FRIEDMAN, Circuit Judge.

These are consolidated appeals from a judgment of the United States District Court for the District of Connecticut, determining the damages in a patent infringement suit. The infringer contends that the district court improperly based a portion of the damages upon the lost profits of the patentee instead of a reasonable royalty. The patentee contends that the district court erred in (1) limiting prejudgment interest to the portion of the damages based upon a reasonable royalty and denying prejudgment interest on the lost profits portion of the damages, and (2) awarding simple interest at the statutory rate rather than compound interest at the market rate. We reverse the denial of prejudgment interest on the lost profits portion of the damages, and affirm the other aspects of the judgment.

I

This case began in June 1972, when Champion Spark Plug Company (Champion) filed suit in the district court seeking a declaratory judgment that United States Patent No. 3,219,276 (the '276 patent) of Gyromat Corporation (Gyromat) was invalid and unenforceable. Gyromat responded by filing a counterclaim alleging that Champion had infringed the patent and seeking damages and injunctive relief.

The patent covers industrial paint spraying machines, in which paint is atomized and sprayed on the surface to be painted. There are various types of industrial paint sprayers currently in use, all having differing costs of operation and operating characteristics, such as the efficiency with which the paint supply is transferred to surfaces of various contours.

Claims 5 and 6 of the patent (the only claims involved in the case) define a spraying machine in which a spray nozzle is mounted on a frame that moves up and down and is supplied with paint from a stationary paint supply. They further recite a specific regulator mounted on the frame which maintains the pressure of the paint supply at the nozzle. The adjustment of the regulator is in turn controlled by a stationary pressure control valve, which permits the pressure at the nozzle to be changed without having to shut down the system. This is known as a reciprocating system.

Although all of Champion's infringing devices involved air atomizing machines, they were of two distinct types. Most of the machines were so-called short-stroke machines which contain two nozzles, mounted one above the other, which move up and down in short strokes. In a short-stroke machine, the spray nozzles are placed in positions that enable the device to paint the entire vertical area evenly, with a proper overlap in the area sprayed by the adjoining nozzle. In the other type, the so-called long-stroke machine, each frame contains a nozzle that moves up and down the entire height of the work and sprays paint on the entire vertical surface to be covered.

Before the district court, Champion conceded that if the '276 patent was valid, Champion had infringed it. After a trial, the district court held that claims 5 and 6 were invalid as obvious under 35 U.S.C. § 103 (1976). The Court of Appeals for the Second Circuit reversed. Champion Spark Plug Co. v. Gyromat Corp., 603 F.2d 361, 202 USPQ 785 (1979), cert. denied, 445 U.S. 916, 100 S.Ct. 1276, 63 L.Ed.2d 600 (1980). The court held the patent valid and remanded the case to the district court.

Following the remand, the district court in 1980 enjoined Champion from infringing claims 5 or 6. It then referred the case to a special master to hear evidence and report to the court on the amount of damages and the amount of a reasonable attorney's fee, if any.

After a three-day hearing, the master rendered a lengthy report to the district court accompanied by proposed findings of fact and conclusions of law. Both parties filed exceptions to the report. In response, the master wrote a detailed letter to the court in which he answered the objections and amplified the reasons for his decision.

The district court "thoroughly reviewed this matter at each stage, both upon receipt of the original Report of the Master, and then upon receipt of the objections filed by the parties, and finally upon receipt of the Master's final report." The court "ACCEPTED, RATIFIED and AFFIRMED" the master's report, findings of fact and conclusions of law "as the ruling of this court . . ." The district court entered final judgment in accordance with the master's recommendations.

The master made the following determinations:

1. Gyromat was entitled to recover its lost profits on the sale by Champion of infringing short-stroke machines. This conclusion rested upon the master's findings that demand existed for the patented product, that because Gyromat and Champion were the only two suppliers of such machines, "[n]on-infringing substitutes were not readily and reasonably available during the accounting period," that Gyromat could have made and sold the infringing systems, and that Gyromat had established the amount of profit it lost on those sales. He calculated Gyromat's lost profits on these sales at $747,944.

