This case, which involves the validity and enforceability of certain patents on products and processes used in seaming carpets, is before this court for the second time. In the first trial, the trial court held the patents-in-suit to be invalid and unenforceable on two grounds: fraud in the procurement of a patent not in suit, which rendered the patents-in-suit invalid under the doctrine of “unclean hands,” and patent misuse and violations of the antitrust laws. 1 On appeal, this court reversed the trial court’s decision and remanded the case for further proceedings on the grounds that the record was not developed fully enough to allow this court to determine the points of law *573 upon which the judgment rested, and that the trial court had failed to articulate and apply- the proper legal standards in assessing the presence of fraud, patent misuse, and antitrust violations. 2

On remand, following a second trial, the trial court again held the patents-in-suit to be invalid and unenforceable on the basis of the same defenses relied upon in its prior decision, and also on the basis of four new defenses: (1) the inventions protected by the patents were “obvious” from the prior art, (2) the inventions were derived from the work of a third party, (3) the failure to disclose to the Patent Examiner the source from which the inventions were derived was fraud, and (4) the patent infringement suits against appellees were barred by laches. The trial court denied appellees’ motion for attorneys’ fees under 35 U.S.C. § 285 , finding that this was not an “exceptional case” warranting such an award.

Appellant challenges the trial court’s rulings with respect to each of the defenses, and cross-appellants challenge the trial court’s denial of their motion for attorneys’ fees as an abuse of discretion.

For the reasons stated below, we affirm the judgment of the trial court that the patents-in-suit are invalid because they are obvious from the prior art. We reverse the rulings of the trial court with respect to the other defenses, however, finding those rulings to be wholly without support in the law or the evidence. Finally, we affirm the denial of appellees’ motion for attorneys’ fees under 35 U.S.C. § 285 as properly within the discretion of the trial court.

Appellant, Carpet Seaming Tape Licensing Corporation (“Carpet Seaming”), a Texas corporation which is the exclusive licensee under the three patents originally issued to Charles Burgess, brought these patent infringement actions against appellees, Best Seam Incorporated (“Best Seam”), a California corporation making and selling hot-melt adhesive carpet seaming tape, Vectron Industries, Inc. (“Vectron”), another California corporation also making and selling hot-melt adhesive carpet seaming tape, and Eugene Tassee, the President of Vectron and, with his wife, the owner of all its stock. The patents-in-suit cover the products and techniques utilized in a hot-melt face-seaming process for installing carpet, which may be briefly described as follows.

The method for seaming carpets most accepted today is a face-seaming process using a hot-melt adhesive tape. Carpet sections are positioned pile-side-up on the floor, and the edges to be joined are rolled back just far enough to allow placement of the tape beneath the seam. The adhesive which is in solid form on the tape is then melted, and the carpet edges pressed down upon it for bonding as the adhesive resolidifies. This system has largely supplanted sewing as well as back-searning (a process in which carpet sections are turned pile-side-down, and the seaming work is performed on the back of the carpet) since it allows the work to be done quickly and without the need for moving cumbersome sections of the carpet once they have been positioned as the installer desires.

The tape utilized in this process is composed of three elements. The upper-most element is a layer of hot-melt adhesive. Hot-melt adhesive is solid and nonadhesive at room temperature. When heated, it becomes molten and forms a bond that it retains when cooled and resolidified. During the manufacturing of the tape, the adhesive layer is bonded to the second element of the tape, a layer of synthetic mesh that lends the tape strength. The seaming process actually involves the bonding of two pieces of carpet to this single strip of mesh. The final element of the tape is a paper barrier that prevents the adhesive layer *574 from bonding to whatever lies beneath it when it is heated.

