PROST, Circuit Judge.

Cisco Systems, Inc. appeals from the final judgment of the United States District Court for the Eastern District of Texas, which was based on the findings of two separate jury verdicts finding that: Cisco directly and indirectly infringed specified claims of Commil USA, LLC's U.S. Patent No. 6,430,395 ("'395 patent"); the specified claims of the `395 patent are not invalid as indefinite, for lack of enablement, or as lacking adequate written description; and that Cisco was liable for $63,791,153 in damages as well as pre-judgment interest and costs. We find that the district court gave the jury a legally erroneous instruction with respect to indirect infringement. Additionally, we find that Cisco's evidence of a good-faith belief of invalidity may negate the requisite intent for induced infringement. However, we find that the district court did not err in granting a partial new trial. Thus, we affirm-in-part, vacate-in-part, and remand for further proceedings consistent with this opinion.

I. BACKGROUND

A. THE PATENT AND ACCUSED PRODUCTS

In a wireless system, mobile devices such as phones and laptop computers communicate with fixed "base stations" according to standardized procedures that govern the way in which data exchanged between devices is formatted, ordered, maintained, and transmitted. These procedures are referred to as "protocols." Effective wireless communication requires that the transmitting device and the receiving device follow the same protocol.

The `395 patent relates to a method of providing faster and more reliable handoffs of mobile devices from one base station to another as a mobile device moves throughout a network area. The `395 patent teaches that the communication protocol is divided based on time sensitivity. The portions of the protocol requiring accurate time synchronization — "real-time capabilities" — are performed at the base station. This part of the protocol is called the "low-level protocol." Other parts of the protocol that are not time-sensitive comprise the "high-level protocol," which is performed on another device called a switch. The base station and switch cooperate to handle a connection with a mobile unit. To implement the full communications protocol, the base station runs an instance of the low-level protocol for the connection and the switch runs a corresponding instance of the high-level protocol.

Cisco is a major supplier of WiFi access points and controllers. Commil alleges that certain Cisco access points and controllers infringe claims 1, 4, and 6 of the `395 patent. Claim 1, the patent's sole independent claim, provides:

B. THE DISTRICT COURT PROCEEDINGS

A jury trial commenced on May 10, 2010. On May 17, 2010, the jury returned a verdict rejecting Cisco's invalidity contentions, finding Cisco liable for direct infringement, and awarding Commil $3.7 million in damages. The jury also found that Cisco was not liable for induced infringement. Commil filed a motion for a new trial on the issues of induced infringement and damages, which the court granted on December 29, 2010.

On April 5, 2011, a second trial was held with respect to indirect infringement and damages. On April 8, 2011, the jury returned a verdict in favor of Commil on both issues and this time awarded $63.7 million in damages. On September 28, 2011, the district court entered an amended final judgment granting $63.7 million in actual damages, $10.3 million in prejudgment interest, and $17,738 in costs. This appeal followed.

II. DISCUSSION

Cisco appeals the district court decision on several grounds. First, Cisco contends that an erroneous instruction allowed the jury to find inducement based on mere negligence. Second, Cisco argues that the district court erroneously precluded Cisco from presenting evidence of its good-faith belief of invalidity to show that it lacked the requisite intent to induce infringement of the asserted claims. Third, Cisco argues that the district court abused its discretion in granting a new trial and that the district court violated the Seventh Amendment by granting a new trial on certain issues, but not others. Fourth, Cisco claims the court erred in construing the term "short-range communication protocol." Fifth, Cisco argues that there is not substantial evidence to sustain the jury verdict on infringement. Sixth, Cisco contends the claims are indefinite, not enabled, and lacking adequate written description. Finally, Cisco objects to the damages award on the grounds that Commil's royalty base violates the entire market value rule. We take each of these issues in turn.

A. THE JURY INSTRUCTION

Before the district court and on appeal, Cisco challenged the second trial's jury instruction on induced infringement. The district court denied Cisco's motion for a new trial on the jury instruction issue. We review the denial of a motion for a new trial under the law of the regional circuit. Riverwood Int'l Corp. v. R.A. Jones & Co., 324 F.3d 1346, 1352 (Fed.Cir. 2003). In the Fifth Circuit, the denial of a motion for a new trial "will not be disturbed absent an abuse of discretion or a misapprehension of the law." Prytania Park Hotel v. Gen. Star Indem. Co., 179 F.3d 169, 173 (5th Cir.1999). Whether a jury instruction on an issue of patent law is erroneous is a matter of Federal Circuit law that is reviewed de novo. Sulzer Textil A.G. v. Picanol N.V., 358 F.3d 1356, 1363 (Fed.Cir.2004) (citing Advanced Display Sys., Inc. v. Kent State Univ., 212 F.3d 1272, 1282 (Fed.Cir.2000)). We will set aside the jury verdict, "if the movant can establish that `those instructions were legally erroneous,' and that `the errors had prejudicial effect.'" Id. (citations omitted). In reviewing jury instructions, we review the trial record and the jury instructions in their entirety. Id.

At the second trial, the court instructed the jury that it could find inducement if "Cisco actually intended to cause the acts that constitute direct infringement and that Cisco knew or should have known that its actions would induce actual infringement." J.A. 6389 (98:24-99:2). The "knew or should have known" language is a verbatim recitation of the standard for showing induced infringement we originally set forth in Manville Sales Corp. v. Paramount Sys., Inc., 917 F.2d 544, 553 (Fed. Cir.1990). This court, sitting en banc, again approved this language in DSU Medical Corp. v. JMS Co., Ltd., 471 F.3d 1293, 1306 (Fed.Cir.2006) (en banc). Cisco alleges that this instruction allowed the jury to find inducement on the showing of mere negligence and, as such, is legally erroneous in view of the Supreme Court's recent decision in Global-Tech Appliances, Inc. v. SEB S.A., ___ U.S. ___, 131 S.Ct. 2060, 179 L.Ed.2d 1167 (2011).

