MARKEY, Chief Judge.

On remand from the Supreme Court. Having considered the record, appealed judgment, district court's findings and application of law, arguments in this court and the Supreme Court, and invited briefs of the parties and amicus American Intellectual Property Law Association, we respond to the remand and again affirm the judgment on the 35 U.S.C. § 102(g) defense and reverse the judgment on the 35 U.S.C. § 103 defense.

The Remand

In remanding, Dennison Manufacturing Co. v. Panduit Corp., ___ U.S. ___, 106 S.Ct. 1578, 89 L.Ed.2d 817, 229 USPQ 478 (1986), vacating, 774 F.2d 1082, 227 USPQ 337 (Fed. Cir. 1985), the Supreme Court said:

Petitioner contends that the Federal Circuit ignored Federal Rule of Civil Procedure 52(a) in substituting its view of factual issues for that of the District Court. In particular, petitioner complains of the rejection of the District Court's determination of what the prior art revealed and its findings that the differences identified between respondent's patents and the prior art were obvious.

Petitioner's claims are not insubstantial. As this Court observed in Graham v. John Deere Co., 383 U.S. 1, 17-18 [ 86 S.Ct. 684, 693-94, 15 L.Ed.2d 545, 556-57], 148 USPQ 459, 467 (1966):

"While the ultimate question of patent validity is one of law, . . . the § 103 condition [that is, nonobviousness] . . . lends itself to several basic factual inquiries. Under § 103, the scope and content of the prior art are to be determined; differences between the prior art and the claims at issue are to be ascertained; and the level of ordinary skill in the pertinent art resolved. Against this background, the obviousness or nonobviousness of the subject matter is determined. Such secondary considerations as commercial success, long felt but unresolved needs, failure of others, etc., might be utilized to give light to the circumstances surrounding the origin of the subject matter sought to be patented. As indicia of obviousness or nonobviousness, these inquiries may have relevancy."

This description of the obviousness inquiry makes it clear that whether or not the ultimate question of obviousness is a question of fact subject to Rule 52(a), the subsidiary determinations of the District Court, at the least, ought to be subject to the Rule.

The Federal Circuit, however, did not mention Rule 52(a), did not explicitly apply the clearly-erroneous standard to any of the District Court's findings on obviousness and did not explain why, if it was of that view, Rule 52(a) had no applicability to this issue. We therefore lack an adequate explanation of the basis for the Court of Appeals' judgment: most importantly, we lack the benefit of the Federal Circuit's informed opinion on the complex issue of the degree to which the obviousness determination is one of fact. In the absence of an opinion clearly setting forth the views of the Court of Appeals on these matters, we are not prepared to give plenary consideration to petitioner's claim that the decision below cannot be squared with Rule 52(a). Instead, we grant the petition for certiorari, vacate the judgment and remand the case to the Court of Appeals for further consideration in light of Rule 52(a).

106 S.Ct. at 1579, 89 L.Ed.2d at 821, 229 USPQ at 479.

Opinion on Remand

Reversal of the district court's judgment on the § 103 defense was and is compelled because of legal error, because a first set of clearly correct and unchallenged findings requires that the conclusion of nonobviousness, 35 U.S.C. § 282, must remain untrammeled and because a second set of findings, some nonprobative, some clearly erroneous, is unable to support, intrinsically and in light of the first set, the district court's conclusion of obviousness.

The Background section of our earlier opinion noted error in the second set of findings but did not label it clearly erroneous, Rule 52(a), and did not expressly indicate that the noted error was not the basis for reversal set forth in the Opinion section, thus raising a question respecting the basis for our judgment. That circumstance, regrettable because it has delayed a just end to this litigation, is here rectified.

On review of a judgment based on a conclusion of obviousness under 35 U.S.C. § 103 , this court must consider not only whether there is legal error, but also whether the underlying findings are either nonprobative, or clearly erroneous, or both. If, unlike the present case, a district court had not made findings necessary to resolution of the § 103 question, and legal error were present, an appellate court would vacate in view of that legal error and remand for the district court to make the missing findings. Loctite Corp. v. Ultraseal Ltd., 781 F.2d 861, 872-75, 228 USPQ 90, 97-99 (Fed. Cir. 1985). If unassailable findings were made but could not support the appealed judgment under a proper application of law, an appellate court may vacate or reverse but not make its own findings. Icicle Seafoods, Inc. v. Worthington, ___ U.S. ___, ___, 106 S.Ct. 1527, 1530, 89 L.Ed.2d 739, 743 (1986).

If findings necessary to support a legal conclusion are clearly erroneous, the conclusion cannot stand. If those facts were such as to permit only one of two possibilities, or if the burden of proving those necessary facts had been on appellee, reversal would not mean the appellate court found facts in defiance of Rule 52(a). If, unlike the present case, appellee did not bear the burden below, a remand-requiring fact-finding function might remain.

To obtain reversal without remand, an appellant-patentee must convince this court that the patent challenger failed at trial to carry its statutory burden, 35 U.S.C. § 282, of proving by clear and convincing evidence sufficient facts to support the obviousness conclusion. One way to do that is to show that a proper application of the law to unassailable findings compels reversal. Another is to show that the findings on which the obviousness conclusion rested are clearly erroneous and that nothing of record warrants a further exercise of the fact-finding function or indicates any possibility that the appealed judgment might be sustained by such exercise. Because the record here establishes that appellant has made both showings, reversal is required and there is no basis for remanding the case for new findings by the district court.

