This is the second time that this suit for the award of attorneys' fees in a civil rights action has been before us. In our first review, we affirmed a jury award against prison officials stemming from their use of excessive force, false reporting, and suppression of communication against prisoners. Furtado v. Bishop, 604 F.2d 80 (1st Cir. 1979) ("Furtado I "). Among the issues that we considered was the district court's decision to calculate the amount of attorney's fee award under 42 U.S.C. § 1988 by dividing the plaintiffs' damage recovery by two. There we held that "(b)ecause the fifty percent of recovery formula ignores time and labor spent, as well as other factors, it cannot stand." Id. at 98. We remanded despite the trial court's alternative finding that counsel had spent $20,000 worth of work on the case "timewise" since summary announcement of this total gave no clue as to its supporting reasoning. We specifically requested the district court's views, taking into account the factors identified in King v. Greenblatt, 560 F.2d 1024 (1st Cir. 1977).

The district court, 84 F.R.D. 671, in a substantial opinion articulating its thought processes, arrived at the same result: an award of $13,750, being one half plaintiffs' recovery of $27,000.1 It began by commenting specifically on the dozen criteria that King identified, based on those originally set forth in Johnson v. Georgia Highway Express, Inc., 488 F.2d 714 (5th Cir. 1974).2 The court then considered the request of plaintiffs' counsel for $24,956 for work in the district court. After criticizing the "presumptively overlarge" month's time spent in commencing suit, and the presence of two lawyers at trial, the court allowed $3000 for services in 1970-1971 and $12,000 for services in 1977-1978. Considering counsel's service on appeal, as to which 200.8 hours were reported for three lawyers, the court felt that (absent any ceiling on total fees), $5000 for the brief and $1000 for preparation and oral argument was reasonable.

Although the reasonable rates for time reasonably spent at trial and on appeal, in the court's view, added up to $21,000, the court felt obliged to limit its total award to one half the $27,500 damages awarded by the jury.3 Its careful explication may be summarized as follows. The court began by asking "whether there should be a ceiling on the fee in a suit such as this, brought for money damages, not to establish some principle or to serve as a public warning beyond the damages themselves." It noted that had this not been a civil rights case, the customary contingent fee understanding would be a fee of one third the recovery. Because counsel did considerable work and because a fee award of one half of what plaintiffs actually receive is in a sense one third of the "recovery", the court adopted its "one third rule". It justified this in part on the criticism that lawyers might attract were they to receive fees exceeding one half of their clients' receipts and in part on the thought that the fee must be appropriate to the case.4 Moreover, presumptively, any services rendered in prosecuting a successful appeal would be included in the one third of total recovery.

In establishing its one third rule or fifty-percent-of-recovery ceiling as a rule of law, the district court recognized that the legislative history of section 1988 "indicates that Congress' concern was that persons whose civil rights had been violated might be deterred from suing by the high cost of representation. The fee award is to be 'what it costs them to vindicate these rights in court.' " (footnote omitted) (internal quotation from Sargeant v. Sharp, 579 F.2d 645, 648 (1st Cir. 1978), quoting S.Rep.No.94-1011, 94th Cong., 2d Sess. 2 (1976), reprinted in (1976) U.S.Code Cong. & Ad.News, pp. 5908, 5910). Nonetheless, the trial court did

"not read this language to mean that Congress intended that every person who could claim an invasion of civil rights, no matter how minor, would be entitled to retain counsel at the defendant's expense, no matter how great. There must be an element of reason; the fee must not be such as to encourage the overpressing of marginal claims, a by no means idle fear."

The district court established its fifty percent ceiling on fees to effectuate this "element of reason". The court recognized that the ceiling would not apply where damages were nominal, or if only injunctive relief was sought, or where a large recovery was out of proportion to work done, or if the appeal had been frivolous. Finally the court left open the possibility that in a case involving less work it "would consider one third to mean one third of the verdict, not one third of the verdict plus the fee."

