In re Industrial Gas Antitrust Litigation

MEMORANDUM OPINION AND ORDER

GETZENDANNER, District Judge:

Plaintiffs in this antitrust action seek to represent a nationwide class of direct purchasers of industrial gas.¹ Each of the thirteen plaintiffs purchased a significant amount óf gas itself. The court, on December 30, 1981, ordered class certification in part, ruling that the case could proceed as a class action with respect to certain issues, but not others. Defendants (who are the country’s five leading producers of industrial gas)² responded to this decision with a motion asking for leave to take an interlocutory appeal. The court reviewed this pleading, and decided, sua sponte, to treat it as a request to reconsider. Plaintiffs countered with a cross-motion to reconsider the court’s refusal to certify the entire case as a class action.

All told, well over three hundred pages of briefs have been submitted throughout the course of this litigation addressing the applicability of Rule 23 of the Federal Rules of Civil Procedure. The court has carefully reviewed these filings as well as numerous affidavits and other documents of record. It is the court’s judgment that its opinion and order of December 30, 1981 should be vacated,³ that defendants’ motion to reconsider should be granted, and that plaintiffs’ cross-motion to reconsider should be denied. Plaintiffs have not carried their burden of satisfying the requirements of Rule 23.

I. The Industrial Gas Industry⁴

The term “industrial gas” refers for purposes of this litigation to oxygen, nitrogen and argon. Each of these products is fungible in that one seller’s merchandise is indistinguishable from that of a competitor. Industrial gases are sold at commercial levels of purity.

Customers utilize these gases in various industrial and other processes. In certain instances, substitute products and methods are also available. Nitrogen, for example, is often used as a freezing agent, but customers can alternatively satisfy their needs with either carbon dioxide or mechanical refrigeration devices. In these cases, a gas merchant must compete not only with rival gas merchants, but with businesses in the substitute fields.

The production of industrial gas occurs at facilities known as air separation plants, and involves a process called fractional distillation. Air is first liquefied through compression and cooling to temperatures less than -300° F. The liquid mixture thereafter vaporizes, and the three gases separate as their respective boiling points are reached.⁵ Electricity costs constitute a significant portion of the production expenses incurred.

Some of the oxygen and nitrogen thus produced flows directly through a pipeline to a customer’s nearby plant. Such “on-site” or “pipeline” purchasers do not fall within the putative class and will not be mentioned further.

All other product that is sold by defendants to end-users is called “merchant” product. Low volume merchant customers often receive their merchandise in gaseous form stored in high pressure steel cylinders. Defendants package “cylinder gas” in the following manner. First, the gases separated at the air separation plant are compressed and individually reliquefied. The liquid product is then transported by truck to a “filling station,” where the liquid is compressed and vaporized back into a gaseous state. This gas is next pumped into cylinders, and thereafter delivered by truck to the customer’s place of business. Most of the actual containers that are used in this distribution process are owned by the gas producer, and are rented by the customer. When a full cylinder is delivered, empty ones are retrieved.

Other low-volume customers purchase “liquid cylinders,” which are fifty gallon thermos-like structures that are specially designed to retard the evaporation of the liquid they contain. Delivery is once again usually made by truck.

High volume bulk purchasers similarly have the option of receiving either gaseous or liquid product. When the former is purchased, the gas arrives in a large tube trailer pulled by a truck tractor. The majority of bulk sales, however, involve liquid product that is transported from an air separation plant in trucks and rail cars that are constructed to hold the product at temperatures low enough to maintain its liquid state. In general, it is much cheaper to transport liquid rather than gaseous product since a given volume of liquid contains many more molecules of product than does a corresponding volume of gas.

Bulk liquid purchasers must have on their premises a cryogenic storage tank that can sustain the product’s liquefieation prior to its use. (Evaporization losses nevertheless occur so rapidly that, even with such a tank, it is uneconomical for a customer to purchase bulk liquid product if the customer does not intend to use the merchandise in a relatively short amount of time.) Such customers also require special devices for injecting the stored liquid into their business operations. If the product is ultimately to be used in a gaseous state, vaporizing equipment is needed. If the product is to be drawn and used as a liquid, vacuum insulated pipe is then required. Often, the company that provides the product also sells or furnishes these necessary devices.

The cost of transporting industrial gas, even in liquid form, is extremely high relative to product value. As a result, neither liquid oxygen nor liquid nitrogen may be economically shipped more than two hundred miles from the air separation plant from which it is produced. Similarly, cylinder product is rarely sold outside of a fifty to seventy-five mile radius centered at the point of distribution. Because of these economies, defendants assert that the industrial gas market in this country consists not of one overall market, but of numerous, localized sub-markets, each largely insulated from the rest. It can hence be misleading, defendants continue, to argue from the admitted fact that defendants collectively sold more than 80% of the industrial gas marketed in the United States during the period of the alleged conspiracy. In certain sub-markets, regional firms played a significant role.

Defendants further point out that they face considerable competition from numerous independent distributors, particularly in the cylinder gas field. These 1400-1500 independent businesses typically purchase bulk liquid product (usually from one of the defendants)⁶ for reconversion to gaseous form and ultimate resale.⁷ Defendants maintain that they themselves directly account for at most only 25% of all cylinder gas sales nationwide.⁸

In general, bulk-liquid purchasers must pay not only for the actual product they obtain, but also for the monthly rental and upkeep on the cryogenic tank they use. These two prices may be quoted separately or collectively, depending upon the customer’s preference. An additional charge may also be added for the cost of transporting the product to the customer’s tank. Other times, this cost is simply built into the basic product charge.

Most bulk liquid transactions are made pursuant to requirements contracts that last up to five years. These contracts are individually negotiated and contain varying provisions regulating whether, and how often, the seller may raise the stated price over the life of the contract. Some contracts bar all increases; others limit the number of increases allowed per year, or the percentage increase that may be tacked on. Numerous other formulae have been employed. Many contracts further authorize the customer to solicit competing bids whenever the seller attempts to raise the contractual price; in these circumstances, however, the seller often retains a right to match any lower bid that is received.

Cylinder customers, on the other hand, often purchase their goods pursuant to individual purchase orders rather than long-term requirements contracts. The amount they pay is also individually negotiated, and usually reflects both product and delivery charges, as well as a rental fee on the cylinders that are used.

It is apparent from the foregoing that numerous factors potentially influence the amount defendants can charge for a particular product distributed in a particular way under non-collusive circumstances. First, the conditions in the various sub-markets differ. The amount of local competition (both from defendants and regional producers) varies from area to area, as does the significant input cost of electricity. Moreover, even within a sub-market, purchasers may not be similarly situated. One may seek product in order to accomplish a goal that can be satisfied using alternative methods or goods. Such a purchaser will have greater bargaining strength than a second customer who does not have ready alternatives. One purchaser’s bargaining strength may also be relatively stronger simply due to greater size. Different contracts further provide, as shown above, varying amounts of protection from inflationary and other upward pressures on price. Finally, different customers require and receive different amounts of technical assistance in integrating the product they purchase into their businesses.

II. Plaintiffs’ Substantive Claims

In broad form, plaintiffs allege a conspiracy to raise “price schedules,” and a conspiracy to allocate customers. Defendants’ desire to obtain higher revenues allegedly motivated both forms of behavior.

An understanding of the former charge requires further background. Plaintiffs contend that defendants perceived and treated the American market for industrial gas as national in scope, notwithstanding the economic reasons why it should have been localized. See supra. Plaintiffs point out that each defendant maintained production and distribution facilities throughout the country. Moreover, gaps in this system were plugged through defendants’ practice of “swapping” gas with one another. For example, if producer A possessed facilities in New York, but not Los Angeles, and producer B was situated in exactly the opposite manner, both merchants could sell in both markets by “swapping”: A would deliver gas to B in New York in return for shipments in Los Angeles. Affidavits submitted by defendants’ own employees confirm that swaps played an important role in the distribution of industrial gas:

To the best of my knowledge, the Linde Division [of Union Carbide] has exchanged these industrial gases with competitive industrial gas producers since as early as 1956. Product exchanges are common in many industries, for many of the same reasons that they have been such a long standing practice in the industrial gas industry. There are several basic reasons that the Linde Division has engaged in product exchange transactions including: (1) to supplement the product supply of a local plant that may be sold out; (2) to secure product to meet peak demand needs; (3) to reduce transportation expenses, and thus the price, that a customer would pay for transportation for long distance shipments from a more distant Linde plant; (4) to secure product in case of liquid production plant shutdowns (either for normal maintenance or unexpected plant outages); (5) to secure product for the backup of an on-site customer when an on-site plant suffers a shutdown; and (6) to secure product to build a baseload of business in a market in which the product recipient has little or no production capacity for the purpose of making the future construction of a plant in. such a market economically viable.

Dobbins Supp.Aff. at ¶2. Accord, Hallet Aff. at ¶ 2 (Air Products); Wong Aff. at ¶ 2 (Liquid Air).

. Further, plaintiffs assert, defendants actually attempted — through their national price schedules — to price their products on a national basis. These schedules were internal price lists each defendant maintained to one extent or another during the relevant period of time. The prices listed on them for each form of product were uniform; they did not contemplate regional variation.

Defendants contend that the prices reflected in these documents stated mere goals or targets that were seldom reached; “schedule prices” bore little or no relation to the actual prices that were charged during the time in question. Plaintiffs do not deny that few, if any, purchasers actually bought at schedule price. Plaintiffs claim instead that the schedules served as “benchmarks” from which all pricing negotiations began. By emphasizing these national prices, defendants attempted, in plaintiffs’ view, to prevent the price erosion that might- otherwise have occurred in numerous localities.

Most importantly, it is plaintiffs’ theory of the case that defendants conspired to raise their price schedules in tandem, thereby raising the “benchmarks” and ultimately the entire market structure. As a subsidiary proposition, plaintiffs further allege that defendants illegally agreed to measures designed to police compliance with the underlying price agreement. Thus, defendants allegedly shared price information and agreed, as part of the conspiracy, to impose in their contracts the clauses requiring that they be notified of any lower bid received by a customer following an attempt to raise contractual prices.

Plaintiffs also assert that defendants conspired to allocate customers, mainly through an agreement not to offer competitive bids to “historical” accounts of rivals. Such quotes were instead to be made at or above the schedule price.⁹ The object of this alleged agreement was obviously to reduce the a,mount of competition each defendant faced in maintaining its client roster.

