RADER

Andrew G. DiNovo, DiNovo, Price, Ellwanger & Hardy, LLP, of Austin, TX, argued for plaintiff-appellant. With him on the brief were Stephen L. Neal, Jr., DiMuro Ginsberg, PC, of Alexandria, VA, Michael P. Adams, Winstead Sechrest & Minick, P.C., of Austin, TX, and Gregory R. Lyons, Wiley Rein, LLP, of Washington, DC. Of counsel were Bernard J. Dimuro, DiMuro, Ginsberg, PC, of Alexandria, VA, and William M. Parrish, DiNovo, Price, Ellwanger & Hardy, LLP, of Austin, TX. Winstol D. Carter, Jr., Morgan, Lewis & Bockius, LLP, of Houston, TX, argued for defendants-appellees. With him on the brief were David J. Levy, C. Erik Hawes, James A. Glenn and David M. Morris.

Before RADER, Chief Judge, LINN and MOORE, Circuit Judges.

Opinion dissenting-in-part filed by Circuit Judge MOORE.

RADER, Chief Judge.

The United States District Court for the Eastern District of Virginia granted The University of Phoenix, Inc. and Apollo Group, Inc.'s (collectively, “Phoenix”) motion for summary judgment of non-infringement of all asserted claims. Because the district court's determination was based in part on an erroneous claim construction, this court vacates the summary judgment of non-infringement with respect to some claims, affirms with respect to others, and remands for further proceedings.

I

This appeal features three patents: U.S. Patent Nos. 6,170,014 (“the '014 patent”), 6,282,573 (“the '573 patent”), and 6,606,664 (“the '664 patent”). These patents are directed to regulating access to content that is delivered through a computer network. All three patents belong to the same family and share the same specification.

The inventors initially filed Application No. 09/272,221 (“the '221 application”), including claims covering two distinct inventions. The Patent and Trademark Office (“PTO”) found that claims 1–15 of the '221 application were drawn to “managing content in a shared environment through a security manager and a funds flow manager,” while claims 16–37 were drawn to a distinct invention involving a “preregistered user accessing content that contains previously treated critical portion.” J.A. 3595. Accordingly, the PTO imposed an election/restriction requirement. Hence, the inventors continued to pursue the first set of claims in the '221 application, which eventually issued as the '014 patent, while separately pursuing the second set of claims in a divisional application, which eventually issued as the '573 patent.

The claimed invention in the '014 patent is a multilevel computer architecture with a registration server level, a content server level, and a client level. '014 patent col.23 ll.14–37. The architecture prevents unauthorized use of content, in part, by keeping the content on a content server, separate from the registration server (which is free of such content). Id.

The '573 patent claims a method and a computer architecture for regulating access to content in a shared use environment. '573 patent col.23 l.47–col.26 l.43. The claimed invention prevents unauthorized use of content, in part, by “treating” (i.e., separating, encrypting, compressing, or otherwise enhancing protection) a critical portion of the content and only delivering this critical portion to registered users. Id.

The '664 patent is a continuation of the application that issued as the ' 573 patent. The '664 patent claims a multi-level computer architecture including a registration server level, a content server level, and a client level and having a particular type of client-server network communication link. '664 patent col.23 l.49–col.25 l.12.

Digital–Vending Services International, LLC (“Digital–Vending”) filed suit in the United States District Court for the Eastern District of Texas against Phoenix, Capella Education Company (“Capella”), and Walden University, Inc. (“Walden”), alleging infringement of the '014 patent, the '573 patent, and the '664 patent. The Texas court transferred the case to the United States District Court for the Eastern District of Virginia. After a Markman hearing, the Virginia district court issued a claim construction order. Capella and Walden entered into a settlement agreement with Digital–Vending. Phoenix filed a motion for summary judgment of non-infringement of all asserted claims. Digital–Vending then filed a motion for “clarification,” which sought reconsideration of the court's construction of the term “registered user.” The district court granted the motion for summary judgment of non-infringement, denied the motion for “clarification,” and entered judgment in favor of Phoenix. Digital–Vending appealed the district court's construction of various claim terms and the resulting grant of summary judgment of non-infringement. This court has jurisdiction under 28 U.S.C. § 1295(a)(1).