We can understand how the trial court's concern to assure a correspondence between the dollar recovery and attorney's fees in cases where damages are the sole remedy may have seemed a viable course. One of the twelve factors endorsed in King, supra, is "the amount involved and the results obtained." 560 F.2d at 1027. In the absence of countervailing considerations, this factor might be read as consistent with a formula that enforces the relationship between damage award and attorneys' fee recovery by means of a percentage ceiling. Compare Perez v. University of Puerto Rico, 600 F.2d 1, 2 (1st Cir. 1979) (permitting consideration of the fact that only nominal damages were awarded) with Copper Liquor, Inc. v. Adolph Coors Co., 624 F.2d 575, 583, 584 n. 16, No. 78-3209, slip op. at 8581-82 & n. 16 (5th Cir. 1980) (modest damage recovery "should not be ignored" but "should not control an attorneys' fee award"). But deciding that an attorneys' fee award is "excessive" by means of an automatic, controlling percentage comparison with the damages recovered is another way of stating that the damage recovery in that case is too small to support such a fee award. We believe that casting this relationship in terms of a percentage ceiling discounts one key object of the legislative intent behind § 1988: the goal of encouraging private enforcement of civil rights law.5

Under the traditional "American rule", successful plaintiffs must bear the expenses of vindicating their rights. Alyeska Pipeline Service v. Wilderness Society, 421 U.S. 240, 95 S. Ct. 1612, 44 L. Ed. 2d 141 (1975). Thus plaintiffs typically will not act to redress injuries unless the expected recovery exceeds the expected cost of so acting.6 Legal services are very expensive. Difficult doctrinal questions can necessitate extensive analysis and research, while application of legal principles requires factual investigation and proof a process that can be very time consuming and hence costly even when the ruling law is settled and straightforward. Consequently it would not be sensible under the American rule to pursue suits involving real and reprehensible invasions of civil rights if they promise only modest recovery or a highly uncertain chance of substantial recovery. By amending section 1988, Congress chose in the case of certain important rights to alter this pattern of prohibitively costly vindication. Congress thus deliberately sought to provide for suits that would be unlikely to occur under the regime of the American rule. It therefore is precisely the civil rights lawsuits whose prospect of modest recovery would not justify the expense of a difficult or acrimonious legal fight the "marginal" suits, in the words of the district court-that Congress intended to make practicable.7

The district court proposed the ceiling for cases "brought for money damages, not to establish some principle or to serve as a public warning beyond the damages themselves." This reading of section 1988 would finance cases that create but not cases that apply legal rules. But pathbreaking holdings that will not be enforced are of limited public value. And the "principle" of enforcement is served by suits that "merely" seek damages. See Knight v. Auciello, 453 F.2d 852, 853 (1st Cir. 1972) (per curiam). We therefore believe that the district court's proposal-even were it supported by the statute's intent-would be of questionable wisdom, as well as conceptually difficult to administer. Furthermore, as we have said in Perez v. University of Puerto Rico, supra, 600 F.2d at 2:

" '(I)n authorizing awards of attorneys' fees to plaintiffs in civil rights actions Congress was concerned with enforcement not only of the civil rights of the public at large and of identifiable groups but also with the rights of individual plaintiffs. Its goal was to remove financial impediments that might preclude or hinder "private citizens," collectively or individually, from being "able to assert their civil rights," Senate Report.... We therefore reject the view that, to be eligible for shifting of attorneys' fees, the civil rights plaintiff is obligated to show that his action resulted in direct benefits to others, rather than in benefits solely to himself.' "

Perez v. University of Puerto Rico, 600 F.2d 1, 2 (1st Cir. 1979) (quoting Zarcone v. Perry, 581 F.2d 1039, 1042 (2d Cir. 1978)). Cf. King v. Greenblatt, 560 F.2d 1024, 1026 (1st Cir. 1977) (rejecting by implication an argument that a smaller attorney's fees award was proper because the case was not "of truly exceptional public importance"); Gibbs v. Town of Frisco City, 626 F.2d 1218, 1221 (5th Cir. 1980) (principles from recent Supreme Court cases "compel the conclusion that classification of a suit as the equivalent in some way of a state wrongful death action is not a special circumstance sufficient to justify denial or reduction of an otherwise proper award of attorney's fees").

The district court's opinion evinces a laudable desire to guard against meritless civil rights suits and undeserved attorney's fees. But these desirable ends can be furthered by means that do not discourage litigation of all but the egregious or easily proved actual civil rights invasions. Of central importance is the fact that the statute provides that only "prevailing part(ies)" can be awarded fees. Attorneys relying on such a contingent fee award thus will be disciplined by their own lack of judgment if they undertake to represent clients with no reasonable chance of victory or if they devote effort out of reasonable proportion to the odds of successfully establishing a right to relief. Moreover, meritless suits are subject to the levying of attorney's fees against those who sue on a "frivolous, unreasonable, or groundless" claim, Christianburg Garment Co. v. EEOC, 434 U.S. 412, 422, 98 S. Ct. 694, 700, 54 L. Ed. 2d 648 (1978), or in bad faith. Roadway Express, Inc. v. Piper, --- U.S. ----, ----, 100 S. Ct. 2455, 2464, 65 L. Ed. 2d 488 (1980); Sargeant v. Sharp, 579 F.2d 645, 647 & n. 2 (1st Cir. 1978); S.Rep.No.94-1011, 94th Cong., 2d Sess. 5 (1976), reprinted in (1976) U.S.Code Cong. & Ad.News 5908, 5912 ("This bill thus deters frivolous suits by authorizing an award of attorneys' fees against a party shown to have litigated in 'bad faith' under the guise of attempting to enforce the Federal rights created by the statutes listed in S. 2278."). See also Nadeau v. Helgemoe, 581 F.2d 275, 279 n. 3 (1st Cir. 1978) (discussing case in which any award of attorney's fee would be inappropriate).

