RICH, Circuit Judge.

These appeals are from a May 19, 1988, judgment of the United States District Court for the Eastern District of Virginia, T.S. Ellis, III, Judge. 688 F.Supp. 206, 8 USPQ2d 1897. Plaintiff Environmental Instruments, Inc. (Eli) brought suit alleging that defendant The Sutron Corporation (Su-tron) had infringed the sole claim of Ell’s U.S. Patent No. 3,900,819 (’819) on a “Thermal Directional Fluid Flow Transducer.” Eli also alleged infringement of its U.S. Patent No. 3,995,481 (’481), entitled “Directional Fluid Flow Transducer,” but dropped that claim for infringement prior to trial. Sutron counterclaimed for, inter alia, a declaration that both patents were invalid and unenforceable.

The district court held that: (1) the ’819 patent had not been proved invalid, but was not infringed; (2) the ’481 patent is invalid for obviousness under 35 U.S.C. § 103; and (3) Sutron failed to prove its charges of inequitable conduct. We vacate that portion of the judgment addressed to the validity of the ’819 patent and otherwise affirm.

I. BACKGROUND

The patents in suit relate to the use of electrically heated films of platinum or other resistive metals to measure fluid flow, particularly wind speed and direction, or “hot film” anemometry. The ’819 claim, quoted infra, is lengthy; we briefly describe the invention for background. Ell’s commercial version of the ’819 invention comprises a pair of “sensing elements:” two ceramic tubes filmed with platinum, arrayed side-by-side with their major axes parallel. An insulator between the elements closes the space between them, and the overall cross-sectional configuration of the sensor or probe is defined in the ’819 claim to be a “figure eight.” Below are perspective and cross-sectional figures from the specification showing the sensing elements 10a and 10b connected by insulator 25.

*1563 [[Image here]]

In operation, the two elements are connected in series. A feedback-controlled heating current is passed through the elements to heat them to an elevated operating temperature, which corresponds to a constant total series resistance across the two elements. Since the elements are at a higher temperature than the surrounding fluid (air), there is a transfer of energy from the elements to the fluid in the form of heat. The energy needed to maintain the elements at the chosen operating temperature is measured as the total difference in potential across the two elements. This difference in potential is a function of, and may be translated into, wind speed. Wind direction may also be determined, simultaneously, by measuring the potential difference across each individual element and comparing the two differences in potential. The sensing element facing the fluid stream, or leading element, is stream-cooled to a greater degree than the sensing element at the trailing edge of the device. Thus, the difference in potential across the leading element will be different from the difference in potential across the trailing element. These two differences may be compared and the resulting value is indicative of direction flow, i.e., wind direction. It appears that a distinct advantage of the sensor configuration claimed in the ’819 patent is its ability to measure wind direction more satisfactorily than previously known sensors. The claimed sensors have no moving parts and have found their main use as cross-wind sensors in battle tank fire control systems.

[[Image here]]

The invention of the ’481 patent differs from that of the ’819 patent in that the ’481 patent claims a sensor in broader terms and includes electronic circuitry in which the sensing elements are connected in series and operated as one arm of a single Wheatstone bridge.

Ell’s dispute with Sutron arose after Su-tron was awarded a contract to supply 1,000 air monitoring systems using wind sensors to the United States Air Force. Sutron had specified in its bid that it would supply Eli sensors to the Air Force. According to Eli, Sutron at some point purchased an Eli sensor and sent it to a Swiss competitor of Eli, Cossonay, for comparative testing. The testing purportedly showed that Cossonay’s sensor was superi- or, and Sutron then selected Cossonay to supply the sensors for the Air Force contract. It is undisputed that Sutron purchased approximately sixty wind-sensing probe heads from Cossonay. This purchase prompted Ell’s complaint for patent infringement, filed June 11, 1987.

II. DISTRICT COURT’S OPINION

The district court’s memorandum opinion, 688 F.Supp 206, 8 USPQ2d 1897, records the court’s findings of fact and conclusions of law in narrative form.

