Defendants Johns-Manville Corporation and Johns-Manville Sales Corporation (collectively, J-M) appeal from a trial court judgment that they infringed a patent held by Lam, Inc. The court awarded Lam treble damages and attorneys' fees, and enjoined J-M from subsequent infringement.

Lam's patent, No. 3,950,638, issued on April 13, 1976, is for a lighting fixture (luminaire) that permits indoor use of a high intensity discharge (HID) lamp in rooms of average ceiling height. The first of the patent's two independent claims-for a luminaire with a fluted reflective bowl-is not at issue here. The second independent claim, number four of the patent, reads as follows:

"A luminaire for a high intensity lighting device said luminaire comprising: an upwardly opening reflective bowl, terminating in a substantially circular outer lip, the diameter of said reflective bowl being greater on the outer lip than at the lower end; a base integrally disposed on the lower end of said bowl; means for holding a high intensity lighting device substantially axially within said bowl and arranged so that the radiation which is emitted substantially below the horizontal plane of said outer lip will be reflected over said outer lip and means to hold said bowl in a fixed spatial relationship relative to said lighting device, the optical shape of said bowl being an elliptical surface of revolution, one focus of which is disposed at said source and the second focus being disposed at a point over said outer lip of said bowl."

The trial court found infringement of that claim and of claims eight and nine, dependent on claim number four:

"8. The luminaire according to claim 4 wherein a refractor is arranged upon the base.

"9. A luminaire according to claim 4 wherein substantially all radiation from said lighting device leaves the luminaire after no more than one reflection, thereby substantially eliminating trapping of the radiation with the luminaire or reflection through said lighting device."

In general, the patent claims describe a fixture intended to permit indoor use of an HID lamp by placing it in an upwardly opening luminaire having an elliptically surfaced bowl. The bowl is designed to reflect light beams out to a focal point just over the lip of the bowl, thereby dispersing the beams as closely to the horizontal as possible. As part of the fixture, a refractor may be placed at the base of the bowl, which will bend downward beams back up and out. The shape of the bowl prevents trapping and inefficiency by reflecting nearly all beams out of the bowl after a single reflection.

The principal dispute focuses upon the reflector's elliptical surface. As perhaps its main argument, J-M contends that its own reflector, otherwise nearly identical, does not infringe because the surface of its bowl is parabolic, or at least "closely approximates a parabolic surface."

It is useful, here, to discuss the nature of the HID lamp and Lam's development of the fixture for the HID lamp. Most indoor commercial lighting today uses fluorescent lamps. Fluorescent lamps, as well as the incandescent bulbs they have generally replaced, have been used both as direct lights (downlights) and indirect lights (uplights). HID lamps have been in general use since the 1930s and, due to recent improvements, can now serve general lighting needs. HID lamps produce extremely bright light, thus requiring considerably fewer fixtures.1 They are also much more efficient than fluorescent or incandescent light sources, consuming only about one-tenth the energy to light the same area. To date, the HID lamps have been widely used outdoors, but because of their extreme brightness and resulting glare they have not been used indoors except in factories, gymnasiums, and other areas with high ceilings.

Beginning in 1971, Lam tried to develop a method for utilizing HID lamps in rooms with conventional ceilings. After its early efforts using direct and indirect fixtures failed, Lam hired a consulting firm headed by Thomas M. Lemons. That firm experimented with both direct and indirect fixtures. Mounting the lamp in an indirect fixture blocks direct glare from below; however, if the indirect fixture allows light rays to shine directly to the ceiling, an HID lamp will produce an objectionable bright spot. Eventually, the consultants placed the lamp in an indirect fixture and pointed it downward, then used a bowl to reflect the light beams back up and out, nearly horizontally. As finally worked out, the lamp uniformly illuminates a wide area of the ceiling; from there, the beams reflect down toward the floor, almost as if the ceiling were a light source. The elliptical surface refocuses the light at a point just above the lip of the bowl, producing a wider angle of distribution than if the beams emerged further above the bowl. The LUXXtra fixture that Lam developed contains a bowl that is 8 inches high, 201/2 inches wide at the base, and 241/4 inches wide at the rim.2 The refractor is placed on the bottom of the bowl. The bowl's shape lets most light beams emerge after only a single reflection and thereby avoids trapping of the beams. As a consequence, the LUXXtra luminaire is very efficient, using over seventy percent of the light emitted by the HID lamp. The final design followed many earlier efforts, and it was late 1972 before Lam was satisfied with the product. Lam applied for its patent in late 1973.3

