This opinion addresses challenges by the Plaintiffs’ Management Committee (“PMC”) and by certain plaintiffs represented by Victor Yannacone to Chief Judge Weinstein’s adoption of a plan for the distribution of the fund established as a result of the class settlement with the defendant chemical companies. See In re “Agent Orange” Product Liability Litigation, 611 F.Supp. 1396 (E.D.N.Y.1985) (“Distribution Opinion ”). Because no party to this litigation is adverse to the PMC, we requested that Special Master Kenneth Feinberg defend the district court’s distribution order essentially in the role of an amicus curiae. A detailed discussion of the development and selection of the distribution plan appears in the first of this series of opinions, 818 F.2d 145, familiarity with which is assumed.

Certain plaintiffs represented by Mr. Yannacone have also filed a petition for writ of mandamus or prohibition to have the PMC removed as class counsel. That issue is also addressed herein.

1. The Timeliness of the Pending Appeals

A party seeking to appeal a final decision of a district court in any case where, as here, the United States is a party must file a notice of appeal within 60 days after entry of the decision. Fed.R.App.P. 4(a)(1). The notice of appeal filed by Mr. Yannacone is concededly untimely. That appeal is therefore dismissed.

[*181] The Special Master argues that the PMC’s pending appeal is also untimely because it was noticed on August 19, 1986, more than 60 days after the distribution plan was adopted on May 28, 1985. However, important aspects of the distribution plan remained to be decided as of the earlier date, including, for example, the means of compensating veterans from Australia and New Zealand, 611 F.Supp. at 1443-45; the criteria for establishing a claimant’s exposure to Agent Orange, id. at 1417; and the entities that were to implement and administer the individual payment program, id. at 1427. Moreover, Chief Judge Weinstein apparently did not view the entire distribution plan as final until July 31, 1986, when he entered an order pursuant to Fed.R.Civ.P. 54(b) designed to “constitute a final judgment upon this Court’s Distribution Opinion of May 28, 1985.”

We do not believe that appellants were faced with the choice of appealing from the May 28 order or not at all. Whether that order was appealable is of great doubt. It was not a collateral order that “did not make any step toward final disposition of the merits of the case and will not be merged in final judgment,” Cohen v. Beneficial Industrial Loan Corp., 337 U.S. 541, 546, 69 S.Ct. 1221, 1225, 93 L.Ed. 1528 (1949). Unlike such a collateral order, the May 28 order could be effectively reviewed as part of the final judgment. Id. See also Coopers & Lybrand v. Livesay, 437 U.S. 463, 468, 98 S.Ct. 2454, 2457, 57 L.Ed.2d 351 (1978); Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 171-72, 94 S.Ct. 2140, 2149-50, 40 L.Ed.2d 732 (1974).

Even if the May 28 order was appealable under Cohen, there is still no reason to bar an appeal from the July 31 order, which was clearly intended by the district court to be final. See 15 C. Wright, A. Miller & E. Cooper, Federal Practice & Procedure § 3909, at 452 n. 38 (1976) (“There is often little reason to deny review on appeal from a clearly final judgment on the theory ... that an earlier order that did not terminate the entire proceeding was nonetheless so final as to have been appealable. Doctrines designed to facilitate intermediate appeals to avoid hardship often do not serve any corresponding interest in protecting opposing parties and the courts against delayed appeals.”). Dickinson v. Petroleum Conversion Corp., 338 U.S. 507, 70 S.Ct. 322, 94 L.Ed. 299 (1950), is a rare case in which the Supreme Court dismissed an appeal on the ground that it should have been filed prior to the entry of final judgment. The instant case is distinguishable from Dickinson in at least two respects, however. First, the order that would have been appealable in Dickinson dismissed all claims raised by the appellant. The Court thus noted that the appellant’s interests “could not possibly have been affected” by any action that remained to be taken by the district court. Id. at 515, 70 S.Ct. at 325. In contrast, the plaintiffs here continued to have an active interest in the litigation after the May 28 decision. Second, the Court recognized in Dickinson that the case had arisen before the adoption of Rule 54(b), a provision with the “obvious purpose” of “reducpng] as far as possible the uncertainty and the hazard assumed by a litigant who either does or does not appeal from a judgment of the character we have here.” Id. at 512, 70 S.Ct. at 324. The Court therefore expressly refused to “try to lay down rules to embrace any case but this.” Id.

