This case requires us to explore the kind of showing a patent licensee needs to make in order to recover royalties already paid under a license agreement when the patent itself is subsequently declared invalid. The main appeal is brought by Transitron Electronic Corporation from a judgment rendered by the district court in favor of Hughes Aircraft Company, the defendant in Transitron's action for restitution and damages arising out of a patent licensing agreement.

On March 31, 1950, Hughes filed a patent application for a "glass-sealed semi-conductor crystal device," which had been invented by Harper Q. North and Justice N. Carman, Jr. The invention is described in this court's opinion in General Instrument Corp. v. Hughes Aircraft Co., 399 F.2d 373, 374 (1st Cir. 1968), as well as in the opinion of the district court, 487 F. Supp. 885, 889 (D.Mass.1980). The patent, No. 2,694,168, was issued on November 9, 1954. On different occasions in the period 1950 through 1954, Hughes amended its application by adding claims purporting to describe in greater detail the invention itself and the process of producing it. The present case involves two of these amendments, filed on July 27, 1954 as claims 102 and 103, which the patent office allowed on September 2, 1954 as claims 60 and 61.

On March 20, 1951, a year after it had first filed for the United States patent, Hughes applied for a British patent on the same invention. The British patent was issued on November 24, 1954, as No. 721,201. The same material embodied in claims 60 and 61 of the United States patent, together with a three-page preamble, was filed in Great Britain, on November 8, 1954, not as a claim under the parent patent, but as a separate improvement in an application for a British Patent of Addition.

Beginning in 1954, Hughes investigated possible infringement of its patent rights by several manufacturers, including Transitron. Notices were sent out charging infringement and offering non-exclusive licenses. On June 1, 1962, Hughes and Transitron executed a five-year license agreement, giving Transitron the right to manufacture diodes covered by claims 60 and 61 of Hughes' patent in exchange for a one percent royalty and for licenses under certain patents held by Transitron. Transitron manufactured under this license until October 1, 1962, at which time it notified Hughes that it had switched to a manufacturing process outside the Hughes patent. In 1966, Hughes discovered that Transitron was once again manufacturing diodes covered by the license. After Hughes demanded back royalties, the parties agreed to a lump sum settlement and renewal of the license agreement. Transitron operated under the license for several months and then again announced that it had switched to another method.

In 1967, Hughes sued General Instrument Corporation for infringement of claims 60 and 61. Hughes Aircraft Co. v. General Instrument Corp., 275 F. Supp. 961 (D.R.I.1967). General Instrument responded by challenging the validity of these claims. Hughes prevailed in the district court; on appeal, however, this court reversed, holding that claim 60 was new matter, not supported by the original patent application, in violation of 35 U.S.C. § 132.1 General Instrument Corp. v. Hughes Aircraft Co., 399 F.2d 373 (1st Cir. 1968).2 Noting that commercial sales of the diode embodying the new matter had commenced over a year prior to filing, we declared claim 60 to be invalid, 399 F.2d at 377, 380. In reaching that conclusion, this court emphasized the preamble that was filed as part of Hughes' application for the British Patent of Addition, but that had not been provided to the United States patent office.3 Id., at 376.

Transitron brought this action in 1970. Count I was in antitrust, resting on the theory that Hughes had used the patent (in particular claims 60 and 61), which it was said to have procured through fraud on the patent office and in violation of the Invention Secrecy Act, 35 U.S.C. §§ 181-188, to coerce Transitron and others to enter into licensing agreements. Treble damages were sought. Count II, relying on much the same factual allegations, sought the return of all royalties for patent misuse. Count III likewise sought the return of royalties, but on the theory that Hughes' false and misleading representations to the patent office as reflected in the file wrapper were material misrepresentations which induced Transitron to take a license. Alternatively, return of royalties was sought on the theory that the parties' belief that claims 60 and 61 were valid constituted a mutual mistake of fact.

Hughes counterclaimed for certain back royalties, asserting that the 1966 settlement was invalid because Transitron had misrepresented to Hughes the extent of its use of Hughes' patent between 1962 and 1966.

