Its two patents1 upon devices for control of an automatic machine tool were infringed, The Bullard Company complained to the District Court, by the General Electric Company.2 Denying the accusation, the latter counterclaimed for a judgment declaring the patents invalid. Decision went for GE solely on non-infringement, without any declaration upon the patents. On Bullard's appeal we affirm.

The patents comprise combination claims only. The master to whom the cause had been referred sustained the patents, and on the finding that GE's mechanism embraced, as equivalents, all of the essential elements of the patents, he held GE to be an infringer. Without passing on the validity of the patents, the District Judge set aside the holding of infringement because, unlike the master, he concluded that GE's electronic means was not an equivalent of the patents' mechanical means of operation.

We do not consider the master's appraisal of the patents or resolve this issue of equivalents, for we think his finding that GE's machine contained all of the vital elements of Bullard's claims is not supported by the evidence and thus clearly erroneous. F.R.Civ.P. 52(a). Because of this failure of the proof we find infringement has not been established.

The Bullard patents successively covered a device which exoterics could describe in the following manner. Its function is to direct and control a machine tool (the movable member) employed in making a part or parts of any mechanical equipment automatically. Further, this direction and control are applicable to general purpose machine tools as contrasted with tools usable for producing only a single item or unit, though repeatedly. Because the device may direct the tool in variable courses it is said to embody an intermittent control. It may be set, in advance, to send the tool into the performance of predetermined missions and, upon completion of one, to enter upon another and so on in sequence without overlapping or resetting. The paths to be travelled by the tool are programmed in the control device by inserting dogs or stops in the slots of a function or selecting drum or like means, and thereafter the dogs or stops, as the drum rotates, engage and activate the appropriate components to set in motion the desired point to point action of the tool. A sample piece of work must first be laid out, and its pattern run off manually, on the machine to ascertain the proper positioning of the dogs for automatically carrying out the intended job.

While the device in each of the patents is automatic, in No. 183 the operation is principally mechanical, while in No. 792 it is largely electronic. The main difference is the use of the motive power in stopping the progressive movement of the tool-- to complete one step and ready ('index') it to undertake the next. This is done in 183 by the use of a companion piece which contacts an aptly positioned dog on a stop or detector drum to terminate that particular movement of the tool, thereupon signalling the function drum to turn to and obey the next set command. In 792 the stopping is achieved through electric circuit impedances.

The novel element, in the master's view, is the machine's 'feed-back' ability, that is its relay to the function drum that it has reached its prescribed endpoint and to set it on the next cycle. On this he says:

'A detailed consideration of Patent '183 leads to the conclusion that the inventive idea of the patent is a method using feed-back of data and comparison of fed-back data with control data for controlling the operation of a machine on an automatic, intermittent, or point-to-point, universally flexible basis, controlling the performance of the machine in the ways in which a human operator would or could control it.'

'The most important aspect of this process in Patent '792 is in the relationship between the movement of the cutter-head (movable member) and the variation in the variable impedance in one leg of the circuit; in Patent '183, the most important aspect was found in the relationship between the cutter head (or movable member), and the movement of its companion element toward the stop on the function (stop or detector?) drum. In the former patent (792), this relationship is made operative by the bridge circuit, or, in other words, by an electronic concept; in the latter patent, this relationship is made operative by a mechanical concept.

'* * * In brief, the Special Master finds as a matter of fact that Patent '792 shares with Patent '183 the inventive idea of effective point-to-point feed back to the control of data provided by the activity of the machine, on which data the control operates to direct the further activties of the machine.'

The accused GE device, known as GENPC, is described by the master in noting its differences from Bullard:

'At first observation, we find a very great difference between the patents in suit and the accused device in the method of giving commands to the machines. The General Electric device uses a punched tape, prepared usually away from the machine the operation of which it is to control, which punched tape is applied to the machine by a 'tape-reader' which translates the symbols on the tape into commands to be followed by the machine. Different feed rates, different axes of motion, and the various requisite command data are indicated to the machine by the punched tape travelling through the tape-reader.

'Other differences of major nature may be elaborated almost toa the length of the transcript of the evidence, but the inquiry here must be directed to the basic idea, or inventive idea, of the accused General Electric device. Inspection of the mode of operation of the General Electric device shows quickly that we find again the idea of data being fed back from the work piece to the machine control. The use of the fedback data in actually controlling the operation of the machine differs somewhat in the General Electric device from that of Patent '792, in that the General Electric device uses the series of selsyns, and it certainly differs in form from that of Patent '183.'