Accordingly, we conclude that the PMC’s appeal from the district court’s distribution plan was timely filed. We therefore need not consider the PMC’s petition for a writ of mandamus, which raises the same issues.

2. General Principles

District courts enjoy “broad supervisory powers over the administration of class-action settlements to allocate the proceeds among the claiming class members ... equitably.” Beecher v. Able, 575 F.2d 1010, 1016 (2d Cir.1978). In reviewing allocations of class settlements, therefore, we will disturb the scheme adopted by the district court only upon a showing of an abuse of discretion.

In the present case, a relatively modest settlement fund must be allocated equitably among a large and diverse group of[*182] claimants. There are 240,000 claimants dispersed throughout the United States, Australia, and New Zealand. They suffer from an immense variety of ailments and have different medical and financial needs. Having pursued a number of often inconsistent goals in this litigation, they are as sharply divided over the distribution of the settlement fund as they are over its adequacy. The PMC seeks what it regards as a conventional scheme for “tort-based” recovery by individuals; Mr. Yannacone’s clients want the fund devoted largely to establishing a foundation; the district court adopted a compensation based scheme to distribute the bulk of the fund with the remainder to be used to establish a foundation. See P. Schuck, Agent Orange on Trial 211-13, 220 (1986).

The district court was not bound to choose among only those plans offered by class members who spoke out. Rather, it had to “exercise its independent judgment to protect the interests of class absentees, regardless of their apparent indifference,” In re Traffic Executive Association— Eastern Railroads, 627 F.2d 631, 634 (2d Cir.1980), as well as to protect the interests of more vocal members of the class. The district judge therefore had discretion to adopt whatever distribution plan he determined to be in the best interests of the class as a whole notwithstanding the objections of class counsel, see, e.g., Distribution Opinion, 611 F.Supp. at 1409 (criticizing distribution plan proposed by PMC on ground that “too great a share of the fund would go to lawyers and medical experts”); Plummer v. Chemical Bank, 668 F.2d 654, 659 (2d Cir.1982) (district courts cannot rely solely on “the arguments and recommendations of counsel” in evaluating propriety of class settlements), or of a large number of class members. See TBK Partners, Ltd. v. Western Union Corp., 675 F.2d 456, 462 (2d Cir.1982) (holding in shareholders’ derivative suit that even “majority opposition ... cannot serve as an automatic bar to a settlement that a district judge after weighing all the strengths and weaknesses of a case and the risks of litigation, determines to be manifestly reasonable”). See also Cotton v. Hinton, 559 F.2d 1326 (5th Cir.1977) (approving settlement over objections of counsel purporting to represent almost 50 percent of class); Bryan v. Pittsburgh Plate Glass Co., 494 F.2d 799 (3d Cir.) (approving settlement over objections of almost 20 percent of class), cert. denied, 419 U.S. 900, 95 S.Ct. 184, 42 L.Ed.2d 146 (1974).

3. Choice of Law

In adopting a distribution plan that departed from traditional tort principles by not requiring “a particularized showing of individual causation and injuries,” 611 F.Supp. at 1402, the district court held that such a plan would be consistent with “the consensus of state law,” id. at 1403, that figured in its certification of a class action. In re “Agent Orange” Product Liability Litigation, 100 F.R.D. 718 (E.D.N.Y.1983).

