TM Patents, L.P. v. International Business MacHines Corp.

121 F. Supp. 2d 349 | S.D.N.Y. | 2000

121 F. Supp. 2d 349 (2000)

TM PATENTS, L.P., and TM Creditors, L.L.C., Plaintiffs,
v.
INTERNATIONAL BUSINESS MACHINES CORPORATION, Defendant.

No. 97 CIV. 1529(CM) (MDF).

United States District Court, S.D. New York.

November 13, 2000.

Stephen B. Judlowe and Dennis J. Modolino, Hopgood Calimafde Judlowe & Modolino, LLP, New York, NY, for Plaintiffs.

Christopher A. Hughes and Christopher K. Hu, Morgan & Finnegan, LLP, New York, NY, for Defendant.

DECISION AND ORDER GRANTING DEFENDANT'S MOTION TO DISMISS PLAINTIFF'S CLAIMS UNDER U.S. PATENT NO. 5,212,773 ("THE '773 PATENT") FOR LACK OF STANDING, OR IN THE ALTERNATIVE, GRANTING DEFENDANT'S MOTION FOR SUMMARY JUDGMENT FOR NON-INFRINGEMENT

McMAHON, District Judge.

Before me are several motions relating to the viability of plaintiff's claims for infringement of U.S. Patent No. 5,212,773 ("the '773 patent"), relating to a store-and-forward system for transmitting messages from input to output circuits within a massively parallel processor. For the reasons stated below, I conclude that (1) IBM's motion for summary judgment dismissing the case on the ground that plaintiff does not have title to the patent, and thus cannot maintain this proceeding, should be granted; and (2) assuming there is standing, IBM is entitled to summary judgment dismissing TM's infringement claims. While the '773 patent is enforceable, IBM's products do not infringe the patent's claims as interpreted by this Court.

Prior Proceedings

1. On November 8, 1999, this Court entered an order following a so-called Markman hearing, see Markman v. West-view Instruments, Inc., 517 U.S. 370, 116 S. Ct. 1384, 134 L. Ed. 2d 577 (1996), 517 U.S. 370, 116 S. Ct. 1384, 134 L. Ed. 2d 577, in which I construed the claims in suit under the '773 patent. A supplemental Markman opinion was issued on December 17, 1999. Familiarity with that decision is assumed. See TM Patents, L.P. v. International Bus. Machs., 72 F. Supp. 2d 370 (S.D.N.Y.1999).

2. On July 28, 2000, this Court entered an order disposing of a number of pending motions with regard to both the '773 patent and the other patent in suit. However, the Court ordered a hearing in connection with TM's motion for a declaration that the '773 patent was in fact enforceable and the parties' cross-motions for summary judgment on the issue of infringement of the '773 patent. That hearing was held on September 21, 2000.

3. On September 12, 2000, IBM filed with the Court a suggestion that TM had never obtained valid title to the '773 patent. I indicated that IBM's submission should be treated as a motion to dismiss for lack for subject matter jurisdiction and directed TM to respond. It filed papers in opposition on September 20. IBM filed a supplemental brief on October 6, addressing issues that were the subject of questions at the September 21 hearing. TM, at the Court's invitation, filed a reply to that brief on October 6. Thereafter, the parties filed written responses to questions propounded by the Court.^[1]

I

IBM'S CHALLENGE TO THE COURT'S SUBJECT MATTER JURISDICTION

Statement of Facts

IBM contends that this Court lacks subject matter jurisdiction because TM procured the '773 patent through assignment from someone who never held valid title to the patent. While the parties have not presented me with extensive affidavit testimony or with live witnesses, they have assembled a plethora of ancient memoranda in order to reconstruct the history of this patent's ownership. The pertinent facts (which for the most part are undisputed, though the conclusions that follow from them are not) are as follows:

The invention under the '773 patent, entitled Wormhole Communications Arrangement for Massively Parallel Processor, was the brainchild of W. Daniel Hillis. At the time of the invention, Hillis was a graduate student at Massachusetts Institute of Technology. His attendance at M.I.T. was funded by the Hertz Foundation, a private foundation. He was not the direct recipient of federal government funding.

On April 28, 1983, Thomas Engellenner of M.I.T. wrote to the Office of Patent Counsel at the Naval Underwater Systems Center in Newport, Rhode Island, and enclosed two patent applications — Hillis' application for what eventually became the '773 patent (entitled "Parallel Processor"), and an application for a second invention entitled "Processor/Memory Circuit." In his letter, Engellenner disclosed to the Navy that M.I.T. was electing to treat the two patents as having arisen under Advanced Research Projects Agency ("ARPA") Contract N-00014-80-C-0505, and was further electing not to apply for the patents itself "because of budgetary constraints and the limited commercial potential perceived in the short term." (IBM 9/12/00 Submission at Ex. 2.) IBM has not supplied the Court with a copy of this contract, and given its extensive efforts (via document requests and Freedom of Information Act inquiries) to locate one from any and every possible source, I am constrained to conclude that no copy exists. However, IBM has located a March 1982 document that appears to be a modification or extension of this contract. (IBM's Oct. 17 Submission at Ex. 4.)

The wording of Engellenner's letter is curious, in that M.I.T. does not assert that work on either invention was actually funded by the Department of the Navy. Indeed, what MIT said was almost deliberately opaque:

The inventors herein are primarily a group of graduate students at M.I.T. but from time to time a number of these inventors have been employed by M.I.T. as research associates funded by ARPA.

(Id.) Engellenner did not identify how many inventors were involved in creating the two inventions and also did not state which (or how many) of those inventors fell within the ambit of "a number of inventors" who had been "employed by M.I.T. as research associates funded by ARPA." In particular, M.I.T. did not specifically identify Hillis as one of those Government-funded graduate students, or the invention for which Hillis alone was responsible as having been ARPA-funded.

Nonetheless, M.I.T. included Hillis' invention in the ambit of its ARPA disclosure. (Id.) And even though his fellowship was privately funded, Hillis, in contemporaneously published articles, admitted that he was working on an invention at MIT's Artificial Intelligence Laboratory, funded in part by Contract N-00014-80-C-0505 (IBM 10/6/00 Submission at Ex. 9, 10.) — the very contract under which Engellenner claimed Government sponsorship for both inventions disclosed in his letter to the Navy. This "connection machine," as he called it, was the invention eventually embodied in the '773 patent can be seen by reading a 1982 article in the International Journal of Theoretical Physics, where Hillis described the proposed architecture of the connection machine as "a locally connected array of processing-memory cells." (IBM 10/6/00 Submission at Ex. 10.)

In addition, documents provided to IBM by M.I.T. include a copy of Hillis' original patent application (Ser. No. 499, 474), filed on May 31, 1983. It contains a handwritten notation "3803" in the corner; "3803" is the case number assigned to Hillis' invention by M.I.T. (Id. at Ex. 6.)

On the same day that Engellenner wrote to the Navy, he also wrote a memorandum to Hillis, summarizing discussions they had held during the preceding weeks (IBM 9/12/00 Submission at Ex. 3.) It appears from the memorandum that Hillis and his patent attorney had approached the M.I.T. Patent, Copyright & Licensing Office on a relatively urgent basis in order to clarify his own ownership rights — and perhaps to induce M.I.T. to accept reassignment of the inventions — so that he could enter into an arrangement with IBM to manufacture prototype chips employing the invention. The first of these discussions apparently took place on March 23, 1983, according to a memorandum that Engellenner wrote to Arthur A. Smith, Jr., the head of M.I.T.'s patent office, on that very day. (TM 9/20/00 Submission at Ex. C.)

Engellenner's memorandum to Smith, if taken at face value, discloses that even Hillis conceded that some portion of his work could be considered Government-funded:

I explained to Hillis M.I.T.'s basic policy on inventions, that being, that if an invention is conceived or reduced to practice at M.I.T. with the use of sponsored research funds or significant amounts of M.I.T. funding or facilities, agreements signed by all employees and visiting scientists require that the invention be assigned to M.I.T. Hillis and his attorney felt that there were probably more than one invention here and that at least some aspect of the patent application being prepared by [the patent attorney] would belong to M.I.T. because of ARPA sponsorship. [sic]

(Id. at p. 1.) (emphasis added). Of course, the highlighted language is hearsay within hearsay,^[2] and I have some difficulty concluding that it would be admissible, even under the liberal standard of Fed.R.Evid. 807 — both because Hillis himself later created documents suggesting that neither patent was created under Government sponsorship or belonged to M.I.T. (IBM 9/12/00 Submission at Ex. 5), and because the memo contains certain representations that appear to be inconsistent with the record. For example, Engellenner's memorandum states that Hillis claimed "a routing circuit" as belonging to him, while acknowledging that "a processing element (which was purportedly developed with ARPA sponsorship)" probably did not. (TM 9/20/00 Submission at Ex. C at 2.) However, M.I.T. Case No. 3802, which was assigned to a "Processor/Memory Circuit" (emphasis added), is listed in the relevant documentation as having been invented by Hillis and seven other graduate students (Id. at Ex. D.), while Hillis' singular invention was assigned M.I.T. case number 3803 and entitled "Parallel Processor." (Id.)^[3]

Nonetheless, no one disputes that M.I.T. lacked interest in owning and exploiting the inventions itself and was more than happy to waive its rights in the invention. However, in its letter to the Navy disclosing the patents (IBM 9/12/00 Submission at Ex. 2.), M.I.T. conceded nothing in terms of Hillis' sole ownership of any patents. Indeed, in his memorandum to Hillis, Engellenner identified Hillis (and another student) as having "from time to time been employed as research assistants funded by ARPA." (Id. at Ex. 3 at p. 1.) Thus, it seems that M.I.T. took the position that any and all patent rights associated with both inventions belonged to it in the first instance, whether Hillis thought so or not. And while the university volunteered to waive its rights in Hillis' invention, M.I.T. also specifically advised Hillis that he would need to file papers with the Department of the Navy in order to protect his title to the invention. (Id. at Ex. 2.)

M.I.T. advised Hillis, at the time it offered to waive its rights in the invention that became the '773 patent, that it did so "subject only to the following conditions:"

first, your patent application should be amended to recite that "the government has rights in this invention pursuant to a contract with the Advanced Research Projects Agency No. N-00014-80-C-0505" and any licenses issued to commercial concerns under these inventions should acknowledge the government's royalty-free license to these inventions.... [other conditions omitted]

(Id. at Ex. 3 at pp. 1-2.) Engellenner then stated, "If these terms are agreeable to you, I will be happy to present your case to the patent committee." (Id.) There is no evidence in the record to indicate whether Hillis did or did not consent to M.I.T.'s conditions, although, as will be seen below, the 1991 patent application that resulted in the issuance of the '773 patent did not contain any of the language demanded by the M.I.T. as a condition of its waiver. (Id. at Ex. 11.)

I surmise that the '773 patent application did not contain MIT's suggested language because Hillis took the position that at least one of the inventions he discussed with Engellenner was his invention and his alone. Indeed, Hillis represented to the Department of the Navy, in a letter he wrote on November 14, 1984, that both inventions were his. Nonetheless, in that letter, Hillis very deferentially asks the Navy to authorize his rights in both inventions:

P.C. Lall, Esquire Office of Patent Counsel Naval Underwater Systems Center Department of the Navy Building 142 Newport, Rhode Island 02840 SUBJECT: Invention Disclosure under ARPA Contract N-0014-80-C-0505 MIT Case 3802 "Processor/Memory Circuit" MIT Case 3803 "Parallel Processor"

Dear Mr. Lall:

I am writing to request authorization to retain the patent rights on the inventions cited above.

Because these inventions were developed when I was a student at MIT supported by a fellowship from the private Hertz Foundation and not through any government grants, it is my understanding that I retain the rights to patents that may be issued. I further understand that without addressing the question of ownership of the patent rights, MIT has elected not to seek patent protection for these inventions and suggested that retention of rights by the inventors is in the best interest of the public, as letters from MIT to your office indicate.

Through my affiliation with Thinking Machines Corporation, I am actively involved in continuing and expanding development of the inventions. We are integrating computer architecture and software innovations that promise important improvements in addressing urgent and very large scale Department of Defense problems. We are also exploring the application of my inventions to solving data processing concerns of other Federal agencies, and are both convinced of and dedicated to realizing the potential for comparable gains for non-Defense computational problems. Further, related to my deep appreciation of the value of the technology transfer to the civilian economy, we are aggressively designing applications software for responding to classes of problems of direct interest to society at large.

I close by repeating my request for your consideration in authorizing my retention of patent rights on the referenced inventions.

Your kind attention to this matter is very much appreciated.

