In this appeal from the entry of partial summary judgment, the parties dispute the district court's ruling with respect to the viability of certain claims brought pursuant to the Employee Retirement Income Security Act (ERISA), 29 U.S.C. §§ 1001-1461. The district court held that while some claims survived the statute of limitations, others did not. We affirm in part, reverse in part and remand.

On May 1, 1987, the Department of Labor (DOL) filed suit against both the trustees of a multi-employer health and welfare fund and Consultants & Administrators, Inc. (C & A), a corporation that supplied the fund with dental services. The welfare fund was operated by the Construction and General Laborers District Council of Chicago and Vicinity to provide medical and related services to affiliated union members. The DOL charged that the trustees had violated certain provisions of ERISA by awarding noncompetitive contracts to C & A, by operating a kickback scheme with C & A and by otherwise violating their fiduciary duties.

The defendants named in the DOL's lawsuit fell into two groups. In the first group were two former trustees--Alfred Pilotto and James Caporale--and the C & A defendants, who were directly implicated in the kickback scheme. The alleged scheme began when C & A was awarded the dental services contract. C & A paid ten percent of its gross revenue to Pinckard & Associates, a company owned by Pilotto's son-in-law, which funneled the money to Pilotto and Caporale in the form of kickbacks. Caporale, Pilotto and four C & A defendants were convicted of charges stemming from their roles in this kickback scheme.1 The DOL sued this group of defendants (whom we refer to as "C & A defendants") to recover the kickbacks.

The second group of defendants in the DOL's action comprised the remaining trustees--or in the district court's phrase, the "innocent trustees" (we refer to this group simply as "trustees"). These trustees claimed to have no knowledge of the kickback scheme with C & A. The DOL maintained, however, that the trustees had violated ERISA in three ways: (1) by failing to monitor adequately C & A's services to ensure that C & A's prices were not excessive (the "monitoring claim"); (2) by twice renewing the dental services contract with C & A (in 1983 and again in 1986) without soliciting comparable bids from other dental service providers (the "bidding claim"); and (3) by failing to take steps to recover the kickbacks paid to Caporale and Pilotto by C & A (the "kickback claim").

The trustees brought their own suit against the C & A defendants on May 29, 1987. Like the DOL's suit filed 28 days earlier, the trustees' action sought recovery of the kickbacks paid to Caporale and Pilotto.2

All the defendants soon filed motions for summary judgment on the ground that the claims were barred by the statute of limitations. In the DOL's suit against the trustees, the trustees asserted that all three claims (the monitoring, bidding and kickback claims) were barred by the three-year statute because the DOL knew about the relevant conduct before May 1, 1984. The district court agreed in part, but held that some of the trustees' allegedly improper activity (specifically the later monitoring and bidding conduct) occurred within the three-year limitations period and therefore was not barred. The court granted summary judgment to the trustees on claims arising before May 1, 1984, but allowed claims based on activity occurring after that date. It thus held that the monitoring and bidding claims were partially barred, but that the kickback claim was fully barred since the trustees' duty to sue to recover the kickbacks arose before May 1, 1984.3

The C & A defendants also moved for summary judgment on statute of limitations grounds in both the DOL's and the trustees' suits for recovery of the kickbacks. The district court, finding a six-year limitations period applicable because the suits involve fraud, held that both the DOL's and the trustees' claims against the C & A defendants survived the statute of limitations.4

We begin by addressing this court's jurisdiction over these appeals. In both the DOL's and the trustees' actions against the C & A defendants, the district court found the suits to be timely and accordingly denied the defendants' motions for summary judgment. In an order dated May 4, 1990, the district court certified this ruling for interlocutory appeal pursuant to 28 U.S.C. § 1292(b), finding that it involved a controlling question of law as to which there is substantial ground for difference of opinion and that an immediate appeal may materially advance the ultimate termination of the litigation. A panel of this court granted leave to appeal on September 4, 1990, concluding that "all questions concerning the statute of limitations" should be resolved in a single appeal. The C & A defendants' appeal in No. 90-3113 is properly before the court.

In the DOL's suit against the trustees, the district court granted summary judgment in favor of the trustees as to certain claims (specifically, the pre-May 1, 1984 bidding and monitoring claims and the kickback claim). The district court entered final judgment on those claims on May 4, 1990, certifying under Fed.R.Civ.P. 54(b) that there was no just reason for delay. We agree with the district court that the three prerequisites for an appealable judgment under Rule 54(b) are met here: first, the action involves separate claims; second, there is a final decision as to at least one of these claims; and third, there has been an express--and proper--determination that there is no just reason for delay. Stearns v. Consolidated Management, Inc., 747 F.2d 1105, 1108 (7th Cir.1984). In particular, allowing this case to proceed to trial before a final resolution of the statute of limitations issues might necessitate a second trial after a later appeal. The DOL's appeal (No. 90-2450) from the order granting partial summary judgment is therefore proper.

That leaves the trustees' cross-appeal from the district court's partial denial of their summary judgment motion. (The court denied summary judgment as to the DOL's post-May 1, 1984 bidding and monitoring claims.) This ruling, however, because it lacks a certification under either Rule 54(b) or 28 U.S.C. § 1292(b), appears not to be a final, appealable order. The parties argue that the trustees' cross-appeal is nevertheless proper under the doctrine of pendent appellate jurisdiction. That doctrine permits us to exercise jurisdiction over an ordinarily unappealable interlocutory order only when it is "inextricably entwined" with an appealable interlocutory order and there are "compelling reasons for not deferring the appeal of the former order to the end of the lawsuit." United States ex rel. Valders Stone & Marble, Inc. v. C-Way Construction Co., 909 F.2d 259, 262 (7th Cir.1990) (quoting Illinois ex rel. Hartigan v. Peters, 861 F.2d 164, 166 (7th Cir.1988)). As we have emphasized, the doctrine of pendent appellate jurisdiction is limited in scope, and will not be held applicable merely because the orders are related. See Patterson v. Portch, 853 F.2d 1399, 1403 (7th Cir.1988). In the present case, we find that the DOL's appeal and the trustees' cross-appeal are indeed inextricably entwined. Both involve the same facts and turn on precisely the same issue: at what point the DOL's knowledge triggered the three-year statute of limitations to bar the DOL's claims. As in Peters, the appeals are thus "the head and tail of the same coin," 861 F.2d at 166, and concerns of judicial economy justify our exercise of jurisdiction over both appeals together. The trustees' cross-appeal (No. 90-2568) is properly before us.

Because the district court's decision was an entry of summary judgment, our review is de novo, and we draw all reasonable inferences in favor of the nonmoving parties. Allensworth v. General Motors Corp., 945 F.2d 174, 178 (7th Cir.1991). Summary judgment is appropriate only if we find that there is no genuine issue as to any material fact and that the moving parties are entitled to judgment as a matter of law. Fed.R.Civ.P. 56(c); Old Republic Ins. Co. v. Federal Crop Ins. Corp., 947 F.2d 269, 273 (7th Cir.1991).

These actions allege fiduciary violations under sections 404 and 406 of ERISA, 29 U.S.C. §§ 1104 & 1106. Plan fiduciaries are personally liable for such violations. 29 U.S.C. § 1109. The applicable statute of limitations provides as follows:

No action may be commenced under this subchapter with respect to a fiduciary's breach of any responsibility, duty, or obligation under this part, or with respect to a violation of this part, after the earlier of--

(1) six years after (A) the date of the last action which constituted a part of the breach or violation, or (B) in the case of an omission, the latest date on which the fiduciary could have cured the breach or violation, or

(2) three years after the earliest date on which the plaintiff had actual knowledge of the breach or violation;

except that in the case of fraud or concealment, such action may be commenced not later than six years after the date of discovery of such breach or violation.

29 U.S.C. § 1113.

In the DOL's suit against the trustees, the three-year "actual knowledge" provision of 29 U.S.C. § 1113(2) is at issue. Because the DOL brought suit on May 1, 1987, the crucial date for statute of limitations purposes is May 1, 1984: if the DOL had actual knowledge of a "breach or violation" before that date, then its suit is time barred as to that breach or violation. We begin by reviewing the DOL's activities that potentially gave rise to actual knowledge of the trustees' alleged violations.