In the mandamus proceeding, we expressed “considerable skepticism” as to whether such a consensus would emerge among the states with respect to the legal rules applicable to the plaintiffs’ claims. In re Diamond Shamrock Chemicals Co., 725 F.2d 858, 861 (2d Cir.), cert. denied, 465 U.S. 1067, 104 S.Ct. 1417, 79 L.Ed.2d 743 (1984). In the first of this series of opinions we have stated that the district court’s conclusion as to the national consensus was to be praised more for its analysis than for its utility as a predictor of what various courts would do.

However, our disagreement with use of the national consensus in certifying a class does not foreclose its use as a method of establishing criteria for distributing a class settlement fund. As another Court of Appeals has observed in the class action context, “the allocation of an inadequate fund among competing complainants is a traditional equitable function, using ‘equity’ to denote not a particular type of remedy, procedure, or jurisdiction but a mode of judgment based on broad ethical principles rather than narrow rules.” Curtiss-Wright Corp. v. Helfand, 687 F.2d 171, 174 (7th Cir.1982) (citation omitted) (citing Zients v. La Morte, 459 F.2d 628, 630 (2d Cir.1972)). Use of a single national stan[*183] dard, regardless of what law various courts might have chosen in Agent Orange cases, is a permissible method of disbursing the fund. An individual claimant state-by-state approach would seriously deplete the portion of the fund going directly to veterans by diverting a substantial amount to lawyers and to the adjudicators necessary to implement the PMC’s complex scheme. The diversion might be so great as to reduce benefits for all claimants, including those who would be subject to the most favorable state laws. We thus agree with the approach of the district court on this question, although on a different rationale.

4. Payments for Death or Disability of Exposed Veterans

The PMC contends that the district court abused its discretion in compensating individual disabled veterans and families of deceased veterans without requiring “a particularized showing of individual causation and injuries.” 611 F.Supp. at 1402. The PMC argues that a portion of the settlement fund will thereby be distributed to undeserving claimants whose injuries were not caused by Agent Orange. Even if that outcome is the case, we do not believe that it is a grounds for altering the distribution scheme.

Chief Judge Weinstein did not deem necessary proof that a veteran’s death or disability resulted from exposure to Agent Orange ^[1] because he found the available evidence insufficient to establish which non-traumatic injuries could have been caused by Agent Orange and which could not. In other words, as between exposed veterans suffering from diseases for which the PMC would provide compensation and exposed veterans suffering from other non-traumatic diseases, the district court con-eluded that the former had no stronger claim for benefits than the latter because “causation cannot be shown for either individual claimants or individual diseases with any appropriate degree of probability.” 611 F.Supp. at 1409.

Chief Judge Weinstein did not abuse his discretion in adopting a distribution plan that reflected this conclusion. He was not obligated to adopt a plan that conformed to a theory of the relationship between Agent Orange and certain diseases that has little or no scientific basis. Further, he could take into account the very substantial countervailing evidence that Agent Orange was not harmful to any personnel in Vietnam. See In re “Agent Orange” Product Liability Litigation, 597 F.Supp. 740, 782-95 (E.D.N.Y.1984) (“Settlement Opinion ”) (reviewing scientific data on effects of Agent Orange and concluding that “all that can be said is that persuasive evidence of causality has not been produced”). He could also consider the substantial difficulty of proving that any particular plaintiff was injured by Agent Orange in making an equitable allocation of the limited settlement fund. See Curtiss-Wright Corp., 687 F.2d at 174-75 (equitable allocation of a class action settlement fund may be accomplished over party’s objection without “resolv[ing] trial-type issues of liability” based on district court’s independent “weighting of] the relative deservedness” of claimants). Moreover, he was correct in seeking a distribution scheme governed by criteria that are relatively easy and inexpensive to apply.

