OPINION OF THE COURT
ALDISERT, Circuit Judge.
The question presented is whether the district court properly applied the guidelines for awarding attorneys’ fees we set forth when this matter was previously before us. 487 F.2d 161 (3d Cir. 1973) (Lindy I). We affirm in part, but vacate the judgment and remand for entry of an order in accordance with this opinion.
This protracted litigation emanates from the plumbing fixtures cases.1 In connection with final court approval of the settlement of claims by the national class2 of builder owners, the district court awarded attorneys’ fees to Harold E. Kohn and David Berger, and their respective firms.3 Those participating in the settlement fund may be categorized into three groups: appellees Kohn and Berger represented the first category; other lawyers, not implicated in this appeal, represented those in the second category; the third consisted of class members not represented by counsel until after court approval of the settlement. Two members of this third category— Friendswood Development Company and [108]*108Humble Oil & Refining Company — objected to the fee award to Kohn and Berger, and appealed. The award was vacated and the cause remanded, fees were awarded again, and Friendswood and Humble have appealed again. Although the three categories were not formally denominated as subclasses, see F.R.Civ.P. 23(c)(4), appellants advance arguments which obtain unerringly for all members of the third category. Accordingly, we treat the contentions as applicable to all members of the third category.4
In its opinion on remand, the district court determined that appellees were entitled to attorneys’ fees of $1,134,765.45 from the settlement fund then valued at $29.3 million. Of this amount, the district court ordered members in the third category — the previously unrepresented claimants — to pay $925,968.61. Thus, the district court ordered members of the third category to pay 81.6 per cent of the attorneys’ fee from the settlement fund, even though their aliquot share of the fund was only about $8,145,400 or 27.8 per cent. The district court ordered no further payment of attorneys’ fees from the settlement fund, but appellees were to receive $861,191 under private contracts with their clients, the members of the first claimant category.
Any understanding of the specific issues in this appeal must start with the rules we enunciated when these proceedings were here before. The district court has properly summarized what we said:
In discussing the proper standards which would govern the award of fees in a case of this sort,. . . . the first inquiry of the Court should be into the hours spent by the attorneys, including how many hours were spent in what manner by which attorneys. 487 F.2d at 167. After determining the time spent, the district court should then undertake to fix an hourly rate of compensation to be applied to the hours worked. While the amount thus found to constitute reasonable compensation should be the “lodestar” of the Court’s fee determination, at least two other factors should be taken into account in computing the value of attorneys’ services, namely the contingent nature of success and the extent, if any, to which the quality of an attorney’s work mandates either increasing or decreasing the amount to which the Court has found the attorney reasonably entitled. 487 F.2d at 168. Finally, after determining the total reasonable value of an attorney’s services in securing recovery of a fund for the class, the Court should determine what portion of that amount should be paid by the unrepresented claimants. [Absent extraordinary circumstances, the unrepresented claimants should pay for the attorneys’ services 'in proportion to their benefit from them — that is, the unrepresented claimants should pay a percentage of the reasonable value of the attorneys’ services to the class equal to their percentage of the class’ recovery.] 487 F.2d at 169.
382 F.Supp. at 1003.
The appellants press four contentions: (1) The district court should have made no award for services performed by the Berger firm; (2) The district court should have excluded certain specific services performed by appellees from the “lodestar” defined in Lindy I; (3) The district court should not have doubled the “lodestar” to reflect (a) contingency and (b) quality factors; and (4) The district court required category three claimants to pay a disproportionate share of the attorneys’ fees from the settlement fund.
I.
Appellants’ contention that the Berger firm should not participate at all in the attorneys’ fee awarded from the fund rests on two premises: (a) that Berger and his firm contributed nothing to the creation of the settlement fund, and (b) that, in any event, Berger and his firm will be adequately compensated for their services in connection with this litigation.
In support, appellants marshal several subsidiary points. First, Berger claimed [109]*109only one hour’s time related to settlement work. See ibid, at 1007. Second, although Berger signed the settlement agreement with the defendants, he only did so because Kohn asked that Berger’s name be included and no defendant objected. Brief for Appellants at 33-34. Third, the district court concluded that the reasonable value of all work performed by the Berger firm was $139,186.85. However, as a result of a 50-50 fee splitting agreement between Kohn and Berger, the Berger firm stood to gross about $893,500 from the litigation. See 382 F.Supp. at 1028, n.34. Finally, the Berger firm did not support its application with accurate, contemporaneously-made time records. Instead, it relied, on a “reconstruction” of the hours expended.
The district court was confronted with, and rejected, the identical arguments. We agree with its disposition. Ibid, at 1009-11. In summary, the district court found that the time expended by Berger and his firm “contributed materially to the creation of the settlement fund”; that Berger consulted with Kohn “at every stage of the proceedings”; that Kohn valued Berger’s judgment and advice; and that Berger’s firm reviewed briefs and pleadings for the Kohn firm during the early stages of the litigation. Upon a review of the record, we conclude that these findings of the district court were not clearly erroneous. Krasnov v. Dinan, 465 F.2d 1298, 1302 (3d Cir. 1972). With respect to the “reconstruction” of time expended, the district court stated:
. [T]ime records, although highly desirable, are not the only means of proving time spent in a multidistrict litigation of this sort. . . . Although mere estimates of time are not acceptable, an allowance of attorneys’ fees may be based on a reconstruction, provided that the time records are substantially reconstructed and are reasonably accurate.