The goal of avoiding awards of undeserved fees is, we think, better advanced by close and systematic scrutiny than by special formulae such as the one half of recovery rule whose application at best involves highly subjective judgments and at worst permits radically disparate fee awards. While broad discretion and subjective views will necessarily weigh significantly in fee scrutiny, perhaps the most hopeful approach to date lies in the approach developed by the Third Circuit in Lindy Bros. Builders, Inc. v. American Radiator & Standard Sanitary Corp., 3 Cir., 487 F.2d 161 (1973), and Lindy Bros. Builders, Inc. v. American Radiator & Standard Sanitary Corp., 3 Cir., 540 F.2d 102 (1976) (en banc), and most recently endorsed and applied in Copeland v. Marshall, --- F.2d ----, No. 77-1351 (D.C.Cir.1980) (en banc). See also Copper Liquor, Inc. v. Adolph Coors Co., 624 F.2d 575 at 583 n. 15 (5th Cir. 1980).

This approach recognizes that commenting on the twelve factors identified in Johnson v. Georgia Highway Express, Inc., may not in any real sense contribute to the rational setting of a fee; the comments are imprecise and the items overlap.8 In an effort to develop a useful analytical framework that can be applied by trial courts in all cases and can also lend itself to meaningful review, the Third and D.C. Circuits recognize two levels or steps in analysis. The starting point is to calculate the "lodestar": "The number of hours reasonably expended multiplied by a reasonable hourly rate." Copeland, at ----. This would involve separating out work done in relation to a firm's hierarchy, from senior partner to junior associate (and, we would add, including work that was or ought to have been assigned to a non-lawyer); eliminating time beyond that consistent with a standard of reasonable efficiency and productivity;9 and, after receiving documentation and possibly holding a hearing, assigning appropriate hourly rates for the kinds of work done by those at different levels of expertise. This results in a "lodestar" fee that then is adjusted upward or downward to reflect the contingent nature of any fee (if such is not reflected in the hourly rate), delay in payment, quality of representation (i. e., an unusually good or poor performance above or below the skill already reflected in the hourly rates), exceptional (and unexpected) results obtained, etc. For a more detailed discussion, see Copeland, at ---- - ----.

We therefore reject the district court's one-third-of-total-recovery rule. We recognize that the amount recovered may be considered, but only as one factor among many others. A one-third-of-total-recovery rule may not be used either as a ceiling or as a floor for the award.

The trial court has not made a finding guided by the rule stated in Part I, supra. Rather, the trial court has determined (a) that $21,000 would be an appropriate attorneys' fee if it were to be calculated on the basis of factors other than its relation to the amount of recovery, and (b) that in this case the factor of relation to the amount of recovery should be applied to produce a form one-third-of-total-recovery ceiling. Whether the trial court would, in view of our holding, allow $21,000, or some amount between $15,750 (i. e., $13,750 plus $2,000) and $21,000, remains undetermined. In these circumstances, we must consider whether to remand to the trial judge for a determination in light of our holding or instead proceed ourselves to determine an appropriate award in this case. Ordinarily the amount of an award of attorneys' fees is to be determined by the trial court and the role of an appellate court is to review for errors of law or abuse of discretion. In the distinctive circumstances of the present case, however, we conclude that, in the interests of expediting final disposition of the issue of attorneys' fees, already twice considered below and twice before this court, it is appropriate that we proceed to determine the appropriate award of attorneys' fees in this case. Consistently with our general practice of reviewing awards of attorneys' fees only for error of law or abuse of discretion, however, we will accept the district court's findings on the hourly compensation as our "lodestar" figure, absent an identified reason for not doing so. We now review the evidence, aided by the district court's findings on the factors identified in King v. Greenblatt, and guided, although not in any rigid sense, by the "lodestar" approach of Lindy I and II and Copeland.