The Global-Tech Court held that induced infringement "requires knowledge that the induced acts constitute patent infringement." Id. at 2068. The knowledge requirement of Global-Tech may be satisfied by showing actual knowledge or willful blindness. Id. at 2072. In reaching this conclusion the Court expressly distinguished actual knowledge and willful blindness from recklessness and negligence explaining that:

Id. at 2070-71 (citations omitted). The Court acknowledged that the facts that must be adduced to find willful blindness prevent such a finding on facts that support only recklessness or negligence. Id. Moreover, the Court rejected the standard set forth by this court, in part, because it permitted "a finding of knowledge when there is merely a `known risk' that the induced acts are infringing." Id. at 2071.

Commil contends that the jury instruction in this case merely allowed the jury to find knowledge based upon circumstantial evidence. Circumstantial evidence can, of course, support a finding of actual knowledge or willful blindness just as it did in Global-Tech. Id. at 2071-72. The jury instruction in this case, however, was not so limited. While the court did instruct the jury that certain circumstantial evidence could support a finding of inducement, the present jury instruction plainly recites a negligence standard, which taken literally, would allow the jury to find the defendant liable based on mere negligence where knowledge is required. J.A. 6389 (98:19-99:15). Therefore, to the extent our prior case law allowed the finding of induced infringement based on recklessness or negligence, such case law is inconsistent with Global-Tech and no longer good law. It is, therefore, clear that the jury instruction in this case was erroneous as a matter of law. This finding, however, does not end our inquiry.

In order to set aside a jury verdict, we must find not only that the jury instruction was legally erroneous, but also that the instruction had a prejudicial effect. Sulzer Textil, 358 F.3d at 1364 ("[I]t is not enough to merely show that a jury instruction is erroneous; [petitioner] also must show that the erroneous jury instruction was prejudicial."). If the erroneous jury instruction "could not have changed the result, the erroneous instruction is harmless." Environ Prods., Inc. v. Furon Co., 215 F.3d 1261, 1266-67 (Fed.Cir.2000). Commil contends that when viewed as a whole, the jury instruction required the jury to find facts that satisfy the Global-Tech standard and, therefore, there is no prejudicial effect. We cannot agree.

A finding of inducement requires both knowledge of the existence of the patent and "knowledge that the induced acts constitute patent infringement." Global-Tech, 131 S.Ct. at 2068; see also DSU Med. Corp., 471 F.3d at 1306 (explaining that an "alleged infringer must be shown ... to have knowingly induced infringement," not merely knowingly induced the acts that constitute direct infringement" (citation omitted)). Here, the jury was clearly instructed that Cisco could not be liable for induced infringement if it was not aware of the `395 patent. The jury was also instructed that Cisco must have actively and knowingly aided and abetted direct infringement. The jury, however, was not instructed that in order to be liable for induced infringement, Cisco must have had knowledge that the induced acts constitute patent infringement. On the contrary, the jury instruction allowed Cisco to be held liable if "Cisco knew or should have known that its actions would induce direct infringement." J.A. 6389 (99:10-11). With respect to whether the induced acts constitute patent infringement, it is clear that the jury was permitted to find induced infringement based on mere negligence where knowledge is required. This erroneous instruction certainly could have changed the result. Facts sufficient to support a negligence finding are not necessarily sufficient to support a finding of knowledge. Accordingly, we vacate the jury's verdict on induced infringement and remand for a new trial. Because we vacate the induced infringement verdict upon which the damages award is based, we also vacate the damages award.

B. CISCO'S GOOD-FAITH BELIEF OF INVALIDITY

Cisco further contends that the district court erred in preventing Cisco from presenting evidence during the second trial of its good-faith belief of invalidity to rebut Commil's allegations of induced infringement. We agree.

Prior to the second trial, Cisco proffered evidence to support its good-faith belief that the `395 patent is invalid. Commil filed a motion in limine to exclude this evidence, which the district court granted without written opinion. It is not entirely clear from the record why the district court precluded Cisco from presenting its evidence. However, during a colloquy with Cisco's counsel at a pretrial hearing, the district court appeared to base its decision on the fact that our precedent indicates that such evidence is relevant where it relates to a good-faith belief of non-infringement, but is silent with respect to invalidity. J.A. 6061-63. It is true, as the district court noted, that we appear to have not previously determined whether a good-faith belief of invalidity may negate the requisite intent for induced infringement. We now hold that it may.

Under our case law, it is clear that a good-faith belief of non-infringement is relevant evidence that tends to show that an accused inducer lacked the intent required to be held liable for induced infringement. See DSU Med. Corp., 471 F.3d at 1307 (finding a demonstrated belief of non-infringement sufficient to support a jury verdict that the defendant did not induce infringement); Ecolab, Inc. v. FMC Corp., 569 F.3d 1335, 1351 amended on reh'g in part, 366 Fed.Appx. 154 (Fed.Cir.2009) (finding that a reasonable belief of non-infringement supported a jury verdict that the defendant lacked the intent required for induced infringement); Kinetic Concepts, Inc. v. Blue Sky Med. Grp., Inc., 554 F.3d 1010, 1025 (Fed.Cir.2009) (holding that defendant's "belief that it can freely practice inventions found in the public domain" supports "a jury's finding that the intent required for induced infringement was lacking"); Bettcher Indus., Inc. v. Bunzl USA, Inc., 661 F.3d 629, 649 (Fed. Cir.2011) (finding opinion of counsel regarding non-infringement "admissible, at least with respect to [defendant]'s state of mind and its bearing on indirect infringement"). We see no principled distinction between a good-faith belief of invalidity and a good-faith belief of non-infringement for the purpose of whether a defendant possessed the specific intent to induce infringement of a patent.

