MINER, Circuit Judge:

Our discussion of the background and procedural history of this litigation appears in Judge Winter's lead opinion, 818 F.2d 145. The nine members of the Plaintiffs' Management Committee ("PMC") and various outside counsel appeal, on a number of grounds, the district court's decision setting attorneys' fees. On June 18, 1985, the district court issued an amended order, awarding over seven million dollars in fees and three million dollars in expenses to eighty-eight attorneys and law firms involved in the action. In re "Agent Orange" Product Liability Litigation, 611 F.Supp. 1296 (E.D.N.Y. 1985) ( "Agent Orange"). The nine members of the PMC, individually and as a group, challenge the district court's use of a national hourly rate in calculating the fee awards under the lodestar formula set forth in City of Detroit v. Grinnell Corp., 495 F.2d 448 (2d Cir. 1974) ( "Grinnell I"), and City of Detroit v. Grinnell Corp., 560 F.2d 1093 (2d Cir. 1977) ( "Grinnell II"), the level of the quality multipliers it set, and its failure to apply a risk multiplier to the fee awards and to credit certain hours and expenses. Four outside counsel challenge the district court's findings as to the value of their work to the class and the decision to abrogate various contingency fee arrangements between counsel and certain class members. For the reasons set forth below, we affirm in part and reverse in part.

I. BACKGROUND

In May of 1984, on the eve of trial, a settlement was reached with the chemical company defendants, calling for the establishment of a $180 million dollar fund for the benefit of the class. By order dated June 11, 1984, the district court required fee petitions to be filed no later than August 31, 1984, and scheduled hearings on the petitions for the early fall. Notice of Proposed Settlement of Class Action, reprinted in In re "Agent Orange" Product Liability Litigation, 597 F.Supp. 740, 867 (E.D.N.Y. 1984). Pursuant to this procedure, well over 100 attorneys and law firms filed petitions, claiming tens of thousands of hours of work performed for the benefit of the class. The fee petitions fell into three categories: those filed by the nine members of the PMC; those filed by members of Yannacone and Associates, the original consortium of attorneys in charge of the action; and those filed by attorneys not connected with any court-appointed entity representing the class.

In reviewing fee petitions, the district court developed guidelines falling into two categories — one covering the hours to be credited for work performed and the other covering the expenses to be reimbursed. The hourly guidelines were as follows:

1. Court Time: One-half of the time requested for review of court orders was permitted on the ground that the majority of court orders were made in open court or after extensive briefing. Telephone conference time with court personnel was awarded in full, except that no time was awarded for conferences relating to internal management difficulties of the PMC. Attendance at, and preparation for, court hearings was awarded in full. Review of hearing transcripts was awarded in full for those attorneys attending the hearing. Nonattending attorneys were awarded for only half such time. Travel to and from hearings and court appearances also was awarded on a fifty percent basis.

2. Management Committee Meetings: All time for PMC meetings on substantive issues was permitted. Travel to and from such meetings was awarded on a fifty percent basis. No time was awarded for meetings on nonsubstantive topics. The same division was made for telephone conferences among PMC members.

3. Educational Reading: Time for review of scientific materials relating to the causation issue and other issues in the case was awarded on a fifty percent basis on the ground that such knowledge could be used by counsel in future cases.

4. Depositions: Half of the time was awarded for travel to and from depositions, for attendance by nonparticipating attorneys, and for review and reading. All time for preparing and summarizing depositions was granted. No limit on the length of depositions was enforced.

5. Document Preparation: All time for review and preparation of legal documents was awarded, except that those hours used to prepare documents concerning internal PMC organizational issues were not credited.

6. Mail: If a short period of time for review of a substantial amount of mail was requested, no time was awarded under the assumption that counsel simply was opening the mail. If a lengthy period of time was claimed for review of only a few letters, all time was credited under the assumption that counsel was reviewing a letter brief.

7. Intra-Firm Conferences: This time was credited on a fifty percent basis when related to substantive issues.

Agent Orange, 611 F.Supp. at 1320-21, 1350-51. The expense guidelines were as follows:

1. Travel: Documented expenses for hotels were reimbursed at ninety dollars per day. Meals were reimbursed at fifty dollars per day and twenty dollars per day if the attorney was in his home city.

2. Paralegal Time: Paralegals were treated as an expense and reimbursed at a rate of twenty dollars per hour.

3. Out-of-Pocket Expenses: Telephone, mailing, duplication and similar expenses were reimbursed in full if adequately documented.

4. Percentage Approval: When counsel submitted adequate documentation to prove expenses but were unable to establish that those expenses were all related to compensable activity, expenses were reimbursed on a percentage basis.

5. Fees for Non-Causation Experts: A cap of $5,000 per expert was set on the ground that their input was not substantial and not reasonably related to class interests.

