This appeal is from a judgment for the defendant entered by the District Court in a suit for infringement of claims 1 and 4 of a patent (No. 2,001,555) issued to one Trebler on May 14, 1935, for a device for the irradiation of milk with ultra-violet rays. The plaintiff is a New Jersey corporation engaged in the business of manufacturing scientific equipment, including mercury vapor lamps. Its title to the Trebler patent is admitted. The defendant, a Massachusetts corporation, is alleged to have infringed this patent by irradiating milk in its dairy with a device known as a “Type Y.N. Milk Irradiator” manufactured by the National Carbon Company, Inc. The National Carbon Company, Inc., primarily a manufacturer of lamps and arc carbons, is openly defending this suit.
The District Court did not consider the question of infringement but based its judgment for the defendant solely on the ground that the Trebler patent was invalid in that it disclosed no patentable advance over the prior art. In its opinion that court said, 39 F.Supp. 646, 651: “The evidence as to prior knowledge, use, and art leads to the conclusion that Dr. Trebler took certain obvious steps in the art and whatever changes he made did not amount to invention. He took no forward step in the art upon which invention could be predicated. There was no new mode of operation in the device described in the patent in suit. No new result was produced. The patentee’s construction reflected nothing more than the mechanical skill of -one versed in the prior art.”
The question for us is whether this conclusion of non-invention is to be sustained, and this raises the further question of how that conclusion is to be regarded, — whether it is to be regarded as one of fact or as one of law. If it is one of law, it comes to us unweighted by the finding of the court below; if it is one of fact, we cannot reverse under Rule 52(a) of the Federal Rules of Civil Procedure, 28 U.S. C.A. following section 723c, unless we conclude that the result reached by the District Court is clearly erroneous.
In the recent case of B. F. Sturtevant Co. v. Massachusetts Hair & Felt Co., 122 F.2d 900, 906, we held, citing Thomson Spot Welder Co. v. Ford Motor Co., 265 U.S. 445, 44 S.Ct. 533, 68 L.Ed. 1098, 1 that the question of patentable invention was one of fact, and, according face value to that statement, concluded that we ought not to set aside the District Court’s answer to that question unless clear error appeared. This approach to the problem is clearly sanctioned by some of the author*889ities,2 but in spite of this the fact remains that there are in the reports a great number of cases in which federal appellate courts3 have themselves, on the records before them, decided the question of patentable invention without according any weight whatever to the conclusion thereon reached below. Cases of this sort are too numerous to cite but the case of Potts v. Creager, 155 U.S. 597, 15 S.Ct. 194, 39 L.Ed. 275, and the recent case of Weidhaas v. Loew’s, Inc., 2 Cir., 125 F.2d 544, are typical.
We have been unable to find any analysis by any court justifying this latter technique of decision in view of the repeated general statements that the question of invention is one of fact, but the Supreme Court has sanctioned it in some situations. In the case of Heald v. Rice, 104 U.S. 737, 749, 26 L.Ed. 910, that court said: “In the present case the question of the identity of the invention in the original and reissued patents is to be determined from their face by mere comparison * * * That is, if it appears from the face of the instruments that extrinsic evidence is not needed to explain terms of art, or to apply the descriptions to the subject-matter, so that the court is able from mere comparison to say what is the invention described in each, and to affirm from such mere comparison that the inventions are not the same, but different, then the question of identity is one of pure construction, and not of evidence, and consequently is matter of law for the court, without any auxiliary matter of fact to be passed upon by a jury, if the action be at law.”
Then later, the same court in Singer Mfg. Co. v. Cramer, 192 U.S. 265, 275, 24 S.Ct. 291, 295, 48 L.Ed. 437, a case in which only the question of infringement was involved, said, citing Heald v. Rice, “As in each of the patents in question it is apparent from the face of the instrument that extrinsic evidence is not needed to explain terms of art therein, or to apply the descriptions to the subject-matter, and as we are able from mere comparison to comprehend what are the inventions described in each patent and from such comparison, to determine whether or not the Diehl device is an infringement upon that of Cramer, the question of infringement or no infringement is one of law and susceptible of determination on this writ of er-TOT.