In March, 1966, Burgess, after consultation with his patent attorneys, applied for and received a patent (“703”), which is the “parent” of the three patents-in-suit, arid which disclosed the process described above, using a two-element tape, without a paper barrier layer. In December, 1966, Burgess filed a continuation-in-part application, disclosing the same process, but using a composite three-element tape incorporating a barrier web (“876”), the structure of the tape (“038”), and a method for producing the tape (“830”). In January, 1968, Burgess sold his business and the pending applications to Giffen Industries, Inc. (“Giffen”), which subsequently transferred them to appellant. Besides obtaining exclusive licenses for the Burgess patents, appellant also obtained exclusive licenses for the so-called “Clymin” and “Winkler” patents, which disclosed certain improvements on the Burgess “038” patent. 3

Much of the evidence introduced at the second trial pertained to the source of Burgess’ idea of utilizing a hot-melt adhesive for face-seaming carpet, and in particular the role of one Buzz Powell and one Robert Walters in introducing Burgess to the properties of Thermo-Grip, a new thermo-plastic adhesive. In February, 1966, Buzz Powell, a carpet accessory salesman, called on Burgess, who was then the owner of a carpet store and installation business in Macon, Georgia. Powell showed Burgess a hot-melt adhesive, Thermo-Grip, and a gun for dispensing the adhesive that had recently been developed by The United Shoe Machinery Company. Powell also informed Burgess that Robert Walters, manager of a carpet store in La Grange, Georgia, had explained and demonstrated to Powell yarious possible applications of Thermo-Grip for use in seaming carpet. The question of precisely what it was that Walters demonstrated to Powell and what Powell subsequently communicated to Burgess is disputed; it is clear, however, that at the least, Powell showed Burgess how to back-seam carpet using Thermo-Grip and a glue gun, . by extruding molten adhesive from the gun along the back side of the carpet and then pressing a burlap strip onto the adhesive.

At the conclusion of the second trial, the court again entered judgment for appellees, holding the patents-in-suit invalid and unenforceable on the basis of the same fraud and misuse defenses as before, and also on the grounds that (1) the inventions were “obvious” over the prior art, (2) the inventions were derived from the work of a third party, (3) the failure to disclose to the Patent Officer the source from which the inventions were derived was fraud, and (4) the suits against appellees were barred by laches.

In support of its decision, the trial court set forth consolidated findings of fact and conclusions of law consisting of the same findings and conclusions relied upon in its prior decision, and of amended further findings of fact and conclusions of law taken virtually in toto from the proposed findings of fact and conclusions of law submitted by the appellees. 4

*575 II

We affirm the ruling of the trial court that each of the patents-in-suit are invalid for obviousness under 35 U.S.C. § 103 , and the standards enunciated in Graham v. John Deere Co., 383 U.S. 1 , 86 S.Ct. 684 , 15 L.Ed.2d 545 (1966).

It is clear that a patent cannot issue if the invention would have been obvious to a person of ordinary skill in the relevant art at the time the invention was made. 35 U.S.C. § 103 . The Supreme Court in Graham set forth a three-part factual inquiry to guide the courts in determining whether an invention is obvious. Under the Graham analysis, the trier of fact must determine (1) the scope and content of the prior art, (2) the differences between the prior art and the claims at issue, and (3) the level of ordinary skill in the pertinent art. Id. at 17 , 86 S.Ct. at 693 . The Court also indicated that “secondary considerations,” such as commercial success, long-felt need, and the failure of others to develop a solution, may be relevant. Although the obviousness of an invention is ultimately a question of law, subject to our independent review, the factual findings made by the trial court pursuant to the John Deere analysis must be upheld by the appellate court unless they are clearly erroneous. Sarkisian v. Winn-Proof Corp., 688 F.2d 647 (9th Cir.1982). Finally, it should be noted that although a patent is ordinarily presumed valid, and this presumption can be rebutted only by clear and convincing evidence, where the obviousness of a patent is in issue and the applicant fails to disclose pertinent prior art, the presumption disappears, unless the undisclosed prior art is merely cumulative of the cited art. See Carson Mfg. Co. v. Carsonite Int’l Corp., 686 F.2d 665 , 667 (9th Cir.1981).

