After years of interesting struggles with patent issues under the tutelage of the distinguished patent bar in this Circuit, this court, by reason of the Federal Courts Improvement Act of 1982, soon must leave these appellate issues to the experts of the United States Court of Appeals for the Federal Circuit.1 Our view of patent matters therefore will be of little future consequence; nevertheless, we hope to leave.the field in good standing.
In a jury trial in the district court, plaintiff Roberts’ patent was found to be valid and infringed; judgment was entered in his favor in an amount in excess of eight million dollars. On appeal, the original panel, in a concise opinion authored by Judge Posner, concluded the patent was invalid as obvious, reversed the district court, and directed that the case be dismissed. This court determined that the case required en banc consideration. A majority now reaches a conclusion at variance with that of the original panel. At issue is the oftentimes confused role of the jury in a patent infringement action in which the invalidity of the patent in suit is raised as an affirmative defense.
I
Those of us who have experimented, not always successfully, with the “do-it-yourself” approach to car or bicycle repairs may have had occasion to use a conventional socket wrench. When we were able to tell the nut from the bolt, wished to remove the former, and had located the socket of the correct size, the problem became how to remove the socket then attached to the wrench. After pulling, prying, muttering, and more pulling, the two-handed operation was completed. Plaintiff Peter Roberts, having personally experienced such frustration, addressed himself to that problem; in 1963, he designed and constructed a prototype socket wrench with a quick-release feature that permitted its user to facilely change sockets with one hand without the customary pulling, prying, and muttering. Roberts filed an application for a United States patent on the wrench in April, 1964,2 which was rejected in March, 1965. He then extensively amended his application and presented a single claim3 upon which a *1328patent formally was issued on September 28, 1965.
Defendant Sears, the assignee of all rights to Roberts’ patent, mounted an advertising campaign which explained, in layman’s terms, the principal advantage of Roberts’ claimed invention: “Push-button ratchet wrench releases without a fight. Ever tried to separate a socket from an ordinary ratchet wrench when your hands were greasy? Forget it. You just press a button on Sears new Craftsman wrench. They separate easily — no yanking.” Roberts’ quick-release wrench was an enormous commercial success.
In 1969, Roberts sued Sears alleging, inter alia, that he was fraudulently induced to assign his rights to the invention to Sears. The jury awarded Roberts one million dollars in damages. This court affirmed the district court’s judgment against Sears and its decision not to alter Roberts’ monetary award, but reversed the district court’s determination that it lacked authority to order rescission of the agreement assigning Roberts’ rights to Sears. Roberts v. Sears, Roebuck & Co., 573 F.2d 976, 986 (7th Cir.1978) (Sears I).
Following the remand, Sears prepared, executed, and tendered to Roberts, through the district court, reassignment of any and all rights in the patent obtained pursuant to the June 15,1965, agreement, the sole equitable relief anticipated by this court in Sears I. The district court, however, further ordered the entire case reopened for an accounting of Sears’ “unjust enrichment” from June 15, 1965. On appeal, this court held that because Roberts had elected to submit his damage claim to a jury, he was precluded from pursuing the equitable remedy of restitution for “unjust enrichment.” Roberts v. Sears, Roebuck & Co., 617 F.2d 460, 465 (7th Cir.1980).
This court further held that Sears was the lawful owner of all patent rights from June 15, 1965, to January 20, 1977, when Roberts, through reassignment, became the lawful owner of all patent rights. We stated that Roberts would be entitled to sue only for infringement occurring after January 20, 1977. Although Sears was precluded from challenging the validity of the patent in the first trial, we made clear that Sears could do so if sued by Roberts for post-January 20, 1977, infringement. Id.
And it came to pass. Roberts instituted an infringement suit against Sears, which defended on the ground that the patent was invalid as both anticipated and obvious. Roberts demanded a jury trial, which was bifurcated at Sears’ request. Following a five-day trial on the issues of infringement, willful infringement, and validity, the jury was instructed on the relevant substantive law and given five “special verdict” forms to be answered “yes” or “no” pursuant to Fed.R.Civ.P. 49. They read as follows:
(1) We, the jury, find that the defendant willfully infringed the Roberts patent by selling the Roberts type wrenches after January 1977.
(2) We, the jury, find that the Orszulak or Z type design which defendant started to sell in 1980 infringes the Roberts patent in suit.
(3) We, the jury, find that the defendant willfully infringed the Roberts patent by selling the Orszulak or Z type wrenches after January 1977.
(4) We, the jury, find the Roberts patent is new and not anticipated by the Carpenter patent or the Gonzalez patent.
(5) We, the jury, find that the subject matter of the Roberts patent considered as a whole was not obvious to one of ordinary skill in the art in the years 1963-64.
All but special verdict number three were answered affirmatively by the jury. Number three is not now an issue. A two-day trial on the issue of damages followed, resulting in a jury award for Roberts of five million dollars.