Again in the very recent case of United States v. Esnault-Pelterie, 303 U.S. 26, 30, 58 S.Ct. 412, 82 L.Ed. 625, the above cases and Sanitary Refrigerator Co. v. Winters, 280 U.S. 30, 36, 50 S.Ct. 9, 74 L.Ed. 147, referred to in note 3 above, were cited by the Supreme Court as authority for “the rule that where, with all the evidence before the court, it appears that no substantial dispute of fact is presented, and that the case may be determined by a mere comparison of structures and extrinsic evidence is not needed for purposes of explanation, or evaluation of prior art, or to resolve questions of the application of descriptions to subject-matter, the questions of invention and infringement may be determined as questions of law.” [303 U.S. 26, 58 S.Ct. 414, 82 L.Ed. 625.]
It seems that the conflict between the frequently reiterated categorical statement that the question of invention is one of fact and the almost innumerable instances in which the courts have dealt with it as though it were one of law can only be explained by breaking the question of patentable invention down into its component parts. In the broad question there is first the double question; the “auxiliary matter of fact” referred to in Heald v. Rice, supra, of what the prior art was and what the patentee did to improve upon it, that is, what the appliances of the prior art were, how they operated and what they did, and what the patentee’s alleged invention is, how it operates and what it accomplishes, — and then, when these questions have been answered, there arises the further question of whether what the patentee did is properly to be classified as an *890invention over the prior art and given protection as such or whether it is not. This is what the Supreme Court called the “ultimate” fact in United States v. Esnault-Pelterie, 299 U.S. 201, 205, 57 S.Ct. 159, 81 L.Ed. 123.
The nature of the prior art and the nature of what the patentee did to improve upon it must always he questions of fact. They may be relatively simple ones or they may be exceedingly complicated and involved, but questions of fact they must always remain. The question of the name to be given to what was done by the patentee, whether it is to be called an invention over the prior art or whether it is not, is a question fundamentally of the meaning of the words used in the statute, 35 U.S.C.A. § 31, and as such would seem to be a question of law. Whatever it is, however, the second part of the question is for us to decide for ourselves, as the great number of authorities referred to above recognize, at least tacitly, and as the above quotations from decisions of the Supreme Court indicate. But, from those quotations it also clearly appears that the first part of the question, although one of fact, is not always for the District Court, subject only to review by us for clear error. The authorities quoted above establish that not only when there is no substantial dispute either as to the devices of the prior art or as to what the patentee did, but also when the prior art and what the patentee did is clearly apparent to us from the record and the exhibits, then it is for us to compare the alleged invention with the prior art and from our own comparison to say whether the patentee’s structure is to be classified as a patentable invention over the prior art or whether it is not. See Butex Gas Co. v. Southern Steel Co., 5 Cir., 123 F.2d 954. On the other hand it is equally well established that when conflicting expert testimony must be heard and evaluated in order to understand the nature of the prior art and the nature of what, if anything, the patentee did to improve upon it, then it is for the court below to compare the devices and, unless its comparison is clearly erroneous, all that remains for us is to say, on that comparison, whether or not the thing patented is entitled to rank as an invention. See Trane Co. v. Nash Engineering Co., 1 Cir., 25 F.2d 267, 268, and cases cited in footnote 2 above.
The only cases in which the statement that the question of patentable invention is one of fact is literally applicable are those in which the findings of the District Court on obscure and disputed issues as to the nature of the device patented and as to the nature of the devices of the prior art are, unless clearly erroneous, decisive of the ultimate question of invention. That is to say, the broad statement that the question of patentable invention is one of fact can, strictly speaking, only be literally applied in cases involving devices which all would have to agree resulted from an exercise of the inventive faculty of their patentees if they in fact are all that their patentees contend for them, but otherwise that they are not, and the answer to the question of whether or not the patentees’ contentions with respect to them are correct requires an evaluation of conflicting expert testimony.
