C-COR Inc. ("C-COR")1 appeals from a final judgment of the U.S. District Court for the District of Delaware ("district court") in favor of Seachange International, Inc. ("Seachange") on Seachange's claim for infringement of its U.S. Patent No. 5,862,312 ("the '312 patent"). Seachange Int'l, Inc. v. nCUBE Corp., No. 00-568-JJF (D. Del. June 30, 2004). On C-COR's challenge to the district court's claim construction rulings, we conclude that the district court erred in construing the limitations "network for data communications," "distributed computer system," and "processor systems." Based on our revised claim construction, we conclude that C-COR does not infringe as a matter of law, and affirm the denial of C-COR's motion for judgment as a matter of law ("JMOL") on invalidity for lack of written description. Because the district court correctly denied C-COR's motion for JMOL on anticipation as to the Frey and Mendelsohn references, we affirm that ruling. However, we vacate the denial of the JMOL motion as to anticipation by Gardner and remand. Because C-COR was prejudiced by the district court's failure to properly instruct the jury on the construction of the "distributed computer system" limitation, we reverse the denial of C-COR's alternative motion for new trial on anticipation and remand. Thus, we affirm-in-part, reverse-in-part, vacate-in-part, and remand.

Seachange and C-COR are competitors in the video server industry. Seachange is the assignee of the '312 patent, which is directed to a method and apparatus for redundantly storing data—in particular, video data—for video-on-demand. The prior art "RAID-5" (Redundant Array of Inexpensive Disks) storage protocol provides redundant storage of video data by splitting a data file into segments and "striping" the segments across multiple disk drives within a single "processor system." The system retrieves the data by accessing the segments in round-robin fashion. In addition, it stores and strips "parity data," which enables recovery of video data if a drive crashes. See '312 patent, col. 1, II. 11-62.

The '312 patent expands the RAID-5 concept to a "distributed computer system." The distributed computer system includes at least three processor systems, yielding redundant data storage at both the processor-system level (within the processor system) and the computer-system level (across processor systems). Id., col. 2, II. 16-32. This "RAID Squared" system results in increased fault tolerance over the RAID-5 system. Id., II. 6-14, 33-43. Furthermore, the '312 patent teaches that connecting every processor system to every other processor system using point-to-point, two-way channel interconnections makes more efficient use of the read and write bandwidth, an object of the invention. Id., II. 7-9; col. 7, II. 63-67. Thus, "the invention relates generally to mass storage device interconnections and in particular, to a method and apparatus for increasing delivery bandwidth, providing fault tolerance, and input/output load balancing in a multiprocessor computer cluster." Id., col. 7, II. 4-8.

On June 13, 2000, Seachange filed suit against C-COR alleging that C-COR's technology infringed claims 37-38, 40-42, 52-53, and 57-58 of the '312 patent. C-COR asserted affirmative defenses and counterclaims on invalidity. The district court bifurcated the trial of liability and damages. On August 29, 2000, the district court issued its claim construction order. SeaChange Int'l, Inc. v. nCUBE Corp., 115 F. Supp. 2d 473 (D.Del.2000). The district court construed the limitation "interconnecting each of said processor systems through a network for data communications with each other one of said processor systems" to mean "establishing data communications between every pair of processor systems in the distributed computer system using any kind of network." Id. at 482. The district court construed the limitation "processor systems" to require that each system have "at least one central processing unit ["CPU"] capable of running application type software, and at least one mass storage subsystem." Id. at 483. In light of the claim construction order, C-COR stipulated to infringement.

In September 2000, the invalidity issues were tried to a jury. At the pre-trial conference and during trial, C-COR requested that the district court construe the preamble phrase "distributed computer system." The district court declined. The jury found the claims not invalid for inadequate written description, not invalid due to anticipation, and not invalid due to obviousness. C-COR appealed, and we dismissed for lack of jurisdiction because final judgment was not entered. Seachange Int'l, Inc. v. nCube Corp., 81 Fed.Appx. 325 (Fed.Cir.2003).

On April 7, 2004, the district court issued an opinion denying C-COR's motion, in the alternative, for new trial on written description, anticipation, and non-obviousness. nCUBE Corp. v. SeaChange Int'l. Inc., 313 F. Supp. 2d 393 (D.Del.2004). The district court admitted that it had erred when it did not construe the limitation "distributed computer system." Id. at 397. The district court found the error to be harmless because it had explained to the jury that "essentially [the phrase] would require a stand-alone computer in each processor system." Id. at 398. The district court explained that if the Federal Circuit decides that C-COR's proposed construction was correct, then C-COR would be entitled to a new trial. Id. The district court rejected C-COR's argument that under the district court's construction of the "network for data communications" limitation, the asserted claims were invalid for failure to comply with the written description requirement. Id. at 400. Finally, the district court reviewed the evidence presented at trial as it related to anticipation, addressing several references, three of which are discussed on appeal. The district court concluded that the jury's finding that Gardner et al., U.S. Patent No. 5,583,995 ("Gardner"), did not disclose "processor systems" was not against the weight of the evidence, SeaChange, 313 F.Supp.2d at 401; that the jury's findings that Frey, Jr. et al., U.S. Patent No. 5,130,992 ("Frey"), did not disclose either a "distributed computer system" or "processor systems" were not against the weight of the evidence, id. at 401-02; and that the jury's finding that Mendelsohn et al., U.S. Patent No. 5,488,731 ("Mendelsohn"), did not disclose "at least three processor systems" was not against the weight of the evidence, id. at 402-03.

On April 29, 2004, the district court denied C-COR's motion for JMOL on the invalidity issues for the same reasons discussed in the April 7, 2004 opinion. On June 30, 2004, the district court entered final judgment. C-COR timely appealed. We have jurisdiction under 28 U.S.C. § 1295(a)(1).

C-COR argues that the "network for data communications" limitation does not cover a system using any type of network other than a network that interconnects every processor system to every other processor system using direct, point-to-point two-way channel interconnections. C-COR argues that under this construction, it is entitled to a judgment of non-infringement as a matter of law. Alternatively, C-COR argues that under the district court's broader construction, it is entitled to JMOL that the asserted claims are invalid for inadequate written description. C-COR asserts that under the district court's construction of "distributed computer system" and "processor systems," it is entitled to JMOL that the asserted claims are invalid as anticipated by Gardner, Frey, and Mendelsohn. C-COR alternatively challenges the district court's construction of these limitations and argues that under the correct construction it is entitled to a new trial on anticipation. C-COR argues that "distributed computer system" does not require "stand-alone" computers in every processor system. C-COR argues that each "processor system" need not have a CPU capable of running application software.