The trial court first found that Powell’s demonstration of back-seaming using Thermo-Grip and a glue gun constituted pertinent prior art, and that Burgess’ failure to disclose the Powell demonstration to the Patent Examiner made it impossible for the Examiner to assess the obviousness of Burgess’ inventions and rebutted the pre-

sumptions that the patents were valid. The trial court then went on to perform the factual analysis required by John Deere, and concluded that Burgess’ inventions simply combined various ■ elements known in the prior art, in a manner which would have been obvious to a person of ordinary skill in the art who was acquainted with the superi- or qualities of Thermo-Grip.

With regard to appellant’s contention that the trial court erred in not adequately distinguishing between the different claims asserted in the three patents-in-suit, it appears from examination of the findings and the record that, although the court did not make three separate and distinct analyses of the prior art applicable to each patent, its analyses and its findings adequately addressed each of the claims made by the patents, particularly in view of the fact that the three patents were issued under later-filed divisional applications of the “parent” patent and the elements of the three claims were closely intertwined. Moreover, this circuit has indicated that although strict adherence to the Graham analysis is necessary, it is not essential that the court make detailed and explicit findings as to every pertinent fact so long as examination of the findings made and of the record as a whole shows that the judge has grappled with the critical issues underlying the determination of obviousness. Mollura v. Miller, 609 F.2d 381, 383 (9th Cir.1979), cert. denied, 446 U.S. 918 , 100 S.Ct. 1852 , 64 L.Ed.2d 272 (1980); National Lead Co. v. Western Lead Products Co., 291 F.2d 447, 451 (9th Cir.1961).

The trial court found that the level of ordinary skill in the art was that of persons knowledgeable in carpet installation techniques and products at the problem-solving level. As to the process patent, the court found that the use of a three-layer hot-melt tape for back-seaming, and the use of liquid latex poured onto a tape for face-seaming, were old in the art, and that the use of a hot-melt tape for face-seaming, though never practiced, had been suggested in the prior art. As to the tape patent, the court found that all of the elements claimed in *576 the tape patent were found in the prior art, though not in the exact combination proposed by Burgess: the art disclosed composite, three-layer hot-melt tapes, and the art disclosed two-layer tapes, that were not precoated with adhesive, but that did possess the feature of a barrier layer wider than the scrim, as claimed in the Burgess patent. From this analysis the court concluded that all Burgess had done was to combine old elements of the art for a purpose that would have been “obvious” to anyone familiar with the properties of the new Thermo-Grip adhesive. 5 The court considered the conflicting testimony of persons knowledgeable in the art as to whether Burgess’ combination was “obvious,” and was apparently not convinced by assertions that the invention was not obvious simply because no one had thought of combining these elements previously. The trial court also rejected Burgess’ contention that the use of a specially-shaped iron (which was never offered for patenting) and the fact that Burgess was not afraid to soil the iron by placing it in direct contact with the adhesive constituted patentable invention.

Recent decisions in this circuit evaluating combination patents — patents that combine various elements each of which are already disclosed in the prior art — indicate that such patents are subject to particularly close scrutiny for obviousness, because a patentable invention is less likely to be found in a combination of known elements. See, e.g., Sarkisian, supra. Moreover, the fact that such a patent achieves great commercial success has not been regarded as convincing evidence of patentability. See, e.g., Sakraida v. Ag Pro, Inc., 425 U.S. 273 , 96 S.Ct. 1532 , 47 L.Ed.2d 784 (1976) (commercial success due to benefits in terms of convenience and cost are not sufficient grounds for finding a combination patent valid); Tveter v. A.B. Turn O Matic, 633 F.2d 831 (9th Cir.1980), cert. denied, 451 U.S. 911 , 101 S.Ct. 1983 , 68 L.Ed.2d 300 (1981) (to be patentable an innovation must embody invention, and invention excludes adjustments, alterations, and improvements that could be expected to result from the exercise of the skill and ingenuity of a mechanic charged with knowledge of all that is disclosed in prior art).

