POOLE, Circuit Judge:

Chen-Cheng Wang, a mechanical engineer, brought suit against his former employer, the FMC Corporation, under the qui tam provisions of the False Claims Act. 31 U.S.C. § 3729 et seq. (1986). Wang claimed that FMC defrauded the government by its performance on various defense contracts, including one for work on a cousin of the Bradley Fighting Vehicle. The district court granted FMC's motion for summary judgment and dismissed Wang's second amended complaint. For lack of pendent jurisdiction, the district court also dismissed Wang's four state law claims without prejudice. Wang's case raises a novel jurisdictional issue: to bring a qui tam suit based on allegations already in the public domain, must a plaintiff have played a part in disclosing those allegations? We answer yes, and affirm.

BACKGROUND

Wang was fired from his job at FMC on December 11, 1986. He filed this action a year later, on December 10, 1987. In addition to his False Claims Act claim, Wang joined a number of state law claims, including a wrongful termination claim. As required by the False Claims Act (the Act), Wang filed his complaint under seal and presented his claim to the United States government. See 31 U.S.C. § 3730(b)(2). After reviewing the evidence presented to it by Wang, the government filed a Notice of Declination of Appearance dated August 30, 1988. Wang proceeded with the action on his own. See 31 U.S.C. § 3730(c)(3). The district court issued an order removing the seal from the file on November 2, 1988 and permitting service of the complaint on FMC. FMC was served with the complaint on November 8, 1988.

After partial victories in several motions to dismiss Wang's suit, FMC filed an answer to Wang's second amended complaint. The parties thereafter engaged in a significant amount of discovery. FMC produced thousands of documents. On August 13, 1990, FMC filed a motion for summary judgment, seeking to dismiss all of Wang's remaining claims. At the district court's request, FMC filed a supplemental brief on January 2, 1991 addressing whether Wang was an "original source" of the evidence supporting his claim under the Act. On April 23, 1991 the district court granted FMC's motion for summary judgment and dismissed Wang's four remaining state law claims without prejudice, for lack of subject-matter jurisdiction. Wang timely appeals.

ANALYSIS (I)

The False Claims Act provides penalties for one who "knowingly presents . . . a false or fraudulent claim" to the government, 31 U.S.C. § 3729(a), and incentives to whistleblowers who expose the fraud. 31 U.S.C. § 3730. Before proceeding with the suit, a qui tam plaintiff must disclose his evidence of fraud to the government, which then has sixty days to intervene in the suit. 31 U.S.C. § 3730(b). If the government chooses not to intervene, the qui tam plaintiff may proceed with the suit, as the government's assignee, unless the action triggers one of the jurisdictional bars laid out in section 3730(e) of the Act.

The jurisdictional bar at issue in Wang's case is section 3730(e)(4), which provides:

(A) 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.

(B) For purposes of this paragraph, `original source' means an individual who has direct and independent knowledge of the information on which the allegations are based and has voluntarily provided the information to the Government before filing an action under this section which is based on the information.

(1986) (emphasis added).

Section 3730(e) circumscribes the power of courts to hear qui tam suits. Federal courts have no power to consider claims for which they lack subject-matter jurisdiction. Bender v. Williamsport Area School Dist., 475 U.S. 534, 541, 106 S.Ct. 1326, 1331, 89 L.Ed.2d 501 (1986). It follows that "[b]efore we may reach the merits, we must first consider whether the district court had subject matter jurisdiction over" the claims. Price v. United States General Services Admin., 894 F.2d 323, 324 (9th Cir. 1990). We must examine whether any of Wang's claims are blocked by the jurisdictional bar of section 3730(e)(4) before we can consider any other question.

The "Public Disclosure" Provision.