It is axiomatic that one cannot infringe an invalid patent. See, e.g., Prima Tek II, L.L.C. v. Polypap, S.A.R.L., 412 F.3d 1284, 1291 (Fed.Cir.2005) ("there can be no ... induced infringement of invalid patent claims"); Richdel, Inc. v. Sunspool Corp., 714 F.2d 1573, 1580 (Fed. Cir.1983) ("The claim being invalid there is nothing to be infringed."). Accordingly, one could be aware of a patent and induce another to perform the steps of the patent claim, but have a good-faith belief that the patent is not valid. Under those circumstances, it can hardly be said that the alleged inducer intended to induce infringement. Thus, a good-faith belief of invalidity is evidence that may negate the specific intent to encourage another's infringement, which is required for induced infringement. Several district courts have considered this question and come to the same conclusion. See VNUS Med. Techs., Inc. v. Diomed Holdings, Inc., 2007 WL 2900532, at *1 (N.D.Cal. Oct. 2, 2007) ("denying plaintiff's motion for summary judgment on induced infringement based, in part, on an opinion of counsel that the patents-in-suit were invalid"); Kolmes v. World Elastic Corp., 1995 WL 918081, at *10 (M.D.N.C. Sept. 18, 1995) (finding, after a bench trial, no intent to induce infringement where defendants "had a good faith belief in the invalidity" of the patent-in-suit); DataQuill Ltd. v. High Tech Computer Corp., 2011 WL 6013022, at *10 (S.D.Cal. Dec. 1, 2011) (indicating that a belief of invalidity may present a triable issue of fact as to intent to induce infringement); see also Lemley, Inducing Patent Infringement, 39 U.C. Davis. L.Rev. 225,-243 (2005) ("[I]t is not reasonable to assume that merely because a defendant is aware of the existence of a patent, he intended to infringe it. He may believe the patent invalid"); but see Applera Corp. v. MJ Research Inc., 2004 WL 367616, at *1 (D.Conn. Feb. 24, 2004); LadaTech, LLC v. Illumina, Inc., 2012 WL 1188266, at *2 (D.Del. Feb. 14, 2012).

We now hold that evidence of an accused inducer's good-faith belief of invalidity may negate the requisite intent for induced infringement.1 This is, of course, not to say that such evidence precludes a finding of induced infringement. Rather, it is evidence that should be considered by the fact-finder in determining whether an accused party knew "that the induced acts constitute patent infringement." Global-Tech, 131 S.Ct. at 2068.

C. THE GRANT OF A SECOND TRIAL

Cisco challenges the district court's grant of a new trial on two fronts. First, Cisco argues that there was no basis for granting a new trial. Second, Cisco argues that even if a new trial was proper, the partial new trial violated the Seventh Amendment. We discuss these issues seriatim.

1. THE FIRST TRIAL

The district court proceedings in this case were unusual. Commil is based in Israel and the inventors of the `395 patent are Israeli. Throughout the trial, according to the district court, Cisco's trial counsel attempted to play upon religious prejudices and ethnic stereotypes.

For instance, during the cross-examination of Jonathan David, a co-owner of Commil who is Jewish, Cisco's counsel attempted to perpetuate the stereotype of Jewish people as greedy opportunists by asking Mr. David if his cousin was a "bottom-feeder who swim[s] around on the bottom buying people's houses that they got kicked out of for next to nothing." J.A. 5823 (139:19-140:1). Later, when Mr. David mentioned dining at a local barbeque restaurant, Cisco's counsel quipped, "I bet not pork." J.A. 5825 (146:4-24). Following the pork comment, the court questioned counsel on the relevance of his statement and issued a curative instruction stating:

J.A. 5838 (2:25-3:9).

Despite the potent curative instruction and the court's clear displeasure, in his closing, counsel again made several irrelevant and prejudicial remarks. Counsel's behavior reached a new low when he began his closing argument with a reference to the trial of Jesus Christ, stating:

J.A. 6038 (16:1-16).2

After discharging the jury, the court again expressed displeasure with Cisco's counsel and informed Commil that should they file it, a motion for a new trial would be entertained. Shortly thereafter, Commil filed a motion for a new trial on the issues of indirect infringement and damages. In ruling on the motion, the court found that when counsel's comments regarding the trial of Jesus were viewed in context with other comments regarding Mr. David and the inventors Jewish heritage, it was clear that counsel was attempting to align his "religious preference with that of the jurors and employs an `us v. them' mentality — i.e., `we are Christian and they are Jewish.'" Commil USA, LLC v. Cisco Sys., Inc., No. 2:07-CV-341, slip op. at 3 (E.D.Tex. Dec. 29, 2010). The court granted the motion, finding that the comments prejudiced the jury with respect to indirect infringement and damages. Cisco filed a motion for reconsideration, which the court denied.

1. GRANT OF A NEW TRIAL

We review issues not unique to patent law, such as the grant of a new trial based on the prejudicial remarks of counsel, under regional circuit law. Riverwood Int'l Corp., 324 F.3d at 1352. The Fifth Circuit reviews rulings on new trial motions for abuse of discretion, with more exacting review applied to orders granting a new trial than to those denying them. Conway v. Chem. Leaman Tank Lines, Inc., 610 F.2d 360, 362-363 (5th Cir.1980). "[A] new trial will not be granted, even if counsel's remarks are improper, unless after considering the record as a whole the court concludes that manifest injustice would result from letting the verdict stand." Gautreaux v. Scurlock Marine, Inc., 84 F.3d 776, 783 (5th Cir.1996). This is particularly the case where, as here, the statements drew no objection from the opposing party: "[I]mproper argument may be the basis for a new trial where no objection has been raised only `where the interest of substantial justice is at stake.'" Hall v. Freese, 735 F.2d 956, 961 (5th Cir.1984) (quoting Edwards v. Sears, Roebuck & Co., 512 F.2d 276, 286 (5th Cir. 1975)).