Id. at 1321-22, 1351.

Following these guidelines and applying the lodestar formula for calculating attorneys' fees in an equitable fund action, see Grinnell I, 495 F.2d at 471, the district court awarded $10,767,443.63 in individual fees and expenses to various counsel who, in the court's view, had performed work beneficial to the class. In arriving at the lodestar figure, the court employed national hourly rates of $150 for the work of a partner, $100 for the work of an associate, and $125 for the work of a law professor. Agent Orange, 611 F.Supp. at 1326. The court, in its discretion, further applied quality multipliers, ranging from 1.50 to 1.75, to the fees allowed various members of the PMC and other counsel who had exhibited exceptional skill in the litigation and settlement negotiations. Id. at 1328. The district judge, however, declined to apply a risk multiplier to the lodestar figure. Id.

Not satisfied with these awards, two groups of attorneys, including the PMC, now raise numerous objections on appeal.

II. DISCUSSION

A. PMC Members

The district court awarded the individual members of the PMC an aggregate of $4,713,635.50 in fees and $650,356.97 in individual expenses. In addition, the court awarded the PMC, as a whole, expenses in the sum of $1,711,155.87. These attorneys now raise four specific challenges to their individual awards.

1. National Hourly Rates

Faced with a flood of fee petitions from counsel located in all regions of the country, the district court utilized national hourly rates for calculating the fee awards for each attorney. While it recognized that the general rule for fee calculation in this circuit requires the use of "the hourly rate normally charged for similar work by attorneys of like skill in the area," Grinnell II, 560 F.2d at 1098, the district court noted that special problems arise "in applying this general standard in a complex multidistrict litigation that is national in scope, involves counsel from all over the country and extends over many years during which the rates for particular lawyers and classes of lawyers are changing," Agent Orange, 611 F.Supp. at 1308.

Specifically, the court pointed out that if the general rule were interpreted to require imposition of the rates normally imposed within the district, the rule would make little sense in the context of this action, given that the vast majority of counsel involved were non-local. Alternatively, if the rule were interpreted to require imposition of varying rates depending upon the location of each counsel's practice, the district judge perceived that such a rule would minimize the court's familiarity with the rates to be awarded, require an almost unworkable case-by-case review of such rates, and consistently benefit non-local counsel at the expense of the class fund. The district judge concluded that in large multiparty litigation, where substantial numbers of specialized non-local attorneys are involved, utilization of a national hourly rate is appropriate because it "recognizes the national character of the lawsuit and of class counsel while retaining a vitally important administrative simplicity together with an essential neutrality of result as between fee applicants and fund beneficiaries." Id. at 1309.

Relying on five separate sources, the district court developed the national rates to be applied in this action. First, the court considered data compiled in the National Law Journal Directory of the Legal Profession (B. Gerson, M. Liss P. Cunningham eds. 1984), a periodical that provided rate information concerning law firms of fifty or more attorneys throughout the country as of March 1983. Second, the court reviewed the submissions of counsel, in particular the defendants' Memorandum Concerning Plaintiffs' Lawyers' Applications for Attorneys' Fees and for Reimbursement of Expenses, which provided further information on national rates. Third, the court reviewed various surveys of law firm economics, dated 1980 through 1984, and other periodicals relating to the manner in which firms bill their clients. Fourth, the court took notice of its own experience in setting fee awards in class actions. Finally, the district judge reviewed recent fee awards by other courts to understand more fully the manner in which other jurisdictions set appropriate rates. Agent Orange, 611 F.Supp. at 1325-28 (citing, inter alia, In re Fine Paper Antitrust Litigation, 751 F.2d 562, 590 n. 22 (3d Cir. 1984); Grendel's Den, Inc. v. Larkin, 749 F.2d 945, 955-56 (1st Cir. 1984)). From an analysis of this data, the district court arrived at national hourly rates of $150 for partners, $100 for associates and $125 for law professors.

The members of the PMC challenge the use of national rates on the ground that they do not comport with the principles governing attorneys' fee awards in equitable fund actions. They assert that the practice in this and other circuits required the court to review independently the hourly rate for each attorney in order to ensure that he was compensated at a level commensurate with that of other counsel of like skill in the area in which he practices. See, e.g., In re Fine Paper, 751 F.2d at 590-91 (classifying application of national hourly rates as legal error on the grounds that the district court presented no evidentiary basis for their establishment and such rates ignored the market rates that the attorneys would command in their respective communities). Relying on large class action cases in other circuits where courts have awarded varying rates to counsel from different localities, e.g., In re Equity Funding Corp. of America Securities Litigation, 438 F.Supp. 1303 (C.D.Cal. 1977), they argue that, while the task may be a difficult one, other jurisdictions routinely undertake it.

