POSNER, Circuit Judge.
Three deputy sheriffs (policemen) of Crawford County, Wisconsin brought suit against the County charging that its mandatory retirement age of 55 for policemen violated the Age Discrimination in Employment Act of 1967, as amended, 29 U.S.C. §§ 621 et seq. 558 F.Supp. 1175 (W.D. Wis.1983). After a bench trial, the district judge gave judgment for the plaintiffs on their age discrimination claim (while dismissing their other claims and the other defendants), ordered the plaintiffs reinstated with back pay of $115,000 (after subtracting the other wages they had earned in this period, unemployment compensation they had received, and the extra retirement benefits they had gotten by being forced to retire at age 55), and awarded them attorney’s fees of $29,000. The County has appealed, raising some novel questions, chiefly about the Act’s two statutes of limitations; and the plaintiffs have cross-appealed, arguing that they should have received double damages because the defendants’ violation of the Act was willful, that they are entitled to prejudgment interest on the back-pay awards, and that the award of attorney’s fees was too small. Some of these arguments, too, raise difficult questions.
The County makes two significant challenges to the finding that it violated the Act: it argues that the suit is barred by either or both of the Act’s statutes of limitations (unless both are satisfied, suit is [1194]*1194barred), and that being under 55 is a bona fide occupational qualification for a Crawford County police officer. A third argument, that the Eleventh Amendment immunizes counties from having to pay federal money judgments, has no possible merit after Mackey v. Stanton, 586 F.2d 1126, 1130-31 (7th Cir.1978). The purpose of the Eleventh Amendment is to shield state treasuries from federal court judgments, and the financial interdependence between state and county government is not great enough to make the County an arm of the state for purposes of applying the Eleventh Amendment. Although Mackey dealt with counties in Indiana, not Wisconsin, Wisconsin counties appear to have a similar relationship to their state government. See Wis.Stat. § 59.01(1). And all this assumes that the Eleventh Amendment is applicable to cases under the Age Discrimination in Employment Act. It is not, because the Act was an exercise of Congress’s powers under section 5 of the Fourteenth Amendment. See EEOC v. Elrod, 674 F.2d 601, 609 (7th Cir.1982); Ramirez v. Puerto Rico Fire Service, 715 F.2d 694 (1st Cir.1983).
The Age Discrimination in Employment Act requires that charges be filed with the Equal Employment Opportunity Commission within 180 days after the alleged violation occurs. 29 U.S.C. § 626(d). The plaintiffs filed charges within 180 days of the dates on which they were taken off the payroll because they had reached the mandatory retirement age, but (with the exception of plaintiff Brown) more than 180 days after they received individual notice that they would be let go upon reaching that age. After Chardon v. Fernandez, 454 U.S. 6, 102 S.Ct. 28, 70 L.Ed.2d 6 (1981) (per curiam), there can be very little doubt that the 180 days run from the date of notice, not discharge. Although Char-don involved a law of Puerto Rico rather than the Age Discrimination in Employment Act, all the Justices who wrote in Chardon, and all subsequent lower-court decisions, have assumed that its holding is applicable to the statutes of limitations in other federal employment-discrimination statutes. Mogley v. Chicago Title Ins. Co., 719 F.2d 289, 290 (8th Cir.1983), for example, cites Chardon en route to the conclusion that notice rather than discharge lights the 180-day fuse in the Age Discrimination in Employment Act. To the same effect see Price v. Litton Business Systems, Inc., 694 F.2d 963, 965 (4th Cir.1982); Leite v. Kennecott Copper Corp., 558 F.Supp. 1170, 1172 (D.Mass.), aff’d without opinion, 720 F.2d 658 (1st Cir.1983), and Elliott v. Group Medical & Surgical Service, 714 F.2d 556, 563 (5th Cir.1983) (dictum).
Admittedly there is a paradox in applying a notice standard to a case challenging a mandatory retirement age. The defendants in such cases could argue that the statute of limitations began to run on the day the employee started work; for, as long as he is told then what the terms of his employment are, that is the day on which he receives “notice” that he will be unlawfully terminated. And maybe, on this view, if the employee learned of those terms before he signed on, the statute of limitations would begin to run before he began to work. Such arguments, if accepted, would unduly constrict employees’ ability to enforce their rights under the Act. People do not want to begin their employment by suing their employer over a mandatory retirement age that is 30 years in the future for them; and in those few cases that were brought, the courts would be deciding disputes prematurely. And yet the argument is a plausible corollary to the idea that notice sets the statute of limitations running. But no court has pressed the logic of the notice approach so relentlessly, or is likely to do so; and we merely want to emphasize that, merely because a specific notice of termination, such as the plaintiffs in this case received, starts the 180-day statute of limitations running, it does not follow that the notice an employee receives when he is first hired would also set the statute to run; it surely would not.
