This is a patent suit under 28 U.S.C. § 1498 in which plaintiff seeks "reasonable and entire compensation" for alleged unauthorized manufacture for and use by the Federal Government of inventions described and claimed in U. S. Patent No. 2,611,813, entitled "Magnetic Data Storage System", issued to joint inventors T. K. Sharpless and E. S. Eickert, Jr. in 1952. Plaintiff Technitrol, Inc. is the record owner of this patent — the Sharpless patent. Plaintiff's petition charges infringement of claims 1-4, 6-14, and 16-24.1

Believing that it is free to use the patented invention, defendant moved to dismiss or, in the alternative for partial summary judgment, on the ground that the United States is licensed under the Sharpless patent. Plaintiff then countermoved for partial summary judgment that the United States is not so licensed, or for an order specifying the facts not in controversy on that issue. Without resolving the question, the court remanded the case to the trial commissioner with instructions to find the facts relevant to this license issue. The case is now before us on the commissioner's report on that question.2

For the reasons which follow we hold that the defendant is fully licensed under claim 16, and is also licensed under all other claims of the Sharpless patent except to the extent that those other claims may be limited to the system's automatic reset feature, explained below. We leave to later proceedings the determination of whether the three representative claims other than 16 (5, 19, 23) embrace the automatic reset feature, and if so whether those claims (and the patent) as so construed are valid. If both questions are answered affirmatively, then the issue of infringement will have to be reached as to those and the remaining claims.

The Sharpless patent, relating to electronic computers,3 discloses a magnetic data storage system, and has particular application "to systems for storing information, especially where it is desired to transmit, receive, and record information". The specification notes that a particular use of the system is "to store information concerning reservations on public carriers such as airplane lines, railway lines, etc." Generally, the system described in the patent comprises (1) a central storage unit which stores on magnetic disks information such as the number of seats available and reserved on various flights of a commercial airline; (2) remote operating stations, such as airline reservation desks at airports or hotels, where an operator through a suitable keyboard can request information of and send information to the central unit; (3) an arithmetic adding unit, for calculating information to be placed into the central storage unit; and (4) a control unit for sequencing various operations performed by the system. The system is designed to permit an operator at a remote station to find out from the central storage unit whether seats are available on a certain flight, and, if so, to record additional reservations up to the limit of the flight's capacity.

A detailed description of the system is contained in the findings of fact, but the salient technical characteristics may be summarized as follows: The central storage unit consists of a number of magnetic disks (information disks) mounted on a common shaft for rotation by an electric motor. Information, such as the number of seats already reserved and the number of seats available, is stored on the disks in the form of discrete areas of magnetization, called magnetic pulses. The pulses are arranged circumferentially in groups, called registers, around the faces of the disks. Each register contains information about a particular airline flight. Magnetic pickup heads are mounted adjacent to the disks so that, as the disks rotate, the heads, through appropriate circuitry, can "read" pulses on the disks, "write" new pulses on the disks, or "erase" existing pulses.

The remote stations are connected to the central storage unit through electrical transmission lines, akin to ordinary telephone lines. Each remote station has three keyboards, presumably one for each of three airline reservation clerks. By setting appropriate punch keys, an operator can designate (a) a particular flight on a particular day about which he desires information and (b) the number of seat reservations needed. In essence, the punched keys, through the necessary circuitry and coding, set off electrical pulse signals which are transmitted to the central station. There, the signal is decoded and split. The portion of the signal representing the number of desired seat reservations is sent to an electronic adder. The portion of the signal representing the particular flight actuates circuitry for locating the proper register on the information disks. When that register is located, the information on it, i. e., the number of seats already reserved, is sent to the adder. The adder then sums the "seats desired" and the "seats already reserved". If the total exceeds the number of seats available, an alarm circuit lights up a lamp at the remote station, so indicating. If the sum does not exceed the number of seats available, then the register is erased, the new sum is recorded, and the remote station is so advised, indicating that the reservation has been accepted. The system is then ready for another sequence of operations on demand from a remote station. The entire process takes about 0.14 seconds.

