Appellee Sands, the senior party, obtained a patent1 on August 5, 1958, based on an application filed January 30, 1953. The present interference involves this patent and an application2 filed by appellant Paivinen on February 4, 1959.

To establish priority of invention, Paivinen took testimony and introduced other evidence purporting to prove actual reduction to practice of the invention defined in the six interference counts prior to the filing date of the Sands patent. Sands did not attempt to prove an actual reduction to practice but relies upon his application as establishing a constructive reduction to practice as of its filing date. Sands did, however, introduce testimony purporting to show the insufficiency of Paivinen's proofs. Paivinen responded with rebuttal testimony.

The Board of Patent Interferences considered only the testimony-in-chief on behalf of Paivinen, and awarded priority of invention to Sands, holding that "Paivinen has not proven reduction to practice of the invention defined by the counts in issue." The board also commented briefly on the nature of Paivinen's burden of proof. We shall deal with these matters separately. First, however, we shall consider briefly the subject matter of the counts, since a basic understanding of the invention defined in the counts is necessary for proper evaluation of the issues.

Counts 1 and 6 are representative and read:

"1. In combination, a body of ferromagnetic material, at least one conductor disposed adjacent the body in position to send flux through the body, a load impedance, and three separate means to drive currents through at least one of said conductors, two of said means being adapted to drive currents through said conductors in such directions and of such magnitude and duration as to magnetize said body to saturation in substantially opposite directions regardless of its previous condition of magnetization, the third of said means having said load impedance in series therewith and being incapable of reversing the condition of saturation of said body.

"6. A circuit for the evaluation of a logical And function involving two variables of possible zero and unit values, said circuit comprising two saturable magnetic cores, two sensing field generators having a switch in common and having separate series-connected windings one on each of said cores, two set field generators including each one winding on one of said cores, and two reset field generators having a switch in common and having series-connected windings one on each of said cores."

Before considering the counts in relation to the disclosures, a brief consideration of certain technical background information will be helpful. The invention in issue is predicated on the known fact that certain magnetic materials become magnetized in such a manner that they are characterized as having a substantially rectangular hysteresis loop, frequently termed a "square loop." Such "square loop" magnetic materials exhibit a very useful property: namely, if they are placed in a saturated magnetic state in one direction, they will remain in that state until a subsequent magnetic field is applied in such manner as to reverse the state. If the subsequent magnetic field is of sufficient magnitude and duration, it will saturate the material in the reverse direction, and it will remain in that state until another field is applied to reset it to the first state, and so on. This property makes such materials useful for many purposes, including the storage of electrical signals, as in certain computer functions.

The invention defined in counts 1 through 5 can be best understood by considering Fig. 1 of the Sands patent:

NOTE: OPINION CONTAINING TABLE OR OTHER DATA THAT IS NOT VIEWABLE

The toroidal magnetic core 1 is composed of "square loop" material, and has three windings: S, a "set" winding; R, a "reset" winding; and S', an "inspect" winding. Each of the windings is connected to a corresponding current generator (numbered 5, 6 and 8, respectively). In addition, the circuit containing the inspect winding S' incorporates a load 7, across which are placed output terminals.

When current of predetermined magnitude and duration flows through S, it creates sufficient magnetizing force to drive the core to saturation in one direction, for example, a clockwise direction, and this regardless of the previous magnetic state of the core. When the current pulse ceases, the core retains this magnetic state and is said to be in a "set" state. To reset the core, a suitable current pulse is applied to winding R, and the core is driven to saturation in the opposite, or counterclockwise, direction. The inspect winding S' is arranged such that, when the core is in the set state, a low impedance path is presented to a current pulse flowing through S', and a signal is thus produced across load 7. On the other hand, if the core is in the reset state, most of the energy of a current pulse flowing through S' is absorbed by the core, and no effective signal appears across the load. Thus, by observing the presence or absence of a signal voltage across the load when an inspect pulse is applied through winding S', it can be determined whether the core was in the set or reset state.

Obviously, it is necessary that the magnitude and duration of the inspect pulse be kept below a value which would drive the core to saturation; otherwise, even if the core were in the reset state, the inspect pulse would switch the state, and a signal would appear across the output terminals.

