BALDWIN, Circuit Judge.

Loctite Corporation (Loctite) appeals from a judgment of the United States District Court for the Eastern District of Wisconsin, 225 USPQ 67 (1984), which denied Loctite’s claims that U.S. Patent No. 3,435,-012 (’012), issued to Nordlander in 1969, and U.S. Patent No. 4,165,400 (’400), issued to DeMarco in 1979, were infringed, and which declared the ’400 patent invalid. Loctite, the assignee of both patents, appeals the invalidity and noninfringement holdings with respect to claim 1 of the ’400 patent and the noninfringement holding with respect to claims 1, 3-5, and 9-10 of the ’012 patent.1

Ultraseal Ltd., Ultraseal America, Inc., Imprex, Inc., and Aluminum Casting and Engineering Co., Inc. (collectively, Ultra-seal), the alleged infringers, cross appeal from the denials of their antitrust counterclaim and request for attorney fees.

We affirm in part, vacate in part, and remand for proceedings consistent with this opinion.

A. The Technology

Porous articles, such as wood, ceramics, and particularly metal, frequently must be impregnated and sealed before use to fill

the article’s pores. The product used to fill the pores is called an “impregnant” or “sealant.” The impregnation process generally involves immersing the porous article into a liquid sealant bath to permit the pores to fill, removing excess sealant from the article, and “curing” (i.e., polymerizing to a solid form) the sealant in the pores.

The patents at issue involve “anaerobic” curing compositions, a type of sealant dating back to the 1950’s. Though the definition of “anaerobic” is at the heart of this litigation, the term generally relates to the ability to cure when oxygen is absent but not cure when oxygen is present.

B. The Patents

Claim 1 of the '012 patent is representative:

[figure]

The hydroperoxide catalyst, according to ’012, is designed to promote the polymerization of the monomer upon exclusion of air.

Claim 1 of the ’400 patent reads:

The ’400 patent emanated from a parent application containing claims to the process and the composition used therein. After a restriction requirement by the United States Patent and Trademark Office (PTO), a divisional application containing the process claim at issue was filed and that claim issued in the ’400 patent. The composition claims issued in U.S. Patent No. 4,069,378 (’378), which was also assigned to Loctite. In February 1980, Loctite filed a reissue application for the composition claims of the ’378 patent. A final rejection of the claims under 35 U.S.C. § 103 was made, the PTO Board of Appeals (Board) affirmed the rejection, and this Court affirmed the Board. In re DeMarco, 714 F.2d 160 (Fed.Cir.1983) (unpublished opinion).

The dispute arose in 1979, when the energy crisis forced major automobile manufacturers to use lighter metals such as aluminum. Those metals frequently contained numerous microscopic pores and cracks that required sealing.

In April 1979, Loctite filed a complaint against Ultraseal in district court charging direct infringement under 35 U.S.C. § 271(a), inducing infringement under § 271(b), and contributory infringement under § 271(c). The charges alleged (1) importation into the United States and sales of Ultraseal’s product, PC504, with instructions for use, and (2) use of PC504 in an impregnation process.

Ultraseal denied infringement, asserted the invalidity of the patents as a defense, and counterclaimed that Loctite was trying to enforce its patents knowing that Ultra-seal was not infringing — conduct allegedly an attempt to monopolize in violation of federal antitrust law, specifically section 2 of the Sherman Act.2 Ultraseal sought treble damages, costs, and attorney fees under the antitrust laws3 and attorney fees under the patent laws.4

In August, 1979, Loctite filed with the International Trade Commission (ITC) a complaint alleging unfair competition based on Ultraseal’s infringement of the ’378 patent. Eventually, Loctite moved to withdraw its ITC complaint on the basis that the pertinent ’378 patent claims were invalid; the motion was granted and the complaint was dismissed with prejudice.

D. District Court Opinion

The only antitrust issue tried below was that of intent as it related to Ultraseal’s allegation of bad faith enforcement of the patents. Earlier, the district court had severed damages relating to the patent claim and all economic issues relating to the antitrust counterclaim. 219 U.S.P.Q. 93 (E.D. Wisc.1982).

On August 8, 1984, the district court issued its opinion denying Loctite’s infringement charges and holding the ’400 patent invalid under 35 U.S.C. § 103. The court upheld the validity of the ’012 patent and denied Ultraseal’s antitrust claim and request for attorney fees. Ultraseal does not challenge on appeal the district court’s holding on the validity of the ’012 patent.

1. Whether the district court clearly erred in finding that claim 1 of the ’400 patent and claims 1, 3-5, and 7-9 of the ’012 patent were not infringed;

3. Whether the district court erred in denying Ultraseal’s antitrust counterclaim; and

4. Whether the district court abused its discretion in denying Ultraseal’s request for attorney fees.

OPINION

A. Infringement

It is uncontested that Ultraseal’s PC504 contains a monomer and an initiator selected from the monomers and initiators of the patent claims. Though PC504 also includes components not required by the claims,5 such inclusion does not avoid infringement if the claims otherwise read on the accused product or process. See, e.g., Radio Steel & Mfg. Co. v. MTD Products, Inc., 731 F.2d 840, 848, 221 USPQ 657, 663-664 (Fed.Cir.), cert. denied, — U.S. —, 105 S.Ct. 119, 83 L.Ed.2d 62 (1984).

Thus, the dispute does not focus on the identity of the PC504 components. Instead, it focuses on whether PC504 is an

“anaerobic” sealant composition. Ultraseal has carefully selected a monomer and initiator, and added other ingredients not required by the claims, to give it a composition and process which, it alleges, are not “anaerobic.” The claimed inventions, however, require anaerobicity.

