FRIENDLY, Circuit Judge.

International Latex Corporation, a Delaware corporation, assignee of Spanel patent No. 2,360,736 for a seamless dipped latex girdle, brought this action against The Warner Bros. Company, a Connecticut corporation, for infringement. The case was tried in the District Court for Connecticut before Judge Anderson who, in a carefully reasoned opinion, 169 F.Supp. 755 held that the patent was valid and that it had been infringed as to claim 3 but not as to claims 1, 2 and 6. We agree that the patent was valid but that claims 1, 2 and 6 were not infringed. We have come to the same conclusion as to claim 3.

I. The History

Spanel’s patent was for an improvement in women’s girdles. The patent application, filed in June, 1940, sets forth the history of the art, most of this in a manner not seriously challenged. It describes the elaborate stiffened corset typical of the Victorian era; this, it states, was modified and ultimately discarded in favor of the girdle. Until Spanel’s invention girdles were “generally made of elastic textile fabric or panels of diverse material such as elastic fabric, textile fabrics and the like.” One-piece girdles made of such fabric are known as “one-way or two-way stretch girdles”; they stretch only with these limitations as distinguished from the omnidirectional stretch of Spanel’s latex girdle. Ac *559 cording to Spanel, while such fabric girdles gave “considerable satisfaction in use,” they were subject to certain disadvantages — among these that such girdles took “a relatively long time to wash and especially to dry, so that the wearer is often required to own at least two of the girdles in order to keep one in clean condition,” that they tended “to leave welts, ridges, cross markings or other markings on the skin,” and that they were “relatively expensive and also tend to lose their shape and elasticity rather rapidly, especially if washed often.”

Latex, the milky fluid procured by tapping the rubber tree, seemed to offer a possible escape from these difficulties. Until the 1920’s the usual procedure was to turn the latex fluid into crude rubber; this was then “masticated” into sheet or liquid form. In 1924 girdles made of sheet rubber panels sewn together achieved a vogue, but, as stated in one of defendant’s publications, “the ‘rubber corset boom’ burst with a violence equal to its original blowing,” since the garments proved uncomfortable or defective in numerous respects. About 1920 improvements in the techniques of preservation and transportation made it possible to keep the milky latex gathered at the tree in liquid form and bring it to the site of manufacture so that articles might be made by dipping a form or mold into a tank containing the liquid latex. Such dipping is an old art.

The novelty of Spanel’s conception, which was the basis of the patent in suit, did not lie in the use of liquid latex rather than sheet rubber and the avoidance of seams, both of which were indicated by the prior art, but in a thought that for the first time made a seamless latex garment feasible as a girdle. Plaintiff’s witness Ambrose testified that before 1940 the girdle art “was oriented in the direction of selectivity applying constriction on the body.” This “tailoring tradition,” as plaintiff’s counsel describes it, was a deterrent to the manufacture of a seamless latex girdle. Not only was it impracticable to manufacture or market latex girdles moulded to every variation of the female form, there was testimony that such a girdle “wouldn’t mold and wouldn’t flatten the bulges out.” As stated by plaintiff’s witness, Spanel “took the reverse twist. He was very much interested in a uniformity of your stresses, a multi-directional type of stretches, a multi-directional type of force, and a uniform type of stresses, instead of trying to intensify in certain areas.” As set forth in the patent, Spanel found that by using a liquid latex material for the walls of the girdle “considerably thicker than has been heretofore usually employed in the manufacture of garments from liquid latex,” in the range of about 16/1000 to 30/1000 of an inch, he obtained a material having a uniform stretch “in every direction in any given square centimeter of area.” He discovered also that by using this material in a flat form following the contours of the body only to a limited extent, he could produce a garment which, as the District Court found, 169 P.Supp. 758, “supports and constricts the wearer’s torso, giving a firm, slimmed appearance, without markedly restricting freedom of movement.” Among the advantages claimed by Spanel for his girdle were higher tensile strength and consequently lighter weight than the crude rubber girdle; the absence of seams; more evenly distributed constricting effect, giving the wearer the lines of her own figure “ ‘slimmed down’ and properly formed”; a smooth outer surface resulting in a smooth contact with garments worn immediately over the girdle so that these would not tend to “ride up”; and a great reduction in the number of size models required, with attendant savings in manufacturing and merchandising costs.