During this time, J-M's research unit was also attempting to develop an indirect fixture that would permit use of HID lamps in conventional rooms. J-M aimed its efforts principally at floor and wall-mounted units, although J-M made some attempt to design a ceiling fixture. The record is contradictory as to why J-M failed to develop an acceptable ceiling fixture; the experimental units' unattractive appearance may have been as important as the technological problems encountered in avoiding unacceptable glare. Whatever the reason, one of J-M's designers testified that it is "highly desirable" yet "rather difficult" to illuminate the ceiling directly above and still achieve a wide angle of light disbursement.4 App. V, 661.

Soon after the Lam patent was issued for the LUXXtra fixture in April 1976, the Broward County, Florida school board issued specifications for remodeling, including relighting, five elementary schools. The specifications called for Lam's LUXXtra fixtures and described the fixture's efficiency and light distribution pattern. The blueprints accompanying the specifications showed the size and shape of the LUXXtra reflector and its mounting. One of J-M's sales representatives had previously discussed the job with the lighting engineer who developed the specifications. Immediately after the County issued the specifications, J-M sent one of its salesmen to Florida to discuss the specifications with the engineer. They met on May 4, 1976, and the next day, through that salesman, J-M's designers had a copy of the blueprint as well as a Lam brochure that in some detail showed the size and shape of the bowl, use of a refractor on the bottom of the bowl, placement of the lamp vertically within the bowl, and use of an integral auxiliary light source to function until the lamp heats up and becomes effective. Neither the specifications nor the brochure referred to a specific shape for the reflector surface. On its cover the brochure stated, "patented optical system," and inside stated, "patents pending."

The J-M engineer who was given the blueprint and the Lam brochure, according to his own testimony, was told to design a fixture "similar to the Lam LUXXtra." He completed the design the very next day. The drawing went to production on May 7, and by May 14 J-M had calculated its costs and a selling price and was prepared to bid on the Broward County job. On May 16 J-M tested its luminaire, which J-M designated the CLASSPAK fixture, and the next day J-M's director of research took it to Florida for demonstration. On May 18 the lighting engineer recommended to the County that it add CLASSPAK to the specifications, which it did. In June the local Florida electrical contractors bid on the school job. Immediately before the contractors submitted their final bids, J-M several times lowered its bid to electrical suppliers, and the winning contractor based its bid on the CLASSPAK fixture. About two weeks later, Lam's attorney gave J-M notice of probable patent infringement. J-M responded by denying infringement and stating that at no time had it been provided with a diagrammatic representation of the LUXXtra fixture. Apparently the J-M official who sent the letter was unaware that J-M possessed the Broward County blueprint and the Lam brochure. The instant suit soon followed.

On appeal J-M contends that (1) the Lam patent is invalid because it was anticipated by prior art, it was obvious, and the subject matter was in public use or for sale for more than one year before the patent application was filed; (2) the patent is unenforceable because of Lam's bad faith misrepresentations to the patent office, including failure to bring to its attention relevant prior art; (3) J-M did not infringe either literally or under the equivalents doctrine, in part because of file wrapper estoppel; (4) the case does not warrant award of treble damages; (5) the case is not "exceptional," within the meaning of 35 U.S.C. § 285, and thus will not permit an award of attorneys' fees; and (6) the injunction was improperly phrased to preclude J-M's use of the trade name "CLASSPAK."

* VALIDITY

J-M challenges the validity of the Lam patent on the following grounds: prior art anticipated Lam's luminaire, which should not be presumed valid since the patent examiner did not consider the most relevant prior art; and the luminaire, even if not anticipated by prior art, was obvious, on which issue the trial court made no finding. An inventor is not entitled to a patent if the invention was anticipated by prior art or, as of the time it was made, would have been obvious to a person having ordinary skill in that art. 35 U.S.C. §§ 102(a)-(b), 103. The factfinder must ascertain the nature of the pertinent prior art, identify the differences between the subject invention and that art, and determine whether those differences are so slight that the invention should be considered "anticipated" by prior art, or, if not anticipated, would be "obvious" to one ordinarily skilled in that art. Graham v. John Deere Co., 383 U.S. 1, 17, 86 S. Ct. 684, 693, 15 L. Ed. 2d 545 (1966). The two questions-anticipation and obviousness-are related, although the latter is a broader, more general test. See, e.g., Escoa Fintube Corp. v. Tranter, Inc., 631 F.2d 682, 692-93 (10th Cir. 1980).

