dissenting.
The court today applies its new presumption of surrender to all equivalents of the claim elements and limitations that originated in dependent claims that were never amended and that were not the subject of prosecution history estoppel. The court held in its previous opinion, Honeywell Int’l, Inc. v. Hamilton Sundstrand Corp., 370 F.3d 1131 (Fed.Cir.2004) that “the surrendered subject matter is defined by the cancellation of independent claims that do not include a particular limitation and the rewriting into independent form of dependent claims that do include that limitation.” Id. at 1144. Thus, although the limitation at issue (the inlet guide vanes used to control surge) had not been the subject of amendment, argument, or any other form of restriction, the court held that all equivalent guide vanes were presumed surrendered, not simply equivalents presumptively barred by prosecution history estoppel. Id. My colleagues held that since there was no narrowing amendment or limiting argument during prosecution of the Honeywell dependent claims, surrender of the entire universe of potential equivalents is presumed when the original independent claim is cancelled. I have previously pointed out the flaws in these generalizations, id. at 1146 (Newman, J., dissenting in part), for neither precedent nor logic requires the presumption of unlimited surrender when the prosecution history does not support unlimited surrender. See Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co., 535 U.S. 722, 122 S.Ct. 1831, 152 L.Ed.2d 944 (2002):
The complete bar [ ] is inconsistent with the purpose of applying the estoppel in the first place — to hold the inventor to the representations made during the application process and to the inferences that may reasonably be drawn from the amendment. By amending the application, the inventor is deemed to concede that the patent does not extend as far as the original claim. It does not follow, however, that the amended claim becomes so perfect in its description that no one could devise an equivalent. After amendment, as before, language remains an imperfect fit for invention. The narrowing amendment may demonstrate what the claim is not; but it may still fail to capture precisely what the claim is.
Id. at 737-38, 122 S.Ct. 1831.
This court, after creating its new presumption of surrender, did not apply it to the Honeywell-Sundstrand issues on appeal, instead remanding to the district court. Indeed, the district court remarked on the absence of guidance, stating that it was “entering uncharted waters” in which it must “knock the rust off the old sextant and hope for the best.” Honeywell Int’l Inc. v. Hamilton Sunstrand Corp., No. 99-309, 2006 WL 2346446, at *6 n. 3 (D.Del. Aug.14, 2006). Now my colleagues, apparently unsatisfied with the district court’s analysis, offer their own, presenting a new rationale whereby they find that Honeywell engineers were indeed capable of solving the Sundstrand “double solution” problem using inlet guide vanes, although Honeywell did not do so, and therefore that the Sundstrand equivalent was “foreseeable” to Honeywell and should have been patented by Honeywell in 1983. On this theory, my colleagues hold that this equivalent (as the jury found) cannot be reached under the doctrine of equivalents. The flawed sweep of this new rule is illustrated as my colleagues apply the rebuttal criteria of “foreseeability” and “tangentialness” to a presumptive surrender with no outer limit, for there is no prosecution history that narrowed the claim element at issue.
*1318My colleagues not only broaden the grounds of presumptive estoppel to estop all equivalents to any of the elements and limitations that were presented in the dependent claims, but they also restrict the grounds of rebuttal by holding that “foreseeability” is established by other than the “readily known equivalents” contemplated by the Court in Festo, 535 U.S. at 740, 122 S.Ct. 1831. Applying this new rule with no limit to the technology to which “foreseeability” is applied, my colleagues find “foreseeable” the Sundstrand guide vane technology that was developed a decade later. My colleagues find that Honeywell engineers were smart enough to have invented 1 the Sundstrand technology a decade earlier, and thus that Honeywell should or could somehow have claimed it, although it was not described and enabled by Honeywell. On this reasoning, the court holds that Honeywell cannot rebut the presumption of surrender of all equivalents using inlet guide vanes. From this unsound analysis, I respectfully dissent.
Whether and when an inventor is entitled to reach beyond the literal scope of what is claimed in the patent has received continuing attention from courts and the technology community, for the issues of law are not easy, and the considerations of equity are in tension. The Court in Festo, and earlier in Warner-Jenkinson, Co. v. Hilton Davis Chem. Co., 520 U.S. 17, 117 S.Ct. 1040, 137 L.Ed.2d 146 (1997), revised the legal and equitable balance, limiting the doctrine of equivalents while preserving its viability. This court’s new presumption of surrender of all equivalents, when there is no prosecution history estop-pel as to elements from dependent claims that are placed in independent form, places new constraints on the patentee’s access to the doctrine of equivalents. This aspect of patent-based innovation requires a more balanced analysis, in the national interest in innovation. As the Court stated in Festo, “if the doctrine is to be discarded, it is Congress and not the Court that should do so.” 535 U.S. at 733, 122 S.Ct. 1831.