After an eight-day trial, the district court granted judgment for Hughes on all of Transitron's claims and for Transitron on Hughes' counterclaim. Both parties have appealed, although Hughes indicates that it chooses not to press its cross-appeal should we now hold against Transitron on Transitron's claim. We affirm the judgment of the district court in both cases.

On appeal, Transitron presses exclusively its tort and contract theories for recovery of back royalties.4 These have been somewhat amplified from their presentation in the district court, and appear to be, in essence, as follows: (1) In causing Transitron to accept a license and pay royalties, Hughes benefited from conduct that was, at least, negligently deceptive and, at worst, deliberately fraudulent, and that, in either event, such conduct amounted to a tort entitling Transitron to the return of its royalties and to consequential damages; (2) Whether or not Hughes' conduct amounted to fraud or another tort, Transitron is entitled to rescind the licensing agreement and have its royalties returned as the agreement was invalid for lack of consideration. Under each of its theories Transitron asserts a right to recover all royalties paid to Hughes under the license. Since the claim of failure of consideration is just another way of saying that the patent is invalid, we do not treat this theory separately. We begin by considering the law on recovery of royalties by a patent licensee.

II. The Legal Standard for Restitution of Royalties

In Lear, Inc. v. Adkins, 395 U.S. 653, 89 S. Ct. 1902, 23 L. Ed. 2d 610 (1969), the Supreme Court held that a patent licensee would no longer be estopped from challenging the validity of the patent under which he was licensed and, if successful, from rescinding the licensing agreement. Circuit courts since Lear are agreed, however, that while such an adjudication of the patent's invalidity will permit rescission of the agreement, it will not, without more, allow the licensee to recover royalties paid prior to bringing suit. St. Regis Paper Co. v. Royal Industries, 552 F.2d 309, 313 (9th Cir.), cert. denied 434 U.S. 996, 98 S. Ct. 633, 54 L. Ed. 2d 490 (1977); Zenith Laboratories v. Carter-Wallace, 530 F.2d 508 (3d Cir.), cert. denied, 429 U.S. 828, 97 S. Ct. 85, 50 L. Ed. 2d 91 (1976); Troxel Manufacturing Co. v. Schwinn Bicycle Co., 465 F.2d 1253 (6th Cir. 1972). See also Nashua Corp. v. RCA Corp., 431 F.2d 220 (1st Cir. 1970); Drackett Chemical Co. v. Chamberlain Co., 63 F.2d 853, 854-55 (6th Cir. 1933). In Troxel the Sixth Circuit discussed the reasons for this rule, emphasizing the equities between licensor and licensee, and the policies favoring early adjudication of patent validity and encouragement of the use of the patent system by protecting the security of royalty income. The Troxel court pointed out how undesirable it would be if a patentee took all royalty payments subject to the danger of having to disgorge them at an uncertain future date to a licensee who, during the period in which the royalties were paid, had himself enjoyed the protection of the patent and immunity from suit under it. Were the law thus, the value of patents might be so diminished as to discourage inventors from procuring them, with the result that inventions would be kept secret rather than publicly disclosed as the patent system requires. We fully agree with the reasoning in Troxel and accept as a general rule that a licensee may not recover royalties he has already paid for use of a patent which is subsequently declared invalid.

Troxel and the above cases mention one exception to this rule. The exception occurs when a patent holder has induced the licensee to enter into the license agreement through fraud; in such event, the licensee is entitled to restitution of his royalty payments. While we have found only one case in which this exception was actually applied, Ampex v. Memorex Corp., 205 U.S.P.Q. 794 (N.D.Cal.1980),5 such an exception plainly exists. In addition to cases cited in the previous paragraph, see, e. g., In re Multidistrict Litigation Involving Frost Patent, 398 F. Supp. 1353 (D.Del.1975), aff'd in part, rev'd in part, 540 F.2d 601 (3d Cir. 1976); SCM Corp. v. RCA, 318 F. Supp. 433, 470-72 (S.D.N.Y.1970). The fraud exception does no violence to the policies described in Troxel. It does not unduly burden a patent holder's reasonable expectations since fraud, if ever alleged, would have to be proven by clear and convincing evidence, Norton v. Carborundum, 530 F.2d 435, 444 (1st Cir. 1976), and the patent holder would have to be proven not merely careless but to have intended to deceive. An honest patentee, therefore, will have little to fear, while a dishonest patentee can scarcely complain if later stripped of ill-gotten gains.