As discussed supra, the district court granted a new trial based on what it viewed as the prejudicial effect of inflammatory statements made by Cisco's counsel during trial. Cisco claims that the statements do not warrant a new trial. Cisco asks us to review the cold record — substituting our judgment for the district court's — and find that there was no manifest injustice in this case. We decline.

In reviewing the district court's ruling, it is clear that the court did not abuse its discretion. There is ample evidence from which the district court could conclude that the jury was biased by Cisco's actions. Throughout trial, Cisco attempted to instill in the jury, through irrelevant references to ethnicity and religion, an "us versus them" mentality. Cisco persisted in its course of conduct even after the court warned counsel and issued a curative instruction. And, in a case involving Jewish inventors and plaintiffs, Cisco's counsel began his closing argument with a reference to the trial of Jesus Christ.

Even if we were inclined to agree with Cisco that there is no manifest injustice in this case — and we are not — we refuse to substitute our judgment for that of a district court whose "on-the-scene assessment of the prejudicial effect, if any, carries considerable weight." United States v. Munoz, 150 F.3d 401, 415 (5th Cir. 1998). Accordingly, we find that the district court did not abuse its discretion in granting a new trial.

2. PARTIAL NEW TRIAL

The Federal Rules of Civil Procedure allow the courts to grant partial new trials so long as the issues are "distinct and separable." Fed.R.Civ.P. 59. A court's authority to grant a partial new trial is likewise constrained by the Seventh Amendment. Gasoline Prod. Co. v. Champlin Refining Co., 283 U.S. 494, 500, 51 S.Ct. 513, 75 L.Ed. 1188 (1931). "Where the practice permits a partial new trial, it may not properly be resorted to unless it clearly appears that the issue to be retried is so distinct and separable from the others that a trial of it alone may be had without injustice." Id. A partial new trial should not be granted where the issues to be retried are "so interwoven" with other issues in the case "that the former cannot be submitted to the jury independently of the latter without confusion and uncertainty." Id. We have explained, however, that the Seventh Amendment "prohibition is not against having two juries review the same evidence, but rather against having two juries decide the same essential issues." In re Innotron Diagnostics, 800 F.2d 1077, 1086 (Fed.Cir.1986) (citation omitted). Trying issues separately is appropriate where "separate trials would not constitute a `clear and indisputable' infringement of the constitutional right to a fair trial." Id. (citing Bankers Life & Casualty Co. v. Holland, 346 U.S. 379, 384, 74 S.Ct. 145, 98 L.Ed. 106 (1953)).

Cisco contends that the district court violated the Seventh Amendment by granting a new trial on the issues of induced infringement and damages, but not direct infringement and validity. Specifically, Cisco contends, under the circumstances of this case, indirect infringement is not distinct and separable from validity, but rather, they are inextricably intertwined. Cisco argues that where the plaintiff alleges induced infringement and the defendant has evidence of a good-faith belief of invalidity, the issues of validity and induced infringement are not distinct and separable. We disagree.

We note at the outset that "patent infringement and invalidity are separate and distinct issues." Pandrol USA, LP v. Airboss Ry. Prod., Inc., 320 F.3d 1354, 1364-65 (Fed.Cir.2003). Indeed, this court routinely orders a partial new trial on infringement, while upholding an earlier verdict on validity. See, e.g., Cardiac Pacemakers, Inc. v. St. Jude Med., Inc., 381 F.3d 1371, 1374 (Fed.Cir.2004); Comaper Corp. v. Antec, Inc., 596 F.3d 1343, 1354-55 (Fed.Cir.2010).

We previously rejected the "argument that, under the Seventh Amendment, a new trial on willfulness would require a new trial on infringement." Voda v. Cordis Corp., 536 F.3d 1311, 1329 (Fed.Cir. 2008). In order to prove that infringement was willful, a plaintiff must show both that "an infringer acted despite an objectively high likelihood that its actions constituted infringement of a valid patent" and that the "objectively-defined risk (determined by the record developed in the infringement proceeding) was either known or so obvious that it should have been known to the accused infringer." In re Seagate Tech., LLC, 497 F.3d 1360, 1371 (Fed.Cir. 2007). "[T]he objective prong of Seagate tends not to be met where an accused infringer relies on a reasonable defense to a charge of infringement. Thus, the question on appeal often posed is whether a defense or non-infringement theory was reasonable." Bard Peripheral Vascular, Inc. v. W.L. Gore & Assocs., Inc., 682 F.3d 1003, 1005-06 (Fed.Cir.2012) cert. denied, ___ U.S. ___, 133 S.Ct. 932, 184 L.Ed.2d 752 (2013). In such a case, in order to find that infringement was not willful, the defendant's non-infringement theory must be reasonable. The question of a non-infringement theory's reasonableness often requires looking at the merits of noninfringement. Yet, a new trial on willfulness does not require a new trial on infringement. See Voda, 536 F.3d at 1329. We believe the situation in the present case to be analogous.