From the foregoing it seems apparent that no general statement can be made concerning the classification of the broad question of patentable invention either as one of fact or as one of law. It is partly one and partly the other. Looking at the way the courts have repeatedly treated it and at what the Supreme Court said about it in the cases quoted from earlier in this opinion, it appears that the question is sui generis and that any attempt to classify it generally either as one of fact or as one of law is not only futile but misleading. About all that can be said about it is that it is a conclusion, to be drawn in the first instance by the court below and in the second instance by us, from basic facts either found by us or found by the court below depending upon whether those facts are apparent from the entire record in the case or whether they are not. Thus the second part of the broad question,— what the thing patented is to be called,— is for us to decide “unfettered by findings and rulings below except for the weight of the intrinsic authority of all lower court opinions.” Deputy v. Dupont, 308 U.S. 488, 499, 60 S.Ct. 363, 369, 84 L.Ed. 416. The first part of the question, — what the prior art was and what the thing patented is,— may or may not be for us to decide in the same way depending upon the factual situation presented in each particular case.
Confusing and illogical as it seems, what the courts have done is to find a practical solution for a practical problem. In so far as the answer to the question of patentable invention hinges upon findings with respect to obscure or disputed issues calling for *891expert testimony either as to the nature of the device patented or as to the nature of the devices of the prior art, the view of the District Judge, who saw the experts on the stand, heard them testify and had the opportunity to ask them questions, is entitled to great weight and we should not venture to disagree with his view as to what these things are unless we are convinced that in some respect he has fallen into clear error. Rule 52(a), F.R.C.P. But on the other hand, when we can clearly comprehend from the record and the exhibits the nature both of the thing patented and of the things of the prior art, that is, when the facts are simple and we have as good access to the source of facts as the court below had, we are in as good a position as that court to compare the thing patented with the prior art, as well as to say whether or not the thing patented should be called an invention.
Thus the nature of each case determines the technique appropriate for its decision. In the Sturtevant case, 122 F.2d 900, both the nature of the machines of the prior art and the nature of the machines covered by the patents there in suit, — what each did and how each operated, — presented complicated and hotly disputed issues of fact which, for intelligent solution, called for an evaluation of the testimony of expert witnesses. The master appointed by the court below heard such witnesses and made his findings, which the District Court affirmed and adopted as its own, and we therefore, quite correctly limited ourselves to the consideration of whether or not those findings were clearly erroneous. Having determined that they were not, we then took the facts as found, supplemented by the facts evident from the record, and, guiding ourselves by the statement of Judge Learned Hand in the case of Kirsch Mfg. Co. v. Gould Mersereau Co., 2 Cir., 6 F.2d 793, 794, decided, according to our own lights and to the best of our ability, whether the patentee was entitled to rank as an inventor or only as one having the skill to be expected from one versed in the art. Possibly our statement that the question of invention was one of fact, although based upon ample authority, was too broad, but we have no doubt that the technique of decision which we adopted was correct.
In the case at bar the patented appliance and the appliances of the prior art are relatively simple and comprehensible from the record and the exhibits. In my view, we should therefore proceed directly to consider those appliances for ourselves and, supplementing our findings with respect thereto by the findings, which we may say here are not clearly erroneous, of the court below on such obscure and disputed issues as may be material, decide what those appliances are and do and, on the basis of the facts thus arrived at, decidSe for ourselves the question of the validity of the patent in suit.
On February 13, 1911, three foreign physicists, Henri, Helbronner and Von Recklinghausen, applied to the United States Patent Office for a patent on an “Apparatus for Treating Liquids with Ultra-Violet Rays”, and Patent No. 1,118,006 was issued on their application on November 24, 1914. They taught that liquids could be sterilized by exposure to ultra-violet light and that the best results could be obtained by exposing the liquid to the light in the form of a thin but turbulent film. In the drawings which accompanied their application they showed several types of apparatus which they claimed accomplished their purpose, but which do not here require description. It appears without dispute that milk could not be sterilized by their method without imparting to it an unpleasant taste and odor, but the method proved successful and came into use both in Europe and in .this country for sterilizing public water supply systems as well as for sterilizing some other liquids.
