Appellants, Ellen L. Mochel, an applicant for a patent, and her assignee, Corning Glass Works, brought two actions under 35 U.S.C. Section 145 seeking a judgment directing the Commissioner of Patents to issue a patent for an invention entitled “Method of Glass Treatment and Product.” The cases were consolidated for trial, and a judgment of dismissal was entered in each case,1 the district court sustaining the finding of the Patent Office that the claimed subject matter was obvious in view of prior art, and therefore unpatentable under 35 U.S.C. Section 103.2 The district court stated, and the parties agree, that the “only issue here is obviousness.”
Scope of Review
Preliminary to a consideration of the merits of the appeal, we recognize that the factual findings of the Patent Office can be overturned by the district court only on the court’s reaching a “thorough *412conviction” that the Patent Office is in error. A finding that the Patent Office’s rejection of an application is “consistent with the evidence” requires af-firmance of the rejection.3 The presumption of validity given to Patent Office findings, however, is weakened “where an issue has not been the subject of a Patent Office finding, or an assumption underlying the patent office findings is demonstrably inaccurate in a material degree.” 4 With respect to the scope of review in this court, the findings of fact of the district court “shall not be set aside unless clearly erroneous.” Rule 52(a), F.R.Civ.P.
This court has recognized also that the question of obviousness under 85 U. S.C. Section 103 is a mixed finding of law and fact. In Stieg v. Commissioner of Patents, 122 U.S.App.D.C. 361, 353 F.2d 899, 901 (1965) the court said:
“There is no question but that rulings of law as such made by the Patent Office are fully reviewable, free from any ‘clearly erroneous’ rule, by the District Court and by this court. It is just as clear, however, that when the Patent Office applies a proper principle of law to its factual findings, the resulting judgment is entitled to ‘great weight’ and that ‘in the absence of new evidence carrying “thorough conviction” that had not been considered by the Patent Office,’ the action of the Patent Office should be affirmed.”
The distinction between questions of fact and law was stated more precisely by Judge Tamm in Higley v. Brenner, 128 U.S.App.D.C. 290, 387 F.2d 855, 857 (1967):
“We note at the outset that a presumption of correctness accompanies the decisions of the Patent Office. California Research Corp. v. Ladd, 123 U.S.App.D.C. 60, 356 F.2d 813 (1966); Fletcher v. Watson, 92 U.S.App.D.C. 205, 204 F.2d 68 (1953), and substantially the same rule applies to the findings of fact of the District Court. What the prior art is and what the claimed invention is are questions of fact. However, whether the standard of obviousness applied to those facts is correct, is a question of law. Graham v. John Deere Co., 383 U.S. 1, 17, 86 S.Ct. 684, 15 L.Ed.2d 545 (1966).” 5
In Graham v. John Deere Co. the Court pointed out that, “While the ultimate question of patent validity is one of law, * * * the § 103 condition, which is but one of three conditions,6 *413each of which must be satisfied, lends itself to several basic factual inquiries.” The Court continued:
“Under § 103, the scope and content of the prior art are to be determined; differences between the prior art and the claims at issue are to be ascertained ; and the level of ordinary skill in the pertinent art resolved. Against this background, the obviousness or nonobviousness of the subject matter is determined. Such secondary considerations as commercial success, long felt but unsolved needs, failure of others, etc., might be utilized to give light to the circumstances surrounding the origin of the subject matter sought to be patented. As indicia of obviousness or nonobviousness, these inquiries may have relevancy.” 383 U.S. at 17-18, 86 S.Ct. 684 at 694.
Findings of Fact
In a memorandum opinion the district court described the invention “in essence” as “predicated upon the asserted fact that certain glass composites containing more than an ordinary amount of aluminum oxide, generally called high alumina glasses, when chemically strengthened by a process called ion exchange, retain their greatly increased strength even after extensive surface abrasion” and that “ordinary glass composites strengthened by the same exchange treatment and exposed to surface abrasion lose virtually all the strength so imparted.”
The court was “convinced that the claimed result thus outlined is the result of a mere routinization and experiment and it would have been obvious to a person of ordinary skill in the art when the prior art references were examined in relation thereto.” Specifically the court further found that “ionization is old and the use in the circumstances of the particular glass exposed to such treatment adds nothing new, and no evidence was adduced at the trial which would warrant the conclusion and thorough conviction that the Patent Office was wrong in its evaluation and conclusion.”
The court also adopted and ratified the findings of fact and conclusions of law in defendant’s post trial memorandum.