Although two of the plaintiffs did receive specific notice of their impending termination more than 180 days before [1195]*1195they filed their charges, they filed within 300 days, which was in time if Wisconsin is a “deferral” state, 29 U.S.C. § 626(d)(2), defined as “a State which has a law prohibiting discrimination in employment because of age and establishing or authorizing a State authority to grant or seek relief from such discriminatory practice,” 29 U.S.C. § 633(b). Wisconsin does have an age discrimination law, Wis.Stat. §§ 111.31 et seq., and an agency empowered to enforce it, see Wis.Stat. §§ 111.33, 111.36 (now §§ 111.-375, 111.39). The problem is that the law as it read when the plaintiffs were forced to retire provided: “The prohibition against discrimination because of age shall not apply to hazardous occupations including, without limitation because of enumeration, law enforcement or fire fighting.” Wis. Stat. § 111.32(5)(e). This certainly appears to exempt the County, especially when the wording of the statute is compared to that of its successor (not involved in this case), which exempts “employment in which the employe is exposed to physical danger or hazard, including, without limitation because of enumeration, certain employment in law enforcement or fire fighting.” Wis. Stat. § 111.33(2)(f) (emphasis added). Since the reason for giving plaintiffs extra time to sue in deferral states is that the federal act requires anyone complaining about age discrimination in such states to begin state administrative proceedings as a precondition to bringing a federal suit, see 29 U.S.C. § 633(b); Oscar Mayer & Co. v. Evans, 441 U.S. 750, 753, 99 S.Ct. 2066, 2070, 60 L.Ed.2d 609 (1979), and since initiating such a proceeding would have been an act of utter futility here, the County has an argument that the 300-day limitation period does not apply.
But not a compelling argument. The bar of the statute of limitations ought not depend on something so uncertain as whether a particular type of age discrimination is within or without the state’s age discrimination law; it should be enough that there is a law at least potentially applicable to the employer (as was not the case in Keitz v. Lever Bros. Co., 563 F.Supp. 230, 234 (N.D.Ind.1983)), and an agency charged with enforcing it (as was not the case in Eklund v. Lubrizol Corp., 529 F.2d 247, 249 (6th Cir.1976)). Otherwise, just to know how much time he had in which to bring a federal suit, a potential plaintiff would have to conduct research, sometimes in great depth, on the state law. True, a potential plaintiff may have to do some research just to know whether he is in a state that has an age discrimination law and an agency with enforcement powers, though state law may — Wisconsin law does, see Wis.Adm.Code § Ind. 88.21 — require employers to notify employees of their rights under the state’s age discrimination law. But under the view pressed by the County the cautious plaintiff in any state would feel he had to file within 180 days, since he could not be positive whether his claim would come within the scope of his state’s law and a mistake would be disastrous. The 300-day provision would be read out of the statute.
Moreover, the purpose of giving a state that has an age discrimination law a chance to respond to any age discrimination claim is to encourage the states to eliminate age discrimination without federal judicial intervention, and is served even in cases such as this where it is pretty clear that the state agency cannot help the claimant. The filing of a claim may alert the agency to an unintended gap or ambiguity in the law. This may lead the agency to propose an amendment to the law, or may lead a state court (not necessarily in the same case) to eliminate the gap or ambiguity by interpretation. The exclusion of law enforcement officers from Wisconsin’s age discrimination law, even as the law read at the time of the plaintiffs’ termination, was not entirely free from ambiguity; a state court might have decided that only hazardous law enforcement jobs were intended to be excluded and might have gone on to find that these plaintiffs’ jobs were not hazardous. This opportunity would be denied the state if the defendants’ argument prevailed. For that argument implies not merely that the plaintiffs had only 180 days within which to file their charges with the [1196]*1196EEOC but also that they had no obligation to complain to the state — an obligation that exists only in deferral states. See 29 U.S.C. § 633(b).
The second statute of limitations applicable to this case, 29 U.S.C. § 255 (a part of the Portal-to-Portal Pay Act of 1947 incorporated in the Age Discrimination in Employment Act by 29 U.S.C. § 626(e)(1)), provides that every action “shall be forever barred unless commenced within two years after the cause of action accrued, except that a cause of action arising out of a willful violation may be commenced within three years after the cause of action accrued.” The district court found that the County’s violation was not willful; and although the plaintiffs have challenged this finding in their cross-appeal, we think it was right (as shall appear). So the two-year rather than the three-year statute of limitations is applicable (along with the 300-day statute), and it was exceeded. We reject the plaintiffs’ argument that the two-year statute was tolled by section 626(e)(2), which provides: “For the period during which the [EEOC] is attempting to effect voluntary compliance with the requirements of this chapter ... pursuant to subsection (b) of this section, the statute of limitations as provided in section 255 of this title shall be tolled, but in no event for a period in excess of one year.” The conciliation referred to — conciliation under section 626(b) — is that preceding suits by the EEOC. Conciliation in private actions is provided for separately in section 626(d). We do not attribute omniscience to legislative draftsmen; but it is hardly likely that in referring in section 626(e)(2) to conciliation only under subsection (b) they simply overlooked subsection (d), which is just a few brief sentences before (e)(2). The conference report dispels any doubt by stating with admirable directness that the tolling provision in 626(e)(2) “does not apply to conciliation required by section [626](d).” H.Conf.Rep. No. 950, 95th Cong., 2d Sess. 13 (1978), U.S.Code Cong. & Admin.News 1978, pp. 504, 534. See also Leite v. Kennecott Copper Corp., supra, 558 F.Supp. at 1173 n. 1, and references there. The point is not that there are different forms of conciliation; it is that only the EEOC can use the period of conciliation to toll the statute of limitations. The private plaintiff is bound to the statutory period. Although as an original matter one might criticize this result as discouraging the settlement of age discrimination claims, the language and history of the statute leave no room for the interpretation that the plaintiffs suggest.