The master held with Bullard throughout, finding the feed-back principle of both its patents as new and as simulated by GE. His reason for finding infringement is thus stated:

'If we limit the discussion for the moment to a comparison of Patent '792 and the accused General Electric device, it is seen immediately that both are operable because of an electronic circuit which controls the point-to-point progress of the work piece by utilization of data fed back to the control from the work piece. It is also seen immediately that the two circuits are not in form the same, nor are the shapes and names of the circuit components the same. * * *'

'* * * But the function of the data feed-back arrangements in all three devices is the same, namely, to provide a system by which the movement of the work piece from point to point feeds information back to the control mechanism, which information is used to control the further activities of the machine, and the Special Master so finds. It might be added that, so far as the evidence shows, this is the only function of the data feed back arrangements.'

'A careful analysis of the evidence offered to show non-infringement reveals it to be more to show or illustrate differences in the names or shapes of the various components and to attempt to show different modes of operation of the various components, but that same evidence shows that the function or purpose of these components within the General Electric device is the same as the purpose or function of the components of the electronic circuit shown in patent '792, and that these functions are accomplished in substantially the same way in the two devices.'

The District Judge apparently distinguished GE from Bullard exclusively on the ground that GE's machine was an adaptation in its operation of electronics in the place of mechanics or partial electronics, representing a radical departure from the Bullard idea. On this premise he concluded that GE's device was not within the range of equivalents of Bullard, and so, GE had not infringed Bullard. GE, however, to sustain its argument of non-infringement, rests on the failure of Bullard to show an imitation by the NPC of the combination patented in 183 and 792.

It is unnecessary in this case, as we have already observed, to decide whether electronic is an equivalent of mechanical operation, for we are not satisfied that GE's device is substantially identical with the patents' prescriptions. The defenses at trial of invalidity and file wrapper estoppel, while not irretrievably waived, were not pressed by GE, leaving its principal reliances as (a) non-infringement and (b) laches and equitable estoppel. As also previously noted, since we find infringement unproved, we have no occasion to pass upon the other defenses.3

A conviction of infringement of a combination claim, it is hornbook, can be procured only upon a showing that the suspect embodies every essential constituent-- identical or equivalent, but not necessarily of the same shape or form-- of the subject of the claim. Entron of Maryland v. Jerrold Electronics, 295 F.2d 670, 677 (4 Cir. 1961). Moreover, the onus of proving corresponding entirety in the assembly is upon the patentee. We are unconvinced that the GE control instrument contains all of the essential features of the Bullard mechanisms.

Fundamentally, GE does not employ the method of programming delineated in the Bullard patents. In 183 and 792 the movement of the tool in accordance with the program is prearranged, as we have noted, by the appropriate placement of dogs or stops upon the function or selecting drum or a similar instrument. In contrast, this type of director is not found in the NPC. Its program executer is a punched tape, in no degree resembling the mechanics of the function or selecting drum. Aside from those enumerated by the master, a marked difference-- a noteworthy advantage-- disproving that NPC apes Bullard's methods is this: the duration of a single project or the number of them that may be stored on 183 and 792 is limited by the number of dogs or stops installable upon the function drum. There is no such physical restriction with tape.

Bullard rejoins that neither the tape nor the drum is within the patented process but is entirely precedent to that operation, and therefore the elimination of the drum by the substitution of the tape is not the omission of an essential part of Bullard's combination. But this argument is not tenable. The function or selecting drum initiating and watching the progress of the program is art and part of the combination. It is integral in the operation; its removal would render the control inoperable. Of course, we recognize that the preparation of the tape, that is its punching before installation, is quite outside the claims of the patent.

Bullard also insists that the function drum is but one of the means to accomplish the flexible intermittent automatic control of the invention, and that the patent claims cover all means to that end. The typical claims in 183 and 792 read as follows:

'13. Apparatus comprising, a member; means for moving said member along a plurality of paths; means for rendering said member-moving means effective; means for independently selecting the path, direction and rate of movement of said member; and a plurality of means adapted selectively to cooperate with means, movable in proportion to the movement of said member, for rendering said selecting means effective, each of said plurality of means being adapted to be rendered ineffective relatively to said proportionally movable means immediately after rendering said selecting means effective.'

'23. In a machine tool, a tool head adapted to be moved in either direction along a path of travel; presettable control means having a plurality of stations, each adapted to be rendered effective by a step-by-step indexing mechanism for controlling the direction, rate and extent of movement of said head along said path; an electronic circuit adapted to cooperate with said presettable means; said pre-settable means being adapted to provide a predetermined signal voltage in said electronic circuit for each station to which it is indexed; means responsive to the movement of said tool head for nullifying said signal voltage; and means responsive to the null condition of said circuit for causing said pre-settable means to be indexed to its next succeeding station.'