Seachange asserts that the district court did not err in its claim construction rulings or in denying C-COR's motions for JMOL and a new trial. Seachange asserts that even if C-COR's construction of "network for data communications" is correct, this court should remand the case to the district court to consider infringement under the doctrine of equivalents. Seachange adds that even if the district court erred in its construction of the "distributed computer system" and "processor systems" limitations, we should not grant C-COR a new trial on anticipation because C-COR did not establish that the references disclosed the "storing data input" limitation.

Claim construction is a question of law reviewed de novo. Cybor Corp. v. FAS Techs., Inc., 138 F.3d 1448, 1451 (Fed.Cir.1998) (en banc). This court applies regional circuit law in reviewing the denial of a motion for JMOL, Summit Tech., Inc. v. Nidek Co., 363 F.3d 1219, 1223 (Fed.Cir.2004), and the denial of a motion for a new trial, Hewlett-Packard Co. v. Mustek Sys., Inc., 340 F.3d 1314, 1318 & 1323 (Fed.Cir.2003). The Third Circuit reviews a denial of a motion for JMOL de novo by reapplying the JMOL standard. See Rinehimer v. Cemcolift, Inc., 292 F.3d 375, 383 (3d Cir.2002). Thus, we must "determine whether viewing the evidence in the light most favorable to the nonmovant and giving [the nonmovant] the advantage of every fair and reasonable inference, there is insufficient evidence from which a jury reasonably could reach the conclusions that it did." Id. (internal quotations omitted). The Third Circuit reviews a denial of a motion for new trial for abuse of discretion. Id. at 383-84. "[A] new trial should be granted only when the verdict is contrary to the weight of the evidence...." Brennan v. Norton, 350 F.3d 399, 430 (3d Cir.2003).

We turn first to C-COR's challenges to the district court's claim construction rulings. The parties treat the disputed phrases appearing in claim 37 as representative:

37. A method for redundantly storing data in a distributed computer system having at least three processor systems, each processor system comprising at least one central processing unit and at least one mass storage sub-system, comprising the steps of:

interconnecting each one of said processor systems through a network for data communications with each other one of said processor systems; and storing data input at any one of said processor systems according to a distributed, redundant storage process with data stored at each of said processor systems and a portion of a redundant representation of the data is stored at each of said processors.

'312 patent, col. 22, II. 17-29 (emphases added).

The district court construed the phrase "network for data communications" to mean "establishing data communications between every pair of processor systems in the distributed computer system using any kind of network." SeaChange, 115 F.Supp.2d at 482. The district court reasoned that the claim language does not command a specific type of network, that claim differentiation doctrine favors a broad interpretation, that the written description does not require a narrow construction, and that the prosecution history does not show clear disavowal. Id. at 478-81.

Claim 37 does not suggest that the claimed "network" is limited to networks employing direct, point-to-point interconnections. The language specifies only that each processor system be interconnected to each other processor system. Moreover, the technical dictionary definition of "network" does not suggest the point-to-point limitation; indeed, it implies that a "network for data communications" can employ direct or indirect interconnections. See The New IEEE Standard Dictionary of Electrical and Electronics Terms 842 (5th ed.1993) (hereinafter "IEEE").

Seachange agrees with this assessment and asks that we apply the doctrine of claim differentiation. The doctrine of claim differentiation stems from "the common sense notion that different words or phrases used in separate claims are presumed to indicate that the claims have different meanings and scope." Karlin Tech. Inc. v. Surgical Dynamics, Inc., 177 F.3d 968, 971-72 (Fed.Cir.1999). Although the doctrine is at its strongest "where the limitation sought to be `read into' an independent claim already appears in a dependent claim," Liebel-Flarsheim Co. v. Medrad, Inc., 358 F.3d 898, 910 (Fed.Cir.2004), there is still a presumption that two independent claims have different scope when different words or phrases are used in those claims, Kraft Foods, Inc. v. Int'l Trading Co., 203 F.3d 1362, 1365-69 (Fed.Cir.2000); see also Tandon Corp. v. U.S. Int'l Trade Comm'n, 831 F.2d 1017, 1023 (Fed.Cir.1987). However, the doctrine "only creates a presumption that each claim in a patent has a different scope; it is not a hard and fast rule of construction." Kraft, 203 F.3d at 1368 (internal quotations omitted). "[T]he doctrine of claim differentiation can not broaden claims beyond their correct scope, determined in light of the specification and the prosecution history and any relevant extrinsic evidence.... [C]laims that are written in different words may ultimately cover substantially the same subject matter." Multiform Desiccants, Inc. v. Medzam, Ltd., 133 F.3d 1473, 1480 (Fed.Cir. 1998).

Claim 1 of the '312 patent requires "interconnecting each one of said processor systems in a point-to-point two way channel interconnection with each other one of said processor systems." '312 patent, col. 17, II. 54-56. Claim 37 is identical to claim 1, except that claim 37 requires only that the interconnection be through a "network for data communications." The doctrine of claim differentiation creates a presumption that these limitations in claim 1 and claim 37 are of different scope and suggests that claim 37 does not require point-to-point, two-way channel interconnections. However, that presumption is not a hard and fast rule and will be overcome by a contrary construction dictated by the written description or prosecution history. Kraft, 203 F.3d at 1368.

As to the written description, C-COR argues that Seachange limited the scope of "network for data communications" to point-to-point networks by implication. C-COR asserts that the written description discloses only point-to-point interconnections and establishes that the point-to-point interconnections achieve a necessary objective of the invention. C-COR adds that the written description eliminates embodiments incompatible with point-to-point wiring. C-COR cites several cases in support of its arguments. E.g., Microsoft Corp. v. Multi-Tech Sys., Inc., 357 F.3d 1340, 1351-52 (Fed.Cir.2004) (construing claim to require feature that was "central to the functioning of the claimed invention[]"); Alloc, Inc. v. Int'l Trade Comm'n, 342 F.3d 1361, 1369-70 (Fed.Cir.2003) (construing claim to include limitation because "very character of the invention" required that the limitation be part of every embodiment); Watts v. XL Sys., Inc., 232 F.3d 877, 882-83 (Fed.Cir.2000) (construing claim to include limitation, in part, because specification limited invention to embodiments with that feature); Bell Atl. Network Servs., Inc. v. Covad Communications Group, Inc., 262 F.3d 1258, 1271 (Fed.Cir.2001) (construing claim to reflect inventor's consistent usage of claim term in specification); Toro Co. v. White Consol. Indus., 199 F.3d 1295, 1300-01 (Fed.Cir.1999) (construing claim to require a particular configuration where specification described the importance of the configuration and did not disclose others).