But although the County may well have had a good defense at one time based on the two-year statute of limitations, it forfeited the defense in the district court and cannot revive it here. The answer to the complaint did plead statute of limitations as an affirmative defense, and though it did not indicate which statute or statutes of limitations barred the plaintiffs’ suit (all the answer said on the point was, “As and for a third affirmative defense the defendants allege that the complaint is barred by the statute of limitations”), that particular omission was not critical. See, e.g., Santos v. District Council, 619 F.2d 963, 967 (2d Cir.1980). The notice-pleading philosophy of the Federal Rules of Civil Procedure excuses lack of particularity at the pleading stage, with the exceptions listed in Rule 9, none of which is applicable to this case. But what is critical is that, as the case progressed, the County and the other defendants (who have since been dismissed out, as we have said) never mentioned the two-year statute of limitations, though they mentioned the 180/300-day period frequently. It is not referred to in the pretrial order, which “shall control the subsequent course of the action unless modified by a subsequent order.” Fed.R.Civ.P. 16(e). (And the order was not modified in any relevant particular.) It was first mentioned in the County’s brief in this court.
That was too late. An appellant in a civil case cannot raise a ground for reversal for the first time on appeal unless it goes to the district court’s subject-matter jurisdiction; and though there are exceptions to this as to virtually every le[1197]*1197gal rule, none of them is applicable here. See, e.g., Capitol Indemnity Corp. v. Keller, 717 F.2d 324, 329 (7th Cir.1983). It makes no difference if the ground is one abandoned in the district court rather than never raised there. See Oberlin v. Marlin American Corp., 596 F.2d 1322, 1325 n. 1 (7th Cir.1979). And a ground raised but not “pressed” in the district court is abandoned. See, e.g., King v. Stevenson, 445 F.2d 565, 571 (7th Cir.1971); Stanspec Corp. v. Jelco, Inc., 464 F.2d 1184, 1187 (10th Cir.1972). Even a ground “not clearly presented” to the district court will be treated as abandoned, at least in the absence of plain error. See Nathan Rodgers Construction & Realty Corp. v. City of Saraland, 676 F.2d 162, 163 (5th Cir.1982). And there can be no showing of plain error here; for, as we shall see, the plaintiffs were denied an opportunity to show that the statute of limitations was tolled by the defendants’ conduct.
Donovan v. Crisostomo, 689 F.2d 869, 874 (9th Cir.1982), a rare case where the defendants were allowed to raise the statute of limitations as a fresh issue on appeal, illustrates the limits of legal generalization but is distinguishable from the present case on three grounds. (Another such case, even more clearly distinguishable, is Val Decker Packing Co. v. Corn Products Sales Co., 411 F.2d 850, 851-52 (6th Cir.1969).) There was only one applicable statute of limitations in Donovan, so there could be no doubt that it was the statute that the defendants had tried to raise in the district court by pleading statute of limitations as a defense; the statute had been mentioned in the pretrial order; and there were no unresolved factual issues to prevent a definitive resolution at the appellate level. Here the two-year statute of limitations was mentioned, if at all, for the last time before appeal in the answer to the complaint (and maybe not even then, as the reference may have been to the 180/300-day statute). More important, had the plaintiffs known that this statute of limitations was in play they might have presented evidence in the district court to show that the defendants had tolled it, as in Ott v. Midland-Ross Corp., 600 F.2d 24 (6th Cir.1979), by conduct distinct from the conciliation efforts of the EEOC, which as we have said could not toll it.
