Upon the conclusion, with which we are in accord, that the Warren U. S. Patent No. 942,989 anticipates claims 8 and 10 of U. S. Patent No. 1,748,952 to Gookin we can see no invention in Claims 6, 11, 15 and 17 of the latter patent. The great value in the discovery was the employment of an eyelet with a blunt end in setting invisible eyelets. When once a blunt ended eyelet was discovered, so formed as to prevent prongs from passing by the shoulder of the setting machine and beyond the surface of the uppers of shoes, a valuable result was attained, and this result was in every substantial sense achieved by the Warren eyelet. It is true that Gookin’s cylindrical eyelet with a blunt end would enable a setting machine to be used having a somewhat narrower shoulder than would have been possible with a Warren eyelet, but cylindrical eyelets are shown in the U. S. Patent No. 53,234 to Richards and U. S. Patent No. 106,938 to Hoxsie and in the U. S. Patent No. 872,237 to Hughes and such eyelets became available for use with blunt edges after blunt edges were found to be desirable. We cannot see why the slight difference between the Warren and the Gookin eyelet involved invention when an eyelet in either form was equally available to the artisan.

The method patent No. 1,748,951 seems to us also invalid. It in effect describes the use of an old machine in the same way in which that machine has long been used to set invisible eyelets and with a specific form of eyelet that Warren disclosed to the art. All that Gookin’s patents disclose is a new merit in the Warren invention.

We think the decree should be reversed and the bill dismissed.

(dissenting).

The following opinion is — with some verbal changes — that written as the opinion of the court when a majority believed that the invention was patentable. It may now serve as a statement of the facts and issues involved, and of my own ideas. The defendant appeals from a judgment, enjoining it from infringing claims 2, 3, and 6 of Patent No. 1,748,951, and claims 6, .8, 10, 11, 15 and 17 of Patent No. 1,748,952, both issued on the same day to the plaintiff upon applications filed by one, Gookin, and assigned by him to it. The application for the earlier patent was filed on October 9, 1922, and was for a method of setting eyelets in the “upper” of a shoe; the application for the second was filed on August 23, 1923, and was for the eyelet itself. To an understanding of the patents it is *606necessary to say something about the art.. The edge of the “upper” of a shoe is made of three or more layers, or plies, the inside and outside being of leather-with a lining, or “stay”, between. The lacing holes are punched through all three but must be protected against wear, and this, as everyone knows, is done by means of metal eyelets — conical, or cylindrical, tubes with a solid flange at one end, and a circular edge at the other, vertically split, or “scored”, part way down. If the eyelets are put through all three plies and then “overset” — clamped down upon the outer ply — they are unsightly; the patents are concerned with the. art by which they are concealed; it is called “invisible eyelet-ting”. Originally this was done in two steps; the outer layer rfas turned back, the eyelets were put throúgh the inner layer and the lining, and overset upon it fastening the two together. Thereupon the outer layer was put back in place and made fast. In. 1914 one, Muther, patented a machine to do this by a single step (No. 1,112,643). He disclosed an oversetting punch with a shoulder which bulged out the edge of the eyelet after it had passed through the inner plies, but before it had reached the outer ply. As this made the bore — so to say — of the eyelet greater than that of the punched hole, it pushed up the outer ply, through the hole in which the eyelet could not penetrate. As the punch continued to descend, it overset the top of the eyelet upon the inner ply, leaving the outer ply free, which fell back into place, when the punch was withdrawn. A patent of the same general kind issued to one, Dou-’ lett (No. 1,434,356), in 1922, but upon an application filed about eight months after Muther’s. Thus the art stood for eight years — from 1914 to 1922 — until Gookin invented his eyelet. Muther’s machine had always caused trouble; the eyelet’s upper edge was rough and more or less sharp, and even when somewhat dulled, if used with certain kinds of leather, it was apt to catch in the outer ply so as not to clear it when the punch overset it. That resulted in botching the lacing hole and requiring the work to be done over.