Although it appears in the preambles of the ’012 patent claims, the term “anaerobic” breathes life and meaning into the claims and, hence, is a necessary limitation to them. See, e.g., Perkin-Elmer Corp. v. Computervision Corp., 732 F.2d 888, 896, 221 USPQ 669, 675-676 (Fed.Cir.), cert. denied, — U.S. —, 105 S.Ct. 187, 83 L.Ed.2d 120 (1984). It is also a limitation on process claim 1 of the ’400 patent, which requires, inter alia, use of an “anaerobic-curing acrylate monomer,” and “permitting the anaerobic sealant to cure.” The specification, claims, and prosecution history of the ’400 patent make clear that the “to cure” step, and the claim in general, require an “anaerobic” cure.

Henceforth in this opinion, “anaerobic” (with quotation marks) refers to the specific definition of that term in the claims, whereas cure(s) or curing anaerobically (without quotation marks) means the property of curing in the absence but not in the presence of oxygen. The district court interpreted “anaerobic” as a composition that rapidly and spontaneously cures anaerobically without the addition of an outside influence such as heat or a transition metal.

The district court found that Loctite’s commercial embodiment, PMSIOe, “cures at room temperature,” that “Ultraseal’s PC504 cures [only] after inducing 9Q°C. (195°F.) of heat into the composition,” and that PC504 “appears to require the addition of an outside influence such as heat or contact with a substantial amount of a transition metal, such as copper, in order to cure.” Hence, the district court determined that the two products were “different” and, implicitly, that there was no literal infringement.

The district court hinged “the test of Loctite’s claim” on “whether the Ultraseal product performs substantially the same function in substantially the same way to obtain the same result as the Loctite product,” citing as support Sanitary Refrigerator Co. v. Winters, 280 U.S. 30, 42, 50 S.Ct. 9, 13, 74 L.Ed. 147 (1929) (a case involving the doctrine of equivalents). 225 USPQ at 69. The district court found that, “[although Ultraseal’s product and process are very similar to Loctite’s” and “[ajlthough Ultraseal’s composition performs substantially the same function with substantially the same result as the Loctite process, I am not persuaded that the two operate in substantially the same way. Thus, I do not believe that Ultraseal is guilty of infringement.” Id. at 71.

The district court read into the definition of “anaerobic” a requirement of being able to cure anaerobically rapidly and spontaneously at room temperature and in the absence of transition metal. We agree with including that requirement for the ’012 claims, but not for the ’400 claim. We also disagree with the district court’s treatment of the doctrine of equivalents, which is relevant to the ’012 (and possibly the ’400) patent under our analysis. Accordingly, we vacate the judgment of no infringement, and remand for further proceedings.

Claim interpretation, a threshold inquiry when resolving infringement, is a question of law. Fromson v. Advance Offset Plate, Inc., 720 F.2d 1565, 1569, 219 U.S.P.Q. 1137, 1140 (Fed.Cir.1983). Hence, when interpreting “anaerobic” we need not defer to the district court under a “clearly erroneous” standard. Titanium Metals Corp. of America v. Banner, 778 F.2d 775, 782 (Fed.Cir.1985). We would use that standard to review the fact question of literal infringement, i.e., do properly interpreted claims read on the accused product or process. ACS Hospital Systems, Inc., v. Montefiore Hospital, 732 F.2d 1572, 1582, 221 U.S.P.Q. 929, 936 (Fed.Cir.1984).

The patent law “allows the inventor to be his own lexicographer.” Autogiro Co. of America v. United States, 384 F.2d 391, 397, 155 U.S.P.Q. 697, 702, 181 Ct.Cl. 55 (1967). To ascertain the true meaning of disputed claim language, resort should be made to the claims at issue, the specification, and the prosecution history. See, e.g., McGill, Inc. v. John Zink Co., 736 F.2d 667, 673-675, 221 U.S.P.Q. 944, 948-951 (Fed.Cir.), cert. denied, — U.S. —, 105 S.Ct. 514, 83 L.Ed.2d 404 (1984); Fromson, 720 F.2d at 1569-1571, 219 U.S.P.Q. 1140, 1141. Moreover, claims should be construed as they would be by those skilled in the art. Fromson, 720 F.2d at 1571, 219 U.S.P.Q. at 1142.

2. Claim Interpretation — “Room Temperature” and “Absence of Transition Metal” Limitations

The ’012 patent discloses that the ingredients of the composition may be mixed and stored at ordinary room temperature, but that disclosure, like that of the ’400 patent, fails to indicate that the composition of the claims must be able to cure anaerobically at room temperature. Both disclosures are devoid of any teaching that suggests the criticality of a particular temperature to effectuate cure — the emphasis is on the presence or absence, of oxygen. Indeed, expert witnesses representing both parties testified at trial that there is no requirement that an “anaerobic” curing composition has to be one that is able to cure anaerobically at room temperature. The ’012 patent itself teaches a cure at room temperature or higher:

That the specific examples set forth in the patents occur at room temperature does not mean that the claims are imbued with a room temperature limitation. Generally, particular limitations or embodiments appearing in the specification will not be read into the claims. See D.M.I., Inc. v. Deere & Co., 755 F.2d 1570, 1574, 225 U.S.P.Q. 236, 239 (Fed.Cir.1985); Fromson v. Advance Offset Plate, Inc., 720 F.2d 1565, 1568, 219 U.S.P.Q. 1137, 1139 (Fed.Cir.1983).

Nor is there any requirement in the specification that the composition be able to cure anaerobically in the absence of transition metal. To the contrary, both specifications suggest the use of the curing composition with metal and non-metal surfaces.