Wang alleges that FMC defrauded the government in four separate projects. A review of the record makes clear that neither the allegations nor the evidence concerning three of those projects has been publicly disclosed: (1) the Integrated Technology Tactical Vehicle Horsepower Estimation Project ("ITTV"); (2) the Submarine Weapon Handling System ("SWHS"); (3) the Lightweight Towed Howitzer Demonstrator ("LTHD"). Neither the district court nor the parties mentions this fact, and they seem not to understand its implications. Whether or not Wang was the "original source" of the evidence concerning these three projects, the jurisdictional bar of section 3730(e)(4) cannot block Wang's prosecution of them. Where there has been no "public disclosure" within the meaning of section 3730(e)(4), there is no need for a qui tam plaintiff to show that he is the "original source" of the information. United States ex rel. Hagood v. Sonoma County Water Agency, 929 F.2d 1416, 1419-20 (9th Cir. 1991); see also United States ex rel. Williams v. NEC Corp., 931 F.2d 1493, 1500 (11th Cir. 1991). A qui tam plaintiff need prove his status as an "original source" under section 3730(e)(4)(B) "only if an exception is sought to the bar of 4(A)." Hagood, 929 F.2d at 1420.

For his allegation of fraud in the ITTV, Wang relies on his report, begun and completed after just one week of work, criticizing the work of a fellow FMC engineer. There is nothing "public" about this squabble between two FMC engineers. The same might be said for Wang's evidence of fraud in the SWHS project. Wang spent less than two weeks on the project. Wang's work appears to be a computer analysis of minor importance. Based on his work, Wang alleges that the project was "fatally defective due to the lack of engineering insight into the mathematical modeling requirements." So far as the record shows, neither Wang's work nor his allegation of such a defect has been publicly disclosed within the meaning of section 3730(e)(4)(A). Whatever its merit, Wang's SWHS claim is not barred by section 3730(e)(4).

The final project for which Wang relies on allegations and information not publicly disclosed is the LTHD project. Wang, who worked on the project for about two weeks, was assigned a problem that the senior project engineer said was "fairly easy to solve": Wang analyzed how much the lightweight howitzer would skid when fired. Wang had no other involvement with the project. However, the evidence supporting Wang's allegation of fraud is not drawn from the work he did on the LTHD project. Instead, he trolled the massive number of documents made available to him during the discovery phase of his qui tam suit and culled one that he thinks shows evidence of fraud: an interoffice memorandum dated April 6, 1987. The memorandum, marked "company private", discusses the lessons learned after the LTHD project was cancelled by the Army. The memorandum focuses on problems with "composite" technology, which Wang had nothing to do with. The memorandum makes clear that it summarizes a meeting called to address the "specific concerns raised by ARDEC (i.e., the Army's Armament Research and Development Center)" after the project had been cancelled. It appears that all of the issues discussed in the memorandum were first raised and considered in meetings with the Army.

The memorandum Wang relies upon to prove his allegation of fraud has only made its way to the public because of the discovery permitted in Wang's qui tam suit. Evidence publicly disclosed for the first time during the discovery phase of a qui tam suit is not barred from use in that same suit by section 3730(e)(4)(A). If it were, qui tam plaintiffs would have little choice but to waive their right to discovery for fear of disclosing information that would bar the claims for which they might wish discovery in the first place. Cf. United States ex rel. LeBlanc v. Raytheon Co., 913 F.2d 17, 20 (1st Cir. 1990) (filing of a qui tam action is not itself a public disclosure; "to hold otherwise would be to . . . bar all qui tam actions"), cert. denied, LeBlanc v. United States, ___ U.S. ___, 111 S.Ct. 1312, 113 L.Ed.2d 246 (1991). But see United States ex rel. Stinson, Lyons, Gerlin Bustamante, P.A. v. Prudential Insurance Co., 944 F.2d 1149, 1157-60 (3rd Cir. 1991) (information revealed through discovery in one civil litigation was "publicly disclosed" for purposes of a subsequent qui tam action).

The "Original Source" Provision.

