This patent infringement case concerns a new form of printing whereby printer's ink is guided onto paper or other recording media by electrical charges applied to ink streaming from small nozzles, rather than by conventional type, press, pen or other contact. The process, called "ink jet printing," is now used by businesses which require high speed printing or printing on soft or other unusual surfaces. Ink jet printers are particularly well-suited for computer print-outs and labeling. The parties to this action advise us that ink jet printing may be the "wave of the future" and that our decision in the case will control the interpretation of key patents in this expanding industry.

A.B. Dick Company ("A.B. Dick"), the owner of the "Sweet" patent, U.S. Patent No. 3,596,275 issued on July 27, 1971, and the exclusive licensee under the "Lewis-Brown" patent, U.S. Patent No. 3,298,030 issued on January 10, 1967, and Gould, Inc. ("Gould"), the owner of the Lewis-Brown patent, assert that the DIJIT printer, manufactured by the Mead Corporation ("Mead"), infringes these two patents. In a well-considered opinion, District Judge Walter Rice1 concluded inter alia that the Lewis-Brown patent is invalid and that the DIJIT printer does not infringe the Sweet patent.2 Mead Digital Systems, Inc. v. A.B. Dick Co., 521 F. Supp. 164 (S.D.Ohio 1981).

The District Court rested its determination of non-infringement largely on the grounds that the Sweet patent is limited to oscillography--the recording of waveforms of electrical signals. Appellants strongly contest this construction of the Sweet patent. While conceding that the illustrative embodiments of the patent depict only oscillographs, they argue that claims 13 and 334 of the Sweet patent, the claims in issue, disclose a method and apparatus for recording a wide variety of images, including alphanumeric characters.

In order to prevail on this issue, appellants, as we see it, must satisfactorily answer two related questions. First, if the Sweet patent contemplated calligraphy as well as oscillography and in application could compose characters as well as curved lines, why does not the patent say so directly and explain the process unambiguously? Why is this claim left unclear if it was intended? Second, if the Sweet patent claimed high speed character writing and needed no significant improvement or fundamental change to do so, why did Sweet join with Dr. Cumming, his supervisor, in subsequently filing a second, separate patent application disclosing a more advanced multiple ink jet system clearly designed for high speed character writing?

Appellants have failed to answer these questions satisfactorily and have failed to carry their burden of proof on the issues. The Mead DIJIT printer, which is based on the very patent application filed by Sweet with his supervisor after the original Sweet application was filed, is a substantial improvement over the invention disclosed in the Sweet patent.

Since we also agree with the District Court that the Lewis-Brown patent is invalid for obviousness, we affirm.

I. THE PATENTS IN ISSUE AND THE ACCUSED DEVICE

Many large and small steps of discovery in physics and electricity from the Nineteenth Century forward have led to the modern practice of ink jet printing. These steps include Michael Faraday's experiments in 1831 in electromagnetic induction and Lord Rayleigh's discovery in 1897 that a stream of liquid droplets issuing from a nozzel can be made uniform in size and spacing by applying cyclic energy or vibrations to the droplets as they form at the nozzle orifice.

C.R. Winston was the first person to incorporate these longstanding discoveries into a commercially successful ink jet printer. His device, patented in 1962 (U.S. Patent No. 3,060,429) and marketed as the Teletype Inktronic, recorded information by forming a single-file train of ink droplets that were imprinted with a uniform electrical charge. The droplets were directed between a pair of "deflection plates" which received variable electrical signals from an input source. The interaction of the constant charge placed on the droplet with the variable charge placed on the deflection plates caused the droplets to be deflected, the magnitude and direction of deflection depending on the charge placed on the deflection plates. After leaving the deflection field, the droplets travelled along their programed trajectories to a recording medium, usually paper, that passed under the nozzle and deflecting apparatus.

Although the Winston device was able to print alphanumeric characters, it could operate only at relatively slow speeds, since only one droplet could be present in the deflection field at a time. If a trailing droplet entered the deflection field before the leading droplet exited, the trailing droplet would be deflected along the identical trajectory as the leading droplet, since all droplets received identical charges and their trajectories were determined by the charge placed on the deflection plates when the droplets were present in the deflection field.