In this opinion:

Part I supplies our opinion on the remand inquiry ("the degree to which the obviousness determination is one of fact").

Part II treats Rule 52(a)'s applicability to § 103.

Part III examines the statutory burden of proof.

Part IV describes inventor Caveney's inventions and the effect of his claims.

Part V discusses our earlier opinion and the district court's approach to resolution of the § 103 issue in this case.

Part VI identifies the patent claims on appeal.

Part VII sets forth the district court's erroneous claim interpretations.

Part VIII discusses the findings in general.

Part IX reviews correct findings that compel a conclusion of nonobviousness.

Part X reviews nonprobative and clearly erroneous findings that underlay the appealed judgment.

The Appendix comments on parts of the Petition for Certiorari and Reply.

Part I The Obvious-Nonobvious Question is One of Law (a) Nature

A § 103 determination involves fact and law. There may be these facts: what a prior art patent as a whole discloses; what it in fact disclosed to workers in the art; what differences exist between the entire prior art, or a whole prior art structure, and the whole claimed invention; what the differences enabled the claimed subject matter as a whole to achieve; that others for years sought and failed to arrive at the claimed invention; that one of those others copied it; that the invention met on its merits with outstanding commercial success.

With the involved facts determined, the decisionmaker confronts a ghost, i.e., "a person having ordinary skill in the art," not unlike the "reasonable man" and other ghosts in the law. To reach a proper conclusion under § 103, the decisionmaker must step backward in time and into the shoes worn by that "person" when the invention was unknown and just before it was made. In light of all the evidence, the decisionmaker must then determine whether the patent challenger has convincingly established, 35 U.S.C. § 282, that the claimed invention as a whole would have been obvious at that time to that person. 35 U.S.C. § 103. The answer to that question partakes more of the nature of law than of fact, for it is an ultimate conclusion based on a foundation formed of all the probative facts. If itself a fact, it would be part of its own foundation.

When findings on the foundational facts were not clearly erroneous this court must then determine whether the § 103 answer of the district court is supportable by those findings. As with other statutes, that determination engages this court in an exercise legal in nature.

(b) Precedent

The Supreme Court, the regional Circuit Courts of Appeals, and this court have consistently treated the question as one of law.

Graham's statement, "the ultimate question of patent validity is one of law", has been widely interpreted as meaning that one answering the § 103 question is drawing a legal conclusion. And rightly so, because the validity issue in Graham turned on that answer and because of what the Court did in Graham. It disagreed with conclusions reached below, did not remand, described no finding as "clearly erroneous," and did not mention Rule 52(a).

The Court in Graham determined that the facts were (in No. 11) and were not (in Nos. 37 and 43) sufficient to support the lower courts' conclusions on the § 103 question. Similarly, unassailed and unassailable findings here compel our legal conclusion on that question.

(c) Scholars

A survey of writings on "the vexing nature of the distinction between questions of fact and questions of law," Pullman-Standard v. Swint, 456 U.S. 273, 288, 102 S.Ct. 1781, 1790, 72 L.Ed.2d 66, 79 (1982), would unduly lengthen this opinion. Scholars who have dealt with the obviousness question unanimously view it as one of law.

(d) Effect

Like all legal conclusions, that under § 103 rests on a factual evidentiary foundation. As said in Graham, 383 U.S. at 17, 86 S.Ct. at 693, 15 L.Ed.2d at 556, 148 USPQ at 467, the question is determined "against" the "background" of answers to factual inquiries, — a description of how legal questions are normally determined.

One effect of considering the § 103 question one of law in this court is to facilitate a consistent application of that statute in the courts and in the Patent and Trademark Office (PTO).

To contribute to consistency in construing § 103, this court has affirmed judgments while noting noncontrolling misstatements of law and cautioning counsel that judgments are appealed, not opinion language. Controlling misstatements of the legal criteria applicable to § 103 determinations, present in the district court's opinion language in this case, confirm the certain presence of reversible legal error. See 774 F.2d at 1091-1100, 227 USPQ at 342-49.

(e) Decisional Process

The decisional process en route to a § 103 conclusion involves more than answers to the fact inquiries in Graham. It also involves: (1) legal determinations; and (2) legal standards for fact-finding.

(1) Legal Determinations

Analysis begins with a key legal question — what is the invention claimed? Courts are required to view the claimed invention as a whole. 35 U.S.C. § 103. Claim interpretation, in light of the specification, claim language, other claims, and prosecution history, is a matter of law and will normally control the remainder of the decisional process. The district court's claim interpretation in this case constituted compound legal error not subject to Rule 52(a). See Part VII.

Another key preliminary legal inquiry is — what is the prior art? Before answering Graham's "content" inquiry, it must be known whether a patent or publication is in the prior art under 35 U.S.C. § 102, — a legal question. In this case, for example, the '869 tie was not prior art and the district court's primary reliance on it was legal error not subject to Rule 52(a). See Part X.

(2) Legal Standards

Clarity in the law requires universal application of the same legal standards to fact-finding functions performed en route to final § 103 conclusions.