We conclude that the case is not time-barred and move on to the question whether Crawford County’s mandatory retirement age is within the Act’s exception for cases “where age is a bona fide occupational qualification reasonably necessary to the normal operation of the particular business ....” 29 U.S.C. § 623(f)(1). Although there is no question that the duties of a Crawford County deputy sheriff — a police officer in a rural county — at times require a quick, strenuous, physical response in hazardous conditions, or that the capacity for such responses diminishes with age, an inverse correlation between age and job fitness is not enough to carry the employer’s burden of proving that youth is a bona fide occupational qualification. This court, laying stress on the statutory words “reasonably necessary,” held in Orzel v. City of Wauwatosa Fire Dep’t, 697 F.2d 743, 752 (7th Cir.1983) (a case that involved the application of a mandatory retirement age of 55 to an assistant fire chief), that it was not enough for the employer “to show that it made a ‘good faith judgment concerning the safety needs’ of its citizens,” and the court therefore rejected the argument that the employer’s “mandatory retirement policy should be upheld as long as that policy is ‘not the result of an arbitrary belief lacking in objective reason or rationale’ ____” If a good-faith judgment, based on a loose fit between the needs of the job and the effects of age, were enough to bring a mandatory retirement policy within the shelter of section 623(f)(1), the Act would have little bite. Older people in our society are not a despised and powerless minority; and mandatory retirement laws, rather than being manifestations of vicious, exploitive, or unreasoning prejudice, are for the most part efforts — crude and arbitrary perhaps but not malicious — to balance the costs of a fixed retirement age in sometimes forcing out perfectly fit employees with the bene[1198]*1198fits in avoiding individualized evaluations of fitness, evaluations both costly for the employer to make and potentially humiliating for the individual employee to receive. To exempt from the Act mandatory retirement policies just because they were adopted in good faith and struck a reasonable balance between the relevant costs and benefits would therefore deprive the Act of much of its intended effect in helping older people retain their jobs longer. The language of the statute, commanding that any mandatory retirement law under 70 be not only reasonable but reasonably necessary, is against such an interpretation, and Orzel rejected it. Orzel requires that the employer put in persuasive evidence that it needed to impose a mandatory retirement age on its employees (see 697 F.2d at 754; see also Monroe v. United Air Lines, 736 F.2d 394, 402 (7th Cir.1984); EEOC v. County of Los Angeles, 706 F.2d 1039, 1042 (9th Cir.1983)), as the employer did for example in Mahoney v. Trabucco, 738 F.2d 35, 37 (1st Cir.1984), where the First Circuit recently upheld mandatory retirement at 50 for uniformed state police in Massachusetts.
Therefore Crawford County cannot succeed merely by showing that this or other courts have on different evidentiary records upheld mandatory retirement ages of 55 or even less for jobs involving public safety. The unfortunate fact for the County is that the evidence that it put in was weak and that the district judge, whose finding that the age-55 rule was not reasonably necessary we review of course under the deferential “clearly erroneous” standard, was not persuaded by it. The County’s case rested very largely on the testimony of two witnesses. The first, a cardiologist, testified to what is anyway obvious, that the incidence of heart disease and other ailments that can affect one’s ability to respond effectively in the type of emergency situations that sometimes confront police officers rises with age. On cross-examination he acknowledged that he was not familiar with the duties of a Crawford County deputy sheriff. The second witness, an exercise physiologist, testified that it would take at least a year to develop tests that the Crawford County sheriff’s office could use to determine individual fitness. But on cross-examination he acknowledged that such tests already existed. These witnesses’ testimony was not much evidence that mandatory retirement at age 55 is “reasonably necessary” for the normal operation of the Crawford County sheriff’s department — certainly not such evidence as could warrant our overturning the district judge’s finding.
Moreover, a telling bit of evidence that mandatory retirement at age 55 is not “reasonably necessary” is that the Crawford County sheriff’s department does not require deputy sheriffs to take annual or other periodic physical examinations at any age, even though many people develop heart disease, spinal-disc disease, or other conditions that disable them from hazardous physical activities long before they reach 55. Of course these tests are very far from being infallible, and this fact can be used to support a mandatory retirement age. But it cannot be the only support. The fallibility of fitness tests is significant only in relation to the consequences of retaining a physically unfit employee. It is one thing for a pilot to collapse at the controls of a plane and another thing for a bookkeeper to collapse while entering debits in a ledger. Obviously the rural policemen in this case are somewhere in between these extremes, and we can use the fact that their employer does not require any physical examinations to help place them nearer one end or the other. Since such examinations are useful both in spotting existing physical disabilities and in identifying high-risk individuals, they are required by most employers who are seriously concerned with the physical fitness of their employees — airlines and the armed forces being the familiar examples. If Crawford County required such examinations it could argue (we need not decide with what success) that this showed the genuineness of its concern with the physical fitness of its deputy sheriffs; and this in turn would be an argument for a mandatory retirement [1199]*1199age since as we have said physical examinations are not infallible tests of continuing fitness. But the premise is missing; the County does not require examinations.
The County, hoping to get some help from the fact that 55 is the mandatory retirement age for federal law enforcement officers (see 5 U.S.C. § 8335(b)), presented evidence that the duties of Crawford County deputy sheriffs are similar to those of federal marshals. This means, it argues, that being younger than 55 must be a bona fide occupational qualification in this case, for otherwise Congress itself was not acting in good faith when in 1974, seven years after enacting the age discrimination law, it enacted the mandatory retirement law for federal law enforcement officers. This argument was accepted recently in Johnson v. Mayor of Baltimore, 731 F.2d 209 (4th Cir.1984), whieh involved a mandatory retirement age of 55 for firemen (the federal statute makes 55 the mandatory retirement age for federal firemen as well as federal law enforcement officers). But Orzel and a number of other cases cited by the Fourth Circuit in Johnson, 731 F.2d at 216 n. 22, reject the argument. See 697 F.2d at 749-50.