The district court, however, was influenced — and so are we — by statements made during prosecution of the ’012 patent. A rejection under 35 U.S.C. § 103 had been made by the PTO using references by Krie-ble and Karo. The Karo reference showed curing compositions containing the chemical compound formamide as a starter. Arguing against Karo, Nordlander expressly defined “anaerobic sealant and adhesive compositions” as “compositions which are stable in the presence of oxygen but which cure rapidly in the absence of oxygen without additional outside influences.” He presented a supporting affidavit stating that the prior art composition would not “perform as a true anaerobic composition, i.e., some driving force other than the removal of oxygen is necessary before the sealant will cure,” and he argued that his composition “is capable of serving as a true anaerobic sealant, and therefore effective without the formamide starter.”

In light of those statements, we agree with the district court that, to be “anaerobic” within the meaning of the ’012 claims, and hence to infringe those claims literally, a composition must be able to cure anaerobically (i.e., in the absence but not in the presence of oxygen) without some outside influence such as heat or a transition metal.

The “no outside influence” limitation, however, should not apply to the '400 patent, for which there is no similar prosecution history. Hence, if a composition cures anaerobically (and does so rapidly, as discussed, infra) in the presence of a transition metal but not in its absence, or at

90°C. but not at room temperature, its use could infringe the ’400 process claim literally (but only in the presence of the metal or at 90°C.). If on remand the district court finds that such curing occurs, liability of the PC504 user for direct infringement and of the PC504 seller for contributory or induced infringement would have to be considered with respect to the '400 patent.

Loctite argues that an “anaerobic” composition requires the presence of transition metal at any temperature, and, because PC504 will cure anaerobically in the presence of a transition metal, it is “anaerobic.” We disagree that “anaerobic” requires the presence of a transition metal.

Loctite focuses on this preamble language in the ’012 claims:

Unlike our reading of the preamble phrase “anaerobic,” which we (and apparently the parties and district court) construe as a necessary limitation to the claims, we interpret “adapted to remain ... metal surfaces” as merely language of intended use, not a claim limitation. See generally Kropa v. Robie, 187 F.2d 150, 152, 88 U.S.P.Q. 478, 481 (CCPA 1951).

Looking to the specifications, the ’012 patent discloses that the sealant mixtures of the invention may be applied to “glass, ceramic, and metal surfaces and combinations thereof.” The ’400 patent discloses, in its “Background” section, the need to seal not only metal but also “wood and ceramics.” Nowhere do the specifications require contact with a metal surface; therefore, there is no reason to limit the claims as such.

The prosecution history of the ’012 patent also cuts against a transition metal requirement. The ’012 patent discloses that, if the metal concentration exceeds 20-50 parts per million, instability results. When the Examiner requested that that limitation be inserted into the claims, applicant stated that “metal ion contamination is not an inherent part of this system”; it is “totally extrinsic” to the claims. The presence of metal being “totally extrinsic” to the claims suggests that transition metal is not a required component of the claims.

We agree with the district court’s implicit conclusion that the claims of both patents require curing anaerobically to occur “rapidly and spontaneously.” The ’012 patent states:

The ’012 patent uses the phrase “rapidly and spontaneously” in numerous other places to describe the cure. The ’400 patent disclosure stresses “the delicate balance between initiation and inhibition of polymerization” that is “destroyed and the composition will cure” when oxygen is removed, suggesting to the district court (and us) that once the “balance” is upset with the removal of oxygen, the cure is rapid and spontaneous.

Moreover, in remarks made to an Examiner in response to a prior art rejection during prosecution of the ’012 patent, the applicant stated: “[tjhis patent application concerns anaerobic sealant and adhesive compositions (compositions which are stable in the presence of oxygen but which cure rapidly in the absence of oxygen without additional influence).”

As for the precise parameters of “rapid and spontaneous,” the ’012 patent states that the amount of catalyst and optional tertiary amine activator added to the monomer should be sufficient “to polymerize the

monomer within about 8 hours when the composition is confined between close fitting metal parts.” It also states that the setting time “may be varied over a wide range” by the proper selection of materials, amounts, and temperature, and that if desired, it “may be reduced to as little as a few minutes or may be extended to the range of an hour or so where rapid setting is not essential or desired.” On the other hand, the specific examples test the cured material for strength after periods of 24 hours, and in one instance 36 hours.

The district court did not expressly define “rapid and spontaneous.” However, when comparing PC504 with Loctite’s PMSIOe, the district court appeared to weigh heavily an experiment in which PMSIOe cured after six hours at room temperature whereas PC504 did not cure within 24 hours at room temperature. The district court appeared to use that as a basis for finding no infringement, and hence, we construe the district court opinion as imposing on the definition of “rapid and spontaneous” an upper limit of something less than 24 hours. In the context of Loctite’s arguments and the record, we cannot disagree with that determination. Hence, the finding that PC504 did not literally infringe the ’012 patent, i.e., was not able to cure anaerobically within 24 hours at room temperature without a transition metal present, is not clearly erroneous. In so holding, we leave open the possibility that the district court on remand, when defining “rapid and spontaneous” in the context of determining possible literal infringement of the ’400 patent, may have to lower its limit of something less than 24 hours if the facts and arguments so warrant. Also, we do not rule out the possibility that use of a composition exceeding the limit, e.g., a 25 hour cure (which would preclude finding literal infringement), may nevertheless infringe the ’400 patent under the doctrine of equivalents, a doctrine we discuss in greater detail below.

