DENISON, Circuit Judge.
[1] This is an infringement suit based on patent No. 908,649, issued January 5, 1909, to John P. Curry, for a “tool for twisting wife ties.” The District Court thought there was no infringement, and dismissed the bill; complainants appeal. The case is one of those where it is clear enough that the defendants’ device departs from the form shown in the patent drawing, and yet where, in a broad sense, there is equivalency. Plence, it is necessary to ascertain whether the patentee’s actual improvement in the art was generic enough to cover the defendant’s form, and if so, then to determine whether the patentee has, by his claims, imposed or accepted, such limitations as negative infringement. The patent in suit was built upon the common idea, most familiar in screwdrivers and' drills, that by connecting an outer sleeve and an interior shaft by a screw thread engagement- of sufficiently 'high pitch, a longitudinal motion of the sleeve upon the shaft will cause the latter to revolve, and, of course, its operating end would carry with it, in the revolution, anything with which it engaged and which was free to revolve.
The only earlier application of this idea which needs mentioning— the only one because it is the closest — is in the patents to Cassidy, Nos. 666,514 and 656,513, dated August 21, 1900. These were for use in connection with wire ties for fastening corks into bottles. The tie ’is a double wire, looped at its center, which thus becomes one end, and then twisted upon itself, and then having, for the remainder of its length, the wires parallel and with free ends. The loop was laid on top of the cork, projecting forward and down over the edge. The twisted wire was passed down behind the cork to a point below a *424shoulder on the neck; the free ends of the wire were then brought forward passing on each side of the neck under the shoulder and tiren were brought together and put up through the loop. At this time, the looped end and the free double end were gripped by Cassidy’s tool, which, except as to the head on the inner end of the revolving shaft, was essentially like Curry’s. This head consisted of a short arm extending one way at approximately a right angle to the shaft. This bent arm was passed through tire loop in the wire, and then tire two free ends were caught in a wedge-shaped slot in the end of the arm. In some forms of Cassidy, both the loop and the double end of the wire were held i>y the head almost or quite together and nearly in the axial line of the shaft; in other forms, they were held both at one side of such line, one a little further away than the other, but close together. The result was that when the shaft was revolved by sliding the sleeve along it, the loop end and the double end of the wire would twist together, and the loop would have a double twist around the free ends. If both were substantially in the axial line of the shaft, they would be left so close together and with such a compound twist as to be difficult to get hold of to untwist; if they were both held on one side of this line, but as near together as illustrated, the untwisting would be difficult and the twisting would be complicated and for some purposes made more or less impracticable by the fact that both wires, instead of merely twisting together, would swing around a circle, and one upon a larger circle than the other. It occurred to Curry that this general plan could be applied to wire ties for the necks of bags holding Portland cement. He proposed to use a single wire tie having a closed eye at each end, and he adapted his twisting tool to this tie by providing at the inner end of the shaft a right-angled arm extending both ways and carrying two hooks, each equally offset from the axial line, and so shaped as to engage the eyes at the end of the wire. It followed that when the tie was put around the neck of a bag and the eyes engaged by the hooks and the shaft revolved, the wire was perfectly twisted, without any of the swinging motion or compound twisting of Cassidy’s forms, and the free ends of the wire, 'each end consisting of an eye, were necessarily left separated from each other the distance which the hooks were apart, and in the most convenient form for untwisting.
The device of the Curry patent rapidly went into general use in cement mills, and to such extent that in the year 1913 it was in use by 80 per cent, of the cement mills jn the country, and applied 160,-000,000 ties. No other device for this purpose appears to have been commercially known in this particular field until defendant put out its tool. This is the same as Curry’s, except that the tie, instead of having an eye at each end, has a head; and the twisting tool, instead of having hooks entering the tie eyes, has in the same location wedge-shaped slots in which the heads enter and by which they are held in engagement while being twisted. For better understanding, we may say that the right-angled arm or cross-bar of defendants’ tool carries, at each of its ends, a double claw like the familiar claw hammer, and that these are engaged with the heads of the wire tie just as a claw *425hammer surrounds and grips the head of a nail when the nail is passed down to the narrow part of the claw-opening. Plaintiff’s and defendants’ forms are here illustrated:
Plaintiff’s: Mg. 3 of tho Curry Patent.