Transitron asks us to broaden the fraud exception in two respects: First, it urges us to recognize a right in a licensee to recover royalties from a licensor following invalidation of a patent solely upon a showing that the licensor committed fraud on the patent office in procuring his patent, without any necessary showing of a fraudulent misrepresentation in the licensing transaction itself. Second, Transitron urges us to accept as fraud, for this purpose, a category of conduct involving something less than knowing and deliberate misrepresentation, such as gross negligence.6 Transitron cites no direct precedent for these proposed rules; it relies on state laws authorizing rescission of contracts7 and on the federal patent law policies requiring candor in dealings with the patent office.

As already indicated, the Troxel rule that we adopt rests on patent law policies and on the equities between licensor and licensee. Any exception must fit within those policies and equities. Transitron argues that the equities between licensor and licensee are shifted whenever a patent holder has committed misconduct against the patent office, and that to require a refund of royalties for such misconduct would punish a wrongdoer, not force the patent holder into the role of guarantor of the debatable legal conclusion of patent validity. But if fraud on the patent office, without more, were ground for recovery of royalties, a licensee who knew, suspected, or later found that a patent was fraudulently procured would be in the same position as one who suspected invalidity on other grounds; he would have an incentive to delay challenging the patent, enjoying the competitive advantage of the license and avoiding the necessity of defending an infringement suit, secure in the knowledge that he could recoup his royalty costs later. The relevant question, we think, is whether there was fraud perpetrated on the licensee at the time he entered into the license. It is when the deception infects the licensing transaction itself, so that one who might otherwise have chosen to risk suit and challenge the patent is induced to take a license, that the policy of encouraging early challenge is unaffected by allowing restitution. We hold that a licensee may recover royalties paid only upon showing that he was induced to accept a license through fraud on the part of the licensor. Proof of fraud on the patent office may, to be sure, be highly material to such a showing; if one defrauds the patent office, the same fraudulent misinformation may carry over to the licensing negotiations. A licensor may expect a potential licensee to rely on information contained in patent office records relative to the application and patent. Still, the ultimate issue is not patent office fraud but fraud on the licensee. Zenith Laboratories v. Carter-Wallace, 530 F.2d at 514.

We further believe that a definition of fraud not requiring proof of actual bad faith would be inconsistent with the policies stated in Troxel. Transitron argues that the equities of the situation, as well as the need for candor in patent disclosures, favor a broad definition of fraud. In recent years, courts in patent matters have characterized something less than knowing and wilful misrepresentation as constituting fraud for some purposes. See W. R. Grace & Co. v. Western U.S. Industries, 608 F.2d 1214 (9th Cir. 1979), cert. denied, 446 U.S. 953, 100 S. Ct. 2920, 64 L. Ed. 2d 810 (1980) (denying enforcement of patent because of patentee's "calculated recklessness about the truth"); Norton v. Curtiss, 433 F.2d 779 (C.C.P.A.1970) (accepting "gross negligence" as basis for striking patent application); Monolith Portland Midwest Co. v. Kaiser Aluminum & Chemical Co., 407 F.2d 288 (9th Cir. 1969) (awarding attorney's fees against a patent holder because of its "calculated recklessness about the truth"). In none of these cases, however, was the question whether back royalties could be recovered.8 We do not disagree that patent holders should be encouraged to be candid and careful in their dealings with licensees as well as with the patent office. But, in practice, it would be too easy, with the benefit of hindsight, to show that certain information should have been disclosed, and to characterize the failure as some species of "technical" fraud. It would be difficult for patent holders to take effective measures to guard against claims of this nature, which might be used as leverage to impose expensive settlements where the alternative would be to risk a ruinous liability for royalties collected over a period of many years. Transitron's proposed rule would blur all too readily with acceptance of a broad principle contrary to Troxel, allowing recovery for mere errors affecting the validity of a patent; the patent holder would then be subject to a constant risk of being required to disgorge royalties, and its creditors would hesitate to rely on royalty income. Such a broadening of the exception for fraud would undermine the security of royalty income and erode the economic value of patent rights, encouraging inventors to avoid public disclosure and resort instead to trade secrets. We hold that a licensee who would recover royalties paid to a licensor must establish nothing less than that he was induced to accept the license as a result of fraud in the normal sense. Mere patent invalidity is not enough, nor is a showing of "technical" fraud on either the patent office or the licensee, nor is a showing based on some theory of restitution or mistake.