In 1961 Richard G. Sweet, then an electrical engineer conducting research at Stanford University, received as a gift an aquarium equipped with a device that sent bubbles through the water to provide oxygen to the inhabitants of the aquarium. Sweet's scientific curiosity got the best of him, and he set out to determine how oxygen was transferred from the bubbles to the water. The pursuit of this inquiry led eventually to U.S. Patent No. 3,596,275, the Sweet patent, by all accounts a monumental step in the march toward commercial application of ink jet printing concepts.

The apparatus disclosed in Sweet's patent controlled the trajectory of the ink droplets by maintaining a constant voltage on the deflection plates and placing a variable voltage on the droplets, the opposite of the system used by Winston. Attached as Appendix A is one of the illustrative embodiments included in the Sweet patent. This arrangement permitted Sweet's device to operate much faster than Winston's; several droplets, each encoded with its own "signal" charge, could be present in the constantly-charged deflection field at the same time. Indeed, the operational speed of Sweet's device was limited only by the rate at which the droplets could be formed. Sweet found that he could generate as many as 120,000 droplets per second, resulting in a recording speed hundreds of times faster than the Winston device could achieve.

The parties have paid particular attention to the direction of deflection disclosed in the Sweet patent.5 Claim 1 of the patent teaches that "charged droplets are deflected laterally" while claim 33 discloses an apparatus for causing charged droplets "to follow trajectories that are a function of the amount of charge on said droplets."6 The District Court concluded that these claims are ambiguous in that the terms "laterally" and "trajectories" are meaningless without a reference plane. 521 F. Supp. at 174. The District Court resolved this perceived ambiguity by examining the illustrative embodiments, specifications, and the prosecution history of the patent and, on the basis of this examination, concluded that the claims disclose only deflection in a direction transverse to the movement of the recording medium relative to the nozzle. Id.

Appellants vigorously dispute this conclusion. They argue that the term "laterally" in claim 1 means "laterally to the initial droplet direction" and that the term "trajectories" in claim 33 does not address the direction of deflection.

Upon hearing of Sweet's work in late 1962, Arling D. Brown and Arthur M. Lewis of Brush Instruments added a character or function generator to Sweet's droplet-printing device thereby enabling the device to print alphanumeric characters. Attached as Appendix B is one of the illustrative embodiments included in the Lewis-Brown patent. Character generators are devices which are able to "store" letters and numerals and to send to the charging electrodes voltage signals corresponding to the letters and numerals. Messages, consisting of a predetermined sequence of letters and numerals, are sent to the character generator which translates the message to a series of voltage signals and then relays the translated message to the charging electrodes which then charge the appropriate droplets to the appropriate level thereby causing the message to be printed.

The first Lewis-Brown application was filed on September 25, 1964.7 Two further applications were filed during 1965; both contained essentially the same disclosures as the initial application. After a brief prosecution, the Lewis-Brown patent issued on January 10, 1967, and A.B. Dick became the exclusive licensee under it.

After the Lewis-Brown patent issued, Sweet, whose application was pending, copied claims from the Lewis-Brown patent in order to provoke an interference proceeding between himself and Lewis and Brown. After a lengthy interference process, the Patent Office Board of Interferences resolved the question of priority of invention in favor of the Lewis-Brown patent, holding that the Sweet patent application did not disclose a character generator and therefore Sweet could not claim it in hindsight in combination with the droplet-printing apparatus which his application did disclose.8

In May, 1963, during the course of Sweet's work on the original Sweet device, Dr. Cumming, Sweet's supervisor, directed Sweet's attention to a recording system that used a dense array of wire styluses situated above and across a specially treated paper. Trial Transcript at 100. When electrical potential was applied to an individual stylus in the array, current passed from the stylus to the specially treated paper, which produced marks that combined to form the desired character. Dr. Cumming realized that the wire styluses could be replaced with ink jets, such as those used in the original Sweet device, thereby dramatically improving the speed of the system.

