WALD, Circuit Judge:

This case presents issues of first impression concerning the 1986 amendments to the qui tam provisions of the False Claims Act, 31 U.S.C. § 3729 et seq. ("FCA"). Under these provisions, private persons acting on behalf of the government may sue those who defraud the government and share in any proceeds ultimately recovered. Since 1943, the FCA has included a jurisdictional limitation on qui tam actions, the most recent form of which bars suits that are "based upon the public disclosure of allegations or transactions in a criminal, civil, or administrative hearing, in a congressional, administrative, or Government Accounting Office report, hearing, audit, or investigation, or from the news media, unless . . . the person bringing the action is an original source of the information." 31 U.S.C. § 3730(e)(4)(A). Springfield Terminal Railway Co. and its President, David Fink (collectively, "Springfield"), brought suit under the qui tam provisions against Dr. Francis X. Quinn, an arbitrator appointed by the National Mediation Board ("NMB") to resolve a labor dispute between Springfield and its union, alleging that Quinn had fraudulently billed the government for services not actually rendered in connection with the arbitration proceedings. The district court dismissed Springfield's action on the grounds that it violated the FCA's jurisdictional limitation, and Springfield filed this timely appeal. The central issue in the case is under what circumstances a qui tam action will be deemed to be "based upon the public disclosure of allegations or transactions" within the meaning of § 3730(e)(4)(A). Because we find that the district court operated under an unduly broad reading of the jurisdictional bar, we remand this case for further proceedings.

I. BACKGROUND

This case originated as a labor dispute between Springfield, a rail common carrier, and the union representing its employees. After the parties proved unable to resolve their disagreement privately, the NMB appointed a panel of federal arbitrators pursuant to the Railway Labor Act, 45 U.S.C. §§ 153- 54 (1988) ("RLA"). An initial panel determined the dispute was properly referable to arbitration under the RLA, and a merits panel consisting of Dr. Francis X. Quinn as neutral member and one representative each from the railroad and its union began hearings in April 1988. Springfield immediately initiated an action in the United States District Court of Maine challenging the necessity for a merits panel on the grounds that the union had not satisfied all of the procedural prerequisites to federal mediation. See Springfield Terminal Ry. Co. v. United Transp. Union, 767 F.Supp. 333, 335-36 (D.Me. 1991). On June 13, 1988, before the district court took any action on Springfield's suit, the merits panel issued a decision in favor of the union. See Public Law Board No. 4462 Decision and Award (June 13, 1988), reprinted in Joint Appendix ("J.A.") at 86-93. The panel left to the parties the resolution of back pay issues. When the parties again were unable to reach agreement, the panel reasserted jurisdiction and issued a series of short decisions pertaining to back pay and benefits.

Springfield filed a supplemental petition in the Maine district court seeking to set aside all of the merits panel's awards. In its complaints, Springfield alleged, inter alia, that both panels had misconstrued the RLA in submitting the case to arbitration, that the court rather than an arbitration board was the appropriate body to decide the dispute, that Springfield was denied due process of law because of arbitrators' ex parte communications, and that narrow judicial review of arbitral awards violated the separation of powers doctrine of the Constitution. After obtaining the arbitrators' pay vouchers through discovery and filing the relevant discovery materials with the court, Springfield appended a claim that neutral merits arbitrator Quinn had billed the government for activities unrelated to the arbitration proceedings. On cross-motions for summary judgment, the Maine district court awarded summary judgment to the union with respect to Springfield's challenges to the necessity for arbitration, but vacated the merits panel's actual awards due to the panel's reliance on an erroneous computer database version of the Federal Railroad Safety Act of 1970, 45 U.S.C. § 441(b)(1). The court then remanded the case to the merits panel for redetermination of the award under the correct version of the statute without reaching the allegations pertaining to Quinn's billing. See Springfield Terminal Ry., 767 F.Supp. at 347.

In August 1991, two months after the Maine court's remand to the arbitration panel, Springfield filed this action in the United States District Court for the District of Columbia under the qui tam provisions of the FCA, 31 U.S.C. § 3729 et seq. In its complaint, Springfield alleged once again that Dr. Quinn had fraudulently billed the government for days on which he had not actually worked on Springfield's dispute. Springfield contended that its suspicions first arose upon inspection of Quinn's pay vouchers, which were produced by civil discovery processes in the Maine litigation. Based upon its own involvement in the arbitration, Springfield recognized that Dr. Quinn had no arbitral function to perform on several of the days for which he sought pay. According to Springfield, it then conducted further investigation on its own, calling several of the numbers listed on Quinn's telephone records that had been disclosed during discovery. Springfield argues that it was this inquiry that brought to light fraudulent behavior on Quinn's part — revealing, for example, that he had been in Canada attending a national arbitration conference for a three-day period during which he had billed the government for services pertaining to the arbitration. As required by the FCA, 31 U.S.C. § 3730(b)(2), Springfield provided the information forming the substance of its complaint to the Department of Justice, which issued notice of its declination to appear on the government's behalf on December 6, 1991.

