The Defendant-Appellant, Interface Architectural Resources, Inc. (“Interface”), seeks review of a final judgment by the United States District Court for the District of Maryland granting a motion by the Plaintiff-Appellees, Tate Access Floors, Inc. and Tate Access Floors Leasing, Inc. (together “Tate”) to preliminarily enjoin Interface from infringing claims 1-4 and 8-10 of United States Patent No. 4,625,491 (the “'491 patent”). Interface contends that the district court erred by granting the preliminary injunction because Tate failed to make the requisite showing of a reasonable likelihood of success on the merits of its infringement claim. Because the district court did not abuse its discretion in granting the preliminary injunction, we affirm.
I. BACKGROUND
A. The '191 Patent
Tate owns the '491 patent, which pertains to raised access flooring panels. '491 patent, col. 1,11. 6-9. Raised access flooring panels provide an elevated floor under which materials such as cables and pipes can be run. Id. at col. 1, 11. 15-19. The panels rest on pedestals. They are placed side-by-side to form an elevated floor. By lifting and removing certain panels, one can access the materials underneath the floor. Id. at col. 1,11. 30-32.
The panels addressed in the '491 patent have a laminated top surface. Id. at col. 1, 11. 57-63. The transparent laminate protects and covers a decorative top surface. Id. These decorative panels are formed by heat bonding a decorative top layer onto a number of layers of underlying kraft paper using resin. Id. The resulting edges of the laminated panels are brittle; without protection they chip, crack, and appear aesthetically unpleasing.
Prior to the claimed invention, the industry employed an add-on trim strip to protect the edges of the panels and to make them look better. Id. at col. 1, 11. 33-36. Such trim was problematic because it increased the cost of panels, and tended to crack and become loose, and to collect dirt and moisture. Id. at col. 1, 11. 42-46. The '491 patent addresses the problems associated with the edges of the panels without employing an add-on trim strip.
Instead of adding trim, the '491 patent solves the problem of brittle edges by removing the top decorative layer from the periphery of the panels to form an integral trim edge. Id. at col. 1,1. 67 to col. 2,1. 4. The two independent claims at issue, claims 1 and 8, read as follows:
1. A floor panel for elevated floors comprising a rectangular base structure adapted to be supported at its corners and providing a load surface operable to support loads thereon, a floor covering mounted on said load surface providing a decorative exposed surface layer on the side thereof opposite said load surface, said floor covering providing an inner body portion having an appearance contrasting with the appearance of said decorative surface layer, said floor covering providing a border along the edges of said panels *1361along which said decorative surface layer is removed to expose said inner body portion and thereby provide an integral contrasting border around said decorative surface layer.
8. A floor panel comprising a rectangular support structure providing a load surface operable to support loads thereon, a layered laminated floor covering mounted on said load surface providing a single visible decorative layer along the side of said floor covering also providing an inner layer contrasting with said decorative layer having a thickness substantially greater than the thickness of the decorative layer adjacent to said decorative layer along the side thereof facing said load surface, said floor covering providing a border extending along the edges of said panel along which the decorative layer is removed to expose said inner layer.
Id. at col. 4,1. 67 to col. 5,1.11 and col. 5,1. 37 to col. 6,1.11 (emphasis added).
The specification includes figures depicting an embodiment of the claimed floor panels. The periphery of the panels has a sloping, or beveled portion, as well as a flat, horizontal portion. For example, Figure 4 of the '491 patent depicts a panel with edges that appear as follows:
B. The Accused Product
Interface’s accused floor panels have a decorative top layer with a beveled edge, but no horizontal, single-layered portion. An example of the accused panels follows,
Interface contends that its panels fall outside the scope of the asserted claims of the '491 patent, as properly construed, by virtue of their simple beveled edges. It argues that the “border” limitations of claims 1 and 8 encompass only the horizontal portion depicted in Figure 4 of the '491 patent. Because Interface’s panels lack a horizontal portion formed of a single layer of material, and contain only a beveled edge, Interface maintains that its panels do not infringe. Tate contends that panels with beveled edges meet the “border” limitations notwithstanding the absence of a horizontal portion formed of a single layer of material.