After several brainstorming sessions, Sweet and Cumming proposed that Stanford's research laboratory develop this idea, but the research proposal was rejected. Undaunted, Sweet approached Honeywell with the idea, since Honeywell had helped Sweet in his initial patent application. On March 25, 1964, Honeywell, acting on behalf of Sweet and Cumming, filed a patent application that disclosed a device utilizing the multiple jet configuration. The original Sweet-Cumming application disclosed only a facsimile recorder, but an amendment was later filed that disclosed a character generator which enabled the Sweet-Cumming device to print characters. The Sweet-Cumming patent, No. 3,373,437, was issued on March 12, 1968. Attached as Appendix C is an illustration of the Sweet-Cumming invention.

The invention disclosed in the Sweet-Cumming patent is described in the patent as "an improvement over the recording system disclosed and claimed" in the Sweet patent. More specifically, the Sweet-Cumming device is described as "utilizing the principles" of the Sweet patent but being "particularly adapted for presenting a display image as distinguished from a line trace."

The record is silent as to why Sweet and Cumming elected to file an entirely separate patent application that expressly claimed the ability to print predetermined characters with ink droplets. The trial court did not expressly draw any inference from this fact, although it did note that the Sweet-Cumming patent acknowledges an indebtedness to Sweet's prior work. 521 F. Supp. at 177.

Neither Sweet nor Cumming ever actually built a machine based on the techniques and apparatus disclosed in the Sweet-Cumming patent. Shortly after the patent issued, Sweet assigned his interest to A.B. Dick while Cumming assigned his interest to Mead.

In 1967 Mead began in earnest a research and development effort which culminated in the Mead DIJIT printer, the accused device. Attached as Appendix D is an illustration of the DIJIT. While the DIJIT printer embodies virtually all of the principles and techniques disclosed in the Sweet-Cumming patent, Mead's research was independent of the research conducted by Sweet and Cumming.9 Shortly after the Sweet-Cumming patent issued, Mead purchased Cumming's interest: the DIJIT can thus be thought of as an application of the Sweet-Cumming patent.10

The DIJIT printer contains a row of orifices situated above and across the recording medium so that the recording medium travels underneath the orifice array in a direction perpendicular to the array. A continuous stream or jet of ink issues from each orifice simultaneously and each jet is broken into evenly sized and spaced droplets by a process similar to that used in the Sweet and Lewis-Brown devices. Unlike the other devices, though, all of the droplets that print in the DIJIT are essentially undeflected: images are formed by deflecting into a collector or interceptor selected droplets from selected orifices and permitting the remaining droplets to fall undeflected onto the recording medium. Droplet interception is controlled by a computer which can be programmed to send instructions appropriate for printing alphanumeric characters as well as facsimile reproductions of video signals.

The process by which droplets are deflected in the DIJIT is similar to that used in the Sweet and Lewis-Brown devices: each droplet passes near a charging electrode and then through a deflection field. The interaction of the charge placed on the droplet with the electrostatic field causes the droplet to change its course.

However, since the sole purpose of deflection in the DIJIT is to prevent selected droplets from reaching the recording medium, all deflected droplets receive a virtually identical charge. Thus, only two charge levels are necessary in the DIJIT: one charge level (one hundred volts) deflects the droplets to the catcher while the other charge level (zero volts) permits the droplets to travel undeflected to the recording medium. Deflection is used in the Sweet and Lewis-Brown devices to direct droplets to various locations on the recording medium as well as to direct droplets to a collector. To accomplish this result, multiple charge levels are used in those devices.

Moreover, deflection in the DIJIT is longitudinal with respect to the movement of the recording medium, that is, the ink droplets are deflected in the same direction that the recording medium travels. Longitudinal deflection is used since all deflected droplets are intended to be intercepted and the interceptor or collector is situated across the recording medium, parallel to the orifice array. As noted earlier, the District Court concluded that the Sweet device discloses only transverse deflection.