Seachange responds that the written description does not show a clear disavowal or a redefinition. Seachange argues that the type of network connection used is not important to the invention and that the invention resides in RAID Squared. Seachange explains that the text of the specification contains little disclosure of the types of network interconnections covered by the patent because that was not significant. Seachange asserts that a patentee is entitled to claims that are broader than embodiments described and that C-COR seeks to erroneously import a limitation from a preferred embodiment.

We agree with C-COR that the written description consistently refers to the network interconnections as point-to-point, e.g., '312 patent, col. 3, II. 30-34; col. 4, II. 36-43; col. 2, II. 15-25; Abstract; col. 5, II. 1-4, 52-58; col. 6, II. 7-9, and suggests an upper limit on the number of processor systems compatible with point-to-point wiring, id., col. 5, II. 58-61; col. 8, II. 30-38 ("[M]ore or less processor systems can be used, although a practical upper limit may be between nine and thirteen.... When more cluster members are employed, the point to point wiring becomes progressively more difficult and expensive. Indeed, nine cluster members would require thirty-six interconnecting channels while thirteen cluster volumes would have 78 interconnecting channels."). Furthermore, point-to-point interconnections achieve an object of the invention in that they increase read and write bandwidth. See id., col. 2, II. 7-9; col. 7, II. 63-67. However, it is unclear whether these references to point-to-point are simply the consistent description of one possible embodiment or a description of the invention itself. In all of the cases cited by C-COR, the conclusion compelled by the written description is that the limitation in issue is a limitation on the invention, not just a feature of a possible embodiment. The issue is unresolved with certainty from the written description in this case. We turn next to the prosecution history for guidance.

On October 24, 1995, Mann et al. ("Applicant") filed the application which matured into the '312 patent. In a September 30, 1996, Preliminary Amendment ("Preliminary Amendment"), Applicant added thirty-six claims, including claim 37, which was numbered as claim 40 throughout examination ("claim 37(40)"). Applicant also added claims 68-72, and 74, which included a limitation directed to an "interconnecting data communications network." Id. at 11-13. Applicant stated that it added claims "to more fully cover the scope of the invention." Preliminary Amendment at 13.

In a December 12, 1996, Office Action ("First Action"), the Examiner rejected claims 1-75. The Examiner grouped claims 1, 37(40), and others, and rejected them under 35 U.S.C. § 103 as unpatentable over Morita et al., U.S. Patent No. 5,502,980 ("Morita"), in view of Benner et al., U.S. Patent No. 5,072,371 ("Benner"). First Action at 3. In a June 12, 1997, Amendment ("Amendment"), Applicant's attorney stated:

With respect to the various prior art rejections, the Examiner grouped various claims and rejected the grouped claims. Applicant submits that with respect to each group of claims that certain of the claims in the group add further patentably distinct features to the invention and thus are further patentably distinct over the applied references. For simplicity, however, Applicant will in general treat a single claim as being representative of the group of claims but reserves its right to later argue that additional ones of the claims are patentably distinct over the combination of references.

Amendment at 29-30. Applicant's attorney then argued:

The Examiner rejected claims 1, 9-12, 19, 20, 23-27, 40, 48-51, 53-55, 63-66 under 35 U.S.C. § 103 as being unpatentable over [Morita] in view of [Benner].

As an illustrative claim in this grouping, Applicant's claim 1 .... recites a method in which at least three processor systems are interconnected using a point-to-point two-way channel interconnection with each one of the other processor systems. That is, any one processor system can communicate directly with any one of the other processor systems. The claim also recites that data is stored at each of the processor systems which also stores a portion of a redundant representation of the data. Neither the point to point two-way channel interconnection nor the arrangement of stored data and redundant data is suggested by the combination of Morita and Benner.

....

Morita, as the Examiner, [sic] admits does not describe at least three processor systems in a distributed computing system and also does not describe a point to point interconnection. Applicant submits that Morita also does not describe or suggest that the processor system stores data in the manner recited in claim 1....

....

Benner describes a parallel computing system of the hypercube type. As such, although Benner does describe more than two processor systems, Benner does not describe that each of the processor systems are interconnected in a point to point two-way channel interconnection with each other one of the processor systems as recited in Applicant's claim 1. The Examiner indicates, however, that Benner suggests such an arrangement.... Although Benner ... describes his connections as "point to point" paths, Benner does not suggest that each of the processors in the hypercube arrangement are coupled to each one of the other processors in the hypercube arrangement as recited in claim 1. Rather, Benner teaches away from such an interconnection scheme....

....

... Accordingly, in view of the fact that neither Morita nor Benner whether taken separately or in combination suggest the above-mentioned elements of Applicant's invention and furthermore since the combination of Morita and Benner does not appear to be suggested from the references nor appear to be logical, it is submitted that the rejection has been overcome by argument.

Therefore, claim 1 and claims 9-12, 19, 20, 23-27, 40, 48-51, 53-55 and 63-66 are all patentably distinct over Morita in view of Benner.

Id. at 29-35. In this amendment, Applicant did not separately argue that claim 37(40) was patentably distinct on any other basis.

In the same December 12, 1996, First Action, the Examiner also grouped claims 68-72 and 74 and rejected them under 35 U.S.C. § 103 as unpatentable over Morita in view of Benner and Madonna, U.S. Patent No. 5,544,163 ("Madonna"). First Action at 14. The Examiner employed reasoning similar to what he employed in rejecting the grouping that included claim 37(40), except that the Examiner relied on Madonna for disclosure of a "switching system" which was not a limitation in claim 37(40). Id. at 14-15. In response, applicant's attorney argued:

Applicant's claim 68 which is representative of this group of claims is patentably distinguished over the art of record since the references neither describe nor suggest ... at least three processor systems ... interconnecting data communications network ... and data storage control as recited in claim 68 as argued previously with respect to the Morita and Benner references.

Amendment at 46-47.

