Opinion for the court filed by Circuit Judge LOURIE. Opinion concurring in part and dissenting in part filed by Circuit Judge NEWMAN.

Ericsson, Inc. and Telefonaktiebolaget LM Ericsson (collectively, “Ericsson”) appeal from the decision of the United States District Court for the Eastern District of Texas granting judgment as a matter of law (“JMOL”) that Harris Corporation and Intersil Corporation (collectively, “Harris”) do not infringe Ericsson’s U.S. Patent 4,961,222. Ericsson, Inc. v. Harris Corp., No. 4:98cv325, 2002 WL 32114135 (E.D.Tex. July 11, 2002) (“JMOL Order”). Harris conditionally cross-appeals from the district court’s denial of its motion for JMOL relating to damages. Ericsson, Inc. v. Harris Corp., No. 4:98cv325 (E.D.Tex. July 30, 2002) (“Final Judgment ”). For the reasons stated below, we reverse the grant of JMOL of noninfringement and affirm the denial of JMOL relating to damages.

BACKGROUND

Ericsson owns the '222 patent, which is directed to an apparatus for supplying power to a telephone set in a telecommunications system. The patent discloses a subscriber line interface circuit (“SLIC”), which acts as an interface between a telephone exchange and individual subscriber telephone sets. The claimed invention is designed to reduce the idling power that is dissipated by the amplifiers that transmit speech signals across the subscriber line. The device thus conserves power by switching between an active mode, during which the relatively large speech signal amplifiers supply power to the telephone set when it is in use, and a low-power standby mode, during which the smaller auxiliary amplifiers supply power to the telephone set when it is not in use. Ac*1372cording to claim 1, a loop sensing circuit detects whether the telephone receiver is in its lifted (“off-hook”) position or its cradled (“on-hook”) position. The loop sensing circuit then sends a corresponding signal to a control circuit, which in turn sends a control signal to both the speech signal amplifiers and the auxiliary amplifiers. When the loop sensing circuit determines that the telephone receiver is in its off-hook position, the control signal “enables the speech signal amplifiers and disables the auxiliary amplifiers.” '222 patent, col. 4,11. 27-28. Conversely, when the loop sensing circuit determines that the telephone receiver is in its on-hook position, the control signal “disables the speech signals [sic] amplifiers and enables the auxiliary amplifiers so that the speech signal amplifiers, which require poiuer, only supply power to the telephone set when the receiver is off its cradle and a call can be made." Id. at col. 4, 11. 33-37 (emphasis added). Claim 2 depends from claim 1.

Harris manufactures and sells three accused devices: the 5513 SLIC, the 5514 SLIC, and the 5518 SLIC. Each of the accused devices has a low-power standby mode.

In November 1998, Ericsson filed suit against Harris for infringement of claims 1 and 2 of the '222 patent. During the claim construction phase of litigation, both parties agreed that the claim language requiring that the speech signal amplifiers “only supply power to the telephone set when the receiver is off its cradle and a call can be made,” id. at col. 4,11. 35-37 (the “ ‘only supply power’ limitation”), should be given its ordinary meaning. JMOL Order, slip op. at 6, 2002 WL 32114135. After a thirteen-day trial, a jury found that Harris’s three accused devices did not literally infringe the '222 patent but did infringe claims 1 and 2 under the doctrine of equivalents. Id. at 2, 2002 WL 32114135. The jury awarded Ericsson damages in the amount of $3.5 million for lost profits due to lost sales; $645,000 for lost profits due to price erosion; and $136,000 as a reasonable royalty.

The district court, however, granted Harris’s motion for JMOL of noninfringement. After reviewing the testimony of several witnesses, the court found that the “uncontroverted evidence” showed that the speech signal amplifiers in the accused devices supply “some power” to the telephone set in the on-hook position. Id. at 21-22, 2002 WL 32114135. In particular, the court cited the testimony of several of Harris’s witnesses, who testified that the speech signal amplifiers in the accused devices have three transistors (the “QRA 23-25 transistors”) that always supply a small amount of power to the subscriber line in order to prevent corrosion and also supply power to enable on-hook functions such as caller-ID. Id. at 15, 2002 WL 32114135. In addition, the court found that Ericsson’s expert witness,- Dr. Thomas Rhyne, acknowledged that the accused devices supply power to the telephone set in the on-hook position, testifying only that the power supplied to prevent corrosion is an insubstantial amount and that any power supplied for on-hook transmission would be present for only a few seconds at a time. Id. at 16, 2002 WL 32114135. However, because claim 1 requires that the speech signal amplifiers only supply power to the telephone set in the off-hook position, the court concluded that a determination that the scope of equivalence included the accused devices would vitiate the “only supply power” limitation. Id. at 22, 2002 WL 32114135. The court accordingly found no infringement as a matter of law and granted Harris’s motion for JMOL of noninfringement. Id. The court then dismissed the case with prejudice and summarily denied Harris’s motion for JMOL *1373relating to the amount of the damages award. Final Judgment, slip op. at 1.