In passing on the efficacy of national hourly rates, we note that fees in this action were awarded under the equitable fund doctrine, which seeks to ensure that counsel who have performed services beneficial to the class receive fair and just compensation for their respective efforts. Trustees v. Greenough, 15 U.S. (15 Otto) 527, 536, 26 L.Ed. 1157 (1882). In order to provide counsel with such compensation and, at the same time, temper these awards to prevent windfalls, we have adopted a lodestar formula for calculating fees in equitable fund and statutory fee contexts. Grinnell II, 560 F.2d at 1099; Grinnell I, 495 F.2d at 469-71. Under the formula, the district court initially multiplies the number of hours reasonably billed by the hourly rate normally charged for equivalent work by similarly-skilled attorneys in the area. Grinnell II, 560 F.2d at 1098. Once calculated, the district court then may, in its discretion, upwardly or downwardly adjust this figure by considering such factors as the quality of counsel's work, the probability of success of the litigation and the complexity of the issues. Id.

While at least one circuit looks to the rates employed in the area in which the attorney practices, Cunningham v. City of McKeesport, 753 F.2d 262, 267 (3d Cir. 1985), we traditionally have interpreted Grinnell I and Grinnell II as requiring use of the hourly rates employed in the district in which the reviewing court sits, Polk v. New York State Department of Correctional Services, 722 F.2d 23, 25 (2d Cir. 1983). We generally have adhered to this rule whether the attorney involved was local or non-local. Id.; accord Donnell v. United States, 682 F.2d 240, 251-52 (D.C. Cir. 1982), cert. denied, 459 U.S. 1204, 103 S.Ct. 1190, 75 L.Ed.2d 436 (1983); Chrapliwy v. Uniroyal, Inc., 670 F.2d 760, 768-69 (7th Cir. 1982), cert. denied, 461 U.S. 956, 103 S.Ct. 2428, 77 L.Ed.2d 1315 (1983); Avalon Cinema Corp. v. Thompson, 689 F.2d 137, 140-41 (8th Cir. 1982) (in banc). We and other circuits have strayed from this rule only in the rare case where the "special expertise" of non-local counsel was essential to the case, it was clearly shown that local counsel was unwilling to take the case, or other special circumstances existed. Polk, 722 F.2d at 25; Avalon Cinema, 689 F.2d at 140-41.

Accordingly, the issue for review here is whether the district court erred in deviating from this established precedent. While we concede that such conduct in the ordinary case would constitute legal error and require recalculation of the lodestar, we conclude that, in an exceptional multiparty case such as this, where dozens of non-local counsel from all parts of the country are involved, public policy and administrative concerns call for the district court to be given the necessary flexibility to impose a national hourly rate when an adequate factual basis for calculating the rate exists.

An examination of the alternatives to the use of national rates in large multiparty class actions of this sort readily establishes the necessity for affording district courts this discretion. Use of our forum rule would distort dramatically the purposes of the lodestar calculation itself — to ensure fair and just compensation to counsel and to prevent the award of windfall fees. This distortion would occur because, in cases in which the vast majority of attorneys involved are non-local, the forum rule necessarily will either overcompensate or undercompensate a substantial number of non-local attorneys. Undercompensation could deny counsel their right to fair and just fees; overcompensation would not be consistent with the need to prevent windfalls. Adherence to the forum rule in cases in which the inherent limitations of the rule are magnified, i.e., where few local counsel and vast numbers of non-local counsel are involved, therefore, makes little sense.

Resort to a varying approach, depending upon the area in which the individual practices, fares no better. In an action of the magnitude of Agent Orange, in which well over one hundred fee petitions were filed by counsel throughout the country, such an approach would pose an administrative nightmare for the district court. As the district judge here noted, "[s]implicity becomes an especially important goal in a complex case involving a hundred or more fee applications and tens of thousands of pages of supporting documentation and requiring a number of years for prosecution during which rates for particular attorneys and geographic locations change in different ways." Agent Orange, 611 F.Supp. at 1308. While administrative interests normally should not be the primary concern of a court in formulating substantive rules of review, we observe that the attorney-by-attorney approach recommended by the PMC simply would over-tax the capacity of a district court to review fee petitions adequately. Cf. New York Association for Retarded Children v. Carey, 711 F.2d 1136, 1146 (2d Cir. 1983) (burden-saving measures may be taken by district court in light of voluminous fee petitions).