Dr. Quinn moved to dismiss the action under the jurisdictional bar of the FCA, 31 U.S.C. § 3730(e)(4)(A), which prohibits qui tam actions that are "based upon the public disclosure of allegations or transactions in a criminal, civil, or administrative hearing . . . [unless] the person bringing the action is an original source of the information," and for a violation of FED.R.CIV.P. 9(b), which requires that averments of fraud "be stated with particularity." Bypassing the particularity charge, the district court granted Quinn's motion to dismiss for lack of subject matter jurisdiction under § 3730(e)(4)(A) in July 1992. The court reasoned that Springfield's qui tam claim was "based upon" information unearthed during discovery in the Maine litigation and so fell under the "public disclosure" ban of the FCA. The court rejected Springfield's contention that the qui tam allegations were based on its first-hand knowledge of events in the arbitration proceeding and its additional investigative efforts sparked by discovery materials that were innocuous in their own right. Turning then to the "original source" exception of § 3730(e)(4)(A), the court held that information obtained through discovery was of necessity obtained through an intermediary, and thus did not fall within the statutory exception. The court viewed the additional information about Quinn's activities gleaned from Springfield's own investigation as mere "background," and so did not engage in "original source" analysis with respect to it.

II. ANALYSIS A.

This case requires us to chart the changed landscape of qui tam actions in the aftermath of the 1986 amendments to the False Claims Act. The FCA's qui tam provisions, which have existed in various incarnations since 1863, provide cash bounties in certain circumstances to private citizens who successfully bring suit against those who defraud the federal government. Qui tam provisions are designed to set up incentives to supplement government enforcement, and at their best may "compare with the ordinary methods as the enterprising privateer does to the slowgoing public vessel." United States v. Griswold, 24 F. 361, 366 (D.Or. 1885). On the downside, overly generous qui tam provisions present the danger of parasitic exploitation of the public coffers, as exemplified by the notorious plaintiff who copied the information on which his qui tam suit was based from the government's own criminal indictment. See United States ex rel. Marcus v. Hess, 317 U.S. 537, 63 S.Ct. 379, 87 L.Ed. 443 (1943). Seeking the golden mean between adequate incentives for whistle-blowing insiders with genuinely valuable information and discouragement of opportunistic plaintiffs who have no significant information to contribute of their own, Congress has frequently altered its course in drafting and amending the qui tam provisions since initial passage of the FCA over a century ago.

The past serves as prologue; some familiarity with their tortuous, wending history is critical to an understanding of the latest set of amendments to the qui tam provisions. The False Claims Act of 1863 was adopted during the Civil War in order to combat fraud and price-gouging in war procurement contracts. See United States ex rel. LaValley v. First Nat'l Bank of Boston, 707 F.Supp. 1351, 1354 (D.Mass. 1988); JOHN T. BOESE, CIVIL FALSE CLAIMS AND QUI TAM ACTIONS 1-3 (1993). In addition to creating stiff civil and criminal penalties for fraud, the Act allowed any person to bring suit against the offending profiteers under its qui tam provisions, promising successful qui tam plaintiffs one-half of the damages and forfeitures ultimately recovered and collected. See S.REP. No. 345, 99th Cong., 2d Sess., at 10 (1986), reprinted in 1986 U.S.C.C.A.N. 5266, 5275. The original qui tam provisions were, in the words of an earlier court,

passed upon the theory, based on experience as old as modern civilization, that one of the least expensive and most effective means of preventing frauds on the Treasury is to make the perpetrators of them liable to actions by private persons acting, if you please, under the strong stimulus of personal ill will or the hope of gain.

United States v. Griswold, 24 F. 361, 366 (D.Or. 1885). The 1863 Act required qui tam plaintiffs to bear the costs of their litigation efforts, in order to discourage frivolous lawsuits. However, it contained no requirement that the allegations of fraud originate from the investigative efforts of the plaintiffs themselves, and it did not prohibit plaintiffs from bringing suits based exclusively on information that was already in the government's possession. Despite this invitation for abuse, the qui tam provisions were used sparingly during their first half-century. See LaValley, 707 F.Supp. at 1354.