Ericsson timely appealed to this court. We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(1).

DISCUSSION

We review a district court’s grant or denial of a motion for JMOL in a patent case de novo, reapplying the JMOL standard used by the district court. Sextant Avionique, S.A. v. Analog Devices, Inc., 172 F.3d 817, 824 (Fed.Cir. 1999). JMOL is appropriate when “a party has been fully heard on an issue and there is no legally sufficient evidentiary basis for a reasonable jury to find for that party on that issue.” Fed.R.Civ.P. 50(a)(1). When reviewing a district court’s denial of JMOL, we must “determine whether ‘viewing the evidence in the light most favorable to the non-moving party,’ and giving the non-movant ‘the benefit of all reasonable inferences,’ there is sufficient evidence of record to support a jury verdict in favor of the non-movant.” Southwest Software, Inc. v. Harlequin Inc., 226 F.3d 1280, 1289 (Fed.Cir. 2000) (citation omitted).

A determination of infringement requires a two-step analysis. “First, the court determines the scope and meaning of the patent claims asserted ... [and second,] the properly construed claims are compared to the allegedly infringing device.” Cybor Corp. v. FAS Techs., Inc., 138 F.3d 1448, 1454 (Fed.Cir. 1998) (en banc) (citations omitted). Step one, claim construction, is an issue of law, Markman v. Westview Instruments, Inc., 52 F.3d 967, 970-71 (Fed.Cir. 1995) (en banc), aff'd, 517 U.S. 370, 116 S.Ct. 1384, 134 L.Ed.2d 577 (1996), that we review de novo, Cybor, 138 F.3d at 1456. Step two, comparison of the claim to the accused device, requires a determination that every claim limitation or its equivalent be found in the accused device. Warner-Jenkinson Co. v. Hilton Davis Chem. Co., 520 U.S. 17, 29, 117 S.Ct. 1040, 137 L.Ed.2d 146 (1997). Those determinations are questions of fact. Bai v. L & L Wings Inc., 160 F.3d 1350, 1353 (Fed.Cir. 1998).

The determination of the amount of actual damages is also a question of fact. Brooktree Corp. v. Advanced Micro Devices, Inc., 977 F.2d 1555, 1578 (Fed.Cir. 1992). When reviewing a district court’s denial of JMOL, we apply the substantial evidence standard to the jury’s damages award. Micro Chem., Inc. v. Lextron, Inc., 317 F.3d 1387, 1394 (Fed.Cir. 2003).

A. Infringement

On appeal, Ericsson argues that the district court erred in granting Harris’s motion for JMOL of noninfringement. Ericsson first contends that the jury reasonably could have found that the “only supply power” limitation was met literally (and that some other limitation was met equivalently) in light of the substantial evidence that it presented to show that the speech signal amplifiers in the accused devices do not supply power to the telephone set when the receiver is in the on-hook position. Ericsson also asserts that the court erred in holding that the “only supply power” limitation is entitled to no scope of equivalence, arguing instead that it is a temporal or range limitation and is not vitiated by a SLIC that is in standby mode 99.9% of the time. In addition, Ericsson maintains that it offered substantial evidence of equivalence by showing that both caller-ID and the “corrosion current” present only insubstantial differences from the claimed invention. Finally, Ericsson argues that the fact that the accused devices can optionally be used for caller-ID does not negate infringement.

Harris responds that the district court did not err in entering JMOL of nonin-*1374fringement. Harris first argues that Ericsson failed to provide substantial evidence that the accused devices literally satisfy the “only supply power” limitation. Harris also contends that the court correctly construed the “only supply power” limitation to exclude the accused devices because a contrary finding would vitiate that limitation. According to Harris, the speech signal amplifiers in the accused devices always supply power to the phone when it is on-hook, both through the QRA 23-25 transistors and when engaging in on-hook transmission for caller-ID.

