RICH, Circuit Judge.
This appeal is from the final amended judgment of March 11, 1985, of the United States District Court for the Northern District of Texas, Mahon, Judge, holding Hughes Tool Company’s (Hughes Tool) U.S. Patent No. 3,397,928 (’928 patent)1 valid and infringed by Dresser Industries, Inc. (Dresser), awarding damages in the amount of $132,096,430.92, and from orders denying various Dresser post-trial motions to reopen the record, amend the judgment, and for a new trial. We affirm in part, vacate in part, and remand.
I. Background
The ’928 patent claims in suit are directed to combinations of elements in a rock bit used in drilling oil and gas wells. Those combinations involve the use of a packing or seal of the O-ring type in the enclosed journal bearings of the rock bit cutters, wherein the O-ring is compressed by not less than 10% and preferably 15% of the thickness of the O-ring prior to its assembly.
A. Rock Bits
The first rolling cutter rock bit was invented in 1909 by Howard Hughes, Sr. The modern rock bit, the product of technical improvements made over the last seventy-five years to Mr. Hughes’ original, is a sophisticated piece of machinery costing up to several thousand dollars apiece. A rock bit, generally, is comprised of a supporting body structure or “head” to be threaded onto the lower end of a string of drill pipe, and carries three rotatable, generally conical cutters having either steel teeth or tungsten carbide insert (“TCI”) cutting elements. These conical cutters turn on internal bearings and are mounted on three shafts or axles on the head.
In drilling wells using rock bits, the drill pipe with the rock bit attached is lowered, section by section, into the drill hole and rotated under the great weight of the string against the bottom. This causes the bit to work against the rock formation, chipping or crushing the rock as the cones drag and roll upon the bottom of the hole. A drilling fluid, usually water or a thin mud, is pumped downward through the pipe, and emerges from nozzles in the bit, picking up cuttings loosened at the bottom of the hole. The cuttings are carried upward from under the bit outside the pipe to the surface.
Much of the time and expense of rotary drilling is spent in the process of “tripping” —removing a spent bit from the end of a drill pipe at the bottom of the hole and replacing it with a new one. When a rock bit fails, as it inevitably will, the drilling crew must raise the drill stem, uncouple and stack the ninety-foot pipe sections one at a time, substitute a new bit, then recou[1551]*1551pie and lower the pipe into the hole and resume drilling. In drilling wells of several thousand feet and more, the trip time required to change a bit may take up to ten or twelve hours, during which time no drilling gets done and injuries to the crew and other catastrophes such as blow-outs and well fires are most apt to occur.
The goal of rock bit engineers has thus been to increase the useful drilling life of the bit. There have been two primary approaches to solving that problem: improving the effectiveness and durability of the teeth or cutters, and prolonging the life of the bearings on which the cutters turn. Failure of either necessitates a “trip.” This case relates to bearings.
B. Bearings
The ’928 patent is directed to solving the bearing portion of the rock bit design problem. The ’928 patent discloses a design providing a bearing capacity which could outlast the life of the cutters.
Historically, there have been two main classes of bearings involved in rock bit technology: “journal” or “friction” bearings and “anti-friction” bearings which contain roller and possibly also ball bearings. Journal bearings are the simplest and sturdier form. The first rolling cutter rock bits dating from 1909 utilized journal bearings which were unsealed against grease loss and the intrusion of abrasive drilling fluid. The bearing was lubricated by a long lubricator sub-assembly from which grease flowed downward into the bearings and outward from the bottom of the bit. Various attempts were made during the early years to seal the journal bearings of Mr. Hughes’ original design. While a sealed bearing would have prevented the drilling fluid from entering the bearing and retained the lubricating grease in the bearing, none of the early attempts to seal journal bearings succeeded.
Due to the perceived inability to seal them, journal bearings were abandoned in the early 1930’s in favor of anti-friction roller bearings, which were better able to withstand the severe drilling environment without being sealed. In the original roller bearing bits, the drilling fluid was permitted to circulate through the roller bearings, a process known as “slush lubrication.” The roller bearing bit soon became the standard in the industry.
