This appeal from a plaintiff class certification of claims brought under § 1 of the Sherman Act, 15 U.S.C. § l,1 alleging illegal tie-in sales, arises from two consolidated actions by franchisees against their franchisor, Dunkin’ Donuts of America, Inc. The district court decided that it was not *1213necessary for each franchisee to prove that he individually was coerced by the franchisor to accept the allegedly tied items; rather it would be sufficient if the franchisees as a group could prove either that the franchisor had a policy to persuade the franchisees to accept the allegedly tied items, or that a large number of franchisees had, in fact, accepted them. Based on this view of the required proof, the district court concluded that common questions would predominate over individual ones and certified a class under Rule 23(b)(3), F.R. Civ.P.2 This interlocutory appeal, challenging the propriety of these rulings, was requested by the district court under 28 U.S.C. § 1292(b)3 and was permitted by a divided panel of this court by order of May 7, 1975. We conclude that the alternative modes of proof which the district court deemed sufficient cannot substitute for proof by each franchisee that he, individually, was coerced to accept the alleged tie-in sales. Accordingly, the class certification— expressly premised on the district court’s contrary view — cannot stand. The district court opinion is reported at 68 F.R.D. 65 (E.D.Pa.1975).
I.
Adversaries will always differ as to the propriety of the grant or denial of class certification and it is to be expected that litigants and district judges may often desire immediate review of that decision. However, this court has taken a strong position that a class certification decision, per se, is not an appealable final order under 28 U.S.C. § 1291. Hackett v. General Host Corp., 455 F.2d 618 (3d Cir.), cert. denied, 407 U.S. 925, 92 S.Ct. 2460, 32 L.Ed.2d 812 (1972). To qualify for interlocutory review in this circuit a class certification decision must be attended by special factors which take it outside the ambit of the general rule. Katz v. Carte Blanche Corp., 496 F.2d 747, 756 (3d Cir.) (in banc) cert. denied, 419 U.S. 885, 95 S.Ct. 152, 42 L.Ed.2d 125 (1974).
The proceedings in the district court were not cursory. The court heard oral argument on the request for class status in March, 1974. Almost a year later it rendered its decision in a 155 page opinion. In a second, 11 page, opinion, setting forth the § 1292(b) criteria, the district court wrote:
[A controlling question of law] exists here on two levels. The matter of class certification itself is “serious to the conduct of the litigation”, both practically and legally. See Katz v. Carte Blanche Corp., 496 F.2d at 755. On the practical level, a number of factors held cognizable by the *1214Court in Katz are present here: (1) the expense to the litigants of discovery on the merits of a class action and of trial of such an action; (2) inhibition of potential settlement caused by uncertainty as to the propriety of class certification; and (3) the saving of time of the District Court. However, even if we were to focus solely upon the legal questions and posit that a “controlling question” does not exist when the sole issue is whether the factual complex of a given case meets the class action requirements of Rule 23, we would find a controlling question here — that is, the propriety of our rejection of the individual coercion doctrine and the correctness of our resultant conclusions (following some 75 pages of discussion of the antitrust law of tying) as to the requisite proof to establish conduct constituting an illegal tie. As we have indicated, our finding of predominance of common questions, essential to determination of the class, was in large measure a function of our conclusion concerning the requisite proof to establish an illegal tie.
While we believe that the conclusions in our opinion regarding the law of tying were correct, the opinion also reflects a rejection of the views of many other District Court Judges who have adopted the individual coercion doctrine. Although we believe we have demonstrated that the individual coercion doctrine initially evolved from a misreliance on certain cases and have also demonstrated that it cannot be maintained in the face of the Supreme Court’s decision in Perma Life Mufflers, Inc. v. International Parts Corp., 392 U.S. 134, 88 S.Ct. 1981, 20 L.Ed.2d 982 (1968), the presence of these differing District Court opinions requires that we declare that there is “substantial ground for difference of opinion” as to the correctness of our decision.
App. at 185a-86a (footnote omitted).
We doubt that this interlocutory appeal would have been accepted by this court had it not involved the question of the district court’s rejection of the doctrine of individual coercion. Accordingly, the focus of our discussion will be on the propriety vel non of that rejection.
II.
Dunkin’ Donuts began as a small chain of company owned coffee and doughnut shops in New England. Today it is the largest coffee and doughnut franchising system in the country. By 1975 — 20 years after Dun-kin’ Donuts began franchising its operation — more than 700 shops dotted the United States, most of them franchises.