We agree with Ericsson that the district court erred in granting Harris’s motion for JMOL of noninfringement. As discussed above, claim 1 of the '222 patent requires that “the speech signal amplifiers, which require power, only supply power to the telephone set when the receiver is off its cradle and a call can be made.” '222 patent, col. 4, 11. 34-37 (emphasis added). Focusing on the word “only,” the district court held that there could be no infringement under the doctrine of equivalents because there was “uncontroverted evidence” that the speech signal amplifiers in the accused devices supply “some power” to the telephone set in the on-hook position. JMOL Order, slip op. at 21-22, 2002 WL 32114135. We agree that the “only supply power” limitation would indeed be vitiated by an accused device whose speech signal amplifiers always supply power to the telephone set when the receiver is on-hook. See Moore U.S.A., Inc. v. Standard Register Co., 229 F.3d 1091, 1106 (Fed.Cir. 2000) (holding that a “minority” cannot be equivalent to a “majority,” its very antithesis). However, for the reasons set forth below, we disagree with the factual premise that the speech signal amplifiers in the Harris devices necessarily behave in that manner. Instead, we conclude that substantial evidence supports the jury’s finding that the accused devices infringe the '222 patent under the doctrine of equivalents and that that finding does not vitiate the “only supply power” limitation.1

To begin with, Ericsson presented substantial evidence to show that the power supplied by the QRA 23-25 transistors in the accused devices does not vitiate the “only supply power” limitation. The “only supply power” limitation refers only to power supplied by the speech signal amplifiers. Although Harris argued that the QRA 23-25 transistors in the accused devices are part of the speech signal amplification circuitry, Ericsson argued that they are instead part of the control circuitry, which merely switches the speech signal amplifiers on and off. To support its position, Ericsson presented the following evidence at trial: expert testimony stating that the QRA 23-25 transistors are part of the control circuitry that enables and disables the speech signal amplifiers,2 a Harris engineer’s notes labeling the disputed transistors as “shutoff transistors,” and Harris data sheets describing the speech signal amplifiers as “disabled” and “powered down” in the standby power state. We conclude that that evidence was sufficient to support a reasonable jury’s finding that any power supplied by the QRA 23-25 transistors in the accused devices is not supplied by the speech signal amplifiers *1375and therefore does not vitiate the “only supply power” limitation.

Nor does the fact that the accused devices could be used in conjunction with on-hook functions such as caller-ID vitiate the “only supply power” limitation. Ericsson points to evidence, which Harris does not dispute, to show that the accused devices switch into active mode when the phone is on-hook for just four seconds during the caller-ID function. That brief change of mode occurs only when a subscriber who has enabled the caller-ID function receives a call, and according to Ericsson’s expert, amounts to less than 0.1% of the time. Such a trivial difference does not vitiate the “only supply power” limitation because, even under the caller-ID scenario, the speech signal amplifiers in the accused devices are very close to only supplying power to the telephone set when the receiver is in the off-hook position. Thus, the jury reasonably could have found that, even when used with caller-ID, the accused devices are insubstantially different from the claimed invention and consequently infringe under the doctrine of equivalents.3

Despite the dissent’s contrary view, Fes-to is not a part of this case because prosecution history estoppel, even if applicable, would not have precluded the jury’s verdict of infringement under the doctrine of equivalents. The jury might still have found that claim 1 was infringed equivalently, even while finding that the amended limitations, of concern to the dissent, were met literally. Under the theory that we attribute to the jury, the QRA 23-25 transistors are not part of the speech signal amplifiers and the accused speech signal amplifiers are thus literally “disabled” when the telephone receiver is on-hook. The only equivalence issue thus concerns when the speech signal amplifiers supply power, not whether they are “enabled” or “disabled” as opposed to “actively connected” or “effectively disconnected,” which was the subject of the amendment. In other words, the equivalence question relates to whether “the speech signal amplifiers ... only supply power to the telephone set” when the receiver is off-hook. That limitation was never amended and therefore cannot be subject to the Festo presumption. The district court even agreed that Festo was inapplicable when it granted JMOL of noninfringement on the basis of the “all limitations rule,” and not prosecution history estoppel, as the dissent states. In fact, the district court never even mentioned the amendments to the “enabled” and “disabled” limitations, but specifically noted that the disputed limitation was not amended during prosecution. JMOL Order, slip op. at 8, 2002 WL 32114135.4

We therefore conclude that the district court erred in determining that a finding of infringement would vitiate the “only supply power” limitation. Moreover, we conclude that substantial evidence supports the jury’s verdict of infringement under the doctrine of equivalents. Accordingly, we reverse the district court’s grant of JMOL of noninfringement.

