*634 MEMORANDUM OF DECISION

ZOBEL, District Judge.

Candela Laser Corporation (“Candela”), the holder of several patents in the field of laser technology, brought this action against Cynosure, Inc. (“Cynosure”) and others for patent infringement, breach of contract, trade secret misappropriation and unfair competition. 1 On defendant’s motion, I ordered that the issue of liability be tried separately from that of damages. The parties subsequently narrowed the trial to infringement of U.S. Patent Nos. 5,066,293 (the ’293 patent) and 5,109,387 (the ’387 patent) and a counterclaim of invalidity as to the ’387. This memorandum shall constitute my findings of fact and conclusions of law on those issues.

I. BACKGROUND

“Laser” is an acronym for Light Amplification by the Stimulated Emission of Radiation. All conventional lasers have certain basic components in common. A quantity of material, called the excitation medium, is exposed to an energy source that causes the medium’s molecules to emit photons, or light. The light resonates throughout the medium, usually by means of reflection off a combination of lenses and mirrors, called the “optics,” becomes amplified as it travels back and forth and stimulates additional molecules. When the light reaches a critical level of intensity, it leaves the medium in the form of a laser beam.

In a dye laser, the excitation medium consists of dye molecules suspended in a liquid. A flashing light, the flashlamp, serves as the energy source, exciting the dye molecules and causing them to emit photons. A “dye cell” holds the dye solution in proximity to the optics. The amplified light finally passes through a specially coated mirror as a useful laser beam.

In a conventional dye laser most of the amplified light is designed to travel back and forth through the dye cell in precise, coherent paths, or “modes.” Light travelling through the cell in a straight line is said to be travelling “coaxially.” The medium does not produce exclusively coaxial light rays, however, because the flashlamp energy creates certain thermal distortions that change the course of the photons, directing them out of their coherent paths. In a conventional dye laser, some of the light traveling noncoaxially is not reflected back into the cell, but out of the cell at various angles. As a result, the non-coaxial light loses energy relative to the coaxial rays which further gain energy by being reflected off the mirrors at each end of the dye cell.

The patents at issue relate to the use of lasers for selective photothermolysis of biological tissue. In this technique, targeted tissue, such as a port wine birthmark on human skin, is heated by the laser light. The color of the light, duration of the laser pulse and intensity of the laser beam must all be carefully regulated in order to damage the tissue causing the skin’s dark pigment, but leave other tissue unaffected. Dye lasers are well suited to selective photothermolysis because the choice of dye can determine the wavelength of the light and because dye lasers can provide high output energy with controlled pulse duration. Conventional dye laser technology, however, created a laser pulse lasting only a few microseconds, whereas optimal photothermolysis requires nearly a millisecond pulse. The ’293 patent addresses the problem of pulse duration and wavelength. The ’387 patent pertains to a second problem, the creation of byproducts in the dye solution and its consequent dilution.

II. U.S. Patent No. 5,066,293

While employed at Candela Laser, Dr. Horace Furumoto developed a dye laser with a pulse duration approaching one millisecond. He realized that light amplification could in *635 fact be achieved with the rays of light reflected off the cell walls. He did so by means of a dye cell with apertures at either end. Optics at each end of the cell image each aperture on itself, thus capturing the light coming from the aperture and returning it to the cell for amplification. In this manner, light is available for amplification whether it travelled coaxially within the cell, or was reflected off the cell walls. Dr. Furumoto patented his method in 1991. He assigned the patent to Candela which marketed an embodiment of the claimed method as the “SPTL-1,” a dye laser used for selective photo thermolysis.

Dr. Furumoto and Harry Ceccon had founded Candela in 1970. At that time it primarily developed lasers for scientific purposes. In the 1980s, however, Candela began developing lasers for medical applications. Dr. Furumoto led the company until 1991, when he was terminated. He then purchased the assets of Candela’s scientific laser business and founded a new company, Cynosure, Inc. The parties agreed that Cynosure would not compete with Candela in the medical laser business before April 15, 1992, but that it would be permitted to engage in medical laser research and development during that time. When the noncom-petition agreement expired, Cynosure introduced the PhotoGenica V laser, which is now accused of infringing the ’293 patent.

The PhotoGenica V is a flashlamp-excited dye laser with a 24-inch dye cell. Approximately 2.5 centimeters from each end of the dye cell is a flat mirror, one of which is coated to allow the laser beam to exit the cell. According to Dr. Furumoto, the PhotoGenica V, unlike the patented method, uses coaxial light in its lasing process. It takes advantage of improvements in the dye medium and flashlamp technology to create the longer pulses formerly impossible in conventional dye lasers.