The evidence here discloses that the reconstruction was carefully and accurately done. . . . Indeed, from the record as a whole, it appears that petitioner Berger and his firm probably spent more time than they are claiming in connection with this litigation.
382 F.Supp. at 1011. Finding the basic facts not clearly erroneous, and determining that there is no error either in the court’s reasoning or in its application of the proper legal precept, we affirm the district court on this point. Implicit in this conclusion is the understanding that the agreement between Kohn and Berger to split fees is simply irrelevant to the considerations mandated by Lindy I.
Insofar as appellants urge that the Berger firm not receive an attorney’s fee from the fund because Berger was not counsel for a court-appointed Class Representative, we reject the argument. We do not read the appellees’ original petition as being on behalf of Kohn and Berger qua counsel for class representatives; rather, they petitioned “jointly as counsel for the plaintiffs and intervenor plaintiffs [where] the plaintiffs were designated by the Court as class representatives . . .” App. at 77. In addition, we are faced with the district court’s findings of fact — that the Berger firm benefited the fund — which we will not upset.
II.
The district court excluded from its “lodestar” determination the time spent by appellees in negotiating fee arrangements with, and in preparing claim forms for, privately retained claimants. Although Kohn and Berger filed a cross appeal, they do not contest the propriety of this ruling. In any event, we would agree with the district court.
Appellants contend that the court also should have excluded from the “lodestar” time expended (A) relating to appellees’ application for attorneys’ fees and (B) on interventions. The district court fixed the reasonable value of appellees’ services relating to the fee application at $29,806.25. In addition, the district court made an award in the amount of $40,654.00 for “time spent since March 20,1974”, all of which was time spent in connection with the fee [110]*110application.5 Finally, with respect to fee application work, the district court awarded $320 to appellees for the services of paraprofessionals and law students. The reasonable value of work on interventions was $13,085, which figure the district court ultimately doubled, so as to award appellees $26,170 for interventions. 382 F.Supp. at 1024. We now address these contentions.
A. Fee Application and Appeal
The propriety vel non of an award from the fund for services relating to the fee application itself6 turns on considerations of the basis for, and the nature of, the award from the settlement fund.
Federal courts have long recognized “the historic power of equity to permit the trustee of a fund or property, or a party preserving or recovering a fund for the benefit of others in addition to himself, to recover his costs, including his attorneys’ fees, from the fund or property itself or directly from the other parties enjoying the benefit.” Alyeska Pipeline Service Co. v. Wilderness Society, 421 U.S. 240, 257, 95 S.Ct. 1612,1621, 44 L.Ed.2d 141 (1975); see, e. g., Sprague v. Ticonic National Bank, 307 U.S. 161, 59 S.Ct. 777, 83 L.Ed. 1184 (1939); Central Railroad & Banking Co. v. Pettus, 113 U.S. 116, 5 S.Ct. 387, 28 L.Ed. 915 (1885); Trustees v. Greenough, 105 U.S. 527, 26 L.Ed. 1157 (1881). See generally 7A C. Wright & A. Miller, Federal Practice and Procedure § 1803 (1972); Dawson, Lawyers and Involuntary Clients: Attorney Fees from Funds, 87 Harv.L.Rev. 1597 (1974) [hereinafter cited as Dawson]. As we said in Lindy I, “[t]he award of fees under the equitable fund doctrine is analogous to an action in quantum meruit: the individual seeking compensation has, by his actions, benefited another and seeks payment for the value of the service performed.” 487 F.2d at 165. Accordingly, “a benefit to the fund is supposedly required. . . . The standard formula [of benefit] mix[es] together three distinct ideas: that a fund can be benefited by being ‘created, increased or protected’ (or ‘preserved’).” Dawson at 1626. Generally, where litigation involves the competing interests of claimants to a common fund, no attorneys’ fees should be awarded. The reason is that, “if the interests are in conflict, success for one side means no benefit for the other and for a charge against a fund a benefit is required.” Ibid, at 1638 (footnote omitted).
We subscribed to these general views in Lindy I, where we said:
[T]here are two possible “causes of action” that may be urged as the basis for award of attorneys’ fees. One of these “causes” belongs to the plaintiff who brought the underlying suit. His claim is that by instituting the suit he has performed a service benefiting other class members. .
The second “cause of action” for award of attorneys’ fees under the equitable fund doctrine belongs to the attorney. The attorney’s claim is that his conduct of the suit conferred a benefit on all the class members, that one or more class members has agreed by contract to pay for the benefit the attorney conferred upon him, and that the remaining class members should pay what the court determines to be the reasonable value of the services benefiting them.
487 F.2d at 165;7 see SEC v. Aberdeen Securities Co., 526 F.2d 603, 607 (3d Cir. 1975); cf. Merola v. Atlantic Richfield Co., 515 F.2d 165 (3d Cir. 1975) (Merola II) (discussing non-pecuniary benefit as factor affecting “lodestar” award).
[111]*111The district court concluded that time spent in pursuing the fee application was “a part of the administrative expense necessarily incurred in administering the fund. . . . Services performed in this connection are indeed of benefit to the class as a whole because they enable the Court to make its determination on the basis of a full and accurate presentation of all relevant facts.” 382 F.Supp. at 1012.