Examination of the record below indicates that the trial court properly performed the factual analysis required by Graham , and that the trial court’s conclusion that the patents-in-suit are obvious over the prior art is fully supported by the law and the evidence. 6 Accordingly, we *577 affirm the judgment below that the patents-in-suit are invalid for obviousness.

Ill

Having concluded that the trial court properly held the patents-in-suit invalid for obviousness, we turn now to a consideration of the other affirmative defenses relied upon by the trial court as additional grounds for its decision. For the reasons stated below, we reverse the trial court’s rulings with respect to each of the following affirmative defenses on the grounds that they are unsupported by the law or the evidence in this case.

The trial court held the patents-in-suit invalid under 35 U.S.C. § 102 (a), (f), and (g), on the grounds that the Burgess inventions were derived from the work of Walters, as communicated to Burgess by Powell.

Sections 102(a), (f), and (g) provide, respectively, that a patent is not available if the invention was known or used by others before the invention thereof by the patent applicant; the applicant did not himself invent the subject matter sought to be patented; or the invention was made by another who had not abandoned, suppressed, or concealed it. Proof of derivation must be demonstrated by particularly clear and convincing evidence, Tucker Aluminum Products v. Grossman, 312 F.2d 293 (9th Cir.1961), that the entire conception was communicated. Hedgewick v. Akers, 497 F.2d 905 (C.C.P.A.1974).

The trial court found that there was clear and convincing evidence that Powell communicated to Burgess the idea of face-seaming carpet with a precoated hot-melt tape, and went on to find that the essential concept underlying the Burgess inventions was derived from Burgess’ knowledge of Walters’ work. An examination of the record below, however, indicates that these factual findings, which are critical to a finding of derivation, are not supported by the evidence. Conflicting evidence was presented at the trial on the question of precisely what information Powell transmitted to Burgess with respect to the work done by Walters. Although Burgess conceded at trial that Powell demonstrated back-seaming with a glue gun, it was sharply disputed whether Powell also disclosed any or all of the following ideas to Burgess: face-seaming with a glue gun, back-seaming with a precoated tape, or face-seaming with a precoated tape. Nothing short of a disclosure of the concept of face-seaming with a precoated tape would suffice to support a finding of derivation, because, as noted above, this is the only idea which was not previously disclosed in the prior art known to the Patent Examiner. The only pertinent testimony on this issue was Walters’ uncorroborated testimony that he used precoated tapes to face-seam; Powell’s testimony that he did not recall whether he saw Walters face-seam with a precoated tape, much less that he communicated that idea to Burgess; and the testimony of Burgess and his employee Hall, who was present during the Powell visit, that all Powell disclosed was the use of a glue gun for back-seaming.

Thus, even assuming that Walters did in fact practice face-seaming with a precoated tape, there was no testimony whatsoever indicating that Powell disclosed this information to Burgess. Accordingly, the trial *578 court’s finding that there was clear and convincing evidence- that the central concept underlying the Burgess inventions was derived from Walters’ work, as disclosed to Burgess by Powell cannot stand, and the trial court’s holding that the patents-in-suit are invalid for derivation is reversed.

The trial court held the patents-in-suit invalid for fraud on the basis of Burgess’ failure to disclose to the Patent Examiner certain information with regard to (1) the necessity for a paper barrier web in order to practice the hot-melt face-seaming process, and (2) the information derived from the Powell demonstration. With respect to the issue of the barrier web, the trial court set forth three alternative conclusions in support of its finding of fraud: (1) Burgess’ failure to disclose the necessity for a barrier web in connection with the application for the 703 patent, the parent to the patents-in-suit, was gross negligence or bad faith; (2) assuming Burgess believed that the need for a barrier was common knowledge in the art, and did not need to be disclosed, he showed gross negligence or bad faith in not so informing the Examiner, so that the Examiner could properly evaluate the 876 patent, which did employ a barrier web; and (3) Burgess was guilty of gross negligence or bad faith in not informing the Examiner that the 703 patent was only operable on concrete, and thus that the patent had only limited utility.