C. The District Court’s Grant of Preliminary Injunction
The district court properly noted that pursuant to this court’s precedent interpreting 35 U.S.C. § 283, Tate’s entitlement to a preliminary injunction hinged on proof regarding four factors: (1) likely success on the merits, including proof of infringement and lack of a substantial question regarding Interface’s invalidity defense; (2) irreparable harm absent an injunction; (3) that the balance of hardships favors granting the injunction; and (4) that public policy favors granting the injunction. Tate Access Floors, Inc. v. Interface Architectural Res., Inc., 132 F.Supp.2d 365, 370 (D.Md.2001) (citing T.J. Smith & Nephew Ltd. v. Consol. Med. Equip., Inc., 821 F.2d 646, 647, 3 USPQ2d 1316, 1317 (Fed.Cir.1987)); see also Hybritech Inc. v. Abbott Labs., 849 F.2d 1446, 1451, 7 USPQ2d 1191, 1195 (Fed.Cir.1988). After finding that Tate had established each of the four factors, the district court granted the preliminary injunction. Tate, 132 F.Supp.2d at 379.
On the first factor, likely success on the merits, the district court found that Tate established likely success on both validity and infringement. Tate, 132 F.Supp.2d at 370-75. In its validity analysis, the district court noted that patents are presumed valid pursuant to 35 U.S.C. § 282, and that the '491 patent withstood a validity challenge based on inventorship in prior litigation. Tate, 132 F.Supp.2d at 371 (citing Tate Access Floors, Inc. v. Maxcess Techs., 222 F.3d 958, 55 USPQ2d 1513 (Fed. Cir.2000) (affirming jury verdict finding '491 patent infringed and not invalid)). The district court then independently rejected Interface’s obviousness argument. It found that prior art textbooks taught beveling the edges of laminate in other contexts, but that ordinary skill in the art *1363of making floor panels excluded making decorative beveled borders. Tate, 132 F.Supp.2d at 371-72. The district court also considered secondary factors of non-obviousness: “the omnipresence of relatively expensive added trims in the market before the issuance of Tate’s patent,” and the commercial success of Tate’s embodiment of the patent. Id. at 372. Prior to concluding its determination that Tate proved likely success on the merits regarding validity, the district court stated:
The remaining evidence of obviousness, under factors 1 and 3, involves the prior art in floor panels. Whether Interface is practicing the prior art constitutes an affirmative defense to infringement. Fiskars, Inc. v. Hunt Mfg. Co., 221 F.3d 1318, 1323 (Fed.Cir.2000). I address this issue in discussing infringement.
Id.
The district court next determined that Tate had proved likely success on its infringement claim. Id. at 372-75. It noted, properly, that infringement analysis proceeds in two steps: 1) claim construction, which begins with an assessment of the intrinsic evidence and considers extrinsic evidence only if the intrinsic evidence is unclear, and then 2) comparison of the accused device to the construed claims. Id. at 372.
Interface asserted that the term “border” in independent claims 1 and 8 must be construed to mean a “recessed horizontal border formed of a single layer.” Id. Under this construction, Interface’s panels would not infringe; their edges lack a horizontal layer, and are formed only of a multi-layered bevel. Interface also argued that the '491 patent could not be reasonably construed to cover its accused device. Id.
The district court rejected Interface’s claim construction. It noted that it was bound to presume the claim terms “mean what they say,” to accord them their ordinary and accustomed meaning, and to turn to extrinsic evidence only if the intrinsic evidence is unclear. Id. (quoting Johnson Worldwide Assocs. v. Zebco Corp., 175 F.3d 985, 989, 50 USPQ2d 1607, 1610-11 (Fed.Cir.1999)). It then held that nothing in the language of the claims, this court’s claim construction in Tate Access, the specification, or the prosecution history of the '491 patent limits the claimed “border” to a horizontal, single layer. Id. at 372-74. It concluded that the accused panel’s edge is lower and decorative, and that, as Interface advertises, the edge contrasts with the surface of the panel. Id. at 374. Accordingly, the district court held that the accused panel could be construed in accordance with Tate’s motion, and that it “has a ‘contrasting’ border, visible from above, and the decorative layer has been cut away to reveal that border.” Id. Thus, it reasoned, “[t]he claims can rationally be construed to cover Interface’s panel.” Id.
The district court also rejected. Interface’s “practicing the prior art” defense, which it treated as a defense to infringement. Interface argued that if the '491 patent covers all beveled edges, it would impermissibly cover prior art panels with vertical edges and that Westinghouse had sold floor panels with edges beveled as in the accused device prior to Tate’s invention. Id. The court rejected this argument because it found that a vertically edged panel would lack the claimed integral contrasting border and decorative trim. Id. at 375. Second, the court noted that there was some testimony that Westinghouse sold “untrimmed laminate panels beveled in the way that the accused panels are beveled,” but found the conflicting testimony regarding Westinghouse’s prior art insufficient to defeat Tate’s likely success on the merits. Id. at 374-75. The district court concluded: “[although the issue is a somewhat close one, I find a likelihood of success for Tate.” Id. at 375.