In June, 1914, Von Recklinghausen read a paper, which was afterwards published, at the Annual Convention of the American Institute of Electrical Engineers, entitled “Sterilization of Water by Ultra-Violet Rays of the Mercury-Vapor Quartz Lamp.” In this paper he described and illustrated a number of forms of apparatus which he said were useful for the purpose. Among these appears one shaped in general like a straight-sided funnel, set vertically, inside of which is suspended a mercury-vapor lamp. Closely attached around the outside of the upper periphery of the funnel-shaped part of the apparatus appears an annular trough, the outer rim of which is higher than the inner, so that when the liquid to be sterilized is poured into the trough it will over-flow down the inside surface of the funnel where the light rays will act upon it and thence flow out through the spout at the bottom. It does not appear whether or not this apparatus ever was used commercially.
*892Ten year's later the next major development in the art came about. In 1924 Dr. Harry Steenbock, a bio-chemist working at the University of Wisconsin, announced his discovery that the Vitamin D content of certain foods, including milk, could be stepped up by exposure to ultra-violet light to such an extent that food so treated became valuable in the cure and prevention of rickets, and on June 20 of that year he applied for a patent for an “Anti-rachitic Product and Process”. Patent No. 1,680,-818 was issued on this application on August 14, 1928. Dr. Steenbock assigned his patent to the Wisconsin Alumni Research Foundation, a corporation, which as owner permits no person to sell or any dairy to use any machine for the irradiation of milk until that machine has been approved by dairy experts of the Foundation as capable of producing milk, not only of a sufficient Vitamin D potency, but also of a flavor acceptable to the public. The Foundation makes periodical tests of each licensed milk irradiating device for the purpose of making sure that at all times the licensed machines produce milk of proper potency and acceptable flavor.4
George C. Supplee took the next step in the art. On June 2, 1926, he applied for a patent for a “Method of Irradiating Milk Solids With Ultra-Violet Rays and the Product of such Method,” and on August 11, 1931, Patent No. 1,817,936 was issued to him. In his application and in the drawings which accompanied it, he described several types of apparatus for the treatment of both liquid and powdered milk. One of his devices for treating liquid milk was made by positioning sources of ultraviolet light on either side of a milk cooler of the ordinary type. This apparatus consists of a horizontal trough along the bottom of which are two parallel rows of holes spaced close together. Beneath the trough is a vertical, horizontally corrugated, metal flow-board and beneath this is another trough from which leads a pipe. The milk to be treated is introduced into the upper trough from which it flows through the holes down over each side of the flow-board and into the collecting trough at the bottom. A series of lamps arranged on either side of the flow-board subject the milk to ultra-violet rays as it passes in front of them. It appears that the above machine went into successful commercial use.
In 1933 the Creamery Package Manufacturing Company put a machine on the market for irradiating milk and this machine also enjoyed commercial success. It, like the Von Recklinghausen device described above, consists primarily of a cylindrical flow-surface with a mercury vapor lamp suspended inside. In detail this “C.P.” machine consists of a vertical cylinder above the top of which is suspended a circular trough with holes positioned close to one another all along its bottom. Milk is introduced into the trough from which it flows through the holes down the inside of the cylindrical flow-board where it is subjected to light from the lamp. After treatment the milk is collected by means of another circular trough at the bottom and thence piped into any convenient receptacle.
Both the Creamery Package machine and the Supplee one were licensed by the Wisconsin Alumni Research Foundation, and they appear to have given satisfactory results in commercial operation, at least, they produced milk of 135 rat units potency with acceptable flavor fast enough to be commercially useful.