*1329The district court composed no findings of its own. Instead, on Roberts’ motion, the district court entered an order adopting as its findings the affirmatively answered special verdicts rendered by the jury finding the Roberts patent not to have been anticipated or obvious, and to have been infringed. The district court concluded that the Roberts patent “is good and valid in law.” The damage award for willful infringement of the Roberts patent by sale of the Roberts-type wrench was increased by a factor of two, bringing the total award to $8,190,254.4 The district court permanently enjoined Sears from making, using, or selling the infringing wrenches through September 28, 1982, the Roberts patent expiration date. Sears’ motion for judgment notwithstanding the verdict or, in the alternative, for a new trial was denied.
Sears appealed, contending, inter alia, the Roberts patent was invalid because its claim was anticipated by, and further would have been obvious in light of, prior art references not before the patent examiner. Sears also claimed the district court erred in submitting to the jury the “legal questions” of obviousness and anticipation. The three-judge panel of this court reversed the district court’s judgment based on the jury verdict, reached its own contrary conclusion that the patent was invalid for obviousness, and directed that the complaint be dismissed. Roberts v. Sears, Roebuck & Co., 697 F.2d 796 (7th Cir.1983). Roberts’ petition for rehearing with suggestion for rehearing en banc was granted, and the panel decision was vacated by order dated March 14, 1983.
We find it necessary to address only one issue: whether the jury instructions and procedure utilized mandate reversal of the judgment and remand for a new trial. We premise our discussion of this issue with a review of relevant patent principles.
II
Article I, § 8, cl. 8 of the Constitution empowers Congress “to promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right5 to their re*1330spective writings and Discoveries.” In the exercise of this power, Congress enacted the Patent Act of 1952 (the Act), 35 U.S.C. § 1 et seq.6 The Act sets forth three criteria of patentability: novelty, utility, and nonobviousness.7 Graham v. John Deere Co., 383 *1331U.S. 1, 17, 86 S.Ct. 684, 693, 15 L.Ed.2d 545 (1966). In a patent infringement suit, the defendant may challenge successfully the validity of the plaintiff’s patent or any claim thereof by proving that any one of these three conditions of patentability was not satisfied; if the patent is proved invalid, the infringement issue is not reached. Moore v. Wesbar Corp., 701 F.2d 1247 (7th Cir.1983); Swofford v. B & W, Inc., 395 F.2d 362, 364 (5th Cir.1968). Sears, conceding the utility of the Roberts quick-release mechanism, claims it is undeserving of patent protection because it fails to satisfy the statutory conditions of novelty, the anticipation feature of which is at issue in this case, and nonobviousness.
Sears contends that the trial court erred in submitting to the jury the two issues of anticipation, which negates novelty, and obviousness; Sears further argues that requiring the jury to determine these issues necessarily required it to interpret the claim in Roberts’ patent, a function that is, with limited exceptions, for the trial court. We premise our discussion of these claims of error by recognizing that to the trial judge presented with an infrequent demand for a jury trial in a patent infringement case falls the most difficult task of determining which of the plurality of issues subsidiary to the ultimate determination of patent validity are subject to jury determination.
The indiscriminate use of inconclusive labels has engendered a great deal of confusion in the field of patent law. The lack of uniform decision and reasoning in circuit codrt opinions leaves the researcher of the issue whether the standards governing the determination of patentability present legal or factual issues with no definitive answer. Chief Judge Hastings, in recognition of this Circuit’s contribution to the disorder, stated in a separate opinion in Armour & Co. v. Wilson & Co., 274 F.2d 143, 155 (7th Cir. 1960):
We have come to speak of questions of “facts,” “primary facts,” “subsidiary facts,” “evidentiary facts,” “ultimate facts,” “physical facts,” “documentary facts,” “oral evidence,” “inferences,” “reasonable inferences,” “findings of fact,” “conclusions,” “conclusions of law,” “questions of fact,” “questions of law,” “mixed questions of law and fact,” “correct criteria of law,” and so on ad infinitum. The simple answer is that we are all too frequently dealing in semantics, and our choice of words does not always reflect the magic we would prefer to ascribe to them.8
We enter this labyrinth of conflicting ease law with one basic guidepost — the Patent Act itself as interpreted by the Supreme Court. In Graham v. John Deere Co., 383 U.S. 1, 86 S.Ct. 684, 15 L.Ed.2d 545 (1966), the Supreme Court pronounced the ultimate question of patent validity a question of law; the determination of patentable invention requires the courts “to give effect to the constitutional standard by appropriate application, in each case, of the statutory scheme of the Congress.” Id. at 6, 86 S.Ct. at 688. The determination requires an exercise in statutory construction, which, as the Supreme Court recognized, is statutory application in practice; the development of the factual content to which the statutory standard is to be applied is within the province of the fact-finder.
To be patentable, a device must be new, useful, and properly classified as invention. The issue of patentability, then, involves several inquiries: what constitutes the prior art and what does it disclose in scope and content; are there differences between the prior art and the claimed invention which render the latter an improve*1332ment on the former; and would the improvement have been obvious to one of ordinary skill in the pertinent art at the time the invention was made. Armour & Co. v. Wilson & Co., 274 F.2d at 156.
