Rather, the district court’s disregard of the governing legal principles strips its likelihood-of-confusion finding of the protection afforded by the clearly erroneous standard. Digits need not add up to a definitive sum. If among one or more of the digits we find a likelihood of confusion, Container’s elimination of a digit here and a digit there does not militate against our result. So long as the trial judge was correct in a sufficient number to establish confusion he has satisfied by his findings all that need be found to convict Container of legal confusion. Even one digit might be the infinity of unfair competition. Our task is therefore to determine on our own whether the surviving digits add up to unfair competition. We have no difficulty concluding that they do. Container’s advertisements invite franchisees to “buy direct;” we think it self-evident that the phrase is intended to mislead franchisees into believing that Container is also a supplier of Kentucky Fried or is affiliated in some respect with Kentucky Fried. Intent to mislead is relevant, though not essential, to a finding of unfair competition. See, e. g., Volkswagenwerk Aktiengesellschaft v. Rickard, 492 F.2d 474, 478 (5th Cir. 1974). Moreover, we also think the phrase is likely to achieve the intended confusion. The phrase “buy direct” arguably informs a potential customer that the product’s source is not Kentucky Fried itself, but the confusion underlying an unfair competition holding need not be confusion with respect to the seller’s identity. When a franchisor legitimately requires franchisees to purchase from approved sources, an unfair competition holding can be based on the likelihood of confusion with respect to whether the product’s source has been approved. In addition, Container’s cartons do not have its own name printed on them, and Container’s advertisements avoid any notice that its products are not approved by Kentucky Fried. Of even greater significance, Container ships its products in packing boxes that bear Kentucky Fried’s trademarks. Use of the marks on packing boxes could have no purpose or effect other than to mislead customers. Any doubt on this score vanishes when Container’s numbering system is considered. Container affixes a seven-digit number to each shipping case; the number has no purpose in Container’s invoicing system; and remarkably enough the number contains exactly the same digits, albeit rearranged, as does one of Kentucky Fried’s standard part numbers for cartons. The trial judge found a bonus of digits even assuming that he was wrong with respect to some of them. Kentucky Fried augmented this showing of Container’s confusing tactics by producing evidence that Container had achieved its result. Franchisees were actually misled. An unfair competition plaintiff succeeds by establishing likelihood of confusion; actual confusion is not necessary. When there is evidence of actual confusion, however, it provides persuasive support for an inference that confusion is likely. There can be no more positive or substantial proof of the likelihood of confusion than proof of actual confusion. Moreover, reason tells us that while very little proof of actual confusion would be necessary to prove the likelihood of confusion, an almost overwhelming amount of proof would be necessary to refute such proof. World Carpets, Inc. v. Dick Littrell’s New World Carpets, 438 F.2d 482, 489 (5th Cir. 1971). Container’s contention that the showing of actual confusion was statistically insignificant therefore misses the point. See Roto-Rooter Corp. v. O’Neal, 513 F.2d 44 (5th Cir. 1975). The evidence that some Kentucky Fried franchisees were actually misled reinforces our conclusion that Container’s activities created a likelihood of confusion. In sum, Kentucky Fried has produced overwhelming proof that Container’s behavior was designed to create confusion, that it was likely to succeed, and that in some instances it actually succeeded. A more solid showing of unfair competition has rarely been assembled. One cannot read this record without concluding that Container poached on Kentucky Fried’s chicken coop and did so overtly and avariciously. Container fell just short of costuming its sales force in the garb of the Colonel, goatee and all. We uphold the district court’s ruling on this issue. IV. Trademark Infringement Kentucky Fried’s next theory is trademark infringement. Trademark infringement is a narrower aspect of unfair competition; both turn primarily on the likelihood of customer confusion. See generally B. H. Bunn Co. v. AAA Replacement Parts Co., 451 F.2d 1254 (5th Cir. 1971); American Foods, Inc. v. Golden Flake, Inc., 312 F.2d 619 (5th Cir. 1963). Trademark infringement, however, occurs only when the defendant brings about the confusion in a particular manner: by using the plaintiff’s name or marks, or marks confusingly similar. In the case at bar Container has concededly used Kentucky Fried’s marks without Kentucky Fried’s consent. Nevertheless, Container presents a two-pronged defense to the trademark infringement claim. First, Container contends that Kentucky Fried has forfeited its marks by engaging in unrestrained licensing. Second, Container argues that its use of the marks on supplies rather than chicken is not likely to confuse prospective purchasers. We deal with each contention in turn. A. Forfeiture Courts have long imposed upon trademark licensors a duty to oversee the quality of licensees’ products. See, e. g., Denison Mattress Factory v. Spring-Air Co., 308 F.2d 403, 409 (5th Cir. 1962). The rationale