In 1983, a member of the Bradley family of vehicles, the Multiple Launch Rocket System ("MLRS"), was experiencing gear failures: the teeth of one gear in the powertrain had broken off in a number of MLRS vehicles. Wang, who was trained in gear technology, was part of the team of FMC engineers called in to study the problem. Wang spent less than two weeks on the project. He prepared a four page report, which focused on the "backlash" that occurred when stress was put upon the transmission's gears. In his report, Wang made four recommendations for further research and analysis. Wang does not know what action, if any, FMC took on his recommendations. In the end, FMC decided to remedy the gear failures by widening the teeth of the troublesome gear. Since that change was implemented in 1984, the gear failures have ceased. Wang had nothing to do with this solution, and appears not to have known how it came about.

The district court ruled that Wang's suit was blocked by the jurisdictional bar of section 3730(e)(4)(B) because Wang was not an "original source" of the information on which his allegation was based. Because one need only be an "original source" of one's information if one's allegation has been publicly disclosed, 31 U.S.C. § 3730(e)(4)(A), the necessary premise of the court's ruling is that Wang's allegation of fraud in the Bradley had been publicly disclosed before Wang brought his suit. Wang has never disputed, and his arguments appear to accept, that his allegation had been publicly disclosed. As appendices to his declaration in opposition to summary judgment, Wang submitted newspaper accounts describing problems with the Bradley's transmission system, apparently published before the date of Wang's complaint. It is true that Wang's allegation about the Bradley is supported by a few factual assertions never before publicly disclosed; but "fairly characterized" the allegation repeats what the public already knows: that serious problems existed with the Bradley's transmission. United States ex rel. Dick v. Long Island Lighting Co., 912 F.2d 13, 14 (2nd Cir. 1990). The district court characterized Wang's allegation and most of his information as a rehash of what already had been publicly disclosed. Wang does not dispute this characterization, and it finds support in the record. See id., 912 F.2d at 16 (accepting that suit is based on allegations already made public because "there is no dispute that appellants' suit is based upon publicly disclosed `allegations or transaction'").

Where an allegation has been publicly disclosed within the meaning of section 3730(e)(4)(A), a plaintiff may only bring a qui tam suit if he can show that (1) he has "direct and independent knowledge of the information on which his allegation is based"; and (2) that he "has voluntarily provided the information to the Government before filing" his qui tam action. 31 U.S.C. § 3730(e)(4)(B). The district court, purporting to follow Houck on Behalf of United States v. Folding Carton Administration Committee, 881 F.2d 494, 505 (7th Cir. 1989), held that Wang's knowledge was not "direct and independent." This was error. Houck stands for the simple proposition that where one would not have learned of the information but for its public disclosure, one does not have "direct and independent knowledge" of the information. See also Stinson, 944 F.2d at 1160. Wang had personal knowledge of the Bradley's transmission problems because he worked (however briefly) on trying to fix them. The fact that someone else publicly disclosed the Bradley's transmission problems does not rob Wang of what he saw with his own eyes. Wang's knowledge of the transmission problems was "direct and independent" because it was unmedicated by anything but Wang's own labor.

Wang had direct and independent knowledge in May, 1983 of what he thought was a fraud on the government. He waited until December, 1987, a year after he was fired by FMC, to bring the alleged fraud to the government's attention. In the interim, someone else publicly disclosed the Bradley's transmission problems. Wang is now revealing what is already publicly known. The difficult question is whether, as a statutory and not only a practical matter, Wang's effort comes too late, because section 3730(e)(4)(A) requires a qui tam plaintiff to have played some part in his allegation's original public disclosure. The Second Circuit has held that, under section (4)(A), a qui tam plaintiff "must have directly or indirectly been a source to the entity that publicly disclosed the allegations on which a suit is based." Dick, 912 F.2d at 16. We agree with the Second Circuit. To bring a qui tam suit, one must have had a hand in the public disclosure of allegations that are a part of one's suit.

