General Instrument Corporation, Defendant, Appellant, v. Hughes Aircraft Company, Plaintiff, Appellee, 399 F.2d 373 (1st Cir. 1968)

Defendant appeals from a judgment of the U. S. District Court for the District of Rhode Island declaring plaintiff's two patents, North and Barnes, valid and infringed. While both patents relate to semiconductor devices, the issues as to each warrant separate discussion.

The North patent (No. 2,694,168, applied for March 31, 1950 by Harper Q. North and Justice N. Carman, Jr., issued November 9, 1954) describes a glass-sealed housing containing a semiconductor crystal¹  connected on one side with a terminal lead wire and on the other with a metal point (a "cat whisker") which is in turn affixed to another terminal wire. The assembly so housed is a diode, or simple semiconductor rectifier with two terminals for use in electrical circuits.² 

The patent proceedings occupied almost five years. Of 106 claims filed, 61 were allowed. Of the 61 allowed, only numbers 60 and 61 are claimed to have been infringed. These claims entered the proceedings for the first time on March 9, 1954, as claims 100-104. After rejection, claims 102 and 103 were rewritten as 105 and 106 and finally accepted as 60 and 61. For present purposes we shall confine our attention, as did the district court, to claim 60. It was this claim, concededly infringed by defendant, which was held valid.

It seems to us particularly important that we not only ask the right questions but ask them in the proper sequence. That sequence would appear to be: (1) what is the invention described by claim 60; (2) does such invention enjoy whatever protection is afforded by the remainder of the North patent; (3) if it does, is the North patent valid against charges of anticipation and obviousness; (4) if so, have plaintiff's rights been limited by estoppel or laches; and (5) even if otherwise valid and not vulnerable to laches, does a newly added portion of the record raise an issue whether or not the patent is invalidated by plaintiff's application for a similar foreign patent without securing a statutorily required license?

The district court defined what it deemed the invention at issue and proceeded to answer questions 2, 3, and 4 favorably to plaintiff. It had no occasion to deal with the fifth issue, which, as we shall see, infra, is one of several significant questions arising from stipulated additions to the record made after the judgment below. Differing with the district court in its answer to the second question, we have no occasion to consider the others.

The definition of the invention described by claim 60 requires almost equal attention to what is and what is not involved. We reproduce in the margin claim 60, as presented in appellant's brief (with the exception of the identifying numbers, which are ours) to distinguish structural from dimensional features.³ 

Paraphrased for convenience, we note the following six structural and four dimensional elements:

 Structural Dimensional 1. Hollow glass uniform tubular 1. Maximum width of container container. "of the order of one tenth inch." 2. Ends of container "solid" and "massive". 2. Ends having substantially same width as rest of container. 3. Two "one piece ductile lead wires", each entering end section from outside, hermetically 3. Each wire to lead through sealed to end section glass end section for a distance through which it leads, and "at least 1.5 times" the diameter of the lead wire. 4. Each terminating within the inner chamber of the glass 4. The ratio of the width of the container. end sections to the diameter of the lead wires to be "at least of the order of" 5 to 1. 5. A semiconductor crystal "affixed to, supported by, and electrically connected to" one wire. 6. And a cat whisker connecting crystal and second wire.

The district court found that the dimensional relationships, rather than being surplusage as plaintiff continues to argue, are "critical * * * contribute in a very significant degree to the commercial success * * * and do define the invention." 275 F. Supp. 961, 968 (D.R.I. 1967). This finding takes on added strength from the fact, not known by the district court at the time of its decision, that plaintiff, one day before issuance of the North patent, had filed for a Patent of Addition in England, seeking to add to the parent British patent (essentially North, without claims 60 and 61) the subject matter of claims 60 and 61. In justifying the Patent of Addition, plaintiff reiterated textually the critical nature of the third and fourth dimensions noted in our schema.⁴ 