Count 6 calls for two cores, each with a set, reset and inspect winding. Fig. 2 of the Paivinen application shows a typical multi-core arrangement (with three cores instead of two):

NOTE: OPINION CONTAINING TABLE OR OTHER DATA THAT IS NOT VIEWABLE

The reset windings 49, 50 and 51 are connected in series so that all three cores can be reset with one pulse. The inspect windings 46, 47 and 48 are likewise connected in series with a load 54. Such an arrangement is termed a logical "AND" circuit. When an inspect pulse is generated from source 52, a signal will appear across load 54 only if core 40 and core 41 and core 42 are all in the set state. (The set windings and their corresponding generators are 43 & 65, 44 & 66 and 45 & 67 for cores 40, 41 and 42, respectively.)

Essentially, the issue raised by the present appeal relates to the problem of the burden of proof as applied to the present record. The legal concepts as to burden of proof have caused much confusion when applied in patent interferences. It is properly recognized that in the usual case the junior party has the burden of proof with respect to the issue of priority of invention. This burden is actually broken down into two separate components: 1) a burden of coming forward with evidence, and 2) a burden of persuasion. E. g., I Jones on Evidence § 176 (4th ed. 1938).

A summary review of certain of the pertinent legal propositions will be helpful in resolving the issues here presented. Normally, the burden of persuasion is said to be satisfied if the junior party is able to make out his case by a "preponderance of the evidence." E. g., Creamer v. Kirkwood, 305 F.2d 486, 50 CCPA 715; Seeley v. Rennick, 314 F.2d 577, 50 CCPA 1214. See generally 3 Revise & Caesar, Interference Law and Practice § 462 (1947).

However, in situations where, as here, the junior party files his application subsequent to the issuance date of the senior party's patent, it is generally held that the junior party must prove priority "beyond a reasonable doubt." Conner v. Joris, 241 F.2d 944, 44 CCPA 772. See also cases cited in 3 Revise & Caesar, supra, § 467. In addition, the application of the junior party must clearly support the interference counts.3 Jepson v. Coleman, 314 F.2d 533, 50 CCPA 1051. The junior party can, however, escape the necessity of proving his case "beyond a reasonable doubt" by establishing for his application an effective filing date prior to the date of issuance of the senior party's patent. Thus, even where a junior party copies claims from an already-issued patent, he nevertheless need only prove priority by a "preponderance of the evidence," if he can show support for the counts in a co-pending or parent application with a filing date prior to the issue date of the patent. Sellner v. Solloway, 267 F.2d 321, 46 CCPA 897; 3 Revise & Caesar, supra, § 464.

In the case at bar, Paivinen's application in interference was filed subsequent to the date of issue of Sands' patent. However, this application is said to be a continuation of his own earlier-filed case4 (hereinafter termed "parent"), which has a filing date prior to the date of issuance of the Sands' patent. Appellant argues that his parent application contains clear disclosure in support of all of the six interference counts, and that his burden of proof is discharged by establishing a mere preponderance of the evidence. Appellee, on the other hand, argues that the parent application does not clearly support each and every limitation of the counts, and asserts that appellant must therefore prove priority beyond a reasonable doubt.

The board disposed of this problem with the rather cryptic statement that:

"The degree of burden of proof upon the junior party presents a peculiarity. It appears that Paivinen is entitled on the record to the filing date of an application filed during the pendency of the Sands application as to count 6, as to which the burden of proof would be the ordinary one of a preponderance of the evidence, but that this earlier application does not support a limitation in counts 1 to 5 inclusive. Sands contends as to these counts that Paivinen is under the burden of proving his case beyond a reasonable doubt. The Primary Examiner declared the interference stating that the involved application of Paivinen was a continuation of the earlier application, but in the course of his ruling on Paivinen's motion to amend also ruled that the involved application is a continuation-in-part, only of the earlier application. There appears to be no reason to disagree with Sands on this point."

We interpret this as meaning that the board held Paivinen's parent application does not support counts 1 through 5 and that he is therefore required to prove priority beyond a reasonable doubt with respect to those counts.