2. Since there was an acceptable non-infringing substitute available for the long-stroke system incorporating the patented controls, for the sale of these infringing devices Gyromat was entitled only to receive a reasonable royalty. The master set this royalty at 17.5 per cent of the selling price of the machines, for a total of $18,200.

3. Gyromat was entitled to prejudgment interest from the dates of infringement, but only for the reasonable royalty portion of the award and not on the lost profits portion. In his subsequent letter to the district court commenting on the parties' exceptions to his report, the master stated that there was "some support in the cases" Gyromat cited for Gyromat's argument that "no distinction between lost profits and reasonable royalty awards is required with respect to prejudgment interest." The master, however, concluded that his limitation of prejudgment interest to reasonable royalties "is consistent with the current weight of authority."

4. The interest rate to be applied was "the statutory interest rate in Connecticut." The master declined to award compound interest at the market rate because "there seems to be little case support" for doing so.

5. This is not an exceptional case under 35 U.S.C. § 285 warranting attorney's fees and treble damages (except for sales Champion made after the injunction, with respect to which the master trebled the damages and awarded Gyromat its attorney's fees for bringing a contempt proceeding).

II

In awarding damages for lost profits, the master applied the four-part test of Panduit Corp. v. Stahlin Bros. Fibre Works, Inc., 575 F.2d 1152, 197 USPQ 726 (6th Cir. 1978), which this court implicitly approved as a method of proving damages in Central Soya v. George Hormel Co., 723 F.2d 1573, 220 USPQ 490 (Fed. Cir. 1983). The master stated that to recover under that theory Gyromat was required to establish:

1. That there be a demand for the product during the period in questions [sic].

2. That during the period there was an absence of acceptable non-infringing substitutes.

3. That Gyromat had the manufacturing and marketing capability to meet the demand for the product covered by the patent.

4. Gyromat must establish the profit it would have made.

Champion does not challenge the master's use of the Panduit test as a basis for determining the existence of lost profits. It argues only that the record does not support the master's factual findings upon which he based his award of lost profits.

Champion's brief and reply brief consist mainly of a presentation of the evidence supporting its position. Champion, however, largely ignores the contrary evidence upon which the master relied. It complains that the master improperly credited and placed too great reliance upon what it characterizes as the self-serving testimony of Gyromat's president, Mr. Wiggins. The credibility of the witnesses and the weight to be given to their testimony and the other evidence in the record, however, is a matter for the trier of the facts. We have no basis for concluding that the master improperly gave too great weight to Gyromat's evidence or too little weight to Champion's.

Based upon our review of the entire record, we cannot say that the master's factual determinations upon which he based his award of lost profits were clearly erroneous. We discuss those findings and the supporting evidence with respect to the four Panduit factors.

1. The master found that "[t]he sales of Champion and Gyromat during the period in question clearly establish the demand for the painting system covered by the patent in suit." He further pointed out that Champion was on notice from December 1965 of Gyromat's contention that Champion's control system infringed the patent in suit, and that Champion's decision to risk infringement liability indicates the value it placed on the patented features.

Champion contends that the evidence "proved that the marketability and demand for the [Champion] products are not dependent upon the presence of [the patented features]." It relies in large part on the successful sale of noninfringing systems it introduced after the injunction.

The substantial number of sales by Champion of infringing products containing the patented features itself is compelling evidence of the demand for the product. See, e.g., Bros Inc. v. W.E. Grace Manufacturing Co., 320 F.2d 594, 138 USPQ 357 (5th Cir. 1963); Transit Development Co. v. Cheatham Electric Switching Device Co., 194 F. 963, 967 (2d Cir. 1912). Champion's sales necessarily meant that there were buyers who wanted the product and were willing to pay Champion's price, which was substantially the same as that of Gyromat. As the master observed.

[t]he patented control features were advertised by Champion and while Champion has shown that painting systems could be made and sold without the patented features, the patented control system was obviously important enough to keep for 15 years on all of its short-stroke reciprocating painting systems. If there was no demand for the patented system, Champion would not have run the risk of infringement.

