dissenting.
The district court correctly construed the only claim limitation in dispute: “[T]he term ‘adjacent’ as used in the ’876 patent means ‘not far off.’ ” The court then found that the openings nearest the end walls of General American Transportation Corporation’s (GATC) accused railcar were “adjacent” to those walls and therefore literally infringed the claim. I see no error in claim construction or in the finding of infringement.
Claim limitations “will be given their ordinary meaning, unless it appears that the inventor used them differently.” ZMI Corp. v. Cardiac Resuscitator Corp., 844 F.2d 1576, 1579 (Fed.Cir.1988). This court’s citations to the claim language and specification to support a meaning other than the ordinary meaning of “adjacent” are misplaced. The claims do require openings adjacent to both end walls and side walls, but they do not define how close to the wall those openings must be to be considered adjacent. Nor do the claims require an opening to be closer to one wall than another to be considered adjacent to a particular wall. The only hint that an opening should be configured in a specific manner comes in a discussion of the inventor’s preferred embodiment. Although we construe claims in light of the specification, we do not import the specification wholesale into the claims. Intervet Am., Inc. v. Kee-Vet Lab., Inc., 887 F.2d 1050, 1053, 12 USPQ2d 1474, 1476 (Fed.Cir.1989). It is improper to read the inventor’s preferred arrangement as a claim limitation here.
This court’s de novo claim construction would necessarily prevent a finding of infringement whenever a vent is nearer a side wall than an end wall. This cannot be. For example, a vent only an inch away from an end wall will not be considered “adjacent” to the end wall as long as the vent is less than an inch from a side wall. Given the dimensions at issue, surely a vent such a short *773distance from an end wall is “adjacent” to that wall and must be infringing. The difference between this hypothetical and our case is only a matter of degree. The vent here was only 33 inches from the end wall of a railcar which is 812 inches long. The district court is in a much better position to evaluate this matter of degree.
The district court also correctly concluded that the claims were not obvious over the cited prior art, which it considered to include U.S. Patent No. 4,593,536 to Fink and Habel, the AFFX 2002 cryogenic railcar and variants thereof, U.S. Patent No. 3,561,226 to Rubin, and British Patent No. 399,678 to Kurth. It looked at the invention as whole and explicitly refused to “us[e] the patent or invention at issue as a guide to combine bits and pieces of earlier designs and then reach a conclusion of obviousness.” See Lindemann Maschinenfabrik GMBH v. American Hoist & Derrick Co., 730 F.2d 1452, 1462, 221 USPQ 481, 488 (Fed.Cir.1984) (“[T]he question is whether there is something in the prior art as a whole to suggest the desirability, and thus the obviousness, of making the combination”). The court concluded that “[tjhere is no clear suggestion of combining different pieces of prior art to produce the invention claimed by the ’876 Patent unless the patent is improperly used as a guide to make the combination.” Again, I find no error in the district court’s analysis of the prior art or in its finding that there was no motivation to combine the cited references. It was mindful that “[c]are must be taken to avoid hindsight reconstruction by using ‘the patent in suit as a guide through the maze of prior art references, combining the right references in the right way so as to achieve the result of the claims in suit.’ ” Grain Processing Corp. v. American Maize-Prods. Co., 840 F.2d 902, 907, 5 USPQ2d 1788, 1792 (Fed.Cir.1988) (citing Orthopedic Equip. Co. v. United States, 702 F.2d 1005, 1012, 217 USPQ 193, 199 (Fed.Cir.1983)).
This court relies principally on a combination of the AFFX 2002 railcar, which admittedly disclosed many of the elements of the ’876 patent claim, and the Kurth patent, which disclosed openings in a dry ice bunker adjacent to both side walls of a railcar. This combination requires the precise measure of hindsight that the district court refused to supply. The evidence at trial showed that no one really understood the gas flow cooling mechanism of the AFFX 2002 cryogenic design. The Fink patent, which is based on the AFFX 2002, disclosed a system in which CO2 gas supposedly exited the overhead bunker, flowed by convection down one side of the ear, traveled underneath the load in the car, and then recirculated back up the other side of the car. If that were true, a second row of vents on the opposite side of the car would not only have been unnecessary, but likely would have been detrimental to the cooling convection flow that was thought to exist. There was thus no motivation to include the Kurth patent which disclosed openings on the opposite side walls of a railcar bunker.
The 1933 Kurth patent also is directed to the loading and cradling of pre-made dry ice. There is no indication that “CO2 snow” which subsequently sublimates is ever created in the bunker shown in Kurth, and the testimony at trial was that Kurth disclosed a “cold plate” bunker which simply provides a large cold surface for cooling the air in the lading compartment below. The extremely small number of vents disclosed in Kurth was unlikely to provide an adequate cooling mechanism for the sublimation of a bank of “CO2 snow.” Given this testimony, I believe the district court could readily conclude that Kurth operated in a fundamentally different way from the other cryogenic cars cited as prior art. Without motivation to combine Kurth, the only reference which disclosed openings on both sides of a bunker, with the AFFX 2002, whose method of heat transfer was at best uncertain, GATC failed to prove by clear and convincing evidence that the ’876 patent is obvious.
The district court’s analysis is bolstered by the secondary considerations of nonobviousness. The court found that Cryo-Trans’ “ ’876 patent design has been accepted by the industry” and that it enjoys a high degree of commercial success. It also found that “[t]he ’876 patent design solved a long felt transportation problem for frozen food processors, who were searching for a viable solution since 1980 to the poor economic and *774maintenance viability problem of old railroad mechanical cars.”
The district court further found that in late 1990, GATC became aware of Cryo-Trans’ commercially successful cryogenic railcar patent, and “decided it had to promptly get into the cryogenic railcar market and copy Cryo-Trans’ high cube car.” To this end, GATC obtained a copy of the original drawings of the Cryo-Trans car from the company which manufactured the railcar shell, and then moved the shell to a second manufacturer which completed the interior of the car along with identical cars that were being completed for Cryo-Trans.
GATC admitted that its original prototype infringed the ’876 patent. After Cryo-Trans filed this suit, GATC merely blocked off several of the openings closest to the ends of its ears. At trial, one of GATC’s own engineers admitted that the accused device did not work any differently than the admittedly infringing first prototype. The trial court was surely correct when it concluded that “GATC copied the design of the Cryo-Trans high cube car after the commercial viability of the technology was proven by Cryo-Trans.” None of these findings are clearly erroneous.
The district court did not err in holding that the ’876 patent “represents a significant improvement from the relevant prior art and is not obvious from a review of the relevant prior art.” Eschewing the aid of hindsight to combine the cited references, and with appropriate deference to the district court’s factual findings, I would affirm the judgment in its entirety.