Appellants’ Contentions
Appellants contend that the district court erred in holding that “no evidence was adduced * * * which would warrant the conclusion and thorough conviction that the Patent Office was wrong in its evaluation and conclusion”, when (1) admittedly the strength results achieved by Dr. Moehel’s discovery are vastly superior to the prior art- results and cannot be achieved as the process was used in the prior art; (2) both the Patent Office and district court ignored undisputed evidence that those skilled in the art could not explain at the time of the invention, or many years later, why the high alumina glasses of Dr. Mochel are permanently strengthened and conventional glass compositions of the prior art are not; (3) the principal prior art author testified that he was “surprised” when he learned of Dr. Mo-chel’s discovery; and (4) the discovery has made the difference between success and failure in the commercial use of the prior art process.
Methods of Strengthening Glass
There are two basic methods of strengthening glass, (1) a mechanical process called “thermal tempering” 7 and *414(2) a chemical process of ion exchange. In turn there are two basic types of ion exchange, the first involving an exchange of ions of a smaller diameter from an exterior substance for ions of a larger diameter within the glass. Both mechanical tempering and the first type of ion exchange require very high temperatures and result in strengthening of the glass only upon cooling.
The second type of ion exchange involves the exchange of a larger ion from an exterior substance for a smaller ion within the glass.8 In this process, the strengthening occurs at the time of transfer at a relatively low temperature and is described as ion “stuffing” or “crowding” because the larger ions are “stuffed” or “crowded” into the space formerly occupied by the smaller ions.
Appellants’ Claimed Invention
The Mochel application for patent involves the second method of low temperature ion exchange and the resulting product. It is claimed that Dr. Mochel discovered that a low temperature ion exchange process produced a product retaining, after substantial surface abrasion, a large proportion of its increased strength when the glass composition contained at least 5%, and preferably 10 to 25% of alumina (aluminum oxide, AI2O3). By contrast, as the district court found, “ordinary glass composition strengthened by the same exchange treatment and exposed to surface abrasion loses virtually all the strength so imparted.” Traditional strengthening has been of soda-lime glass, ordinary window glass, with a low alumina content (1-2%).
Appellants seek a patent on two types of claims: method claims 1, 4, 5, 10 and 11, and product claims 6, 9, 12 and 13. Method claims 1, 4, 10 and 11 require about 5 to 25% alumina, and claim 5 calls for 10 to 25%. Product claims 6, 12 and 13 define alkali aluminosilicate glass articles containing 5 to 25% alumina, while product claim 9 is limited to a composition containing 10 to 25% alumina.
The district court and appellee accept appellants’ statement that claim 1 is the broadest method claim and claim 6 the broadest product claim. These claims, as set out in the findings read:
“1. In a method for strengthening an alkali metal containing glass in which the alkali metal ions in a surface of the glass are replaced by larger monovalent metal ions selected from the group consisting of the alkali metals, copper, and silver by bringing the surface of the glass into contact with a source of said larger mon-ovalent metal ions while retaining the glass at an elevated temperature but below the strain temperature of the glass until the surface of the glass to a depth of at least 5 microns is placed in compression, the improvement which comprises using as the initial glass to be strengthened by said treatment an alkali aluminosilicate consisting essentially, by weight on the oxide basis, of about 5-25% alkali metal oxide, about 5-25% AI2O3, and Si02, the total of alkali metal oxide, A1203, and Si02 constituting at least 80% by weight of the glass composition.
“6. An alkali aluminosilicate glass article having substantially increased strength after being subjected to surface abrasion said article containing ions of an alkali metal and of a monovalent mental of larger size selected *415from the group consisting of the alkali metals, copper, and silver and having a surface compressive stress layer of a depth of at least 5 microns, the concentration of the ions of a monovalent metal of larger size being greater in the surface layer than in the interior portion of the article and the concentration of the smaller alkali metal ions being greater in the interior portion of said article than in the surface layer thereof, said differences in concentration creating the compressive stress in the surface layer, the interior portion of said glass article having a composition consisting essentially, by weight on the oxide basis, of about 5-25% alkali metal oxide, 5-25% A1203, and Si02, the total of alkali metal oxide, AI2O3, and Si02 constituting at least 80% by weight of the glass composition.”