Since determining whether an invention was either anticipated or obvious requires inquiry into the state of the art and the art may be quite technical and difficult to understand, a court may also consider secondary characteristics that suggest whether the invention was obvious. These include whether the invention quickly met with commercial success, whether it met long-recognized but unsolved needs, and whether others had attempted to solve those needs but failed. See, e.g., Graham v. John Deere Co., 383 U.S. 1, 17-18, 86 S. Ct. 684, 693-694, 15 L. Ed. 2d 545 (1966).

By statute, a patent is presumed to be valid, 35 U.S.C. § 282, but the presumption is considerably weakened if the patent examiner did not consider relevant prior art. E.g., Sidewinder Marine, Inc. v. Starbuck Kustom Boats and Prods., Inc., 597 F.2d 201, 206 (10th Cir. 1979). While the ultimate issue of the patent's validity is a question of law, subissues such as nature of prior art, ordinary skill in the art, and obviousness are questions of fact not to be overturned unless clearly erroneous. E.g., Central Soya Co. v. Geo. A. Hormel & Co., 645 F.2d 847, 850 (10th Cir. 1981); Norfin, Inc. v. International Business Machs. Corp., 625 F.2d 357, 364-65 (10th Cir. 1980).

J-M argues that because elliptical and parabolic surfaces are commonly used in downlights, because downlights can be turned upside down and used as indirect lights, and because HID units had been used previously in that way, Lam was remiss in failing in its patent application to cite any downlight fixtures as relevant prior art. J-M also argues that the patent examiner never considered the most relevant prior art: J-M's own Holophane '485 fixture with its elliptical reflector, initially used as a downlight but later used as an indirect light; the Wendel patent, which utilizes an elliptical reflector in a downlight; and the Harling patent, which uses in a downlight a reflector surface with both parabolic and elliptical surfaces of revolution.

Coinventor Thomas M. Lemons, testifying as an expert witness for Lam, said that in the patent application he did not cite any downlights as prior art because he believed they were irrelevant. A downlight can be turned upside down and directed at the ceiling, but it distributes light poorly, and with an HID lamp it would produce unacceptable ceiling glare. According to Lemons, developing a fixture that used an HID unit in a low-ceilinged indoor room required him to "unlearn" many principles relevant to downlights. The patent examiner appears to have agreed that downlights were irrelevant since, according to the trial court, the examiner "considered downlights in the first action (the processing of the original patent application) but thereafter dropped all reference to them." App. I, 173. The trial court found that not only was Lemons familiar with the properties of elliptical and parabolic surfaces, but they were so well-known that the patent examiner must have known them too. As we discuss in part III B, the patent office records show the examiner's objection to the patent claims was not due to the previous use of elliptical or similar surfaces in downlights.

Equally significant, the trial court found that the patents J-M cited were not relevant prior art. The trial court found that the light distribution of the Holophane '485 and the Wendel were "far different and much less desirable." App. I, 167. Similarly, it found that the Harling patent was not relevant prior art, concluding that downlights simply have nothing to do with indirect lighting fixtures. Id. at 166. As the trial court points out, J-M's own inability to design a fixture suitable for low-ceilinged use of HID's and its quick copying of the LUXXtra luminaire undermine J-M's argument that its own Holophane '485 patent anticipated the Lam patent and made it obvious. Similarly, McGraw Edison, owner of the Harling patent, did not just turn it over and use it as an uplight, but instead employed outside consultants to do "a tremendous amount of photometric computation." App. V, 801. When the patent examiner does not dwell on prior downlight art that must have been known to him and the trial court finds such art irrelevant, the party charged with infringement has not overcome the presumption that the patent office made a correct determination. Cf. Saf-Gard Prods., Inc. v. Service Parts, Inc., 532 F.2d 1266, 1271 (9th Cir.) (presumption of patent validity strengthened when the trial court examined "the pertinent prior art and concluded that the patent was validly issued."), cert. denied, 429 U.S. 896, 97 S. Ct. 258, 50 L. Ed. 2d 179 (1976).