Both parties devote a large portion of their briefs to the burden of proof issue thus raised. Both seem to agree that there was confusion on the part of the board with respect to which counts were unsupported by the Paivinen parent application. As appellant points out, the Primary Examiner, in ruling on appellant's motion to amend, held that the parent application did not disclose the switches recited in counts 2 through 6. The board, however, (in the statement quoted above) indicated that the lack of supporting disclosure was with regard to counts 1 through 5 and apparently found a supporting disclosure for count 6. However, the board made no distinction as to appellant's burden of proof as it was related to the count for which it found support in the parent case.

In view of this apparent confusion over precisely which of the counts are unsupported by the parent disclosure, and in view of the unsatisfactory state of the record with regard to the merits of the issue of support, we must remand this issue to the board for further consideration.5 However, since we feel that the decision on the issue of support and, concomitantly therewith, the determination of the nature of Paivinen's burden of proof, might well have a substantial impact on the final outcome of this case upon its reconsideration by the board, we will indicate briefly our concept of the relation of the burden of proof to the merits of the priority issue. We do this also because, as will become apparent later in this opinion, we feel that appellant made out a prima facie case of priority, for which reason we are obliged to reverse the holding of the board in that regard.

It is clear that the question of "burden of persuasion" is concerned with the weight and sufficiency of the evidence. The question resolves itself into two related considerations: 1) are the appellant's allegations of fact believable?; and 2) if so, are they sufficient, as a matter of law, to prove his case? In order to discharge his burden of proof here, appellant must obtain affirmative answers to both of these questions. The second we shall answer in our consideration of the merits of the priority issue. The first remains for the board to determine on remand. If the board should find that, as to certain of the counts, appellant's burden is to show priority beyond a reasonable doubt, then what this will really mean is that appellant's alleged facts must be credible beyond a reasonable doubt. Likewise, if the burden is found as to others of the counts to be one of a preponderance of the evidence, then appellant's allegations of fact need be supported only by a preponderance of the evidence.

In order to show prior invention, Paivinen must prove a completed conception and reduction to practice prior to the filing date of the Sands patent, for, he admits in his brief, he "has introduced no evidence to support a finding of reasonable diligence over the period in question." He states therefore that "he must establish a complete conception and an actual reduction to practice prior to the Sands filing date of January 30, 1953."

While the total testimony contained in the record is lengthy, the relevant portion, i. e., the positive testimony on behalf of Paivinen, is relatively short and consists of the direct testimony of Paivinen and his corroborating witness Allen, and the cross examination testimony of Allen. Allen and Paivinen were co-workers, both being employed by Burroughs Corporation at the time of the acts testified to.

Paivinen introduced evidence to establish his conception of the invention of the counts prior to Sands' January 1953 filing date. Among the exhibits introduced in appellant's behalf are several pages from a laboratory notebook kept by him during the latter months of 1951 and the early months of 1952. The pertinent pages bear the date notation "2-25-52." Paivinen testified that he conceived the invention on February 25, 1952, and referred to the several pages of his notebook as fixing the date. He also testified that he showed these pages to Allen on February 25 and to one Avery on February 26. The last page of the group bears the notation:

  Witnessed and Understood L.G. Allen 2-25-52
  Witnessed and understood R.W. Avery 2-26-52

Allen testified that he had witnessed and understood the material contained in the notebook pages on the date indicated thereon.

Page 25 of Paivinen's notebook contains a circuit drawing which closely corresponds to the circuit shown in Fig. 1 of his application, which is reproduced below.

NOTE: OPINION CONTAINING TABLE OR OTHER DATA THAT IS NOT VIEWABLE

The board agreed that the circuit at page 25 of Paivinen's notebook is the same as that shown in Fig. 1 above, except for a minor matter not here pertinent. We shall, therefore, hereinafter refer to the circuit on page 25 of the notebook, the circuit which was conceived, built and tested and which embodies the invention in issue, as "the circuit of Fig. 1." It should be noted that the circuit of Fig. 1 contains a plurality of cores, and is thus a representation of the arrangement defined in count 6 (except that count 6 requires but two cores).