In sum, we interpret “anaerobic” for both patents as the ability to cure anaerobically, i.e., in the absence but not in the presence of oxygen, and to do so rapidly and spontaneously. The district court added as a requirement the ability to do so at room temperature without the addition of transition metal. We read that added limitation into the ’012 claims but not into the ’400 claim. Hence, we affirm the district court’s holding that the ’012 claims are not literally infringed, but we remand for a determination of literal infringement of the ’400 patent using the proper definition of “anaerobic.” The remand should also address infringement by equivalence of the ’400 patent (if there is no literal infringement) and of the ’012 patent (discussed below).

PC504 does not literally infringe the ’012 claims because, though it cures at 90°C., it cannot rapidly cure anaerobically at room temperature without transition metal. Nevertheless, PC504 can infringe the ’012 claims under the “doctrine of equivalents” if it and the claimed invention perform substantially the same function in substantially the same way to give substantially the same result. See, e.g., Atlas Powder Co. v. E.I. DuPont DeNemours & Co., 750 F.2d 1569, 1579, 224 U.S.P.Q. 409, 416 (Fed.Cir.1984). Whether PC504 and the claimed invention are equivalent is a question of fact to be resolved in the first instance by the district court and reviewable here under a clearly erroneous standard. Id.6

The district court found that PC504, though performing substantially the same function to give substantially the same result, did not operate in “substantially the same way” as PMSIOe. However, the district court’s approach in making that finding was incorrect as a matter of law.

As a threshold matter, the district court compared PC504 with Loctite’s commercial embodiment, and that is error. The comparison of PC504 must be with the claimed invention. See, e.g., Nestier Corp. v. Menasha Corp., 739 F.2d 1576, 1579, 222 U.S.P.Q. 747, 749 (Fed.Cir.1984), cert. denied, — U.S. —, 105 S.Ct. 1756, 84 L.Ed.2d 819 (1985).

Comparing PC504 with the claimed invention, and assuming it is found on remand that PC504 rapidly cures anaerobically at 90°C., the two would differ in only one respect. The claimed composition could rapidly cure anaerobically at room temperature, whereas PC504 — through Ultraseal’s selection of claimed monomer and initiator, and of other ingredients not required by the claims — could rapidly cure anaerobically only at 90°C. In essence, it was precisely because of that difference that the district court found that the claimed invention and PC504 do not perform in substantially the same way. That finding, however, would allow the difference itself to dictate a finding of no equivalence, and if that were the law, one could never have infringement by equivalence. The analysis must go further, and the question the district court should consider on remand is this: given the difference, would the accused composition at 90°C. and the claimed invention at room temperature perform substantially the same function {e.g., filling the pores of the treated material with solid material) in substantially the same way {e.g., by rapidly curing in the absence but not in the presence of oxygen) to give substantially the same result {e.g., a filled material).

That analysis assumes a finding on remand that PC504 rapidly cures anaerobically at 90°C. No such finding has yet been made. Though the district court found that PC504 cures at 90°C., it did not state that (1) it did so rapidly and spontaneously in the absence of oxygen and (2) in the presence of oxygen at that temperature it will not cure rapidly and spontaneously. Those conditions determine whether the composition rapidly cures anaerobically at 90°C., and, hence, whether PC504 infringes the ’012 claims by equivalence (or the ’400 claim literally).

There are limitations to the doctrine of equivalents. The doctrine has been “judicially devised to do equity” in situations where there is no literal infringement but liability is nevertheless appropriate to prevent what is in essence a pirating of the patentee’s invention. Hughes Aircraft Co. v. U.S., 717 F.2d 1351, 1361, 219 U.S.P.Q. 473, 480 (Fed.Cir.1983). Concomitantly, two policy oriented limitations, applied as questions of law, have developed. First, the doctrine will not extend to an infringing device within the public domain, i.e., found in the prior art at the time the patent issued; second, prosecution history estoppel will not allow the patentee to recapture through equivalence certain coverage given up during prosecution. See, e.g., Perkin-Elmer Corp. v. Computervision Corp., 732 F.2d 888, 900, 221 U.S.P.Q. 669, 678 (Fed.Cir.1984), cert. denied, — U.S. —, 105 S.Ct. 187, 83 L.Ed.2d 120 (1984).

Prosecution history estoppel is relevant here to the statements made by Nordlander during the ’012 patent prosecution to circumvent prior art showing a for-mamide starter. Though the same statements were used, supra, to interpret claims in view of the prosecution history, the two doctrines are to be distinguished. Interpreting claims in view of the prosecution history applies as a preliminary step in determining literal infringement. Prosecution history estoppel applies as a limitation to the doctrine of equivalents after the claims have been properly interpreted and no literal infringement is found. Estoppel

does not involve a reinterpretation of the claims. Id.

Because of the Nordlander statements, we interpreted “anaerobic” in the ’012 patent, and hence literal infringement, to require that the composition have the ability rapidly to cure anaerobically in the absence of an outside influence such as heat or metal. In the prosecution history estoppel context, though, different considerations apply. The issue now is whether Nordlan-der’s statements on “outside influence” preclude the non-literally infringed claims from covering under the doctrine of equivalents compositions curing anaerobically with “outside influences,” more specifically with heat or metal.