Typically, Dunkin’ Donuts provides its franchisees with a “turn-key” operation. It selects suitable sites for doughnut shops and either purchases or leases them. It builds a doughnut shop, and offers the prospective franchisee a lease on the land and building. It also offers to sell him the equipment needed to operate the shop. Appellant contends that it is this package deal aspect of the franchise that often attracts the prospective franchisee: the new businessman has only to turn the key, and he is in business. Dunkin’ Donuts does not manufacture or sell the products, or the ingredients of the products, sold in the shops. Rather, to assure uniform quality of products sold under the trademark, Dunkin’ Donuts maintains a quality control system for the most important supplies used in the operation, and supervises an approved supplier system as to those supplies.
Two actions, seeking damages as well as declaratory and injunctive relief, were filed in 1972 against Dunkin’ Donuts by 14 of its present and former franchisees. One action, Ungar, arises under franchise agreements signed before November 1, 1970, the date Dunkin’ Donuts changed its franchise agreement form. The other, Rader, arises under franchise agreements signed after that date. The actions were consolidated below; insofar as this appeal is concerned, the plaintiffs do not rely on the terms of their franchise agreements. We treat the two actions as one.
The claims asserted against Dunkin’ Donuts were wide-ranging, including counts *1215based on violations of the antitrust,4 patent and securities laws, as well as counts alleging fraud, breach of fiduciary duty, breach of contract and tortious interference with business expectations. Class action status was sought with respect to most, but not all, of the claims; the proposed class consisted of all present and former franchisees, a group estimated to number over 600. The district court certified a class as to three aspects of the case only: “(1) the antitrust tying claims with respect to equipment, signs, real estate and supplies; (2) the contractual claim with respect to the Advertising Fund; and (3) the antitrust claims with respect to the in-term restrictive covenant.” 68 F.R.D. 150 (footnote omitted). The class certification was only on the issue of liability. The district court reserved the question of damages, and the question whether certain individual claims were barred by the statute of limitations, for a separate proceeding should liability be established.
Neither side has briefed the issue of class certification of the Advertising Fund claim; both sides have briefed the issue of certification of the in-term restrictive covenant claim. The district court’s § 1292(b) certification does not include either of those issues. The sole question presented on this appeal is the propriety of class certification of the antitrust tying claims, which question subsumes review of the district court’s conclusions as to the proof requisite to establish an illegal tie-in.
Summarized, it was appellees’ contention that Dunkin’ Donuts had a policy of granting a license to use its trademark only on the condition that the licensee accept certain other items from Dunkin’ Donuts, and that this practice constituted a tying arrangement illegal under the antitrust laws. Specifically, appellees contended that four kinds of items were illegally tied to the trademark: real estate, equipment, signs and supplies. Appellees contended that Dunkin’ Donuts prevented its franchisees from using their own premises, requiring them to lease or sublease the shop premises from Dunkin’ Donuts on onerous terms. With respect to equipment, appellees contended that, prior to 1970, the franchise agreement required the franchisee to purchase his “equipment package” from Dun-kin’ Donuts. In 1970, allegedly in response to the Siegel v. Chicken Delight litigation,5 *1216Dunkin’ Donuts changed the standard form franchise agreement deleting the equipment purchase provision. But, the argument continued, Dunkin’ Donuts’ de facto policy of tying the equipment purchase to the trademark license did not change. Franchisees were given a “woefully inadequate” 30 day option to obtain equipment elsewhere, and those who inquired about it were pressured not to exercise it. Moreover, there was only one approved equipment vendor, and 'there was evidence that it would not sell directly to franchisees. The signs required for a Dunkin’ Donuts store could be purchased from two sources: directly from Dunkin’ Donuts, or from an approved sign vendor. Appellees argued that the apparent freedom of choice was wholly illusory because the only approved vendors were those who secretly paid to Dunkin’ Donuts substantial sums for each sign purchase, raising prices to the franchisee accordingly and eliminating any economic savings that might otherwise have been available.
Dunkin’ Donuts does not now sell or manufacture the supplies necessary to operate a doughnut shop. Instead, with the asserted purpose of assuring uniform quality of products sold,6 it approves certain suppliers to sell to franchisees and, in the alternative, provides quality control specifications which non-approved vendors must satisfy. Appellees contended that this is not a quality control system at all, but a disguised tying arrangement: the quality control specifications are useless, designed only to insure that non-approved vendors do not qualify to sell to franchisees; approved suppliers make payments to Dunkin’ Donuts as quid pro quo for being approved, and raise their prices accordingly. The effect of this system, appellees argued, is to tie to the trademark license the purchase of supplies from a limited group of sellers whose prices pass on to the franchisee the cost of the payments to Dunkin’ Donuts. Appellees’ arguments with respect to the tying aspect of the case stressed that the absence of express contractual tying provisions could have no importance if illegal tying policies could be satisfactorily established in other ways. And appellees emphasized that the focus of their case was not individual instances of illegal conduct, but a pervasive company policy, “firm and resolutely enforced”, to tie the real estate, equipment, signs and supplies to the trademark license.