The infringement issue in this case turns on the application of the "doctrine of equivalents," a canon of patent construction sometimes used in infringement cases to determine the scope of patent claims. This doctrine holds that a device is in infringement even if it does not "literally" infringe the patent "if it performs substantially the same function in substantially the same way to obtain the same result." Sanitary Refrigerator Co. v. Winters, 280 U.S. 30, 42, 50 S. Ct. 9, 13, 74 L. Ed. 147 (1929). However, as Justice Jackson noted in Graver Mfg. Co. v. Linde Co., 339 U.S. 605, 608-609, 70 S. Ct. 854, 856, 94 L. Ed. 1097 (1950), the doctrine of equivalents is a two-edged sword: "where a device is so far changed in principle from a patented article that it performs the same or similar function in a substantially different way, but nevertheless falls within the literal words of the claim, the doctrine of equivalents may be used to restrict the claim and defeat the patentee's action for infringement." The doctrine thus may be applied in favor of as well as against a patentee. It can be used to hold a device in infringement even though the device does not literally infringe the patent claims; it can be used to hold a device not in infringement even though the device falls within the literal words of the claim.

The general purpose of the doctrine is to encourage patentable inventions as well as genuine improvements of existing inventions but to discourage what Justice Jackson in Graver calls "unimportant and insubstantial changes and substitutions in the patent which, though adding nothing, would be enough to take the copied matter outside the claim, and hence outside the reach of the law" in the absence of a flexible doctrine designed to protect the patentee from insubstantial changes. 339 U.S. at 607, 70 S.Ct. at 855.

The doctrine "is not the prisoner of a formula," and "a finding of equivalents is a determination of fact.... It is to be decided by the trial court and that court's decision, under general principles of appellate review, should not be disturbed unless clearly erroneous." 339 U.S. at 609-10, 70 S.Ct. at 856-57. Judge Hand referred to the doctrine as "in misericordiam to relieve those who have failed to express their complete meaning" in the patent and stated that its application in "each case is inevitably a matter of degree." Claude Neon Lights, Inc. v. E. Machlett's Sons, 36 F.2d 574, 576 (2d Cir.1929). Many courts have said that a patent should be accorded a wider range of equivalents if it represents a major step in the field such that it may be called a "pioneer patent." See, e.g., Boyden Power-Brake Co. v. Westinghouse, 170 U.S. 537, 569, 18 S. Ct. 707, 723, 42 L. Ed. 1136 (1898); Acme Highway Product Corp. v. D.S. Brown Co., 473 F.2d 849 (6th Cir.1973).

Appellants direct us to the following passage in Graver as controlling: "If accused matter falls clearly within the claim, infringement is made out and that is the end of it." 339 U.S. at 607, 70 S. Ct. at 855. This so-called doctrine of "literal infringement" continues to live in the cases despite repeated pronouncements that infringement is not a mere matter of words. See CHISUM, Patents Sec. 18.04 (1983) and cases cited therein. Indeed, in Graver the Supreme Court indicates that the doctrine of literal infringement is not to be taken literally: two paragraphs after announcing that doctrine the Court recognizes that the doctrine of equivalents applies even when the accused device "falls within the literal words of the claim." 339 U.S. at 608-09, 70 S.Ct. at 856.

Courts, however unfortunately, continue to pay lip service to the doctrine of literal infringement as though it were the rule in Shelley's Case. Perhaps we are embarrassed to expose the "wholesale realism" which controls many infringement cases, see 339 U.S. at 608, 70 S.Ct. at 856, and we choose instead to present the facade of precision and certainty which attends the doctrine of literal infringement.

As Judge Hand pointed out in Claude Neon Lights, there is no doubt that the doctrine of equivalents "violates in theory the underlying and necessary principle that the disclosure is open to the public save as the claim forbids, and that it is the claim and that alone which measures the monopoly." 36 F.2d at 575 (citations omitted). As Hand stated, the doctrine is an "anomaly" needed to soften and render flexible the rule of literal infringement.