The DOL began investigating the fund's relationship with C & A as early as 1977, but nothing developed after initial phases of the investigation concluded. The first DOL investigation began in January 1977 but was terminated in June of the same year; the records of this investigation were apparently destroyed, so its findings are unclear. A second DOL investigation began in 1978, but was called off in 1979 at the request of the U.S. Attorney. At this time the Department of Justice and its Chicago and Miami strike forces were conducting an independent investigation which ultimately led to the indictments and convictions of former trustees Caporale and Pilotto. Caporale and Pilotto were indicted for their role in the kickback scheme on June 3, 1981, and convicted on October 13, 1982.

The DOL officially opened its third investigation of the fund in January 1983, although the investigation did not actually begin until September. During September the DOL conducted an onsite audit of the fund's records and interviewed its attorney, Marc Pekay. DOL investigator Sammiesteen Haynes-Green conducted the interview, which focused on the relationship between the fund and C & A. Haynes-Green was informed of the methods that the trustees used in selecting C & A for its dental services contract and in monitoring C & A's performance and fees. She learned that the trustees retained the Martin E. Segal Company to prepare, solicit and analyze bids, and that the trustees monitored C & A primarily in two ways: through quarterly reports of a panel of independent dentists (the "Peer Review Committee") on the quality of performance and the reasonableness of fees; and through "utilization reports" prepared by C & A, which showed the value of services compared to the amount of premium paid. In a memo prepared before January 16, 1984, Haynes-Green concluded that, in their bidding procedures for the previous contract (the 1980-83 contract), the trustees had not considered bids from plans comparable to that of C & A--i.e., other "closed-panel" plans.5 The minutes of trustee board meetings made available to the DOL in February 1984 revealed that the trustees decided to continue selecting bids and monitoring service providers as it had done in the past. In December 1983 a new contract (the 1984-86 contract) was awarded to C & A.

On April 5, 1984, Haynes-Green filed a report in which she compared the fee schedules under the C & A contract with a schedule of standard fees supplied by the American Dental Association. On May 2, 1984, the fund gave the DOL a bid analysis it had prepared for its new (1984-86) dental services contract. The analysis showed that the trustees had in fact continued to conduct business as usual in their award of the new contract: the only closed-panel bid was submitted by C & A.

The district court, applying the three-year statute of limitations of 29 U.S.C. § 1113(2), held that any bidding or monitoring claim that arose after May 1, 1984 was viable. Thus the DOL's bidding claim based on the most recent contract (the 1987 contract, which was renewed in 1986) and its claim based on post-May 1, 1984 monitoring activity survived. But the DOL's claim for bidding on the 1984 contract (which occurred in 1983) and the pre-May 1, 1984 monitoring claim were held to be barred. On appeal, the battle lines are clearly drawn: the DOL argues that none of its claims are barred by the statute of limitations, while the trustees argue that all of the DOL's claims are barred.

The thrust of the DOL's argument is that the district court failed to apply the standard of "actual knowledge," but instead allowed mere constructive knowledge to trigger the limitations period.6 At no time before May 1, 1984, contends the DOL, did it have actual knowledge of an ERISA violation on the part of the trustees. In addition, the DOL maintains that the district court improperly lumped together distinct violations, finding the DOL to have had knowledge of an overly general transaction which the court characterized as "continued careless dealings with C & A."

The trustees respond by arguing that "actual knowledge" does not require knowledge of every detail of a transaction, and that the DOL certainly had sufficient knowledge to start the limitations period. Further, the trustees urge that since the DOL knew all the details of their dealings with C & A before May 1, 1984--and since those dealings never really changed after that date--all the DOL's claims are time barred, including those premised on conduct after May 1, 1984.

It is difficult to say in the abstract precisely what constitutes "actual knowledge" of a "breach or violation." As the DOL correctly insists, actual knowledge must be distinguished from constructive knowledge. " 'To charge [plaintiff] with actual knowledge of an ERISA violation, it is not enough that he had notice that something was awry; he must have had specific knowledge of the actual breach of duty upon which he sues.' " Radiology Center, S.C. v. Stifel, Nicolaus & Co., 919 F.2d 1216, 1221 (7th Cir.1990) (quoting Brock v. Nellis, 809 F.2d 753, 755 (11th Cir.), cert. dismissed, 483 U.S. 1057, 108 S.Ct. 33, 97 L.Ed.2d 821 (1987)). At the same time, the relevant knowledge for triggering the statute of limitations is knowledge of the facts or transaction that constituted the alleged violation. Consequently, it is not necessary for a potential plaintiff to have knowledge of every last detail of a transaction, or knowledge of its illegality. See Ziegler v. Connecticut General Life Ins. Co., 916 F.2d 548, 552 (9th Cir.1990); Blanton v. Anzalone, 760 F.2d 989, 992 (9th Cir.1985). The line between actual and constructive knowledge of a violation is thus not as bright as the DOL suggests, and in cases near the border the distinction may well be merely semantic.7 Suffice it to say that to have actual knowledge of a violation to trigger ERISA's three-year statute of limitations, a plaintiff must know of the essential facts of the transaction or conduct constituting the violation.

But how does one characterize the relevant transaction and its essential facts? Like so many areas of law, this matter turns on the level of generality employed in characterizing the transaction at issue. We know that somewhere between "every last detail" and "something was awry" lies the requisite knowledge of an ERISA violation. The proper characterization will usually turn in part on the complexity of the underlying factual transaction, the complexity of the legal claim and the egregiousness of the alleged violation. Beyond such generalizations, however, we can only say that judges, faced with particular contexts and relying on their "situation sense,"8 must make the determination.

We first address the bidding and monitoring claims. According to the DOL, these violations involved complex and intricate financial transactions that required lengthy study before a suit could be filed. It follows from this characterization that each of its claims of fiduciary breach should be analyzed individually. Under the district court's analysis, however, the bidding and monitoring claims involve relatively simple transactions of "continued careless dealings" with C & A (or "imprudence," to use the trustees' term). The court found that the DOL had knowledge of the careless dealings before May 1, 1984.

We think that the district court's characterization of the relevant factual transaction was overly general. The underlying facts are too complex to be characterized simply as "careless dealings." Such an approach is too close to a rule under which knowledge that "something was awry" constitutes actual knowledge of a violation, a formulation we rejected in Radiology Center. Moreover, it is not immediately obvious that the conduct known to the DOL was in fact careless until the transactions are examined in greater detail. While we therefore agree with the DOL that the bidding and monitoring claims should be analyzed separately, we nevertheless also agree with the result ultimately reached by the district court.

As for the DOL's bidding claim on the 1984 contract, the evidence reveals that the DOL did have knowledge of the essential facts before May 1, 1984. First, by 1981 the DOL knew of the indictment (and by 1982 the convictions) of two fund trustees for taking kickbacks from C & A. By itself, this knowledge does not translate into knowledge of improper bidding in connection with the award of the 1984 contract to C & A, but it is surely a piece of knowledge that is relevant when combined with other knowledge. Second and more important is the DOL's third investigation, which began in 1983 and was conducted by Haynes-Green. Early in the investigation, Haynes-Green reviewed the fund's contractual arrangement with C & A and the bidding procedures conducted by the Segal Company. By February 1984, she had analyzed the bid proposals and analysis for the 1980-83 contract. In a handwritten memorandum, Haynes-Green noted that C & A's was the only closed-panel plan considered for the 1980 contract, since the only two other such plans had withdrawn their bids. The minutes of the trustees' 1983 meetings, which had also been reviewed by February 1984, revealed that the trustees had nevertheless decided to continue soliciting bids and selecting a dental service provider in the same manner as had been used in the past. Also in February 1984, the DOL converted the case from a "limited review" case to a "fiduciary violation" case; Haynes-Green and supervisory investigator John Ryan had concluded that they were faced with fiduciary violations. Finally, the DOL noted that by April 1984 it had accomplished review of the bidding activities. We find, therefore, that before May 1, 1984, the DOL had knowledge of the essential facts underlying the bidding claim on the 1984 contract.9

The bidding activities leading up to the 1987 contract, however, involve a new transaction and a distinct violation. The trustees contend that it is essentially the bidding procedure that is being challenged, and since the bidding procedure did not materially change between 1984 and 1987, the 1987 bidding claim is barred along with the 1984 claim. The flaw in the trustees' argument is that it ignores the continuing nature of a trustee's duty under ERISA to review plan investments and eliminate imprudent ones. See 29 U.S.C. § 1104(a)(1)(B); Morrissey v. Curran, 567 F.2d 546, 549 n. 9 (2d Cir.1977). If knowledge of an ERISA violation barred claims based on similar future conduct, this continuing fiduciary duty would be severely weakened, and trustees would be left free to engage in repeated violations, so long as they have once been discovered but not sued. See Buccino v. Continental Assurance Co., 578 F.Supp. 1518, 1521 (S.D.N.Y.1983). On the other hand, we must not allow the "continuing violation" theory to be abused by plaintiffs who delay bringing suit in the hope of racking up damages. Such abuse is avoided here because we have found that the DOL may not recover for bidding activities on the 1984 contract. Strictly speaking, we are faced here with a repeated, rather than a continued, violation. The bidding activity on the 1987 contract (renewed in 1986) is more accurately characterized factually as a distinct transaction. First, it involved a new and separate contract. And second, although the trustees employed similar bidding procedures for the 1984 and 1987 contracts, those procedures, rather than being part of a formal policy, were separately considered and decided upon with respect to each contract.10 On the DOL's bidding claims, then, we draw the line where the district court did: the 1984 bidding claim is barred by the statute of limitations, but the 1987 bidding claim survives.