III. Plaintiffs’ Class Action. Request

Plaintiffs request declaratory, injunctive and damage relief on behalf of themselves and the putative class they seek to represent. This class consists of:

All persons, firms and corporations in the United States that purchased industrial gas from any defendant (including their respective subsidiaries or affiliates), at any time during the period of July 1, 1974 to and including June 30, 1980 but excluding (a) any defendants, or other industrial gas manufacturers and their respective subsidiaries and affiliates and excluding (b) all on-site and pipeline purchasers. '

IV. The Requirements of Rule 23(a)

Rule 23 of the Federal Rules of Civil Procedure governs class action certification in the federal courts. All class proponents must first establish compliance with the requirements of subsection (a) of the rule. Accordingly they must show that:

(1) the class is so numerous that joinder of all members is impracticable, (2) there are questions of law or fact common to the class, (3) the claims or defenses of the representative parties are typical of the claims or defenses of the class, and (4) the representative parties will fairly and adequately protect the interests of the class.

Defendants for the most part concede that these four tests have been satisfied. The court’s discussion will be brief.

At oral argument held on the initial motion to certify, plaintiffs’ counsel estimated that the proposed class contains at least 10,000 members. Defendants did not challenge this figure, and the court presumes it to be accurate. The requisite numerosity is clearly present.

The analysis found in section V. of this opinion confirms the undisputed proposition that at least some of the issues to be resolved in this case can be tried in a common fashion. The requirements of subsection (a)(2) do not bar prosecution of this suit as a class action.

Similarly, the requirements of subsection (a)(3) present no obstacle. As one court has noted, “Since the representative parties need prove a conspiracy, its effectuation, and damages therefrom — precisely what the absentees must prove to recover— the representative claims can hardly be considered atypical.” Minnesota v. United States Steel Corp., 44 F.R.D. 559, 567 (D.Minn.1968).

The requirement of “adequate representation” contains two components: “(a) the plaintiff’s attorney must be qualified, experienced, and generally able to conduct the proposed litigation, and (b) the plaintiff must not have interests antagonistic to those of the class.” Wetzel v. Liberty Mutual Insurance Co., 508 F.2d 239, 247 (3d Cir.), cert. denied, 421 U.S. 1011, 95 S.Ct. 2415, 44 L.Ed.2d 679 (1975). The previous discussion of “typicality” establishes that the latter half of this test is satisfied. Counsel for plaintiffs, moreover, are both able and experienced in the prosecution of antitrust class actions. Plaintiffs have successfully crossed the threshold of subsection (a).

Y. The Requirements of Rule 23(b)(3)

Class proponents must further demonstrate that their claims fit within a subpart of subsection (b) of Rule 23. Plaintiffs seek large damage awards on behalf of the class, and accordingly rely on subpart (b)(3). This section authorizes class actions when:

the court finds that questions of law or fact common to the members of the class predominate over any questions affecting only individual members, and that a class action is superior to other available methods for the fair and efficient adjudication of the controversy. The matters pertinent to the findings include: (A) the interest of members of the class in individually controlling the prosecution or defense of separate actions; (B) the extent and nature of any litigation concerning the controversy already commenced by or against members of the class; (C) the desirability or undesirability of concentrating the litigation of the claims in the particular forum; (D) the difficulties likely to be encountered in the management of a class action.

From the very commencement of this suit, the parties have sharply disputed whether common questions “predominate,” whether the proposed class action is “superior,” and whether it is “manageable.”

The Seventh Circuit, in Simer v. Rios, 661 F.2d 655, 665 (7th Cir.1981), cert. denied, 456 U.S. 917, 102 S.Ct. 1773, 72 L.Ed.2d 177 (1982), recently explained the proper method of analyzing “predomination” questions:

Our inquiry into the predomination analysis must take two steps. Our first focus must be on the substantive elements of plaintiffs’ cause of action and inquire into the proof necessary for the various elements. Second, after examining the proof necessary we must inquire into the form that trial on these issue would take. At this point it also becomes necessary to examine the procedural devices and alternatives available in trying class actions. This discussion interfaces with many aspects of the issue of manageability ....

Id. at 672. To carry out this analysis, the court must first determine the substantive issues likely to arise at trial. Well established doctrine holds that all plaintiffs suing under section 4 of the Clayton Act, see note 1, supra, must establish three facts in order to recover: (1) that defendants violated the antitrust laws; (2) that the alleged violations caused plaintiffs to suffer some injury to their “business or property”; and (3) that the extent of this injury can be quantified with requisite precision. See, e.g., Windham v. American Brands, Inc., 565 F.2d 59, 65 (4th Cir.1977) (en banc), cert. denied, 435 U.S. 968, 98 S.Ct. 1605, 56 L.Ed.2d 58 (1978). Plaintiffs’ extensive damage prayer injects several additional issues into the case as well. Specifically, since the initial filing occurred on July 3, 1980, all claims arising prior to July 3, 1976 are “forever barred,” 15 U.S.C. § 15, absent a showing that defendants “fraudulently concealed” their illegal conduct.¹⁰ The trier of fact must thus determine (with respect to the pre-1976 claims) whether defendants affirmatively attempted to conceal their wrongdoing, and whether plaintiffs either knew or reasonably should have known of their claims prior to the filing of the initial complaint. See, e.g., City of Detroit v. Grinnell Corporation, 495 F.2d 448, 460 (2d Cir.1974). Overall, five issues of fact — violation, fact of damage, extent of damage, attempt to conceal, and due diligence — are present.

The analysis next proceeds under Simer to an examination of the type of “proof necessary for the various elements.” In particular, the court must determine which elements may be tried in “common” fashion. An issue meets this standard when there exists generalized evidence that proves or disproves the element on a simultaneous, class-wide basis. Such proof obviates the need to examine each class member’s individual position, and permits the suit to go forward in a truly representative manner. As Judge Wyzanski has written, an issue is appropriately deemed “common” when there would not “be a substantial difference in the quantum or character of the requisite proof if the plaintiffs included all 20,000 in a class action or only the 6 named plaintiffs.” Windham v. American Brands, Inc., 539 F.2d 1016, 1019 (4th Cir. 1976) (Wyzanski, J., sitting by designation), reversed on different grounds, 565 F.2d 59 (4th Cir.1977) (en banc), cert. denied, 435 U.S. 968, 98 S.Ct. 1605, 56 L.Ed.2d 58 (1978).

When so phrased in terms of general principles, the scope of the necessary inquiry seems clear. Difficulties often arise, however, when the court turns to a case at hand. Suppose the class proponent proffers a description of a generalized approach it intends to follow in proving an element of its case. Can the opponent defeat the proffer on the.ground that it is so substantively flawed it should be disregarded? Plaintiffs have consistently argued throughout this litigation that no such scrutiny is permissible given the Supreme Court’s admonition that Rule 23 deprives the court of “any authority to conduct a preliminary hearing into the merits of a suit in order to determine whether it may be maintained as a class action.” Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 177, 94 S.Ct. 2140, 2152, 40 L.Ed.2d 732 (1974). In plaintiffs’ view, if defendants think plaintiffs’ approach is invalid, they should make their arguments at trial, not in their briefs opposing certification.

Plaintiffs read the delphic statements in Eisen too broadly. Rule 23(b)(3) specifically enjoins a court ruling upon a certification motion to make certain “findings.” To carry out this judicial mandate, a court must of necessity function like one; it must be able to engage, for example, in independent analysis. Yet if plaintiff’s position were accepted, a court would be obliged to accept all assertions of the class proponent at face value. The court’s function would not be to “find,” but to “ratify.” Rule 23(b)(3) contemplates a different scheme.

Stated somewhat differently, the class proponent bears the burden of demonstrating compliance with Rule 23(b)(3). See, e.g., Eggleston v. Chicago Journeymen Plumbers, 657 F.2d 890, 895 (7th Cir.1981), cert. denied, 455 U.S. 1017, 102 S.Ct. 1710, 72 L.Ed.2d 134 (1982). To the extent such a showing requires a demonstration that certain issues can be handled with generalized proof, the proponent bears the burden with respect to this sub-point as well. Acceptance of plaintiffs’ interpretation of Eisen would, however, virtually eliminate this burden. All proposed common methods would have to be credited so long as the class proponent set forth its belief that the method was probative of the ultimate issue in question; cognizance of any challenge to the latter qualifier would amount, under plaintiff’s theory, to an “impermissible scrutiny of the merits.” In this court’s view, however, the ultimate determination of whether a proponent has sufficiently discharged its burdens under Rule 23 must rest with the court, not the proponent. If a proposed common method of proof is so insubstantial that it realistically amounts to no method at all, the court must have the power to rule accordingly. Judge Pointer was clearly right when he stated that courts are “not impotent under Rule 23, bound to accept any common methodology ... which the plaintiffs may submit when seeking class certification.” Moore v. Southeast Toyota Distributors, Inc., No. CV 81-P-0491-S, slip op. at 8 (N.D.Ala. Feb. 4, 1982).¹¹

It cannot be denied that while engaging in the type of review just described, the court may touch upon issues that have relevance for the “merits” of plaintiffs’ suit as well. But, for the reasons already stated, such scrutiny is necessary if the court is to fulfill its responsibilities under Rule 23(b)(3). These complications were not at issue in Eisen, and the decision’s seemingly broad language must be interpreted in this light:

Nor is the class issue separable from the merits in all cases (including this one). The common questions, typicality, conflicts and adequacy of representation, Fed.R.Civ.P. 23(a), and predominance tests, Fed.R.Civ.P. 23(b)(3), are determinations (unlike, for example, the notice question involved in Eisen IV) which may require review of the same facts and the same law presented by review of the merits.

Blackie v. Barrack, 524 F.2d 891, 897 (9th Cir.1975), cert. denied, 429 U.S. 816, 97 S.Ct. 57, 50 L.Ed.2d 75 (1976). Indeed, in one post-Eisen decision, the Supreme Court has itself quoted similar remarks with approval:

Evaluation of many of the questions entering into the determination of class action questions is intimately involved with the merits of the claims. The typicality of the representative’s claims or defenses, the adequacy of the representative, and the presence of common questions of law or fact are obvious examples. The more complex determinations required in Rule 23(b)(3) class actions entail even greater entanglement with the merits.