*1376B. Damages

Having reversed the JMOL of nonin-fringement, we turn next to the issue of damages. Harris conditionally cross-appeals from the court’s denial of its motion for JMOL relating to the amount of the damages award for lost profits due to lost sales and price erosion. Harris does not, however, contest the jury’s award of $136,000 as a reasonable royalty.

1. Jurisdiction

As a preliminary matter, Ericsson challenges the propriety of Harris’s cross-appeal. Citing Bailey v. Dart Container Corp., 292 F.3d 1360 (Fed.Cir. 2002) (order), Ericsson argues that Harris’s cross-appeal should be dismissed because Harris does not seek to expand its rights under the district court’s judgment. Ericsson is mistaken. A cross-appeal is appropriate “when a party seeks to enlarge its own rights under the judgment or to lessen the rights of its adversary under the judgment.” Id. at 1362. In other words, a party may file a cross-appeal “when acceptance of the argument it wishes to advance would result in a reversal or modification of the judgment rather than an affirmance.” Id. That is precisely the situation before us: Harris requests that we reverse the district court’s denial of its motion for JMOL and reduce the amount of the damages award entered against it. Harris thus seeks to lessen Ericsson’s rights under the district court’s judgment by reducing Ericsson’s monetary recovery. Harris’s cross-appeal, although conditional, is therefore properly before us.

2. Lost Profits Due to Lost Sales

On the merits of the cross-appeal, Harris objects to the jury’s award of $3.5 million for lost profits due to lost sales. First, Harris argues that Ericsson failed to provide “sound economic proof’ of its market definitions and market share estimates. Harris contends that Ericsson’s damages expert, an accountant named Daniel Jackson, improperly based his definition of the so-called “Harris market” on one witness’s testimony, rather than on market research and recognized economic principles such as the cross-elasticity of demand between the infringing product and available substitutes. With respect to the narrower “Ericsson market,” Harris maintains that Mr. Jackson again failed to calculate the cross-elasticities of demand, improperly assumed that switching costs were so high as to prevent customers from purchasing SLICs that were not pin-compatible with their existing line cards, and ignored that the SLIC industry is characterized by “second-sourcing.” Second, Harris argues that Jackson arbitrarily allocated Harris’s sales between the “Harris market” and the “Ericsson market,” neglecting to consider that many of Ericsson’s customers purchased from Harris as a “second source” to Ericsson and that customers might redesign their line cards. Finally, Harris asserts that Ericsson failed to establish the “but for” causation necessary to recover damages for lost profits because Jackson’s analysis failed to take into account available low-power standby SLICs that were noninfringing substitutes for those covered by the '222 patent.

Ericsson responds that Jackson used an approved methodology for determining lost profits, that set forth in Panduit Corp. v. Stahlin Bros. Fibre Works, Inc., 575 F.2d 1152 (6th Cir. 1978), and that his expert opinions were independently supported by testimonial and documentary evidence. First, Ericsson argues that Jackson appropriately segmented the market to reflect differing barriers to entry depending on whether a customer had already designed-in a Harris SLIC or an Ericsson SLIC. Ericsson defends Jackson’s “Harris market” definition and his market share figures as supported by the estimate of a Harris employee. Similarly, *1377Ericsson argues that Jackson’s “Ericsson market” definition and his corresponding market share calculations were supported by actual sales records and evidence relating to the costs associated with redesigning a line card. Ericsson also maintains that a lost profits analysis need not include cross-elasticity calculations. Second, Ericsson argues that it proved, causation of lost profits. Ericsson adverts to internal Harris emails, customer testimony, Harris marketing materials, and Harris’s sales of infringing products as evidence of customer demand for the patented low-power standby feature. Ericsson further argues that Jackson accounted for nonin-fringing substitutes in his analysis of the “Harris market” but that noninfringing substitutes were necessarily excluded from the “Ericsson market” because that market segment consisted only of SLICs that were pin-compatible and function-compatible replacements of the Ericsson SLIC.

