Plaintiff, the owner of a combination patent relating to a rotary seal assembly (Hatch '843),1 appeals from a judgment holding the patent invalid as obvious. Defendant has cross-appealed, claiming that plaintiff's failure, during the prosecution of the patent application, to cite two prior patents described in its own internal memorandum as "the patents of most concern," and its use of a misleading affidavit to avoid another prior art reference, made this an exceptional case entitling it to an award of fees. Defendant has raised other issues which we do not reach because we affirm on the issue of validity.

Plaintiff does not challenge any of the findings of fact, arguing instead that the district court misapplied the law, particularly by minimizing the importance of the presumption of patent validity and of the commercial success of the patented seal. Our statement of the case is based largely on those findings and on the text of the patent.

Rotary mechanical seals are used in equipment which contains either a stationary housing through which a rotating shaft extends, or a stationary shaft about which a roller turns. In either situation, the relatively moving parts need lubrication; the rotary seal provides a leak-proof barrier which keeps the lubricant within the machine and dirt and grime out.

Years ago oil seals were nothing more than a lubricating material called a "packing" crammed in a "stuffing box" which surrounded a shaft. Such packing needed frequent replacement, tended to leak, and was generally unsatisfactory. By 1961, when plaintiff's Hatch device was developed, the rotary seals in common use were more complex. Typically they included four features worthy of separate mention.

(1) The primary seal enclosing the lubricant was affected by engaging a sealing ring (which might rotate with a shaft) with a mating ring or mating surface (which might be affixed to the stationary housing). (2) The sealing ring, and sometimes the mating ring as well, was provided with a spring means which applied pressure in an axial direction i. e., parallel to the rotating shaft to maintain the sealing ring in "end face" running engagement with the mating surface. (3) An interlocking means transmitted driving torque to the sealing ring. And finally (4), a secondary seal protected the spring and the interlock from outside dirt. From even this abbreviated description it is evident that such seals required a multiplicity of parts.2

In 1959, Caterpillar Tractor Company began to use a much simpler and more durable seal called the Duo Cone. The Duo Cone used neither a spring means to apply pressure to the sealing ring in an axial direction (or to the mating ring or surface) nor any mechanical interlock to attach the ring to the shaft. Instead, Caterpillar shaped the back of the sealing ring (and also the back of the mating ring) more or less like an irregular, horizontal "V," to receive a doughnut shaped rubber ring (a "torus") in such a way that pressure against the rubber ring would simultaneously apply axial force to push the rings together and also radial force affixing the sealing ring to the shaft and the mating ring to the housing. The separate parts which previously formed an axial spring, a radial interlock, and a secondary seal, were all replaced by a simple rubber torus which, when compressed, continuously applied the requisite force to hold the ring in place.

The Hatch seal, invented in 1961, was also a simple combination of a metal ring and a rubber ring. Unlike the Duo Cone seal, however, the rear of the metal ring contained a right angled seat and a cross section of the rubber ring was shaped like a parallelogram rather than a circle. Compression of the rubber ring, upon installation, pressed the metal ring in an axial direction against its opposing radial face to form the primary seal and also provided the torque to cause rotation. The invention covered both a single face seal including one metal ring and one rubber ring and a double face seal consisting of two identical seals mounted with their metal rings face to face.3

The specifications in the Hatch patent primarily emphasize the advantages of using the simple rubber ring to perform three entirely different functions previously performed by three different elements in the rotary seals then in common use.4 Essentially the same advantages were achieved by the Duo Cone seal which was described in the patents owned by Caterpillar Tractor Company.5

The district court's conclusion that the Hatch invention was obvious was reached after making an accurate and detailed determination of the scope and content of the prior art, considering the differences between that art and the Hatch claims, and concluding that the level of skill in the rotary seal art was "quite high" at the time of the Hatch invention since both plaintiff and defendant were then aware of the advantages of the Duo Cone assembly.