Appellant both denied the existence of any illegal tie-ins, and insisted that, if there were any illegality, the absence of express contractual tying provisions robbed the case of significant common issues and required that the focus of proof be on the individual franchisee.
Whether a franchisee was attracted by a “turn-key” offer, or compelled to accept a tie, could only be determined on an individual basis. Defendant emphasized the explicit statements in its promotional material and in its form contracts which reflected franchisees’ freedom of choice. It emphasized also the variety of experiences admitted to by plaintiffs in their depositions and interrogatory answers. Some acknowledged dealing with defendant because they could not afford to acquire their own real estate or build their own shop, or buy equipment independently. They never considered dealing with anyone but defendant — the topic never came up. Some acknowledged they purchased equipment from other sources. Some acknowledged they had been persuaded by defendant’s salesmen, not coerced, to deal with defendant. And a few plaintiffs testified that their franchise salesman had told them they must *1217deal with defendant if they wanted the franchise.
Concerning the “approved supplier” claim, defendant argued that the issue was not whether in retrospect some less restrictive means of quality control could be devised, or whether defendant’s specifications were good or bad, but rather whether defendant’s specifications and procedures had been fairly applied. That issue was an individual one.
Appellant’s Brief at 6-8 (footnotes omitted).
III.
Faced with the request for class action certification, the district court had to consider the tying claims within the narrow compass of Rule 23, F.R.Civ.P. The central issue on this appeal is the correctness of the district court’s conclusion that the requirement of Rule 23(b)(3) — that “questions of law or fact common to the members of the class predominate over any questions affecting only individual members” — was satisfied under the circumstances of this case. The district court based its conclusion that common questions predominated on its analysis of the substantive law of tying. Specifically it rejected the so-called doctrine of individual coercion which, broadly stated, requires that a plaintiff alleging an illegal tie-in establish that his acceptance of the tied item was coerced and not voluntary.
The district court’s analysis of the law of tying consumed nearly two-thirds of its 155 page opinion — as the trial court put it: “[W]e have not presented a formulation with the precision of a jury charge.” 68 F.R.D. at 116. Under the rubric of “synthesizing the use-coercion dialogue” the district court drew the following conclusions about the law of tying:
(1.) [T]he individual coercion doctrine is not properly a part of tying law. It is not necessary to prove coercion in order to establish an illegal tie; concomitantly, there can indeed be a “voluntary,” yet illegal, tie. .
(2.) [T]he principal significance of the coercion notion in this area is as one mode of proving use of economic power. .
(3.) [Pjroof of individual coercion is but one of several means of establishing the use of economic power or leverage. We agree with defendant that the individual coercion mode of proof of use is inapplicable in a class context. .
(4.) [In a class context] use of economic power may be established by evidence of a firm and resolutely enforced company policy to influence the franchisees to purchase from the franchisor or its designated sources. [P]ersuasion or influence may be the virtual equivalent of coercion where there is an unequal relationship between the parties, as there is here. .
[Or use of economic power in a class context may be inferred] from the acceptance by large numbers of buyers of a burdensome or uneconomic tie.
Ibid, at 114-15.
Having decided that the individual coercion doctrine was not properly a part of the law of tying but was really just one mode of proof and that the individual coercion mode of proof was inapplicable in a class context, the conclusion that a 23(b)(3) class action was proper — that common questions predominated over individual ones — followed directly. If franchisees were not required to prove that they were individually coerced to accept the tied product, the individual questions presented by that kind of proof would drop out of the liability determination. And the common questions raised by proof of company policy, or by proof that a large number of franchisees accepted a burdensome tie, would predominate. The district court concluded with respect to class certification:
[W]e find that common questions of fact and law exist and predominate in the *1218area of plaintiffs’ equipment, sign, real estate and supplier tying claims. This conclusion is premised mainly upon the emergence from the class action discovery of sufficient evidence of Dunkin’ Donuts company policy in these areas to provide a prima facie substitute for the Siegel II express contractual tie. Needless to say, had we accepted rather than rejected the individual coercion doctrine, we could not have made this finding, for individual questions would clearly have predominated. In the absence of the individual coercion doctrine, we find that the principal issues for trial are issues which involve evidence common to the entire putative class. It is important to add that where a company policy emerges, matters of individual exception (and there may be such matters here) will not alter the result.