To describe the federal statute as requiring retirement at age 55 is not completely accurate. The statute provides: “A law enforcement officer or a firefighter who is otherwise eligible for immediate retirement under section 8336(c) of this title shall be separated from the service on the last day of the month in which he becomes 55 years of age or completes 20 years of service if then over that age. The head of the agency, when in his judgment the public interest so requires, may exempt such an employee from automatic separation under this subsection until that employee becomes 60 years of age.” 5 U.S.C. § 8335(b). So if the employee is not eligible for immediate retirement under section 8336(c), or has not put in 20 years of service, or is exempted by the head of his agency, he can continue working after he reaches 55. Moreover, to make the fact that the federal government has imposed a mandatory retirement age on its own law enforcement officers conclusive on the question whether such a retirement age is reasonable for state or local officers having similar duties would apply section 8335(b) far beyond its scope — would make a statute applicable only to federal officers applicable in effect to state and local officers as well.
The legislative history of section 8335(b), which explains the purpose behind the choice of age 55, may, it is true, be some evidence of what a “reasonably necessary” retirement age is for such officers. Legislative history is not a conventional form of “evidence”; but the age discrimination act is not a source of conventional legal duties. The determination whether a particular retirement age is reasonably necessary is sufficiently remote from the usual kinds of factual determinations courts are called on to make to justify the broadest possible conception of what is “evidence.” In suggesting that the evidentiary showing made by the County was inadequate we hope we will not be misunderstood to mean that its inadequacy lies in its being unconventional.
In the legislative history of section 8335(b) we read “that these occupations [federal law enforcement and fire control] should be composed, insofar as possible, of young men and women physically capable of meeting the vigorous demands of occupations which are far more taxing physically than most in the Federal Service. They are occupations calling for the strength and stamina of the young rather than the middle aged. Older employees in these occupations should be encouraged to retire.” S.Rep. No. 948, 93d Cong., 2d Sess. 2 (1974); see also H.R.Rep. 463, 93d Cong., 1st Sess. 2-4 (1973), U.S.Code Cong. & Admin.News 1974, p. 3698. This is helpful to the County, but not very; for the failure to require annual physical examinations of its deputy sheriffs rather belies any claim that it thinks the sheriffs department should be composed of “active, vigorous, physically capable men.” Id. at 3. Moreover, although the age-55 law applicable to federal law enforcement and firefighting personnel [1200]*1200was on the books when Congress in 1978 amended the Age Discrimination in Employment Act to raise minimum mandatory retirement ages from 65 to 70, the legislative history of the amendments is inconsistent with the proposition that the age-55 law had established a precedent of general applicability to law enforcement officers. The Senate Report states that “in certain types of particularly arduous law enforcement activity, there may be a factual basis for believing that substantially all employees above a specified age would be unable to continue to perform safely and efficiently the duties of their particular jobs, and it may be impossible or impractical to determine through medical examinations, periodic reviews of current job performance and other objective tests the employees’ capacity or ability to continue to perform the jobs safely and efficiently.” S.Rep. No. 493, 95th Cong., 2d Sess. 10-11 (1977), U.S.Code Cong. & Admin.News 1978, pp. 513-514. The County failed in this case to persuade the district judge that there was “a factual basis” for believing that “substantially all employees” aged 55 or older would be unable to perform their duties safely and efficiently because of the “particularly arduous” nature of the employment, and that medical or other objective tests to determine the capacity of older workers on an individual basis were infeasible.
In sum, and quite apart from the testimony given by the plaintiffs’ expert witnesses (one of whom, we note, gave rather incredible evidence — that'police officers ought to be allowed to work until they are 85), we think the district judge’s finding that being under 55 is not a bona fide occupational qualification for a Crawford County deputy sheriff is not clearly erroneous and must therefore be upheld along with his conclusion that the County violated the Age Discrimination in Employment Act by forcing the plaintiffs to retire when they reached 55. But we also agree with the district court that it was not a willful violation.