The first two inquiries determine novelty under § 102; in simple terms, there must be a difference between each prior art reference and the device sought to be patented for the latter to be. deemed “new.” Only if such differences exist does § 103 require a determination whether the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time of invention to one of ordinary skill in the art. Nonobviousness, then, distinguishes those useful innovations that are capable of sustaining a patent from those that are not. Graham v. John Deere Co., 383 U.S. at 3-4, 11, 86 S.Ct. at 686, 690.9
We turn first to a more detailed discussion of novelty. A device is “new” if its essence has not been disclosed in a prior art device. Section 102(a), (e), and (g) provide, in distilled form, that a device lacks novelty if it has been anticipated by a prior patent or publication in this or a foreign country or by prior use, knowledge, or invention in this country. This court has stated that anticipation is strictly a technical defense; “[ujnless all of the same elements [of the patented device] are found [in a single prior art device] in exactly the same situation and united in the same way to perform an identical function,” the former is not anticipated by the latter. Illinois Tool Works, Inc. v. Sweetheart Plastics, Inc., 436 F.2d 1180, 1182-83 (7th Cir.1971). We also have stated, however, that “[w]hen the only features distinguishing the purported invention from a prior art product are insubstantial, the earlier may properly be said to anticipate the later product.” Shelco, Inc. v. Dow Chemical Co., 466 F.2d 613, 614-15 (7th Cir.), cert. denied, 409 U.S. 876, 93 S.Ct. 125, 34 L.Ed.2d 129 (1972). Thus, in this Circuit, the test has been one of substantial identity;10 “[i]t is sufficient if the general aspects are the same and the difference in minor matters is only such as would suggest itself to one of ordinary skill in the art.” Amphenol Corp. v. General Time Corp., 397 F.2d 431, 438 (7th Cir.1968). *1333See also Popeil Brothers, Inc. v. Schick Electric, Inc., 494 F.2d 162, 164 (7th Cir.1974); Deep Welding, Inc. v. Sciaky Brothers, Inc., 417 F.2d 1227, 1234 (7th Cir.1969).
Substantial identity is determined by reference to the language of the patent claims, which define the ambit of the claimed invention. Thus, the determination whether a patented device has been anticipated by a prior art reference requires a two-step analysis: (1) the identity of the patented device, as well as its scope, must be determined by the claims submitted to and allowed by the Patent Office; and (2) the patented device, as so defined, must be compared with each prior art reference.
Construction of the patent claims is a question of law, Super Products Corp. v. D P Way Corp., 546 F.2d 748, 756 (7th Cir.1976), and thus ultimately the responsibility of the trial judge in a patent infringement case tried to a jury. If, however, extrinsic evidence is needed to explain a term of art in a patent claim, the meaning of which is disputed, a narrow factual issue for jury determination is presented. Id.
The issue whether a prior art reference and the patented device are substantially identical may be treated as a question of pure construction in a case in which no substantial disputes of fact are presented and the clarity of the patent documents is such that the court can determine from mere comparison of the descriptions contained therein whether the devices at issue are substantially identical.11 However,where material factual disputes exist as to the scope and content of the prior art and the differences between each prior art reference and the patented device, as identified by the trial court, issues for jury determination, if a jury is to be the fact-finder, are presented. This is not to say that the trial court does not maintain ultimate control over the issue of anticipation. To hold otherwise would be to vitiate the Supreme Court’s determination that patent validity ultimately is a question of law. Novelty, like utility and nonobviousness, is a condition precedent to patentability. In any given case, lack of novelty may be the ultimate determinant of patent validity; as such, the issue must be deemed to be within the province of the court.
Assuming differences between the patented device and each prior art reference preclude a finding of anticipation, under the broader obviousness test, the disclosures of the prior art references may negative invention because in their light the patented device would have been obvious at the time the ’ invention was made to a person having ordinary skill in the pertinent art.
Graham v. John Deere Co., 383 U.S. 1, 86 S.Ct. 684, 15 L.Ed.2d 545 (1966), established the exclusive means by which to measure nonobviousness under section 103. Republic Industries, Inc. v. Schlage Lock Co., 592 F.2d 963, 972 (7th Cir.1979). In reviewing a trial court’s application of the Graham standard, we are mindful of the Court’s admonition that “strict observance of the requirements laid down here will result in that uniformity arid definiteness which Congress called for in the 1952 Act.” Graham v. John Deere Co., 383 U.S. at 18, 86 S.Ct. at 694.
While the ultimate question of patent validity is one of law, ... the § 103 condition ... lends itself to several basic factual inquiries. Under § 103, the scope and content of the prior art are to be determined; differences between the pri- or 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 *1334or nonobviousness of the subject matter is determined. Such secondary considerations as commercial success, long felt but unresolved needs, failures 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 relevance.
Id. at 17-18, 86 S.Ct. at 693-694 (emphasis added).