Defendant’s: Mgs. 8 and 5 of the Beam Patent.
The thing which Curry did, as compared with Cassidy, was to provide a tool of this class carrying upon its revolving shaft a crosshead provided, at points opposite each other and equidistant from the shaft’s axial line, with means for engaging the two free ends of a wire tie provided with corresponding engaging means. Cassidy fell short of this; and while it now seems a rather obvious thing to space the two engagement points opposite each other and equidistant from the axis for engaging the two free ends of a tie, it never had been clone with any closely analogous tool, it simplified Cassidy’s tool, it produced a stronger, surer and more accurate twisting, and it left the tie better adapted for untwisting and removing. This was invention, and entitled him to a patent of the scope just stated; but his claims are said to be limited to less than the real invention for the reasons to be mentioned; and the case must turn upon the application of the doctrine of equivalents.
The three claims of the Curry patent, each of which is said to be infringed, are quoted in the margin. In the epurt below the limita*426tions of the first claim — particularly as imposed during the application —formed the subject chiefly considered; and it was thought the second and third carried the same limitations. From the history of tire case, we conclude that the first claim is, as to the form of the “re-curved hooks,” more limited than either of the others; and as we think the broader claims’ (in this respect), the second and third, are infringed, it will not be necessary to determine tire precise scope of the first. The Patent Office record is peculiar. The first claim, as first presented, was as follows:
“Claim 1 — The tool for twisting wire ties, consisting of the spindle having a spiral body with handle adapted to rotate the same by a longitudinal movement, the end of the spindle having two recurved hooks equidistant from the axis of the spindle and at a suitable distance apart to hold the ends of a wire tie in position for untwisting.”
The second and third claims were originally in their final form. The only (now material) difference between the original first claim and the second claim is that the first omits the reference letters used in the second and that the first specifies that the hooks shall be “recurved.” The examiner, in his first action, required the correction of several errors in the specification, and rejected claim 1 on reference to Casr sidy, saying further that “claims 2 and 3 are deemed allowable, subject to the foregoing.” It is suggested that this action is also a rejection of claims 2 and 3 upon the same ground; but this is incorrect. It is the fixed practice of the Patent Office not formally to allow any claim until all objections, to description or drawings are removed and until all the claims are ready for allowance. Prior to that time, it is customary to indicate that certain claims are allowed or allowable subject to the doing of whatever is necessary to make the entire case ready for allowance; and, in this case, the examiner’s language means that claims 2 and 3 are found to be allowable, but cannot be formally granted until the specification has been corrected and until claim 1 is either canceled or put in acceptable form.
[2] Obviously, this action indicates that the examiner considered claim 1 as broader than claim 2, f or he would not intend to allow the, broader claim and reject the narrower. This inference that the first was broader must have' resulted, mainly, on the absence therefrom of reference letters. We have recently had occasion to refer to the effect of such letters, in Houser v. Starr, 203 Fed. 264, 269, 121 C. C. A. 462, .and we have given the subject further ‘ consideration in this case. We cannot regard their presence or absence as being of any controlling effect. Sometimes, they may have some importance in construction, and sometimes not; just as with reference to the formula “substantially as described.” National Co. v. Mark, 216 Fed. 507, 515, 133 C. C. A. 13. The name of a part, as an element of a claim, of necessity carries us to the specification to see what that part is, and *427the reference letter in connection with the part does not naturally do more. In either case, the claim refers to the named part, in the form shown in the specification and drawings or in any other form which is the equivalent of the form shown; and in'determining equivalency, we cannot, usually, get aid from any supposed limitation arbitrarily to be implied from the use of a reference letter. We are aware that sentences may be found among some of the opinions of the Supreme Court, and, indeed, of this court, which, taken by themselves, imply the existence of such an arbitrary limitation, but in no case do we find this to have been the essential ground of decision or to have been more than a makeweight. In Knapp v. Morss, 150 U. S. 221, 228, 14 Sup. Ct. 81, 37 L. Ed. 1059, it is said that the patent must be restricted to the specific form indicated by the letters of reference, because “otherwise, the effect would be to make the claim coextensive with what was rejected in the Patent Office”; and so it appears there that, as usually, the limitation was in truth imposed by the state of the art and not by the reference letters. So, also, in Ross Co. v. Randall, 104 Fed. 355, at page 359, 43 C. C. A. 578, this court, in an opinion by Judge Day, reviewed some of the cases which touched on the subject, and said:
“An analysis of the cases will show that the conclusion reached depends upon the character of the improvement under consideration.”