In 1943, however, after a decade in which New Deal and World War II government contracts boomed and qui tam suits correspondingly surged, the Supreme Court in United States ex rel. Marcus v. Hess, 317 U.S. 537, 63 S.Ct. 379, 87 L.Ed. 443 (1943), spotlighted the pitfalls of the overly generous qui tam provisions then in effect. In Hess, a qui tam relator brought suit under the 1863 Act alleging that electrical contractors had engaged in collusive bidding on government contracts. Prior to the plaintiff's qui tam action, the same defendants had been criminally indicted for the same behavior and fined, after a plea of nolo contendere, a sum of $54,000. The Supreme Court nonetheless permitted the qui tam plaintiff's suit (and concomitant share of the bounty), reasoning that, even if the qui tam plaintiff had brought no new information to the fore and had merely copied the government's indictment, the suit advanced the purposes of the FCA by affording a recovery to the government in excess of the criminal penalty. See id. at 545, 63 S.Ct. at 385. The Court found neither express impediment to the suit in the text of the Act nor any intent to impose one in its legislative history, see id. at 546, 63 S.Ct. at 385, and declined to establish a judicial bar on its own initiative.

Hess inspired public outcry over the liberality of the qui tam provisions that prompted speedy congressional response. Eleven months after the opinion was handed down, President Roosevelt signed tightening amendments to the False Claims Act. The amendments were the product of careful compromise: the House bill would have repealed the qui tam provisions altogether, while the Senate bill barred qui tam jurisdiction for suits based on information already in possession of the government unless the information was "original with such person." 89 CONG.REC. 510,744 (daily ed. Dec. 16, 1943). The Senate's original source provision, however, was dropped in conference without explanation. See United States ex rel. Stinson, Lyons, Gerlin Bustamante, P.A. v. Prudential Ins. Co., 944 F.2d 1149, 1153 (3d Cir. 1991); United States ex rel. LaValley v. First Nat'l Bank of Boston, 707 F.Supp. 1351, 1355 (D.Mass. 1988); see also JOHN T. BOESE, CIVIL FALSE CLAIMS AND QUI TAM ACTIONS 1-12 n. 38 (1993). The final version of the amendment thus barred qui tam suits that were "based on evidence or information the Government had when the action was brought." 31 U.S.C. § 3730(b)(4) (1982) (superseded).

Once again, the passage of time revealed that Congress, in its attempt to evade Scylla, had steered precipitously close to Charybdis. The new statutory barriers substantially decreased the use of qui tam provisions to enforce the FCA, see BOESE at 1-12, and courts greeted those qui tam suits that did arise with considerable caution. The case of United States ex rel. Wisconsin v. Dean, 729 F.2d 1100 (7th Cir. 1984), marked the nadir of the qui tam action. In Dean, the Seventh Circuit held that the FCA barred a qui tam suit "whenever the government has knowledge of the `essential information upon which the suit is predicated' before the suit is filed, even when the plaintiff is the source of that knowledge." Id. at 1103 (citing United States ex rel. Weinberger v. Florida, 615 F.2d 1370, 1371 (5th Cir. 1980)). The court dismissed the State of Wisconsin's qui tam action alleging Medicaid fraud that the state's own investigation had uncovered, because the state had already reported the fraud to the federal government as required under the terms of its participation in the Medicare reimbursement program, and had further used the information to prosecute the defendants on state criminal grounds. Noting that the original source provision in the 1943 Senate bill had mysteriously vanished in conference, the court found no clear intent to preserve it in the legislative history. The court concluded that, "[i]f the State of Wisconsin desires a special exemption to the False Claims Act because of its requirement to report Medicaid fraud to the federal government, then it should ask Congress to provide the exemption." Dean, 729 F.2d at 1106.

Predictably, the National Association of Attorneys General immediately adopted a resolution urging Congress "to rectify the unfortunate result" of the Dean case. S.REP. No. 345, 99th Cong., 2d Sess., at 13, reprinted in 1986 U.S.C.C.A.N. at 5278. Congress responded with the False Claims Amendment Act of 1986, the avowed purpose of which was "to enhance the Government's ability to recover losses sustained as a result of fraud against the Government." Id. at 1, reprinted in 1986 U.S.C.C.A.N. at 5266. Concerned about "sophisticated and widespread fraud" depleting the national fisc, the Senate Report concluded that "only a coordinated effort of both the Government and the citizenry will decrease this wave of defrauding public funds. [Accordingly, the Senate bill] increases incentives, financial and otherwise, for private individuals to bring suits on behalf of the Government." Id. at 1-2, reprinted in 1986 U.S.C.C.A.N. at 5266-67.