Spanel’s girdle had a stormy voyage through the patent office; as will later appear, the details of this have an importance requiring a somewhat tedious summary.

As filed, the application contained 35 claims. Most of these concerned the basic invention, described in what was then claim 1 as “a seamless, one-piece, *560 latex girdle in the form of a hollow, flattened, truncated-cone shaped member in normal position, but adapted when pulled into position on the wearer’s body to be stretched and to serve as an enclosing flesh-supporting sheath”; claims 6-14 inclusive recited an integral crotch (“panty-girdles”) and claims 25, 26 and 27 spoke of the nature of the inner surface, as to both of which more hereafter. In November, 1941, the examiner rejected all the garment claims on the ground of undue multiplicity and also in view of Charnaux’ patent No. 1,729,502 for the use of perforated rubber in binding garments, together with Spanel’s own patent No. 2,185,527 for seamless latex baby pants.

An amendment of June, 1942, cancel-led all previous claims and filed new claims 36 through 44. These new claims again related primarily to the basic invention; the crotch remained in several of the claims but all reference to the different interior surface disappeared. All were again rejected on the basis of the Charnaux and Spanel patents mentioned above, and also Barnes No. 2,125,482 and an Italian patent, Guiseppe No. 258,394, which were stated, along with Spanel’s baby pants, to have taught “the use of an integral crotch member.”

A further amendment of April, 1943, cancelled all previous claims and added new claims 45-49. Claim 45 contained a more limited statement as to the crotch; claim 49 coupled the basic invention with claims relating to perforation and slight roughening of the interi- or surface. Spanel’s solicitors, in addition to arguing non-anticipation of the crotch by Barnes et al. and of the basic invention by Charnaux and the baby-pants, said of Claim 49, “No prior art has been cited pertinent in any respect to this feature.” In July, 1943, the Examiner again rejected all claims, adding a citation to Work No. 1,632,017, alleged to have taught “the use of small perforations for applicant’s purpose.”

Nothing daunted, applicant was back in November, 1943, with three new claims, 50, 51 and 52; the two latter contained references to roughening of the inner surface and are identical with Claims 1 and 2 in the patent as issued; Spanel’s solicitors elaborated on the advantages of the roughening and emphasized that “None of the prior art citations suggests any comparable construction.” In December, 1943, all these were rejected, the examiner stating that “the rejection may be taken as final for the purpose of appeal, in case applicant so elects.”

Applicant did not so elect. Instead he filed, on April 17, 1944, new claims 53 to 56 replacing all the prior claims except 51 and 52. Claim 53 is the final version of the crotch claim and became Claim 3 of the patent as granted. Claim 56 contains reference to pinpoint perforation and roughened inner surface, in substantially the same language as rejected claim 49; it constitutes claim 6 of the patent as granted. Quite inexplicably, this succeeded when all else had failed. On April 28, 1944, the patent issued, including the two “roughening” claims, 51 and 52, previously rejected, the “roughening-perforation” claim 56, formerly rejected when it had been presented as claim 49 in almost the same words, and the narrowed claim 53 as to the crotch. We set forth in the margin claims 1, 2, 3 and 6 of the patent. 1 As will be seen, *561 claims 1 and 2 relate to an integral one-piece supporting latex girdle with roughened inner surfaces, claim 6 refers to such a garment also having pinpoint perforations distributed on the front and back areas, and claim 3 is for such a girdle with a particular type of integral crotch.

Plaintiff placed the seamless latex girdles on the market early in 1940. Exploitation was necessarily curtailed in 1942 due to the shortage of rubber. When manufacture was resumed after the war, the girdles proved a tremendous success. Sales achieved a volume of 3,-996,000 units and $10,631,000 in 1950. However, all girdles manufactured and sold by plaintiff during this period had a smooth inner surface, and therefore did not practice the invention as stated in claims 1, 2 and 6 of the patent. The record does not show how many were within claim 3.