We therefore accept the district court's finding that dimensions are critical. While this defines the scope of claim 60, it is also necessary to define what is not included. The district court appropriately noted that such parts of the North patent as referred to a crystal "buried in glass" were not involved in claim 60. Id. at 970. What it did not apparently recognize was that the technique of sealing a metal wire to a glass enclosure by means of, first, encompassing the wire with a bead of glass and, second, fusing the beaded lead with the surrounding glass at a conveniently low temperature was not part of claim 60 and had in fact been surrendered by plaintiff with the cancellation of its original claim 55.⁵  Notwithstanding this, the court makes much of "glass to glass sealing" and the fact that "North used beaded leads fused into the ends of a tubular glass envelope and this resulted in the great strength and ability to withstand the stresses of lead beading and low-temperature glass to glass final sealing." Id. at 972. While it is not clear to us how central a role this factor played in the district court's judgment, we shall assume that it was surplusage. Otherwise, we would have to reverse on the ground that the claim in question had been upheld because of a feature which it did not embrace.⁶ 

Coming, then, to what claim 60 covers — and no more, we face the threshold question: are the structural elements and dimensions of claim 60 implicit in and supported by the specifications of the North patent as originally filed? Defendant, in brief and argument, treats this as breaking down into three questions: (a) are the requirements for exact description of 35 U.S.C. § 112 complied with; (b) can claim 60 be read as not introducing new matter; and (c) does the public use (beginning in January 1953) of the concepts articulated in claim 60 reflect the first disclosures of North or does such use disclose features which were not formally covered until claim 60 was subsequently applied for?

We are drawn to the conclusion that claim 60 is invalid in that it is not linked integrally or essentially enough with what had been set forth in the prior disclosures of the North patent. Since what was old did not describe what was added, the latter must be considered new matter. Harries v. Air King Products Co., 183 F.2d 158 (2d Cir. 1950). And since commercial sales of the "Hughes commercial diode", which embodied all the "new matter" embraced in claim 60, had commenced over a year prior to the filing thereof the claim fails for "late claiming". 35 U.S.C. § 102(b); Muncie Gear Works, Inc. v. Outboard Marine & Mfg. Co., 315 U.S. 759, 62 S. Ct. 865, 86 L. Ed. 1171 (1942).

Our conclusion rests in part on the interpretation of words used in the North patent which we have felt compelled to make and which differs from that adopted by the district court, in part on our view that claim 60 is not merely a narrower version of broader original claims, and in part on the significance of the nearly contemporaneous filing in Great Britain for the Patent of Addition covering substantially the identical ground of claim 60.

Of the six structural and four dimensional elements we have identified in claim 60, we pass as having no significance here structural items 1, 2,⁷  and 6, and dimension 2. On each of the remaining items, we differ from the findings below — not on credibility of witnesses but in the interpretation of patent documents. We do not say that each difference indicates error which would alone be reversible, but only that the aggregation of these six differences, together with the remaining grounds relied on, compel the result.

Commencing with the findings on structural elements, we note that the court below concluded as to element 3 that two wires, 10 and 12,⁸  of differing diameters and composition, constituted when welded together, a "one piece structure". But claim 60 describes two "one piece ductile lead wires". If a two-wire "structure" can qualify as a "one piece wire", we cannot imagine what a "two piece wire" might be. What confirms us in our view is data not available to the district court. In seeking its foreign Patent of Addition, plaintiff referred to a defect in its parent patent (i. e., the British patent identical to North) in calling for "composite leads * * * of Dumet wire * * * welded to thicker copper wires", the weld resulting in a weak point. It therefore sought "one piece flexible Dumet leads * * * thereby eliminating the difficulties * * * with composite leads." To accomplish this change it used the "one piece ductile" language of claim 60.

As to structural element 4, the district court found that the North patent, excluding claim 60, disclosed each lead wire as terminating within the inner chamber of the glass envelope. It reached this conclusion by reasoning that (1) wire 12 terminated at or in a layer of silver paste; (2) the paste, being a conductor of electricity, cannot be part of the glass envelope — which does not conduct electricity; (3) therefore, "it can be agreed beyond doubt that the lead wire does in fact terminate within the chamber that is defined by the glass house." Id. at 968. The court may well be justified in its conclusion that the wire did not terminate at the glass envelope. But to reason that all that is not envelope must be inner chamber is to overlook the fact that "chamber" connotes enclosed space and that the wire which the court had in mind was separated from the inner space by (1) a glass embedded crystal (#16 in Fig. 14 of North patent), (2) a layer of copper (#18), (3) a layer of silver (#20), (4) a silvered glass bond (#22), and (5) a copper plate (#15) on the end of the wire.

