OPINION
ANDEWELT, Judge.In this patent action filed pursuant to 28 U.S.C. § 1498, plaintiff, Albert L. de Graffenried, seeks compensation from the United States for the unauthorized use of a device covered by certain claims of plaintiff’s patent, United States Patent No. 3,217,568 (the ’568 patent). In a prior decision, this court held that (1) the pertinent claims of the ’568 patent are valid, (2) those claims cover a controller system used in the guided boring of large gun barrels at the Watervliet Arsenal in Watervliet, New York (the Arsenal), and (3) plaintiff’s suit is not barred by the doctrine of laches. De Graffenried v. United States, 20 Cl.Ct. 458 (1990) (De Graffenried I). The sole remaining issue is the amount of compensation to which plaintiff is entitled. This decision sets the parameters of plaintiff’s compensation award.
I.
Section 1498 provides, in pertinent part:
(a) Whenever an invention described in and covered by a patent of the United States is used or manufactured by or for the United States without license of the owner thereof or lawful right to use or manufacture the same, the owner’s remedy shall be by action against the United States in the United States Claims Court for the recovery of his reasonable and entire compensation for such use and manufacture.
To determine “reasonable and entire compensation” under Section 1498, the court must determine first which machines used at the Arsenal should be part of the calculus when assessing compensation and second what is the appropriate level of compensation for the unauthorized use of those machines.
Plaintiff contends that he is entitled to compensation for the Arsenal’s use of twelve different guided boring lathes that employ a controller system covered by the ’568 patent. Defendant disagrees and contends that compensation should be allowed only for the following four:
Identification Size of Gun Barrel Date of First Use in the
Number Produced Manufacture of Gun Barrels
WV 11235 155 mm December 1974
WV 11205 105 mm June 1976
WV 11225 8" January 1977
WV 11190 105 mm October 1978
*212The remaining eight guided lathes in dispute can be divided into three groups. The first group consists of two lathes, referred to by the parties as the Bergen device and the modified Bergen device. These lathes were developed in part by Arsenal personnel and were used in assessing whether guided boring was a viable technology. The second group consists of the following two lathes:
Identification Size of Gun Barrel Date of First Use in the
Number Produced Manufacture of Gun Barrels
WV 2488 175 mm November 1967
WV 11187 175 mm October 1972
The third group consists of four lathes that were ordered from Cincinnati Milacron, Inc., prior to the expiration of the ’568 patent but were not fully assembled by Cincinnati Milacron or delivered to the Arsenal until after the ’568 patent had expired on November 16, 1982.
Defendant contends that no compensation is appropriate for the first group of lathes for a series of reasons, one of which is that any such claim for compensation would be barred by the statute of limitations. For the second group, defendant relies exclusively upon a statute of limitations bar. As to the third group, defendant contends that the prerequisites for government liability under Section 1498 did not occur until after the ’568 patent had expired and, therefore, no compensation is warranted.
II.
As explained above, the first group consists of two guided lathes, the Bergen device and the modified Bergen device. As to the Bergen device, the trial evidence indicates that the Arsenal was no longer using the device on the date the ’568 patent was issued. Since the United States did not manufacture or use the device during the time the device was “covered by a patent” (i.e., during the patent term), compensation is not available under Section 1498.
Turning to the modified Bergen device, the Arsenal did use that device during the ’568 patent term, but only for about four months and only to bore a small number of gun tubes. Defendant contends that this limited use should be considered de minimis and therefore should not be subject to compensation under Section 1498. But Section 1498 mandates reasonable and entire compensation “whenever” (emphasis added) there is an unauthorized manufacture or use of a patented invention by or for the United States. Section 1498 does not contain any exception for short-term use of a patented invention. Hence, in each case, the court must engage in a factual inquiry to determine what constitutes “reasonable and entire compensation.” See, e.g., Decca, Ltd. v. United States, 225 Ct.Cl. 326, 336, 640 F.2d 1156, 1167 (1980), cert. denied, 454 U.S. 819, 102 S.Ct. 99, 70 L.Ed.2d 89 (1981). As explained in Section VI, infra, sometimes that inquiry results in compensation being based on the amount or the nature of the use of the patented invention, but sometimes it does not. The court cannot exclude the possibility of compensation for limited use without first undertaking the required factual inquiry.
III.
Next, as to all guided lathes in the first and second groups, defendant contends that no compensation is available because any claim based on the Arsenal’s use of these lathes would be barred by the statute of limitations. Defendant argues that since the instant action was filed more than six years after the Arsenal first began using all of the devices in the first and second groups, suit on each device is *213barred by the six-year statute of limitations applicable to Claims Court actions.1
Defendant is correct that, typically, the date of the first use of a patented invention is controlling when determining whether a claim filed under Section 1498 is barred by the statute of limitations. In Starobin v. United States, 229 Ct.Cl. 67, 71, 662 F.2d 747, 749 (1981), the Court of Claims explained that-under Section 1498, a distinct claim arises for each patented device used by the United States, and as to each such device, only one right to recovery against the government arises, i.e., each additional use of the device does not give rise to a new cause of action. Hence, a cause of action under Section 1498 can be brought upon the government’s first use of the patented invention, and, typically, the Claims Court statute of limitations under 28 U.S.C. § 2501 would start to run on that date. See id.