W. Daniel Hillis

(Id. at Ex. 5.) (emphasis added)

It is not true, as TM now asserts (See TM 9/20/00 Submission at n.4.), that the Navy "dropped the matter" after receiving the above-quoted letter containing Hillis' rather veiled claim of right to the patents. What is true is that the Government was perfectly willing to authorize Hillis to retain ownership of the patents. Indeed, the Navy forwarded the necessary forms on at least two occasions: once to M.I.T. in November 1984, together with a request to provide the forms to those inventors who wanted to retain rights in the ARPA contract inventions (IBM 9/12/00 Submission at Ex. 4.), and once directly to Hillis in April 1986 (IBM 9/12/00 Submission at Ex. 7.). However, the Navy, having been notified by M.I.T. that it elected to treat the patents as having arisen under Government sponsorship, took the position that Hillis' rights in the invention were subject to the same conditions as M.I.T.'s rights. These, of course, included a royalty-free license to the Government.

Hillis — by now employed by plaintiffs' predecessor-in-interest, the now-defunct Thinking Machines Corporation — was apparently no more interested in adjudicating this matter of ownership with the Navy than he had been with M.I.T. So he took the path of least resistance. He did not sign the forms the Navy required. Neither did he send a letter to the Navy advising them that he disputed the Government's assertion that he needed to sign anything at all.

I can only speculate about why Hillis, having asked the Navy for permission to keep his inventions, did not sign the forms that were sent to him by both M.I.T. and the Navy. The record is devoid of evidence on the subject; it contains no affidavit from Hillis, and his deposition testimony demonstrates a total lack of recall about the forms. There is no dispute, however, that he did not sign them. M.I.T. sent a set of confirmatory instruments (possibly those sent to M.I.T. by the Navy in November 1984) to Dr. Hillis at least as early as May 3, 1985. That they had not been signed and returned some 11 months later is evidenced by his receipt of a letter from M.I.T. dated April 3, 1986, warning Hillis that he was endangering his title by failing to return them:

We have recently been called by Ms. Ina Griffith of the Office of Naval Research who told us that you have not yet returned the instruments to ONT. Ms. Griffin asked us to contact you on this matter. We strongly advise you to complete these instruments, since the lack of them can endanger your title to the inventions.

(IBM 9/12/00 Submission at Ex. 6.) (emphasis added) In spite of this, Hillis did not return the forms, as reflected in M.I.T.'s June 24, 1987 "Report of Inventions and Subcontracts (Final)" for Contract No. N-00014-80-C-0505, which indicated that confirmatory instruments for Case No. 3803 had not been forwarded to the Navy as Contracting Officer. (IBM 10/6/00 Submission at Ex. 4.)

Ultimately, the '773 patent application, under the title "Wormhole Communications Arrangement for Massively Parallel Processor," was filed on Feb. 22, 1991. Hillis listed himself as the sole inventor. He did not disclose any governmental interest in the patent, including any compulsory license. The application for the '773 patent explicitly relates back to the May 31, 1983 application, Ser. No. 499,474 ("Parallel Processor") ("the great grandfather application"), which the parties agree is M.I.T. Case No. 3803, the original Hillis invention. As discussed below, the '773 patent is the result of successive applications that could not, as a matter of law, contain additional information or a different invention than that described in the original application.^[4]

In view of the foregoing, IBM contends that Hillis lacked title to assign to Thinking Machines, because he failed to file the paperwork required by 35 U.S.C. § 201(d). Of course, if Hillis did not have valid title, then he could not convey good title to Thinking Machines, which in turn could not convey good title to plaintiffs — or so IBM argues. Since only a patent owner has standing to sue for infringement (except in a few discrete instances that are not relevant here), such a finding would be the end of the matter where the '773 patent is concerned.

Plaintiffs do not dispute that Hillis never sent in the forms that would have perfected his title as against the Government. They read the documents to suggest that Hillis never agreed with M.I.T. about Government sponsorship of the invention that became the '773 patent, and argue that Hillis needed no waiver by the Government because the invention was not Government-sponsored. Plaintiffs do admit, however, that there was never any formal adjudication of this controverted issue. As happens all too often in these situations, both Hillis and his assignor, Thinking Machines, were content to leave the situation murky. IBM, which has a lot riding on being able to knock out TM's claims under '773 patent, is not.

Plaintiffs do not suggest that this Court embark on the time-consuming task of resolving whether '773's wormhole routing message network was invented with ARPA sponsorship back in the early 1980s — at least, not at this juncture. Rather, they make a number of purely legal arguments about why this Court has jurisdiction over the '773 infringement claims.

First, TM argues that it has sufficiently pled jurisdiction and standing, and that the resolution of any factual disputes must abide the trial of this action.

Second, plaintiffs contend that IBM is estopped to challenge TM Creditors' title, because it has repeatedly admitted TM's title in documents previously filed in this action — indeed, as recently as last summer.

Third, plaintiffs assert that the Bankruptcy Court's award of title to the '773 patent to TM Creditors during Thinking Machines' bankruptcy proceeding — a proceeding to which both IBM and the Government were parties, and in which TM contends the issue of defective title could and should have been raised — is res judicata on the question.

Fourth, TM claims that its title to the patent was secured by virtue of its recording of same following the bankruptcy proceeding, citing the patent recording statute, 35 U.S.C. § 261.

Fifth, if all else fails, TM asserts that IBM has the burden to offer sufficient competent evidence to warrant a finding that it lacks ownership of the patent, and that it has failed to do so. At a minimum, TM asserts that there is a disputed issue of fact on the question.

Discussion

A. The Issue of Standing is Jurisdictional and Must Be Determined by the Court at the Outset

Before a federal court can consider the merits of a legal claim, the person seeking to invoke the jurisdiction of the court must establish the requisite standing to sue. See Whitmore v. Arkansas, 495 U.S. 149, 154, 110 S. Ct. 1717, 109 L. Ed. 2d 135 (1990). "Whether a claimant has standing is the threshold question in every federal case, determining the power of the court to entertain the suit." In re Gucci, 126 F.3d 380, 387-88 (2d Cir.1997), (quoting Warth v. Seldin, 422 U.S. 490, 498, 95 S. Ct. 2197, 45 L. Ed. 2d 343 (1975)). It remains open to review at all stages of the litigation. See National Org. for Women v. Scheidler, 510 U.S. 249, 255, 114 S. Ct. 798, 802, 127 L. Ed. 2d 99 (1994).

Standing requirements drawn from the Constitution look to whether the plaintiff has made out a `case or controversy' between himself and the defendant within the meaning of Article III. See Warth v. Seldin, 422 U.S. 490, 498-499, 95 S. Ct. 2197, 2205, 45 L. Ed. 2d 343 (1975). The plaintiff can do this by showing he has an "injury in fact," or a "distinct and palpable injury." Association of Data Processing Serv. Orgs. v. Camp, 397 U.S. 150, 90 S. Ct. 827, 25 L. Ed. 2d 184 (1970); Valley Forge Christian College v. Americans United for Separation of Church and State, 454 U.S. 464, 102 S. Ct. 752, 70 L. Ed. 2d 700 (1982). If TM is not the true owner of the patent, it suffers no injury in fact as a result of any alleged infringement. To bring suit for patent infringement, a plaintiff must have legal title to or be the exclusive licensee of the patent in suit. See Enzo APA & Son, Inc. v. Geapag A.G., 134 F.3d 1090, 1093 (Fed.Cir. 1998); Abbott Lab. v. Diamedix Corp., 47 F.3d 1128, 1130 (Fed.Cir.1995); Arachnid, Inc. v. Merit Indus., Inc., 939 F.2d 1574, 1578-79 (Fed.Cir.1991). Thus, patent ownership is a prerequisite for application of the infringement statute, and also confers standing.

There is some confusion as to whether the question of patent ownership should properly be characterized as a challenge to TM's standing to bring suit, or to the court's subject matter jurisdiction. Two other courts in this district have dismissed infringement claims when the plaintiff did not own title to the patent, on the basis that plaintiff "lacked standing to invoke the subject matter jurisdiction of the court." RAD Data Communications, Inc. v. Patton Elecs. Co., 882 F. Supp. 351, 352 (S.D.N.Y.1995); Imatec, Ltd. v. Apple Computer, Inc., 81 F. Supp. 2d 471 (S.D.N.Y.2000); see also GAIA Techs., Inc. v. Reconversion Techs., Inc., 93 F.3d 774, 780 (Fed.Cir.1996) (calling the standing question "jurisdictional" and dismissing plaintiff's infringement claims on standing grounds). These holdings may have improperly conflated the issue of standing with subject matter jurisdiction. Cf. DaSilva v. Kinsho Int'l Corp., 229 F.3d 358 (2d Cir.2000)^[5]; However, I need not delve into this theoretical problem at the present time. It does not matter whether this Court should properly treat the defendant's challenge to ownership as a motion for lack of subject matter jurisdiction under Fed.R.Civ.P. 12(b)(1), or for failure to state a claim due to lack of standing under 12(b)(6), or even as a motion for summary judgement on the merits of the question of ownership. As will be seen below, TM has failed to make the necessary showing of ownership under any theory.

B. Judicial Estoppel Does Not Bar IBM From Raising the Issue of Standing

IBM has consistently admitted during the course of this litigation that TM Creditors LLC has title to the '773 patent. (See Def's Answer to the First Am. Compl., filed March 6, 1998, ¶ 15; Def's Am. Answer to the First Am. Compl., filed August 28, 2000, ¶ 15.) IBM also moved to add TM Creditors, which was not originally a party to this action, as a party plaintiff, asserting in connection with its motion that title to the '773 patent belonged to TM Creditors. (See Def.'s Mem. in Supp. of Def.'s Mot. To Dismiss, filed May 1, 1997, at p. 2, 12.) While it is arguable that the first answer, and even the joinder of TM Creditors, predates IBM's awareness of the Hillis issue, it is quite clear that IBM's amended answer, filed August 28, 2000, came long after defendant became aware of a possible cloud on TM's title. Indeed, IBM first raised the jurisdictional issue with counsel for TM by letter dated April 4, 2000. (IBM 9/12/00 Submission at Ex. 10.) Nonetheless, it admitted TM Creditors' ownership of the '773 patent almost five months later — and a mere two weeks before it filed the suggestion of lack of jurisdiction with the Court. As far as TM is concerned, that should end the matter.

TM argues that these admissions should prevent IBM from challenging TM's ownership of the patent under the theory of judicial estoppel. An admission made during the course of the litigation, however, is not the same as judicial estoppel, which "prevents a party from asserting a factual position in a legal proceeding that is contrary to a position previously taken by [the party] in a prior legal proceeding." Bates v. Long Island R. Co., 997 F.2d 1028, 1037 (2d Cir.1993); see also 18 Charles A. Wright, Arthur R. Miller & Edward H. Cooper, Fed. Pract. & Proc. § 4477 (1981). The theory may be invoked where, (1) the party against whom the estoppel is asserted took an inconsistent position in a prior proceeding, and (2) that position was adopted by the first tribunal in some manner. See Bates at 1038. The doctrine clearly does not apply in this case. IBM asserted TM Creditors' ownership of the patent in this proceeding for the purposes of joinder.

The purposes of judicial estoppel — to "preserve the sanctity of the oath" and to "protect judicial integrity by avoiding the risk of inconsistent results in two proceedings" — are not at issue here. See Simon v. Safelite Glass Corp., 128 F.3d 68, 71 (2d Cir.1997) (internal quotation marks omitted) (emphasis added). IBM has not benefitted from some prior judicial determination that TM owns the '773 patent, only to claim the opposite is true here for other purposes. Furthermore, the Second Circuit has recognized that "judicial estoppel does not apply when the first statement resulted from `a good faith mistake or an unintentional error.'" Id. at 73 (citations omitted).

Alternatively, TM argues that IBM's admissions of TM's ownership are formal judicial admissions that are conclusive against IBM in this action. See Western World Insurance Co. v. Stack Oil, Inc., 922 F.2d 118, 122 (2d Cir.1990). However, "[s]ince an objection to subject matter jurisdiction goes to the power of the court to hear and decide the case, parties may not create or destroy jurisdiction by agreement or by consent." 5A Wright & Miller, Fed. Prac. & Proc. § 1350 at 204 (2d ed.1990); see also Members for a Better Union v. Bevona, 152 F.3d 58 (2d Cir. 1998). Similarly, the parties cannot by stipulation confer standing on a non-patent owner in contravention of the Constitution and the statute. IBM is thus correct that it cannot create otherwise non-existent jurisdiction by admission. See Barhold v. Rodriguez, 863 F.2d 233, 234 (2d Cir.1988).

C. The Question of Ownership Is Not Res Judicata By Virtue of the Bankruptcy Court's Judgment

Of equal or greater interest to this Court is the impact of the disposition of the '773 patent by the Bankruptcy Court in the District of Massachusetts in connection with Thinking Machines' bankruptcy proceeding. TM argues that the claim of ownership is res judicata, or precluded, by virtue of the determination of the Bankruptcy Court in the Thinking Machines bankruptcy. In that proceeding, to which IBM was a party, TM Creditors paid more than $27 million in creditor claims in exchange for title to Thinking Machines' patents. The '773 patent was one of those patents. The Bankruptcy Court entered an order confirming Thinking Machines' Reorganization Plan, and ¶ 12 of that order provided that all transfers of property to the Plan Entities (of which TM Creditors was one) as provided in the Plan "... shall be legal, valid and effective and shall constitute the transfer to and shall vest in the Plan Entities good title to such property free and clear of all liens, charges, Claims, encumbrances, Administrative Claims, interests and rights of offset, except as expressly provided in the Plan or this Confirmation Order...." In re Thinking Machines Corp., No. 94-15405 (Bankr. D.Ma. Feb. 2, 1996) (excerpts attached as TM 9/20/00 Submission at Ex. K.)