Allen testified that he built and tested the circuit of Fig. 1 under the direction of Paivinen in March of 1952. Among the exhibits are several pages from a laboratory notebook issued to and kept by Allen. Page 7, which bears the date "3-18-52", contains the following statement: "A voltage-driven magnetic gate was built & tested, following the circuit on pg. 25 of workbook #000124 [Paivinen's notebook]."

It should be noted that the circuit of Fig. 1 contains, instead of a rectangle marked "LOAD", a fourth core, 13, which is termed the output core. In this regard Paivinen testified:

"One other point that could also be made is that we show on Fig. 25 [sic: Page 25] the use of an output core. That is output core D. On the other hand, as noted on page 24 of my notebook, this is not the only manner in which the circuit could be utilized, in that we could use some other type of impedence across which this voltage could be generated, and we specifically note, I specifically note on page 24, that a resistor could be used in place of the output core.

"Under this latter implementation, the same two magnitudes of transfer current would still be expected to arise or would arise in response to the impedences offered by the input cores, but the voltage would simply be observed across a resistor instead of providing an actual input signal to the output core D."

It appears that an output core was selected as the load so as to simulate closely the possible practical uses for the circuit. As Allen testified:

"Q237. You mentioned possible uses of a circuit which you built and tested. Could you indicate one or more such uses? A. The particular tests we conducted would best be used as an input device to a magnetic equipment where outside information comes in from various and sundry sources, this outside information being standardized by means of such things as a current drivers [sic] to read into cores and this information then being transferred magnetically into another core which could be the input cores of successive stages of logic.

"Q238. Is this a practical use of the Paivinen circuit which you built and tested? A. Yes.

"Q239. What would be the nature of the subsequent logic, magnetic logic blocks or magnetic blocks into which this circuit feeds?

"Mr. Anderegg: Objected to as irrelevant in connection with the interference.

"A. It would be of similar type circuits, I mean constant-voltage gates and constant-voltage buffers or OR circuits. The two together would comprise a logical family from which it would be possible to design and build a wide variety of computing type devices.

"Q240. Does the output core D in the circuit which you actually built and tested simulate an input core to such a logic block? A. Very well.

"Q241. Might this circuit also feed into a magnetic core shift register?

"Mr. Anderegg: Objected to as speculative and irrelevant.

"A. Certainly. A core is a core, and what is connected on the output of this output core could be more logic, it could be a magnetic shift register, it could be any variety of things.

"Q242. Is this the practical use of this circuit, to feed into a magnetic core shift register? A. Surely.

"Q243. Would output Core D simulate the input core of such a shift register? A. Yes.

"Mr. Anderegg: Objected to.

"Q244. You mentioned that input information fed in to such a circuit would be standardized through current generators in a practical circuit. Does the test which you performed and the current generators which you used simulate such standardized equipment? A. Yes. The current generators that I used were such that, as long as the input was above a given threshold, it would put out this value of current, so that whether it was barely above the threshold or whether it was well above the threshold would be of no consequence to the current generator.

"Q245. And would this simulate the current generators used in the practical use of this circuit which you described? A. Yes, indeed."

After the circuit was built, Allen performed a series of tests in the presence of Paivinen. The equipment was arranged so that a cycle of operations could be repeated at high speed and various waveforms could thus be observed on an oscilloscope. The operations were: inspect pulse (to the inspect windings); reset pulse (to the reset windings); readout of output core; set pulse (to one or more of the set windings); and repeat the sequence. Oscilloscopes were placed across the inspect loop and across the output winding of the output core.

Various combinations of set impulses to the input cores were tested. Paivinen and Allen used the term "coincidence" to denote the condition where all three input cores had been set and "non-coincidence" to indicate the condition where any one or more of the input cores was in the reset state. Thus, under the coincidence condition, all cores would offer low impedance to the inspect pulse and a signal would appear across the output winding of the output core. As summarized by the board:

"* * * It is understood that a switching arrangement was provided so that signal pulses to all the input cores or to some lesser number of them could be selected as desired. The repetition of a condition enabled currents to be examined with an oscilloscope and photographed. * * Some of the oscilloscope photos are in Allen's note book record. * *"

Paivinen testified as follows with respect to the necessary goals of the test program:

"The tests that were conducted had to show two things: first of all that we could prevent information from being developed in or transferred to the output core under the situation where one or more of the input cores was in a high impedence condition; and then as a second condition we had to show that we could transfer information to the output core under the circumstance where all the input cores were in the low impedence condition."