The Prior Art
As set forth in appellee’s brief, the Patent Office examiner, the Board of Appeals, and the district court all relied upon the following publication and United States patents to show the obviousness of the plaintiffs’ claimed improvement:
(1) Kistler, J. Am. Ceram. Soc., Feb., 1962, pp. 59-68
(2) Weber, 3,218,220 Nov. 16, 1965
(3) Hood et al., 2,779,136 Jan. 29, 1957
The Board of Appeals sustained the examiner’s rejection of all claims under 35 U.S.C. Section 103 on the ground of unpatentability of claims 1, 4 to 6 and 9 over Kistler or Weber, in view of Hood et al. and Kurz; 9 and the unpatentability of claims 10 to 13 over Weber in view of Hood, et al. and Kurz. In addition the Board of Appeals relied upon The Glass Engineering Handbook10 and The Manufacture of Glass11 as further references supporting the examiners’ rejection. Neither of these publications nor the Kurz patent are relied upon or discussed in either the court’s findings (incorporating defendant’s post trial memorandum) or appellee’s brief.
a. Weber Patent
The Weber patent involved the use of an electrical current to speed ion exchange at temperatures below the strain point of glass exchanging potassium (larger) ions with sodium (smaller) ions in the glass, placing the surface layer of the glass in compression. “The invention particularly relates to the tempering of soda glass, or more specifically soda-lime-silica glass.” App. at 209. But since the primary emphasis of the patent is upon the electrolysis method, the composition of the glass involved was not at issue nor is it mentioned except to identify the glass used. Weber dismisses ion exchange without the use of electricity as “impractically slow” and devotes no further attention to it. It was held in Kistler v. Weber, 412 F.2d 280, 283 (C.C.P.A.1969), however, that *416Weber did teach ion diffusion without electrical current.
b. Hood-Stookey Patent
The Hood-Stookey patent involved the chemical strengthening of glass by ion exchange, through the transfer of smaller (e.g. lithium) ions for the larger alkali metal ions of the glass (e.g. sodium) . The transfer in the Hood patent occurs above the strain point of glass but below the softening point. The strengthening that occurs is the result of rapid cooling which places the surface layer of the glass in a “compressional stress together with an internal tensional counter stress” since the inner portion of the glass has a higher thermal expansion coefficient than the surface layer of the glass. App. at 213. The lower thermal expansion coefficient of the surface layer is the result of the ion exchange which places the smaller lithium ions in the spaces occupied by the larger metal ions of the glass. Since the tranfer occurs above the strain point the glass is allowed to flow enough to compensate for the difference in size of the ions, but since it is below the softening point, the flow is not sufficient to distort the surface of the glass. Thus the strengthening that occurs in the glass is not the result of “crowding” or the formation of a compression layer, but of cooling and a differential in the thermal expansion coefficients of the internal and the external layers of glass.
The Hood-Stookey process must be carried out at temperature above the strain point of glass- — the opposite of the Weber, Kistler and Mochel processes, which must be carried out below the strain point. Altogether, it is clear that Hood-Stookey describes an entirely different ion exchange process. This was recognized by the Patent Office in allowing the Weber patent and also by the Court of Customs and Patent Appeals in the recent case of Application of Garfinkel, 437 F.2d 1000 (C.C.P.A.1971) where in referring to the Hood-Stookey patent the court said that, “It is disclosed that high abraded tensile strength is obtained ; however, this is due to a mechanism unrelated to the claimed subject matter (the low temperature ion exchanges) .” 437 F.2d at 1002.12
The Hood patent did discuss the use of alumina content glass. The use of alumina, however, was to combine chemically with silica and lithium to form “crystals of betaspodumene” to create an even greater differential between surface and interior thermal coefficients of expansion. The formation of these crystals also causes the glass to become translucent or opaque.
The Hood patent includes eleven examples of experiments, with results indicating the strength of the glass as a result of the treatment. The findings adopted and ratified by the trial court discuss two of the examples, 9 and 10, but a third, example 5, is also relevant.
Example 5
Example 9
Example 10
% alumina In glass..... 20 10 20
Temp. oF Ion exchange treatment____ 825°C. 750°C. 730°C.
M.O.R. of strengthened glass.......... 23,000 psi 58,000 psi 67,000 psi
(Modulus of Rupture, or breaking point)
Appellant's Br. at 11; App. at 216.
*417The findings, relying on examples 9 and 10, conclude that a high alumina content that results in high strength is clearly disclosed. But example 5 discloses the opposite, and it is apparent that the strength variations in the Hood-Stookey patent occur from different elements and a different process.