III. OPINION

A. Infringement Of The ’819 Patent

1. No Literal Infringement

We need not elaborate details of Ell’s literal infringement argument because it is plainly unsound. Eli asks us to *1564 decide that “figure eight” means something other than “figure eight.” That we will not do. We agree with the trial judge that the accused oval- or racetrack-configured sensors do not fall within the first “figure eight” limitation of the claim. Furthermore, claim 1 (see infra) defines “the conductors” as “being exposed to ventilation over at least a majority of their surface.” These words have meaning and must be given effect. Pennwalt Corp. v. Durand-Wayland, Inc., 833 F.2d 931, 935, 4 USPQ2d 1737, 1739-40 (Fed.Cir.1987), cert. denied, — U.S. -, 108 S.Ct. 1226, 1474, 99 L.Ed.2d 426 (1988). The district court made a finding, not shown to be clearly erroneous, that the conductors of Sutron’s accused sensors “are not exposed to ventilation over a majority of their surface.” The import of this finding is that there can be no literal infringement. Claim 1 does not read on Sutron’s sensors because the conductors in them are not “exposed to ventilation over at least a majority of their surface,” as the claim requires.

As to Ell’s suggestions that the district court should have found literal infringement because the accused sensor employs the “principal teachings of the ’819 patent” and is with one exception virtually “a Chinese copy of Figure 13 of the ’819 patent,” we think the district court prudently rejected any such theories. The disclosure of a patent is in the public domain save as the claims forbid. The claims alone delimit the right to exclude; only they may be infringed. SRI International v. Matsushita Electric Corporation of America, 775 F.2d 1107, 1120-22, 227 USPQ 577, 585-86 (Fed.Cir.1985).

2. Infringement By Equivalents

Ell’s argument that the Sutron sensors infringe under the doctrine of equivalents requires more extended discussion.

Claim 1 as originally presented for examination clearly read on Sutron’s sensors. The examiner, however, rejected the claim under 35 U.S.C. § 102 in view of the Haya-kawa patent alone, and under 35 U.S.C. § 103 in view of the same patent in combination with others. Hayakawa discloses a “Tandem-Type Hot-Wire Velocity Meter Probe” for measuring “flow velocity of liquids and/or gases,” comprising two “hot-wire members disposed in tandem at upstream and downstream ends of an insulating support member.” Cross-sectional views of five embodiments of Hayakawa’s improved probe are reproduced below, the main body 2 being the insulating support member and the “hot wires” being mounted along its sides at la and lb, which are embedded wires and at 2e, 2f, 2g, and 2h, which are films.

[[Image here]]

Eli amended its claim to avoid Hayaka-wa. As once amended, the clause at issue is set forth in the margin, additions underscored, deletions in brackets. 1 Despite this prolix amendment the examiner adhered to his rejections, added a § 112 rejection for indefiniteness, and made them final. As to the § 112 rejection, the examiner’s com *1565 plaint was that the term “major shape” in the claim as once amended lacked antecedent basis and what it “really define[d]” was “not clear.” Eli then amended the claim a second time. Twice amended, it then read in its entirety:

1. A directional fluid flow sensor comprising:

(a) at least two similar thermally and physically separated resistive electrical conductors;

(b) each of said conductors having a length at least equal to the largest cross section dimension of the conductor;

(c) each of said electrical conductors including a hollow electrically non-conductive refractory cylindrical substrate supporting body extending the length of the conductor, and a conductive resistance film having a non-zero temperature coefficient adhered to the outer surface of the substrate body and extended over the length of the substrate body;

[(c)] (d) a thermo insulating bridging means operatively disposed between, and closing the gap between said electrical conductors over the length of the conductors, thereby preventing connected flow around one conductor independent of the other conductor, the [major] overall shape of the sensor being figure eight in cross section, [defined by the conductors and not by the bridging means,] with [each] the conductors being [conductor] exposed to ventilation over at least a majority of [its] their surface, and with the conductor pair cross section itself used to define the fluid dynamic cross section of a figure eight cross section which is exposed to the fluid stream where the resulting local stagnation region caused by impinging flow at its point of separation against an electrical conductor is therefore small with respect to the conductor cross section, each conductor exhibiting a change in electrical resistivity as a function of temperature; and,

(e) each of said electrical conductors being provided with electrical connection means, whereby each electrical conductor can be electrically heated by an electrical current passing through each conductor.