If it had been, the plaintiffs would be entitled to an award of liquidated damages equal in amount to their actual damages— or so at least the statutory language indicates. See 29 U.S.C. § 626(b), incorporating by reference section 16(b) of the Fair Labor Standards Act of 1938, 29 U.S.C. § 216(b). The inference from the language is strengthened by the fact that Congress, some years after enacting the Fair Labor Standards Act, adopted as part of the Portal-to-Portal Pay Act of 1947 a provision allowing courts to make a smaller award of liquidated damages in a suit under section 16(b) if the employer was acting reasonably and in good faith. See 29 U.S.C. § 260. This provision was not incorporated in the Age Discrimination in Employment Act. From this one can argue that the original section 16(b) (now 216(b)) gave the courts no discretion and that by not incorporating the ameliorative provision, section 260, Congress expressed its intention not to allow them that discretion in age discrimination suits either. Against this it can be argued that for all that appears the failure to incorporate section 260 was an oversight, and that in any event there is an element of unjust enrichment in a minimum-wage, child-labor, or maximum-hour case — the sort of ease to which the Fair Labor Standards Act applies — that is not found in most age discrimination cases.
Although the Fifth Circuit has held that there is discretion in an age discrimination case to award less than the full amount of liquidated damages provided for in section 216(b), Hays v. Republic Steel Corp., 531 F.2d 1307, 1311-12 (5th Cir.1976); Hedrick v. Hercules, Inc., 658 F.2d 1088, 1095-96 (5th Cir.1981); Elliott v. Group Medical & Surgical Service, supra, 714 F.2d at 558 n. 2, there is contrary authority, see Goodman v. Heublein, Inc., 645 F.2d 127, 129 (2d Cir.1981), and cases cited there, including a Third Circuit case that appears to overrule sub silentio an earlier Third Circuit case that had followed Hays, the leading Fifth Circuit case. Compare Wehr v. Burroughs Corp., 619 F.2d 276, 279 n. 5 (3d Cir.1980), with Rodriguez v. Taylor, 569 F.2d 1231, 1244 (3d Cir.1977). Two of our cases state, but without any elaboration, that an award of liquidated damages [1201]*1201under section 626(b) “essentially doubles” the award for back pay and benefits. Orzel v. City of Wauwatosa Fire Dep’t, supra, 697 F.2d at 757 n. 27; Syvock v. Milwaukee Boiler Mfg. Co., 665 F.2d 149, 156 (7th Cir.1981). These were observations made in passing, not holdings. The issue (rightly described as unsettled in Schlei & Grossman, Employment Discrimination Law 524 (2d ed. 1983)) is not directly presented in this case. But as we shall see when we come to discuss the issue of prejudgment interest, it has tangential relevance.
The County’s violation was not “willful” just because it knew it was forcing the plaintiffs to retire before the age of 70. The plaintiffs must show that the County should have known it was violating the Act. See Orzel, supra, 697 F.2d at 757-58; Syvock, supra, 665 F.2d at 156 and n. 10. With so many laws on the books requiring law enforcement officers to retire at 55, the County had no reason to think its mandatory retirement ordinance unlawful. It has been found unlawful only because the evidence that the County introduced at trial was not strong enough to persuade the trier of fact that the ordinance came within the exemption for jobs for which youth is a bona fide qualification. The County could not know when it discharged the plaintiffs in 1979 that it would not be able to gather enough evidence to put on a convincing case at trial. Moreover, until 1982, when this court decided Orzel, the County had reasonable grounds for believing that the law making 55 the mandatory retirement age for federal law enforcement officers would carry the day for them. And until last year, when the Supreme Court decided EEOC v. Wyoming, 460 U.S. 226, 103 S.Ct. 1054, 75 L.Ed.2d 18 (1983), it also had reasonable grounds for believing that the Age Discrimination in Employment Act might be held unconstitutional as applied to public officers engaged in law enforcement. Now it is true that in Orzel we upheld a finding of willful violation without reference to any of these sources of legal doubt, which were as much present in that case as they are in this. See 697 F.2d at 758-59. But there was much stronger evidence in Orzel than there is here that the employer had been advised by counsel that it was violating the law. And we were applying the clearly-erroneous standard to a finding that the violation had been willful, not as here to a finding that it had not been willful. The Age Discrimination in Employment Act does not require us to award liquidated damages — damages punitive in consequence though (as we shall see) not in purpose — of more than $115,000 against a small county whose total budget, we were told at argument, is only $3 million a year and whose total population is only 16,000, just because the County’s lawyers failed to steer a straight course through rapidly changing legal currents.
A related issue is whether the plaintiffs were entitled to prejudgment interest on the amounts awarded as back pay. Although the Age Discrimination in Employment Act is silent on the matter, we have held that the district judge has the discretion to award prejudgment interest or deny it. See Syvock v. Milwaukee Boiler Mfg. Co., supra, 665 F.2d at 162; see also Kelly v. American Standard, Inc., 640 F.2d 974, 982-83 (9th Cir.1981). The power to make such an award is derived by combining the statutory language empowering the court “to grant such legal or equitable relief as may be appropriate to effectuate the purposes of the” Act, 29 U.S.C. § 626(b), with the common law principle (on which see, e.g., General Motors Corp. v. Devex Corp., 461 U.S. 648, 103 S.Ct. 2058, 2060, 76 L.Ed.2d 211 (1983); Union Bank of Benton v. First National Bank, 677 F.2d 1074, 1077, 1079 (5th Cir.1982); Dobbs, Handbook on the Law of Remedies 165-66 (1973)) allowing prejudgment interest to be awarded where the obligation on which the defendant defaulted was a fixed sum. The obligation here was to pay the plaintiffs the wages to which they would have been contractually entitled if not forced to retire at 55.
