BIGGS, Circuit Judge.

The court below, at its No. 273, found that Minnesota Mining Manufacturing Company and Byron J. Oakes, were entitled to claims 1 to 6, inclusive, and claim 10 of Patent No. 2,111,006 issued March 15, 1938, to Norman P. Robie. At its No. 252 the court below found that Minnesota Mining and Oakes were entitled to claims 1 and 2 of Patent No. 2,111,272 issued March 15, 1938, to Paul M. Paulson. Minnesota Mining and Oakes were the plaintiffs in the two actions brought under R.S. Section 4915, 35 U.S.C.A. § 63, to compel the Commissioner of Patents to issue a patent embodying the claims enumerated to Minnesota Mining and Oakes. The defendants in the respective actions, the Carborundum Company at No. 273, and the Carborundum Company and E.I. DuPont De Nemours Company at No. 252, have appealed. The appeal from the judgment of the District Court at No. 273 is at our No. 8923 and the appeal from the judgment at No. 252 is at our No. 8922. Since the cases were consolidated for trial in the court below it is appropriate to dispose of the appeals in one opinion.

The findings of fact, conclusions of law and discussion of the court below, reported in D.C., 59 F. Supp. 709-720, are so fully set forth that to avoid needless repetition reference is here made to that opinion. We think it is sufficient to state that questions of priority of invention having arisen between Oakes and Robie and Oakes and Paulson there were extended proceedings in the Patent Office. Oakes claimed that he was entitled to copy into his original application filed August 24, 1931, claims 1 to 10 of Robie's patent and claims 1 and 2 of Paulson's patent. Robie's patent was for "Abrasive Products and Methods of Manufacturing the Same". Paulson's patent was for "Abrasive Material". Oakes' patent application No. 558,872 filed on August 24, 1931, was for "Vinyl Resin Adhesive Sandpaper" and claimed, inter alia, a bond, consisting of a "polymerized vinyl compound", to hold the abrasive material in place. Robie's claims 5 and 10 and Paulson's claims 1 and 2 appropriated polyvinyl alcohol as a binder. Robie's claims 1 to 4 inclusive and claim 6 appropriated a polymerized vinyl compound or a polyvinyl compound as the bond.

The Patent Office tribunals decided that Oakes was not entitled to preempt Robie's or Paulson's claims concluding in the Robie interference that Oakes by his 1931 application could not "disclose a small number of chemical agents, state that they fall in a certain genus and later claim specifically a previously unnamed unspecified member of that genus which somebody else has found to possess desirable qualities.", and that Oakes' general monomeric and polymeric formulas, discussed hereinafter in some detail, could not be deemed to include polyvinyl alcohol. A like decision on substantially similar grounds was reached in the Paulson interference which was based on an application filed by Oakes in 1938, Oakes' 1938 application containing the substance of his 1931 application and being a continuation in part thereof.

As we have stated only two of Robie's claims in issue, viz., 5 and 10, appropriated polyvinyl alcohol as a bond. The others claimed a "polymerized vinyl compound". None the less the parties to the Oakes v. Robie interference and the tribunals of the Patent Office seem to have agreed that if Oakes' 1931 application had disclosed polyvinyl alcohol, Oakes would have been entitled to all of Robie's claims sought to be preempted whether appropriating polyvinyl alcohol or not. Like positions seem to have been taken by the parties in the court below. The court concluded that Oakes' 1931 application did disclose the use of polyvinyl alcohol as the bond or binder and accordingly it adjudged Oakes to be the first original and lawful inventor of the invention described in claims 1 to 6 inclusive and claim 10 of Robie's patent and in claims 1 and 2 of Paulson's patent.