C. Kistler Publication
The final and most important article of prior art relied upon is “Stresses in Glass Produced by Nonuniform Exchange of Monovalent Ions” by S. S. Kistler of the College of Engineering, University of Utah, published in the Journal of the American Ceramic Society, Vol. 45, No. 2. (Feb.1962), 59-68, App. at 223. Kistler’s article discloses the basic low temperature ion exchange process. Kistler’s work was an attempt to measure the stress of compression that resulted from the ion exchange. He used a variety of glasses identified both by manufacturer’s number and by composition and reported the results of experiments with three glasses: Corning 0211, with a content of 1.93 percent AI2O3; and Leitz slide covers, with an AI0O3 content of 1.0 percent. The third glass was identified as Corning 0088 with an alumina content of 5.12 percent. By affidavit and at trial it was established that the actual content of Corning 0088 was approximately 2 percent. Kistler’s article discloses an increase in the strengths of the glasses as a result of the low temperature ion exchange process. Kistler also separated samples for the first two glasses he used, one of which was handled very carefully to reduce any possible abrasions, while the other was handled “carelessly” in order to subject it to the normal abrasions that would occur in laboratory work. His findings indicated that the strength obtained by the ion exchange process was rapidly lost as a result of the abrasion. His publication did not disclose or indicate that he separated the samples for Corning 0088. Neither did it disclose or indicate that Kistler expected any different result from the Corning 0088, whether the alumina content was at 5.12 percent or at 2.0 percent. The principal thrust of Kistler’s article was to demonstrate the means of the ion exchange, and to demonstrate the means to measure the stress created by the resulting compression in the surface of the glass.
The Mochel Application and Prior Art The Mochel claims admittedly involve the use of the Kistler ion exchange process at low temperatures to create a compression layer on the surface of the glass.13 What is claimed as patentable is the use of high alumina glass with a resulting strength that is not only present at the time of the transfer but which is retained to a significant degree after purposeful abrasion of the glass to simulate actual wear and tear (abraded strength). The Kistler article published a composition of Corning Glass 0088 at 5.12 percent alumina which would place it within the range of the patent applied for by Mochel. As noted supra, however, the testimony at the trial and affidavit submitted to the Patent Office indicate that the 5.12% figure was an error and that the 0088 glass could not have contained more than-2.09% of alumina. This is not seriously challenged by appellee.14
In any event, it is clear from Kistler’s article that he did not anticipate nor demonstrate any substantial strength increase in Corning 0088 as compared to the other two types of glass he used. This can be accounted for in two ways: first, the shapes of the glasses used were different, and since the geometry *418of glass has a bearing on its mechanical strength, the three glasses cannot be satisfactorily compared; second, Kistler was not interested in achieving or demonstrating the results of experiments that involved abrasions or in findings for abraded strength and did not separate the Corning 0088 sample. Thus, the difference between the Kistler process and the Mochel application is limited to the use of high alumina content glass by Mochel, with its attendant product.
Level of Ordinary Skill in the Art
It is difficult to determine exactly what level the skill in the art had reached when Mochel made application for her patent. Relying upon what is identified as prior art, however, it is apparent that there had been a good deal of experimentation involving the use of the ion exchange process to strengthen glass. It seems clear also that the principal means of experimentation involved variations of process and external source materials, with lesser reliance upon variations in the composition of the glass subjected to the processes. Mochel herself testified that her earlier experimentation had involved the use of gases, primarily sulphur and sulphate, to achieve increased durability, and that it was only upon hearing about Kistler’s ion exchange process that she was able to make her discovery. TR. at 48, App. at 56, 57. Her earlier work had, however, involved the use of high alumina glasses, and upon disclosure of the basic ion exchange process she subjected her already-in-use high alumina glass to the new process. An applicant’s own earlier work is not prior art as to him. Illinois Tool Works, Inc. v. Solo Cup Co., Inc., 461 F.2d 265 (7th Cir. 1972).
Secondary Considerations
a. Commercial Success, Long Felt Need, Failure of Others
Appellants introduced uncontested testimony of the commercial success of Dr. Mochel’s invention. The glass was first made commercially by Corning Glass Works in 1963 and up to the time of trial in 1971 had resulted in sales of over $14,000,000. It is used for laboratory equipment, strengthened automobile windshields, windshield for high speed planes, and windows for many space capsules. All lunar modules which had landed on the moon before the time of the trial had windows made from this glass.15
Appellee does not question the commercial success of the Mochel invention but argues that the district court correctly found that commercial success is inadequate to overcome the prior art showing of obviousness. We agree that commercial success in itself is not sufficient to overcome a showing of obviousness ; but, as the Supreme Court held in Graham v. John Deere Co., supra, “commercial success, long felt but unsolved needs, failure of others, etc.” may “be utilized to give light to the circumstances surrounding the origin of the subject matter” and may be relevant “as indicia of obviousness or nonobviousness.” 383 U.S. at 17-18, 86 S.Ct. 684, at 694.
b. Unexplainable Results
It is undisputed that glass experts are unable to explain the increase in abraded strength as a result of using high alumina content glass. Dr. Kistler testified that he was unable to give a scientific explanation; that he had “no theory to offer” and was “surprised that it existed.”16 While this testimony was *419not before the Board of Appeals, the Board did have an article by Dr. Fred M. Ernsberger of Pittsburg Plate Glass Company, published in The Glass Industry in October, 1966 in which it is said: “The magic ingredient in these preferred compositions is alumina. * * * entirely satisfactory explanation has been offered for the effect of alumina * * App. at 159, 161.