A. Claim Interpretation

Plaintiff alleges that the PhotoGenica V laser nevertheless infringes on Claims 1 and 5 of the ’293 patent. Claim 1 is for:

A method of phothermolysis [sic] of biological tissue comprising:

energizing a liquid dye medium in a dye cell including an aperture at each end thereof to an energy level at which the medium has net optical gain thereby producing an output beam from said dye cell;

from each end of the cell returning into the cell substantially all light other than the output beam, within a wavelength band emanating from the cell, such that the light traverses the cell with internal reflections off cell walls and the cell amplifies the light, the light being returned including other than spatially coherent light traveling coaxially in the cell, the output beam being of at least one hundred microsecond duration and at least one-tenth joule; and

illuminating the biological tissue with the output beam from the cell. (Emphasis added.)

Claim 5 substitutes “subcutaneous vascular lesions” for “biological tissue” but is identical in all other respects.

The parties do not dispute that the PhotoGenica V laser embodies a method of photothermolysis of biological tissue or subcutaneous vascular lesions in which a liquid dye medium is energized in a dye cell with an aperture at each end thereby producing an output beam. The controversy centers around the meaning and application of the claims’ second paragraph. Specifically, the parties differ regarding the meaning of “the light” in the second and third of the three steps delineated above.

Plaintiff contends that “substantially all the light” refers to the type of light, i.e., both coaxial and noncoaxial light, rather than some quantity of light. Thus, both coaxial and noncoaxial light from each end returns into the cell, but only some of that light reflects off cell walls to become amplified. Plaintiff claims that the PhotoGenica V returns both coaxial and noncoaxial light to the cell and uses some of each in its lasing process. It does not claim that the PhotoGenica Y reflects and amplifies “substantially all” the returned light.

*636 Defendant reads this section to describe a method in which “substantially all light” returns into the cell and reflects off cell walls and is amplified for use as a laser. It maintains that the PhotoGenica V does not infringe because most of the light it amplifies is coaxial, i.e., it resonates back and forth in a straight line within the cell; only a very small proportion traverses the cell with internal reflections off cell walls.

A determination of patent infringement involves a two-step analysis; the court first interprets the language of the claim and then compares the claims to the accused device. Standard Oil Co. v. American Cyan-amid Co., 774 F.2d 448, 452 (Fed.Cir.1985). When the meaning of key terms is disputed, courts may consider extrinsic evidence, including testimony of witnesses, as well as reference to the specification, the prosecution history and prior art. Tandon Corp. v. I.T.C., 831 F.2d 1017, 1021 (Fed.Cir.1987). The claims, not the specification or the embodiments, define the invention. Uniroyal, Inc. v. Rudkin-Wiley Corp., 837 F.2d 1044, 1053 (Fed.Cir.1988) (citing Loctite Corp. v. Ultraseal Ltd., 781 F.2d 861, 870 (Fed.Cir.1985)).

The claims’ clear meaning is that substantially all the light from each end of the dye cell is returned to the cell and substantially all the light is amplified by travelling through the dye cell with reflections off the cell walls. This interpretation comports with credible witness testimony, the specification, the prosecution history and the prior art. Plaintiffs own witness, Dr. Ralph Jacobs, testified that one with ordinary skill in the art would understand the phrase “substantially all light other than the output beam” to mean all light, both coaxial and noneoaxial. He explained that the remainder of the phrase referred to that coaxial and noneoaxial light, which is then amplified by travelling with internal reflections off cell walls. Dr. Jacobs did not assert that one ordinarily skilled in the art would read the phrase to include a method by which only a portion of the coaxial and noneoaxial light is internally reflected and amplified.

The patent specification also supports this reading; it stresses the significance of returning both coaxial and noneoaxial light into the cell but at no point refers to only a portion of the returned light being amplified and reflecting off cell walls. Rather, when it describes the invention’s operation, it states:

The light emanating from the cell is collected by the optical system and reflected back into the cell. The light traverses the cell in a number of total internal reflections off other cell walls. The dye solution in an excited state amplifies the light rays traversing the cell.

If the specification was referring to some part of the light emanating from the cell when it addressed reflection and amplification, it could have easily said so. It therefore must contemplate that all the returned light is also internally reflected and amplified.

The ’293 patent’s lengthy prosecution history provides several clues to the meaning of the claims, and to their distinction over prior art. The Patent and Trademark Office (“PTO”) initially rejected the ’293 application as being unpatentable over the prior art. A 1983 publication of R. Anderson 2 described the use of dye lasers for photothermolysis. In 1971, J.B. Marling 3 published his invention of a conventional dye laser which had some internally reflected light. According to the PTO, it would have been an obvious development to combine the two references to include Marling’s dye laser into Anderson’s invention.

