This is a bill for infringment of four patents for improvements in circular knitting machines: Mayo patent, No. 363,528, dated May 24, 1887; Mayo patent, No. 461,357, dated October 13, 1891; Johns patent, No. 600,788, dated March 15, 1898; Ames patent, No. 600,761, dated March 15, 1898. The complainant has not pressed the Mayo 1887 patent on this appeal, thereby limiting the alleged infringement to the three remaining patents. The two Mayo patents relate to improvements in the machine proper, and the Johns and Ames patents to improvements in an attachment to the machine for introducing a supplementary thread. The defendants’ machine is constructed under two patents issued to Rowe — No. 570,059, dated October 27, 1896, and No. 581,887, dated May 4, 1897.
The Circuit Court dismissed the bill on the ground of noninfringement, except as .to one claim, which was held void for want of invention.
A circular knitting.machine of the Mayo type comprises a needle cylinder, an encircling rotating cam cylinder, and latched needles, with butts or lateral projections on their -shanks. The needle- cylin*528der has a series of longitudinal parallel grooves on its outer surface, and the cam cylinder is provided with suitable cams or inclines on its inner surface. The needles lie in the grooves of the needle cylinder, with their butts projecting outwardly. As the cam cylinder revolves, the butts come in contact with and slide up and down the inclines or cams, and so cause the requisite up- and-down movement of the needles in the operation of knitting.
In knitting the tubular portions of a stocking, all the needles are in operation, and knit one course at each revolution of the cam cylinder. In knitting the heel or toe of a stocking, only a portion of the needles are in operation, and the cam cylinder is reciprocated, or moved first in one direction and then in the opposite direction. The heel and toe are in the form of a pouch or pocket, and they are knit by what is known as the narrowing and widening operation. In this operation about one half of the needles are first raised to a higher level, or into an idle position, so that their butts will not be operated upon by the stitch cams, and the other half of the needles remain in an active position, where their butts may be operated upon by the stitch cams. In the operation of narrowing, which then takes place, one of the active needles is shifted to a higher level, or to the idle series, at each reciprocation of the cam cylinder, by raising or elevating the butt of the needle. When the narrowing is completed, the widening takes place by exactly the reverse operation; one of the idle needles is shifted to a lower level, or to the active series, at each reciprocation of the cam cylinder, by lowering or depressing the butt of the needle. This method of shifting one individual needle as the cam cylinder moves in one direction, and another individual needle as the cam cylinder moves in the opposite direction, is the generic mode of operation in knitting the heel or toe of a stocking. There is a modification of this method, which is used only in widening. This modification consists in first throwing one needle out of operation, and then throwing two needles into operation on each reciprocation of the cam cylinder. In other words, it consists in continuing the narrowing operation of carrying one needle out of operation through the widening, and in adding two needles to the active series during the widening, in place of one. The net result, of course, is the same, namely, that one needle is added to the active series at each reciprocation of the cam cylinder. This modification, from the point of view of the result accomplished, is also described as throwing two needles into operation as the cam cylinder moves in one direction, and throwing one of these needles out of operation as the cam cylinder moves in the opposite direction. To distinguish these two modes of operation, the first is called the “one-and-one method,” and the second the “two-and-one method.”
The two-and-one method is useful for the following reasons: In the one-and-one method the last needle to knit in the widening operation, as the cam cylinder moves in one direction, is the first needle to knit on the next course on the return reciprocation of the cam cylinder in the opposite direction. This needle therefore draws two loops, with the result that a series of small holes is left *529along the line or seam where the narrowed and widened portions are joined in the heel or toe pocket. This disadvantage is overcome by the two-and-one method, by which the last needle to knit in a given course is thrown out of operation on the return reciprocation of the cam cylinder, and therefore it is not the first to knit on the next course. To express the effect in another form, in knitting each course one needle knits beyond the needle which is the first to knit in the next returning course.
It may be observed in this connection that both these methods produce commercial products, and that machines embodying each method have gone into extensive commercial use. Upon a comparison of the relative position occupied by these two methods in the art, the most that can be fairly claimed for the two-and-one method is that it is a preferred mode of operation, because it produces a somewhat better product.