*1364After concluding that Tate had established the requisite likelihood of success on both infringement and validity, the district court analyzed the remaining three preliminary injunction factors. It found that Tate had established these factors as well. Tate established with objective evidence that it would suffer irreparable harm absent a preliminary injunction despite also being entitled to a presumption of irreparable harm due to its “strong showing of validity,” and “quite ‘clear’ ” showing on infringement. Id. at 375-78. Interface failed to rebut this presumption. Id. at 378. The district court next concluded that the balance of harms between Tate and Interface weighed slightly in Tate’s favor. Id. at 378-79. Finally, because it found no public health issue at stake and because of the strong public interest in protecting patent rights, the district court found that the final factor-public interest-also failed to preclude an injunction. Id. at 379. Accordingly, the district court preliminarily enjoined Interface from infringing claims 1-4 and 8-10 of the '491 patent. Tate Access Floors, Inc. v. Interface Architectural Res., Inc., No. 00 CV 2543 (D.Md. Mar. 9, 2001) (preliminary injunction order).
Interface appeals. We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(1).
On appeal, Interface challenges only the district court’s determination regarding likely success on the merits of Tate’s infringement claim. Interface does not challenge the validity of the '491 patent, nor does it contend that the district court erred in its analysis of the remaining preliminary injunction factors. Accordingly, the crucial question is whether the district court properly construed the “border” limitations in independent claims 1 and 8 to include beveled edges without a single, horizontal layer such as those contained in Interface’s accused panels.
Interface argues that the district court erred in determining that Tate had demonstrated the likely success of its infringement argument for two reasons. Interface’s first and primary argument is that regardless of the claim construction, its accused products cannot literally infringe the asserted claims of the '491 patent because Interface merely practices the prior art. In addition, Interface argues that, as properly construed, the “border” limitations exclude simple beveled edges such as those found in its accused floor panels.
II. STANDARD OF REVIEW
The decision to grant a preliminary injunction under 35 U.S.C. § 283 rests within the district court’s sound discretion. Amazon.com, Inc. v. Barnesandnoble.com, Inc., 239 F.3d 1343, 1350, 57 USPQ2d 1747, 1751 (Fed.Cir.2001). This court sustains such a grant absent an abuse of discretion. Id. Abuse of discretion requires a showing that the district court committed a clear error of judgment in weighing the relevant factors, or based its decision upon a clearly erroneous factual finding or an erroneous legal standard. Id.; see also Mentor Graphics Corp. v. Quickturn Design Sys., Inc., 150 F.3d 1374, 1377, 47 USPQ2d 1683, 1685 (Fed. Cir.1998). To the extent a decision to grant a preliminary injunction rests on questions of law, including claim construction, our review is de novo. Mylan Pharms., Inc. v. Thompson, 268 F.3d 1323, 1329, 60 USPQ2d 1576, 1580 (Fed.Cir. 2001) (citing Globetrotter Software, Inc. v. Elan Computer Group, Inc., 236 F.3d 1363, 1367, 57 USPQ2d 1542, 1545 (Fed. Cir.2001)).
III. DISCUSSION
The moving party, here Tate, is entitled to a preliminary injunction if it shows the following four factors:
*1365(1) a reasonable likelihood of success on the merits;
(2) irreparable harm absent an injunction;
(3) that the balance of hardships tips in its favor; and
(4) that the public interest favors an injunction.
Amazon.com, 239 F.3d at 1350, 57 USPQ2d at 1751. Individually, no factor is dispositive; the district court must weigh the factors against each other and against the form and magnitude of requested relief. Id. (quoting Hybritech, Inc. v. Abbott Labs., 849 F.2d 1446, 1451, 7 USPQ2d 1191, 1195 (Fed.Cir.1988)).
In order to demonstrate likely success on the merits, Tate must show that, in light of the presumptions and burdens applicable at trial, it will likely prove that Interface infringes the asserted claims of the '491 patent and that the patent will likely withstand Interface’s challenges to its validity. See id. (citing Genentech, Inc. v. Novo Nordisk, A/S, 108 F.3d 1361, 1364, 42 USPQ2d 1001, 1003 (Fed.Cir.1997)). If Interface raises a substantial question concerning infringement or validity, meaning that it asserts a defense that Tate cannot prove “lacks substantial merit,” the preliminary injunction issued improperly. See id. at 1350-51, 57 USPQ2d at 1751.