[1202]*1202If liquidated damages had been awarded, it could be argued that an award of prejudgment interest would be duplicative. The Supreme Court held long ago that the liquidated-damages provision of section 216(b) is “compensation, not a penalty or punishment by the Government,” Overnight Motor Transport Co. v. Missel, 316 U.S. 572, 583, 62 S.Ct. 1216, 1223, 86 L.Ed. 1682 (1942), and this holding is still the law despite the subsequent enactment of section 260, Thompson v. Sawyer, 678 F.2d 257, 281 (D.C.Cir.1982). Among the items being compensated for is delay. Marshall v. Brunner, 668 F.2d 748, 753 (3d Cir.1982). Although these are cases under the Fair Labor Standards Act, their analysis seems applicable to the Age Employment in Discrimination Act. The only form of compensatory damages that the Act explicitly allows is back pay. This makes it plausible to think that the liquidated-damages provision was intended in whole or part to compensate victims of willful violations for other items of damage such as delay, rather than simply to punish the willful violator. A number of decisions accept this argument. See Blim v. Western Electric Co., 731 F.2d 1473, 1479-80 (10th Cir.1984) (per curiam), and cases cited there; Spagnuolo v. Whirlpool Corp., 641 F.2d 1109, 1114 (4th Cir.1981); cf. Brooklyn Savings Bank v. O’Neil, 324 U.S. 697, 715, 65 S.Ct. 895, 906, 89 L.Ed. 1296 (1945); contra, Criswell v. Western Airlines, Inc., 709 F.2d 544, 556 (9th Cir.1983). But liquidated damages were not awarded here and we do not assume that by creating a liquidated-damages remedy for willful violations Congress meant to deprive the district courts of all power to award prejudgment interest in cases of nonwillful violation. Cline v. Roadway Express, Inc., 689 F.2d 481, 489 (4th Cir.1982). When double damages are awarded, they usually will be far greater than would be necessary to compensate the plaintiff for delay — far greater, that is, than an award of prejudgment interest would be. Of course interest can mount up, but the short outside statute of limitations (two or three years, unless tolled) makes it unlikely that it will mount up to the point where it is equal to the principal.
The only problem with the analysis is that, as we noted earlier, under the age discrimination law it is not certain that a plaintiff who proves a willful violation will automatically get double damages, that is, will be awarded the maximum liquidated damages. If he gets less than the maximum it may well be less than the amount of prejudgment interest that would be awarded in a nonwillful-violation case, so that as a practical matter a plaintiff would get more compensation if he did not prove a willful violation than if he did prove one. Congress could not have intended this result. But maybe the way around it is to hold that if the judge can award less than double damages in a case of willful violation (an issue we need not decide, as there was no willful violation here), he can award prejudgment interest as well in such a case, provided that the total award does' not exceed double damages, and provided also that the liquidated-damages component is not intended to compensate the plaintiff for delay, since the award of prejudgment interest is designed to do that. Cf. McClanahan v. Mathews, 440 F.2d 320, 325 (6th Cir.1971). This seems a preferable solution to holding that there is no power to award prejudgment interest in any case of nonwillful violation. And all this assumes that an award of double damages is not automatic in a case of willful violation; many courts, as we have noted, think that it is automatic.
The only reason the district judge gave for not awarding prejudgment interest in this, a case of nonwillful violation, was that “the benefits of a later lump sum settlement which may be received and which can be invested offsets the diminished value of those dollars which may have been paid at an earlier date____” In other words, if the plaintiffs had not been discharged but had continued receiving their salaries, they would in all likelihood have spent all or most of those salaries on living expenses and would have had little or nothing left over to invest, while they can in[1203]*1203vest the large, lump sum that they will receive as back pay in this proceeding and the investment income will be a benefit that they would not have gotten if the defendants had not forced them to retire. This is an interesting point but it does not convince us without explanation. We do not know how the plaintiffs lived after they lost their jobs, though we do know that all three had some earnings from other jobs, some unemployment compensation, and some retirement benefits from the County — all these sources of income having been deducted in figuring the award of back pay. If the plaintiffs borrowed money to maintain their previous standard of living and the back pay award will go largely to retiring those loans, then the interest on the loans will be an expense that the plaintiffs would not have borne if they had not been discharged. And if they did not borrow, but let their standard of living drop, that is a cost to them for which the interest income on the lump-sum back-pay award that they are now to receive could be viewed as reasonable, and certainly not excessive, compensation.