In the earlier machines of the Mayo type, all the rotary knitting of the body of the stocking was automatic, while the reciprocating knitting of the heel and toe was done by hand. When the rotary knitting was partially or wholly completed, or when the heel or toe were to be knit, the machine was stopped, and about one-half the needles shifted to an idle position. In the narrowing and widening operation which followed, the individual needles were raised by a pick or hook in the hand of the operator, and depressed by the finger of the operator. As the result of this use of a pick to shift the needle, the term “picker” has come to designate the means by which the needles are shifted in an automatic machine, and the machine itself is called the “picker machine.” The pickers which raise the needles are known as elevating or lifting pickers, or lifters, and the pickers which depress the needles, as the depressing pickers, or droppers. Two lifting pickers are used in automatic reciprocating knitting, one operating when the cam cylinder moves in one direction, and the other when the cam cylinder moves in the opposite direction; and the same, of course, is true of the depressing pickers. The necessity of having two lifting pickers and two depressing pickers is because there must be two stitch-cam arrangements on the cylinder in reciprocating knitting; one operating when the cylinder is moved in one direction, and the other operating when the cylinder is moved in the opposite direction.
While the old picker machine, with the needles raised and depressed by hand during reciprocating knitting, went into extensive commercial use, the efforts of inventors were naturally directed to making the whole machine automatic, by devising automatic pickers.
The first automatic arrangement of this character disclosed in the record is the Hollen patent, of October 10, 1876. This machine was limited to automatic lifting pickers. Then follows the Haddan patent, of 1881, which disclosed both automatic lifting and depressing pickers. Haddan was followed by the Branson patent, of December 29, 1885, which also covered both kinds of automatic pickers. The next in point of time is the Mayo automatic picker *530machine, covered by his 1887 patent in suit, but not pressed on this appeal. There are also found in the prior art two complete automatic circular knitting machines of a different type — the Shaw machine, described in the Shaw patent of June 8, 1880, and the Aiken machine, described in the Aiken patent of August 25, 1885. In the Shaw patent there is also disclosed, for the first time, the two-and-one method.
The Mayo 1891 patent, in suit, is for specific improvements in the automatic picker machine of the Mayo 1887 patent. In order to understand the Mayo 1891 patent, and the position it occupies in the art, we must first understand the Mayo 1887 patent.
The problem Mayo sought to solve in his 1887 patent was the construction of automatic pickers in a circular knitting machine which was already automatic in respect to rotary knitting. The difficulty of the problem resided in the fact that, where any obstacle like a picker was interposed in the path of the needle butts in a moving cam cylinder, it was necessary to get that obstacle out of the way of the following needle butts. Further, since there are two lifting and two depressing pickers, the problem involved not only getting the active picker out of the way of the following needle butts, but also providing that the needle butts should not bring up against the back part of the inactive picker.
An invention is the embodiment of a conception. In order, therefore, properly to comprehend the Mayo invention, or the way in which Mayo solved this problem, we must first start with his conception of an automatic picker. The Mayo conception of an automatic picker was a picker with a single movement, as distinct from a picker with a compound movement — a picker with a sliding movement, as embodied in his 1887 patent, or with a swinging movement, as embodied in his 1891 patent. Mayo never reached the conception of an automatic picker having a compound movement — both sliding and swinging.
On the other hand, the Rowe conception of an automatic picker, which is found in the defendants’ machine, was a picker with a compound movement — both sliding and swinging. In a comparison of the Mayo machine with the defendants’ machine, it is necessary always to keep in mind this fundamental difference in conception between the Mayo picker and the Rowe picker, since it is this difference which accounts in large measure for the difference in the organization of the two machines.
Again, it is apparent that a Mayo picker, with only a single' movement, either sliding or swinging, which is adapted to operate in the path of the needle butts, must have some positive locking! device for holding the picker in an inoperative position except-when it is required to perform its work. In the absence of such a device, it is obvious that the pickers would remain in the path of the needle butts, and that the needle butts would not ride over them, because the movement of the pickers is restricted to a movement, in one direction.
It is also apparent that a Rowe picker, with its compound sliding and swinging.mpvem.ents, is adapted to operate in. the path of-dhe, *531needle butts without locking devices for holding the pickers in an inoperative position. Such devices are not necessary, because the pickers may remain in the path of the needle butts, and the needle butts will ride over them by reason of their compound movement.
It is thus seen that the locking devices of the Mayo 1887 patent are the essential and vital means employed in the embodiment of his conception of an operative automatic picker, and, further, that the employment of such means is not essential in the embodiment of the Rowe conception of an operative automatic picker.
We come now to the consideration of the three claims of the Mayo 1891 patent which are in issue.