We agree with Ericsson that substantial evidence supports the jury’s damages award for lost profits due to lost sales. To recover lost profits, a patent owner must prove “a causal relation between the infringement and its loss of profits.” BIC Leisure Prods., Inc. v. Windsurfing Int’l, Inc., 1 F.3d 1214, 1218 (Fed.Cir. 1993). More specifically, the patentee must show “a reasonable probability that ‘but for’ the infringing activity, the patentee would have made the infringer’s sales.” Crystal Semiconductor Corp. v. TriTech Microelecs. Int’l, Inc., 246 F.3d 1336, 1353 (Fed.Cir. 2001). To show “but for” causation, the patentee must reconstruct the market to determine what profits the patentee would have made had the market developed absent the infringing product. Grain Processing Corp. v. Am. Maize-Prods. Co., 185 F.3d 1341, 1350 (Fed.Cir. 1999). Such market reconstruction must be supported by “sound economic proof of the nature of the market and likely outcomes with infringement factored out of the economic picture.” Id.

We have affirmed lost profits awards based on “a wide variety of reconstruction theories in which the patentee has presented reliable economic evidence of ‘but for’ causation.” Crystal Semiconductor, 246 F.3d at 1355. We believe that Ericsson has done just that. Ericsson presented its damages theory at trial through Jackson’s expert testimony. Jackson reconstructed the “but for” market by segmenting the market and determining Ericsson’s lost profits based on its market share, a method that has met with this court’s approval on previous occasions. See, e.g., id. at 1354-56; State Indus., Inc. v. Mor-Flo Indus., Inc., 883 F.2d 1573, 1577-80 (Fed.Cir. 1989). Specifically, Jackson identified two markets: the “Harris market” and the “Ericsson market.” Jackson defined the “Harris market” as consisting of sales to customers that had not previously designed-in an Ericsson SLIC. Jackson identified Ericsson, AMD, and others as the competitors in that market and, based on documents.and the testimony of various witnesses, estimated that Ericsson would have had a 40% market share in the “Harris market.” Jackson defined the narrower “Ericsson market” as consisting of only those SLICs that were compatible with a customer’s existing line card. Jackson identified Ericsson and AMD as the only competitors in that market and, based on actual sales records and the switching costs that a customer wpuld incur to redesign a line card, estimated that Ericsson would have had a 97% market share in the “Ericsson market.” Jackson then allocated Harris’s infringing sales in the reconstructed market as follows: 24% to the “Harris market” and 76% to the “Ericsson market.” Finally, Jackson calculated Ericsson’s lost profits due to Harris’s infringement to be $3.5 million.

*1378Based on our careful review of the record, we conclude that Ericsson’s market definitions and allocations were supported by substantial and economically sound evidence. We also conclude that Ericsson’s market share analysis adequately compensated for the effects that any noninfringing substitutes would have had on Ericsson’s lost profits. See BIC Leisure Prods., 1 F.3d at 1219 (stating that a patentee may rely on proof of market share in lieu of proof of noninfringing substitutes in a lost profits analysis); State Indus., 883 F.2d at 1577-78 (same). Moreover, we point out that Ericsson’s failure to present all of the economic evidence that Harris now identifies does not mean that Ericsson failed to present sound economic evidence. Harris was entitled to present its own damages theory regarding, for example, how cross-elasticity calculations and second-sourcing would have affected the “but for” market. It was ultimately up to the jury, however, to weigh the credibility of the parties’ opposing theories and evidence. See Micro Chem., 317 F.3d at 1394. We will not overturn a jury’s determination as to the amount of a damages award when, as in this case, that verdict was supported by substantial evidence.

3. Lost Profits Due to Price Erosion

Harris also challenges the jury’s award of $645,000 for lost profits due to price erosion, arguing that Ericsson failed to provide sound economic proof for those damages. In particular, Harris contends that Jackson improperly concluded that both.,the “Harris market” and the “Ericsson market” were “totally inelastic” and that he failed to consider that Ericsson would have lost sales in response to an elevated price. Harris also asserts that Jackson relied on a flawed benchmark, the obsolete Ericsson 3762 SLIC, in his price erosion calculation.