Among legal standards for determining scope and content of the prior art, for example, are: a prior patent must be considered in its entirety, i.e., as a whole, including portions that would lead away from the invention in suit, W.L. Gore Associates, Inc. v. Garlock, Inc., 721 F.2d 1540, 1550, 220 USPQ 303, 311 (Fed. Cir. 1983), cert. denied, 469 U.S. 851, 105 S.Ct. 172, 83 L.Ed.2d 107 (1984); elements of separate prior patents cannot be combined when there is no suggestion of such combination anywhere in those patents, ACS Hospital Systems, Inc. v. Montefiore Hospital, 732 F.2d 1572, 1577, 221 USPQ 929, 933 (Fed. Cir. 1984); and a court should avoid hindsight, W.L. Gore Associates, Inc., 721 F.2d at 1553, 220 USPQ at 313. Failure here to employ established legal standards led to the second set of findings nonprobative and clearly erroneous under Rule 52(a). See Part X.

Conclusion on Part I

A determination that an invention would have been obvious when it was made to one of ordinary skill in the art under § 103 is thus a conclusion of law based on fact. The "degree to which" it is one of fact is solely that degree required to erect a foundation of facts capable of supporting the conclusion, those facts having been found by applying correct legal standards and expressed in findings free from clear error and based on clear and convincing evidence.

A determination that an invention would not have been obvious when it was made to one of ordinary skill in the art is also a conclusion of law. At trial's outset, that conclusion is in place. See § 282 and Part III. The factual questions are those raised by the patent challenger in an effort to replace the existing with a contrary conclusion, and those, if any, raised by the patentee in rebuttal of the challenger's case.

Part II Applicability of Rule 52(a) to the § 103 Issue

Graham, Anderson, Rule 52(a), and the clearly erroneous standard have, before the remand, been consistently applied by this court in affirming judgments that upheld a patent and judgments that struck down a patent. In the case at bar, our earlier opinion having noted that reversal was required by legal error and the district court's correct and controlling findings on objective evidence of nonobviousness, explicit reference to Rule 52(a) was not made.

Rule 52(a) is applicable to all findings on the four inquiries listed in Graham: scope and content of prior art; differences between prior art and claimed invention; level of skill; and objective evidence (secondary considerations). The last may include: commercial success due to the invention; failure of others; long felt need; movement of the skilled in a different direction; skepticism of experts ( see United States v. Adams, 383 U.S. 39, 52, 86 S.Ct. 708, 714, 15 L.Ed.2d 572, 580, 148 USPQ 479, 484 (1966)); copying the invention in preference to the prior art ( see Diamond Rubber Co. v. Consolidated Rubber Tire Co., 220 U.S. 428, 441, 31 S.Ct. 444, 450, 55 L.Ed. 527, 534 (1911)); and other events proved to have actually happened in the real world (hence the description "objective").

Rule 52(a) is thus fully applicable to all findings on which the § 103 issue is decided, but findings must be made on, and the § 103 issue must be decided in light of, all the probative evidence.

Part III The Statutory Burden of Proof

It is neither necessary nor appropriate for a court to declare a patent valid. A trial court is required by Congress, 35 U.S.C. § 282, supra note 1, to say only whether the patent challenger carried its burden of establishing invalidity in the particular case before the court. 774 F.2d at 1096, 227 USPQ at 346. When the burden has not been carried, a court need only so state. When that burden has been carried, the court should declare the patent invalid. The statute makes a patent presumptively valid when issued and imposes the burden of establishing invalidity on the challenger. It is the judiciary's duty to follow statutes that requires a trial court lacking a conviction of obviousness to hold that the challenger's burden was not carried. Thereupon, the patent simply remains valid until another challenger carries the § 282 burden.

The presumption mandated by § 282 is applicable to all of the many bases for challenging a patent's validity. When, as here, the sole challenge is an allegation of obviousness, the presumption is that the invention would not have been obvious.

A determination of anticipation under 35 U.S.C. § 102 may be reached by comparing only the claims and a prior art disclosure submitted by the patent challenger. A determination that a patent challenger has not carried its § 282 burden under § 103 may be reached on consideration of only the evidence submitted by the patent challenger. A determination that a patent challenger has carried its burden of establishing obviousness under § 103, however, requires full consideration of any objective evidence of nonobviousness presented in rebuttal. That § 103 issue cannot fairly be decided on only one party's part of the evidence (e.g., patent challenger's prior art), while disregarding the compelling impact of the other party's part (e.g., patentee's objective evidence). Nor is there warrant for singling out § 103 as an area in which courts may disregard the probative force of any part of the evidence.

As noted at 774 F.2d at 1099 n. 24, 227 USPQ at 349 n. 24, patent challenger Dennison told the trial court it should disregard the objective evidence of nonobviousness unless it were "undecided" after viewing only the prior art. Dennison mischaracterized this court's opinion in W.L. Gore Associates v. Garlock, Inc., 721 F.2d 1540, 220 USPQ 303 (Fed. Cir. 1983), cert. denied, 469 U.S. 851, 105 S.Ct. 172, 83 L.Ed.2d 107 (1984), as support for that misstatement of law. The district court accepted and applied that erroneous legal analysis. Thus the court decided on only part of the evidence, disregarded the controlling impact of its own unchallenged findings on the objective rebuttal evidence of nonobviousness (evidence that may be, as here, the most probative), and shifted the statutory burden of proof.

One sued for patent infringement has over twenty possible defenses. Respecting the "would have been obvious" defense, misperceptions in pre-1982 court opinions have led this court to say the judicial process requires that a court withhold a conclusion of obviousness until it has fully assessed the impact of any objective evidence of nonobviousness. Only then can a court base its judgment, as it must, on all probative evidence of record.