2. In determining the availability of alternate sources for the product, the master found that only air atomized, short-stroke systems should be considered because "other paint spray systems had different characteristics and were more expensive than the Gyromat/Champion short-stroke systems." After discussing the products of each of the electrostatic spray equipment manufacturers alleged by Champion to be its competitors, the master found that "Champion and Gyromat were the only two manufacturers regularly offering short-stroke reciprocators to the users of industrial paint sprayers" and that no acceptable non-infringing substitutes for this type of equipment were readily available.

Champion first argues that because claims 5 and 6 of the '276 patent do not refer either to long- or short-stroke systems, it was error to limit acceptable substitutes to the short-stroke products, and that since other manufacturers supplied long-stroke systems, the court's finding of the lack of available substitutes cannot stand.

This contention is irrelevant. The critical inquiry in determining whether to award lost profits is whether Gyromat has shown a reasonable probability that it would have made the infringing sales that Champion made. Lam, Inc. v. Johns-Manville Corp., 718 F.2d 1056, 219 USPQ 670 (Fed. Cir. 1983); Milgo Electronics Corp. v. United Business Communications, Inc., 623 F.2d 645, 206 USPQ 481 (10th Cir.), cert. denied, 449 U.S. 1066, 101 S.Ct. 794, 66 L.Ed.2d 611 (1980); Livesay Window v. Livesay Industries, Inc., 251 F.2d 469, 116 USPQ 167 (5th Cir. 1958). The master's finding that no acceptable noninfringing substitutes were available is not undermined by, but is fully consistent with, the fact that long-stroke systems also infringed the patent.

Champion next attacks the finding on the ground that the evidence clearly established that the long-stroke systems were in direct competition with Champion's infringing system.

The record supports the master's contrary finding. Mr. Wiggins testified that the initial cost of a short-stroke system such as Champion's was significantly lower than other systems. Mr. White, Champion's manager of engineering, testified as follows:

Q: And in selecting various kinds of applicators for a particular job, what parameters would be relevant?

A: Number one, I'd have to look at the competition, the budget that the customer has to spend. We'd look awfully silly if we quoted a rotational atomizer system, which is most efficient in the marketplace today, to a customer who only had a few thousand dollars to spend. We'd lose the job.

We'd look awfully silly if we're competing with Gyromat with a rotational atomizer . . . .

* * * * *

Q: I take it that's because rotational atomizer systems are more expensive?

A: Yes it is.

THE MASTER: By a factor of about how much?

THE WITNESS: Possibly fifteen percent.

This evidence supports the master's finding that "the low pressure short-stroke systems of Champion and Gyromat cost substantially less to install than other types of electrostatic paint systems . . . . If a customer for low pressure short-stroke painting systems with the control system covered by the '276 patent had been unable to purchase [such a system], he would have had to pay more for a system that may not have been as adaptable to his needs." These findings, together with Wiggins' testimony that other systems had different operating characteristics and were inferior to air atomizing equipment for certain work, support the master's finding that only air atomized, short-stroke systems were acceptable noninfringing substitutes.

Champion challenges the master's finding that "Champion and Gyromat . . . were substantially the only sources of short-stroke reciprocators" on the ground that "the Ransburg Corporation [a major manufacturer of paint spraying systems] will produce and sell air atomized vertically reciprocating paint sprayers having fixed regulators upon customer request."

The master rejected Ransburg as an alternative supplier because of testimony that it only rarely sold air atomized, short-stroke systems. Champion's witness, Mr. Kidd, a former employee of Ransburg, testified that although he was aware of more than 2,700 paint spraying systems installed by Ransburg, he had actual knowledge of only four or five installations of the air atomized, short-stroke variety. Mr. Wiggins corroborated this testimony, stating that Ransburg was involved with bell and disk-type systems.