In its earlier opinion, this court reversed a similar ruling by the trial court on the grounds that it had not used the proper standards in assessing fraud, noting that fraud requires an element of subjective culpability in the form of bad faith or at least gross negligence, and that proof of that element must be clear and convincing. Carpet Seaming Tape Licensing Corp. v. Best Seam Inc., 616 F.2d 1133, 1138-40 (9th Cir.1980). This court further noted that in order for inadequate disclosures pertaining to the 703 patent to justify a finding that the patents-in-suit were invalid under the doctrine of “unclean hands,” the cases emphasize the need for fraudulent or deceptive intention. Id. at 1140-41 .

The record does not contain evidence of gross negligence or bad faith sufficient to meet the clear and convincing standard. On the contrary, the evidence suggests that Burgess’ disclosures were made in good faith and were in fact adequate. There is ample support in the record for a finding that Burgess’ failure to disclose the need for a paper backing in connection with his application for the 703 patent was justifiable because the need for some sort of paper barrier to keep the adhesive from adhering to the floor was common knowledge in the art. Burgess’ failure to disclose this fact when he applied for the subsequent patents cannot support' a finding of fraud because the Patent Examiner was fully aware that the barriers had been used in the prior art, because several of the patents that he cited in searching the application disclosed such barriers. Finally, the failure to disclose the fact that the 703 patent was only operable on concrete floors unless some sort of paper barrier, such as newspaper, was used is essentially a restatement of the court’s first theory, which we rejected above. Moreover, the fact that an invention has only limited utility and is only operable in certain applications has not been regarded as grounds for finding a patent invalid for lack of utility. Freedman v. Overseas Scientific Corp., 248 F.2d 274 (2d Cir.1957).

The trial court also found fraud arising from Burgess' failure to disclose the Powell demonstration to the Patent Examiner on the grounds that Burgess thereby deprived the Examiner of information which would have led to the discovery that Burgess’ inventions were obvious and were derived from the work of a third party.

The trial court’s findings with respect to this issue cannot stand, for two reasons. First, the discussion above of the trial court’s findings with respect to derivation establishes that the Powell demonstration was not pertinent prior art because the information imparted to Burgess was considerably less pertinent to the Burgess inventions than other prior art that was fully *579 known to the Patent Examiner. Second, the record is devoid of evidence that Burgess and his attorneys, to whom Burgess did disclose the Powell demonstration, did not make a reasonable and good faith conclusion that the demonstration lacked the materiality to require disclosure. Although the trial judge is clearly empowered to weigh the evidence and to decide issues of credibility, including that of Burgess’ state of mind in failing to disclose the Powell demonstration, the trial court’s discretion is constrained by the requirement that fraud be demonstrated by clear and convincing evidence. We find that the evidence relied on by the trial court is, as a matter of law, insufficient to meet the clear and convincing requirement, and accordingly, reverse the trial court’s ruling that the patents-in-suit were invalid for fraud.

The trial court found the patents-in-suit to be invalid and unenforceable for patent misuse and violations of the antitrust laws, arising from certain activities engaged in by appellant with regard to the accumulation and pooling of patents, the imposition of a uniform royalty rate for licensing patents to competitors, and attempts to enforce the patents against competitors by institution of infringement actions. The trial court held that appellant had attempted to monopolize the carpet seaming tape industry in violation of § 2 of the Sherman Act, finding that appellant, through its predecessor Giffen Industries, had manifested a specific intent to monopolize the carpet seaming tape industry, had engaged in anti-competitive conduct towards that end, and had possessed a dangerous probability of success. The court further found that the activities engaged in by appellant would constitute patent misuse even without a showing of specific intent and anticompetitive impact.

