Whether an antitrust action with potentially 300,000 claimants is manageable as a class action is the question directed to us in this § 1292(b) appeal. We decline to answer on the ground that the matter is one of fact subject to determination by district court. A second query asks whether there may be separate juries utilized during the liability and damage phases of a bifurcated trial. In the absence of a definitive order on that subject by the district court, the inquiry is essentially a request for an advisory opinion, which we may not honor.

Plaintiffs Link and Katz owned Mercedes automobiles which were repaired at various times by authorized dealers. Believing the costs to be excessive, plaintiffs filed this suit under the Sherman and Clayton Acts alleging a conspiracy to fix prices. They requested a class action certification and sought injunctive relief as well as treble damages and attorneys' fees.

The plaintiffs contend that defendant Daimler-Benz A.G., the parent corporation located in West Germany, and an American subsidiary, Mercedes-Benz of North America, conspired with Mercedes dealers in the United States to maintain high prices for nonwarranty repairs. It is alleged that there was an illegal agreement to base repair rates on artificially-maintained prices of parts and flat labor repair times set out in the manufacturer's manual.

Pursuant to Fed.R.Civ.P. 23(b)(3), the district court certified the plaintiff class, consisting of some 300,000 persons in the United States who had Mercedes cars repaired during the four years in question.1 The court also listed as controlling questions of law under 28 U.S.C. § 1292(b):

A. Whether it is proper to certify a class of approximately 300,000 members where the proof of damages will vary for each member of the class;

B. Whether there can be a bifurcated trial in this case of liability and damages with separate juries for each segment of the case.

A panel of this court allowed the appeal, and two weeks later the district court filed its "Memorandum Opinion" explaining the reasons for its belief that immediate appellate review was indicated.

We have held that a grant or denial of class action certification is not a final order and, hence, not appealable under our general jurisdictional statute, 28 U.S.C. § 1291. Kramer v. Scientific Control Corp., 534 F.2d 1085 (3d Cir. 1976); Hackett v. General Host Corporation, 455 F.2d 618 (3d Cir.) cert. denied, 407 U.S. 925, 92 S. Ct. 2460, 32 L. Ed. 2d 812 (1972). However, an interlocutory order of this nature may qualify for accelerated appeal under 28 U.S.C. § 1292(b):

"When a district judge, in making in a civil action an order not otherwise appealable under this section, shall be of the opinion that such order involves a controlling question of law as to which there is substantial ground for difference of opinion and that an immediate appeal from the order may materially advance the ultimate termination of the litigation, he shall so state in writing in such order. The Court of Appeals may thereupon, in its discretion, permit an appeal to be taken from such order . . . ."

This Court does not follow a policy of freely accepting an appeal from the grant of a class action certification where such action is grounded in the discretionary power of the district court. We have taken the position that "(t)o qualify for interlocutory review in this circuit, a class certification decision must be attended by special factors which take it outside the ambit of the general rule. Katz v. Carte Blanche Corp., 496 F.2d 747, 756 (3d Cir.), cert. denied, 419 U.S. 885 (95 S. Ct. 152, 42 L. Ed. 2d 125) (1974)." Kramer v. Scientific Control Corp., 535 F.2d at 1087; Ungar v. Dunkin' Donuts of America, Inc., 531 F.2d 1211, 1213 (3d Cir.), cert. denied, 45 U.S.L.W. 3250 (U.S. Oct. 4, 1976).

We recognize that class action determination has significant practical effects on the litigation and an aggrieved party may have a very real interest in securing early appellate review. But the same considerations apply to many other types of interlocutory orders and we cannot sanction an erosion of the prohibition against "piecemeal" appellate review. Our constantly increasing caseload2 reinforces the other more philosophical reasons for that policy.

Further, as we noted in Johnson v. Alldredge, 488 F.2d 820 (3d Cir. 1973), cert. denied, 419 U.S. 882, 95 S. Ct. 148, 42 L. Ed. 2d 122 (1974), 28 U.S.C. § 1292(b) is not designed for review of factual matters but addresses itself to a "controlling question of law." In the cases where we have considered and reversed class action certification, there were other overriding legal issues: e. g., Ungar v. Dunkin' Donuts, supra ("individual coercion" and tying arrangements); Katz v. Carte Blanche, supra (superiority of test case in Truth in Lending context); Kauffman v. Dreyfus Fund, Inc., 434 F.2d 727 (3d Cir. 1970), cert. denied, 401 U.S. 974, 91 S. Ct. 1190, 28 L. Ed. 2d 323 (1971) (eligibility of class representative). Because Rule 23(c) provides that the district court's determination to permit a class action "may be conditional, and may be altered or amended before the decision on the merits," that determination, in and of itself, does not present a "controlling question of law" to which this court should be hospitable under § 1292(b). If the district court has qualms about determining a class, because it has a serious question whether it is "apply(ing) the correct criteria to the facts of the case," Katz v. Carte Blanche Corp., 496 F.2d at 756, (a) it should hesitate in determining the class until reasonably assured of the correctness of its ruling and (b) it should not certify for § 1292(b) consideration without stating persuasive reasons why the particular class action question is so unusual as to demand the intervention of an appellate court. In affording immediate appellate review of "controlling questions of law," § 1292(b) was not designed to substitute wholesale appellate certainty for trial court uncertainty under circumstances where, as here, the Rule gives broad discretion to the district court to revise its class action determination at any time prior to the decision on the merits.