But this is not a typical case. In Japanese War Notes Claimants Ass’n of the Philippines, Inc. v. United States, 178 Ct.Cl. 630, 373 F.2d 356, cert. denied, 389 U.S. 971, 88 S.Ct. 466, 19 L.Ed.2d 461 (1967), the Court of Claims explained that under specified circumstances, the statute of limitations on a claim would not begin to run until the aggrieved party reasonably should have known about the existence of a claim. The court stated:
Plaintiff must either show that defendant has concealed its acts with the result that plaintiff was unaware of their existence or it must show that its injury was “inherently unknowable” at the accrual date____ In this situation the statute will not begin to run until plaintiff learns or reasonably should have learned of his cause of action.
Id. at 634, 373 F.2d at 359 (footnote omitted). See also Hopland Band of Pomo Indians v. United States, 855 F.2d 1573, 1577 (Fed.Cir.1988).
The Japanese War Notes prerequisites for delaying the commencement of the running of the statute of limitations are present here. The pertinent facts were presented by the parties during trial and were discussed to some extent in that portion of this court’s prior opinion discussing defendant’s laches defense. De Graffenried I, 20 Cl.Ct. at 486-90. The most reasonable inference to draw from the trial evidence, and the one the court draws here, is that government personnel deliberately concealed from plaintiff material facts concerning the government’s use of guided boring devices. Inter alia, Arsenal personnel “represented that there had not been any change in the facts upon which defendant based its denial of plaintiff’s 1968 administrative claim even though, in fact, the Arsenal had developed a guided boring device which already had been used to control the deep boring of an object rotating about an axis.” Id. at 488. In addition, the government cancelled a previously planned tour of the Arsenal which would have disclosed the structure of the controller device used at the Arsenal.
Defendant contends that even assuming the government did conceal material facts, the statute of limitations commenced to run at least by 1971, because by that time plaintiff reasonably should have learned of his cause of action. Defendant relies upon a letter to plaintiff dated May 28, 1971, from William A. Folsom, an associate of a small company of which plaintiff was a principal. But, as the court explained in De Graffenried I:
[W]hile this reference indicates that in 1971 plaintiff knew there were runout systems being used at the Arsenal, there is no evidence in the record indicating that plaintiff had direct knowledge at that time, or any time prior to 1978, of the physical structure of the controllers used at the Arsenal. Plaintiff could not conclude that his patent was being infringed simply from the Arsenal’s use of a runout controller. Plaintiff originated and patented one possible concept for controlling runout, and plaintiff was not in a position to conclude that any runout controller developed by the Arsenal *214would necessarily or even likely be covered by his patent.
Id. at 488-89.
Clearly, notwithstanding defendant’s actions, if plaintiff had sufficient information suggesting that the Arsenal was using his patented device, the statute of limitations would have commenced to run and plaintiff would have been obliged to commence suit to determine the remainder of the facts. But here, there was no significant evidence suggesting to plaintiff that the Arsenal device was covered by the ’568 patent. As the court explained in its prior discussion of laches:
[P]laintiff took all of the steps reasonably possible, short of instituting suit, to determine whether the Arsenal was being forthright in its allegations of non-infringement. The structure of the Arsenal device could not be determined by examining a finished gun barrel. Therefore, to determine whether the controllers used at the Arsenal were covered by his patent, plaintiff had to inspect the actual Arsenal device or blue prints depicting its structure. But the Arsenal is a restricted area and plaintiff could not make such an inspection without the Arsenal’s authorization. Plaintiff had arranged for an inspection on February 2, 1967, but when plaintiff and his counsel arrived at the Arsenal, the previously granted authorization was revoked. Another request had been denied in 1968.
Id. at 489.
Based upon the trial record, the court cannot conclude that plaintiff reasonably should have known about the use of his patented invention at the Arsenal prior to 1978. Hence, the statute of limitations does not bar plaintiff’s claim for compensation based on the Arsenal’s use of the guided lathes in the first two groups.
IV.
Turning to the third group of guided lathes used at the Arsenal, the four lathes manufactured by Cincinnati Milacron were initially ordered by the Arsenal during the term of the ’568 patent but were not delivered to the Arsenal until early 1984, more than one year after the ’568 patent had expired. It is well established that the mere entry into a contract to purchase a device covered by a patent does not itself create liability under 28 U.S.C. § 1498(a). See, e.g., Tektronix, Inc. v. United States, 216 Ct.Cl. 144, 151, 575 F.2d 832, 836, cert. denied, 439 U.S. 1048, 99 S.Ct. 724, 58 L.Ed.2d 707 (1978). Section 1498 reaches only “use or manufacture” of a patented device by or for the United States, and the execution of a contract to manufacture a device is not the same as the actual manufacture. Id.