In our earlier opinion reversing the rulings of the trial court with respect to patent misuse and antitrust violations, this court indicated that the trial court had failed to make the findings of specific intent and probability of success necessary for a violation of § 2 of the Sherman Act, and had incorrectly applied the law pertaining to the accumulation and enforcement of patents against competitors. Carpet Seaming, supra, 616 F.2d at 1141-43 .

Following the second trial, the trial court made express findings of intent to monopolize and probability of success; in all other respects, however, the trial court’s analysis of the misuse and antitrust issues mirrors the analysis contained in its first opinion.

The trial court based its findings of anti-competitive intent upon what it apparently considered to be the inherently anticompetitive nature of certain activities engaged in by appellant. Specifically, the trial court found that the agreement between appellant’s predecessor, Giffen, and Bruck Industries, Inc. (“Bruck”), whereby those companies pooled the Burgess patents and the Winkler and Clymin patents (which disclosed modifications of the hot-melt tape developed by Burgess) was illegal. The court found that the terms of the agreement, and particularly the imposition of a uniform royalty rate for licensees under the pooled patents and attempts to enforce the patents against competitors by the bringing of infringement actions, gave Bruck and Giffen an unfair competitive advantage. The court did find that the Burgess patent “blocked” the Winkler patent, but went on to suggest that a nonexclusive license to Giffen under the Winkler patent would have been a more satisfactory method for resolving the dispute between Giffen and Bruck than the agreement described above.

This court noted in its earlier opinion that, absent additional evidence of anticompetitive purpose or impact, an agreement to pool patents and to exploit those patents in the manner described above is legally permissible. Id. at 1142 . We also noted that a finding that the Burgess patents blocked either the Clymin or the Winkler patents would provide additional support for the inference that the pooling agreement was for a legitimate, rather than an anticompetitive purpose. Id. Finally, this court indicated that attempts to enforce rights under *580 a patent against competitors are perfectly proper, and that infringement actions are presumed to be lawful and in good faith, a presumption which can only be rebutted by clear and convincing evidence. Id. at 1143 .

Thus, the mere fact that appellant engaged in these activities cannot, without more, support a finding that appellant possessed the specific intent required for a violation of § 2 of the Sherman Act, particularly in view of the trial court’s express finding that the Burgess and Winkler patents occupied a “blocking” relationship. Similarly, absent evidence that the agreement between Bruck and Giffen, which was patterned on a settlement agreement approved by this circuit in Cutter Laboratories v. Lyophile Cryochem, Inc., 179 F.2d 80 (9th Cir.1949), was itself improper, the trial court’s speculation as to other preferable alternatives cannot support a finding of illegality.

The trial court’s finding that appellant possessed a “dangerous probability of success” is likewise unsupported by the evidence. While the court did find that four companies had either gone out of business, gone into bankruptcy, or ceased making carpet tape in the past few years, neither the findings of fact nor the record as a whole demonstrate any causal connection between this result and appellant’s activities, and the record is devoid of evidence with respect to the market shares of the parties and their competitors.

In short, no additional evidence was presented at the second trial that would support a finding of patent misuse or antitrust violations in accordance with the standards set forth in our earlier opinion, and the trial court’s findings of specific intent to monopolize the carpet seaming industry and of dangerous probability of success are not supported by the record. Accordingly, the holdings of the trial court with respect to the patent misuse and antitrust defenses are reversed.

The trial court held that, even if the patents-in-suit were found to be valid, appellant’s patent infringement claim was barred by the doctrine of laches. The court based this holding on a finding that appellant had waited more than six years from the time it first learned of the alleged infringement before filing suit, thus triggering a presumption of laches, and that appellant had failed to rebut the presumption by proving an excuse for the delay.