A final judgment of a United States Bankruptcy Court is, of course, res judicata as to all matters properly before it that are necessary to that final judgment and that were or could have been litigated in the proceeding. See St. Pierre v. Dyer, 208 F.3d 394, 399 (2d Cir.2000); National Labor Relations Bd. v. United Techs. Corp., 706 F.2d 1254, 1259 (2d Cir.1983); In re Teltronics Servs., 762 F.2d 185, 190 (2d Cir.1985) (applying res judicata rules to bankruptcy matter). Under this doctrine, also referred to as "claim preclusion," a subsequent action will be barred where a claim between the parties has been litigated and decided. In determining whether claim preclusion applies, a court may consider whether (1) the prior decision was a final judgment on the merits; (2) the litigants were the same parties; (3) the prior court was of competent jurisdiction; and (4) the causes of action were the same. See Corbett v. MacDonald Moving Servs., Inc., 124 F.3d 82, 87-88 (2d Cir.1997); In re Teltronics Servs., Inc. at 190. In the bankruptcy context, the court also asks whether an independent judgment in a separate proceeding would "impair, destroy, challenge, or invalidate the enforceability or effectiveness" of the reorganization plan. Sure-Snap Corp. v. State St. Bank and Trust Co., 948 F.2d 869, 875-76 (2d Cir.1991), although this particular inquiry "may also be viewed as an aspect of the test for identity of the causes of action." Corbett at 88.

However, the catch for our purposes is that the res judicata effects of a judgment extend only to matters that were properly before the Bankruptcy Court or that could properly have been brought there.^[6]See In re Teltronics at 190. Therein lies the rub.

The general principles are not disputed. Under the Bankruptcy Code, a debtor's estate consists of, inter alia, "all legal or equitable interests of the debtor in property as of the commencement of the case." 11 U.S.C. § 545(a)(1). The filing of a bankruptcy case does not, and cannot, give a debtor or its creditors greater rights in property than the debtor had prior to bankruptcy. See In re Brown, 734 F.2d 119, 124 (2d Cir.1984) (nonetheless affirming debtor's entitlement to avoid creditor's lien and enjoy exemption under bankruptcy Code); Old Stone Bank v. Tycon I Bldg. Ltd. Partnership, 946 F.2d 271, 276 (4th Cir.1991). As a result, the Bankruptcy Court's jurisdiction "does not extend to property that is not part of a debtor's estate," Rutherford Hosp., Inc. v. RNH Partnership, 168 F.3d 693, 699 (4th Cir.1999), and even a confirmed bankruptcy plan "cannot furnish anyone rights to what was not property of the debtor's estate." Terry Oilfield Supply Co. v. American Sec. Bank, N.A., 195 B.R. 66, 73 (S.D.Tex.1996). Thus, the Bankruptcy Court could not have passed title to the '773 patent if Thinking Machines had no title to pass.

IBM contends that no attempt was actually made in the bankruptcy proceeding to transfer any rights in Thinking Machines' patents beyond those the debtor actually owned. IBM is correct. The First Amended Joint Plan of Reorganization transferred only "the Debtor's right title and interest in and to... patents and patent applications of the Debtor" (IBM 10/6/00 Submission at Ex. 24 at 38.), while the Collateral Patent Assignment assigned "all of Assignor's right, title and interest" in the patents to TM Creditors. (Id. at Ex. 25 at frame 843.) IBM observes, correctly, that nothing in the "free and clear" language quoted above from the Final Judgement of the Bankruptcy Court is inconsistent with that, or expands on the inherent limitation of the Plan and the Collateral Assignment — namely, that they convey and/or assign whatever interest the Debtor has, and nothing more.

Both parties acknowledged at the hearing that no one involved in the Bankruptcy proceeding believed that the court actually adjudicated whether Thinking Machines owned the '773 patent.^[7] And there is no question in my mind that the Bankruptcy Court had jurisdiction to resolve disputes over the debtor's ownership to any particular estate asset. See 11 U.S.C. §§ 363(f) and 1123; Celotex Corp. v. Edwards, 514 U.S. 300, 307-08, 115 S. Ct. 1493, 1498-99, 131 L. Ed. 2d 403 (1995). Thus I have no doubt that a challenge to ownership of the '773 patent could have been raised in the bankruptcy context. The question is whether the failure to do so bars litigation of that issue before me.

IBM asserts that it neither challenged nor questioned Thinking Machines' title to the patent because it could not have done so, for several reasons. First, the issue was not "identical" to any issue raised in the bankruptcy, given the special meaning of "identity" for former adjudication purposes. Second, nothing in the Plan alerted it to the possibility of a title defect. Third, IBM had no incentive to raise the issue in the absence of notice. Fourth, IBM lacked standing to raise the issue because TM deliberately declined to create a justiciable controversy over any specific patent during the bankruptcy, or to identify any specific patent that they would later contend was infringed by IBM.

IBM is in substantial part correct.

"Under the doctrine of res judicata or claim preclusion, `a final judgment on the merits of an action precludes the parties or their privies from relitigating issues that were or could have been raised in that action.'" St. Pierre v. Dyer, 208 F.3d, 394, 399 (2d Cir.2000) (quoting Federated Dep't Stores, Inc. v. Moitie, 452 U.S. 394, 398, 101 S. Ct. 2424, 69 L. Ed. 2d 103 (1981)). IBM was a party to the Thinking Machines' bankruptcy. However, that did not automatically confer standing on it to challenge the debtor's title to the '773 patent. And "res judicata is inapplicable if formal jurisdictional or statutory barriers precluded the plaintiff from asserting its claims in the first action." Computer Assocs. Int'l, Inc. v. Altai, Inc., 126 F.3d 365, 370 (Fed.Cir.1997).

In a bankruptcy case, "a party who is not directly `aggrieved' by the construction of a provision of the Plan would lack the requisite standing to be heard." In re Johns-Manville Corp., 68 B.R. 618, 624 (Bkrtcy.S.D.N.Y.1986) ("Thus, no party may successfully prevent the confirmation of a plan by raising the rights of third parties who do not object to confirmation."), afield in part and rec'd. in part on other grounds, 78 B.R. 407 (S.D.N.Y.1987), afield sub. nom, Kane v. Johns-Manville Corp., 843 F.2d 636 (2d Cir.1988). "Only parties adversely affected by provisions of a plan may raise an objection to confirmation based on such provisions." In re Gaston & Snow, Nos. 93 Civ. 8517, 93 Civ. 8628, 1996 WL 694421, at * 7 (S.D.N.Y. Dec.4, 1996) (quoting In re Johns-Manville Corp., 68 B.R. at 623-24).

IBM argues that the only entity that could arguably have been "aggrieved" by any attempt to usurp title to the '773 patent was the U.S. Government (the true owner of the patent). And in fact, IBM, wearing its hat as competitor of TM, did lack standing to challenge TM's title to the '773 patent. It had not been charged with infringement of the '773 prior to the bankruptcy, and it was not charged with infringement during the bankruptcy. It was not even threatened with infringement of the patent during the bankruptcy. Indeed, TM deliberately refrained from identifying IBM as a potential target of future infringement actions, in order to avoid IBM's commencement of a declaratory judgment action touching on the validity of the patent. (IBM 10/6/00 Submission at Ex. 30.) Plaintiffs have admitted in response to IBM's interrogatory that IBM was not on notice of alleged infringement of the '773 patent until November 1, 1996, at the earliest. (See id. at Ex. 34.)

Without even a hint that Thinking Machines planned to charge IBM with infringement of the '773 patent, IBM had no standing to bring a declaratory judgment action against the '773 patent. It could not have adjudicated the issue, even if it had been aware of the alleged defects in Thinking Machines' title while the bankruptcy was pending. Justiciability "... requires an explicit threat or other action by the patentee, which creates a reasonable apprehension on the part of the declaratory plaintiff that it will face an infringement suit...." Phillips Plastics Corp. v. Kato Hatsujou Kabushiki Kaisha, 57 F.3d 1051, 1052 (Fed.Cir.1995) (refusing to issue declaratory judgment on patent ownership in light of mere license negotiations and no threat of suit).

TM counters by noting that IBM internal documents reveal defendant's fear that it might be infringing the patent. (TM 10/6/00 Submission at 8 n. 10; Pl.'s Mem. in Opp'n to Def's Mot. for Summ. J. on Non-Infringement at Ex. A.) However, under Phillips Plastics, it is a threat of suit by the patentee, not simply the inchoate fear that the patentee might sue, that confers standing. It is simply indisputable that IBM, in its position as competitor/potential infringer, could not have litigated the question now before me in the bankruptcy proceeding.^[8]

However, to the extent IBM suggests that it could not have raised the issue of ownership because it had no interest whatever in uncovering defects in Thinking Machines' title to its purported patent during the bankruptcy, I am constrained to disagree. While TM's deliberate failure to raise the spectre of allegation rendered IBM disinterested in its status as a Thinking Machines competitor, IBM stood before the Bankruptcy Court wearing another hat as well — that of trade creditor. (See IBM's 10/6/00 Submission at 22 n. 5.) As I understand matters, TM Creditors, the entity that ultimately took whatever patent title Thinking Machines was able to pass, was formed for benefit of the debtor's trade creditors — that is, they would have to look to the results of patent infringement litigations for recoupment of the amounts owed. While IBM's trade claims were apparently infinitesimal in relation to what was at stake in connection with the enforcement of the parties' various patents, it had the same incentive as any other trade creditor to ensure that it was getting something of value in exchange for giving up its trade debt. The trade creditors were getting the patents, but a patent without good title is not worth the paper it is written on.

Nonetheless, IBM's failure to raise the issue of ownership in the bankruptcy proceeding is not fatal to its claim here, because IBM could not have known about the possibility of a defect in the title. As a general rule, newly discovered evidence does not preclude application of res judicata. See Saud v. Bank of New York, 929 F.2d 916, 920 (2d Cir.1991). However, res judicata does not apply when the existence of the ownership issue "could not have been discovered with due diligence." L-Tec Electronics Corp. v. Cougar Electronic Org., Inc., 198 F.3d 85, 88 (2d Cir.1999) (finding new claims based on different legal theories rather than facts).

IBM contends that Thinking Machines had a duty to disclose all relevant facts concerning its assets prior to confirmation of the Plan. See 11 U.S.C. § 1125 (Disclosure Statement must contain "adequate information"). It is undisputed that no notice was given of any defect in or question about title to the '773 patent, and the issue never arose in the proceedings. And there is no evidence before me from which I could conclude that IBM should have been on notice of the alleged defect in title. Indeed, in the context of this lawsuit, it took years of discovery from multiple parties, as well as FOIA requests addressed to the Government, before the issue crystallized. In contrast, the court in Sure Snap found that the debtor, who was bringing subsequent claims against its creditors, "had adequate information about lender liability claims, prior to commencement of the... bankruptcy proceedings." Sure-Snap Corp. v. State St. Bank and Trust Co., 948 F.2d 869, 873 (2d Cir.1991).

I agree with IBM that, in the absence of notice from Thinking Machines of possible clouds on title, the Bankruptcy Court would never have allowed a wholesale investigation to discover whether there existed some defect in TM's title in the more than 100 patents and applications Thinking Machines purported to own. See In re Eagle-Picher Indus., 169 B.R. 130, 134 (Bkrtcy.S.D.Ohio 1994) ("one seeking to conduct [discovery] has the burden of showing good cause for the examination which it seeks.")

So while IBM would have been fatally dilatory had it sat on knowledge of a potential defect in a patent's title, it cannot be foreclosed from litigating the patent's ownership now, precisely because it had no such knowledge, and no way of obtaining it.

Finally, res judicata is also inapplicable because the claim raised in this proceeding was not identical to the claims raised in the Bankruptcy Court.

For purposes of claim preclusion, a claim "could have been raised" in a prior proceeding "only where the transaction or connected series of transactions at issue in both suits is the same, that is, where the same evidence is needed to support both claims, and where the facts essential to the second [transaction] were present in the first." Interoceanica Corp. v. Sound Pilots, Inc., 107 F.3d 86, 91 (2d Cir.1997) (quoting SEC v. First Jersey Sec., 101 F.3d 1450, 1463-64 (2d Cir.1996)); see also Computer Assocs. Int'l, Inc. v. Altai, Inc., 126 F.3d 365, 369 (2d Cir.1997). What constitutes a single transaction "is a factual question to be determined `pragmatically' based on factors such as `whether the facts are related in time, space, origin, or motivation, whether they form a convenient trial unit, and whether their treatment as a unit conforms to the parties' expectations." NBN Broad., Inc. v. Sheridan Broad. Networks, Inc., 105 F.3d 72, 78 (2d Cir.1997) (quoting Restatement (Second) of Judgments, § 24(1) (1982)).