Services performed in connection with the fee application are necessary to the attorney’s recovery. They benefit him, for without them, the attorney cannot, since Lindy I, recover. But such services do not benefit the fund — they do not create, increase, protect or preserve it. Accordingly and in the circumstances of this case, we accept the prevailing rule for litigation involving the competing interests of claimants to a common fund. See page 110, supra. There being no benefit to the fund from services perfdnHed~~by appellees in connection With their fee application, there should be no attorneysilee-award from the fund for those sqrv-iees: The district court, therefore, erred in awarding $29,806.25 for the fee application work prior to March 20, 1974, in awarding $40,654 for work after March 20, 1974, and in awarding $320 for fee application work by paraprofessionals and law students.
Having made a determination of the applicable rule of law for the case before us, it is important to emphasize that which is not before us. We do not decide today what the rule should be where the fee application is pressed by the client himself, and competing interests among claimants surface. Cf. Dawson 1634 n.126. Nor do we decide the rule for the situation where a competing claimant to a common fund persuades the court in the first instance to set an attorney’s fee inadequate under Lindy I, thereby necessitating an appeal.8 We specifically reserve these issues.
B. Intervention.
Appellants challenge the district court’s award for time expended on interventions contending that this work in no way benefited the class. They argue that there was no reason to intervene in this multidistrict class action because class members dissatisfied with the representation of the class might opt out and thereby avoid the binding effect of a 23(b)(3) action. F.R.Civ.P. 23(c)(2).
The district court considered and rejected appellants’ argument. It found that time spent on interventions “was beneficial to the class as a whole”, and continued:
Faced with many additional litigants with sufficient financial resources to finance the cost of notice to the class and also the cost of a protracted trial, defendants in this case were under much more pressure to settle than otherwise. In its recent decision in Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 94 S.Ct. 2140, 40 L.Ed.2d 732 (1974), the Supreme Court held that class representatives must pay the cost of the notice required by Rule 23 in a case such as this one. The cost of notice in this case was approximately $150,000, a sum which was paid from the settlement fund. Had there been no settlement, plaintiffs and intervenors would have been required to advance this sum if this litigation were to go ahead as a national class action.
382 F.Supp. at 1013; see ibid, at 1010.
We agree with appellants that, since the 1966 amendments to Rule 23, there is no necessity for intervention in a 23(b)(3) class action. Indeed, Berger admitted as much during his deposition. But this truism cannot end the analysis. While intervention may no longer be necessary, it is certainly still permissible. The federal rules allow intervention “of right” when an applicant’s interest is not “adequately represented by existing parties.” F.R.Civ.P. 24(a). Moreover, Rule 23 itself states that the district [112]*112court may make appropriate orders, inter alia, “requiring, for the protection of the members of the class or otherwise for the fair conduct of the action, that notice be given ... of the opportunity of members to signify whether they consider the representation fair and adequate, to intervene and present claims or defenses, or otherwise to come into the action.” F.R. Civ.P. 23(d)(2) (emphasis added).
As rehearsed previously, the propriety of a particular award for an attorney’s services from a common fund depends on whether the specific services benefited the fund — whether they tended to create, increase, protect or preserve the fund. Appellants argue that, because intervenors did not advance the costs of notice, their presence in the case did not benefit the fund. This argument misses the mark. The point is, and the district court so found, that, when confronted by the intervenors and the financial strength they added to the plaintiff class, “defendants . . . were under much more pressure to settle.” This finding was not clearly erroneous. Krasnov v. Dinan, supra. We must conclude, therefore, that the services performed on intervention helped to create the fund — they helped to force the settlement.
In the instant case, there was never a formal class certification. Brief for Appellants at 7; see n.2 supra. Thus, the interests of those that intervened were not formally represented by the named plaintiffs who sought to be class representatives. In these circumstances, it may not be said that the interventions served only to clutter the action. Accordingly, the district court did not err when it awarded Kohn and Berger fees for services performed on interventions.
We express no opinion on whether attorneys’ fees for services on interventions should be awarded from an equitable fund where the interventions occur after a class certification. In such circumstances, it might be argued that intervention serves no essential purpose, for class certification requires a finding of adequate representation. See F.R.Civ.P. 23(a)(4), 23(c)(2)(C), 23(d)(2), 24; 7A C. Wright & A. Miller, Federal Practice and Procedure § 1799 (1972).
III.
Lindy I instructs that, in addition to the “lodestar”, the court’s computation of the value of attorneys’ services should reflect two other factors — the contingent nature of success and the quality of the attorney’s work. 487 F.2d at 168. In Merola II we observed that the second factor is “evidenced by the work observed, the complexity of the issues and the recovery obtained. In settled cases, the second additional factor [quality] is reflected largely in the benefit produced. It permits the court to recognize and reward achievements of a particularly resourceful attorney who secures a substantial benefit for his clients with a minimum of time invested, or to reduce the objectively determined fee where the benefit produced does not warrant awarding the full value of the time expended.” 515 F.2d at 168-69.
Although the district court did not have the benefit of Merola II, its analysis generally followed these lines. With certain exceptions 9 that need not detain us, the district court doubled the amounts deemed reasonable compensation on the basis of hourly rates “because of the contingency and quality factors”. 382 F.Supp. at 1024. In reaching this determination, the district court relied on the complexity of the litigation, the skill and experience of appellees Kohn and Berger, and the $29 million recovery.