The "wholesale realism" of the doctrine of equivalents means that precedents have "little or no value" since a decision is "bound to have an arbitrary color, as in all close cases of interpretation, and it is difficult to give it greater authority than an appeal to the sympathetic understanding of an impartial reader." Claude Neon Lights, 36 F.2d at 576. This result will not set well with those who demand rules in this area and treat patent law problems as questions of semantics. Settled expectations are an important goal, but the survival of the doctrine of equivalents in this specialized field demonstrates that "unsparing logic" must be tempered with "wholesale realism." See Royal Typewriter Co. v. Remington Rand, Inc., 168 F.2d 691, 692 (2d Cir.1948).

With these general legal principles in mind, we now turn to the arguments advanced by appellants to support their assertion that the DIJIT printer infringes the Sweet patent.

Appellant's principal argument is that the District Court improperly circumscribed the Sweet patent by its illustrative embodiments: a patent is measured, appellants argue, by its claims and not its specifications or embodiments. Claims 1 and 33 of the Sweet patent, properly understood, extend far beyond oscillography, presumably to all ink jet printers that use the Sweet technique of selectively charging ink droplets. They argue that Sweet's invention consists of five elements--an ink jet generator, an ink jet stimulator, a droplet-charging mechanism, a deflecting mechanism, and a droplet-collection mechanism--and that the DIJIT printer contains and uses all five of these elements "in precisely the manner specified" in the Sweet patent. Appellant's Brief at 46. Appellants conclude therefore that the Sweet claims "read on" the DIJIT printer and that infringement follows as a matter of law.

Appellants are wrong on two counts. First, a finding of literal infringement is not the end of an infringement inquiry, despite numerous judicial statements to the contrary.11 As this Court stated in National Rolled Thread Die Co. v. E.W. Ferry Screw Products, Inc., 541 F.2d 593, 600 (6th Cir.1976):

Infringement should not be determined by a mere decision that the terms of a claim of a valid patent are applicable to the defendant's device. Two things are not necessarily similar in a practical sense because the same words are applicable to each....

There is no magic in a name, nor in a claim; that the words preferred by a patentee to define his invention apply literally to another's device suggests, but does not prove, infringement; there must be a substantial identity, to justify that conclusion of law.

(citations omitted).

Put differently, the doctrine of literal infringement is informed by the doctrine of equivalents: in infringement actions the court must consider the substance of the invention along with the form of the claims. Thus, even were we to determine that the DIJIT printer "reads on" claims 1 and 33 of the Sweet patent,12 we would still be required to consider whether those claims should be restricted by the doctrine of equivalents.

When we make this inquiry, we see that appellants mischaracterize the relation between the DIJIT printer and the Sweet patent: The DIJIT printer does not use the elements disclosed in the Sweet patent in "precisely the manner" specified in the patent. Rather, the DIJIT printer--to use the time-worn and conclusory formulation of the doctrine of equivalents--uses the elements disclosed in the Sweet patent in a substantially different way to achieve a substantially different result.

The Sweet patent does not contemplate a high speed character printer with coordinated multiple jets and a deflection system whereby all charged droplets are deflected into a collector and uncharged droplets are deposited on the recording medium to form the desired characters. The DIJIT printer, quite simply, is a more sophisticated device, embodying inventive insights not part of the Sweet patent. While it does rely on Sweet's fundamental concept of ink jet charging and deflection, it also incorporates those concepts of the Sweet-Cumming patent which Sweet himself described in that later patent as improvements over his original device. Those concepts include the coordination of multiple jets, interception for creating an apparent discontinuity in the image, and a charging and deflection system whereby the final picture is not characteristic of the charging signals.

The District Court thus correctly concluded that the Sweet claims must be restricted so as not to "read on" the DIJIT printer. This interpretation of the claims results from the application of the doctrine of equivalents, not from a mistaken reliance on the illustrative embodiments of the patent, as appellants believe.

In reaching this conclusion, we are heavily influenced by the fact that after his initial research, Sweet joined with Cumming to file and obtain a second patent disclosing, in essence, the Mead DIJIT printer. The fact that Sweet filed for a new patent with Cumming represents a contemporaneous implied admission that the Sweet-Cumming complex system of charging an array of jets and using uncharged droplets for writing constitutes a significant advance on Sweet's original work. This is the only reasonable explanation for Sweet's conduct. Appellants cannot deny that a pioneer in the field--Sweet himself--clearly appears to have believed in the early 1960s that the Sweet-Cumming patent--and thus the DIJIT printer--was a significant advance and should be independently patented. Actions sometimes testify more truly about a man's purpose than his words. We are more persuaded by Sweet's conduct at invention time than by his testimony at trial when his interests may have changed because of the way he had sold his patent rights.

