MARKEY, Chief Judge.

SRI International (SRI) appeals from a final judgment of the United States District Court for the Northern District of California granting summary judgment of non-infringement to Matshushita Electric Corporation (MEI). 591 F. Supp. 464, 224 USPQ 70 (1984). We reverse.

BACKGROUND

Familiarity with the district court's opinion being assumed, only those factors necessary to an understanding of this court's decision will be discussed here.

The district court's opinion contains an excellent explication of the technology involved, reflecting a clear recognition of the undisputed fact that the filter and camera embodiment described in the specification of SRI's United States Patent No. 3,378,633 (the '633 patent) issued to SRI as assignee of the inventor Dr. Albert Macovski on April 16, 1968, operate somewhat differently from the MEI filter and camera accused of infringement.

The Claims

Each claim charged to have been infringed is drawn to a structure. Because of its central role as the broadest, claim 1 is set forth:

1. A spatial filter for affording a monochrome recording from which upon subsequent scanning, information for reproducing an image in color of the object photographed may be derived, said filter comprising a first grid of parallel spaced lines having the color of a subtractive primary, a second grid relatively angularly superimposed over all of the said first grid, said second grid having parallel spaced lines having the color of another subtractive primary, each grid having the same line density.

Claim 2 depends from claim 1, adding only "the relative angle between the first and second grids is 45°." Claim 7 and dependent aim 8 are drawn to a camera in combination with the filter of claim 1. Claim 9 and dependent claim 10 are drawn to "apparatus for generating the color representative electrical signals required for a color television receiver to reproduce a scene in color," and include the filter of claim 1.

Claim 3 was not charged to have been infringed. It is set forth because it affects the question of claim interpretation:

3. In a camera of the type wherein light from a scene being photographed is focused by a lens on black and white, monochromatic sensitive film, the improvement for affording a recording of the color information in the scene being photographed on said film in a manner so that said color information may be electronically derived therefrom comprising a filter adjacent said film, said filter having a first grid of vertical spaced lines which are colored cyan, and superimposed over all of said first grid a second grid having its lines at a 45° angle to the lines of the first grid, said second grid lines being yellow in color, both said grids having the same line density. [Emphasis added.]

Operation

The district court's Memorandum Opinion included helpful illustrations, that of SRI's filter being based on Figure 1 of the '633 patent drawings. Figure 1 shows an embodiment of the invention set forth in claim 1, in which one grid is superimposed over all of the other grid, the equi-width stripes of one are vertical and those of the other are angled to the vertical. In the operation of that embodiment, the horizontal distance travelled by the scanning beam in a single pass ("scan line") across and between angularly disposed stripes is greater than that across and between vertical stripes. Thus, the frequency range generated by interruptions in a single pass across the angular grid is lower than that across the vertical grid, and two basically discrete (red and blue) carrier frequency ranges are generated at that point. The frequencies are then decoded by low and band pass filters.

In the MEI filter, one grid is superimposed over all of the other grid and both grids have equi-width stripes, but the stripes of the two grids are at equal and opposite angles to the vertical. In operation, the horizontal distance travelled by the scanning beam in a single pass across and between stripes of both grids is the same, and the red and blue carrier frequency ranges thereby generated are the same at that point. However, the phase difference between successive, timed scanning passes produces a concentration of red and blue components at alternating, overlapping, or interleaved frequencies. To delay successive passes and decode the interleaved frequencies, MEI employs a 1-Hertz (1-H) delay comb line filter.

SRI has consistently maintained, and it is apparently undisputed, that there must be two frequencies separated at some point; otherwise no filter will work.

District Court Proceedings (a) Preliminaries

On July 19, 1982, SRI filed a complaint and jury demand, alleging MEI's willful infringement of claims 1, 2, and 7-10 and seeking treble damages, costs, attorney fees, and an injunction. That same day, the district court scheduled a status conference for October 22, 1982. On September 29, 1982, MEI filed answer, alleging invalidity, unenforceability, and non-infringement, and a counterclaim for declaratory judgment of invalidity and non-infringement.

In April, 1983, the parties filed extensive briefs.

On October 19, 1983, the district court issued a comprehensive Second Order for Pretrial Preparation, setting April 30, 1984 for "Trial before the Jury".

The October 19, 1983 order required the filing of exhibits, witness lists, summaries of all proposed testimony, statements of experts' theories with conclusions and bases therefor, experts' curriculum vitae, and reports prepared for testimony. The order form also contained a requirement applicable "In non-jury cases", i.e., the filing of a written narrative statement of the proposed direct testimony of each witness.

The October 19, 1983 order also required the filing of objections to testimony or exhibits, with the ground for each objections. Pursuant to another order of the same date, the parties were required to prepare statements of fact which each party contends should be submitted to a jury for determination.

Each side filed numerous objections. The record does not reflect that the court ruled on all of those objections.

In response, on January 20, 1984, counsel for SRI wrote the court, listing among the issues for the jury the question of whether MEI's filter is "so far changed in principle" that SRI's and MEI's filters are different inventions (MEI's position) or different versions of the same invention (SRI's position), quoting Westinghouse v. Boyden Power Brake Co., 170 U.S. 537, 568, 18 S.Ct. 707, 722, 42 L.Ed. 1136 (1898).

On February 2, 1984, counsel for MEI wrote the court, saying that: no issue required jury determination; the issue is whether the claims can cover MEI's filter; that issue is one of claim language scope and a matter of law; and whether MEI's camera functions in a basically different manner is a legal question, citing Kalman v. Kimberly Clark Corp., 713 F.2d 760, 218 USPQ 781 (Fed. Cir. 1983).

(b) February 10, 1984 Pretrial Conference

A February 10, 1984 pretrial conference to narrow the issues began with a late attempt by MEI's counsel to bar the inventor (Dr. Macovski) as an expert witness because of his financial interest in the outcome of the litigation. MEI's counsel said they had originally considered objecting on the ground that "[t]he subject matters [sic] as to which Dr. Macovski is going to testify is so far outside of anyone's real world experience that it's almost impossible to evaluate without appropriate technical background and credentials," and was therefore not appropriate "to go to a jury." That objection was not made. The court overruled the objection with respect to financial interest on the ground that SRI would be prejudiced if denied its witness on the eve of trial.