The 1986 amendments, representing still another congressional effort to reconcile avoidance of parasitism and encouragement of legitimate citizen enforcement actions, wrought several changes in the qui tam provisions. On the one hand, Congress clearly acted upon its expressed intention to "encourage more private enforcement suits," S.REP. No. 345, 99th Cong., 2d Sess., at 23-24, reprinted in 1986 U.S.C.C.A.N. at 5288-89, by enacting the "original source" exception that had somehow slipped through the cracks in 1943 and that probably would have saved the qui tam action in Dean. To enhance incentives, the amendments also increased monetary awards, adopted a lower burden of proof, and allowed qui tam plaintiffs to continue to participate in the actions after intervention by the government. See United States ex rel. Stinson, Lyons, Gerlin Bustamante, P.A. v. Prudential Ins. Co., 944 F.2d 1149, 1154 (3d Cir. 1991). On the other hand, Congress left in place barriers to parasitic lawsuits. The 1986 amendment changed the language of the jurisdictional bar, however, so that it now reads:

No court shall have jurisdiction over an action under this section based upon the public disclosure of allegations or transactions in a criminal, civil, or administrative hearing, in a congressional, administrative, or Government Accounting Office report, hearing, audit, or investigation, or from the news media, unless the action is brought by the Attorney General or the person bringing the action is an original source of the information.

31 U.S.C. § 3730(e)(4)(A).

The history of the FCA qui tam provisions demonstrates repeated congressional efforts to walk a fine line between encouraging whistle-blowing and discouraging opportunistic behavior. The 1986 amendments inevitably reflect the long process of trial and error that engendered them. They must be analyzed in the context of these twin goals of rejecting suits which the government is capable of pursuing itself, while promoting those which the government is not equipped to bring on its own.

B.

The 1986 qui tam amendments set up a two-part test for determining jurisdiction. First, the reviewing court must ascertain whether the "allegations or transactions" upon which the suit is based were "public[ly] disclos[ed]" in a "criminal, civil, or administrative hearing, in a congressional, administrative, or Government Accounting Office report, hearing, audit or investigation, or from the news media." 31 U.S.C. § 3730(e)(4)(A). If — and only if — the answer to the first question is affirmative, see Wang v. FMC Corp., 975 F.2d 1412, 1416 (9th Cir. 1992), will the court then proceed to the "original source" inquiry, under which it asks whether the qui tam plaintiff "has direct and independent knowledge of the information on which the allegations are based." 31 U.S.C. § 3730(e)(4)(B). Under these circumstances, if the qui tam plaintiff qualifies as an original source, the action may proceed; if she does not, the action is barred.

In applying the jurisdictional bar to Springfield's suit, the district court answered yes to the first (public disclosure) question by focusing principally on the material Springfield received during discovery in the Maine litigation. The court first equated discovery with "public disclosure" in a "civil hearing." It then determined that Springfield's suit was "based upon" the arbitrator's pay vouchers and telephone records disclosed in discovery. Over Springfield's objection that its own investigative efforts and personal knowledge of the arbitration proceedings had substantially contributed to the exposure of fraud, the court insisted that it was "not enough that some of the information that forms the basis of the suit came from non-public sources. . . . The non-public information must be `substantive information about the particular fraud, rather than merely background information which enables a putative relator to understand the significance of a publicly disclosed transaction or allegation.'" United States ex rel. Springfield Terminal Ry. Co. v. Quinn, Civ. Action No. 91-2081, at 8 (D.D.C. July 14, 1992) (citing United States ex rel. Stinson, Lyons, Gerlin Bustamante, P.A. v. Prudential Ins. Co., 944 F.2d 1149, 1154 (3d Cir. 1991)). The court characterized Springfield's contribution as "no more than any other would-be qui tam litigant with the same publicly available information could have done." Id. at 9. Turning to the second question, the court determined that Springfield did not qualify as an original source for the discovery materials, because information produced in discovery was "by its very nature" obtained through the intermediary action of the producing party. Id. at 10. Accordingly, the court dismissed Springfield's action under § 3730(e)(4)(A).