We have no problem with the adequacy of the trial court's findings on obviousness. When, as here, the state of the art is extensive and technical, we do not expect the trial court to describe it fully. Nor do we expect the court to describe fully the knowledge of one ordinarily skilled in the field. It is enough that the trial court consider the evidence presented as to the state of the art, including prior patents, grapple with the issues, then make findings sufficient to provide the appellate court with a clear understanding of the trial court's reasoning. See Sidewinder Marine, Inc. v. Starbuck Kustom Boats and Prods., Inc., 597 F.2d 201, 209 (10th Cir. 1979); Price v. Lake Sales Supply R. M., Inc., 510 F.2d 388, 391 (10th Cir. 1974). Upon review, we will not disturb these factual findings unless they are clearly erroneous.

In the case of the Lam patent, the trial court adequately considered the state of the art (including all prior art J-M cited), the differences between the prior art and the Lam invention, and the knowledge of those ordinarily skilled in the field. The court specifically concluded that LUXXtra "was not obvious to a person of ordinary skill in the art." App. I, 168. It found that while the laws of physics and geometric principles utilized were not new, "the total result of (Lam's) combination of those known laws and principles was new." Id. Secondary characteristics that support a finding of nonobviousness are Lam's, J-M's, and McGraw Edison's difficulty in developing a luminaire that would allow use of an HID unit in low-ceilinged rooms, and the commercial need for the unit, as evidenced by J-M's immediate design of the CLASSPAK product line, the obvious efficiencies of the unit, and its early sales success. We hold that the trial court's findings of nonanticipation and nonobviousness were not clearly erroneous.

An inventor is not entitled to a patent if the invention was "in public use or on sale in this country more than one year prior to the date of the application for patent in the United States." 35 U.S.C. § 102(b). J-M argues that an evaluation of the fixture by Jack DeVeuve, just over one year before Lam applied for its patent, constituted "public use." DeVeuve was a lighting engineer who subsequently sought to purchase the LUXXtra fixtures for a school project.

It is well settled that "(t)he good faith use of the device or apparatus for experimental purposes is not a public use within the intent and meaning of the statute." McCullough Tool Co. v. Well Surveys, Inc., 343 F.2d 381, 394 (10th Cir. 1965), cert. denied, 383 U.S. 933, 86 S. Ct. 1061, 15 L. Ed. 2d 851 (1966). Although J-M says that DeVeuve had no obligation to Lam to treat as confidential the information about the LUXXtra fixture, J-M characterizes the visit as "to inspect and make light measurements of the LUXXtra fixture suspended at varying heights from the ceiling." Appellants' Br. 51. The record makes clear that DeVeuve inspected a mock-up fixture Lam kept in a locked room. The trial court characterized the visit as "intended to obtain technical engineering comment," "to get his professional opinion on the qualities of the experimental fixture," "a visit and inspection (that) were in no way sales oriented," and noted that "sale of the fixture was not discussed nor was it in anyone's mind at that point." App. I, 160. When DeVeuve, impressed by the luminaire, asked permission for the school board members to see the mock-up, Lam refused "because the luminaire was still in the design process." Id.

The record clearly supports the trial court's finding that to the extent DeVeuve's visit constituted public use, it was for experimental purposes and does not invalidate the patent.

J-M argues that we should refuse to enforce the Lam patent because its inventors acted in bad faith in prosecuting the patent application. Specifically, J-M argues unenforceability because the patent applicants failed to disclose the most relevant prior art and misrepresented that the luminaire could be used in rooms as low-ceilinged as eight to nine feet, when it cannot be used where ceilings are lower than about ten feet.

A court will refuse to enforce a patent if the applicant obtained it through fraud or bad faith. An applicant must fully disclose all matters reflecting adversely upon the application, "and he risks non-enforcement of his monopoly on later discovery of a failure to fulfill that obligation." True Temper Corp. v. CF&I Steel Corp., 601 F.2d 495, 501 (10th Cir. 1979). In particular, data furnished to the patent office to compare performance of the invention with prior art must be fairly and accurately presented since the patent office has no means of verifying the data. Id. In this Circuit, a patent will not be enforced if the applicants obtained the patent by intentional fraud, or by misrepresentations or failures to disclose amounting to gross negligence. Id. at 502.

Regarding the Lam patent, J-M is not alleging that the applicants submitted improper test data. Rather, J-M's principal contention is that the applicants failed to cite downlight luminaires as prior art. However, as we have noted, Lam's consultant Lemons testified that he did not cite downlight patents because he believed they were irrelevant, a conclusion with which the trial court agreed, and the patent examiner evidently agreed as well. The trial court found "(n)othing in the record (that) suggests intent, reckless conduct or negligence on the part of the applicant in failing to cite (prior downlight art)," App. I, 173; Lemons "acted in good faith and ... was guilty of no negligence," id.; and the "(a)pplicants for the patent disclosed everything concerning prior art to the examiner a reasonable person would be expected to disclose," id. at 175. The record supports the trial court's findings, and we do not find them clearly erroneous.