Like the other secondary considerations to be taken into account, the fact that neither appellant nor appellee can explain the unexpected superiority of the Mochel invention is not sufficient in itself to sustain a finding of unobviousness. However, as was held in the Seventh Circuit, “the fact that defendants cannot explain how it happens [the operation of the claimed invention in this case was a small personal towel that both washes and dries] even after it has been done, is strong proof of the unobviousness of such a combination.” Canaan Products, Inc. v. Edward Don & Company, 388 F.2d 540, 545 (7 Cir. 1968). See also Kolene Corporation v. Motor City Metal Treating, Inc., 307 F.Supp. 1251, 1261, (E.D.Mich.1969), aff’d. 440 F.2d 77 (6 Cir. 1971) cert. denied, 404 U.S. 886, 92 S.Ct. 203, 30 L.Ed.2d 169.
It is significant too that the first examiner’s rejection of the Mochel application was in part based upon a misconception of how the process worked. He described the result of the ion exchange process as being dependent upon the “holding ability” “of the glass to hold the large substituting alkali ion firmly in its exchanged, ‘squeezed’ place in the glass surface so that the strength reducing surface flaws or cracks in the glass surface are kept compressed and ineffective.” In doing so he equated increases in mechanical strength, chemical resistance, antiweathering, and abrasion resistance, and relied primarily upon the British Kurz patent as teaching that “adding AI2O3 to the glass, gives the glass this holding ability.” (Examiner’s Answer, Appeal No. 738-33, filed June 6, 1966, App. at 122, 124). That reliance upon the British patent by Kurz has been largely abandoned. The testimony offered by appellant as to the unexplainable superiority of the strength of the high alumina glass is not challenged by the Patent Office.
Obviousness
With all of the foregoing consideration in mind we reach the critical question of obviousness as to both the method and product claims. Was it obvious to a person skilled in the art that a high alumina content glass subjected to a low temperature ion exchange process would yield grealy superior abraded strength?
a. Method Claims
We agree with appellee that the process set forth in the preambles of method claims 1, 4, 5, 10 and 11 is disclosed in the Kistler publication and Weber patent. Their admission of prior art in their application and subsequent briefs may not be denied by their submission of an affidavit under Rule 1.131 (Patent Office Practice Rules 35 U.S.C. A. Appendix I, 37 C.F.R. § 1.131) purporting to place their own invention prior to that of Kistler. Application of LoPresti, 333 F.2d 932, 934, 52 C.C.P.A. 755 (1964). Since the Kistler process is precisely that followed by Mochel, the mere substitution of starting material cannot make the process patentable. Application of Kanter, 399 F.2d 249, 251, 55 C.C.P.A. 1395 (1968); Application of Fahrni, 210 F.2d 302, 303, 41 C.C.P.A. 768 (1954).
b. Product Claims
The findings adopted and ratified by the district court include this *420statement: “Product claims 6, 9, 12, and 13 clearly define no more than the characteristics one would expect to find in a high alumina glass after it had been subjected to the ion exchange treatment of Kistler and Weber.” App. at 28. Does the evidence support a finding that this result was obvious from the prior art?
In resolving this question, we must keep in mind that, “The test of obviousness * * * must be applied as of the time of the invention and not retrospectively as of the time of the suit. ‘Many things may seem obvious after they have been made, and for this reason courts should guard against slipping into use of hindsight. We must be careful to “view the prior art without reading into that art the teachings of appellant’s invention.” Application of Sporck 301 F.2d 686, 689, 49 C.C.P.A. 1039 (1962). * * * Further, even if all the elements used are old, a new result, an unexpected result, a far more efficient result, or a more economical result will satisfy the requirements of patenta-bility.” Higley v. Brenner, supra, 387 F.2d 855, at 858.