In a September 16, 1997, Office Action ("Allowance"), the Examiner allowed claims 1-28 and claims 32-75. Allowance at 1. His statement of reasons for allowance noted that "the prior [art] of record fail to teach `storing data input at any one of the processor systems according to a distributed redundant storage process with data stored at each of the processor systems and a portion of the redundant representation of the data stored at each of said processor system[s].'" Id. at 3.

On November 26, 1997, after the Examiner had concluded his examination and issued his notice of allowance, a protest petition ("Protest") was filed. In that Protest, the protester argued that Gardner anticipated claims 1-28, 36-67, and 72-75. Protest at 2. The protester asserted that claims 29-35 and 68-71 were obvious over Gardner in view of Madonna and other references. Id. at 3. The protester emphasized that Gardner taught the use of RAID-5 at the system level and asserted that such feature could not be the point of novelty. However, the protester conceded that "[i]f there exists a difference between [Gardner] and [Seachange's application], the difference may relate to the selected network." Id. at 3-4.

In a January 1998 Office Action, the Examiner requested that Applicant respond to the protest. On May 4, 1998, Applicant responded by asserting that "[n]owhere in [Gardner] does the patentee describe or suggest a store/retrieve two-way point-to-point configuration." Protest Response at 2 (internal citation omitted). The response further argued that Gardner "actually recites ... a network which may, for example, comprise an Ethernet™, Fiber Distributed Data Interchange (FDDI), Asynchronous Transfer Mode (ATM), a small computer system interface (SCSI) or any other network used for transporting data amongst the media service." Id. (internal quotation omitted). Applicant's attorney added that Gardner "relates to an allocation scheme for a plurality of media servers connected in a conventional client-server computer network." Id. at 3. The attorney said that Gardner does not suggest "interconnecting each of the processors in a point to point configuration (claims 1-15); .... [or] interconnecting each one of said processor systems through a network for data communications with each other one of said processor systems (claims 40-54)...." Id. at 3-4. The Examiner then allowed the claims to issue.

C-COR argues that Applicant's arguments made during prosecution narrowed the scope of the "network for data communications" limitation in claim 37(40) to cover only a point-to-point network. Seachange counters that Applicant's arguments did not amount to a clear and unambiguous disclaimer of claim scope. We agree with C-COR.

"The prosecution history constitutes a public record of the patentee's representations concerning the scope and meaning of the claims, and competitors are entitled to rely on those representations when ascertaining the degree of lawful conduct, such as designing around the claimed invention." Hockerson-Halberstadt, Inc. v. Avia Group Int'l, Inc., 222 F.3d 951, 957 (Fed.Cir.2000). Thus, in construing the claim, we consider the prosecution history to determine "whether the patentee disclaimed or disavowed subject matter, narrowing the scope of the claim terms." Nystrom v. Trex Co., 374 F.3d 1105, 1113 (Fed.Cir.2004) (internal quotation omitted). In doing so, we examine the entire prosecution history, which includes amendments to claims and all arguments to overcome and distinguish references. Rheox, Inc. v. Entact, Inc., 276 F.3d 1319, 1326 (Fed.Cir.2002); Elkay Mfg. Co. v. Ebco Mfg. Co., 192 F.3d 973, 979 (Fed.Cir.1999). Where an applicant argues that a claim possesses a feature that the prior art does not possess in order to overcome a prior art rejection, the argument may serve to narrow the scope of otherwise broad claim language. Rheox, Inc., 276 F.3d at 1325 ("Explicit arguments made during prosecution to overcome prior art can lead to narrow claim interpretations...."); Ekchian v. Home Depot, Inc., 104 F.3d 1299, 1304 (Fed.Cir.1997) ("[S]ince, by distinguishing the claimed invention over the prior art, an applicant is indicating what the claims do not cover, he is by implication surrendering such protection."). A disclaimer must be clear and unambiguous. Omega Eng'g, Inc. v. Raytek Corp., 334 F.3d 1314, 1323-25 (Fed.Cir. 2003).

The Examiner grouped several claims together, including claims 1 and 37(40), and rejected them as a group as being obvious over Morita in view of Benner. First Action at 4. Applicant responded, stating, inter alia, that applicant will in general treat a single claim as being representative of the group. Amendment at 29. Applicant then selected claim 1 as "an illustrative claim" and argued that Morita and Benner do not suggest connecting each processor to each other processor via point-to-point, two-way channel interconnections. Id. at 30-32. Applicant also argued that Morita and Benner do not disclose "the arrangement of stored data and redundant data" required by claim 1. Id. at 30-31. Applicant concluded that "in view of the fact that neither Morita nor Benner ... suggest the above-mentioned elements of Applicant's invention ..., it is submitted that the rejection has been overcome by argument." Id. at 35. Even though Applicant "reserve[d] its right to later argue that additional ones of the claims are patentably distinct over the combination of references," id. at 29-30, Applicant made no separate patentability argument for claim 37(40). Because Applicant provided "clear notice of th[e] linkage" between claim 1 and claim 37(40) for the purpose of its argument to overcome the prior art rejection on the basis of the "point-to-point" and "redundant storage" limitations, it would be improper to now broadly construe claim 37(40) not to contain those limitations. See Elkay, 192 F.3d at 980 (holding that arguments made with respect to a claim during the prosecution of an earlier patent applied to a claim in a later patent where the claims were "affirmatively linked" by the applicant); Digital Biometrics, Inc. v. Identix, Inc., 149 F.3d 1335, 1347 (Fed.Cir.1998) (holding that a general statement distinguishing prior art applied to all claims linked to the statement).

Seachange argues that Applicant made two arguments to overcome Morita and Benner, and because the "point-to-point" argument relates directly to the "point-to-point" language of claim 1 rather than the "network for data communications" language of claim 37(40), and because the "redundant storage" argument relates to all claims of the grouping, the public was on notice that claim 37(40) did not necessarily contain the "point-to-point" feature. Seachange buttresses this argument by citing the notice of allowance in which the Examiner allowed all claims based only on the "redundant storage" feature. However, the conclusion that Seachange would have us reach-that Applicant responded to the First Action and distinguished claim 37(40) over Morita and Benner based only on the "redundant storage" feature, and not based on the point-to-point feature—is unwarranted. Nothing in the prosecution history suggests that the point-to-point argument did not apply to all of the grouped claims. Applicant did not indicate that the "point-to-point" argument applied only to claim 1. Instead, the natural reading of Applicant's statements suggests that the "point-to-point" argument applied to each claim in the grouping.