About 1950 the defendant, long a prominent manufacturer of girdles, became aware of plaintiff’s increasing sales of seamless latex girdles. It arranged to have certain such girdles manufactured for it, and determined, on the advice of counsel, that these would not infringe Spanel’s patent. Defendant’s panty girdle had a somewhat different form of crotch; and defendant’s girdles accomplished the departure from smoothness in the interior surface, desired to promote evaporation and provide a more comfortable “feel,” not by abrading the surface of the latex, as Spanel had taught, but by a process, known as “flocking,” which applies fine particles of fabric such as cotton, rayon or wool before the latex has hardened. Plaintiff likewise has used “flocking” on all its girdles which do not have smooth inner surfaces; most of the girdles it has made since 1950 have been flocked. As might have been expected, plaintiff did not agree that defendant’s variations avoided infringement. Hence this suit.

II. Validity

Defendant contends Spanel’s patent is invalid. It claims there was no invention in Spanel’s basic concept in view of the prior art and particularly because of Charnaux, Young and Hemm No. 1,-923,524, and Spanel’s own patent for baby pants, and certainly none in roughening the inner surface, with or without perforations, or in an integral crotch. We agree with the District Court that the patent is valid because of the invention in Spanel’s basic concept, although methods for roughening the inner surface of latex were known, perforation also was known and had been specifically taught by Charnaux, and the *562 precise form of integral crotch taught in claim 3 would scarcely meet the statutory test of patentable invention, 35 U. S.C. §§ 101, 103, even though, as the District Court held, 169 F.Supp. 764, it had not been anticipated by Barnes or the baby-pants.

Although the prior art had disclosed the advantages of using rubber in garments for constricting purposes and the feasibility of seamless latex objects in general, Spanel was the first to develop a seamless latex girdle having such features of thickness, tensile strength, shape and size that the garment would produce the desired shaping effect on the wearer’s body although, or rather because, the garment was not tailored to the particular form of the body. None of the patents cited by the examiner or relied on by defendant had anticipated this. Charnaux had taught in 1929 that perforating rubber garments would enhance the advantages of rubber for constricting purposes by overcoming the hindrance to cutaneous evaporation presented by smooth unperforated rubber and that varying the size and arrangement of the holes would permit variation in the forces. But this last was the precise opposite of Spanel’s concepts of uniformity of force and avoidance of “tailoring”; Charnaux said nothing as to the thickness or strength of the material; and his teaching was the construction of various articles “by means of india-rubber in the form of bands or sheets.” Young and Hemm taught only processes for making perforations — not, as was conceded by defendant’s expert, Noble, the art of making a seamless latex girdle. Indeed, Dr. Noble’s book “Latex in Industry,” published in 1936, three years after Young and Hemm’s patent, does not mention the use of latex in making girdles or other foundation garments. Spanel’s baby-pants patents were for a seamless latex garment but constriction was neither intended nor desired. While in this field each case must necessarily depend on its own facts, the patentability of Spanel’s basic concept is supported by such decisions as Kurtz v. Belle Hat Lining Co., 2 Cir., 1922, 280 F. 277 and Franc-Strohmenger & Cowan v. Arthur Siegman, Inc., 2 Cir., 1928, 27 F.2d 785, and is not negated by Deering, Milliken & Co. v. Temp-Resisto Corp., 2 Cir., 1960, 274 F.2d 626, and the cases there cited.

Spanel’s acceptance of a patent limited to claims wherein his invention was coupled with features not themselves patentable does not render the patent invalid but simply restricts the scope of the monopoly. As said in International Cellucotton Products Co. v. Sterilek Co., 2 Cir., 1938, 94 F.2d 10, 12, “He has done nothing which need prevent him from insisting in support of the claim as allowed that his invention was broader than the examiner supposed.” See also Stromberg Motor Devices Co. v. Zenith Carburetor Co., 7 Cir., 1918, 254 F. 68, 77, certiorari denied 1919, 249 U.S. 605, 39 S.Ct. 288, 63 L.Ed. 798; Ethyl Gasoline Corporation v. Coe, 1943, 78 U.S.App.D.C. 233, 139 F.2d 372; Texas Co. v. Globe Oil & Refining Co., D.C.N.D.Ill. 1953, 112 F.Supp. 455, 477-479, affirmed 7 Cir., 1955, 225 F.2d 725.