Plaintiffs' counsel's work in the district court was performed during two time periods, one remote and one recent. The first, in 1969-1971, was so long ago that only estimates of time spent were available. The total time estimated was 212 hours and Attorney Stern, then a young salaried attorney with a legal services group, was paid at the rate of $23 an hour. This is the basis of the present request. The court commented that the time seemed "presumptively over-large" for commencing suit and concluded that $3000 instead of the requested $4876 was fair for this segment. For the 1977-1978 period of preparation, trial, and post-trial brief, Attorney Stern sought compensation at $100 an hour for 141.5 hours or $14,150, and Attorney Avery sought $100 an hour for 59.3 hours or $5930. The district court took exception to the adding of a second attorney for trial, and particular exception to a claim by both attorneys for $2000 of time spent waiting for a verdict. It nevertheless allowed $12,000. Coincidentally, the court added, this approximated what the total time spent would bring at the attorneys' customary billing rate of $60 an hour ($60 X 200.8 hours = $12,048). The court, however, said that it had factored into the hourly figures a modest amount to reflect the contingent nature of success, had allowed something more than $60 for Attorney Stern, but had reduced Attorney Avery's hours and rate.

While it is impossible to trace the actual adjustments made by the court, we determine that its final, pre-ceiling adjustment figure of $15,000 for district court services is justified and fair. Just as the court in Copeland, at ---- -----, felt able to accept the final result of the district court even though its "calculations do not precisely conform to the procedures identified in earlier cases and elaborated upon in this opinion" so do we under these circumstances feel comfortable with $15,000 as the "lodestar" figure for services in the district court.10

When we analyze the alternative fee computations relating to the appeal, we have a different problem and a different solution. We are not able to adopt the reasoning of the district court, and we conclude that the appellate component of the fee award on the record before us, including, of course, the briefs submitted, see Souza v. Southworth, 564 F.2d 609 (1st Cir. 1977), should be somewhat larger than was awarded by the district court.

The total time claimed by three attorneys for plaintiffs' 90 page brief on appeal was 216 hours at $100 an hour. The district court does not give us a basis for its judgment other than to say that (1) although there were some close questions of law, the facts were important, and defendants had filed a lengthy brief, nevertheless plaintiffs' brief might be thought excessive; (2) although persuading the appellate court on the merits appears not to have been easy, plaintiffs lost a large amount of their recovery below on the prejudgment interest issue, which occupied only a small part of their brief; and (3) a party should not have to pay more than $5000 for a brief "unless it had gained in value by being in support of a much larger verdict."11 Not only do we lack a basis for assessing the validity of the final figure but the court's concluding comment, in effect imposing a maximum tied to the monetary recovery, leaves the result vulnerable for the same reasons we have expressed in rejecting the one-third rule.

Conducting, therefore, our own review of the appellate briefs we begin with the premise that successful appellate work should be compensated according to the same principles that govern section 1988 fee awards for trial work. See Lund v. Affleck, 587 F.2d 75, 77 (1st Cir. 1978); Bond v. Stanton, 630 F.2d 1231, 1235-1236 (7th Cir. 1980). Special limitations on appellate compensation could introduce unsavory incentives if civil rights attorneys could be forced into protracted and uncompensated efforts to defend awards for their trial victories.

As we have stated, the first inquiry in determining fee awards requires that a "lodestar" amount be figured. Turning initially to a determination of the reasonable number of hours to be compensated, we observe the plaintiffs to make the following requests:

Mr. Stern    122.35 hours
Mr. Avery     43.20 hours
Mr. Shapiro   50.75 hours
             ------------
             216.30 hours

This hourly request is documented by affidavits giving a general breakdown of hourly efforts and based on contemporaneous time records. In general, these records are presented in commendable detail, but we do note two problem areas. First, there are some ambiguities and inconsistencies between the records of the three lawyers. For example, Mr. Stern makes one three hour entry labeled "Conf. G. Sousa and travel." Since we are disinclined to compensate an attorney at professional rates for travel time-especially when it is conceivable that other billable work may sometimes be conducted during this time-this aggregation leaves us with no indication of the amount of billable time actually spent in conference. Similarly, there are inconsistencies between the different attorneys' recorded conference times with one another.12 A total of 5.5 hours is billed in this inconsistent or ambiguous fashion. We exclude this time, leaving a total of 210.8 hours.

Second, the plaintiffs claim that their tripartite authorship of the brief involved "virtually" no duplication of research and writing, "since the legal issues were easily divisible." But the submitted records do not indicate how responsibility for issues in the brief actually was divided, if indeed it was split along "easily divisible" lines. We believe that convincing description of the division of labor must accompany reports of contemporaneous or identical work performed by several attorneys if deductions for duplication are to be avoided. In this respect the plaintiffs' records are wanting and their claim is weakened. Cf. King v. Greenblatt, 560 F.2d 1024, 1027 (1st Cir. 1977) (requiring "full and specific accounting").