Claim 4 relates to the two-and-one method; claim 6, to a pivoted picker, as distinct from the sliding picker of the 1887 patent; and claim 11, to an improved detail of construction.
Claim 4 is as follows:
“In a knitting-machine, the needle-cylinder and the needles carried thereby, combined with a cam-cylinder provided with stitch-cams and needle elevating and depressing cams [pickers] having fingers or hooks the fingers of the elevating-cams engaging but a single needle and the fingers of the depressing-cams being of such length as to engage two needles, latches to normally hold the said elevating and depressing cams in inoperative position, and devices through which the cam-cylinder may be reciprocated, whereby in the operation of the machine two needles may be returned into action at each reciprocation of the cam-cylinder and one of said needles carried out of action at the next reciprocation thereof in the opposite direction, and so on, to effect the widening of the fabric, substantially as and for the purpose specified.”
This claim is for the combination for performing the widening operation by the two-and-one-method. Mayo had before him his 1887 patent, and he wished to incorporate this modification of the one-and-one method into the machine of that patent. He also, in the eye of the law, had before him the Shaw patent, wherein the two-and-one method in a different type of machine is described as the “preferred” form. The two-and-one method involved the throwing of one needle out of operation and two needles into operation on each reciprocation of the cam cylinder during the widening. To accomplish this, Mayo continued the narrowing operation by keeping the narrowing tripping cams in operation, during the -widening, and by lengthening the fingers of the two depressing pickers so they would depress two needles instead of one. It may be that this change- would not have suggested itself to the skilled mechanic, with knowledge of the two-and-one method as described in the Shaw patent, and with the Mayo 1887 patent before him. In other words, this way of solving the problem may have involved invention. That question, however, is not before us. Claim 4 is a claim for a combination of certain elements, and the question of infringement turns upon whether the defendants’ machine contains this combination.
Shaw accomplished the two-an-one method in one way, and Mayo in another way. Of course, Mayo, in view of Shaw, could not claim broadly the two-and-one method, nor does claim 4 purport to be for a method. It is for a combination to accomplish exactly what Mayo undertook to accomplish — the introduction into *532the latched picker machine of his 1887 patent of the two-and-one method.
Except as to the changes already mentioned, all the elements of claim 4 appear in the Mayo 1887 patent, and are therefore a part of the prior art. The question of the infringement of this claim is narrowed to the inquiry whether the defendants’ machine contains the element, “latches to normally hold the said elevating and depressing cams [pickers] in inoperative position.” We have already referred to the circumstance that latches or locking devices of some kind were essential in the embodiment of the Mayo conception of a single-movement picker, and that latches or locking devices were not necessary in the embodiment of the Rowe conception of a compound-movement picker. It is necessary now to be more specific. Do we find in the defendants’ machine any locking devices which correspond to the Mayo spring-actuated arms pivoted to the outside of the cam cylinder, and provided with suitable shoulders against which the pickers rest and are positively held in a locked position, or any devices which perform substantially the same function? Let us turn first to the defendants’ depressing pickers. These pickers slide in carriers which are pivoted to the outside of the cam cylinder. The result is that they have a compound movement, a sliding diagonal movement, and a vertical movement, owing to the carrier being pivoted. The effect of having this compound movement is that the needle butts, following the butt which is operated upon, ride over the picker, and there is consequently no necessity for any latch or locking device to hold it in an inoperative position. This also applies to the inactive picker, against the back of which the needle butts may strike. The needle butts will also ride over these pickers, although, in order to avoid the wearing of the butts, there is in defendants’ machine mechanism for holding down these pickers. It thus appears that latches or locking devices for the depressing pickers, in the sense of the Mayo invention, are absent from the defendants’ organization.
AYe come next to the lifting pickers. The two lifting pickers in defendants’ machine are carried by a T-shaped carrier pivoted on the outside of the cam cylinder, and their hooked ends project through diagonal slots in the cylinder. The pickers move up and down in these slots. They have consequently a sliding movement, and also a swinging movement upon the pivoted T piece. Both pickers move with the carrier, and both swing about the same pivot. It follows that, as one picker operates, and is thus carried into an inoperative position, the other picker is carried into an operative position. Neither picker is at any time latched or locked in an inoperative position, and one is always in an operative position. The spring attached to the carrier is in no sense a locking device, within the meaning of the Mayo patents. Further, the defendants’ lifting pickers are located above the stitch cams, instead of at the sides, as in the Mayo. In consequence of this, the needle butts are never brought up against the back of the inactive lifting picker, but are carried under and around it.