Seeking to bar the witness entirely, MEI's counsel did not say whether the court had the "appropriate technical background and credentials." In the end, MEI accepted "summary judgment", though the court heard no witness, but read and evaluated the written narrative statements of witnesses and Dr. Macovski's affidavit.

The court then turned to the purpose of the conference, i.e., determining "whether there are any issues here involving conflicts of facts." Having read the papers then on file, the court concluded:

It seems to me, under the particular recent decisions under the Federal Circuit, it's a pure question of deriving ultimate facts from undisputed evidence which is properly done by the judge. Mr. Gregg, I have read your papers carefully, but I don't see any factual issues that would be proper for a jury.

SRI's counsel then said: "That if there are no evidentiary facts in dispute, this is a case that is ripe for summary judgment. And I would suggest under those circumstances and with your Honor's consent, that the defendants bring a motion for summary judgment." When the judge asked whether SRI wished also to bring a summary judgment motion, SRI's counsel said it did not because counsel believed there were factual disputes present.

Asked to identify the disputes, SRI's counsel responded:

The dispute with respect to infringement largely evolves around this: There is no question but what the words of the claim, as we say, "read on," cover, define, not only the structure in the 633 patent, which is in suit here, but the structure which the defendants used.

And in accordance with the Supreme Court's decision in Graver vs. Linde, you look to the words of the claim, and do you determine whether or not they cover what the accused device is; and if it does, infringement is made out, and that is the end of it.

But there is another side of the coin, and the other side of the coin is this: That if the invention, if whatever the defendant has, if his device is so utterly different in principle from the device which is described in the patent, then regardless of the wording of the claims, the claims have not validly covered it. And that is the issue with respect to infringement, the principle [sic] issue as to whether or not the system which is used by the defendants in this case is so different in principle that it's an altogether different invention. One is an apple; the other is an orange.

And that seems to me to be mainly a fact of determination, and so on down the line.

To that statement the court responded:

Well, I don't think so. . . . [I]f the issue in this case is whether a thing which is colored orange, has [a] peel about an eighth-inch thick, . . . and once it's peeled, it's sectioned, those facts are undisputed. The question is whether the ultimate fact may be derived from that evidence that it is an orange.

That doesn't strike me as a jury issue. That strikes me as an issue that is appropriate for a judge to decide, particularly if oranges are rather obscure kind [sic] of things in the common experience of a jury. There is no dispute it has certain kinds of seeds inside. There is no conflict, in other words, in the evidence. There is only a conflict in the interpretation of the evidence.

It strikes me that is appropriate for a judge to decide. Whether it should be decided on summary judgment or not is another question. I wouldn't keep anybody from making a motion for summary judgment, but it would be useful in this case for the court to hear the testimony of the experts because I think the court needs some enlightenment on some rather technical questions.

After some discussion of evidence of obviousness, the court stated:

I just don't see any dispute here, and it seems to me that these are all legal issues, that is, they are issues of ultimate fact concerning the application of principles such as obviousness, anticipation, infringement, to undisputed facts.

Of course, Mr. Gregg, the two sides will present different evidence, but that doesn't mean there is any conflict in the evidence. And in the absence of an evidentiary conflict, I don't see a jury issue in a patent case.

MEI's counsel at that point said he would move for summary judgment "if it were done in a manner which I think would be permitted by the rules, where on the motion the court could hold an evidentiary hearing which would permit the court to hear the experts at least, and in a live situation develop an education as to the subject matter."

Evidentiary hearings to determine whether a genuine issue of material fact is present are appropriate on summary judgment motions. See 6 MOORE'S FEDERAL PRACTICE ¶ 56.11[8] (1985).

The court then described its procedure in bench trials, under which, to preclude surprise and facilitate preparation for trial, written narrative statements of testimony are exchanged, and the record after trial would become those narrative statements plus testimony adduced at trial. The court added:

And that way the record will be a combination of the narrative statement of testimony, plus the testimony adduced here. Now, that is not summary judgment, but it is a bench trial, and it anticipates the testimony and provides everybody with a complete statement of testimony and exhibits in advance.

Now, that will be my way of proceeding in this case. I don't think it's appropriate for summary judgment just because I would want the live witnesses. But I do think we can resolve these questions fairly and intelligently in a reasonable fashion in this way, and also I think more economically and expeditiously.

Usually when trials are conducted in this fashion, they don't take as long as people anticipate.

The February 10, 1984 conference ended with the setting of March 30 for submission of pretrial briefs and April 30 for a bench trial.

After the court stated at the February 10, 1984 conference, that it would conduct a bench trial, SRI's counsel did not orally insist on SRI's right to a jury trial. On the record here, all concerned appear to have expected at this point that there would be a bench trial.

On February 23, 1984, the court issued a Third Order for Pretrial Preparation and Trial in which it denied SRI's request for jury trial and ordered the bench trial discussed in the foregoing quote:

It appearing that there are no disputed issues of fact requiring determination by a jury; it is hereby

ORDERED, ADJUDGED, AND DECREED that the liability phase of this case shall be tried to the Court in accordance with the following schedule . . .

The order then repeated the bench-trial requirement for filing narrative statements of testimony, exhibits, and objections.

(c) April 13, 1984 Pretrial Conference

On April 13, 1984, at what proved to be the last pretrial conference and last courtroom event, the court conducted an extended dialogue with SRI's counsel, beginning with the court's demand that counsel state the "essence" of the '633 patent and "the novel element on the basis of which that patent was issued." The court asked SRI's counsel whether the assertions in MEI's brief (claims are limited to filters so operated that the angle difference of the stripes creates different frequencies) stated "the essence of the patent".

Counsel for SRI began correctly by stating the structural limitations of the claims. When the court said the "essence" was "the difference in the angles of the two sets of stripes creates different frequencies for the transmission of red and blue color," counsel said "stripes at an angle to one another and creating different frequencies" was old and that the "novel feature" was the equal width and spacing of the stripes.

Pressed by the court on whether "the basis on which the patent was issued" was "two sets of stripes being at an angle to one another" and "for the purpose of generating different frequencies," SRI's counsel finally, and inexplicably, said, "[t]hat's correct."

Reliance on a finding that a "novel element", or "essence" (or "gist", or "key") of a structural invention lies in the operation of a specification-described embodiment of the claimed structure would render meaningless the statutory requirement for claiming, 35 U.S.C. § 112, the statutory requirement for treating claims individually, 35 U.S.C. § 282, and the entire examination system centering on the allowance or rejection of claims. Inexplicably, SRI's counsel did not present these statutory considerations firmly to the district court at the April 13, 1984 pretrial conference.