The case held, in substance, that when the invention has already been decided to be. of a very limited character, the use of reference letters may re-enforce that conclusion. In two opinions of this court, written by Judge Taft (McCormick Co. v. Aultman, 69 Fed. 371, 393, 16 C. C. A. 259, and Bonnette Co. v. Koehler, 82 Fed. 428, 431, 27 C. C. A. 200), it is held and reaffirmed that the patentee describing his elements by reference to letters on the drawing does not thereby deprive himself of the benefit of the doctrine of equivalents otherwise applicable to his patent. See, also, National Co. v. Interchangeable Co. (C. C. A. 8), 106 Fed. 693, 715, 45 C. C. A. 544; Kelsey Co. v. Spear Co. (C. C. Pa.), 155 Fed. 976, 980, 981; Walker on Patents (4th Ed.) § 117a.
Rarely, if ever, can a claim using reference letters be saved from anticipation by limiting it to specific form, when the same result would not follow from the mere naming of the part with the implied reference to specification and drawing for identification. To hold that a limitation, beyond what would otherwise exist, is necessarily to be implied from the presence of reference letters, would be so contrary to the underlying principle of interpreting claims by reference to the specification that it cannot be accepted as a general rule. We may add that, just as with “substantially as described,” it was, 40 years ago, very common — almost universal — to find reference letters in a claim, while now it is unusual; and it cannot be supposed that patents have grown broader as their number, complexity and refinements have increased. National Tube Co. v. Mark, supra, at page 521 of 216 Fed., 133 C. C. A. 13.
This application, as originally filed, showed an exception, or a possible exception, to the general view just stated. The natural ground *428for assuming, as the examiner did, that the first claim was broader than the second, was that -the second used reference letters and tire first did not. If two claims differ from each other only in this respect, it must be assumed that the letters are intended to imply a smaller range of equivalents than is intended by the claim without the letters; and we may concede that the examiner was justified (as to most of the elements) in taking this view of the first claim. In response to the office action, plaintiff amended his first claim by inserting reference letters, thus making it exactly like the second (save for the addition of “recurved,” and the omission of “extending laterally in the same direction”), and then he added the functional reference to use “for twisting wire ties having eyes upon their opposite ends” which is said to limit this claim to the particular form of tool specially adapted to engage with that particular form of tie. At any rate, it is clear that the first claim took such form that it might be'narrower than the second, as to the element “hooks.” The claim, as so amended, was at once allowed. We see nothing extraordinary, nor, indeed, unusual in this result. The examiner, from the beginning, saw that a claim which covered this tool and not Cassidy’s was allowable. He feared, rightly or wrongly, that the language of tire original first claim was so vague that it might cover Cassidy’s, and so he required some limitation^ the plaintiff could not have two claims alike, and so, rather than abandon the first claim altogether and rest the case upon his second, he limited his first claim so as to be not as broad in this respect as the second.
In all this, there is nothing which should limit the second claim beyond the fair import of its words. The second claim is for:
“The head, d, upon the end of the spindle wfth two hooks, 0, †, extended in the same direction laterally upon the head, equidistant from the axis of the spindle and at a suitable distance apart to hold the ends of a wire tie in position for untwisting.”