III. Infringement

(A) We shall first consider plaintiff’s appeal from Judge Anderson’s conclusion that defendant’s girdles that did not contain a crotch-piece did not infringe claims 1, 3 and 6 since the roughening of their inner surface was produced by flocking rather than by abrading the latex. Plaintiff asserts that a proper interpretation of the patent would include roughening by applying a foreign substance as well as by creating changes in the surface of the latex and in the alternative that it is entitled to relief under the dbctrine of equivalents.

Plaintiff’s interpretation of the claims to include flocking is inadmissible. Claim 1 refers to a girdle made of “deposited latex sheet material * * * the external surfaces of said sheet material being smooth and the inside surfaces thereof being of a slightly roughened matted character.” It is the inside surface of the deposited latex which is *563 to be roughened; nothing suggests the roughened surface is to be part latex and part something else. When claim 2 refers to “the internal surfaces of said sheet material being of a matted character” and claim 6 to “the inside surfaces of the latex being of a slightly roughened matted character,” the meaning is the same. 2 3 Moreover, as said in Carl Braun, Inc. v. Kendall-Lamar Corp., 2 Cir., 1941, 116 F.2d 663, 665, “Infringement is not proved simply by the language used in a claim without regard to the specifications. It must be proved by showing that the claim covers what is alleged to infringe when the claim is read upon the specifications which describe the invention.” Here the specifications state that “according to the invention, the inside surface of the latex walls of the girdle preferably has a less smooth texture than the outside, which texture, being of a more ‘matted’ character, tends to produce a better feel to the skin and better moisture absorbent qualities,” and, even more significantly, that “This roughened inner surface of the girdle may be produced by roughening the surface of the form on which the girdle is generated. The form may be either sand-blasted, etched, or given a rough surface in any suitable manner; e. g., it might be made of unglazed porcelain.” There was evidence that in May and June, 1940, plaintiff was experimenting with sand-blasting; nowhere, in the 114 pages of the file, would a searcher encounter a single reference to flocking. Plaintiff can therefore prevail as to infringement on claims 1, 2 and 6 only if this is a proper case for applying the doctrine of equivalents.

Judge Anderson found, D.C., 169 F.Supp. at page 763, that “even if the doctrine were not ruled out as a matter of law, ‘flocking’ as used in the accused girdles is not in fact the equivalent of the roughened or matted internal latex surfaces of the girdles as described in the patent claims.” Plaintiff attacks this finding with force. Admittedly the two processes serve the same objectives —promoting evaporation, lessening too close a sticking to the skin, and giving a softer and more comfortable feel. The results are sufficiently indistinguishable that defendant’s expert, who was called to demonstrate the difference between them, mistakenly identified a flocked piece of latex as sandblasted and was uncertain whether a sandblasted piece was sandblasted or flocked. And neither the superiority of flocking nor its being more expensive, in an unstated degree, would require a determination that it passed the bounds of equivalence in a proper case. Johns-Manville Corporation v. National Tank Seal Co., 10 Cir., 1931, 49 F.2d 142, 146, certiorari denied 1931, 284 U.S. 654, 52 S.Ct. 33, 76 L.Ed. 555. We should thus have difficulty in saying that under no circumstances could flocking foe deemed an equivalent for abrading the inner surface of latex; however, that issue is an unreal one, arising out of the attempted dichotomy of “equivalence” as between fact and law. The only question of interest to the parties is whether flocking is to be treated as an equivalent under the circumstances here. We think not.