The court then said MEI's filter does not generate frequency differences. When SRI's counsel said it did, the court demanded an explanation. Pleading that inventor Macovski could explain it better, counsel said MEI's 1-H delay resulted in frequency differences without which no camera would work.

Pointing to a statement in a brief filed a year earlier (April 29, 1983) by SRI's counsel (that MEI's was a "phase" system), the court deemed counsel's statement in the April 13, 1984 conference (that MEI's system generated two frequencies, albeit in a different way) an improper and possibly punishable change of position.

"Phase" and "frequency" are differing electrical phenomena. That difference appears to have influenced the district court. Counsel for SRI did not point out to the court that claim 1, for example, says nothing whatever about "phase" or "frequency", but calls only for "affording a monochrome recording from which upon subsequent scanning, information for producing an image in color of the object photographed may be derived." (Emphasis added.) Whether those skilled in the art would immediately recognize that "information" is equally applicable to "phase" and "frequency" is a contested issue (see Narrative Statement of Ball, App. Vol. II, 527-28) resolvable by live witnesses at trial. It is not resolvable on this appeal.

Counsel said he had learned more during the year, and that his expert witnesses and a paper by MEI's employee, Nakabe, would at trial show the correctness of his present statement.

When the court pointed out that "tenth grade geometry" shows no difference in spacing across and between MEI's stripes, SRI's counsel agreed, saying that, "[t]here is no frequency difference on a given scan line," but that "MEI's 1-H delay produced frequency differences."

When the court asked whether the real issue was whether the '633 patent "protects a system" which generates color differences by a time delay in rate of scanning between different lines, counsel answered, "Yes." The court then asked whether the patent or file wrapper said anywhere that "an essence or novel idea in this patent to be protected is the generation of frequency differences through time delays." Counsel for SRI sought to answer by citing references in the specification to frequency overlap.

Invited to comment, MEI's counsel simply agreed, in three sentences, with what the court had been saying, i.e., that use of "phase" and "frequency" was mere "semantics" and that nothing in the '633 patent suggests MEI's "technique", at which point the court said, "Well, I don't see why I shouldn't grant summary judgment in this case."

Referring repeatedly to the statement about "phase" in the year-old brief of SRI's counsel, the court rejected counsel's promise of what Dr. Macovski would show at trial and counsel's assertions that those skilled in the art would immediately on reading the '633 patent recognize the availability of what counsel called the minor changes made by MEI.

The court then asked MEI's counsel whether he desired a trial or summary judgment. MEI's counsel responded by offering to have his narrative witness statements sworn. The court said there was no need for any statements, because MEI's exhibits and the year-old statement of SRI's counsel removed all controversy.

Recognizing that it had surprised SRI, the court set a schedule for briefing on summary judgment. SRI accompanied its brief with an affidavit of Dr. Macovski. MEI filed no affidavit.

On August 1, 1984, after considering counsels' voluminous submissions, the district court entered summary judgment in favor of MEI.

242 exhibits and narrative statements were filed, along, apparently, with six briefs. The appendix on appeal consists of over 1,746 pages.

Issue

Whether the district court erred in granting summary judgment of non-infringement.

OPINION [49] (1) Introduction

The Federal Rules of Civil and Appellate Procedure and of Evidence are and should be applied in patent cases no differently from their application in any other type of case. Hence summary judgment under Rule 56, Fed.R.Civ.P., is entirely appropriate, in a patent as in any other case, where there is no genuine issue of material fact and the movant is entitled to judgment as a matter of law. See Molinaro v. Fannon/Courier Corp., 745 F.2d 651, 653-54, 223 USPQ 706, 707 (Fed. Cir. 1984); Chore-Time Equipment, Inc. v. Cumberland Corp., 713 F.2d 774, 778, 218 USPQ 673, 675 (Fed. Cir. 1983). The movant bears the burden of demonstrating absence of all genuine issues of material fact, Cooper v. Ford Motor Co., 748 F.2d 677, 679, 223 USPQ 1286, 1287-88 (Fed. Cir. 1984), the district court must view the evidence in a light most favorable to the nonmovant and draw all reasonable inferences in its favor, United States v. Diebold, Inc., 369 U.S. 654, 655, 82 S.Ct. 993, 994, 8 L.Ed.2d 176 (1962), Martin v. Barber, 755 F.2d 1564, 1566, 225 USPQ 233, 234 (Fed. Cir. 1985), and must resolve all doubt over factual issues in favor of the party opposing summary judgment, Palumbo v. Don-Joy Co., 762 F.2d 969, 973, 226 USPQ 5, 7 (Fed. Cir. 1985).

The record does not indicate that MEI actually filed a motion for summary judgment. MEI appears to have merely acquiesced when the district court announced, at the end of the April 13, 1984 pretrial conference, an intention to grant summary judgment (unless convinced otherwise by SRI's forthcoming brief).

The party opposing the motion must point to an evidentiary conflict created on the record; mere denials or conclusory statements are insufficient. Barmag Barmer Maschinenfabrik AG v. Murata Machinery, Ltd., 731 F.2d 831, 836, 221 USPQ 561, 564 (Fed. Cir. 1984). Because, however, infringement is itself a fact issue, a district court must approach a motion for summary judgment of infringement or noninfringement with a care proportioned to the likelihood of its being inappropriate. D.M.I., Inc. v. Deere Co., 755 F.2d 1570, 1573, 225 USPQ 236, 238 (Fed. Cir. 1985); Palumbo v. Don-Joy Co., 762 F.2d 974, 226 USPQ at 8.

Though speedy and inexpensive, summary judgment is nonetheless a "lethal weapon" capable of "overkill". Brunswick v. Vineburg, 370 F.2d 605, 612 (5th Cir. 1967). See generally Louis, Federal Summary Judgment Doctrine: A Critical Analysis, 83 Yale L.Rev. 745 (1974). It denies the non-movant its "day", i.e. a trial, in court. Moreover, experience has shown that a trial often establishes facts and inferences not gleanable from papers submitted pre-trial. Further, a reversal of summary judgment on appeal is not dispositive, but merely remits the case for trial. If reversal be error, that error will manifest itself at close of the non-movant's case, when the district court may direct a verdict or enter judgment accordingly.