The system is designed so that, through its scanning and selector equipment, only one keyboard of any remote station can communicate with the central station at a time. This feature avoids the possibility of simultaneous duplicate requests being made to a register from more than one keyboard, and thus prevents a particular flight from oversubscription.

A particular aspect of the Sharpless system, as alleged by plaintiff,4 is the automatic or memory reset feature by which position volatility is avoided. On the common shaft with the information disks is a master or clock disk which has recorded circumferentially about its face two channels of magnetic pulses, one channel has 160 evenly-spaced pulses, the other has one pulse. The purpose of the clock disk, as its name implies, is to serve as a timing device to coordinate the timed-position of the information disks with other elements and circuits of the computer system. A pickup head is located over each channel. One head produces 160 evenly-timed pulses per revolution of the clock disk, the other one pulse per revolution. The 160-pulse head is connected to a scale-of-ten electronic counter which produces one output signal for each 10 input pulses. The counter therefore generates 16 pulses (160/10) for each revolution of the clock disk, and it thereby divides the information disks into 16 10-pulse registers, or information segments. Pulses from the scale-of-ten counter are applied through electronic circuitry to a digital counter which in turn produces 16 unique voltage combinations, each thus representative of one register on an information disk. When one of the voltage combinations matches up, or coincides, with a similar combination from a remote station (generated by an operator seeking information about a particular flight or register), circuitry is activated by which the register is "read".

It is essential that the clock disk and the scale-of-ten counter stay in synchronization. Otherwise, the counter will not produce signals representative of each register, or, to put it another and more technical way, the system would be position volatile. During normal operation, with power on and the equipment functioning properly, the clock disk and counter will stay synchronized. However, at start-up, either initially or after a power failure, the clock disk and counter may lose synchronization. For example, the clock disk may continue to rotate momentarily, through inertia, after electric power to the counter and the motor have ceased. Also, when power is cut off, the counter, a purely electronic device, loses count; and on re-start, its counts may not be synchronized with the start of a register on an information disk, so that such counts would be arbitrary with respect to register positions. Plaintiff asserts that this problem of potential loss of synchronization, or position volatility, is solved by the one pulse per revolution channel of the clock disk, which, it contends, is disclosed in the patent as connected through appropriate circuitry to reset the counters to zero after each revolution of the clock disk, thereby synchronizing the clock disk and counter if they are out of synchronization.

The 24 patent claims are combination claims which define the invention in terms of the elements making up the system. In general, they describe the magnetic memory central storage device, the remote stations, and associated circuitry by which the system is made operable. Claim 16 differs from the other claims in that it does not recite the remote stations and is directed primarily to details of the central station.

The Government's computer contracts with the University of Pennsylvania (ENIAC and EDVAC)

Defendant contends that it is licensed under the Sharpless patent because the invention claimed was developed in the course of government-sponsored research at the Moore School of Electrical Engineering of the University of Pennsylvania.

Mr. T. K. Sharpless graduated from the Moore School with a bachelor's degree in electrical engineering, and was thereafter hired by the Research Division of the Moore School in February 1943 as a full-time, non-academic, professional engineer to work on outside contracts negotiated with the School, which consisted preponderantly of those with the United States. In the early 1940's, the Army was interested in developing an electronic computer with a broad capability of solving mathematical problems relating to ballistic studies. Analog computers used for this purpose had proven unsatisfactory. So, after consultation with the Moore School, the Army in 1943 decided to underwrite the creation of the world's first general purpose electronic digital data processor or computer. At that time, digital computer technology was a virgin and unexplored universe charted only by untested theories unaccompanied by much practical experience. The research and development contract which the United States made with the University of Pennsylvania therefore called broadly for the Moore School "to engage in research and experimental work in connection with the development of an electronic numerical integrator and computer", later to become known (from its initial letters) as the ENIAC. The Government specified the capabilities it desired in the completed machine — what the ENIAC was supposed to be able to do — but it did not otherwise limit the scope or direct the path of the research to be conducted. In particular, the United States did not suggest a mode of operation, the speed of calculation, or the type of circuitry to be used; indeed, the Government was not in a position to furnish guidance in any of these areas. The ENIAC contract, in short, represented a first broad-scale foray into the digital computer field, and no one knew for certain what would be encountered. Sharpless was a key research engineer on the ENIAC project and played an important role in the overall planning of the contract performance. The ENIAC program ultimately produced a patented machine5 to which the Government received a non-exclusive license under the contract.6