The conclusion that Applicant narrowed the meaning of the word "network" as used in claim 37(40) to require point-to-point interconnections is consistent with similar treatment of claims 68-72 and 74. Claims 68-72 and 74 each recite an "interconnecting data communications network" in language that Seachange admits is "almost identical" to the "network for data communications" language of claim 37(40). In arguing for patentability of claims 68-72 and 74 (of which claim 68 was representative), Applicant stated that "the references neither describe nor suggest ... interconnecting data communications network... and data storage control as recited in claim 68 as argued previously with respect to the Morita and Benner references." Amendment at 46-47. Applicant's only previous argument relevant to the "data communications network" limitation was the argument in response to the rejection of the group of claims including claims 1 and 37(40) that Morita and Benner did not disclose that each of the processor systems are interconnected in a point-to-point two-way channel interconnection with each other one of the processor systems. See id. at 28-48. By arguing for patentability in both groups of claims based upon an interpretation of "network" as requiring point-to-point interconnections, the "inescapable conclusion" is that claims 68-72 and 74, and likewise, claim 37(40), must be so limited. See SciMed Life Sys., Inc. v. Advanced Cardiovascular Sys., Inc., 242 F.3d 1337, 1342-45 (Fed.Cir.2001) (drawing the "inescapable conclusion," from reading together portions of the intrinsic record, that the inventor disavowed claim scope and that a claim term has a narrow meaning).

We find unpersuasive Seachange's argument that the Examiner cited the "redundant storage" feature as the reason for allowance. The fact that the Examiner did not indicate reliance on the point-to-point distinction is of no consequence. An applicant's argument made during prosecution may lead to a disavowal of claim scope even if the Examiner did not rely on the argument. Microsoft, 357 F.3d at 1350.

Seachange further argues that to distinguish Benner, Applicant needed only to argue that claim 1 requires each processor be interconnected to each other processor, and did not need to argue that each processor be interconnected to each other processor by point-to-point connections. Seachange asserts that Applicant's "as argued previously" statement referred only to the argument that Benner did not disclose a system in which each processor system was interconnected to each other processor system. However, Applicant did not argue that claim 1 was patentable merely because it featured a system in which each processor system was interconnected to each other processor system. Applicant argued that Benner does not describe that each of the processor systems are interconnected in a point-to-point two-way channel interconnection with each other one of the processor systems as recited in Applicant's claim 1. Amendment at 32. Applicant made no separate patentability argument for claim 37(40) at that time. Consequently, Seachange cannot now rewrite the prosecution history to distinguish claims 37(40), 68-72 and 74, based only on the limitation that each processor be interconnected to each other processor, and thereby erase the requirement that all connections be point-to-point. Hockerson-Halberstadt, 222 F.3d at 957 (explaining that an argument "that would erase from the prosecution history the inventor's disavowal of a particular aspect of a claim term's meaning" is "inimical to the public notice function provided by the prosecution history").

Seachange also argues that the file wrapper shows that Applicant added claim 37(40) "to more fully cover the scope of the invention," Preliminary Amendment at 13, and that Applicant would not have added claim 37(40) if it was of the same scope as claim 1. However, Seachange inappropriately emphasizes Applicant's subjective intent. Courts must "view[] the prosecution history not for ... applicant's subjective intent, but as an official record that is created in the knowledge that its audience is not only the patent examining officials and the applicant, but the interested public." Biogen, Inc. v. Berlex Labs., Inc., 318 F.3d 1132, 1139 (Fed.Cir.2003).

Seachange's final argument is that Applicant's response to the protest shows that claim 1 is of different scope than claim 37(40). Protest Response at 3-4. On the one hand, Applicant stated that Gardner did not disclose either a point-to-point network as in claim 1 or a network for data communications as in claim 37(40), implying a difference in the scope of claim 1 and claim 37(40). On the other hand, Applicant distinguished over Gardner based on the "network" element and contended that Gardner disclosed a number of different types of networks—Ethernet™, FDDI, ATM, and SCSI networks used for transporting data among the media—but did not disclose "a store/retrieve two-way point-to-point configuration." Id. at 2. C-COR cites to Digital Biometrics for the proposition that this global statement limits claim 37(40) to point-to-point networks. In Digital Biometrics, the applicant made a global statement distinguishing its invention from the prior art and applying specifically the remarks to "all of the pending claims [that] stand rejected under 35 USC 102(b) or 35 USC 103." 149 F.3d at 1347 (internal quotation omitted). Although the applicant then distinguished each claim on narrower grounds, we held that the argument with respect to individual claims did not "eliminate [the import] of global comments made to distinguish applicants' `claimed invention' from the prior art." Id.

In the final analysis, the Protest Response does not alter our conclusion that the statements Applicant made in responding to the First Action and in distinguishing over Morita and Benner inescapably narrowed the meaning of the data communications network in claim 37(40) to a point-to-point network. Applicant's response to the First Action was "a deliberate surrender of claim scope, unmistakable in its effect because it is not suitable to multiple interpretations." Omega Eng'g, 334 F.3d at 1327. The Protest Response addressed Gardner, did not alter the bases upon which the Applicant distinguished Morita and Benner, and did not undo this disclaimer already made. See Hockerson-Halberstadt, 222 F.3d at 957.

The district court erred in construing the "network for data communications" element of claim 37(40) and the claims dependent therefrom. The presumption attendant to claim differentiation doctrine is rebutted. The phrase "network for data communications" is limited to networks in which every processor system is connected to every other processor system via direct, point-to-point, two-way channel interconnections.

The district court construed "distributed computer system" to require that there be "a stand alone computer in each processor system." SeaChange, 313 F.Supp.2d at 398. The phrase "distributed computer system" appears in the preamble of method claim 37. Thus, a preliminary question exists as to whether "distributed computer system" is a claim limitation. See Catalina Mktg. Int'l v. Coolsavings.com, Inc., 289 F.3d 801, 807-08 (Fed. Cir.2002). "In general, a preamble limits the invention if it recites essential structure or steps, or if it is `necessary to give life, meaning, and vitality' to the claim." Id. at 808 (quoting Pitney Bowes, Inc. v. Hewlett-Packard Co., 182 F.3d 1298, 1305 (Fed.Cir.1999)). "[I]f the preamble helps to determine the scope of the patent claim, then it is construed as part of the claimed invention." NTP, Inc. v. Research In Motion, Ltd., 392 F.3d 1336, 1358 (Fed.Cir. 2004).