In the present case, the admirable and extensive pretrial preparation system designed and vigorously managed by the district court, moved from preparation for a jury trial, through determination of whether the jury demand should be granted, to preparation for a bench trial, to the court's examination of unsworn counsel, to the court's study of exhibits and unsworn narrative statements, and finally to a grant of summary judgment. Counsel did not supply the clear, candid, and uncompromising explication of the law required to protect the pretrial preparation system in this instance from error.

As was stated in Cable Electric Products, Inc. v. Genmark, Inc., 770 F.2d 1015 (Fed. Cir. 1985), "the circumstances appropriate to summary judgment are those in which a district court is able to conclude that, with regard to any factual issues material to granting judgment as a matter of law, no genuine dispute exists." Here there was no dispute on the structure of MEI's filter and camera, and no dispute on the manner in which that structure is operated, but there was a dispute, as discussed infra, on whether that manner of operation meant that MEI's structure was so far changed in principle that it performed the same or similar function as the claimed invention in a substantially different way. As discussed, infra, that disputed question raises a genuine issue of material fact and the answer to that question controls the issue of infringement in this case.

Focusing exclusively on the way the structure of the Figure 1 embodiment in the specification operates, the district court allowed that focus to control its claim interpretation and its grant of summary judgment. In so doing, it erred.

(2) The Claims Read on MEI's Structure

The claims are structural, not method, claims. MEI did not in its brief before us concede that the claims as written read directly on its filter and camera. Because MEI's argument is based entirely on its "technique", i.e., on how its structure and that disclosed in Figure 1 of SRI's specification (or SRI's commercial embodiment) are operated, a concession that its structure falls literally within the words of the structural claims in suit would have saved time and cost for all concerned by focusing on the sole actual infringement issue present, which, as discussed below, centers on the so-called "reverse doctrine of equivalents", i.e. no infringement in law, in spite of literal correspondence with the claim language, because the product or process alleged as an infringement is not actually the invention patented.

At oral argument before the panel here, the following colloquy occurred:

THE COURT: Now the next question. If you ignore the specification — which, of course, is another question, hypothetical, but if you ignore it — don't you have literal infringement here, don't [the claims] read right on [the MEI device]?

COUNSEL FOR MEI: Oh, I suppose if you ignore the specification . . .

THE COURT: That's the hypothetical.

COUNSEL FOR MEI: Then you can take a meaning, as the judge said, of relatively angularly superimposed that would mean crossed at any angle.

THE COURT: Yes.

COUNSEL FOR MEI: Certainly you would do that, but I think what the judge, and I think going into this on that issue, we had the harder case to present to the district judge because what we had to do in order to show him what that language meant was teach him the art, go through the specification, show him the prior art, go through the prosecution history, which gives that terminology a special meaning. I agree, your honor, without the special meaning created by the prosecution here and the prior art, that language could as well read on the MEI system as the SRI system. [Emphasis added.]

To understand what is being claimed in each claim one must often refer to the specification, prosecution, and prior art. One must do so to resolve any ambiguity in claim language. Use may be made by either party of materials extraneous to the words of a claim, to give them a "special meaning" which will support an assertion of actual infringement/non-infringement under the doctrines of equivalents/reverse equivalents. That use does not change the answer to the question of whether the claim language as written "reads on" the accused device, which is synonymous with what has long been called " literal infringement". As the text makes clear, determination that the claim words read literally on the accused device is but an "initial hurdle". See Autogiro Co. of America v. United States, 384 F.2d 391, 399, 181 Ct.Cl. 55, 155 USPQ 697, 704 (1967).
Total candor on both sides could have produced a stipulation at the outset that the sole infringement issue before the district court, as explained infra, was whether the structure of MEI's filter and camera was "so far changed in principle" that it "performs the same or a similar function in a substantially different way." Graver Tank Mfg. Co., Inc. v. Linde Air Prods. Co., Inc., 339 U.S. 605, 608-09, 70 S.Ct. 854, 856-57, 94 L.Ed. 1097, 85 USPQ 328, 330 (1950).

The factual issue is whether MEI's filter and camera infringe properly interpreted claims. Fromson v. Advance Offset Plate, Inc., 720 F.2d 1565, 1569, 219 USPQ 1137, 1140 (Fed. Cir. 1983). Nothing in the present record would justify a determination that literal correspondence with claim 1, for example, is avoided by a composite filter structure formed of grids with equiwidth stripes superimposed in their entireties at equal and opposite angles to the vertical to derive information for reproducing an image in color, i.e., the MEI filter.

The language of the claims themselves is clear, definite, and unambiguous. It reads directly, unequivocally, and word-for-word on MEI's structure. As above indicated and as the district court correctly noted, there is no factual dispute concerning the structure of MEI's filter. It is undisputed that MEI's filter employs "relatively angularly superimposed" grids. The asserted claims, as the district court appeared to sense, are therefore literally met. If literal correspondence with the language of the claims were the sole consideration, SRI would on the present record be entitled to summary judgment, because there is no issue of genuine material fact that could affect that consideration.

Determination of actual infringement is not in every case, however, limited to literalism. An accused infringer, as here, may argue that the claim language, as properly interpreted, does not actually read on his device. See Leesona v. United States, 530 F.2d 896, 906, 208 Ct.Cl. 871, 887-88, 185 USPQ 156, 163 ( aff'd 192 USPQ 672) (1976) ("more than a literal response to the terms of the claims must be shown to make out a case of infringement"). Where, as here, a defendant says the claim language, as interpreted in light of the specification, prosecution history, and prior art, does not actually read on his device, claim interpretation is required. If the presently asserted claims must be interpreted as limited to a structure that is operated in the precise manner in which Figure 1 of the specification is operated, a structure not operated in that manner would not be an actual infringement. As appears below, however, nothing of record requires that the asserted claims be so limited.

(3) Claim Interpretation

Claim interpretation or construction is a legal matter subject to review free of the clearly erroneous standard applicable to fact findings. Raytheon Co. v. Roper Corp., 724 F.2d 951, 956, 220 USPQ 592, 596 (Fed. Cir. 1983); Fromson, supra, 720 F.2d at 1569, 219 USPQ at 1140. In construing claims, the prosecution history is a useful tool, McGill Inc. v. John Zink Co., 736 F.2d 666, 673, 221 USPQ 944, 949 (Fed. Cir. 1984), as is the patent specification, Fromson, 720 F.2d at 1569, 219 USPQ at 1140.