The actual litigation in the bankruptcy case between TM and IBM involved IBM's assertion of patent infringement by the Debtor. It was in this context that the set-off issue arose — whether IBM would be free to enforce its infringement claims against the successors to Thinking Machines should those successors decide to sue IBM. IBM objected to the use of the "free and clear" language in the conveyance of the patents to TM Creditors, not because it was challenging the validity of the title that would pass to TM Creditors under the Plan, but because it feared that the use of this phrase would preclude it from asserting counterclaims. (TM 9/20/00 Submission at Ex. M.) IBM eventually dropped the objection. Nothing in that sequence of events suggests that I should find the "identity of issues" prong of a res judicata analysis to be satisfied.^[9]

In sum, I conclude that res judicata does not bar IBM's challenge to TM's ownership of the patent.

D. Nothing in the Patent Recording Statute Creates Ownership in TM if its Assignor Lacked Ownership

Plaintiffs argue that, even if Dr. Hillis's failed to obtain title to the '773 patent, TM Creditors managed to get title simply because they recorded the assignment to them from TM Patents of the '773 patent.

The patent recording statute, 35 U.S.C. § 261, applies the common law bona fide purchaser rule to patent transfers. When the legal title holder of a patent transfers his or her title to a third-party purchaser for value without notice of an outstanding equitable claims for title, that purchaser takes entire ownership of the patent, free of any prior equitable encumbrances. See FilmTec Corp. v. Allied-Signal Inc., 939 F.2d 1568, 1573 (Fed.Cir.1991). In addition, § 261 adopts the principles of the real property recording acts, in that it "is intended to cut off prior legal interests, which the common law rule did not." Id. at 1573-74.

Under the statute:

An assignment, grant or conveyance shall be void as against any subsequent purchaser or mortgagee for a valuable consideration, without notice, unless it is recorded in the Patent and Trademark Office within three months from its date or prior to the date of such subsequent purchase or mortgage.

35 U.S.C. § 261. By its very terms, § 261 is only a recording statute. It governs the rights of competing assignees. The statute does not address the validity of the rights purportedly transferred by an assignment.

In fact, the companion regulation, 37 CFR § 3.54, expressly provides that "The recording of a document ... is not a determination by the [Patent and Trademark] Office of the validity of the document or the effect that document has on the title to an application, a patent, or a registration." (emphasis added) More to the point, the Federal Circuit has held that "the mere fact that an assignment was recorded in the PTO does not, without more, prove that a valid assignment actually took place." GAIA Techs., Inc. v. Reconversion Techs., Inc., 93 F.3d 774, 778 n. 3 (Fed.Cir.), amended on other grounds by, 104 F.3d 1296 (Fed.Cir.1996) (reversing lower court's finding of standing). Clearly, § 261 does not grant an assignee any title better than the assignor had.

When a statute operates to vest patent title in the Government, subsequent transfers by the inventor or anyone else, whether recorded or unrecorded, will not trump the Government's rights in the invention. See FilmTec Corp. v. Allied-Signal, Inc., 939 F.2d 1568 (Fed.Cir.1991) ("FilmTec"); FilmTec Corp. v. Hydranautics, 982 F.2d 1546 (Fed.Cir.1992) ("Hydranautics"). Both of the FilmTec infringement cases arose from the same set of facts, and their similarity to the present case merits close analysis. The inventor, Cadotte, conducted his research while at MRI, a non-profit research organization, under contract to the Government. The contract, issued under authority of two federal statutes, provided that MRI:

agrees to grant and hereby does grant to the Government the full and entire domestic right, title and interest in any invention, discovery, improvement or development (whether or not patentable) made in the course of or under this contract or any subcontract (of any tier) thereunder.

FilmTec at 1570.

Sometime between September 1977, when FilmTec was organized, and the February 1979 filing of the patent application, Cadotte made the invention that led to the patent. Cadotte left MRI in February 1978. He assigned the patent on his invention to his company, and Filmtec recorded the assignment.

Cadotte claims he made the invention the month after leaving. Defendant-appellant Allied claimed Cadotte formed his invention while he was still at MRI. The trial judge did not address the question of when the invention was made, because he reached his decision in favor of granting a preliminary injunction to FilmTec on the grounds that the contract quoted above conveyed no more than equitable title to the Government. See id. at 1570. The Federal Circuit disagreed, and found the contract "expressly granted to the Government MRI's rights in any future invention." Id. at 1573. Thus, "[i]f a similar contract provision existed between Cadotte and MRI, as MRI's contract with the Government required, and if the invention was made before Cadotte left MRI's employ..., Cadotte would have no rights in the invention or any ensuing patent to assign to FilmTec." Id. If that was the case, Cadotte's purported assignment to Filmtec would be "a nullity." Id. at 1572. The court remanded for findings on both questions.

On remand, the district court found that the invention was made while Cadotte worked at FilmTec, and reinstated the preliminary injunction. See Filmtec v. Allied-Signal, Inc., 988 F.2d 129, 1993 WL 2309 at *1 (Fed.Cir.1993) ("Allied") (unpublished decision) (adopting findings of fact in Hydranautics). While Allied's appeal was pending, the Federal Circuit resolved the issue of ownership in Hydranautics, a different infringement case brought on the same patent. The court concluded that the statutes governing the MRI contract clearly provided for title to any invention made or conceived under contract to vest with the Government, and that the invention conceived while Cadotte was at MRI was the same as that described in the patent. See Hydranautics at 1550-51. The Allied court concurred in the Hydranautics analysis and adopted the findings, invalidating FilmTec's title. See Allied, 1993 WL 2309 at *1.

The FilmTec cases control here. Hillis admittedly developed the inventions in the '773 patent while he was at M.I.T. If his work was under the Government funding agreement, and if he did not comply with the requirements necessary for him to retain title, title passed to the Government, Hillis did not have anything to transfer to TM Patents, and the subsequent "transfer" to TM Creditors could not be rendered valid by recording it.

The cases cited by plaintiff do not support the proposition that recording cures the defect in title. In CMS Indus., Inc. v. L.P.S. Int'l, Ltd., 643 F.2d 289 (5th Cir. 1981), a holder of patent rights executed two transfers on the same day: a recorded assignment to a subsidiary of the title, and an unrecorded reservation of the majority of rights in the parent company. See id. at 291-92. The court found that the intended effect of the two transfers was to keep the benefits of the patent as though the parent had retained title, while still recording the assignment of title. The court held that the unrecorded reservation of rights was an attempted assignment, and was thus ineffective against the bankrupt subsidiary's creditor, who later acquired the patent rights in liquidation by assignment, and was a bona fide purchaser for value without notice of the attempted assignment. Id. at 289. This case only shows the general rule under § 261 that a recorded assignment is valid as against an unrecorded assignment. The Government did not need to record its assignment because took title by operation of law.^[10]

Plaintiffs also rely on Heidelberg Harris, Inc. v. Loebach, 145 F.3d 1454, 1458 (Fed.Cir.1998). Plaintiff Loebach assigned his rights in his invention to his employer in 1980. In 1983 that company entered an agreement with defendant Heidelberg Harris in which defendant would receive an unrestricted license to the patent in 1990. See id. at 1456. In 1991 Loebach won back his right to the patent in a suit against his former employer. See id. at 1457. He then sued defendant for infringement. The Appeals court upheld Heidelberg Harris' bona fide purchaser defense, agreeing with the district court that its title to the unrestricted license vested in 1983. See id. at 1459. Here again, defendant purchased the patent rights from an entity that, at least until 1991, had valid title. TM Patents did not.

If TM's argument were correct, any charlatan could fraudulently convey the patents of another to any assignee, who could subsequently gain good title by simply recording the fraudulent assignment. The patent recording statute is no bar to IBM's claim.

E. TM has not raised a genuine issue of fact to controvert IBM's showing that TM does not have title to the '773 patent.

As none of TM's arguments enables it to avoid the merits of the dispute, I turn to the facts. Both sides agree that the party seeking to invoke federal jurisdiction bears the burden of establishing standing. See Lujan v. Defenders of Wildlife, 504 U.S. 555, 561, 112 S. Ct. 2130, 119 L. Ed. 2d 351 (1992); Jaghory v. New York State Dept. of Educ., 131 F.3d 326, 329 (2d Cir.1997); Imatec, Ltd. v. Apple Computer, Inc., 81 F. Supp. 2d 471, 480 (S.D.N.Y.2000); Ortho Pharm. Corp. v. Genetics Inst., Inc., 52 F.3d 1026, 1032-33 (Fed.Cir.1995). However, TM argues that because it has pleaded ownership, the circumstances of this case trigger an immediate shift of that burden to defendants. I disagree.

1. TM bears the burden of proving ownership of the patent

The burden of establishing TM Creditor's valid title does not shift to IBM just because IBM raised the jurisdictional issue. Cedars-Sinai Med. Ctr. v. Watkins, 11 F.3d 1573, 1583-84 (Fed.Cir.1993). TM cites Lockheed Aircraft Corp. v. United States, 213 Ct. Cl. 395, 553 F.2d 69, 88-89 (1977), for the proposition that defendants must prove Government ownership. They argue that "[t]he `heavy' burden of proving the gravamen of a Government entitlement defense, viz., conception and first actual reduction to practice within the scope of a Government procurement contract, is on the party raising that defense." (TM 10/18/00 Submission at 2.) But Lockheed does not apply to this stage of the case. There the Government raised the defenses of invalidity and license in response to the infringement charge, and had the burden of proving those defenses. However, before addressing the defenses, the court stated that Lockheed "has been and continues to be the sole owner of the patent in suit." Lockheed, at 72. Thus, the jurisdictional question of ownership was not at issue. See FilmTec at 939 F.2d at 1572-73 (cautioning against confusing plaintiff's right to maintain infringement action with in jus tertii, or title in a third person, defense). Thus, to establish standing, TM must demonstrate that Hillis and TM Patents held title to the '773 patent.

TM must do more than merely plead title ownership at this stage of the case. Where a standing issue has gone beyond the pleadings to summary judgment or trial, "the plaintiff must do more than plead standing, he must prove it." Glover River Org. v. U.S. Dept. of Interior, 675 F.2d 251, 254 n. 3 (10th Cir.1982); see also Lujan v. Defenders of Wildlife, 504 U.S. 555, 112 S. Ct. 2130, 119 L. Ed. 2d 351 (1992). In Lujan the Supreme Court stated:

Since they are not mere pleading requirements but rather an indispensable part of plaintiff's case, each element must be supported in the same way as any other matter on which plaintiff bears the burden of proof, i.e. with the same manner and degree of evidence required at the successive stages of the litigation.

504 U.S. at 561, 112 S. Ct. at 2136. Therefore, to survive IBM's motion for summary judgment, TM must come forward with specific facts to support its claim of ownership, and this evidence must be sufficient evidence for a reasonable fact-finder to infer that TM holds title to the patent.

2. TM has not met its burden of proving subject matter jurisdiction

In a motion to dismiss challenging the court's jurisdiction, the district court may engage in fact-finding and "may dismiss a facially sufficient complaint for lack of subject-matter jurisdiction if the court finds, based on affidavits or other evidence outside the complaint, that the asserted basis for federal jurisdiction is not sufficient." RAD Data Communications, Inc. v. Patton Elecs. Co., 882 F. Supp. 351, 352 (S.D.N.Y.1995) (dismissing patent infringement actions on grounds that plaintiff did not have legal title to patent-in-suit at time of alleged infringement). Here, however, there is really no need for fact-finding; the material facts are undisputed. The parties differ only in the conclusion they draw.

The parties do not dispute that Hillis' invention as claimed in the patent was developed and reduced to practice while he was at M.I.T. Neither do they dispute that the invention in the '773 is M.I.T. Case No. 3803, "Parallel Processor," which lists Hillis as the inventor. Nor do they dispute that the original '464 patent application, which was filed in 1983, pertains to the invention-in-suit. Plaintiff argues, however, that Case No. 3803 was not created under a federally-funded MIT/ARPA contract, and that Hillis' research on this invention is thus exempt from statutory requirements that he sign executory instruments in order to retain title as against the Government. Not only do they offer absolutely no evidence to support this, but IBM has provided extensive evidence to the contrary.

A patentee has presumptive title to an invention. See Beech Aircraft Corp. v. EDO Corp., 990 F.2d 1237, 1248 (Fed.Cir.1993); Arachnid, Inc. v. Merit Indus., Inc., 939 F.2d 1574, 1578 n. 2 (Fed. Cir.1991) ("The entity to whom the grant of a patent is made by the PTO [or that entity's successor in title] holds the `legal title' to the patent."). However, the United States has title to all "subject inventions" made in performing work under a funding agreement with a research organization such as M.I.T. By statute, "subject invention" is defined as "any invention conceived or reduced to practice in the performance of work under a funding agreement." 35 U.S.C. § 201(e) (emphasis added).^[11] This means that the United States had title to any and all inventions that were either conceived or reduced to practice as a result of any work carried out at MIT's Artificial Intelligence Laboratory under the MIT/ARPA contract.