A delay of six or more years between the time at which the plaintiff learns of the alleged infringement and the time at which suit is brought triggers a presumption that the delay is unreasonable and unexcused; at this point, the burden shifts to. the patentee to prove that the delay was excusable, and the patentee may not rely on other litigation as an excuse for the delay in the absence of notice to the alleged infringers that the patentee was delaying enforcement of the patent until conclusion of the pending litigation. Jensen v. Western Irrigation & Manufacturing, Inc., 207 U.S.P.Q. 817 (9th Cir.1980). Conversely, where suit is brought within six years of learning of the alleged infringement, the burden is upon the defendant to show that the delay was unexcused and that the defendant suffered injury as a result of the delay.

The trial court’s ruling of laches rested on a finding that it was “probable” that appellant had learned of the infringement more than six years before instituting infringement actions. The record is devoid of any evidence that would support such a finding; indeed, the only direct evidence presented on this issue at trial indicates that appellant learned of the infringement less than six years before the suits were filed. The trial court found that the activities alleged to have infringed the patents-in-suit had commenced by 1969. The suits against the appellees were filed in 1976. The only evidence as to the time at which appellant’s predecessor, Giffen, actually learned of the infringement was the testimony of one of Giffen’s employees suggesting that the infringement was discovered in June, 1971, just prior to the time that Gif *581 fen sent out letters giving notice of infringement.

Under these circumstances, where the only evidence actually presented at trial establishes that appellant learned of the infringement in 1971, less than six years before the patent infringement suits were brought, the trial court’s finding that it was “probable” that appellant learned of the infringment more than six years earlier cannot be upheld, even under a “clearly erroneous” standard. Because the trial court’s holding of laches rested entirely upon the presumption triggered by a delay of over six years, rather than upon proof by appellees that the delay was unexcused or that appellees suffered actual injury as a result of the delay, the holding that appellant’s claims are barred by laches is reversed.

Best Seam and Vectron, as cross-appellants, challenge the denial of their motion for attorneys’ fees under 35 U.S.C. § 285 , which authorizes the court to award reasonable attorneys’ fees to the prevailing party in an “exceptional case.” The trial judge found that the instant case was not an “exceptional case” justifying an award of attorneys’ fees under § 285, but did not set forth its reasoning on this issue.

In the recent case of Mayview Corp. v. Rodstein, 620 F.2d 1347, 1357 (9th Cir.1980), this court restated the two basic principles which may be gleaned from this circuit’s numerous decisions under 35 U.S.C. § 285 as follows: (1) a district court’s determination as to whether a case is to be deemed “exceptional” is discretionary, and is not to be overturned unless it results from an abuse of discretion or an erroneous conception of the law; and (2) the district court’s discretion in awarding attorneys’ fees in patent cases may be invoked only upon a finding of bad faith or inequitable conduct on the part of the losing party that would make it grossly unjust for the prevailing party to be left with the burden of his litigation expenses.

The court’s decision to reverse the trial court’s rulings with respect to the affirmative defenses of fraud and antitrust violations, which are the only bases upon which bad faith or inequitable conduct could be found, is sufficient to dispose of cross-appellants’ claim under the second principle stated in Mayview. Moreover, even if a finding of bad faith or inequitable conduct could be supported by the record, the decision to award attorneys’ fees under § 285 is committed to the discretion of the trial court. Cross-appellants are unable to cite any cases where this court has decided to award fees under § 285 where the district court has denied such fees in the first instance, and this case does not present any special and compelling circumstances that would suggest that this court should depart from its long-standing policy. Accordingly, the denial of cross-appellants’ motion for attorneys’ fees under § 285 is affirmed.

Conclusion

For the reasons stated above, we affirm the judgment of the trial court on the grounds that the patents-in-suit are invalid for obviousness; we reverse, however, the rulings of the trial court with respect to each of the other affirmative defenses relied upon as additional grounds for holding the patents to be invalid and unenforceable, namely, derivation, fraud, patent misuse and antitrust violations, and laches, finding each of these rulings without support in the law or the evidence. We affirm the trial court’s denial of attorneys’ fees under 35 U.S.C. § 285 .

The judgment of the trial court is AFFIRMED in part, and REVERSED in part.