Both Paivinen and Allen stated, at many points in their testimony, that these goals had been successfully met and that the circuit operated as intended. The oscilloscope photographs bear this out. Photograph #1, which depicts voltage developed across the output winding of the output core, clearly shows a signal for the coincidence condition and none for the non-coincidence condition. Photograph #3 shows the same relationship for magnetic flux in the output core.

Photograph #7 shows flux developed in the output core under a condition where the inspect signal duration was intentionally made too great. As testified by Allen:

"Q227. Did this source [of inspect signal] serve to completely switch the core or only partially switch the core in your tests?

"Mr. Anderegg: Objected to as leading.

"A. In the successful tests and as exhibited in Figure 3 [photograph 3], page 9 of Exhibit B, this source did not completely switch the input cores, and it would have been impossible for this circuit to function logically as an AND circuit if it did so.

"In later tests where the duration of the pulse was extended beyond the point where it did completely switch the cores as exhibited in figure 7 [photograph 7], page 10, Exhibit B, we demonstrated improper operation of the gate circuit where the time duration of the inspect pulse was too great."

On the basis of this evidence, the board held that "Paivinen has not proven reduction to practice of the invention defined by the counts in issue." The board found three deficiencies in Paivinen's proof. We shall consider these in the order raised by the board.

"The test pulse source did not operate upon the Paivinen circuit or cause the latter to operate in the manner it would be expected to in practical application. In the tests the same condition was repeated periodically so that current forms produced could be observed on the oscilloscope in manifold superposition. Then the conditions were changed, perhaps stopping the equipment completely, and then when the operation stabilized sufficiently for a steady representation on the scope, that would be taken and recorded as the operation under the changed conditions. This kind of testing might obscure a failure of the device to immediately respond to a change in condition, which it would be required to do in practical application." We think the board was in error here. As appellant points out, there is no evidence that the equipment was shut off for the purpose of altering input conditions. And even if the equipment were shut off, there is nothing to indicate that the circuit would delay in its response to the changed conditions. The board's statement thus does not appear to be based on any supporting evidence and we are unwilling to accept it as anything more than a mere speculative conjecture.

Moreover, as stated in appellant's brief, "there are many uses of the Paivinen circuit which would not even require an immediate response to a change in conditions." As brought out in the record, one of the many possible uses for the circuit could be as a control device for a bottling operation. In commenting on this possibility in his brief, appellant states:

"* * * Considering the fact that the repetition rate employed during the Allen tests was 1200 cycles per second, it can readily be computed that even if the Paivinen circuit in fact failed to respond to a new condition (such as the bottle being capped) for even 12 successive sensing operations (each of which takes 1/1200 of a second) the total time lost would be only about one hundreth of a second. Clearly, in a mechanical operation such as this, a time delay of one hundreth of a second is negligible."

We might also point out, as a matter of judicial notice, that a high rate of repetition would appear to be necessary in order to observe the waveforms on an oscilloscope, at least, as was done here, for the purpose of making photographic records of the waveforms. Such use of the oscilloscope is recognized by those of ordinary skill in this art as an acceptable, if not the only practical means for obtaining a clear picture showing the nature of the waveform produced in the output winding of the output core.

The board's second objection to Paivinen's proof was stated as follows:

"More important is the fact that the device was not shown to operate an output device, as it would be required to do in practice in computational or logical systems. All that was done at the output end was to connect an oscilloscope across the output terminals which were presumptively otherwise unconnected."

In this regard, we agree with appellant that the board was clearly incorrect. The load used in the inspect loop of the tested circuit was, as we have indicated, an output core. Such a core is not recited in any of the six counts. We have already quoted a portion of Allen's testimony wherein he stated that the output core simulated an input core to a subsequent logic block "very well." He also testified that the output core would simulate the input core of a shift register. The testimony clearly points out that among the practical uses for the Paivinen circuit would be the feeding of information into a shift register or into other logic circuits.