C. Seals
There were many problems in designing a sealed bearing rock bit, because of the severe conditions in which such bits operate — the very limited space available in a rock bit, the extreme weight and high temperatures at the bottom of the drill hole, and the rapid and peculiar movements of the cutters and shafts of the bits when slightly worn. An effective seal would have to possess both the resiliency to follow the erratic movements of the cutters but also be able to maintain an effective seal against changes in relative hydraulic pressures of the grease and drilling fluid.
Although the first patents disclosing sealed bearings were issued as early as 1918, it was not until the late 1950’s that the first successful sealed-bearing rock bits were developed. Those first commercially successful sealed bearing rock bits utilized the then conventional roller bearings, an internal supply of grease, and the so-called “Belleville” seal, disclosed in Hughes Tool’s Atkinson et al. U.S. Patent No. 3,075,781. The Belleville seal makes use of a saucer-shaped, annular, steel spring covered with rubber or plastic that surrounds the bearing shaft, providing continuous pressure against the opposed sealing surfaces of the head and cutter throughout the compound movements of the cutters during operation of the bit.
Although the Belleville-sealed bearing bits increased bit life dramatically, rock bit design engineers continued to work on improving bearing performance, hoping to design a bearing that would outlast the useful life of the cutters. Although synthetic rubber O-ring seals, which had developed rapidly in the World War II period, were known to be effective seals in other industrial applications, they were usually recommended for use only when the bearing surfaces to be sealed were free of abrasive substances and where movements such as [1552]*1552wobbling between the parts to be sealed would be small.
O-rings had been used as seals in rock-bit bearings before the present invention. Neilson U.S. Patent No. 3,127,942, issued in April, 1964 (Neilson patent) discloses a roller-bearing bit having an O-ring seal. Both U.S. Patent No. 2,075,997, issued to Clarence Reed in 1935 (Reed patent) and U.S. Patent No. 2,814,461, issued in 1957 to William Green (Green patent) relate specifically to a packing ring in a journal bearing bit. Green discloses the placement of a squeezed “packing ring” or “sealing gasket” (neither shown as nor stated to be O-rings) on the bearing shaft and between the opposing head and cutter surfaces, while Reed shows a packing ring similarly positioned that “would be compressed when placed in position.”
D. The Squeeze
Although Dresser, Smith International, and other competitors as well as Hughes Tool experimented with O-ring seals in the 1950’s and early 1960’s, none of these efforts was successful. The first successful O-ring seal in a rock bit was developed at Hughes Tool by Edward M. Galle, patentee of the ’928 patent. Although Galle was not the first to make use of an O-ring seal in a rock bit, the primary distinction between the Galle invention and the work of his predecessors was the amount of cross-sectional “interference” or “squeeze” imparted to the O-ring in Galle’s rock bit journal bearing design.
Although it was understood that relatively high-speed rotary operations, such as those involved in rock bit drilling, called for a small amount of squeeze, the ’928 patent teaches a squeeze of “not less than substantially ten percent” and perferably not less than fifteen percent of the cross-sectional thickness of the O-ring in its relaxed condition. This is considerably higher than the amount of squeeze recommended by manufacturers of O-rings. Such amounts of squeeze were believed to increase friction and overheat the O-rings, thus decreasing their life span. The ’928 patent disclosed that the heat generated by such large interferences apparently “dissipates in the drilling fluid that turbulently flows around the drill bit.”
E. Hughes Tool’s New Rock Bits
Hughes Tool introduced new rock bits embodying the seal invention of the ’928 patent with journal bearings in 1969. The new bits were very successful, demonstrating a capacity to endure for extremely long runs, on the order of 100 hours or more, which was double or triple the performance of the old slush-lubricated roller bits. The new sealed journal bearing bits were particularly successful because they unlocked the full potential of the TCI cutting structures, which typically had outlasted the bearings.
Hughes Tool’s competitors reacted immediately to the introduction of the new bits, commencing research projects to develop competitive bits. Although the engineers at Hughes Tool’s main competitors, Dresser and Smith International, Inc., were skeptical of the teachings of the ’928 patent and particularly of the usefulness of O-rings in a bearing seal, both companies ultimately decided to copy the O-ring seal of the ’928 patent and began selling rock bits that included Hughes Tool’s patented O-ring seal. Over $3.5 billion worth of the new sealed journal bearing rock bits were ultimately sold by Hughes Tool and its compete itors. Dresser’s total sales of such rock bits in the years 1973 through 1982 amounted to $343,354,994.00.