At the trial in the District Court the plaintiffs offered as new evidence the testimony of certain experts, to be discussed more at length hereinafter. By this testimony the plaintiffs sought to show that Oakes in the specification of his 1931 application did disclose the use of polyvinyl alcohol as a bond. The defendants objected to the receipt of this evidence, relying on Barrett Co. v. Koppers Co., 3 Cir., 22 F.2d 395. The defendants assert in substance that the expert testimony offered in the District Court was available to the plaintiffs at the time of the proceedings in the Patent Office and therefore should not have been received by the District Court. But as the learned District Judge pointed out the circumstances of the cited case are clearly distinguishable from those of the cases at bar for in the Koppers case the plaintiffs in the R.S. Section 4915 proceeding had instructed their witnesses in the interference not to answer questions which would disclose subsequent commercial practices and, thereafter, had endeavored to offer this very testimony in the District Court. R.S. Section 4915 provides that the trial before the District Court "shall be without prejudice * * * to the right of the parties to take further testimony". It should be noted that prior to the 1927 amendment to R.S. Section 4915, 44 Stat. 1336, Sec. 11, the evidence given in the Patent Office in an interference proceeding could be introduced in a District Court only as secondary evidence, after proper foundation laid. General Talking Pictures Corp. v. American Tri-Ergon Corp., 3 Cir., 96 F.2d 800, 811-812. To adopt the view contended for here by the defendants would be to change the nature of an R.S. Section 4915 proceeding and to rewrite the statute. Accordingly we reject it.

After stating that the narrow issue presented is whether page 5 of Oakes' 1931 application "makes a disclosure of PVA [polyvinyl alcohol], i.e., a polyvinyl compound containing sufficient hydroxyl groups to be self-dispersible in water and usable as a bond for sandpaper.", and that "Other issues are substantially similar and will be decided by the determination of the issue posited.", the trial court went on to say, "This being so the issues of law drop from the case. What we have at bottom, is a fact question." We conclude that the court below, to some degree at least, misunderstood the principle of Morgan v. Daniels, 153 U.S. 120, 125, 14 S.Ct. 772, 773, 38 L.Ed. 657. Mr. Justice Brewer stated the heart of the rule as follows: "Upon principle and authority, * * * it must be laid down as a rule that, where the question decided in the patent office is one between contesting parties as to priority of invention, the decision there made must be accepted as controlling upon that question of fact in any subsequent suit between the same parties, unless the contrary is established by testimony which in character and amount carries thorough conviction."

Morgan v. Daniels states a rule of law whereby a fact question must be judged. The question therefore is whether all competent evidence, "new" and "old", offered to the District Court carries "thorough conviction" that the Patent Office erred. General Talking Pictures Corp. v. American T. Corp., supra, at page 812 of 96 F.2d. A proceeding under R.S. Section 4915 is an anomaly. The trial in a District Court is a trial de novo. But the decision of the Patent Office may not be reversed unless the evidence before the court carries thorough conviction that the Patent Office erred. In other words, a District Court may not reverse the decision of the Patent Office on a mere preponderance of the evidence. More is required. The District Court must find that the decision of the Patent Office was without substantial basis in the evidence or was wrong as a matter of law. In the cases at bar the trial court does not seem to have applied such a rule. Though it carefully weighed all the testimony the district court seems to have reached its conclusions on what it deemed to be a preponderance of the evidence.

But quite apart from the foregoing the words of a patent or a patent application, like the words of specific claims therein, always raise a question of law for the court and may not be determined by the opinion of experts. Sanitary Refrigerator Co. v. Winters, 280 U.S. 30, 50 S.Ct. 9, 74 L.Ed. 147; Solomon v. Renstrom, 8 Cir., 150 F.2d 805, 807. The fundamental question in the cases at bar is whether Oakes' 1931 application fulfilled the requirements of R.S. Section 4888, 35 U.S.C.A. § 33; viz., whether Oakes in his 1931 application described his invention or discovery "in such full, clear, concise and exact" terms as to enable one skilled in the art to ascertain its nature. This is a question of law, open to this court, precisely as it was open to the court below.