Non-profits conducting research under these government funding agreements can, under 35 U.S.C. § 202, elect to obtain title to any inventions developed under or derived from those agreements. A contractor must notify the Government of inventions made under the contract within a "reasonable time." 35 U.S.C. § 202(c)(1). Failure to comply with the conditions of § 202 results in the Government's acquiring title. See Thermalon Indus., Ltd. v. U.S., 34 Fed. Cl. 411, 414 (Fed.Cl.1995) ("NSF [a Government entity] acquires title to a patent, rather than merely a license, inter alia, in the event the grantee fails to disclose within a reasonable time that the patented invention resulted from the grant.").

Under 35 U.S.C. § 202(d), an individual inventor can retain title to the patent providing he or she satisfies certain explicit conditions:

If a contractor does not elect to retain title to a subject invention in cases subject to this section, the Federal agency may consider and after consultation with the contractor grant requests for retention of rights by the inventor subject to the provisions of this Act and regulations promulgated hereunder.

The pertinent "regulations" are the Federal Acquisitions Regulations (FAR), which include specific requirements that the contractor (or the inventor if the contractor does not elect to retain title) affirmatively execute confirmatory instruments and place an appropriate statement in the patent. 48 C.F.R. § 52.227-11(f)(1), 11(f)(4).^[12]

M.I.T. Case No. 3803, which corresponds exactly to the invention described in the '464/773 application, was part of the ARPA contract. The best evidence of this comes from Hillis himself. In his 1984 letter he requested the Navy's authorization to retain the patent rights on both the 3803 and 3802 inventions, which can only be seen as an acknowledgment of the Navy's rights. He also admitted in contemporaneously published articles that his work on the inventions was funded by ARPA. Furthermore, Hillis has never explicitly stated that his invention was not developed under the contract. Aside from the self-serving innuendo in his 1984 letter to the Navy, in which Hillis intimated that the inventions were not developed with Government funding, TM offers no evidence that this invention was developed outside of ARPA sponsorship.

Hillis was thus obligated to execute the necessary forms with the Navy in order to obtain title. The evidence in the record is conclusive that he did not.^[13] Hillis received the proper forms from the Navy on three separate occasions. M.I.T.'s final report in 1987 on Case No. 3803 in 1987 showed Hillis had never forwarded the confirmatory instruments. TM has not been able to produce them in spite of the obvious advantage in doing so. Hillis himself has no recollection of signing or submitting them. Indeed, Hillis has not even submitted an affidavit other than testimony clearly asserting that he, not the Navy, was entitled to the patent in the first place. His veiled suggestion to that effect in the November 14, 1984, letter does not suffice to raise a disputed issue of material fact in this regard — especially in light of his admission in the aforementioned 1982 article in the International Journal of Theoretical Physics.

TM argues that the Navy's decision to sign off on the contract without the confirmatory documents means it was conceding title. TM cites no authority for this proposition, no doubt because it is completely counterintuitive. TM in effect argues that the Navy had an obligation to take some affirmative action if it wanted to obtain title as against Hillis. However, § 202 clearly requires the inventor working under a funding agreement to take some affirmative action if he wants to obtain title as against the Government. The Navy made its position clear: in view of M.I.T.'s disclosure, it owned the inventions; it was willing to cede those inventions to Hillis, in accordance with the provisions of § 202(d), but Hillis had to sign the forms for that to happen. He did not do so. As a result, he never took title to the invention.

Finally, there is no question that, if the Government owned the patent described under the parent application, it owned all divisional and continuation applications derived from the '474 application.

First, as set forth above, divisional applications by definition are based on the same written description as the original application and do not contain subsequently developed inventions, as a matter of law. The invention claimed in the '773 patent must, as a matter of law (35 U.S.C. § 120), have been fully described in the original '474 application in accordance with the first paragraph of 35 U.S.C. § 112, because it is the result of divisional and continuation applications. Section 112 (¶ 1) requires that a patent application "shall contain a written description of the invention ..." (emphasis added). "An applicant complies with the written description requirement by describing the invention with all its claimed limitations...." Gentry Gallery, Inc. v. Berkline Corp., 134 F.3d 1473, 1479 (Fed.Cir.1998) (emphasis added)(quoting Lockwood v. American Airlines, Inc., 107 F.3d 1565, 1572 (1997)); see also Mendenhall v. Cedarapids, Inc., 5 F.3d 1557, 1566 (Fed.Cir.1993) ("A patentee cannot obtain the benefit of the filing date of an earlier application where the claims in issue could not have been made in the earlier application.")

Divisional and continuation applications may not contain any new subject matter:

`Continuation' and `divisional' applications... are both ... based on the same disclosure as an earlier application. They differ however, in what they claim. A continuation application claims the same invention claimed in an earlier application, although there may be some variation in the scope of the subject matter claimed. A `divisional' application ... [is] carved out of an earlier application .... [It] claims only one or more, but not all, of the independent inventions of the earlier application.

Transco Prods. Inc. v. Performance Contracting, Inc., 38 F.3d 551, 555 (Fed.Cir. 1994) (emphasis added) (citations omitted).

Here, the text of the original '474 patent application, filed May 31, 1983 (IBM 10/6/00 Submission at Ex. 5.), is identical to that of the application which resulted in the '773 patent — i.e., Application Serial No. 660,323 filed February 22, 1991. (Id. at Ex. 1.) The '773 patent on its face reveals it is derived directly from the '474 application through a series of divisional and continuation applications originating with the '474 application:

Division of Ser. No. 489,179, Mar. 5, 1990, Pat. No. 5,123,109, which is a division of Ser. No. 478,082, Feb. 9, 1990, Pat. No. 5,152,000, which is a division of Ser. No. 184,739, Jun. 27, 1988, Pat. No. 5,008,815, which is a continuation of Ser. No. 499,474, May 31, 1983, Pat. No. 4,814,973.

(IBM 10/6/00 Submission at Ex. 14.) (emphasis added)

The written description of the '474 application (and thus the '773 patent) was complete as of the original filing date of May 31, 1983. Dr. Hillis' inventive activities with respect to the subject matter claimed in the '773 patent necessarily ceased as of that date. See Mendenhall, 5 F.3d at 1566. If Dr. Hillis wished to claim new or later-developed subject matter, he would have been required either to file a completely new and separate application or to insert additional written description into the '474 application and to identify specifically the modified application as a continuation-in-part ("C-I-P"). See Transco, 38 F.3d at 555 ("A `CIP' application is a continuing application containing a portion or all of the disclosure of an earlier application together with added matter not present in that earlier application. See MPEP § 201.08") (emphasis added). There is, however, no such continuation-in-part application or separate application in the '773 patent lineage, nor was there any new disclosure added to the '474/773 written description. Consistent with the identity of invention in the '474/'773 written description, plaintiffs admit that the invention claimed in the '773 patent was conceived and reduced to practice "as of May 31, 1983." (IBM 10/17/00 Submission at Ex. 2.) Clearly, nothing claimed in the '773 patent was invented by Dr. Hillis after the '474 application was filed on May 31, 1983. See Mendenhall, 5 F.3d at 1566. And at that time, he was a student at M.I.T.

Furthermore, the Manual of Patent Examining Procedure ("MPEP") makes clear that the assignment of an original application automatically carries with it ownership of all divisional, continuation, or reissue applications. See MPEP § 201.12. This automatic ownership principle makes particular sense here since the written description of the original '474 application describes the subject matter of Fig. 6B — the very embodiment of the '773 patent claims — which was "actually reduced to practice" under the terms of, and during the term of, the MIT/ARPA contract. (IBM 10/1700 Submission at Ex. 3 at ¶ 15-20.)

In view of the foregoing, TM has failed to adduce any evidence raising a disputed issue of material fact on the question of ownership/standing. There is simply no basis in this record to deny IBM's motion for dismissal/summary judgment on the ground that TM lacks standing to pursue the instant infringement claim.

II

ASSUMING THIS COURT HAS JURISDICTION, IBM IS ENTITLED TO SUMMARY JUDGMENT DECLARING THAT ITS PRODUCTS DO NOT INFRINGE THE '773 PATENT

All parties and the Court have treated IBM's challenge to TM's ownership as a challenge to the Court's subject matter jurisdiction. And, of course, a lack of subject matter jurisdiction would void all Court action with respect to the '773 (including the Markman decision interpreting the disputed claims). However, as noted in Section I.A. above, I have elected not to resolve the fascinating question of whether lack of standing goes to this Court's subject matter jurisdiction or is simply an element of its case-in-chief.^[14] Assuming arguendo that the issue of patent ownership does not go to the Court's power to act (i.e. subject matter jurisdiction), I reach the merits of the parties' cross motions for summary judgement of noninfringement as an alternative ground for decision. On the undisputed facts, I conclude that IBM is not infringing the '773 patent as I have interpreted its claims. Therefore, in the alternative, I grant IBM's motion for partial summary judgment of non-liability and deny TM's cross-motion for a judgment of liability.

A. The Patent is Enforceable

Before reaching the parties' cross-motions for summary judgment on the issue of infringement, I must dispose of a preliminary matter. Last summer, TM moved for partial summary judgment declaring that the '773 patent was enforceable. It anticipated that IBM would challenge the validity of the patent on the ground of fraud on the Patent Office. The relevant facts are undisputed and can be summarized quickly.

The '773 patent is a successor to applications that reach back as far as May 1983. In particular, it is a divisional application of Application Serial No. 489,179. In four related applications leading to the '773 application, including Serial No. 489,179, TM had cited the patent examiner to the "Adams" patent. Moreover, the Adams patent was relied upon by the examiner in disallowing each of those applications. However, the attorney who prepared the '773 application failed to disclose the Adams patent in the '773 application. IBM contended in its Answer to the First Amended Complaint and Counterclaims that this constituted inequitable conduct on TM's part, rendering the patent void and unenforceable.

The short answer to IBM's contention is that TM was not required to disclose Adams in the '773 application, since that patent was disclosed in the four prior applications leading to the '773 patent. The Manual of Patent Examination Procedure (MPEP) states that an applicant need not resubmit references cited in a parent application because the examiner is required to consider information in a parent application when examining a continuing application (such as a divisional application). See MPEP § 609. The Federal Circuit recently ruled, "In view of MPEP § 609, it can not be inequitable conduct for an applicant not to resubmit, in the divisional application, the information that was cited or submitted in the parent application." ATD Corp. v. Lydall, Inc., 159 F.3d 534, 537 (Fed.Cir.1998). This is true regardless of the materiality of the prior art reference to the patentability of the invention. See Transmatic Inc. v. Gulton Industries, Inc., 849 F. Supp. 526, 542 (E.D.Mich.1994).

IBM argues that ATD does not apply here — even though it resulted in summary judgment for the patentee against a charge of inequitable conduct due to nondisclosure — because § 609 did not become part of the MPEP until well into the prosecution of the '773 patent. It urges that the law in effect prior to November 1992, when § 609 became effective, was to the effect that an applicant was required to disclose all relevant prior art in a divisional application and could not rely on the patent examiner to locate that prior art in prior related applications. It further argues that TM was on notice that the particular examiner who examined the '773 patent was unaware of Adams, so that regardless of the state of the law, it was inequitable not to apprize him of this particular prior art.

However, IBM's argument does not hold. There is no indication in ATD, or in the case it cited on the point — Transmatic Inc. v. Gulton Indus., Inc., 849 F. Supp. 526, 539 (E.D.Mich.1994) — that the patent applications in suit postdated the final enactment of § 609. Indeed, the patent in Transmatic was issued on June 7, 1983, long before November 1992. And in any event, even under the MPEP prior to the addition of § 609, it was the patent examiner's responsibility to review parent applications for pertinent prior art. See MPEP § 705.05. A patentee cannot be penalized for a patent examiner's dereliction of duty. See Nintendo of America, Inc. v. Magnavox Co., 707 F. Supp. 717 (S.D.N.Y.1989).

The Federal Circuit has long decried the use of charging inequitable conduct in patent litigation. See Preemption Devices, Inc. v. Minnesota Mining & Mfg. Co., 732 F.2d 903, 908 (Fed.Cir.1984) (charges of inequitable conduct are a "much-abused and too often last-resort allegation."). As the PTO itself has determined that the conduct alleged by IBM is not inherently inequitable, this Court will certainly not second-guess it. TM's motion for summary judgment declaring the patent enforceable is granted.

B. IBM Is Not Infringing the '773 Patent

The '773 Patent Claims in Suit

While I have no wish to rewrite the Markman opinion, I must set forth a few pertinent facts about the '773 patent before I address the issue of infringement.

The '773 patent teaches a particular computer message routing system that uses a technique referred to by the parties as "wormhole routing." A wormhole routing system, which is described more fully in the Markman opinion, 72 F. Supp. 2d at 392, is characterized by the ability to move the initial segments of a message from one node to the next appropriate node without waiting for the rest of the message to arrive at the first node — a sort of continuous stream routing of the bits of data that comprise a single message. Wormhole routing represents an improvement over so-called "store and forward" routing, in which an entire message must arrive at a particular node before any portion of it can be transmitted to the next node, as well as over other forms of message routing.