Consideration of all the evidence may or may not require a conclusion that what may appear obvious to the court would not have been obvious (because for years it was not) to those of ordinary skill when the invention was made. When a district court has admitted evidence and made findings establishing nonobviousness, disregard of their legal impact is a jurisprudentially improper approach gravely damaging the concepts of fair trial and principled decisionmaking under the statutes, §§ 103, 282.

This court's many judgments of patent invalidity show that § 282 and a requirement to consider all evidence do not unduly burden patent challengers. Nonobviousness evidence must be weighed along with closeness of the prior art and other evidentiary factors such as nexus with the claimed invention; see for example the discussion in cases in which this court has affirmed a district court's assignment of little weight to objective evidence that was insufficient for varying reasons.

The district court here recognized that the prior art was not close, expressly faulting Panduit for relying on that fact. See supra note 15 and 774 F.2d at 1089-90, 227 USPQ at 341. Evaluated under established legal standards, the prior art here was so remote as not to have suggested anything like the claimed inventions to those working for years and at great expense to design a successful cable tie. As discussed in our earlier opinion, the district court erred in holding to the contrary.

The strong evidence of nonobviousness, and its inescapable refutation of the view that the claimed inventions would have been obvious from the prior art, are what distinguish this case from some others. As the district court said, "This is a case which presents that issue [effect of nonobviousness evidence] as clearly as any I've seen." That evidence, coupled with the remoteness of the prior art, also resolves in this case the "difficulties" seen in Graham, 383 U.S. at 18, 86 S.Ct. at 694, 15 L.Ed.2d at 556, 148 USPQ at 467.

According to the uncontroverted evidence, with the prior art before its researchers for years, Dennison conducted a long, expensive research project and produced its "ladder" tie, a tie entirely different from those of the claims in suit. Then, as the court found, Dennison copied Caveney's inventions. When sued, it resurrected that same art for litigation purposes, saying what it copied from Caveney had been obvious to Dennison all along. Even then, knowing from Caveney's patents where it wanted to go at trial, Dennison could only misconstrue the claims and reconstruct the prior art to make it look like what it copied. Dennison's obviousness defense is clearly refuted in this case by the unrefuted record of its own long and frustrating experience and that of others. Dennison clearly did not carry its § 282 burden of proving that Caveney's claimed inventions would have been obvious under § 103.

Part IV Caveney's Inventions — Effect of His Claims

The drawings may be deceptive in seeming to show merely a simple, small, plastic tie. The contrary, however, is established by the undisputed record, including the testimony of Dennison's own engineer witnesses quoted at 774 F.2d 1096, 227 USPQ 346, e.g., "a good tie is a damn difficult thing to design, extremely difficult". Caveney's particular claimed cable ties constitute important contributions achieved not easily but only after years of effort and investment of a million research dollars, — and never achieved in years of research by Dennison, whose tie designs did not remotely resemble the claimed ties until it copied them.

It was undisputed that the ties of Caveney's '538 patent, for example, are applied to groups of parallel cables by an installing machine, reliably and safely holding together the many cables in critical environments such as jet airliners. Caveney's contribution to the safety of the airline riding public is a social benefit.

Though technology has burgeoned, the patent system is not limited to sophisticated technologies and powerful corporations. Nowhere in the statute or the Constitution is the patent system opened only to those who make complex inventions difficult for judges to understand and foreclosed to those who make less mysterious inventions a judge can understand after hearing, as here, the inventor's explanation of his invention and the engineering principles he employed. The constitutional purpose is to encourage disclosure of patentable contributions to "progress in the useful arts", all the useful arts, not just the esoteric. The statute requires utility, novelty, and nonobviousness, not complexity.

That cable ties look alike to a casual observer who sees only straps, heads, and locking elements is not determinative. What is determinative and unchallenged on this record is that those elements were available to workers in the art for years before Caveney made his inventions (and remain available today). In viewing the prior art, the district court improperly treated all cable ties as virtually interchangeable (after hearing all about Caveney's inventions and then viewing the ease with which creative defense counsel reconstructed the prior art at his easel). The fact is clear and unchallenged on the record, however, that the crucial structural distinctions set forth in the claim limitations gave the claimed inventions the distinction that led the industry and Dennison to prefer Caveney's structure over anything in the prior art.

Dennison submitted no tests and no testimony challenging the superiority of the claimed ties found by the district court. Nor did it submit evidence that any prior cable tie suggested Caveney's crucial claimed structural limitations that accounted for the success of his inventions.

The claims are not mere semantic exercises and the present inventions are not insignificant variations or innovations of a commonplace sort. Graham, 383 U.S. at 16, 86 S.Ct. at 693, 15 L.Ed.2d at 555, 148 USPQ at 466. The district court found that Caveney achieved and disclosed a "preeminent commercial product" long sought without success by others. Thus Caveney trod well the long and failure-strewn path not only to the Patent Office, id. at 19, 86 S.Ct. at 694, 15 L.Ed.2d at 557, 148 USPQ at 467, but to a successful and useful product.