II. The Present Action
Hughes Tool originally filed this action against Dresser in 1976, alleging infringement of its '928 patent as well as U.S; Patent Nos. 3,476,195 (’195 patent) and reissue patent Re. 28,625. The reissue patent was dismissed from the suit voluntarily before trial in a consent judgment of invalidity.
On July 11,1980, the district court granted Dresser’s motion for summary judgment, dismissing this action on collateral estoppel grounds, after both the ’928 and the ’195 patents were held invalid for obviousness by the United States District Court [1553]*1553for the Middle District of California in an earlier-filed suit between Hughes Tool and Smith International. On appeal, the Court of Appeals for the Ninth Circuit reversed the holding of invalidity of both the '928 and ’195 patents, holding that Smith had not sustained its burden of showing invalidity. Smith International, Inc. v. Hughes Tool Co., 664 F.2d 1373, 215 USPQ 592 (9th Cir.), cert. denied, 456 U.S. 976, 102 S.Ct. 2242, 72 L.Ed.2d 851 (1982).
The present case was thereupon reactivated in May, 1982. Following a bench trial in February, 1983, the district court entered a Memorandum Opinion and Order on February 6, 1984, finding the ’928 patent valid and infringed, but finding the ’195 patent invalid.
Dresser filed post-trial motions under Fed.R.Civ.P. 59, asking for a new trial on the grounds of newly discovered evidence regarding the validity of the ’928 patent as well as the size of the damage award. Those motions were denied by the district court.
Dresser has appealed the district court’s holding of validity of the ’928 patent and the amount of damages awarded to Hughes Tool for the infringement. Dresser also appeals the court’s denial of its post-trial motions. Hughes Tool does not appeal the district court’s holding that the ’195 patent is invalid, but cross appeals on two issues relating to the sufficiency of the damage award for the infringement of the ’928 patent.
III. The Decision Below
The lower court opinion being unreported, we will summarize briefly its major findings and conclusions with the general preliminary comment that the court was thorough, fair, and comprehensive in its analysis. It wrote 42 pages exclusive of formal findings and conclusions which occupy an additional 22 pages.
A. Validity
The district court, in its extensive memorandum opinion, began its analysis by noting that the ’928 patent “involves an O-ring or packing ring positioned in a groove, thus forming a seal for a rock bit. The court noted that claim 1 calls for the O-ring to be “compressed by not less than substantially ten percent of its thickness prior to assembly.”
The court then turned to an examination of the scope and content of the prior art and a comparison of the differences between the prior art and the claims at issue, finding that the prior art revealed all of the elements of the '928 patent except the requirement of a squeeze of “not less than substantially 10%.” Although Dresser submitted evidence of articles and catalogs showing that squeezes of more than 10% were commonly used for slow rotary speed seals, the court found that rock bit bearings are run under high speed and high temperature rotary conditions. The court also noted that “[wjithout exception, the catalogs recommended higher squeezes for slow rotary speeds and lower squeezes for fast rotary speeds.” The court thus concluded that rock bit engineers in the 1960’s would have consulted the charts for fast rotary speeds and would have concluded that squeezes well below 10% were proper for O-ring bearing seals in rock bits. The court also noted in conclusion that the prior art would probably have discouraged rock bit engineers from considering the use of O-rings as bearing seals because of the abrasive conditions and the heat, pressure, and friction encountered at the bottom of the drill hole. The court thus found that the inventor, Galle, proceeded “contrary to the accepted wisdom of the art” by using an 0-rir.g seal in a rock bit and by requiring a minimum of a 10% squeeze.
With respect to the criticality of the minimum 10% squeeze, the district court found that the successful results of the greater squeeze were unexpected and “contrary to the thinking of rock bit engineers in the 1960’s.” Moreover, the criticality was substantiated by the fact that the minimum 10% squeeze was the one factor that both Hughes and all of its competitors found to be essential to a longer-life bit. The court concluded that there was “ample evidence showing the criticality” of the minimum 10% squeeze.