Ibid, at 141 (footnote omitted).
We disagree with the district court’s novel approach to the law of tying. The “controlling question of law” certified to this court pursuant to 28 U.S.C. § 1292(b) has not been narrowly formulated; however, we understand it to involve, primarily, the district court’s rejection of the necessity of proof of individual coercion and, secondarily, the district court’s acceptance of the two alternative modes of proof mentioned above. We turn now to these issues.
IV.
A.
In view of the district court’s thorough analysis of the law of tying, we are surprised by the statement that “the Supreme Court has not set forth a coercion requirement in the tying cases.” 68 F.R.D. at 98. We believe that coercion is implicit — both logically and linguistically — in the concept of leverage upon which the illegality of tying is premised: the seller with market power in one market uses that power as a “lever” to force acceptance of his product in another market. If the product in the second market would be accepted anyway, because of its own merit, then, of course, no leverage is involved; in the language of the district court, there is no use of the seller’s market power. We find no ambiguity in the Supreme Court’s language on this point. In Times-Picayune Publishing Co. v. United States, 345 U.S. 594, 605, 73 S.Ct. 872, 878, 97 L.Ed. 1277, 1288 (1953) the Court observed: “By conditioning his sale of one commodity on the purchase of another, a seller coerces the abdication of buyers’ independent judgment as to the ‘tied’ product’s merits and insulates it from the competitive stresses of the open market. But any intrinsic superiority of the ‘tied’ product would convince freely choosing buyers to select it over others, anyway.” The same opinion summarized the law of tying: “The common core of the adjudicated unlawful tying arrangements is the forced purchase of a second distinct commodity with the desired purchase of a dominant ‘tying’ product, resulting in economic harm to competition in the ‘tied’ market.” Ibid, at 614, 73 S.Ct. at 883, 97 L.Ed. at 1293.
Northern Pacific Ry. v. United States, 356 U.S. 1, 5-6, 78 S.Ct. 514, 518, 2 L.Ed.2d 545, 550 (1958), perhaps the fountainhead of tying law under § 1 of the Sherman Act, provided the following exposition of the theory of the law of tying:
For our purposes a tying arrangement may be defined as an agreement by a party to sell one product but only on the condition that the buyer also purchases a different (or tied) product, or at least agrees that he will not purchase that product from any other supplier. Where such conditions are successfully exacted competition on the merits with respect to the tied product is inevitably curbed. Indeed “tying agreements serve hardly any purpose beyond the suppression of competition.” Standard Oil Co. of California and Standard Stations v. United States, 337 U.S. 293, 305-306, 69 S.Ct. 1051, 1058, 93 L.Ed. 1371. They deny competitors free access to the market for the tied product, not because the party imposing the tying requirements has a better prod*1219uct or a lower price but because of his power or leverage in another market. At the same time buyers are forced to forego their free choice between competing products.
The same kind of language and analysis respecting leverage, force and coercion can be found in other leading Supreme Court tying cases. Fortner Enterprises, Inc. v. United States Steel Corp., 394 U.S. 495, 503-04, 89 S.Ct. 1252, 1258-59, 22 L.Ed.2d 495, 505 (1969); United States v. Loew’s Inc., 371 U.S. 38, 49, 83 S.Ct. 97, 104, 9 L.Ed.2d 11, 20 (1962). In view of these teachings, we simply cannot accept the district court’s view that the Supreme Court has not set forth a coercion requirement in tying cases.
B.
Saying that coercion is an element of a tying claim is, of course, not the same thing as saying that it must be established on an individual basis, but we think that the two statements are logically related. What is sufficient to coerce one buyer’s choice may not be sufficient to coerce another buyer’s choice; an item that one buyer might accept voluntarily, another might accept only if forced to do so. The expressions “coercion” and “individual coercion”, in our view, reflect different perspectives on the same phenomenon. In the context of substantive tying law, “coercion” is the relevant term. But in the context of a class action, the question is whether “individual coercion” must be established by each plaintiff. We proffer this distinction with the modest hope of admitting into the semantic darkness of this realm of law some small ray of light, and we will use the two expressions accordingly.
The district court explained the individual coercion doctrine as follows:
Although the individual coercion doctrine has nowhere been explicated in detail, it appears to require proof by the plaintiff of such events and circumstances surrounding the relationship between the franchisor and each franchisee as will demonstrate that the franchisee was coerced into agreeing to an anticompetitive tie, usually of equipment or supplies.