One other observation is in order. Section 1292(b) is not intended to grant the appellate courts power to give advice on speculative matters. While counsel and the district court might believe it helpful to have the appellate court's view on proposed alternate courses of action, our jurisdiction extends only to orders of the district court. These orders must be definitive, effective, and in a posture capable of affirmance or reversal. Requesting that the Court of Appeals choose from a number of alternatives submitted is not the type of appeal envisioned by § 1292(b).

With these concepts in mind, we turn to an analysis of the case at hand. The district court's memorandum discusses the troublesome problems anticipated in proving the damages of 300,000 separate class members and notes that a number of courts have held a class of that size unmanageable. See Ralston v. Volkswagenwerk A.G., 61 F.R.D. 427 (W.D.Mo.1973), Boshes v. General Motors Corp., 59 F.R.D. 589 (N.D.Ill.1973). Contrary authority is also cited for the position that sheer size does not defeat the superiority of a class action. In re: Master Key Antitrust Litigation, 70 F.R.D. 23 (D.Conn., filed May 27, 1975), 1975-1 Trade Cases P 60,377. However, earlier in the memorandum, the judge stated:

". . . in the event liability is found against the defendant, a further determination will be made as to the continuation of the class action on the damage issue."

The district court therefore viewed the scope of its order as limited and contingent. It contemplated but did not order: (1) a bifurcated trial on liability, and (2) a re-examination of class certification before the damage phase begins.

We have some doubt that the anticipated difficulties of proving damages on behalf of the large class are actually before us. The district court mused that counsel might be able to devise methods of solving the problem before that time arrived, and listed several possibilities. In any event, we conclude that the issue is not one which this court should decide on interlocutory review.

It is obvious that the district court's concern a justifiable and serious one is with manageability. This is a practical problem, and primarily a factual one with which a district court generally has a greater familiarity and expertise than does a court of appeals. Consequently, it is an area in which the trial court must of necessity be granted a wide range of discretion. The housekeeping problems involved in notifying an extremely large group of people, reviewing replies, answering inquiries, and processing of claims as well as the effect on other cases on the court's docket, are all matters which can best be evaluated by the man on the scene. These are often difficult judgments and that fact has not been overlooked in the rules.

Rule 23(c)(1) provides that the determination must be made "as soon as practicable." That does not necessarily mean at the onset of the litigation. The designation should be made at a time when the trial court has had adequate opportunity to acquaint itself with the case and the complexities likely to be encountered in its disposition. Moreover, as we noted earlier, a class certification, once made, is not irrevocable. The rule in a pragmatic approach states that "an order under this section may be conditional, and may be altered or amended before the decision on the merits." Thus, after a determination of liability, the district court is free to decertify the class for a proper reason,3 and unmanageability would be such a circumstance.

The district court also has the option of creating subclasses if that appears helpful in handling the litigation. In short, the rule is flexible and recognizes that whether the case can be managed must be left to the informed discretion of the trial court which will manage it. There are no circumstances in this case which makes manageability a controlling question of law, and we therefore will not answer the question submitted.

The second question certified by the district court is the propriety of having separate juries for each segment of the bifurcated trial. The court, however, entered no order directing separate juries and whether it will is a matter of pure speculation at this juncture. As the memorandum points out, the parties may be able to develop a method of proving damages which both satisfies legal standards and permits the efficient use of court time. The utilization of a master is a possibility. A jury waiver would not be unexpected in the damage phase of a trial where the evidence consisted only of the mechanical application of mathematical formulations.

A consideration important to our decision is that the case is in a preliminary stage. At the time the appeal was taken, the defendants had not yet filed their answers, the early skirmishing having been confined to discovery on jurisdictional issues. Those matters have since been resolved against the defendants, and the parties are now in a position to prepare the case on the merits. We believe that further development of the facts will aid the trial court in choosing among the alternative procedures discussed in its memorandum. By the time of the pretrial conference, the issues should be more sharply defined, and the district court may be in a better position to make positive rulings rather than merely "thinking out loud," as its memorandum may fairly be characterized.

The better appellate practice is to rule on those issues which have been decided, and where a decision will "materially advance the ultimate termination of the litigation." We read the district court's order and memorandum as failing to meet those criteria. In short, we have been asked for an advisory opinion on an interesting legal proposition but one which may never be invoked in this case, see Nickert v. Puget Sound Tug & Barge Co., 480 F.2d 1039 (9th Cir. 1973); Control Data Corp. v. International Business Mach. Corp., 421 F.2d 323 (8th Cir. 1970). We must decline the invitation.4

Accordingly, we remand this case to the district court.

SEITZ, Chief Judge, concurring.

I believe that the majority have not adequately explained their refusal to answer either of the questions certified by the district court. It is not incumbent upon the court to state reasons why it would or would not vote to accept an appeal under § 1292(b). Senate Report No. 2434, 85 Cong. 2d Sess., pp. 3, 4, U.S.Code Cong. & Admin.News, 1958, p. 5255. However, I consider it useful to do so in this case, especially as a prior panel of our court granted permission to appeal.

There is an important distinction between the functions of the panel passing on the original request to allow an appeal under § 1292(b) and the panel which thereafter considers an appeal which has been allowed. The original panel is not required to do more than agree that the certified matter appears to be worthy of immediate appellate consideration. On the other hand, the panel which considers the matter after full briefing and consideration of the issues may well be in a better position to determine that what seemed proper for interlocutory review really should not be so reviewed.1 I therefore do not consider the majority's present decision to be in impermissible conflict with the decision of the earlier panel. Nor do I say this because the matter is now considered in banc.