Applying this standard, we uphold the district court's determination that Transitron is not entitled to recover back royalties. As stated in Norton v. Carborundum, 530 F.2d at 444, "(a)n allegation of fraud must be proved by clear, unequivocal and convincing evidence and not by a mere preponderance of the evidence which leaves the issue in doubt." As the question of fraud ordinarily is treated as a question of fact, an appellate tribunal must accept the district court's finding unless it is "clearly erroneous." Edward Valves, Inc. v. Cameron Iron Works, Inc., 286 F.2d 933, 947-48 (5th Cir.), modified on other grounds, 289 F.2d 355, cert. denied, 368 U.S. 833, 82 S. Ct. 55, 7 L. Ed. 2d 34 (1961).

In the district court, Transitron sought to prove that Hughes knew at the time of filing claims 60 and 61 that they embodied new matter and that it deliberately withheld that information from the patent office. Transitron sought to show that the material filed as claims 60 and 61 had its source in a February 1952 invention by Harper Q. North, which North submitted to Hughes' patent department on disclosure form "PD 437." According to Transitron, Hughes put North's improved diode into immediate production, and produced it at the rate of 1,500 per day from the end of February 1952, compared with 10,000 to 20,000 per year under the earlier North/Carman process. Transitron asserts that Hughes' patent attorneys decided in May 1952 to apply for a patent on North's invention, but that the attorney in charge "simply forgot" until two years later, when it was too late to file a new application because the process had already been in public use for more than a year.9 Transitron asked the district court to infer that Hughes, upon realizing that it was too late to apply for a patent on the North invention, sought to obtain protection for this new invention by slipping it in as a claim under the 1950 North/Carman application. To support this assertion, Transitron cites Hughes' application for the British Patent of Addition, which expressly treated North's improvement as a development beyond the scope of the parent patent. Transitron contends that Hughes deliberately concealed the British Patent of Addition (BPA) from the United States Patent Office and from licensees in order to avoid the conclusion that the claims contained new matter. Transitron also cites as a misrepresentation Hughes' March 8, 1954 statement, in filing claims 60 and 61, that Hughes had actively sold its diodes for only nine months.

As we have indicated, charges of fraud on the patent office would not, in themselves, suffice to state a claim for restitution. However, Transitron's assertions, if true, might show that Hughes committed fraud on Transitron by representing, contrary to knowledge and belief, that its patent was valid, or by continuing to conceal from Transitron, when the license was negotiated, information material to validity which it had knowingly concealed from the patent office.

The district court, however, rejected Transitron's theory of deliberate fraud on the patent office, finding that "(t)here is no evidence that any of the Hughes patent counsel who worked on the prosecution believed the claims were novel but persisted in this effort to gain patent coverage by deliberately concealing that novelty from the examiner." 487 F. Supp. at 898. The district court relied particularly on the testimony of Harper Q. North and of David Doody, Hughes' Director of Patents since 1957. North testified that he considered claims 60 and 61 as simply "engineering improvements" on the original invention, which was the diode itself. Id. Doody, whom the district court found to be "candid and helpful," id., had not been present during the time of the patent prosecution. Later, Doody had reviewed material submitted by possible infringers asserting that the claims might not be supported by the original application, and he and his staff, as well as outside counsel, concluded that the claims were not new matter. It was Doody who negotiated the settlement with Transitron.