The district court found that the basic elements of the Hatch device were disclosed by the prior art. The concept of a simple juxtaposition of a rubber ring and a metal sealing ring with compression upon installation providing both axial and radial force and also performing the secondary sealing function was revealed by Duo Cone. Combinations of a metal ring containing an angled rear seat with a rubber washer shaped like the one used by Hatch had been used on more than one occasion before.6 Indeed the rubber ring with a cross section resembling a parallelogram a frusto-conical shape had even become known as a "Belleville washer." Based on the prior use of such a washer and a right angled rear seat on a metal ring, together with the simple concept illustrated by Duo Cone, the district judge concluded that the Hatch patent lacked the inventiveness required by § 103; he said that the determination of obviousness was "based on clear and convincing evidence."

The Hatch device enjoyed prompt and widespread commercial success. The district court discounted the significance of this success, however, because Caterpillar had refused to grant a license to manufacture the Duo Cone seal, and apparently had limited its use to Caterpillar's own heavy equipment. The district court drew the inference that the demand for the Hatch device would not have been so high if Caterpillar had commercially marketed Duo Cone. Plaintiff criticizes the district court's analysis as inadequately supported by the record and as contradicted by evidence identifying superior features of the Hatch device, particularly its efficient installation.

The district court refused to find plaintiff guilty of fraud on the Patent Office either because of the failure to cite pertinent prior art or because of the somewhat misleading character of an exhibit attached to an affidavit submitted by one of the inventors.

A patent issued in 1958 (Payne '226) taught the use of a mating ring supported upon a frusto-conical member.7 This was the first patent discussed in detail in the district court's analysis of the prior art.8 It was one of two patents which one of the inventors of Hatch described, before the Hatch application was filed, as "the patents of most concern." Unquestionably it was relevant prior art, but it was not cited to the Patent Office. Without condoning plaintiff's omission, and after stating that it considered this a "close case," the court declined to find fraud because another somewhat similar Payne patent had been cited and the omission of Payne '226 appeared to be an unintentional isolated act.

Prior to issuance, the Hatch claims had been rejected on the basis of an earlier patent, Krug '274. Krug had disclosed a seal in which compression of "a centrally apertured connecting disc" composed of synthetic rubber material applied axially directed force against a metal sealing ring.9 In order to avoid the Krug reference, one of the inventors of Hatch prepared an affidavit describing tests comparing the flexibility of the Krug disc and the Hatch frusto-conical washer. An exhibit to that affidavit contained a pictorial representation of the Krug installation which made it appear that the unconfined portion of the rubber disc was significantly greater (and presumably, therefore, more flexible) than it actually was in the test model used by the affiant. However, the actual dimensions of the test device were accurately set forth on the exhibit and a careful examination of the diagram would have revealed its distortion. For this reason, together with a reluctance to infer that the Patent Examiner actually relied on tests of only one device when the Krug patent made it clear that variations in construction would produce variations in actual pressure, the district court refused to find that the affidavit was fraudulent.

Defendant's accused device was first marketed in 1970. Apparently initial attempts to make a seal using a Duo Cone metal ring and a rubber ring shaped like a doughnut with an axial lip on it failed. Later defendant copied plaintiff's metal ring with a right-angled rear seat and substituted a washer with a cross sectional appearance only slightly different from a parallelogram. Defendant's theory of non-infringement, rejected by the district court, was that plaintiff's claims were limited to the precise parallelogram shape and, further, that pressure on defendant's washer produced more of a rolling action than spring-like compression. In any event, we do not reach the issue of infringement raised by defendant's cross appeal since we accept the district court's analysis on the obviousness issue.

Plaintiff's principal argument is that "the district court erred as a matter of law in failing to heed properly the mandate of the statute 35 U.S.C. § 28210 that a patent is presumed valid."

There are four steps in plaintiff's argument that, as a matter of law, the statutory presumption requires that the Hatch patent be accepted as valid. First, plaintiff points out that in this jurisdiction the presumption is "not an idle gesture" and that the burden of overcoming that presumption "rests heavily" upon the person asserting invalidity.11 The basis for the presumption according to plaintiff, is the recognition that patent examiners have technical expertise that is entitled to the same deference as determinations by other administrators. Second, plaintiff asserts that the prior art presented to the district court was the same as the prior art considered by the Patent Office, and therefore the presumption is strengthened. Third, plaintiff urges that the presumption is further strengthened by the fact that Patent Office prosecution lasted for over four years and the claims were considered by three different patent examiners. And finally, the commercial success of the Hatch seal makes the presumption even stronger. In sum, in this circuit the statutory presumption is strong, and in this case it is very, very, very strong.