In response, Dr. Furumoto amended his application to include the requirement that “the light traverses the cell with internal reflections off cell walls.” That new provision is the essence of this invention; it distinguishes the ’293 laser from the Marling laser which may have had some internally reflected light. Because the PTO rejected a patent that covered lasers with some inter *637 nally reflected light but allowed a patent with the stipulation that “the light” traverse the cell with internal reflections, the conclusion is inescapable that “the light” refers to “substantially all light.”

B. Infringement

To infringe the ’293 patent the PhotoGenica V must return, amplify and reflect off cell walls substantially all light emanating from the apertures. Plaintiffs infringement argument is based on the fact that the PhotoGenica V optics reflect substantially all' light back into the cell. Its analysis concentrates on the claims’ return step. In plaintiffs view, Claim 1 covers any optical configuration of mirrors and lenses that returns into the cell substantially all light other than the output beam. The PhotoGeniea V places mirrors approximately 2.5 centimeters from the ends of the dye cell. Because there is so little space between the end of the cell and the mirror, light does not escape from the cell. Accordingly, substantially all light, other than the output beam returns to the cell.

Once the PhotoGeniea Y satisfies the claims’ return step, plaintiff insists the infringement question is closed. After the optics return substantially all light to the dye cell, it follows that the returned light either reflects off cell walls and is amplified or it passes through the dye cell walls. Because Candela reads the patent not to require “substantially all” light to be reflected and amplified, it concludes that Cynosure must literally infringe.

Defendant differs regarding the return step, the reflecting step and the amplifying step. First, the patent specification describes a return process in which “an optical system at each end of the cell images each aperture upon itself.” Cynosure argues that regardless of whether its laser returns substantially all light to the dye cell, infringement is impossible unless the optics “reimage” the light. The PhotoGeniea V’s flat mirror does not reimage' the light, thus it does not read on the claim.

Next, according to Cynosure, the main difference between the ’293 patent and the PhotoGenica V is in the method of reflecting and amplifying the light. The ’293 method reimages the returning light in order to produce internal reflections. In fact, defendant claims that other than a theoretical coaxial beam through the middle of the dye cell, the ’293 method contemplates no coaxial light. The light travels through the cell with internal reflections and is amplified as it progresses.

In contrast, the PhotoGeniea V does not utilize internal reflections to amplify light. Instead, as in conventional lasers, light returned to the dye cell resonates coaxially through the dye cell between the mirrors. Defendant claims that it is impossible for internal reflections to be used in the amplification process of conventional lasers. According to defendant’s analysis, before the laser action starts, the flashlamp is fired and imparts energy into the dye molecules which then fluoresce and emit light in every direction. The coaxial light reflects off the flat mirrors and bounces back and forth between them. That resonating coaxial light increases in energy; as it travels through the dye cell it “gains” energy from the fluorescing dye molecules. Only that amplified light produces a useful laser beam.

The noncoaxial light, however, loses energy as it strikes the cell walls because part of the light passes through them. Consequently, coaxial light is more powerful than noncoaxial light. As lasing begins, the more powerful coaxial light “steals the gain” from any noncoaxial light. The noncoaxial light quickly dies out but coaxial light remains to resonate between the mirrors, be amplified, and finally leave the cell to form the laser beam.

Plaintiffs argument concerning the return step is faithful to the language of the claim. Any optical configuration that functions to return substantially all light, whether it “reimages” the light or not, reads on Claim 1. Although the specification of the ’293 patent discloses two curved optical systems, the language of Claim 1 is simply not limited to any particular optical configuration. That the PhotoGeniea V laser employs a physically different optical system to perform the claim limitation is therefore of no consequence.

But because substantially all light must be internally reflected and amplified, the deei *638 sion that the PhotoGenica Y reads on the return step does not settle the ultimate infringement question. In the PhotoGenica V, substantially all the light may be returned, but it is returned via a flat mirror. Thus most of the light remains coaxial. Any non-coaxial light returned to the cell either leaves through the cell walls or is changed to coaxial light due to the gain-stealing phenomenon.

In fact, plaintiff never claims that substantially all returned light in the PhotoGenica V is internally reflected and amplified. Both parties submitted extensive empirical test results to buttress their analyses. Not only are the results inconclusive; they were inconsistent through the course of the trial. More important, however, they succeed only in suggesting that some noncoaxial light exists in the PhotoGenica V system. As determined above, the question is not whether some noneoaxial light exists or if some light is internally reflected and amplified but whether substantially all light is internally reflected and amplified. Plaintiff has simply failed to demonstrate, or even to assert that is the case. Accordingly, I conclude that the PhotoGenica V does not infringe the ’293 patent.