It is clear that the construction and arrangement in defendants’ *533machine avoids the necessity of any locking devices for the pickers. That machine does not embody “latches to normally hold the said elevating and depressing cams [pickers] in inoperative position,” or any equivalent thereof, and consequently does not infringe claim 4 of the Mayo patent. Starting with a fundamentally different conception of an automatic picker, we find in defendants’ machine a different organization of pickers and cams, and a different mode of operation.
The next claim of the Mayo 1891 patent in issue is claim 6:
“In a knitting-machine, a needle-cylinder and needles carried thereby, and cam-cylinder provided with a slot, combined with a needle-operating cam [picker] pivoted outside of the said cam-cylinder and extended through the said slot to operate upon the said needles and be operated thereby, substantially as described.”
This claim covers an improved form of the Mayo conception of a picker with a single movement. Mayo first embodied that conception in his 1887 patent, in a sliding picker on the inside of the cam cjdinder, while here it is embodied in a swinging or pivoted picker on the outside of the cam cylinder, and extending through a slot in the cylinder. The same fundamental distinction, however, still exists between this picker and the compound-movement picker of the defendants’ machine.
Respecting the infringement of this claim, it may first be observed that the Mayo 1887 patent was in issue in the court below. It was there contended that the depressing pickers in defendants’ machine infringed claim 2 of that patent. This was upon the theory, of course, that these pickers were sliding, and not pivoted. Upon this issue the Circuit Court held there was no infringement, because the Mayo picker moved in a fixed plate, while the picker in defendants’ machine had a swinging movement on a pivoted carrier. There is no contention, therefore, that the depressing pickers infringe claim 6 of the 1891 patent, but the charge of infringement is limited to the lifting pickers.
The lifting pickers in defendants’ machine manifestly come within the broad language of claim 6. The first question, therefore, to be determined on reading this claim, is whether Mayo is entitled to the broad invention covering all pickers which are pivoted outside the cam cylinder and extending through slots in the cylinder, or whether the claim must be restricted to substantially his way of pivoting.
A brief examination of the organization of a circular knitting machine will throw light on this question. In such a machine, as already described, we have a needle cylinder in which the shanks of the needles lie in grooves, with their butts projecting outwardly, an encircling revolving cam cylinder in which the needle butts ride up and down on the cams in the operation of knitting, and notched cams or pins, called “pickers,” which are interposed in the path of the needle butts to shift the individual needles in making the heel and toe of a stocking.
Now, in the first place, it is evident that these pickers must, either have a sliding movement or a swinging movement in order *534to do their work and get out of the way of the following needle butts. In other words, they must either slide in guideways or swing on pivots. Neither of these general conceptions would be patentable, for they are manifest to any mechanic. The inventive faculty would be exercised, and only exercised, in finding out a practical operative way in which to embody either of these conceptions, for therein lies the difficulty, and consequently the solution of the. problem..
Again, it is manifest that the picker must be mounted somewhere —that is, either on the inside of the cam cylinder, or on the outside of the cam cylinder, or above the cam cylinder; and it is clear that no one can patent the broad idea of mounting a picker in either of these positions.
The conception that it would be better to mount a pivoted picker on the outside instead of the inside of the cylinder would not involve the exercise of invention, first, because it is a most common thing in the mechanic arts to mount a mechanical part on the outside'dnstead of the inside of the Cylinder of a machine; and, secondly, because it would occur to any one familiar with the organization of cam cylinders and needle cylinders in these machines that it would be better to mount the pickers on the outside of the cam-cylinder, in order to get them out of the way of other working parts, if this could be done. The real difficulty lies in doing it, and invention lies in the conception of the means for doing it, or in the special mode of pivoting. We cannot say, therefore, that there was anything patentable in the broad idea of pivoting a picker on the outside of the cam cylinder, and cutting a slot in the cylinder to afford an opening for the worldng end of the picker. The mounting of a picker on the outside of the cam cylinder necessarily involved either cutting a hole in the cylinder, or an arm extending over the top of the cylinder, as would appear to any mechanic.