3. The master relied upon testimony by Mr. Wiggins, president of Gyromat, and an accounting exhibit detailing the pattern and substance of Champion's infringing sales, as a basis for finding that "during the accounting period Gyromat would have had the manufacturing and marketing capability to make and sell the air atomizing, short stroke paint spraying systems sold by . . . Champion."

In challenging this finding, Champion points to testimony purporting to establish that Gyromat did not have a substantial sales and advertising budget, did not employ any direct salesmen, and was unaware of the sales techniques necessary to obtain orders from large corporate buyers. Champion also stresses that Gyromat bid against it on only seven of the 152 infringing sales.

Mr. Wiggins, however, testified that Gyromat produced about 160 short-stroke machines during the period of infringement (1965 to 1980), which was approximately 40 percent of the number of short-stroke machines Champion sold during that period. He stated that not only could Gyromat itself have handled the increased production, but that substantial portions of the work could have been subcontracted without serious adverse effect on Gyromat's over-all profit margin. This and other evidence in the record supports the master's finding that "Gyromat, throughout the period of infringement by Champion, could have supplied the needs of the marketplace for short-stroke electrostatic remote controlled reciprocators."

Champion's argument that Gyromat would have been unable to sell to Champion's customers is unconvincing. The record establishes that Gyromat had substantial sales, and it was known in the market. Mr. Wiggins testified that Gyromat's equipment was used by several divisions of General Motors, and the master specifically found that "there are relatively few manufacturers in the electrostatic paint spray field, and Champion's customers could have readily found Gyromat as a source of the [patented device]." The fact that Gyromat bid against Champion on only seven of the 152 infringing sales does not show that Gyromat could not and would not have made those sales if Champion had not infringed. "It is impossible . . . for the patent owner to negate every possibility that the purchaser might not have bought another product." Milgo, 623 F.2d at 663, 206 USPQ at 495. See also American Hoist Derrick v. Sowa Sons, Inc., 725 F.2d 1350, 220 USPQ 763, 775 (Fed. Cir. 1984).

4. In determining the profits Gyromat would have made on the sale of the infringing products, the master relied upon a study by an independent accounting firm projecting Gyromat's estimated profits if it had sold the infringing products, calculated on a year-to-year basis, and the testimony of Gyromat's own accountant. Champion argues that the accounting analysis was an unreliable basis for establishing the amount of Gyromat's estimated profits. It relies primarily on the testimony of its witness, Mr. Herron, who disputed the reliability of the accounting methods employed when applied to the timespan of 15 years and characterized the resulting profit margin analysis as "extremely optimistic" and "totally unreasonable."

"[W]hen the amount of the damages cannot be ascertained with precision, any doubts regarding the amount must be resolved against the infringer." Lam, 718 F.2d at 1065, 219 USPQ at 675. On the basis of the evidence in the record, the master's finding of the amount of lost profits is not clearly erroneous. The accounting exhibit was prepared by an outside accounting firm, and was supported by the testimony of Mr. Lisy, Gyromat's own accountant. Moreover, Mr. Herron admitted that the analysis was proper on a year-by-year basis.

"In proving his damages, the patent owner's burden of proof is not an absolute one, but rather a burden of reasonable probability." Id; see also Livesay, supra. The master's findings that Gyromat met that burden here are not clearly erroneous.

III

In limiting prejudgment interest to the royalty segment of the damages, the master recognized that there was "some support in the cases" for Gyromat's position that such interest also should cover lost profits. He concluded, however, that his more limited award of interest was "consistent with the current weight of authority."

Subsequent to the district court's judgment awarding damages on the basis of the master's report, the Supreme Court decided General Motors Corp. v. Devex Corp., 461 U.S. 648, 103 S.Ct. 2058, 76 L.Ed.2d 211 (1983). The principles announced in that case call for payment of prejudgment interest on the lost profits portion of the damages awarded to Gyromat.

In Devex, the Supreme Court affirmed the judgment of the United States Court of Appeals for the Third Circuit upholding, under the statute governing damages in patent infringement cases, 35 U.S.C. § 284 (1982), a district court award of prejudgment interest on damages based upon royalties. The court of appeals sustained the award on the ground that it was not an abuse of discretion, and the Supreme Court affirmed on that basis.