This finding of the district court is closely associated with its finding as to the fifth structural element that the semiconductor crystal is disclosed in North as "supported by" a lead wire. The court, after making its finding, above noted, that North shows the lead wire terminating within the chamber stated: "This being so, the semiconductor element is supported by said first end of said lead wires. This court, without hesitancy, finds this as an established fact." Id.

Even were we to accept the court's definition of "chamber", this would not answer the question whether North shows the lead wire as "supporting" the crystal. In the first place, as we have just noted, there are no fewer than four layers of metal, paste, and bond between wire and crystal. In the second place, the North disclosure is of a rectangular crystal with five of its six surfaces completely embedded in glass. When, pursuant to the court's questioning, defendant's witness went through mental gymnastics which involved the imaginary cutting away of all the glass and conceded that conceivably there was some support given the crystal from the wire, the resulting concession comes as close to being of the scintilla variety as we can imagine. Again, we have an unfair advantage over the district court in having available plaintiff's British Patent of Addition which, in explaining the claim 60 "supported by" language, states that "the semiconductor body is bonded only to the end of its associated lead wire and has no direct connection to the glass envelope." Moreover, the drawing accompanying the Patent of Addition shows clearly single-piece wires extending well into the inner chamber of the glass envelope, one of such wires providing the sole support of the semiconductor crystal.

Moving now to the first dimension (which confines the entire envelope to a maximum width "of the order of one tenth inch") we are referred to two echoes of this part of claim 60 in the rest of the North patent. One statement refers to the fact that the invention makes possible sizes of glass envelopes reduced to "the order of .09" " in diameter. The other reference is to a drawing illustrating the sealing of bead to cylinder, which merely states that the cylinder "in the illustrated example" has "an outside diameter of .09" ". The district court noted that a defense witness conceded that the references to .09" gave support to the "order of one tenth inch" language in claim 60.⁹  The only further explanation for the finding as to this dimension is the holding itself: "The court reads this dimension in direct relation to the word `maximum' * * *." Id. at 967. But saying so does not make it so if in truth there is, as here, no concept of maximum previously articulated. Appellee has "drawn a line where no line existed before." Girdler Corp. v. E. I. DuPont de Nemours & Co., 59 F. Supp. 583 (D. Del. 1945), aff'd 152 F.2d 757 (3d Cir. 1946).

The district court based its finding that the third noted dimension of claim 60 — as to length of seal being "at least 1.5 times" the diameter of the lead wire — was adequately supported in the North patent on the admission of defendant's witness in cross-examination that "the patent does disclose * * * a seal * * * equal or greater than 1½ times the diameter of the lead." The court was satisfied that "this is the very essence and meaning and purpose of the patent claim in this regard." 275 F. Supp. at 968. To say that a ratio computed from measurements (partly taken from observation of a drawing) which proves to be equal to or larger than 1.5 to 1 is also "at least" 1.5 to 1 is to state the observed relationship accurately. But such an observation says nothing about a requirement that 1.5 to 1 be the smallest acceptable ratio. See Harries v. Air King Products Co., supra, 183 F.2d at 161. To assume a requirement from a relationship which, so far as the patent's teaching is concerned, is fortuitous is to make the same impermissible leap noted in discussing the first dimension.

Here, however, there is lacking even the predicate of the diameter of "the" lead wire. As we have noted, claim 60 seems to contemplate a lead wire which is a unit in all respects, rather than a welded assembly of two different wires. That conclusion is strengthened here, since we have, in the North patent prior to the introduction of claim 60, an illustrative (not a normative) diameter for only the external segment of the lead wire, and no dimension for the internal segment.¹⁰  Moreover, the length of seal — the numerator of the ratio — is derived only after making assumptions based on the observed shape and cut of a bead. Finally, as in other elements of claim 60, we note that plaintiff, in its British Patent of Addition, made explicit the importance of a long seal preferably "in excess of one and one-half times the diameter of the Dumet lead", pointing out that shorter seals cause variations in the contact between whisker and semiconductor. What is left to assumption, measurement of illustrations, and approximation in North is accomplished directly, explicitly, and precisely in the British proceedings.