Hughes Aircraft Co. v. U.S., 717 F.2d 1351, 1362, 219 USPQ 473, 481 (Fed.Cir.1983), acknowledged the view of some courts “that virtually any amendment of the claims creates a ‘file wrapper estoppel’ effective to bar all resort to the doctrine of equivalents.” However, Hughes “reject[ed] that view as a wooden application of estoppel, negating entirely the doctrine of equivalents and limiting determination of the infringement issue to consideration of literal infringement alone.” Id. The court concluded:

Id. Thus, whenever the doctrine is evoked, “a close examination must be made as to, not only what was surrendered, but also the reason for such a surrender;” the fact that claims were narrowed “does not always mean that the doctrine of file history estoppel completely prohibits a patentee from recapturing some of what was originally claimed.” Bayer Aktiengesellschaft v. Duphar Internat'l Research, 738 F.2d 1237, 1243, 222 U.S.P.Q. 649, 653 (Fed.Cir.1984).7

Nordlander’s argument was in response to prior art showing use of a chemical (formamide) as a starter. A starter’s role is akin to the activating function of a transition metal, which is also a chemical. Consequently, Nordlander’s argument weighs heavily against allowing Loctite to cover under the doctrine of equivalents an accused product that depends on a transition metal activator to cause a composition rapidly to cure anaerobically. However, unlike the formamide used in the reference, and unlike transition metal, heat is not a chemical. Because of that, and because the ’012 patent contemplates curing anaerobically at temperatures higher than room temperature, because Ultraseal’s product was found to be “very similar” to Loctite’s, and because Ultraseal may be using the most important aspect if not the gist of Loctite’s inventions, i.e., the ability to fill pores by rapidly curing anaerobically, the prosecution history will not necessarily prevent the district court from applying the doctrine of equivalents to compositions that rapidly cure anaerobically with heat. We decline to make a final conclusion now because the district court may, on remand, want to take additional evidence (if it wishes) and make additional findings underlying the question of prosecution histo

ry estoppel with respect to both transition metal and heat.

B. Validity of ’400

In holding the ’400 patent invalid under 35 U.S.C. § 103,8 the district court committed these legal errors: (1) the court appeared to have ignored the statutory presumption of validity; (2) the court made insufficient Graham findings; (3) the court gave questionable weight to objective evidence of nonobviousness; and (4) the court failed to consider the claimed process as a whole, inasmuch as it relied primarily on the Board decision respecting another invention as evidence of obviousness. Consequently, we vacate the holding of invalidity, and remand for further proceedings.

A patent is presumed valid. 35 U.S.C. § 282. The burden is on the party asserting invalidity to prove it with facts supported by clear and convincing evidence. SSIH Equipment, S.A. v. USITC, 718 F.2d 365, 375, 218 USPQ 678, 687 (Fed.Cir.1983). Though the district court “believe[d] there is clear evidence to support a finding of obviousness,” there is no indication in the district court’s opinion that it recognized the statutory presumption of validity, that the burden was on Ultraseal to overcome the presumption, or that Ultra-seal actually carried its burden. Recognition of the presumption is particularly important here, where the district court acknowledged that the obviousness question was “close.” Cf. Panduit Corp. v. Dennison Manufacturing Co., 774 F.2d 1082, 1097 (Fed.Cir.1985) (“The court’s uncertainty should have, in view of § 282, ended the obviousness inquiry.”)

Obviousness under 35 U.S.C. § 103 is a question of law based on the underlying factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 17, 86 S.Ct. 684, 693, 15 L.Ed.2d 545, 148 U.S.P.Q. 459, 467 (1966): (1) the scope and content of the prior art; (2) the differences between the prior art and the claims at issue; (3) the level of ordinary skill in the art; and (4) objective evidence of secondary considerations. See, e.g., Simmons Fastener Corp. v. Illinois Tool Works, Inc., 739 F.2d 1573, 1575, 222 U.S.P.Q. 744, 746 (Fed.Cir.1984), cert. denied, — U.S. —, 105 S.Ct. 2138, 85 L.Ed.2d 496 (1985).

Loctite argues that the district court made no findings on the first three inquiries of Graham. Without those findings, Loctite urges, there is no factual basis to permit a conclusion that the invention would have been obvious. We agree.

Rule 52(a) of the Federal Rules of Civil Procedure states that, in a non-jury trial, “the court shall find the facts specially and state separately its conclusions of law thereon.” “Where the trial court fails to make findings, the judgment will normally be vacated and the action remanded for appropriate findings to be made.” ACS Hospital Systems, Inc. v. Montefiore Hospital, 732 F.2d 1572, 1578, 221 U.S.P.Q. 929, 933 (Fed.Cir.1984). The rule does not place a severe burden upon the trial judge, for he “need only make brief, definite, pertinent findings and conclusions upon the contested matters.” 5A J. Moore & J. Lucas, Moore’s Federal Practice H 52.06[1] at 138 (2d ed. 1985). Moreover, “[wjhere a full understanding may be had without the aid of separate findings ... we recognize a narrow exception to the general rules.” ACS Hospital Systems, 732 F.2d at 1578, 221 U.S.P.Q. at 933. Also:

found those inferences that are necessary to support the ultimate finding....

The ultimate test of the adequacy of-findings is whether they are sufficiently comprehensive and pertinent to the issue to form a basis for the decision (and whether they are supported by the evidence). 5A Moore’s Federal Practice, II 52.06[1] at 142. In patent cases, the need for express Graham findings takes on an especially significant role because of an occasional tendency of district courts to depart from the Graham test, and from the statutory standard of unobviousness that it helps determine, to the tempting but forbidden zone of hindsight. Thus, we must be convinced from the opinion that the district court actually applied Graham and must be presented with enough express and necessarily implied findings to know the basis of the trial court’s opinion.