This Court has characterized the doctrine of equivalents as a principle, peculiar to patents, whereby “after all aids to interpretation have been exhausted and the scope of the claims has been enlarged as far as the words can be stretched, on proper occasions courts *564 make them cover more than their meaning will bear.” Royal Typewriter Co. v. Remington Rand, Inc., 2 Cir., 1948, 168 F.2d 691, 692. Judge Hand there emphasized the necessarily imprecise nature of the doctrine — “All patents are entitled to its benefit to an extent, measured on the. one hand by their contribution to the art, and on the other by the degree to which it is necessary to depart from the meaning to reach a just result.” His own opinions have shown how small the benefit appropriate in a particular case may be. In Deitel v. Unique Specialty Corp., 2 Cir., 1931, 54 F.2d 359, 360, he refused to apply the doctrine to a patent for a vanity case in the form of a book where the accused article was within the claims save only that the channel members clamping the facing and bottom covers to the frame were integral rather than separate, although “Ordinarily we should not insist upon so slight a difference.” This refusal was because “the invention is at best extremely narrow, and we can see no ground for resorting to the doctrine of equivalents when the claims speak plainly and the invention is a trifling step forward.” We think a similar result ought follow when, as here, although the invention may have been broader, the applicant drastically narrowed his claims to overcome difficulties in the patent office. Indeed, Judge Hand’s opinion with respect to claim 19 in International Cellucotton Products Co. v. Sterilek Co., supra, 94 F.2d at pages 11-12, is at least' a sign-post along that road. Plaintiff is right in asserting that when it asks us to apply the doctrine of equivalents to include flocking within claims 1, 2 and 6, it is not seeking to produce the result, forbidden by the cases cited below in our discussion of defendant’s appeal, of recovering ground that Spanel had ceded in the patent office — as would be the case if it were contending that a smooth inner surface girdle was the equivalent of a roughened one. Nonetheless we think it would be incongruous to apply an equitable doctrine, which is “in misericordiam to relieve those who have failed to express their complete meaning,” Claude Neon Lights v. E. Machlett & Son, 2 Cir., 1929, 36 F.2d 574, 576, in such a way as to grant to a patentee, who has deliberately accepted a patent limited to a single application of his invention, protection against a defendant who had made a different application— particularly when neither the claims nor even the specifications contain language general enough to comprehend what defendant has done.

We do not read Graver Tank & Manufacturing Co. v. Linde Air Products Co., 1950, 339 U.S. 605, 70 S.Ct. 854, 94 L.Ed. 1097, as leading to a. contrary conclusion. There the patentees had invented an entirely new type of flux for use in electric welding. The Court applied the doctrine of equivalents to a flux that contained some of the elements included in the claims along with one substance which was not included in the claims held valid, although, as appears from the opinion of the District Court, D.C. N.D.Ind.1947, 86 F.Supp.- 191, it had been disclosed in the specifications and was embraced in a broad claim granted by the patent office but ultimately declared invalid as including inoperative elements, 1949, 336 U.S. 271, 276-277, 69 S.Ct. 535, 93 L.Ed. 672. See Note, The Doctrine of Equivalents Revalued, 19 Geo.Wash.L.Rev. 491, 502, footnote 49 (1951). 3 The facts here are quite different.

(B) We come now to defendant’s appeal from the district court’s decision that its panty girdle infringed claim 3. The final clause in that claim refers to a “crotch area being shaped and of a length to extend down substantially *565 below the lower margin of the back portion when the girdle is in a flat position.” The crotch piece in the defendant’s girdle is not of that nature, as the District Judge found, D.C., 169 F.Supp. 760; defendant’s girdle extends further down over the buttocks, thereby giving, as defendant claimed, greater control, and provides the necessary crotch length by a higher curvilinear cut for the front leg openings. The District Judge nevertheless concluded there was infringement by applying, in this instance, the doctrine of equivalents which he had rejected as to claims 1, 2 and 6.

If the original and final wording of claim 3 had been the same, we would not disagree, for the difference here is surely minor. However, we must look at the patent office history and, when we do, we find that even so modest an expansion of the claim by the doctrine of equivalents is precluded by “file-wrapper estoppel.” The description of the crotch in allowed claim 3 (originally claim 53) differs from that in rejected claim 45 only in the quoted reference, in the last clause of claim 3, to the crotch area extending below the lower margin of the back. This was a feature not contained in the Barnes patent the examiner had cited and one which the final letter of applicant’s patent solicitors urged “particularly in its relationship to the other features is clearly novel to the art and important to the success of the invention.” It is settled that the rejection of a claim forbids “any interpretation of those secured which leaves them identical with that rejected” or any use of the doctrine of equivalents to that end. Deitel v. Unique Specialty Corp., supra; I. T. S. Rubber Co. v. Essex Rubber Co., 1926, 272 U.S. 429, 443-444, 47 S.Ct. 136, 71 L.Ed. 335; Smith v. Magic City Kennel Club, 1931, 282 U.S. 784, 790, 51 S.Ct. 291, 75 L.Ed. 707; Moon v. Cabot Shops, Inc., 9 Cir., 1959, 270 F.2d 539, certiorari denied 1960, 80 S.Ct. 596.