Although not a panacea, the use of national hourly rates in exceptional multiparty cases of national scope, where dozens of non-local counsel are involved, appears to be the best available method of ensuring adherence to the principles of the lodestar analysis. The risk of overcompensation or undercompensation on a large scale, apparent under the forum rule, is somewhat neutralized, while, at the same time, the administrative burden on the district court, apparent under the varying rate rule, is reduced to a manageable level. In granting the district court this discretion, however, we caution that such rates should be employed only in the exceptional case presenting problems similar to those presented here. We further caution that, even in similar cases, national hourly rates should be employed only when the district court is presented with an adequate evidentiary basis on which to fix such rates. Once the court is satisfied with the evidence, it should make clear, factual findings that support its determination.

We are aware that at least one circuit has rejected the imposition of national hourly rates on the ground that they do not comport with the lodestar principle. In re Fine Paper, 751 F.2d at 591. To the extent, however, that the Third Circuit's decision was based upon the fact that the national rates employed did not comport with that circuit's rule requiring the hourly rate to reflect the rate normally charged in the local in which counsel practices, we already have rejected its analysis by following a forum rate rule. See Polk, 722 F.2d at 25. In addition, In re Fine Paper, though not entirely clear on this point, may be read to condemn only national hourly rates not based on an adequate evidentiary record. The Third Circuit, in reversing the district court's adoption of such rates, indicated that the district court there had not referred to any evidence supporting the existence of such rates, 751 F.2d at 590, and noted that "the subject is not one on which judicial notice is appropriate," id. If read in that context, our decision is in accord with that of the Third Circuit, since we limit the utilization of national rates to those instances in which an adequate evidentiary basis exists. Finally, even assuming that In re Fine Paper stands for an absolute prohibition on the imposition of national hourly rates, we note that, subsequent to that decision, the Third Circuit Task Force on Court Awarded Attorney Fees, organized at the behest of the Chief Judge of that Circuit, recommended that the court permit the utilization of such rates in exceptional cases. Court Awarded Attorney Fees, Third Circuit Task Force, 108 F.R.D. 237, 260-62 (Oct. 8, 1985).

Given our determination that the utilization of national hourly rates in limited circumstances is proper, we further conclude that the district court did not abuse its discretion in calculating the specific hourly rates in the present case. In its decision, the court set forth the five bases upon which it computed these rates. The PMC does not challenge specifically those bases and we find little reason to question them. Hourly rates for counsel in this action were difficult to calculate because the majority of attorneys involved normally would have been compensated through contingency fee arrangements rather than on an hourly basis. Difficulties aside, however, the district judge, in our view, took adequate steps to ensure a fair and just hourly rate of compensation. We therefore hold that the national hourly rates of $150 for partners, $100 for associates and $125 for law professors constituted an element of fair and just compensation for counsel in the context of this case.

2. Quality Multipliers

Having computed the initial lodestar figure, the district court awarded discretionary quality multipliers of 1.5, and in one case 1.75, to six members of the PMC on the ground that these attorneys had exhibited exceptional skills in the litigation and settlement negotiations. The six PMC recipients now challenge the level of the multipliers as being unjustifiably low and further challenge the district court's failure to award quality multipliers in connection with the fees of the three other PMC members.

The decision to allow a quality multiplier rests in the sound discretion of the district court, Hensley v. Eckerhart, 461 U.S. 424, 437, 103 S.Ct. 1933, 1941, 76 L.Ed.2d 40 (1983); Grinnell II, 560 F.2d at 1098, due to "the district court's superior understanding of the litigation and the desirability of avoiding frequent appellate review of what essentially are factual matters." Hensley, 461 U.S. at 437, 103 S.Ct. at 1941. The Supreme Court, however, in Blum v. Stenson, 465 U.S. 886, 899, 104 S.Ct. 1541, 1549, 79 L.Ed.2d 891 (1984), and more recently in Pennsylvania v. Delaware Valley Citizens' Council for Clean Air, ___ U.S. ___, 106 S.Ct. 3088, 92 L.Ed.2d 439 (1986), has severely restricted those instances in which a district court may allow such a multiplier.

In Blum, a decision concerning application of the lodestar analysis to a fee award under 42 U.S.C. § 1988, the Court determined that factors such as quality of representation are presumed to be fully reflected in the initial lodestar figure, derived by multiplying the number of hours reasonably billed by the court-established hourly rate. Blum, 465 U.S. at 899, 104 S.Ct. at 1549. Accordingly, the Court concluded that an adjustment to the lodestar figure for such a factor would only be proper in "the rare case where the fee applicant offers specific evidence to show that the quality of service rendered was superior to that one reasonably should expect in light of the hourly rates charged and that the success was `exceptional.'" Id. (emphasis added). In Delaware Valley Citizens' Council, a decision concerning application of the lodestar analysis to a fee award under section 304(d) of the Clean Air Act, 42 U.S.C.A. § 7604(d) (West 1983), the Court reaffirmed the narrow approach taken in Blum, declaring that calculating fee awards under the lodestar analysis "leaves very little room for enhancing the award based on [counsel's] post-engagement performance." Delaware Valley Citizens' Council, 106 S.Ct. at 3098.