Coopers & Lybrand v. Livesay, 437 U.S. 463, 469 n. 12, 98 S.Ct. 2454, 2458 n. 12, 57 L.Ed.2d 351 (1978) (quoting 15 C. Wright, A. Miller, & E. Cooper, Federal Practice and Procedure § 3911, p. 485 n. 45 (1976)). The Court of Appeals for this Circuit has similarly instructed that Eisen does not “foreclose inquiry into whether plaintiff is asserting a claim which, assuming its merit, will satisfy the requirements of Rule 23 ... . ” Eggleston v. Chicago Journeymen Plumbers, supra, 657 F.2d at 895. In short, while the court must assume that plaintiffs’ claim is meritorious in the sense that defendants’ liability ultimately could be established in some fashion, the court need not accept blindly every assertion that a particular common approach has validity. Nichols v. Mobile Board of Realtors, 675 F.2d 671, 678 (5th Cir.1982). Some inquiry into the latter type of claim is necessary for sound application of Rule 23(b)(3), and is not barred by the Supreme Court’s assertions in Eisen.¹²

[II] Yet even if a court finds, as this court ultimately will, that certain issues must be tried with individualized proof, the analysis does not end. The Simer Court squarely held that “[t]he use of separate trials on individual issues of fact should not be an absolute bar to the use of the class action device. Rather we must examine whether any procedural devices can be utilized that would alleviate the burdens of such a trial procedure.” Id., 661 F.2d at 674. A contrary ruling would “essentially preclud[e] class treatment whenever separate issues had to be tried. Such an approach has not been accepted since it would gut Rule 23(b) of its basic policy.” Id. at 672 n. 29.¹³ Thus, after examining which issues are common, the court must go on to examine exactly how the individualized issues will be tried. The court must assess each proposal that is offered to simplify the process of individualized adjudication. As before, the court need not accept all suggestions that are made, but if it finds at the conclusion of its analysis that various proposals are sufficiently viable, and that the individualized inquiries can thus be handled in a manageable fashion, class certification should be ordered.¹⁴

The remainder of this opinion constitutes an application of the Simer approach. In subpart V.A., the court examines each of the five issues implicated by the complaint in order to ascertain which are “common” as that term has been defined. After applying the form of analysis it believes is proper under Eisen, the court concludes that the issues of conspiracy, fact of damage and concealment are common, but that the remaining issues of damages and due diligence are not.

In subpart V.B., the court next examines the precise manner in which damages (the more significant of the two individualized issues) must be tried. The court starts from the premise that the proposed class action would be unmanageable if calculation of each purchaser’s injury necessitated an individualized, full-scale “minitrial” held before this court. The court thus examines the propriety of various procedural methods of alleviating the strain, testing each proposal, as before, for both minimal substance and legality. Finding no method acceptable, the court denies class certification.

A. Which Issues are Common?

1. Violation

Both sides agree that the violation issue is common. Resolution of this question depends solely upon an analysis of defendants’ challenged behavior. Since this conduct did not vary from class member to class member, proof relevant to any one purchaser must of necessity apply equally to the case of each and every remaining member of the class as well.

2. Fact of Damage

The court, in its initial class action opinion, ruled that fact of damage was an issue susceptible to common proof. The court noted that plaintiffs had at their disposal common expert testimony supporting their assertion that each class member felt the effects of defendants’ allegedly anti-competitive behavior. Particularly relevant were the following statements by Dr. Sam Peltzman plaintiffs’ expert:

7. The materials which I have reviewed indicate that the five defendants account for approximately 80% of the industrial gas market; that defendants consider the merchant market a national market and accordingly market, sell and price their fungible products on a national basis; that defendants, during the conspiracy, utilized national price schedules which served as benchmarks for their pricing in the merchant market; that defendants, during the term of the conspiracy, implemented general nationwide price increases in tandem fashion; and that these general price increases were implemented as to the class as a whole on an ongoing continuing basis.

8. In view of the characteristics of the industrial gas merchant market and defendants’ dominant position in that market, I conclude that defendants’ nationwide conspiratorial conduct, would over the term of the conspiracy, result in a general increase in the prices charged all class members throughout the United States.

Peltzman 1st Aff. at ¶¶ 7-8. In the court’s view, plaintiff’s ability to proffer such testimony adequately demonstrated the existence of a common method of going forward on the issue of fact of damage. The court specifically ruled that the general topic was suited for expert testimony, that plaintiff’s expert was amply qualified to render an opinion, and that Dr. Peltzman’s statements could not be rejected at this stage of the lawsuit.

In their motion to reconsider, defendants challenge only the latter finding. Defendants contend that Dr. Peltzman’s assertions completely lack merit, and that plaintiffs therefore do not have a viable plan for proving impact in a common manner. For the reasons previously stated, such an attack cannot be rejected out of hand. The court will consider its particulars.

Experts can be challenged on various grounds. Opponents may, for example, concede the factual premises of an expert’s testimony, but dispute the inferences he draws. Conceivably, such an attack will justify rejection of the expert’s opinion, even at the certification stage of the lawsuit. An expert’s bald face assertion that ultimate fact “X” follows from given fact “Y” simply does not prove or disprove the existence of “X” when no discernible reason justifies the suggested relationship. See Fed.R.Evid. 401. An irrational belief remains irrational even when voiced by an expert.

But this form of attack has not been mounted by defendants against Dr. Peltzman. Defendants appear to concede (most clearly, in their reply brief in support of their motion to reconsider) that Dr. Peltzman’s ultimate conclusion as to class-wide impact is tenable if the “assumptions” listed in paragraph seven of his affidavit are supportable.¹⁵ However, defendants continue, at least two of these beliefs cannot be credited in light of the “wealth of evidence” offered by defendants in opposition. Defendants specifically challenge Dr. Peltzman’s “assumptions” that defendants’ price schedules “served as benchmarks for their pricing in the merchant market,” and that schedule increases “were implemented as to the class as a whole on an ongoing continuing basis.”

Defendants mischaraeterize the record. The quoted statements are not mere “assumptions” of Dr. Peltzman, but are summaries of factual claims the plaintiffs intend to prove in common fashion — i.e., through the testimony of former employees of the defendants as to how defendants utilized their price schedules, both generally and following increases in their rates. To reject Dr. Peltzman’s testimony on the ground asserted by defendants, the court would thus have to make factual determinations as to what exactly transpired during the period in question. The court would have to determine the “true” role played by the price schedules in the industrial gas industry.

As a theoretical matter, even factual inquiries may be permissible during the course of a certification ruling. In an oft-cited opinion, Judge Godbold, writing for an en banc court, postulated such an instance:

“On the merits” is an elusive term, not wholly suitable as a guideline in this situation. In a sense much that a plaintiff alleges with respect to his individual claim goes to the “merits” of his claim. A plaintiff who seeks to represent all employees on a claim of discriminatory denial of promotions might, if the case goes to trial, lose on his individual claim because it turns out that he was never employed by the defendant. Retrospectively, one might say he lost “on the merits.” Yet there is no doubt that the court could lay bare at a preliminary stage this fact concerning the plaintiff, and, having done so, conclude plaintiff was not a proper representative because he lacked the nexus with the class and its interests and claims which is embraced in the various requirements of 23(a) and (b). It is inescapable that in some cases there will be overlap between the demands of 23(a) and (b) and the question whether plaintiff can succeed on the merits.

Huff v. N.D. Cass Company of Alabama, 485 F.2d 710, 714 (5th Cir.1973) (en banc) (emphasis added); see also Nichols v. Mobile Board of Realtors, 675 F.2d 671, 678 (5th Cir.1982). The type of factual determinations the defendants ask this court to make are, however, far more difficult and complicated than those suggested in Huff. The court’s error quotient is accordingly high, particularly since it is being asked to rule without the benefit of traditional trial safeguards, and the court must move with trepidation in resolving these disputes in a way that effectively precludes further litigation, at least insofar as many of the absent class members are concerned. “[T]here is a danger,” Judge Marshall has written, “in delving too far into the merits of a case prior to a class action determination.” McCray v. Standard Oil Co., 76 F.R.D. 490, 500 (N.D.Ill.1977). These dangers are particularly acute here.

Simply put, the evidence before the court pertaining to Dr. Peltzman’s “assumptions” is not nearly so one-sided as defendants apparently believe. The various “impact studies” submitted by the defendants¹⁶ can be impeached on statistical grounds,¹⁷ and do not in any event establish that the price schedules were completely divorced from actual pricing decisions. Even if it is true that many purchasers did not experience price increases during the period the schedules were rising, such evidence does not exclude the possibility that these same individuals would have paid even less in the absence of the alleged conspiratorial behavior. The jury, of course, may ultimately agree with defendants that the factual bases for Dr. Peltzman’s conclusion of class-wide impact do not exist. But this possibility is not fatal. Plaintiffs have suggested how they intend to present their evidence in a common manner, and the issue for the court is merely whether these showings are substantial enough to justify further proceedings in this direction. In this court’s view, they are.

Furthermore, defendants attack a straw man to the extent they claim that the price schedules lacked market significance since few purchasers actually paid the pertinent schedule price. As set out before, plaintiffs have never argued that transaction prices precisely tracked the schedules; their claim is rather that these schedules influenced the actual prices paid in more subtle ways; the schedules are alleged to have been “benchmarks,” nothing more.¹⁸

It thus appears that plaintiffs have a sufficiently viable, common method of going forward on the question of impact. The only issue that remains on this phase of the case is whether the class format will similarly accommodate defendants’ right to present their side of the story. See, e.g., In re Transit Company Tire Antitrust Litigation, 67 F.R.D. 59, 72 (W.D.Mo.1975) (In ruling upon a class certification request, a court must consider both the affirmative evidence likely to be put forth by the plaintiffs and the defensive evidence likely to be offered by the defendants.). Not surprisingly, defendants answer in the negative, arguing that since each merchant price resulted from numerous factors of varying strength, see supra, conclusive proof that these prices were not influenced by the alleged conspiracy will necessarily require an individualized examination of each transaction. But defendants do not bear the burden of proof on the question of impact, and they accordingly need not “prove” anything in an affirmative sense. Defendants need only refute plaintiffs’ presentation, and the relevant inquiry therefore turns on whether common proof will reasonably suffice for that purpose. The court believes it will.

• Plaintiffs, as shown above, will attempt to establish class-wide impact by presenting Dr. Peltzman’s testimony, and by proving the predicate facts underlying it, e.g., that defendants colluded to raise or stabilize the prices reflected on the price schedules, and that these schedule prices were in fact “benchmarks” influencing the actual prices charged in the merchant market. Defendants can rebut this case by presenting evidence either that no collusion occurred or that the price schedules were routinely disregarded when actual pricing decisions were made. Defendants also can impeach Dr. Peltzman by presenting examples of actual purchasers who, according to defendants’ experts, paid no more over the life of the alleged conspiracy than they would have, éven if defendants had refrained from the actions now challenged. All of these presentations fall under the “common” heading because they refute the evidence of impact relied upon by each class member, and thus tend to undermine each purchaser’s cause of action simultaneously.