The fourth dimension — that the ratio of width of end sections to diameter of the lead wires be "at least of the order of" 5 to 1 — was, as in the case of the first and third, found to be prefigured in the North patent from the coincidence of the ratio derived from illustrative measurements with the prescription in claim 60. At best, indulging in all favorable assumptions for plaintiff, we have an external end section width of .09" and a wire diameter of .02", or a ratio of 4.5 to 1 which does not seem to us to teach very clearly a ratio "at least of the order of" 5 to 1.¹¹  Again, however, there is the the ambiguity caused by the differing diameters of the lead wire components. Further, there is no basis for reading illustrative dimensions as having a normative teaching. And here too we find explicit explanation for the necessity of the minimum ratio in plaintiff's British Patent of Addition.

In justifying resort to the North patent drawings in the above manner to find support for the dimensional recitations in claim 60, the district court cited Permutit Co. v. Graver Corp., 284 U.S. 52, 52 S. Ct. 53, 76 L. Ed. 163 (1931), and National Latex Products Co. v. Sun Rubber Co., 274 F.2d 224 (6th Cir.), cert. denied 362 U.S. 989, 80 S. Ct. 1078, 4 L. Ed. 2d 1022 (1960). But Latex merely held that reference to another patent, together with a drawing, was sufficient to disclose the invention described in later claims. And Permutit seems to us to be authority for rejecting a drawing even though it is consistent with a later asserted concept if there is no contemporaneous suggestion of the concept.

Apart from finding that the elements of claim 60 all found support in the prior disclosures of the North patent, the district court also held that they "did not constitute initial claiming of subject matter previously unclaimed but merely narrower definition of subject matter originally disclosed." 275 F. Supp. at 973. We need not concern ourselves with the specific claims cited by the court (44, 46, and 48) or by the plaintiff (45, 50). Our holding that claim 60 goes substantially beyond the invention claimed in the original application necessarily applies to all its parts.

If the features of claim 60, as we hold, are not supported by other parts of the North patent and are not merely narrower exemplifications of the earlier claims, then under the doctrine of Muncie Gear, Inc. v. Outboard Mfg. Co., supra, the claim must fail. For, as the district court found, a diode manufactured by plaintiff and reflecting the features of claim 60, has been in public use or on sale for more than a year prior to the filing of claim 60.

In view of our conclusions that the claim 60 disclosures are really new matter, not narrower claims foreshadowed or supported by prior disclosures, we have no occasion to consider the questions of anticipation or obviousness or, a fortiori, the weight to be given the commercial success enjoyed by plaintiff's product; or defendant's arguments as to estoppel and laches. Nor do we consider defendant's assertion that an additional reason for invalidity might lie in plaintiff's action in acquiring a British patent if, contrary to 35 U.S.C. §§ 184, 185, no license had been first obtained.

Finally, defendant seeks a remand to the district court for the award of reasonable attorney fees under 35 U.S.C. § 285, principally because of plaintiff's failure to reveal at an early date the existence of the British patent proceedings. In answer to defendant's interrogatories, plaintiff responded on March 19, 1964 that an application for a patent corresponding to North had been filed in Great Britain in 1951 and that patent 721,201 had been granted in 1955. Omitted was the salient fact that the British Patent of Addition had been applied for on November 8, 1954. Only in February 1968 did this come to the attention of plaintiff's counsel who immediately made disclosure to defendant. The only explanation proffered was that "* * * no one now at Hughes was aware of its existence until the patent was cited in an interrogatory recently served on Hughes by ITT." We are told in plaintiff's brief that this was a "good-faith error caused by passage of time, personnel turnover, and lack of sufficient cross-referencing in plaintiff's files."