Rebuttal of the presumption of surrender — foreseeability
Now, although I accept as the law of this case that a presumption of surrender arose with respect to the use of inlet guide vanes to control surge, my colleagues apply the Festo rebuttal criteria in significantly more restrictive ways than were established by the Supreme Court. The Court had explained that “[tjhough prosecution history estoppel can bar a patentee from challenging a wide range of alleged equivalents made or distributed by competitors, its reach requires an examination of the subject matter surrendered by the narrowing amendment.” Id. at 737, 122 S.Ct. 1831. However, my colleagues do not examine the surrendered subject matter, and indeed they can not, for there is no narrowing amendment to define surrendered subject matter. The dependent claims were not rejected on any ground; they were simply “objected to” because they were in dependent form. The use of the inlet guide vanes was never discussed in the prosecution, and the guide vane clause was not amended in any dependent or independent claim. There was no prosecution for this issue based on prior art.2 Unlike the Court’s Festo presumption of prosecution history estoppel, which is limited by narrowing amendment, the panel majority’s presumption of surrender is without limit. *1319Honeywell, 370 F.3d at 1144 (“Under such circumstances, the surrendered subject matter is defined by the cancellation of independent claims that do not include a particular limitation and the rewriting into independent form of dependent claims that do include that limitation. Equivalents are presumptively not available with respect to that added limitation.”).
In the remand decision from which this appeal is taken, the district court had recognized that the element for which equivalency was charged — -the inlet guide vanes used to adjust a set point to avoid surge— had not been narrowed and thus did not fit the Festo mold. Nonetheless, the district court ruled that the Sundstrand apparatus and method with respect to the inlet guide vanes were “foreseeable” and that the rationale underlying the amendment bore more than a “merely tangential” relation to the Sundstrand equivalent. However, the district court declined to give its reasoning, stating that the record was “too extensive and complex” for the court to identify the facts on which its conclusion rested, and that it would be “unduly burdensome” to explain which evidence it credited. Honeywell, 2006 WL 2346446 at *6 n. 2. The district court stated that “it seems quite intuitive to this lay court that measuring IGY position ... is a reasonably obvious way — both at present and in 1982-83 — to determine whether the APU is experiencing high flow or low flow.” Id. at *6. The court also mentioned “credibility,” although it did not identify who or what it did not believe. The court concluded that the Sundstrand use of inlet guide vanes to control surge was “foreseeable,” although witnesses for both sides agreed that the use of inlet guide vanes for surge control was not known at the time of the Honeywell invention.
Equivalency is determined as of the time of litigation, while foreseeability for equivalency purposes is determined at the time of the patent application. The Court in Festo explained that foreseeability means “readily known equivalents,” 535 U.S. at 740, 122 S.Ct. 1831, not unknown equivalents developed a decade later. Honeywell’s use of inlet guide vanes was developed before 1982, and the Sundstrand inlet guide vanes were developed between 1990 and 1995. Sundstrand had argued at the jury trial that its guide vane system does not infringe, either literally or under the doctrine of equivalents, because it was designed to meet what Sundstrand calls the “double solution” problem, and that its system makes a different use of inlet guide vane position than does Honeywell. These differences were not disputed. My colleagues nonetheless find the Sundstrand device and method “foreseeable” based on a combination of prior art surge control devices, and hold that Honeywell could have described and claimed the Sundst-rand technology a decade before it was developed by Sundstrand.
In deciding whether the presumption of surrender has been rebutted, the question is whether the technological equivalent is embraced by subject matter that was relinquished during prosecution of the patent. This requires analysis of the prosecution history leading to cancellation of the original independent claim and rewriting of the original dependent claim in independent form. Patent claims are customarily presented in independent and dependent form, a practice encouraged by the Patent and Trademark Office, for it simplifies examination. When an original independent claim is cancelled, it is obligatory that the next dependent claim be placed in independent form. Even on the majority’s analysis of the rebuttal criteria of foreseeability and tangentialness, the proper approach is to determine for equivalency was implicated in that cancellation.