Turning now to Oakes' 1931 application we find, as has been stated, that it is for a "Vinyl Resin Adhesive Sandpaper", the bond for which is to be supplied by "* * * condensation or polymerization products of organic compounds having the ethylene linkage which exhibits the property of polymerization under the influence of heat, light, pressure, etc., or combinations of the same." The application teaches that the compounds employed as bonds shall be waterproof and not water soluble; that they shall be thermoplastic, i.e., that they shall soften on heating, and that they shall be soluble in volatile organic solvents.

The application goes on to state: "In general, it is contemplated by my invention to provide an abrasive article in the nature of sandpaper, including a filler, sealing, backing or sizing, impregnated or binder coats of a polymerization or resinified product of the organic compounds characterized by the ethylene linkage, such as the aliphatic organic compounds, characterized by an ethylene linkage or the aromatic substituted aliphatic organic compounds including an ethylene linkage of which vinyl compounds, styrols, crotonylenes may be cited as examples.

"These polymerized products may include the following: —

"Vinyl chloride CH₂=CHCl, vinyl acetate CH₂=CH (C₂H₃O₂); B-chlor-propylene CH₂=CCl CH₃; styrene CH=CH₂, the above allowing the following general description:

"RCH=CH — R, the polymerized product thereof, having the general formula (RCH=CH — R)_n; R being either hydrogen, hydroxyl, or any organic radical, or in the case of esters, one R may be hydrogen, one R chloride, acetate, bromide, etc., radical."

As we have stated claims 1 to 4 and claim 6 of Robie's patent appropriate a binder comprised of a polymerized vinyl compound. Robie's claims 5 and 10 and Paulson's claims 1 and 2 appropriate a binder comprised of polyvinyl alcohol. Robie's claim 1, which for the purposes of this opinion may be treated as typical of Robie's claims appropriating a polymerized vinyl compound, is as follows: "An abrasive article comprising abrasive grains and a solidified binder comprising a polymerized vinyl compound containing sufficient hydroxyl groups to be self-dispersible in water." Robie's claim 5, which for the purposes of this opinion may be treated as typical of Robie's claim 10 and Paulson's claims 1 and 2 appropriating polyvinyl alcohol, is as follows: "A coated abrasive article comprising a backing material and a layer of abrasive grains attached thereto by a binder comprising polyvinyl alcohol." We must decide first whether Oakes' specification will read upon the claims of Robie and of Paulson sought to be preempted by him. For the sake of clarity we will deal first with claims 1 to 4 and claim 6 of Robie's patent which appropriate a binder comprised of a polymerized vinyl compound. We will deal second with Robie's claims 5 and 10 and Paulson's claims 1 and 2 which appropriate a binder comprised of polyvinyl alcohol.

Oakes' specification will read upon those claims of Robie which appropriate a binder comprised of a polymerized vinyl compound. This will avail Oakes little, however, for it will be observed that, with the exception of the four monomers enumerated in the third paragraph of the quotation from Oakes' application, the description of the products to be employed as binders is couched in the most general terms and would include many, indeed hundreds, of chemical substances, some of which might serve as a binder for abrasive materials for sandpaper but many of which could not be thus employed. Even if we add the qualifications of the specification that the compounds employed as bonds shall be waterproof, thermoplastic and soluble in volatile organic solvents, a great number of unusable compounds remain. Oakes has disclosed a small number of monomers in the third paragraph of the quotation ostensibly as examples, but in reality to serve as a springboard for the claiming of an entire genus of polymerization products. He has not even limited his position to the aliphatic organic compounds characterized by ethylene linkage but seeks to include as well substituted aliphatic organic compounds including those of ethylene linkage. Any claim based upon Oakes' specification unless it limits the bond to one or more of the polymers named in the third paragraph quoted from the specification, vinyl chloride, vinyl acetate, B-chlor-propylene, or styrene, would be invalid under R.S. § 4888. In re Collins, 75 F.2d 1000, 22 C.C.P.A., Patents, 1053; In re Mraz, 36 App. D.C. 435. This becomes clear when one realizes that to determine which compounds, disclosed by Oakes (other than those specifically named), would be useful as a bond or a binder for the abrasive material of sandpaper out of the very large group designated in Oakes' application, would require even a person skilled in the art to embark on many experiments in a comparatively little known field of organic chemistry.