Although the district court found that Tate had established likely success on the merits regarding both infringement and validity, only infringement is at issue in this appeal. As Interface confirmed at oral argument, Interface did not appeal and does not challenge the district court’s determination that Tate would likely prevail on the issue of validity. We will therefore affirm the district court’s order preliminarily enjoining Interface from infringing the asserted claims of the '491 patent unless the district court determined, improperly, that Tate would likely prevail on its contention that Interface’s panels infringe the asserted claims of the '491 patent.
Infringement analysis involves two steps: the court first construes the scope of the asserted claims and then compares the accused device to the properly construed claims to determine whether each and every limitation of a claim is present, either literally or equivalently, in the accused device. Amazon.com, 239 F.3d at 1351, 57 USPQ2d at 1751; Kahn v. Gen. Motors Corp., 135 F.3d 1472, 1476, 1478, 45 USPQ2d 1608, 1610 and 1612 (Fed.Cir.1998). Claim construction is an issue of law, reviewed de novo. Markman v. Westview Instruments, Inc., 52 F.3d 967, 979-81, 34 USPQ2d 1321, 1328-29 (Fed.Cir.1995) (en banc), aff'd, 517 U.S. 370, 116 S.Ct. 1384, 134 L.Ed.2d 577 (1996). Comparing the accused product to the asserted claims is a question of fact that we review with substantial deference. Glaxo Grp. Ltd. v. Ranbaxy Pharms., Inc., 262 F.3d 1333, 1335, 59 USPQ2d 1950, 1952 (Fed.Cir.2001).
A. Practicing the Prior Art
Interface first argues that regardless of the proper claim construction, its panels cannot literally infringe because they merely practice the prior art, or that which would have been obvious in light of the prior art. This contention lacks merit. See Baxter Healthcare Corp. v. Spectramed, Inc., 49 F.3d 1575, 1583, 34 USPQ2d 1120, 1126 (Fed.Cir.1995).
This court made unequivocally clear in Baxter that there is no “practicing the prior art” defense to literal infringement. See 49 F.3d at 1583, 34 USPQ2d at 1126. Baxter involved a jury verdict that the defendant, Spectramed’s accused products infringed, either literally or under the doctrine of equivalents, the asserted claims of Baxter’s patent. Baxter, 49 F.3d at *13661579 & n. 6, 34 USPQ2d at 1123 & n. 6. The district court then entered judgment as a matter of law that Spectramed did not infringe the asserted claims of Baxter’s patent, which claimed a method of calibrating blood pressure transducers. Id. at 1580, 49 F.3d 1575, 34 USPQ2d at 1124. This court reversed. We held that substantial evidence supported a finding of literal infringement notwithstanding Spec-tramed’s argument that its accused devices could not infringe because they were constructed using only the teachings contained in the prior art. Id. at 1582-83, 49 F.3d 1575, 34 USPQ2d at 1125-26. We explained:
Implicit in Spectramed’s argument is that Baxter, in order to establish literal infringement, must prove by a preponderance of the evidence that Spec-tramed’s accused devices embody all the limitations in the asserted claims, and in addition, Spectramed’s accused devices must not be an adoption of the combined teachings of the prior art. This is not a correct statement of the law governing patent infringement. There is no requirement that the accused device be nonobvious in light of the prior art, or otherwise be itself patentable. Literal infringement exists if each of the limitations of the asserted claim(s) read on, that is, are found in, the accused device, Questions of obviousness in light of the prior art go to validity of the claims, not to whether an accused device infringes.
Id. at 1583, 49 F.3d 1575, 34 USPQ2d at 1126 (footnote omitted).
Thus, our decision in Baxter directly forecloses Interface’s “practicing the prior art” argument. Interface attempts to distinguish Baxter on the grounds that it “merely addresses the parties’ respective burdens of proof when the prior art defense is raised.” It also seeks to glean authority for the putative “practicing prior art” defense to literal infringement from the doctrine of equivalents, the reverse doctrine of equivalents, and an extension of a Supreme Court case predating the 1952 Patent Act. We address these arguments below. Even if they had merit— which they do not — this panel would not be free to overrule Baxter. See Vas-Cath Inc. v. Mahurkar, 935 F.2d 1555, 1563, 19 USPQ2d 1111, 1117 (Fed.Cir.1991) (“[W]e note that decisions of a three-judge panel of this court cannot overturn prior prece-dential decisions.”); Kimberly-Clark Corp. v. Ft. Howard Paper Co., 772 F.2d 860, 863, 227 USPQ 36, 37 (Fed.Cir.1985) (“Counsel is apparently unaware that a panel of this court is bound by prior prece-dential decisions unless and until overturned en banc.”); see also Yunus v. Dept. of Veterans Affairs, 242 F.3d 1367, 1372 n. 1 (Fed. Cir.2001) (citing Newell Cos., Inc. v. Kenney Mfg. Co., 864 F.2d 757, 765, 9 USPQ2d 1417, 1423 (Fed. Cir.1988), for the proposition that upon direct conflict between Federal Circuit decisions, “the prec-edential decision is the first”).