The majority have refused to determine whether the class was properly certified because they fail to discern a "controlling question of law." I might agree with their conclusion were we untrammeled by precedent. However, in Katz v. Carte Blanche Corp., 496 F.2d 747 (3d Cir.), cert. denied, 419 U.S. 885, 95 S. Ct. 152, 42 L. Ed. 2d 125 (1974), a majority of this court sitting in banc held that whether a district court has committed an abuse of discretion in certifying a class is a controlling question of law for purposes of § 1292(b). I nevertheless concur in the majority's refusal to discuss the propriety of class certification, since there has not been a sufficient showing that the issues raised admit of a "substantial ground for difference of opinion." As the majority noted in Katz, it is unlikely that a district court would certify that there is "substantial ground for difference of opinion" as to the correctness of one of its orders unless it thought so. But it can hardly be assumed that district courts will always properly review the criteria mentioned in § 1292(b): as the majority stated in Katz, "(d)enial of permission to appeal may be based upon a different assessment than that of the district court as to any of the three (mentioned) criteria." 496 F.2d 747, 754.

Appellants have questioned the class certification order on several grounds,2 including that the district court failed to properly identify the factors relevant to the class certification decision and that it abused its discretion in evaluating the factors which it did identify. But the only allegation of error which poses a substantial and out of the ordinary question is that the certification is invalid because it was premised on the assumption that the court could decertify the class after the initial stage of the proceedings.

Were it clear that the court did rest its decision on this assumption, I might conclude that there is "substantial ground for difference of opinion" as to the propriety of the class certification. But it is far from clear that the district court was thinking of decertification under F.R.Civ.P. 23(c)(1) rather than the possibility of holding bifurcated proceedings within the overall class action suit. The portion of the court's order which sets forth the "controlling questions of law" mentions a "bifurcated trial in this case of liability and damages," and does not mention decertification, which would seem to relegate proof of any damages to an entirely separate lawsuit. The court's Memorandum Opinion states that "(t)he most appropriate way to proceed in this case is to bifurcate the issues of liability and damages," and indicates how discovery would proceed in the event there was a finding of liability. While the court also said that it had the "right to decertify the class following a determination of liability against the defendants," it cited for this assertion Ungar v. Dunkin' Donuts of America, Inc., 68 F.R.D. 65 (E.D.Pa.1975), rev'd on other grounds, 531 F.2d 1211 (3d Cir.), cert. denied (1976), which refers to a bifurcated trial, not decertification. He also cites 3B Moore's Federal Practice, Para. 23.45(2), which speaks generally of separate "proceeding(s)" for assessing individual damages, and does not specifically discuss decertification.

In sum, the question of whether the district court could decertify the class after the initial stage of the proceedings is not clearly before the court. Since the other allegations of error do not raise a "substantial ground for difference of opinion" within the statutory meaning and since this court has a legislative mandate to ensure that § 1292(b) does not undermine the proper scope of the finality rule, I conclude that we should not address the propriety of the class certification, even though that question, if entertained, might be easily disposed of. See Milbert v. Bison Laboratories, 260 F.2d 431 (3d Cir. 1958).

The majority also decline to discuss whether it is permissible under the Seventh Amendment to have a bifurcated trial of liability and damages with separate juries for each segment of the case. Their decision is based on the fact that the district court has not at this point ordered such a trial. But I do not think this is a complete response to the issue posed, since one can still argue that the Seventh Amendment claim may be entertained on this appeal because it is raised by the class certification order.

I conclude that the Seventh Amendment problem should not be considered because it is not sufficiently implicated in the class certification decision. We should not consider the Seventh Amendment questions which might inhere in decertification procedures, since, as noted, it is not clear that the court based its decision to certify on the assumption that it could decertify before proof of individual damages. Moreover, as to bifurcating the present action into liability and damage stages, the court's Memorandum Opinion mentions several techniques other than separate juries by which the proof of damages might be handled in a bifurcated proceeding, such as using a Master to calculate damage to individual class members, or using "expert testimony, statistical computations and computer analysis."

I therefore concur in the disposition by the majority if, as I understand it, the majority is vacating the earlier order of our court granting permission to appeal and remanding the case to the district court.

Had I been a member of the panel that was presented with the question whether to grant a certificate under section 1292(b), I might have voted against entertaining an interlocutory appeal. This is so because I believe that the position of this Court has been one of reticence in reviewing district court orders certifying suits as class actions, at least in the absence of an intertwined controlling legal issue.1

However, this case is now before us in a markedly different posture.2 Permission to appeal was in fact granted, a significant amount of time has elapsed since certification,3 substantial efforts have been expended in proceedings before this Court, and it now appears that a serious and controlling legal question has emerged with respect to the issue of separate juries to try the liability and damage phases. Given these factors, it would seem to me to be improper to vacate, summarily, the grant of permission to appeal. Consequently, I believe that the Court should proceed to a consideration of the substantive points that have been tendered.