The sole effect of the grant to Caveney of the property right, 35 U.S.C. § 261, to exclude others for a limited time from unauthorized use of his inventions is to require that others avoid the claimed structure newly disclosed to the public in the patent documents. Thus Caveney's claims do not remove existing knowledge from the public domain, for there was no such knowledge until Caveney disclosed it in his patents. Nor do those claims restrict access to materials earlier available, for those materials remain available. Contrary to the misstatement by Dennison to the trial court and to the Supreme Court, see Appendix, the patent law is and has always been clear, certain, and immutable that, during the short life of the patents in suit, Dennison and all workers in the art remain perfectly free to employ and combine straps, heads, and pawls as those elements were before the Caveney inventions. They need only avoid Caveney's own novel combination and structure as claimed. Connell v. Sears, Roebuck Co., 722 F.2d 1542, 1548, 220 USPQ 193, 198 (Fed. Cir. 1983).

Under our economic and patent systems, valuation of the worth of an inventor's contribution is left to the public, not to the judiciary in determining patentability. A judge is nowhere authorized to declare a patent invalid on his or her personal evaluation. Graham's statement, "emphasis on nonobviousness is one of inquiry, not quality," 383 U.S. at 17, 86 S.Ct. at 693, 15 L.Ed.2d at 556, 148 USPQ at 467, can be seen as a recognition that judges are not to decide whether they think a particular invention "ought" to be patented. And that is well. Were it otherwise, a letter to the court would do, and trials, Title 35, and the PTO would be unnecessary.

Though not all inventions achieve public acceptance, many millions in annual sales ($50 million for Panduit alone), with accompanying taxes and jobs, resulted from Caveney's inventions in the hands of assignee Panduit and admitted copier Dennison. It is undisputed that Caveney's patented ties are novel and useful. The district court found that Caveney's '538 tie leads the industry, and the unchallenged record shows the industry centering on Caveney's inventions, which made owner Panduit No. 1 and copier Dennison No. 2 or 3 in the industry. Caveney's inventions meet the constitutional purpose, for they are classic examples of what the Framers intended when they referred to "Progress of . . . useful Arts." U.S. Const. art. I, § 8, cl. 8.

Part V This Court's Earlier Opinion — District Court's Approach

In its earlier Opinion section, this court focused on the law governing obviousness and burden of proof, §§ 103, 282, and on the impact of the district court's uncontestably correct findings compelling a conclusion of nonobviousness. We expand, somewhat, that focus here.

The law must be the same for all patents and types of inventions. A level playing ground for the marketplace of ideas is as necessary for technological innovation as it is for politics and social policy. If the district court's decisional approach in this case were deemed permissible, it would defeat those goals, for it would destroy every patent, frustrate Congress' effort to effectuate its constitutionally granted power and the Framer's intent to promote progress in the useful arts, and disserve the nation's need to encourage research investment. It would also rend asunder the entire body of § 103 law established in this court's precedents since its creation on October 1, 1982.

No effective, uniform, reliable patent system could long survive if the law permitted a decisional approach to § 103 determinations like that here employed by the district court and suggested in Dennison's Petition for Certiorari: (1) interpreting claims by redrafting them to one word; (2) implying that that word describes the "differences"; (3) picking from a prior patent an item describable by that word (in effect finding no differences); (4) focusing on isolated minutiae in a prior art patent while disregarding its scope, i.e., its entire disclosure, and how its disclosed structure works; (5) making no finding of a suggestion (because there was no evidence thereof) that items found separately in prior patents could or should be shaped positioned, related, and combined as in the claim; (6) considering as prior art what was not; (7) considering not the problem solved by the invention (here a successful cable tie), but speculating on a "problem" of how prior devices might be reconstructed to match the claimed structure, with the benefit of hindsight aided by the inventor's engineering testimony about the inventions in suit; (8) determining, years later, that an invention had been obvious all along from the prior art to one of ordinary skill, in the face of a conclusive finding that it had not, and thus would not have, been obvious from that same art to others, including the many highly skilled engineer-cable designers of Panduit and Dennison who had diligently tried and failed for years, at great expense, to design a successful cable tie; (9) disregarding the impact on the § 103 conclusion of the unchallenged finding that others aware of the prior art, including the infringer's cable tie designers, did not find the claimed inventions obvious but were proceeding in directions entirely different from that of the inventor; (10) disregarding the impact on the obviousness conclusion of the unchallenged conclusive evidence that the infringer, with the prior art before it, after years of research (during which, as its own witness said, it was "tantalizingly close," 774 F.2d at 1096, 227 USPQ at 346), and after patenting many of its entirely different "ladder" type ties, did not find the inventions obvious but deemed it necessary to study Caveney's patents and, as the court unequivocally found, "finally came up with" its copy of the inventions; and (11) holding an invention obvious on the basis of "engineering principles" taught the court by the inventor and a subjective view of "common experience" in the face of its own unchallenged finding that the same "principles/experience" did not make that invention obvious to others who, with the prior art before them, went for years in other directions, which others included the highly skilled engineers at Panduit and Dennison who were fully aware of those same principles and who had even more relevant experience in the art.

Virtually all inventions are necessarily combinations of old elements. The notion, therefore, that combination claims can be declared invalid merely upon finding similar elements in separate prior patents would necessarily destroy virtually all patents and cannot be the law under the statute, § 103.

As explained in our earlier opinion and more fully here, see Parts III, IX, the district court's finding that Dennison's effort to carry its § 282 burden with clear and convincing evidence left "room for some difference of opinion," see Part IX, should have ended the obviousness inquiry, and, infringement having been conceded, should have resulted in judgment for Panduit. In holding otherwise, the court gave but lip service to § 282 and effectively transferred the burden of proof fixed by the statute.