Judge McLaren, on the other hand, stated that notwithstanding the statutory presumption of validity, he had a duty to give the patent close scrutiny and concluded that a clear and convincing showing of obviousness was sufficient to warrant a holding of invalidity.12 We find no error in the court's treatment of the presumption.

We agree, of course, that the presumption is not an idle gesture, either in this circuit or elsewhere. But we are not persuaded that Judge McLaren failed to give the presumption all the weight it deserved.

There are two aspects to the presumption of validity. First, as a matter of procedure, § 282 places the burden of persuasion on the party attacking the validity of the patent. This burden remains upon the alleged infringer throughout the proceeding and is in no sense dependent on the character of the proceedings before the Patent Office or the amount of prior art cited to, or considered by, the Patent Examiner. In its first aspect, the presumption is constant.

There is an additional aspect to the presumption which relates to the deference due to the technical expertise possessed by the Patent Office and not generally possessed by federal judges. Notwithstanding the ex parte character of the normal Patent Office prosecution,13 if it appears that the prior art which is cited to establish invalidity in a judicial proceeding had already been considered by the Patent Office, the burden of overcoming the presumption of validity then rests heavily on the person attacking the patent. He must make a "clear and cogent" showing in order to prevail.

The presumption, however, never becomes so strong that a patent is completely unassailable. A demonstration that the art considered by the Patent Office is the same as the art considered by the court strengthens the presumption in the sense that it provides the justification for the requirement that invalidity be established by clear and convincing evidence, but such a demonstration is not also a bootstrap that lifts the presumption to a still higher level.

The basis for the requirement that invalidity be established by clear and convincing evidence is largely, if not wholly, dissipated when pertinent prior art is shown not to have been considered by the Patent Office.14 For then the Examiner's expertise may have been applied to an incomplete set of data and there can be no certainty that he would have arrived at the same conclusion in the face of the evidence and argument presented to the court. Nor may we safely assume that the Examiner has considered art which is not cited. On the contrary, we have held that the failure to cite pertinent prior art implies that it was overlooked by the Examiner.15

In this case Judge McLaren concluded that defendant had made a clear and convincing showing of invalidity. The presumption never imposes a heavier burden than that. It is surprising to find the plaintiff arguing in this case that an even stronger showing was required in view of the fact that the art considered most relevant by one of the inventors, and also considered highly relevant by the district court namely, Payne '226 was not cited to the Patent Office.

Nor do we attach any significance to the fact that three different examiners rejected prior drafts of plaintiff's claims, or to the not unusual, though distressing, period of time which the Patent Office required to complete its task. A patent application which is so obviously meritorious as to result in an immediate allowance might well command more respect than one that ultimately survived a series of rejections followed by a series of narrowing redrafts of pertinent claims.

We agree with plaintiff that the evidence of commercial success is significant, but not because it enhances the importance of the presumption of validity. Rather, commercial success is persuasive circumstantial evidence bearing directly on the issue of non-obviousness. Commercial success demonstrates that there was a market for the patented device and implies that persons skilled in the art had an economic incentive to make it as soon as they could; the failure to produce a device satisfying a known demand indicates that the inventor's solution was not obvious.16

The validity of the conclusion depends upon the actual availability of the pertinent prior art to the persons seeking a solution to the problem.17 If the critical reference came into existence, or was discovered, only a short time before the patented conception, commercial success may merely demonstrate nonobviousness as compared to the earlier, well known art, rather than when compared to the recent reference. In this case, commercial success of the Hatch seal strongly indicates that it was a nonobvious improvement over the complex devices described in the patent specifications and generally available on the open market in 1961. The question, however, is whether that success is equally persuasive on the question whether Hatch was a nonobvious improvement over Caterpillar's Duo Cone seal. Since applications for the patent covering Duo Cone were not made until 1959, and since there appears to have been no effort to market Duo Cone generally, the commercial success of the Hatch seal does not necessarily demonstrate its nonobviousness in light of Duo Cone.