[1554]*1554The court also discussed evidence of the skepticism of the experts at both Smith International and Dresser, the overwhelming commercial success of Hughes Tool’s new rock bits, and the fact that both Dresser and Smith International chose to copy the patented feature of the new bits. The court then concluded that the evidence “clearly established] that the use of an O-ring with a minimum of a 10% squeeze in a rock bit would not have been obvious to a person having ordinary skill in the art____”
B. Damages
Although Hughes Tool sought damages based on its lost profits with respect to the TCI journal bearing bits Dresser sold to customers within the United States, the court concluded that Hughes Tool “failed to demonstrate with ‘reasonable probability’ an approximation of the profits it would have made on the infringing Dresser sales.” The court then set out to determine, “under the ‘willing buyer — willing seller’ rule, what a reasonable royalty” would have been.
In assessing the evidence on the damages issue, the district court found that customer demand for the patented rock bits was strong and that no acceptable non-infringing alternative existed. However, the court also found that Hughes Tool failed to demonstrate the manufacturing and marketing ability to meet the demand for at least part of the period of Dresser’s infringement.
Turning specifically to the commercial situation as it existed in 1973, the district court noted that Hughes Tool was experiencing a rising demand for its patented journal bearing bits and that Hughes Tool believed that those bits were more profitable than its sales of certain other tools. Although Hughes Tool had a company policy of not licensing its patented inventions to major competitors, the court did note evidence of forty-five prior licenses and offers of licenses under other patents owned by Hughes Tool at royalty rates of six percent or less. The court did not give much weight to those prior royalty rates, however, because those licenses were for widely diverse products. Similarly, the court acknowledged the existence of one license at four percent to a non-competitor, Gault, and specifically held that it would not consider a fifteen percent license agreement made in 1983 in settlement of litigation with Hughes Tool’s competitor, Rock Bit Industries. The court ultimately concluded that Hughes Tool showed no desire to license direct competitors under the ’928 patent.
The court also found that Dresser was concerned about a potential loss of fifty to sixty percent of its rock bit market to Hughes Tool; that when Dresser considered beginning to manufacture the patented journal bearing bits itself in 1973, it projected total profits of 60% on sales of the new bits; and that its total profits on all sales the year before had been 7.2%. The district court also noted that Hughes Tool asserted profits of 49-54% on sales of the patented journal bearing insert bits.
The district court found, however, that Hughes Tool did not realize that the demand for rock bits other than sealed journal bearing bits would drop as sharply as it did. In addition, Hughes Tool was aware that its competitors were working diligently to produce competitive rock bits. Finally, the district court found that the ’928 patent did not give the new rock bits their essential market value, but rather that the patented O-ring seal was only one of numerous parts that contributed to the overall value of the rock bit.
Although Hughes Tool’s expert at trial asserted that 52.8% was a reasonable royalty for both the patents in suit, that assertion was based on the assumption that the patented combination with the O-ring seal was responsible for most of the value of the bit. The district court disagreed, stating that the '928 patent contributed only part of the total value of the bit, and concluded that 25% of the gross sales price would have been a reasonable royalty had the parties negotiated for a license in 1973.
The district court went on to find that a ten percent interest rate was appropriate for awarding prejudgment interest. The court declined to increase the damages, [1555]*1555finding that Dresser did not willfully infringe the ’928 patent. Although Dresser “deliberately copied the ideas and designs in the ’928 patent,” the court found that Dresser did so only after a reasonable investigation into the scope of the patent and a good faith belief that the patent was invalid.
In its amended judgment of March 11, 1985, the district court awarded damages in the amount of $132,096,430.92 to Hughes Tool. That amount included prejudgment interest of $29,439,158.91.
The court did not award prejudgment interest for the period between November 30, 1979 and February 22, 1982, when the ’928 patent “was considered invalid and unenforceable by the California federal court.” That was the period between the district court holding of invalidity in the Smith suit and its reversal by the Ninth Circuit Court of Appeals.
The court held further hearings on February 15, 1985 and March 1, 1985, on Dresser’s motions to reopen the record, to alter or amend the judgment, and for a new trial. After stating that it did not find the allegedly newly discovered evidence warranted reopening of the case or a reduction in the damages award, the court denied Dresser’s motions in an order dated May 2, 1985.