Although the Sweet-Cumming patent may not represent the basic advance, the quantum leap in insight, represented by the Sweet patent, it is a useful improvement not self-evident from Sweet's first discovery. Sweet should not be relieved of his failure to express this separate advance in his original patent by Judge Hand's "misericordiam." The two insights, separately patented, should not be conflated into one, although the first was the more pioneering effort.

The District Court held the Lewis-Brown patent invalid under 35 U.S.C. Sec. 10313 on the grounds that in light of the prior art the subject matter of the patent taken as a whole would have been obvious to a person of ordinary skill in the art at the time the invention was conceived. 521 F. Supp. at 185. Appellants argue that in reaching this conclusion the District Court disregarded the presumption created by 35 U.S.C. Sec. 28214 that a patent is valid over the prior art considered by the Patent Office. Since the District Court considered the same prior art as did the Patent Office, appellants argue that the District Court was bound by the Patent Office's determination of validity.

Appellants misapprehend the nature of the presumption created by 35 U.S.C. Sec. 282.15 As this Court noted in Universal Electric Co. v. A.O. Smith Corp., 643 F.2d 1240, 1245 (6th Cir.1981), the presumption merely places the burden of proof of invalidity on the party raising the issue and it has no independent evidentiary value. It does not prevent a court "save in cases of clearest error" from invalidating a patent on the prior art the Patent Office considered, as appellants would have us hold.

In Graham v. John Deer Co., 383 U.S. 1, 17, 86 S. Ct. 684, 693, 15 L. Ed. 2d 545 (1966), the Supreme Court established the factual inquiries that must be made when obviousness is alleged: "the scope and content of the prior art are to be determined; differences between the prior art and the claims at issue are to be ascertained; and the level of ordinary skill in the pertinent art resolved." Once these factual determinations are made, the court must then decide as a matter of law whether the subject matter of the patent in issue is obvious. The findings of the District Court on the factual issues are subject to the clearly erroneous standard. Universal Electric, 643 F.2d at 1247 and cases cited therein.

Regarding the scope and content of the prior art, the District Court found that the use of character generators in connection with ink drop printing was known in the art prior to Lewis' and Brown's work: the Winston patent discloses a character generator of sorts and Sweet used a sine-wave generator, a kind of function generator, with his machine. 521 F. Supp. at 183.

Regarding the differences between the prior art and the claims at issue, the District Court found that the Lewis-Brown patent discloses, in essence, a function generator like that used in the Winston device attached to an ink droplet charging apparatus like that used in the Sweet device. 521 F. Supp. at 176.

Regarding the level of ordinary skill in the pertinent art at the time of Lewis' and Brown's work, the District Court found that artisans in the field of ink jet printing were familiar with the use of character generators. 521 F. Supp. at 183. The court also cited the short time that passed between Sweet's invention and the Lewis-Brown invention as evidence that the use of a character generator with the Sweet device would have been obvious to a person of ordinary skill in the art. Id.

Based on our review of the record, we cannot say that any of these findings are erroneous. A consideration of these findings leads to but one conclusion: attaching a character generator to the Sweet device--which appellants concede is the Lewis-Brown "invention", Appellant's Brief at 19,--would have been obvious to a person of ordinary skill in the art at the time of Lewis' and Brown's work.

Since we agree with the District Court that the Lewis-Brown patent is invalid for obviousness, we need not consider the District Court's conclusion that the DIJIT printer does not infringe this patent.16 Universal Electric, 643 F.2d at 1241.

Accordingly, the judgment of the District Court is affirmed.

APPENDIX "A"

NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLEAPPENDIX "B"

NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLEAPPENDIX "C"

NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLEAPPENDIX "D"

NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLE