This is an appeal from a decree which adjudged the device described in the specification of letters patent No. 621,219, issued to George A. Burwell on March 14, 1899, to be an infringement of the fifth, seventh, eighth, ninth, and tenth claims of letters patent No. 595,696, issued to George J. S. Collins on December 21, 1897. The patented structure is a combination of old elements, and the decree is assailed upon the grounds that there was no invention in the combination, and that Burwell’s device was not an infringement upon it. Both structures are for housings or means for holding the roller elements in roller bearings in proper relations to each other while they are in operation, and for keeping them together in the same relation when they are removed from the roller race.
When Collins conceived his device, roller bearings were in common use. The main problem that was before those engaged in improvements in the art was to construct these bearings so that they would develop the least friction, and many machines had been made and patented in which an-annular roller race between the journal and the journal box or the hub, main supporting rollers • therein, and smaller separating rollers were common and familiar elements. On January 20, 1896, Collins had applied for a patent, which was subsequently issued to him on July 6, 1897, as No. 585,909, for an improvement in roller bearings, which disclosed a bearing consisting of an annular roller race in a journal box, main supporting balls, and separating devices therein. Between each two of the main supporting balls a separating device, consisting of two plates of triangular form, with a cylindrical roller between them at each angle, was so placed that when the wheel was in operation the balls could not touch each other, and there was no sliding contact among any of the parts, save only the trunnions of two of the little cylinders, which were extended so that their en'ds might come in contact with the side walls of the roller race, and thus prevent any sliding contact of the triangular plates therewith. This patent shows another separator, which consists of two cylindrical rollers, engaged at their ends with parallel side members and two small balls placed between the cylinders and their plates and the inner circumferential wall of the journal box. But in the structures shown in this patent the main supporting balls and the separating devices were not held together in any fixed relation to each other, and hence they could not be kept in their proper relation when removed from the roller race. There was no caging or housing to contain them and keep them in place.
The desideratum which Collins sought to reach in the combination of the patent here in suit was a housing or caging which would hold the main supporting balls and the cylinders or rollers which separated *453them in such a relation to each other and the cage that when in operation there would be no sliding friction between any of the walls of the roller race and any of its contents, that the balls and the separating devices should have a rolling, and not a sliding, contact, and that all the contents of the roller race could be removed bodily therefrom without disturbing their fixed relations to each other. He says in his specification:
“My present invention relates more particularly to means for connecting the several separating devices, and also means whereby all the roller elements are rotatably held in a housing, and are thus capable of being bodily removed from the journal simultaneously without disturbing the relative position of said roller elements.”
Figures 4 and 5 accompanying the specification, and claims 8 and 10 of this patent, which are fair samples of the claims in suit, fairly show the device which Collins conceived and perfected to accomplish his purpose. They are here reproduced:
“(8) A housing for the roller elements of a ball-bearing comprising two ring plates, a plurality of ball-separating devices rotatably secured at intervals between the plates at or near their inner margins, said rings being provided with a socket or recess between each separating device, a plurality of roller elements in said sockets and means for securing the plates at a suitable distance apart, whereby said roller elements may partially project beyond the plane of the outer surfaces of said plates, and also project peripherally between them as and for the purpose specified.”
“(10) In a roller bearing, the combination with an axle provided with suitable bearing surfaces, a hub provided with a ball cup and roller elements located in the race formed thereby, ball-separating devices between the roller elements of a housing for confining the roller elements, whereby said housing and roller elements may be removed from the bearing without disturbing the relation of the balls to each other.”
Two great advantages which this combination secured are that it so held the roller elements of a ball-bearing wheel in operation that every bearing upon any of the walls of the roller race and between the balls and the separating devices was a rotating, and not a sliding, bearing, and (2) that it enabled the operator to remove all the contents of the roller race bodily therefrom without any change in their relations to each other. The housing which Collins constructed is so made that the bearing balls project on each side beyond the cage so that the latter cannot come in contact with the journal, the side walls, or the inner circumferential wall of the roller race.
BurwelFs combination, which the defendant uses, accomplishes the *454same results' by a similar caging, preferably made, the patentee says in his specification, out of a single sheet of stamped metal, bent into the same form of housing which Collins constructed out of two annular rings. Below is a copy of figure I of the drawings attached to the Burwell patent, in the upper part of which the stamped plate is shown bent into its final shape, while the lower part discloses a part of the sheet before it is bent.
The brackets or arms, F, are made integrally with the main body of the housing, and are first bent at right angles at the outer periphery of the main body of the cage, and then again at right angles toward its center, so as to embrace the ends of the separating rollers, D, which are mounted on pins, d, the ends of which are journaled in the bent plate. When the plate has been thus bent into final shape, the main supporting balls protrude through the cage in four directions, exactly as in the combination of Collins, so that every contact with the walls of the roller race is a rotating, and not a sliding, contact. The separating devices are held in the same relative position as in the device of Collins, and the housing holds the balls and the separating rollers in such a way that they can all be removed from the race without losing their fixed relation to each other. Thus it will be seen that the same result is attained by practically the same means in the two combinations. There are, it is true, minor differences in the construction of the two housings. Collins’ cage is made of two annular ring plates, while Burwell’s is made of a single plate, so stamped and bent as to take the exact form and have the exact effect of the two plates of Collins. Collins’ plates are held apart, and secured in fixed relation to each other at their outer margins, by spreading blocks or rivets, while Burwell accomplishes the same result by the arms, H, H, of his bracket, *455F, when it is bent in position. The supporting balls in the device of Collins have a greater diameter than the distance between the separating cylinders, so that the latter hold the balls in the cage; while the balls in Burwell’s device have a less diameter than the distance between the cylinders, and are held in their cage by the corners of the overturned brackets in the housing. But these differences relate to mere details of construction. They are the result of the substitution for one mechanical device of its obvious equivalent. Collins’ combination consisted of a cage, supporting balls, and separating devices so constructed that the housing should hold the other elements in such a fixed relation, both when within and when without the roller race, that when they were in operation the supporting balls should have no contact with each other, that their contact with the separating devices should be a rolling, and not a sliding, contact, and that no part of the combination should have a sliding contact with any of the walls of the race. Every essential element of this device is found in the combination used by the defendant. The latter accomplishes the same result by means which are the obvious mechanical substitutes of those described by Collins, and hence, in the eyes of the law, are the same means which he used. The use of a different, but mechanically equivalent, method or material to construct some of the elements of a patented combination will not avoid infringement where the principle or mode of operation is adopted, and the elements, when completed, perform the functions by the same means as, or by mechanically equivalent means to, those described in the patent. Ide v. Trorlicht, Duncker & Renard Carpet Co., 115 Fed. 137, 149, 53 C. C. A. 341; National Hollow Brake-Beam Co. v. Interchangeable Brake-Beam Co., 106 Fed. 693, 711, 45 C. C. A. 544, 562; Watch Co. v. Robbins, 64 Fed. 384, 396, 12 C. C. A. 174, 187; New Departure Bell Co. v. Bevin Bros. Mfg. Co. (C. C.) 64 Fed. 859. The defendant has adopted the principle, the exact mode of operation, shown by the patent to Collins, and his device is a plain mechanical equivalent of the combination described by the plaintiff. It is too slavish a copy in principle and in mechanical means to escape infringement if there was any novelty in Collins’ combination and any invention in conceiving and perfecting it.
This conclusion has not been reached without a consideration of the admitted rule that the subsequent issue of letters patent No. 621,219 to Burwell upon this infringing device raises the legal presumption that the latter does not impinge upon the prior patent to Collins. But in the case at bar this presumption is overcome not only by the facts to which reference has already been made, but by the further fact that the court below, after careful consideration, has decreed infringement. This finding and decree raises a counter presumption of at least equal force to that based upon the subsequent patent, and, in the absence of an obvious error of law or a serious mistake of fact on the part of the chancellor, it must prevail. Brammer v. Schroeder, 106 Fed. 918, 928, 46 C. C. A. 41, 51.
We turn to the question of the validity of the patent to Collins. In support of the contention that there was no patentable novelty in the device which Collins described counsel rely upon the following' *456letters patent: No. 23,704, April 19, 1859, to William S. Pratt; No. 224,971, February 24, 1880, to George W. Thomas; No. 242,006, May 24, 1881, to James H. Langley; No. 422,903, March 11, 1890, to Bruno Beaupre; No. 427,301, May 6, 1890, to Charles G. Deming; No. 449,959, April 7, 1891, to George F. Simonds; No. 449,964, April 7, 1891, to George F. Simonds; No. 449,968, April 7, 1891, to George F. Simonds; No. 550,100, November 10, 1895, to Marie Victor Theodore Dubreuil; No. 559,537, May 5, 1896, to Charles W. Pierce; No. 574,472, January 5, 1897, to Conrad Bender and Charles G. S. Mueller; and No. 585,909, July 6, 1895, to George J. S. Collins. There are two easily distinguishable and well-defined classes of roller bearings—those in which the main supporting rollers are cylinders or wheels, and those in which they are spheres or balls. In the former class these main rollers are ordinarily held in place while in operation by axles or journals protruding from their ends which have bearings in or against rings within the roller race. The patents to Pratt, Thomas, Langley, Beaupre, and Deming concern bearings of this class, and the combinations they disclose fail to anticipate the device of Collins (1) because the problem of finding and constructing a housing for the roller elements of such bearings is radically different from that of inventing a serviceable cage for balls and their separating devices, and (2) because none of the devices shown in these patents accomplish the result which Collins sought to reach, and did attain, by the combination which he made. It is plain that an operative housing for cylinders or wheels used in a roller race would be futile as a housing for spheres or balls, and the patents' under consideration only serve to' demonstrate this fact. There is not one of the devices shown in them which would retain in their proper relation to their separating rollers, balls or spherical rollers, 'and the combination of Collins would be equally useless if an attempt were made to apply it to the main cylindrical rollers. Moreover, there is not one of the devices described in these patents which accomplishes the desideratum which Collins sought to reach, and which he did attain. He conceived and perfected a construction which eliminated all sliding friction between any of the contents of the roller race and its side walls. In every one of the patents in which cylindrical main supporting rollers or wheels are shown, there is a sliding friction between one or more of the contents of the roller race and its side walls when the bearing is in operation. For these reasons the patents to Pratt, Thomas, Beaupre, Langley, and Deming are not entitled to farther consideration, and they will be here dismissed.
The three patents to Simonds describe a construction of ball bearings in which three series of balls inclosed in three cages are indispensable to prevent friction between the walls of the roller race and its contents. One cage incloses the main bearing balls, which protrude through it peripherally, and toward the journal. The two other cages are placed between the bearing balls and the side walls of the race, and they include two series of balls to take the end thrust and to prevent friction between the housing of the main supporting balls and the side walls of the cage. In one of these patents a wire cage for the bearing balls is disclosed. But this housing is not so constructed that *457friction between it and the side walls can be avoided without the interposition of other balls at its ends, and there are no rotatable separating devices between the bearing balls. . It is a far cry from these complicated structures with their three cages and their three series of balls, one in the middle of the roller race to bear the burden of the bearing, and one at each end to prevent friction on the side walls, to the simple and effective construction of Collins; too far a cry to deprive him of the credit of his invention. Not only this, but the structures of Simonds show no separating rollers between the bearing balls,, so that they must constantly rub against each other and against their cage, and his combinations, with all their complications, fail to attain one of the great purposes of Collins’ invention. They fail to hold the contents of the roller race in such relation that there can be no sliding friction between the bearing balls or between them and the devices which separate them.
The patent to Dubreuil reveals a yet more complicated and impractical plan to attain the results which Collins reached so simply. It comprises a loose double ring, placed upon the axle or journal, and inclosed in a box. In this double ring balls and wheels alternate, both of which bear upon the axle. The wheels are journaled on pin's fixed in the inside of the double ring. To the outer sides of this ring other wheels are affixed, which bear against the interior walls of the journal box or roller race for the purpose of keeping the ring in its place and of preventing friction against the sides of the race. Clearly, there is nothing in this double ring, inclosing wheels and balls, with its additional set of wheels affixed to its sides to prevent the friction of the ring against the side walls of the roller race, to anticipate or suggest the simple and practical combination of Collins by which the main bearing balls are held in a position where they support the bearings and prevent all sliding friction upon the side walls at the same time.
The structure described in the patent to Pierce has a housing made of a perforated cylindrical ring and of two perforated annular rings through which the ba^ls protrude and form a rolling contact with the four walls of the roller race. The cage is so constructed that the balls may be removed from the race in this housing without disturbing their relation to each other or to the cage, But the cylindrical ring which forms the periphery of the housing is wider than the diameter of the balls, so that its edges are inserted in grooves in the sides of the roller race, and no roller separating devices between the bearing balls are provided, so that there is constant sliding contact between the housing which is the only separating device in the combination and the rotating balls. The construction of Pierce fails to anticipate the combination of Collins, because it lacks the roller separators and the narrow housing borne upon the bearing balls, which respectively prevent sliding friction between the balls and the separating devices and all contact between the housing and the walls of the roller race.
The patent to Bender and Mueller does not answer the claims of Collins, because it was not issued until Collins had conceived and perfected his invention. This patent was issued on January 5, 1899, *458and the application for it was filed on November 3, 1896. Collins-perfected his invention as early as March, 1896, and he applied for the patent in suit on July 30, 1896. The Bender and Mueller patent is in evidence here to establish the statutory defense that Collins’ combination had been patented prior to his invention or discovery thereof. U. S. Comp. St. 1901, p. 3394, § 4920; Act July 8, 1870 (16 Stat. 208, c. 230) § 61. But it is indispensable to the maintenance of this defense that the defensive patent should have been issued before the patentee under the patent in suit made his invention. As the patent to Bender and Mueller was not issued until after Collins conceived and perfected his patented device, nor until after he had applied for a patent for it, it is neither competent nor material evidence upon the issues in this case. Bates v. Coe, 98 U. S. 31, 33, 34, 25 L. Ed. 68; Du Bois v. Kirk, 158 U. S. 58, 64, 15 Sup. Ct. 729, 39 L. Ed. 895; Walker on Patents, § 55, p. 53.
The first patent to Collins (No. 585,909) is not an anticipation of the claims of his subsequent patent here under consideration (1) because the applications for the two patents were pending in the patent office at the same time, and (2) because the combination described and claimed in the second patent is not described or claimed in the application for the first patent. This application was filed on January 20, 1896, and the patent upon it was issued on July 6, 1897. The application for the second patent was filed on July 30, 1896, and it ripened into a patent on December 31, 1897. Both applications were pending in the patent office from July 30, 1896, until July 6, 1897. The specification and claims of the application for the first patent failed to disclose essential elements of the combination of the second patent, to wit, (1) the housing or cage, and (2) the means by which the separating devices and the main supporting balls are secured and held in the housing. Even where each of several applications which subsequently ripen into patents to the same inventor discloses all the inventions claimed in all the applications, and they are all pending at the same time, no one of the applications or patents can be used to anticipate any of the claims of the others which it does not itself claim and secure. Ide v. Trorlicht, Duncker & Renard Carpet Co., 115 Fed. 137, 145, 53 C. C. A. 341; Walker on Patents, § 180; Westinghouse Electric & Mfg. Co. v. Dayton Fan & Motor Co. (C. C.) 106 Fed. 724, 726; Suffolk Mfg. Co. v. Hayden, 3 Wall. 315, 318, 18 L. Ed. 76; Graham v. McCormick (C. C.) 11 Fed. 859; Graham v. Mfg. Co. (C. C.) 11 Fed. 138, 141. Much less can such an application and a patent upon it which fails to describe essential elements of a combination subsequently patented to the same inventor upon an application pending at the same time have the effect of anticipation.