This reporting problem is compounded by a more substantive issue. Five and one quarter person-weeks-one tenth of a lawyer's year-is a large amount of time to spend on one appellate brief. This is especially true in light of the attorneys' statement that here an "extensive amount of research, cite checking, proof-reading and other similar tasks necessary for preparation of the appellate brief were performed by two law students...."13 This suggests that a substantial cut in the total claim might be necessary to reach a reasonable sum for this amount of work.

On the other hand, the plaintiffs were required to respond to a 75 page brief by defendants that attacked the trial judgment on eight separate points. The plaintiffs responded with a thoroughly researched and well organized brief that cited 127 cases and extended over 93 pages. The appellate panel's treatment of the case divided it into four broad areas. The first was the propriety of recovery for the plaintiffs' segregated time in prison. This question combined relatively straightforward procedural, damage, and evidentiary issues together with difficult constitutional problems. In the latter regard, the panel noted that while the "plaintiffs' theory that recovery for segregated confinement could also be based upon the Eighth and Fourteenth Amendments is not commonplace and widely accepted", the panel "certainly could not say that plaintiffs' theory of recovery for segregated confinement was plainly erroneous", especially "(g)iven the state of the law in this area ...." 604 F.2d at 88.

The second area that the panel considered was a group of evidentiary issues, one of which presented a very close question. On this matter-the issue of the admissibility of the "Cross affidavit", see id. at 90-93-the court expressed some "reservations" about the trial court's decision. But the court ultimately upheld the decision on reasoning that tracked closely the argument presented in the plaintiffs' brief. The other four specific issues in this area also were resolved in favor of the plaintiffs, although "with greater dispatch." Id. at 93.

The third area was comprised of challenges to two jury instructions. The court held for the plaintiffs on both questions, indicating by the tone of its reasoning that it considered these issues to be no more than mildly difficult. Once again the court employed arguments advanced by the plaintiffs.

The fourth area was the question of prejudgment interest and attorneys' fees. The plaintiffs encounter their first appellate loss on the prejudgment interest issue. The pecuniary result of this was significant; the total award was decreased by about one quarter. However, the court acknowledged that this was an uncharted area, and that its ruling broke new ground. See id. at 97. Furthermore, the plaintiffs devoted only five pages of their brief to this question, and those five pages presented succinct and reasonable advocacy on the point.14 Finally, the plaintiffs were successful in convincing the appellate panel that the trial judge had erred in adopting a fifty percent formula for calculating attorneys' fees. Indeed, it was the remand for recalculation of fees that led to the present appeal.

This review convinces us that the plaintiffs' appellate brief presents no problem of supererogation.15 We consequently have no occasion to consider the appropriate response to advocacy that engaged in efforts beyond those required by the demands of reasonably diligent representation. But while we believe the product of the appellate effort was appropriately tailored to the demands of the task at hand, the amount of time claimed to create this product still troubles us. In our judgment, the number of hours claimed should be reduced by one half to reflect possible duplication of effort and diseconomies of committee authorship that may occur in a case where there is no demonstration to the contrary. We thus decide that a total of 105.4 hours is a reasonable amount of time to use to calculate the lodestar figure.16

We next determine a reasonable hourly rate. The plaintiffs originally requested the district court to award at $100 an hour. This request was accompanied by affidavits and copies of orders establishing that these attorneys previously have been awarded fees figured from rates in the range of $60-75. Before this court, plaintiffs have revised their request to "rely ( ) upon a $77-80 hourly rate" as "reasonable."

We decide that a $75 an hour rate is reasonable for the effort that had been displayed in the appeal at issue. This results in a lodestar figure of $7905.

The second step suggested by the Copeland approach is to adjust the lodestar for factors that it does not yet include. We view this step as a residual category that is relevant only when the parties point out factors that do not easily fit into the initial calculation of the lodestar. In this regard, we have no indication that the affidavit figures produced by the plaintiffs reflect any such justifications for departure from the lodestar norm as work of lower quality or a less contingent nature than the appellate work we evaluate here. We therefore see no need to make further adjustments in this $7905 sum. See Copeland, slip op. at 22 (burden of justifying deviation from the lodestar rests on the party proposing the deviation).

Plaintiffs thus are entitled to a total attorneys' fee award of $22,905, in addition to the uncontested $1000 allowed by the trial court for preparation and delivery of oral argument and the unconstested $2000 allowed for opposition to the petition for certiorari.

So ordered.