Although the differing burdens of proof for infringement and validity are indeed important, Baxter in no way “merely addresses the parties’ respective burdens of proof when the prior art defense [to literal infringement] is raised.” The discussion in Baxter cited above makes clear that literal infringement is determined by construing the claims and comparing them to the accused device, not by comparing the accused device to the prior art. Baxter, 49 F.3d at 1583, 34 USPQ2d at 1126 (“There is no requirement that the accused device be nonobvious in light of the prior art, or otherwise be itself patentable. Literal infringement exists if each of the limitations of the asserted claim(s) read on, that is, are found in, the accused device.”).
Interface cites several doctrine of equivalents cases in an attempt to bolster its “practicing the prior art” defense to literal infringement. They hold that the scope of equivalents may not extend so far as to *1367ensnare prior art. See Fiskars, Inc. v. Hunt Mfg. Co., 221 F.3d 1318, 1323, 55 USPQ2d 1569, 1573 (Fed.Cir.2000) (holding that the argument that a hypothetical claim literally covering the accused device would be unpatentable is a defense to infringement under the doctrine of equivalents, for which the defendant bears the burden of proof); Sextant Avionique, S.A. v. Analog Devices, Inc., 172 F.3d 817, 827, 49 USPQ2d 1865, 1871 (Fed.Cir.1999) (stating that the defendant would have had a “complete defense” if practicing the prior art, because “the prior art and prosecution history estoppel provide independent ‘policy oriented’ limitations on the doctrine of equivalents ”) (emphasis added); Wilson Sporting Goods Co. v. David Geoffrey & Assocs., 904 F.2d 677, 685, 14 USPQ2d 1942, 1948-49 (Fed.Cir.1990) (“In this context it is important to remember that the burden is on Wilson to prove that the range o'f equivalents which it seeks would not ensnare the prior art Uniroyal ball.”).
With respect to literal infringement, these cases are inapposite. The doctrine of equivalents expands the reach of claims beyond their literal language. That this expansion is guided and constrained by the prior art is no surprise, for the doctrine of equivalents is an equitable doctrine and it would not be equitable to allow a patentee to claim a scope of equivalents encompassing material that had been previously disclosed by someone else, or that would have been obvious in light of others’ earlier disclosures. But this limit on the equitable extension of literal language provides no warrant for constricting literal language when it is clearly claimed.
Moreover, just as the doctrine of equivalents cannot extend so broadly as to ensnare prior art, claim language should generally be construed to preserve validity, if possible. See, e.g., Whittaker Corp. v. UNR Indus., Inc., 911 F.2d 709, 712, 15 USPQ2d 1742, 1744 (Fed.Cir.1990). Prior art is relevant to literal infringement only to the extent that it affects the construction of ambiguous claims. Where the meaning of claim language is clear in light of the specification, however, and properly supported by the patent’s disclosure, the situation differs. Fairness and the public notice function of the patent law require courts to afford patentees the full breadth of clear claim language, and bind them to it as well. Consequently, where such claim language clearly reads on prior art, the patent is invalid. See Rhine v. Casio, Inc., 183 F.3d 1342, 1345, 51 USPQ2d 1377, 1379 (Fed.Cir.1999) (“[I]f the only claim construction that is consistent with the claim’s language and the written description renders the claim invalid, then the axiom does not apply and the claim is simply invalid.”).
Our law requires patent challengers to prove invalidity by clear and convincing evidence. Where an accused infringer is clearly practicing only that which was in the prior art, and nothing more, and the patentee’s proffered construction reads on the accused device, meeting this burden of proof should not prove difficult. Nevertheless, accused infringers are not free to flout the requirement of proving invalidity by clear and convincing evidence by asserting a “practicing prior art” defense to literal infringement under the less stringent preponderance of the evidence standard. Moreover, as in the doctrine of equivalents-ensnarement cases, and in the “construction to preserve validity” cases, the fact that the patentee happens to be practicing the pri- or art is irrelevant; it is the presence of the prior art and its relationship to the claim language that matters for invalidity.1
*1368Neither the reverse doctrine of equivalents nor the remaining precedent Interface cites alters this result. Interface argues that the existence of the reverse doctrine of equivalents proves that the literal infringement inquiry cannot end upon a finding that a claim, as properly construed, reads on the accused device. The reverse doctrine of equivalents, Interface contends, may result in a finding of nonin-fringement even when the accused device possesses each and every limitation of an asserted claim. Interface therefore urges this court to adopt its “practicing the pri- or art” defense as a rebuttal similar to its conception of the reverse doctrine of equivalents.