Plaintiff contends that the production of one of the four machines was sufficiently far along when his patent expired so that the machine should be considered to have been “manufactured” by Cincinnati Milacron for the Arsenal during the patent term. But the court concludes to the contrary. The general rule is that until a device covered by a patent is actually assembled, the device has not been “manufactured.” Deepsouth Packing Co. v. Laitram Corp. 406 U.S. 518, 528, 92 S.Ct. 1700, 1707, 32 L.Ed.2d 273 (1972); see also Decca, 225 Ct.Cl. at 337, 640 F.2d at 1168. In Paper Converting Machine Co. v. Magna-Graphics Corp., 745 F.2d 11, 19 (Fed.Cir.1984), the court articulated what amounts to a narrow exception to that general rule. But that narrow exception is not broad enough to encompass plaintiff’s claim herein.
In Paper Converting, Magna Graphics had contracted with a company to deliver a particular device. But prior to manufacture and delivery, a district court held that Magna Graphics had infringed Paper Converting’s patent by a previous sale of a similar device to a third party. The district court enjoined Magna Graphics from any future infringement. In an effort to fulfill its obligations under its pending contract without violating the injunction against infringement, Magna Graphics manufactured and then separately tested the two major subassemblies of the device to assure that the device would work. The subassemblies were then delivered to the customer more than six months before the patent expired, *215but, pursuant to an agreement, the customer did not assemble the machine until after the patent expired. The court labeled Magna Graphics’ plan a “scheme to avoid patent infringement,” id. at 15, and upheld a finding of infringement. The court stated:
Given the amount of testing performed here, coupled with the sale and delivery during the patent-term of a “completed” machine (completed by being ready for assembly and with no useful noninfringing purpose), we are not persuaded that the district court committed clear error in finding [infringement].
Id. at 19.
Herein, plaintiff has not demonstrated any analogous “scheme to avoid infringement.” Cincinnati Milacron did not deliver the guided boring machines in unassembled form during the patent term and there is no evidence suggesting that either completion of the manufacture of the machines or the date the machines were first used by the Arsenal was influenced in any way by the expiration of the patent. While Cincinnati Milacron did manufacture parts of the final product — a boring bar and a boring head — prior to the expiration of the patent, the bar was not yet ready to receive the boring head because certain piping was not installed until after the patent had expired. Moreover, the product had in no sense been tested to the point where it could be predicted to work. Unlike Paper Converting, there was no “operable assembly” (id.) pri- or to the expiration of the patent. Indeed, there was no successful testing of the lathes at Cincinnati Milacron until six months after the patent had expired, and there was no successful boring of a useable gun tube until well over a year after the patent had expired. Paper Converting considerations are not remotely involved here.
Plaintiff presents an alternative argument for compensation for all four Cincinnati Milacron lathes. Plaintiff argues that had he known about Cincinnati Milacron’s contract with the Arsenal, he could have sued Cincinnati Milacron for patent infringement on the grounds that the contract represented a sale of a patented invention within the scope of 35 U.S.C. § 284. Had plaintiff sued Cincinnati Milacron, plaintiff reasons, the government would have compensated Cincinnati Milacron for the damages Cincinnati Milacron would have had to pay to plaintiff. Hence, plaintiff concludes, “reasonable and entire compensation” under Section 1498 should include compensation for these “sales.”
This reasoning must be rejected. First, it is based on a faulty premise. It does not appear that plaintiff could have successfully brought suit against Cincinnati Milacron for an infringing “sale” until the product was constructed and ready for use. Ecodyne Corp. v. Croll-Reynolds Engineering Co., 491 F.Supp. 194, 198 (D.Conn.1979). See also Lang v. Pacific Marine & Supply Co., 895 F.2d 761, 764 (Fed.Cir.1990); Lemelson v. United States, 752 F.2d 1538, 1548 (Fed.Cir.1985) (“Generally, infringement can occur only when the claimed combination has been assembled and is used or is available for use.”)
Second, and more fundamentally, the question here is one of federal government liability under Section 1498, not private liability under 35 U.S.C. § 284. Government liability under Section 1498 arises from the “use[] or manufacture[ ] by or for the United States.” There is no mention of liability for a “sale” to the United States of a device covered by a patent. In contrast, with respect to private liability for patent infringement, the “sale” of a patented device is specifically defined in 35 U.S.C. § 271 as an act of infringement and hence is compensable under 35 U.S.C. § 284.2
By authorizing suit against the United States, Section 1498 constitutes a waiver of sovereign immunity (Decca, 225 Ct.Cl. at 335, 640 F.2d at 1167). It is well established that any such waivers must be strictly construed. See United States v. Mitchell, 445 U.S. 535, 538, 100 S.Ct. 1349, 1351, *21663 L.Ed.2d 607 (1980). Since Section 1498 does not specifically authorize suits against the United States in instances of a sale of a patented device, it would violate the principles of sovereign immunity to interpret the statute as mandating compensation by the United States for such a sale. Hence, the court concludes that no compensation is appropriate under Section 1498 for the four guided lathes manufactured by Cincinnati Milacron.