Eli argued in accompanying remarks that the amendment to the newly designated clause (d) overcame the § 112 rejection. Eli also argued that this amendment overcame the § 102 rejection on Hayakawa:

Claim 1 furthermore defines over the Hayakawa structure by defining that the insulating bridge means closes the gap between applicant’s conductors so as to provide the overall figure-eight in cross section shape of the sensor, and such structure is not taught or suggested by Hayakawa whose overall structure is determined by his support member 2 in Fig. 1(A) through 1(E).

The claim as amended satisfied the examiner, he withdrew his rejections, and allowed the claim.

The district court concluded that in view of the prosecution history, claim 1 should be construed “to encompass only the figure eight cross-sectional configuration.” The court then held that there was no infringement by equivalents because the Sutron sensors have a “racetrack,” or oval configuration, not a figure eight configuration. The district judge drew a picture of his concept of the accused Sutron sensor in cross-section in his published opinion, 688 F.Supp. at 212, 8 USPQ2d at 1901, q.v.

We think the district court’s sole error in discussing infringement was its statement, with reference to Graver Tank & Manufacturing Co. v. Linde Air Products Co., 339 U.S. 605, 70 S.Ct. 854, 94 L.Ed. 1097 (1950), that the doctrine of equivalents “is not designed to permit a patent to exceed the scope of its claims.” The Supreme Court said a number of things in Graver Tank, but that was not one of them. The essence of the doctrine of equivalents is that it permits recovery for infringement where the accused device does not fall within the literal scope of the claims and is, therefore, outside their literal scope.

As to result, we think the district court properly estopped Eli from asserting that *1566 the accused sensors infringe under the doctrine of equivalents. We have made, in accordance with our precedent, e.g., Mannesmann Demag Corp. v. Engineered Metal Products Co., 793 F.2d 1279, 1284-85, 230 USPQ 45, 48 (Fed.Cir.1986), an inquiry as to what was surrendered during prosecution and why. Eli made two significant changes in the claim in the second amendment. First, Eli replaced the phrase “the major shape of the sensor being defined by the conductors and not by the bridging means,” with the phrase “the overall shape of the sensor being figure eight in cross section.” Second, Eli emphasized that the conductor pair cross section itself defines the fluid dynamic cross section “of a figure eight cross section” by the insertion of the quoted phrase. When Eli added these figure eight limitations it argued that they overcame the § 102 and § 112 rejections. To overcome the former would be a predicate to overcoming the § 103 rejection, so the figure eight limitations were asserted to overcome that rejection too, albeit indirectly. Thus, the reasons for the amendment were to overcome: (1) the § 102 rejection; (2) the § 103 rejection, and (3) the § 112 rejection. The first two reasons, the amendment itself, and the examiner’s allowance of the claim, convince us that the figure eight limitations fall on the highly limiting end of the “spectrum” of “limiting effect.” See Hughes Aircraft Co. v. United States, 717 F.2d 1351, 1363, 219 USPQ 473, 481 (Fed.Cir.1983). That a second, concurrent amendment was made, and that the figure eight limitations made the claim more definite, do not alter our conclusion. On these facts, we think the district judge was right to limit Eli to a reading of “figure eight” that does not cover the accused sensor having the “racetrack” or oval configuration. Cf. Exhibit Supply Co. v. Ace Patents Corp., 315 U.S. 126, 135-37, 62 S.Ct. 513, 518-19, 86 L.Ed. 736, 52 USPQ 275, 279-80 (1942); Prodyne Enterprises, Inc. v. Julie Pomerantz, Inc., 743 F.2d 1581, 1583, 223 USPQ 477, 478 (Fed.Cir.1984). See also LaBounty Mfg., Inc. v. United States Int’l Trade Comm’n, 867 F.2d 1572, 1575-76, 9 USPQ2d 1995, 1998-99 (Fed.Cir.1989), and Loctite Corp. v. Ultraseal Ltd., 781 F.2d 861, 870, 228 USPQ 90, 96, note 7 (Fed.Cir.1985). We therefore affirm the finding of no infringement by application of the doctrine of equivalents.