1. "Public disclosure"

In reviewing the district court's conclusions with respect to the issue of public disclosure, we agree that discovery material, when filed with the court (and not subject to protective order), is "public[ly] disclos[ed]" in a "civil hearing" for purposes of § 3730(e)(4)(A)'s jurisdictional bar. See United States ex rel. Kreindler Kreindler v. United Technologies Corp., 985 F.2d 1148, 1158 (2d Cir.), cert. denied, ___ U.S. ___, 113 S.Ct. 2962, 125 L.Ed.2d 663 (1993). Such was the status of the pay vouchers and phone records at issue here. Although Springfield argues that the word "hearing" in common parlance suggests relatively formal proceedings open to the general public, see Appellants' Brief at 27, this contention is belied by the fact that courts frequently employ the term to connote informal, "paper" proceedings. See United States v. Florida East Coast Ry. Co., 410 U.S. 224, 239, 93 S.Ct. 810, 818, 35 L.Ed.2d 223 (1973) (allowing paper hearings after concluding that "[t]he term `hearing' in its legal context undoubtedly has a host of meanings."); KENNETH C. DAVIS, ADMINISTRATIVE LAW TREATISE § 6-22 (1978) ("[T]he word `hearing' no longer means receiving sound through the ears. One may now `hear' the written words on a piece of paper."). It is clear from statutory context that the term "hearing" was intended to apply in a board context of legal proceedings under § 3730(e)(4)(A). Otherwise, the indictment in United States ex rel. Marcus v. Hess, 317 U.S. 537, 63 S.Ct. 379, 87 L.Ed. 443 (1943), would not constitute public disclosure within the meaning of the new amendments, see United States ex rel. Stinson, Lyons, Gerlin Bustamante, P.A. v. Prudential Ins. Co., 944 F.2d 1149, 1155 (3d Cir. 1991), a result that cannot reasonably be attributed to Congress. If court documents could be copied at will to provide the basis for qui tam suits, a half-century of precedent would be swiftly refuted without a flicker of recognition from Congress. We agree with the district court that for purposes of § 3730(e)(4)(A), "hearing" is roughly synonymous with "proceeding."

We do, however, restrict that interpretation to discovery material such as that involved here which is actually made public through filing, as opposed to discovery material which has not been filed with the court and is only theoretically available upon the public's request. See Stinson, 944 F.2d at 1162 (Scirica, J., dissenting). Until discovery materials are filed with the court, we doubt that the discovery process conducted between two private litigants could itself constitute a public disclosure within the meaning of § 3730(e)(4)(A). Cf. Seattle Times Co. v. Rhinehart, 467 U.S. 20, 33, 104 S.Ct. 2199, 2208, 81 L.Ed.2d 17 (1984) ("[Discovery] proceedings were not open to the public at common law, and, in general, they are conducted in private as a matter of modern practice."). Certainly the incentive structure adopted by Congress indicates otherwise. If discovery materials are not filed with the court, they are only potentially in the public eye. If they are not yet in the public eye, no rational purpose is served — and no "parasitism" deterred — by preventing a qui tam plaintiff rom bringing suit based on their contents. To bar a qui tam suit under these circumstances would prevent the utilization for enforcement purposes of allegations or transactions that may not otherwise come to the attention of the authorities. Although different filing rules for discovery materials in different jurisdictions may result in some disparity as to when discovery materials fall under the public disclosure bar, we believe the aims of the FCA are best served by this dividing line.

2. "Based upon . . . allegations or transactions"

A finding that the filed discovery material was publicly disclosed, however, does not by any means end the inquiry. The jurisdictional provision bars only those qui tam actions that are "based upon the public disclosure of allegations or transactions." 31 U.S.C. § 3730(e)(4)(A). Congress did not prescribe by mathematical formulae the quantum or centrality of nonpublic information that must be in the hands of the qui tam relator in order for suits to proceed. When, as here, some information relied upon by the qui tam plaintiff is undeniably in the public domain, the task of ensuring that qui tam suits are limited to those in which the relator has contributed significant independent information can prove tricky.

We find, however, that in the context of this case the district court's interpretation of "based upon the public disclosure of allegations or transactions" erected too high a threshold for the qui tam plaintiff. In dismissing Springfield's suit, the district court assumed without analysis that the pay vouchers and telephone records disclosed during discovery constituted "allegations or transactions" within the meaning of the jurisdictional bar. We disagree with that assumption. As the Ninth Circuit recently recognized, "Courts sometimes speak loosely of barring a qui tam suit because it is based on `publicly disclosed information.' But the Act bars suits based on publicly disclosed `allegations or transactions,' not information." Wang v. FMC Corp., 975 F.2d 1412, 1418 (9th Cir. 1992) (citing United States ex rel. Dick v. Long Island Lighting Co., 912 F.2d 13, 17 (2d Cir. 1990); Houck v. Folding Carton Admin. Committee, 881 F.2d 494, 504 (7th Cir. 1989), cert. denied, 494 U.S. 1026, 110 S.Ct. 1471, 108 L.Ed.2d 609 (1990)). We too find a distinction between "allegations or transactions" and ordinary "information" as a matter of common usage and sound interpretation of the FCA. The pay vouchers and telephone records disclosed during discovery — the only public information considered by the district court — were not in and of themselves sufficient to constitute "allegations or transactions" of fraudulent conduct within the meaning of the FCA jurisdictional bar.