After starting work on the ENIAC, the Research Division of the Moore School came to believe that a new machine, using techniques of operation and circuitry different from those contemplated for the ENIAC, could be built which would better perform the tasks desired by the Army. In October 1944, the scope of the ENIAC contract was broadened to include "research and experimental work in connection with the development of an Electronic Discrete Variable Calculator", or EDVAC (following the initial letters of the name). Work on the EDVAC was performed concurrently with that on the ENIAC until the completion of the ENIAC contract. In April 1946 the Army entered into a second contract with the University of Pennsylvania; this one called for full-scale research and development of an EDVAC.

Sharpless, who had previously worked on the EDVAC under the ENIAC contract, became technical director of the EDVAC project. He remained on the Moore School's payroll through October 15, 1947.7 Eickert, the co-inventor of the Sharpless patent, worked at the Moore School from September 1945 to April 1947, but did not participate in either the ENIAC or EDVAC projects.

Technitrol, Inc.

In April 1947, Sharpless, Eickert, and two other persons, incorporated plaintiff Technitrol, Inc. to exploit the technology of high speed computers in the military and industrial fileds. Although still employed by the Moore School in the summer and fall of 1947, Sharpless worked for Technitrol on occasional evenings and during the two-week summer recess at the School.8

Beginning in the summer of 1947, Technitrol negotiated an agreement with, and developed a computerized airlines reservations system (called "Reservisor") for, American Airlines. Sharpless did some work on the Reservisor system during August 1947, and devoted his efforts to it after joining Technitrol full-time in October 1947. Eickert, who had left the Moore School to join Technitrol in April 1947, aided Sharpless occasionally on the project during that summer, and worked with him full-time from December 1947. On May 26, 1948 a patent application, which matured into the patent in suit, was filed, naming Sharpless and Eickert as joint inventors.

Defendant's license defense is bottomed on the "Patent Rights" article of the EDVAC contract in which the University agreed to grant to the Government:

* * * an irrevocable, nonexclusive, nontransferable and royalty-free license to practice, and cause to be practiced for the Government, throughout the world, * * * each Subject Invention in the manufacture, use and disposition according to law of any article or material, and the use of any method. * * *.

"Subject invention" was defined as

* * * each invention, improvement and discovery (whether or not patentable) conceived or first actually reduced to practice (i) in the performance of this contract * * *, or (ii) in the performance of any research and development work relating to the subject matter hereof which was done upon the understanding that this contract or any subcontract hereunder would be awarded * * *.

Both Sharpless and Eickert knew or should have known of this patent license clause, as well as the comparable article in the ENIAC contract, and agreed, actually or in effect, to conform to their requirements. Plaintiff is, of course, likewise bound.

Since the defendant does not urge that the invention, improvement or discovery embodied in the Sharpless patent was actually reduced to practice in the course of performance of the EDVAC contract, our problem is to decide, first, what was the date of initial conception, and, second, whether that conception occurred "in the performance of" the EDVAC agreement or "in the performance of any research and development work relating to the subject matter hereof [i. e., the EDVAC contract] which was done upon the understanding that [the EDVAC] contract * * * would be awarded."