B. Validity Of The ’819 Patent

Our affirmance of the district court’s decision that the sole claim of the ’819 patent is not infringed resolves the controversy as to that patent. Accordingly, we need not pass on its validity. The judgment that Sutron has not proven claim 1 of the '819 patent invalid is vacated and the appeal from that judgment is dismissed as moot. Fonar Corp. v. Johnson & Johnson, 821 F.2d 627, 634, 3 USPQ2d 1109, 1114 (Fed.Cir.1987), cert. denied, — U.S. -, 108 S.Ct. 751, 98 L.Ed.2d 764 (1988).

C. Validity of the ’481 Patent

1. Introduction

On the eve of trial, Eli announced it would not pursue its claim for infringement of the ’481 patent. Eli, however, did not move to dismiss its claim for infringement of this patent. A dismissal at that juncture would have required an order of the court, see Fed.R.Civ.P. 41(a)(2), and no such order was entered. Ell’s claim for infringement of the ’481 patent therefore remained in the case. We conclude there was a continuing, actual controversy between the parties with respect to both validity and infringement of the '481 patent. Accordingly, the district judge did not abuse his discretion in deciding its validity and in entering judgment on Sutron’s counterclaim for a declaration of invalidity.

2. Obviousness

The judge decided that the invention of the ’481 patent would have been obvious in view of a July 1950 paper, Design And Applications Of Hot-Wire Anemometers For Steady-State Measurements At Transonic And Supersonic Airspeeds, by Herman H. Lowell.

It appears that Lowell’s paper is well known in this field. The patent examiner, however, did not refer to it. We accept the *1567 judge’s findings concerning what Lowell’s paper discloses. We quote these relevant passages from Lowell:

V-array. — A V-combination consists of two wires (of as nearly equal dimensions as practicable) generally intersecting at an angle of between 30° and 90°.

* * * * * *

A V-array may be used as a single wire by placing the two wires in series; the pair comprise one arm of a bridge. So connected, the array is well adapted to the making of mass-flow measurements. Alternately, the two wires comprise two adjacent arms of a bridge....

$ $ ‡ $ sfc sjc

When connected in series as for mass-flow determination, V-array behavior, as far as the flow component in the array plane is concerned, is nearly the same as that of a single wire running through the center of gravity of the array parallel to the base of the array triangle.

♦ ♦♦♦♦♦

[T]he principal direction of a V-array is the vertex-angle bisector. In general, temperature and mass-flow rate measurements are made when the angle bisector and flow vector have been made to coincide as nearly as is practicable in the given situation.

Lowell thus discloses, as the judge found, what Lowell calls a “V-array” for making mass flow measurements. The two wires or legs of the V may be connected in series and operated as one arm of a Wheatstone bridge.

The judge made a side by side comparison of the elements of claim 1 of the ’481 patent to what is disclosed in Lowell, but did not make express findings whether or not each element of the claim is identically disclosed. On review, we think there can be no dispute that what Lowell discloses differs in two significant respects from the ’481 invention. First, the legs of Lowell’s V are not a pair of “closely spaced” elements “extending side by side along the sensor’s longitudinal axis,” as called for in claims 1 and 2. 2 The longitudinal axis of the V-array is the bisector of the angle defined by the legs of the V. The legs intersect on that axis at the V’s vertex and then diverge from it and each other; they do not “extend side by side” along it. As to Lowell’s statement that his V-array is equivalent to a single-wire sensing element “running through the center of gravity of the array parallel to the base of the array triangle,” that also differs from the claimed pair of sensing elements. The second notable difference from what is claimed is that Lowell does not provide any means to prevent fluid flow between the legs of the V-array. The V-arrays are naked wires in free space. In contrast, what is claimed is a paired-element sensor “having an electrically non-conductive thermal insulator disposed between the two elements which prevents fluid flow there-between.” (Our emphasis).

In light of these differences, we fail to see how the invention of the ’481 patent would have been obvious to one of ordinary skill in view of Lowell alone. Elsewhere in his paper, Lowell does disclose an array comprising a parallel pair of wires extending side by side, but, in contrast to the V-array, he does not suggest connecting the parallel wires in series to serve as a *1568 single arm of a Wheatstone bridge. Nor does Lowell suggest using an insulator to prevent fluid flow between his bare wires. Since Lowell fails to suggest the desirability of the claimed arrangement of sensing elements and insulator, Lowell necessarily fails to suggest the desirability, and thus the obviousness, of the invention as a whole. 35 USC 103. Accordingly, we think the judge was wrong to decide that the invention of the ’481 patent would have been obvious in view of Lowell alone.