II

This court reviews claim constructions without deference. Cybor Corp. v. FAS Techs., 138 F.3d 1448, 1451 (Fed.Cir.1998) (en banc). However, “a party may not introduce new claim construction arguments on appeal or alter the scope of the claim construction positions it took below.” Conoco, Inc. v. Energy & Envtl. Int'l, L.C., 460 F.3d 1349, 1358–59 (Fed.Cir.2006). Contrary to Digital–Vending's suggestion, Blackboard, Inc. v. Desire2Learn, Inc., 574 F.3d 1371 (Fed.Cir.2009), does not stand for the proposition that a party is free to seek a claim construction on appeal substantially different from the construction it proposed below simply because the district court “construed the claim language in a manner different from the construction proposed by either party.” Id. at 1378. In Blackboard, the appellee's claim construction position on appeal was consistent with its earlier proposed construction but inconsistent with a statement made during a colloquy at the Markman hearing. Id. This court explained that, in light of the specific factual circumstances in that case, the question of waiver was a “difficult one” and ultimately held that the appellee had not waived its validity challenge, which relied upon an issue of claim construction. Id. This court did not hold that a party was free to argue a claim construction different from both its earlier proposed construction and the district court's construction simply because the district court had not adopted either party's proposed constructions. To the contrary, this court has often stated that a party may not, as a general rule, change the scope of its claim construction on appeal. See, e.g., CCS Fitness, Inc. v. Brunswick Corp., 288 F.3d 1359, 1371 (Fed.Cir.2002) (“Our precedent makes clear that in the context of claim construction, a waiver may occur if a party raises a new issue on appeal, as by, e.g., presenting a new question of claim scope.”); Interactive Gift Express, Inc. v. Compuserve, Inc., 256 F.3d 1323, 1347 (Fed.Cir.2001) (concerns about the proper role of appellate review “preclude a party from changing its claim construction, that is, the scope of its claim construction, on appeal”).

All of the claims in the '014 patent, claims 23–37 of the '573 patent, and all of the claims of the '664 patent refer to “computer architecture for managing content,” with a registration server “free of content managed by the architecture” and a content server “containing content managed by the architecture.” See, e.g., '014 patent col.23 ll.14–38. During claim construction briefing and argument before the district court, Digital–Vending took the position that “content managed by the architecture” either required no construction or should be construed as “courseware and other content such as audio, video, text and interactive software.” J.A. 54. Phoenix proposed construing the term to mean “digital data controlled by any part of the architecture.” Id. The district court rejected both parties' proposed constructions and instead construed “content managed by the architecture” to mean “digital material capable of being transmitted over a computer network that is being sold or licensed through the architecture.” J.A. 55.

Digital–Vending argued before the district court that summary judgment of non-infringement was improper even under the district court's construction of “content managed by the architecture.” On appeal, however, Digital–Vending has not challenged the district court's grant of summary judgment of non-infringement under the district court's construction. Instead, Digital–Vending now seeks to overturn the district court's grant of summary judgment of non-infringement based on a construction of “content managed by the architecture” that is different from both the construction it proposed below and the district court's construction. Digital–Vending argues that this court should construe “content managed by the architecture” to mean “the digital product being commercialized, that is delivered over a computer network, and that the network architecture protects.”