Leaving bad faith aside, it strains credulity to believe that a substantial business enterprise not only could err in 1964 but, during important litigation over the subsequent four years in which the subject matter of the missing patent application was at the very heart of the issues, could remain oblivious to its existence. To us this is the type of exceptional oversight which makes an exceptional case under 35 U.S.C. § 285. The lack of such critically important information effected a grave disservice both to the opposing party and to the district court, as we have indicated by our frequent references to subordinate issues of interpretation as to which the contemporaneous British filing was relevant. It seems to us that the ineptitude here is fully as serious as was that in Russell Box Co. v. Grant Paper Box Co., 203 F.2d 177 (1st Cir. 1953), and requires the same result. See also Plymouth Rubber Co. v. Minnesota Mining & Mfg. Co., 203 F. Supp. 595 (D. Mass. 1962), aff'd 321 F.2d 151 (1st Cir. 1963). Reasonable attorneys' fees will be awarded defendant on this branch of the case.

The second patent here in issue, Barnes,¹²  deals with a semiconductor device in which a P-type material (i. e., germanium so treated that it is characterized by a deficit of electrons) is fused with an alloy of a donor impurity containing a donor substance such as antimony. This donor element contributes negative electrons to germanium so as to produce an N-type germanium in the area of interaction, the result being a P-type germanium adjacent to an N-type region, or, if another N-type region is created, an N-P-N junction transistor.

The basic concept of the Barnes patent lies in using a metallic alloy including an active impurity of a donor type and an element which dissolves germanium, rather than a "pure" impurity. The alloy specifically taught in Barnes is comprised of a donor impurity such as antimony (the function of which is to provide sufficient donor atoms to the germanium to produce an N-type region), and germanium "wetting agents" such as bismuth and tin (the twin functions of which are to produce a melting point for the alloy which will be high enough to protect the completed device from damage due to high temperatures met in operation or subsequent manufacturing steps, and low enough to avoid damage through excessive heating of the germanium). The alloy also must be capable of melting or dissolving germanium so that the alloy and germanium will coalesce resulting in a fusion of the two substances and some diffusion of donor atoms into the P-type germanium region.

A pellet containing the alloy is placed on a crystal of P-type germanium and subjected to controlled heat which melts both the alloy and part of the crystal. After cooling, an indented area under the pellet is now an N-type germanium region, containing donor atoms fused with germanium, with some additional diffusion of such atoms into that part of the crystal which was not molten. By repeating the process on the reverse side of the crystal, an N-P-N device is created.

The claims of Barnes held to have been valid and infringed by defendant, are described in the margin.¹³  Defendant's infringing products were transistors using alloy buttons consisting of lead (90%) and antimony (10%).

Of the findings made by the district court, the following are in issue on this appeal: that the Barnes patent is valid; that transistor claims 10 and 11, when read in the light of Barnes' basic teaching, are valid; and that defendant's transistors infringe the Barnes patent.

On the issue of validity, defendant contends that three prior patents fully anticipated Barnes and also that these patents together with similar experiments by others using alloys with germanium to form P-N junctions, rendered Barnes obvious. In reviewing the conclusions of the district court, we acknowledge that the problem area attacked by Barnes in his patent was at the time an active field. The questions of novelty and obviousness raised by Barnes' contributions are close ones. On balance, we conclude that their resolution by the district court must stand, whatever might have been our views had we been the triers in the first instance.

The thrust of Barnes' teaching, insofar as it added to prior art, is not merely that alloys containing a donor impurity and other metals coalesce more effectively with germanium than a "pure impurity". It is also a combination of increased understanding of the process (principally fusion, secondarily diffusion)¹⁴  of that coalescence, the consequent prerequisites of the mix of the alloy and of the timing and temperatures of the heating operation, resulting in and aimed at devices capable of being mass produced and of versatile use. These teachings find their expression in the language of claims 2 and 4 which refers to an "indentation" in the germanium specimen and to an alloy or alloy button containing "an element in which germanium is soluble when said alloy is molten."