It follows that defendants’ assertion that they will need to examine the situation surrounding each sale in the merchant market is flawed. Defendants need only point to examples of instances in which impact was not present, for the very existence of each example undermines a case like plaintiffs’ which depends upon the testimony of an expert who denies the possibility that any such example might occur. Of course, as the number of examples expands, so does the strain on the court’s ability to try the lawsuit. But without now deciding the outer limits of what is permissible, the court is confident that defendants’ right to present an effective defense can be preserved within the context of a manageable proceeding.

3. Damages

The question of damages is individualized since each class member must step forward and make separate showings at least as to the amount of product purchased. Whether these are the only individualized showings that are needed is addressed at length infra.

4. Attempt to Conceal

All sides agree this issue is common for the same reasons that govern the question of violation.

5. Due Diligence

In the court’s initial opinion, it found due diligence to be an individualized issue since resolution required specific knowledge of what each class member knew — or should have known — and when. See Simer v. Rios, supra, 661 F.2d at 673 (question of the impact of an invalid regulation “extremely individualized” since it “depends on each individual’s state of mind.”) Plaintiffs did not challenge this assertion in their written cross-motion to reconsider, but did express disagreement with the court when questioned at oral argument. Plaintiffs’ position is that once an attempt to conceal is shown, “due diligence does not become something which has to be proved affirmativelyQ] it has to be disproved.” (Transcript of December 3, 1982 hearing at 54) But this contention, even if correct as a matter of substantive law, is irrelevant for present purposes. Whether due diligence is a matter to be proved by plaintiffs or disproved by defendants, it remains an issue that potentially must be litigated — and litigated individually — whenever a class member seeks damages on a claim that arises before the statutory period.¹⁹

6. Conclusion

Three of the five issues to be resolved in this case fall within the “common” eategory. Only damages and due diligence require individualized proof, and the inquiry thus shifts, in accordance with Simer, to a determination of whether even they may be handled in a manageable fashion.

B. Can the Burden of Trying the Individual Questions be Alleviated?

The court’s primary concern is with the issue of damages.²⁰ Underlying this concern is a belief that this lawsuit cannot be deemed manageable if this court must hear an unending series of plenary “mini-trials” devoted to the question of damages. Common questions simply do not “predominate” when each purchaser’s specific injury must be calculated in a separate, time-consuming hearing — each conducted before this court, each specific to a given class member, and each replete with discovery, experts, cross-examination and documents. See, e.g., State of Alabama v. Blue Bird Body Co., 71 F.R.D. 606, 616 (M.D.Ala.1976), approved on this point, 573 F.2d 309, 328 (5th Cir.1978); Ralston v. Volkswagenwerk, 61 F.R.D. 427, 433 (W.D.Mo.1973).

In their briefs supporting the original motion to certify, plaintiffs disputed this view, arguing that even if a separate “mini-trial” was needed to determine the extent of each class member’s injury, the class could still be certified since the damage question was simply “irrelevant.” This claim finds no support in Rule 23: “[A] trial judge cannot, in determining the manageability of a proposed class action, look exclusively to only one aspect of the case ...; he can and must look at the case as a whole and, as we have seen, consider proof of damages as well as other issues in the ease.” Windham v. American Brands, Inc., supra, 565 F.2d at 71. Indeed, when read closely, most of the decisions cited by plaintiffs do not take a different position. They merely assert the nearly tautological proposition that “the necessity for calculation of damages on an individual basis should not preclude class determination when the common issues which determine liability predominate." Bogosian v. Gulf Oil Corp., 561 F.2d 434, 456 (3d Cir.1977), cert. denied, 434 U.S. 1086, 98 S.Ct. 1280, 55 L.Ed.2d 791 (1978) (emphasis added); accord, e.g., In re South Central Bakery Products, 86 F.R.D. 407, 419 (M.D.La.1980); In re Sugar Industry Antitrust Litigation, [1977-1] Trade Cases (CCH) ¶ 61,373 (N.D.Ca.1976), at 71,335. Damage questions can thus be deemed “irrelevant” only if it is first determined that the remaining issues “predominate.” Under Simer, this predicate determination itself requires a finding that damages can be handled in a manageable fashion.

In the alternative, plaintiffs offer several proposals designed to simplify the damage phase of the case. Each suggestion is flawed, however, and the court is compelled to conclude that, on the basis of the present record, damage adjudication appears likely to assume the unmanageable character that is feared. Therefore, class certification must be denied.

Plaintiffs’ main proposal involves use of a formula. Plaintiffs postulate that each class member’s injury can be calculated through a simple process of multiplying the amount of the individual’s purchases by the uniform overcharge figure applicable to the time and product under scrutiny. Plaintiffs further explain that the uniform overcharge figure is the difference between the average FOB market price (for the relevant time and product) and the governing “swap price.” The “swap price” refers to the price defendants charged one another when settling imbalances in their “swap accounts”— accounts that recorded the net amount of product supplied to competitors through swap transactions of the type described before.²¹

Clearly, if this methodology is sufficiently viable, manageability concerns would greatly subside, for individualized proof would be needed only to establish the amount of each class member’s purchases; the applicable overcharge figure would be set without reference to considerations that vary from purchaser to purchaser. Recognizing these facts, defendants challenge the economic logic underlying the swap price theory. The issue for the court is whether these attacks undermine plaintiffs’ theory to such an extent that the formula may now be rejected.

To have any significance, this formula must be programmed to determine the difference between the prices actually paid and the prices that would have been paid in the absence of defendants’ allegedly illegal behavior; such is the conceptual measure of damages applicable in this case. See P. Areeda & D. Turner, Antitrust Law ¶ 344 (1978). Defendants argue from this definition that since plaintiffs intend to subtract swap prices from actual merchant prices (i.e., prices that were allegedly the product of a conspiracy), the swap price theory “works” only if swap prices approximate the prices that would have been paid by the class members in the absence of the alleged conspiracy. However, defendants continue, no such equation can be drawn, for while swap prices were uniform throughout the country (for a given time and product), pricing in a hypothetically non-collusive merchant market would most certainly have been otherwise. Defendants base the latter assertion on their belief that the pricing actually observed in the purportedly price-fixed merchant market was “chaotic” itself; it makes no sense, defendants reason, to assume that greater uniformity would have prevailed under uncontrolled, competitive conditions.

Plaintiffs do not deny that the swap prices were nationally uniform. Nor do they dispute the factual assertion that non-collusive pricing would have been variable. (Indeed, one of plaintiffs’ substantive contentions appears to be that defendants artificially reduced such diversity by utilizing their national price schedules. See supra.) Defendants’ attack nevertheless fails, for it is based on a misunderstanding of the manner in which the swap price formula operates. Defendants’ argument would have force if plaintiffs were claiming that each class member was overcharged by the difference between that individual's actual price and the applicable swap price. Under this approach, the resulting difference could reasonably be said to approximate the purchaser’s injury only if the swap price itself approximated the price that purchaser would have paid but-for the conspiracy. As this would be true equally in the case of each class member, it would then follow, given the uniform nature of the swap price, that plaintiffs’ theory implicitly assumed a non-collusive world in which price variation was absent.

But plaintiffs proffer an entirely different approach. They propose instead to subtract the swap price from the average price paid for the relevant product during the year of the conspiracy in question. In this way plaintiffs seek to determine a single unit overcharge figure to be applied against each purchaser who bought during that time period. See note 21, supra. Furthermore, plaintiffs argue, the use of a uniform overcharge figure is appropriate regardless of whether non-collusive pricing was variable: Even if it was, collusive pricing was presumably disparate to the exact same degree; each price in the latter market simply was higher than its corresponding measure in the non-collusive market by the uniform amount. Dr. Peltzman explains:

Contrary to Professor Markham’s assertion, my proposed use of swap prices as a non-conspiratorial benchmark does not assume that there must be a single “competitive price” for the entire country and for all customers (¶ 18). Nor does geographic diversity of prices make it impossible to have a single benchmark (¶ 23). In a conspiracy which is national in scope, customers in all areas will pay prices which are higher than they would have been without the conspiracy. Those high conspiratorial prices may differ among customers, just as it might be the case that the lower non-conspiratorial prices differed among customers. My methodology is designed to provide an estimate of the excess of the conspiratorial price. For this purpose, I proposed a comparison of a competitive benchmark to an average conspiratorial price. Nothing in this methodology requires any assumption that all customers pay the same conspiratorial price or that all would have paid the same price absent conspiracy. And, unless it can be convincingly shown that the effects of the conspiracy differed among customers, an attempt at individual calculation of damages is not only not required [but] would be invalid methodologically.

Peltzman 4th Aff. at ¶ 16 (emphases in original). In short, plaintiffs’ theory rests not on an assumption of uniform non-collusive pricing, but on a presumption that defendants’ illegal activities presumptively influenced each sale transaction to the exact same extent. According to Dr. Peltzman, the trier-of-fact should find, in the absence of contrary evidence put forth by defendants, that each purchaser (of a given product during a given time) paid a uniform surcharge.

This presumption is an absolutely crucial lynch pin in the analysis justifying the swap price formula as a lawful determinant of damages. Without it, there is little, if any, reason to think that plaintiffs’ formula adequately measures the extent of each class member’s individual injury, as the antitrust laws demand.²² Clearly, if members of the class paid varying amounts of overcharge, a theory that assumes the contrary is hardly programmed to deal with the complexities of each plaintiff’s situation.²³ Such a theory would appear instead to be capable of nothing more than mere arbitrary assignments of injury.