Part VI The Claims Identified

The parties stipulated that decision on validity of four "representative" claims would be determinative for all. Trial was conducted on only those four claims. Only three of those claims were considered by this court on appeal.

The district court, in "Supplemental Findings and Conclusions" prepared by Dennison, listed 30 claims as "tried to the court." (See the Appendix for Dennison's presentation to the Supreme Court of claim 26, which was listed but was not among the four "representative" claims tried to the court and was not presented to this court.)

Part VII Erroneous Claim Interpretations

In interpreting the claims, the district court committed fundamental legal error when it analyzed each by a single word description of one part of the claimed tie. In patent law, a word ("teeth"; "hinge"; "ledge") means nothing outside the claim and the description in the specification. A disregard of claim limitations, as here, would render claim examination in the PTO meaningless. If, without basis in the record, courts may so rewrite claims, the entire statutory-regulatory structure that governs the drafting, submission, examination, allowance, and enforceability of claims would crumble.

The court's interpretations were doubly contrary to Congress' statutory scheme. First, the claim must particularly point out and distinctly claim the invention. 35 U.S.C. § 112. Second, the "subject matter" that must have been obvious to deny patentability under § 103 is the entirety of the claimed invention, a concept Congress nailed down with the next statutory phrase "as a whole".

When the prior art is compared with erroneously interpreted claims, findings of differences between the prior art and the claims will necessarily be clearly erroneous. When the sole error lies in claim interpretation, a remand for comparison of the prior art with properly interpreted claims is required. Here, however, the error occurred also in: disregard of clear findings of such legal significance as to have left undisturbed the presumption of nonobviousness that attached when the patents issued; the application under § 103 of erroneous criteria, see Parts V, IX; and reliance on prior art so remote from the properly interpreted claims at issue as to preclude a conclusion of obviousness from that art.

The district court erroneously interpreted claim 24 of the '146 patent as drawn only to multiple "teeth":

I conclude that this [claim limitations] is a description of one element, not three.

The district court erroneously interpreted claim 1 of the '538 patent as drawn only to a "hinge":

And that is the essence of '538 patent, the discrete hinge.

The district court erroneously interpreted Claim 10 of the '869 patent as drawn only to a "ledge":

It seems to me, and I so find, that to increase the compressive strength of the cable tie, and specifically the pawl and hinge of the cable tie, you would set the pawl on a ledge.

The district court improperly dismissed the novel structural claim limitations that defined the disposition, positioning, relationship, and operation of the elements in the claims. The ignored claim limitations, however, define crucial structural elements of the invention as a whole, and clearly distinguish (and bind the inventor to the distinction) the whole claimed device from those in prior patents. The prosecution history and the 15 prior patents cited in the PTO establish that no claim was allowed on the mere presence of multiple "teeth", or of a "hinge", or of a "ledge". The claims, with their present detailed content, resulted from extensive interaction with the Patent Examiner. The district court pointed to nothing in the prior art that suggested any of the claimed constructions.

Interpreted in the light of the specification, claim language, prosecution history, and the other claims, the claims at bar are drawn to entire combinations as set forth at 774 F.2d 1102-04, 227 USPQ 531-52. Each of the numerous limitations of each claim was fully considered in the PTO. If Dennison had avoided in its ties the presence of any one limitation in a claim, it would have avoided infringement of that claim. It was a clear error of law for the district court to have ignored the limitations clearly set forth in the claims.

Part VIII The Findings In General

As above indicated, the district court's opinion indicates that it made two sets of findings. It referred to its reality-based first set (clear, fully supported findings on undisputed concrete objective real-world evidence of nonobviousness) not in terms of its operative legal significance, but only in terms of the uncertainty that set created. See Part IX. It decided the issue on its speculation-based second set (nonprobative and clearly erroneous findings on prior art modified by Dennison and disputed by Panduit at trial). See Part X.

We cannot see why the district court's first set of findings did not require a conclusion that Caveney's inventions, which had for years escaped others who sought them, "would not have been obvious" under § 103; nor why Panduit and Dennison wasted research resources for years if Caveney's inventions were obvious to all throughout those years; nor how the prior art made Caveney's eminently successful inventions obvious to the court in 1984 when it had not made them obvious to skilled engineers (each more skilled than the "ordinary mechanic" referred to in Hotchkiss v. Greenwood, 52 U.S. (11 How.) 261, 13 L.Ed. 683 (1851)) who had been designing unsuccessful or far less successful cable ties for years when Caveney's inventions were made in the 1960's.

The district court did deal with the six most relevant patents (all but one were cited by the Patent Examiner). Not one of those patents, however, disclosed a tie remotely approaching the structure or success of the claimed ties. Not one disclosed or suggested all the claimed structural limitations. Not one disclosed or suggested Caveney's claimed combinations. Not one disclosed teeth, a hinge, or a ledge as claimed. Nor was there evidence or finding that the patents together suggested the claimed combination.

Though the district court said it "found" the combination obvious, the sole support for that "finding" (sic, conclusion) was the presence in separate prior patents of individual elements differing greatly in shape, position, relation, combination, and function. In relying thereon, the court made a fundamental mistake. A holding of invalidity on that basis, as above indicated, is contrary to statute and would defeat the congressional purpose in enacting Title 35.

Indeed, that the elements noted by the court lay about in the prior art available for years to all skilled workers, without, as the court found, suggesting anything like the claimed inventions, is itself evidence of nonobviousness.