This new construction is substantially different, having a far more limited scope, compared to the construction Digital–Vending proposed below. Digital–Vending's earlier proposed construction, “courseware and other content such as audio, video, text and interactive software,” J.A. 54, was very broad and merely listed examples of types of content included, without excluding any types of content or otherwise delineating the outer reaches of the claim term. The district court's construction limits “content managed by the architecture” to “digital material capable of being transmitted over a computer network that is being sold or licensed through the architecture.” J.A. 55. Despite the fact that the district court adopted a claim construction narrower in scope than the construction Digital–Vending itself proposed, Digital–Vending now argues that the district court's construction is too broad because it includes material that is licensed without charge to registered users. Because Digital–Vending's newly proposed construction for “content managed by the architecture” is substantially different in scope from the construction it sought below, this court will not attempt to review an argument not presented first to the trial court.

III

All of the asserted claims, other than claims 13–22 of the '573 patent, explicitly require a “registration server.” The district court construed “registration server” as “[a] server for performing new user registration that contains at least a portion of a remote registration manager and a registration database. Registration servers must be separate from content servers and be free of content managed by the architecture.” J.A. 38.

The district court erred in requiring a registration server to be free of content managed by the architecture. This court has noted that “the context in which a term is used in the asserted claim can be highly instructive.” Phillips v. AWH Corp., 415 F.3d 1303, 1314 (Fed.Cir.2005) (en banc). The context in which the term “registration server” is used in the claims strongly suggests that a registration server does not have the inherent characteristic of being free of managed content. Many of the asserted claims specifically require the registration server to be free of content managed by the architecture, while other asserted claims merely require a registration server, without stating this additional limitation. In particular, all of the claims in the '014 patent, all of the claims in the '664 patent, and claims 23–37 of the '573 patent explicitly require “at least one registration server, each registration server comprising a remote registration manager and a registration database for new user registration, and each registration server being further characterized in that it is free of content managed by the architecture.” '014 patent col.23 ll.17–22 (emphasis added); '664 patent col.23 ll.52–57; '573 patent col.25 ll.17–22. In contrast, claims 1–22 of the '573 patent do not contain this language. Claims 1–12 of the ' 573 patent simply refer to a “registration server” without further description, while claims 13–22 of the '573 patent do not contain the phrase “registration server.”

“Registration server” is presumed to have the same meaning throughout all of the claims in the absence of any reason to believe otherwise. See Fin Control Sys. Pty, Ltd. v. OAM, Inc., 265 F.3d 1311, 1318 (Fed.Cir.2001) (“the presumption [is] that the same terms appearing in different portions of the claims should be given the same meaning unless it is clear from the specification and prosecution history that the terms have different meanings at different portions of the claims”). If “registration server” were construed to inherently contain the “free of content managed by the architecture” characteristic, the additional “each registration server being further characterized in that it is free of content managed by the architecture” language in many of the asserted claims would be superfluous. This construction is thus contrary to the well-established rule that “claims are interpreted with an eye toward giving effect to all terms in the claim.” Bicon, Inc. v. Straumann Co., 441 F.3d 945, 950 (Fed.Cir.2006). In Phillips, this court reinforced the importance of construing claim terms in light of the surrounding claim language, such that words in a claim are not rendered superfluous. 415 F.3d at 1314. For example, when a claim refers to “steel baffles,” this “strongly implies that the term ‘baffles' does not inherently mean objects made of steel.” Id. In this case, the reference in some claims to a “registration server being further characterized in that it is free of content managed by the architecture” strongly implies that the term “registration server,” standing alone, does not inherently mean a server that is free of managed content.

Parts of the specification describe the registration server as being free of managed content. However, it is worth noting that claims 1–22 of the ' 573 patent relate to methods for protecting content, while claims 23–37 of the ' 573 patent relate to a computer architecture for protecting content. As noted above, the architecture claims explicitly require the registration server to be free of managed content, as this is one way in which the architecture prevents unauthorized use of the managed content. See ' 573 patent col.25 l.13–col.26 l.43. The method claims, on the other hand, do not recite any requirement of keeping the registration servers free of managed content; rather they describe treating a critical portion of the managed content, as a means for preventing unauthorized use. See id. col.23 l.47–col.25 l.12. Notably, the inventors only referred to the requirement of keeping registration servers free of managed content when describing embodiments of the claimed architecture and never referred to such a requirement in the parts of the specification describing the claimed methods. Thus, the specification used the term “registration server” in a manner consistent with the differences in claim language. Moreover the careful distinctions in specification descriptions avoid any hint that the inventors clearly disavowed claim scope with respect to the method claims. See Epistar Corp. v. Int'l Trade Comm'n, 566 F.3d 1321, 1334 (Fed.Cir.2009) (requiring “expressions of manifest exclusion or restriction, representing a clear disavowal of claim scope” in order to limit claims based on language in the specification).