III. U.S. Patent No. 5,109,387

Since the beginning of dye laser technology, “dye degradation” has been a persistent problem. The solution in a dye cell is composed of a small amount of dye dissolved in a large volume of liquid solvent. Each time the laser is fired, the flashlamp energizes the dye molecules, converting them to their “excited” state. When they revert to their “resting” state, the molecules emit the photons amplified in the lasing action. A small number of the excited molecules, however, do not shift back, but remain energized and are transformed or “degraded” into byproducts unfit for the lasing process because they may themselves absorb light of the same wavelength as that of the laser. Thus, accumulation of byproducts in the dye solution impairs laser performance; eventually, the laser can no longer realize selective photothermolysis. At that point the dye supply, the “reservoir,” must be replaced to rid the solution of the byproducts.

After several unsuccessful attempts to remedy the problem, Candela developed a “batch recovery” technique. The laser operated until dye degradation made it unusable. Lasing was suspended and the system attached to the recovery apparatus. A pump passed the solution through a charcoal filter which removed all dye molecules, both degraded and undegraded. The operator then added new, undegraded dye to the resulting clean solvent. Finally, the solution was reconnected to the laser and operation resumed. This process was not fully satisfactory, it required frequent filter replacement and suspension of lasing during solvent recovery.

In 1989, Dr. Gerome Garden and his colleagues claimed to have invented a system to address these problems. They devised a filter which acted as a “regeneration medium”; dye solution with byproducts would go into the filter and clean solution would come out. They attached the regeneration medium to a circulation system which operated continuously during the laser operation, eliminating the need to stop lasing during the filtering process.

These findings followed their work with a Candela laser at Northwestern University. Instead of discarding the filter in the batch recovery system after it became filled with dye, as Candela’s instructions recommended, they added more and more new dye to the system. Eventually, the filter became filled with dye, at which point dye began to leave it. They determined that the dye coming out of the filter was not degraded and that the “regenerated” solution could be recirculated and used to operate the dye laser. They patented the discovery in 1992. 4

The ’387 patent is for a dye laser system that improves performance by extending dye solution lifetime, dispensing with periodic dye replacement and ensuring consistent energy output.

Claim 21 teaches:

*639 A system for restoring dye solution degraded by use in dye laser comprising: a circulation path for conveying the dye solution;

a pump connected to said circulation path and operable to pump the dye solution; and

a regeneration medium adapted to absorb impurities and connected to said circulation path, said regeneration medium containing dye at a saturated concentration whereby the dye solution in said recirculation path may be regenerated by circulation through said regeneration medium.

Claim 25 covers:

A method of modifying a dye laser system having a circulation path for providing a dye solution to a laser cavity comprising the step of:

connecting a regeneration medium having a saturated concentration of dye solution to the circulation path so that the dye solution passes through both the laser cavity and the regeneration medium whereby the dye solution can be regenerated by circulation through the regeneration medium. 5

According to plaintiff, Dr. Garden patented a way to increase dye longevity by filling a charcoal filter with dye to a concentration in equilibrium with the dye in solution passing through the filter. The filter eliminates only degraded dye molecules from the solution and supplies new dye to the system in place of the removed byproducts. The dye solution leaving the filter is consequently at the desired dye concentration for lasing and the system is virtually free of byproducts. The solution is thereby “regenerated.” 6

Plaintiff asserts that Cynosure’s PhotoGenica V laser incorporates each element of Claim 21. It includes a system for restoring degraded dye solution made up of a pump and a “circulation path” for conveying the solution to the laser and the reservoir. Within the path is a charcoal filter that contains dye at a saturated concentration and changes degraded dye solution into usable dye solution, thus acting as a “regeneration medium.” Candela also claims the PhotoGenica V infringes on Claim 24 of- the patent because it implements a method of modifying a dye laser system to include the requisite circulation path and regeneration medium.

Defendant claims that the patent is invalid, and in any event is not infringed by the PhotoGenica V. The patent is invalid for the reasons stated herein; for the sake of completeness, however, I include an infringement analysis as well.

A. Validity

A patent is presumed to be valid, the party asserting invalidity carries the burden of proof by clear and convincing evidence. 35 U.S.C.A. § 282 (West 1984). In addition to the presumption of validity, a presumption of administrative correctness attaches to the decision by the PTO to issue a patent. This is particularly true when the party attacking validity relies on prior art that the PTO considered before .awarding the patent. Standard Mfg. Co. v. United States, 25 Cl.Ct. 1, 50 (1991).