V.
As the court explained in Sections II-IV, supra, plaintiff is entitled to compensation under Section 1498 for the Arsenal’s use of seven of the twelve guided lathes used at the Arsenal. The court now turns to the issue of calculating the appropriate “reasonable and entire compensation” for the use of those seven lathes.
Government liability for the unauthorized use or manufacture of a patented invention has its roots in the fifth amendment’s prohibition against the unauthorized taking of private property for public use. Leesona Corp. v. United States, 220 Ct.Cl. 234, 599 F.2d 958, cert. denied, 444 U.S. 991, 100 S.Ct. 522, 62 L.Ed.2d 420 (1979). Pursuant to 28 U.S.C. § 1498, a patent owner whose patented invention is used or manufactured by or for the United States without a license may bring a suit in the Claims Court and recover from the United States “reasonable and entire compensation.” “Reasonable and entire compensation” has two components. Decca, 225 Ct.Cl. at 336, 640 F.2d at 1167. The first component is the value of the license that, in effect, was taken by the government. The value is “determined ordinarily as of the time the Government takes the license.” Id. The second component involves compensation for the government’s delay in paying for that license. Decca, 225 Ct.Cl. at 337, 640 F.2d at 1168. The parties herein agree as to the proper measure of delay compensation. The central dispute involves the value to be placed on the patent license that in effect was taken here.
In Decca, after a review of pertinent precedent, the Court of Claims concluded that “[t]o appraise a patent license taken by the Government, the court utilizes one of three methods of valuation: (1) determination of a reasonable royalty for the license; (2) awarding a percentage of governmental cost savings arising from governmental use of the patented invention; or (3) awarding lost profits.” Id. at 336, 640 F.2d at 1167 (footnotes omitted). The lost profit method is not a viable alternative here because plaintiff neither manufactured nor sold the patented apparatus and, therefore, did not directly lose any profit as a result of defendant’s use of the patented invention.
As to the two remaining alternatives, the Decca court classified calculation of a reasonable royalty as the “preferred method” and used that method where “the facts of [the] case [did] not constrain [it] to use a method of valuation different from the preferred method.” Id. at 345, 640 F.2d at 1172. Herein, plaintiff favors use of the “percentage of governmental cost savings” method, but argues that even if the reasonable royalty approach is employed, the reasonable royalty should be calculated with reference to cost savings. The court will first address the calculation of a reasonable royalty and then address plaintiff’s contention that cost savings should be used instead of a reasonable royalty to the extent that the two methods result in different levels of compensation.
VI.
When determining a reasonable patent royalty, a court first assesses whether there is an established royalty rate for the patent. If there is an established rate, that rate typically constitutes the reasonable royalty and is used as “the best measure of reasonable and entire compensation.” Tektronix, Inc. v. United States, 213 Ct.Cl. 257, 552 F.2d 343 (1977). In the instant case, there is no such established royalty rate because, despite his best efforts, plaintiff never succeeded in securing a licensee under the ’568 patent.
In the absence of an established royalty rate, courts typically set the reasonable royalty by determining the royalty *217rate that the parties would have negotiated had they willingly engaged in license negotiations at the time of the taking. See, e.g., Tektronix, 213 Ct.Cl. 257, 552 F.2d 343; Amerace Esna Corp. v. United States, 199 Ct.Cl. 175, 181, 462 F.2d 1377, 1380 (1972) (“In the absence of an existing royalty rate, courts often resort to a ‘willing seller-willing buyer’ approach to establish what a reasonable royalty should be under the particular facts with which they are faced. In effect, this requires the synthesization of a hypothetical negotiation at the time of defendant’s infringement” (footnote omitted)). The “willing seller-willing buyer” test permits consideration of a broad set of factors, including, in appropriate situations, facts that were not known at the time of the hypothetical negotiation. Fromson v. Western Litho Plate & Supply Co., 853 F.2d 1568, 1574-76 (Fed.Cir.1988).
Here, the government was using the patented technology on the date the patent was issued, November 16, 1965, and the parties have stipulated that this date should be the date of the hypothetical negotiation. However, the parties disagree as to the type of royalty structure that plaintiff, as a willing seller of a patent license, and the Arsenal, as a willing buyer, would have negotiated on that date.
Plaintiff contends that the parties would have negotiated a license in which the royalty varied based on the Arsenal’s actual cost savings resulting from use of the patented invention. The parties would have resorted, plaintiff argues, to a “throughput” or “per-use” royalty pursuant to which defendant’s royalty payments would vary based on the number of times it used the patented invention, i.e., a separate royalty payment, in effect, would have been required for each cannon produced using lathes that employed a control system covered by the ’568 patent. Since plaintiff’s control system saved money on each cannon bored, a payment for each use of the patented device would tend to result in the total royalty varying with the actual cost savings to the Arsenal.