For example, the “Overview of the Architecture” section, which is devoted to describing the claimed architecture in Figure 1, states that “each registration server 108 is free of courseware or other deliverable content that is managed by the architecture 100.” '573 patent col.8 ll.28–30. This section explains further that “a content server 110 and a registration server 108 may not reside on the same computer because that would violate the requirement that registration servers 108 not contain courseware.” Id. col.9 ll.12–15. Likewise, the “Content Server” section, which describes the role of the content server in the claimed architecture, provides that “[u]nlike the registration server 108, the content server 110 contains courseware and/or other managed content 400.” Id. col.13 ll.36–37. In the section discussing security “in the architecture 100,” the specification states, “[b]ecause content is not stored on the registration server 108, security precautions can be taken that might not otherwise be available.” Id. col.22 ll.61–63.

In contrast, the portions of the specification describing embodiments of the claimed methods do not suggest that the registration server cannot contain any managed content. See '573 patent col.16 l.37–col.22 l.45. Figures 6 and 7, which “illustrate methods of the claimed invention,” also do not provide any reason to believe that the registration server used in the claimed methods must be free of managed content. Id. col. 16 ll.39–40.

In the “Brief Summary of the Invention,” the specification lists various ways in which the present invention provides additional security: “For instance, additional security is provided by separating registration information from content, by identifying and treating critical portions, and by monitoring the connection over which content is supplied to a client.” '573 patent col.7 ll.10–14. This statement merely enumerates three different ways in which the claimed invention protects content—(1) keeping the registration server free of managed content, (2) treating critical portions, and (3) monitoring the content server-client connection. This statement—again far from a disavowal of claim scope—does not even suggest that every embodiment of the invention must contain all three features.

Similarly, the prosecution history does not provide any basis for reading a “free of content managed by the architecture” limitation into the stand-alone phrase “registration server.” “[B]ecause the prosecution history represents an ongoing negotiation between the PTO and the applicant, rather than the final product of that negotiation, it often lacks the clarity of the specification and thus is less useful for claim construction purposes.” Phillips, 415 F.3d at 1317. For this reason, it is particularly important not to limit claim scope based on statements made during prosecution “[a]bsent a clear disavowal or contrary definition.” August Tech. Corp. v. Camtek, Ltd., 655 F.3d 1278, 1286 (Fed.Cir.2011) (quoting Home Diagnostics, Inc. v. LifeScan, Inc., 381 F.3d 1352, 1358 (Fed.Cir.2004)).

The PTO initially rejected claims 1–3, 7–15, and 16–37 of the '221 application as obvious in light of U.S. Patent No. 5,974,409 (the “Sanu” reference). The PTO later imposed an election/restriction requirement on the '221 application, finding that it covered two distinct inventions. Accordingly, claims 1–3 and 7–15 of the '221 application eventually issued as claims of the '014 patent, while claims 16–37 later issued as claims of the '573 patent (based on a divisional application). In response to the PTO's rejection of claims 1–3 and 7–15, the inventors argued that Sanu did not render these claims obvious because: (1) it did not teach a registration server; (2) it did not teach a registration server for remote registration of a new user; (3) it did not teach a content-free registration server; and (4) it did not teach content servers that serve content only to registered users. This account shows that the inventors unequivocally disavowed claim scope covering registration servers that were not content free for claims 1–3 and 7–15, which eventually issued as claims of the '014 patent. Consistent with this disavowal, these issued claims also explicitly contain the “free of content managed by the architecture” claim limitation.