With respect to the first of appellants' contentions, the decision of the trial judge to certify this case as a class action, in my judgment, would not appear to be, at least on the facts available at this time, an abuse of discretion. However, I believe that it would have been preferable if he had undertaken the species of detailed analysis recommended by Judge Van Dusen.4

The thrust of the appellants' claim that the order of the district judge should be reversed appears to be that the suit is unmanageable as a class action. Yet, a district judge is ordinarily in a better position to determine whether class action treatment is appropriate on grounds of manageability, and his conclusion in that regard should not be disturbed in the absence of compelling circumstances. It is true that the contemplated class of 300,000 is an admittedly large one, but other cases have approved classes of comparable numbers. Moreover, the size of the class does not appear to make the substantive issues, at least in the liability stage of the case, intractable. This would seem to be so because the central contention of the class representatives, so far as the merits of the controversy are concerned, is that the two principal defendants entered into a single conspiracy with their dealers to fix the cost of repairs.

As to the jury issue, I am in accord with Judge Gibbons that review of this problem would not, in the context of the case at hand, constitute an advisory opinion.5 The trial judge did not speculate about using the device of a bifurcated trial and separate juries. Instead, a fair reading of the district judge's opinion indicates that the assumed propriety of that mechanism appears to have been one of the prime predicates of his decision to certify the case as a class action,6 at least in the first instance. Inasmuch as the trial judge's determination turned in large measure on the plan to use separate juries, I believe that this Court should decide this issue.

Since the majority has chosen not to confront the jury problem, I do not believe that it would be appropriate, at least as this time, to set forth my views on the constitutionality of the separate jury procedure proposed by the district judge. Under the disposition of the appeal made today, it may well be that we will have to face this issue in the future. At that juncture, there will be ample opportunity to resolve definitively this difficult and important question, in light of the precise factual matrix in which the matter will then arise. The thorny problems of waiver and the possible designation of a master will by that time have been stripped away, and a direct encounter with the issues in a more pristine form will be possible.7

Circuit Judge ROSENN joins in this opinion.

VAN DUSEN, Circuit Judge, dissenting.

I respectfully dissent from the majority opinion's holding that a district court order, which certifies a class and adopts a bifurcated trial of a private civil antitrust action for treble damages under 15 U.S.C. § 15, does not qualify under 28 U.S.C. § 1292(b). My dissent is based on the district court's failure to make findings or show consideration of these factors,1 among others:

(a) that the liability aspect of the trial required a decision not only of violation of §§ 1 and 2 of the Sherman Act, 15 U.S.C. §§ 1 and 2, but also determination under 15 U.S.C. § 15 of the fact of damage2 (an element which would give rise to problems in proving the latter item on a class basis where the class consists of "potentially 300,000 claimants" located in all 50 states, Puerto Rico, Guam, and the Virgin Islands);

(b) that there would be a substantial difference in the quantum and character of liability proof under § 4 of the Clayton Act, 15 U.S.C. § 15, if the plaintiffs included all the above 300,000 individuals alleged to comprise the class, rather than only the named plaintiffs;

(c) that determination of liability itself under § 1 of the Sherman Act might also vary as to each class member, depending upon whether the particular dealer-defendant participated in the conspiracy, so that inquiry is required into whether the common questions predominate over the individual questions making a class action superior under F.R.Civ.P. 23(b)(3);

(d) that, on this record, there would appear to be no theoretical or practical formula or method to aid in the computation of damages sustained by different individuals, potentially requiring each class member to produce voluminous documentary evidence of his transactions in order to secure judgment;

(e) that, in view of the above, there is a question as to whether the named plaintiffs are typical as required by F.R.Civ.P. 23(a)(3);3

(f) that, if more than one class is necessary, the named plaintiffs might not adequately represent many members of the group of 300,000 persons treated as members of various classes and subclasses by the district court, see F.R.Civ.P. 23(d)(4); and

(g) that class suits have an in terrorem effect in forcing settlement (cf. Blue Chip Stamps v. Manor Drug Stores, 421 U.S. 723, 740-42, 95 S. Ct. 1917, 44 L. Ed. 2d 539 (1975)).

This court has consistently recognized that if the district court "has not properly identified the issues and not properly evaluated which are common, the (district court order) is not entitled to such deference." Katz v. Carte Blanche Corporation, 496 F.2d 747, 756-57 (3d Cir. 1974), and cases there cited,4 where Judge Gibbons also quoted Professor Moore as follows (at 757):

" 'In determining whether an action brought as a class action is to be so maintained the trial court should carefully apply the criteria, set forth in Rule 23 . . ., to the facts in the case; and if it fails to do so its determination is subject to reversal by the appellate court when the issue is properly before the latter court.' "

I agree with part I of Judge Gibbons' opinion.5 As to part II, I do not believe that class action treatment is permissible on this record, where a " § 4 damage action charging price fixing" is involved (see page 877 of Judge Gibbons' opinion), although I agree with Judge Gibbons that the Seventh Amendment is no bar to a bifurcation of the trial on this case. In view of Bruszewski v. United States, 181 F.2d 419 (3d Cir. 1950), a bifurcated trial of the § 1 Sherman Act claim apparently can be more appropriately tried in a test action brought by the named plaintiffs. See Katz v. Carte Blanche Corporation, 496 F.2d 747, 759-61 (3d Cir. 1974) (en banc).

I would vacate the district court order and remand for district court (1) consideration of the factors mentioned above in light of the antitrust issues set forth in Judge Gibbons' opinion, and (2) appropriate findings and conclusions after such consideration.

GIBBONS, Circuit Judge, dissenting.