The district court also reviewed the evidence of the decision-making process among Hughes' attorneys regarding PD 437, and found it susceptible of a far more innocent interpretation than Transitron gave it. The court found no clear indication that the patent attorneys had decided how to prosecute PD 437 whether as a claim under the original invention or as a new invention. The court found the two-year hiatus in action on the claims attributable not to oversight, as Transitron asserts, but to Hughes' inability to obtain from Dupont information concerning the composition of its gold thermosetting compound, which North used to make his improved diode. The district court found that Hughes' patent attorney viewed the use of this compound, rather than any dimensional change, as the gist of North's improvement.

The district court also considered and rejected Transitron's urging that it find fraudulent misrepresentation in Hughes' March 1954 statement to the effect that it had been active in selling its diode for only nine months. The court understood the statement as an effort to demonstrate that the diode had achieved "commercial success" by showing that Hughes had sold a large quantity in a short time; the court implicitly rejected Transitron's view that the statement related to patentability under the one-year rule of 35 U.S.C. § 102(b). 487 F. Supp. at 901. The court noted also that Hughes' attorney may have intended this statement to refer back to September 1953, the date of the amendment to which the March 1954 statement was a supplement. If so, the reference to nine months would designate February 1953, a date when Hughes' diode was advertised widely. 487 F. Supp. at 900 n.4.

The district court also rejected Transitron's theory that Hughes' treatment of claims 60 and 61 as a separate improvement rather than as a clarification of the original patent disclosure, in its application for the British Patent of Addition, indicates that Hughes believed that the claims were new matter. The district court noted that Hughes could not have treated the material as claims under the parent patent in Britain, because under British law a patent application must be put in order within one year of filing. Hughes might also have been unable to obtain an independent patent for the material, since the diode was so similar to that disclosed in the main patent. Relying on its examination of correspondence between Hughes' British and American patent counsel, the district court concluded that a Patent of Addition, which protects inventions, was the best way under British law for Hughes to protect this subject matter.10

In sum, the district court concluded that "the Hughes lawyers and scientists had a good faith belief that these developments and modifications were improvements in the existing invention and were not new matter which required disclosure as such to the patent office." 487 F. Supp. at 900.11 The court noted that reasonable lawyers may differ on what constitutes new matter, and that the closeness of that question in this case is demonstrated by the fact that this court differed with the district court in General Instrument, supra, 399 F.2d 373.12 It is not the function of this court to review the evidence de novo, or to make our own judgment of the credibility of witnesses. Merrill Trust Co. v. Bradford, 507 F.2d 467, 468 (1st Cir. 1974); Custom Paper Products Co. v. Atlantic Paper Box, 469 F.2d 178, 179 (1st Cir. 1972). Particularly in light of the heavy burden of proof on the plaintiff to show fraud, we do not regard as clearly erroneous the district court's finding that Hughes did not commit knowing and wilful fraud in obtaining patent protection for claims 60 and 61.

B. Concealment of the British Patent of Addition

The district court did not make a specific finding on whether Hughes committed fraud on Transitron in connection with the licensing transaction, but we think a negative finding on this matter is implicit in the court's other conclusions. Since we accept the court's conclusion that Hughes had no knowledge or belief that its claims were new matter, Transitron's best case for fraud in the licensing transaction comes from Hughes' failure to disclose to Transitron the existence of the British Patent of Addition. This court's opinion in General Instrument demonstrates the persuasive power of the preamble, 399 F.2d at 376-80. Reliance was placed on the preamble and accompanying drawing for clarification of the exact nature and importance of the various structural and dimensional aspects of the claims and of the product and process which had been disclosed in the original filing. The explanation provided in the British filing played no small role in our conclusion that claim 60 was new matter, not supported by the original application. For example, in relation to the "one piece ductile lead wires" referred to in claim 60, the district court had concluded that the language of the original application could be interpreted as referring to one-piece wires. Relying on the BPA's reference to "composite leads * * * of Dumet wire" as a defect in the original diode, this court concluded that the original application did not support the "one piece wires" feature of claim 60. Id., at 377-78.