As noted in the Markman opinion, TM originally took the position that the '773 patent patented the concept of wormhole routing. I have already rejected that notion, ruling that the '773 teaches a computer system consisting of certain hardware (two or more processors and a message router), not a process of continuous message streaming, which is what wormhole routing is. Thus, I have already ruled that "TM has not patented wormhole routing, but rather a system for accomplishing that result." Id. at 393.

For our purposes, the most important language in the claims reads as follows:

(iii) a switch connected to said input circuits for, for each message received by said input circuits, decoding one address element of the message to identify therefor an output circuit, said switch establishing a path for said message between the input circuit which received the message and the identified output circuit to facilitate the transfer of message elements of said message there between, said switch maintaining the path until the last of the serially-received message elements for the message have been transferred to the identified output circuit.

In the Markman decision, I ruled that the phrase "said switch establishing the path" meant that the switch would decide where the message would go (to which node it would next be routed) — or, in other words, "... the switch sets up a path for the message to travel from the input circuit that received the message to the output circuit that was identified during the decoding process. When it does that, the switch `establishes' the path [of the message]. It is fixed and will not be changed." Id. at 397, 116 S. Ct. 1384.

I further ruled that the phrase "said switch maintaining the path until the last of the serially-received message elements for the message have been transferred to the identified output circuit" meant the following: (1) the path between Node 1 and Node 2 that is set for Message A, once established, cannot be changed; (2) the path, once established, must be kept free for the exclusive use of Message A until that message has arrived at its destination node; and (3) no other message may cut across, or "interleave," the path that is being maintained for Message A during its transmission. Id. at 397-98, 116 S. Ct. 1384. For our purposes, the most important of those rulings is the notion that "interleaving" of messages is inconsistent with the concept of maintaining a path.

Under Independent Claims 1 and 9, the switch must establish and maintain a path for each message that goes through the system. This, too, was stipulated by the parties in connection with the Markman hearing. (IBM 10/6/00 Submission at Ex. 5.)

In the Markman decision, I held that these limitations, considered together, "are the heart of the computer system contemplated in the invention." Id. at 395, 116 S. Ct. 1384. After reviewing the materials submitted in connection with these motions, I reemphasize that holding. This is what the '773 patent is all about. If IBM's products do not do these things, then they do not literally infringe the patent, since if an accused product fails to satisfy even a single limitation of a claim, there is no literal infringement. See Kahn v. General Motors Corp., 135 F.3d 1472, 1477 (Fed. Cir.1998); Laitram Corp. v. Rexnord Inc., 939 F.2d 1533, 1535 (Fed.Cir.1991).

IBM's Products vs. The Patent

Exhibit 1, appended to this opinion, is a visualization of the system contemplated by the '773 patent, copied from the patent file. Messages are received at the input circuits (depicted on the left side of the page) and pass over wires to the designated output nodes. Once a particular wire "belongs" to a particular message, it cannot be used for any other message until the message that first "claimed" the message has been fully transmitted from the input to the output circuit. This is what the patent claim means by "establishing and maintaining" a path (which was defined, per stipulation of the parties, as a wire and its associated logic).

In Claim 2, which is a dependent claim, the '773 adds a buffer that connects to the switch and stores messages destined for an output node that is already receiving a message. If there is competition between two messages for a particular destination, the first to arrive at the switch will proceed directly to that destination, while the other is stored in the buffer until the output circuit designated by the switch is once again free. Consistent with the notion of wormhole routing, the head of the buffered message will proceed to its ultimate destination as soon as that destination is free to receive it; it need not wait until the entire message has arrived at the buffer before it can move forward.

In the most general terms, the parties have identified two sets of IBM products that are alleged to infringe the patent: they identify them as "pre-Springwood" and "post-Springwood," with "Springwood" referring to a particular change that IBM made in message routing structure (described below). Exhibit 2 appended to this opinion is an exemplar of a pre-Springwood product; Exhibit 3 is a post-Springwood product. At the left of each of the illustrations is the input circuit; at the right, the destination, or output circuit. Between the two is something called the "central queue." TM describes this as the "buffer" described in independent Claim # 2 of the '773 patent — that is, a place where messages can be queued up or stored until they can be transmitted to their final destination — and it does indeed perform that function, at least in the so-called pre-Springwood products.

Under IBM's technology, messages are made up of constituent pieces called "flits" (for "flow control digits"). Flits are either 8 bits long (working with the so-called "Vulcan" switch) or 16 bits long (working with the so-called "Trailblazer" switch). A group of 8 flits is referred to as a "chunk." In pre-Springwood products, messages could be up to 255 flits long, while post-Springwood, message length is theoretically unlimited.

The difference between the pre-Springwood and post-Springwood products is the existence in the former of a working 8 by 8 "crossbar," or direct connection, that allows messages to travel between the input and the output circuits without going through the central queue. In the post-Springwood products, the crossbar has been disabled (by causing it to "think" that all output circuits are busy), so all flits move into the central queue and are routed from there to their destination node.^[15] This change was made in 1997. TM contends that the change was made to avoid liability for patent infringement.

The absence of the crossbar, or direct path, between input and output circuits, is a major difference between the post-Springwood IBM products and the '773 claims. As will be seen below, this modification makes IBM's post-Springwood chips rather unlike the chips envisioned by the '773 patent. However, the pre-Springwood IBM products operate remarkably like the preferred embodiment of the '773 patent, reading under both Claims 1 and 2. They have a direct path between the input and output circuits (Claim 1). And they have a buffer where messages can wait when the direct path is unavailable (Claim 2).

Messages arrive at the input circuits on a flit by flit basis. In both the pre-Springwood and post-Springwood products, the first flit of the first message to reach the input circuit (Message A) is assigned by the switch to a path to an identified output circuit (along the right side of the page), whereupon it is sent to that output circuit. Whenever there is message contention in the pre-Springwood products, and in the case of all messages in post-Springwood products, the flit is diverted to the central queue (the buffer), from which it will be sent on when its designated output circuit is free.^[16]

Nonetheless, behind this superficial (and appealing) similarity lies a substantial difference in the way the two systems operate when there is contention for the same output node — a difference that also carries over into the post-Springwood era products.

IBM contends that none of its products infringe the '773 because operation of the central queue must be considered when one defines what is going on in the portion of the IBM product that provides the "switching function" — i.e. transferring each and every message received on an input port (i.e., receiver) to its respective destination output port (i.e., transmitter).^[17] (See IBM 10/12/00 Submission at 1). An inherent operational characteristic of the switch as defined by IBM is to interleave chunks of different messages with each other when multiple messages are in the process of transiting the central queue (either because they cannot transit the crossbar in pre-Springwood products, or because all messages go via the central queue in post-Springwood products). The logic ("receiver arbitration logic") that controls the movement of the flit between the input circuit and the central queue employs a methodology known as the Least Recently Used ("LRU") algorithm. Under that algorithm, the central queue "reaches" for the flit that is waiting in whatever input circuit was least recently connected to it. So, for example, if Message A is arriving at the queue from Input Circuit # 1, Message B is arriving from Input Circuit # 2, and Message C from Input Circuit # 3, the entirety of Message A (assuming that to be the first message that arrived on the scene) will not be transmitted through the queue before either of the later-arriving messages flows through. Rather, after taking a flit from Message A, the switch will check to see whether there is anything in another input circuit, and if there is, will process the flit that is waiting in whatever input circuit was least recently connected to it.^[18] As flits emerge from the central queue, there is a parallel LRU logic that causes the output port that was least recently used to read its incoming flit ahead of a more recently used output circuit. To IBM, this inherent characteristic of the central queue means that its products, whether p re-or-post-Springwood, cannot possibly infringe Claim 1, because different flits of different messages are continually being interleaved, both as they come into the central queue and as they emerge therefrom.^[19]

TM responds by arguing that pre-Springwood products infringe the '773 literally to the extent that messages that are free to do so transit the crossbar. TM referred to this at oral argument as "parttime infringement" of the patent. Put otherwise, TM claims that the crossbar is the switch in the pre-Springwood products, and argues that the independent claim (Claim 1) is necessarily infringed whenever there is no competition for a particular output circuit, because to that extent IBM's products operate exactly like the '773 patent says they should. And to the extent that pre-and-post-Springwood products send messages through the central queue, TM argues that they infringe by equivalents, because IBM's use of a process known as dynamic multiplexing means that a de facto discrete path is established and maintained for each message within the central queue.

Infringement Analysis

To determine infringement, the claims as construed are compared to the allegedly infringing device. See Texas Instruments Inc. v. Cypress Semiconductor Corp., 90 F.3d 1558, 1563 (Fed.Cir.1996). To infringe a claim, each claim limitation must be presented in the accused product, literally or equivalently. See Sofamor Danek Group, Inc. v. DePuy-Motech, Inc., 74 F.3d 1216, 1220 (Fed.Cir.1996). On motions for summary judgment, the Court is not sitting as a finder of fact, however. I am limited at this juncture to determining whether there exists any disputed issue of material fact, and whether, on the undisputed facts, judgment is warranted as a matter of law.

I consider first whether there is any evidence from which a reasonable jury could find that IBM's products infringe the '773 patent literally, and then whether they could be said to infringe under a doctrine of equivalents. I elect to consider IBM's motion first, and therefore view the evidence most favorably to TM for purposes of this motion.

Literal Infringement

On the undisputed facts, IBM's products, whether pre-or-post Springwood, do not literally infringe the '773 patent.

The issue is easiest to understand with the post-Springwood chips, so I will address them first. Because the crossbar has been disabled in these chips, all portions of all messages proceed to a single central location, from whence they are transmitted to their respective output nodes. Not all flits of a single message run into or out of the central queue over the same path, and the transmission of one message to its output circuit can be (and frequently is) interrupted by the transmission of flits from other messages, based on the LRU algorithm. Thus, the switch employed in the post-Springwood products does not "establish" or "maintain" a path as I have construed those terms from the '773 patent.

This is easiest to comprehend if one focuses on the concept of "maintaining the path until the last of the serially received message elements for the message have been transferred to the identified output circuit." The term "maintaining a path" has already been defined to preclude interleaving — based in significant part on a statement made by TM's counsel at the Markman hearing, to the effect that "If you interrupt that path and you make the original message stop and let that path be used by somebody else, then you're not maintaining the path." (Markman Hr'g Tr. at 295.) But IBM's products employ a shared-connection approach, in which multiple messages traverse the same path elements within the switch between the input and output nodes. Particular segments are never maintained for the exclusive use of a particular message. Indeed, because the central queue has only one in-port and one out-port, all flits of all messages enter and exit the central queue along the same path, with segments of different messages interleaving constantly. More significant, in view of the Markman decision, is the fact that any particular set of wires and logic elements remains dedicated to a particular message only as long as it takes a particular chunk (eight flits) of a message to traverse that particular leg of the path. Once the chunk has passed, the path becomes free, and the next thing that crosses it is as likely to be a chunk from a different message as the next chunk of the original message. There is nothing in IBM's system, as I understand its workings, to prevent a chunk from Message B to break the continuity between two chunks of Message A along the same path (again, defined as a combination of wires and logic elements).

IBM also argues persuasively that its post-Springwood system does not "establish a path" for a message as I defined that term. To satisfy the "establishing" element of the claims, the arrangement of wires and logic elements that carry a given message must be set and fixed for all elements of a given message from the input circuit all the way to the designated output. The route the message takes from the former node to the latter cannot be altered during the message's transmission. In all IBM post-Springwood switches, however, every message passing through the central queue is broken up into discrete chunks, which are repeatedly interleaved with the chunks of any other messages that also happen to be passing into or out of the central queue. To the extent that any set of wires and logic elements might be set, it is only for the particular path segment running from the input to the central queue, and subsequently the segment running from the queue to an output. No path is established at the outset from beginning to end. That is not what this Court meant when I interpreted the phrase "establishing the path."

I therefore conclude that no reasonable juror could find that the operation of IBM's central queue literally infringes the '773 patent claims as construed by my Markman decision.

Matters are trickier when considering the pre-Springwood products, because there is no evidence contradicting TM's conclusion that IBM's and TM's switches operate identically to the extent that a particular message's designated output circuit is free. TM is of course correct that there is a doctrine known as part-time infringement, and that if a claim reads on a part of an accused device, the device infringes. See Suntiger, Inc. v. Scientific Research Funding Group, 189 F.3d 1327, 1336 (Fed.Cir.1999), citing Bell Communications Research, Inc. v. Vitalink Communications Corp., 55 F.3d 615, 623 (Fed.Cir. 1995) ("[I]t should be noted any future infringement analysis respecting the assigning step should be undertaken with due attention to the principle that an accused product that sometimes, but not always, embodies a claimed method nonetheless infringes.")