Defendant disagrees and contends that the parties never would have entered an arrangement that varied the royalty based upon the extent the patented apparatus was used. Rather, defendant argues, the parties would have negotiated a license calling for an up-front payment of a sum certain, plus a one-time royalty for each system the Arsenal used that was covered by the ’568 patent.
Review of the evidence presented at trial, including evidence as to industry practice and the parties’ subjective intent at the time of the hypothetical negotiation, and an analysis of the potential benefits and costs involved with the use of throughput and per-use royalties, indicate that defendant’s view of the outcome of the hypothetical negotiation is correct. The parties would not have negotiated a patent license containing a throughput or per-use royalty.
A.
First, the prac!:ce in the machine tool industry at the time of the hypothetical negotiation was that throughput and peruse royalties were not included in patent licenses where, as here, the patent covers an apparatus and the product produced using the patented apparatus is not itself patented. For example, Cincinnati Milacron, a major machine tool manufacturer, has never entered a license on an apparatus patent in which the royalty was based either on the output of the apparatus or on the estimated cost savings that would result from use of the apparatus. Bethlehem Steel Corporation, a user of machine tools to whom plaintiff turned as a possible patent licensee, has a similar licensing history. It has never entered a license on a machine patent in which the licensee was obliged to pay royalties based upon cost savings from the machine or to make continuing payments based on the value of unpatented products produced by the machine.
During trial, in addition to presenting evidence from individuals employed at firms in the machine tool industry, defendant also presented testimony from a licensing expert who has had significant experience in negotiating licenses in the machine tool industry. The expert, who the court found highly credible, testified that *218throughput and per-use royalties were virtually never employed in licenses involving patented machine tools.
B.
Second, plaintiff has failed to demonstrate that the parties herein were in any way inclined at the pertinent time to depart from this general industry practice. The evidence does not suggest that plaintiff or defendant had considered throughput or per-use royalties, much less that they would have consented to employ one in licensing the ’568 patent. To the contrary, plaintiff’s efforts to license his patent were quite consistent with general industry practice. For example, on December 14, 1963, following a meeting with representatives of Westinghouse Electric Corporation, plaintiff prepared a letter summarizing an offer to Westinghouse and responding to some Westinghouse questions. At the time, the controller patent had not yet been issued, but plaintiff anticipated its issuance. In an exhibit to the letter, plaintiff set forth “ball-park” licensing terms. He offered Westinghouse the “exclusive complete shop rights” to the controller patent (the ’568 patent) as well as the “exclusive commercial shop rights” to the indicator patent (issued to plaintiff and Folsom as co-inventors in 1962). Consistent with the industry norm, rather than suggesting a throughput or per-use royalty, the proposal called for a $50,000 initial payment plus a royalty of $2,000 for each device Westinghouse sold that was covered by either patent.3
Similarly, in August 1969, plaintiff authorized William Folsom to produce a single runout controller system covered by plaintiff’s patents and to sell that system either to Bethlehem Steel or to Streak, Inc. The authorization anticipated that Folsom would charge a royalty rate of six percent of the gross sales price of the system, with five percent going to plaintiff and one percent to Folsom. The six-percent royalty, based on gross sales price, further indicates plaintiff’s intention to license plaintiff’s patents based on a single royalty payment for each patented apparatus rather than based on a royalty that varies depending on the extent the apparatus is used.
Plaintiff also had negotiations directly with the Arsenal and at no time suggested a royalty based on the extent a patented controller was used. In its 1962 response to the Arsenal’s request for proposals, Flight Command Association, a small company of which plaintiff was a principal and Folsom an associate, bid $46,000 for the manufacture, installation, and testing of a runout controller. Though plaintiff had already filed a patent application on its run-out indicator, there was no suggestion in the bid that the Arsenal should pay a throughput or per-use royalty in the event the controller ultimately built was covered by a subsequently issued patent. In 1968, plaintiff offered to license the controller patent to the Arsenal for $1, provided that plaintiff would be able to demonstrate the device to prospective customers. In March 1970, plaintiff offered a license under the ’568 patent under the condition plaintiff were given a contract to build a second runout controller for the Arsenal. There was no suggestion in 1968 or 1970 that the license would include a throughput or peruse royalty.4
C.
Third, the nature of the technology and the surrounding circumstances similarly do *219not suggest that the hypothetical negotiation between the parties would have ended with a license containing a throughput or per-use royalty. In evaluating whether, in the course of the negotiation, the parties would have been drawn to the use of throughput or per-use royalties, it is appropriate first to review some of the potential advantages and disadvantages inherent in these types of royalties and then to predict the extent to which these advantages and disadvantages would have affected the hypothetical negotiators’ decisions.
In appropriate circumstances, throughput or per-use royalties can be highly efficient and bring benefits to both the licensor and the licensee. For example, a per-use royalty can permit a licensor efficiently to vary the royalty charged to different licensees under the same patent. Generally, each licensee will place a different value on a license which would depend upon the financial advantage the patented technology would bring to the licensee, i.e., the impact on the licensee’s costs, sales, profits, etc. The more a licensee gains from access to the patented technology, the more it will be willing to pay for a license. One typically significant variable in determining the value of a patent license is the frequency with which the licensee will use the patented invention. Thus, if a patented technology permits the manufacture of gun barrels at a lower cost, the more gun barrels manufactured, the greater the cost savings and, hence, the more valuable the license. A licensing scheme that employs throughput or per-use royalties, therefore, can be an efficient way for the licensor to vary the royalties charged to different licensees according to the value that the license represents to each licensee.