68 F.R.D. at 78.
Notwithstanding its view that no coercion requirement had ever been authoritatively set forth, the district court perceived that a substantial number of district courts and courts of appeals had adopted the rule that individual coercion had to be proved.7 Ibid, at 105. However, it was of the opinion that Federal Trade Commission v. Texaco Inc., 393 U.S. 223, 89 S.Ct. 429, 21 L.Ed.2d 394 (1968) and Perma Life Mufflers, Inc. v. International Parts Corp., 392 U.S. 134, 88 S.Ct. 1981, 20 L.Ed.2d 982 (1968) established that the rule was unsound. We disagree.
The district court concluded:
In terms applicable here, Texaco holds that proof of use of economic power does not require evidence of coercion, but rather that evidence of per*1220suasion or influence will suffice where there is dominance in bargaining power of a franchisor over a franchisee, or, what is essentially the same, where there is evidence of an inherently coercive marketing system. We are satisfied that the same principle applies in the traditional tying situation which is involved here. The importance of this conclusion is that it is a further indication of the unsoundness of the individual coercion doctrine.
68 F.R.D. at 108-09.
Its discussion indicates that the district court was fully aware that the factual and legal complex there presented was very different from the complex involved here. Texaco was an enforcement proceeding brought by the Federal Trade Commission, not a treble damage action brought by private plaintiffs; it was brought under Section 5 of the Federal Trade Commission Act, 15 U.S.C. § 45, not under Section 1 of the Sherman Act, 15 U.S.C. § 1. See Belliston v. Texaco, Inc., 455 F.2d 175 (10th Cir.), cert. denied, 408 U.S. 928, 92 S.Ct. 2494, 33 L.Ed.2d 341 (1972) (finding no Sherman Act violation in a private action by Texaco dealers). It involved an agreement between two major corporations, Texaco and Goodrich, under which Texaco would induce its service station dealers to buy Goodrich tires, batteries and accessories, in return for a commission on the sales. The FTC found:
(1) that Texaco has dominant economic power over its dealers; (2) that Texaco exercises that power over its dealers in fulfilling its agreement to promote and sponsor Goodrich products; and (3) that anticompetitive effects result from the exercise of that power.
393 U.S. at 226, 89 S.Ct. at 431, 21 L.Ed.2d at 398. Based on these ultimate findings, the FTC concluded that the arrangement was an “unfair method of competition” unlawful under Section 5 of the FTC Act and ordered the companies to refrain from entering into such arrangements. The Supreme Court granted enforcement.
We cannot fail to note the very significant legal distinctions between Texaco and the case at bar. Section 5 of the FTC Act is broader in its coverage than the antitrust statutes. It subsumes not only antitrust violations but other conduct as well:
[T]he Federal Trade Commission does not arrogate excessive power to itself if, in measuring a practice against the elusive, but congressionally mandated standard of fairness, it, like a court of equity, considers public values beyond simply those enshrined in the letter or encompassed in the spirit of the antitrust laws.
Federal Trade Commission v. Sperry & Hutchinson Co., 405 U.S. 233, 244, 92 S.Ct. 898, 905, 31 L.Ed.2d 170, 179 (1972); see Times-Picayune Publishing Co. v. United States, supra, 345 U.S. at 609, 73 S.Ct. at 880, 97 L.Ed. at 1290. Furthermore, an enforcement proceeding like Texaco necessarily implicates in some degree the principle of judicial deference to administrative expertise. In fact, in Texaco the Court stressed this limitation on its review function:
While the ultimate responsibility for the construction of [Section 5 of the FTC Act] rests with the courts, we have held on many occasions that the determinations of the Commission, an expert body charged with the practical application of the statute, are entitled to great weight. FTC v. Motion Picture Advertising Serv. Co., 344 U.S. 392, 396, 73 S.Ct. 361, 364, 97 L.Ed. 426 (1953); FTC v. Cement Institute, 333 U.S. 683, 720, 68 S.Ct. 793, 812, 92 L.Ed. 1010 (1948). This is especially true here, where the Commission has had occasion in three related proceedings to study and assess the effects on competition of the sales-commission arrangement for marketing TBA. With this in mind, we turn to the facts of this case.
393 U.S. at 226, 89 S.Ct. at 431, 21 L.Ed.2d at 398. Finally, and most fundamentally, the legal focus was not on the relationship between Texaco and its service station dealers, it was on the agreement between Texaco and Goodrich. The issue was whether Texaco and Goodrich were engaged in unfair competitive practices vis-á-vis Good*1221rich’s competitors. That is a very different issue from the issue whether a franchisor is engaged in illegal tying vis-a-vis its franchisees.7a We simply cannot equate the factual and legal complex in Texaco with the case at bar in any material respect. Accordingly, we are unable to accept the district court’s reliance on that case as authority for the unsoundness of the individual coercion doctrine.