On September 21, 1933, Supplee, in conjunction with one Merrill. J. Dorcas, applied for a patent for a “Method of and Apparatus for Irradiating Liquids” and Patent No. 2,202,610 was issued to them on May 28, 1940. In this patent they adhered to the teaching that the best method for irradiating milk is by flowing it in a film while subjecting it to ultra-violet rays, but they for the first time taught that turbulence in the film was undesirable. This is clear from the patent and is conceded by counsel for the plaintiff. The apparatus which they designed to attain their object consisted of a smooth, flat, vertical flow-board with a trough at the bottom for collecting the milk after it has been irradiated, and a vented horizontal pipe at the top into which the milk to be irradiated is introduced and from which it flows onto the flow-board through a narrow horizontal slot cut in the side of the *893pipe. The patentees emphasize that “The width of the slot is important.” They say that it should always he “of such width that the liquid to be treated will always form a film across the slit before the liquid can flow through the slit,” and that “Due to the fact that the slot is of such width that a surface film can form across it there is no tendency of the liquid to flow until the liquid has reached a height in the chamber (we called it a pipe in describing the machine) above the top of the slit.” It appears that this device produced a very smooth, uniform, and non-turbulent film on the flow-board, but it does not appear that it ever was used in any way except experimentally.
This was the state of the art, so far as material here, when Dr. Trebler entered the field.
Dr. Trebler entered the employ of a manufacturer of dairy products called the National Dairy Products Company in 1930 as a commercial engineer and “troubleshooter.” In 1931 he was assigned by his employer to investigate existing methods for irradiating milk and immediately found the two commercial methods then on the market which may be described as the Supplee “washboard” machine and the Creamery Package or “C.P.” machine. Upon testing these machines he, as an expert milk taster, thought that the irradiated milk produced by them was of unsatisfactory flavor and so in his laboratory he sought to discover a better form of apparatus. After over two years of unsuccessful experiments with various metals and various sources of ultra-violet light, he decided that off-taste was due to over-irradiation of thin spots in flowing films of milk. To quote the plaintiff’s brief “He reasoned that a thin spot or area on a moving film due to a bubble or other irregularity might be subject to over-irradiation because its velocity would be slower than that of the rest of the film since ‘it would be retarded by the adhesion of the milk to the metal surface, or to the underlying layers’5 and hence might be exposed to the irradiation for an unduly long period.” He decided therefore to devise an apparatus which would produce as smooth and uniform a film as possible, that is, an apparatus which would produce a film without thin spots due to foam, impurities or other irregularities.
The apparatus which he devised and for which the patent in suit was issued, consists of a smooth, either vertical or inclined, flow-surface fed by the overflow over a “knife edge” dam or weir from a trough above it. In the trough he placed a baffle or flange positioned close to the edge of the trough over which the milk flows and extending both above and below the surface of the milk in the trough. He shows the usual trough at the bottom of the flow-board to collect the milk after it has been irradiated. He says that the function of his flange or baffle is to break up or hold back foam or foreign matter in the trough and also to insure an even flow over the dam, and that the dam is of “knife edge” form for the purpose of breaking up any bubble which might escape the baffle.