The district court did not follow established legal standards (Part I) in finding the content of the prior art. 774 F.2d at 1087-89, 227 USPQ at 339-41. It treated no claim, nor the entire prior art, nor any prior patent "as a whole," but selected bits and pieces from prior patents that might be modified to fit its legally incorrect interpretation of each claim as consisting of one word.

Instead of making express findings on the structural differences as claimed, the district court took the legal steps of erroneously interpreting: the '146 claim as a one-word (teeth) "solution" to a "problem" (looseness) not disclosed in the Emery patent; the '869 claim as a one-word (ledge) "solution" to a "problem" (compression) not disclosed in any prior patent; and the '538 claim as a one-word (hinge) "solution" to a "problem" (rigidity) it saw in the nonprior-art tie of the '869 patent, a tie that did not exist when the invention defined in the '538 claim was made.

The court specifically found, see Part IX, that Caveney's patents disclosed the best ties in the art. It then spoke of "analogous" items, "expected results" from "ordinary steps," "engineering principles," and "common experience." It nowhere reconciled those evaluations with its contrary findings that no one skilled in the art had for years been led to those evaluations by the prior art. The court's treatment emasculated the crucial, critically important structural limitations that distinguish the claimed inventions from the prior art and that were the key to success of Caveney's inventions.

Part IX Findings On the Evidence of Nonobviousness

As recognized by this court and by the Supreme Court in the remand, the district court did articulate much of the correct approach. The court also said "I have done everything I am supposed to do." The court's characterization of its action, and its statement that it knew of the presumption of validity, are not findings and not subject to Rule 52(a).

As noted in our earlier opinion, in any event, the court totally abandoned the approach it had articulated. With admirable candor, the court expressed the uncertainty created by the conflict between its conclusion and its clear findings on the evidence of nonobviousness:

Now I want to return to where I started and say that I think Mr. Caveney and Mr. Moody and Panduit have designed and perfected an excellent and, indeed, preeminent commercial product, and the fact that they have done so contributed greatly to the difficulty of this case. It is not easy to find something obvious in the face of those facts. And I do so with great reluctance and great resistance. And in fact I have been going back and forth on this thing in my mind throughout the trial.

It gives me great comfort to know that I am just the first stop on this trip. Everything I have said here can be analyzed just as well by the Court of Appeals for the Federal Circuit.

All the documents are here, the testimony is there for them to read, and if they come to a different conclusion than I, so be it.

This is a case in which the credibility of the witnesses has really played a very small part. It is primarily, it seems to me, a matter of applying your experience to the undisputed facts. It's a matter of interpretation. What do these things teach? What would they teach a person of ordinary skill in the art?

* * * * *

I think there is room for some difference of opinion here based largely upon those secondary considerations that have given me such trouble in this case: failure by others, copying by Dennison, and unquestioned commercial success of the plaintiff. (Emphasis added.)

In addition, the unchallenged record contains these express statements and findings based on undisputed evidence, not clearly erroneous, and further compelling a conclusion of nonobviousness:

(1) "Obviously, no previous patent showed the combination at issue here. I am not persuaded that this invention is similar to any one piece of prior art."

(2) "[O]ther were going in different ways and did not arrive at the Caveney structure or anything substantially identical to it."

(3) "I am satisfied that in plaintiff's patents we see the best of the art at the times in question. The '146 patent [sic, invention] was better than anything else that then existed and the second two patents [sic, inventions], the '869 and '538 patents, were the best at the time."

(4) "In fact, plaintiff's commercial embodiment of the '538 patent appears to be the leader in the industry."

(5) "[T]he evidence is simply overwhelming that Dennison copied the '538 patent."

(6) "And it was only after the issuance of the '538 patent that [Dennison] finally came up with something that looks very much like the '538 patent [sic, tie]."

(7) "[T]his commercial success, the failure of others, and the copying by Dennison. . . ."

Part X Nonprobative and Clearly Erroneous Findings

The district court nowhere pointed to anything in the prior art that would have suggested the desirability, and thus the obviousness, of making the distinctive structural elements and combinations Caveney invented and claimed. Nor did the court succeed in the difficult task of casting its mind back into that of a person of ordinary skill in the art who had no preknowledge of the crucial structural differences that vitalize Caveney's inventions. That person would not have ignored how the devices disclosed in the prior patents worked, or the rigidity of the pawl in Fein, or the releaseability of the pawls in Litwin and the Japanese patents, or the absence of wedging or an abutment wall or anything but air opposing the teeth in Fein and Litwin, or the absence of anything in Emery to suggest a design the court found so much better than what Emery disclosed.

Because the district court's opinion is an admixture of claim misinterpretation and fact finding without express reference to "differences", it is necessary to glean those portions that may be treated as implied findings on the differences. As above indicated, erroneous claim interpretations led the district court to implied findings that the difference re claim 24 was only multiple teeth in any arrangement, re claim 6 was only a hinge of any type or sort, and re claim 10 was anything that could be called a ledge, but whether a prior patent showed some sort of teeth, or hinge, or ledge, is alone legally insignificant and nonprobative in this case.

Beyond the nonprobative findings, there were these additional "findings", that were clearly erroneous:

The '146 Patent

"The problem presented by Emery was the so called release problem."

No problem at all is presented in the Emery patent. Like all inventors, Emery presented his invention as flawless. Problems with the Emery tie itself were not seen until years after the patent issued, and, as indicated below, those directly concerned (the distributor and manufacturer of the Emery tie) saw the problem and its solution as entirely distinct from that envisioned by the court and entirely distinct from Caveney's invention.