J-M contends that the applicants misrepresented their invention as usable in rooms with ceilings as low as eight to nine feet, when in fact it cannot be used in rooms with ceilings lower than ten feet high. The only evidence J-M cites to support this argument consists of the notes of Jack DeVeuve, who visited the mock-up of the LUXXtra fixture. His notes apparently recommend a clearance of seven feet four inches between the base of the fixture and the floor. Pl. Ex. 82. We do not construe those notes as proof that the fixture cannot be utilized when ceilings are lower than ten feet, nor are they evidence substantially contradicting the claim that the luminaire is suitable for "relatively low ceilings such as found in schools, offices and shops," App. I, 9, or for "rooms of average height even as low as 8 or 9 feet," id. at 8. There is no evidence any misrepresentation here was made intentionally or with gross negligence. We hold that the patent is fully enforceable.

In determining whether there is infringement, a court looks first "to the words of the claims contained in the patent and if the accused device falls clearly and definitely within those claims, infringement is made out." McCullough Tool Co. v. Well Surveys, Inc., 343 F.2d 381, 401 (10th Cir. 1965), cert. denied, 383 U.S. 933, 86 S. Ct. 1061, 15 L. Ed. 2d 851 (1966). A device may infringe either "literally" by matching each feature of the patent claim, or it may infringe as the "equivalent" of the patented device. A device is "equivalent" if it "performs substantially the same function in substantially the same way to obtain the same result," Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 608, 70 S. Ct. 854, 856, 94 L. Ed. 1097 (1950) (quoting Sanitary Refrigerator Co. v. Winters, 280 U.S. 30, 42, 50 S. Ct. 9, 13, 74 L. Ed. 147 (1929)). However, a patent holder is estopped from asserting equivalency if he "gave up" the accused device's mode of operation in order to meet the patent office's prior art objections. This is determined by reviewing the history of the patent application as noted on the "wrapper" of the patent file; it is customarily termed "file wrapper estoppel." J-M argues that its luminaire does not literally infringe and is not the equivalent of LUXXtra, and that based upon the file wrapper history, Lam is estopped from a construction of its patent claim that would embrace a parabolic surface.

Whether a device literally infringes or is the equivalent, and whether the patent holder is subject to file wrapper estoppel, are questions of fact. We will not set aside the trial court's findings unless they are clearly erroneous. E.g., Milgo Elec. Corp. v. United Business Communications, Inc., 623 F.2d 645, 656 (10th Cir.), cert. denied, 449 U.S. 1066, 101 S. Ct. 794, 66 L. Ed. 2d 610 (1980); Eimco Corp. v. Peterson Filters & Eng'r Co., 406 F.2d 431, 438 (10th Cir. 1968), cert. denied, 395 U.S. 963, 89 S. Ct. 2105, 23 L. Ed. 2d 749 (1969). We must determine, then, whether the record supports the findings of infringement and no application of file wrapper estoppel.

A. Infringement: Literal or as an Equivalent

J-M characterizes the upper portion of its reflector as parabolic and the lower portion as a "special surface." While the trial court questioned the accuracy of J-M's demonstration, J-M purportedly showed that the upper part of its bowl reflects light in beams parallel to one another. The special surface reflects beams in a less orderly manner: they criss-cross one another and cross rays from the upper surface, and they do not converge at a single focal point or a single focal arc. Even though J-M's chief of domestic patent operations, Joseph J. Kelly, testified that as a whole the side wall of the CLASSPAK reflector is not a parabola (App. VI, 932), J-M still argues that the reflector "approximates a parabola," and that the upper part reflects beams in a parallel manner. Lam argues that standard ellipse formation techniques can reproduce J-M's reflector surface design and that the CLASSPAK fixture is a literal infringement.

We do not address the question whether the infringement was literal, however, since we find ample evidence that CLASSPAK is the equivalent of LUXXtra and that Lam is not estopped from arguing equivalency. As to equivalency, the trial court found as follows:

"Classpak is a remarkably close copy of LUXXtra, and, if it isn't an exact copy, the doctrine of equivalents applies.... No more does Classpak escape the brand of an infringing device by simply 'approximating a parabola.' I find that Luxxtra is built in accordance with the allowed claims of the patent, and I find that Classpak is an equivalent of Luxxtra, infringing allowed claims 4, 8 and 9 of Lam's patent."