Admittedly there were prior uses of high alumina content glass with varying quantities of alumina, but there is no proof of any indication that the use of' from 5 to 25 percent alumina in the Kis-tler ion exchange process would yield anything more than the standard increases in strength, and certainly nothing to suggest that the strength obtained would be retained upon abrasion.17
The strengths obtained by Moehel in her use of the Kistler process are in fact clearly superior — so superior as to constitute a difference in kind and not merely degree. Higley v. Brenner, supra, 387 F.2d at 859. The patent application contains test results comparing strengths obtained that demonstrate the difference. For example, glasses containing 18 percent AI2O3 yielded strength after abrasion measured at 52 X 103 psi M.O.R. (Modulus of Rupture, or breaking point) to 77 X 103 psi M.O. R., or 52,000 to 77,000 psi M.O.R. Glasses containing the usual 1 to 2 percent AI2O3 yielded, after the same abrasion test, M.O.R.’s of 12,000 to 37,000 psi.
The Patent Office did not challenge the superiority of the product, but concluded that the result was obvious. In doing so both the examiner and Board of Appeals relied in part upon Kistler’s disclosure of 5.12% AI2O3 glass used in his ion exchange experimentation to show that Moehel had performed the obvious and obtained the expected. Even after the filing of an affidavit stating that this was an error, the Board concluded that it was reasonable to assume that the 5.12% figure “was probably correct.” The proof at the trial that the 0088 glass could not have contained more than 2.09% alumina is undisputed. As noted supra, the district court did not rely on the 5.12% figure, and appellee argues that the reliance on the erroneous figure is not necessary to support the rejection by the Patent Office.18 *421It does, however, constitute an assumption which is “demonstrably inaccurate in a material degree.” California Research Corp. v. Ladd, supra, 356 F.2d 813 at 819. To the extent that the Examiner and Board of Appeals relied upon Kistler’s reported use of 5.12 percent alumina glass, their conclusions were inconsistent with the evidence.
Appellee now relies primarily upon the Hood-Stookey patent. Recognizing the “the Hood, et al., process involves the replacement of a larger alkali ion with the smaller lithium ion, and is a “high temperature” process, appellee argues that “the record clearly demonstrates that these differences, vis-a-vis the Kistler or Weber processes have no significance.” It is argued further that “the fact that in the Hood, et al., process, alumina reacts with silica and the lithium ion is likewise “entirely without significance.” Appellee’s Br. at 9.
Clearly the Hood-Stookey patent does not state that the high strength is due solely to the higher proportion of alumina. Rather it is stated that the higher strengths are due to lithium reacting with the A1203 and Si02 in the glass “to form finely divided crystals” (App. at 214). In the Mochel process the higher strengths are attributable solely to the surprising and unexplained effect of alumina in the glass. Unlike the Hood patent which reacts directly with the A1203 to precipitate crystals, there is no direct action on the A1203 in the Mochel process, since the product glass remains completely transparent, and the A1203 in the glass remains unchanged. The effect of the direct action between the lithium and the A1203 in the Hood patent, because of the crystallization, is to render the glass translucent or opaque, clearly not an acceptable result in such usages as lenses. The resulting products are totally different. There is no evidence that the Hood-Stookey patent has ever been used commercially.
Finally, it appears clear from the expert testimony offered by appellants at the trial that the result achieved by Dr. Mochel was not obvious to one skilled in the art. In particular, it was not obvious to either Kistler or Weber. Appel-lee argues that this is “clearly irrelevant” and that the “question is what the Kistler publication and Weber and Hood, et al. patents taken collectively would suggest to those of ordinary skill in the art.” It is clear, however, that Weber knew about the Hood-Stookey process at the time he filed his application. While it does not appear from the evidence when Kistler first learned of the Hood-Stookey process, he did know about it when he testified at the trial. He was still “surprised” at the result and unable to explain it.
While the Patent Office Board of Appeals did not have the benefit of Kis-tler’s testimony at the trial,19 it did have before it the article written by Dr. Ernsberger in 1966, in which it was recognized that even at that time no satisfactory explanation had been offered for the effect of alumina. The Ernsberger article discussed the Hood-Stookey patent. No mention is made in the Board’s opinion of the Ernsberger article; nor did it discuss any of the proof indicating that Mochel’s discovery was unexplained even after it was made.
*422There can be no doubt that Dr. Kis-tler, Dr. Ernsberger, and Dr. Garfinkel were all of at least ordinary skill in the glass art. Yet neither the Hood-Stookey patent nor any other prior art aided any of them in explaining why the higher alumina glasses used by Dr. Mochel were permanently strengthened without forming crystals.