The parties and the trial commissioner have attacked these questions by first seeking to define the full scope of the patent claims, particularly with respect to the position volatility feature. We have decided, however, not to take that route and, instead, to advance toward the solution of the license problem without, at this time, determining the exact reach of the patent claims. We choose to avoid, for the present, this troublesome issue because our tentative appraisal suggests that, in this instance, the question of patent scope cannot well be resolved without inquiring simultaneously into patent validity — and the issue of validity has not yet been tried out in this litigation. The controversy over patent scope turns mainly on whether the claims (except for claim 16) diclose the automatic or memory reset feature counteracting position volatility. This aspect is not expressly recited in any of the claims, but the trial commissioner found, and plaintiff agrees, that under 35 U.S.C. § 112 the "means" clauses of the claims incorporate this feature.9 Defendant disagrees very strongly and contends, also, that the specifications and drawings do not support the claims as construed in this way. The parties battle strenuously over these issues, and the artillery from both sides is heavy. As we see it now, this conflict should not be resolved apart from, or prior to, a consideration of the validity of the claims if so read. Since it is the patent claims which determine the legally protected subject matter of a patent (Continental Paper Bag Co. v. Eastern Paper Bag Co., 210 U.S. 405, 419, 28 S. Ct. 748, 52 L. Ed. 1122 (1908); White v. Dunbar, 119 U.S. 47, 51-52, 7 S. Ct. 72, 30 L. Ed. 303 (1886)), in construing the claims (with respect to the automatic reset feature) in order to define the invention for license purposes we would be in large part predetermining the patent's validity. Any construction of the claims which we adopted now would govern in subsequent proceedings on validity and infringement (see Dominion Magnesium Ltd. v. United States, 320 F.2d 388, 162 Ct. Cl. 240 (1963)), and that reading would be measured against the requirements for patentability: novelty, utility, and non-obviousness. See 35 U.S.C. §§ 101-103 (1964). In determining validity, as the Dominion case and many others illustrate, courts must frequently choose between a narrow construction of the claims that upholds the patent and a broad construction that strikes it down. We do not wish to make that choice at this stage, believing that it would be both unwise for the court and unfair to the parties to affix a meaning to the claims in the absence of further proceedings, including the making of a record, on validity.

Moreover, it is unnecessary to delimit fully the scope of the patent claims in order to move toward resolution of the license issue. The contractual provision does not tie the Government's license to the patent laws; any invention, improvement or discovery conceived in the course of performance of the EDVAC contract, whether or not patentable, is licensed to the Government.10 We may therefore begin to determine the scope of the license by comparing the substantive elements of the Sharpless system with the work done under the ENIAC and EDVAC contracts. As our discussion will show, we can block out much of the answer to the license problem even though we do not, and cannot, decide it entirely at this time.

Conception during the period of the ENIAC contract

In deciding whether the Government has a license, one basic question is when was the Sharpless invention (or inventions) conceived. Conception is a pivotal if somewhat nebulous notion in patent law, which is defined (1 Walker, Patents § 45 (Deller's 2d ed.) at 191-92) as "* * * the formation in the mind of the inventor of a definite idea of a complete and operative invention as it is thereafter to be reduced to practice * *. The date of conception is the date when the inventive idea is crystallized in all of its essential attributes and becomes so clearly defined in the mind of the inventor as to be capable of being converted to reality and reduced to practice by the inventor or by one skilled in the art." In this instance there is no real difficulty with the timing of "conception". The record makes it plain, in our view, when the various components of the Sharpless invention or inventions (however broadly or narrowly defined) first became crystallized.