One can readily see that the factual inquiries necessary to the determination of anticipation (i.e., the scope and content of the prior art and differences between each prior art reference and the claims of the patent in suit) also compose part of the tripartite factual inquiry upon which the determination of obviousness must rest. The anticipation inquiry, itself largely factual in nature, Hughes Tool Co. v. Ingersoll-Rand Co., 437 F.2d 1106, 1108 (5th Cir.), cert. denied, 403 U.S. 918, 91 S.Ct. 2230, 29 L.Ed.2d 696 (1971), is subsumed within the obviousness inquiry. Under the obviousness test, however, an additional factual inquiry is to be made: the level of ordinary skill in the pertinent art must be determined.
Several crucial concepts, as expounded upon in Graham and its progeny, are embodied in the section 103 standard.
The nature of the problem confronting the would-be inventor defines the relevant prior art. Republic Industries, Inc. v. Schlage Lock Co., 592 F.2d at 975. Pertinent art has been defined as that art to which one can reasonably be expected to look for a solution to the problem that the patented device attempts to solve. Morpul, Inc. v. Crescent Hosiery Mills, 265 F.Supp. 279, 303 (E.D.Tenn.1967). Inquiry into the level of ordinary skill in the pertinent art is necessary because a patentable invention must evidence more ingenuity and skill than that possessed by an ordinary mechanic acquainted with the business. Republic Industries, Inc. v. Schlage Lock Co., 592 F.2d at 975.
Obviousness is measured not by determining what would have been obvious to actual artisans, but by considering whether a hypothetical person, possessing reasonable skills in the pertinent art and knowledge of all prior art, would have found the same solution when addressing himself to the same problem. Id. Invalidity should not be found on the theory that if a development is obvious to the court, it must have been obvious to a person having ordinary skill in the art. Buzzelli v. Minnesota Mining & Manufacturing Co., 480 F.2d 541, 542-43 (6th Cir.1973). Judge Learned Hand put it this way:
To judge on our own that this or that new assemblage of old factors was, or was not, “obvious” is to substitute our ignorance for the acquaintance with the subject of those who were familiar with it.
Reiner v. I. Leon Co., 285 F.2d 501, 504 (2d Cir.1960), cert. denied, 366 U.S. 929, 81 S.Ct. 1649, 6 L.Ed.2d 388 (1961).
It also should be borne in mind that obviousness can only be determined by reference to the precise facts presented; intuitive analysis distorted by the invention’s simplicity and retrospective self-evidence must be avoided. Simplicity is not to be equated with obviousness. Skee-Trainer, Inc. v. Garelick Manufacturing Co., 361 F.2d 895, 899 (8th Cir.1966). Experience has shown that some of the most simple advances have been the most nonobvious. Van Veen v. United States, 151 U.S.P.Q. 506 (Ct.C1.1966). Patentability does not depend upon an astonishing breakthrough or creation of a new technology. Sauer Machine v. Corrugated Finishing Products, Inc., 642 F.2d 203, 206 (7th Cir.1981). In addition, a court must be careful not to declare an innovation obvious because it has become obvious through hindsight.” Id. “ ‘Obvious to try’ is not the same as ‘obviousness.’ ” Novo Industri A/S v. Travenol Laboratories, Inc., 677 F.2d 1202, 1208 (7th Cir.1982), quoting In re Goodwin, 576 F.2d 375, 377 (C.C.P.A.1978), aff’d, 599 F.2d 1061 (C.C.P.A.1979).
*1335Nor is the focus on the manner in which the invention was achieved or on the quality of the mind behind it. “[I]t is immaterial whether ‘the invention resulted from long toil and experimentation or from a flash of genius.’ ” Graham v. John Deere Co., 383 U.S. at 16 n. 8, 86 S.Ct. at 693 n. 8, quoting section 103 Reviser’s Note.
Although prior to 1966 courts often dealt with obviousness as a question of fact, Graham established patent validity, of which nonobviousness is the ultimate determinant, as a question of law. This court consistently has viewed obviousness as a question of law, but that legal determination must rest upon the tripartite factual inquiry set forth in Graham. See, eg., Dickey-john Corp. v. International Tapetronics Corp., 710 F.2d 329 (7th Cir.1983); Dual Manufacturing & Engineering, Inc. v. Burris Industries, 619 F.2d 660 (7th Cir.) (en banc), cert. denied, 449 U.S. 870, 101 S.Ct. 208, 66 L.Ed.2d 90 (1980). With the Graham -mandated factual determinations in hand,12 the court must draw the inferences that the findings reasonably induce, and reach the ultimate legal question whether the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the pertinent art and knowledge of all prior art. 35 U.S.C. § 103; Sakraida v. Ag Pro, Inc., 425 U.S. at 280, 96 S.Ct. at 1536; Moore v. Wesbar Corp., 701 F.2d at 1250.
Graham, in setting forth the analytical steps to be taken in determining obviousness, necessarily clarified the respective functions of judge and jury, as well as the scope of appellate review: disputes as to the Graham subsidiary facts are within the province of the jury; however, responsibility for the ultimate determination of obviousness lies with the trial judge, who must determine whether the facts as found by the jury -fall within the legislative standard.