Phoenix argues that the inventors also disavowed this same claim scope with respect to claims 16–37, which later issued as claims of the '573 patent, even for claims without the “free of content managed by the architecture” language. Phoenix's argument rests solely on the following statement by the inventors: “The rejections of claims 16–37 under 35 U.S.C. § 103(a) rely on the same reasons set forth in the rejection of claims 1–3, 7–15. For the reasons explained above, the rejection of claims 16–37 should therefore be withdrawn as well.” J.A. 3958. To the contrary, the inventors went on to explain that “claims 16–37 have limitations that are significantly different from the limitations in claims 1–3, 7–15” and performed additional analysis regarding the rejection of claims 16–37. In the overall context of the prosecution history, the inventor's statements do not clearly disavow claim scope. As noted above, the “reasons explained above” include four alleged differences between Sanu and the claimed invention. Some of these alleged differences were clearly relevant to claims 16–37, such as the requirement of a registration server. Hence, this general reference to arguments from the “Remarks on Claims 1–3, 7–15” section applies to arguments about claim language that was repeated in claims 16–37, such as Sanu's failure to teach a registration server or a registration server for remote registration of a new user, as opposed to its failure to teach a content-free registration server. This reading of the inventor's statements is particularly reasonable in light of the language of the later issued claims. Moreover the inventors never argued that the registration servers had to be free of content when pursuing the divisional application that gave rise to the '573 patent. When the inventors' statements “are considered in the context of the prosecution history as a whole, they simply are not clear and unmistakable enough to invoke the doctrine of prosecution history disclaimer.” Ecolab, Inc. v. FMC Corp., 569 F.3d 1335, 1343 (Fed.Cir.2009). This court therefore declines to read the “free of content managed by the architecture” limitation that is explicitly recited in various claims into the stand-alone phrase “registration server.”

IV

Claims 13–22 of the '573 patent do not contain the words “registration server” but require a “registered user.” '573 patent col.24 ll.40, 44–45. The parties stipulated that the term “registered user” should be construed as “a person who has previously been registered with a registration server.” J.A. 62. Before the Markman hearing, Phoenix made clear to Digital–Vending on two occasions that this construction would mean that the claims having a “registered user” limitation would inherently also require a registration server. Digital–Vending did not object, and the district court accepted their stipulation, construing the term “registered user” accordingly.

After Phoenix moved for summary judgment of noninfringement, relying in part on this stipulated construction of “registered user,” Digital–Vending filed a motion for “clarification,” which in substance was a motion for reconsideration, asking the court to construe “registered user” as “a person who has an existing account with the architecture.” J.A. 1841. The district court denied this motion, explaining that it was not “appropriate to provide [Digital–Vending] a second bite at the apple regarding Markman disputes at such a late stage in the litigation, as [Phoenix] and its experts have relied on the court's Markman ruling in completing discovery and formulating a position on summary judgment.” J.A. 18.

On appeal, Digital–Vending takes the position that the district court erred in construing “registered user” as requiring a registration server and argues that this claim term should be construed more broadly to include users registered by other means, such as by telephone or by mail. By stipulating to the construction that the district court adopted, Digital–Vending waived its right to challenge this construction on appeal. See SuperGuide Corp. v. DirecTV Enters., Inc., 358 F.3d 870, 889 (Fed.Cir.2004) (finding waiver of claim construction argument where the party agreed to a construction before the district court). Moreover, Digital–Vending has not challenged the district court's denial of its motion for reconsideration. Therefore, this court shall not entertain Digital–Vending's claim construction argument regarding this claim term.