1. Anticipation.

A patent is invalid if it is anticipated, i.e. disclosed elsewhere prior to the patent application. 35 U.S.C.A. § 102(a), (b), (g) (West 1984). The first step in an anticipation analysis is a determination that all the elements of the claimed invention were described in a single prior-art reference. In re Spada, 911 F.2d 705, 708 (Fed.Cir.1990). It is immaterial to the analysis that an earlier embodiment was “makeshift” as long as it is “complete,” that is, each element of the claim is found in the earlier device. Barmag Barmer Maschinenfabrik A.G. v. Murata Mach., Ltd., 731 F.2d 831, 838 (Fed.Cir.1984).

*640 The patent defines a regeneration medium as a saturated filter which retains degraded dye while good dye passes through. The parties differ on the meaning of “saturation.” According to plaintiffs expert, Dr. Lloyd Snyder, however, a system is in equilibrium whenever the solution entering the filter is in the same concentration as that leaving the filter even if only for a short period of time. At that point, the filter is not absorbing any molecules of dye; it is therefore “saturated.”

Defendant contends that if, as plaintiff insists, a “regeneration medium” is merely a saturated filter which removes byproducts and returns usable solution to the system, and a “saturated” filter is one that will hold a steady dye/solvent concentration for some relatively short period, then the patent was anticipated 'by Dr. Furumoto’s work while employed at Candela in 1982. He described his system in an instruction manual he sold and delivered tó NASA with a Candela LFDL-2 laser 7 . The manual illustrates a dye laser with a circulation path and pump connected to a charcoal filter. Page three explains the system’s operation:

We have found that poor dye laser performance can often be attributed to impure solvents. Running new solvent through the solvent reclamation filter will clean any alcohol, and consistent performance is assured. The filter can be used to remove dye from ■ the solvent. After several reclaiming cycles, the filter will load up and appear not to extract dye. This is especially true of pure alcohol solvents____ [W]e have discovered that even under the “loaded up” condition, the filter is still effective in removing degraded products and returning performance to the best recorded.

There are questions yet to be resolved when the dye laser solution is run under the “loaded up” condition. There must be a steady state, dye concentration once the filter has loaded up. This steady state dye concentration is not known but appears to be between 5 x 10 (-5) and 10 (-4) M____ A spectrophotometer reading should be taken of the dye solution to measure its concentration.

Dr. Furumoto sold variations of the LFDL laser with similar instruction manuals, in fact, the manual included with the LFDL-1, the laser Dr. Garden and his colleagues used at Northwestern, contained a similar passage:

Once the dye has degraded to an unusable level, the solution can be cleaned by sending it through the solvent recovery filter. The initial cleaning with a new filter will remove all trace of dye. Later batches will not clean up as well as the filter loads up. However, it is our experience that dye may leach through the filter but the degraded and harmful byproducts are removed. Replace filter only if laser output is not reproducible to its maximum output.

According to defendant, Dr. Furumoto’s earlier work and the manuals describing it explicitly disclosed every element of Claims 21 and 25.

Plaintiff insists that Cynosure’s evidence does not clearly, or convincingly demonstrate that Dr. Furumoto’s work embodied the claimed invention, that, it was not “on sale,” and that it was abandoned.

a. Section 102(a)

Section 102(a) prohibits a patent on a device that was “known or used by others in this country ... before the invention thereof by the applicant for patent, ...” 35 U.S.C.A. § 102(a) (West 1984). ■ “Prior knowledge and use by a single person is sufficient” to preclude patentability. Spalding & Evenflo Cos. v. Acushnet Co., 718 F.Supp. 1023, 1031-32 (D.Mass.1989) (quoting Coffin v. Ogden, 85 U.S. (18 Wall.) 120, 124, 21 L.Ed. 821 (1874)). According to Cynosure, Dr. Furumoto’s early work represents the patented method; therefore, the ’387 patent is invalid.

Plaintiff states that the LFDL-2 manual teaches that only a single “steady state” concentration exists, and fails to advise that the concentration can be varied by the quantity of dye loaded into the filter. That omission *641 is immaterial, however, because Claims 21 and 25 make no mention of a variable concentration and consequently such is not part of the invention.

Plaintiff also asserts that the manual’s reference to the use of a spectrophotometer to gauge dye concentration refers to the possible use of the loaded up filter with a batch dye circulator, not to its use “on line” for supplying solution at the required concentration for lasing. The Furumoto circulation system is “connected” to the laser, which is all the patent requires. Moreover, other parts of the manual discuss running the laser with or without the dye circulation assembly in operation “on line.”