We acknowledge that the current case presents a unique situation where a jury considering induced infringement, but not validity, may be asked to consider evidence of invalidity in order to decide whether Cisco possessed a good-faith belief of invalidity. Nonetheless, the fact that a second jury will consider evidence of invalidity that supports Cisco's position on the good-faith belief issue does not compel the conclusion that the second jury will decide the ultimate issue of invalidity. Indeed, the issue of whether Cisco possessed a good-faith belief of invalidity is distinct and separate from the issue of whether the patent claims are invalid. In order to determine that Cisco had a good-faith belief of invalidity, the jury must merely decide whether Cisco possessed that belief in good-faith. The jury need not decide whether the underlying position was meritorious. Thus, although the two juries will review the same evidence of invalidity, they will not decide the same essential issues. Therefore, we cannot say that separate trials on invalidity and induced infringement would constitute a clear and indisputable infringement of the constitutional right to a fair trial. Accordingly, we find that holding separate trials on the issues of invalidity and induced infringement does not violate the Seventh Amendment.

D. CLAIM CONSTRUCTION AND THE MERITS

Cisco challenges the district court's construction of the term "short-range communication protocol." The court construed this term to mean "a set of procedures required to initiate and maintain short-range communication between two or more devices." Commil USA, LLC v. Cisco Sys., Inc., 2:07-CV-341, slip op. at 1 (E.D.Tex. Nov. 6, 2009). Cisco's argument is without merit. Cisco does not contend that a "short-range communication protocol" is not "a set of procedures required to initiate and maintain short-range communication between two or more devices." Rather, Cisco asks this court, as it did the district court, to limit the term to only those specific short-range communication protocols listed in the patent. We decline. See Phillips v. AWH Corp., 415 F.3d 1303, 1323-24 (Fed.Cir.2005) (cautioning against importing limitations from the specification into the claims).

Cisco also appeals the district court's findings regarding validity, infringement, and damages. Cisco argues that the claims are invalid for reasons of indefiniteness, non-enablement, and lack of written description. We find these contentions without merit. Because we remand for a new trial, we do not reach the issues of infringement and damages. Rather, we leave them to be decided by the district court in first instance.

AFFIRM-IN-PART, VACATE-IN-PART, AND REMAND.

NEWMAN, Circuit Judge, concurring in part, dissenting in part.

I agree that remand is appropriate, and I agree that in this case a partial retrial was within the district court's discretion.

However, I respectfully dissent from the change of law set forth in Part II.B of the court's opinion. The court holds that if the inducer of infringement believes in good faith that the patent is invalid, there can be no liability for induced infringement, although the patent is held valid. The opinion makes clear that the court intends to adjust the law, as in statements including:

Maj. op. at 1368-69 (emphasis added). This change in the law of induced infringement is inappropriate.

A good-faith belief of patent invalidity may be raised as a defense to willfulness of the infringement, but it is not a defense to the fact of infringement. Patent invalidity, if proved, eliminates an invalid patent and thus is a total defense to infringement. However, a "good-faith belief" in invalidity does not avoid liability for infringement when the patent is valid.1 No rule eliminates infringement of a valid patent, whether the infringement is direct or indirect.

The "inducement" statute, 35 U.S.C. § 271(b), serves a different purpose. The inducement statute is designed to allow remedy against an entity that provides an infringing product or method to direct infringers, but is not itself a direct infringer. The inducement statute does not import a validity criterion, or a "good faith belief" about validity, into proof of the act of infringement. See Akamai Techs., Inc. v. Limelight Networks, Inc., 692 F.3d 1301, 1308 n. 1 (Fed.Cir.2012) (en banc) ("Because liability for inducement, unlike liability for direct infringement, requires specific intent to cause infringement, using inducement to reach joint infringement does not present the risk of extending liability to persons who may be unaware of the existence of a patent or even unaware that others are practicing some of the steps claimed in the patent.").

The majority's view that a belief in invalidity can negate infringement is contrary to the principles of tort liability, codified in the inducement statute. Liability for induced infringement is akin to "liability... under a theory of joint tortfeasance, wherein one who intentionally caused, or aided and abetted, the commission of a tort by another was jointly and severally liable with the primary tortfeasor." Hewlett-Packard Co. v. Bausch & Lomb Inc., 909 F.2d 1464, 1469 (Fed.Cir.1990); see also Global-Tech Appliances, Inc. v. SEB S.A., ___ U.S. ___, 131 S.Ct. 2060, 2067, 179 L.Ed.2d 1167 (2011) (recognizing that § 271(b) codified pre-1952 case law, wherein induced infringement "was treated as evidence of `contributory infringement,' that is, the aiding and abetting of direct infringement by another party"); H.R.Rep. No. 82-1923, at 9 (1952) (explaining that the new subsection (b) "recites in broad terms that one who aids and abets an infringement is likewise an infringer.").

A mistake of law, even if made in good faith, does not absolve a tortfeasor. "Our law is ... no stranger to the possibility that an act may be `intentional' for purposes of civil liability, even if the actor lacked actual knowledge that her conduct violated the law." Jerman v. Carlisle, McNellie, Rini, Kramer & Ulrich LPA, 559 U.S. 573, 130 S.Ct. 1605, 1612, 176 L.Ed.2d 519 (2010). "If one intentionally interferes with the interests of others, he is often subject to liability notwithstanding the invasion was made under an erroneous belief as to some legal matter that would have justified the conduct." Id. (quoting W. Keeton, D. Dobbs, R. Keeton & D. Owen, Prosser and Keeton on Law of Torts 110 (5th ed.1984)). A trespass "can be committed despite the actor's mistaken belief that she has a legal right to enter the property." Id. (citing Restatement (Second) of Torts § 164, & cmt. e (1963-1964)). "[P]atent validity is a question of law," CSIRO v. Buffalo Technology (USA), Inc., 542 F.3d 1363, 1375 (Fed.Cir.2008), and an "erroneous belief" of the "legal matter" of validity does not excuse the violation. Jerman, 130 S.Ct. at 1612.