Notwithstanding this error, we will sustain the judgment on alternative grounds. Sutron asserts, in an argument improperly styled as part of its cross-appeal, that Hayakawa (1) discloses the elements of the ’481 invention not found in Lowell, and (2) suggests their incorporation into Lowell’s arrangement. We agree.

As the judge found, Lowell characterizes his V-array as equivalent to a single wire. Lowell states: “A V-array may be used as a single wire by placing the two wires in series; the pair comprise one arm of a bridge.”

Hayakawa, like Lowell, recognizes the use of single “hot-wire” sensors as anemometers. Hayakawa, however, rejects the use of such single-element sensors in favor of his improved device comprising two parallel sensing elements, either wires or films, separated by an insulator. This is, no doubt, why Hayakawa was the principal reference during prosecution; substitution of a dual-element device for a single element is the essence of his disclosure. After detailing the drawbacks of anemometers using a single hot wire, Hayakawa states:

The most characteristic feature of the tandem-type hot-wire velocity meter probe of the invention resides, of course, in the use of two hot-wire members in tandem arrangement, i.e., at upstream and downstream ends of an approximately flat shaped support member made of an electrically and thermally insulating material.

Hayakawa plainly suggests to one skilled in the art that Lowell’s V-array may be replaced with what claim 1 of the '481 patent recites as “a dual element sensor” having a pair of “closely spaced” elements, “extending side by side along the sensor's longitudinal axis and having an electrically non-conductive thermal insulator disposed between the two elements which prevents fluid flow therebetween.” We therefore conclude that claim 1 of the ’481 patent would have been obvious to one of ordinary skill in view of the combination of Lowell and Hayakawa and sustain the judgment on that basis. Dependent claim 2 of the ’481 patent, whose patentability has not been argued separately, falls with claim 1. See Gardner v. TEC Systems, Inc., 725 F.2d 1338, 1350, 220 USPQ 777, 786 (Fed. Cir.), cert. denied, 469 U.S. 830, 105 S.Ct. 116, 83 L.Ed.2d 60 (1984).

3. Inequitable Conduct And Attorney Fees

We reject as meritless Sutron’s argument that the patents in suit were procured through inequitable conduct, as well as its request for attorney fees.

IV. CONCLUSION

The judgment that Sutron failed to prove the ’819 patent invalid is vacated. In all other respects, the judgment is affirmed.

Costs

Sutron shall have its costs.

AFFIRMED-IN-PART, VACATED-IN-PART.

Notes

. (c) a thermo insulating bridging means opera-tively disposed between, and closing the gap between [connecting] said electrical conductors over [a majority of] the length of the conductors, thereby preventing connected flow around one conductor independent of the other conductor, the major shape of the sensor being defined by the conductors and not by the bridging means, with each conductor exposed to ventilation over at least a majority of its surface, and with the conductor pair cross section itself used to define the fluid dynamic cross section which is exposed to the fluid stream where the resulting local stagnation region caused by impinging flow at its point of separation against an electric conductor is therefore small with respect to the conductor cross section, each conductor exhibiting a change in electrical resistivity as a function of temperature.

. Claim 1, the only independent claim, reads:

1. Apparatus for measuring fluid flow comprising

a. a dual element sensor having a pair of closely spaced, elongate, temperature sensitive elements whose electrical resistance is a function of temperature, the pair of elements extending side by side along the sensor’s longitudinal axis and having an electrically non-conductive thermal insulator disposed between the two elements which prevents fluid flow therebetween,

b. means connecting the two elements of the dual element sensor in series,

c. means connected to the dual element sensor and forming a bridge circuit with the two series connected elements of the sensor being in one arm of the bridge,

d. means for causing a current to flow through the two series connected elements to cause them to be heated above the ambient temperature, and

e. regulating means connected to the bridge circuit for regulating the current to cause the total resistance of the series connected elements to be held constant for constant ambient temperature.