Finally, in plaintiffs view, Dr. Furumoto and Candela envisioned and promoted their filter as a device for purifying solvent, not for regenerating the dye solution. What Dr. Furumoto may have thought of the system at the time of its discovery is, however, irrelevant to an anticipation analysis. W.L. Gore & Assoc. v. Garlock, Inc., 721 F.2d 1540, 1548 (Fed.Cir.1983), cert. denied, 469 U.S. 851, 105 S.Ct. 172, 83 L.Ed.2d 107 (1984) (if prior invention was consistent and reproducible it is irrelevant if those using it may not have appreciated its results).

The evidence demonstrates no discernab'le difference between the invention claimed in the ’387 patent and Dr. Furumoto’s earlier work; therefore, defendant has carried its burden of proving invalidity by clear and convincing evidence. See Avia Group Int'l v. L.A. Gear, 853 F.2d 1557, 1562 (Fed.Cir.1988). Moreover, the presumption of validity is readily overcome because the prior art on which defendant relies was not before the PTO. See Standard Mfg., 25 Cl.Ct. at 50.

b. Section 102(b)

A patent is invalid under § 102(b) if the claimed invention was in public use or on sale prior to the date of the application for patent. 35 U.S.C.A. § 102(b) (West 1984). In defendant’s view, Dr. Furumoto’s early work, coupled with the sale of the laser system and manual to NASA in 1982, invalidates the ’387 patent.

The statutory definition of “on sale” includes a “single sale or offer to sell.” Intel Corp. v. I.T.C., 946 F.2d 821, 830 (Fed.Cir.1991). As discussed above, the LFDL manuals fully describe the ’387 invention. Because there is no significant difference between the ’387’s method and that. of the device Candela sold to NASA in the early 1980s, the device was “on sale” before the patent application. Accordingly, the patent is invalid under § 102(b).

c. Section 102(g)

Finally, an invention may not be patented if it discloses a discovery made by another person who had not abandoned, suppressed or concealed it. 35 U.S.C.A. § 102(g) (West 1984). Defendant argues that because Dr. Furumoto sold his early work, it was not “abandoned.” Plaintiff points out that Dr. Furumoto failed to pursue his findings that a solvent recovery filter could remove byproducts and pass dye.

To defeat a charge of abandonment, an inventor need only make a complete discovery public or use it commercially. Pall Corp. v. Micron Separations, Inc., 792 F.Supp. 1298, 1306 (D.Mass.1992) (citing Friction Div. Prods. v. E.I. DuPont de Nemours & Co., 658 F.Supp. 998, 1014 (D.Del.1987), aff'd, 883 F.2d 1027 (Fed.Cir.1989)). Candela included the discovery in the NASA manual and in manuals for its other flashlamp-excited dye lasers. The evidence further demonstrates that although Dr. Furumoto did not envision wide applicability for his discovery, he did not conceal or suppress it. He consulted Dr. Michael Drake, a filtering systems scientist, regarding further applications of the carbon filter. Using other dye/solvent combinations, he continued to investigate his discovery but found only limited uses. Thus, he implemented the invention commercially and in the laboratory and did not abandon it. Section 102(g) consequently invalidates the patent.

2. Obviousness.

Even if not identical to earlier work, a patent is invalid if, in light of earlier disclosures, the device would have been' obvious to one ordinarily skilled in the art. 35 U.S.C.A. *642 § 103 (West 1984). Defendant challenges the validity of the ’387 patent on the grounds that it was obvious in view of Dr. Furumoto’s instruction manuals, as well as the work of other experts in filter technology.

Defendant first argues that there is no difference between the ’387 patent and the device disclosed in Furumoto instruction manuals. Furthermore, it says the ’387 method was obvious given U.S. Patent No. 4,364,015, and articles by V.A. Mostovnikov 8 and B.A. Knyazev. 9 According to defendant’s expert, Dr. Michael Drake, these references teach placing a filter in a closed circulation path with a dye laser so that they may remove contaminants from the solution passing through them. All filters absorb some undegraded dye from solution as well and therefore all filters will eventually reach some equihbrium with the surrounding solutions.

Instead of addressing the specifics of the prior art, plaintiff relies on the legal test for obviousness which requires the court to determine “(1) the scope and content of the prior art; (2) the level of ordinary skill in the pertinent art; (3) the differences between the elaims(s) of the patent in suit and the prior art; and also to review (4) the existence of objective criteria for nonobviousness, such as commercial success, long felt, but unsolved need, failure of others, acceptance by experts in the field and copying.” Standard Mfg. Co. v. United States, 25 Cl.Ct. 1, 51 (1991). According to plaintiff, the fourth factor is most important, and it is therefore significant that Dr. Furumoto persisted in experiments to improve dye longevity even after his findings regarding a “loaded up” filter, and that throughout the 1980s he wrote letters, memos and publications indicating solvent recovery was the best available answer to dye degradation.