This is an appeal, pursuant to 28 U.S.C. § 1292(b),1 of an order of the district court determining that the suit in question could be maintained as a Fed.R.Civ.P. 23(b)(3) class action. Plaintiffs seek to represent a class of all persons, firms, or corporations who owned or leased a Mercedes-Benz automobile during the period from March 1970 to March 1974, and had non-warranty repairs performed by authorized Mercedes dealers during that period. The defendants are Daimler-Benz A.G., the West German automobile manufacturer, and its wholly-owned United States distributor, Mercedes-Benz of North America, Inc. The complaint charges that the defendants conspired with individual Mercedes retail dealers, referred to as co-conspirators but not named as defendants, to fix the retail prices of parts and the flat labor repair times for non-warranty repairs, in violation of § 1 of the Sherman Act. 15 U.S.C. § 1. The complaint seeks on behalf of the class both money damages under § 4 of the Clayton Act2 and injunctive relief under § 16 of the Clayton Act.3

A substantive element of the § 4 cause of action is proof of actual injury to the business or property of each plaintiff resulting from the alleged violation.4 Since the § 4 cause of action is for money damages the defendant is entitled to, and has demanded, a jury trial.5 The injunctive remedy in § 16, on the other hand, requires only proof of threatened loss or damage.6 Since the § 16 cause of action looks only to equitable relief it presents no jury trial issue.7

On September 9, 1974, the plaintiffs moved, pursuant to Fed.R.Civ.P. 23(c), for a class action determination. Affidavits filed in opposition to that motion suggest that there are approximately 300,000 members in the proposed class, that the individual Mercedes dealers number about 588, and that the number of different parts, the prices of which were allegedly fixed, is over 26,000. On July 8, 1975, the district court, without making any distinction between the § 4 damage claim and the § 16 claim for injunctive relief, entered an order that the action "shall be maintained as a class action pursuant to Fed.R.Civ.P. 23(b)(1), 23(b)(2) and 23(b)(3)." Upon the entry of that order defendants moved for an order amending the July 8, 1975 order, to provide that in the district court's opinion an immediate appeal is warranted under 28 U.S.C. § 1292(b) as it would materially advance the ultimate termination of the litigation.8 Acting on that motion, on August 7, 1975, the district court vacated the July 8, 1975 order and entered a new order providing that the "action shall be maintained as a class action pursuant to Fed.R.Civ.P. 23(b) (3) (only)" and that an immediate appeal is authorized in accordance with § 1292(b).

In support of this August 7, 1975 order the district court filed an opinion. This opinion set forth common questions of law and fact which, in the court's belief, predominated over questions which were not common to the class members.9

A prerequisite to a 23(b)(3) class action is the existence of questions of law or fact common to the class . . . .

The common questions for determination include: (1) whether defendants combined and conspired to establish retail parts prices to be charged to the class by the dealers; (2) whether defendants combined and conspired to establish designated labor times to be utilized by dealers in setting prices to the class for nonwarranty repairs of Mercedes-Benz vehicles, (3) whether documents establishing prices for parts and the amount of designated labor time were circulated among the dealers; (4) whether defendants enforced compliance by the dealers in regard to the prices for parts for designated labor time; (5) whether the defendants conspired with the Mercedes-Benz National Dealer Council in regard to establishment of fixed prices for parts and designated labor time for repair work in the performance of nonwarranty auto repairs; (6) whether the plaintiff class was harmed by the alleged acts of the defendants and whether the alleged acts of the defendants, if proved, violate the antitrust statutes of the United States.10

It can be seen immediately that questions (1) through (5) go to the establishment of a § 1 Sherman Act violation. But proof of such a violation does not establish civil liability under § 4 or § 16 of the Clayton Act. There must under § 4 be proof of injury to business or property,11 and under § 16 proof of threatened loss or damage.12 Awareness of the distinction between acts which violate § 1 of the Sherman Act, and the elements which establish liability in private party litigation under § 4 and § 16 of the Clayton Act is vital. A conspiracy to establish maximum price fixing, for example, would clearly violate the Sherman Act,13 but would not be a cognizable action under § 4 on behalf of retail consumers since they would have suffered no injury as a result of such acts.

In the district court opinion the only fact question referring to the essential element of either private cause of action is question (6), "whether the plaintiff class was harmed by the alleged acts." The formulation of this question, however, demonstrates the court's misapprehension of the nature of § 4 liability in a Rule 23(b)(3) context, for obviously only individual members of the class, not the class, can be injured in business or property by a retail price fixing conspiracy.14 This misapprehension is further reflected in the district court's statement that a common legal question included: "whether the alleged acts of the defendants, if proved, violate the antitrust statutes of the United States."15 The only substantive antitrust statute alleged to have been violated is § 1 of the Sherman Act. That a price fixing conspiracy, if proved, violates § 1, hardly seems an open legal question.16