Given these pronouncements, the issue, in our view, is not whether the quality multipliers awarded by the district court here were set too low, but rather whether they should have been awarded at all. In what we consider to be a close case, we conclude that the district court did not abuse its discretion in awarding the multipliers for quality to six of the PMC members, or in failing to award them to the other three members.

The district court specifically found that these six attorneys, as well as several outside counsel who have not appealed, deserved to be awarded quality multipliers at various rates because each had "demonstrated an unusual degree of skill in presenting complex and often novel issues to the court," Agent Orange, 611 F.Supp. at 1328, or had "shown a level of organization and efficiency that goes beyond what is usually expected," id. Under ordinary circumstances, even assuming the high level of work performed by counsel here, we would be constrained to reverse the district court's award in light of the severe restrictions set forth in Blum and Delaware Valley Citizens' Council. While the work indeed may have been of high quality, the presumption is that such factors already are reflected in the initial lodestar figure.

In this case, however, we find that the use of a national hourly rate skews the normal lodestar analysis enough to require consideration of quality factors in order to satisfy the requirements of just and fair compensation. While we affirm the use of national rates in the present case, we realize that such rates inherently cannot be calculated as precisely as those under the forum rule, or those under the varying locale rule. Consequently, the Blum and Delaware Valley Citizens' Council presumption of inclusion of quality factors within the initial lodestar figure should not, in our view, apply to those instances in which the district court utilizes this less precise analysis.

3. Risk Multiplier

The district court declined to award a risk multiplier to any attorney involved in the case. It reasoned that risk of success should not be judged solely from the vantage point of whether a complete recovery at the conclusion of the action is viable, but also should include an evaluation of the likelihood that the parties will reach a settlement. In this regard, the court noted that it was probable that the defendant chemical companies would settle the case "to avoid the further burden of litigation and to improve their respective financial pictures." Agent Orange, 611 F.Supp. at 1311. The court also recognized that awarding risk multipliers in a case such as Agent Orange, which held out little chance for a victory on the merits but a significant chance of settlement, would fuel the filing of nuisance litigation "in which settlement becomes the main object and attorney fee awards an overpowering motivating force." Id.

Furthermore, the court indicated that strict application of inversely proportionate risk multipliers to cases such as Agent Orange, which it described as a high-risk case of highly questionable merit, would lead to a confounding disparity in the treatment of cases falling just above and just below the standard for frivolousness under Fed.R.Civ.P. 11. Attorneys in successful cases bordering on the frivolous, yet falling just above the proscriptions of Rule 11, would be awarded the highest risk multipliers, since the risk of success in such cases obviously would be great. In contrast, counsel in similar cases falling just below Rule 11's proscriptions, would not only receive no risk multiplier, but also would be subject to court-imposed sanctions for having brought such a case.

Finally, the court took note that, as a matter of public policy, the need to utilize a risk multiplier in a given case must be viewed in relation to the equally important concerns of judicial administration and legal morality. To this end, the refusal to allow a multiplier here would force the legal community "to think at least twice before initiating sprawling, complicated cases of highly questionable merit that will consume time, expense and effort on the part of all concerned, including the courts, in a degree vastly disproportionate to the results eventually obtainable." Id. at 1312. While such a policy would not reward the filing of these questionable cases, the court did note that counsel's entitlement to a lodestar award without a multiplier would nonetheless serve adequately to encourage attorneys to represent plaintiffs in cases of this nature.

The PMC challenges the district court's failure to allow a risk multiplier on the ground that it does not comport with principles of just and fair compensation. While conceding that plaintiffs' case would have been difficult to prove, the PMC members strongly take exception to the district court's description of the action as being of dubious or questionable merit. As to the probability of the parties reaching a settlement in the action, the PMC members point to the fact that such a settlement was not reached until the eve of trial, and label as "economic suicide" the notion that they advanced funds and spend thousands of hours working on the case with some inner assurance that defendants would make a reasonable settlement proposal because of the bothersome nature of the litigation.

We have labeled the risk-of-success factor as "perhaps the foremost" factor to be considered under the second prong of the lodestar analysis. Grinnell I, 495 F.2d at 471. The multiplier takes into account the realities of a legal practice by rewarding counsel for those successful cases in which the probability of success was slight and yet the time invested in the case was substantial. Id.; see 7B C. Wright, A. Miller M. Kane, Federal Practice and Procedure § 1803, at 524-27 (1986). As the chance of success on the merits or by settlement increases, the justification for using a risk multiplier decreases. Grinnell I, 495 F.2d at 471. The need for this type of multiplier is magnified when the "diminutive character of the individual claims" forces counsel to bring the action on a class basis. 7B C. Wright, A. Miller M. Kane, supra, § 1803, at 527. Without the prospect of some consideration for the risks and uncertainties of the action, "the necessary incentive [for prosecuting such a suit] would be lacking and a major weapon for enforcing various public policies would be blunted." Id.