IBM does not deny the existence of the part-time infringement doctrine (which was recognized in this Circuit nearly a century ago, see Wright Co. v. Herring-Curtiss Co., 211 F. 654, 655 (2d Cir.1914)). Nor does it deny that its pre-Springwood products infringe TM's patent if one focuses solely on the operation of the crossbar in non-contention situations. Instead, IBM argues that part-time infringement is precluded in the case of the '773 patent by the very terms of the claim limitation language.

IBM's argument derives from Claim 1's use of the phrase "for each message." As noted above, the claim language does require that the switch perform the decoding, establishing and maintaining functions for each message that arrives at an input circuit. IBM contends that this claim language requires that one look at how the product functions in connection with each and every message that passes through the system (i.e., in both contention and non-contention situations), so that if its products do not literally infringe all of the time (i.e., in connection with the routing of each message), they do not infringe at all. Since IBM argues that there is no evidence from which a reasonable juror could conclude that messages which are diverted to the central queue literally infringe Claim 1 — a proposition with which I have already indicated agreement — IBM concludes that its product does not infringe Claim 1 of the '773 patent. In IBM's view, only a device that infringes with respect to each message can infringe the '773 patent.

These arguments point to a gaping hole in the Court's Markman decision. At the Markman hearing, the parties did not focus on the meaning of the word "each" in the phrase "for each message," and the Court did not discuss that term in its opinion (having no idea that it was disputed). Indeed, the parties appeared to agree that "all of the recited functions of the `switch' claim must be performed on each and every message, without exception." (IBM 10/4/00 Letter at Ex. 5, citing, inter alia, Pl's Reply Mem. at 39, n. 34.) However, it has become clear to me as I worked my way through the papers submitted on the instant motion that the parties do dispute, quite vigorously, what is meant by the term "each" in the phrase "for each message."

IBM contends that the word must be read literally; that is, that the switch, having decoded the address element of each and every message, must designate its path to the output circuit ("establish" the path) and maintain the same. Under such a reading, as far as I can discern, the only embodiment that would fall under the claim language is one in which a message where a direct path to the output circuit is in use would nonetheless have a path established by the switch, would then be buffered, and once released from the buffer would proceed to the path set for it by the switch before it was buffered.

TM argues that IBM's reading of the term "each" is overly restrictive. It contends that the invention claimed in Claim 1 will work only in the non-contention situation, when in fact each message can be routed directly to its designated output along a maintained path, and that one must proceed to Claim 2 in order to address the message contention situation. Indeed, TM notes that even the preferred embodiment shows a system in which the switch (1) established a path to the designated output when there is no contention for that output circuit, but (2) established a path to the buffer (the structure added in dependent Claim 2) in contention situations. TM's affidavits reveal that, when the designated output circuit is free, the message does not necessarily proceed out of the buffer and to the output along a path that was set by the switch prior to the message's arrival at the buffer. (See, e.g., Kuszmaul Aff. ¶ 5.)

This newly-crystallized dispute adds a wrinkle to the situation. However, I must side with IBM on this issue, because it is not possible to read the word "each" in a way that supports TM's claim of part-time infringement. Doing so would violate the first maxim of the Markman decision.

The first principle undergirding the Markman hearing was set forth as follows:

In construing the patent ... I have adopted the principle that, where the language of the claim is clear and unambiguous, I will read nothing additional into it.

TM Patents, LP v. IBM Corp., 72 F.Supp.2d at 380. There is nothing unclear of ambiguous about the words "for each message." Each is not synonymous with "every." Thus, had I been aware at the Markman phase that the parties disputed the meaning of the word "each," I would have given it a literal interpretation — notwithstanding the fact that the preferred embodiment apparently does not establish and maintain a path to the output circuit for each message in a contention situation. The Markman decision is deemed amended accordingly.

Reading "each" as "some" would also allow TM to recapture certain elements that it disclaimed in order to obtain allowance of the '773 patent. In claim element iii of Claim 1 of the '773 patent, TM originally used the following language:

(iii) a switch connected to said input circuits for decoding at least one address element of each message received by said input circuits to identify therefor an output circuit, said switch establishing a path from at least some of said input circuits to the output circuits identified for the messages received thereby to facilitate the transfer of message elements there between....

As allowed, the relevant claim language reads as follows:

(iii) a switch connected to said input circuits for, for each message received by said input circuits, decoding one address element of the message to identify therefor an output circuit, said switch establishing a path for said message between the input circuit which received the message and the identified output circuit to facilitate the transfer of message elements of said message there between....^[20]

TM added the words that limited claim (iii) to "each message received by said input circuits" after the patent had been rejected by the Examiner on the basis of prior art — specifically, the Lawrence patent. This rejection is discussed in some detail in the Markman opinion. See TM Patents, L.P. v. IBM Corp., No. 97-1529 at 3 (S.D.N.Y. filed Dec. 17, 1999) ("Supplemental Markman Ruling").

In making this and other changes to the claim in order to avoid rejection, TM made the following representation to the Examiner:

One benefit of the invention recited in claim 35 [now claim 1] is that a router node can, when it receives enough of a message to determine an output circuit over which it will transmit the message, will establish a path from the input circuit to the output circuit, which it will maintain until the entire message has been transmitted....

It should be noted that this explanation does not focus on the use of the phrase "for each message." But, fairly read, it supports the Court's prior finding that paths must be established all the way from input circuit to output circuit as soon as the router node receives enough of a message to decode its address. The claim does use the language "for each message," and there is no indication in either the claim or in the prosecution history that this benefit of the invention applies only to some of the messages that are passing through the system.

The Examiner, as noted in the Markman decision, distinguished Lawrence when it finally issued the patent. However, he did not do so on the basis of the addition of the "for each message" language. He did so because Lawrence taught a "store and forward" system of message routing, rather than a wormhole routing system. See Supplemental Markman Ruling at 483-484. While he sets forth the very language from TM's "persuasive" response to his prior rejection that I have quoted above, he says nothing explicit about the need for this system to apply to "each" message. Thus, nothing in the language on which I relied in the Markman decision would support a finding that the "for each message" language was inserted to overcome Lawrence.

However, Lawrence is not the only prior art of relevance. IBM points the Court to additional prior art that was neither focused on at the Markman hearing nor specifically mentioned by the Examiner. Specifically, it calls my attention to an article by Lermani and Kleinrock, "Virtual Cut-Through: A New Computer Communication Technique," (IBM 11/13/00 at Ex. 5), which TM cited to the Patent Office in a September 24, 1992 Information Disclosure Statement. Having reviewed Kermani and Kleinrock, I conclude that it disclosed a system in which a path is established for at least some messages. (See id. at "Abstract" and p. 268). Just two months after receiving this prior art citation from TM, the Examiner stated, "The art of record fails to teach or remotely suggest the claims in their verbatim as they states and/or as argued by the applicant in his response...." (emphasis added) "In their verbatim," the claims apply to each message, not just to some messages. In view of the fact that the art of record includes a system that establishes a path for some messages, I see no way around the conclusion that TM's invention is limited to a system that establishes a path "for each message," and that the insertion of that precise language was done to overcome the impact of prior art. It has long been settled that a patentee cannot recapture subject matter that was surrendered during prosecution of the patent, and that post-hoc, litigation inspired argument cannot be used to reclaim subject matter that the record in the PTO clearly shows has been abandoned. See Desper Prods., Inc. v. QSound Labs, Inc., 157 F.3d 1325, 1340 (Fed.Cir.1998).

In its briefs and at the hearing, TM revealed that the switch in the preferred embodiment does not establish a path between the input and output circuits when there is message contention, but only establishes a path from the input circuit to the buffer. But this does not resurrect its abandoned claim; it merely indicates that the preferred embodiment set forth in the patent file teaches both Claims 1 and 2, and does not show a system that embodies only Claim 1. To the extent that the Court made any statement in its Markman opinion which suggested it was adopting the logical fallacy TM propounds here,^[21] I can only apologize and correct my error here.

Therefore, I must reject TM's "part-time infringement" argument. IBM is correct that no reasonable trier of fact, in view of the claimed requirement that the switch establish and maintain a path all the way to the output circuit for each message, rather than for some messages, could define the "switch" as the crossbar, or could look to the operation of IBM's message routing products in non-competition situations alone. The logic used in the pre-Springwood chips when there is message contention and the central queue is used is identical to that used in the post-Springwood chips — that is, messages are transmitted in chunks; paths consisting of wires and logic are established and maintained only as long as a particular chunk is in transit, and certainly are not maintained until the "last of the serially-received message elements for the message have been transferred to the identified output circuit." Therefore, because the part-time infringement doctrine cannot be applied in this unique case, there is no literal infringement of the patent by IBM's products.^[22]

I emphasize that it is only because I reached my conclusion about the meaning of the phrase "for each message" and the reason for its use in the patent that I find no literal infringement by the pre-Springwood products. If I am wrong about the inapplicability of part-time infringement in this case, then I would have reached the opposite result, because no reasonable juror could possibly conclude that the patent was not literally infringed in non-contention cases. Indeed, I search IBM's voluminous papers in vain for a single suggestion that this is not the case.

Infringement by Equivalents.

Of course, a patent may be infringed otherwise than literally. Infringement by equivalents occurs when ever limitation of the claim is equivalently present in an accused device. See Pennwalt Corp. v. Durand-Wayland, Inc., 833 F.2d 931, 934-35 (Fed.Cir.1987). Under the function-way-result test one considers whether the element of the accused device at issue performs substantially the same function, in substantially the same way, to achieve substantially the same result, as the limitation at issue in the claim. See Hilton Davis Chem. Co. v. Warner-Jenkinson Co., 62 F.3d 1512, 1518 (Fed.Cir. 1995) (en banc), rec'd. on other grounds, 520 U.S. 17, 117 S. Ct. 1040, 137 L. Ed. 2d 146 (1997); Graver Tank & Mfg. Co., Inc. v. Linde Air Prods. Co., 339 U.S. 605, 608, 70 S. Ct. 854, 856, 94 L. Ed. 1097 (1950). It is well-settled that a product does not infringe an invention by equivalents simply by performing the same function — it must do so in fundamentally the same way as is disclosed in the patent. As the United States Supreme Court held in its most recent discussion of the doctrine of equivalents, an over-broad definition of the concept of equivalence "conflicts with the definitional and public-notice functions of the statutory claiming requirement," such that the doctrine "must be applied to individual elements of the claim, not to the invention as a whole." Warner-Jenkinson Co. v. Hilton Davis Chemical Co., 520 U.S. at 29, 117 S. Ct. 1040. Thus, the doctrine of equivalents "prevents an accused infringer from avoiding infringement by changing only minor or insubstantial details of a claimed invention while retaining their essential functionality." Sage Products, Inc. v. Devon Indus. Inc., 126 F.3d 1420, 1424 (Fed.Cir.1997).

Ordinarily, the issue of equivalence is for the jury. However, "Where the evidence is such that no reasonable jury could determine two elements to be equivalent, district courts are obliged to grant partial or complete summary judgment." Warner-Jenkinson, 520 U.S. at 39, n. 8, 117 S. Ct. 1040.

On the undisputed facts before this Court, TM's claims of infringement by equivalents fare no better than did its claims of literal infringement. Indeed, the fact that TM makes these claims simply underscores its belief that any message routing system employing wormhole routing will infringe its patent. I held in the Markman opinion that TM did not own the concept of wormhole routing. Therefore, it cannot bar IBM, or any other competitor, from coming up with a different wormhole routing system. And that is exactly what IBM has done here.

TM's argument can be summarized as follows: in or about 1924, scientists developed a system known as multiplexing, which reconstructs waves by using sampling points. Multiplexing permits several different messages to travel through the same wire at the same time. Different data in the wire relate to different messages, with particular sampling points containing data pertaining to Message A, different sampling points containing data comprising Message B, and so forth. The most obvious example of multiplexing is a telephone wire, which carries thousands of conversations simultaneously through the same medium, with the medium's logic set in such a way that each message is on a different time sequence. Thus, my call to you reaches your telephone receiver, while Bill's call to Jane reaches Jane's telephone receiver, even though they are carried over the same wire.

In a nutshell, TM claims that IBM employs a form of multiplexing to run messages from various input nodes through the central queue and on to their output modes simultaneously (or virtually so). TM urges that IBM's use of multiplexing infringes its patent because each message is on an "exclusive path," that path being its own time sequence. As TM describes IBM's use of time multiplexing, each receiver in a multi-receiver system is allotted a recurring sequence of time cycles during which it can transmit a portion of a message over a single shared connection. TM insists that the sequential ordering of the receivers is on an unchanging pattern, thereby creating a "path" that is "established" and "maintained." TM has submitted affidavits from individuals who, for purposes of this motion will be deemed to be skilled in the art, in which those individuals purport to reconstruct what must be occurring inside IBM's central queue, based on their knowledge of the phenomenon of time-division multiplexing. That is the only evidence offered in support of TM's contention that the operation of the central queue results in the establishment and maintenance of a path.

IBM argues in response that TM is wrong, because TM's witnesses have guessed wrong about how the central queue works, and because the way the system actually does work is not equivalent to the teaching of the '773.

IBM is right on both counts.