A claim is construed in the light of the claim language, the other claims, the prior art, the prosecution history, and the specification, not in light of the accused device. Contrary to what MEI's counsel wrote the district court, claims are not construed "to cover" or "not to cover" the accused device. That procedure would make infringement a matter of judicial whim. It is only after the claims have been construed without reference to the accused device that the claims, as so construed, are applied to the accused device to determine infringement. See Palumbo v. Don-Joy Co., 762 F.2d at 974, 226 USPQ at 8.

(i) Prosecution History

The court said SRI "had a lot of difficulty getting the Patent Office [sic] to issue the patent." The patent issued after two Office Actions and an interview. Some claims, including asserted claims 9 and 10, were allowed after one Office Action.

Responding to the examiner's initial rejection of certain claims in view of Kell, U.S. Patent No. 2,733,291, SRI entered extended Remarks in which it distinguished Kell on a number of different bases. Among those Remarks were these:

Claim 1, as amended, distinguishes over Kell in calling for a spatial filter which is comprised of a first grid of parallel spaced lines having the color of a subtractive primary, and a second grid which is relatively angularly superimposed over all of said first grid, said second grid having parallel spaced lines which are the color of another subtractive primary. . . . Claim 1 further distinguishes over Kell in calling for each grid to have the same line density. Every teaching in Kell, both in the specification and in the drawings, ... is that the line density is different for each grid. [Emphasis in original.]

The reason why Kell makes his line density different for each grid is that that is the way that he generates his two different modulating frequencies. This applicant insures two different modulating frequencies by virtue of placing one grid at an angle to another grid. As a result, the scanning beam, which traverses the line grids in successive parallel lines, is interrupted at two different frequencies by the two angularly disposed lines of the two grids.

At the April 13, 1984 conference, the court intensively examined SRI's counsel, focusing on and reading aloud the second of the two paragraphs above quoted. Continuing, the court said:

Do you accept that statement?

MR. GREGG: That's correct.

THE COURT: And that's the essence of the '633 patent, is it not?

MR. GREGG: That's part of the essence, bearing in mind what the statute ... 35 U.S.C. § 103, has to say, ... that you must look at the invention as a whole. And in connection with claim one, it is the filter; in connection with claim seven, for example, it is the entire camera, with the filter in it.

The district court in its Memorandum Opinion quoted the middle sentence of the paragraph it read aloud: "[T]his applicant insures two different modulating frequencies by virtue of placing one grid at an angle to another grid," and held that the claims must therefore be interpreted as limited to composite filters with angled stripes that themselves generate "two different modulating frequencies" when scanned in one pass. 591 F. Supp. at 470, 224 USPQ at 75. Because it found that MEI's angled stripes themselves generate the same modulating frequency in one pass, the court thought there could be no infringement of the structural claims in suit.

Unlike the patentees in Prodyne Enterprises, Inc. v. Julie Pomerantz, Inc., 743 F.2d 1581, 223 USPQ 477 (Fed. Cir. 1984) and Builders Concrete Inc. v. Bremerton Concrete Prods. Co., 757 F.2d 255, 225 USPQ 240 (Fed. Cir. 1985), SRI did not amend a claim to avoid a prior art reference by insertion of a structural limitation. SRI did not rely on incorporation by reference, as did the patentee in Envirotech Corp. v. Al George, Inc., 730 F.2d 753, 760, 221 USPQ 473, 477 (Fed. Cir. 1984). Nor, as did the patentee in SSIH Equipment S.A. v. U.S. Intern. Trade Com'n, 718 F.2d 365, 376, 218 USPQ 678, 687 (Fed. Cir. 1983), did SRI amend a claim to escape prior art by inserting language which could be understood only by reference to the prosecution history. SRI not only did not amend to escape prior art, it did not insert "relatively angularly superimposed" to do so, and at no time did it insert language into either the claims or the specification that would limit the "information" (set forth in claim as produced by its crossed grids) to that consisting of two rather than one frequency.

Nor did SRI insert in any claim any limitation to a particular manner of operation of its claimed structure. Nor in its Remarks did it distinguish Kell on the basis of Kell generating only one frequency in his single scan pass. On the contrary, SRI expressly recognized in the above-quoted Remarks that Kell generated two ("that is the way he generates his two different modulating frequencies").

Nowhere in the prosecution history is there any discussion of a composite filter that generates one modulating frequency. As the first above-quoted paragraph indicates, SRI recognized that Kell angularly superimposed parts of striped grids and that Kell got two frequencies in a single pass, but pointed out that Kell did so with partially superimposed grids of different line density. SRI could not have properly relied for patentability over Kell on admittedly old features.

The claims being structural, the quoted Remarks distinguished them over Kell, as those Remarks were required to do, on the basis of structure, i.e., grids angularly superimposed in part (Kell) vs. grids angularly superimposed in their entireties (SRI) and different line densities (stripe widths) (Kell) vs. same line densities (SRI). The claims were necessarily allowed on the basis of those structural differences. SRI's Remarks cannot be read, therefore, as strictly limiting (actually rewriting) the claims as though they were drawn to a method of operation in which angled stripes themselves generate two frequencies. Cf. Hughes Aircraft Co. v. United States, 717 F.2d 1351, 1361, 219 USPQ 473, 480 (Fed. Cir. 1983). SRI's explanation that Kell and SRI used different structures to get two frequencies illustrated the claimed structural distinction between partial vs. entire superimposition and different vs. same stripe widths; it did not convert the claims in suit to method claims.

In determining that SRI had "argued to the Patent Office a narrow construction of the claims," and was "precluded from arguing a broader construction for purposes of infringement," 591 F. Supp. at 470, 224 USPQ at 75, the district court misconstrued the nature and effect of the prosecution history. Nothing appearing in that history in relation to Kell required a "narrow construction" limiting the structural claims in suit to a method of operation. It was therefore not necessary for SRI to argue "a broader construction" to prove infringement.