The problem with risk multipliers, however, is that they tend to reward counsel for bringing actions of dubious merit. If such multipliers are awarded on a perfectly proportionate basis, i.e., the greater the chance that the case would not succeed the higher the multiplier, "the net effect . . . would be to make a marginal case as attractive to bring as a very strong case." Laffey v. Northwest Airlines, Inc., 746 F.2d 4, 27 (D.C.Cir. 1984), cert. denied, 472 U.S. 1021, 105 S.Ct. 3488, 87 L.Ed.2d 622 (1985). This, in turn, would provide an incentive for counsel to flood "the courts with unmeritorious litigation," McKinnon v. City of Berwyn, 750 F.2d 1383, 1392 (7th Cir. 1984), "leading . . . to a situation in which every conceivable claim would be litigated, subject only to the ability of the courts to handle the burden," Laffey, 746 F.2d at 27; accord Leubsdorf, The Contingency Factor in Attorney Fee Awards, 90 Yale L.J. 473, 491 (1981). The net result, of course, would be a dilution of the judiciary's ability to handle those cases with potentially meritorious claims.

A court, therefore, in adjudging whether to award a risk multiplier, should examine closely the nature of the action in order to determine whether, as a matter of public policy, it is the type of case worthy of judicial encouragement. In our view, the case here clearly is not and, consequently, we agree with the district court's decision not to impose a risk multiplier.

From the outset, the factual and legal difficulties hindering the successful prosecution of plaintiffs' case have been staggering. Factual evidence of causation has been at best tenuous and, if not for the last-minute settlement, the military contractor defense would have prevented class members from realizing any recovery at all. When these significant weaknesses in plaintiffs' case are viewed in light of the sheer magnitude of the action and the thousands of hours of court time that this type of action requires, it becomes clear that the federal courts should not actively encourage the bar to file such dubious actions in the future.

Besides matters of public policy, the settlement itself presents a rational for denying counsel's request. While today we hold that the settlement falls within the range of reasonableness under Fed.R.Civ.P. 23, we are aware that the $180 million settlement provides a very small return to the class in light of the claims asserted. In our estimation, the relatively small size of the settlement reflects class counsel's realization of the extreme difficulty they would incur in overcoming the inherent weaknesses of their case, in particular the military contractor defense, and the defendant chemical companies' realization that they could end a burdensome litigation at very low cost. Award of a risk multiplier in such circumstances, as the district court reasoned, only would further the unwelcome prospect of nuisance litigation being brought in federal courts.

In denying class counsel their requested multiplier, we note that each attorney has received the fair value of his services to the class under the lodestar analysis. An additional award of a risk multiplier not only would provide excessive compensation but would encourage counsel to accept similar matters for litigation in the future. We find no reason to do more to encourage litigation that could substantially occupy the federal judiciary in matters of little merit.

4. Hours and Expenses

The PMC members challenge the district court's guidelines on the grounds that they improperly failed to credit certain hours and reimburse certain expenses. Specifically, they challenge the court's decision to disallow fifty percent of the time spent on reading scientific literature, to disallow fifty percent of the time spent on travel, to disallow a portion of the time spent reviewing mail and on the telephone, to disallow fifty percent of the time spent reviewing depositions, and to disallow a substantial amount of post-settlement work. As to expenses, they challenge the court's decision to reduce expenses by a percentage when such expenses could not be connected with compensable activity, to set a maximum fee for noncausation expert witnesses, and to treat paralegals as a cost. In sum, they allege that, taken together, if not separately, such radical deductions in their hours and expenses billed constituted an abuse of the court's discretion.

The district court is given broad discretion in setting fee awards. Hensley, 461 U.S. at 437, 103 S.Ct. at 1941, Carey, 711 F.2d at 1146. We cannot reverse a district court's finding in this regard merely because we might have weighed the information provided in the fee petitions differently or might have found more of the hours billed as being beneficial to the class. Cf. Anderson v. Bessemer City, 470 U.S. 564, 573-74, 105 S.Ct. 1504, 1511-12, 84 L.Ed.2d 518 (1985). The district judge is in the best position to weigh the respective input of counsel, considering its "superior understanding of the litigation." Hensley, 461 U.S. at 437, 103 S.Ct. at 1941. Accordingly, we will reverse a district court's findings as to which hours to compensate "only when it is apparent that the size of the award is out of line with the degree of effort reasonably needed to prevail in the litigation." Carey, 711 F.2d at 1146.