Throughput and per-use royalties can also benefit licensees. For example, a throughput or per-use royalty benefits the licensee by giving it additional control over the total license cost. The licensee has the flexibility to reduce the amount of the royalty payment under the license simply by using the patented technology less.5 Throughput and per-use royalties also can result in the patented technology becoming available to a greater number of licensees. If the licensor chooses to charge all licensees a single identical fee for each apparatus covered by the patent regardless of the amount of use, the fee chosen by the licensor in an effort to maximize its profits could well prove too high for certain potential licensees who anticipate comparatively little use of the patented apparatus. These same potential licensees may find a license financially beneficial if instead a throughput or per-use royalty were offered. In this way, throughput and per-use royalties potentially can result in patented technologies becoming available to licensees who otherwise would not have access.6
While there are potential benefits, there also are potential problems and costs associated with throughput and per-use royalties. One significant potential disadvantage is an increase in transaction costs for administering the license. A license that involves a fixed payment at the time the license is executed, plus a fixed payment for each patented apparatus used, is typically relatively easy to administer. To monitor compliance, the licensor need only *220identify each apparatus used by the licensee so as to determine which apparatus are covered by the patent. A throughput or per-use royalty, however, can raise far more complex and costly administration problems. To monitor compliance, the licensor would have to monitor the licensee’s actual use of each patented apparatus. Developing such a monitoring system can prove difficult and costly. Indeed, firms typically do not want outsiders either on their factory floor continually watching production or looking at the firm’s confidential records that discuss quantities of production.
Another potential disadvantage of throughput and per-use royalties involves the inherent unpredictability of the timing and total amount of royalty payments. For example, when a throughput or per-use royalty is employed in licensing a patent covering an apparatus, the total amount of payment for the license is indefinite at the time the license is signed and remains indefinite through the time all patented machines have been purchased and used up until the patent expires. Some licensors, including small operations like plaintiff’s, may prefer a more immediate and more definite payment of royalties.
A major disadvantage of throughput and per-use royalties is that unless the cost savings resulting from the patented technology is known with some certainty at the time of the license, it can prove difficult for the parties to agree as to the amount of likely cost savings and, hence, as to the amount of the royalty that should be paid for each use of the patented apparatus. On this point, defendant’s expert testified to the effect that throughput and per-use royalties are more commonly used in the licensing of chemical processes, where the licensor has fully developed the patented process and can demonstrate expected savings to the licensee. The cases upon which plaintiff and his licensing expert relied are generally consistent with this testimony of defendant’s expert. These cases typically involve licensing of patents that contain process or method claims in addition to apparatus claims and cover developed technologies. In Horvath v. McCord Radiator & Mfg. Co., 100 F.2d 326 (6th Cir.1938), cert. denied, 308 U.S. 581, 60 S.Ct. 101, 84 L.Ed. 486 (1939), the patent included method claims and, in addition, the patent owner had previously licensed the patents using a throughput royalty. In Binks Mfg. v. Ransburg Electro-Coating Corp., 122 U.S.P.Q. 74 (S.D.Ind.1959), aff'd in part and rev’d in part, 281 F.2d 252 (7th Cir.1960), the patent contained process as well as apparatus claims and the process had been fully developed at the time of licensing. Indeed, the patent owner was able to support a per-use royalty by demonstrating the cost savings to the licensee in a pilot operation. The patents involved in Brulotte v. Thys Co., 379 U.S. 29, 85 S.Ct. 176, 13 L.Ed.2d 99 (1964), and Dixie-Vortex Co. v. Paper Container Mfg. Co., 43 F.Supp. 518, 522-23 (N.D.Ill.), aff'd in part and rev’d in part, 130 F.2d 569 (7th Cir.1942), similarly covered inventions that were fully developed in working machines prior to the alleged infringer’s use of them.
When the potential benefits and costs associated with throughput and per-use royalties are evaluated in the context of the particular facts of this case, the court cannot conclude that during the hypothetical negotiation, the parties would have been drawn to the use of throughput or per-use royalties. First, at the time of the negotiation, plaintiff was seeking to maximize the initial payment in order to get his fledgling business going. It is not apparent that plaintiff would have agreed to a throughput or per-use royalty in view of the indefiniteness as to the timing and amount of payment. Second, the number of cannons manufactured at the Arsenal is potentially sensitive information and it is not clear that the Arsenal would have agreed to a monitoring system that would have provided this information to plaintiff on an ongoing basis.