*1320On remand, the district court attempted to apply the instructions from the Supreme Court and this court as to rebuttal of the presumption of prosecution history estoppel. The Court has guided that the foreseeability analysis must be “directed to the particular equivalent in question,” Id. at 740, 122 S.Ct. 1831. The particular equivalent is the Sundstrand device and method using the inlet guide vane positions to solve the so-called “double solution” problem that attends Sundstrand’s choice of the DELPQP flow-related parameter, by blocking the control signal during high-flow conditions. Honeywell states that this is an equivalent use of inlet guide vanes, in the context of Honeywell’s use of inlet guide vane position to adjust the set point against which the flow-related parameter is measured across the spectrum of flow conditions. The panel majority finds the Sundstrand device and method foreseeable and thus not reachable under the doctrine of equivalents.
The panel majority cites two references that it designates as prior art that rendered the Sundstrand device foreseeable a decade before Sundstrand developed it. My colleagues focus first on a device called the L1011 developed in the late 1970s; this device made no use of inlet guide vanes, but instead used a totally different system that approached the problem of the ambiguity of a flow parameter at high flow conditions by use of a “shock switch.” I suppose the purpose of this reference is to show that the problem was not new; however, recognition of the problem does not render foreseeable Sundstrand’s equivalent that was developed more than a decade later using inlet guide vanes.
The panel majority also relies on the Glennon patent, and finds that in 1979 it was known that adjusting guide vane position has an impact on airflow. However, the issue is not the use of louvered vanes to control airflow, a known concept; the issue is whether it was foreseeable, at the time of the Honeywell patent application, to use inlet guide vanes for surge control in an accessory power unit in the manner of the Sundstrand device, as part of the decision logic for high-flow conditions when the DELPQP parameter produces an ambiguous signal.
My colleagues suggest that testimony from Honeywell’s witnesses Clark and Muller establish foreseeability of the Sundstrand equivalent, for they testified that Honeywell was working on the use of inlet guide vanes to improve the operation of a surge controller, and that if they knew about the Sundstrand problem they might have figured out how to solve it in the way that Sundstrand solved it a decade later. But these witnesses did not testify that Honeywell already had this knowledge, or that it was known to the prior art or otherwise was a known equivalent or “reasonably foreseeable.” It was undisputed that the Sundstrand equivalent was developed a decade later in full view of the Honeywell patents.
My colleagues also state that the district court’s findings are predicated on witness credibility, and thus are untouchable. The district court did not tell us what aspects strained its credulity, or indeed why the district court had sustained the jury verdict in 2001 if tainted by the incredibility of witnesses. Deference to a district court’s findings on credibility is generally warranted, but when the issue is one of objective science/technology, more is required than a terse announcement by the court that no reasons will be given for its ruling. As explained in Anderson v. Bessemer City, 470 U.S. 564, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985), deference is warranted only when competing witnesses have “told a coherent and facially plausible story that is not contradicted by extrinsic evidence,” and when the finding is “not *1321internally inconsistent.” Id. No deference is owed to a district court’s conclusion predicated upon findings it did not state or explain. See Edwards v. Wyatt, 335 F.3d 261, 274-75 (3d Cir.2003).
The district court’s reluctance to state the facts on which its conclusion rests, as Federal Rule of Civil Procedure 52 requires, negates the traditional deference. I accept that the record in this lengthy litigation is difficult to summarize, but I also accept — for it is uncontradicted — that the record is devoid of any evidence that the Sundstrand equivalent previously existed or was the “readily known equivalent” required by Festo. It was uncontra-dicted that the Sundstrand equivalent was developed years after the Honeywell application was filed and prosecuted, and only after considerable effort. Nothing in the record supports my colleagues’ finding that the Sundstrand equivalent was foreseeable at the time the Honeywell patent applications were prosecuted.
Simply presenting claims of varying scope, whether in independent or dependent form, is not a narrowing amendment or argument. The patent examining rules encourage, through lowered fees, the use of dependent claims, for it facilitates examination. It is routine for broader claims to be cancelled and dependent claims to be rewritten in independent form. This protocol has no relevance to whether a claimed element is amended or narrowed or argued during prosecution. It is the narrowing of scope during prosecution that produces the presumptive estoppel; claim elements and limitations that were not the subject of amendment or argument do not raise the Festo presumption, and should not be deemed to raise this court’s new presumption. Cancelling an independent claim is not an estoppel-generating act as to elements whose scope was not amended or otherwise restricted during prosecution.