We find no abuse of discretion here. The critical inquiry when reviewing hours billed to the common fund in a class action is whether the work performed resulted in a benefit to the class. See Grinnell II, 560 F.2d at 1099. In determining which hours were beneficial, we note that there "are no hard-and-fast rules," Seigal v. Merrick, 619 F.2d 160, 164 n. 9 (2d Cir. 1980), but that "[a]mple authority supports reduction in the lodestar figure for overstaffing as well as for other forms of duplicative or inefficient work," id. Moreover, we and other circuits have held that in cases in which substantial numbers of voluminous fee petitions are filed, the district court has the authority to make across-the-board percentage cuts in hours "as a practical means of trimming fat from a fee application." Carey, 711 F.2d at 1146; accord Ohio-Sealy Mattress Manufacturing Co. v. Sealy Inc., 776 F.2d 646, 657 (7th Cir. 1985); Copeland v. Marshall, 641 F.2d 880, 903 (D.C.Cir. 1980) (in banc). But see In re Fine Paper, 751 F.2d at 596 (court must identify with some specificity any disallowed hours). Under such circumstances, no item-by-item accounting of the hours disallowed is necessary or desirable. Ohio-Sealy, 776 F.2d at 658.

Here, the fee petitions, to say the least, were voluminous, consisting of tens of thousands of pages of billing sheets and other exhibits. To suggest that the district court could not take advantage of percentage reductions in such a context would be absurd. In reviewing these across-the-board cuts, we find nothing that we could classify as an abuse of discretion. Moreover, it is not unusual for hours of travel time, deposition time and other quasi-administrative items to be compensated at lower rates. E.g., Sun Publishing Co. v. Mecklenburg News, Inc., 594 F.Supp. 1512, 1520 (E.D.Va. 1984); Steinberg v. Carey, 470 F.Supp. 471, 479-80 (S.D.N.Y. 1979). But see Crumbaker v. Merit Systems Protection Board, 781 F.2d 191, 193-94 (Fed. Cir. 1986) (reasonable travel time should be compensated at the same rate as other working time). The district judge gave reasons, though somewhat generalized, for each percentage cut that he made. We find these to be an adequate reflection of the benefit that the class derived from counsel's work.

We also find no abuse of discretion in the district court's guidelines for expenses. Counsel are entitled to reimbursement only for those expenses incurred in the course of work that benefitted the class. In re Armored Car Antitrust Litigation, 472 F.Supp. 1357, 1388-89 (N.D.Ga. 1979), modified and remanded on other grounds, 645 F.2d 488 (5th Cir. 1981). Overstaffing and other extravagances are not recoverable. Id.

Given this standard, the district court's finding that the reports of the noncausation witnesses were of only marginal use to the class and were "uniformly inadequate" suggests that the court in fact was generous in setting the cap for fees to these experts at $5,000 each. Report and Recommendation of United States Magistrate, Re: Fee Petitions, appendixed to and incorporated in Agent Orange, 611 F.Supp. at 1351. We also find no abuse of discretion in the district court's determination that expenses connected with those hours disallowed as not being beneficial to the class should not be reimbursed. See In re Fine Paper Antitrust Litigation, 98 F.R.D. 81, 85 (E.D.Pa. 1983), rev'd on other grounds, 751 F.2d 562 (3d Cir. 1984). Finally, although we concede that under certain circumstances it may be appropriate not to treat paralegal time as an expenses in a large class action, see Dorfman v. First Boston Corp., 70 F.R.D. 366, 374-75 (E.D.Pa. 1976), we note that the district court in so doing was simply following our prior directive, see Grinnell I, 495 F.2d at 473. We decline to reevaluate that rule here.

B. Outside Counsel

1. Ashcraft Gerel

The Magistrate's recommendation, adopted by the district court, also reflected a negative quality multiplier of .25 on the ground that in 1983 the firm had withdrawn from the litigation when the PMC refused its request to be given exclusive control of the action. When the firm withdrew, other counsel involved were forced to perform numerous services that Ashcraft Gerel already had performed. The Magistrate thus concluded that the firm "failed to discharge [its] burden when it decided to cease work on the case, thereby requiring other attorneys to duplicate its work." Agent Orange, 611 F.Supp. at 1367.

In adopting the Magistrate's recommendations, however, the district court offset the fee awarded to Ashcraft General against the benefits obtained by the firm's many opt-out clients from "the use of discovery materials assembled through the multidistrict discovery process and paid for by the class." Id. at 1343. The district court further found that the value of such services for the opt-outs far exceeded the firm's services to the class. Consequently, the court abrogated any fee award to the firm, but maintained the modified expense award.