Both plain meaning and a consideration of the aims of the statute direct this conclusion. The legislative history is admittedly silent with regard to the significance Congress attached to its choice of words in this much-amended section. However, in common parlance, the term "allegation" connotes a conclusory statement implying the existence of provable supporting facts. See WEBSTER'S THIRD NEW INTERNATIONAL DICTIONARY 55 (1976). The term "transaction" suggests an exchange between two parties or things that reciprocally affect or influence one another. See id. at 2425. On the basis of plain meaning, and at the risk of belabored illustration, if X + Y = Z, Z represents the allegation of fraud and X and Y represent its essential elements. In order to disclose the fraudulent transaction publicly, the combination of X and Y must be revealed, from which readers or listeners may infer Z, i.e., the conclusion that fraud has been committed. The language employed in § 3730(e)(4)(A) suggests that Congress sought to prohibit qui tam actions only when either the allegation of fraud or the critical elements of the fraudulent transaction themselves were in the public domain.

Our reading, however, does not rest on plain meaning alone. "In determining the meaning of the statute, we look not only to the particular statutory language, but to the design of the statute as a whole and to its object and policy." Crandon v. United States, 494 U.S. 152, 158, 110 S.Ct. 997, 1001, 108 L.Ed.2d 132 (1990). Many potentially valuable qui tam suits would be aborted prematurely by a reading of the jurisdictional provision that barred suits when the only publicly disclosed information was itself innocuous. In terms of the mathematical illustration, when X by itself is in the public domain, and its presence is essential but not sufficient to suggest fraud, the public fisc only suffers when the whistle-blower's suit is banned. When X and Y surface publicly, or when Z is broadcast, however, there is little need for qui tam actions, which would tend to be suits that the government presumably has chosen not to pursue or which might decrease the government's recovery in suits it has chosen to pursue. The cogent analysis in United States ex rel. Joseph v. Cannon, 642 F.2d 1373 (D.C. Cir. 1981), cert. denied, 455 U.S. 999, 102 S.Ct. 1630, 71 L.Ed.2d 865 (1982), construing the pre-1986 qui tam provisions, is instructive in illuminating this tension:

To require that the evidence and information possessed by the United States be a mirror image of that in the hands of the qui tam plaintiff would virtually eliminate the bar. On the other hand, to permit the bar to be invoked when the United States possesses only rumors while the qui tam plaintiff has evidence and information would be to permit the bar to repeal effectively much of the False Claims Act. Between these extremes lies the answer.

More precisely, the answer rests in that area where it is possible to say that the evidence and information in the possession of the United States at the time the False Claims Act suit was brought was sufficient to enable it adequately to investigate the case and to make a decision whether to prosecute.

The question, properly, then, is whether the information conveyed [to the government] could have formed the basis for a governmental decision on prosecution, or could at least have alerted law-enforcement authorities to the likelihood of wrongdoing. . . .

Id. at 1377 (quoting Pettis ex rel. United States v. Morrison-Knudsen Co., 577 F.2d 668, 674 (9th Cir. 1978)). We believe that Congress, by its careful choice of terminology in the 1986 amendments, navigated between the two extremes recognized in Cannon to create a standard under which qui tam actions are barred only when enough information exists in the public domain to expose the fraudulent transaction (the combination of X and Y), or the allegation of fraud (Z). When either of these conditions is satisfied, the government itself presumably can bring an action under the FCA and there is no place in the enforcement scheme for qui tam suits.

Section 3730(e)(4)(A)'s bar focuses more on the quantum of information already in the public sphere than on the amount or quality of information brought forward by the qui tam plaintiff. Nonetheless, the "based upon . . . allegations or transactions" prohibition has certain additional consequences for the plaintiff's case. Obviously, if the elements of the fraudulent transaction (X + Y) are already public, plaintiff's additional information, even if nonpublic, cannot suffice to surmount the jurisdictional hurdles. Thus, a qui tam action cannot be sustained where all of the material elements of the fraudulent transaction are already in the public domain and the qui tam relator comes forward with additional evidence incriminating the defendant. See United States ex rel. Precision Co. v. Koch Indus., Inc., 971 F.2d 548, 552-53 (10th Cir. 1992), cert. denied, ___ U.S. ___, 113 S.Ct. 1364, 122 L.Ed.2d 742 (1993). Similarly, there may be situations in which all of the critical elements of fraud have been publicly disclosed, but in a form not accessible to most people, i.e., engineering blueprints on file with a public agency. Expertise in the field of engineering would not in itself give a qui tam plaintiff the basis for suit when all the material elements of fraud are publicly available, though not readily comprehensible to nonexperts. See United States ex rel. Stinson, Lyons, Gerlin Bustamante, P.A. v. Prudential Ins. Co., 944 F.2d 1149, 1160 (3d Cir. 1991) ("[T]he relator must possess substantive information about the particular fraud, rather than merely background information which enables a putative relator to understand the significance of a publicly disclosed transaction or allegation."). However, where only one element of the fraudulent transaction is in the public domain ( e.g., X), the qui tam plaintiff may mount a case by coming forward with either the additional elements necessary to state a case of fraud ( e.g., Y) or allegations of fraud itself ( e.g., Z).