We do not say, of course, that when an applicant distinguishes prior art and obtains allowance of a structural claim on the basis of the manner in which the claimed structure operates, he may not thereby have so limited his claim. When that happens, the examiner will normally require insertion of limitations to the argued operation. Nor do we say here that claim I may not be limited by prior art not before us on this appeal. Nor do we say that the claims must be interpreted to cover all products that literally infringe, no matter how far those products may be changed in principle to operate in a substantially different way from the way in which the claimed invention operates. We hold only that the single sentence on which the district court focused did not, in view of the structural distinctions argued and claimed, form the basis for allowance of claim 1. Because it was not relied on for patentability, as was the insertion in Builders Concrete, Inc. v. Bremerton Concrete Products Co., 757 F.2d 255, 225 USPQ 240 (Fed. Cir. 1985), that sentence did not constitute a prosecution history estoppel. A sentence pointing out how SRI's different structure can be operated to achieve the same result as Kell did not limit the claim solely to structures operated as the Figure 1 embodiment of the claimed invention is operated. MEI does not, of course, claim to employ Kell's structure or anything similar thereto.
Further, our present holding is limited to the present record and the grant of summary judgment. It does not foreclose the introduction by the parties at trial of any evidence, including that reflected in the prosecution history, on which the fact question raised by the reverse doctrine of equivalents, infra, may be resolved.

We review the district court's construction of the claims in light of the prosecution history as a question of law and we hold it to have been in error.

(ii) The Specification

The district court also focused on the specification, saying, 591 F. Supp. at 470, 224 USPQ at 74:

The specifications of the '633 patent make clear that the claims allegedly infringed envision a filter with grids superimposed at different angles from the vertical to generate frequency variations for the two primary color signals.

Quoting the specification's description of a composite filter structure comprised of two grids having lines at "relative angles such that . . . an adequate separation is obtained between carrier sidebands derived from said scanning," the district court said "the specifications assume one of the two grids to be oriented vertically with the other diagonally imposed on it to form a 45° angle." Id. at 470, 224 USPQ at 74. The district court correctly described the specification, but the difficulty is this: claims are infringed, not specifications.

When claim construction is required, claims are construable, as above indicated, in light of the specification, United States v. Adams, 383 U.S. 39, 49, 86 S.Ct. 708, 713, 15 L.Ed.2d 572, 148 USPQ 479, 482 (1966), yet "[t]hat claims are interpreted in light of the specification does not mean that everything expressed in the specification must be read into all the claims." Raytheon Co. v. Roper Corp., 724 F.2d at 957, 220 USPQ at 597. If everything in the specification were required to be read into the claims, or if structural claims were to be limited to devices operated precisely as a specification-described embodiment is operated, there would be no need for claims. Nor could an applicant, regardless of the prior art, claim more broadly than that embodiment. Nor would a basis remain for the statutory necessity that an applicant conclude his specification with "claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention." 35 U.S.C. § 112. It is the claims that measure the invention. Aro Mfg. Co., Inc. v. Convertible Top Replacement Co., Inc., 365 U.S. 336, 339, 81 S.Ct. 599, 600-01, 5 L.Ed.2d 592, 128 USPQ 354, 356-57 (1961); Bandag, Inc. v. Al Bolser's Tire Stores, Inc., 750 F.2d 903, 922, 223 USPQ 982, 996 (Fed. Cir. 1984); Jones v. Hardy, 727 F.2d 1524, 1528, 220 USPQ 1021, 1022 (Fed. Cir. 1984). Here the claims, which must be read as a whole, set forth structures distinguished from the prior art and were allowed on the basis of structural differences (e.g., claim 1's limitation to superimposition of entire grids, each having equi-width stripes).

The district court said, "[r]ead alongside the specifications, then, the allegedly infringed claims plainly teach a filter which positions red and blue signals at different carrier frequencies so that they can be decoded by electrical bandpass filters." 591 F. Supp. at 470, 224 USPQ at 74. Specifications teach. Claims claim. There is nothing in the asserted claims themselves about frequencies or band-pass filters. Claim 1, for example, is drawn to a filter affording a recording from which, upon scanning, "information" for reproducing an image may be derived. That a specification describes only one embodiment does not require that each claim be limited to that one embodiment.

Infringement, literal or by equivalence, is determined by comparing an accused product not with a preferred embodiment described in the specification, or with a commercialized embodiment of the patentee, but with the properly and previously construed claims in suit. See ACS Hospital Systems, Inc. v. Montefiore Hospital, 732 F.2d 1572, 1578, 221 USPQ 929, 933 (Fed. Cir. 1984); Teledyne McCormick Selph v. United States, 558 F.2d 1000, 1007, 195 USPQ 261, 267 (Ct.Cl. 1977).

The law does not require the impossible. Hence, it does not require that an applicant describe in his specification every conceivable and possible future embodiment of his invention. The law recognizes that patent specifications are written for those skilled in the art, and requires only that the inventor describe the "best mode" known at the time to him of making and using the invention. 35 U.S.C. § 112.

Whether SRI was entitled to a claim of the breadth of claim 1, for example, depends on what appears in the prior art as a whole, not on whether one or more embodiments appear in the specification. Entitlement is a question of patentability, and, as above indicated, is not before us. That the claims in suit are supported by the specification's disclosure has not been here challenged.

As the Supreme Court said in Smith v. Snow, 294 U.S. 1, 11, 55 S.Ct. 279, 283, 79 L.Ed. 721, 24 USPQ 26, 30 (1935):

We may take it that, as the statute requires, the specifications just detailed show a way of using the inventor's method and that he conceived that particular way described was the best one. But he is not confined to that particular mode of use since the claims of the patent, not its specifications, measure the invention. Paper Bag Patent Case, 210 U.S. 405, 419 [ 28 S.Ct. 748, 751, 52 L.Ed. 1122]; McCarty v. Lehigh Valley R. Co., 160 U.S. 110, 116 [ 16 S.Ct. 240, 242-43, 40 L.Ed. 358]; Winans v. Denmead, 15 How. 330, 343 [ 56 U.S. 330, 14 L.Ed. 717]. While the claims of a patent may incorporate the specifications or drawings by reference, see Snow v. Lakeshore R. Co., 121 U.S. 617, 630 [ 7 S.Ct. 1343, 1351, 30 L.Ed. 1004] and thus limit the patent to the form described in the specification, it is not necessary to embrace in the claims or describe in the specifications all possible forms in which the claimed principle may be reduced to practice.