As it would certainly be an extreme position for any court to hold that the first person who mounted a sliding or a pivoted picker on the inside of the cam cylinder was entitled to a monopoly of all pickers which slide or are pivoted on the inside of the cylinder, so, with equal truth, it would be an extreme position for the court to hold that the first person who mounted a pivoted picker on the outside of the cam cylinder, with a slot in the cylinder, or an arm extending over the top of the cylinder, was entitled to a monopoly of all pickers mounted in that way.
At this time Mayo was devising improvements in the automatic machine covered by his 188" patent, and, admittedly, there existed in the art both sliding and pivoted pickers, which were located on the inside of the cam cylinder. In his 1887 patent there were sliding pickers on the inside of the cam cylinder, and it occurred to him it would be better to have the pickers pivoted on the outside of the cam cylinder and extend through slots in the cylinder. At this stage of his mental conception he had made no invention, for this would have suggested itself to any skilled mechanic. The important and difficult question, however, remained, how can this change be made? In successfully solving that question, Mayo *535made his real and only invention. It follows that claim 6 must be limited to the mode of pivoting invented by Mayo, or to what may be fairly held to be the equivalent of his mode of pivoting.
Holding that this is a valid claim when so limited, the question of infringement turns upon whether the lifting pickers in defendants’ machine contain the Mayo mode of pivoting, or a mode of pivoting which performs the same function in substantially the same way.
In the Mayo construction, both the pivoted support for the picker and the pivot are located outside the cam cylinder. The slots through which the pickers extend are inclined slots, and the pivotal axis about which the picker swings extends in an inclined direction on the outside of the cylinder. The result is that the picker, in operating, swings radially in and out of the cylinder in an inclined direction. The pivoting of the picker is such that there is no other movement, except a swinging movement, to cause the end of the picker to travel in the proper path to shift the needles.
In the defendants’ machine the two lifting pickers are not separately pivoted to the cam cylinder, but both slide in a T-shaped carrier which is pivoted to the outside of the cam cylinder. The pivot itself extends through the wall of the cam cylinder, and is arranged radially to it. The defendants’ pickers swing upon a radial axis. They do not move radially, however, by reason of their swinging movement. In order to give them a radial movement, it is necessary to give them a sliding movement as well as a swinging movement about the radial axis. In a word, the Mayo picker operates by a simple radial or swinging movement, and the defendants’ picker by a compound movement — both swinging and sliding.
The defendants’ pickers do not infringe claim 6 of the Mayo patent, because, starting with a different conception, they are different in construction, arrangement, and mode of operation. Mayo solved the problem of pivoting a picker on the outside of the cam cylinder in one way, and Rowe in a radically different way.
We come next to Claim 11 — the remaining claim of the Mayo 1891 patent in issue:
“In a knitting-machine, the needle-cylinder grooved for the reception of jacks, a series of jacks, a surrounding ring correspondingly grooved for the reception of jacks, and a cam-ring, as a**, to act upon and move the said jacks, combined with an independent ring, b, overlapping the inner ends of the jacks, the hooks or portions of the jacks engaging the yarn being at the proper space thereof between their ends, substantially as described.”
This claim relates to thin blades of metal, called sinkers or jacks or web holders, which extend inward between the needles, and which co-operate with the needles to hold down and press forward the fabric during the knitting operation. These sinkers are provided with hooks to engage the fabric. They are located on the top of the needle C3dinder, and are arranged to slide radially between the needles. They are actuated by the cam ring, which also: serves to hold down .tlie-ir/outer ends against any upward pull.. There is also a second ring;,;-which overlies, and thus serves to. hold *536down, the inner ends of the sinkers. The improvement covered by this claim is for this additional ring described in the claim as “an independent ring, b, overlapping the inner ends of the jacks.” The problem is here simply to hold down the sinkers, and the essence of the Mayo improvement lies in the location of the second ring to hold down their inner ends.
It is not every minor change or improvement in construction in a machine composed of many different parts, like a circular knitting machine, that rises to the dignity of invention. The work of holding down the sinkers had been previously done by two rings, both located on the outer side of the hooks of the sinkers. In view of what already existed in the art, and the nature and character of this improvement, there was no invention or patentable novelty in simply changing the location of one of the rings to the other side of the hooks, or to the extreme inner ends of the sinkers.
We come now to the second branch of the case at bar, which relates to an attachment to the machine for the purpose of introducing an extra thread during certain portions of the knitting, especially the heel and toe. This mechanism is the subject-matter of the Johns patent and of the Ames patent. The latter is for an improvement on the Johns patent. These patents are known as the “winder patents,” since their main feature is a rotary winder for winding the end of the extra thread several times about the main thread.