Courts sometimes speak loosely of barring a qui tam suit because it is based on "publicly disclosed information." See e.g. Dick, 912 F.2d at 17 ("one must have been a source to the entity that first publicly disclosed the information on which a suit is based"); Houck, 881 F.2d at 504 ("[t]he information upon which Houck based his complaint was publicly disclosed"); Hagood, 929 F.2d at 1419 ( qui tam suit not barred because based upon "information [that] was not publicly disclosed"). But the Act bars suits based on publicly disclosed "allegations or transactions," not information. 8 U.S.C. § 3730(e)(4)(A). The point is not mere semantics: the Act distinguishes between "allegations" and the "information on which the allegations are based." 8 U.S.C. § 3730(e)(4)(B). The Act appears to be invoking the common logical distinction between an assertion and its proof. Although not empty, this distinction rarely matters in applying the Act, because where the public knows of information proving an allegation, it necessarily knows of the allegation itself. But the reverse is not always true. An allegation can be made public, even if its proof remains hidden. No doubt when courts speak of "publicly disclosed information," they mean both the allegation of fraud and all information proving the allegation that has made its way to the public. In Wang's case, as in the typical qui tam suit, the distinction between an allegation and its proof makes little difference: it is undisputed that both Wang's allegation and most of its supporting information have been publicly disclosed.

The Second Circuit contends that its construction of section (4)(A), requiring a qui tam plaintiff to play some part in the original public disclosure of the allegations made in his suit, is "the most natural reading" of the Act as amended in 1986. Dick, 912 F.2d at 17. Perhaps it is. But undoubtedly the text remains ambiguous. Where a statutory term is ambiguous, courts must "construe it to contain that permissible meaning which fits most logically and comfortably into the body of both previously and subsequently enacted law." West Virginia University Hospitals, Inc. v. Casey, ___ U.S. ___, 111 S.Ct. 1138, 1148, 113 L.Ed.2d 68 (1991). The history of the False Claims Act and the legislative history of its most recent amendment make clear that qui tam jurisdiction was meant to extend only to those who had played a part in publicly disclosing the allegations and information on which their suits were based.

The history of the False Claims Act has been considered in detail elsewhere. See Hagood, 929 F.2d at 1420; Stinson, 944 F.2d at 1152-54, 1162-68. For our purposes, that history can be presented summarily. In United States ex rel. Marcus v. Hess, 317 U.S. 537, 63 S.Ct. 379, 87 L.Ed. 443 (1943), the Supreme Court held that the Act as then written did not require a relator to bring any original information to his qui tam suit and could rely wholly on allegations and information produced by the government itself. In response to Hess and the outcry it generated, Congress passed a bill restricting qui tam suits. As the Senate Report for the 1986 amendments to the False Claims Act says, the Senate version of the 1943 amendments "specifically provided that jurisdiction would be barred on qui tam suits based on information in the possession of the Government unless the relator was the original source of that information." S.Rep. 345, 99th Cong., 2d Sess. (1986) at 12, reprinted in 1986 U.S.C.C.A.N. 5266, 5277 ("Senate Report"). Inexplicably, says the Senate Report, the clause requiring that the relator be the original source of the Government's information was dropped just before those amendments became law. Id.

Congress' effort in 1943 to narrow Hess' expansive reading of the qui tam provisions led to severely restrictive interpretations of those same provisions in later cases. Courts read the amended Act as prohibiting all qui tam suits where the government already possessed the information, even where the relator had independently uncovered fraud against the government and the government knew of that fraud only because the relator had been decent enough to tell the government about it. See e.g. Pettis ex rel. United States v. Morrison-Knudsen Co., 577 F.2d 668 (9th Cir. 1978). In a decision that stirred controversy, the Seventh Circuit held that even though the state of Wisconsin had investigated and uncovered Medicaid fraud, it could not maintain a qui tam suit because the government already had learned of the information when Wisconsin reported its discovery as required by an unrelated regulation. United States ex rel. State of Wisconsin v. Dean, 729 F.2d 1100, 1104 (7th Cir. 1984). Having told its tale of fraud to the government, as the regulation required, the relator was barred from suing under the Act.