Our review of the record also reveals that the DOL had knowledge, prior to May 1, 1984, of the essential facts underlying its monitoring claim based on pre-May 1, 1984 activity. During Haynes-Green's extensive investigation of the fund beginning in 1983, she learned the specific methods by which the trustees monitored C & A's performance. By April 1984, Haynes-Green had reviewed the minutes of the trustees' meetings, which revealed what the trustees did to monitor C & A. Also by that time, she had reviewed copies of the actual Peer Review Committee reports and C & A's utilization reports which were relied upon by the trustees. In addition, she had completed a review of C & A contracts and had compared C & A schedules with the American Dental Association's schedule of standard fees. In fact, Haynes-Green discussed her findings in a memorandum prepared by February 1984, specifically noting that the Peer Review Committee does not determine whether C & A's charges are reasonable.

The DOL maintains that despite all this research, it still did not know as of May 1, 1984 "whether or not C & A's charges were reasonable in light of its services," DOL Br. at 37, and more study was needed on the monitoring issue. But to have actual knowledge of the relevant facts, the DOL did not have to know the effects of the transaction, or the specific fact of its illegality. Ziegler, 916 F.2d at 552. Hence, to have knowledge of the monitoring violation, the DOL did not need specific knowledge that C & A's fees were unreasonable. The claim is that the trustees' monitoring was inadequate: once the DOL knew about all the essential procedures used by the trustees in monitoring C & A, the three-year limitations period began running. Further, the DOL's argument that it did not know precisely how each different monitoring procedure was relied upon by the trustees is unpersuasive. The DOL may not have known "every last detail" of the trustees' monitoring procedures, but it knew all the essentials of those procedures by April 1984.

Turning to the monitoring activity after May 1, 1984, we agree with the district court that claims premised upon this activity survive the statute of limitations. The recent monitoring claims present a tougher question than the recent bidding claims, because the later monitoring activity does not appear as factually distinct from the earlier activity, but follows more closely a continuous pattern. Nevertheless, we think that the post-May 1, 1984 monitoring activity is not barred. First, this monitoring activity was conducted pursuant to a new contract with C & A. And because the trustees could have changed their monitoring methods, the post-May 1, 1984 monitoring conduct is fairly characterized as a separate violation. Second, given ERISA's imposition of a continuing fiduciary duty, past knowledge of a past violation generally should not be held to preclude a suit for a repeated or continued violation. Buccino, 578 F.Supp. at 1521. And third, plaintiffs' abuse through delay in bringing suit is again avoided here, where the monitoring claims have been limited to those arising after May 1, 1984.

Turning to the kickback recovery claim, we disagree with the district court's conclusion that the claim is barred by the statute of limitations. The district court found that the DOL knew before May 1, 1984 that the trustees had failed to file suit to recover the kickbacks from Pilotto and Caporale. Instead of suing, however, the DOL waited until 1987, and in the district court's view this "failure to bring suit sooner amounted to a negligent and unreasonable delay, for which [the DOL] offers no explanation." Mem.Op. & Order at 16. The court rejected the DOL's theory that the trustees' failure to sue for recovery of the kickbacks was an ongoing violation:

To hold that the failure to take a single affirmative act, which creates a single injury, is a continuing violation would be to render the statute of limitations meaningless for such a violation.... In this case, as it was, almost six years had passed from the time of Pilotto and Caporale's indictment, without the Trustees bringing suit against them, by the time the DOL brought its suit in 1987. If this failure to sue were a "continuing violation," the DOL could have waited even longer, and brought suit against the Trustees whenever it chose, there being nothing to start the statute of limitations running.... While at this point we cannot say when the failure to sue became a violation, if it did become a violation this happened before May 1, 1984.

Id. at 18.

We disagree with the district court's assumption that there is nothing to trigger the statute of limitations for such failure-to-sue claims. As the DOL points out in its brief, it is not until the trustees' delay in bringing suit has precluded (or at least prejudiced) their ability to recover that the DOL has a claim--and hence knowledge of a violation. Absent some other evidence of unreasonable delay, then, the end-point of the statute of limitations on the primary claim will mark the start of the limitations period on the derivative claim; when the trustees could no longer sue to recover the kickbacks, only then did the DOL have knowledge of a failure-to-sue violation. The trustees had actual knowledge of the kickback scheme in June 1981, when Pilotto and Caporale were indicted. Section 1113 gave them three years from that time to sue for recovery of the kickbacks.11 The three-year limitations period on the DOL's failure-to-sue claim therefore did not begin to run until June 1984, so its claim against the trustees is timely.12 The DOL's kickback recovery claim against the trustees must therefore survive the statute of limitations.

The trustees argue, in the alternative, that the DOL's claims are barred by the doctrine of laches. As a general rule, the United States is not subject to the equitable defense of laches in enforcing its rights. United States v. Summerlin, 310 U.S. 414, 416, 60 S.Ct. 1019, 1020, 84 L.Ed. 1283 (1940); Silverman v. Commodity Futures Trading Comm'n, 549 F.2d 28, 34 (7th Cir.1977). An exception to this rule has been created in cases brought by the Equal Employment Opportunity Commission asserting the rights of individual claimants. EEOC v. Massey-Ferguson, Inc., 622 F.2d 271, 275 (7th Cir.1980); United States v. Georgia Power Co., 474 F.2d 906, 923-24 (5th Cir.1973); see also Occidental Life Ins. Co. v. EEOC, 432 U.S. 355, 373, 97 S.Ct. 2447, 2458, 53 L.Ed.2d 402 (1977) (federal courts have power to restrict or deny backpay relief if defendant is prejudiced by inordinate delay of EEOC). The district court held that the laches exception does not apply to a suit brought by the DOL under ERISA, because such a suit is brought primarily in furtherance of the public interest rather than on behalf of an individual or a discrete group of workers. See Georgia Power Co., 474 F.2d at 923. Indeed, we have recognized that the government's capacity to sue under ERISA advances important public interests tied to the purposes of ERISA itself, such as maintaining public confidence in funds serving thousands of employees. See Secretary of Labor v. Fitzsimmons, 805 F.2d 682, 690-94 (7th Cir.1986) (en banc).

The trustees argue, however, that the DOL is simply suing on behalf of the individual fund beneficiaries and no government funds are at stake; therefore, they contend, this case is indistinguishable from the EEOC cases which applied laches against the government. The trustees' argument is a strong one. The EEOC, when it sues under Title VII of the Civil Rights Act of 1964 on behalf of individuals, is surely promoting important public interests as well as seeking recovery for individuals; yet the courts have found laches applicable. In addition, since the fund involved in the present case is not a pension fund but a health and welfare fund, it is not backed by government guarantees and any recovery would apparently go to the fund.