The general problem of automatically inserting an extra thread in a circular knitting machine has been solved in two ways, which are represented by two distinct types of mechanism: (1) The extra thread may be carried forward to the needles by frictional engagement with the main thread by means of a carriage containing a guide through which the threads passes. The carriage is moved forward to bring the extra thread into engagement with the main thread, and, when the extra thread is to be withdrawn, the carriage is moved back, and the extra thread drawn to one side, and between the blades of a cutting device which severs it. This is the frictional type of extra thread-inserting mechanism illustrated in the Shaw and other patents. (2) Instead of depending on frictional contact, the extra thread may be carried around the main thread by a winding device, the purpose being to make a more secure engagement of the threads than is made by frictional contact. This is the winder type of extra-thread-inserting mechanism illustrated by the patents in suit.
The rotary winder method may be properly called a preferred method, since it would apparently operate with more certainty in the case of fine threads with a smooth surface.
Both the automatic frictional contact devices, as well as the rotary winder, have gone into extensive commercial use.
There are also two distinct types of rotary winding devices. In one type, the clamping mechanism which clamps and holds the free end of the extra thread during the winding operation rotates with the winding mechanism, or winder proper. In the other type, only the winder proper rotates. While the clamping mechanism coacts with the winder, it is separate from it, and does not rotate with it. *537The Johns and the Ames patents belong to the first rotary winder type, and the defendants’ machine, constructed under one of the Rowe patents, to the second rotary winder type. The mechanism which embodies these distinct conceptions of a rotary winder is necessarily different in construction and mode of operation.
On this branch of the case, the important question on the issue of infringement is whether the Johns patent should be limited to cover the particular type of winder in which both the winder and the clamping mechanism revolve, or whether, in accordance with the broad language of the claims relied upon, it should be construed to cover both forms of rotary winder.
At the time of the Johns invention, the conception of a rotary winder was not new in the art. The prior British patent to Marshall & Hewitt, and the prior American patent to Swinglehurst, certainly disclosed rotary winders. While a strong case has been made out against the operativeness of the Marshall & Hewitt mechanism, and while, in the Swinglehurst device, the entire bobbin of extra thread rotates around the main thread, still it is manifest that the broad idea of the rotary winder is present in these mechanisms. Further, automatic devices for the insertion of an extra thread were in common use, although of a different type. Again, the Johns patent is not for a generic invention, in which the device of the patent is only one form in which the invention may be embodied. These considerations are sufficient to show that Johns is not entitled to a broad patent covering every kind of rotary winder in a circular knitting machine. The most that the patentee can ask, under the circumstances, is a liberal application of the doctrine of equivalents in the construction of his patent. There is no warrant for holding that the invention can be so enlarged as to include all forms of rotary winder.
The fundamental and characteristic feature of the Johns conception of a rotary winder is a clamping device for the free end of the thread which is attached to the winder and rotates with it. This is apparent upon an inspection of the device. There is a stationary boss, with two holes — one in the center, for the passage of the main thread, and the other at one side, for the passage of the supplementary thread. Mounted on the stationary boss, and capable of rotation thereon, is a hollow ring or shell, which forms the rotary winder. To this winder is rigidly fixed a curved arm with a projecting finger; also a pivoted lever operated by a spring. The arm, in conjunction with the lever, forms a clamp, which is opened or closed to grasp the additional thread. When this thread is not in use, its free end is held by the clamp. When the time arises for the introduction of this thread into the knitting, the winder is rapidly rotated, with the result that the free end of the thread is carried several times around the main thread. Before the revolution of the winder ceases, the outer edge of the revolving lever is struck by a projection on the lever which operates to open the clamp and release the free end of the thread, so that it may pass to the needles with the main thread.