It is possible, therefore, that the BPA would have influenced Transitron's conclusion about the validity of the claims and its choice between accepting a license and challenging the patent. Thus had Transitron established that Hughes deliberately concealed the BPA, we might consider fraud to be shown. But the district court found no such wilful concealment; rather, it found Hughes' failure to disclose the BPA attributable to defects in Hughes' internal recordkeeping. 487 F. Supp. at 907.13 Hughes asserted that those among its staff who prosecuted the United States application and who handled licensing negotiations were simply unaware of the BPA. This assertion may seem implausible; this court expressed skepticism in General Instrument, 399 F.2d at 381. But the district court accepted it,14 and the record contains no evidence that Hughes' staff knew of the BPA or that they deliberately concealed it. There was, moreover, undisputed testimony from David Wolf, who represented Transitron in its negotiations with Hughes, that Transitron never requested a list of Hughes' foreign patents. Rather Transitron requested either a license under any corresponding foreign patents or a release from liability for infringement of any foreign patents. Hughes chose to grant the latter. Transitron asks us to infer that Hughes made this choice in order to avoid revealing the existence of the BPA. But fraud must be proven by clear and convincing evidence; we cannot say the district court clearly erred in refusing to make this inference.

Transitron also puts forward a theory of fraud based on the Invention Secrecy Act, 35 U.S.C. §§ 181-188. This statute allows certain government officials to screen patent applications before the information disclosed in them is published and to issue secrecy orders when they determine that publication of the information may be detrimental to national security. Under section 184, one who wishes to file a patent application in a foreign country must either wait six months from the date of filing the application in the United States or obtain a license from the Commissioner of Patents. Violation of this requirement is punishable by fine or imprisonment under section 186, as well as, under section 185, by denial of a United States patent on the invention or invalidity of a patent which has been issued.15 Transitron asserts that Hughes violated section 184 by filing its application for the British Patent of Addition without either obtaining a license or filing the identical information with the United States Patent Office at least six months before filing in Britain, so that Hughes' patent was invalid from the time of its issuance. To complete this theory of fraud, Transitron must argue not only that Hughes' patent was invalid because of this violation, but that Hughes knew or believed that its patent was invalid on this ground when it negotiated the license agreement with Transitron, and that it deliberately concealed this fact from Transitron.16

The district court considered this theory and rejected it. 487 F. Supp. at 903. The court held that Hughes was not required to seek a license for foreign filing because claims 60 and 61, which it found to be "virtually identical" to the British filing, were filed (as claims 102 and 103) on March 8, 1954, more than six months before the filing of the British application on November 8, 1954. Transitron does not dispute the dates cited by the district court; its challenge is to the court's holding that Hughes' United States filing and the British filing were sufficiently identical for the United States filing to trigger the six month waiting period.17

We turn therefore to the question of identity between claims 60 and 61, as filed in the United States, and Hughes' application for the British Patent of Addition. Neither the statute itself, 35 U.S.C. § 184, nor the regulations under it provide much guidance on the question of what constitutes an application for purposes of triggering the six-month waiting period, although the regulations related to the scope of licenses for foreign filing do suggest an effort by the patent office to ensure its opportunity to review in advance the details of any disclosure to be made in a foreign application.18 A few cases have addressed this issue, but the courts involved have adopted conflicting standards. In In re Gaertner, 604 F.2d 1348 (C.C.P.A.1979), the Court of Customs and Patent Appeals adopted a strict standard, indicating that any and all information filed abroad must first be submitted for national security review. Other courts, however, have held that only discrepancies "essential to the identity of the invention" could trigger operation of the Act. Beckman Instruments, Inc. v. Coleman Instruments, Inc., 338 F.2d 573, 576 (7th Cir. 1964); Control Systems Research, Inc. v. Aerotech, Inc., 429 F. Supp. 914, 917 (W.D.Pa.1977); Blake v. The Bassick Co., 245 F. Supp. 635, 636 (N.D.Ill.1965). In Control Systems the court indicated that the Act would not apply to differences necessitated by a foreign language or by the "idiosyncracies of practice in a foreign patent office." Id., at 917.