Dealing now with Robie's claims 5 and 10 and Paulson's claims 1 and 2 which appropriate a binder of polyvinyl alcohol we think that clarity will be served if we juxtapose the two questions which we have just answered in respect to claims 1 to 4 inclusive and claim 6 of Robie's patent and assume arguendo that Oakes' specification will read upon these claims. It will be observed that after describing in a very general way in his specification the polymerized products to be employed, Oakes names four monomers beginning with vinyl chloride and ending with styrene. Since Oakes has named them, assuming a favorable state of the prior art, we entertain no doubt but that he would be entitled to claim them. But after designating them, Oakes goes from the particular to the general and, stating "the above allowing the following general description", sets forth by a general polymeric formula (RCH=CH — R)_n, thereby attempting to embrace within his specification substantially all the polymerization products described by him in the first paragraph quoted from his specification. This he may not do for the reasons hereinabove stated since such an embrace transgresses the mandates of R.S. Section 4888. Again the man skilled in the art would be required to resort to experiment unless he employed the polymers of the monomers specifically named by Oakes in the third paragraph quoted from his specification.

We think, however, that we should state our reasons for concluding that a claim appropriating polyvinyl alcohol as the bond cannot possibly be read upon Oakes' specification as well as our reasons for concluding that Oakes did not intend to disclose polyvinyl alcohol in his 1931 application and may not now claim it.

Oakes' application, as we have repeatedly stated, teaches that the polymerization products employed shall be waterproof, thermoplastic and soluble in volatile organic solvents. Polyvinyl alcohol is soluble in water and is not soluble in volatile organic solvents. It is not thermoplastic.

Construing the language of the specification as it should be construed and as in fact the Patent Office construed it, we think that when Oakes employed a general monomeric formula and a general polymeric formula in the fourth paragraph of his specification as quoted he indicated the polymers to be used as binders were to be formed from their respective monomers. The internal evidence of the 1931 application supports this view. Moreover, every one of the four monomers named in the third paragraph, as quoted from the application, can be polymerized. But vinyl alcohol cannot be polymerized by any process presently known. Indeed vinyl alcohol does not exist as a separate compound. If it exists at all, it is but momentarily and in solution. One authority states that "Any attempt to prepare vinyl alcohol invariably leads to the formation of acetaldehyde." Polyvinyl acetate can be hydrolyzed to polyvinyl alcohol. This method ordinarily is employed to create polyvinyl alcohol. The foregoing constitutes convincing evidence that Oakes did not intend to disclose polyvinyl alcohol in his 1931 application.

Referring now to the fourth paragraph quoted from the application and in particular to the general monomeric formula and the general polymeric formula stated by Oakes and to the words which follow them (after the last semicolon in that paragraph), it will be observed that omitting "the case of esters", Oakes stated that the R must be "either hydrogen, hydroxyl, or any organic radical * * *". This phrase considered alone would seem to be susceptible of the meaning, however strained, that one might employ in the formula at the same time hydrogen and hydroxyl, hydrogen or any organic radical, or hydroxyl or any organic radical; but the phrase immediately following, viz., "or, in the case of esters, one R may be hydrogen, one R chloride, acetate, bromide, etc., radical." destroys that illusion. The English language is indeed flexible and at time uncertain, but we think that it is clear that Oakes in employing the phrase last quoted intended that R should represent the same element or compounds in the general formulas except in the case of esters.