This restrictive interpretation was too much for Congress. In part, the 1986 amendments to the Act were "aimed at correcting restrictive [court] interpretations of the act's . . . qui tam jurisdiction" provisions. Senate Report, at 4, 1986 U.S.C.C.A.N. 5269. The 1986 amendments to qui tam jurisdiction were remedial, not innovative. Congress wanted in 1986 what it apparently thought it had in 1943: a law requiring that the relator be the original source of the government's information. Seeking only to "correct" opinions like Dean, Congress permitted one who publicly disclosed the information to bring a qui tam suit. There is nothing to suggest that Congress meant to do any more than that, and some evidence that it meant to do less. See § 3730(e)(2)(A) ( qui tam suits against political officials still barred if involving information already possessed by the government).

This interpretation of Congress' intent accords with the Act's purpose of encouraging "private individuals who are aware of fraud being perpetrated against the Government to bring such information forward", H.R.Rep. No. 660, 99th Cong., 2d Sess. 22 (1986), because the interpretation:

is most likely to bring `wrongdoing to light' since, by barring those who come forward only after public disclosure of possible False Claims Act violations from acting as qui tam plaintiffs, it discourages persons with relevant information from remaining silent and encourages them to report such information at the earliest possible time.

Dick, 912 F.2d at 18 (emphasis in original). See also Senate Report at 2 1986 U.S.C.C.A.N. 5266 (the 1986 amendments are meant "to encourage any individual knowing of Government fraud to bring that information forward"). The paradigm qui tam plaintiff is the "whistleblowing insider." Stinson, 944 F.2d at 1160. Qui tam suits are meant to encourage insiders privy to a fraud on the government to blow the whistle on the crime. In such a scheme, there is little point in rewarding a second toot.

It is important to note that under the rule we adopt today, all those who "directly or indirectly" disclose an allegation might qualify as its original source. Dick, 912 F.2d at 18. Anyone who helped to report the allegation to either the government or the media would have "indirectly" helped to publicly disclose it. If, however, someone republishes an allegation that already has been publicly disclosed, he cannot bring a qui tam suit, even if he had "direct and independent knowledge" of the fraud. He is no "whistleblower." A "whistleblower" sounds the alarm; he does not echo it. The Act rewards those brave enough to speak in the face of a "conspiracy of silence," and not their mimics. Senate Report, at 6, 1986 U.S.C.C.A.N. 5271.

Wang knew of the Bradley's transmission troubles. He sat quietly in the shadows and breathed not a word about them until he was fired. While Wang was silent, some other conscientious or enterprising person bravely brought the transmission problems to the attention of the media and the Army. If there is to be a bounty for disclosing those troubles, it should go to one who in fact helped to bring them to light. Because he had no hand in the original public disclosure of the Bradley's troubles, Wang's claim regarding the MLRS is blocked by the jurisdictional bar of section 3730(e)(4)(A).

(II)

The jurisdictional issue in this case is complicated. The merits are not. To survive summary judgment, Wang must establish " evidence on which a reasonable jury could find for the plaintiff." Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 252, 106 S.Ct. 2505, 2512, 91 L.Ed.2d 202 (1986) (emphasis added). Viewing, as we must, the evidence in the light most favorable to him, Wang nonetheless has failed "to make a showing sufficient to establish the existence of an element essential to [his] case, and on which [he] will bear the burden of proof at trial." Celotex Corp. v. Catrett, 477 U.S. 317, 322, 106 S.Ct. 2548, 2552, 91 L.Ed.2d 265 (1986). Having searched through thousands of documents and conducted extensive discovery, Wang has failed to produce sufficient evidence to support an inference of fraud by FMC. All reasonable inferences defeat Wang's claims. White v. Roper, 901 F.2d 1501, 1505 (9th Cir. 1990).