We reject this argument. The Supreme Court referred to the reverse doctrine of equivalents in Graver Tank & Manufacturing Co. v. Linde Air Products Co., 339 U.S. 605, 608-609, 70 S.Ct. 854, 94 L.Ed. 1097, 85 USPQ 328; 330 (1950). It applies “where a device is so far changed in principle from a patented article that it performs the same or a similar function in a substantially different way, but nevertheless falls within the literal words of the claim.” Id. at 609, 339 U.S. 605, 70 S.Ct. 854, 94 L.Ed. 1097, 85 USPQ at 330. In such a case, the reverse doctrine of equivalents “may be used to restrict the claim and defeat the patentee’s action for infringement.” Id.
Not once has this court affirmed a decision finding noninfringement based on the reverse doctrine of equivalents. And with good reason: when Congress enacted 35 U.S.C. § 112, after the decision in Graver Tank, it imposed requirements for the written description, enablement, definiteness, and means-plus-function claims that are co-extensive with the broadest possible reach of the reverse doctrine of equivalents. See In re Donaldson Co., 16 F.3d 1189, 1194 & n. 5, 29 USPQ2d 1845, 1849-50 & n. 5 (Fed.Cir.1994) (rejecting Commissioner’s argument that by enacting § 112 Congress intended only to codify the reverse doctrine of equivalents and thereby to render the requirements applicable only in the litigation context, and therefore holding that the requirements of § 112, ¶ 6 apply to all interpretations of means-plus-function claim language, but noting that one result of enacting § 112 may have been to codify the reverse doctrine of equivalents); cf. Hilton Davis Chem. Co. v. Warner-Jenkinson Co., 62 F.3d 1512, 1569, 35 USPQ2d 1641, 1688-89 (Fed.Cir.1995) (en banc) (Nies, J., dissenting) (noting that the reverse doctrine of equivalents was originally used by the courts to reduce the scope of broad “means” claims -to “cover only what the inventor discloses and equivalents thereof’), rev’d and remanded, 520 U.S. 17, 117 S.Ct. 1040, 137 L.Ed.2d 146, 41 USPQ2d 1865 (1997).
Even were this court likely ever to affirm a defense to literal infringement based on the reverse doctrine of equivalents, the presence of one anachronistic exception, long mentioned but rarely applied, is hardly reason to create another. We therefore decline the invitation to adopt Interface’s “practicing the prior art” defense on these grounds.
We also will not do so on the basis of Interface’s argument that the Supreme Court sanctioned the “practicing prior art” defense in Scott Paper Co. v. Marcalus Mfg. Co., 326 U.S. 249, 66 S.Ct. 101, 90 L.Ed. 47, 67 USPQ 193 (1945).2 Inter*1369face’s argument is misguided. Scott Paper announced no broad “practicing the prior art” defense to literal infringement.
In Scott Paper, the Supreme Court carved out an exception to the doctrine of assignor estoppel. Id. at 257, 66 S.Ct. 101, 67 USPQ at 197 (“[T]he patent laws preclude the petitioner assignee from invoking the doctrine of estoppel” where the assign- or’s accused device was disclosed in a prior patent.); see also Diamond Scientific Co. v. Ambico, Inc., 848 F.2d 1220, 1222-23, 6 USPQ2d 2028, 2029 (Fed.Cir.1988) (noting that in Scott Paper the Supreme Court carved an exception to the rule that an assignor was estopped from challenging the validity of a patent). The plaintiff-assignee accused the defendant of infringing a patent the defendant had assigned to the plaintiff in exchange for valuable consideration. Id. at 250, 66 S.Ct. 101, 67 USPQ at 194. The defendant-assignor maintained that his accused device was a copy of that disclosed in an expired prior art patent. Id. The district court held that the assignor was estopped from challenging the patent’s validity under the doctrine of assignor estoppel, and could not resort to an argument of noninfringement based on the prior art to “accomplish the same result by indirection.” Id. The court of appeals reversed, allowing the assignor to “measure the extent of anticipation for the purpose of limiting the claims of the assigned patent, and thus avoid infringement.” Id.