The alloy contains "wetting agents" such as tin or tin and bismuth (which melts at a high enough temperature to be impervious to such subsequent steps of manufacturing as soldering and to the heat expectable in using the device, and yet which melts at a temperature well below the melting point of germanium) and a donor impurity such as antimony (which provides an adequate supply of donor atoms within the germanium). The controlled application of heat to a button or plate of the alloy juxtaposed to a germanium wafer or crystal, followed by a cooling period, results in the germanium melting and fusing with the donor impurity and also being distributed in diminishing concentration toward the exterior surface of the alloy button or plate. The end product, if an alloy button or pellet is used is a junction (or boundary) between a P region and a N region surrounding the button which is now firmly embedded in the indented crystal. Repeating the process on the other side of the crystal yields an N-P-N transistor.

Some of the advantages of this end product over prior types of N-P-N junction devices are: (1) that the protruding button — not having been subjected to the high temperature required to melt pure antimony and the subsequent stresses of the different rates of contraction of pure antimony and germanium — was less likely to fracture, was firmly embedded, and provided a convenient area for the attaching of electrodes; (2) that the button could be ground flush with the surface of the crystal, permitting a flat surfaced transistor; and (3) that such buttons imbedded on both sides of a crystal would form a biconcave transistor, having the advantages of adequate surface area on the sides to attach emitter and collector electrodes, a minimum thickness (and resistance) of the central P region, and yet an adequate P region at an end to allow the attachment of a base electrode.

While, arguably, all of this can be described as mechanical improvements over prior art falling short of invention, we think this falls in the crepuscular area where a district court's judgment to the contrary is permissible. The major relevant patent in the prior art was granted to one Dunlap.¹⁵ 

Dunlap's patent related to the use of germanium as the photosensitive element of a photocell, utilizing the photo-electric effect created when light strikes a germanium P-N junction unit. The principal teaching is that a very thin film of an acceptor impurity placed on one face of a P-type germanium wafer will, under the application of heat, accomplish an "alloying and diffusion" of the impurity film into the wafer, creating a P-N junction. Heat applied to a plate or solder of an opposite (donor) impurity adjacent to the reverse face of the wafer will result in a "very slight alloying and diffusion of the impurity into the wafer." The purpose of this limited alloying and diffusion is not to create a junction but to furnish or absorb electrons as may be necessary to provide good conduction (i. e., an ohmic contact). The purpose of the device is to utilize and control the voltage produced by the impingement of light upon the P-N junction.

While this is Dunlap's principal teaching, he moves into the field claimed by Barnes by stating (1) that his process can, by reversing the donor and acceptor, be used on N-type germanium; (2) that a substitute for a film of pure impurity to create the P-N junction can be a drop, which (3) may be conveniently in the form of a solder of 85 per cent lead and 15 per cent antimony, which (4) can be diffused into the wafer by a suitable heating cycle.

Dunlap is like the chapters in a good detective story of which Barnes is the final chapter. That is, having read the denouement in Barnes, one can now find most of the clues in Dunlap. But they are not so explicit, so instructive, or so obvious that Dunlap must be held to anticipate Barnes or render the teaching of Barnes obvious. The P-N junction formed in Dunlap was not in the shape of a concave indentation in the crystal. It was either a planar area, lying under and parallel to the impurity film remaining on the surface of the crystal, or, if none of the film remained, on the surface. An N-P-N device formed by using an impurity film would not enjoy the advantages of the biconcave device described by Barnes. Dunlap also describes using an impurity drop but this is not associated with an indentation in the crystal except for one created by a separate process of etching and for a different purpose.

There is, moreover, no teaching directed to the temperature which a lead and antimony alloy must reach to form a P-N junction with germanium. Dunlap gives an instruction that such an alloy need be heated only to 250° C. to produce the "very slight alloying and diffusion" necessary to insure a good conductive contact on the side of the wafer opposite to that containing the P-N junction. He also gives an instruction relating to the time and temperature required for the heating of a film of antimony to form a P-N junction. But he gives no instruction pertinent to the heating of an alloy film or alloy drop with germanium to form a rectifying junction. The objective of melting germanium at a sufficiently low temperature to avoid damage to it and the fracturing of a button caused by differential rates of contraction of antimony and germanium is not addressed either directly or by necessary implication.