A defendant's ultimate liability for induced infringement, as for direct infringement, is subject to various defenses including patent invalidity and unenforceability. However, whether there is infringement in fact does not depend on the belief of the accused infringer that it might succeed in invalidating the patent. Such a belief, even if held in good faith, does not negate infringement of a valid and enforceable patent. This rule applies, whether the infringement is direct or indirect. My colleagues err in holding that "evidence of an accused inducer's good-faith belief of invalidity may negate the requisite intent for induced infringement." Maj. op. at 1368.

The Court stated in Global-Tech that "induced infringement under section 271(b) requires knowledge that the induced acts constitute patent infringement." 131 S.Ct. at 2068. The Court did not include a belief in patent validity as a criterion of infringement. Global-Tech does not hold that if the inducer "believed" that the patent is invalid, the inducer avoids infringement when, as here, validity is sustained.

Validity of the Commil patent was sustained by the jury, sustained by the district court, and sustained by this court. Whatever Cisco's "belief" as to invalidity of the patent, this belief is irrelevant to the fact and law of infringement. A belief of invalidity cannot avoid liability for infringement of a patent whose validity is sustained. The panel majority's contrary holding is devoid of support in law and precedent.

The district court applied the correct law, and excluded the issue of validity from its retrial of the issue of infringement. My colleagues now hold that although validity was found and sustained at trial, "the district court erred in preventing Cisco from presenting evidence during the second trial of its good-faith belief in invalidity to rebut Commil's allegations of induced infringement." Maj. op. at 1367. The court mis-cites the Global-Tech ruling concerning the inducer's knowledge "that the induced acts constitute patent infringement," maj. op. at 1369, for Global-Tech relates to knowledge of infringement, not knowledge of validity.

The fact of infringement does not depend on whether the inducer's view of patent validity is held in good faith or bad faith. Validity and infringement are distinct issues, bearing different burdens, different presumptions, and different evidence. Although the court now acknowledges that "patent infringement and invalidity are separate and distinct issues," maj. op. at 1371, the court holds that on this third infringement trial the jury "may be asked to consider evidence of invalidity." Maj. op. at 1372. If the jury is required to consider evidence of invalidity, as the court holds, it strains fairness to deny Cisco's request for redetermination of the issue of validity.

I respectfully dissent from the court's incorrect statement of the law of induced infringement, and from the holding that a showing of a good faith belief in patent invalidity can avoid all liability for induced infringement of a valid patent.

O'MALLEY, Circuit Judge, concurring in part and dissenting in part.

I agree in large measure with the majority's thoughtful opinion in this case. First, I agree that the induced infringement judgment and award must be vacated. Like Judge Prost, I believe this is so both because the trial court instructed the jury to apply an incorrect legal standard during its deliberations and because the court erred in excluding evidence regarding Cisco's alleged good faith belief in the invalidity of the asserted claims of the `395 patent. On the latter point, I agree that an accused inducer's good faith belief of invalidity of a patent claim is relevant to its intent to induce infringement of that claim and is, thus, admissible for that purpose.

I also agree that Cisco's objections to the trial court's construction of the claim term "short range communication protocol" are not well taken. Contrary to Cisco's contentions, the `395 patent's written description does not limit the short-range communication protocol to Bluetooth and related protocols. Instead, the patent consistently refers to such protocols as exemplary. See, e.g., `395 patent at col. 18, ll. 23-24 ("Bluetooth wireless technology is an example of such a short-range communication protocol."); col. 8, ll. 41-46. We have expressly stated that we do not limit the scope of the claims to the preferred embodiment described in the written description. See Phillips v. AWH Corp., 415 F.3d 1303, 1323 (Fed.Cir.2005) ("[W]e have expressly rejected the contention that if a patent describes a single embodiment, the claims of the patent must be construed as being limited to that embodiment.").

Finally, I agree that the trial judge acted within his discretion in granting a new trial following the first trial involving these parties and these patent claims. Cisco does not deny that its local counsel's conduct was reprehensible, nor does it debate whether some curative action was appropriate. Cisco simply asks that we substitute our judgment for that of the trial court with respect to whether that curative action should have been a new trial. I agree with the majority that it is appropriate to defer to the trial court's first-hand assessment of whether counsel's conduct was sufficiently improper as to call into question the integrity of the jury's verdict. Accordingly, while it may be true that a different jurist might have refused to set aside the jury's verdict in these circumstances, the decision to do so here should not be reversed.1

Having set out those places where I agree with the majority's resolution of this appeal, I turn to the two decisions over which I must part company with my colleagues. First, I cannot endorse the majority's refusal to address Cisco's potentially dispositive arguments regarding whether Commil did or ever could prove the third-party direct infringement which is a necessary predicate to Commil's induced infringement claim. Next, I disagree with the majority's conclusion that the trial court's decision to order only a partial retrial of the issues presented is defensible; I believe the partial retrial order deprived Cisco of its right to a jury trial as guaranteed by the Seventh Amendment to the United States Constitution and must be reversed. I believe, therefore, that we should not address the strength of Cisco's validity arguments based on the record from the first trial and should leave those questions to the third jury to visit this matter.

I address these disagreements in turn.