Appellants argue that Lindy I resolved the contingency issue adversely to the attorneys.[113]*11310 They also question the quality of the work performed. On these hases, they urge that the district,court erred in increasing, by any amount, the sum deemed a reasonable hourly compensation for attorneys’ services.
A. Contingency
When appellants argue that Lindy I forecloses the contingency issue they not only misread the language of Lindy I upon which they rely, but they also belittle the district court’s painstaking comparison between the limited effect of the criminal prosecutions and the breadth- of the civil litigation.
Chief Judge Seitz said in Lindy I:. “The court may find that the contingency was so slight . . . that an increased allowance for the contingent nature of the fee would be minimal.” 487 F.2d at 168 (emphasis added). The' language was permissive, not mandatory.
The district court specifically found the contingency factor was not slight, but rather justified “a substantial increased allowance”. 382 F.Supp. at 1017. Success of the civil suits, the district court stated, depended on several contingencies:
(1) the uncertainty of the class members’ ability to prove liability and damages;
(2) the existence of other categories of claimants in this litigation contending that they and not the builder-owners were entitled to any recovery against the defendants;
(3) this Court’s prior decision in Maricopa County v. American Radiator & Standard Sanitary Corp., 323 F.Supp. 381 (E.D.Pa.1970), and Mangano v. American Radiator & Standard Sanitary Corp., 50 F.R.D. 13 (E.D.Pa.1970), aff’d, 438 F.2d 1187 (3d Cir. 1971);
(4) the fact that the time period and range of products encompassed within the government’s evidence in the criminal case was drastically more limited than the allegations in the complaints herein; and
(5) the settling defendants’ denials of any violation of the antitrust laws and any overcharges, and the fact that they could have been expected to litigate all issues fully, including appeals if necessary.
Ibid, at 1015. Thus, assuming plaintiffs could prove that defendants had violated the- antitrust laws, plaintiffs would have had to prove they, as distinguished from an alternate potential class of claimants in the distribution scheme, were entitled to collect damages. Moreover, at the outset of the civil litigation, defendants challenged the propriety of a national class action on the ground that it was unmanageable. Perhaps most important, however, and as more specifically set forth by the district court, ibid. at 1015-16, only three of the 16 defendants were convicted after original not guilty [114]*114pleas. The 13 other defendants pleaded nolo contendere, thus imposing on plaintiffs the evidentiary burdens to which Chief Judge Seitz alluded in his footnote 12 to Lindy I, supra at n.10. This analysis led the district court to conclude:
Thus, this was not a ease where plaintiffs received from the government on a silver platter the evidence necessary to prove their claims against the three convicted defendants, much less against the thirteen which had entered pleas of nolo contendere. Substantial problems of proof were present as to both liability and damages. Yet petitioners were able to negotiate settlements with all defendants which yielded funds producing $10.75 for each bathroom unit purchased by claimants during the full four-year period.
Ibid, at 1016.
B. Quality
Appellants contend further: that “[t]he bulk of the time claimed in this ease was on routine matters”, that this “was not a complex antitrust case involving novel substantive issues”, and that “the result achieved was not outstanding”. Brief for Appellants at 45 — 46.
The answer to the first argument is that Lindy I does not dictate consideration of the proportion of time spent on routine matters in assaying the quality factor. The “lodestar” concept reflects such considerations, encompassing the amount of time spent on various projects, the status of individual attorneys, and the reasonable hourly rate for each person for each service performed. See Lindy I, 487 F.2d at 167.
We might agree with appellants that the basic issue in the case — a price-fixing conspiracy — did not involve novel substantive issues. But that agreement would not detract from the complexity of this massive multidistrict litigation, nor from the novelty of the myriad attendant subsidiary issues. On this score, the district court stated:
Petitioners were here participating in no run-of-the-mill class suit. The plumbing fixture antitrust litigation has involved the largest number of cases of any matter that has ever come before the Judicial Panel on Multidistrict Litigation. 374 cases are included in this one massive litigation. Over 10,000 claims were filed in the builder-owner settlement alone.
382 F.Supp. at 1019 (footnote omitted). We do not consider this finding of complexity clearly erroneous, but, as hereinafter set forth, we question whether this factor should be considered under the rubric of quality of professional performance.
The court also detailed the skill and experience of appellees Kohn and Berger, and the members of their firms, to support its finding of quality service. The court concluded:
Petitioners Kohn and Berger and various members of their firms have filed many briefs and pleadings which have been reviewed by the undersigned judge and have personally appeared before the undersigned on many occasions since June of 1970. The record here also includes various pleadings, briefs and memorandum filed before that date. From a review of these written materials and from personal observations, this Court concludes that petitioners have exhibited an unusual degree of skill in conducting these proceedings. They were aggressive and imaginative throughout. Because the work has been efficient and of an atypical quality, the aggregate hourly compensation should be increased. Indeed, the total amount of time spent by petitioners in this case is relatively low when one considers what was accomplished and the number of years that have elapsed since the litigation was commenced. Certainly, economy of effort should not be penalized. Less experienced and less skillful attorneys would undoubtedly have expanded much more time in achieving the same result than did petitioners.
Ibid, at 1021-22.