The analytical deficiencies of the district court's approach are further revealed later in its opinion, when it writes: "I find that the claim of the plaintiffs as individuals is typical of the claims of the class, especially on the liability issue."17 This statement is interesting since the court has never properly identified the liability issue. Possibly what was intended was a reference to the Sherman Act § 1 violation. Certainly there is nothing in the district court opinion, or in the affidavits which have been called to our attention, suggesting that the conspiracy had a like effect upon the business or property of every repair customer of all 588 individual Mercedes dealers.18 It is arguable that proof of the mere existence of the conspiracy might suffice to establish threatened loss or damage, and thus to sustain injunctive relief under § 16.19 But the district court opinion draws no such distinction, and its discussion is focused primarily on establishing liability under § 4 rather than § 16. The district court's misconception of the substantive elements of § 4 liability are clearest in its treatment of the distinction between § 4 "liability" and "damages." Nowhere does the court show an awareness that some degree of injury to business or property damages if you will is a substantive element of § 4 liability. The court, however, recognizing that some proof of the amount of the overcharges resulting from the conspiracy will have to be proved at some point, and that the necessity for such proof has something to do with the class action determination, concludes:

(T)he troublesome question in certifying the class involves the problems (sic) anticipated in proving damages on behalf of 300,000 separate class members. This argument was the basic contention made by the defendants to resist the certification of the class. The parties should not be subjected to the trial of an antitrust matter involving a class of 300,000 members if (the) . . . anticipated problems in proving damages render the certification inappropriate. Considering the question the district court has decided the problem can best be approached by bifurcating the issues of liability and damages.

Despite the majority opinion of this court, it is clear beyond dispute that the district court's decision in favor of Rule 23(b)(3) class action certification was based upon a decision to try the liability and damage issues before separate juries. In its order of August 7, 1975 the order which we are reviewing the district court in certifying its class action determination for § 1292(b) review, stated:

7. The undersigned is of the opinion that an immediate appeal from this Order is authorized by 28 U.S.C. § 1292(b), in that:

1. It involves controlling questions of law as to which there is substantial ground for difference of opinion, namely

B. Whether there can be a bifurcated trial in this case of liability and damages with separate juries for each segment of the case; . . .

Disingenuously, the majority suggests that because the district court has not yet entered a formal order providing for a trial before separate juries our discussion of the propriety of such a procedure would be an "advisory opinion."20

The district judge will surely be surprised to learn that this court's review of his conclusion on a proposition of law would amount to an advisory opinion. The decision to have separate juries for each segment of the bifurcated trial was an essential postulate of the district court's reasoning in support of its order granting class certification. It is clear beyond peradventure that the district judge would not have entered the class action order except for his belief in the legal proposition that he could try liability and damages in a § 4 Clayton Act case before separate juries.

Under 28 U.S.C. § 1292(b) this court reviews an order which because it is appealable is a judgment. Fed.R.Civ.P. 54(a). We do not review opinions or certified questions.21 But when the district court advances a legal proposition in support of its judgment this court's review of the correctness of that legal proposition is not advisory. Such review is simply a natural consequence of reviewing any lower court judgment. We have repeatedly recognized that in a § 1292(b) appeal we consider any legal issues which suggest that the judgment before us can be affirmed or should be reversed.22 The majority's reference to an "advisory opinion" is manifestly insupportable.

Equally insupportable is the majority's attempt to justify its unwillingness to decide the significant issue tendered by this appeal by reference to "(o)ur constantly increasing caseload."23 In another context I have warned of the pernicious influence of the "fatigue" factor in appellate review.24 In this instance that factor cannot be said to be even remotely relevant. For § 1292(b) appeals from class action determinations, while highly significant for the litigants and the district court in particular cases, are an insignificant part of the appellate caseload. A leading commentator writes:

Though a good deal has been written about § 1292(b), numerically the statute has not been of great importance. In the fiscal year 1974, 16,436 appeals were taken to the eleven courts of appeals. By contrast trial court certificates under § 1292(b) are made in only about 100 cases a year and the courts of appeals allow interlocutory appeal in about half of those 100 cases.

Wright, Federal Courts 518-19 (3d ed. 1976) (footnotes omitted).

Since the holding in Katz v. Carte Blanche Corporation, supra, on March 15, 1974, that class action determinations may be reviewed by the § 1292(b) certificate route, the Third Circuit experience has not been different from the national experience to which Professor Wright refers. Between March 15, 1974 and December 31, 1976 the district courts in this Circuit have filed § 1292(b) certificates in 67 cases, of which 29 have been rejected and 35 granted by this court. But of the 67 petitions only five involved class action issues, and two of those were denied. The instant case is one of the three in which the petition was granted. The other two are Braden v. University of Pittsburgh (Civ. No. 75-1657, presently pending en banc); and Ungar v. Dunkin' Donuts of America, Inc., 531 F.2d 1211 (3d Cir.), cert. denied, 429 U.S. 823, 97 S. Ct. 74, 50 L. Ed. 2d 84 (1976). A parsimonious attitude toward our review of § 1292(b) appeals would have excluded these two highly significant cases. And the consequences might well have been waste of district court time and effort, and cost to the litigants vastly disproportionate to the time spent by this court in disposing of them. We cannot let our feeling of being overworked distort our sense of proportion. This appeal should be decided not dismissed.