In essence the facts are undisputed.20 We accept the findings of the Patent Office and district court with respect to “what the prior art is,” i. e.,the Weber and Hood-Stookey patents and Kistler article and “what the claimed invention is.” We conclude, however, that (1) the Patent Office was clearly in error in accepting the published A1203 content of Kistler’s 0088 Corning Glass sample; (2) the Hood-Stookey patent taught an unrelated process which produced a different product by a different method; and (3) under undisputed testimony of the expert witnesses the Mochel product resulting from the use of high alumina glass was not obvious to a person of ordinary skill in the art. Whether the standard of obviousness applied was correct is a question of law. Higley v. Brenner, supra. On the basis of our conclusions and judged by the standards laid down by the Supreme Court in Graham v. John Deere Co., supra, we hold that the district court was in error in concluding that the Mochel product “would have been obvious to a person of ordinary skill in the art when the prior art references were examined in relation thereto.”
The judgments of the district court are affirmed as to the method claims (1, 4, 5, 10 and 11) and reversed as to the product claims (6, 9, 12 and 13) of appellants’ application Serial No. 754,311.
. Appellants seek only one patent, i. e., on method claims 1, 4, 5, 10 and 11 and product claims 6, 9, 12 and 13 of their application Serial No. 754,311. This application is the subject of Civil Action No. 1488-70 and was a continuation in part of application Serial No. 555,617, which is the subject of Civil Action 2599-67. All claims had been rejected by the examiner and the Patent Office Board of Appeals. '
. 35 U.S.C. § 103 reads:
Ҥ 103. Conditions for patentability; non-obvious subject matter
“A patent may not be obtained though the invention is not identically disclosed or described as set forth in section 102 of this title, if the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains. Pat-entability shall not be negatived by the manner in which the invention was made. July 19, 1952, c. 950, § 1, 66 Stat. 798.”
. Morgan v. Daniels, 153 U.S. 120, 14 S.Ct. 772, 38 L.Ed. 657 (1894); National Distillers and Chemical Corporation v. Brenner, 128 U.S.App.D.C. 386, 389 F.2d 927, 928 (1967); Zenith Radio Corporation v. Ladd, 114 U.S.App.D.C. 54, 310 F.2d 859, 862 (1962); Hays v. Brenner, 123 U.S.App.D.C. 96, 357 F.2d 287, 289 (1966); Abbott v. Coe, 71 App.D.C. 195, 109 F.2d 449, 451 (1940); Esso Standard Oil Company v. Sun Oil Company, 97 U.S.App.D.C. 154, 229 F.2d 37, 39, cert. denied, 351 U.S. 937, 76 S.Ct. 1027, 100 L.Ed. 1491 (1956).
. California Research Corporation v. Ladd, 123 U.S.App.D.C. 60, 356 F.2d 813, 819 (1966); Stradar v. Watson, 100 U.S.App.D.C. 289, 244 F.2d 737, 739 (1957); Scripto, Inc. v. Ferber Corporation, 207 F.2d 308 (3 Cir. 1959); Dole Refrigerating Company v. Amerio Contact Plate Freezers, Inc., 265 F.2d 627, 629 (3 Cir. 1959); Murray Company of Texas, Inc. v. Continental Gin Company, 264 F.2d 65, 69 (5 Cir. 1959).
. Flour City Architectural Met. v. Alpana Alum. Prod., Inc., 454 F.2d 98, 106 (8 Cir. 1972) contains a comprehensive analysis of the cases in all circuits decided subsequent to Graham v. John Deere Co., supra, and concludes that “the ultimate question of obviousness vel non [is] a matter of law, not fact.”
. The other conditions, novelty and utility, were set out in the Patent Act of 1793. The test of nonobviousness was the result of judicial interpretation of the constitutional requirement for patent-ability announced in Hotchkiss v. Greenwood, 52 U.S. 248, (11 How. 248, 266) 13 L.Ed. 683 (1851) and codified in the Patent Act of 1952, presently 35 U.S.C. Section 103. While Hotchkiss emphasized “inventiveness”, Congress and the Patent Act of 1952 described the test to be “nonobviousness”, specifically in*413tended to abolish the test of “the flash of creative genius” announced in Cuno Corp. v. Automatic Devices Corp., 314 U.S. 84 (1941). Graham v. John Deere Co., supra, 383 U.S. at 14, 15, 86 S.Ct. 684.
. Thermal tempering is accomplished by rapidly cooling the surface of a glass article while it is very hot. The rapid cooling causes the outer surface to “freeze” or become rigid, while the interior portions are still hot. As the interior portions cool to room temperature they contract, putting the outer layers of the article under considerable compressive stress. Mechanical strengthening of glass is not involved in this patent application but is a part of the prior art.