In this case, the preamble notes that the claimed method is for storing data in a "distributed computer system" which has "at least three processor systems." '312 patent, col. 22, ll. 17-18. The preamble further describes the components of a "processor system" to be "at least one central processing unit" and "at least one mass storage subsystem." Id., col. 22, ll. 19-20. The body outlines the claimed method, the steps of which repeatedly involve "said processor systems." Id., col. 22, ll. 22-29 (emphasis added). The preamble provides the only antecedent basis and thus the context essential to understand the meaning of "processor system"; therefore, the preamble, including the phrase "distributed computer system," limits the scope of the claimed invention. NTP, 392 F.3d at 1358-59 (relying on preamble of method claim to define scope of limitations that derived their antecedent basis from preamble).

The issue then is how, if at all, the preamble phrase "distributed computer system" impacts the scope of "processor system." The parties agree that the ordinary meaning of "distributed computer system" is a "computer system in which several interconnected computers share the computing tasks assigned to the system." IEEE 375 (defining "distributed system"). The parties dispute whether a "distributed computer system" requires that each processor within the system operate on its own power supply and clock, i.e., that each processor "stand-alone." Seachange argues that the IEEE definition implies multiple "stand-alone" computers and that "stand-alone" computers are necessary to implement RAID-5 at the system level. However, the IEEE definition does not say that computers must have a separate clock and power supply. Furthermore, Seachange points to nothing in the specification to suggest that the processors must "stand-alone" in order to achieve redundancy at the system level. Indeed, neither the claim nor the written description states such a requirement. See, e.g., '312 patent, col. 22, ll. 17-29; col. 2, ll. 17-23; col. 3, ll. 27-35; col. 4, ll. 36-44; col. 5, ll. 52-58. The written description describes a system where if one processor fails, another can perform the task that the failed processor was performing. See, e.g., id., col. 16, ll. 23-27. Such a system is consistent with the requirement that the processors "share computing tasks."

Because it is improper to import a limitation into a claim where the limitation has no basis in the intrinsic record, Ecolab, Inc. v. Envirochem, Inc., 264 F.3d 1358, 1366 (Fed.Cir.2001), we conclude that the district court erred in requiring that each processor system "stand-alone," i.e., have a separate clock and power supply. "Distributed computer system" should be given its ordinary meaning, which both parties agree is "a computer system in which several interconnected computers share computing tasks assigned to the system."

The district court construed "processor systems" to require that each system have "at least one [CPU] capable of running application type software, and at least one mass storage subsystem." SeaChange, 115 F.Supp.2d at 483. C-COR argues that the district court erroneously imported the "capable of running application type software" limitation from a preferred embodiment. Seachange counters that C-COR's position is inconsistent with the patent's objective of applying RAID-5 at the system level. Seachange asserts that because the system runs application software, if each processor system did not run the application software, then the claims would simply cover a prior art RAID-5 arrangement. Seachange points us to the language that a component of a "single processor system" is a "remote file provider," '312 patent, col. 8, II. 39-42, that "[t]he remote file provider represents any third party application or device driver that might use the cluster technology," id., col. 9, II. 7-9, and that "[e]xamples include Lotus Notes, medical applications, or database systems," id., II. 9-11. C-COR responds by noting that nothing in the ordinary meaning of CPU or in the intrinsic record requires that the claimed processor system be capable of running application software. C-COR adds that Seachange's citation to the written description bolster's C-COR's own position because the citation suggests that processors can simply execute device driver instructions.

The claim states that "each processor system compris[es] at least one [CPU] and at least one mass storage sub-system." '312 patent, col. 22, II. 19-20. The claim requires that "data [be] stored at each of [the] processor systems," id., col. 22, II. 25-28, presumably within the mass storage sub-system. However, the claim provides no more detail about the CPU and does not state that it must be capable of running application-type software. The ordinary meaning of CPU is "[t]he unit of a computing system that includes the circuits controlling the interpretation of instructions and their execution." IEEE 171. That definition likewise says nothing about a CPU being capable of running application-type software. Moreover, although the written description teaches an embodiment in which the CPU runs a "third party application," that is not the only embodiment. The written description says that the remote file provider can be any third-party application or device driver. '312 patent, col. 9, II. 7-9. A device driver need not be the same as application-type software. See IEEE 387 (defining "driver"). Thus, C-COR is correct that the written description supports a broader construction. Because we do not import limitations from a preferred embodiment, Fuji Photo Film Co., Ltd. v. Int'l Trade Comm'n, 386 F.3d 1095, 1106 (Fed.Cir.2004), each "processor system" need not have a CPU capable of running application software.

Seachange's final argument is a plea for the import of a limitation to help preserve the validity of the asserted claims. However, we must decline Seachange's invitation to import a limitation to preserve validity because this is not a case in which "the court concludes, after applying all the available tools of claim construction, that the claim is still ambiguous." Liebel-Flarsheim, 358 F.3d at 911. The limitation finds no support in the claims, written description, prosecution history, or technical dictionary.

For the foregoing reasons, we hold that the district court erred in its construction of the "processor system" limitation. A "processor system" must have a CPU but it need not be capable of running application-type software.

"A determination of infringement is a two-step process. The court must first correctly construe the asserted claims, and then compare the properly construed claims to the allegedly infringing devices, systems, or methods." NTP, 392 F.3d at 1364. "To prove infringement, the patentee must show that the accused device meets each claim limitation either literally or under the doctrine of equivalents." Catalina Mktg., 289 F.3d at 812.

In its post-argument brief, Seachange conceded that literal infringement would be precluded if we were to construe "network for data communications" to require that every processor be connected to every other processor point-to-point because C-COR's hypercube connects only certain pairs of nodes point-to-point. However, Seachange argues that the prosecution history does not preclude application of the doctrine of equivalents as a matter of law and seeks a remand. Seachange asserts that in distinguishing over Benner, the only thing that Applicant disclaimed was Benner's partially interconnected hypercube. Seachange argues that Applicant did not disclaim coverage of a system in which only some of the interconnections are point-to-point so long as each processor interconnects to each other processor, with those channels that are not point-to-point passing through intermediate nodes.