Whatever mental process Dr. Trebler may have gone through in arriving at his conclusion that a smooth, even flow, free from irregularities due to foam, bubbles or impurities, was essential in order to produce satisfactory irradiated milk, he is not entitled to rank as an inventor therefor because he was not the first to claim it in the patent office. Supplee and Dorcas, in the application for their patent which they filed on September 21, 1933, taught what Trebler found by experimentation and claimed in the application for his patent filed on March IS, 1934. So, if Trebler is to be ranked as an inventor it must be for the form of the apparatus which he devised. ■
But his form of apparatus consisted in only a combination of old parts. He used a smooth, flat flow-board similar to the one shown in the Supplee-Dorcas patent, fed by the overflow from a trough positioned above it as in the device depicted in the Von Recklinghausen publication of 1914. This latter device also shows a “knife edge” dam or weir over which the liquid flows onto the flow-board. The only feature of Trebler’s device not shown in any device of the prior art of irradiating milk is the baffle, but the use of a baffle for the purpose for which Trebler employed it was old and well known. Trebler himself testified that long before he became interested in the art of milk irradiation he knew of the use of baffles for smoothing out the flow of liquids over a weir and for holding back foam, bubbles and impurities. Furthermore the upper edge *894of the slot in the Supplee-Dorcas machine did just what Trebler’s baffle did; it acted as a skimmer and flow-evener because, being a capillary slot, it was below the surface of the milk in the supply pipe or chamber. In our opinion it was no more an act of invention for Trebler to have added the baffle to his apparatus than it was an act of invention for the patentee in the patent considered by the Supreme Court in Altoona Public Theatres, Inc., v. American Tri-Ergon Corp., 294 U.S. 477, 55 S.Ct. 455, 79 L.Ed. 1005, to have added a fly wheel to a pre-existing machine. See, also, Dorr Co., Inc., v. Yabucoa Sugar Co., 1 Cir., 119 F.2d 521, 522, 523.
The plaintiff here lays great stress upon the superiority of the result which Trebler’s device produced over the result produced by the earlier machines, in that his machine, it claims, produced a milk of 400 rat unit potency without offensive taste faster than they did. It also stresses the commercial success of Trebler’s device. The findings of the court below hardly warrant the conclusion of superiority and success which the plaintiff would have us draw, but even if we asume all that the plaintiff asserts in these respects, still no different result is indicated. Commercial success is only a make weight to be thrown into the scale when invention is in doubt (Textile Machine Works v. Louis Hirsch Textile Machines, 302 U.S. 490, 498, 58 S.Ct. 291, 82 L.Ed. 382), and we are not here in doubt. It is too well established to require citation of authorities that a mere improvement in result does not makte paten^tabile that which otherwise would not be.
Considering all the facts we conclude that the District Court was correct in ruling that claims 1 and 4 of Dr. Trebler’s patent disclose no patentable advance over the prior art.
The judgment of the District Court is affirmed with costs to the appellee.
MAGRUDER and MAHONEY, Circuit Judges. Judge WOODBURY has made an interesting analysis of the vexed question whether the issue of invention is one of fact or of law. We prefer to reserve judgment on the correctness of this analysis, because the present case does not require a decision on the point. For the reasons stated by Judge WOODBURY, we agree that the District Court was clearly right in ruling that the claims in suit are invalid for want of invention. Since this is so, it does not matter whether that ruling is to be deemed a finding of fact or a conclusion of law.See, also, the cases cited in United States v. Esnault-Pelterie, 299 U.S. 201, 205, footnote 6, 57 S.Ct. 159, 81 L.Ed. 123.
See Reinharts, Inc., v. Caterpillar Tractor Co., 9 Cir., 85 F.2d 628, 630, and Strong-Scott Mfg. Co. v. Weller, 8 Cir., 112 F.2d 389, 395, and cases cited in each. See, also, Minnesota Mining & Mfg. Co. v. Coe, 1941, App.D.C., 125 F.2d 198, 199.
We do not refer to cases taken by the Supreme Court because of a conflict between decisions below when necessarily that court is called upon to “consider independently the question as to which of the decisions on this question is based upon the sounder reasoning and is correct.” Sanitary Refrigerator Co. v. Winters, 280 U.S. 30, 36, 50 S.Ct. 9, 10, 74 L.Ed. 147. See, also, Thomson Spot Welder Co. v. Ford Motor Co., 265 U.S. 445, 447, 44 S.Ct. 533, 68 L.Ed. 1098; Concrete Appliances Co. v. Gomery, 269 U.S. 177, 180, 46 S.Ct. 42, 70 L.Ed. 222.
Potency is determined by feeding tüe milk to be tested to rats suffering from rickets and then observing the rate and extent of their recovery. For the purpose of comparison standard units of potency called “rat units” have been agreed upon by those who work in the art. . Expert milk tasters determine the question of flavor.
This is Trebler’s testimony on the stand.