But even more significant than these first two points is that at the time of the hypothetical negotiation, the patented technology was still in the process of being developed. Hence, it was not possible for the parties, with any reasonable degree of certainty, to predict the cost savings that ultimately would result from use of the *221patented technology in the boring of large tube cannons. In this setting, the parties likely would not have been able to agree as to the appropriate royalty to be charged for each use of the patented device or each cannon put through the device.
Given the industry practice, plaintiffs subjective intent, and the benefits and costs of throughput and per-use royalties, the court concludes that had the hypothetical negotiation occurred, the parties would not have departed from industry practice and employed a throughput or per-use royalty. Instead, the hypothetical negotiation would have resulted in the type of license typically used in the machine tool industry where there is an up-front payment, of a significant specified sum, plus a royalty payment for each apparatus covered by the patent that was manufactured or used by the United States during the patent term. Hence, the court will use such a compensation structure for determining a reasonable royalty here. See, e.g., Stickle v. Heublein, Inc., 716 F.2d 1550, 1561-63 (Fed.Cir.1983), where in a patent infringement action, the Court of Appeals for the Federal Circuit reversed the district court’s award of damages based on a per-use royalty calculated from cost savings where the industry practice was not to use such royalties and the evidence did not support a conclusion that the hypothetical negotiation would have resulted in a per-use royalty.
VII.
Next, plaintiff contends, in effect, that assuming the reasonable royalty analysis does not result in a royalty based on cost savings, the court should reject the reasonable royalty approach and instead base the royalty directly on an appropriate percentage of the cost savings that did result. But assessing the reasonable royalty is the preferred method and plaintiff has not presented any compelling facts that would “constrain [the court] to use a method of valuation different from the preferred method.” Decca, 225 Ct.Cl. at 345, 640 F.2d at 1172.
Moreover, a reasonable royalty analysis based on a hypothetical negotiation seems particularly appropriate in Section 1498 cases. Unlike with private litigants, a patent holder cannot prevent the government from using its patented invention. “The Government has a right to take patent licenses and cannot be enjoined from doing this.” Decca, 225 Ct.Cl. at 335, 640 F.2d at 1166. Therefore, with respect to the federal government, a patent owner, in effect, has no choice but to be a licensor; the sole issue is the appropriate royalty payment due. Thus, the Section 1498 suit, in a sense, is a substitute for royalty negotiations that should have taken place at the time the invention was first manufactured or used by or for the government. In this context, it seems particularly appropriate to base the royalty on the outcome of a hypothetical negotiation that would have occurred at that time.7
VIII.
Next comes the task of setting the appropriate compensation, i.e., establishing *222and applying the reasonable royalty. The best indication of the royalty that would have resulted from the hypothetical negotiation is plaintiff’s 1963 letter to Westinghouse in which he offered a license with considerable exclusivity for $50,000 plus a royalty of $2,000 per system. In light of Flight Command Association’s 1962 offer to the Arsenal to manufacture and install a runout controller for $46,000, this proposed $2,000 royalty would represent approximately five percent of what plaintiff had believed to be the cost of a runout controller.8 The court concludes that had the hypothetical negotiation between the Arsenal and plaintiff occurred, the parties generally would have followed this formula of the Westinghouse letter.
Hence, the hypothetical negotiation would have ended with plaintiff receiving an up-front payment of $50,000 adjusted for inflation between the date of the Westinghouse letter and the date of the negotiation, plus five percent of the cost the Arsenal would have had to pay an outside firm to manufacture and install a controller system on the existing lathes at the Arsenal. Since the Arsenal did the work in-house, it is appropriate to add to the in-house costs a profit margin of ten percent to account for profit that would have had to be paid had the work been done by an outside firm. Based upon the best available information in the record concerning such costs, the court concludes that the five percent royalty should be applied as follows:
Cost of Profit Total 5%
Conversion Factor Conversion Royalty
Modified Bergen Device $ 60,000 $ 6,000 $ 66,000 $ 3,300
WV 2488 79,523 7,952 87,475 4,374
WV 11187 183,650 18,365 202,015 10,101
WV 11235 75,000 7,500 82,500 4,125
WV 11205 91,203 9,120 100,323 5,016
WV 11225 98,164 9,816 107,980 5,399
WV 11190 105,173 10,517 115,690 5,785
Thus, the court is awarding compensation for the modified Bergen device even though the device was used only for a short period of time. As explained above, defendant contends that there was only de minimis use of the modified Bergen device and therefore no compensation is warranted. But in making this argument, defendant in a sense wants it both ways. Defendant argues against plaintiff’s efforts to employ a per-use royalty on the grounds that the parties would not have focused on the amount of use had they entered negotiations for a license. But then defendant turns around and seeks to avoid any compensation for the modified Bergen device based on the low amount of use. In effect, defendant wants the amount of use considered when it benefits defendant, but ignored when it works to its detriment.
The court rejects this approach. As summarized above, the appropriate method of analysis is with reference to a hypothetical licensing negotiation in November 1965. The court concludes that during such a negotiation, the parties would not have carved out a compensation exception for controllers used for a comparatively short *223period of time, even those used exclusively for developmental purposes. The evidence indicates that plaintiff would have sought remuneration for each device covered by the patent that was manufactured or used and it therefore is appropriate to treat the modified Bergen device in the same manner as all other devices for compensation purposes.