V

Digital–Vending argues the district court erred by construing “server” as limited to a single computer, rather than including a combination of computers. However, the district court did not construe the term “server.” The district court did not construe the term “server” in its claim construction order and also did not reach this issue in its summary judgment order. Phoenix made several arguments in moving for summary judgment of non-infringement, one of which was that its system did not infringe because its registration manager and registration database were stored on separate computers in different buildings. For Phoenix to prevail on this non-infringement argument, “server” must be construed as excluding a combination of computers that are housed separately. The district court did not reach this issue because it granted summary judgment of non-infringement on another basis. See J.A. 17 (“the court does not squarely reach the issue of [Phoenix's] registration manager and registration database being housed on computers physically located in different buildings”); see also J.A. 18 (“based on the complexity of the issues and because the court's above determination is dispositive, the additional grounds presented in support of summary judgment are not discussed herein”).

As a general rule, “a federal appellate court does not consider an issue not passed upon below.” Interactive Gift Express, 256 F.3d at 1344 (quoting Singleton v. Wulff, 428 U.S. 106, 120, 96 S.Ct. 2868, 49 L.Ed.2d 826 (1976)). “This is because appellate courts are courts of review and no matter how independent an appellate court's review of an issue may be, it is still no more than that—a review.” Id. at 1344 (internal quotations omitted). Digital–Vending has not presented any reason to deviate from this general rule, and therefore, this court shall not construe “server” in the first instance.

VI

The district court granted summary judgment of non-infringement based on its construction of the terms “registration server” and “content managed by the architecture.” As discussed above, the district court construed “registration server” as requiring the server to be “free of content managed by the architecture,” J.A. 38, and construed “content managed by the architecture” as “digital material capable of being transmitted over a computer network that is being sold or licensed through the architecture,” J.A. 55. Because every asserted claim requires a “registration server” (including claims 13–22 of the '573 patent, based on the stipulated construction of “registered user”), the district court found that infringement of each claim required a registration server free of digital material capable of being transmitted over a computer network that is being sold or licensed through the architecture.

Digital–Vending identified Phoenix's Apply Web computers as part of the alleged “registration server.” Phoenix presented undisputed evidence that these Apply Web computers contain the following digital content: the Phoenix logo, a course catalog, a financial options guide, and various documents relating to student privacy, tuition, and financial aid. The district court found that these materials were “content managed by the architecture” because the Phoenix website contained a “Terms of Use” digital document indicating that website users were granted a limited license to access these materials. Accordingly, the district court held that the alleged “registration server” was not free of “content managed by the architecture,” as required for every asserted claim under its claim constructions, and granted summary judgment of non-infringement of all asserted claims on this basis.

For the reasons explained above, under the correct claim construction, a “registration server” does not have to be free of content managed by the architecture. Accordingly, Digital–Vending need not show that the registration server is free of managed content to prove infringement of claims 1–22 of the ' 573 patent, which refer to a “registration server” or a “registered user” but do not contain the “free of content managed by the architecture” limitation. Therefore, this court vacates the grant of summary judgment of non-infringement of claims 1–22 of the '573 patent.

In the alternative, Phoenix asks this court to affirm the district court's grant of summary judgment of non-infringement based on the alleged failure of the accused system to meet the requirement of having a registration server that contains both a portion of a remote registration manager and a registration database. The record shows that Phoenix's computers hosting Apply Web (the alleged portion of a remote registration manager) and Phoenix's computers hosting the V3 database (the alleged registration database) are physically separate and located in different buildings several miles apart. Phoenix's non-infringement argument relies upon a construction of “server” that excludes a combination of computers that are housed separately. Because the district court did not construe the term “server” and did not reach this issue, this court remands to allow the district court to consider this alternative non-infringement argument in the first instance. See Interactive Gift Express, 256 F.3d at 1344.