Examples may make the matter clearer, employing the four monomers named in the third paragraph as quoted from the application. Vinyl chloride, CH₂=CHCl, is an ester. Consequently, under the language of the fourth last paragraph as quoted from the application one may substitute for the first R of the general formulas hydrogen or chloride or acetate, etc., and for the last R one may substitute any other of the substances named. The formula of polyvinyl chloride is (CH₂=CHCl)_n. It follows that in this instance Oakes' general formulas work well. Vinyl acetate is also an ester and by substituting hydrogen for the first R and the acetate radical, C₂H₃O₂, for the second R the Oakes' general formulas are effective. B-chlor-propylene, CH₂=CCl CH₃, is not an ester and to make the general formulas effective it will be observed that one must substitute hydrogen for the first R and a chloride for the second. Styrene is not an ester and to render the formulas effective one must substitute the benzine ring for the first R and hydrogen for the second.

Vinyl alcohol is an alcohol, not an ester, and if one substitutes hydrogen for the first R and an hydroxyl for the second in the general formulas, one arrives at the symbol for vinyl alcohol, H₂C=CHOH, in the monomeric formula and for polyvinyl alcohol (H₂C=CHOHO)_n in the polymeric formula. If one substitutes hydrogen for both Rs in the monomeric formula one arrives at an unstable compound of no practical use possessing the formula H₂C=CH — H. If one substitutes an hydroxyl for both Rs in the monomeric formula, the symbol for ethylene, CHOH=CHOH, a gas, appears. Ethylene may be polymerized. Polyethylene has the formula (CHOH= CHOH)_n, and it will be observed that if hydroxyl is substituted for both Rs in Oakes' general polymeric formula, the symbol (CHOH=CHOH)_n appears. Polyethylenes are water resistant and are soluble in hot aromatic solvents. Polyethylenes are liquids, however, except when subjected to substantial pressures. It seems clear that Oakes did not intend to make any reference, however indirect, to either ethylene or polyethylene in his application.

From the foregoing it would appear that one would have to make the substitutions for the Rs designated by Oakes for esters in order to make the formulas effective even in the cases of the two non-esters designated by Oakes in the third paragraph quoted from his application. That Oakes intended substitutions for the Rs to be made differently in the cases of esters and non-esters seems indisputable from the words and symbols which he has employed in the fourth paragraph.

The plaintiffs' expert witnesses testified that it was not an unusual practice among organic chemists to employ R as a symbol for one substance at one end of a general formula and for another substance at the other end of the same formula. These experts encountered little difficulty in substituting the symbol for hydrogen in lieu of the first R in both the general monomeric formula and in the general polymeric formula and substituting the symbol for hydroxyl, OH, for the other R in each formula, thus arriving at the symbol for vinyl alcohol, (H₂C=CHOH), and for polyvinyl alcohol, (H₂C-CHOH)_n. In arriving at this solution of Oakes' difficulties they, of necessity, were compelled to avoid the impact of the final clause quoted from the application, "or, in the case of esters, one R may be hydrogen, one R chloride * * *". These experts testified that the symbol R, unlike the symbol X in an algebraic equation, might mean different things in the same formula. They pointed out a number of instances wherein, they insisted, R or X had been thus used in general organic chemical formulas instead of employing primes (R₁, R₂ etc.) or other letters to indicate different chemical elements or compounds. No documentary evidence was produced, however, squarely supporting these assertions. The defendants' witnesses testified to the contrary.

Great emphasis is laid by the plaintiffs on the fact that their counsel wrote to Dean F.C. Whitmore of the School of Chemistry and Physics, Pennsylvania State College, to Dr. Allen P. Colburn, head of the Division of Chemical Engineering of the University of Delaware, and to Dr. Walter M. Lauer, Professor of Chemistry at the University of Minnesota (who were among the plaintiffs' expert witnesses testifying in the court below), stating: "A patent specification which contains the words `vinyl resin' as a part of its title has the following language in its specification: `* * * the general formula (RCH=CHR)_n; R being either hydrogen, hydroxyl, or any organic radical, or in the case of esters, one R may be hydrogen, one R chloride, acetate, bromide, etc., radical.' What compound or compounds do you think the inventor had in mind or intended to cover by his reference to `hydroxyl' in the above quoted passage."