Gookin ended this difficulty by the simple expedient of turning in the upper edge of the eyelet. The result was that, as the punch descended and began to split the eyelet at the scorings, it pushed out not edges, but flat segments, which did not engage in the inner surface of the outer layer, so that the punch overset them upon the lining free and clear. Gookin’s . first patent was for this method; his second was for the eyelet itself. All the claims in suit of the first patent include the use of a turned in’ eyelet; all the claims in suit of the second patent, except 15 and 17, are similarly for a turned in eyelet. The defendant concedes that it infringes, but challenges the validity of all the claims. That challenge is in four parts: first, it says that the change in the form of the eyelet did not demand invention; second, that before 1922 eyelets had been turned in at their edge by “tumbling” so as to perform in precisely the same • way as Gookin’s; third, that the exact invention was embodied in an exhibit known as the “Regal Shoe” in 1918; fourth, that it was disclosed in a patent issued to one, Warren, in 1909, and in Gookin’s earlier patent No. 1,327,033. It further argues that claims 15 and 17, which do not comprise turned in tops are invalid for other reasons; and that all the method claims are invalid because of the eyelet claims. The judge overruled all the defences and held all the claims valid.

If the art did not independently arrive at Gookin’s method — a question which for the moment I reserve — it appears to me that his contribution was an invention. Everyone agrees that, although Muther made a great step forward in operation-his machine would frequently miscarry; it was a standing complaint of the industry that it so often did. Gookin’s method, if not infallible, is nearly so, and has altogether displaced other methods when leather is used in which the edge is likely to catch. During the eight years between the issuance of Muther’s patent and Gook-in’s application, no less than ten efforts were made, having at least as pari; of their, object the improvement of “invisible eye-letting” ; these were all the product of experts in the plaintiff’s employ, itself one of the largest and most competent manufacturers of shoe machinery. Gookin himself made two of these before he filed his method patent in 1922. It is difficult for me to conceive a more propitious setting for an invention; all the customary features are present; a need existing over a substantial period, experiments seeking to fill it, eventual success which went far to. displace what had been in the field. To argue, as the defendant does, that in the face of this it was an obvious expedient *607to turn in the edge of the eyelet, is to disregard the facts. The books are full of such instances, and uniformly refuse to judge the degree of ingenuity required by the apparent simplicity of the solution, once achieved.

The next defence is that the defendant had itself practiced the method before October 9, 1922, by “tumbling” the old kind of eyelet. “Tumbling” means putting a great number of eyelets into a drum and rotating it so that they keep rubbing upon each other. It was an old practice in this art, and is not confined to eyelets; primarily it is to polish small articles or .to cover them with lacquer; it was so used by _the defendant. Moreover, it will dull the rough or sharp edges of the eyelet, and perhaps it was used for that purpose too before 1922. If continued long enough, it will do more than dull the edges; it will actually turn them in just as Gookin’s eyelet is turned in; the defendant still turns in the greater part of its eyelets by this process; only about ten per cent of its output is made mechanically. The relevant question here is whether it ever did so before 1922; that is, did so deliberately, and with that purpose in mind, for an unrecognized and accidental prior user would of course not count. Eibel Process Co. v. Minnesota & Ontario Paper Co., 261 U.S. 45, 66, 43 S.Ct. 322, 67 L.Ed. 523; Regar & Sons v. Scott & Williams, 2 Cir., 63 F.2d 229, 231; McKee v. Graton & Knight Co., 4 Cir., 87 F.2d 262, 264. As to that the evidence is in conflict and the judge has found against the defendant. Quite aside from that finding, the testimony would not have answered the severe standard of The Barbed Wire Patent Case, 143 U.S. 275, 12 S.Ct. 443, 36 L.Ed. 154, which the Supreme Court has as yet shown no disposition to relax. Smith v. Hall, 301 U.S. 216, 222, 57 S.Ct. 711, 81 L.Ed. 1049. There was no documentary confirmation of the recollection of the defendant’s witnesses, who were testifying to matters then fifteen years past. Douglas, its chief witness, did indeed say that he had at times, when dealing with customers’ complaints, hammered down eyelets till they were bent over enough not to catch, and had used the result as a standard for “tumbled” eyelets; but he had nothing but his memory to fix the dates. Smyth and Baker, former employees of the defendant, contradicted him, and Baker at least had contemporaneous documentary corroboration of his testimony, in letters showing that satisfactory “new blunted eyelets” had first appeared in 1924. It is improbable that, if ■these had been part of the defendant’s common practice, he would never have heard of them before. As is usually the case upon this issue, the positive testimony of those who have little reason to fabricate is impressive; but the burden is so heavy against uncorroborated testimony that we should have decided against the defendant even without the finding; and the finding lays any possible doubts.