As one of our predecessor courts stated in Autogiro Co. of America v. United States, 384 F.2d 391, 398, 181 Ct.Cl. 55, 64, 155 USPQ 697, 703 (1967):

The specification "set[s] forth the best mode contemplated by the inventor of carrying out his invention." 35 U.S.C. § 112. This one embodiment of the invention does not restrict the claims. Claim interpretation must not make use of "best mode" terms inasmuch as the patentee need not guard against infringement by listing every possible infringing device in the specification. Adams v. United States, 330 F.2d 622, 165, Ct.Cl. 576 (1964), aff'd 383 U.S. 39 [citations omitted] (1966) . . .

The specification language on which the district court focused was excerpted from that describing and explaining the operation of the structure shown in Figure 1, the embodiment employing a set of vertical grid stripes and a second set of grid stripes disposed at a relative angle to the vertical of 45° . The specification, however, expressly states that Figure 1 "by way of example" shows a version of the claimed invention in which a filter has vertical stripes and stripes angled at 45° to the vertical.

The language of no asserted claim requires that any stripes be vertical. Claim 2 is expressly more narrow than claim 1 in that it specifies "45° " as the relative angle between the stripes of the two grids. Non-asserted claim 3, supra, however, does contain a limitation requiring that one set of grid stripes be "vertical".

If, as SRI contends, MEI's stripes are each at 22 1/2° to the vertical, (i.e., that MEI merely "tilted" the grids of Figure 1) their relative angle would be 45° . MEI tells us it does not rely on the absence of vertical stripes from its filter as an indication of non-infringement.

The district court thus read, via the specification, the "45° " limitation of narrower claim 2, and the "vertical" limitation of narrower claim 3, into broader claim 1 and the other asserted claims. It is settled law that when a patent claim does not contain a certain limitation and another claim does, that limitation cannot be read into the former claim in determining either validity or infringement. Fromson, supra, 720 F.2d at 1570, 219 USPQ at 1141; Environmental Designs, Ltd. v. Union Oil Co. of California, 713 F.2d 693, 699, 218 USPQ 865, 871 (Fed. Cir. 1983); Palumbo v. Don-Joy Co., 762 F.2d at 977, 226 USPQ at 10.

The reason for the rule is well illustrated here. Non-asserted claim 3 is specifically drawn to the structure of Figure 1. To interpret the asserted claims as though they were also drawn to that structure (or to its manner of operation) would be to render the asserted claims superfluous.

In interpreting the asserted claims in light of the specification, the district court read limitations into them from other claims. That was error as a matter of law.

(4) The Reverse Doctrine of Equivalents

As counsel for SRI told the court, the "principal issue" of fact was whether MEI's filter is "so different in principle that it's an altogether different invention," an apparent attempt to paraphrase the judicially stated reverse doctrine of equivalents. As a basis for its grant of summary judgment, the district court found non-infringement under that doctrine. Because in this case there is a genuine issue of material fact respecting whether MEI's filter and camera structure is so far changed in principle that it performs the same or a similar function in a substantially different way, a trial of that issue is required.

At various points, SRI's counsel was, however, unhelpful to the district court. Under extended questioning by the court, counsel responded affirmatively to a question of whether the specification of the '633 patent contained material describing MEI's filter operation, leading the court to state a nonexistent requirement, "I don't think anywhere in the '633 patent do you find a picture of filters that are identical to the MEI."

Availability of the reverse doctrine of equivalents was set forth by the Supreme Court in Graver Tank Mfg. Co., Inc. v. Linde Air Prods. Co., Inc., 339 U.S. 605, 608-09, 70 S.Ct. 854, 856-57, 94 L.Ed. 1097, 85 USPQ 328, 330 (1950):

The wholesome realism of [the doctrine of equivalents] is not always applied in favor of a patentee but is sometimes used against him. Thus, where a device is so far changed in principle from a patented article that it performs the same or a similar function in a substantially different way, but nevertheless falls within the literal words of the claim, the doctrine of equivalents may be used [in reverse] to restrict the claim and defeat the patentee's action for infringement.

As made clear in Graver Tank, in which the Court upheld a finding of infringement under the doctrine of equivalents as "supported by the record" and "not clearly erroneous", the ultimate finding of infringement, or "actual infringement," may depend on facts and circumstances unique to the case.

Infringement is defined in 35 U.S.C. § 271(a) as the unlicensed making, using, or selling of a patentee's claimed invention. Recognizing frailties of language and limitations on human foresight, the law acknowledges that one may appropriate another's patented contribution not only with a product precisely described in a patent claim (literal infringement) but also with a product that is not quite so described but is in fact "substantially the same thing, used in substantially the same way, to achieve substantially the same result" (doctrine of equivalents). The law also acknowledges that one may only appear to have appropriated the patented contribution, when a product precisely described in a patent claim is in fact "so far changed in principle" that it performs in a " substantially different way" and is not therefore an appropriation (reverse doctrine of equivalents). One who takes a claimed structure and merely uses it in a way that differs from that in which a specification-described embodiment uses it, does not thereby escape infringement. Smith v. Snow, supra.

The patentee bears the burden of proving infringement by a preponderance of the evidence. Envirotech Corp. v. Al George, Inc., 732 F.2d 1572, 221 USPQ 929 (Fed. Cir. 1984); SSIH Equipment S.A. v. United States International Trade Commission, 718 F.2d 365, 218 USPQ 678 (Fed. Cir. 1983). Initially that burden is carried when literal infringement has been proved. See Graver Tank, 339 U.S. at 607, 70 S.Ct. at 855-56, 85 USPQ at 330. When a patentee establishes literal infringement the accused infringer may undertake the burden of going forward to establish the fact of non-infringement under the reverse doctrine of equivalents. If the accused infringer makes a prima facie case, the patentee, who retains the burden of persuasion on infringement, must rebut that prima facie case.

Because products on which patent claims are readable word for word often are in fact the same, perform the same function in the same way, and achieve the same result, as the claimed invention, a defense based on the reverse doctrine of equivalents is rarely offered.

The doctrine of equivalents raises a fact question determinable upon inquiry into whether a product is substantially the same thing used substantially the same way to achieve substantially the same result. Graver Tank, 339 U.S. at 609, 70 S.Ct. at 856-57, 85 USPQ at 330 ("A finding of equivalence is a determination of fact."). See also Autogiro Co. v. United States, 384 F.2d 391, 181 Ct.Cl. 55, 155 USPQ 697 (1967). As above indicated, the reverse doctrine of equivalents also raises a fact question, determinable on inquiry into whether a product has been so far changed in principle that it performs the same or similar function in a substantially different way. Though the second inquiry differs (literal infringement being present) it is nonetheless directed to a fact issue.