The parties in this case do not dispute either the amount or nature of evidence disclosed in discovery or the fact that Springfield relied in part upon this evidence — pay vouchers and telephone records — in formulating its complaint. Resolution of the "allegations or transactions" issue thus does not turn on "a determination of fact which only a fact-finder could make but which has not been made." See United States v. Garrett, 720 F.2d 705, 710 (D.C. Cir. 1983) (citation omitted), cert. denied, 465 U.S. 1037, 104 S.Ct. 1311, 79 L.Ed.2d 708 (1984). Instead, this case asks us to review legal conclusions pertaining to the "allegations or transactions" language that the district court drew from undisputed facts. In such circumstances we engage in independent review of the legal sufficiency of the district court's views and of its application of the law to undisputed facts. See Herbert v. National Academy of Sciences, 974 F.2d 192, 197 (D.C. Cir. 1992); Hohri v. United States, 782 F.2d 227, 241 (D.C. Cir. 1986), rev'd on other grounds, 482 U.S. 64, 107 S.Ct. 2246, 96 L.Ed.2d 51 (1987); cf. Auerbach v. Sverdrup Corp., 829 F.2d 175, 181 (D.C. Cir. 1987) (ordering remand where additional facts could be introduced and inferences drawn by trial court), cert. denied, 485 U.S. 905, 108 S.Ct. 1075, 99 L.Ed.2d 234 (1988).

Turning then to the case at hand, we conclude that the information disclosed via discovery cannot rise to the level of "allegations or transactions" so as to prevent the exercise of jurisdiction. Fraud requires recognition of two elements: a misrepresented state of facts and a true state of facts. The presence of one or the other in the public domain, but not both, cannot be expected to set government investigators on the trial of fraud. In pertinent part, liability under the FCA is triggered when a person "knowingly presents, or causes to be presented, to an officer or employee of the United States Government or a member of the Armed Forces of the United States a false or fraudulent claim for payment or approval," 31 U.S.C. § 3729(a)(1) (emphasis added), or "knowingly makes, uses or causes to be made or used, a false record or statement to get a false or fraudulent claim paid or approved by the Government," 31 U.S.C. § 3729(a)(2) (emphasis added). Knowledge of the allegedly misrepresented state of affairs — which does not necessarily entail knowledge of the fact of misrepresentation — is always in the possession of the government. Just so here. Publication of the facially valid pay vouchers, already in the government's pocket, cannot by itself be sufficient to defeat qui tam jurisdiction. Indeed, the entire qui tam regime is premised on the idea that the government's knowledge of misrepresented claims against the federal fisc (without knowledge that they are misrepresented) does not in itself translate into effective enforcement of the laws against fraud. See S.REP. No. 345, 99th Cong., 2d Sess., at 2, reprinted in 1986 U.S.C.C.A.N. at 5267 ("[O]nly a coordinated effort of both the Government and the citizenry will decrease this wave of defrauding public funds.").

Nor does the additional presence of publicly disclosed telephone records alter the calculus here. The telephone records did not suggest any misrepresentation on Quinn's part in submitting pay vouchers. At most, the records provided a convenient vehicle by which Springfield, already suspicious because of its personal experience with the arbitration, could pursue its independent investigation. Recognizing that the task of determining whether "allegations or transactions" have been "public[ly] disclos[ed]" will never be cut-and-dried, we nonetheless are confident in this case that the information put in the public domain by the discovery filing did not present so clear or substantial an indication of foul play as to qualify as either an allegation of fraud or a fraudulent transaction upon which a qui tam suit could be based.

3. Original Source

Because we believe the district court erred in concluding that the pay vouchers and telephone records represented publicly disclosed allegations or transactions within the meaning of § 3730(e)(4)(A), we do not find it necessary to critique its original source analysis with respect to these documents. Nonetheless, we are still required to proceed to our own original source analysis for allegations of fraud placed in the public domain by Springfield itself. Springfield first brought its claim alleging Quinn's fraud to the fore in the Maine litigation after conducting its own investigation of the circumstances surrounding the pay vouchers obtained in discovery. Because the actual allegations of fraud were thus publicly disclosed, Springfield must now qualify as an "original source" of those allegations in order to bring its qui tam suit. See JOHN T. BOESE, CIVIL FALSE CLAIMS AND QUI TAM ACTIONS, 4-34 (1993).