The Rowe winder used in defendants’ machine is different in construction and mode of operation. Its important parts consist of a cylinder or winder mounted upon a boss. This cylinder is provided *538with a large opening in its side, which acts as a thread eye for the extra thread when it is wound around the main thread. The thread lies loosely in this opening, and the vertical wall of the slot, when the cylinder is rotated, acts to throw the end of this thread about the main thread. The clamp or gripper comprises a fixed jaw and a movable jaw, which are moved in and out through the opening in the cylinder or winder. A cutting plate is secured to the fixed jaw, and, when this jaw is moved to clamp the end of the extra thread, it cuts the thread. When the extra thread is not passing to the needles, the gripper extends through the opening in the cylinder or winder, and the end of the extra thread is clamped and held therein. When the extra thread is to be inserted, the gripper swings upon its pivot, and draws the extra thread through the opening in the cylinder. The cylinder is then rotated. During the greater part of the first revolution of the cylinder the gripper retains its hold upon the end of the extra thread, thus causing it to be wrapped around the cylinder. The jaws of the gripper are then opened to release the end of the thread, and the continued revolution of the cylinder causes the projecting end of the thread to be thrown around the main thread by the engagement of the extra thread with the vertical wall of the opening in the cylinder. When the extra thread is to be withdrawn, the gripper is moved inward through this opening. The extra thread comes between the jaws of the gripper, and is severed by the closing of the movable jaw at the same time that the free end is clamped between the jaws.
The claims of the Johns patent in issue are as follows:
“(1) In a knitting-machine, the combination with a source of plural thread-supply, of a rotary winder for carrying the free end of one thread one or more times about another.
“(2) In a knitting-machine, the combination with a source of plural thread-supply, of a rotary winder for carrying the free end of one thread one or more times about another, and means for severing said thread.
“(3) In a knitting-machine, the combination with a guide device for a plurality of threads, of a rotary winder for carrying the free extremity of one thread one or more times about another.
“(4) In a knitting-machine, the combination with a guide device for a plurality of threads, of a rotary winder for carrying the free end of one thread one or more times about another, and means for holding said thread end previous to winding.
“(5) In a knitting-machine, the combination with a guide device for a plurality of threads, of a rotary winder for carrying the free extremity of one thread one or more times about another, means for holding said thread end previous to winding, and means for severing one end of said thread.”
These claims must be limited to a rotary winder, which contains the characteristic feature of the Johns patent, namely, a winder in which the clamping mechanism rotates with the winder; and they cannot be enlarged to include the mechanism employed in the defendants’ machine, in which the gripper does not rotate, and which requires an essentially different structure and organization of parts. It is clear that, when the claims are limited to the actual invention of the Johns patent, they are not infringed.
It may be noticed in this connection that these broad claims are not found in the original application for the Johns patent. They were not inserted until five years after that application, and subsequent to the *539date of' the *Rowe patent, and to the time when the Rowe device was put upon the market.
It is unnecessary to consider the Ames patent in great detail. It is for an improvement on the Johns type of rotary winder. In Ames, we find a rotary winder and gripping mechanism in the form of a sleeve with a series of wire projections like a brush. These wires or projecttions extend parallel with the axis of the winder. The end of the extra thread is engaged between two of the wires of the brush, and when the brush rotates it carries the extra thread several times around the main thread. In the subsequent movement of the extra thread with the main thread towards the needles, the extra thread is drawn from between the two wires of the brush or holder. A finger with a hooked end then engages the extra thread, and pulls it to one side between the blades of the cutting device. In the movement of this finger in drawing the extra thread to one side, it is again drawn between two of the brush-like wires, and, after the thread is cut, its free end is left between the wires in readiness for another winding. It will be noted that in this arrangement the clamp or holder does not move to the extra thread and grasp it, but the extra thread, in the operation of cutting, is moved into engag-ement with the clamp, or between the opening of two of the wire projections. This is the fundamental distinction between the Ames and Johns devices.
The claims in issue are as follows:
“(8) In a knitting machine, the combination with a guide device for a plurality of threads, of a rotary winder through which said threads are passed, one or more projections on said winder, and means to draw one of said threads to one side and leave its end in the path of movement of said projection or projections, to be rotated thereby.”
“(10) In a feed mechanism for a knitting machine, the combination of a tubular twister slotted at one end, means for rotating said twister, and means for throwing a thread into engagement with the slotted end of said twister.
“(11) In a feed mechanism for a knitting-machine, the combination of a tubular twister slotted at one end, means for throwing a yarn into engagement with the slotted end of said twister, means for rotating said twister, and means for cutting off said yarn.”
These claims cannot he construed to cover every winder in which the thread is drawn to one side, and into a position to be engaged by the winder. This is a patent for an improvement in a highly advanced art, in which not only the Johns patent, but the Williams patent of February 7, 1893, must be included. These claims must therefore be limited to the specific combinations of devices set forth in the claims; in other words, to the Ames organization of twister or holder and winder. From the description already given of the defendants’ winder, it is manifest .that it differs in structure, arrangement, and mode of operation. For these reasons, as in the case of the Johns patent, the defendants’ winder does not embody the invention covered by these claims of the Ames patent.