It follows that plaintiffs’ theory cannot be accepted at this time, notwithstanding the court’s rejection of defendants’ prior challenge. See pp. 296-297, supra. Plaintiffs have offered neither argument nor citation in support of their claim that they are entitled to invoke the crucial presumption in question. Plaintiffs have simply made the bald face assertion that a presumption of uniform impact embodies “the best approximation of the real world that can be made today after defendants conspired.” (Transcript of September 2, 1981 hearing at 57) The court cannot accept this proposition — given the nature of this case — without further elaboration. Plaintiffs are not alleging that defendants’ conspiratorial behavior simply generated a higher plateau from which defendants entered into the very same pricing activities they would have commenced even in the absence of the conspiracy; plaintiffs are not claiming, in other words, that “everything that went on would have gone on anyway, except at a lower level.” Plaintiffs’ theory of the case is much broader. It encompasses the notion that certain forms of behavior occurred {e.g., forbearance from competitive solicitations of rivals’ “historical” customers) that simply would not have been observed under non-collusive circumstances. Under plaintiffs’ own theory of the case, defendants’ allegedly illegal behavior thus added to the pricing universe several factors, all of which might not have affected each transaction. Under these circumstances, and without meaningful argument from plaintiffs, the court cannot now assume that uniform impact was present.²⁴ The fact of such impact could be proven at trial, but plaintiffs have offered absolutely no indication of how, or, for that matter, whether, they will lay a foundation for Dr. Peltzman’s conclusory assertions; plaintiffs are silent as to their method, if any, of going forward on this crucial point.²⁵

The nature of the court’s holding should not be misunderstood. The court is not making a final, binding determination that the alleged conspiracy must have influenced the merchant market, if at all, in a non-uniform manner. Nor is the court even foreclosing the possibility that plaintiffs might be entitled to a contrary presumption. The court is simply holding that, on the present record, plaintiffs have failed to offer even a minimal explanation of why the court should accept the assumption anchoring their theory.²⁶

Moreover, even if the concept of uniform impact is accepted, the specific figure generated by the swap price method must additionally bear some reasonable relation to the “true” amount of uniform damage. Under Dr. Peltzman’s assumptions, this relation holds only if the applicable swap price is equal to or greater than the average price that would have been paid in the absence of the conspiracy for the product at issue. See Appendix. (This assertion is obviously a more refined version of defendants’ earlier-rejected claim. See pp. 296297, supra.) Since plaintiffs have made sufficient showings that non-collusive prices would have tended in this industry towards marginal cost,²⁷ the necessary relation may be restated as follows: Swap prices must have been equal to or greater than the average marginal cost of producing the product in question during the relevant period of time.

Defendants vigorously argue that no such relation can be proven since swap prices were mere accounting devices that bore no market significance. See In re Transit Company Tire Antitrust Litigation, 67 F.R.D. 59, 77-78 (W.D.Mo.1975).²⁸ Defendants claim that they intended all swaps to be pure barter, that swap accounts were carefully monitored to prevent buildups of large imbalances either way. Most importantly, a swap “purchaser” had no assurance that it could compensate the “seller” with money rather than return product; it was the seller’s option either to demand a money settlement determined by the swap price (usually at the end of the year) or to carry its receivable into the future in the hope that additional transactions would eliminate the imbalance. Swap prices thus could not have served as market signals guiding the purchaser’s decision to enter into the swap arrangement. Cf. Airweld v. Airco, 576 F.Supp. 676 at 679 (D.Or.1983) (swap transactions in the industrial gas industry are not “sales” for purposes of the Robinson-Patman Act).

Whatever the merits of this argument, it cannot be held, on the basis of this record, that swap prices had no relevance for the seller:

The plain and important fact is that swaps are not pure barter. Even if defendants wish it would work out that way, there is no obligation that a recipient of x cubic feet of oxygen supply x cubic feet to the sender. Once this elemental fact is grasped, the consequences of truly arbitrary swap prices are easily grasped.... [If] the swap price is set arbitrarily low, there would of course be many requests for swapped gas, but few would be accommodated, since potential senders would understand the recipient’s incentives to settle in cash. Again, there would be a supply-demand imbalance. This time much would be demanded but little offered. As in any market, there will be some price intermediate between those which are so high as to provoke excess supply and those so low as to provoke excess demand at which the swap market will clear. Such a price will convince potential suppliers that when they honor a swap request, their request will be honored in the future or they will be adequately compensated in money.

Peltzman 4th Aff. at ¶ 13 (emphases added). Plaintiffs might thus be able to construct a viable explanation as to why swap prices could not possibly have been set at a level under the average marginal cost. The simple truth, however, is that no such explanation has been tendered. During the original round of briefing and argument, plaintiffs did contend that the swap price must have exceeded the marginal cost. See Transcript of September 2, 1981 hearing at 64. But no consideration was given then, or has been given since, to average measures.

There are additional problems with the swap price methodology. First, plaintiffs intend to estimate average FOB merchant prices from publicly available Census figures. However, these figures do not list average FOB prices since they are based on price submissions that often reflect built-in transportation charges. (Transportation assessments do not come into play in swap deals; the purchaser simply fills its truck up at the seller’s plant.)²⁹ Dr. Peltzman has assured the court that

[t]he methodology for [making the necessary adjustment] is surely simple to comprehend: estimate the portion of revenues per 100 cubic feet as reported to the Census which is non-product charge and either deduct this from the gross revenues per 100 cubic feet or add it to the non-collusive benchmark price. Given this, the question becomes whether it is practical to do so with some reasonable degree of accuracy. While the precise way in which this cannot be specified at this stage of the case, without further data, it is clear that nothing in my methodology requires an individualized invoice-by-invoice calculation of damages.

Peltzman 4th Aff. at ¶ 15 (emphasis added).³⁰ This response is insufficient. For the court to accept Dr. Peltzman’s claims, it needs to know at least the rough outlines of how he would approach the problem at hand. No such guidance has been given, however, and the court is left merely with his unexplained assurance that he can proceed in a common fashion. Courts should not certify class actions on this basis. In re Hotel Telephone Charges, 500 F.2d 86, 90 (9th Cir.1974).

Moreover, the rationality of applying the swap formula to cylinder sales is seriously in doubt. The materials before the court³¹ indicate that cylinder products were not swapped, and that the various swap prices hence did not pertain to them. The relevant Census figures also do not segregate cylinder sales of argon from liquid sales. Because these two forms of product sold at differing price levels, such joint reporting diminishes the usefulness of the Census data as a determinant of the average FOB price of either form. In short, all of the data needed to operate the swap price theory in a proper manner cannot be obtained easily. Plaintiffs’ formula approach does not ease the court’s concern that the proposed class action is unmanageable. See Klein v. Henry S. Miller Residential Services, 94 F.R.D. 651, 660-61 (N.D.Tx. 1982); Chmieleski v. City Products Corp., 71 F.R.D. 118, 162 (W.D.Mo.1976); In re Transit Company Tire Antitrust Litigation, supra, 67 F.R.D. at 77.³²

Nor is the court aware of any other viable method of handling the damages question. The solution adopted by the court in its initial opinion was to attack the very premise that a problem existed. The court reasoned that the danger of burdensome “mini-trials” was merely speculative, because the case might be settled or otherwise resolved long before damages had to be reached, and because the very complexity of the damage question — the complexity which rendered the “mini-trials” unmanageable in the first place — would, in any event, likely reduce the number of plaintiffs willing to prove up damages. See generally note 13, supra. The court now recognizes that Simer implicitly forecloses such an approach. Plaintiffs in Simer sued the federal Community Services Administration for improperly denying cash assistance payments to members of the proposed class of low-income individuals. Without ever ruling on the propriety of the class certification request, the district court granted summary judgment for the plaintiffs on the purely legal issue of whether the agency’s authorizing statute supported its position. The parties then devised a settlement plan under which the funds not spent on the assistance program were to be funneled into various projects purportedly of interest to the class. Subsequently, the district judge vacated the settlement and dismissed the suit, finding that the settlement had been improperly obtained, that the named plaintiffs had been fully compensated, and that a class could not be certified. In affirming the latter ruling, the Seventh Circuit stressed that resolution of the class members’ individual damage claims would necessitate an unmanageable series of “mini-trials.” The Court “gave no consideration to the possibility of settlement — which was very high, inasmuch as one settlement had already been reached — nor did it consider the likelihood that individual class members would pursue their individual claims— which was very low, in that the class members were poor people with claims of no more than $250.” (Reply Memorandum in Support of Defendants’ Motion to Certify a § 1292(b) Appeal from the Class Certification Order at 3). The considerations which originally motivated this court were present in Simer, but ignored.

Sub-classing pursuant to subsection (c)(4)(B) of Rule 23 is, by contrast, a recognized solution to problems of manageability in certain cases. See Simer v. Rios, supra, 661 F.2d at 674. Plaintiffs, however, have not suggested this solution, and the court is at a loss as to how sub-classing would be of assistance in this particular lawsuit.

What plaintiffs do suggest is that they be permitted to obtain a “fluid recovery” on behalf of the class. The Court in Simer explained this term:

The fluid recovery is used where the individuals injured are not likely to come forward and prove their claims or cannot be given notice of the case. In a fluid recovery the money is either distributed through a market system in the way of reduced charges or is used to fund a project which will likely benefit the members of the class.

Id., 661 F.2d at 675 (citation and footnotes omitted). When a fluid recovery is sought, no attempt is made to correlate class members with actual injuries. Damages are distributed in an imprecise manner that eliminates the need for individualized determinations.

The problem with this “solution” — aside from plaintiffs’ failure to specify exactly what type of fluid recovery they envision— is that it is simply impermissible as a matter of law. Fluid recoveries have been condemned as illegal in numerous decisions. See, e.g., Windham v. American Brands, Inc., supra, 565 F.2d at 72; In re Hotel Telephone Charges, supra, 500 F.2d at 90; Eisen v. Carlisle & Jacquelin, 479 F.2d 1005, 1018 (2d Cir.1973), vacated on other grounds, 417 U.S. 156, 94 S.Ct. 2140, 40 L.Ed.2d 732 (1974); Al Barnett & Son, Inc. v. Outboard Marine Corporation, 64 F.R.D. 43, 55 (D.Del.1974). But see In re Sugar Industry Antitrust Litigation, 73 F.R.D. 322, 353-54 (E.D.Pa.1976). To be sure, Simer rejects the breadth of these precedents and authorizes á fluid approach when “the use of such a mechanism is consistent with the policy or policies reflected by the statute violated.” Id. at 676. The Court explicitly recognized, however, that this form of reasoning might lead to an absolute ban on fluid recoveries in antitrust cases. Id. at 676 n. 42.

Such a per se approach is entirely correct. The Clayton Act limits recovery to those actually injured by defendants’ conduct,³³ and further limits the extent of the recovery to the amount the claimant has approximately suffered. See note 22, supra, and accompanying text. Fluid recoveries disregard these limitations, and hence are not “consistent with the policy or policies reflected by the statute violated.”

The final question is whether the case could be made manageable by bifurcating the trial. See Fed.R.Civ.P. 42(b). A positive response might be possible if the court coupled bifurcation with an order under subsection (c)(4)(A) of Rule 23³⁴ certifying the case as a class action solely with respect to the common issues raised by the complaint; the court would then have before it all common questions (those arising out of the actions of the named plaintiffs and the absentees), as well as the individual questions presented solely by the complaints of the named plaintiffs. Trial would be bifurcated between the common and individual issues. If the class prevailed in the initial, common phase of the proceeding, the named plaintiffs would then continue to litigate their individual questions in this court; the remaining class members could also proceed, but since no class action would then exist, they would not be obligated to pursue their individual claims here. Rather, they would be entitled to sue in any district court in which jurisdiction and venue could be obtained.