The EEOC cases might nevertheless be distinguishable on a different ground. Title VII contains no express limitation on the time within which the EEOC may bring an enforcement action, Massey-Ferguson, 622 F.2d at 275, and the Supreme Court had suggested that laches would therefore apply to such actions, Occidental Life, 432 U.S. at 373, 97 S.Ct. at 2458. In contrast, Congress has specified the limitations period governing ERISA actions; in fact, it has specified that period in a variety of circumstances, providing that a plaintiff must sue within six years from the last date of a breach, within three years from the date on which the plaintiff had actual knowledge of a breach or, in the case of fraud or concealment, within six years from the date of discovery of a breach. 29 U.S.C. § 1113. Courts are often hesitant to apply laches where a plaintiff has sued within the time period expressly provided by the applicable statute. See, e.g., United States v. Mack, 295 U.S. 480, 489, 55 S.Ct. 813, 817, 79 L.Ed. 1559 (1935); United States v. RePass, 688 F.2d 154, 158 (2d Cir.1982). A plaintiff might reasonably rely on section 1113's specific limitations periods; correspondingly, defendants, in the face of the statute, may have difficulty making the requisite showing of reasonable reliance on the DOL's failure to file suit. See Bennett v. Tucker, 827 F.2d 63, 69 (7th Cir.1987).

Moreover, the general principle that laches is not available against the government would seem to have particular application where Congress has specified a limitations period. That general principle was originally founded on concerns of public policy and sovereignty: "It was deemed important that, while the sovereign was engrossed by the cares and duties of his office, the public should not suffer by the negligence of his servants." United States v. Thompson, 98 U.S. 486, 489, 25 L.Ed. 194 (1879). The principle is thus closely related to the principle that estoppel does not lie against the government.13 See Mendrala v. Crown Mortgage Co., 955 F.2d 1132, 1139-40 (7th Cir.1992). An increasingly invoked rationale for the no-estoppel principle involves the separation of powers. Allowing the federal government to be estopped, so the argument goes, would frustrate federal statutes and infringe upon Congress' exclusive authority to make law. See, e.g., Schweiker v. Hansen, 450 U.S. 785, 788, 101 S.Ct. 1468, 1470, 67 L.Ed.2d 685 (1981); Mendrala, 955 F.2d at 1140; McCauley v. Thygerson, 732 F.2d 978, 982 (D.C.Cir.1984); Portmann, 674 F.2d at 1159. Applying laches against the federal government may similarly implicate separation of powers concerns where Congress has specifically provided for limitations periods. In a context where Congress has spoken clearly and completely on a limitations period, the application of laches may tend to frustrate the scheme it has codified.

Despite these concerns, there is authority for applying laches in cases governed by a statute of limitations. See, e.g., Maksym v. Loesch, 937 F.2d 1237, 1248 (7th Cir.1991); K-Mart Corp. v. Oriental Plaza, Inc., 875 F.2d 907, 911 (1st Cir.1989); Azalea Fleet, Inc. v. Dreyfus Supply & Machinery Corp., 782 F.2d 1455, 1459 (8th Cir.1986). But see Tandy Corp. v. Malone & Hyde, Inc., 769 F.2d 362, 365 (6th Cir.1985) (presumption against application of laches to cut down statute of limitations). Given these cases, we hesitate to declare that laches can never be applied against the government in an ERISA case simply because Congress has codified a statute of limitations. Moreover, the concerns of sovereignty and separation of powers that are ordinarily said to prevent estopping the government may be misplaced in a case in which the government is suing not on its own behalf but as the representative of private interests. So it is not clear, after all, that a DOL suit under ERISA is different from the EEOC cases in which laches has been held to apply.

In any event, we find that the elements of laches have not been met in the case before us. Laches is a species of estoppel and thus requires a showing of (1) unreasonable delay and (2) harm or prejudice to the defendant. See Maksym, 937 F.2d at 1248; United States v. Kairys, 782 F.2d 1374, 1384 (7th Cir.), cert. denied, 476 U.S. 1153, 106 S.Ct. 2258, 90 L.Ed.2d 703 (1986). As for those claims that we have found not precluded by the statute of limitations, the DOL's delay was not unwarranted in the circumstances (which include the complexity of the investigation and the fact that the DOL sued within section 1113's specific time limits, see Tandy Corp., 769 F.2d at 365). Further, the trustees have failed to make an adequate showing of harm, beyond mere "inconvenience," resulting from the DOL's failure to file suit. See Bennett v. Tucker, 827 F.2d 63, 69 (7th Cir.1987).

The DOL's suit against the C & A defendants seeks recovery of the kickbacks improperly paid under the kickback scheme involving C & A and convicted former trustees Caporale and Pilotto. The C & A defendants sought summary judgment on the ground that the DOL had the crucial knowledge of the kickback scheme outside the applicable statute of limitations. The DOL's alleged knowledge of the scheme comes essentially from two related criminal investigations into the activities of C & A, Caporale, Pilotto and others. The district court, applying the six-year limitations period for cases of "fraud or concealment," found the DOL's suit timely and denied the C & A defendants' motion for summary judgment.

In September of 1976, FBI agents interviewed Daniel Milano, Jr., who was an officer of C & A and the son of its founder. Milano gave the agents information regarding kickback schemes involving C & A, and a grand jury was convened in Chicago. During 1978 a DOL Pension Task Force was set up in Miami to conduct civil ERISA investigations in and around Florida. The Task Force attempted to obtain materials from the Chicago grand jury, but was granted access only to documents involving a Miami fund and C & A. The Chicago grand jury was eventually terminated, and by 1980 a grand jury in Miami had taken over the entire investigation. The DOL Task Force also attempted to gain access to the Miami grand jury materials, but obtained only documents unrelated to the ongoing criminal investigation.

By this time, the Department of Justice had established the Miami Organized Crime Strike Force, which was conducting the grand jury investigation in Miami. The Miami Strike Force was staffed by investigators from different government agencies, including William R. Gamble, an employee of the DOL. Gamble had been present at the FBI agents' interview of Milano and later reviewed Milano's grand jury testimony. Milano described a kickback arrangement involving a Miami fund and a subsidiary of C & A, naming Pilotto as a participant; he also stated that C & A's contract in Chicago was based on a kickback scheme. Gamble prepared a report describing the Chicago kickback scheme on April 7, 1978. The report contained a warning on its cover stating that its contents were not to be disclosed and were to be used "solely for the purpose of the Grand Jury Investigation" in the absence of a court order.

The Office of the Inspector General, a division of the DOL, investigated all Strike Force cases that were labor related, and hence was investigating Caporale and Pilotto by 1979. In addition, the OIG assisted in the grand jury investigation and prepared a draft of the Miami indictment, which detailed the elements of the kickback scheme and mentioned Caporale, Pilotto and other participants. The DOL had also opened a civil investigation in Chicago in 1978 (the second of the DOL's three civil investigations of the Chicago fund, see supra section III.A), but this investigation was closed at the request of the U.S. Attorney so that it would not interfere with the grand jury investigation.

As noted above, the district court rejected the C & A defendants' argument that the DOL's claim was barred by the statute of limitations. The court applied the six-year limitations period of 29 U.S.C. § 1113 for cases of "fraud or concealment" and held that the DOL had not discovered the kickback scheme before May 1, 1981 (six years before the DOL filed suit). Ordinarily, we would begin by considering whether the district court applied the correct statute of limitations. It appears, however, that this issue has been waived: on appeal, the C & A defendants do not challenge the district court's application of the six-year statute of limitations to the DOL's claim for recovery of the kickbacks. Instead, they argue only that the DOL discovered the kickback scheme before May 1, 1981.14 We need not rely solely on C & A's waiver, however, because we believe that the six-year period for cases of "fraud or concealment" is appropriate in the DOL's case, though on a different basis than that relied on by the district court. Recent case law has clarified the applicability of the six-year "fraud or concealment" limitations period, and we proceed to address the question of the applicable statute of limitations.15

The relevant portion of 29 U.S.C. § 1113 provides: "[I]n the case of fraud or concealment, such action may be commenced not later than six years after the date of discovery of such breach or violation." The district court assumed that this six-year period applied because the underlying claim involved allegations of fraud. Several months after the district court issued its decision, however, this Court rejected the district court's approach in Radiology Center, S.C. v. Stifel, Nicolaus & Co., 919 F.2d 1216 (7th Cir.1990). In that case we held that "the phrase ["fraud or concealment" in section 1113] refers to steps taken by the defendant to hide the fact of the breach rather than to the underlying nature of plaintiffs' claim." Id. at 1220. In other words, the six-year "fraud or concealment" provision, as the Eighth Circuit had concluded, "incorporates the fraudulent concealment doctrine." Id. (quoting Schaefer v. Arkansas Medical Soc'y, 853 F.2d 1487, 1491 (8th Cir.1988)).