There are two other patents in the prior art on which defendant relies, both granted to W. G. Pfann — No. 2,792,538, applied for on September 14, 1950 and issued on May 14, 1957, and No. 2,950,425, applied for on September 14, 1950 and issued on August 23, 1960. '538, insofar as relevant here, describes a conductor in the form of a wire, composed of an impurity alloy, which is bonded with the germanium by being heated and pressed into its body. This is a point contact translating device. While the result is a P-N junction, in a concave declivity in the crystal, the teaching not only contemplates quite a different mode of manufacture from that described in Barnes, but is limited to P-N junctions with an area not much larger than the diameter of the wire used. Barnes has no such limitation on its versatility.¹⁶  We cannot say, as a matter of law, that '538 teaches or makes obvious the process or device claimed in Barnes.

Pfann '425 is even more remote from the problem attacked by Barnes. '425 is concerned with varying the composition of collector and emitter wire electrodes to control their operating characteristics. While discussing various combinations of alloys to control impedance, multiplication and the injection of electrical carriers, it does not purport to deal with the technique of joining the electrodes to the semiconductor body.

We therefore affirm the district court's conclusion that these three prior art patents did not anticipate or make obvious the teaching of Barnes claims 2 and 4. Other evidence relevant to obviousness included the experimental work of Dr. Jenny (whose application for a patent involving the use of an antimony-lead alloy was rejected as reading on Dunlap) and other work of personnel of the General Electric Company. We have considered this evidence, including entries in notebooks. We conclude that, while a number of researchers were working with alloys in an effort to produce an effective N-P-N junction device, "In this situation, although, within limits, a question of law, the determination whether a discovery of a new combination is or is not obvious must be a question of fact. The [defendant] has failed to persuade us that the district court employed a wrong rule, or exceeded proper limits." Koppers, Inc. v. Foster Grant, Inc., 396 F.2d 370 (1st Cir., filed June 6, 1968).

We have exempted from our affirmance the district court's finding that claims 10 and 11 of the Barnes patent are valid. Claim 10, see n. 13 supra, referred to the metallic button as containing only germanium and a donor impurity. It omitted any reference to an alloy or to an element thereof in which germanium is soluble. Claim 11 reinstated a reference to an alloy but omitted any reference to an element in which germanium was soluble, apparently substituting as the sole requirement of the alloy that its melting point be between 150° C. and 500° C.

The district court dealt only with claim 10 and held that "read in the light of Barnes' specification this language ["a donor impurity"] refers to the `donor alloy' defined in the specification * * *." 275 F. Supp. at 981. It noted that "To interpret the claim language as including an alloy consisting solely of antimony and germanium would be inconsistent with the patent's basic teaching that buttons of substantially pure antimony are unsatisfactory." Id.

We agree with the court's observation about the patent's basic teaching but we cannot make the same jump. Where ten of the twelve claims made clear that an alloy including a donor impurity was the invention, we do not treat lightly the exclusion of the concept from one of them. The cases cited by plaintiff, Carnegie Steel Co. v. Cambria Iron Co., 185 U.S. 403, 22 S. Ct. 698, 46 L. Ed. 968 (1902), and National Development Co. v. Lawson-Poster Shoe Machinery Corp., 129 F.2d 255 (1st Cir. 1942), do not go so far as to allow us to draw upon specifications to supply a claim with a missing crucial ingredient. We make the same observation as to claim 11. That an alloy may melt between certain temperatures does not convey the essential information that it must also be able to dissolve germanium between those temperatures.

We come now to the finding of the district court that defendant infringed all four claims of the Barnes patent here in issue. We are obliged to reverse.¹⁷  We do so on the basis that Barnes, in the course of the Patent Office proceedings resulting in issuance of his patent, limited the scope of his invention to exclude the kind of alloy — antimony and lead — which defendant uses in its product.

We regret that this issue was not specifically called to the district court's attention. The issue of infringement was argued below on the factual ground, i. e., whether lead was or was not an element in which germanium was soluble. The court found that lead "fully satisfies the conditions for a suitable alloy that the Barnes specification prescribes, and they (sic) respond in every respect to the Barnes' claims." Id. at 978. In other words the district court found that lead was covered by the doctrine of equivalents. The ground for our decision, "file wrapper estoppel", is, as we have said before, one of those "several limiting principles which strike a compromise between the doctrine [of equivalents] and the alternative of holding a patentee to the language of his claims." Borg-Warner Corp. v. Paragon Gear Works, Inc., 355 F.2d 400, 404-405 (1st Cir. 1965), cert. dismissed 384 U.S. 935, 86 S. Ct. 1461, 16 L. Ed. 2d 536 (1966).