Much has been said in argument and written in the briefs to show that the tenth claim of the patent in suit is so broad and general in its terms that it covers devices in which the main bearing elements are cjdinders or wheels, and in which the separating devices are not rotatable; and that, when it is read in this way, the combination it describes is anticipated by several of the patents to which reference has been made. But the claims of a patent must be read in the light of the specification which accompanies them. The specification may *459always be referred to to restrict, though not to expand, the claims. And, when this tenth claim is read in the light of these familiar rules, its true construction limits it to a housing for the balls and the_ rotatable separating devices between them described in the specification: and when it is thus read no anticipation can be found in the prior patents.
When Collins conceived and constructed the device secured to him by the patent here in suit, the field for improvement in the art he was studying was narrow. Ball bearings were numerous and well known. Housings for the contents of roller races were common. But no housing had yet been constructed which held the main supporting balls and the rotatable separating devices between them both while they were within and while they were without the roller race in such a fixed relation to each other, to the housing, and to the walls of the roller race that, when the balls were in operation, there would be no sliding friction between any of the walls of the race and any of its contents, and no sliding friction between the balls themselves, or between them and the separating devices between them. Collins made a combination which first accomplished these results. His device may not have been a great improvement, but it was certainly a marked advance in the art, and, in view of the numerous less successful attempts to take the step he took which were rewarded with patents by the government, and in view of the evident inability of those inventors and of the proprietors of mere mechanical skill to reach the point to which Collins attained, the title of inventor ought not to be denied him. His invention is one of those intermediate steps in the gradual progress of a useful art toward perfection which are evidenced by the great majority of patented inventions, and it falls within the familiar rules that a new combination of old elements by which an old result is attained in a more facile, economical, and efficient way may be protected by a patent (National Hollow Brake-Beam Co. v. Interchangeable Brake-Beam Co., 106 Fed. 693, 707, 45 C. C. A. 544, 557; Seymour v. Osborne, 11 Wall. 516, 542, 20 L. Ed. 33; Rees v. Gould, 15 Wall. 187, 189, 21 L. Ed. 39; Thompson v. Bank, 53 Fed. 250, 252, 3 C. C. A. 518, 520; Ide v. Trorlicht, Duncker & Renard Carpet Co., 115 Fed. 137, 143, 53 C. C. A. 341), and that, where the advance in the art is gradual, and several inventors make valuable improvements and form different combinations, which accomplish the desired result with varying degrees of success, each is entitled to his own combination, so long as it differs from those of his competitors, and does not include theirs (National Hollow Brake-Beam Co. v. Interchangeable Brake-Beam Co., 106 Fed. 693, 712, 45 C. C. A. 544, 563; Railway Co. v. Sayles, 97 U. S. 554, 556, 24 L. Ed. 1053; McCormick v. Talcott, 20 How. 402, 405, 15 L. Ed. 930; Stirrat v. Mfg. Co., 61 Fed. 980, 981, 10 C. C. A. 216, 217; Griswold v. Harker, 62 Fed. 389, 391, 10 C. C. A. 435, 438; Adams Electric Co. v. Lindell Ry. Co., 77 Fed. 432, 440, 23 C. C. A. 223, 231; Ide v. Trorlicht, Duncker & Renard Carpet Co., 115 Fed. 137, 143, 53 C. C. A. 341).
The combination of claims 5, 7, 8, 9, and 10 of letters patent No. 595,696 was new, it was useful, it was the product of the intuitive *460genius of an inventor. These claims are infringed by the device constructed in accordance with the description in letters patent No. 621,219, to George A. Burwell, wliich the defendant is using, and the decree below is affirmed.