Upon the argument on both branches of this casé, we were impressed with the wide difference, in the mechanism of the complainant’s and defendants’ machines. That impression has only been confirmed by a more careful and thorough examination of the exhibits, the record, and the briefs. .The machine of the Rowe patents, which is the der fendants’ machine, has not borrowed, nor is it built upon,, .the .specific *540inventions, which are actually covered by the several claims in issue of the patents in suit. The most Rowe has done has been to solve, in an advanced art, the same, specific problems, along substantially different lines.
In the argument on this appeal, and especially in its supplemental brief, the complainant has laid special emphasis on the point that Mayo, in his 1891 patent, was the first to produce a successful commercial automatic picker machine. The difficulty with this argument, conceding that such an argument has weight in some cases, is it indefiniteness. The complainant manufactures the Acme machine, which has gone into extensive commercial use. But in certain particulars, including the arrangement of stitch cams and pickers, the Acme machine differs from the Mayo 1891 patent, in suit. Mayo himself testifies that this machine is more automatic than the machine of that patent. Then, again, the 1891 patent covers other inventions than those which are found in the claims in issue. There may also have been other things which have helped to give the Acme machine its commercial success. There is no satisfactory proof in the record that the introduction of the two-and-one method into the picker machine, supplemented by pickers pivoted on the outside of the cylinder, differentiates a successful commercial machine from an unsuccessful noncommercial machine. Mayo’s testimony is only to the effect that these features were improvements, and made the machine work satisfactorily.
The two-and-one method is simply an improvement in the circular knitting-machine art, and nothing more. It has nothing to do with the automatic character of the machine, its only office being to produce a somewhat better product. It does not separate success from failure in the art. Machines with the one-and-one method are successful commercial machines, producing commercial products. The following excerpts from Mayo’s testimony fully corroborate this position:
“Int. 23. To what extent, generally speaking, is what you have called the ‘two-and-one arrangement’ employed in the circular, seamless knitting machines now on the market?
“Ans. Nearly all the machines now on the market, either full automatic or partially so, are using a two-and-one method of knitting the heel and toe.”
“Cross-Int. 25. How many circular, seamless knitting machines are now on the market or in use? I mean, how many different makes?
“Ans. Those that I recall at this moment are the Standard, the Invincible, the Excelsior, the Boss, the Keystone, the Paxton & O’Neill, one made by H. Brinton & Co., the Acme, the Branson, the Victor, the Scott & Williams, the Hepworth (which I think has, however, lately been withdrawn from the market), and the National; the last named being a spring-needle machine.”
“’Cross-Int. 30. And in how many, or what ones, of the machines you have mentioned, is the two-and-one method of knitting the heel and toe employed?
“Ans. It is employed on the Standard, Invincible, Excelsior, Boss, Paxton & O’Neill, Acme, Victor, and, I think, on some of the others, but cannot state positively.”
The history of this art discloses that neither a fully automatic machine, as distinct from a partially automatic machine, a pivoted picker on the outside of the cylinder, as distinct from a sliding picker on the inside, nor the two-and-one method, as distinct from the one-and-one method, are essential to a successful commercial machine producing a commercial product.
*541For these reasons the case at bar does not come within the principle laid down in Consolidated Valve Company v. Crosby Valve Company, 113 U. S. 157, 5 Sup. Ct. 513, 38 L. Ed. 939; Keystone Manufacturing Company v. Adams, 151 U. S. 139, 14 Sup. Ct. 395, 38 L. Ed. 103, and other cases in which an inventor has added to previous failures just what was necessary to make a practical commercial machine.
The main grievance of the complainant respecting the decision below is that the Circuit Court failed to recognize the doctrine of equivalents in its application to the Mayo inventions. In view of the scope and character of these inventions, we know of no doctrine of equivalents which would permit us to hold that the butts of the following needles, or the spring of the T-shaped carrier, in the defendants’ machine, are the equivalents of the positive locking devices conceived by Mayo, and embodied in his automatic picker.
By reason of the nature of the machine and the number of patents involved, this case has called for an unusual amount of time, and labor. Upon full consideration, we have reached the conclusion that the decree of the Circuit Court must be affirmed.
The decree of the Circuit Court is affirmed, and the appellees recover their costs of appeal.