We cannot grant so comprehensive a reading of 'means' in the claims; it would preempt all and every conceivable form of programming. The patentee is not entitled to that unlimited exclusiveness. True 'the claims measure the invention', Continental Paper Bag Co. v. Eastern Paper Bag Co., 210 U.S. 405, 419, 28 S. Ct. 748, 52 L. Ed. 1122 (1908), but the claims must be construed in the light of the specifications. Carl Braun, Inc. v. Kendall-Lamar Corporation, 116 F.2d 663, 665 (2 Cir. 1941). Again, the specifications are protective only to the point they give such 'full, clear, concise and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same.' 35 U.S.C. 112; Continental Paper Bag Co., supra, 210 U.S. 419, 28 S. Ct. 748; Smith v. Snow, 294 U.S. 1, 11, 55 S. Ct. 279, 79 L. Ed. 721 (1935). We are not persuaded that the claims and specifications would suggest to the statute's skilled person a tape with the attributes of the NPC's.

When Bullard's programmer is supplanted with a plainly distinguishable substitute, the new aggregation cannot be said to be an infringing duplication of the Bullard combination. Kay Patents Corp. v. Martin Supply Co., 202 F.2d 47, 51 (4 Cir. 1953). Furthermore, even if the two assemblies perform the same functions-- and this is disputed-- GE's cannot be branded a copy, for it does not reach the result, i.e. the programming, in the same way as does Bullard. Without this parallel there is not infringement. Union Paper-Bag Machine Co. v. Murphy, 97 U.S. 120, 125, 24 L. Ed. 935 (1877).

Bullard urges, and the master concluded, that in any event infringement was established by the inclusion in the GE machine of the feed-back principle. Undeniably, this feature is common to Bullard and GE. But the patents are not confined to this element alone. They claim patentability in its combination with many others. This, of course, was Bullard's privilege but the letters patent must be judged by the law of combinations. A later machine omitting an essential of a patented combination is not an infringement of it because the later combination includes an element of the earlier which singly is unpatented, such as the feed-back here, though the element may possess patentability.

As Justice Whittaker said for the Court in Aro Mfg. Co. v. Convertible Top Replacement Co., 365 U.S. 336, 344, 81 S. Ct. 599, 604, 5 L. Ed. 2d 592 (1961):

'* * * For if anything is settled in the patent law, it is that the combination patent covers only the totality of the elements in the claim and that no element, separately viewed, is within the grant. See the Mercoid cases (Mercoid Corp. v. Mid-Continent Investment Co.), supra, 320 U.S. (661) at page 667 (64 S. Ct. 268 at page 272, 88 L. Ed. 376); (Mercoid Corp. v. Minneapolis-Honeywell Regulator Co.) 320 U.S. (680), at page 684 (64 S. Ct. 278 at page 280, 88 L. Ed. 396). The basic fallacy in respondent's position is that it requires the ascribing to one element of the patented combination the status of patented invention in itself. Yet this Court has made it clear in the two Mercoid cases that there is no legally recognizable or protected 'essential' element, 'gist' or 'heart' of the invention in a combination patent.'

For these reasons we do not find infringement.

The procedure pursued in the District Court bred difficulties. At the outset a master was appointed with directions 'to be present and participate in the trial of this case and after the conclusion of the trial and after the submission of briefs and final arguments' to file a report of his findings of fact and conclusions of law. However, the District Judge and the master sat together in hearing the evidence and arguments. Hence it is difficult to identify the 'trial' as that of the master or the judge. To distinguish between the functions of the two officials also is not simple.

Illustrating some results of the confusion, when evidence was rejected by the master in the presence of the District Judge, the ruling was not later put in an exception to the report by the proffering party because it was thought that the point had already been made to the District Judge. Another complexity developed in the appraisal of the findings contained in the master's report. They do not, we are told by the appellee, have the stature before the court accorded by F.R.Civ.P. 53(e)(2), since the trial judge had seen and heard the witnesses for himself, and naturally he must have weighed their credibility independently, quite without reference to the master's assay.

The Rules contemplate a separate and distinct proceeding before the master. If it is a genuine reference, they should be strictly followed; otherwise the trial stages before and subsequent to the report become blurred. Of course, the District Court has the right on an intricate subject of suit, as here, to engage an advisor to attend the trial and assist the court in its comprehension of the case. Danville Tobacco Ass'n v. Bryant-Buckner Associates, Inc., 333 F.2d 202, 206 (4 Cir. 1964). But when there is a merging of master and advisor the result may have a hybrid status. In our consideration we have endeavored to give the master and his proceedings their orthodox place.

Further uncertainty is introduced by the District Judge in neither expressly approving nor disapproving the master's report. The opinion below says only that 'the Court agrees with much of the Special Master's report.' In effect the only departure from the master's conclusion seems to be the Court's holding that GE's were not equivalents of the Bullard devices. A more explicit ruling would have avoided all misunderstanding.

As stated, we conclude infringement was not proved.

Affirmed.