On May 31, 1985, Dresser appealed the district court’s judgment with respect to the validity of the ’928 patent and the amount of damages, as well as the district court’s denial of its post-trial motions. Hughes Tool filed its cross-appeal on June 13, 1985, seeking review of the district court’s failure to award increased damages as well as its failure to award prejudgment interest for the period the ’928 patent had temporarily been under the cloud of a district court holding of invalidity.
IV. Opinion
The issue here is whether the district court erred in concluding that Dresser did not carry its burden of proving that the invention claimed in the ’928 patent, as a whole, would have been obvious to one of ordinary skill in the art at the time the invention was made.
Under 35 U.S.C. § 282 the ’928 patent is presumed valid. That presumption places the burden of establishing invalidity on the attacker, Dresser. It is for the district court to determine whether the attacker has carried the burden. American Hoist & Derrick Co. v. Sowa & Sons, Inc., 725 F.2d 1350, 1360, 220 USPQ 763, 771 (Fed.Cir.), cert. denied, 469 U.S. 821, 105 S.Ct. 95, 83 L.Ed.2d 41 (1984). The attacker must provide clear and convincing evidence of the facts leading to the conclusion of invalidity of the claims at issue in order to carry that burden. Lindemann Maschinenfabrik GMBH v. American Hoist & Derrick Co., 730 F.2d 1452, 1459, 221 USPQ 481, 486 (Fed.Cir.1984).
After considering all of the evidence before it, including prior art before the Patent and Trademark Office (PTO) during prosecution as well as evidence that was not before the PTO, the district court concluded that the evidence did not support Dresser’s assertion that the prior art would have suggested a minimum 10% squeeze of O-rings used to seal rock bit bearings. Dresser’s evidence included catalogs and articles which suggested that a squeeze of greater than 10% would be appropriate for low speed rotary applications. However, the evidence before the court also indicated that the sliding velocity on the surface seal of a rock bit bearing is as high as 540 feet per minute, a concededly “high” rotary speed. The district court thus properly found that rock bits are sometimes run at high speeds and that rock bit engineers in the 1960’s would have designed the bits for high speed rotary operations, thus leading them to conclude that squeezes well below 10% would have been appropriate.
Dresser also challenges the district court’s finding that a squeeze of not less than 10% is “critical,” arguing that the burden should have been on Hughes Tool to establish the criticality of that claim limitation. Although the district court expressed “concern” over the evidence of criticality before the PTO, the presumption of [1556]*1556validity placed the burden of establishing non-criticality on Dresser. Dresser has not pointed to any evidence tending to show that the minimum of 10% squeeze was not critical. We are not persuaded that the district court was clearly erroneous in concluding that Dresser had not carried its burden of persuasion on the criticality issue. If it was not critical to success, why was it copied?
Finally, Dresser argues that the district court erred by considering objective indicia of nonobviousness (the so-called “secondary considerations”), namely, evidence of skepticism of experts, long-felt need, commercial success, and copying. Dresser conceded that rock bits embodying the features claimed in the ’928 patent attained enormous commercial success and satisfied a long-felt need. Dresser also conceded that while its engineers were initially skeptical, it eventually copied the patented features of Hughes Tool’s new rock bits, including the amount of squeeze imparted to the O-ring seal. The district court noted that, despite Dresser’s assertions that other aspects of Hughes Tool’s new rock bits were responsible for Hughes Tool’s commercial success, the one feature that Hughes Tool and all of its competitors continued to use was the minimum 10% squeeze called for in the claims in suit of the ’928 patent. Such continuous use of the patented feature while other features were not copied gives rise to an inference that there is a nexus between the patented feature and the commercial success. We are not persuaded that the district court erred in concluding that there was such a nexus in this case.
We conclude, on the record before us, that the district court properly held that Dresser failed to carry its burden of proving that the ’928 patent is invalid for obviousness of the claimed invention.
Dresser has not denied its infringement of the ’928 patent. Thus the only issues remaining on appeal and cross-appeal are those pertaining to the amount of the damage award.