App. I, 165.

The CLASSPAK luminaire performs the same function, reflection of light beams, to achieve the same result, lighting of a room with an HID lamp without objectionable glare.4 In concentrating on its parabolic surface, J-M evidently argues that reflecting light beams in a parallel manner is a "different way" to achieve the same result. But we consider the "way" to be the downward placement of a lamp in an upwardly opening bowl designed to reflect the beams at an angle close to the horizontal so as to cover a wide area of the ceiling. A slight deviation is not a "different way." Our conclusion might not be the same if J-M's deviation-a surface that "approximates a parabola" rather than an ellipse-served a function not met by the patent, but none has been shown. The reflector surface serves only to get the light beams up and out of the bowl at a low angle with no more than a single reflection. Thomas Lemons testified that while an elliptical surface is best because it brings the beams out to a single focal point just above the lip, a parabola or a special curve could approximate the results achieved with an ellipse. App. II, 307-308; VI, 823. J-M would have us read Lam's claim narrowly so that a surface that was parabolic or "special" would not be considered infringing, but this we decline to do.5 The trial court observed that the CLASSPAK and LUXXtra surfaces are so similar that "a sharp pencil can hardly differentiate between the lines of the two bowls," App. I, 177, and we see no utility to a parabolic surface not met by an elliptical surface. The record supports the trial court's finding that in this context a parabolic surface is "essentially equivalent" to an "elliptical surface," and we therefore affirm the trial court's conclusion that J-M's CLASSPAK fixture infringed the Lam patent.

J-M contends that the file wrapper history of the Lam patent shows Lam explicitly limited its claim to a luminaire with an elliptical surface in order to overcome the patent examiner's prior art objections. The test for "file wrapper estoppel" is as follows:

"The law is settled that the extent of an invention is to be determined by the patent claims, together with the 'file wrapper' history in the Patent Office, and a claim which has been narrowed for the purpose of obtaining a patent cannot be expanded to include that which was eliminated.... The application of this doctrine, however, is limited to changes made to overcome rejections by the Patent Office for the reason that the claim was anticipated in prior art similarities. It is not applicable where the patentee encountered difficulties in the Patent Office because of particular wording and indefiniteness of the claims."

Eimco Corp. v. Peterson Filters and Eng'r Co., 406 F.2d 431, 438 (10th Cir. 1968) (citations omitted), cert. denied, 395 U.S. 963, 89 S. Ct. 2105, 23 L. Ed. 2d 749 (1969); accord, Central Soya Co. v. Geo. A. Hormel & Co., 645 F.2d 847, 851 (10th Cir. 1981); Sears, Roebuck & Co. v. Jones, 308 F.2d 705, 708 (10th Cir. 1962), cert. denied, 371 U.S. 952, 83 S. Ct. 509, 9 L. Ed. 2d 501 (1963).

The file wrapper history of the Lam patent began on November 14, 1973, with the filing of the patent application. The basic claim, which after amendment emerged as claim four of the issued patent, read as follows:

"A luminaire for a high intensity lighting device adapted to be suspended from the ceiling, said luminaire comprising: an upwardly opening reflective bowl, terminating in a substantially circular outer lip, the diameter of said reflective bowl being greater on the outer lip than at the lower end; a base integrally disposed upon the lower end of said bowl; means to hold said device substantially axially within said bowl and arranged so that the radiation which is emitted substantially below the horizontal plane will be reflected over said outer rim; means to hold said bowl in a fixed spatial relationship, relative to said lighting device."

App. VII, 996. That claim did not attribute any particular optical shape to the surface of the reflector bowl. Lam added several claims dependent upon the basic luminaire claim; these included using flutes in the reflective bowl and placing a refractor at the base of the bowl. Id. Two original dependent claims added the provisos that the luminaire emit substantially all the light after no more than one reflection and that the optical shape be an elliptical surface of revolution. By a notice dated June 15, 1974, the patent examiner rejected all claims in the initial application on the grounds that they were "vague and indefinite" and that "no patentable significance is attached to the recitation of the 'shape.' " Id. at 1003. Additionally, the examiner stated that " 'the optical shape' ... does not indicate what shape is referred to." Id.