Perma Life, the district court said, “emasculates the individual coercion doctrine.” 68 F.R.D. at 110. Not only do we fail to discern in that case any attempt by the Court to impugn the virility of the doctrine, but we fail to perceive in that case any consideration of the doctrine. Perma Life was not a class action. Perma Life was an action by “Midas” muffler dealers seeking treble damages under the antitrust laws for losses suffered by virtue of restrictive provisions — inter alia, tying, exclusive dealing, resale price fixing — contained in their dealership contracts. The central issue was whether the doctrine of pari delicto applied to bar the dealers’ action because they had participated in the illegal contracts. Holding that the action was not barred, the Court made the following observations about the franchisees’ participation:
Although petitioners may be subject to some criticism for having taken any part in respondents’ allegedly illegal scheme and for eagerly seeking more franchises and more profits, their participation was not voluntary in any meaningful sense. They sought the franchises enthusiastically but they did not actively seek each and every clause of the agreement. Rather, many of the clauses were quite clearly detrimental to their interests, and they alleged that they had continually objected to them. Petitioners apparently accepted many of these restraints solely because their acquiescence was necessary to obtain an otherwise attractive business opportunity. . . . Moreover, even if petitioners actually favored and supported some of the other restrictions, they cannot be blamed for seeking to minimize the disadvantages of the agreement once they had been forced to accept its more onerous terms as a condition of doing business.
392 U.S. at 139-40, 88 S.Ct. at 1985, 20 L.Ed.2d at 990 (emphasis added).
The Supreme Court granted certiorari in Perma Life because the lower court rulings “seemed to threaten the effectiveness of the private action as a vital means for enforcing the antitrust policy of the United States.” Ibid, at 136, 88 S.Ct. at 1983, 20 L.Ed.2d at 988. If a franchisee were barred from suing by virtue of the fact that he had participated. in the illegal franchising scheme, then, obviously, no action could ever be maintained by a franchisee against a franchisor. Dunkin’ Donuts does not assert that any franchisee is barred from suit by virtue of participation in the franchising *1222system. No threat here is posed to the effectiveness of the private action. The issue here is whether a class action can be maintained: the danger posed is not to antitrust plaintiffs, it is to antitrust defendants.8 Accordingly, we find no support for the district court’s view of Perm a Life.
In sum, we disagree with the district court’s view that the Supreme Court has not set forth a coercion requirement in tying cases; and with its reliance on Texaco and Perma Life as establishing the unsoundness of the doctrine of individual coercion. We believe that coercion has been and continues to be an integral part of the law of tying as established by the Supreme Court.9 With regard to the doctrine of individual coercion we believe it is open to us to reject it, as the district court did, or to accept it. Having in mind the particular factual complex before us, we choose the latter course. Our reasons follow.
V.
We are genuinely concerned that the law of tying is becoming a kind of semantic shell game, resting more on key words than on careful analysis. For that reason, we feel compelled to take a pragmatic and prudential, as well as a jurisprudential, view of the problem before us, taking into our perspective both the nature of the franchising industry and the effects which may be perceived to flow from the decision under review.
A.
The franchise system in this country today is not free from problems. Most, if not all, of these arise from the disparity in power and sophistication between franchisor and franchisee. Abuses such as arbitrary franchise terminations and fraudulent promotional schemes have been the object of legislative concern. See Bohling, Franchise Terminations Under the Sherman Act: Populism and Relational Power, 53 Texas L.Rev. 1180, 1180-81 (1975) (citing legislative materials); Ozanne & Hunt, 92d Cong., 1st Sess., The Economic Effects of Franchising (Comm. Print 1971) (recommending, inter alia, legislation to protect franchisees from arbitrary terminations, and full disclosure legislation). See generally, Mariniello v. Shell Oil Co., 511 F.2d 853 (3d Cir. 1975). But, apart from abuses of the system, significant advantages — both economic and social — are perceived to flow from the franchise system of distribution. Franchising is said to: (1) provide opportunity for individuals, including persons of modest means, to go into business for themselves; (2) decrease the risk of failure of individual businesses; (3) decrease economic concentration *1223by providing an alternative to vertical integration of business; and (4) benefit consumers by providing standardized products to an increasingly mobile public. Hunt, The Socioeconomic Consequences of the Franchise System of Distribution, 36 Journal of Marketing 32, 33-35 (1972). Moreover, the sizeable, and apparently expanding, position that franchising occupies in our national economy cannot be gainsaid. Retail sales of franchising firms totaled more than $130 billion in 1972, almost 30% of all retail sales, according to estimates of the U.S. Department of Commerce. Franchising in the Economy 1971-73. The fast food franchises are among the most dynamic and successful in the industry. Ibid.