One might jump to the conclusion that Radiology Center's narrow construction of the six-year "fraud or concealment" provision prevents the DOL from utilizing the six-year limitations period. And if the six-year period does not apply, then the three-year "actual knowledge" provision is certainly fatal to the DOL's claim: the DOL concedes that it had actual knowledge of the kickback scheme in June 1981, when the indictment was returned. But we must first decide whether the six-year period applies, and that requires consideration of the fraudulent concealment doctrine.

The court in Radiology Center did not specify the elements needed to show fraudulent concealment.16 In Schaefer, the Eighth Circuit summarized the doctrine as requiring "that plaintiffs show (1) that defendants engaged in a course of conduct designed to conceal evidence of their alleged wrongdoing and that (2) they were not on actual or constructive notice of that evidence, despite (3) their exercise of due diligence." 853 F.2d at 1491-92 (quoting Foltz v. U.S. News & World Report, Inc., 663 F.Supp. 1494, 1537 (D.D.C.1987), aff'd, 865 F.2d 364 (D.C.Cir.), cert. denied, 490 U.S. 1108, 109 S.Ct. 3162, 104 L.Ed.2d 1024 (1989)). That summary was based on the extensive discussion of fraudulent concealment in the case of Hobson v. Wilson, 737 F.2d 1, 33-36 (D.C.Cir.1984), cert. denied sub nom. Brennan v. Hobson, 470 U.S. 1084, 105 S.Ct. 1843, 85 L.Ed.2d 142 (1985). The Hobson court noted that the key to a finding of fraudulent concealment is some act on the part of the defendant designed to avoid detection; it quoted the Supreme Court's decision in Wood v. Carpenter, 101 U.S. (11 Otto) 135, 25 L.Ed. 807 (1879): " 'Concealment by mere silence is not enough. There must be some trick or contrivance intended to exclude suspicion and prevent inquiry.' " Id. 101 U.S. (11 Otto) at 143, 25 L.Ed. at 809, quoted in Hobson, 737 F.2d at 33. In other contexts, the doctrine of equitable tolling is applied where there is fraudulent concealment. Cases generally distinguish between two types of acts that can constitute fraudulent concealment:

acts that are self-concealing (such as frauds)[;] and acts where, absent a subsequent act of concealment, only the perpetrator, but not the fact that a cause of action might exist, would be unknown (such as a burglary). In the former case, concealment is established by the nature of the act; in the latter case, additional acts of concealment are required to trigger the tolling doctrine.

Id. (footnote omitted). In the first type of case--self-concealing acts--it is not enough that an act be a substantive fraud; rather, the defendant must engage in "some misleading, deceptive or otherwise contrived action or scheme, in the course of committing the wrong, that is designed to mask the existence of a cause of action." Id. at 34 (emphasis deleted) (citing Wood, 101 U.S. (11 Otto) at 143, 25 L.Ed. at 809). Our cases have generally placed both categories of acts under the heading "fraudulent concealment" or "equitable tolling." See, e.g., Teamsters Local 282 Pension Trust Fund v. Angelos, 815 F.2d 452, 456 & n. 4 (7th Cir.1987); Suslick v. Rothschild Securities Corp., 741 F.2d 1000, 1004 (7th Cir.1984); Tomera v. Galt, 511 F.2d 504, 509-10 (7th Cir.1975); Sperry v. Barggren, 523 F.2d 708, 710-11 (7th Cir.1975).

Radiology Center creates a wrinkle in this scheme. On the one hand, it held that a mere underlying claim of fraud is not enough to qualify for ERISA's six-year limitations period. (This would seem to cast doubt on the "self-concealing acts" category of concealment.) On the other hand, Radiology Center simply "incorporated" the fraudulent concealment doctrine. (This would seem to imply that the doctrine was incorporated as it stands in this and other circuits, including the category of self-concealing acts.) The view that the six-year statute does not encompass "self-concealing acts" appears to be supported by dicta in a recent case. In Cada v. Baxter Healthcare Corp., 920 F.2d 446 (7th Cir.1990), cert. denied, --- U.S. ----, 111 S.Ct. 2916, 115 L.Ed.2d 1079 (1991), the court expressed the view that fraudulent concealment "must not be confused with efforts by a defendant in a fraud case to conceal the fraud."17 The question, then, is whether Radiology Center's incorporation of "the fraudulent concealment doctrine" into ERISA's statute of limitations includes so-called self-concealing acts, or is limited to active concealment that is separate from the underlying wrongdoing.

We think that fraudulent concealment as incorporated in section 1113 is not so limited, but can include genuine acts of concealment committed in the course of the underlying wrong. First, Radiology Center 's incorporation of the fraudulent concealment doctrine is not specifically restricted to non-fraud or "active concealment" cases; the case merely holds that a fraud claim does not trigger the six-year period by virtue of the "underlying nature of plaintiffs' claim." 919 F.2d at 1220. Its language is broad, extending to "steps taken by the defendant to hide the fact of the breach." Id. Second, Radiology Center incorporates the line of cases expounding the fraudulent concealment doctrine, and the bulk of those cases include self-concealing acts as concealment that tolls the statute of limitations. Indeed, the Eighth Circuit's Schaefer decision, which the court in Radiology Center expressly followed, relied on the expansive version of the doctrine as formulated in Hobson. Schaefer, 853 F.2d at 1491-92 (citing Hobson, 737 F.2d at 33-34 & nn. 102-03). Third, the language of the statute's six-year provision ("in the case of fraud or concealment") seems to support the expansive version: its use of the disjunctive and separate inclusion of "fraud" would be in tension with an interpretation of fraudulent concealment that excluded concealment that occurs in the course of the fraud. Finally, in the case that is the keystone of federal fraudulent concealment doctrine, the Supreme Court included both categories, holding that the limitations period does not begin to run until the plaintiff discovers the cause of action where "the fraud has been concealed or is of such character as to conceal itself." Bailey v. Glover, 88 U.S. (21 Wall.) 342, 349, 22 L.Ed. 636, 639 (1875) (emphasis added).18 We emphasize that nothing we have said here disturbs Radiology Center: fraud claims do not receive the benefit of ERISA's six-year statute of limitations simply because they are fraud claims. There must be actual concealment--i.e., "some trick or contrivance intended to exclude suspicion and prevent inquiry." Wood, 101 U.S. (11 Otto) at 143, 25 L.Ed. at 809.

Such actual concealment was present in this case. Caporale, Pilotto and the C & A defendants committed fraud in the kickback scheme; they concealed the fraud actively, by channeling the kickbacks through a dummy corporation, James H. Pinckard & Associates, where the kickbacks were falsely labeled as payments for services supposedly rendered in checking members' eligibility. See Caporale, 806 F.2d at 1496-97. To be sure, the concealment occurred in the course of the fraud itself rather than independently of it. But it surely consisted of "steps taken by wrongdoing fiduciaries to cover their tracks." Radiology Center, 919 F.2d at 1220. We also conclude that the DOL has shown the requisite diligence and was not on notice of the defendants' fraud before June 1981.19 As we discuss below, the DOL suspended its investigations to allow the criminal investigation to proceed, and DOL employees' participation in the criminal investigation is not attributable to the DOL.

We conclude that section 1113's six-year "fraud or concealment" limitations period applies to the DOL's suit against the C & A defendants.

The next question is what the DOL knew before May 1, 1981, for its suit must have been commenced within six years "after the date of discovery" of the violation. If the DOL discovered, or could have discovered through due diligence,20 the kickback scheme before May 1, 1981, then its claim is barred even under the six-year limitations period.

C & A's principal argument is that the DOL had discovered the scheme through DOL employee William Gamble, who assisted the Miami Strike Force with the grand jury investigation and admittedly knew the details of the kickback scheme before 1981. Gamble, however, worked under the supervision of an Assistant U.S. Attorney and was under specific instructions that grand jury materials could not be used for any purpose other than assisting the Justice Department's criminal investigation. Gamble, who was essentially acting as an agent of the grand jury, treated the information he obtained in his work with the Strike Force as confidential. C & A maintains that much of Gamble's information did not constitute "matters occurring before the grand jury" under Fed.R.Crim.P. 6(e), so his knowledge must be imputed to the DOL. We do not think that the parameters of Rule 6(e)'s secrecy requirement are determinative of the DOL's knowledge through Gamble (and others on the Strike Force). If an agency employee assisting in a grand jury investigation makes a reasonable judgment that materials are confidential, then the agency should not be held to have discovered the information in those materials for purposes of a statute of limitations. Otherwise agencies would be penalized for cautiously following the dictates of federal law.