. This procedure of strengthening glass by ion exchange involves immersing glass in a molten bath of the salt formed from soft metals at a temperature at which the ion exchange will be facilitated but which is not so high as to cause the glass itself to flow. The point at which a glass will flow is its “strain point”, a temperature at which glass has a given viscosity. Ion exchange that occurs below the strain point of glass with a batli containing larger ions than those in the glass results in the “crowding” or “stuffing” of the larger ions into the glass, but without a flowing of the surface, and yields a strengthened glass article. The metals principally used in this process are potassium, sodium and lithium.
. Kurz, British Patent No. 835,820, May 25, 1960, involves the coating of glass with a “sort of skin which contains resistance improving glass components.” Joint Appendix (App.) at 219. The glass article (moisture free and as free as possible of alkalis) is subjected to acidic gases such as sulphur dioxide with oxygen at temperatures between 1,000 to 450 degrees centigrade. The patent is for “a method of treating articles of alkali' — lime — silicate glass” to “improve the chemical resistance of the glass,” and it identifies the addition of AI2O3 only as a factor that would make a “melt” (if a melting hath instead of the gas treatment were to be used) “very difficult, or even impossible.” Kurz’s patent is a method of enabling ordinary glass to retain its original strength with a chemically changed composition of the surface forming a layer resistant to air and moisture.
. The Glass Engineering Handbook, E. B. Shand, (1958) at 4 identifies a glass containing 20.5% A12C>3 as a commercial glass and describes its mechanical hardness in terms of resistance to impact abrasion and indentation hardness. App. at 236-239.
. In The Manufacture of Glass by L. M. Angus-Butterworth, at 38, this sentence appears: “Alumina improves the durability of the glass. It gives increased resistance to acids and is therefore used in laboratory glassware.” The text goes on briefly to explain other uses of alumina not related to increase of strength or durability. App. at 233-235.
. At the trial of this case Dr. Hannon Garfinkel, an expert in glass research employed by Corning Glass, testified: “In my opinion, there is no parallel at all (between the Moeliel discovery and the Hood-Stookey patent) and both techniques depend upon completely different phenomena.” App. at 103, Transcript (TR.) at 126.
. AVhile the AVeber patent is aimed at speeding the ion exchange through the use of electrical current, it is also essentially the same process disclosed by Kistler and used by Mochel.
. Appellee’s brief states that the disputed disclosure was not relied upon by the district court and is not needed to support the rejection made in the Patent Office. Appellee’s Br. at 17.
. The specific advantages of this glass were described as a high strength to weight ratio, the capacity to strengthen a very complex shape, the capacity to allow a trade off of strength and break patterns (as in a windshield which can be strengthened to resist breakage but when it does break will break into fine particles), and abrasion resistance (TR. at 87-89).
. TR. at 85, App. at 78. The district court found, and appellee argues, that the testimony of Dr. Kistler “is merely cumulative and clearly no more per*419suasive * * * than the evidence before the Board of Appeals.” Appellee does not challenge the evidence that there is no scientific explanation of the superiority in strength of the high alumina glass, but argues that this is irrelevant to the issue of obviousness.
. Essentially the same results were obtained from all three glasses used by Kistler. Dr. Garfinkel testified witli respect to the Corning 0088 Glass: “Well, the results were essentially the same as those obtained with the two preceding glasses, that is, the window glass and the 0211 glass and that is that in every case the effect of the ion exchange strengthening was lost by the tumble abrasion.” (App. at 92) Moreover, had the superiority actually achieved by Mo-chel through an alumina content of at least five percent been obvious or expected, it is difficult to explain why Kistler did not expect greater superiority from his own 5.12 percent sample, and, not obtaining it, question the result.
. The Board of Appeals suggested that appellants’ “dismissal of Kistler’s analysis as being inaccurate seems to be unrealistic or even unreasonable”, but concluded that even if it were accidental, it would have “disclosed the treatment of glass of the analysis given ‘ * * * to others unmotivated by applicant’s interest in going behind what was said to de*421termine what was probably meant.’ ” (App. at 198) The Board noted also that the Kistler publication, “interpreted as we believe it must be interpreted, illustrates this invention as applied to a glass of 5.12 percent A1203 at the most.” (App. at 199)
. The Board did suggest that Kistler’s failure to appreciate the advantage of the five percent glass in retaining its un-abraded strength was a “dubious basis for unobviousness of a product (even neglecting the process considerations) to argue that a prior art worker did not appreciate all the advantages of his product but even that state of fact is not shown.” (App. at 198-199) This fact was of course clearly shown at the trial in the district court.
. The Patent Office presented no witnesses at the trial in district court, but relied upon its cross-examination of appellants’ witnesses.