C-COR counters that prosecution history estoppel precludes a patentee from asserting as an equivalent subject matter that was disclaimed during prosecution. C-COR argues that because Seachange disclaimed coverage of all networks except those in which every processor is connected to every other processor with a point-to-point interconnection, Seachange cannot recapture through doctrine of equivalents coverage of a network that employs indirect connections. Alternatively, C-COR asserts that a finding of equivalents where only some processors are interconnected via point-to-point interconnections would "vitiate" the limitation that every processor be connected to every other processor by point-to-point interconnection.

Because Seachange concedes that C-COR does not literally infringe, there can only be infringement under doctrine of equivalents. "[T]he determination of infringement under the doctrine of equivalents is limited by two primary legal doctrines: (1) prosecution history estoppel and (2) the `all elements' rule." Lockheed Martin Corp. v. Space Sys./Loral, Inc., 324 F.3d 1308, 1318 (Fed.Cir.2003). The application of these doctrines is a question of law. Id. "[T]he `all elements rule' provides that the doctrine of equivalents does not apply if applying the doctrine would vitiate an entire claim limitation." Asyst Techs., Inc. v. Emtrak, Inc., 402 F.3d 1188, 1195 (Fed.Cir.2005). As noted supra Part II.B.1, we construed "network for data communications" to cover only "those networks in which every processor system is connected to every other processor system via direct, point-to-point, two-way channel interconnections." Seachange's equivalents theory implies that a network in which every processor is connected to every other processor through indirect interconnections can be equivalent to a network in which every processor is connected to every other processor by direct, point-to-point interconnections. However, equivalents under such a theory would vitiate the requirement that every processor be connected to every other processor point-to-point, and therefore must fail as a matter of law. See Asyst, 402 F.3d at 1195 (explaining that finding "unmounted" equivalent to "mounted on" would vitiate the limitation).

Because there is no infringement either literally or under doctrine of equivalents, C-COR is entitled to judgment of non-infringement as a matter of law. Because there is no need to remand, we reverse the judgment of infringement. CVI/Beta Ventures, Inc. v. Tura LP, 112 F.3d 1146, 1161-62 (Fed.Cir.1997); see Exxon Chem. Patents, Inc. v. Lubrizol Corp., 64 F.3d 1553, 1560-61 (Fed.Cir.1995). However, we must now consider C-COR's invalidity counterclaims. Cardinal Chem. Co. v. Morton Int'l, Inc., 508 U.S. 83, 95-103, 113 S. Ct. 1967, 124 L. Ed. 2d 1 (1993).

C-COR's challenge to the denial of JMOL was premised on the district court's construction of "network for data communications." Because we adopted C-COR's alternative argument that "network" in the asserted claims is limited to point-to-point networks, the premise of C-COR's written description challenge no longer exists. We thus affirm the district court's denial of C-COR's motion for JMOL.

"[A] claim is anticipated if each and every limitation is found either expressly or inherently in a single prior art reference." Celeritas Tech., Ltd., v. Rockwell Int'l Corp., 150 F.3d 1354, 1361 (Fed.Cir.1998). Whether a prior art reference anticipates a claimed invention is a question of fact. Advanced Display Sys., Inc. v. Kent State Univ., 212 F.3d 1272, 1281 (Fed.Cir.2000).

When we determine on appeal that a trial judge has misinterpreted a patent claim, we independently construe the claim to determine its correct meaning, and then determine if the facts presented at trial can support the judgment as a matter of law. See Exxon, 64 F.3d at 1559-60. The jury in this case found the claims in issue, as construed by the district court, not anticipated or rendered obvious by Gardner, Frey, and Mendelsohn. C-COR filed a motion for JMOL, and the district court denied the motion. C-COR appeals the denial of its motion as to anticipation based on each of these references. We review the district court's decision in light of our claim construction rulings.

We have broadened both the "distributed computer system" and the "processor systems" limitations. If these were the only limitations at issue and C-COR were properly entitled to JMOL of anticipation under the district court's narrower construction, our broader construction of those limitations would not change the outcome. See Teleflex, Inc. v. Ficosa N. Am. Corp., 299 F.3d 1313, 1328 (Fed.Cir.2002). However, we have more narrowly construed the "network for data communications" limitation to cover only point-to-point interconnections. If the jury was not presented with any evidence that the references disclosed point-to-point interconnections, JMOL would be precluded. See Exxon, 64 F.3d at 1561.

The district court's opinion does not discuss the details of the evidence before the jury on the "network for data communications" limitation, nor do the parties in their briefs in this appeal elucidate the evidence presented to the jury to establish that this limitation was disclosed by Gardner, Frey, or Mendelsohn. However, the district court did explain that

Gardner does not disclose the '312 Patent's claimed "processor systems" as the court has defined it. Specifically, the Court concludes that Gardner does not teach or suggest a system in which each computer of the distributed computer system runs both client applications and has disks. Therefore, the Court concludes that the jury's rejection of [the] anticipation claim was not against the weight of the evidence.

SeaChange, 313 F.Supp.2d at 401.

C-COR argues on appeal, as it did in the district court, that Gardner discloses the "processor systems" limitation, as evidenced by the following passage:

each client [application] may be a process running on any CPU, and need not be limited to executing on a CPU different from that of the media server. Although such a "combined" configuration could have potential drawbacks, it is of course possible to have a media client [application] and media server executing on the same machine.

Gardner, col. 15, II. 16-21. C-COR asserts that Seachange's own expert, Dr. Rhyne, admitted at trial that "if you had one computer serving as the client and the server, you would be running an application on a processor that had a disk. You would meet the processor requirement." (Rhyne Test. of Sept. 21, 2000, at 826, II. 13-17.) Seachange responds that Dr. Rhyne explained that Gardner also teaches a system in which a computer runs only the "media portion" (the application), with the "disk interface server" located elsewhere. (Id. at 837, II. 14-16.)