Conclusion
For the reasons set forth above, plaintiff is entitled to “reasonable and entire compensation” for seven of the twelve guided lathes used at the Arsenal. As discussed above, there are two components to “reasonable and entire compensation”: (1) the value of the patent; and (2) the compensation due as a result of the government’s delay in paying for its use of the patented device. As to the value, plaintiff is entitled to $50,000 adjusted for inflation between December 14, 1963, and November 16, 1965, plus a five percent royalty for each of the seven lathes as set forth in Section VIII, supra. As to delay compensation, the parties have agreed as to the appropriate method of calculation. Accordingly, on or before March 2, 1992, the parties shall file a stipulation as to the appropriate delay compensation and the appropriate total award based on the findings and conclusions set forth above. In a separate order, the court will order that judgment be entered for that amount.
IT IS SO ORDERED.
. 28 U.S.C. § 2501 provides: “Every claim of which the United States Claims Court has jurisdiction shall be barred unless the petition thereon is filed within six years after such claim first accrues.”
. Section 284 provides for "damages adequate to compensation for ... infringement.” Section 271 provides that "[e]xcept as otherwise provided in this title, whoever without authority makes, uses or sells any patented invention within the United States during the term of the patent therefor, infringes the patent” (emphasis added).
. Plaintiff does not remember actually sending this letter to Westinghouse. But the reasonable conclusion is that he did send it. Plaintiff impressed the court as a person who had a strong desire to license his patents and who worked diligently, even doggedly, toward that end. It is not reasonable to conclude that plaintiff would not have followed up on his meeting with Westinghouse representatives with a written response. There is no evidence to suggest that plaintiff drafted any response different from this letter. In any event, even if plaintiff never sent the letter, its contents would seem indicative of plaintiffs conclusions at that time as to an appropriate licensing scheme.
. Plaintiffs son discussed the possibility of employing a throughput or per-use royalty in a license of the ’568 patent in an academic thesis submitted in 1981. But this is long after the date of the hypothetical negotiation and, in any event, the proposal was unsuccessful. No entity indicated a willingness to enter such a license.
. For example, if after entering the license the licensee determines that a new technology is superior to the patented technology for certain uses, the licensee can simply cease using the patented technology for those uses and, in effect, lessen its royalty obligations under the license.
. A throughput or per-use royalty also permits the licensor and licensee to share the risk of the parties’ estimations as to the success of the patented technology turning out to be incorrect. If the patented technology turns out to be not as successful as predicted, it will be used less, the licensee will benefit less from the license, and the patentee will receive lower royalties. On the other hand, if the technology turns out to be more successful, the technology typically will be used more, the licensee will benefit more, and the patentee will receive higher royalties. A license based on a single up-front payment would involve no such sharing of risk. In such cases, if the patented technology turns out to be more successful than the parties anticipated, only the licensee benefits. It agreed to pay a fixed fee and is required to pay no more. On the other hand, if the patented technology turns out to be worthless, only the licensor benefits. It keeps the license payment while the licensee in effect secures no financial benefit from the license.
. One additional point on cost savings is worthy of discussion, though the point was not decisive in this court's decision to rely upon the reasonable royalty method. Contrary to plaintiffs allegations, quantifying the cost savings directly attributable to plaintiffs patented invention would not be a simple and straightforward matter here. To support its contention of cost savings, plaintiff relies in substantial part upon a report by an Arsenal employee who helped design and manufacture the guided boring devices used at the Arsenal. The report, written in February 1975, estimated that the new method employed at the Arsenal using a guided bore had saved the Arsenal $2.8 million in the manufacture of gun barrels. But there are two problems with relying on the $2.8 million figure as a measure of cost savings here. First, the "new method” did not involve simply modifying the "old method” so as to use the technology covered by the '568 patent. In order to produce the guided boring device that it ultimately used, the Arsenal had to solve a series of complex technological problems that were not addressed in plaintiffs patent. Neither party quantified the extent to which the $2.8 million in cost savings was attributable to the Arsenal’s own technological developments rather than to plaintiffs invention.
Second, the $2.8 million figure was based on a comparison of costs associated on the one hand with a new guided boring system used at the Arsenal and on the other with the conventional boring system previously used at the Arsenal for manufacturing 175 mm gun barrels. Apparently, a third technology, rapid boring, was developed at the Arsenal and was more efficient than conventional boring. Any calcu*222lation of cost savings that resulted from the Arsenal’s use of plaintiffs patented technology should compare plaintiffs patented technology with the next most efficient technology available to the Arsenal. At least during some of the relevant time, rapid boring was viable, available, and the second most efficient technology.
. The court concludes that the parties would not have included in the royalty base the cost of the lathes upon which the patented bore guidance devices were installed. The evidence, including the Westinghouse license offer and the Folsom authorization, indicates that plaintiff envisioned the controller being marketed as an add-on to existing lathes.