. Carpet Seaming Tape Licensing Corp. v. Best Seam, Inc., 197 U.S.P.Q. 230 (C.D.Cal.1977). The sole issue at the first trial was the validity and enforceability of the patents, the parties having stipulated in advance to the fact of infringement if the patents were found to be valid and enforceable. At the conclusion of appellant’s case-in-chief the trial judge, pursuant to Federal Rule of Civil Procedure 41(b), granted appellees’ motion to dismiss on the grounds that the patents-in-suit were invalid and unenforceable, and, therefore, that upon the facts and the law appellant had shown no right to relief.

. Carpet Seaming Tape Licensing Corp. v. Best Seam, Inc., 616 F.2d 1133 (9th Cir.1980). In conjunction with this point, we noted that Rule 41(b) motions should be granted only in clear cases, and that in the interest of obtaining a full and complete record for both the trial and the appellate court, it is generally advisable to put the defendant to its proof and to decide the case after all the evidence has been adduced.

, The Clymin patent, issued to Conso Engineering Company (“Conso”), a subsidiary of Consolidated Foods Corporation, in 1969, disclosed a three-element tape containing a layer of spaghetti-like tracks of hot-melt adhesive which run together upon melting to form a sheet, thus providing the installer with a quick visual index for gauging adequate heating of the adhesive. The Winkler patent, issued to Brack Industries, Inc. (“Brack”) in 1973, dis-closed a three-element tape in which embossed ribs of solidified adhesive, joined at their base regions by a thin layer of adhesive, extend continuously on the tape.

In 1976, following several years of litigation between Giffen, Conso, and Brack, a settlement agreement was reached pursuant to which Giffen formed Carpet Seaming Tape Licensing Corporation, the appellant corporation herein, and received exclusive licenses under the Burgess, Clymin, and Winkler patents. The history of the litigation and the terms of the settlement agreement are more fully set forth in this court’s earlier opinion. Carpet Seaming, supra note 2 , 616 F.2d at 1136 -37 nn. 1-2.

. In our earlier opinion, we stressed the importance of accurate and independent fact finding by the trial judge, and admonished the court below not to rely mechanically upon findings of fact submitted by the parties. Id. at 1137-38 n. 3.

. The only difficulty with the trial court’s analysis is its reliance upon Powell’s demonstration of back-seaming with Thermo-Grip and a glue gun as pertinent prior art which Burgess failed to disclose to the Patent Examiner, and which rebuts the presumption of the patents’ validity. The trial court’s conclusion that Thermo-Grip was the “missing link” which, once discovered, would make the Burgess patent obvious, is without any support in the record. Neither the trial court’s findings of fact nor the record as a whole contain any indication as to how Thermo-Grip represented an advance over previously used hot-melt adhesives. Moreover, there is no factual support for the conclusion that the use of a hot-melt glue gun to back-seam carpet is any more pertinent to the Burgess invention than other prior art that was fully disclosed to the Patent Examiner; on the contrary, it appears less pertinent than the use of precoated three-layer hot-melt tape for back-seaming carpet, for example, that was disclosed in the Dildilian-Bigelow-Sanford patent. See note 4 supra.

*577 The trial court did make another finding as to the prior art, however, that is directly relevant to the issue of obviousness. The trial court found that Walters had, in the course of his experiments with Thermo-Grip, made up precoated tapes and used them both for back-seaming and face-seaming. It cannot be said that this finding is clearly erroneous, because the trial judge had the opportunity to hear testimony from Walters and other witnesses and was free to conclude that Walters’ version of the events was the most credible, even though Walters’ testimony was only partially corroborated. Although the trial court did not expressly rely upon this finding in concluding that Burgess’ invention was obvious, Walters’ practice of back-seaming and face-seaming with precoated tapes is clearly pertinent prior art sufficient to rebut the presumption that the Burgess patents were valid, and together with the other evidence noted above, supports the conclusion that the invention is obvious.