In so holding, however, the Court stated that "prejudgment interest should be awarded under § 284 absent some justification for withholding such an award," repeated the same statement without reference to section 284, and twice stated without the "justification for withholding" qualification that prejudgment interest "should ordinarily be awarded." 103 S.Ct. at 2062-2064. The Court further stated:

We do not construe § 284 as requiring the award of prejudgment interest whenever infringement is found. That provision states that interest shall be "fixed by the court," and in our view it leaves the court some discretion in awarding prejudgment interest. For example, it may be appropriate to limit prejudgment interest, or perhaps even deny it altogether, where the patent owner has been responsible for undue delay in prosecuting the lawsuit. There may be other circumstances in which it may be appropriate not to award prejudgment interest. We need not delineate those circumstances in this case.

Id. at 2063-2064, footnote omitted.

Champion, relying on the second sentence of this passage, argues that under Devex the award of prejudgment interest is within the discretion of the district court and that the district court here did not abuse its discretion in denying such interest on the lost profits portion of the award. We do not read the Supreme Court's decision as announcing the principle that Champion discerns in it. To the contrary, the Court's repeated statements that prejudgment interest "should ordinarily be awarded" indicates that that is the governing principle the Court enunciated. This conclusion is further supported by the Court's statement at the beginning of its opinion that "[t]his case concerns the proper standard governing the award of prejudgment interest in a patent infringement suit under 35 U.S.C. § 284" and its explanation that it "granted certiorari to consider the standard applicable to the award of prejudgment interest under 35 U.S.C. § 284." at 2060.

Indeed, this court recently recognized, citing Devex, that "[p]rejudgment interest is typically included as part of the patentee's recovery to ensure compliance with the statutory mandate of 35 U.S.C. § 284 that damages be `adequate to compensate for the infringement.'" Stickle v. Heublein, Inc., 716 F.2d 1550, 219 USPQ 377 (Fed. Cir. 1983). See also Deere Co. v. International Harvester Co., 710 F.2d 1551, 218 USPQ 403 (Fed. Cir. 1983).

The fact that Devex involved prejudgment interest on royalties and the present case involves prejudgment interest on lost profits does not justify a different result here. The Court pointed to "Congress' overriding purpose [in § 284] of affording patent owners complete compensation" and explained why an award of prejudgment interest generally is necessary to accomplish that objective:

In the typical case an award of prejudgment interest is necessary to ensure that the patent owner is placed in as good a position as he would have been in had the infringer entered into a reasonable royalty agreement. An award of interest from the time that the royalty payments would have been received merely serves to make the patent owner whole, since his damages consist not only of the value of the royalty payments but also of the foregone use of the money between the time of infringement and the date of the judgment.

Id. at 2062-2063, footnote omitted. In light of these policy considerations, there is no convincing reason to treat the two types of damages differently.

Indeed, the principal case upon which the Supreme Court relied in Devex, Waite v. United States, 282 U.S. 508, 51 S.Ct. 227, 75 L.Ed. 494 (1931), was a patent infringement suit against the United States in which the Court of Claims had denied prejudgment interest on damages based on lost profits. The Supreme Court reversed on the ground that "an award of prejudgment interest to the patent owner was necessary to ensure `complete justice as between the plaintiff and the United States,' id., 282 U.S. at 509, 51 S.Ct. at 227, even though the statute governing such suits did not expressly provide for interest." Devex, 103 S.Ct. at 2063. Taken together, Devex and Waite indicate that ordinarily prejudgment interest should be awarded on both the lost profits and the royalty portions of the damages awarded for patent infringement.

Champion argues that this case falls within the exception that Devex recognized for special circumstances that would make an award of prejudgment interest inappropriate. Without attempting to define or suggest the boundaries of such circumstances, we conclude that nothing in this case warrants a denial of prejudgment interest on the lost profits. Indeed, Champion's argument on this aspect of the case is inconsistent with its failure to appeal from the award of prejudgment interest on the royalty segment of the damages. If there were something in this case that would make an award of prejudgment interest inappropriate, it would apply to both segments of the damage award and not just to the lost profits portion.