Ericsson responds that an elasticity calculation is not required to recover for price erosion when the patent owner instead establishes barriers to entry. Ericsson points out that it presented evidence of two barriers to entry: the '222 patent and the costs associated with redesigning a line card, the latter of which prevented customers from switching to other manufacturers’ SLICs that were not pin-compatible. In any event, Ericsson argues, any issues with elasticity were presumably resolved by the jury when it awarded Ericsson $645,000 in damages for price erosion instead of the requested $8.1 million. Finally, Ericsson maintains that the Ericsson 3762 SLIC was an ideal benchmark for the price erosion calculation because it experienced the same market conditions as the Ericsson 3764 SLIC, except that there was no Harris clone of the 3762 SLIC.

We agree with Ericsson that substantial evidence supports the jury’s damages award for lost profits attributable to price erosion. To recover lost profits on a theory of price erosion, a patentee must show that “but for” infringement, it would have sold its product at a higher price. See BIC Leisure Prods., 1 F.3d at 1220. The patentee must also “present evidence of the (presumably reduced) amount of product the patentee would have sold at the higher price.” Crystal Semiconductor, 246 F.3d at 1357. Moreover, “the patentee’s price erosion theory must account for the nature, or definition, of the market, similarities between any benchmark market and the market in which price erosion is alleged, and the effect of the hypothetically increased price on the likely number of sales at that price in the market.” Id.

To make out its theory of price erosion, Ericsson again relied on the expert testimony of Jackson, who used a “benchmark methodology” to assess price erosion. Jackson compared the performance of the patented product, the Ericsson 3764 SLIC, in the market affected by in*1379fringement with that of a similar product, the Ericsson 3762 SLIC, in a market free of infringement. He ultimately determined that Ericsson was entitled to $8.1 million in damages for price erosion.

Despite Harris’s contention, we cannot say that Ericsson’s theory of an inelastic market precludes it from recovering price erosion damages. In Crystal Semiconductor Corp. v. TriTech Microelectronics International, Inc., 246 F.3d 1336 (Fed.Cir. 2001), we stated that “a patentee must produce credible economic evidence to show the decrease in sales, if any, that would have occurred at the higher hypothetical price.” Id. at 1359 (emphasis added). Ericsson addressed that point at trial in its price erosion analysis, submitting that the unique market conditions in this case would not have resulted in decreased sales at an increased price. Namely, Ericsson presented evidence of the high switching costs associated with redesigning a line card, the relatively low costs of SLICs, and Jackson’s expert opinion that those two conditions would produce an inelastic market. Although we have recognized that an inelastic market may be “rare,” id., it was for the jury to determine whether this was such a case based on the evidence before it.

Moreover, Harris has not shown that the 3762 SLIC was an inappropriate benchmark for the 3764 SLIC. On the contrary, Ericsson offered substantial evidence of the similarities between the two products and their markets. For example, Jackson testified that the only technological difference between the 3764 SLIC and the 3762 SLIC — i.e., that the former is “constant current” and the latter is “resistive feed” — would not have affected the price of the products. He also explained that the 3762 SLIC was an ideal benchmark because both AMD and Harris were competitors in the 3764 SLIC market, whereas AMD was the only competitor in the 3762 SLIC market. Thus, according to Jackson, removing Harris’s infringing sales from the 3764 SLIC market would result in a market very similar to the one used as a benchmark.

The jury obviously did not accept Ericsson’s theory of price erosion damages in its entirety, as it awarded only $645,000 of the $8.1 million that Ericsson requested. To the extent that the jury did accept Ericsson’s theory of damages attributable to price erosion, however, we find that award to be supported by substantial evidence. Accordingly, we affirm the district court’s denial of Harris’s motion for JMOL relating to the amount of the damages award.

CONCLUSION

For the foregoing reasons, we reverse the district court’s grant of JMOL of non-infringement and affirm the court’s denial of JMOL relating to the amount of the jury’s damages award.

COSTS

Costs to appellants.

AFFIRMED-IN-PART and REVERSED-IN-PART.

concurring in part, dissenting in part.

I

In this post-Fesio era it is no longer available, for an amended claim limitation, simply to apply the classical criteria of equivalency as if the Festo holdings did not exist. From the irregular and inconsistent application of the law in this opinion, I respectfully dissent.