It is clear to us that the elements of the Sharpless system, except for the automatic reset feature (dealing with position volatility), were all conceived prior to Sharpless' cessation of employment with the Moore School (October 15, 1947) and during the performance of the EDVAC contract. First, the defendant introduced a set of documentary exhibits made up of sketches and written material prepared by Sharpless in the first half of August 1947. Defendant's expert witness testified, without substantial contradiction by plaintiff, that the drawings collectively disclosed an information storage system with sufficient clarity to have enabled its reduction to practice by a person skilled in the art existing in August 1947, without the need for exercise of any new or unobvious faculties, techniques and means. The witness also said that every element and combination contained in the Sharpless patent was revealed in the drawings.11 This evidence shows that, except for the automatic reset facet, the Sharpless system was crystallized by the end of August 1947.

These drawings are themselves sufficient to establish conception by Sharpless before August 15, 1947, but there is more. The disclosures in these drawings are confirmed by a set of drawings made by Sharpless in the first two days of October 1947 while still on the Moore School payroll (see footnote 7, supra). These later drawings reveal further specific details of the keyboard and scanning devices shown in the earlier drawings, and are fully consistent with them and with the defendant's expert's testimony as to what the August drawings taught one skilled in the art.

In addition to this convincing documentary material, there are significant admissions by plaintiff which date conception of the invention before Sharpless left the university. These prior statements cannot be discounted on the ground that their substance has not been otherwise proved; all that is necessary to constitute an admission is a previous statement by an adversary party which is inconsistent with the position he is taking in litigation. Employers Mutual Casualty Co. of Des Moines v. Mosqueda, 317 F.2d 609, 612 (C.A. 5, 1963); Cox v. Esso Shipping Co., 247 F.2d 629, 632 (C. A. 5, 1957). The trier of fact can rightly take account of such admissions, although they are not, of course, conclusive if there is opposing evidence. La Flare v. Chase, 8 App.D.C. 83 (1896).

The first set of admissions occurred in a patent office interference proceeding between the Sharpless patent application and one filed by another. An interference is a procedure initiated by the Commissioner of Patents which seeks to determine the priority of invention between two applicants who are claiming the same invention, or between an applicant and a patentee. 35 U.S.C. § 135 (1964). The Sharpless application was the later filed, so Sharpless was designated the "junior party" and was required to file a "Preliminary Statement," a sworn document substantiating its claim to priority. See Patent Office Rules of Practice 201, 215, 216, 37 C.F. R. §§ 1.201, 1.215, 1.216 (1970). Sharpless and Eickert swore in this statement: (1) that the first drawing of the invention defined by the count in interference was made during August 1947; (2) the invention was first disclosed to others during July 1947; and (3) the active exercise of reasonable diligence towards reducing the invention to practice began during July 1947.

The interference was later amended by adding several new counts, including one which eventually issued as Claim 16 of the Sharpless patent. Sharpless' attorneys filed a so-called Rule 233(e) statement (as amended, 37 C.F.R. § 1.231(f) (1970)) which said that the dates originally sworn to in the preliminary statement would govern the new claims. The Rule 233(e) statement therefore reaffirmed the dates earlier sworn to and is equivalent to a new sworn statement.

These interference proceeding admissions, made fourteen years before the filing of the petition in the present action, are strong evidence of the time of conception. Their weight is buttressed by the failure of both Sharpless and Eickert to repudiate them in their testimony in this case. And they are not the only admissions in the record. In the course of an infringement proceeding in a federal district court between Technitrol Engineering Co., Inc. and the Sperry-Rand Corporation,12 those parties agreed to exchange formal statements as to the dates of conception on which they would rely. Eickert, in his capacity as President of Technitrol, said:

As presently advised, the earliest dates of conception of the inventions which are the subject of the Sharpless, et al. patent No. 2,611,813 upon which the plaintiff will rely at the trial of this action is during the period from June 25, 1947 to July 6, 1947. (Emphasis added.)

Sharpless was, of course, still working on the EDVAC contract for the Moore School during that period.