Furthermore, as became clear at oral argument, the PMC itself would no longer require proof that a veteran was actually[*184] exposed to Agent Orange in order to qualify a claimant for benefits under its distribution plan. Thus, servicepersons who spent their entire tour of duty far away from sprayed areas could receive payments under the PMC plan merely by developing any of the 24 medical conditions that the PMC claims are associated with Agent Orange. In contrast, the district court's plan would require some evidence of exposure. ^[2] Even if the district court’s distribution plan is overbroad with regard to ailments, that fact hardly renders it less desirable than the PMC’s plan, which is clearly overbroad with regard to exposure.

We further note that the distribution plan adopted by the district court does not entirely disregard traditional tort principles of causation. For example, it provides payments only to veterans who have become disabled from non-traumatic, non-accidental, non-self-inflicted causes and to the survivors of veterans who have died from such causes. Consequently, a veteran who died or became disabled as a result of an auto collision, a gunshot wound, or a narcotic overdose, all causes clearly unrelated to Agent Orange exposure, would have no claim to payments from the settlement fund.

In sum, given the inconclusive state of the scientific evidence as to what injuries, if any, were caused by Agent Orange, the district court did not abuse its discretion in holding that all exposed veterans who have suffered nontraumatic death or disability have stated “colorable legal claims against defendants ... [sufficient] to allow them to share in the settlement fund.” In re Chicken Antitrust Litigation American Poultry, 669 F.2d 228, 238 (5th Cir.1982), quoted in Distribution Opinion, 611 F.Supp. at 1411.

We emphasize that the district court is free to alter the distribution plan in the future to simplify it even more or to clarify standards as concrete issues arise. We also ask the district court to review its procedures for establishing exposure to Agent Orange in light of Attachments 2 and 3 to the PMC’s reply brief and recent news reports concerning the possible discovery of a biological “fingerprint” left in veterans’ blood by dioxin. See Researchers Report Finding Telltale Sign of Agent Orange, N.Y. Times, Sept. 18, 1986, § A at 28, col. 3 (late city final ed.).

5. Class Assistance Programs

We turn now to the district court’s proposal to establish “a class assistance foundation ... to fund projects and services that will benefit the entire class.” 611 F.Supp. at 1432. The PMC contends that use of the settlement fund for class assistance programs would contravene the decisions of this court in Eisen v. Carlisle & Jacquelin, 479 F.2d 1005 (2d Cir.1973), vacated and remanded on other grounds, 417 U.S. 156, 94 S.Ct. 2140, 40 L.Ed.2d 732 (1974) (remedy proposed before finding of liability in order to make class manageable; rejected because it benefitted future odd-lot investors rather than past investors who had suffered loss), and Van Gemert v. Boeing Co., 553 F.2d 812 (2d Cir.1977) (rejecting proposal that would have permitted unclaimed portion of damage award to be paid to class members who had already been made whole).

We do not believe that the district court was necessarily foreclosed by Eisen and Van Gemert from using a portion of the settlement fund to provide programs for the class as a whole. The instant case is, of course, distinguishable from Eisen and Van Gemert in several important respects.

[*185] First, the class that will benefit from the district court’s distribution plan is essentially equivalent to the class that claims injury from Agent Orange. That was not the case in either Eisen or Van Gemert. In Eisen, the proposed recovery scheme would primarily have benefitted not the class of persons who claimed injury from prior odd-lot transactions but instead a class of persons who would engage in such transactions in the future. In Van Gemert, the proposal at issue would have distributed the unclaimed portion of a damage award to class members who had already recovered their losses in full, a group the court characterized as a “next best class.” 553 F.2d at 815. Hence, the distribution plan adopted by Chief Judge Weinstein simply lacks the sort of “fluidity” between the class claiming injury and the class receiving recovery that existed in Eisen and Van Gemert.