The legal issue which is inextricable from the merits of the district court's class action determination arises from the demand by defendants for a unitary jury trial of the § 4 Clayton Act damage claim. In making that demand the defendants rely upon the Seventh Amendment which, they urge, guarantees a trial before a single jury rather than separate juries as contemplated by the district court. But defendants somewhat overstate the reach of that Amendment. In the seminal case, Gasoline Products v. Champlin Refining Co., 283 U.S. 494, 51 S. Ct. 513, 75 L. Ed. 1188 (1931), the Supreme Court held that the Seventh Amendment right to jury trial is not abrogated when an appellate court orders a new trial only on one of the issues originally tried before a unitary jury in the court below. The Court reasoned that the Seventh Amendment guaranteed a unitary jury trial only of issues, not of cases:

Lord Mansfield, in applying the common law rule where the verdict, correct as to one issue, was erroneous as to another, said: ". . . For form's sake, we must set aside the whole verdict. . . ." Edie v. East India Co., 1 W.Bl. 295, 298. But we are not now concerned with the form of the ancient rule. It is the Constitution which we are to interpret; and the Constitution is concerned not with form, but with substance. All of vital significance in trial by jury is that issues of fact be submitted for determination with such instructions and guidance by the court as will afford opportunity for that consideration by the jury which was secured by the rules governing trials at common law. See Herron v. Southern Pacific Co., 283 U.S. 91 (51 S. Ct. 383, 75 L. Ed. 857). Beyond this, the Seventh Amendment does not exact the retention of old forms of procedure. See Walker v. Southern Pacific R. Co., 165 U.S. 593, 596 (17 S. Ct. 421, 41 L. Ed. 837). It does not prohibit the introduction of new methods for ascertaining what facts are in issue (see Ex parte Peterson, 253 U.S. 300, 309 (40 S. Ct. 543, 64 L. Ed. 919)), or require that an issue once correctly determined, in accordance with the constitutional command, be tried a second time, even though justice demands that another distinct issue, because erroneously determined, must again be passed on by a jury. . . .

Here we hold that where the requirement of a jury trial has been satisfied by a verdict according to law upon one issue of fact, that requirement does not compel a new trial of that issue even though another and separable issue must be tried again.25

Defendants urge that the Gasoline Products holding is limited to retrials following an initial unitary jury trial of the entire case. They point out that in a recent class action opinion of this court we referred specifically to the problems of the right to a unitary trial26 which is a question of first impression for this circuit.

Gasoline Products does not suggest that its distinction between issues as opposed to cases is limited to retrials, and given the analysis which the Court made, no Seventh Amendment policy suggests such a limitation. Thus, I agree with the Fifth and Tenth Circuits,27 and with the commentators,28 that the Seventh Amendment guarantees a unitary trial of issues, not of cases, even in the first instance.

That conclusion does not end the inquiry, however, for in Gasoline Products the Court also made it clear that a corollary to its distinction between issues and cases was the rule that any issues to be tried before separate juries must be clearly distinct and separable.

Where 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.29

Such a 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.

Defendants contend that the issues of liability and damages are so interwoven in a price fixing action seeking damages under § 4 that they cannot be heard by separate juries. Admittedly, in an action under § 4 the task of defining which issues are so separate and distinct as to satisfy the test of Gasoline Products is complex. Since "liability" under § 4 necessarily includes proof of injury to business and property, 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.

A way out of this possibly constitutional dilemma in having separate juries hear overlapping issues in private antitrust class actions was suggested recently by Judge Wyzanski while sitting by designation in the Fourth Circuit, in Windham v. American Brands, Inc., 539 F.2d 1016 (4th Cir. 1976), rehearing en banc granted December 13, 1976. Recognizing the class manageability problems flowing from the necessity to prove both a Sherman Act § 1 violation, and the resulting impact of that violation in order to establish § 4 liability, he ordered that the trial issues be bifurcated between the Sherman Act § 1 violation and the injury to business or property. Bifurcating these issues eliminates problems of overlapping proof, since all evidence relating to the fact and extent of damages will be relegated to the second stage of trial. The first stage of the trial will be concerned only with the defendants' alleged violation of § 1.

This method of trial raises an issue of the propriety of allowing a class representative to establish that a defendant has committed a Sherman Act violation, without also proving, in the same trial, that he has been injured in his business or property. Such a class representative would have performed the role of a private attorney general by obtaining a violation holding benefiting all of the other class members, even though he has not yet met his burden in proving all of the elements of § 4 liability. Thus Judge Wyzanski's procedure might result in a violation verdict on behalf of the class being established by a class representative who ultimately failed to prove injury to his business or property. Such a scenario is somewhat troubling because the language of § 4 suggests that private parties should not have standing to litigate Sherman Act violations unless they have been actually injured by those violations.30

Judge Wyzanski relies heavily on the analogy between his suggested bifurcation procedure and a private antitrust damage suit following a successful government antitrust suit.31 In such a private action the judgment obtained by the government is prima facie evidence that the alleged defendants have violated the antitrust laws. 15 U.S.C. § 16(a).32 The analogy to the government action is incomplete in two respects. First, if in the violation stage the case is maintained as a class action, the violation determination will be given a much greater effect than would a government judgment. While the government judgment is only prima facie evidence against a defendant in a later private damage action, the class action violation determination becomes law of the case, and eventually res judicata, in favor of or against all class members who have not opted out.33 Secondly, the class representative is not a perfect analogy to the Justice Department, or even to a State attorney general.34 Congress has not authorized a general roving commission for the enforcement of the antitrust laws. Instead, it has authorized private enforcement only by persons suffering actual or threatened injury to their business or property.35 Under Judge Wyzanski's approach, however, a party who never really suffered any injury to his business or property might nevertheless obtain a violation determination having a more significant effect than a decree in a government action. Perhaps concern about such a possibility explains the Fourth Circuit's recent vote to rehear Windham v. American Brands, Inc., supra, en banc.