Were this doctrine only that of "estoppel" and nothing more, we would be inclined to treat this as a defense which, not having been asserted below, is deemed waived. But this doctrine is more. "It is a rule of patent construction consistently observed that a claim in a patent as allowed must be read and interpreted with reference to claims that have been cancelled or rejected, and the claims allowed cannot by construction be read to cover what was thus eliminated from the patent." Schriber-Schroth Co. v. Cleveland Trust Co., 311 U.S. 211, 220-221, 61 S. Ct. 235, 239, 85 L. Ed. 132 (1940). A reviewing court — and a trial court — must therefore scrutinize the Patent Office proceedings with as much care as it does the disclosures and claims in the patent itself. Moreover, what is at issue is not merely a contest between the parties but a public interest that cancelled or rejected claims not be revived and restored. Schriber Co. v. Cleveland Trust Co., supra at 221, 61 S. Ct. 235. We therefore, as in the past, feel bound to review the file wrapper history even though not ventilated in the district court. Cf. Hap Corp. v. Heyman Mfg. Co., 311 F.2d 839, 844 (1st Cir. 1962), cert. denied 373 U.S. 903, 83 S. Ct. 1290, 10 L. Ed. 2d 198 (1963); Unistrut Corp. v. Power, 280 F.2d 18, 21 (1st Cir. 1960).¹⁸ 

Barnes' claim 28 (which became claim 2) originally called for an alloy button containing "germanium and an active impurity of the donor type." This was first rejected by the examiner "on Dunlap alone or Dunlap in view of Pfann of record. The use of antimony as a donor in the button of Dunlap to give a reversed conductivity is clearly suggested by Dunlap as well as Pfann." Barnes then changed his language by adding to that above quoted "and an element in which germanium is soluble when said alloy is molten." Supporting the amendment, counsel argued that "Dunlap in describing the formation of a P-N junction utilizes a layer or drop of impurity (column 4, lines 71 to 75, column 5, lines 1 to 10) such as indium (column 9, lines 42 to 43) which is listed to cause diffusion. * * * Dunlap * * * has no alloy button containing an element in which germanium is solvent * * *."

Barnes' reference to Dunlap's "layer" was quite correct. The layer, or film, was either a pure donor such as indium or pure acceptor such as antimony. Barnes faced a dilemma. If he confined himself to distinguishing his invention only from Dunlap's method of creating a rectifying junction by using a single impurity element, he would have been subject to rejection as not taking account of Dunlap's provisions contemplating the use of solder in forming such a junction.¹⁹  On the other hand, if he attempted to distinguish Dunlap's alternative method, he would have to reckon with an alloy. He chose the latter course, apparently trying to distinguish Dunlap's specific alloy from the generality of alloys he thought suitable on the ground that Dunlap's alloy lacked a germanium solvent. His reference to "drop of impurity", therefore, was not restricted to a drop of indium, but must be held to include Dunlap's teaching that a drop may also be "in the form of a solder" and that a suitable solder consisted of "85% lead and 15% antimony".²⁰  When, therefore, he justified his invention by distinguishing his new language (added also to what became claim 4) as involving ingredients differing from those described in Dunlap, he effectively took lead out of the category of elements in which germanium is soluble so far as infringement of his patent was concerned. Defendant's device falls within the description of what Barnes surrendered.²¹ 

That part of the judgment which held valid claims 2 and 4 of the Barnes patent, No. 2,742,383, will be affirmed. All other parts of the judgment will be reversed, and the case remanded to the district court with directions to enter a decree in accordance with this opinion. The defendant is awarded full costs in this court, together with reasonable attorneys' fees for that part of this appeal relating to the North patent. On this issue we will receive written submissions, the first to be filed by defendant in ten days.