Second, we were particularly concerned in Eisen that the availability of “fluid class recovery” would have allowed plaintiffs to satisfy the manageability requirements of Rule 23 where they otherwise could not. The damages to the average class member in Eisen were estimated at no more than $3.90, see 479 F.2d at 1010, and, as counsel for the named plaintiff conceded, “[i]f each [member] had to present his own personal claim for damages, the class, indeed, would not be manageable.” Id. at 1017. We foresaw that such an unwarranted relaxation of the manageability requirements would have induced plaintiffs to pursue “doubtful” class claims for “astronomical amounts” and thereby “generate ... leverage and pressure on defendants to settle.” Id. at 1019. However, the instant case, unlike Eisen, was maintainable as a class action regardless of the form of recovery available to the plaintiff class. Accordingly, our concern in Eisen that the availability of a particular form of recovery would vastly enlarge the number of class actions in the federal courts is not present in the instant case.

Finally, the instant case, unlike Eisen and Van Gemert, arises out of a pretrial settlement. As the Supreme Court has recognized, a district court may “provide[] broader relief [in an action that is resolved before trial] than the court could have awarded after a trial.” Local Number 93, International Association of Firefighters v. City of Cleveland, — U.S. -, 106 S.Ct. 3063, 3077, 92 L.Ed.2d 405 (1986). Indeed, we have previously recognized that some “fluidity” is permissible in the distribution of settlement proceeds. See Beecher v. Able, 575 F.2d at 1016 n. 3; West Virginia v. Chas. Pfizer & Co., Inc., 314 F.Supp. 710, 728 (S.D.N.Y.1970), aff'd, 440 F.2d 1079 (2d Cir.), cert. denied, 404 U.S. 871, 92 S.Ct. 81, 30 L.Ed.2d 115 (1971).

We thus conclude that a district court may, in order to maximize “the beneficial impact of the settlement fund on the needs of the class,” 611 F.Supp. at 1431, set aside a portion of the settlement proceeds for programs designed to assist the class. However, we believe that the district court must in such circumstances designate and supervise, perhaps through a special master, the specific programs that will consume the settlement proceeds. The district court failed to do so in the instant case. Instead, it provided that the board of directors of a class assistance foundation would control, inter alia, “investment and budget decisions, specific funding priorities, ... [and] the actual grant awards,” id. at 1435, and that the court would retain only “[a] comparatively modest supervisory role” in such decisionmaking. Id. at 1436.

We are unwilling for several reasons to permit the distribution of any settlement proceeds to a largely independent foundation. First, while a district court is permitted broad supervisory authority over the distribution of a class settlement, see Beecher v. Able, 575 F.2d at 1016, there is no principle of law authorizing such a broad delegation of judicial authority to private parties. We perceive no assurance that the “self-governing and self-perpetuating” board of directors of the class assistance foundation, or any other such body that might be devised by the court, will possess the independent, disinterested judgment required to allocate limited funds to benefit the class as a whole. One of the district court’s prime functions in distribut[*186] ing such a fund is to protect the less vocal and less activist members of the class. The proposed foundation is not well designed to perform that function. Moreover, given the very evident discord among various veterans as to the use of the settlement fund, we see great hazards in transferring that discord to a foundation having permanent control over portions of that fund. There is a great danger that the fund would be expended in ways that generate more controversy than benefits and would create even more frustration among a group already frustrated enough by perceived political and legal setbacks. However unique it may be, this is an action for personal injuries, and we believe that only direct judicial supervision can assure that the settlement fund is expended for appropriate purposes.

We acknowledge the strong sentiment among some veterans for the creation of such a foundation. We also note, however, their great expectations for the foundation are similar to the expectations that prompted this class action litigation. Those latter expectations were frustrated when confronted with the reality of legal proceedings. Great expectations underlying the foundation proposal still exist because the concrete tasks to be undertaken by it remain unclear, and the reality of hard and controversial choices concerning use of the fund has not yet been confronted.

Moreover, we are concerned that the broad mandate given the class assistance foundation, which must remain an arm of the court however loosely connected, would permit settlement proceeds to be expended on activities inconsistent with the judicial function. For example, activities to “help class member veterans better obtain and utilize VA services” and to “increase public awareness of the problems of the class,” id. at 1440, might include political advocacy. We do not believe that the proceeds of a court-administered settlement ought to be used for such a purpose.