The Supreme Court affirmed. It explained:
The judgment is affirmed for the reason that we find that the application of the doctrine of estoppel so as to foreclose the assignor of a patent from asserting the right to make use of the prior art invention of an expired patent, which anticipates that of the assigned patent, is inconsistent with the patent laws which dedicate to public use the invention of an expired patent. The assignor has a complete defense to an action for infringement where the alleged infringing device is that of an expired patent.
Id. at 257-58, 66 S.Ct. 101, 67 USPQ at 197 (emphasis added).
As the above paragraph indicates, the Scott Paper rationale was based on the premise that the assigned patent would be invalid due to anticipation were it coextensive with the expired prior art patent. Id. The statement that, in such circumstances, the assignor has a “complete defense to an action for infringement ” must be understood in context. Id. (emphasis added). Of course, under the present scheme, invalidity is a defense to an action for patent infringement. This general statement provides no license for asserting that a particular defense, if successful, results in nonin-fringement as opposed to invalidity.
Moreover, the court of appeals decision that the Supreme Court affirmed in Scott Paper was premised on claim construction to preserve validity, not on noninfringement. Id. at 251, 66 S.Ct. 101, 67 USPQ at 194-95 (“The Court of Appeals reversed, holding that the prior art may be resorted to by the assignor to measure the extent of anticipation for the purpose of limiting the claims of the assigned patent, and thus avoid infringement.” (emphasis added) (citation omitted)). This is consistent with our case law holding that, if possible, claims are construed to preserve validity. It certainly announces no general rule that “practicing prior art” is an affirmative defense precluding literal infringement.
*1370B. Claim Construction
Interface’s second argument is that the district court erred in construing the “border” limitations in independent claims 1 and 8 to encompass the simple beveled edges in its accused floor panels. This argument also fails.
Claim interpretation begins, as always, with the language of the claims. Johnson Worldwide Assocs., Inc. v. Zebco Corp., 175 F.3d 985, 989, 50 USPQ2d 1607, 1610 (Fed.Cir.1999). “[A] court must presume that the terms in the claim mean what they say, and, unless otherwise compelled, give full effect to the ordinary and accustomed meaning of claim terms.” Id. (citing Nike Inc. v. Wolverine World Wide, Inc., 43 F.3d 644, 646, 33 USPQ2d 1038, 1039 (Fed.Cir.1994); E.I. du Pont de Nemours & Co. v. Phillips Petroleum, 849 F.2d 1430, 1433, 7 USPQ2d 1129, 1131 (Fed.Cir.1988); Envirotech Corp. v. Al George, Inc., 730 F.2d 753, 759, 221 USPQ2d 473, 477 (Fed.Cir.1984)). This strong presumption in favor of the ordinary meaning of claim language as understood by one of ordinary skill in the art may be overcome where: 1) the patentee has chosen to become his or her own lexicographer by clearly and explicitly defining the claim term; or 2) where the claim term would render the claim devoid of clarity such that there is “no means by which the scope of the claim may be ascertained from the language used.” Bell Atl. Network Servs., Inc. v. Covad Communications Group, Inc., 262 F.3d 1258, 1268, 59 USPQ2d 1865, 1870 (Fed.Cir.2001) (quoting Johnson, 175 F.3d at 990, 50 USPQ2d at 1610).
One ordinarily skilled in the art of raised access floor panels would understand the term “border” to refer to the area or zone forming an edge or trim framing the decorative surface of the panel. Independent claim 1 recites “a border along the edges of said panels along which said decorative surface layer is removed to expose said inner body portion.” '491 patent, col. 4, 1. 67 to col. 5, 1. 11. Independent claim 8 recites “a border extending along the edges of said panel along which the decorative layer is removed to expose said inner layer.” Id. at col. 5,1. 37 to col. 6,1. 11. Nothing in the language of these claims requires the “border” to be horizontal or formed of a single layer.
The “border” in claim 1 is formed by exposing an “inner body portion.” A portion may certainly consist of one or more layers.
The “border” in claim 8 is formed by exposing “said inner layer.” That language refers to the phrase “an inner layer,” which first appears earlier in claim 8. '491 patent, col. 6, 1. 5. It is well settled that the term “a” or “an” ordinarily means “one or more.” Tate Access Floors, Inc. v. Maxcess Techs., Inc., 222 F.3d 958, 966 n. 4, 55 USPQ2d 1513, 1518 (citing Elkay Mfg. Co. v. Ebco Mfg. Co., 192 F.3d 973, 977, 52 USPQ2d 1109, 1112 (Fed.Cir.1999)). Moreover, additional language in claim 8 suggests that “an inner layer” means one or more layers. The relevant portion of claim 8 reads as follows:
a single visible decorative layer along the side of said floor covering also providing an inner layer contrasting with said decorative layer having a thickness substantially greater than the thickness of the decorative layer adjacent to said decorative layer along the side thereof facing said load surface....