IV

The final issue in this case is whether the master improperly determined that prejudgment interest is to be calculated as simple interest at the state statutory rate rather than, as Gyromat urges, as compound interest at the market rate. The master rejected Gyromat's argument because "there seems to be little case support" for it.

This court has recognized that the district court has substantial discretion to determine the interest rate in patent infringement cases. Railroad Dynamics, Inc. v. A. Stucki Co., 727 F.2d 1506, 220 USPQ 929 (Fed. Cir. 1984); see also Lam, supra. In Railroad Dynamics, we upheld a district court's award of interest at 6 percent; in Lam, we upheld an award of interest at the prime rate. In both cases, the basis of our decision was that it had not been shown that the district court abused its discretion in setting the rate.

We conclude that the determination whether to award simple or compound interest similarly is a matter largely within the discretion of the district court. In Railroad Dynamics we upheld, as not an abuse of discretion, 6 percent compound interest. Gyromat's argument on this point would in effect require as a matter of law that prejudgment interest always be compounded. We decline to create such a rule.

Gyromat relies upon the recent amendment of 28 U.S.C. § 1961(b) (1982) that provides for compounding of interest upon federal district court judgments. It urges us to follow the lead of Congress in that amendment and to announce the same principle for prejudgment interest. Properly viewed, however, that provision cuts against, rather than supports, Gyromat's position.

The fact that Congress restricted the award of compound interest to judgments itself strongly suggests that the legislature did not intend that there should be the same rule for prejudgment interest. This is confirmed by the legislative history of the provision, which shows that Congress failed to adopt a proposal that would have provided for prejudgment interest on a compound basis.

The amendment of 28 U.S.C. § 1961(b) was part of the Federal Courts Improvement Act of 1982, Pub.L. No. 97-164, § 302, 96 Stat. 55-56. The Senate bill, S. 1700, originally authorized the district court to award prejudgment interest where such an award is appropriate to afford the prevailing party complete relief (section 302(b)(1)) and provided that such prejudgment interest "shall be compounded annually" (section 302(c)). This was one of several provisions dealing with interest.

The provision was eliminated by a floor amendment in the Senate. Senator Grassley, who submitted the amendment, stated that the subject of prejudgment interest

was not addressed in testimony or evidence submitted to the courts subcommittee during the course of its hearings on S. 1700 this year, but was a carryover from prior versions of the bill. In response to the concerns expressed by OMB and on further examination by committee staff, it seemed that this matter should be given more thorough attention before enactment into law.

Thus, the entire prejudgment interest provision has been stricken by this amendment. . . . It is anticipated that the subcommittee will consider the question in the coming months so as to be able to explore all facets of the matter and give it careful consideration prior to any further legislative activity.

127 Cong.Rec. S14699 (daily ed. Dec. 8, 1981).

This legislative history shows that in amending 28 U.S.C. § 1961(b), to provide for compound interest on judgments, Congress intended to take no action with respect to prejudgment interest, but to leave that matter for another day. The provision for compound interest on judgments in section 1961(b) provides no basis for compounding prejudgment interest.

Since the Courts Improvement Act, Congress has not enacted any legislation dealing with prejudgment interest. Unless and until Congress provides for the compounding of such interest, we leave it to the informed discretion of the district courts to decide in each case whether it is appropriate to compound prejudgment interest.

Considering all the circumstances of this case, we cannot say that the district court abused its discretion in computing prejudgment interest on the basis of the simple statutory rate, not compounded.

CONCLUSION

The judgment of the district court is affirmed insofar as it awarded damages for lost profits and simple interest at the statutory rate on the reasonable royalties portion of the award. The judgment is reversed insofar as it refused to award prejudgment interest on the lost profits portion of the damages, and the case is remanded to award such interest in accordance with our opinion.

AFFIRMED in part, REVERSED in part, and REMANDED.