The district court here addressed the factual inquiries listed in Autogiro, but said it was doing so "as a matter of law" because, as it construed the claims in suit, the "gist of the . . . patent is the creation of two carrier frequencies to encode color information." But a court's election to ignore the structural claims, and to substitute a "gist" drawn from the operation of a disclosed embodiment, cannot convert the fact question raised by the reverse doctrine of equivalents into a legal question of claim construction. That fact question is simple and direct: Is the accused product so far changed in principle that it performs the function of the claimed invention in a substantially different way?

There is no question, as the district court found and as SRI from the outset admitted, that MEI's filter and camera are so operated as to perform their functions in a way that differs from the way in which the structure of SRI's Figure 1 performs those functions. SRI and MEI had the right, however, to adduce documentary and testimonial evidence at trial in an effort to convince the fact finder that the MEI filter is or is not "so far changed in principle" that it performs "in a substantially different way."

The test mandated in Graver Tank leaves room for the fact finder's application to varying circumstances. Words like "so far", "principle" and "substantially" are not subject to rigid pre-definition; nor will the "principle" of a structural invention be always and immediately apparent. It is precisely the role of a trial to apply the test in light of all the live testimony and physical evidence adduced.

At trial, SRI and MEI can cross-examine each other's witnesses on their narrative statements, and may sustain all or some of their objections filed pursuant to the October 19, 1983 order. (Those objections asserted conflict with witnesses' other statements and writings.) The SRI and MEI filters and cameras may be demonstrated. The principle of the claimed invention, and that of MEI's product, and the degree of change, if any, in the latter, may be determined in light of all the evidence, as may the question of whether the latter performs in a substantially different way. In sum, the key infringement question, i.e., whether a principle has been so far changed that the way is substantially different, depends on the evidence.

The Supreme Court supplied guidance useful here in Graver Tank, 339 U.S. at 609, 70 S.Ct. at 856-57, 85 USPQ at 330 (e.g., equivalence is not the prisoner of a formula, nor an absolute to be considered in a vacuum; it does not require identity in every purpose and respect; it is determined against the context of the patent, the prior art, and particular circumstances of the case; one factor is whether the accused product resulted from independent research; an important factor is whether persons reasonably skilled in the art would have known of the interchangeability of an ingredient not contained in the patent with one that was). A major factual dispute in this case, for example, is whether the parties employ substantially different ways of encoding color (MEI's position) or readily interchangeable ways long known to those skilled in the art for whom patents are written (SRI's position). See Thomas Betts Corp. v. Litton Systems, Inc., 720 F.2d 1572, 1580, 220 USPQ 1, 6 (Fed. Cir. 1983) (known interchangeability of elements is evidence of equivalence).

(5) Reason for In Banc Action

MEI relies on one phrase in a single case, Kalman v. Kimberly-Clark Corp., 713 F.2d 760, 218 USPQ 781 (Fed. Cir. 1983), for the proposition that non-infringement under the reverse doctrine of equivalents is a legal question. In Kalman, this court determined that the district court's fact finding of identity of invention (reached after a four day bench trial) was not clearly erroneous, and that "the stipulation by the parties, coupled with [Kimberly Clark's] failure to counter Kalman's affidavits and evidence submitted in his motion for summary judgment" dictated a finding of infringement. On that undisputed record, this court affirmed the judgment based on that finding. In response to an argument of appellant, the question of the reverse doctrine of equivalents — in a case involving that particular record of stipulation and unchallenged affidavits — was termed "a legal question". Id. at 771, 218 USPQ at 788.

We have found no case indicating that the fact question of infringement or non-infringement becomes "a legal question" when the determination rests on a disputed application of the reverse doctrine of equivalents.

Because the statement respecting the reverse doctrine of equivalents, i.e., that it raises a legal question, may be considered a holding to that effect, the court has sua sponte taken this case under consideration in banc. See South Corp. v. United States, 690 F.2d 1368, 1370 n. 2, 215 USPQ 657, 658 n. 2 (Fed. Cir. 1982). To the extent that the statement constituted a holding in Kalman, it must be and is hereby overruled.

As indicated in the text, counsel cited the statement in Kalman to the district court. Though the statement does not appear in the district court's Memorandum Opinion, it could have contributed to its grant of summary judgment. To the extent the district court may have been misled, this court regrets the necessity of correcting the statement appearing in Kalman.

It is settled that the question of infringement (literal or by equivalents) is factual. Graver Tank, supra. It would be anomalous to hold that the question of noninfringement (literal or by the reverse doctrine of equivalents) is not factual. Equivalence and non-equivalence, as the terms indicate, are but opposite sides of the same coin. In determining each, a court must say whether an accused device operates in substantially the same way (equivalence) or in a substantially different way (non-equivalence). The doctrine of equivalents may be employed, as was said in Graver, to expand or restrict a claim.

In both instances, one is dealing with whether a particular product falls within a range of products which the patent entitles the patentee to exclude others from making, using, or selling.

Indeed, the Supreme Court in Graver Tank, supra, and one of our predecessor courts in Decca Limited v. United States, 420 F.2d 1010, 1014, 190 Ct.Cl. 454, 459-463, 164 USPQ 348, 351, cert. denied, 400 U.S. 865, 91 S.Ct. 102, 27 L.Ed.2d 104 (1970), said "the doctrine of equivalents may be used to restrict the claim," (emphasis added), and this court in Kalman, 713 F.2d at 771, in rejecting appellant's argument on the reverse doctrine of equivalents, employed the classic statement of equivalence, saying: "the Berlyn devices do the same work, in substantially the same way, to accomplish substantially the same result."

The question of non-infringement under the reverse doctrine of equivalents, as above indicated, raises in this case a genuine issue of material fact which renders the issued summary judgment inappropriate. See Martin v. Barber, 755 F.2d 1564, 225 USPQ 233 (Fed. Cir. 1985).

(6) Conclusion

The genuine and material fact question of whether MEI's filter and camera are "so far changed in principle" that they "perform the same or a similar function in a substantially different way" being present and disputed, the district court erred in granting summary judgment.

A genuine issue of material fact being present, a trial on that issue is required. Presumably, jury trial was here denied on the premise that no such issue was present. If SRI renews its request for jury trial, the district court will consider that request in light of the foregoing opinion.

REVERSED.