Plaintiff’s counsel, in his effective argument before us, contended the decision ef the District Court had left Spanel with the glory of his invention but, so far as concerns defendant, with little else; ours leaves him with even less. Plaintiff’s disappointment is understandable. But the result stems from Spanel’s decision not to appeal the examiner’s rejection of the claims embodying his basic invention but rather to accept a patent with exceedingly narrow and restricted claims. It is not appropriate for us either to speculate as to the causes for his choice or to relieve him from its consequences.

The judgment of the District Court dismissing so much of the complaint as relates to claims 1, 2 and 6 is affirmed; the judgment in favor of plaintiff in respect of claim 3 is reversed with directions to dismiss the complaint.

Notes

. “1. An integral one-piece supporting girdle formed of deposited latex sheet material of a strength and shape throughout to apply constrictive force to the flesh around the lower torso of the wearer, the external surfaces of said sheet material being smooth and the inside surfaces thereof being of a slightly roughened matted character for contact with the skin of the wearer.

“2. An integral one-piece supporting girdle formed of deposited latex sheet material of a strength and shape throughout to apply constrictive force to the flesh around the lower torso of the wearer, the internal surfaces of said sheet material being of a matted character and such as to provide in contact with the skin an interconnected cellular formation.

“3. As an article of manufacture, a one-piece deposited latex supporting girdle formed of constricting walls, compris *561 ing a constricting front portion and a constricting back portion each formed as normally flat areas the sides of which substantially conform to the outlines of the lower female torso as viewed from the front or back, and narrow areas extending along each side of the girdle, said latter areas being formed with a smoothly curved outwardly convex substantially uniform cross section and serving to integrally join the corresponding sides of said front and back portions and smoothly merging therewith, a readily stretchable narrow crotch area of the deposited latex, extending as a continuous and integral connection between the lower central regions of said front and back portions, the edges of the front of said crotch area extending upwardly and gradually outwardly to merge with and form the lower edges of the middle front portion of the girdle from whence said edges continue outwardly and downwardly to the region of said side areas of the girdle, thereby forming curvilinear cut-out areas at each side of the lower middle front portion, the upper margins of said cut-out areas substantially conforming with the line of juncture o'f the upper thighs and body of the wearer and said cut-out areas also increasing the effective length of said stretchable crotch area, said crotch area also being shaped and of a length to extend down substantially below the lower margin of the back portion when the girdle is in flat position.

“0. An integral one-piece supporting girdle formed of deposited latex of a strength and shape throughout to apply generally uniform constrictive force to the flesh around the lower torso of the wearer, the front and back areas being formed with ‘pin-point’ perforations distributed thereover, and the inside surfaces of the latex being o'f a slightly roughened matted character.”

. Judge Anderson was riglit in holding, D. G., 169 F.Supp. at 765, that the references to perforation in Claim 6 did not placo it on any bettor basis, with respect to infringement, than Claims 3. and 2. For, as stated in the letter of Spanel’s solicitors to the patent office dated November 5, 1943, “This claim defines a novel combination of features * * * viz., the ‘pin point’ perforation feature in conjunction with the feature of forming the inside surfaces of the latex so as to bo of a slightly roughened or matted character.” See, e.g., Sears, Roebuck & Co. v. Minnesota Mining & Mfg. Co., 4 Cir., 1957, 243 F.2d 336, 140. The examiner’s repeated references to Charnaux and Work had made it abundantly clear that ho would not allow a claim for I>erforation alone.

. The same note concludes (p. 497) that where the hardship to the patentee relied on as a basis for invoking the doctrine of equivalents stems from limitation of claims as a result of prosecution before the patent office, the case “is not such as to warrant aid from the doctrine but instead is more properly viewed as a condition resulting from the patentee’s own mistakes of judgment * * * even though the extent of financial loss to the patentee may be extreme.”