Finally, appellants deprecate the $29.3 million settlement as being “merely an ‘acceptable’ ” one, Brief for Appellants [115]*115at 46, now representing about $10.75 per bathroom unit sold during the term of the price-fixing conspiracy. In answer to this argument, the district court observed that the per unit settlement figure was “two to three times the actual price increase” resulting from the conspiracy, 382 F.Supp. at 1022, and that the builder-owner settlement was “a much larger per unit figure than” the contractor or wholesaler settlements, ibid. The court concluded, “the original settlement was a most favorable one [as] indicated by the fact that only one attorney opposed it” when presented for approval. Ibid. We quite' agree.
We must now determine whether the district court committed error by using these considerations to increase its “lodestar” award for several categories by 100 per cent.
C. Doubling
To evaluate the district court’s doubling of the “lodestar” we first must recognize distinctions in the juridical functions of trial and reviewing courts, and the concomitant differences in their respective responsibilities. As we have often said, the award of a reasonable attorney’s fee is within the district court’s discretion.11 Merola v. Atlantic Richfield Co., 493 F.2d 292, 295 (3d Cir. 1974) (Merola I); Lindy I, supra, 487 F.2d at 166; Tranberg v. Tranberg, 456 F.2d 173, 175 (3d Cir. 1972). Unhappily, this formulation does not advance the analysis; it merely restates the problem, for one must understand what one means by “discretion”.
Ordinarily in the law, as the Ninth Circuit observed several years ago, “discretion”
is defined as: “The power exercised by courts to determine questions to which no strict rule of law is applicable but which, from their nature, and the circumstances of the case, are controlled by the personal judgment of the court.” 1 Bouv. Law Diet., Rawles’ Third Revision, p. 884. Judicial action — discretionary in that sense — is said to be final and cannot be set aside on appeal except when there is an abuse of discretion. A common example is a court’s ruling on the extent of cross-examination. Alford v. United States, 282 U.S. 687, 694, 51 S.Ct. 218, 75 L.Ed. 624. Discretion, in this sense, is abused when the judicial action is arbitrary, fanciful or unreasonable, which is another way of saying that discretion is abused only where no reasonable man would take the view adopted by the trial court. If reasonable men could differ as to the propriety of the action taken by the trial court, then it cannot be said that the trial court abused its discretion.
Delno v. Market St. Ry., 124 F.2d 965, 967 (9th Cir. 1942); see Napolitano v. Campania Sud Americana de Vapores, 421 F.2d 382, 384 (2d Cir. 1970). Thus, one may attack an exercise of discretion as “irrational”. One seeking to establish such an abuse of discre[116]*116tion, however, assumes a heavy burden. Almquist v. Almquist, 214 Kan. 788, 791, 522 P.2d 383, 386 (1974). Lindy I and Merola I both emphasize that the appellate court may also find an abuse, or more accurately a “misuse”, of discretion where the trial court utilizes improper standards or procedures in determining fees. Lindy I, supra, 487 F.2d at 166; Merola I, supra, 493 F.2d at 295. Similarly, a clearly erroneous finding of fact may require reversal. Ibid.
Stated negatively, the appellate court may not upset a trial court’s exercise of discretion on the basis of a visceral disagreement with the lower court’s decision. Similarly, the appellate court may not reverse where the trial court employs correct standards and procedures, and makes findings of fact not clearly erroneous. In sum, “[i]f the district court has applied the correct criteria to the facts of the case, then, it is fair to say that we will defer to its exercise of - discretion.” Katz v. Carte Blanche Corp., 496 F.2d 747, 756 (3d Cir.) (in banc), cert. denied, 419 U.S. 885, 95 S.Ct. 152, 42 L.Ed.2d 125 (1974). But if the trial court has not properly identified and applied the criteria, the court’s determination will not be entitled to such deference.
As previously rehearsed, the district court went to great lengths to articulate clearly the reasons for its decision. Cf. Greater Boston Television Corp. v. FCC, 143 U.S.App.D.C. 383, 444 F.2d 841, 851 (D.C. Cir. 1970) (Leventhal, J.), cert. denied, 403 U.S. 923, 91 S.Ct. 2229, 29 L.Ed.2d 701 (1971). The court found that the contingent nature of success alone justified “a substantial increased allowance”. 382 F.Supp. at 1017. Inasmuch as we understand appellants to challenge the amount by which the district court increased the “lodestar” award, we decline the invitation to upset the exercise of discretionary authority. Had any one of us been the trial judge acting on this fee application, we might well have acted differently; we might have increased the basic award by 20, 50, or 75 per cent. But as appellate judges, we do not find the district court’s action “so unreasonable or so arbitrary as to amount to a prejudicial abuse of . discretion.” Napolitano v. Compania Sud Americana de Vapores, supra, 421 F.2d at 384; see Rosenberg, Judicial Discretion of the Trial Court, Viewed From Above, 22 Syracuse L.Rev. 635, 662 (1971).
After thoroughly reviewing appellants’ contentions, we have concluded that the district court followed the guidelines stated in Lindy I; accordingly, we cannot say that the district court abused its discretion in doubling the amount deemed a reasonable “lodestar” award for most categories of services performed.
The district court’s struggle with this case on remand and the difficulties we encountered on review, however, persuade us that additional direction is necessary so that the district courts within this circuit may properly implement the teachings of Lindy I. We now turn to that task.
IV.
Before detailing the factors relevant to what we, in Lindy I, termed “the contingent nature of success” and “the quality of an attorney’s work”, certain principles must be emphasized.