An appellate court does not sit to adjudicate de novo the factual issues underlying the determination of obviousness. When made in a bench trial, the Graham factual determinations are reviewed under the clearly erroneous standard of Rule 52(a), Federal Rules of Civil Procedure. When these factual determinations have been made by a jury, our review is limited, assuming the instructions are not an issue, to ascertaining whether the denial of a motion for a new trial or a motion for a judgment notwithstanding the verdict was an abuse of discretion. An appellate court’s independent review of the trial judge’s conclusion of law is predicated upon the factual determinations made in the district court. See generally 2 D. Chisum, Patents § 5.04[3] n. 32 (1983).
Before turning to an examination of how decisional responsibilities were allocated between judge and jury in this case, we must determine whether there was a total absence of subsidiary factual disputes that would have required the trial judge to dispose of the entire case as a matter of law. Dual Manufacturing & Engineering, Inc. v. Burris Industries, 619 F.2d at 663.
Ill
Shown below in Figure 1 is a conventional socket wrench. It consists of a handle (3) *1336and a detachable socket (2). The pushbutton (1) on the head of the wrench corresponds with the pushbutton (also 1) on Roberts’ quick-release mechanism shown below in Figure 2. Dr. Youngdahl, Roberts’ expert witness, explained the structure and operation of the Roberts quick-release mechanism. The principal object of Roberts’ device was to provide a means for releasing a socket from a conventional wrench and engaging another in a one-handed operation. Although the Roberts quick-release mechanism retains the ball-type detent (4) utilized in the conventional wrench, in place of a spring directly behind the ball, Roberts uses a springloaded pin (5) that moves in the center of the stud (6). The pin contains a specially-shaped camming recess (7). “Selective alignment,” a key feature of the Roberts device, refers to the' type of ball and pin positioning that results from the “camming action” between the sloped recess of the pin and the curved surface of the ball. In the socket retaining position, small longitudinal movements of the pin result in changes of position of the ball with respect to the pin; the ball and pin can assume one of an infinite number of positions relative to one another depending upon the depth of the socket dimple. The ball will be cammed outward only as far as is necessary to engage the socket dimple. This “selective alignment” feature is claimed to offer unique advantages: it results in a smooth, reliable quick-release operation; it allows use of sockets with various dimple depths; it automatically compensates for socket wear; it prevents “hang up” of the ball detent; and it allows for one-handed operation in both removing and replacing sockets. To release the socket, the pushbutton (1) is depressed, moving the pin to the position where the ball can begin to move along the sloped recess of the pin. Once the ball is in the recess, the socket can fall of its own weight. With the pushbutton still depressed, a new socket may be slipped on and securely retained by releasing the pushbutton, all in a one-handed operation. It is obviously easier to use the wrench than it is to understand what makes it easier to use.
*1337Sears’ expert witness, Smyers, testified that nothing in Roberts’ claim denominates his device as a quick-release mechanism; he further testified that Roberts’ claim does not specify a particularly-shaped recess.
The trial transcript discloses numerous factual disputes as to the structures, operations, and functions of, and the differences between, the prior art and the patent in suit.
Dr. Youngdahl testified that the Carpenter device, a locking device for a socket wrench patented in 1928 (U.S. Patent No. 1,660,989) was not a quick-release mechanism, but was designed to effect a positive lock of the socket on the wrench through the use of a spring arm.13 He further testified to the Carpenter structure and operation as follows: Carpenter’s essential element, the spring arm upon which the ball usually rests,14 was designed to prevent the ball from moving into and becoming stuck in the recess, thereby inadvertently allowing unlocking of the socket. Although Carpenter contains a long, narrow, triangularly-shaped slot, its purpose is to receive the spring arm; to release the lock, the user pushes the button inward to place the recess under the ball; the ball, however, does not move into the recess because the spring arm continues to bear it outward. Therefore, it is claimed the socket must be removed forcefully; it will not fall of its own weight. Insofar as the Carpenter device discloses a pushbutton, pressing it merely unlocks the socket but does not effect release or removal of the socket from the wrench.
Youngdahl further testified that if the spring arm is removed, a “camming action” between the inclined surface of the recess and the ball may or may not force the ball toward the locking position; it “may cam” or it “may jam”; in Youngdahl’s opinion, the Carpenter geometry cannot provide for selective alignment. On cross-examination, Youngdahl, asked to assume that the Carpenter device contained a very weak spring arm, agreed that the force generated by the “camming action” might overcome the spring force and allow the ball to move into the recess, thereby effecting release of the socket.15 He further stated, however, that a weak spring arm will not éffect the locking of the socket on the wrench, the primary goal of the Carpenter device.
Smyers contrarily testified that Carpenter was a quick-release wrench. Although Smyers conceded that the text of the Carpenter patent did not so specify, he stated that he believed the spring arm was an optional feature. Smyers testified that when the pushbutton is depressed, the recess in the Carpenter device will be aligned with the ball. The weight of the socket will force the ball into the recess, releasing the socket. He further testified that a camming action between the slightly slanted sides of the recess and the ball will force the ball to its locking position.