The district court’s award of damages to Hughes calculated as a royalty of 25% of the dollar amount of Dresser’s total sales of certain infringing bits is based on a finding of fact with respect to Dresser’s projected profits which Dresser asserts is clearly erroneous.2 The basic predicate for the district court’s 25% royalty determination is that Dresser projected “total profits of 60% as a percent of sales (PX 82, 93, Response to Request for Admission 11).” That 60% “profit” figure was then contrasted with Dresser’s 1972 actual profits of 7.2% for all of its sales. (PX 82, Response to Request for Admission 16.) The 60% profit projection is referenced repeatedly in the court’s analysis and is clearly critical to its damage award:
“... the projection of 60% profit in the present case, admitted by Dresser”
“Hughes’s 49% and 54% royalty requests would be 81.6% and 90% of Dresser’s projected 60% profits in this case.”
“15% profit” in other Hughes litigation is “considerably lower profit than those projected in the present case.”
“the parties in 1973 projecting 49% [Hughes] to 60% [Dresser] profits.”
“Dresser projected 60% profits on the sale of bits with the ’928 seal.”
“Hughes suggests it realized profits of 49% and 54% on its ’928 bits sold in the United States and abroad.”
“This figure [25% of sales price] would still have allowed Dresser to realize a large profit on its sales of the bits.”
“Of these various considerations [relevant to setting a reasonable royalty] the expected profit from making and selling the patented invention is the primary consideration.”
[1557]*1557Before this court Hughes Tool argues: “The magnitude of this ‘large profit’ of 60% less the royalty award of 25%, left Dresser a 35% profit from sales — as contrasted with its usual profit of 7.2% of sales on all of its products.” Hughes’s simple arithmetic, however, is a gross distortion of reality.
The evidence of Dresser’s projected profitability does not show, as the trial court found and Hughes asserts here, that, in 1973, Dresser was projecting 60% profit on the proposed new product line. Rather, Dresser’s projection was a 60% return on the additional investment required to make the bits. Thus, the 60% projection of such return cannot be compared meaningfully with the 7.2% actual profits of 1972.
The evidence on this point from PX 82 and PX 93 on which the district court relied for its finding is set forth below in pertinent part. In Requests for Admissions propounded to Dresser (PX 82) the following pertinent requests were made and answered:
REQUEST NO. 11 [by Hughes]
During the year 1973, defendant, Dresser Industries, anticipated that its incremental future sales of Insert Rock Bits with Journal Bearings would provide defendant with profits, on an earnings before tax basis, of not less than 60% of dollar sales of such Rock Bits.
RESPONSE TO REQUEST NO. 11 [by Dresser]
Denied. During the year 1973, for the purposes of capital appropriations only, personnel of Defendant at one time forecast that its incremental future sales ascribable to that particular appropriation request of insert Rock Bits with Journal Bearings would provide Defendant with profits, on an earnings before tax basis, of not less than 60% of dollar sales of such Rock Bits. [Emphasis ours.]
REQUEST NO. 16 [by Hughes]
During its fiscal years 1972 through 1981, defendant, Dresser Industries, was making profits, on a before tax basis, as a percentage of net dollar sales of all of its products and services as set forth below:
[[Image here]]
RESPONSE TO REQUEST NO. 16 [by Dresser]
Admitted.
The documentary evidence (PX 93) which supports the answer to Interrogatory 11 is a 1973 internal Dresser report entitled “CAPITAL REQUIREMENTS FOR COST REDUCTION & EXPANSION OF OILFIELD & MINING PRODUCTS.” Page 1 of this document follows:
PROJECT SUMMARY
Approval is requested for an investment of $4.6MM in new equipment for our Division’s rock bit manufacturing plants. The projected return on investment is 60.1%
This investment is required for two distinct reasons:
—Expansion
Equipment investment $2.8MM — R.O.I. 67.1%
[1558]*1558To expand productive capability to meet current and projected Marketing demand for sealed bearing bits, blasthole bits, and drilling tool cutters.
—Quality and Cost
Equipment investment $1.8MM — R.O.I. 27.4%
To replace existing old equipment with modern, high precision machine tools in order to meet the exacting specifications of today’s high-technology bits and thereby establish a uniform quality level of our products.