Taken together, these drawings and admissions are ample to establish that every element of Sharpless' "Magnetic Data Storage System", except the automatic reset feature, was conceived before Sharpless left his employment at the Moore School. There are also other, lesser, indications in the record supporting this conclusion, and there is very little to the contrary. Plaintiff's argument for a later conception date rests mainly on its claim that the automatic reset feature is central and necessary to the invention (a position which, as we have noted, the defendant strongly disputes). And it is quite true that the record just as clearly discloses that the automatic reset feature (the solution to position volatility) was not conceived until 1948 after both Sharpless and Eickert had left Moore and the University of Pennsylvania. This is established by a drawing by Sharpless showing the Reservisor system (developed by Technitrol for the air carrier) as envisioned on January 1, 1948. The central storage system consisted of, among other things, information disks and a clock disk mounted on a common shaft. However, the clock disk did not have the one pulse per revolution channel by which the memory could be reset to avoid position volatility. Recognizing the problem, Eickert, sometime after January 1, 1948, suggested modifying the clock disk with the single pulse channel.

"In the performance of this contract" for federal license purposes

We now come to consider whether the components of the Sharpless conception — again, excluding the automatic reset feature — may still fail to have been conceived "in the performance of" the EDVAC contract, although temporally developed before Sharpless separated from the Moore School. It is possible, of course, for an invention to lie outside the performance of the government contract, even though crystallized during the period of that project, because its subject matter is distinct from the government work. We have stated, in large part, the principles which govern construction of government license clauses, depending on conception "in the performance of [the] contract," in Mine Safety Appliances Company v. United States, supra, 364 F.2d 385, 176 Ct. Cl. 777 (1966). The inventors in that case were employed by a university in a government-sponsored aviation medicine program designed to study the effects of acceleration on the human body. The subject matter of the patent was an energy-absorbing crash helmet, which, it was claimed, was beyond the scope of the government contract. In rejecting that position, the court said that the license provision was operative whenever the "private" invention bore a "close connection" or a "close and umbilical connection" to the government-underwritten research. 364 F.2d at 389, 391, 176 Ct. Cl. at 783, 787. Furthermore, the court declined to restrict the license clause to circumstances where every component of the invention had been discovered in the course of contract performance. The Government would be licensed (364 F.2d at 391, 176 Ct. Cl. at 787-788), "[a]t least in those instances in which the invention was conceived or practiced during the existence of the contract, * * [where] an important factor in the invention was itself within the contractual scope, or resulted directly from the course of contract performance."

What Mine Safety teaches is that the issue of license vel non should be approached liberally by asking what the United States (acting for its taxpayers) can fairly be said to have purchased through its sponsorship of the contract project. The Federal Government has the right to use, royalty-free, those ideas, improvements, discoveries, and inventions — crystallized during performance of the federal contract — which have a "close and umbilical relationship" to the work and research funded by the United States. Having borne the expense of that effort, the public is entitled to enjoy the fruits without further charge. Accordingly, as we said in Mine Safety, "[i]t is fitting to read the license-grant of Section 17 [which is the same as the license clause of the EDVAC contract] liberally, * * * and not to confine it severely within a narrow compass." 364 F.2d at 392, 176 Ct.Cl. at 789.

The original ENIAC contract was for pioneer research in the digital computer field. The ENIAC machine which was ultimately developed incorporated a great number of vacuum tubes, including a small capacity internal vacuum tube memory, and could be programmed in a variety of ways through manual switches and plugboards to perform a number of different functions. The ENIAC's capabilities were limited, however, by its clumsy programming techniques, and even before the ENIAC machine was completed the Research Division of the Moore School began considering design modifications, including the use of magnetic disks and drums for the internal storage of information. These improvements formed the basis of the EDVAC, a second generation general purpose computer. The amendment of the ENIAC contract, previously mentioned, which broadened the scope of work to include experimentation on the EDVAC, also required that a complete work report be delivered to the Government. This report, entitled "Automatic High Speed Computing, a Progress Report on the EDVAC", was prepared by Eckert and Mauchly, employees of the Moore School and joint inventors of the ENIAC. Dated September 30, 1945, the report disclosed proposed characteristics of the EDVAC, among the most important of which was a high speed, large capacity memory. The report specifically referred to magnetic storage devices, first proposed in 1944, as an area for exploration, although it indicated that an acoustic mercury delay line system looked more promising at the time.