Youngdahl’s testimony on the Gonzalez device, a ball socket attachment for impact tools and socket wrenches patented in 1965 *1338(U.S. Patent No. 3,172,675), may be summarized as follows: Gonzalez was designed to effect a positive lock of the socket on the wrench under the adverse conditions encountered with impact tools; the flat surface of the flange, with which the ball is normally engaged, “holds the ball out positively”; therefore, as planned, small longitudinal movements of the pin do not result in a change of position of the ball. Youngdahl stated that Gonzalez designed his device so that there would be a sharp discontinuity between the pin and flange, giving rise to the flat edge which supports the ball. When the pushbutton is depressed to release the socket, the ball drops into a space behind the flange. In Youngdahl’s opinion, the open space behind the flange in Gonzalez performs only one function — receiving the ball in the socket removing position; it will not invariably perform the function of camming the ball outward into the socket retaining position,16 as Roberts’ recess will. In order to perform the dual function that Roberts’ recess does, a sloping face must be cut in the flange, and the space behind the flange enlarged. It was Youngdahl’s further opinion that realistic limitations on the size of a socket wrench would preclude the necessary enlargement of this space.
Smyers, however, was of the opinion that the space behind the flange in Gonzalez will receive the ball and allow the socket to fall of its own weight or be manually removed. It was his further opinion that a camming action between the sharp edge of the flange and the ball will move the ball upward to engage a new socket. He also stated that Gonzalez was designed to achieve “selective alignment”; the ball and pin can assume an infinite number of positions to engage shallow or deep-dimpled sockets.
Sears has referred to the factual disputes raised by the expert testimony as “semantic quibbles,” “linguistic sophistry,” “petty details of terminology,” “a grab-bag of contentions,” and “imaginary distinctions.” If that is all it may actually amount to after the differing views are weighed, it is for the jury to say as a factual matter, not for this court acting as a fact-finder. We, like the trial judge who denied without comment Sears’ motion for summary judgment on the issue of patent validity and its subsequent motion for a directed verdict, believe the evidence presented was sufficient to raise questions of fact for jury determination on the elements, operations, and functions of the pri- or art references and the differences between them and the Roberts mechanism. We are not prepared to resolve the factual disputes between the experts and to come to our own factual conclusions.
IV
Application of the principles enunciated in Section II, supra, to the “special verdict” procedure and attendant instructions utilized by the trial court mandates the conclusion that the jury was impermissibly allowed to be the final arbiter of patent validity.
Initially, we note that the jury was required to interpret the scope and identity of the Roberts patent claim.17 That was asking too much of the jury. Construction of the patent claim, a prerequisite to the determinations of validity and infringement, is a matter of law for the court. Only a factual dispute as to the meaning of a term of art used in the patent claim, the resolution of which required resort to expert testimony, properly would *1339have been submitted to the jury.18 Super Products Corp. v. D P Way Corp., 546 F.2d at 756. Because the trial court’s construction of Roberts’ claim was necessary to the jury’s resolution of the factual disputes on the differences between the subject matter of the claim and the prior art, that error has shadowed the entire decisional process.
In addition, the jury was explicitly instructed, over Sears’ objection, to determine the ultimate legal question of obviousness. An attendant instruction merely identified the Graham underlying factual inquiries and instructed the jury to “consider the evidence” presented by the parties with respect to them.19 The jury, which evidénced confusion about the proper application of the legal standard of obviousness,20 proceeded to return a “special verdict” that Roberts’ patent would not have been obvious to one of ordinary skill in the art in 1963-64. On Roberts’ motion, the district court adopted as its own findings the “special verdicts” of the jury finding the Roberts patent was neither anticipated nor obvious and was infringed. The district court then added that the patent was “good and valid in law.” Sears’ motion for a judgment notwithstanding the verdict or, in the alternative, for a new trial was denied.
Sears argues it was error to instruct the jury to determine the legal question of obviousness; Sears further contends that the jury made no factual findings subsidiary to the determination of obviousness, but rendered a finding solely on the legal issue, which then was adopted by the trial court. In apparent reliance upon this court’s decisions in Panther Pumps & Equipment Co. v. Hydrocraft, Inc., 468 F.2d 225 (7th Cir. *13401972), cert. denied, 411 U.S. 965, 93 S.Ct. 2143, 36 L.Ed.2d 685 (1973), and Dual Manufacturing & Engineering, Inc. v. Burris Industries, Inc., 619 F.2d 660 (7th Cir.1980), Roberts contends that the “special verdict” procedure did no more than ask the jury to resolve the material factual disputes; because a jury may not be asked to find facts in a legal vacuum, it was the trial court’s duty to instruct the jury on the law applicable to the case. Roberts also appears to contend that the necessary factual findings should be presumed from the jury’s verdict, and further that the trial court ruled on obviousness as a matter of law in denying Sears’ motion for judgment notwithstanding the verdict.