A.
Preliminarily, we reaffirm the standards enunciated in Lindy I, supra, 487 F.2d at 167-69. Nothing in the augmentation hereinafter set forth should be considered as a dilution or diminution of that basic formula.
We find it necessary also to observe that we did not and do not intend that a district court, in setting an attorneys’ fee, become enmeshed in a meticulous analysis of every detailed facet of the professional representation. It was not and is not our intention that the inquiry into the adequacy of the fee assume massive proportions, perhaps even dwarfing the case in chief. Once the district court determines the reasonable hourly rates to be applied, for example, it need not conduct a minute evaluation of each phase or category of counsel’s work. In Lindy I we said: “Any increase or decrease in fees to adjust for [117]*117the quality of work is designed to take account of an unusual degree of skill, be it unusually poor or unusually good. If the district judge determines that particular work was of a typical quality, he should, in increasing or decreasing the fee, be cognizant of the amount of time devoted to that given activity.” 487 F.2d at 168-69 (emphasis added). The italicized words were not intended to carry the implication that, in assaying the quality of work that might affect the lodestar award, the district court should inquire separately into the components of the legal representation, to-wit, pleadings, discovery, court appearances, etc. Rather, we meant only that consideration of the quality factor should relate to the overall conduct of the specific case before the court, to-wit, “that particular work”, “that given activity” for which the fee is awarded. See also Part IY C infra.
Finally, we underscore that once a district court determines the “lodestar” it should inquire separately into the contingency and quality factors, and make specific findings of fact as to each.
B.
Under the rubric of “the contingent nature of success” the district court should appraise the professional burden undertaken — that is, the probability or likelihood of success, viewed at the time of filing suit. The court may increase the amount established in the computation of the “lodestar” as a reasonable fee on the basis of a careful evaluation of the following factors:
1. Analysis of plaintiff’s burden. Subsumed in this category are the following considerations: (a) the complexity of the case, — legally and factually; (b) the probability of defendant’s liability, — whether it is clear or dubious; whether it has been previously suggested by other civil or criminal proceedings; whether it is asserted under existing ease law or statutory interpretation, or is advanced as a novel theory; (c) an evaluation of damages, — whether the claims would be difficult or easy to prove.
2. Risks assumed in developing the case. This category subsumes consideration of: (a) the number of hours of labor risked without guarantee of remuneration; (b) the amount of out-of-pocket expenses advanced for processing motions, taking depositions, etc.; (c) the development of prior expertise in the particular type of litigation; recognizing that counsel sometimes develop, without compensation, special legal skills which may assist the court in efficient conduct of the litigation, or which may aid the court in articulating legal precepts and implementing sound public policy.
3. The delay in receipt of payment for services rendered.
If, having considered the foregoing or other relevant criteria, the district court desires to increase the “lodestar” award, it should identify those factors supporting its conclusion, state the specific amount by which the basic fee should be increased due to the contingency of success, and give a brief statement of reasons therefor. We reiterate that any such increment in the “lodestar” award is to be considered and applied apart from the evaluation of the quality of services rendered in the particular proceedings.
C.
Under the rubric of “the quality of an attorney’s work”, the court should appraise the manner in which counsel discharged his or her professional responsibilities. The district court may use this factor to increase or decrease the “lodestar” calculation.
As a first principle, the court must recognize that a consideration of “quality” inheres in the “lodestar” award: counsel who possess or who are reputed to possess more experience, knowledge and legal talent generally command hourly rates superi- or to those who are less endowed. Thus, the quality of an attorney’s work in general is a component of the reasonably hourly rate; this aspect of “quality” is reflected in the “lodestar” and should not be utilized to augment or diminish the basic award under the rubric of “the quality of an attorney’s work”.
[118]*118Lindy I, then, permits an adjustment to the “lodestar” — up or down— based on the all-round performance of counsel in the specific case: “Any increase or decrease in fees to adjust for the quality of work is designed to take account of an unusual degree of skill, be it unusually poor or unusually good.” 487 F.2d at 168. By this is meant simply that the district court may determine that the lawyer discharged the professional burden undertaken with a degree of skill above or below that expected for lawyers of the caliber reflected in the hourly rates. As previously rehearsed, we do not intend that this evaluation entail a detailed analysis of the lawyer’s performance in each category of services rendered. Rather, the increase or decrease reflects exceptional services only; it may be considered in the nature of a bonus or penalty. The heavy burden of proving entitlement to such an adjustment is on the moving party.
In determining whether to adjust the “lodestar” for quality work or not, the district court may consider, inter alia:
1. The result obtained by verdict or settlement, evaluated in terms of (a) the potential money damages available to the class member, i. e., a comparison of the extent of possible recovery with the amount of actual verdict or settlement; (b) the benefit — monetary or non-monetary — conferred on the class, i. e., permitting the court “to recognize and reward achievements of a particularly resourceful attorney who secures a substantial benefit for his clients with a minimum of time invested . .” Merola II, supra, 515 F.2d at 168.
2. An evaluation of the professional methods utilized in processing the ease, — rewarding the use of efficient methods to expedite the case and penalizing the use of methods the predominant purpose of which was to delay or obstruct the proceedings.