The original, separate EDVAC contract called for "research and development" on a "preliminary model of a small Electronic Discrete Variable Calculator". The small model was to serve to "demonstrate the feasibility" of a full-scale EDVAC having the properties described in the Eckert and Mauchly report. The feasibility model did not exhaust the goals of the contract. Indeed, amendments to the EDVAC grew from a small preliminary model to a versatile full-size machine capable of solving a variety of sophisticated problems, and its growth in complexity was paralleled by the increase in its cost to the Government, from an original contract price of $100,000 to one of $409,700. The EDVAC became the subject of a patent,13 which disclosed and claimed several different central or internal memory elements including, it should be noted, both acoustic mercury and magnetic disk devices.

The record therefore fully supports the conclusion that a magnetic information storage device — the system used in the Sharpless patent — was well within the scope of the EDVAC contract. Plaintiff contends otherwise, and the trial commissioner agreed, on the ground that the EDVAC machine actually built employed an acoustic mercury delay line device. But the ENIAC and EDVAC contracts were for research and development and their scope cannot fairly be limited to the features embodied in the machines physically produced. The Government spent nearly $900,000 on the ENIAC and EDVAC contracts; it was paying not merely or even primarily for specific machines but for the advancement of knowledge in computer technology produced by the research. Indeed, the ENIAC machine was obsolete before it was even completed — the knowledge gained in producing it gave engineers at the Moore School the ability, which they recognized, to create a much more useful computer, the EDVAC. There was a continuous line of research running from the ENIAC project through the EDVAC contract, devoted to increasing human understanding, as well as the actual performance, of high speed computers. The particular machines built under the two interrelated programs were not the sole aim of the contracts, nor the only results for which the Government paid. It paid, too, for the exploration of the field and the acquisition of new knowledge,14 and it is entitled to the crystallized ideas, improvements, and inventions emerging from that process of ongoing study, inquiry, and creation. Scientific research is not a straight and narrow journey, with blinders, along a single path to a known destination. In the process of discovery and invention under a research and development program many roads are uncovered; some are pursued immediately and some must wait for another day. One "product" of the research is the overall accretion in knowledge, and this encompasses the untraveled ways sketched out, as well as the trodden ones. The possible use of a magnetic memory system was revealed by the ENIAC-EDVAC research and was well within the scope of the EDVAC contract.15

True, the EDVAC program ultimately produced a model machine based on the acoustic mercury system, rather than the magnetic data storage system preferred by Sharpless. But this specific end-result did not limit or characterize the broad scope the EDVAC project had had from its inception, nor did it mean that the United States abandoned or gave up its claim on the research that had been done on other systems or avenues, or its right to pursue those paths under the EDVAC contract. We have no doubt that if it had been decided by the EDVAC project managers, in July-August 1947, that it would be useful to do further work on a magnetic data storage system (as well as an acoustic mercury system) that such further research would have been within the general scope of the EDVAC agreement as it stood at the time. So also, the Sharpless patent cannot be isolated from the EDVAC on the ground that the former is specifically designed as an inventory control system while the latter is not so limited. The EDVAC was a general purpose computer capable of being programmed to perform many tasks, including inventory control, and that function is hardly remote or unlikely for the Federal Government, which has a unique problem in keeping track of its far-flung property.