"[T]he text of Emery makes abundantly clear that Emery's intention was to achieve wedging against the abutment wall . . . inside the head. . . ."

That the Emery patent as a whole disclosed to those skilled in the art a pawl that extends outside the head in use is necessarily confirmed by the unchallengeable documentary evidence of the Orban patent, the Bourne patent, and the latter's prosecution history, which together refute defense counsel's argument that Figure 2 was a "draftsman's error."

Two years after the Emery patent issued, the distributor of the Emery patent tie filed application for the Orban patent. That application and patent described the problem with the Emery tie as the cutting off of the chisel edge (one tooth) when, as was common practice, the protruding strap was cut off. That could happen only if the one tooth were outside the head in use as shown in Figure 2. The distributor's "solution", as Dennison's expert testified, was to raise the end wall of the head.

Four years after the Emery patent issued, the manufacturer of the Emery tie further confirmed that the patent as a whole disclosed the pawl outside the head in use. In prosecuting the application filed for the Bourne patent, the manufacturer's attorney told the PTO that the problem with the Emery tie (at even that late date) was "bending of the pawl and tongue [strap] in opposite directions." That could happen only if Emery's pawl was outside the head in use as shown in Figure 2. The manufacturer's "solution" was to keep the pawl inside the head at all times.

Dennison has not mentioned to us the purpose or function of the serrations (teeth) on the curved surface of Emery's pawl. It was able to get its expert witness to say only that one might force the Emery pawl back into the head if one applied enough load to bend the end wall, but then it might fall through the head.

"And in order to achieve the parallel kind of relationship between the pawl and the abutment wall that will be necessary for effective wedging, it follows that the teeth of the pawl must be in a straight line and equidistant from the surface of the abutment wall."

The district court learned that effective wedging required a parallel-straightline-equidistant relationship only from Caveney's engineering testimony, and from the hindsight reconstruction of the prior art by Dennison's witness, not from the Emery patent or from any other prior art patent. Dennison, and others the court found were going in different directions, did not learn this from anything in the prior art and the court cited no prior art basis for its statement. As a finding on the prior art, the statement is clearly erroneous.

The '538 Patent

"Now the '869 patent had its own problems."

"Now, there are numerous examples in the prior art of hinges and discrete hinges of the kind claimed in the '538 patent. The web in Emery is one example. Litwin has a hinge. Fein, if hinging is a matter of degree, has a hinge, and I refer to Column 2, Line 37 of Fein." "A contemporaneous development that is pertinent is the Bourne patent that had a discrete hinge".

I find nothing in the '538 patent, or at least claim 1 thereof, other than a discrete hinge, that is new over the '869 patent.

This and the court's extended ensuing discussion of the '538 invention in light of '869 "problems" are simply irrelevant. Neither the '869 patent nor the '869 invention existed when the '538 invention was made. The '869 invention was the last made of all inventions before the court. It is prior art to nothing in the case.

It is clear on the face of Emery, Litwin, Fein, Bourne, and the '146 patent that no hinge remotely "of the kind claimed in the '538 patent" is shown or suggested in the drawings or specifications of any of those patents. On appeal, Dennison made no effort to support this finding. At trial, it abandoned Bourne in view of the PX 77 showing that Bourne disclosed no hinge. Uncontroverted exhibits PX 68, 14D established that Fein disclosed no hinge, despite its incomprehensible reference to "flexibility pivotable ratchet buckle." It is undisputed that the Emery and '146 hinges were horizontal and in shear. The finding that the prior art disclosed a hinge of the kind claimed is clearly erroneous.

The '869 Patent

"The problem addressed by that ['869] patent was insufficient strength in compression."

"The Japanese patent and the Litwin patent, which are prior art to the '869 patent, also showed structures analogous to the '869 ledge in that they absorbed compressive force, force exerted by the strap as it was tightened about the bundle."

"But I, nonetheless, find the '869 patent to be obvious both on general principles of physics and in light of the Japanese patent and the Litwin patent."

There is not a word in the '869 patent about compression, or about any "problem" of "insufficient strength," in compression or otherwise. Nor is there a word about absorbing compressive force. The court was here repeating back to Caveney his engineering testimony on the advantages achieved by the structural distinctions in the claim. The court then turned to its legally improper and impossible "general principles" standard of patentability. See 774 F.2d at 1097-98, 227 USPQ at 347-48.

The Japanese and Litwin patents disclosed structures entirely distinct from and nonsuggestive of the ledge and pawl structure set forth in claim 10. Both disclosed releasable pawls. Neither disclosed a ledge and pawl inside a frame. Whether those patents showed structures "analogous" because something in them "absorbed compressive force" is irrelevant, claim 10 being drawn to an entirely different structure, not to "absorbing compressive force." If the court's discussion be considered an implied finding that the content of the Japanese and Litwin patents constitute disclosures capable of rendering obvious the invention set forth in claim 10, such a finding would be clearly erroneous.

CONCLUSION

The portion of the district court's judgment rejecting Dennison's defense under 35 U.S.C. § 102(g) is affirmed. The portion of the district court's judgment upholding Dennison's defense under 35 U.S.C. § 103 is reversed. The case is remanded for further proceedings consistent with this opinion.

AFFIRMED IN PART, REVERSED IN PART, AND REMANDED

APPENDIX