How many and specific the findings need be are questions to be resolved on a case-by-case basis. Sometimes, the failure to make findings is characterized as a “dereliction of duty.” Seattle Box Co. v. Industrial Crating & Packing, 756 F.2d 1574, 1578, 225 U.S.P.Q. 357, 360 (Fed.Cir.1985). In Jones v. Hardy, 727 F.2d 1524, 1529, 220 U.S.P.Q. 1021, 1025 (Fed.Cir.1984), the failure to make specific Graham findings constituted error. In Cable Electric Products, Inc. v. Genmark, Inc., 770 F.2d 1015, 1023, 226 U.S.P.Q. 881, 883-884 (Fed.Cir.1985), the failure to make a finding on the pertinency of prior art was held distinguishable from Jones because: Cable was a summary judgment case and Rule 52(a) typically does not require findings in such cases; a finding on pertinency (which is merely one aspect of Graham findings) is not strictly a Graham requirement; and the Jones failure to make findings was accompanied by other, more major flaws. King Instrument Corp. v. Otari Corp., 767 F.2d 853, 857, 226 U.S.P.Q. 402, 405 (Fed.Cir.1985), was distinguished from Jones because there was not a complete void of Graham findings as there was in Jones; also, Jones involved a “parade of legal errors.”

The present case is more like Jones than King or Cable. This description of the district court action in Jones could apply here:

727 F.2d at 1529, 220 U.S.P.Q. at 1025.

Here, the district court begins its opinion on validity by stating that 35 U.S.C. § 108 requires three inquiries to determine obviousness. Those inquiries, however, are virtually abandoned in the opinion, as the district court weighs very heavily the Board decision on the application for reissue of DeMarco’s '378 patent.

The district court set forth the Graham test as a three-part inquiry. As indicated above, however, it has a fourth prong, so-called secondary considerations which, when present, must be considered. Simmons Fastener Corp. v. Illinois Tool Works, Inc., 739 F.2d 1573, 1575, 222 U.S.P.Q. 744, 746 (Fed.Cir.1984), cert. denied, — U.S. —, 105 S.Ct. 2138, 85 L.Ed.2d 496 (1985). It does not appear that that was done.

One such secondary consideration is commercial success. However, despite Loc-tite’s argument that sales of the composition embodied in the ’400 invention soared when put on the market, the court did not make findings on whether the argued commercial success existed or whether such success was caused by the patented invention. Instead, it stated that:

time Loctite’s [product] was introduced, mainly due to the automobile manufacturers’ switch to lighter cars and lighter metals.

225 U.S.P.Q. at 75.

The district court also said that Loctite’s argument on sales should be discounted because it “did not value it highly enough to present it to the Board of Appeals.” However, we need not speculate on why such evidence was not presented to the Board. The important point is that it was presented here and, hence, should be considered.

There were other secondary considerations the court overlooked in its validity discussion. Earlier, in the “Facts” section of the opinion, the district court stated:

Id. at 68-69. The district court appeared to ignore those vital facts.

The court stated its belief that the “most critical evidence came from the inventor herself.” Her testimony was to the effect that she was presented with a problem, devised an idea to try, and thought the idea would be successful. However, that testimony should not itself compel a conclusion of obviousness. First, the inventor may be of a higher level of skill than someone of ordinary skill in the art, and what would have been obvious to her would not necessarily have been obvious to one of ordinary skill in the art. See, e.g., Interconnect Planning Corp. v. Feil, 774 F.2d 1182, 1138 (Fed.Cir.1985) (“The invention must be evaluated not through the eyes of the inventor, who may have been of exceptional skill, but as by one of ‘ordinary skill.’ ”); Standard Oil Co. v. American Cyanamid Co., 774 F.2d 448, 454 (Fed.Cir.1985) (“The actual inventor’s skill is irrelevant to the inquiry.”) Here, there were no findings as to the level of ordinary skill. Second, this court has often said that the test of § 103 is not “obvious to try.” See, e.g., Milliken Research Corp. v. Dan River, Inc., 739 F.2d 587, 593, 222 U.S.P.Q. 571, 576 (Fed.Cir.1984). Third, “[patentability shall not be negatived by the manner in which the invention was made.” 35 U.S.C. § 103. In any event, the district court here should have considered all the evidence mandated by Graham, not just one piece of evidence.

The claimed invention must be considered as a whole in determining obviousness. Litton Industrial Products, Inc. v. Solid State Systems Corp., 755 F.2d 158, 164, 225 U.S.P.Q. 34, 38 (Fed.Cir.1985); W.L. Gore & Associates, Inc. v. Garlock, Inc., 721 F.2d 1540, 1548, 220 U.S.P.Q. 303, 309 (Fed.Cir.1983), cert. denied, — U.S. —, 105 S.Ct. 172, 83 L.Ed.2d 107 (1984).

Though the district court cautioned against focusing on a “gist,” it did not heed its own advice. It divided the process claim at issue into composition and process components, stated that the composition was the “core” of the invention, and then adopted the Board’s discussion of the reissue application, which involved the ’378 composition. The court’s reasoning was that, because the Board found the reissue composition obvious, “nothing appears in the claims of the ’400 patent itself which would alleviate the obviousness weakness that it has.”

Though consideration of an invention’s “gist” is appropriate in some contexts, e.g., in determining infringement under the doctrine of equivalents, when determining obviousness there is no legally recognizable or protected “essential,” “gist,” or “heart” of the invention. See, e.g., Medtronic, Inc. v. Cardiac Pacemakers, Inc., 721 F.2d 1563, 1567, 220 U.S.P.Q. 97, 101 (Fed.Cir.1984). Had the district court analyzed the invention in its entirety, it would have had to consider whether it would have been obvious to use the composition defined in the claims in the process defined by those claims. Even if a composition is old, a process using a known composition in a new and unobvious way may be patentable. In re Zierden, 411 F.2d 1325, 1329, 162 U.S.P.Q. 102, 104 (CCPA 1969).