I

Cisco argues that the trial court erred in denying its request for judgment as a matter of law on Commil's induced infringement claim because Commil failed to prove — and allegedly cannot prove — that any third-party practices all of the steps of the method claimed in claim 1 of the `395 patent. Specifically, Cisco asserts that its customers do not perform (1) the step of "dividing" the communications protocol into two smaller protocols or, (2) the step of "running an instance" of the low-level protocol at the base station and "running an instance" of the high-level protocol at the switch. Cisco asserts that only it — and not its customers — practices the "dividing" step because the WiFi protocol is divided into high- and low-level protocols at the factory when the products are created. Because of this, Cisco asserts that its customers cannot perform this step and, thus, do not directly infringe claim 1. With respect to the "running an instance step," Cisco asserts that, in its WiFi system, the access points run only a single copy of the protocol that communicates with all of the connected devices, and do not run separate copies of the low-level protocols with respect to each mobile unit or device to which the access point is connected. If one or both of these steps of the claimed method are not, in fact, practiced by its customers, Cisco is correct that Commil's claims of induced infringement — predicated as they are on claims of single party direct infringement — must fail.2

On appeal, Cisco asks that we reverse the trial court's denial of its motion for judgment as a matter of law on these grounds and enter judgment in its favor. It contends that we need not reach the propriety of the trial court's inducement instructions or evidentiary rulings, or the need for a new trial based on errors therein, because Commil failed to prove the direct infringement predicate for its induced infringement claim. Cisco contends that these questions are dispositive of the induced infringement claims and that the need for a new trial could be obviated by their resolution.

Whatever the merits of Cisco's argument regarding the direct infringement aspect of Commil's induced infringement claim, Cisco is correct that it is within the scope of our authority on this record to resolve them now. Indeed, I believe it is our obligation to do so. If we send this matter back for a new trial on induced infringement without resolving these issues, we likely will see the case return in much the same posture. If a new panel ultimately concludes Cisco is correct as to either one of the required steps of claim 1 — and that no finding of induced infringement can stand on that ground — we will have forced the parties and the trial court to go through a new trial when none was necessary.

I do not purport to prejudge Cisco's arguments; Cisco may be wrong on both points.3 I urge us, however, to judge them one way or another. We do no one any favors by kicking these potentially dispositive cans down the road and may well be requiring undue expense and wasting scarce judicial resources in the process. I believe our appellate function requires that we avoid such inefficiency whenever possible. For these reasons, I dissent from the majority's refusal to resolve these properly preserved issues on appeal.

II

I turn next to the majority's conclusion that the partial new trial order entered by the district court in this case — one which we now effectively reinstate — did not violate the Seventh Amendment. I do not agree. Accepting the proposition, as I have, that the district court acted within its discretion to find a new trial warranted in this case, I believe that nothing other than a full retrial on all issues can be justified under the law.4

While partial retrials are permissible in appropriate circumstances, the Supreme Court has set forth a strict standard for determining when such circumstances exist. In Gasoline Prods. Co. v. Champlin Refining Co., 283 U.S. 494, 500, 51 S.Ct. 513, 75 L.Ed. 1188 (1931), the Supreme Court explained that a court's authority to order a new trial is constrained by the Seventh Amendment such that "a partial new trial ... may not properly be resorted to unless it clearly appears that the issue to be retried is so distinct and separable from the others that a trial of it alone may be had without injustice." Applying that standard to the case before it, the Supreme Court concluded that a partial retrial on damages alone would violate the Seventh Amendment because the facts and issues relating to the merits of the contract action were not sufficiently separable from those relating to damages. Gasoline Prods., 283 U.S. at 500, 51 S.Ct. 513; see also Witco Chemical Corp. v. Peachtree Doors, Inc., 787 F.2d 1545, 1549 (Fed.Cir. 1986) (quoting standard from Gasoline Products and finding that "it is inappropriate... to have one jury return a verdict on the validity, enforceability, and contract questions while leaving the infringement questions to a second jury.").

The Fifth Circuit — the regional circuit from which this case arises and whose law we are to apply to this non-patent basic right — repeatedly has cautioned against resort to partial retrials, citing to the guidance from Gasoline Products. See, e.g., Nissho-Iwai Co. v. Occidental Crude Sales, Inc., 729 F.2d 1530, 1539 (5th Cir. 1984) (approving refusal to order partial retrial of fraud claim alone because "the fraud claim arose out of the acts surrounding the breach of contract [claim]" and understanding of one required understanding of the other); Davis v. Safeway Stores, 532 F.2d 489, 491 n. 3 (5th Cir.1976) (noting that granting of a partial new trial over particular issues requires those issues to be clearly separable from other issues in the case); Vidrine v. Kansas City S. Ry. Co., 466 F.2d 1217, 1221 (5th Cir.1972) (observing that a partial retrial is only appropriate where issues are "so distinct and separable" that there will be "no injustice or prejudice" to either party); Williams v. Slade, 431 F.2d 605, 609-10 (5th Cir.1970) (noting that if a verdict is a "product of passion or prejudice" a new trial on all issues must be ordered).

The requirement that issues in multiple trials be separable and distinct protects parties' rights under the Seventh Amendment by guarding against circumstances which threaten those rights. See, e.g., Pryer v. C.O. 3 Slavic, 251 F.3d 448, 454-58 (3d Cir.2001) (collecting cases disapproving of partial retrials and outlining factors counseling against such partial retrials). One such circumstance is where the issues relating to the separated claims overlap, causing the potential for "confusion" or "uncertainty" when one issue is submitted to the jury without the other. Id.; See also, Nissho-Iwai Co., 729 F.2d at 1539 (where claims arise from same transactional facts, jury cannot understand one fully without understanding the other and having both presented jointly); FIGA v. R.V.M.P. Corp., 874 F.2d 1528, 1534 (11th Cir.1989) (partial retrial just on damages is not appropriate where damages evidence was not fully separable from evidence of insured's alleged intent to cause fire); United States ex rel. Miller v. Bill Harbert Int'l Constr., Inc., 865 F.Supp.2d 1, 10 (D.D.C.2011) (partial retrial inappropriate because jury needs thorough knowledge of underlying conspiracy in order to understand and assess whether particular defendants joined that conspiracy).