The sum of it is that we find the Mine Safety requirement of a "close and umbilical connection" between the government-sponsored research and the "private" invention to have been fully satisfied in this instance: the ENIAC spawned the EDVAC which in turn revealed the concept of a magnetic data storage system in an unbroken chain of descent. All of the Sharpless invention (leaving aside the automatic reset feature) was therefore conceived during and in the performance of the EDVAC contract. We are not saying this is so merely because Sharpless happened to use some knowledge or idea gained through his work on that project.16 We are saying, rather, that the discrete "invention, improvement and discovery" which Sharpless conceived before he left the project — his conception of a magnetic storage data system, embodied in the patent in suit — had a very close and integral connection, as a whole, with the government-financed project. His "invention, improvement and discovery" was not separate from, or independent of, the ongoing EDVAC program; on the contrary, it was inseparable from it. Accordingly, we hold that the United States is licensed as to all elements of the Sharpless invention (excluding, for the time being, the automatic reset feature).

As shown above, and practically conceded by defendant, the automatic reset phase of the Sharpless patent was conceived after Sharpless left the Moore School. Defendant argues that, nevertheless, it is licensed (assuming arguendo that it forms part of the patented invention at all) because it is, at most, a minor part of the invention, the major portion of which was conceived while the inventor was doing government work. In Mine Safety, supra, we did not have to decide whether an invention could be licensed — although not conceived or practiced until after the end of the contract's life — because of its close connection with the prior federal program. That problem does arise here, at least potentially, with respect to the automatic reset feature. We now determine that that feature, if it is validly covered by the patent, is not licensed since it was conceived after performance and not "in" the performance of the EDVAC contract. We make this holding on the assumption (which will have to be tested in later proceedings) that the automatic reset feature is a non-obvious improvement or extension of the magnetic data storage system conceived by Sharpless before he separated from the University of Pennsylvania. We think that the terms of the license clause before us preclude the blanketing of a non-obvious "invention, improvement and discovery" conceived after the termination of the inventor's federal connection, even though that separate invention may have a close connection with the preceding government work. The clause, as worded here, does not purport to extend its reach chronologically, by mortmain, into the indefinite future. The unqualified phrase "in the performance" of the federal contract has both temporal and substantive connotations; in the former aspect, it excludes post-contract developments which are new and not obvious.

Conclusions and directions for further proceedings

As we have already said, we do not now decide the scope of the Sharpless patent (with respect to the automatic reset feature); that issue should be canvassed and resolved in further proceedings along with the question of patent validity. In the light of that deferral of final decision, our present decision on the license defense is, first, that claim 16, which was found by the trial commissioner and is agreed by the parties to be entirely independent of the automatic reset device, is fully licensed to the Government, and, second, that the other three representative claims (5, 19, 23) are licensed to the Government except to the extent they may be found hereafter to be validly limited to the automatic reset feature. In other words, the only possible area of infringement liability concerns the solution to position volatility, i. e. the automatic reset feature; all other aspects of the Sharpless system and patent are licensed to the Government.

Before any conclusion of infringement is reached, it must be determined in the subsequent proceedings that (1) the patent properly claims the solution to position volatility (the automatic reset feature); (2) the patent is valid as so construed; and (3) the defendant has appropriated the automatic reset feature to its own use. We also emphasize that, even if the plaintiff is successful in proving infringement of the reset feature, such a showing can impose no infringement liability on defendant for use of any or all of the other elements of the magnetic data storage system embodied in the patent, which we now hold are licensed to the United States. Any recovery will be limited to a reasonable royalty for the reset feature alone. Cf. Cover v. Chicago Eye Shield Co., 130 F.2d 25 (C.A.7, 1942). If it turns out that a proper royalty limited to this particular aspect cannot be separately established, the trial commissioner should consider whether plaintiff is entitled to any recovery at all. Cf. New Jersey Zinc Co. v. Singmaster, 4 F. Supp. 967, 980-981 (S.D.N.Y.1933), modified, 71 F.2d 277 (C.A. 2, 1934); U. S. Colloid Mill Corp. v. Myers, 6 F. Supp. 283, 285-286, 268 (S.D.N.Y.1934).

The case is remanded to the trial commissioner for further proceedings consistent with this opinion.