Lam amended its application on October 9, 1974, retaining the general references to "optical shape" and merely particularizing inner references such as "said bowl" and "said lip." Id. at 1007-08. On April 8, 1975, the claims examiner allowed the other independent claim, for a bowl with flutes, but disallowed the remaining claims. As to the basic claim for an HID luminaire, using a bowl not necessarily fluted, the examiner rejected it as unpatentable in light of the prior Einstein, Schaffer, and Ainsworth patents. At the same time the examiner said he would allow the dependent claims if Lam redrafted them to include the basic claim's limitations. As to prior art, the examiner found it teaches use of a "reflective, upwardly opening bowl ... shaped in a manner to prevent radiation ... from being directly emitted substantially below the horizontal plane of the bowl's upper lip ...." App. VII, 1021. We believe this finding means only that prior art teaches use of an indirect fixture to avoid direct emission below the horizontal.

On July 25, 1975, Lam's patent attorney submitted a second amendment, which added further description to the reflective bowl as follows: "the optical shape of said bowl being an elliptical surface of revolution, one focus of which is disposed at said light source and the second focus being disposed at a point over said outer lip of said bowl." Id. at 1027. Once this change was made, the patent examiner approved all claims. The general claim was renumbered as claim four of the patent as issued and is the sole independent claim Lam asserts was infringed by J-M.

Based upon our review of the file wrapper history, we are satisfied that Lam added the reference to "elliptical surface" to overcome an objection of vagueness or uncertainty, not prior art. Nowhere in the patent examiner's rejection of the general luminaire claim does he suggest that he rejected the claim because of any relevant prior art relating to the shape of the reflector surface. Therefore, Lam is not estopped from asserting that a parabolic or similar reflector surface infringes its patent.

The patent laws permit the court to award damages up to three times the amount found necessary to compensate for the infringement. 35 U.S.C. § 284. Courts have limited the increase to instances in which the infringement was willful, and even then it is committed to the trial court's discretion. E.g., Milgo Elec. Corp. v. United Business Communications, Inc., 623 F.2d 645, 665 (10th Cir.), cert. denied, 449 U.S. 1066, 101 S. Ct. 794, 66 L. Ed. 2d 610 (1980). An infringer who reasonably doubted that the patent was valid has not willfully infringed the patent. E.g., Eltra Corp. v. Basic Inc., 599 F.2d 745, 757 (6th Cir.), cert. denied, 444 U.S. 942, 100 S. Ct. 297, 62 L. Ed. 2d 308 (1979). The trial court concluded that J-M willfully infringed Lam's patent, a conclusion we will not disturb unless we find that court abused its discretion.

We have already described the haste with which J-M marketed a fixture "similar" to LUXXtra. After spending several years trying to develop a satisfactory indirect HID fixture, J-M's engineer, working with the Florida blueprint and the Lam brochure, designed the CLASSPAK fixture, virtually identical to Lam's, in a single day.

J-M personnel must have known at that time J-M was about to infringe on another's patent. The Lam brochure bore two references to patent protection. The brochure said nothing about an elliptical surface to the reflector. In designing the CLASSPAK fixture, J-M did not attempt to avoid Lam's patent claims: J-M's marketing manager for commercial lighting, Clarence C. Keller, testified that "no particular optical design went into drawing those curves." App. III, 409.

According to Keller, J-M "decided we would do the things that were necessary to get the job, or at least compete for the job ...." App. IV, 566. J-M competed hard to make the sale in Florida, taking the CLASSPAK there for a personal demonstration and lowering its price to suppliers several times. The picture is one of haste to compete with Lam. In its defense J-M says it would never have filled an order for the units if it had believed there was an infringement. App. VI, 914-15. As it turns out, Lam did claim infringement before J-M made the Florida sale. J-M then found that the patent read, "an elliptical surface of revolution," determined that its CLASSPAK fixture had a parabolic surface, and obtained an opinion of no infringement from both in-house and outside counsel. Presented to us then is a case of J-M copying Lam's design,6 but before making the first sale, investigating the claims of the Lam patent, and arriving at a colorable assertion of no infringement.