The usual assumption in economics as in life is that a dollar today is worth more than a dollar tomorrow, both because it can be invested and earn interest and because the future is uncertain. On that assumption the plaintiffs have lost something by having to wait years to receive the salary payments that it now turns out the defendants owed them, even if they did not borrow during the interim. But maybe there is something special about this case that overrides that assumption; this will be a matter for the district court to explore on remand along with any other equitable factors that might argue for or against an award of prejudgment interest, such as the County’s finances. Unlike a violation of the minimum-wage law, this is not a case of withholding wages that are due to an employee for work actually performed; so there is not the element of unjust enrichment that persuaded the court in McClanahan v. Mathews, supra, 440 F.2d at 325-26, to adopt a strong presumption in favor of awarding prejudgment interest. Cf. Argonaut Ins. Co. v. Town of Cloverdale, 699 F.2d 417, 421 (7th Cir.1983); Myron v. Chicoine, 678 F.2d 727, 733-34 (7th Cir.1982). The County did not have the benefit of the plaintiffs’ work after forcing them to retire. All we hold, therefore, is that the judge has not yet adequately exercised his discretion in the matter of prejudgment interest.
We have left to decide only the attorneys’ fees issues. The plaintiffs asked for fees and disbursements (excluding costs allowable under 28 U.S.C. § 1920) of (in round numbers) $67,000. The judge excluded all disbursements and awarded 50 percent of the attorney’s fees requested, which made an award of $29,000, and the plaintiffs have appealed. The County concedes as it must that the plaintiffs, having prevailed on their age discrimination claim, are entitled to a reasonable attorney’s fee. See 29 U.S.C. §§ 216(b), 626(b). It also acknowledges the applicability to this question of the case law that has developed under the more frequently invoked Civil Rights Attorney’s Fees Awards Act of 1976, 42 U.S.C. § 1988. Although the Age Discrimination in Employment Act is not a civil rights act within the meaning of section 1988, age discrimination cases commonly cite section 1988 cases on fee questions, see, e.g., Vocca v. Playboy Hotel of Chicago, Inc., 686 F.2d 605 (7th Cir.1982) (per curiam); Spagnuolo v. Whirlpool Corp., supra, 641 F.2d at 1115; Cleverly v. Western Electric Co., 594 F.2d 638, 642-43 (8th Cir.1979), and as the County does not question this practice neither shall we. Since the oral argument of this appeal, this court, joining the other circuits that have considered the question, has held that expenses of litigation that are distinct from either statutory costs or the costs of the lawyer’s time reflected in his hourly billing rates — expenses for such things as postage, long-distance calls, xeroxing, travel, paralegals, and expert witnesses — are part of the reasonable attorney’s fee allowed by the Civil Rights Attorney’s Fees Awards Act. Henry v. Webermeier, 738 F.2d 188, 191-192 (7th Cir.1984). As we noted in that decision, any other conclusion would be arbitary, since a lawyer’s billing rate includes [1204]*1204expenses that are indistinguishable from those that lawyers usually bill separately under the heading of disbursements or expenses rather than attorney’s fees proper.
We also agree with the plaintiffs that the district judge should not have simply halved the amount of the fee that they requested. The judge thought the fee request excessive and he may well have been right. But rather than indicating which parts of the request he thought excessive, the judge simply cut the whole thing in half. This would be an appropriate way of proceeding in a case where the amount requested was small. If the plaintiffs were requesting, say, $1,000, it would not be worth the judge’s time to pore over the details of the request. As a former practitioner, he could quite properly use his impression of prevailing market rates in his district to cut down the request. But where as in this case the request is for a large amount of money, here more than $50,000, the judge must do more than eyeball the request and if it seems excessive cut it down by an arbitrary percentage. He has to make a judgment — considering the nature of the case and the details of the request, taking evidence if need be, and defending his judgment in a reasoned (though brief) opinion — on what the case should have cost the party who submitted the request. In this case some of the items seem excessive; among other things, $6,000 for preparing and presenting the attorney’ s-fee request itself seems much too large (nevertheless, by giving the plaintiffs half of what they requested, the judge in effect gave them $3,000 for preparing the fee request — which still seems like too much, though this is a matter for exploration on remand). The total bill seems altogether disproportionate to the scale of the case and to the County’s attorney’s fees, if they have been correctly represented to us (a matter the district judge can explore on remand). So in remanding we do not want to be understood as suggesting that the district judge cut the plaintiffs’ fee request by too much. Maybe he cut it by too little; but he went about the cutting in the wrong way. In deciding what if any deduction he should make because the plaintiffs dismissed several defendants and abandoned several legal theories in the course of the proceeding, he should of course be guided by the Supreme Court’s recent decision in Hensley v. Eckerhart, 461 U.S. 424, 103 S.Ct. 1933, 76 L.Ed.2d 40 (1983).
To summarize, the judgment of the district court on the merits is affirmed except with respect to prejudgment interest; as to that, the case is remanded for further proceedings. The district court’s award of attorney’s fees is vacated and that matter also remanded for further proceedings. No costs will be awarded in this court.
So Ordered.