The fact that the Duo Cone seal was not sold generally does not undermine the significance of its contribution to the art. An especially attractive feature of the Hatch seal was its ease of installation, a factor of less importance to Caterpillar than to other users because Caterpillar had such an excellent service organization. The development of a seal which would generally be salable was also of less importance to Caterpillar because its own equipment provided an adequate demand for Duo Cone to justify the capital expenditures necessary to produce it. Accordingly, the fact Hatch was more useful and more salable than Duo Cone does not necessarily imply that it was not obvious in the light of Duo Cone.18

One feature of the Hatch device which no doubt contributed to its commercial success was the "strip-like closure means" which holds the two halves of the double face seal together, keeping the precisely machined faces free from grit prior to installation.19 This feature contributed to the ease of installation of the Hatch seal. However, this band was described in the patent as well known; if one assumes that a seal which could otherwise be installed as a unit was not patentable, the addition of the closure strip would appear obvious.

We also agree with plaintiff that the issue of obviousness raises a question of law and, therefore, we are free to disagree with the district court even if we cannot brand his conclusion as clearly erroneous. Nevertheless, as we read the transcript of the testimony in this case, we are impressed with the importance of having live witnesses, subject to cross-examination, explain the operation of physical exhibits in a way which enables a district judge to understand what is before him and to interrupt with proper questions when he does not understand. Our study of the testimony, the documents, and the physical exhibits has, we believe, led us to a fair understanding, at least of the Hatch device, the accused device, the Duo Cone seal, and the use of the Belleville washer in the prior art. But our understanding of the written record, even when aided by briefs and oral argument, is comparable to that of a student who has taken a correspondence course instead of attending classes on a daily basis with a laboratory available for experiment when needed. The trial judge really is in a better position to evaluate the obviousness issue than we are. Therefore, although we frankly acknowledge respect for plaintiff's advance in the art, plaintiff has not persuaded us that the district judge made an error of law. In this court it is appellant's burden to persuade us that an intelligent district judge, who has demonstrated a thorough understanding of the relevant art, committed an error requiring reversal. This plaintiff has not done. We accept the district judge's conclusion that, after the significant breakthrough achieved by the Duo Cone seal, it was obvious for a person skilled in the art to experiment with a configuration employing a Belleville washer and a metal ring with a right-angled seat opposed to another right angle to compress the rubber ring upon installation of the seal and, therefore, that the Hatch device does not disclose the degree of inventiveness required by § 103.20

For similar reasons, and with comparable misgivings, we accept Judge McLaren's appraisal of the defendant's argument that it should be awarded fees because plaintiff was guilty of fraud, or at least inequitable conduct, in the Patent Office prosecution.

Certainly Payne '226 should have been cited. We are not persuaded that it was an irrelevant reference because it described a Belleville washer made of Teflon rather than rubber.21 Nor do we find any merit in the suggestion that the examiner was presumptively aware of Payne '226 since other Payne patents were cited.22 Nor do we condone what we assume was a lack of diligence in failing to cite prior art which one of the inventors recognized as highly pertinent before the application was filed. Nevertheless, the district judge could properly conclude that there was insufficient evidence of scienter to support a finding of fraud.

We also share Judge McLaren's appraisal of the exhibit to the Morely affidavit. Surely, if the Patent Office proceedings had been of an adversary character, the discrepancy between an actual groove width of .118 inches,23 and a schematic representation, not drawn to scale, that appears to portray a groove about five times as wide, would have been made apparent to the examiner. On the other hand, if the examiner studied the affidavit and the exhibit with sufficient care really to appreciate its relevance to the Krug reference, he must also have recognized the discrepancy between the actual and the apparent width of the groove, and, further, understood its relationship to flexibility of the washer. We think this exhibit illustrates the hazards inherent in ex parte technical presentations, but we accept Judge McLaren's appraisal of its significance in the context of the entire case.

Based on our review of the record and the briefs, we conclude that the case was well presented in the district court by both sides, that the trial judge correctly found the facts and accurately applied the law, and that his ultimate conclusions should not be disturbed. In all respects the judgment is

Affirmed.