Comparing claims 60 and 61 (Appendix I) with the specifications for the BPA (Appendix II), we note that the language of the actual claims in the two filings is very similar. The most noticeable difference between the two filings is not in the claims themselves, but in the filing with the British application of a three-page preamble, which explains in detail the difficulties with the earlier diode production and the advantages of the claimed improvement, and an accompanying drawing, both of which were not filed in the United States. The district court found that this preamble was submitted on the advice of British patent counsel. 487 F. Supp. at 902. Our opinion in General Instrument, 399 F.2d 373, establishes that the differences between claims 60 and 61 and the BPA were of considerable significance for the purpose of evaluating the claims to decide whether they involved new matter. Whether the differences are significant for purposes of the Invention Secrecy Act is another question, however. We think it clear that the discrepancies go beyond the language and style of practice differences that Control Systems would permit; we are less certain that they are essential to the identity of the invention, so as to fail the test applied in that case and others agreeing with it. If we were to apply the stricter test of Gaertner, condemning any unauthorized foreign disclosure without considering its significance for national security, we think a strong case could be made that Hughes committed a violation of the Act. But Transitron's task in this case is to show not only that Hughes violated section 184, but also that the violation was so clear that Hughes must have known of it and of the consequent invalidity of its patent at the time it agreed to the license. Given the lack of consensus on the legal standard, and in light of Transitron's burden of proving fraud by clear and convincing evidence, we think the question of violation is close enough that we cannot impute to Hughes the requisite knowledge.

Transitron urges us to make an exception to the general rules against restitution for cases involving violation of the Invention Secrecy Act, even where the violation of the Act is not so clear as to form a basis for a finding of fraud. Transitron cites no authority, but relies on the policy underlying section 185 that compliance with the Act be strongly encouraged by depriving violators of any of the benefits of patent protection. Were a clear and knowing violation shown, we would have little hesitation in applying the punitive policy of section 185; but in such a case the licensor would be guilty of fraud, and the licensee would be entitled to restitution without the need for any special exception for the Invention Secrecy Act. In this type of case, where the question of violation of section 184 depends on a debatable legal conclusion as to whether the foreign filing and the corresponding United States filing were similar enough to meet the requirements of the Act, we think the reasoning of Troxel is applicable. Allowing recovery of royalties already paid would tend to encourage delay in challenging the validity of a patent, whether the ground of challenge is based on the Invention Secrecy Act or on other provisions of the patent laws. Similarly, a patentee who acts in the good faith belief that his patent is valid should not be subject to the continuing risk of liability for the return of royalties upon a finding of invalidity, no matter what the technical basis for invalidity of the patent.

Since we reject Transitron's theories of fraud, we do not review the district court's finding that Transitron did not rely on misrepresentations by Hughes, or its ruling that fraud in inducing agreement to a patent license cannot give rise to a tort claim.

Affirmed.

CLAIMS 102(60) AND 103(61) AS FILED WITH THE UNITED STATES

102. A semiconductor device comprising a vitreous envelope having an inner chamber and a longitudinal axis, said envelope including a tubular body section of substantially uniform external cross-section at right angles to said axis, and first and second end sections, at least the major portion of each of said end sections constituting a solid vitreous member having a cross-section at right angles to said axis equal to said external cross-section of said tubular section; first and second solid, one-piece lead wires extending through said first and second end sections, respectively, along said longitudinal axis and hermetically and directly sealed to said end sections, each of said lead wires having a first end terminating within said chamber, a semiconductor element mounted on, supported by and electrically connected to the first end of said first lead wire; and a resilient metallic element mounted on and supported by the first end of said second lead wire and contacting said semiconductor element; the length of the seal between each of said lead wires and the respective end section along said longitudinal axis being at least 1.5 times the maximum cross section dimension of the sealed portion of the lead wire.

103. A semiconductor device as defined in claim 102 in which transverse, outside dimension of said envelope is of the order of five times the maximum cross-sectional dimension of said lead wire.APPENDIX II

Inventors: HARPER QUA NORTH and JUSTICE NEAL CARMAN, Jr.

(SEAL) 768,740

Complete Specification: Nov. 8, 1954.

No. 32309/54.

(Patent of Addition to No. 721,201,

dated March 20, 1951).

Complete Specification Published: Feb. 20, 1957.