This plan would ease manageability concerns by decentralizing the second phase of the litigation. The individual questions would be tried before many courts, and the burden experienced by any one tribunal would correspondingly diminish.

Defendants argue that Simer rejects this “solution” to manageability problems. Relying on one sentence in one footnote, defendants assert that under Simer, “a bifurcated trial would not solve the problem of individual issues ... since it would merely delay the issue to a later date and at that point the complications would arise.” Id., 661 F.2d at 673-74 n. 33. However, it is not clear from the context of the quoted statement whether it reflects the Simer Court’s own view, or merely its summary of a prior decision. It is also not clear whether the Court had in mind bifurcation involving several courts, or simply bifurcation in the context of a lengthly trial conducted in phases before one court. Defendants must point to more than an ambiguous sentence in a footnote before this court will conclude that the Seventh Circuit has rejected the procedural set-up proposed here.³⁵

But while Simer does not necessarily bar the bifurcation plan described above, constitutional considerations might. By the plain, though unnecessary, terms of Rule 42(b), a bifurcation order must “preserv[e] inviolate the right of trial by jury as declared by the Seventh Amendment to the Constitution or as given by a statute of the United States.” The concern created by the proposed scheme is that it contemplates separate juries trying different aspects of single causes of action, those of the unnamed class members. In Gasoline Products Co. v. Champlin Co., 283 U.S. 494, 51 S.Ct. 513, 75 L.Ed. 1188 (1931), the Supreme Court dealt with an analogous situation. The Court of Appeals had found error in the district court’s jury instructions on the damage aspects of a breach of contract counterclaim, and had remanded for a retrial limited to a recalculation of the damages due. The Supreme Court reversed, holding that the proceedings on remand had to encompass not only the question of damages, but also the underlying issue of whether the counter-defendant was liable at all. The Court stated that “[wjhere the practice permits a partial new trial, it may not properly be resorted to unless it clearly appears that the issue to be retried is so distinct and separable from the others that a trial of it alone may be had without injustice.” Id. at 500, 51 S.Ct. at 515. No sharp line could be drawn in the case under review because a proper damage assessment necessitated a prior understanding of what the pertinent contract specifically provided. But the initial jury’s verdict definitively revealed only a finding that a contract had existed and had been breached; it offered no guidance as to what that jury had specifically understood the contract to require. Thus, it was impossible for the district court to instruct the jury on remand as to the details originally determined; the members of the second panel could only guess as to what determinations had been originally made and were hence binding. The damage question was “so interwoven with that of liability that the former [could not] be submitted to the jury independently of the latter without confusion and uncertainty, which would amount to a denial of a fair trial.” Id.³⁶

The principles announced in Gasoline Products apply outside the context of retrials; they similarly govern instances, like here, where bifurcation is proposed as an initial matter. 9 C. Wright & A. Miller, Federal Practice & Procedure, § 2391, p. 303 (1971). The question for the court is thus whether the issues to be separated are distinct enough that they may be considered fairly in isolation. In particular, the court must consider whether damages can be properly assessed apart from inquiries into violation or fact of damage.

The latter separation — between fact of damage and damages — is the most troubling. Though there is precedent upholding this form of bifurcation against Seventh Amendment challenge, see In re Ampicillin Antitrust Litigation, 88 F.R.D. 174, 179 (D.D.C.1980); In re Master Key Antitrust Litigation, supra, 70 F.R.D. at 29, the weight of authority suggests caution:

In an antitrust action brought under the Clayton Act, it is most important to be aware of the Seventh Amendment limitations on a Rule 42(b) bifurcation. This awareness is important because liability under § 4 necessarily includes proof of injury to business and property. Therefore, bifurcation to separate juries of liability and damages in a § 4 case inevitably introduces the possibility that in the liability phase the first jury might find that there was such injury, while the second jury might on the same evidence of injury in the damage phase, find none.

State of Alabama v. Blue Bird Body Co., supra, 573 F.2d at 318; accord, Windham v. American Brands, Inc., supra, 565 F.2d at 71; Link v. Mercedes-Benz of N. Am., Inc., 550 F.2d 860, 875 (3d Cir.) (Gibbons, J., dissenting)³⁷ cert. denied, 431 U.S. 933, 97 S.Ct. 2641, 53 L.Ed.2d 250 (1977); A1 Barnett & Son, Inc. v. Outboard Marine Corporation, supra, 64 F.R.D. at 56-57; Silberman, The Defense View — Litigating With Just a “Touch of Glass,’’ 49 Antitrust L.J. 1529, 1543-44 (1980); Simon, Class Actions — Useful Tool or Engine of Destruction, 55 F.R.D. 375, 386 (1973); cf. In re Transit Company Tire Antitrust Litigation, supra, 67 F.R.D. at 75.

Having contended from the start of this litigation that damages are susceptible to common proof, plaintiffs have for the most part ignored the possibility of individualized damage “mini-trials.” Little attention has been focused in this direction, and the court consequently lacks much, if any, guidance as to the issues and questions that would arise in such proceedings. Accordingly, it is impossible to determine with much certainty the degree to which bifurcated damage and fact of damage inquiries would overlap. But certainly, as the Blue Bird Court noted, the potential is very great; both inquiries concern themselves with the presence of an alleged overcharge; one inquiry is simply a more particularized variant of the other. See also United Air Lines, Inc. v. Wiener, 286 F.2d 302 (9th Cir.), cert. denied, 366 U.S. 924, 81 S.Ct. 1352, 6 L.Ed.2d 384 (1961). Thus, were bifurcation ordered, the court might well learn at the end of the common phase of the trial that the issues to be raised in the individualized proceedings were so “interwoven” with those already tried that the second stage could not go forward as planned. Instead, the damage “mini-trials” would have to expand to encompass a re-exploration of the overlapping issues necessary for a constitutionally fair assessment of damages. A major systemic goal underlying the concept of class actions — the realization of final, classwide resolutions of disputed factual questions— would thus be frustrated.

The danger that the foregoing scenario might materialize is simply too great for this court to ignore. Moreover, to the extent the question is merely uncertain in large degree, plaintiffs once again fail. It is they who bear the burden of persuading the court that the proposed bifurcation plan is viable.³⁸

Nor can these problems be avoided by limiting the common phase of the trial to the question of violation, the issues of fact of damage and damages both being reserved for individualized adjudication. As Professor Areeda has noted, “an antitrust damage assessment cannot be divorced from thoughtful attention to the rationale for liability and the internal logic of the liability holding.” Areeda, Antitrust Violations Without Damage Recoveries, 89 Harv. L.Rev. 1127, 1139 (1976). Judge Wyzanski, though ultimately finding this point nondispositive, evidently agrees: “[P]roof as to what such violations, if they occurred, caused in the way of damage may become relevant to, and even to some degree intertwined with, issues of violation .... ” Windham v. American Brands, Inc., 539 F.2d 1016, 1022 (4th Cir.1976), approved on this point en banc, 565 F.2d 59, 71 (4th Cir.1977). Seventh Amendment concerns might thus require a retrial of nearly all issues litigated in the initial phase of a bifurcated version of this lawsuit. Under these circumstances, it is difficult to see how a class suit, with its attendant costs of notice and management, would be “superi- or” to a series of individual cases brought by purchasers who desired to proceed. This difficulty is particularly pronounced when the individual actions may be consolidated, as here, for discovery and trial.

There are additional difficulties with a trial procedure that limits the absent class members’ formal interest to a determination of liability. First, the only discovered opinion squarely endorsing this form of bifurcation — Judge Wyzanski’s panel opinion in Windham — was promptly reversed en banc. See also In re Electric Weld Steel Tubing Antitrust Litigation, [1980-81] Trade Cases (CCH) ¶ 63,783 (E.D.Pa.1980), at 78,183. But cf. Bogosian v. Gulf Oil Corp., 561 F.2d 434, 455 (3d Cir.1977), cert. denied, 434 U.S. 1086, 98 S.Ct. 1280, 55 L.Ed.2d 791 (1978) (suggesting in dictum— prior to the Fourth Circuit’s en banc reversal — that Judge Wyzanski’s approach might have merit); Link v. Mercedes-Benz of N. Am., Inc., supra, 550 F.2d at 876-78 (Gibbons, J., dissenting) (approving Judge Wyzanski’s approach, but with serious reservations).

Further, the doctrine of standing requires that a private antitrust claimant be one who has suffered injury to his business or property on account of the defendants’ illegal acts. It is seriously open to question whether an absent class member may “join” a class suit if allegations of injury are not simultaneously pressed on his behalf. See generally Windham v. American Brands, Inc., supra, 565 F.2d at 71 n. 38; Link v. Mercedes-Benz of N. Am., Inc., supra, 550 F.2d at 875-76 (Gibbons, J., dissenting).

For all the reasons which have been given, the court cannot make the findings necessary for certification under Rule 23(b)(3). Plaintiffs have failed to propose a sufficiently viable and permissible method of alleviating the burden of calculating and distributing damages.

VI. Conclusion

Throughout the course of this opinion, the court has examined the manageability of a class action trial as a vehicle for resolving plaintiffs’ claims. The court has sought to “survey the factual scene on a kind of sketchy relief map, leaving for later view the myriad of details that cover the terrain.” Professional Adjustment Systems of America v. General Adjustment Bureau, Inc., 64 F.R.D. 35, 38 (S.D.N.Y.1974). This review has been at times searching, but it has not been, in this court’s view, improper. Plaintiffs’ proposals cannot survive the level of scrutiny necessary for principled application of Rule 23, and cannot anchor a finding in favor of class certification.

The court labors under no illusion that its approach can be reconciled with every decision ever written on the subject of antitrust class actions. Indeed, this opinion differs significantly from the original class action opinion issued in this very case. This opinion reflects nothing more than the court’s best understanding of the applicable law and facts. To the extent this understanding diverges from that reflected in prior decisions, such divergence is, with respect, intentional.

Perhaps this divergence is most acute in the treatment of ambiguity in the record. Many courts have simply ignored obvious failures by plaintiffs to specify how they intend to proceed. These courts have nevertheless granted class certification seemingly on the hope that things would eventually work out. This court cannot endorse such an approach. If it is unclear whether plaintiffs’ proposed method of proceeding is sufficiently viable to justify further action in that direction, such ambiguity must be construed against the plaintiffs. It is they who bear the burden on these questions.