Paivinen's testimony likewise contains statements to the same effect. We think the testimony clearly shows, contrary to the board's opinion, that the circuit did in fact operate a typical output device. The fact that the output winding of the output core was unconnected to any further device is immaterial. Further connections to further devices could be made ad infinitum, but it is the first connection, to the output core, that is, significant.

Thus, Elmore v. Schmitt, 278 F.2d 510, 47 CCPA 958, relied on by both the board and appellee, is not controlling. In that case, which involved subject matter very similar to that here in issue, we held that the tests there relied upon to show reduction to practice were insufficient because it was not shown that the test arrangement was designed to simulate any practical application of the invention. In other words, we did not hold that laboratory testing is unacceptable as a means of proving a reduction to practice, but rather that, for such tests to be acceptable proof of such reduction to practice, a relationship must be established which relates the test conditions to the intended functional setting of the invention. In the present case, as distinguished from the situation in Elmore v. Schmitt, we think such a relationship has been clearly established.

The board concluded its objections to Paivinen's proofs by stating:

"Finally, counts 1 to 5 contain a requirement that the pulse supplying means that has a load impedance in series therewith (the inspect pulse circuit of Paivinen) is `incapable of reversing the condition of saturation of said body (core).' The tests described by Allen do not appear to positively establish that condition. * * *"

The board went on to state that, in its opinion, the 80 micro-second inspect pulse "may have carried one or more of the input cores to opposite saturation and stopped there, in which case the limitation in the counts 1 to 5, which is express and positive, would not be met." The point of this argument seems to be that the inspect pulse may have switched one or more of the input cores to the set state (cores which, in the non-coincidence condition, would initially have been in the reset state) without any indication to that effect being observed on the oscilloscope.

This is, perhaps, a possibility, but in none of the evidence we have considered is there any indication that such in fact was the case. Arrayed against this absence of support for the board's conjecture, on the other hand, we have the testimony of Paivinen as follows:

"The most critical condition here is if only one input core is offering such a high impedence, because under that circumstance none of the other cores are assisting in maintaining a very low level of current during this critical time when essentially zero information should be transmitted to the output core.

* * * * * *

"So it is necessary that the circuit be so designed that the inspect pulse duration and amplitude not completely switch a single input core, if that is the only element which is offering a high impedence, during the time that the transfer of information is expected to occur between the input cores A, B, C and the output cores [sic] D.

* * * * * *

"Q106. Was any source employed which was insufficient in amplitude and duration to completely switch the input cores A, B and C?

"Mr. Anderegg: Objected to as leading.

"A. Yes, sir, as discussed previously, this had to be one of the conditions that the inspect pulse fulfills. That is to say, the inspect pulse must be sufficiently short in duration that even under the circumstance where only one input core is offering a high impedence, that that input core does not completely switch during the inspect pulse."

We have previously quoted testimony of Allen to the same effect. We have been presented with no good reason to disbelieve the testimony of either of these men.

A large portion of appellee's brief is given over to an extremely technical discussion which purports to support his doubt that the inspect pulse was, as testified to by Paivinen and Allen, of insufficient duration to completely switch a single core. Paivinen also devotes substantial space in his brief to answering these arguments. We will not discuss these matters here for, as we have indicated, nothing in the evidence we have considered has convinced us that the testimony of Allen and Paivinen was false or erroneous. We are not here passing upon the testimony and documents on behalf of Sands nor Paivinen's rebuttal testimony, since they are not relied upon by the board to support its decision. It is possible therefore that the board may, on remand, find evidence which would prompt it to consider fully and pass upon the technical arguments concerning inspect pulse duration. For the present, however, we are satisfied that the limitations of counts 1 through 5 with regard thereto were fully met by the tests of Allen and Paivinen.

Thus we hold that, on the basis of the evidence actually considered by it, the board erred in concluding that Paivinen had not made a satisfactory showing of actual reduction to practice of the invention of the counts prior to January 1953.

The appealed decision is reversed and the case remanded for a determination of priority based upon a consideration of all the relevant evidence.

Reversed and remanded.