The next, and apparently the most hopefully urged, defence is the “Regal shoe”, Exhibit X. This is supposed to have been made at the “Whitman factory” of the Regal Company in 1918, and is now put forward as an anticipation. The defendant’s case, taken at its optimum, is that it was one of 482 “Sample and Trial” shoes made at that time, of whose disposition and of the details of whose fabrication nothing is known. This is the only survivor; it was found among a number of other shoes of all sorts, apparently kept as a sort of museum of antiquities, such as are not' uncommon in large industrial plants. The eyelets in it appear to have been turned in before the top was overset, and arguendo I shall assume that they were the same as Gookin’s eyelet. Yet, even so, and if I further assume that the “upper” was made and “eyeletted” in 1918, and for that matter that the whole shoe was then completed — though that is directly contrary to the fact — the exhibit would not be an anticipation. So far as appears, it may have been the only one of its kind ever made; and whether it was an experiment, thrown aside as a memento of what not to do; or whether, the eyelets having been separately turned in by hand, it was decided that the expense was prohibitive; or whatever else was the reason for not producing it in quantity; it was, so far as we can tell, a mere experiment in which event it could not be an anticipation. Nor is this possibility met, if all 482 of the “Samples and Trials” had such eyelets, which we do not know. The whole run was — at least from its name — an experiment.

Moreover, that aside, the evidence leaves at large the date of the shoe’s manufacture. The defendant’s theory originally was that it had been kept in the Regal Company’s “Boston office” until that was closed in 1920, when, along with other relics, it was *608brought to the room where it was found in 1937. Nobody asserted any actual acquaintance with it before 1923. Two of the Regal Company officials quite positive-, ly declared that it had not since 1920 left that storage room where it had been thrown with other old shoes. They were, however, conclusively shown to be mistaken ; absolutely indisputable evidence (which I need not set out) shows that the rubber heel now on the shoe was put there after February, 1922. If the “upper” was in fact made and “eyeletted” in 1918, as part of a run of 482 “Samples and Trials”, it was taken out for some wholly unknown reason at least four years later and a rubber heel was put on it. It is extremely hard to conjure up any adequate motive for this; nobody has suggested why such a relic should have been removed, completed and restored to its asylum. The documentary evidence of the production of the “upper” in 1918 is indeed not implausible; the number pressed upon the lining is within the limits of those carried upon a sheet which bears the date “9/30/19” and which purports to record “Spring 1918 Case Number Records”. Although nobody knows who typed this sheet, it may be proper to assume that it was done on September 30, 1919, and even that it was made from accurate records of what had happened eighteen months before. But what are we then to suppose? Did the run consist only of “uppers”? The defendant does not suggest as much; that would clearly not be enough for an anticipation. If not, for what conceivable motive was the shoe reheeled? Besides, the number is by no means conclusive; other shoemakers have the custom of numbering their shoes; and finally, nobody knows that this was a Regal shoe at all. The room where it was found was open to all comers, and it may have been thrown there by anybody at any time. Thus, even if a discarded experiment were an anticipation, this will not serve; courts have consistently refused to accept such evidence in anticipation of patents.

The next defence is the patent to Warren (No. 942,989) issued on December 14, 1909; and Gookin’s earlier patent of 1920 (No. 1,327,033). I need not-pause over the second; the theory is that it showed a turned in edge, but it did not. The “short tapering portion at the leading end” (p. 1, 1. 95) was something quite different and differently intended. Warren’s patent was for two dies to form an eyelet, and the eyelet was of two kinds; one with a solid top, and the other with a perforated. The defendant says that the second form anticipated. Obviously it did not and could not anticipate the method patent, because it appeared five years before Muther’s machine, and indeed nobody before Gookin thought of using such an eyelet for “invisible eyeletting”. The same is not however true as to the eyelet patent. Claim 10 of that patent does not require that the- eyelet shall have any cylindrical “barrel” to grasp the shoulder of the oversetting tool, nor that it shall be “scored”, and the reference seems to us final. A patent for a “manufacture” cannot rest upon a new use of an old article for the statute does not permit patents for new uses unless they be processes (§ 31, Title 18, U.S.Code). Traitel Marble Co. v. Hungerford B. & C. Co., 2 Cir., 18 F.2d 66, 68; H. C. White Co. v. Morton E. Converse & Son Co., 2 Cir., 20 F.2d 311, 313; Regar & Sons, Inc. v. Scott & Williams, Inc., supra, 63 F.2d 231. Claims 6, 11, 15 and 17 have, however, as an element that the “barrel” of the eyelet must be in part cylindrical, and this, as I shall try to show, was itself an invention, so that they .do not have to depend upon the turned in edge. Claim 8 did not however contain this, though it did prescribe that the “barrel” must be “scored”, which was not in itself a patentable distinction. The question arises whether, since this difference prevents Warren’s eyelet from being literally an anticipation, it preserves the claim. If the difference had been necessary to adapt the eyelet to the new use, that would be true, for very slight physical changes are enough, if dictated by a genuine inventive thought. Topliff v. Topliff, 145 U.S. 156, 163, 164, 12 S.Ct. 825, 36 L.Ed. 658; C. & A. Potts & Co. v. Creager, 155 U.S. 597, 608, 15 S.Ct. 194, 39 L.Ed. 275; Rockwood v. General Fire Extinguisher Co., 2 Cir., 8 F.2d 682, 686; Gorden Form Lathe Co. v. Walcott Machine Co., 6 Cir., 32 F.2d 55, 58. But Gookin expressly said that “scoring” was only “preferable”, i. e., it was not essential (p. 3, 1. 15); hence it follows that Warren anticipated both claims 8 and 10.