An "original source" is defined in § 3730(e)(4)(B) as "an individual who has direct and independent knowledge of the information on which the allegations are based." Other courts have — we think correctly — construed this definition to impose a conjunctive requirement — direct and independent — on qui tam plaintiffs. See United States ex rel. Stinson, Lyons, Gerlin Bustamante, P.A. v. Prudential Ins. Co., 944 F.2d 1149, 1160 (3d Cir. 1991); United States ex rel. Dick v. Long Island Lighting Co., 912 F.2d 13, 16 (2d Cir. 1990); Houck v. Folding Carton Admin. Committee, 881 F.2d 494, 505 (7th Cir. 1989), cert. denied, 494 U.S. 1026, 110 S.Ct. 1471, 108 L.Ed.2d 609 (1990). "Direct" signifies "marked by absence of an intervening agency." Stinson, 944 F.2d at 1160. "Independent knowledge" is knowledge that is not itself dependent on public disclosure. See Houck, 881 F.2d at 505.

Significantly, the "original source" provision requires the relator to possess direct and independent knowledge of the "information" underlying the allegation, rather than direct and independent knowledge of the "transaction" itself. On the basis of plain meaning, then, we find that § 3730(e)(4)(B) does not require that the qui tam relator possess direct and independent knowledge of all of the vital ingredients to a fraudulent transaction. Thus, to revert to previous analysis, the term "information" in the "original source" context is not confined to the combination of X and Y, the two essential elements to the transaction. This distinction also conforms to our understanding of the purposes of the Act. Rare indeed would be the case in which relators could gain "original source" status, if such were the standard, because the misrepresented state of affairs, e.g. X, would almost always have been disclosed to the government independently by the alleged defrauder. If we were to equate "information" with "transaction," we would undo both Congress' careful choice of wording and its manifest intent to reverse the result in United States ex rel. Wisconsin v. Dean, 729 F.2d 1100 (7th Cir. 1984), a case in which the misrepresented state of affairs (paperwork seeking reimbursement for indigent psychiatric care) had already been presented to the government by an alleged perpetrator of Medicaid fraud. We think it clear, in light of the aims of the statute, that "direct and independent knowledge of information on which the allegations are based" refers to direct and independent knowledge of any essential element of the underlying fraud transaction ( e.g., Y).

Applying these principles, it appears beyond question that Springfield qualifies as an original source for the allegations of fraud it made in the context of the Maine litigation. Springfield had direct and independent knowledge of essential information underlying the conclusion that fraud had been committed. Because, as stated above, the pay vouchers and phone records did not themselves suffice to indicate fraud, Springfield had to have bridged the gap by its own efforts and experience, which in this case included personal knowledge of the arbitration proceedings and interviews with individuals and businesses identified in the telephone records. Springfield started with innocuous public information; it completed the equation with information independent of any preexisting public disclosure. As such, Springfield is an original source within the meaning of the jurisdictional bar, and federal courts are not statutorily precluded from entertaining its suit.

III. CONCLUSION

The district court relied upon an excessively broad reading of "based upon the public disclosure of allegations or transactions" in dismissing Springfield's qui tam action. The remedial purposes of the FCA are inadequately served by an expansive interpretation of the jurisdictional bar that prevents qui tam suits when only innocuous or spotty information — insufficient in itself to constitute an allegation of fraud or to reveal the essential elements of a fraudulent transaction — exists in the public domain. The goal of avoiding suits that merely drain the public fisc is amply advanced by a construction of § 3730(e)(4)(A) that bars suit only when specific allegations of fraud or the vital ingredients to a fraudulent transaction exist in the public eye.

Concerns about abuses stemming from this broader reading are partially allayed by the fact that the FCA provides for the defendant's recovery of all reasonable attorneys' fees if the qui tam plaintiff's claim is "clearly frivolous, clearly vexatious, or brought primarily for purposes of harassment." 31 U.S.C. § 3730(d)(4). Moreover, as in any suit, if the qui tam plaintiff's action does not present valid legal issues, the courts may dismiss on conventional grounds. But the courthouse doors do not swing shut merely because innocuous information necessary though not sufficient to plaintiff's suit has already been made public. We vacate the district court's dismissal and remand with instructions to proceed with the suit.

Vacated and remanded.