On appeal, Digital–Vending has only challenged the district court's grant of summary judgment of non-infringement based on erroneous claim construction and has not challenged the grant of summary judgment of non-infringement under the district court's claim construction. The only claim construction error this court has found on appeal is the improper addition of a “free of content managed by the architecture” limitation to the term “registration server.” However, all of the asserted claims, other than claims 1–22 of the ' 573 patent, explicitly require not only a registration server but a “registration server being further characterized in that it is free of content managed by the architecture.” See, e.g., '573 patent col.25 ll.20–22. Accordingly, the district court's construction of “registration server” is harmless error with respect to these claims. This court thus affirms the grant of summary judgment of non-infringement for claims 23–37 of the '573 patent and all asserted claims in the '014 patent and the '664 patent.

AFFIRMED–IN–PART, VACATED–IN–PART, AND REMANDED

Costs

No costs.

MOORE, Circuit Judge, dissenting-in-part.

I join the majority opinion in all respects but one. Because I conclude that the district court correctly construed the term “registration server,” I would affirm the district court's grant of summary judgment of non-infringement of claims 1–22 of the '573 patent. The district court correctly held that the claimed “registration server” must be “free of content managed by the architecture.” This case presents one of the rare instances where a patentee clearly disavowed claim scope through limiting language in the specification. The majority errs and allows the patent owner to reclaim surrendered claim scope, thus subverting the public notice function of patents.

Our case law is clear that a claim term should be accorded the full breadth of its plain and ordinary meaning as understood by a person of ordinary skill in the art when read in the context of the specification and prosecution history. See Phillips v. AWH Corp., 415 F.3d 1303, 1313 (Fed.Cir.2005) (en banc). But there are two limited exceptions to this rule: if the patentee acted as its own lexicographer or clearly disavowed the plain and ordinary meaning of a claim term. Thorner v. Sony Computer Entm't Am. LLC, 669 F.3d 1362 (Fed.Cir.2012).

The standard for disavowal is exacting. “Where the specification makes clear that the invention does not include a particular feature, that feature is deemed to be outside the reach of the claims of the patent, even though the language of the claims, read without reference to the specification, might be considered broad enough to encompass the feature in question.” SciMed Life Sys., Inc. v. Advanced Cardiovascular Sys., Inc., 242 F.3d 1337, 1341 (Fed.Cir.2001). Language giving rise to disavowal must amount to “expressions of manifest exclusion or restriction, representing a clear disavowal of claim scope.” Teleflex, Inc. v. Ficosa N. Am. Corp., 299 F.3d 1313, 1325 (Fed.Cir.2002).

The plain and ordinary meaning of the term “registration server” does not require that the server be free of content managed by the architecture. But the patentee disavowed the full scope of this claim term with its repeated statements in the specification to the contrary. In its initial description of the registration server, the specification states that “each registration server 108 is free of courseware or other deliverable content that is managed by the architecture 100. In particular, courseware is not stored on the registration server 108.” '573 patent col.8 ll.29–33. That data is stored on the content server. Id. col.13 ll.36–37. The patentee clearly disavowed the full scope of the term “registration server” when it stated:

A given computer may host several content servers 110, or it may host several registration servers 108, but a content server 110 and a registration server 108 may not reside on the same computer because that would violate the requirement that registration servers 108 not contain courseware. Id. col.9 ll.10–15 (emphasis added). In discussing the content server, the specification again emphasizes that “[u]nlike the registration server 108, the content server 110 contains courseware and other managed content.” Id. col.13 ll.36–37. The specification describes the reason for separating the registration functions from content, stating “additional security is provided by separating registration information from content.” Id. col.7 ll.10–11; col.22 ll.61–63 (“Because content is stored on the registration server 108, security precautions can be taken that might not otherwise be available.”).

It is difficult to imagine a clearer case of disavowal. The patentee states that it is a “requirement” that a registration server include no content managed by the architecture. This is exactly the type of clear and unmistakable language that is sufficient to place the public on notice that the patentee has surrendered the full scope of a claim term. We have found disavowal in cases with less restrictive language. For example, in The Toro Co. v. White Consolidated Industries, Inc., 199 F.3d 1295, 1301–02 (Fed.Cir.1999), we held there was disavowal when the specification disclosed a single embodiment and described the relevant structure as “important to the invention.” The instant case is much more compelling as the specification expressly states that it is a “requirement” that the registration server be free of content managed by the architecture.