In patent cases particularly, the type of verdict and instructions utilized by the trial court are keys to effectuation of the proper distribution of decisional responsibilities among judge and jury. In Dual, in a rare exercise of this court’s supervisory powers, Judge Pell writing for the en banc court, Judge Sprecher dissenting, directed that one of two alternative procedural methods be utilized in those instances where jury trials of patent issues cannot be avoided. The preferable method was stated to be the use of special verdicts: “[bjecause only issues of fact subsidiary to the legal question of obviousness are within the province of the jury, its resolution of those issues of fact should ordinarily be articulated in special verdicts under Rule 49(a), Fed.R.Civ. P.”21 Alternatively, it was stated that the “same result may be achieved by special interrogatories returned with a general^ verdict under Rule 49(b), a device primarily designed to test the jury’s application of the law in reaching a general verdict.” Dual Manufacturing & Engineering, Inc. v. Burris Industries, Inc., 619 F.2d at 667. But, it was added that the use of a general verdict, with or without special interrogatories, ordinarily will serve no purpose because the trial court still must decide the legal issue of obviousness. It was pointed out that although a general verdict standing alone will give rise to the presumption that the factual issues have been resolved in favor of the prevailing party, presumptions are of little use to a trial court fulfilling its obligation to decide the ultimate legal issue of obviousness. The Dual court concluded that “the special verdict device may allow the jury, if one is utilized, to serve a useful function in resolving specific contested issues as to the concrete facts, and it should be used.” Id. (emphasis added).
In this case the directions of Dual were not followed. The verdict form utilized was “special” only in the sense that the legal issue of patent validity was broken down into the components of obviousness and anticipation. Proper special verdicts, however, are to be addressed only to the subsidiary questions of fact that compose the Graham tripartite inquiry upon which the legal determination of validity must rest, not to the obviousness and anticipation components of patent validity themselves.
In light of this court’s instructions in Dual that special verdicts should be used, the question arises whether the failure to utilize proper special verdicts constitutes an abuse of discretion. At least one circuit court has suggested it may.22 In Panther Pumps & Equipment Co. v. Hydrocraft, Inc., 468 F.2d 225 (7th Cir.1972), this court declined to find in the circumstances of that case an abuse of discretion in the trial court’s refusal to submit thirty-two special *1341interrogatories to the jury. However, since the decision in Panther Pumps this court has exercised its supervisory powers to direct that certain procedures be followed. What was done in the present case is exactly what Dual sought to avoid. The court undertook to utilize Special verdicts, but in doing so turned the whole case over to the jury. Having chosen the special verdict procedure, it was error to ignore Dual’s instructions. But, even apart from Dual, the procedure utilized did not comport with Panther Pumps.
In Panther Pumps, both parties and the trial court had confused issues of fact with issues of law; all issues relating to validity and infringement were submitted to the jury. Four “special interrogatories” asked whether the two patents were valid and whether they were infringed. It is important to note that the issue whether the trial court erred in permitting the jury to decide the legal question of obviousness was not specifically before the court in Panther Pumps. Rather, defendant argued that in view of the' way the jury verdict was reached, it was impossible for a reviewing court to determine whether the three-step analysis required by Graham had been properly made.
The Panther Pumps court reiterated that obviousness is a question of law and that in a patent case, as in any other case tried to a jury, questions of law are for the court and questions of fact are for the jury. Allocation between judge and jury of their respective decisional responsibilities may be accomplished by the use of special verdicts or special interrogatories or “by the court’s instructions to the jury before it returns a general verdict.” Id. at 228. That the jury returns a general verdict may not preclude appellate review of the legal conclusion on validity under the Graham guidelines if the jury was properly instructed.
As in any other jury trial, the court’s rulings on questions of law are subject to review. If the ultimate issue of validity depends on subsidiary fact questions, it is the court’s duty to instruct the jury that it should return one verdict if the facts are found one way and a different verdict if the facts are found otherwise.
Id. On appeal, the reviewing court will presume that “the disputed matters of fact have been resolved favorably to the prevailing party in accordance with the trial judge’s instructions.” Id. Because defendants did not challenge the propriety of the trial judge’s instructions, they were not an issue before the court in Panther Pumps.
Panther Pumps did make clear, however, that a compulsory instruction must be given when a general verdict is utilized: “[I]t is the court’s duty to instruct the jury that it should return one verdict if the facts are found one way and a different verdict if the facts are found otherwise.” Id. In other words, the jury must be instructed that if it finds, facts A, B, C, and D, it must render a certain verdict. Anything less than strict adherence to this procedure by a trial court constitutes an abdication of its active duty to retain ultimate control over the issue of obviousness.23
*1342Panther Pumps itself does not suggest that the trial judge, having given the required compulsory instruction, also must rule independently on the issue of obviousness after the general verdict is returned. Under Panther Pumps, the trial judge retains ultimate control over the legal question of patent validity through the compulsory instruction supporting a general verdict. In Dual, it was stated that on the basis of the factual findings presumed from a general verdict, the court must then decide the issue of obyiousness. It amounts to the same thing. Under the Panther Pumps procedure, the jury, in finding certain facts, also is permitted to express its opinion on the legal issue, but the responsibility for deciding the legal issue remains with the trial judge, guided by the jury’s factual determinations. The procedure and instructions utilized in this case were deficient in every respect.