Figure 1 of Gardner discloses a system employing a plurality of media servers, with the media clients located elsewhere. Gardner, Figure 1. The text of Gardner at column 15, lines 16-21, discloses that a media client could be combined with a media server on one machine. According to Dr. Rhyne, such an embodiment would meet the "processor requirement." To be anticipating, Gardner need not disclose a separate figure depicting a plurality of machines with the combined media client and media server; disclosing the embodiment in textual form is enough. The fact that another embodiment is disclosed does not detract from the remainder of the disclosure. See Hewlett-Packard, 340 F.3d at 1324 & n. 6 ("The anticipation analysis asks solely whether the prior art reference discloses and enables the claimed invention, and not... whether alternatives are also disclosed."). Indeed, Seachange acknowledges that Gardner discloses that "a single client could combine a client and a server," but argues that the reference "teaches away" from the combination of a client and a server, and that the testimony of C-COR's expert, Dr. Wilkes, was unclear. Seachange's argument that Gardner teaches away from a computer with a combined media client and media server is misplaced. Teaching away is irrelevant to anticipation. Celeritas, 150 F.3d at 1361; Bristol-Myers Squibb Co. v. Ben Venue Labs., Inc., 246 F.3d 1368, 1378 (Fed.Cir. 2001).

In addition, Seachange's challenge to Dr. Wilkes' testimony lacks support in the record. In response to Dr. Rhyne's testimony suggesting that the "processor system" limitation, as construed by the district court, was the only limitation missing in Gardner and that an embodiment showed a separated media client and server (see Rhyne Test. of Sept. 21, 2000, at 835, I. 10-838, I. 6), Dr. Wilkes testified:

Figure 1, which is the one [Dr. Rhyne] used to talk about a client server picture... does indeed have separate clients and servers connected by a network. So yes, from that figure, you might conclude that client server was the only thing that was taught. And given that the way this stuff is put together, Claim 37 requires, that's a no, no, fails that test.

But if you turn to Column 15, Lines 16 through 21, you find some interesting text. Line 16. It should be noted that each client may be a process running on any CPU. It's just the software thing that matters and need not be limited to executing on a CPU different from that of a media server.

That's what it says. Although such a combined configuration could have potential drawbacks. It is, of course, possible to have a media client and a media server executing on the same machine.

The claim here merely requires that it be possible, not that it be necessary [sic], the best thing to do in the world.

(Wilkes' Test. of Sept. 22, 2000, at 1063, I. 3-1064, I. 5.).

Seachange, in the alternative, argues that C-COR did not prove that Gardner disclosed either the "distributed computer system" or the "storing data input" limitations. As to the "distributed computer system" limitation, Dr. Rhyne twice testified that Gardner "is a distributed computer system." (Rhyne Test. of Sept. 21, 2000, at 837, II. 9-11, II. 20-22.) As to the "storing data input" limitation, Dr. Wilkes testified that "[Gardner] also talk[s] about using RAID-5 across the media servers, plural, in order to provide redundancy in case one of the media servers breaks.... They talk about using RAID-5 in each of the media servers for extra redundancy." (Wilkes' Test. of Sept. 20, 2000, at 606, II. 3-8). Indeed, Gardner describes a system with redundancy within a media server, Gardner, col. 2, II. 24-26, and across media servers, id., II. 34-37.

Because we have concluded that the district court's claim constructions of both the "distributed computer system" and "processor systems" limitations were in error, because several of the district court's conclusions on the evidence presented on anticipation by Gardner are contrary to the record, and because the evidentiary record as to the disclosure of Gardner on the "network for data communications" limitation has not been fully presented in this appeal, we vacate the district court's denial of C-COR's motion for JMOL on anticipation. We remand the issue for reconsideration in light of our claim construction rulings and the foregoing discussion of the evidence. If the district court determines that, on the basis of clear and convincing evidence in the record presented to the jury, no reasonable juror could not find anticipation based on Gardner, the district court should grant JMOL to C-COR on that ground. If not, JMOL based on Gardner should be denied.

As for the Frey and Mendelsohn references, the district court reviewed the evidence and concluded that C-COR had failed to establish that the weight of the evidence proved that the claims were anticipated. We have considered the parties' arguments and agree with the district court's assessment that substantial evidence supported the jury's verdict with regard to the Frey and Mendelsohn references. The denial of JMOL based on the lack of anticipation by Frey or Mendelsohn is affirmed.

An erroneous instruction on claim interpretation that affects the jury's decision on anticipation is grounds for a new trial. See Advanced Display, 212 F.3d at 1281-82; cf. Ecolab, Inc. v. Paraclipse, Inc., 285 F.3d 1362, 1373 (Fed.Cir. 2002). A party seeking to set aside a judgment based on erroneous jury instructions must establish that "(1) it made a proper and timely objection to the jury instructions, (2) those instructions were legally erroneous, (3) the errors had prejudicial effect, and (4) it requested alternative instructions that would have remedied the error." NTP, 392 F.3d at 1365. "[W]hen the error in a jury instruction `could not have changed the result, the erroneous instruction is harmless.'" Id. (quoting Environ Prods., Inc. v. Furon Co., 215 F.3d 1261, 1266-67 (Fed.Cir.2000)). For prejudicial error to exist, there must have been sufficient evidence proffered below to support anticipation under a correct construction. See id. at 1365-66; Ecolab, 285 F.3d at 1374.

Over C-COR's timely objection, see SeaChange, 313 F.Supp.2d at 393-99, the district court instructed the jury based upon its claim construction rulings but refused C-COR's request for an instruction on the construction of the "distributed computer system" limitation. We have now agreed with C-COR's proposed construction of that limitation. Because the district court has already held that its failure to instruct the jury on the "distributed computer system" limitation was error, and because the district court also has held that such error would be prejudicial and would require a new trial if we decide, as we have, that it should have adopted C-COR's proposed construction, we reverse the denial of C-COR's motion in the alternative for new trial and remand for such further proceedings as may be deemed appropriate by the district court consistent with this opinion.

Because the district court erred in interpreting "network for data communications" and C-COR does not infringe as a matter of law, we reverse the judgment of infringement. Because we have construed the "network" limitation to be limited to point-to-point interconnections, we affirm the denial of C-COR's motion for JMOL on inadequate written description. Because the district court correctly denied C-COR's motion for JMOL on anticipation as to the Frey and Mendelsohn references, we affirm that ruling. For the reasons stated in this opinion, we vacate the denial of the JMOL motion as to anticipation by Gardner and remand the same to the district court for reconsideration. Because the district court erred in its interpretation of "distributed computer system" and because C-COR was prejudiced by the court's failure to properly instruct the jury on the construction of the "distributed computer system" limitation, we reverse the district court's denial of C-COR's motion in the alternative for new trial and remand for further proceedings consistent with this opinion.

Each party shall bear its own costs.

AFFIRMED-IN-PART, REVERSED-IN-PART, VACATED-IN-PART, AND REMANDED.