While we find that the district court's award of fees and expenses prior to abrogation reflects fair and just compensation for Ashcraft Gerel's services to the class, we conclude that abrogation of the fee award constituted an abuse of discretion. In analyzing the general problem of individual use of discovery materials, the district court properly determined that, in return for the use of discovery materials obtained in the multidistrict litigation, such individual plaintiffs "could be assessed a reasonable fee, to be paid back into the fund as their fair share of the legal expenses assumed by the class." Id. at 1317. The court then suggested two ways in which this could be done. First, the court could require counsel in the opt-out cases to report to the district court any fee received from the opt-out plaintiffs so that the court could deduct the appropriate amount. Id. Second, the court could assess the opt-out plaintiffs for the cost of the discovery at the time they made use of it. Id.

Neither of these means of assessment permitted the court to offset Ashcraft Gerel's opt-out clients' payments for use of discovery materials against fees awarded to the firm for its representation of class members. The fee awarded the firm here has no relation to services performed for the opt-outs. Abrogation of the fee, therefore, has the net effect of relieving the class from its responsibility to pay Ashcraft Gerel fair and just compensation for services it provided, rather than assessing the opt-out plaintiffs for use of the discovery materials.

Accordingly, we conclude that Ashcraft Gerel should be awarded the fee that the district court, accepting the Magistrate's recommendation, determined to be fair and just.

2. Sullivan Associates

Sullivan Associates, a law firm primarily involved in the litigation during the early days of the action, challenges the district court's fee award on the ground that the court improperly determined that much of its work was not beneficial to the class. The district court awarded the firm $52,311 in fees and $20,573.08 in expenses. The court, upon the recommendation of the Magistrate, denied the firm's motion to supplement the award. The court found that the hours requested were excessive and that the firm had spent most of its time furthering the interests of its opt-out clients.

After reviewing the district court's calculations, we conclude that there was no abuse of discretion. The district court was in a much better position to determine whether the work performed by the firm benefitted the class. For the same reasons as given in section II(A)(4), supra, we find no basis upon which to question the district court's figures.

3. Australian Counsel

William T. McMillan, Ross V. Lonnie, Paul J. Davison, Roger L. MacLaren, and Michael S. Bigg, all Australian attorneys, appeal the district court's awards of fees and expenses. The district court awarded McMillan $3,650 in fees and $27,178.34 in expenses, Lonnie no fees and $3,055.93 in expenses, Davison no fees and $2,042.08 in expenses, MacLaren no fees and $3,683.39 in expenses, and Bigg $5,700 in fees and $22,561.76 in expenses. The basis for the challenge to these awards is that they do not adequately reflect the services that counsel performed for the class.

We again find no abuse of discretion. Appellants have given us no adequate reason to question the district court's calculations and we decline to do so.

4. Kraft Hughes

Kraft Hughes, a New Jersey firm peripherally involved in the litigation, challenges the district court's award. The court awarded the firm $2,425 in fees and $3,935.48 in expenses. The firm now argues that this is no more than the out-of-pocket costs of its involvement and substantially undercredits its contribution to the litigation. Moreover, the firm contends that it was improper for the district court to abrogate the contingency fee agreements that the firm had with a number of class members.

Kraft Hughes concedes in its presentation to this court that it cannot establish the factual findings of the district court to be clearly erroneous. Consequently, the firm bases its appeal primarily on the ground that its fee agreements with its clients, as a matter of law, should not have been abolished. We find this argument, however, to be without merit.

It is well established that a district court, pursuant to its rulemaking authority or on an ad hoc basis, may review a contingency fee agreement. Boston and Maine Corp. v. Sheehan, Phinney, Bass Green, P.A., 778 F.2d 890, 896 (1st Cir. 1985); Dunn v. H.K. Porter Co., 602 F.2d 1105, 1108 (3d Cir. 1979). When dealing with an equitable fund action, "the court has an even greater necessity to review the fee agreement for [Fed.R.Civ.P. 23(e)] imposes upon it a responsibility to protect the interests of the class members from abuse." Dunn, 602 F.2d at 1109. That is exactly what the district court did here in requiring counsel, prior to receiving fees from the settlement, to certify that he or it had retained no fees or expenses from any class members. We find no basis to overrule the district court's decision in this regard.

III. CONCLUSION

To summarize: we affirm the district court's utilization of national hourly rates and conclude that they may be used in the circumstances revealed here. We further affirm the district court's award of quality multipliers to various counsel, and the district court's denial of risk multipliers. We affirm the district court's decision regarding hours credited and expenses reimbursed to the PMC. We reverse the decision to offset Ashcraft Gerel's fee against the use of the multidistrict discovery materials by the firm's opt-out clients and order the reinstatement of the previously approved fee without allowance for a risk multiplier. As to all other aspects of the district court's decision respecting attorneys' fees, we affirm.