for this requirement is that marks are treated by purchasers as an indication that the trademark owner is associated with the product. Customers rely upon the owner’s reputation when they select the trademarked goods. If a trademark owner allows licensees to depart from its quality standards, the public will be misled, and the trademark will cease to have utility as an informational device. A trademark owner who allows this to occur loses its right to use the mark. Container argues that Kentucky Fried has forfeited its marks by inadequately controlling quality in the 47 states for which it owns the marks and by failing to control quality at all in the other three states. With respect to the 47 states in which Kentucky Fried owns the trademarks, however, the record belies Container’s assertion that Kentucky Fried’s program for controlling quality is a sham. Retention of a trademark requires only minimal quality control, for in this context we do not sit to assess the quality of products sold on the open market. We must determine whether Kentucky Fried has abandoned quality control; the consuming public must be the judge of whether the quality control efforts have been ineffectual. We find that Kentucky Fried has sufficiently overseen the operations of its franchisees. Container has failed to carry the heavy burden placed on a party seeking to establish a forfeiture. See American Foods, Inc. v. Golden Flake, Inc., 312 F.2d 619, 625 (5th Cir. 1963). We also find unpersuasive Container’s contention based on Kentucky Fried’s lack of any control at all over Kentucky Fried Chicken outlets in the other three states. Kentucky Fried does not own the marks there; Colonel Sanders conveyed the trademark rights to others before selling the remaining business to Kentucky Fried’s predecessors. The trademark owners in those three states therefore are not Kentucky Fried’s licensees, as Container erroneously contends, but rather concurrent owners. Concurrent ownership of marks in separate geographical territories is clearly permissible. See 15 U.S.C. § 1052(d). Indeed, we have ourselves approved decrees expressly establishing such concurrent usage of identical marks. See American Foods, Inc. v. Golden Flake, Inc., 312 F.2d 619, 625-26 (5th Cir. 1963); Sheila’s Shine Products, Inc. v. Sheila Shine, Inc., 486 F.2d 114 (5th Cir. 1973); cf. Dawn Donut Co. v. Hart’s Food Stores, 267 F.2d 358 (2nd Cir. 1959). Kentucky Fried Chicken customers in 47 states buy the product of plaintiff Kentucky Fried Chicken Corporation, backed by its reputation. Kentucky Fried Chicken customers in Florida, one of the remaining states, buy the product of the Florida owner, backed by its reputation. That many consumers are undoubtedly oblivious to the corporate structures does not undermine the trademark system; Florida customers will associate the trademarks with the Florida product, and national customers will associate the marks with the national product. That an occasional purchaser will travel between geographical districts is not a problem sufficient to justify outlawing concurrent trademark ownership. We conclude that Kentucky Fried has not forfeited its trademarks through unrestrained licensing. The marks are valid. B. Likelihood of Confusion Container’s next contention in defense of the trademark claim is that its use of the marks was not likely to mislead potential customers. Preliminarily, we reject Container’s formulation of its argument. Container articulates its contention in terms of the trademarked products: Kentucky Fried’s marks, says Container, apply only to chicken, not to paper products. We reject this assertion for three reasons. First, although Kentucky Fried’s registration refers only to chicken (and other food products), we believe that closely related products are included as well. We deal here not with Kentucky Fried spaceships or Kentucky Fried writing pens, but with the paper products necessary to the operation of a store that sells Kentucky Fried chicken. As the district court noted, Kentucky Fried could hardly emboss its trademarks upon the chicken itself. There is a symbiotic relationship between Kentucky Fried chicken and its cartons and accoutrements. Under these circumstances we hold that the registration is sufficient to encompass the tangential supplies. The second reason that Container’s argument fails is that Kentucky Fried clearly possesses common law trademarks with, respect to the supplies themselves. Kentucky Fried concededly uses the marks on supplies, and it was the first party to do so. It therefore has the common law marks on supplies even if the statutory registration applies only to chicken. Kentucky Fried’s cartons may not be “finger-lickin’ good,” but that is the cartons’ trademark nonetheless. Finally, Container’s focus on the trademarked products erroneously assumes that infringement occurs only when the alleged infringer uses the mark on the same category of product with respect to which the trademark owner obtained the mark. We have squarely and repeatedly rejected that position. See PGA v. Bankers Life & Casualty Co., 514 F.2d 665 (5th Cir. 1975); Beef/Eater Restaurants, Inc. v. James Burrough Ltd., 398 F.2d 637 (5th Cir. 1968); Continental Motors Corp. v. Continental Aviation Corp., 375 F.2d 857 (5th Cir. 1967). Even if Container were correct that the Kentucky Fried marks applied only to chicken, therefore, Container’s use of the marks on supplies might nonetheless constitute infringement. Container’s attempt to limit the products for which Kentucky Fried owns trademarks provides no