Each statutory class of claims should be considered independently on its own merits. In re Durden, 763 F.2d 1406, 1410, 226 U.S.P.Q. 359, 361 (Fed.Cir.1985). It follows also that each ease should be decided on the basis of its own particular fact situation. Id. Certainly, each claim of a patent “shall be presumed valid independently of the validity of other claims.” 35 U.S.C. § 282. Cf. Interconnect Planning Corp. v. Feil, 774 F.2d 1132, 1137 (Fed.Cir.1985) (“When a patent has been reissued with claims that are not substantially identical to the original claims, the invention as a whole, as now claimed, must be evaluated in terms of 35 U.S.C. § 103. The original claims are not prior art against the reissued claims.”).

In sum, the district court failed to make sufficient Graham findings and otherwise improperly assessed obviousness. Rather than make Graham findings at the appellate level, which we cannot do, we remand for the district court to make appropriate findings and itself assess obviousness under the guidelines enunciated herein.

C. Ultraseal’s Antitrust Counterclaim

Ultraseal alleges that Loctite has brought suit in bad faith, i.e., without any belief in infringement, and that such action constitutes an attempt to monopolize in violation of section 2 of the Sherman Act.

We must approach a federal antitrust claim as would a court of appeals in the circuit of the district court whose judgment we review. See Atari, Inc. v. JS & A Group, Inc., 747 F.2d 1422, 1438-1440, 223 U.S.P.Q. 1074, 1086-1087 (Fed.Cir.1984) (in banc). Using that approach here, we apply Seventh Circuit law primarily and look to guidance (albeit non-binding) from other circuits, especially in uncharted areas.

To establish an illegal attempt to monopolize, plaintiff must prove: (1) a specific intent to monopolize; and (2) a dangerous probability that the attempt would be successful in achieving a monopoly in the relevant market. Photovest Corp. v. Fotomat Corp., 606 F.2d 704, 711 (7th Cir.1979), cert. denied, 445 U.S. 917, 100 S.Ct. 1278, 63 L.Ed.2d 601 (1980).

Here, the district court dismissed the counterclaim because it found that Loctite did not bring suit in bad faith. Had it found bad faith, it would have had to make specific findings defining the relevant geographic and product markets, and specifying the mai’ket share possessed by Loctite in the relevant market. See, e.g., American Hoist & Derrick Co., v. Sowa & Sons, Inc., 725 F.2d 1350, 1366, 220 U.S.P.Q. 763, 775-776 (Fed.Cir.), cert. denied, — U.S. —, 105 S.Ct. 95, 83 L.Ed.2d 41 (1984); Photovest Corp. v. Fotomat Corp., 606 F.2d at 711; Mercantile National Bank of Chicago, et al. v. Quest, Inc., Paul Magers, 431 F.2d 261, 267, 166 U.S.P.Q. 513, 517 (7th Cir.1970), cert. denied, 401 U.S. 956, 91 S.Ct. 977, 28 L.Ed.2d 239 (1971).9

Following Handgards, Inc. v. Ethicon, Inc., 601 F.2d 986, 202 USPQ 342 (9th Cir.1979), cert. denied, 444 U.S. 1025, 100 S.Ct. 688, 62 L.Ed.2d 659 (1980), the district court required Ultraseal to prove by “clear and convincing” evidence that Loctite brought its infringement suit in bad faith. Ultraseal contests that standard, urging us to accept the standard used in proving other elements of a civil antitrust cause of action — preponderance of the evidence. We decline to do so, and more importantly, we believe the Court of Appeals for the Seventh Circuit would decline to do so.

Handgards involved allegations that a patentee prosecuted an infringement action knowing the patent was invalid. It stated:

601 F.2d at 993, 202 U.S.P.Q. at 348. The problem recognized by Handgards “is to provide the means whereby the bad faith infringement action can be identified post hoc with a sufficiently high degree of certainty to make it highly improbable that the action in fact was brought in good faith.” 601 F.2d at 933, 202 U.S.P.Q. at 348. Imposing treble damages under the antitrust laws, “a sanction strongly punitive ... dictates that such means exist.” Id.

Handgards sought a solution to the problem. The court cited Walker Process Equipment, Inc. v. Food Machinery & Chemical Corp., 382 U.S. 172, 86 S.Ct. 347, 15 L.Ed.2d 247 (1965), as requiring a “clear and convincing” standard for the type of conduct involved there, fraudulent procurement of a patent as a basis for an antitrust claim. 601 F.2d at 996, 202 U.S.P.Q. at 3 51.10 The Handgards court saw in that standard a barrier “to prevent frustration of patent laws by the long reach of antitrust laws,” which “suggests our proper course,” i.e., “to erect such barriers to antitrust suits as are necessary to provide reasonable protection for the honest patentee who brings an infringement action” to protect his patent. The barrier identified by Handgards is “that a patentee’s infringement suit is presumptively in good faith and that this presumption can be rebutted only by clear and convincing evidence.” Id. Handgards specifically rejected a “preponderance standard” because that “would eliminate a barrier we hold necessary, and were it accepted as proper ‘might well chill’ legitimate patent enforcement efforts ‘because of fear of the vexatious or punitive consequences of treble damage suits.’ ” Id.