We do not imagine that many persons are, in any meaningful sense, forced to enter into franchise agreements. It may be that some do so because they have been deceived as to the terms, or the potential profitability, of the franchise; it may be that others do so because they lack sophistication and do not understand or appreciate the details of the bargain. More realistically, we would expect to find that an arrangement apparently reasonable at its inception begins to seem burdensome to the franchisee as the business is successfully established. Only from the successful business can the franchisor effectively seek a continuing return on investment; yet as the venture prospers, the franchisee, in time, may come to regard the arrangement as onerous, restricting his profitability. Viewed this .way, it is apparent that the underlying issues are economic as much as legal. What price may a franchisor properly demand for its “product”?10 By what means may it properly exact that price? Of course, within the narrow compass of this appeal we have no occasion to address such fundamental issues, or even to consider whether such issues are properly within the ambit of judicial cognizance. We have rehearsed these issues, not because they control our decision, but because we ought not to be oblivious to the mainsprings of this litigation.
B.
To prove a per se illegal tie-in, a plaintiff must establish three things. First, *1224he must establish that the conduct in question was a tie-in: “an agreement by a party to sell one product but only on the condition that the buyer also purchases a different (or tied) product.” Northern Pacific Ry. v. United States, supra, 356 U.S. at 5, 78 S.Ct. at 518, 2 L.Ed.2d at 550. Second, he must establish that the seller “has sufficient economic power with respect to the tying product to appreciably restrain free competition in the market for the tied product.” Ibid. at 6. And third, he must establish that “a ‘not insubstantial’ amount of interstate commerce is affected.” Ibid.
Obviously, with respect to the first element, a formal agreement is not necessary, although it is sufficient.11 But, in the absence of a formal agreement, a plaintiff must establish in some other way that a tie-in was involved and not merely the sale of two products by a single seller. This can be done by proof that purchase of one product, the tied product, was not voluntary, i. e., by proof of coercion. The effect of the district court’s formulation is to remove the necessity for proving a tie-in. Under the district court’s formulation it is only necessary to prove salesmanship in connection with the sale of two products by a seller-franchisor with dominant economic power over a buyer-franchisee, and a tie-in is established. The practical effect of this is to substitute proof of economic power for proof of tie-in. Proof of economic power must, perforce, focus on the seller; but proof of a tie-in must focus on the buyer, because a voluntary purchase of two products is simply not a tie-in.
When we consider the district court’s formulation in the context of a franchise class action, we are further persuaded that it lacks prudential as well as jurisprudential support. The district court stated “that the individual coercion mode of proof of use [of economic power] is inapplicable [unnecessary?] in a class context.” 68 F.R.D. at 115. Under the district court’s theory a class of franchisees could prove a per se illegal tying arrangement by establishing: (1) that the franchisor was economically dominant over its franchisees; (2) that the franchisor offered for sale, and sold, more than one product to its franchisees; and (3) that the franchisor had a policy to persuade franchisees to buy its products.
It is difficult to conceive of a franchisor-franchisee relationship in which this could not be established. The very nature of the franchise institution contemplates the presence of a comparatively strong financial entity — the franchisor — that makes available to a relatively weaker financial entity— the franchisee — an attractive business opportunity at a relatively modest initial investment. One of the primary attractions of the franchise system to franchisees is that the franchisor offers a package deal including many, if not all, of the items necessary to operate the business.12 And we cannot imagine any seller that would not have a firm and resolutely enforced policy to persuade buyers to buy what is offered for sale. Thus, inherent in any franchise arrangement are the factors the district court considered sufficient to estab*1225lish prima facie liability, in class action form, for a tie-in. This is not a reductio ad absurdum extension of the district court’s formulation; it is a stark, routine application of it.
As an alternative to proving a policy to persuade, the district court would allow illegal use of economic power to be inferred from proof of “acceptance by large numbers of buyers of a burdensome or uneconomic tie.” 68 F.R.D. at 115. This test, in our view, has the same pragmatic drawback as the policy to persuade test. Assuming that what is “economic” from a franchisor’s point of view is “uneconomic” from a franchisee’s, this test would render prima facie illegal virtually every franchise system involving “large numbers” of franchisees.