The majority does not deny the disavowal in the specification, but rather concludes that it is limited to claims 23–27—the architecture claims. The majority holds that the disavowal does not clearly and unmistakably apply to the corresponding method claims 1–22. The majority also concludes that there is no disavowal because a holding of disavowal would render superfluous the language in claim 23, which states “each registration server being further characterized in that it is free of content managed by the architecture.” Majority Op. 1273–76.

The majority's claim that the portions of the specification discussing the claimed methods do “not suggest that the registration server cannot contain any managed content,” Majority Op. at 1276, is with all due respect incorrect. The distinction the majority draws between the “architecture” disclosure and “method” disclosure is contradicted by the same portion of the specification the majority cites. For example, the specification plainly states that “FIG. 7 illustrates methods for operating architecture 100,” including locating “the service provider Web site, which is hosted by the registration server 108.” ' 537 patent col.20 ll.39–44 (emphasis added); see also id. col.22 ll.29–30 (“content server 110 communicates that acceptance to the funds flow manager 308 in the registration server 108 ”). This is the same “registration server 108” that was expressly limited by the patentee earlier in the specification. When the specification refers to this registration server, by reference number, it is the one that has already been limited to a server that is free of content managed by the architecture.

Our precedent does not require that a patentee restate disavowal language repeatedly throughout the specification every time it references the same element; such a rule would be nonsensical. The method in this case is unequivocally the method of operating the defined architecture. The portion of the specification disclosing the method expressly refers by reference number to the architecture and its components. In short, there are not two different registration servers: one in the architecture claims and a different one in the method claims. At all times, the methods use a specific registration server— registration server 108—the very one the majority acknowledges must be free of managed content. And if any doubt remained, the specification expressly states that the “methods of the present invention” use the architecture described in the specification. Id. col.16 ll.48–53 (“Unless otherwise expressly indicated, the description herein of methods of the present invention therefore extends to corresponding systems and configured storage media, and the description of systems and configured storage media of the present invention extends likewise to corresponding methods.”). To me, this is a clear and unmistakable disclaimer.

The majority's second justification relies on claim differentiation with regard to claim 23, which explicitly states that the registration server must be “free of content managed by the architecture.” The majority reasons that if the term “registration server” alone means a server with no content managed by the architecture, then the explicit limitation in this claim is superfluous. Majority Op. 1274–75. I agree. It does appear that the proper construction of “registration server” renders this language superfluous. I further agree that the doctrine of claim differentiation counsels against doing this. However, when faced with a clear case of disavowal and a claim differentiation argument, the court must always hold that the clear and unmistakable disavowal trumps. Our case law is quite clear on this choice.

[T]he doctrine of claim differentiation does not serve to broaden claims beyond their meaning their meaning in light of the specification, and does not override clear statements of scope in the specification and the prosecution history. Toro Co., 199 F.3d at 1302 (citations omitted). And, this is the only reasonable result. To hold otherwise would allow a patentee to game the system by filing a continuation including limitations that would be superfluous thus eliminating the clear disavowal in the earlier filed specification. The majority's logic allows this patentee to recapture claim scope that was surrendered in the application, and violates the public notice function of patents. Indeed, in the present case, claim 23 (the only claim with this language that would be rendered superfluous) of the '573 patent was submitted to the PTO after the filing of its parent application. Once the patentee has clearly disavowed the plain and ordinary meaning of a term, the patentee should not be able to change the rules by recapturing that scope through creative claim drafting.

This is to me a clear and unmistakable case of disavowal. The district court's claim construction is correct. Thus, we should affirm its summary judgment of non-infringement of claims 1–22 of the '573 patent.