I am persuaded, however, that the complete congruence of Dr. Furumoto’s earlier work and the ’387 patent compels a finding of obviousness. Further, the evidence shows that even the ’387 inventors understood no difference between Mostovnikov’s filter and their regeneration medium. Dr. Garden’s colleague, Dr. Abnoeal Bakus, noted that the regeneration medium would eventually fill up with byproducts and would have to be discarded. In deposition, Dr. Garden was asked to identify the difference between the Mostovnikov device and a regeneration medium. He testified that “if it was just a matter of getting rid of byproducts, then it would do the exact same thing as Mostovnikov’s filter which is — it would fill up with byproducts and then it would have to be discarded, and it doesn’t do that.” The fact that Mostovnikov’s filter was alumina rather than activated charcoal and that it was not placed in the circulation path also fails to create a distinction over the prior art. The ’387 patent itself allows that materials other than activated charcoal are suitable for a regeneration medium. The Mostovnikov article notes that his filter could be connected to a circulation path. The identity between the two devices, as tacitly recognized by the inventors, voids the patent.

Accordingly, I find the ’387 method obvious in light of the prior art.

3. Indefiniteness.

Section 112 of the patent statute requires that all claims be sufficiently specific to enable a person ordinarily skilled in the art to make and use the invention. 35 U.S.C.A. § 112 (West 1984). A determination of nonobviousness depends on whether those ordinarily skilled in the art would understand what is claimed. Amgen, Inc. v. Chugai Pharmaceutical Co., 927 F.2d 1200, 1217 (Fed.Cir.1991), cert. denied, —U.S.-, 112 S.Ct. 169, 116 L.Ed.2d 132. Defendant reasons that the parties’ greatly contrasting interpretations of the patent, particularly regarding the definitions of “saturation” and “regeneration medium,” demonstrate that the claims are indefinite. Plaintiff does not refute defendant’s specificity argument.

*643 The ’387 inventors themselves were vague as to how the regeneration medium worked, thus, I can only conclude that a person reasonably skilled in the art would not be certain as to the scope and meaning of the claims. The patent is therefore also void for indefiniteness.

4. Inequitable Conduct.

Cynosure claims that the patent is unenforceable because its inventors failed to disclose information to the PTO which may have had a bearing on its decision to issue the patent. Plaintiff argues that defendant has not met its burden to prove, by clear and convincing evidence, that the inventors acted with intent to deceive.

The mere allegation that documents of some materiality were not disclosed to the PTO does not establish inequitable conduct. FMC Corp. v. Manitowoc Co., 835 F.2d 1411, 1415 (Fed.Cir.1987). One who alleges this form of inequitable conduct must offer clear and convincing proof of (1) prior art or information that is material; (2) knowledge chargeable to applicant of that prior art or information and of its materiality; and (3) failure of the applicant to disclose the art or information resulting from an intent to mislead the PTO. Id. Defendant presents no evidence of intent to deceive and thus fails to carry its burden of proving inequitable conduct.

B. Infringement

According to defendant, the PhotoGenica V dye circulation system is comprised of two loops. Dye solution circulates continually from a main reservoir through the first loop. Some of that solution is diverted to a second loop wherein a filter eliminates byproducts and also some usable dye from solution. The solution, thus purified, then returns to the reservoir to be used again. Because some undegraded dye is removed from solution, the device is also equipped with a means to add dye, without which the dye concentration in the solution leaving the filter would become too low for lasing. The dye addition means operates approximately every 5,000 laser pulses.

Cynosure argues that this configuration does not infringe because it does not include a “regeneration medium.” Its filter removes degraded and undegraded dye from solution, but does not “regenerate” it as evidenced by the need for a frequently used dye addition means. Because, however, a regeneration medium as defined in the patent is merely a saturated filter which retains degraded dye while allowing good dye to pass through, and because defendant admits that its filter selectively removes undesirable molecules, the infringement question turns on whether the filter is “saturated.”

As noted above, the parties disagree as to the definition of “saturation.” Plaintiffs expert, Dr. Snyder, testified that a saturated filter is one that is not absorbing any molecules of dye at a given point in time. At that point, the concentration of solution entering the filter is the same as that leaving the filter; thus, the system is in equilibrium. The filter’s saturation point depends on time, temperature, and the amount of dye in the solution. The plaintiff does not specify the number of laser pulses through which the system must remain in equilibrium in order to fall under this definition of “saturation.”