A qui tam plaintiff may file a civil action alleging that another has "knowingly" presented or caused to be presented "a false or fraudulent claim", 31 U.S.C. § 3729(a)(1), or has "knowingly" made, used, or caused a false record or statement to be made to get a false or fraudulent claim paid, id. (a)(2), or has "knowingly" made, used, or caused a false record or statement to be made to "decrease an obligation" to pay the government. Id. at (a)(7). See Hagood, 929 F.2d at 1421.

The Act's scienter requirement is laid out in section 3729(b). One violates the Act if one has "actual knowledge" that one is submitting a false or fraudulent claim for payment or approval, "acts in deliberate ignorance of the truth or falsity" of one's false claim, or "acts in reckless disregard of the truth or falsity" of one's false claim. As held in Hagood:

Innocent mistake is a defense to the criminal charge or civil complaint. So is mere negligence. The statutory definition of `knowingly' requires at least `deliberate ignorance' or `reckless disregard' . . . [W]hat constitutes the offense is not intent to deceive but knowing presentation of a claim that is either `fraudulent' or simply `false'. The requisite intent is the knowing presentation of what is known to be false.

929 F.2d at 1421 (citation omitted). The Act's scienter requirement is something less than that set out in the common law, where "the words `to defraud' commonly refer `to wronging one in his property rights by dishonest methods or schemes', and `usually signify the deprivation of something of value by trick, deceit, chicane or overreaching.'" McNally v. United States, 483 U.S. 350, 358, 107 S.Ct. 2875, 2880, 97 L.Ed.2d 292 (1987) (quoting Hammerschmidt v. United States, 265 U.S. 182, 188, 44 S.Ct. 511, 512, 68 L.Ed. 968 (1924)). As at common law, however, "innocent mistakes" and "negligence" are not offenses under the Act.

For each of his surviving claims, Wang has no evidence that FMC committed anything more than "innocent mistakes" or "negligence," if that. Virtually all of Wang's evidence consists of his own affidavit, which assesses and condemns FMC's performance on various defense projects.

The ITTV Project. Wang's entire claim is based on an allegedly faulty calculation by another FMC engineer. But Wang's criticism of the engineer's calculations, even if accurate, proves no more than an "innocent mistake." Wang only says that the miscalculation reflected the engineer's "very low level of understanding." Bad math is no fraud.

The SWHS Project. The evidence supporting Wang's allegations about this project is not clearly laid out. He only says, however, that FMC suffered a "lack of engineering insight." Again, this might be proof of a "mistake" or even of "negligence" in performing the work. But there is no evidence that FMC showed "deliberate ignorance" of false claims for payment based upon that work. Proof of one's mistakes or inabilities is not evidence that one is a cheat.

The LTHD Project. Wang says no more than that FMC's engineering work was of "low quality", and that the design for the lightweight howitzer was "faulty." He bases his claims on FMC's own self-critical assessment of its work after the project was ended by the Army. The memorandum relied on by Wang was part of a dialogue with the Army. The government knew of all the deficiencies identified by Wang, and discussed them with FMC. The fact that the government knew of FMC's mistakes and limitations, and that FMC was open with the government about them, suggests that while FMC might have been groping for solutions, it was not cheating the government in the effort. Without more, the common failings of engineers and other scientists are not culpable under the Act.

Wang's case betrays a serious misunderstanding of the Act's purpose. The weakest account of the Act's "requisite intent" is the "knowing presentation of what is known to be false." Hagood, 929 F.2d at 1421. The phrase "known to be false" in that sentence does not mean "scientifically untrue"; it means "a lie." The Act is concerned with ferreting out "wrongdoing," not scientific errors. Dick, 912 F.2d at 18. What is false as a matter of science is not, by that very fact, wrong as a matter of morals. The Act would not put either Ptolemy or Copernicus on trial.

Because Wang has not produced evidence that FMC acted with the intent requisite for liability under the Act, his surviving claims were properly dismissed. The district court judgment granting summary judgment for FMC and dismissing Wang's state law claims for lack of subject matter jurisdiction is AFFIRMED.