It is true as a general matter, of course, that knowledge in the possession of a government agent with a duty to disclose is imputed to the government. In re "Agent Orange" Product Liability Litigation, 597 F.Supp. 740, 796 (E.D.N.Y.1984), aff'd, 818 F.2d 145 (2d Cir.1987). The DOL agents here were loaned out to the Justice Department's Strike Force to assist in the grand jury's criminal investigation. They were thus wearing two hats, officially employed by the DOL but reporting to Justice Department attorneys and essentially acting as agents of the grand jury in Miami. They did have some duty (under DOL regulations) to continue reporting to DOL superiors, at least as to the "status" of their work. But they had a conflicting--and superior--duty not to report anything that would tend to reveal confidential grand jury matters. Rule 6(e) itself states that government personnel who are called on to assist government attorneys "shall not utilize that grand jury material for any purpose other than assisting the attorney for the government in the performance of such attorney's duty to enforce Federal criminal law." Fed.R.Crim.P. 6(e)(3)(B).

The general rule is that Rule 6(e)'s nondisclosure requirement applies to anything that may reveal what occurred before the grand jury. See, e.g., In re Grand Jury Matter (Catania), 682 F.2d 61, 63 (3d Cir.1982); 8 James W. Moore, Moore's Federal Practice p 6.05 (2d ed. 1991). Whether a given document or interview is sufficiently "grand jury related" to fall within Rule 6(e) is a difficult question in many cases. See In re Special February 1975 Grand Jury (Baggot), 662 F.2d 1232, 1237-38 (7th Cir.1981), aff'd, 463 U.S. 476, 103 S.Ct. 3164, 77 L.Ed.2d 785 (1983). As for Gamble's participation in the FBI's 1977 interview of Milano, we think that Gamble made a reasonable judgment in assuming that the information was confidential. He attended the interview only because of his participation in the Miami grand jury investigation. Moreover, he could not know at the time whether the information revealed at the interview would go before a grand jury, and in fact Milano testified before the grand jury a few months later. Similarly, Gamble's report was prepared for the grand jury investigation, contained material that was presented to the grand jury and had an explicit warning on its cover:

CAUTION: Grand Jury Investigation--In the absence of a F R C Pro Rule 6(e) Order, information indicated in this report shall be used solely for the purpose of the Grand Jury Investigation. The information contained herein was obtained by Grand Jury 76-4, Southern Judicial District of Florida.

See In re Grand Jury Matter (Garden Court Nursing Home), 697 F.2d 511, 513 (3d Cir.1982). Gamble's report also incorporated portions of an earlier FBI report, which C & A argues was independent of the grand jury investigation and not protected under Rule 6(e). While it is true that such FBI reports do not automatically fall within the scope of Rule 6(e), see Catania, 682 F.2d at 63-64, such reports can fall within the Rule's protection under appropriate circumstances, such as where they are closely related to the grand jury's investigation itself and where disclosure would reveal the identities of targets and other witnesses. See, e.g., Baggot, 662 F.2d at 1238; In re Subpoena to Testify Before Grand Jury (University of Florida), 864 F.2d 1559, 1564 (11th Cir.1989); Garden Court Nursing Home, 697 F.2d at 513. There is no need to decide definitively at this stage whether the FBI report was actually protected under Rule 6(e)'s secrecy requirement. The report was closely tied to Milano's grand jury testimony and apparently was itself submitted to the Miami grand jury. Moreover, it contained the names of confidential grand jury witnesses. Gamble's judgment call to keep the entire report confidential was a reasonable one, and his discovery of the kickback scheme should not be imputed to the DOL.

C & A also contends that the DOL discovered the kickback scheme through the DOL's Office of the Inspector General (OIG). Two OIG employees also participated in the Miami Strike Force investigation and assisted in the preparation of a draft indictment. Like Gamble, however, the OIG employees were agents of the grand jury and reasonably withheld grand jury materials. We reject C & A's argument that Catania, 682 F.2d at 62, calls for disclosure of the draft indictment. Disclosing the draft indictment during the grand jury's ongoing investigation would likely reveal the grand jury's decision processes and alert potential targets. And the OIG employees could make a reasonable judgment that they should not disclose such material under Rule 6(e). We decline to impute the OIG employees' knowledge to the DOL for purposes of triggering the statute of limitations. Like Gamble, the OIG employees' knowledge was wholly derived from their temporary participation in a sensitive Justice Department criminal investigation. Without the benefit of any clear legal dividing lines, they treated the materials involved in the investigation as confidential. Imputing their knowledge to the DOL would risk creating an incentive to err on the side of disclosure of grand jury materials rather than on the side of caution, and might discourage such interagency cooperation in criminal enforcement.

Finally, the DOL did not otherwise discover the kickback scheme before the June 1981 indictment. The additional documents discussed by C & A, including an article in Mother Jones magazine, do not establish actual or constructive discovery of the Chicago kickback scheme. Assuming that these documents were sufficient to give rise to suspicion on the DOL's part, the DOL exercised the required diligence, attempting to gain access to grand jury materials and Justice Department information. The information that the DOL succeeded in obtaining, however, did not reveal the Chicago scheme. Moreover, the DOL suspended its civil investigation at the specific request of the U.S. Attorney to avoid interference with the criminal investigation. The DOL's suit against the C & A defendants may proceed.21

Shortly after being sued by the DOL, the trustees filed suit against the C & A defendants for recovery of the kickbacks. The only question we must answer in regard to the trustees' claim is which limitations period applies--the six-year "fraud or concealment" period or the three-year "actual knowledge" provision. The three-year limitations period, if applicable, would be fatal to the trustees' claim, because it is undisputed that the trustees had actual knowledge of the kickback scheme in June 1981, when the indictment was returned. The trustees' complaint was filed on May 29, 1987.

The district court, applying the six-year limitations period, held that the trustees' suit was timely. While finding that the trustees had "serious reason to suspect wrongdoing on the part of C & A, Pilotto, and Caporale" before May 1981, the court found that the trustees had not actually discovered the kickback scheme until after that time. But the sort of "discovery" that triggers section 1113's "fraud or concealment" limitations period is not limited to actual discovery; it encompasses constructive discovery in cases such as this one. First, the fraudulent concealment doctrine, incorporated into section 1113 by Radiology Center, requires diligence on the part of the plaintiff in order to be applied at all, at least where self-concealing acts are concerned.22 See cases cited supra note 19. And second, the six-year "fraud or concealment" provision of section 1113 is not triggered by "actual discovery" or "actual knowledge" (like the neighboring three-year provision), but simply by "discovery." Given the provision's use of the general term, its contrast to the three-year provision's "actual" language and Radiology Center 's incorporation of the line of fraudulent concealment cases, we conclude that diligence must be required here. If plaintiffs would have been on prior notice of a violation through their exercise of due diligence, they may not now invoke the six-year "fraud or concealment" period.23

On the undisputed facts, there is no real question that the trustees ignored danger signals that were sufficient to put them on notice of the unlawful scheme.24 Unlike the DOL, therefore, the trustees cannot show the requisite due diligence to qualify for the six-year statute of limitations. Our treatment of the trustees' suit for recovery of the kickbacks diverges from that of the DOL's parallel suit for several reasons. First, the DOL's knowledge of the scheme was in the form of information obtained by certain employees who were loaned out to assist in the Justice Department's Strike Force investigation. Those DOL employees reasonably treated the information they received as confidential, and their knowledge may not therefore be imputed to the DOL. In contrast, the trustees themselves, in their capacity as fiduciaries, were put on notice of the scheme. Second, the DOL's civil investigation was suspended at the request of the U.S. Attorney in order to prevent interference with the grand jury investigation. Not only did the trustees have no justification for failing to conduct further inquiry, but they had a continuing affirmative duty to ensure that funds were not being misappropriated. Indeed, they proceeded to continue their dealings with C & A for several years after the scheme came to light. The trustees did not even sue to recover the kickbacks until after they had been sued by the DOL for failing to take action. Third, the trustees were closer to the situation than the DOL at all relevant times, and clearly had more direct potential access to the activities of fellow trustees Caporale and Pilotto and of C & A.