The entire structure thus described in this part of the claim was novel over Cassidy; it is the essential basis for a very successful tool; the “hooks e, are not limited to a recurved form as in the first claim, and as this recurved form was specially intended for engagement with an eye, the hooks, when spoken of without that limitation must be-thought of as capable of other engagement also; and for every purpose and every function here important, both in twisting and in untwisting, the engagement of a double claw hook with a head on the wire is the equivalent of the engagement of a single claw hook with an eye on the wire. This is emphasized by the fact that the two forms of tie — with eye and with head — were well-known alternatives before Curry selected one of the forms, and by the further fact that although the tie with the head, the defendants’ form, could not be used in the Curry tool constructed as shown in his drawing, because the head would slip off from the hook; yet the Curry tie, the form with the eye, can be used in defendants’ tool, since, if the tie just below the'eye is passed into the double claw, the base of the eye serves as'a head. This method of use is not convenient nor very practical, but that it is possible strongly indicates the essential similarity between the two tools. We are satisfied that claims 2 and 3 are fairly entitled *429to such a range of equivalency that defendants’ tool, Complainant’s Exhibit 5, infringes.
Among the exhibits, we find a tool marked “Defendants’ Exhibit 2.” We can find no reference to this exhibit in the record, save when it was mentioned by expert witnesses, and we infer that it was a tool constructed only to illustrate a theory. If there were evidence of the making or selling of this form, it would be necessary to decide whether it escaped claims 2 and 3, on the theory that its two hooks did not extend in the same lateral direction from the head, and, if so, then whether it was within claim 1; but, as to these questions, there is neither evidence nor argument, and, as was said in Grever v. Hoffman Co., 202 Fed. 923, 927, 121 C. C. A. 281, it will be time enough to determine such questions, when the device is put into actual use and becomes important enough to justify an infringement suit.
[3] The decision below was in part rested upon the inference to be drawn from the fact that defendants’ tool was made pursuant to a later patent granted to one Beam, and it was said that the issuing of the later patent to Beam was either a mistake or a Patent Office decision that the Beam tool did not infringe the Curry patent. In view of what was said on this general subject by this court in Herman v. Youngstown Co., 191 Fed. 579, 584, 112 C. C. A. 185, we infer that the court below thought this case came within the exception suggested in the Herman-Youngslown Case, rather than within the general rule there stated. We are not able to accept that view. If it were to be assumed that the Curry patent was so limited that it could reach nothing except the precise form shown in the drawing or other forms equivalent in every detail and aspect, then it would be true that the later patent could be valid only on the theory that its device was not that kind of an equivalent of the device of the earlier patent; but, obviously, if it is assumed, as we here conclude, that the earlier patent is entitled to a range of equivalency broad enough to reach the later form, the question whether the later form also embodies patentable modification or improvement is quite immaterial. Such confusion as there is on this subject, must come from failing to observe that the word “specific” has only a relative meaning. An invention may be most properly characterized as specific with reference to what has gone before; but if there ever was a specific improvement or construction which was worthy the name of invention and so entitled to a patent, and yet which might not turn out to be relatively generic — and so enlitled to the benefit of equivalents — as against some possible future construction, such a case does not readily suggest itself in concrete form; and, at any rate, this is not one. It is a natural thought that if device b is the mechanical equivalent of patented device a and so infringes, then, because it is the equivalent of what is old in the art, it cannot be patentable, and the finding that it is patentable implies that it is not an equivalent; but this also fails to observe the relative rather than the absolute meaning of equivalency. Its existence depends on its range or scope, and device b may be the equivalent of a, when the latter is broadly considered, and yet a not be the equivalent of b, when the latter is defined as narrowly as it must be.
*430The decree below is reversed and the usual injunction and accounting decree will be entered as to claims 2 and 3. It may be silent, as to claim 1; but since the matter of infringing claim 1 is not decided either way, but is thought to be immaterial, we do not see that the normal disposition of costs should be affected.
Urschel’s personal liability has not been considered in the court below or here¡ and it will be determined by the District Court.