If, on the basis of the quality of services rendered, the court is persuaded that an increase or decrease in the “lodestar” is warranted, it should identify those factors supporting its conclusions, state the specific amount by which the basic fee should be altered due to the quality of work, and give a brief statement of reasons therefor.
D.
Because we view the foregoing as an implementation of the Lindy I formulation, rather than as a modification thereof, we will require that it be employed only for those fee applications not already adjudicated in the district courts. Moreover, in the interest of terminating the lengthy proceedings at bar — now in their fifth year — we will not require the district court here to reconsider its determination. Although we do not disturb the district court’s treatment of the contingency and quality factors, it should be apparent that we do not necessarily endorse the methods or the reasoning employed to reach its result.
Henceforth, awards of equitable fund attorneys’ fees should be made in accordance with the guidelines of Lindy I as refined and implemented today. If these guidelines are followed, we will review such awards only for abuse, or misuse, of discretion.
V.
We turn now to the question of determining what portion of the fee award, set in accordance with the Lindy I criteria, the unrepresented class members should bear.12
[119]*119
A. Allocating the Award
We take as a starting point the settled principle that passive , members of a class who “accepted the fruits” of the labors of others are obligated to contribute to the attorney for the active members who created the fund. Central Railroad & Banking Co. v. Pettus, 113 U.S. 116, 127, 5 S.Ct. 387, 28 L.Ed.2d 915 (1885); Trustees v. Greenough, 105 U.S. 527, 26 L.Ed. 1157 (1881). “Absent extraordinary circumstances, the unrepresented claimants should pay for the attorneys’ services in proportion to their benefit from them — that is, the unrepresented claimants should pay a percentage of the reasonable value of the attorneys’ services to the class equal to their percentage of the class’ recovery.” Lindy I, supra, 487 F.2d at 169. With these guidelines, the district court began its analysis of the allocation problem:
For purposes of making [the] determination in this case, the entire settlement fund must be divided into three separate' categories.27 The corrected total number
27 in their original application, petitioners divided the fund into four separate categories. However, in its earlier opinion in this case, this Court found that suggested categories 2 and 3 should be treated alike. See 341 F.Supp. at 1085-1086. Thus, the portion of the fund to be distributed to claimants represented by attorneys other than petitioners will now be treated as one category, not as two. of units in the settlement (excluding claims against Celotex only) is now 2,704,329. The three categories of the settlement fund are as follows:
1. The claims of those litigants represented by petitioners Kohn and Berger, totaling 497,254.4 units, or 18.4% of the total fund;
2. Thé claims of those litigants represented by counsel other than petitioners Kohn and Berger, totaling 1,445,916.2 units, or 53.8% of the total fund; and
3. The claims of the unrepresented claimants, including those of the objectors, totaling 751,158.4 units, or 27.8% of the total fund.
Therefore, the distributions of the fund will be approximately $5,391,200 to claimants in Category 1; $15,763,400 to claimants in Category 2; and $8,145,400 to claimants in Category 3.
This Court has found that petitioners aré entitled to an award of $1,134,765.45, representing reasonable compensation for the creation of a settlement fund to be distributed among all claimants. Relying on the Third Circuit’s opinion in remanding this case, the objectors contend that the portion of the fee to be paid by them should be 27.8% of the total award, a sum which would amount to $315,464.80 under the determinations which this Court has made herein.
382 F.Supp. at 1025 (footnote omitted). The court, however, ordered the unrepresented claimants to pay, not 27.8 per cent of the fee, but 81.6 per cent. Ibid, at 1027. If we are to affirm this allocation, therefore, we must be able to find “extraordinary circumstances”.
The essence of the district court’s implicit finding of extraordinary circumstances appears to have been its concern that the unrepresented claimants would contribute a disproportionately small percentage of their settlement recoveries to attorneys. That is, if the unrepresented claimants paid only their aliquot share of the equitable fund award — 27.8 per cent of $1,134,765.45 or $315,464.80 — they would be paying only 3.87 per cent of their settlement recoveries to attorneys. By contrast, claimants in category 1 had contracted to pay their attorneys contingency fees of 33V3 per cent of any recovery, and claimants in category 2 had contingency agreements calling for payments of 20 to 40 per cent of their recoveries. Ibid, at 1025.
We believe the district court's concern was misplaced.13 The court should not have [120]*120mixed considerations of the fee determined by equitable principles (the equitable fund award) with considerations of the fees dependent on private contracts. The fee with which the court was concerned is a child of equity — not of express contract. The task before the court was to determine the fee from the fund to which Kohn and Berger, having rendered services that benefited all members of the settlement class, were entitled on the basis of quantum meruit. Once that fee was determined, the general rule is that it be allocated pro rata among the claimants. Private arrangements individual members of the class may have with counsel are simply irrelevant. Were the rule otherwise, “extraordinary circumstances” would seem to inhere in each class action that results in an equitable fund from which an award of attorneys’ fees may be made. This is not what we had in mind in Lindy I — when we reserved a limited exception to the general allocation rule.14
B. “Equitable Set-Off”
In concluding that category 3 claimants should pay 81.6 per cent of the applicable award, the district court followed a circuitous route. The first signpost was the observation: “[Petitioners quite properly recognized that no additional fees should be charged against their clients, who will be paying petitioners $861,191 under the contingent fee agreements in effect. Moreover, this Court has previously ruled that none of the other litigants should be required to pay any additional fees beyond what they have agreed to pay their own attorneys in this litigation.” Ibid, at 1025-26. Under normal circumstances, the court reasoned, appellees Kohn and Berger could recover 53.8 per cent of the applicable award from category 2 claimants, for the benefits the attorneys conferred on them. Attorneys representing category 2 claimants, however, had also performed services benefiting the fund, including category 3 claimants, and, accordingly, had a theoretical quantum meruit right of action for attorneys’ fees against the category 3 claimants. The court then employed a device-styled “equitable set-off” — whereby it ordered category 3 members to pay 53.8 per cent of the award directly to appellees, rather than requiring that Kohn and Berger seek that amount from category 2 claimants, who might then seek to recover the like amount from category 3 claimants.