Partial retrials must also be avoided where it is possible that the very error that is deemed to warrant a new trial may have impacted the jury's determination of other issues. Pryer, 251 F.3d at 455 (3d Cir.2001) (partial retrial is inappropriate whenever it is not clear that the error that crept into one element of the verdict "did not in any way affect the determination of any other") (additional citations omitted). This is especially so where a retrial is prompted by a finding that comments of counsel may have unduly "inflamed" the jury because such a finding "implies that the jury made its decision on an improper basis." United States ex rel. Miller, 865 F.Supp.2d at 10 (D.D.C.2011) (finding partial retrial inappropriate where new trial was based on prosecutor's improper comments on one issue because that error "might well have affected the jury's determination of other issues"). And, partial retrials should be avoided whenever circumstances indicate "there is reason to think that the verdict may represent a compromise among jurors with different views on whether defendant was liable." Pryer, 251 F.3d at 455 (additional citations omitted); See also Stanton v. Astra Pharm. Products, Inc., 718 F.2d 553, 576 (3d Cir.1983) (citing Vizzini v. Ford Motor Co., 569 F.2d 754, 759-60 (3d Cir.1977)) (finding that suggestion of possible compromise verdict was such that permitting retrial of damages alone, absent reconsideration of liability issues was inappropriate, noting "[i]s difficult to say that `allowing a second jury to determine the issue of damages in isolation from the whole of the circumstances surrounding the case was not an injustice...'"). All of these circumstances are presented here.

As the majority explained, the trial court ordered a retrial based on statements by Cisco's local counsel which the trial court believed were so insidious as to call into question the integrity of the jury's verdict. See Commil USA, LLC v. Cisco Sys., Inc., No. 2:07-cv-341 slip. op. at 3-4 (E.D. Tex. Dec. 29, 2010). If the trial court believed the verdict truly was compromised, how could he — and how can we — assume the misconduct infected only a portion of their deliberations? Indeed, it could be that the jury was so incensed by its counsel's conduct that they held it against Cisco by refusing to invalidate Commil's patent, despite a contrary view of the evidence.

Once the partial retrial began, moreover, the trial court's evidentiary rulings themselves reflect the awkward posture in which he had placed the case. Most pointedly, as the majority discusses, the trial court excluded — incorrectly — evidence of Cisco's alleged good faith belief in the invalidity of the claims of the `395 patent. What the majority fails to mention, however, is that Commil itself expressly argued that it would unduly confuse the jury to admit such evidence without also submitting the validity determination to it to decide. See Commil USA, LLC v. Cisco Sys., Inc., No. 2:07-cv-341, Doc. No. 398 at 4-5 (April 1, 2011). And, it was in response to this argument that the evidence was excluded. Where the court closest to these matters saw the potential for confusion because of the interwoven nature of the invalidity claims and Cisco's good faith defense to induced infringement, how can we ignore that potential when we now order the excluded evidence to be admitted? I do not believe we can. I do not believe we can differentiate the circumstances here from the overlapping nature of the issues in Gasoline Products and the host of cases cited above finding a partial retrial improper based on the principles described therein.

The practical implications of the judgment we render today highlight the Seventh Amendment problems we create thereby. Judge Prost and I agree not only that it was error to exclude proffered evidence of Cisco's good faith belief in the invalidity of the claims of the '395 patent, but that the error was not harmless. In other words, we find that Cisco was denied the right to a fair trial on Commil's induced infringement claim because it was denied the opportunity to pursue a valid defense. In the same breath, however, Judges Prost and Newman conclude it is appropriate to retry the case in a posture that would dilute that defense.

When this case returns to Texas for a third trial, the trial court will need to craft instructions that tell the jury that, while Cisco claims it had a good faith belief in the invalidity of the claims of the '395 patent, Cisco was wrong. The jury will need to be told that it is not permitted to conclude it agrees with Cisco's belief. The jury will, thus, begin its deliberations already suspect about Cisco's beliefs and the good faith nature of the same. It is precisely these circumstances against which the Supreme Court insists we guard. Importantly, given the significance of the Seventh Amendment guarantees it is our job to protect, we are not to ask whether it is conceivable that a jury could fairly assess Cisco's case in these circumstances; we are to assume that, where it is not clear that "the issue to be retried is so distinct and separable from the others," it cannot. See, e.g., Gasoline Prods. Co., 283 U.S. at 500, 51 S.Ct. 513; Witco Chemical Corp., 787 F.2d at 1549

Finally, I do not believe we can discount the possibility that the first verdict may have represented a series of compromises by the jury. How can we know that the jury did not agree not to invalidate the claims of the `395 patent only because it found no induced infringement and understood that its direct infringement finding carried with it a smaller damages award? We cannot.

I do not contend that issues in patent cases can never be tried to separate juries, particularly after an appeal reveals that only one issue was adjudicated erroneously. Whether and when a new trial on all issues is required must be determined "only after considering the totality of the circumstances and by answering: `How may the ends of justice be served?'" Witco Chemical Corp., 787 F.2d at 1549 (citations omitted). Here, all circumstances indicate that a partial retrial of Commil's induced infringement claim without retrial of the validity issues is not appropriate.5

While Cisco's counsel's conduct was reprehensible and warranted curative action, action which compromises Cisco's Seventh Amendment right to a jury goes too far. I believe the trial court abused its discretion by only ordering a partial retrial of the claims asserted in this case and that we perpetuate that error by ordering yet another partial retrial.