There was no attempt in the present case to utilize the Panther Pumps -sanctioned general verdict procedure.24 The jury merely was instructed to determine obviousness; it was not instructed that if the evidence supported a particular set of factual findings, a certain Verdict on obviousness was required. The trial court abdicated its control over the legal issue.
Nor are we willing to construe the trial court’s determination that the patent was “good and valid in Jaw” as an indication that it fulfilled its duty to “give effect to the constitutional standard, by appropriate application, in each case, of the statutory scheme of the Congress.” Graham v. John Deere Co., 383 U.S. at 6, 86 S.Ct. at 688. The trial court’s determination was a meaningless pronouncement. It is the compulsory instruction mandated by Panther Pumps that permits the presumption that the factual disputes have been resolved in favor of the prevailing party. Under the circumstances of this case, the trial court could not presume, nor may we, that the facts were found in favor of, Roberts merely because the jury ultimately made the legal determination that his patent would not have been obvious. The trial court had no sufficient factual basis, presumed or otherwise, for ruling that the patent was “good and valid in law.”
*1343We cannot sanction the procedure utilized by labeling the jury’s determination of obviousness as a “general verdict.” Affixing that label provides an insecure basis for presuming the Graham factual determinations have been made. That would be nothing less than sophistry. We believe the placement of a “general verdict” label on a procedure that merely asked the jury to pass upon the ultimate legal question, guided only by general instructions, would in effect dispense with the specific factual determinations called for in Graham as a predicate for the court’s determination of the legal question of obviousness. We hold the procedure and instructions utilized impermissibly allowed the jury to be the final arbiter of the legal issue of patent validity. This constitutes reversible error.
V
Having concluded that the district court committed reversible error, the question remains whether remand for a new trial is the appropriate disposition. Citing Dual, Sears contends that the question of patent validity is one of law for this court to decide; moreover, Sears argues, the patent in suit and the prior art are not complex or highly technical, and are easily understandable from the drawings and descriptions in the documentary evidence.25 Sears also appears to suggest that the erroneous procedure utilized by the trial court, which resulted in “all but worthless” jury verdicts, militates in favor of determination by this court of the legal issue of patent validity.
Although, from the record we have before us, we might be able, after reviewing and weighing the evidence, to arrive at what we consider to be the correct decision on anticipation and obviousness, we would have to find certain facts, as well as determine and apply the law. We believe that the judicial dispute resolution system established by our Constitution should not be infringed by the ad hoc shortcut of dispensing with trial judge and jury. In light of this constitutional system and the judicially-created doctrines that relegate an appellate court to its proper role in the system, Sears’ arguments on the extent to which we may examine and evaluate the conflicting evidence afford no basis for this court to proceed further. We do not intend as a court to make a jury’s factual decisions. It is elementary that the documentary evidence rule cannot be invoked in a case, such as this, in which technical experts have been used to interpret and clarify the structures, operations, and functions of the prior art and the patent in suit. See Wolens v. F.W. Woolworth Co., 703 F.2d 983 (7th Cir.1983); LaSalle Street Press, Inc. v. McCormick & Henderson, Inc., 445 F.2d 84, 87 (7th Cir.1971). In a patent case, as in any other, due regard must be given to the jury’s opportunity to judge the credibility of the witnesses and to appraise the significance of conflicting testimony. Reese v. Elkhart Welding and Boiler Works, Inc., 447 F.2d 517, 521 (7th Cir.1971).
In addition, Sears’ reliance on Dual is misplaced; the record in Dual revealed no subsidiary factual disputes, a circumstance that permitted this court to apply the legislative standard to the uncontradicted facts. Conversely, as demonstrated in Section III, supra, numerous subsidiary factual disputes were present in this case. It remains a mystery to this court how these factual disputes may have been resolved by the jury or what part Graham actually played in the decision of this case. The procedure utilized by the trial court precludes us from confidently presuming that these factual disputes were resolved in favor of Roberts.
This court has long eschewed examination of the question of obviousness in a factual vacuum. An early expression of this court’s desire that “verbal expression” be given to each analytical step mandated by Graham can be found in Cloud v. Standard Packaging Corp., 376 F.2d 384 (7th Cir.1967). The Cloud court refused to imply the Graham findings in the district court’s *1344general conclusion that the defendants had failed to carry their burden of proving the invalidity of the presumptively valid patents in suit. In that case, the evidence was comprised of the uncontradicted explanation of a number of prior art patents by one expert witness. This court suggested that “perhaps” it could make the validity determination “without trespassing upon the area of fact finding.” Id. at 391-92. Nevertheless, even in those less complicated circumstances, we held it was sounder for the initial determination to be made by the trial court. If we undertake to do it all here, no court remains where the parties can obtain review as a matter of right. Id. at 391.
We see no alternative but to remand this case to the district court for a new trial in accordance with this opinion. Nothing we have said is intended to foreshadow the final result to be reached in this case on the merits.
Each party shall bear its or his own costs.