The trustees were on notice of improper payments in several ways but nevertheless chose to bury their heads in the sand. First, the trustees knew during the late 1970s that criminal investigations involving their contract with C & A were being conducted. Second, the grand jury's questioning gave further notice to the trustees and the plan administrator that ten percent of C & A's gross revenue was being paid to Pinckard in exchange for no services. A trustee was also asked about Caporale's and Pilotto's role in the C & A contract, and particularly whether they were the "major movers" behind amending the fund's contract with C & A. Third, as our discussion of the DOL's suit against the trustees indicates, the trustees continued to select C & A as the fund's dental services provider after failing to consider comparable bids and continued to deal with C & A despite indications of wrongdoing. Finally, even if the trustees did not have sufficient notice of the particular kickback scheme that was actually orchestrated, they surely did have notice of the wrongful payment of funds. Yet they failed to exercise diligence in investigating and recovering the wrongful payments.

The trustees are not entitled to application of the six-year "fraud or concealment" limitations period of section 1113. ERISA's three-year "actual knowledge" provision therefore bars their claim against the C & A defendants.

In the DOL's suit against the trustees, we affirm the district court's disposition of the bidding and monitoring claims. The claims based on bidding for the 1984 contract and monitoring activities before May 1, 1984 are barred and summary judgment was properly granted as to those claims; the claims based on bidding for the 1987 contract and monitoring activities after May 1, 1984 should be allowed and summary judgment was properly denied. As for the DOL's kickback recovery claim against the trustees, we find that this claim survives the statute of limitations and we reverse the entry of summary judgment.

In the DOL's suit against the C & A defendants for recovery of the kickbacks, we find that the DOL has met the criteria for application of the six-year "fraud or concealment" limitations period and that its suit is timely. We therefore affirm the district court's denial of summary judgment on this claim. In the trustees' suit against the C & A defendants, however, we find that the six-year limitations period is inapplicable and that their suit is time barred as a matter of law.

AFFIRMED IN PART, REVERSED IN PART AND REMANDED.

POSNER, Circuit Judge, concurring.

My reservations concern Judge Cudahy's analysis of the suit by the Department of Labor for breach of fiduciary obligation under the federal pension law (Employee Retirement Income Security Act--ERISA), and specifically his discussion of laches and fraudulent concealment. I join the rest of his opinion.

1. He appears to be of two minds on whether the doctrine of laches, which bars a plaintiff from maintaining a suit if he unreasonably delayed in filing it and as a result harmed the defendant, should apply to an ERISA case in which the plaintiff is the government. I think it should apply and I would like us to say it should apply, though I recognize that the case can be decided without resolving the issue.

Many decisions hold, on the basis of a hint in Occidental Life Ins. Co. v. EEOC, 432 U.S. 355, 373, 97 S.Ct. 2447, 2458, 53 L.Ed.2d 402 (1977), that laches is applicable to suits by the Equal Employment Opportunity Commission. They are the same kind of suit as this; they are suits on behalf of private individuals in which any damages collected in the suit are paid over to the individuals rather than retained by the agency. EEOC v. Massey-Ferguson, Inc., 622 F.2d 271, 275 (7th Cir.1980); EEOC v. Dresser Industries, Inc., 668 F.2d 1199 (11th Cir.1982); EEOC v. K-Mart Corp., 694 F.2d 1055, 1060-61 (6th Cir.1982); EEOC v. Bethlehem Steel Corp., 765 F.2d 427 (4th Cir.1985); see also EEOC v. Vucitech, 842 F.2d 936, 942-43 (7th Cir.1988); Houghton v. McDonnell Douglas Corp., 716 F.2d 526, 528 (8th Cir.1983); Annot., "Laches or Other Assertion of Untimeliness as Defense to Action Under Title VII ... Brought by Equal Employment Opportunity Commission," 67 A.L.R.Fed. 381 (1984); Mary Lynn Kelly, "Preventing Trial by Ambush: The Laches Defense in Title VII Suits," 8 Rev. Litigation 227, 239-40 and n. 90 (1989).

Judge Cudahy points out that there is no statute of limitations on suits by the EEOC, as there is on suits by the Department of Labor to enforce claims under ERISA, and he suggests that courts are and perhaps should be reluctant to invoke laches when a plaintiff is suing within the period specified by the legislature. The point could be used to support the opposite conclusion: a legislature that places no deadline on suits is presumably not worried about the consequences for defendants of having to defend against suits brought long after the alleged wrongdoing. However that may be, there is plenty of authority for applying laches in cases governed by a statute of limitations. Illustrative federal cases are Maksym v. Loesch, 937 F.2d 1237, 1248 (7th Cir.1991); K-Mart Corp. v. Oriental Plaza, Inc., 875 F.2d 907, 911 (1st Cir.1989), and Azalea Fleet, Inc. v. Dreyfus Supply & Machinery Corp., 782 F.2d 1455, 1459 (8th Cir.1986). Although Tandy Corp. v. Malone & Hyde, Inc., 769 F.2d 362, 365 (6th Cir.1985), says that the presumption is against invoking laches in such a case, it adds that the presumption is rebuttable. A few cases, illustrated by United States v. Mack, 295 U.S. 480, 489, 55 S.Ct. 813, 817, 79 L.Ed. 1559 (1935), and United States v. RePass, 688 F.2d 154, 158 (2d Cir.1982), hold that laches can never be used to cut down a statute of limitations, but, as will appear, Mack can be distinguished.

Laches is a species of estoppel and can therefore be applied whenever the requirements for estoppel are satisfied, and thus whenever there is delay (1) unwarranted in the circumstances that (2) causes harm to the defendant. Maksym v. Loesch, supra, 937 F.2d at 1248. This observation may seem to clinch the case against applying laches here, since ordinarily the federal government cannot be estopped to enforce its rights. It is no surprise, therefore, to read that "laches is not a defense against the sovereign," Costello v. United States, 365 U.S. 265, 281, 81 S.Ct. 534, 543, 5 L.Ed.2d 551 (1961), a declaration that led the Third Circuit in EEOC v. Great Atlantic & Pacific Tea Co., 735 F.2d 69, 80-81 (3d Cir.1984), to question whether laches was a defense to a suit by the EEOC. Judge Cudahy acknowledges, however, that laches will lie against the EEOC, while questioning whether it will lie against the Department of Labor, although the EEOC is no less a part of the federal government than the DOL is.

The familiar language of sovereignty, separation of powers, and important public interests that is used to rebuff claims to estop the government is misplaced in a case in which the government is suing not on its own behalf--as it was doing in Mack, a suit for forfeiture, or in Costello, a denaturalization proceeding--but instead as the representative of private interests. In an ERISA suit, just as in a suit by the EEOC, the invoking of laches to bar the government's suit would not take money out of the U.S. Treasury or interfere with the government's operations. It would not even deprive the government of a financial expectancy. Any money it won in this case would be paid into the pension plans against which the defendants committed a breach of trust. If any public agency has a stake in this case it is the Pension Benefit Guaranty Corporation, which has not filed an amicus brief or otherwise participated.

As the Great Atlantic & Pacific Tea Co. case illustrates, there is a tension in the lower-court cases, and the Supreme Court has not spoken with a single voice either. Compare United States v. Beebe, 127 U.S. 338, 8 S.Ct. 1083, 32 L.Ed. 121 (1888), and Occidental Life Ins. Co. v. EEOC, supra, with NLRB v. International Association of Ironworkers, 466 U.S. 720, 724-25, 104 S.Ct. 2081, 2083-84, 80 L.Ed.2d 715 (1984), and McAllister v. Magnolia Petroleum Co., 357 U.S. 221, 226 n. 7, 78 S.Ct. 1201, 1204 n. 7, 2 L.Ed.2d 1272 (1958). But it seems to me that either we accept the principle of the cases that allow laches to be asserted against the EEOC and apply it to the Department of Labor, or we reject it in both contexts. They cannot be distinguished. As should be apparent, I would accept it. I agree, however, with the government's alternative ground, that the elements of laches were not shown here.

2. The statute allows six years to bring suit from the date on which the violation is discovered, provided it is a "case of fraud or concealment." 29 U.S.C. § 1113. We have interpreted this to mean not that a fraud plaintiff has six years in which to sue but that any ERISA plaintiff can avail himself of the doctrine of fraudulent concealment. Radiology Center, S.C. v. Stifel, Nicolaus & Co., 919 F.2d 1216, 1220 (7th Cir.1990). I do not think it promotes clear thinking about fraudulent concealment to distinguish between "self-concealing wrongs" and "active concealment."