Equivalency is determined element by element, Warner-Jenkinson, 520 U.S. at 29, 117 S.Ct. 1040, yet on the panel majority’s presumption of surrender, there is no restricting action by which to measure what was surrendered. In Festo the Court was explicit that the burden is to “show[ ] that the amendment does not surrender the particular equivalent in question.” 535 U.S. at 740, 122 S.Ct. 1831. My colleagues err in now holding that all equivalents of an element presented by dependent claim are presumed surrendered by simply cancelling the independent claim.
Tangential Relation
The Court explained that when the asserted equivalent related to “aspects of the invention that have only a peripheral relation to the reason the amendment was submitted,” the presumption of estoppel would be rebutted. Id. at 738, 122 S.Ct. 1831. Precedent illustrates that the “tangential” criterion is rebutted when “[tjhere is no indication in the prosecution history of any relationship between the narrowing amendment and [the process] which is the alleged equivalent in this case.” Insitu-form Tech., Inc. v. CAT Contracting, Inc., 385 F.3d 1360, 1370 (Fed.Cir.2004). That analysis well fits this case, for there is no indication in the prosecution history of any relationship between Honeywell’s cancellation of the independent claim and the alleged equivalent Sundstrand apparatus or method.
The panel majority states that “when Honeywell rewrote the application dependent claims into independent form, it ‘effectively add[ed] the [IGV] limitation to the claimed invention’,” maj. op. at 1315 [alteration in original], and therefore, it provided “a direct, not merely tangential, relation to the equivalent,” id. at 1316. This is not the tangential relationship contemplated by Festo, where the Court de*1322manded review of the reason for the narrowing amendment and the relation of that reason to the accused equivalent. It does not suffice to say that by narrowing a claim by adding an element, ipso facto the relation to the accused equivalent element is not tangential. This criterion relates to why an amendment was made; it does not become irrebuttable simply when the accused equivalent concerns the same element that was added by amendment.
Exploration of this criterion requires, at a minimum, review of the reason for cancellation of the independent claim and rewriting the dependent claim in independent form. The panel majority makes no mention of the prosecution of the original independent claims, except to state that the examiner rejected them on the ground of obviousness. Although the panel majority states that “the key to this inquiry is the content of the original dependent claims,” this obscure statement leads to no analysis. In a situation where an originally dependent claim is asserted against an infringer under the doctrine of equivalents, and the asserted limitation was first introduced in a dependent claim, then according to the panel majority there is always more than a tangential relationship between the reason for the amendment and the accused equivalent, even when there is no finding as to the reason for the amendment. This renders the presumption of estoppel irre-buttable, whatever the reason for the amendment. However, the question is whether the subject matter of the accused equivalent was relinquished by the paten-tee during prosecution. This question should be resolved on its facts, not converted into a complete bar.
Why does it matter?
These nine years of proceedings do not reflect an optimum mode of patent dispute resolution. The jury verdict of infringement under the doctrine of equivalents was rendered in February 2001, and all outstanding issues were resolved by the district court by opinion rendered in 2001. The jury was instructed in accordance with precedent:
Instruction No. 3.5.3.... The test for determining equivalency is whether, at the time of the asserted infringement, one of ordinary skill in the art would consider the differences between the accused product or method and the element or elements of the patent claim not literally present to be insubstantial. If such a person would consider those differences not to be substantial, then infringement exists under the doctrine of equivalents.
Honeywell must prove infringement under the doctrine of equivalents by a preponderance of the evidence. In determining whether Honeywell has proven that a feature of the Hamilton Sundstrand product is equivalent to an element of a patent claim, you should focus on the individual element and maintain a special vigilance against allowing the concept of equivalence to eliminate that element entirely.
* * * * * *
It is not a requirement under the doctrine of equivalents infringement that those of ordinary skill in the art knew of the equivalent when the patent application was filed or when the patent issued. The question of whether Hamilton Sundstrand’s product is equivalent to that defined in Honeywell’s claims is to be determined as of the time of the alleged infringement.
The correctness of the instruction was not in dispute. The verdict of no literal infringement, but of infringement under the doctrine of equivalents, was sustained by the district court. Honeywell Int’l, Inc. v. Hamilton Sundstrand Corp., 166 F.Supp.2d 1008 (D.Del.2001). The court’s holding today takes another step to re*1323move patent issues from the jury, but adds nothing to consistency or stability or predictability of the law. I respectfully dissent.
. The Sundstrand technology is patented to Sundstrand.
. Sundstrand stated to the district court that “during the prosecution of the Patents-in-Suit, the Examiner did not reference any Pri- or Art that disclosed inlet guide vanes or the use of their position as part of a surge control system.” Appellee’s Br. at 10.