In determining whether an infringer acted in such bad faith as to merit an increase in the damages awarded against him, we consider three factors of special significance. The first is whether the infringer deliberately copied the ideas or design of another. Obviously, this would suggest intentional infringement. The second is whether the infringer, when he knew or should have known of the other's patent protection, reasonably investigated the scope of the patent and formed a good faith belief that the patent was invalid or that it was not infringed. Here, J-M's consultations with patent counsel are especially pertinent; but a "no infringement" letter from counsel is insufficient by itself. See Celebrity, Inc. v. A & B Instrument Co., 573 F.2d 11, 13-14 (10th Cir.), cert. denied, 439 U.S. 824, 99 S. Ct. 94, 58 L. Ed. 2d 117 (1978); Hartford Nat'l Bank and Trust Co. v. E. F. Drew & Co., 188 F. Supp. 353, 361 n.41 (D.Del.1960), aff'd per curiam, 290 F.2d 589 (3d Cir.), cert. denied, 368 U.S. 825, 82 S. Ct. 45, 7 L. Ed. 2d 29 (1961). The timing of the inquiries to counsel, before or after manufacture and sale of the infringing device, and the strength of counsel's arguments are important. J-M's timing of its inquiries was bad, and we are not impressed with the reasons counsel gave: invalidity because of the prior art, and no infringement because the CLASSPAK fixture has a parabolic surface.

A third factor is the infringer's behavior as a party to the litigation. Strictly speaking, although an infringer's resort to meritless defenses and dilatory tactics is not evidence of willful infringement, it suggests bad faith. See, e.g., Jenn-Air Corp. v. Penn Ventilator Co., 394 F. Supp. 665, 676 (E.D.Pa.1975). Some evidence in the record suggests that J-M litigated in good faith by diligently responding to Lam's allegations, furnishing Lam a CLASSPAK fixture, generally responding to discovery requests, and raising defenses that were at least plausible. But evidence also supports the trial court's finding that J-M "stonewalled" when Lam first mentioned infringement and threatened Lam with litigation expenses of at least $150,000. While we hesitate to attribute too much to statements made during heated discussions, like the trial court we find offensive such threats by a large company to a small one.

Reviewing the factors important to whether J-M's conduct merits an increase in damages, we believe the record supports the finding that J-M's copying was intentional.7 J-M did not even first obtain a copy of Lam's patent and attempt to deviate meaningfully from the patent claims. Since J-M had only the blueprint and the Lam brochure, for all J-M knew, Lam's surface may have "approximated a parabola." Later, when Lam claimed infringement, J-M was able to develop plausible arguments for no infringement. But J-M did not develop those arguments until confronted by Lam. A company intending to avoid patent infringement would have investigated Lam's patent claims earlier. The record contains sufficient evidence to support the trial court's finding of intentional infringement, and we find no abuse of the trial court's discretion.

The court may award the prevailing party attorneys' fees "in exceptional cases." 35 U.S.C. § 285. As with treble damages, the award of attorneys' fees is subject to the trial court's discretion, to be overturned only for abuse of discretion. E.g., Milgo Elec. Corp. v. United Business Communications, Inc., 623 F.2d 645, 667 (10th Cir.), cert. denied, 449 U.S. 1066, 101 S. Ct. 794, 66 L. Ed. 2d 610 (1980). Attorneys' fees are awarded to compensate the prevailing party because the losing party's misconduct was so unfair and reckless as to make it unconscionable for the prevailing party to sustain the expense of counsel. Q-Panel Co. v. Newfield, 482 F.2d 210, 211 (10th Cir. 1973). Courts generally award attorneys' fees only when they have increased damages. In this Circuit we recognize as bases for an award of attorneys' fees either willful infringement or the assertion of sham or frivolous defenses that increase significantly the patent holder's legal expenses through unduly protracted litigation. Id.

We have upheld the trial court's finding that J-M willfully infringed Lam's patent. Therefore, we find no abuse of the trial court's discretion in its award of attorneys' fees.

THE INJUNCTION AGAINST SALE OF THE CLASSPAK FIXTURE

J-M argues that the trial court's injunction is so broad that it improperly prohibits J-M from using its trademark or trade name "CLASSPAK." We think the trial court did not intend that broad a restriction. The injunction prohibited J-M "from making, using or selling the CLASSPAK fixture."8 The injunction was aimed at the J-M luminaire that was modeled on the Lam LUXXtra fixture and properly prohibited J-M from infringing the Lam claims. J-M should be permitted to use the name CLASSPAK on a product altered to avoid infringement. See Plymouth Rubber Co. v. Minnesota Mining and Mfg. Co., 321 F.2d 151, 158 (1st Cir. 1963). Since we construe the trial court's use of "CLASSPAK" as merely identifying the specific infringing fixture, and not as prohibiting J-M's continued use of that name on products that do not infringe Lam's patent, we see no need to modify the trial court's judgment.

AFFIRMED.