Claims 6, 11, 15 and 17 are independently valid. I will assume that the defendant made eyelets, marked with the numbers 36 and 107, before October, 1922, though even that depends, upon oral testimony; but Exhibits D and YY which it puts forward *609as examples of these, were not adequately proved ro be the same as Nos. 36 and 107. Some of the witnesses did indeed so testify, but that is not enough, as I have said. The evidence which I have rejected to prove that the defendant ever turned in the top edge of its eyelet, is stronger than that which proves that it had ever made eyelets with a cylindrical “barrel”. All we know is that certain witnesses produced samples which they thought were like those made in 1915 or 1916. That is far from enough. Nor do any of the prior patents cited anticipate these claims. It is true that Richards (No. 53,234) spoke of a “cylindrical portion” of his eyelet, but the invention was for a sharp edge, equal to a chamfered edge. Nobody reading the disclosure would understand that it made the least difference whether the eyelet was cylindrical or not; one could practice the invention equally well with a conical eyelet like Warren’s. The chance and inessential elements of the form in which an earlier inventor happens to embody his invention is not an anticipation; they do not enrich the art pro tanto, because it is told in substance that they are inessential. The same applies to Iloxsie (No. 106,938) and Hughes (No. 872,237). If the disclosures had necessarily involved the production of Gookin’s eyelet, ignorance of its use would not matter; but that was not the case.

It is another question whether claims 6, 11, 15 and 17 were valid as inventions even if they were not anticipated. The patent explains the value of the cylindrical part of the eyelet (p. 2, 11. 106-130; p. 3, 11. 1-8). The theory is that “a smaller clenching shoulder can accordingly be used” ; and this is because a cylindrical eyelet will continue to be overset, as it moves upward upon the shoulder, by the same circle of contact with the shoulder. But a conical eyelet, as it moves upward, will find its circle of contact--i. e., its oversetting circle — at progressively larger and larger circles on the shoulder. Thus a shoulder of less flare will be necessary with a cylindrical eyelet. It is not necessary to decide how important this feature is; the operation appears to be as the patent discloses it, and if the defendant thinks it useless, it need not make its eyelets in that way. Infringement is ordinarily good enough evidence of utility.

Lastly as to the method claims. The defendant says that even if the eyelet claims are good, these are invalid because they are no more than for the operation of the old Shaw machine (No. 1,205,277) with the new eyelet, and that this could not be patented, not only because it is old as a process, but because the operation of a machine cannot be patented anyway. That argument would, however, be equally an answer to the method claims, if the eyelet patent had never issued; and those claims would certainly have been valid, if they stood alone. Shaw did indeed disclose one method of “invisible eyeletting”, but Gookin disclosed another; the fact that part of the process in each case was to use Shaw’s machine with an eyelet, did not make Gookin’s process invalid, for his eyelet was also part of the process, a variant and a variant whose production required invention. Nor is it true that the process was not more than the inevitable operation of Shaw’s machine, just for the reason that the machine may use different eyelets. Moreover, the method claims are absolutely necessary to protect Gookin’s invention when the eyelet used is like Warren’s, for I have held the eyelet claims bad except for the cylindrical “barrel”.

In my opinion the judgment should therefore be affirmed as to claims 6, 11, 15 and 17, and reversed as to claims 8 and 10, of Patent No. 1,748,952; it should also be affirmed as to claims 2, 3 and 6 of Patent No. 1.748,951.