The court recognizes, though, that some of the deficiencies highlighted in this opinion were not contested vigorously in the briefs filed in connection with the cross-motions to reconsider. Therefore, the court may permit plaintiffs to supplement the record in order to bring a renewed motion for class certification based upon firmer ground.

Accordingly, this court’s opinion and order of December 30, 1981 is vacated, defendants’ motion to reconsider is granted, and plaintiffs’ cross-motion to reconsider is denied. This case may not proceed as a class action. Liaison counsel are directed to report for status conference at 9:30 A.M. on August 4, 1983 to discuss the future of this litigation.

It is so ordered.

APPENDIX

Dr. Peltzman assumes that a uniform overcharge figure is appropriate, and that the difference between the average FOB (conspiratorial) price and the governing swap price generates a uniform figure that is no more than the constant overcharge number. (If the actual overcharge figure exceeds the number dictated by the subtraction, plaintiffs’ method is, if anything, defendant-biased, and defendants have no grounds for complaint.)

Assuming the above constraints, the swap price can be related to the other relevant variables. In the equations to follow, S represents the swap price; X represents the “actual” uniform overcharge; Y represents the number of purchasers in the relevant market (it is assumed there are more than one); and P¡ represents the non-eollusive price charged customer i.

In light of these definitions, Dr. Peltzman’s assumptions can be described algebraically as follows:

Solving for S reveals the following:

Thus, under Dr. Peltzman’s assumptions, the swap price must equal or exceed the average non-conspiratorial price.

SUPPLEMENTAL OPINION

[20] On July 1,1983, the court entered a Memorandum Opinion and Order denying class certification. However, the court specifically permitted the plaintiffs “to supplement the record and to bring a renewed motion for class certification based upon firmer ground.” (Op. at 305). The principal difficulty the court perceived was that “plaintiffs have failed to offer even a minimal explanation of why the court should accept the assumption anchoring their [damage] theory.” (Op. at 297-298).

Plaintiffs’ proposed common method of proof of damages rested on the proposition that it can properly be presumed that the defendants’ alleged illegal conduct influenced the price of each sale transaction to the same extent. (Op. at 297). The theory rested on two assumptions: first, that the swap prices at which gas was supplied to fellow defendant firms were equal to or exceeded the average marginal cost of providing the industrial gases, and second, that defendants set uniform markups above marginal cost in fixing their prices. (Op. at 298-299). The court concluded that but for the conelusory affidavit of Dr. Sam Peltzman, there was no support in the record for either critical assumption.

In response to the court’s decision, the plaintiffs submitted the affidavit of University of Chicago Professor Robert E. Lucas.¹ The defendants countered with the affidavit of University of Chicago Professor George J. Stigler and a further affidavit of Dr. Jesse W. Markham. The plaintiffs concluded with a further affidavit of Dr. Peltzman. It is in this last affidavit that Dr. Peltzman aggressively defends his assumptions against the attacks by Professors Stigler and Markham. Although most of his energy is directed toward criticizing his colleagues’ analyses rather than offering affirmative reasons in support of his assumptions, Dr. Peltzman obviously is prepared to defend his assumptions against intelligent economic attacks.

In determining whether or not to certify a class, the court is merely looking to see if the plaintiffs’ “proposed common method of proof [of damages] is so insubstantial that it realistically amounts to no method at all.” (Op. at 289). Dr. Peltzman has satisfied the court’s concerns at this stage of the proceedings.

Accordingly, the court finds that this action may proceed as a class action on behalf of the following class:

All persons, firms and corporations in the United States that purchased industrial gas from any defendant (including their respective subsidiaries or affiliates), at any time during the period of July 1, 1974 to and including June 30, 1980 but excluding (a) any defendants, or other industrial gas manufacturers and their respective subsidiaries and affiliates and excluding (b) all on-site and pipeline purchasers.

It is so ordered.

1 . Plaintiffs’ substantive claims arise under Section 1 of the Sherman Act, 15 U.S.C. § 1. See section II., infra. Section 4 of the Clayton Act, 15 U.S.C. § 15, authorizes plaintiffs’ private right of action.

2 . Defendants are Union Carbide Corp.; Air Products & Chemicals, Inc.; Airco, Inc.; Chemetron Corporation; and Liquid Air Corp. of North America.

3 . Portions of the court’s first opinion will, however, be reincorporated in this decision verbatim.

4 . The sketch of the industrial gas industry that follows purports to be only that, a broad overview. Some of the court’s statements are subject to qualifications that have been omitted in the interest of brevity.

5 . Argon must be processed further to raise its purity level to commercial standards.

6 . The price paid by distributors for such liquid product is known as the “recompression” price. .

7 . Not all distributors fit this mold. Some purchase for resale cylinders that have already been filled with gas by the defendants. Others purchase liquid product and resell it as such, either in cylinder or bulk form. Defendants’ affidavits indicate that distributors consummate a large percentage of the liquid cylinder and small-volume bulk liquid sales that are made. In their briefs, however, defendants stress only the influence of the distributors in the cylinder gas end of the market.

8 . Defendants also run a few distributorships on their own. These distributorships are operated in the same manner as their independent counterparts.

9 . The statement in the text evidences the interrelationship between the alleged conspiracies to raise price schedules and allocate customers. The two alleged wrongs are, however, conceptually distinct. See infra.

10 . In general, a claim under section 4 of the Clayton Act “arises before” a specific date when the conduct causing the damage occurs prior thereto. See Zenith Radio Corp. v. Hazeltine Research, Inc., 401 U.S. 321, 338-42, 91 S.Ct. 795, 806-08, 28 L.Ed.2d 77 (1971).

11 . The court has no quarrel with the proposition that “care must be taken not to convert the preliminary consideration of whether the [predomination] requirement of Rule 23(b)(3) has been satisfied into a predetermination on the merits. Rather, if plaintiffs’ assertion of class status is at least colorable, the court should allow the action to proceed under Rule 23.” 7A C. Wright & A. Miller, Federal Practice and Procedure, § 1781, p. 84 (1972) (emphasis added). The court understands that there are reasons to restrict the intensity of its scrutiny. See infra. The discussion in the text is intended solely to rebut plaintiffs overbroad contention that no scrutiny whatsoever is permissible.

12 . The court also finds it noteworthy that the Eisen “ban” on inquiries into the merits has often been justified on rationales sensitive to the rights of class opponents. Thus, in Eisen itself, the Court was worried “that a preliminary determination of the merits may result in substantial prejudice to a defendant, since of necessity it is not accompanied by the traditional rules and procedures applicable to civil trials. The court’s tentative findings, made in the absence of established safeguards, may col- or the subsequent proceedings and place an unfair burden on the defendant.” Id., 417 U.S. at 178, 94 S.Ct. at 2153. It has also been suggested that the Eisen “rule” is needed to protect class opponents from de facto “one-way intervention.” See Peritz v. Liberty Loan Corporation, 523 F.2d 349, 353-54 (7th Cir. 1975); see also Jimenez v. Weinberger, 523 F.2d 689, 698, 700 (7th Cir.1975), cert. denied, 427 U.S. 912, 96 S.Ct. 3200, 49 L.Ed.2d 1204 (1976).

Plaintiffs’ subsidiary contention is that the court should abstain from all scrutiny of plaintiffs’ proposed methods of proof since class determinations must be made “[a]s soon as practicable after the commencement of an action brought as a class action,” Fed.R.Civ.P. 23(c)(1), and thus prior to discovery. But the propriety of pre-certification discovery is now firmly established. Eggleston v. Chicago Journeymen Plumbers, 657 F.2d 890, 895 (7th Cir. 1981), cert. denied, 455 U.S. 1017, 102 S.Ct. 1710, 72 L.Ed.2d 134 (1982). If plaintiffs have failed to obtain the information they need to construct methods of proof capable of withstanding the limited form of scrutiny employed here, they have no one to blame but themselves. See also Pointer, “The Judge’s Viewpoint on Issues and Problems Arising in Antitrust Class Actions,” 49 Antitrust L.J. 1581, 1583 (1980).

13 . Indeed, the Simer Court rejected the idea that common questions “predominate” only when “the total time spent trying common issues would be greater than time spent on trial of individual issues.” Id., 661 F.2d at 672. The Court believed that this approach “would unduly block class actions from going forward because only the most complex of common questions would require more litigation time than a series of individual trials.” Id. When these statements are read in conjunction with those reprinted in the text, it becomes apparent that a class action may go forward even if individual trials are necessary, and even if these trials will take more time to resolve than the common questions, provided that in an absolute sense the individual trials do not render the class action unduly burdensome on the court and parties. Thus, if the class is relatively small, or if the individual trials are particularly easy to resolve, certification of the class would seem appropriate.

14 . There are undoubtedly situations in which a “manageable” class action, containing common questions that “predominate,” might not be “superior to other available methods for the fair and efficient adjudication of the controversy.” In light of the ultimate conclusion reached herein, the court need not explore this possibility.

15 . There is one qualification to this general statement. Defendants do challenge Dr. Peltzman’s internal logic to the extent he believes that defendants’ behavior inflated the prices charged end-users of cylinder gas. Defendants argue that they controlled such a small share of this sub-market, see supra, that their alleged activities could not possibly have affected cylinder prices, notwithstanding their control of over 80% of the overall market. See In re Electric Weld Steel Tubing Antitrust Litigation, [1980-81] Trade Cases (CCH) ¶ 63,783 (E.D.Pa. 1980), at 78,182-83. Plaintiffs counter that “the character and effect of a conspiracy are not to be judged by dismembering it and viewing its separate parts, but only by looking at it as a whole.” Continental Ore Co. v. Union Carbide & Carbon Corp., 370 U.S. 690, 699, 82 S.Ct. 1404, 1410, 8 L.Ed.2d 777 (1962) (quoting American Tobacco Co. v. United States, 147 F.2d 93, 106 (6th Cir.1944)). Plaintiffs further argue that “conspirators controlling a seemingly small percentage of the market can through an effective conspiracy cause classwide impact. If the conspirators are the largest producers and the price leaders, their price leadership can cause their conspiracy to have market wide impact.” (Plaintiffs’ Reply Memorandum in Support of Motion for Class Certification at 75 n. 26) What is missing from this response, however, is an indication of how, or even whether, plaintiffs intend to prove at trial that “price leadership” was present, or that the merchant market was so interrelated that the “conspiracy as a whole” must have influenced the gas cylinder sub-portion. In light of the court’s ultimate holding, this point need not be pressed further. Plaintiffs should be aware, though, that, in light of the present record, the court might not have included merchant purchasers of cylinder gas in the class had one been certified at all.