defense to the infringement claim. We do believe, however, that the true thrust of Container’s contention poses a far more difficult issue. We think Container’s position can be reformulated in terms of likelihood of confusion. Container apparently argues that its use of the marks on supplies rather than chicken is not likely to confuse potential customers. We deem this the most difficult aspect of the trademark infringement claim. Trademark infringement occurs only when the use sought to be enjoined is likely to confuse purchasers with respect to such things as the product’s source, its endorsement by the plaintiff, or its connection with the plaintiff. See, e. g., World Carpets, Inc. v. Dick Littrell’s New World Carpets, 438 F.2d 482, 488 (5th Cir. 1971); Continental Motors Corp. v. Continental Aviation Corp., 375 F.2d 857 (5th Cir. 1967); Aloe Creme Laboratories, Inc. v. Estee Lauder, Inc., 533 F.2d 256 (5th Cir. 1976). Our cases demonstrate unbroken insistence upon likelihood of confusion, and by doing so they reject any notion that a trademark is an owner’s “property” to be protected irrespective of its role in the operation of our markets. See American Foods, Inc. v. Golden Flake, Inc., 312 F.2d 619, 625 (5th Cir. 1963). Trademarks convey to purchasers a variety of information, and when a competitor’s use of the same or similar marks interferes with this informational function, trademark infringement is established. These principles were not altered by our recent decision in Boston Professional Hockey Association, Inc. v. Dallas Cap & Emblem Manufacturing, Inc., 510 F.2d 1004 (5th Cir. 1975), cert. denied, 423 U.S. 868, 96 S.Ct. 132, 46 L.Ed.2d 98 (1975). There we held that Dallas Cap infringed Boston Hockey’s trademark — the team emblem of the Boston Bruins — by selling shoulder patches depicting the Bruins’ emblem. Acknowledging that the confusion question there was conceptually difficult, we found the confusion requirement satisfied by the “certain knowledge of the buyer that the source and origin of the trademark symbols were in [Boston Hockey].” 510 F.2d at 1012 (emphasis added). In the case at bar the buyers undoubtedly possess certain knowledge that the source and origin “of the trademark symbols” is in Kentucky Fried. By emphasizing this one phrase from our comprehensive opinion, Boston Hockey could therefore be read to dispose of the confusion issue here. We decline, however, to adopt that reading. Boston Hockey also reiterated our unbroken insistence on a showing of confusion, and we believe that our opinion must be read in that context. Under the circumstances there — involving sales to the consuming public of products bearing trademarks universally associated with Boston Hockey — the fact that the buyers knew the symbols originated with Boston Hockey supported the inescapable inference that many would believe that the product itself originated with or was somehow endorsed by Boston Hockey. Buyers may have had no reason to expect Boston Hockey to possess expertise in manufacturing shoulder patches, but to a Bruins fan the club’s endorsement would be much more important than the quality of the stitchery. And Boston Hockey had every right to reap the rewards of its trademark’s popularity. Several factors suggest that the finding of infringement in Boston Hockey need not control here. First, the buyers with whom we deal are not the consuming public but franchisees fully familiar with the corporate configuration underlying the products. The inference is therefore much weaker that the buyers will believe that the source of a trademark-bearing product is the same as the source of the mark itself. Second, the strictures of the antitrust laws deprive Kentucky Fried of any right to preserve for itself all the rewards of its trademarks’ popularity. As our antitrust discussion makes clear, so long as Kentucky Fried requires franchisees to use supplies bearing the marks, it cannot prevent all competing suppliers from using them; to do so would constitute an illegal tie-in. Finally, we cannot conclude that a seller of boxes bearing Kentucky Fried’s marks is necessarily trading on the good name built by Kentucky Fried. Affixing those marks, after all, is utilitarian; the buyer needs boxes that not only hold chicken but advertise its business as well. Under these circumstances we- do not believe Boston Hockey equates knowledge of the symbol’s source with confusion sufficient to establish trademark infringement, and we deem the confusion issue unresolved by our existing decisions. We decline, however, to resolve this intriguing and tantalizing confusion issue that would be presented if we dealt with a seller whose only asserted transgression was the use of Kentucky Fried’s marks. For here we confront trademark usage in conjunction with an elaborate scheme for misleading franchisees into believing that Container was connected with or approved by Kentucky Fried. We need not recount the facts detailed in our discussion of unfair competition. We reiterate only that they conclusively demonstrate that a likelihood of confusion existed. Container’s use of the trademarks was but a part of the larger scheme, and the use of the trademarks contributed to the confusion. When such confusion exists, trademark infringement exists regardless of whether the use of the marks, standing alone, would have created confusion. We therefore hold that Container infringed Kentucky Fried’s trademarks. We note in this respect that the district court enjoined Container from using the marks, and under the injunction Container w