But there is another, equally serious, problem with the district court’s alternative theory. Whether we call it “petitio principii ” or “arguing in a circle” or “begging the question”, the brute fact is that this test is based on circular reasoning.13 Obviously, if the question is whether there is a “tie”, proof that large numbers of buyers accepted a burdensome or uneconomic “tie” is not helpful. The “proof” assumes the answer rather than proving it. We understand the argument that proof of acceptance of a burdensome or uneconomic offer of a secondary (“tied”) product is some evidence of coercion. We cannot, however, accept the proposition that such proof, alone, would suffice to establish, prima facie, the coercion element of an illegal tie-in claim.14 Establishing that buyers purchase products A and B from the seller does not establish that the seller ties the sale of product A to the purchase of product B. It merely establishes that buyers purchase products A and B from the seller. Cf. Northern Pacific Ry. v. United States, supra, 356 U.S. at 17, 78 S.Ct. at 524, 2 L.Ed.2d at 556 (Harlan, J., dissenting). It might be that proof of a policy to persuade, or proof of purchase by large numbers of buyers, would be relevant to other issues, under other circumstances. But we cannot agree with the district court that these modes of proof — taken singly or in combination — are proper substitutes here for proof of individual coercion by each franchisee. Moreover, the district court’s proffered alternatives would have the practical effect of shifting the burden of proof in the case, relieving the plaintiff of his burden to prove illegality and imposing on the defendant the burden of disproving illegality.
VI.
The basic assumption inherent in the district court’s fomulation is “that persuasion or influence may be the virtual equivalent of coercion where there is an unequal relationship between the parties, as there is here.” 68 F.R.D. at 115. And the district court’s basic fear seems to be that an “al*1226most metaphysical analysis [would] be required if we were to adopt [a coercion standard]; courts would be obliged to parse the human personality in the most sophisticated terms in an effort to determine the franchisee’s state of mind vis-a-vis the putative (and ill-defined) coercion standard.” Ibid. at 112. We are the first to admit that there can be no bright line distinguishing influence, persuasion, and aggressive salesmanship on the one hand, from coercion, on the other. But it is difficult to conceptualize a more damning denunciation of the private free enterprise system than the thesis that there is no distinction where there is an unequal relationship between buyer and seller. The purpose of the antitrust laws is to stimulate economic competition, the essence of which is the presence of many competing sellers; salesmanship — the art of persuasion and influence — is inherent in competition among sellers. It is only when the buyer’s freedom to choose a given product is restricted that the tying doctrine comes into play: so long as “the buyer is free to take either product by itself there is no tying problem.” Northern Pacific Ry. v. United States, supra, 356 U.S. at 6 n. 4, 78 S.Ct. at 518, 2 L.Ed.2d at 550.
Moreover, we are aware of no principle that courts will not inquire into state of mind where that is relevant; the fact that a given legal line is hard to draw does not excuse judges, and juries, as each case arises, from doing their best to draw it. Consider, for example, the problem of determining degrees of culpability: was the conduct “intentional”, “knowing”, “reckless”, “willful”? These kinds of questions are metaphysical, perhaps, but it is the unfortunate province of the courts to struggle with them. See, e. g., Shaffer v. Schlesinger, 531 F.2d 124 (3d Cir. 1976) (involving the question whether an individual’s conscientious objector beliefs were sincerely held).
While we have found indications of legislative concern over abuses in franchising, we have found no Congressional expression of policy suggesting that franchising, per se, violates the letter or the spirit of the antitrust laws. On the contrary, we are conscious of a general Congressional approbation of the franchising system, particularly in light of the opportunity it provides for small businesses.15 Nor has the attention of this court been drawn to any individual or social interest that is offended by the franchise system such that it might justify the massive judicial pronouncement that we perceive would be implicated by an approval of the district court’s decision; a pronouncement of molar, rather than molecular, consequences that in our view would threaten the viability of the franchise system in our economy.16
We answer the controlling question of law as follows: The district court erred in rejecting the individual coercion doctrine under the circumstances of this case. Where, as here, plaintiff franchisees place no reliance on express contractual tie-ins, each, individually, must prove that his purchases were coerced as an element of establishing a prima facie case of illegal tying.17 The district court’s class certification with respect to the tying claims was expressly premised on its erroneous rejection of the individual coercion doctrine. See page 1218, supra. Accordingly, the certification should not have been granted.
The order certifying a class with respect to the antitrust tying claims will be re*1227versed, and the cause remanded for further proceedings consistent with this opinion.