In contrast, defendant’s expert, Dr. Michael Drake, defined a “saturated” filter as one that can retain no more dye under any conditions. , In his opinion, the solution is only at equilibrium when the dye going into the filter and that flowing out are at the same concentration for a prolonged period of time. A charcoal filter in the patented system can absorb about 70 grams of dye before becoming “saturated.” To Dr. Drake, only such a fully saturated filter could hold the system in equilibrium without the addition of fresh dye.

Dr. Drake points out that the PhotoGenica V filter holds only 6-8 grams of dye, not the 70 grams required for saturation. It is therefore capable of retaining more dye. To him, the presence of a dye addition means, which adds more dye than would be necessary if only degraded molecules were removed, proves that the filter is indeed removing good dye. Thus, according to defendant, the filter is not “saturated” as the patent requires.

*644 It is a well-established axiom in patent law that a patentee is free to be his or her own lexicographer. Hormone Research Found. v. Genentech, Inc., 904 F.2d 1558, 1563 (Fed.Cir.1990). Plaintiffs definition of “saturation” is technically consistent with the word’s ordinary use and not contradicted by the patent specification. First, and most significantly, although the patent itself and the specifications refer to the saturated filter as “no longer capable of removing dye from solution” the specifications also acknowledge that the medium absorbed only 6 grams of dye. Because 70 grams of dye must be absorbed before the filter will no longer hold more dye under any circumstances, the patentees plainly intended Dr. Snyder’s definition of saturation to control. Second, defendant’s own expert agreed that “saturated” need not always refer to a completely utilized absorption capacity. Finally, the PhotoGenica Vs dye addition means is not constantly in operation, rather, the system monitors the dye concentration and adds dye only when it becomes too low for effective lasing. Thus, for approximately 5,000 pulses, the solution is in equilibrium; it passes through a saturated filter which does not remove undegraded molecules. During that period, the filter acts as a regeneration medium, accepting unusable dye and returning usable dye solution. Although the prosecution history suggests the ’387 method allows lasing to continue for 4-600,000 pulses, the patent itself simply does not include such a requirement.

Thus, the Court reluctantly concludes Cynosure’s PhotoGenica V literally infringes the ’387 patent. The fact that the patent lacks a specific definition of saturation requires this troublesome result when in other circumstances infringement could have been avoided.

CONCLUSION

For the reasons above, Cynosure’s PhotoGenica V laser does not infringe on U.S. Patent No. 5,066,293. U.S. Patent No. 5,109,387 is invalid because it was anticipated, obvious and indefinite; however, defendant’s device literally infringes on the patent as written. Judgment may be entered accordingly.

. At various points before trial, plaintiff had alleged that defendant, and its president, Dr. Horace Furumoto, infringed three patents and violated state laws related to the use of methods and devices allegedly owned or controlled by Candela. Defendant requested a judicial declarations of noninfringement on those patents as well as one additional patent. All state law claims were dismissed or settled prior to trial, and Dr. Furumoto was dropped from the suit, leaving only those parties and claims discussed herein.

. R. Rox Anderson & John A. Parrish, "Selective Photothermolysis: Precise Microsurgery by Selective Absorption of Pulsed Radiation,” 220 Science 524 (Apr. 29, 1983).

. J.B. Marling et ah, “Long-Pulse Dye-Laser Emission Across the Visible Laser Spectrum,” I.E.E.E. Journal of Quantum Electronics, Oct. 1971, at 198-99.

. The inventors assigned the patent to the Gaelis Corporation, a named plaintiff in this action. Candela Laser Corporation owns exclusive rights to use the patent.

. Plaintiff had, before trial, accused defendant of infringing also Claims 22, 23 and 24, but then withdrew those claims. Only Claims 21 and 25 now remain in the case.

. Although the patent specifications and prosecution history suggest the inventors believed that the medium actually converted degraded dye molecules back to useful molecules, plaintiff has now disclaimed that interpretation.

. The parties agree that this manual was not before the PTO during the prosecution of the patent.

. V.A. Mostovnikov et al.. Recovery of Lasing Properties of Dye Solutions after their Photolysis, 6 Sov.J.Quantum Electronics 1126 (1976).

. B.A. Knyazev et al., Photochemical Effects in a High-power Flashlamp-pumped Laser Utilizing Solutions of Rhodamine 6G in Isopropyl Alcohol, 9 Sov.J.Quantum Electronics 1191 (1979).