There can be little question that the '773 patent does not teach multiplexing. Plaintiff admits that the concept has been known and used widely since the 1920s, yet there is no mention of it in the patent itself or the prosecution history. Nor did it come up at the Markman hearing.

Nonetheless, time multiplexing could qualify as an equivalent if it establishes and maintains discrete paths for each message, as long as it does so in fundamentally the same way that the '773 patent does. I must again emphasize this point: the fact that IBM's products transmit messages via wormhole routing is insufficient to warrant a finding of equivalence. Equivalence requires wormhole routing as taught by the '773 patent — via a path established at the outset of message transmission and maintained for the exclusive use of a particular message until the last bit of data is received at its identified destination.

There is nothing in the record before me from which a reasonable juror could conclude that IBM's products, whether pre-Springwood or post-Springwood, accomplish that result. Indeed, on the record before me, I am constrained to agree with IBM that its chips cannot possibly be deemed the equivalent the invention claimed in the '773 patent, because, while they transmit messages via wormhole routing, they do so in a way that is absolutely antithetical to the patent's disclosures.

As noted above, TM submits conclusory affidavits from individuals who understand time multiplexing. Those individuals draw conclusions about how IBM's queue works based on that understanding. IBM, however, submits affidavits from Craig Stunkel and Derrick L. Garmire — persons who work on the accused products — which establish that TM's witnesses are in error. IBM's products do not assign a recurring sequence of time slots over a single conductor to each individual message, as TM's witnesses hypothesize. Instead, the central queue, at both its input and output ports, operates in accordance with the aforementioned LRU algorithm. When additional message chunks simultaneously are received at additional receivers, the order in which receivers send chunks into the central queue is not sequential. Rather, the receiver that was not serviced for the longest time becomes the next one to be serviced. If this means that a chunk of new message needs to "cut in" ahead of a chunk belonging to a message that is already in transit, then in it cuts — and in doing so, it takes away the time slot to which the earlier message would otherwise have been "assigned." The more messages that arrive simultaneously or in close temporal proximity to one another, the more such interleaving takes place — and with each interleaving, the time sequence of each message then in the system changes.

The only competent evidence before me demonstrates that, far from establishing a fixed path (assuming arguendo that a time sequence could be a "path" within the meaning of the '773 patent claims — itself a highly dubious proposition^[23]), IBM's switch operates in a manner that precludes any finding of a message's "path" being established at the outset or maintained during transit for its exclusive use. Thus, its products do not operate in fundamentally the same way as the '773 patent does, and TM's claim of infringement by equivalents must be dismissed.

I am buttressed in this conclusion by my reading of two recent Federal Circuit cases. For example, in Hoganas AB v. Dresser Industries, Inc., 9 F.3d 948 (Fed. Cir.1993), the Court of Appeals found no infringement by equivalents where both the patent and the allegedly infringing product performed the same overall function — promoting the rapid drying of refractory linings by reducing the risk of explosions — that did so "in a different way." The patent claim utilized elements that were hollow, which "provide in their natural state channels for the release of steam," and which were at least partly burned away during the process. The allegedly infringing products, by contrast, worked through capillary action, and the elemental fibers melted as the process worked, rather than being burned away. Similarly, in Zodiac Pool Care, Inc. v. Hoffinger Industries, Inc., 206 F.3d 1408 (Fed.Cir.2000), the Court of Appeals affirmed judgment as a matter of law in favor of the accused infringer whose product did not embody a certain "clear structural limitation" in the patent in suit — namely, that a stop be located substantially inward of the peripheral edge of a disc. The Court ruled that, if infringement were allowed under the doctrine of equivalents, despite the substantial structural differences between the patent and defendant's product, the claim limitation would be reduced to a "functional abstract[], devoid of meaningful limitations on which the public could rely." Id. at 1416 (quoting Sage Products, 126 F.3d at 1424).

If the differences between the infringing products and the patents in suit in Zodiac Pool and Hoganas are sufficient to preclude application of the doctrine of equivalents as a matter of law, then there can be no question that summary judgment ought to be granted to IBM, given the utter lack of similarity between its process for the wormhole routing of messages and that disclosed in the '773 patent.

Accordingly, I rule that nothing in the record before me raises a disputed issue of material fact, and that IBM is entitled to a judgment of non-infringement as a matter of law. TM's claims of patent infringement relating to the '773 patent are dismissed.

We will proceed to trial on the claimed infringement of the '342 patent only. The parties are directed to participate in a telephone conference related to scheduling on November 15 at 9 a.m.

This constitutes the decision and order of the Court.

NOTES

[1] The Court has considered: (1) TM and IBM's October 6 submissions responding to the Court's October 4 inquiry, (2) the parties' October 12 submissions responding to the Court's October 11 inquiry, and (3) IBM's October 17 and TM's October 18 submissions in response to the Court's final October 13 inquiry. I have also considered and relied on letters submitted by both sides on November 13, 2000. The Clerk is directed to docket all these letters to insure that they are part of the record.

[2] All of Engellenner's memoranda and correspondence are hearsay, though it would appear that some portions of those documents could be admitted into evidence under the business records exception to the hearsay rule if a proper foundation were laid. Fed.R.Evid. 803(6).

[3] I am the first to acknowledge that this may not be an inconsistency, the language of patents being what it is.

[4] See Part I.E.2, infra.

[5] In DaSilva, the Second Circuit considered whether the district court's dismissal of an employee's Title VII complaint based on the employer's not having 15 employees was for failure to state a claim or for lack of subject matter jurisdiction. The Second Circuit agreed with the district court, Judge Berman, that Title VII's "15-employee" rule was a component of the cause of action, not a prerequisite to the exercise of subject matter jurisdiction. See id. at 360-61. DaSilva would offer more guidance here if the question of plaintiff's standing to bring suit were also at issue in that case.

[6] The question of title to the '773 patent was not actually litigated or determined in the bankruptcy proceeding. Accordingly, collateral estoppel, or issue preclusion (as opposed to res judicata, or claim preclusion), does not apply to bar IBM from raising the issue here. See Interoceanica Corp. v. Sound Pilots, Inc., 107 F.3d 86, 91 (2d Cir.1997).

[7] In its Supplemental Brief on this issue, TM asserts that the Bankruptcy Court did actually adjudicate the issue (TM 10/6/00 Submission at 13.), but it offers no evidence to support that assertion — an assertion that contradicts its prior statements to the Court.

[8] While the Government could have raised the issue in the bankruptcy, IBM is not in privity with the Government. Therefore, the former adjudication bar — which applies only to parties and their privies, see St. Pierre v. Dyer, 208 F.3d, 394, 399 (2d Cir.2000) — does not apply.

[9] The instant situation differs from the situation in the cases relied on by TM. In Sure-Snap Corp. v. State St. Bank and Trust Co., 948 F.2d at 872-73, the focus of the bankruptcy proceeding encompassed the entire lender-debtor relationship, and the bank's post-lending conduct was also at issue in the subsequent proceeding. The trustee's motion to expunge plaintiff creditor's claims in In re American Preferred Prescription, 2000 U.S. Dist. LEXIS 10150, *5-6 (E.D.N.Y. March 20, 2000), was barred by the previous bankruptcy plan, which subordinated the creditor's claims. Stoll v. Gottlieb, 305 U.S. 165, 59 S. Ct. 134, 83 L. Ed. 104 (1938), was an action to recover on a guaranty of a corporation's bonds. The Supreme Court held that prior orders of the bankruptcy court denying plaintiff's petition to set aside a decree providing for release of a guaranty were res judicata on the issue of guarantor liability. See id. at 170-71, 59 S. Ct. 134. In each of those cases, res judicata applied because the issues in the second proceeding were identical to those that were or could have been raised in the first. Finally, in Computer Assocs. Int'l Inc. v. Altai, 126 F.3d at 369, the court found res judicata inapplicable.

[10] There was no document of assignment to record. If there had been, the assignment, if unrecorded, would be void as against subsequent assignees for value without notice. The fact that the government's title was vested before the bankruptcy would not save it.

[11] Under 35 U.S.C. § 201(d), "invention" is defined as "any invention or discovery which is or may be patentable under this title ..."

[12] TM claims (in one of many footnoted substantive arguments) that inventors are not subject to the reporting requirements of Section 202(c); that as non-parties to the contract they are separately treated under 202(d) and exonerated from having to provide confirmatory instruments called for by 48 C.F.R. § 52.227-11(f)(1). TM offers no support for this proposition. M.I.T.'s own policies required that Hillis assign to the university inventions conceived or reduced to practice at M.I.T. with the use of sponsored research funds or significant amounts of M.I.T. funds or facilities. There is no evidence from which I could conclude that M.I.T. funds and/or facilities were not used by an M.I.T. graduate student in the creation of his invention. Hillis certainly does not deny it; his 1982 article confirms it. Thus, whatever Hillis did at M.I.T. belonged to the school unless the school waived its rights in the invention, as was clear from Engellener's April 1983 letter to Hillis. (IBM 9/12/00 Submission at Ex. 3.) And if the invention known as Case 3803 belonged to M.I.T. rather than Hillis, Hillis had no standing to complain if M.I.T. elected to treat the invention as ARPA-sponsored.

[13] Hillis also failed to place the required notice in the patent of the Government's rights in the invention.

[14] See discussion of the DaSilva opinion at n. 5, supra.

[15] To be precise, the pre-Springwood products are the Vulcan chip switch and the Trailbralzer chip switch without the Springwood alteration. The post-Springwood products are Trailblazer chip switches with the Springwood alteration.

[16] The central queue is part of a switch structure that is itself the subject of an IBM patent, U.S. Patent No. 5,546,391 ("Hochschild").

[17] I refer to this component of IBM's product as a "switch" for convenience. IBM maintains that there is no "switch" or "switch equivalent" to the '773 patent.

[18] The easiest analogy I can think of for the LRU algorithm is the Big Ten Football Conference's long-standing method for selecting a Rose Bowl representative when two or more teams are tied for the conference championship: the team that has gone the longest without making a Rose Bowl appearance gets the nod. This rule was not modified by the conference's participation in the BCS ranking system. See "Method to Determine Rose Bowl Participant" (visited Nov. 4, 2000) .

[19] TM does not claim that its patent teaches anything like the LRU logic employed by IBM.

[20] I here highlight the language that is relevant to my analysis of the parties' arguments. I note that, at the Markman hearing, the parties focused on the meaning of the word "one," used elsewhere in this limitation. That issue is discussed extensively in the Markman opinion and is also the subject of a supplemental order, dated December 17, 1999 (which supplemental order ought to be appended to the published Markman decision — an omission the Court will rectify). That language is omitted from this discussion, because it has nothing to do with the dispute here identified.

[21] TM points to the following Markman opinion sentence, "If that identified circuit is busy, then the head of the message must proceed to a message buffer, which will be discussed later," to support its claim that the Court, either explicitly or sub silentio, recognized that the word "each" meant "some." If my language left this impression, I apologize. As noted above, this aspect of the claim language was not the subject of argument by the parties and was not thought about, let alone addressed, by this Court. Frankly, I was too busy trying to make sense of what the parties were arguing to hypothesize about what they might have argued.

[22] In addition to patent estoppel, IBM argues that TM is trying to stand the concept of independent and dependent claim on its head, by positing a situation in which the dependent claim was broader, not narrower, than its associated independent claim. It argues that, because the dependent claim covers both contention and non-contention situations, while the independent claim covers only non-contention situations, the dependent claim is broader than the independent claim.

Of course, it is well settled that a dependent patent claim must be narrower than its associated independent claim, see 35 U.S.C. § 112, ¶ 4, and must incorporate by reference all of the limitations of the independent claim from which it refers. See Slater Elec., Inc. v. Thyssen-Bornemisza Inc., 650 F. Supp. 444, 464 (S.D.N.Y.1986). However, IBM's superficially appealing argument is logically fallacious. A dependent claim may well be "narrower" (i.e., more restrictive) than its associated independent claim while nonetheless expanding on the type of situation in which the patent would be infringed. For example, suppose a patent has the following two claims:

1. An automobile having an engine, four wheels and a drive train for transmitting driving power to at least one opposed pair of said wheels.

2. An automobile as described in Claim 1 in which said drive train includes both forward and reverse gears and means for shifting between said gears to cause the driven wheels to be driven in either the forward or reverse direction as desired.

Obviously Claim 2 is narrower than Claim 1. Yet Claim 1 will be infringed by an automobile which can be driven only in the forward direction, while Claim 2 is infringed (only) by an automobile which can be driven in either direction.

Thus a narrower (dependent) claim may, and frequently does, cover a device capable of more types of operation (handling more "situations") than a device that would be covered by a broader (independent) claim.

[23] At the Markman hearing, the parties stipulated that a "path" consisted of a wire and its associated logic. I suppose it is possible that logic for assigning time sequences within a single wire to a particular message could constitute a "path" within that definition, although that is certainly not what the Court had in mind during the Markman process. However, the process of message interleaving pursuant to the LRU algorithm used by IBM does not create a "path" that is either established or maintained within the Markman definition.