’491 patent, col. 6,11. 2-9 (emphasis added). The language “an inner layer” must be read in the context of the language claiming “a single visible decorative layer;” where the patentee meant to constrict the claim to one and only one particular layer, he said so clearly. Moreover, the reference to substantially greater thickness also suggests that the claimed inner layer may consist of one or more layers, because *1371increasing the layers would increase the thickness.3
Like the claim language, the remainder of the specification supports construing the term “border” to include multiple layers framing the decorative paper. Of course, limitations from elsewhere in the specification will not be read in where, as here, the claim terms are clear. See, e.g., Toro Co. v. White Consol. Indus., Inc., 266 F.3d 1367, 1371, 60 USPQ2d 1437, 1439 (Fed.Cir.2000) (citing E.I. du Pont de Nemours & Co. v. Phillips Petroleum, 849 F.2d 1430, 1433, 7 USPQ2d 1129, 1131 (Fed.Cir.1988) (holding that it is improper to import an extraneous limitation from elsewhere in the specification into the claim)); see also Kahn, 136 F.3d at 1476, 45 USPQ2d at 1611 (distinguishing means-plus-function claims from the ordinary situation in which claims may not be limited by functions or elements disclosed in the specification, but not included in the claims themselves).
Nevertheless we note that this is not a case in which the patentee acted as his own lexicographer, and that the written description supports our construction. In pertinent part, it reads:
Rearwardly of the decorative paper, the floor covering material is provided with layers of material having a contrasting color with respect to the decorative paper.
In accordance with the present invention, the surface layer of protective material and the layer of decorative paper are cut away along the edge of the floor covering to expose the inner layers and provide a contrasting color integral border within the floor covering material itself.
’491 patent col. 1, 1. 65 to col. 2, 1. 4 (emphasis added). This language contradicts Interface’s position that the description requires construing the claimed “border” as a single, horizontal layer. Interface emphasizes the statement that “a uniform exposure of the layers of black paper is obtained.” '491 patent, col. 3, 11. 53-54. This language, however, merely underscores that the border is formed by uniformly exposing multiple layers, not by exposing a single layer. Thus, the written description plainly contemplates that the claimed contrasting color border can be formed of multiple layers of material.
Interface argues that notwithstanding this language, Figures 3 and 4 of the '491 patent compel its construction of the “border” limitations. This is allegedly so because the arrow corresponding to the “integral border or edge trim 14,” '491 patent, col. 2, 1. 54, aims at a horizontal, single layer of exposed trim. This argument fails. Even were we to accept the proposition that the arrow in Figures 3 and 4 points only to the horizontal, single-layered portion of the border rather than to the beveled portion (which is also clearly exposed, and which is formed of multiple layers of material) additional limitations from these figures cannot be imported into the unambiguous claim language.
Contrary to Interface’s next contention, nothing in the prosecution history, or elsewhere in the prior art of record,4 renders proper a construction that *1372limits the claimed “border” to a single, horizontal layer. Interface would have us adopt its narrow construction in order to preserve the validity of the '491 patent. It contends that the prior art would render the asserted claims obvious if they cover floor panels with simple beveled edges such as its accused products.
Even if this were so, where claim language is clear we must accord it full breadth even if the result is a claim that is clearly invalid. Apple Computer, Inc. v. Articulate Sys., Inc., 234 F.3d 14, 24, 57 USPQ2d 1057, 1064 (Fed.Cir.2000) (explaining that claims should be read to preserve validity but only where it is possible to do so, and holding that it was impossible and that therefore the claims were simply invalid) (citing Rhine v. Casio, Inc., 183 F.3d 1342, 1344, 51 USPQ2d 1377, 1379 (Fed.Cir.1999) (Where “the only claim construction that is consistent with the claim’s language and the written description renders the claim invalid, then the axiom does not apply and the claim is simply invalid.”)).
IV. CONCLUSION
The district court committed no abuse of discretion in determining that Tate demonstrated its entitlement to a preliminary injunction, including a reasonable likelihood of success on the merits of its infringement claim. The grant of preliminary injunction barring Interface from infringing claims 1-4 and 8-10 of the '491 patent is therefore affirmed.
AFFIRMED.
No costs.