To reduce variable production cost, curtail work-in-process inventories and lead-time, and retain and enhance our position as low cost producer in the industry.
The project is split into three separate appropriation requests, each one of which is an integrated package:
AR 73-124A — Illinois Avenue Expansion
AR 73-124B — Manchester Expansion
AR 73-124C — Illinois Avenue Cost Reduction
The machine tools requested in each individual A.R. are complementary in the production process. Segregation of individual tools out of an A.R. will seriously impair the meeting of the objectives set forth.
New investment in buildings or land is not required.
The testimony of Dresser’s controller, Mr. Munnerlyn, and Hughes Tool’s Chief Accounting Officer, Mr. Willis, is consistent with the above evidence that the expected profit on incremental sales would yield a 60% return on incremental investment. There is no evidence to the contrary.3
Thus, this case closely parallels Georgia-Pacific Corp. v. U.S. Plywood-Champion Papers, Inc., 446 F.2d 295, 170 USPQ 369 (2d Cir.), cert. denied, 404 U.S. 870, 92 S.Ct. 105, 30 L.Ed.2d 114 (1971). The question in Georgia Pacific was the reasonableness of a royalty of $50 per thousand square feet of certain plywood sales. There, the district court (as did the district court here) accepted that a reasonable royalty must be fixed so as to leave the infringer a reasonable profit. Georgia Pacific Corp. v. United States Plywood Corp., 318 F.Supp. 1116, 1122, 166 USPQ 235, 239 (S.D.N.Y.1970). To calculate the royalty, the Georgia Pacific trial court erroneously used a figure of $159.41, which represented Georgia Pacific’s average “realization” on plywood sales [possibly margin], rather than a figure of $50.00, which represented Georgia Pacific’s expected profit on such sales. Thus, the award of a $50.00 royalty, which would leave no profit, was set aside on appeal as excessive. 446 F.2d at 299, 170 USPQ at 372.
The royalty of 25% set in this case, being based on a clearly erroneous finding of projected profits of 60% as a percent of sales, is ipso facto arbitrary and must be set aside.
C. Prejudgment Interest
An award of prejudgment interest to a patent owner for the period during which his patent was declared invalid is appropriate if, after exercising its discretion, the district court decides that such an award is necessary to put the patent owner in as good a position as he would have been in had the infringer entered into a reasonable royalty agreement when the infringement began. The district court here decided not to award prejudgment interest for the time during which the ’928 patent was considered invalid and unenforceable. The court’s decision must be reviewed for abuse of discretion. But to the extent that decision was based on the misunderstanding that prejudgment interest can never be awarded in such a circumstance, it is contrary to law and should be reconsidered on remand. See General Motors Corp. v. De[1559]*1559vex Corp., 461 U.S. 648, 657, 103 S.Ct. 2058, 2063, 76 L.Ed.2d 211, 217 USPQ 1185, 1188 (1983).
Lastly, we affirm the trial court’s May 2, 1985, denial of Dresser’s post-trial motions.
Conclusions
The judgment of the district court with respect to validity is affirmed. The judgment on damages is vacated and the case is remanded for a redetermination of damages based on a reasonable royalty computed in accordance with this opinion and a reconsideration of the amount of prejudgment interest.
Dresser’s Post-hearing Motion
On November 24, 1986, ten months after this case was heard, Dresser filed a Motion to Remand, to Vacate, or to Reopen the Record, or Obtain Relief from Judgment because of a document it became aware of contained in an appendix in a case, Smith International, Inc. v. Hughes Tool Co., Nos. 86-1139, 86-1125, see note 1, supra, which was argued in this court in December of 1986. Dresser argues that this newly discovered evidence has a material effect on the validity of the ’928 patent because it indicates that Hughes Tool was still searching, in 1974, for a bit that would operate efficiently at high RPM. Dresser asserts that this document was withheld. Hughes Tool retorts that it twice provided the document but that Dresser never requested that it be copied, and further that the document does not affect the outcome of this case. Given the factual disputes bound up with this motion, we cannot, as an appellate tribunal, pass on its merits and therefore deny Dresser’s motion.
AFFIRMED IN PART, VACATED IN PART, AND REMANDED.