16 . These studies detail the percentage of prices in a given universe that, over various periods of time, either decreased, remained unchanged, or increased less than the full amount of the applicable schedule increases.

17 . See Madansky Aff. at ¶ 9.

18 . Defendants’ more powerful argument on the “benchmark” question is that no price schedules existed for various types of sales. Were a class otherwise certifiable, the court might have refined its definition accordingly. See generally note 15, supra.

19 . Assuming that the burden of showing the absence of due diligence is in fact properly placed upon defendants, this burden differs significantly from that faced by defendants in attempting to refute plaintiffs’ proposed proof of impact. For the reasons already given, defendants could successfully negate plaintiffs’ allegation of class-wide impact with a limited number of showings to the contrary. By contrast, an equally limited number of demonstrations of no due diligence would have relevance only for the class members in question. In order to prevail oh the question of due diligence against all pertinent members of the class, defendants would have to offer separate proof with respect to each relevant purchaser’s knowledge.

20 . The need to engage in individualized examinations of due diligence will not appreciably extend the litigation if complicated damage assessments must in any event be made. See In re Fine Paper Antitrust Litigation, 82 F.R.D. 143, 155 (E.D.Pa.1979); In re Corrugated Container Antitrust Litigation, 80 F.R.D. 244, 248 (S.D.Tx.1978); In re Plywood Antitrust Litigation, 76 F.R.D. 570, 586 (E.D.La.1976).

21 . An example may help clarify the formula plaintiffs have in mind. Assume that during 1978 the average FOB price for all liquid oxygen sold in the merchant market was 10 cents per 100 cubic feet, and that the swap price for the same product and time period was 7 cents per 100 cubic feet. Assume further that during 1979, the relevant figures (again with reference to liquid oxygen) were 12 cents per 100 cubic feet (average FOB merchant) and 8 cents per 100 cubic feet (swap). Under plaintiffs’ approach, the uniform overcharge figure governing 1978 would be 3 cents per 100 cubic feet (the difference between 10 and 7); the 1979 figure would be 4 cents per 100 cubic feet (the difference between 12 and 8).

Assume further that customer A purchased 400 cubic feet of liquid oxygen during 1978 at 12 cents per 100 cubic feet, and that customer B purchased 400 cubic feet of liquid oxygen during the same year at 10 cents per 100 cubic feet. Finally, assume that in 1979 customer A again bought 400 cubic feet at 12 cents per 100 cubic feet.

Under plaintiffs’ methodology, customer A would thus be entitled to damages equalling 12 cents (400 cubic feet times 3 cents per 100 cubic feet) for his purchases during 1978, and to 16 cents (400 cubic feet times 4 cents per 100 cubic feet) for his purchases in 1979. Customer B would receive 12 cents (400 cubic feet times 3 cents per 100 cubic feet) for his 1978 transactions.

Note that customers A and B receive the exact same award for their 1978 purchases even though they paid varying amounts in the merchant market. This equality results from the fact they both purchased the same amount of product, and from the fact that plaintiffs’ formula rests on a belief that each contemporaneous sale in the merchant market was influenced by defendants’ illegal activities to the same extent (i.e., 3 cents per 100 cubic feet in 1978). In other words, under plaintiffs’ assumptions, customer A would have paid — in the absence of defendants’ conspiratorial behavior — 9 cents per 100 cubic feet in 1978, and customer B would have paid 7 cents per 100 cubic feet. See infra.

22 . Although a private claimant under the antitrust laws need only quantify the extent of his injury with reasonable approximation, see, e.g., Story Parchment Co. v. Paterson Parchment Paper Co., 282 U.S. 555, 566, 51 S.Ct. 248, 251, 75 L.Ed. 544 (1931), such approximation must relate to the injury felt by that claimant. See, e.g., Windham v. American Brands, Inc., supra, 565 F.2d at 66; Kline v. Coldwell, Banker & Co., 508 F.2d 226, 236 n. 8 (9th Cir.1974); In re Hotel Telephone Charges, 500 F.2d 86, 90 (9th Cir.1974); A1 Barnett & Son, Inc. v. Outboard Marine Corporation, 64 F.R.D. 43, 54-55 (D.Del.1974); Landers, Of Legalized Blackmail and Legalized Theft: Consumer Class Actions and the Substance-Procedure Dilemma, 47 So. Cal.L.Rev. 842, 869-70 (1974); Handler, Twenty-Fourth Annual Antitrust Review, 72 Colum.L.Rev. 1, 37 (1972).

23 . It is conceivable that the overcharges might differ, but within a range small enough that the common number generated by the swap price method would be a sufficiently reasonable approximation of each. See generally note 22, supra. Alternatively, the overcharges might differ greatly, but each might be smaller than the swap price number. In that case, the inaccuracies in plaintiffs’ method would redound to defendants’ advantage. Neither of these possibilities have been suggested or advocated by plaintiffs.

24 . Cf. In re Master Key Antitrust Litigation, 70 F.R.D. 23, 26 (D.Conn.), appeal dismissed, 528 F.2d 5 (2d Cir.1975) (“in few class actions is there a simple per capita measure of recovery.”).

25 . The present situation thus differs greatly from what was observed on the fact of damage issue, where it was evident that plaintiffs intended to offer common proof supporting Dr. Peltzman’s “assumptions.” See supra. The record before the court offers no such assurance with respect to the presumption of uniform impact.

26 . Plaintiffs’ alternative damage theory is based upon the post-conspiracy decrease in merchant prices. It suffers from an identical infirmity since it is similarly designed to generate a common overcharge figure. See Transcript of December 3, 1982 hearing at 50.

27 . See Peltzman 2d Aff. at ¶¶ 4, 8.

28 . Of course, even if defendants’ economic logic were correct, swap prices might coincidentally equal or exceed average marginal costs. The determination of such a coincidental relationship would, however, require an initial calculation of each relevant marginal cost, and the manageability problems the swap price method was designed to cure would thus resurface.

29 . Defendants have argued that swap transactions differ from merchant sales in several additional ways, and that for this reason, FOB merchant prices can never be meaningfully compared with swap prices. In re Transit Company Tire Antitrust Litigation, 67 F.R.D. 59, 76 (W.D.Mo.1975). Dr. Peltzman has adequately answered these remaining assertions. See Peltzman 3d Aff. at ¶¶ 5-8.

30 . See generally Martino v. McDonald’s System, Inc., 81 F.R.D. 81, 91-92 (N.D.Ill.1979).

31 . Only defendants placed evidence in the record on this question. Plaintiffs explicitly declined to take discovery on how the swap prices were actually set. See Transcript of September 2, 1981 hearing at 19.

32 . Other courts have rejected damage formulae proffered by class proponents. See, e.g., Moore v. Southeast Toyota, No. CV 81-P-0491-S, slip op. at 7 (N.D.Ala. Feb. 3, 1982); Yanai v. Frito Lay, Inc., 61 F.R.D. 349, 352 (N.D.Ohio 1973); Ralston v. Volkswagenwerk, 61 F.R.D. 427, 432 (W.D.Mo.1973).

33 . Indeed, the courts have grafted even further limitations on this rule. See, e.g., In re Industrial Gas Litigation, 681 F.2d 514 (7th Cir.1982), cert. denied,-U.S.-, 103 S.Ct. 1261, 75 L.Ed.2d 487 (1983).

34 . Under subsection (c)(4)(A), “an action may be brought or maintained as a class action with respect to particular issues.”

35 . Indeed, at least two members of this Court have relied in part upon concepts of bifurcation while certifying (b)(3) classes in the aftermath of Simer. Ridings v. Canadian Imperial Bank, 94 F.R.D. 147, 151 (N.D.Ill.1982) (Aspen, J.); Board of Education v. Admiral Heating, Nos. 79 C 3046, 79 C 5253 (N.D.Ill. Aug. 10, 1982) (Shadur, J.).

36 . Stated otherwise, to the extent there is uncertainty as to the original panel’s findings, the Gasoline Products “rule is dictated for the very practical reason that if separate juries are allowed to pass on issues involving overlapping legal and factual questions the verdicts rendered by each jury could be inconsistent.” State of Alabama v. Blue Bird Body Co., supra, 573 F.2d at 318.

37 . “The majority believes that the dissents’ discussion of [the Gasoline Products] issues is premature in this case, and on this procedural point we differ. Therefore, even though some opinions are styled as dissents, their reasoning on those issues is not necessarily at variance with the views of a majority of this court.” Link v. Mercedes-Benz of N. Am., Inc., 550 F.2d 860, 865 n. 4 (3d Cir.), cert. denied, 431 U.S. 933, 97 S.Ct. 2641, 53 L.Ed.2d 250 (1977).

38 . It should be apparent that this analysis is based upon considerations unique to the Clayton Act. This opinion does not address the permissibility of bifurcation in other contexts, such as when violations of the federal securities laws are alleged.

It must be emphasized as well that the court’s discussion refers solely to cases in which inquiries into fact of damage and damages are completely bifurcated. If the second jury need only determine the appropriate values to plug into a generalized damage formula established at the first trial, Seventh Amendment concerns disappear. See, e.g., In re Plywood Antitrust Litigation, 655 F.2d 627, 636 (5th Cir.1981), cert. granted, 456 U.S. 971, 102 S.Ct. 2232, 72 L.Ed.2d 844 (1982); Arthur Young & Co. v. United States Dist. Court, 549 F.2d 686, 696 (9th Cir.), cert. denied, 434 U.S. 829, 98 S.Ct. 109, 54 L.Ed.2d 88 (1977).

1 . The plaintiffs also submitted the recent case of United National Records, Inc. v. MCA, INC., 99 F.R.D. 178, decided by Judge Nicholas J. Bua on September 30, 1983. That case did not involve any of the class manageability issues with which the parties and the court have been concerned. A much closer case decided September 28, 1983, is Abrams v. Interco Incorporated, 719 F.2d 23 (2d Cir.1983). In that case, Judge Friendly discussed the serious problem of manageability as it related to the class damage issue in that case:

An even more serious problem of manageability relates to damages. Each member of the class would be entitled not to a refund of three times what he or she paid but rather three times the amount by which such payment exceeded the prices that would have prevailed in a free market. Such determination would be complicated by the scores of different products involved, varying local market conditions, fluctuations over time, and the difficulties of proving consumer purchasers after a lapse of five or ten years, (footnote omitted)

The plaintiffs had made no showing of a common method of proof of damages and the Court affirmed the district court’s denial of class certification.