Whatever merit this equitable set-off device might have in the abstract, we hold that the district court erred in applying it to this case. The court “conclude[d] that the value of the benefits conferred on the unrepresented claimants by the many attorneys who represented litigants in Pategory 2 and who contributed to the creation of the settlement fund [was] at least $610,503.81.” Ibid, at 1028 (emphasis added). We find the record devoid of the type of evidence necessary since Lindy I to support an award of attorneys’ fees (or their set-off).15 There were no affidavits, no stipulations, no testimony adduced at a hearing to substantiate this evaluation. See Estien v. Christian, 507 F.2d 61, 64 (3d Cir. 1975).16
[121]*121
C. Adequate Compensation
Nor can we accept as an “extraordinary circumstance” the district court’s concern that Kohn and Berger would not be adequately compensated unless the 53.8 per cent usually allocable to category 2 claimants were added to the 27.8 per cent chargeable to category 3 claimants. Without determining whether the total amount of attorney compensation from private agreements and equitable fund awards is a valid consideration in setting an equitable fund fee, we note that in the peculiar circumstances of this case, the actual total compensation to the Kohn and Berger firms will bear an uncanny resemblance to the total amount laboriously calculated by the district court.
Accepting as a starting point the district court’s gross award from the fund of $1,134,765.45, and deducting therefrom the sums of $29,806.25, $320 and $40,654 as more particularly set forth in Part II, supra, we conclude that the appellees were entitled to an award of $1,063,985.20 from all members of the settlement class. The district court concluded that claimants of categories 1 and 2 should shoulder no part of this burden. The propriety of that ruling is not before us. We have found no extraordinary circumstances justifying departure from the general rule of pro rata allocation; accordingly, category 3 claimants should pay 27.8 per cent of the equita-1 ble fund attorneys’ fee award — or $295,-787.88. Consequently, the Kohn and Berger firms will receive total fees of $1,156,978.88 —including $861,191 under private contingent agreements. The district court’s concern that they be adequately compensated should be satisfied completely, for their total compensation will exceed by $22,213.43 the “final award” of the district court, see 382 F.Supp. at 1024, and by $546,929.03 the district court’s “lodestar” calculation, see ibid, at 1014, which “constitute^] reasonable compensation” for services rendered, Lindy I, supra, 487 F.2d at 168, exclusive of quality and contingency factors.
In the circumstances of this case, we perceive no basis for finding that the benefits conferred on the category 3 claimants exceeded the proportional benefits conferred on the claimants in categories 1 and 2. We have previously indicated that the considerations which appear to have influenced the district court’s deviation from the general rule of pro rata allocation do not rise to the requisite level of “extraordinary circumstances”. Accordingly, we hold that the district court erred in departing from the general rule of pro rata allocation.
VI.
Two additional contentions, neither of which need detain us long, remain for consideration.
First, appellants urge that the district court erred in awarding $10,000 for “future time”. The court arrived at this award by multiplying the estimated 100 hours of additional work that remained to wind up settlement administration by the $100 rate applicable to attorney Fine, appellee Kohn’s partner. The court found that the estimate upon which it based the award was “reasonable”. 382 F.Supp. at 1023. It did so only after hearing testimony. See App. at 1142-43. Having reviewed the record, we cannot say that this finding was clearly erroneous, Krasnov v. Dinan, supra, and we will not upset the award. Rather, we defer to the district judge “who was there”. See n.ll supra.
Second, appellees urge that they are entitled to interest on the fee award from the date of the district court’s 1972 award. This argument is frivolous. Lindy I vacated the original attorneys’ fee award. See n.12 supra. Consequently, there was no predicate award upon which-to base interest. Moreover, because of the result reached herein, category 3 claimants will pay less to the Kohn and Berger firms than [122]*122the $473,712.01 which was paid to appellees on December 13, 1972, out of the first distribution made pursuant to settlement order No. 41. See App. at 1009.
VII.
We conclude as follows:
1. We reverse the district court’s award of $29,806.25 for fee application work prior to March 20, 1974, the award of $40,654 for work after March 20, 1974, and the award of $320 for fee application work by paraprofessionals and law students.
2. We find that the appropriate award to appellees from the settlement fund chargeable to all members of the settlement class was $1,063,985.20.
3. Category 3 claimants should pay 27.8 per cent of the equitable fund attorneys’ fee award instead of 81.6 per cent as ordered by the district court. Based on an award of $1,063,985.20, this would amount to $295,787.88 instead of $925,968.61 as ordered by the district court.
The judgment of the district court will be vacated and the proceedings remanded for entry of an appropriate order in accordance with the foregoing.
Each party to bear his own costs.