Even though the analogy between a class representative and the government is not complete in all respects I see no statutory or due process barriers to the use of Judge Wyzanski's procedure. Realistically, law firms are the prime movers in most consumer class actions. Any class representative who initiates an action under § 4 must still satisfy all of the threshold pleading requirements. If a law firm has a client on whose behalf it is willing to plead the necessary § 4 standing allegations there should be no objection to recognition of that client's standing to litigate a § 1 violation in a class action through a violation verdict. See Fed.R.Civ.P. 11. Permitting bifurcation between issues of violation and issues of injury to business or property seems to me perfectly consistent with the remedial purposes of Rule 23(b)(3).36

Even under Judge Wyzanski's approach, management problems in the case sub judice would remain following a violation determination. If they were insurmountable the foregoing discussion would amount to no more than an academic exercise. They do not appear so to me. The chief source of the difficulty is the defendants' perfectly proper insistence upon a jury trial. If in an initial jury trial violation of the Sherman Act was established, the class members could be notified of that verdict and afforded the opportunity to file, with a designated depositary, claims for any injury they allege. The court could control discovery with respect to such claims, reserving to the defendant the right to try before a jury all those as to which the fact or amount of injury was disputed. Whether this involved one, several or many juries would make no difference, since each jury would be passing upon a discrete set of issues.

Daimler and Mercedes contend that such an approach would ultimately involve the trial of several hundred thousand separate fact issues. Certainly that would not be the case if they obtained a verdict in their favor on the issue of violation of § 1. If the verdict should go against them on that issue it does not seem overly likely that they will dispute, through trial, anything near that number of claims. But more to the point is the observation that if they have engaged in a price fixing conspiracy there may well be several hundred thousand victims with potential claims. If the claims were pressed individually the court would have to hear them. At least if violation is established in a single trial individual claimants will not have to relitigate that part of the case. The district court did not think the multiplicity of claims presented an insurmountable problem. Neither do I.

This court has never definitively determined whether a Rule 23(b)(3) class action is available for a large class in a § 4 Clayton Act case. The Ninth Circuit, recognizing the difficulties presented by the statutory language of § 4, has in a § 1292(b) appeal held that class certification in such a case was improper. In re Hotel Telephone Charges, supra, 500 F.2d at 89-90. The Fifth Circuit has also affirmed the denial of class action status in a § 4 case involving a large number of plaintiffs. Shumate & Co., Inc., supra, 509 F.2d at 155. District Court opinions present no consistent pattern.37 The closest we have come to addressing the issue is Judge Aldisert's opinion in Ungar v. Dunkin' Donuts of America, Inc., supra. But Ungar is distinguishable because the violation alleged was not, as here, a price fixing conspiracy, but a tie-in. Our rejection of class action determination was based upon the conclusion that a substantive element of a tie-in violation was proof of individual coercion. Since individual coercion had to be shown we concluded that individual questions of proof predominated over common questions, even at the violation stage of the case.38 A § 1 Sherman Act price fixing charge is an entirely different matter. Proof of a price fixing conspiracy establishes the violation regardless of its effect upon individual customers. Thus I would distinguish Ungar, and hold that class action treatment is permitted at least in a § 4 damage action charging price fixing.

My conviction that the approach outlined above is consistent with current Congressional intention with respect to antitrust enforcement is reinforced by an examination of the Hart-Scott-Rodino Antitrust Improvement Act of 1976 (Antitrust Improvements Act).39 It should be recalled that in Eisen v. Carlisle & Jacquelin, supra, and Zahn v. International Paper Co., 414 U.S. 291, 94 S. Ct. 505, 38 L. Ed. 2d 511 (1973), the Court for all practical purposes precluded an interpretation of Rule 23 which would permit a fluid class recovery.40 And, it had previously held that a state could not maintain a parens patriae suit on behalf of its residents. Hawaii v. Standard Oil Co., 405 U.S. 251, 92 S. Ct. 885, 31 L. Ed. 2d 184 (1972). Congressional reaction to these holdings was the enactment of § 301 of the Antitrust Improvements Act which amends § 4 of the Clayton Act by adding a provision permitting a State attorney general to bring a parens patriae class action on behalf of residents of his state to secure money damages authorized by § 4. The statute provides for proof of aggregate damages without the necessity of separately proving the individual claim of or amount of damage to persons on whose behalf the suit is brought. It also sets forth a method of distribution. Since the § 301 cause of action does not apply to injuries sustained prior to September 30, 1976 it is inapplicable to the instant suit.41 But the enactment of a statute providing for a parens patriae class action with a fluid class recovery suggests to me that Congress would prefer an imaginative approach to class action litigation in § 4 cases.

I recognize that the argument based on the Antitrust Improvements Act can be turned against me by urging that the new § 301 cause of action occupies the field, that its explicit nonretroactivity shows an intention not to extend the utilization of Rule 23 in the antitrust field. The message one reads into the enactment probably will be essentially a product of his own conviction as to the social desirability of consumer class actions. It seems to me that they serve as a useful adjunct to public enforcement.42

Class action treatment in this case presents no insurmountable Seventh Amendment problem. As to its manageability, we should, I think, defer to the district court which will be faced with the problem if a verdict establishing a § 1 violation is returned. I would affirm the order allowing the case to proceed as a Rule 23(b)(3) class action, but with a direction that the issue of violation of § 1 of the Sherman Act, rather than the issue of § 4 liability to the entire class, be tried to a separate jury.