Finally, we are concerned that, even given the expressed intention to allow the foundation great latitude, the district court and this court would repeatedly be asked to intervene in foundation decisions alleged not to benefit the class. When such claims are made, they call for greater scrutiny than is contemplated by the district court’s exercise of only a “modest supervisory role.” In addition, endless legal argument over the disbursement of the settlement fund would simply prolong the suffering and frustrations of the class.

We explicitly note, however, that the district court may in the exercise of its discretion and after consultation with veterans’ groups undertake to use portions of the fund for class assistance programs that are consistent with the nature of the underlying action and with the judicial function. Accordingly, the district court on remand may designate in detail such programs and provide for their supervision. A reserve fund for as yet undefined programs may be established. Alternatively, the court may reallocate any or all of the funds earmarked for the class assistance foundation to augment the awards to individual class members. The court may choose either to increase the awards to disabled veterans and the survivors of deceased veterans or to provide awards to other class members who have suffered less than total disability-

6. Yannacone Petition for Writ ofMandamus/Prohibition

The petition for a writ of mandamus or prohibition filed by Mr. Yannacone seeks the removal of the PMC as lead counsel. Mr. Yannacone contends that a “conflict of interest” exists between the PMC and the plaintiff class, as evidenced by the differences between the distribution plan submitted by the PMC and the plan submitted by Mr. Yannacone. He also argues that the plaintiffs are entitled to “a reasonable opportunity to be heard through counsel of their own choosing who can and will speak independently on their behalf.” The petition is frivolous.

We note that Mr. Yannacone was among the attorneys who first sought class certification and that he served for some time as the lead counsel for the class. Nevertheless, his present petition reveals a funda[*187] mental misunderstanding of the nature of a class action. A plaintiff who joins in a class action, as many plaintiffs did through Mr. Yannacone, gives up his or her right to control the litigation in return for the economies of scale available under Fed.R.Civ.P. 23. In the related context of a shareholders’ derivative suit, we have rejected any notion that “each individual plaintiff and lawyer must be permitted to do what he pleases in litigation as complex as this, and can behave in total disregard of the interest of other litigants and of the class.” Farber v. Riker-Maxson Corp., 442 F.2d 457, 459 (2d Cir.1971) (per curiam).

The selection of lead counsel for the plaintiff class is left to the discretion of the district court “guided by the best interests of [the class], not the entrepreneurial initiative of the named plaintiffs’ counsel.” Cullen v. New York State Civil Service Commission, 566 F.2d 846, 849 (2d Cir.1977). “Unless there are exceptional circumstances, ... the exercise of discretion should be left untouched by the appellate court.” Id. See also Weight Watchers of Philadelphia, Inc. v. Weight Watchers International, Inc., 455 F.2d 770, 775 (2d Cir.1972) (“ ‘we do not—indeed may not—issue mandamus with respect to orders resting in the district court’s discretion, save in most extraordinary circumstances’ ”) (quoting Donlon Industries, Inc. v. Forte, 402 F.2d 935, 937 (2d Cir.1968)).

Mr. Yannacone has failed even to suggest, much less establish, any “exceptional circumstances” that might warrant removal of the PMC as lead counsel. Indeed, he has suggested nothing more than a difference of opinion between the PMC and himself with respect to the appropriate distribution of the settlement fund. Moreover, these differences were fully aired before the district court, which thoroughly evaluated the merits of each plan in the course of its distribution opinion. See 611 F.Supp. at 1403-10.

Finally, even if we were to order the removal of the PMC as lead counsel, we have no reason whatsoever to expect the district court to appoint Mr. Yannacone to take its place. We have even less than no reason to expect the district court to abandon its own distribution plan in favor of the plan proposed by Mr. Yannacone. Accordingly, the petition is denied.

Affirmed in part, reversed in part, and remanded for further proceedings in accordance with this opinion.