We agree with the above reasoning of Handgards. The patent system, which is rooted in the United States Constitution (Art. I, § 8, cl. 8), serves a very positive function in our system of competition, i.e., “the encouragement of investment based risk.” Patlex Corp. v. Mossinghoff, 758 F.2d 594, 599, 225 U.S.P.Q. 243, 247, modified on other grounds, 771 F.2d 480, 226 U.S.P.Q. 985 (Fed.Cir.1985). By so doing, it “encourages innovation and its fruits:

new jobs and new industries, new consumer goods and trade benefits.” Paulik v. Rizkalla, 760 F.2d 1270, 1276, 226 U.S.P.Q. 224, 228 (Fed.Cir.1985) (in banc). In that sense, therefore, and because the underlying goal of the antitrust laws is to promote competition, Brunswick Corp. v. Riegel Textile Corp., 752 F.2d 261, 266, 224 U.S.P.Q. 756, 758 (7th Cir.1984), cert. denied, — U.S. —, 105 S.Ct. 3480, 87 L.Ed.2d 615 (1985), the patent and antitrust laws are complementary.11 Consequently, the treble damage threat of antitrust liability should not be used to thwart good faith efforts at patent enforcement.

Ultraseal argues that Handgards should apply only to suits where the plaintiff knew that the patent was invalid, not to those where the plaintiff knew there was no infringement. The alleged basis for that argument is a statement in Handgards that the presumption of good faith attached to a patentee’s infringement suit “accords the patentee a presumption commensurate with the statutory presumption of patent validity.” 601 F.2d at 996, 202 U.S.P.Q. at 351. Ultraseal argues that, though the presumption of validity warrants a presumption that the patentee has a good faith belief in validity there is no similar presumption of infringement warranting a presumption that the patentee has a good faith belief in infringement.

Ultraseal’s argument is unpersuasive. The cited, gratuitous statement in Hand-gards was not the sole, not even the primary, reason for Handgards establishing the presumption of a good faith infringement suit. The primary, if not sole, reason was the public policy of erecting a barrier against thwarting patentees from asserting legitimate patent rights. Because that policy applies to a good faith belief in infringement as well as validity, there is no persuasive reason to apply the presumption to one but not the other.

Ultraseal argues that Ramsey v. United Mine Workers of America, 401 U.S. 302, 91 S.Ct. 658, 28 L.Ed.2d 64 (1971), requires a “preponderance” standard. We disagree. Ramsey involved an antitrust suit against a labor union and various coal producers. Based on section 6 of the Norris-LaGuardia Act, the court of appeals upheld a clear and convincing standard for all elements of the cause of action but the Supreme Court reversed, stating that section 6 made the clear and convincing standard applicable to only one element. The Court recognized that normally a preponderance of the evidence standard applies in civil antitrust actions, but it did not rule out use of a clear and convincing standard when appropriate and did not limit such use to situations expressly created by statute. To the contrary, degree of proof “has traditionally been left to the judiciary to resolve.” Santosky v. Kramer, 455 U.S. 745, 769-770, 102 S.Ct. 1388, 1403, 71 L.Ed.2d 599 (1984); Woodby v. Immigration and Naturalization Service, 385 U.S. 276, 284, 87 S.Ct. 483, 487, 17 L.Ed.2d 362 (1966).

The district court did not find clear and convincing evidence that Loctite’s infringement claims were “dishonestly” pursued. We cannot conclude that the district court’s finding was clearly erroneous.

The district court gave careful consideration to these findings underlying its ultimate finding of no bad faith: (1) Loctite performed chemical analyses, identifying the chemicals in PC504 as those in Loctite’s patented anaerobic curing compositions; (2) Loctite conducted tests which, to their satisfaction, demonstrated that PC504 exhibited the anaerobic characteristic of curing in

the absence of air; and (3) Loctite sought advice by outside patent counsel. Although the full opinion of counsel was missing from Loctite’s files, its loss did not indicate to the district court anything but negligence. Moreover, an internal Loctite memorandum contained a statement capsulizing outside patent counsel’s opinion — it stated that outside counsel had confirmed that the Nordlander patent had been infringed and that Loctite’s in-house counsel concurred.

We review those findings under a clearly erroneous standard. “Clearly erroneous” was recently explained in Anderson v. City of Bessemer City, N.C., — U.S. —, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985), as a standard governed by certain principles, the foremost being that “a finding is ‘clearly erroneous’ when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a violation has been committed.” 105 S.Ct. at 1511. Moreover:

105 S.Ct. at 1512. Using those principles, we conclude that the district court’s “no bad faith” finding, and its underlying findings, were not shown by Ultraseal to be clearly erroneous. Our conclusion in that regard is bolstered by our determination that the district court misinterpreted the claims and that, under the correct interpretation, there might be infringement.

D. Attorney Fees

Ultraseal asked for, but was denied, attorney fees under the antitrust laws and under 35 U.S.C. § 285.

The district court’s denial of the request under the antitrust laws is affirmed as part of our affirmance of dismissal of the antitrust counterclaim.

The denial of the § 285 request must also be affirmed. That section allows the district court to exercise its discretion and award attorney fees only in “exceptional” cases.12 Though prosecution of an infringement action in bad faith can be “exceptional,” the district court here found no bad faith and, as previously explained in discussing the antitrust counterclaim, that finding is not clearly erroneous. Hence, the district court did not abuse its discretion in denying Ultraseal’s request for attorney fees.

E. Conclusion

That part of the district court’s judgment denying Loctite’s infringement claims is vacated and remanded for proceedings consistent with this opinion.

That part of the judgment holding the ’400 patent invalid is vacated and remanded for proceedings consistent with this opinion.

That part of the judgment denying Ultra-seal’s antitrust counterclaim is affirmed.

That part of the judgment denying attorney fees sought under 35 U.S.C. § 285 and the antitrust laws is affirmed.

AFFIRMED IN PART, VACATED IN PART, AND REMANDED.