The claims of the Ericsson patent were amended as to the limitation that is at issue for equivalency, thereby raising the Festo presumption of estoppel. The requisite procedure is now to determine whether that presumption has been rebutted with respect to the alleged equivalent. *1380My colleagues on this panel state that the presumption did not arise (and therefore need not be rebutted) because the words “only supply power” were not amended. That is, the court holds that the Festo presumption of estoppel did not arise because these functional words remained unchanged, although the supply of power — -the limitation for which equivalency is alleged- — was restricted by amendment. The claim clause that states the conditions under which power is supplied to the telephone set when the receiver is on its cradle was amended for reasons of patentability, after the examiner persisted in rejections for anticipation and under section 112. Thus “effectively disconnects” was amended to “disables,” and “actively connects” was amended to “enables.”

1. ... which, by the control signals effectively disconnects disables the speech signals amplifiers and actively connects enables the auxiliary amplifiers so that the speech signal amplifiers, which require power, only supply power to the telephone set when the receiver is off its cradle and a call can be made.

Amendment E. This is the clause for which application of the doctrine of equivalents is sought, for in the Harris device a small amount of power is supplied when the receiver is on its cradle. The amendment narrowed the conditions under which power is supplied to the telephone set, producing a rebuttable presumption of es-toppel.

My colleagues state that the jury could have found that the Harris on-cradle power came from a source other than the amplifiers, for example, from transistors; if so, that could indeed be grounds for rebuttal of the Festo presumption, for example on the ground that the reason for the amendment was tangential to the alleged equivalent. Perhaps a court could alternatively find that the alleged equivalent was unforeseeable, thus rebutting the presumption on that ground. However, it can only mislead and confuse the public, to hold that no Festo presumption arises on the narrowing amendment. Tangentialness and unforeseeability are grounds of rebuttal established in Festo; they are not a mechanism for avoiding the presumption. The distinction now drawn does not add clarity to this jurisprudence or impart guidance to practitioners.

The district court granted JMOL of non-infringement, invoking prosecution history estoppel as barring the equivalency of the Harris provision of small amounts of power while the receiver is on the cradle. The district court observed that it was undisputed that the Harris amplifiers supply “some power” to the telephone set in the on-hook position. The district court held that Ericsson was estopped, by the amendment of “effectively disconnected” to “disabled,” from now reaching the Harris devices through equivalency. The majority’s holding that there is no estoppel or no presumption of estoppel because the three words “only supply power” were not amended, although the words concerning the supply of power were amended, produces a dramatic loophole in the Festo decisions.

This avoidance of the Festo presumption of estoppel is interesting, for had the claims in the Festo patent been so read, they too would have avoided the Festo presumption. The result in Festo would have been a simple reinstatement of the jury verdict, as is here done, on the unencumbered question of equivalency in fact. The court’s departure today from its holding in Festo ignores our responsibility to apply that law to all cases pending at the time of the Festo decisions.5

*1381This court’s remand decision in Festo imposed interpretations with which I was not in full sympathy, as I stated at the time. See Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co., 344 F.3d 1359 (Fed.Cir. 2003) (en banc) (Newman, J., dissenting) (criticizing the majority opinion for its application of the presumption of estoppel to claim limitations that were not amended). However, I am even less sympathetic to this inconsistency in the application of precedent. This court had been seeking a brighter line with respect to the doctrine of equivalents, and presumably found it in the Court’s holding in Festo, 535 U.S. 722, 736, 122 S.Ct. 1831, 152 L.Ed.2d 944 (2002) (a narrowing amendment made for reasons of patentability raises the rebuttable presumption of estop-pel). It is our obligation now to apply this precedent with clarity and uniformity. I discern no support for the court’s position that the Festo estoppel does not apply when there is a narrowing amendment to the clause whose equivalency is in dispute, if there are unamended words in the amended clause.

On the panel majority’s holding that no estoppel arose, I would agree that a reasonable jury could have found the facts of equivalency under the classical Graver Tank function/way/result criteria. However, the approach required by precedent is to recognize the Festo presumption of es-toppel, and remand to the district court to determine whether the presumption has been or can be rebutted on any of the grounds established in the Festo cases. Consistent guidance is required for stable law.

II

I concur in Part B of the court’s decision, on the cross-appeal.