Dr. Colburn replied that as the formula was written "I would suppose" that the inventor had in mind derivatives of vinyl alcohol but "since vinyl alcohol has never been isolated it seems more likely that he had in mind derivatives of polyvinyl alcohol * * *". Dean Whitmore answered: "Ordinarily I do not invest in `pigs in pokes' but I am willing to make a guess that what the inventor * * * had in mind was the polymer of vinyl alcohol. Do I win the $64?" Dr. Lauer made a reply in substance similar to that of Dr. Colburn. Dr. Elizabeth Dyer and Dr. Cecil Cameron Lynch, Professors of Chemistry at the University of Delaware, who also were witnesses for the plaintiffs, to whom Dr. Colburn showed the letter written to him, also concluded that the writer of the specification had polyvinyl alcohol in mind.

It is conceded that at the times when the letters were written and the replies were made these witnesses did not know that a suit was about to be tried or who the parties to it were. The witnesses, including Dr. Dyer and Dr. Lynch, reiterated the substance of their replies on their examination-in-chief in the court below. We think that different positions might have been taken by these witnesses had the whole of Oakes' 1931 specification been put before them in the letters written by counsel for there can be no question that polyvinyl alcohol is soluble in water, is not soluble in volatile organic solvents and is not thermoplastic. All else aside, these facts should have served to eliminate polyvinyl alcohol from consideration.

On reading the material first submitted to these experts and the rest of Oakes' application one might decide as a solution to a problem in chemical detection that Oakes had polyvinyl alcohol in mind. He was claiming a vinyl resin, a polymerized product, and he spoke of "hydroxyl". As the expert witnesses for the plaintiff testified, when the word "hydroxyl" is mentioned to a chemist he usually thinks of an alcohol. But we are not concerned with a problem in chemical detection but with whether Oakes set out in his 1931 application such a "full, clear, concise and exact" description of polyvinyl alcohol as to enable him to base a claim upon it in the light of R.S. Section 4888. One possibly might deduce that Oakes intended that there should be substituted for the Rs of his general formulas, whether in the cases of esters or otherwise, the element hydrogen or any compound named in the fourth paragraph quoted from his application, thus arriving at the polymers of the monomers named in the third paragraph. But to deduce that Oakes meant to disclose polyvinyl alcohol by a rearrangement of the Rs of the general formulas, a rearrangement not in accord with the written word, is a matter of much greater difficulty. To make such a deduction verges on the impossible in view of the conditions set out in the application that the bond or binder should be insoluble in water, soluble in volatile organic solvents and thermoplastic. But the possibility of deducing that Oakes meant to include polyvinyl alcohol in his application is insufficient to bring his disclosures within the purview of R.S. Section 4888, for there can be no doubt that the terms used by Oakes are susceptible of very different interpretations than those put upon them by him and his witnesses. In re Crowell, 84 F.2d 206, 23 C.C.P.A., Patents, 1246, is in point. We believe that if a patent were granted to Oakes embracing the claims which he seeks to preempt "the great ends" intended to be accomplished by R.S. Section 4888 would fail in that the United States could not know what it had granted under its monopoly or what would become public property when the term of the patent had expired. See Gill v. Wells, 22 Wall. 1, 25, 89 U.S. 1, 25, 22 L.Ed. 699.

Applying the rule of Morgan v. Daniels to the testimony in these causes, including that offered by the plaintiffs, we conclude as a matter of law that the evidence was not such as should carry thorough conviction that the Patent Office erred in refusing to allow Oakes to preempt the claims sought by him.

The judgments of the court below are reversed and the causes are remanded with the direction to dismiss the complaints.