As another panel of this court explained in Cada v. Baxter Healthcare Corp., 920 F.2d 446, 450-53 (7th Cir.1990), federal statutes of limitations begin to run from the date on which the plaintiff discovered that he was injured (not when he discovers that he has a claim against the injurer--that would be contrary to United States v. Kubrick, 444 U.S. 111, 122, 100 S.Ct. 352, 359, 62 L.Ed.2d 259 (1979)). There are two grounds on which the plaintiff can, after discovering that he has been injured, delay the running of the period of limitations. The first is equitable tolling: even if the plaintiff knows that he has been injured, if despite all due diligence he is not able to file his complaint within the statutory period he can take such additional time to file as he needs. Equitable tolling, in other words, allows the plaintiff additional time (if necessary) to discover his claim after he discovers his injury --the latter being the date that triggers the running of the statute of limitations. Discovery of the injury starts the running of the statute; equitable tolling allows the plaintiff to arrest the running of the statute to give him time to discover that he has a claim.

The second ground for arresting the statute of limitations, equitable estoppel, of which fraudulent concealment is one type, means that the defendant has done something to prevent the plaintiff from filing his claim--for example, has promised not to plead the statute of limitations if the plaintiff will delay suing while the parties attempt to work out a settlement. The something does not have to be done after the plaintiff discovers he has a claim, although in the example I just gave it was. Suppose the plaintiff invested money with the defendant, and suffers a huge loss, and when he goes to the defendant and complains the defendant fobs him off with some elaborate, and elaborately fraudulent tale, as a result of which the plaintiff loses months in discovering that the defendant's wrongdoing was responsible for his financial disaster. If he had lost this time without fault on either his part or that of the defendant, the case would be governed by the doctrine of equitable tolling. The plaintiff would have to have exercised diligence and he would get no more time than he needed to bring his suit after learning the facts necessary to formulate his complaint. But since (in my example) the defendant concealed his wrong, the doctrine of equitable estoppel, not equitable tolling, is in play and the plaintiff does not have to demonstrate diligence (though as in other cases of estoppel he will have to show reasonable reliance) and he gets the full statutory period in which to sue.

Concretely, then, if the statute of limitations is five years, and the plaintiff discovers the facts he needs in order to file his complaint after four and a half years from when he learned that he had sustained an injury, if the defendant is guilty of fraudulent concealment the plaintiff will have a full five years from the date on which he discovered the facts he needed, while if the defendant was not guilty of fraudulent concealment or some other form of equitable estoppel (such as my first example, of a promise not to plead the statute of limitations) the plaintiff will have to file his suit as soon as he can. If fraudulent concealment had begun after the plaintiff discovered all the facts he needed in order to file his suit, the period of concealment would have to be added on to the statute of limitations. So if the statute of limitations began to run at the end of year 1 and fraudulent concealment began in year 3 and lasted until the same day in year 5, the plaintiff must (assuming a five-year statute of limitations), sue by the end of year 8 (1 + 5 + (5 - 3)).

Judge Cudahy is thus correct in footnote 17 that I "classif[y] fraudulent concealment as a species of estoppel (rather than equitable tolling) and would thereby dispense with the requirement of diligence on the part of the plaintiff." But I do not think he is correct in suggesting that in the petty interest of "tidy classification" I, along with the panel in Cada, have departed from the "scheme" of previous cases, a scheme to which Judge Cudahy, employing the royal "we" (for it is unclear whether Judge Ripple joins this part of his opinion), declares his adherence. For at the end of this footnote and in the following one Judge Cudahy makes clear his own view that a plaintiff need not prove diligence in a case of fraudulent concealment as I define it, which is equivalent to what he calls "active concealment that is separate from the underlying wrong." So there is no substantive disagreement. The confusion arises from Judge Cudahy's use of the term "fraudulent concealment" to embrace both what I mean by the term and what he calls "self-concealing fraud" and I would call simply "fraud." A plea to extend the statute of limitations in a simple fraud case in which there has been no effort to conceal the fraud beyond what was implicit in committing fraud in the first place is an appeal to the doctrine of equitable tolling. The plaintiff must act with due diligence to discover that the defendant's fraud was responsible for his injury, but if despite due diligence he can't discover this in time to sue within the statutory period he can take such additional time as is necessary. If he didn't even know he had been injured, the case would be covered by the discovery rule.

I don't know why Judge Cudahy insists on using the term "fraudulent concealment" to cover two categories of conduct with different legal consequences. It is not under the compulsion of the cases cited in his note 17. The reference to diligence in Bailey v. Glover, 88 U.S. (21 Wall.) 342, 348, 22 L.Ed. 636, 638 (1875), is to the plaintiff's discovering that he has a fraud claim (equitable tolling). Likewise Teamsters Local 282 Pension Trust Fund v. Angelos, 815 F.2d 452, 456 (7th Cir.1987). Suslick v. Rothschild Securities Corp., 741 F.2d 1000, 1004 (7th Cir.1984), confusingly uses the term "equitable tolling" to cover equitable estoppel as well, but makes clear that diligence is required of the plaintiff only if the defendant has not interfered with the plaintiff's ability to bring a timely suit. Tomera v. Galt, 511 F.2d 504, 510 (7th Cir.1975), is the same. Sperry v. Barggren, 523 F.2d 708, 711 (7th Cir.1975), states clearly that only if there is no concealment is there a duty of diligence. Nothing in these cases is inconsistent with either Cada or this concurrence.

Except the terminology of some of the cases. Cada tried to make the terminology in this difficult area consistent. Clarification is not disagreement and should not create "tension." It is the distinction between "self-concealing acts" and "active concealment" that is unhelpful. "Active concealment" is redundant; concealment is not a passive state. "Self-concealing" sounds good, but what does it mean? The important distinction, in a case in which the plaintiff's claim is itself for fraud, is between the fraud that constitutes the claim and the fraud designed to conceal the knowledge of the original fraud from the victim or other potential enforcer. The distinction is straight-forward enough in this case. The original fraud was the paying of kickbacks to the welfare plan's trustees by the provider of dental services to the beneficiaries of the plan. The acts of concealment were the acts by which the parties to the kickbacks sought through the use of dummy corporations and the like to make it difficult for anyone to discover that kickbacks were being paid. Of course people who commit frauds often try to conceal them, as distinct from merely hoping that no one will notice so long as they don't advertise the fraud. But there is no reason why such persons should benefit from compounding their frauds, as they would if their efforts at concealment, to the extent successful, shortened the period of limitations. We should bear in mind the distinction between the discovery of the loss or other injury and the discovery of the facts necessary to formulate a complaint, the discovery in other words of the claim. The statute of limitations begins to run from the former date, not the latter, unless tolled. Fraudulent concealment is one ground for tolling, and in a fraud case it requires proof that the defendant did something (beyond merely praying) to prevent the world in general or the plaintiff in particular from discovering the fraud.

There is an added wrinkle in an ERISA case because the statute specifies a six-year period of limitations for a case of fraudulent concealment, rather than just tolling the period for so long as the concealment lasts. This seems to be the substitution of a mechanical rule for an elastic common law doctrine: however long or short the concealment lasts, concealment extends the statute of limitations from three years to six. Whether a plaintiff can use equitable tolling to extend the six years we fortunately need not decide in this case. (I can find no cases on the point.) Thus the statement in the Radiology Center opinion that section 1113 applies fraudulent concealment to ERISA cases requires careful qualification. Were it not for the statute, the doctrine would apply in the way that I have been describing. With the statute, it appears that the statute of limitations is automatically extended by three years--but just three years--regardless of the length of time that the defendant fraudulently conceals his wrong, although we can leave open the question whether if the fraudulent concealment persisted throughout the entire six-year period the plaintiff might be able to invoke the common law doctrine to extend the period further. Another question we can leave open for now is whether the statute permits a plaintiff to plead forms of equitable tolling that do not involve concealment, such as a promise not to plead the statute of limitations.

I regret the failure to use this case as a vehicle for clarifying issues that have been a recurrent source of uncertainty and confusion.

RIPPLE, Circuit Judge, concurring.

I join the judgment of the court and the essential reasoning of the principal opinion. Because the elements of laches have not been met in the case before us, there is no necessity to confront definitively whether such a defense is ever available when the Secretary of Labor brings an ERISA action. I note, however, that it would be an overreaction to read the principal opinion as suggesting the inevitability of a holding that the doctrine of laches does apply. As the principal opinion notes, there are several good reasons that justify treating the ERISA situation as substantially different from EEOC cases.