WIDENER, Circuit Judge:

This patent case concerns an automatic reset feature, a device to prevent information loss in a magnetic data storage system. Technitrol, Inc., the patent owner, brought an infringement suit for an injunction and damages against Control Data Corp., a manufacturer, seller, and user of such systems. Control Data filed a counterclaim asking the court to declare Technitrol’s patent invalid and moved for summary judgment for failing to comply with 35 U.S.C. § 112, para. 2, which reads in pertinent part:

“The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention. ...”

Looking to an affidavit by an expert filed by Control Data, the record of an allied case in the Court of Claims, and the language of the patent itself, the district court held claims 1-15 and 17-24 invalid as not “particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention,” 394 F.Supp. at 520, relying on paragraph 2 of § 112 and rejecting the defendant’s position that the claims are adequate under paragraph 3 of § 112. Sua sponte, the court also granted summary judgment for the defendant on the remaining claim 16 based largely on its holding as to the other claims. Technitrol now appeals.

We are of opinion that the claims adequately describe the invention as we consider them here. We therefore vacate the district court’s grant of summary judgment and remand for further proceedings.

Technitrol’s patent was originally issued to T. K. Sharpless and E. S. Eichert in 1952 as No. 2,611,813. By assignment, Techni-trol is, and at all pertinent times has been, the owner of the patent.

The patent describes a computer system used, for example, in making airline reservations, composed of a central storage unit where information about various flights is stored, and a number of remote stations, such as airport reservation desks, which communicate electrically with the central unit. An operator, by punching certain keys on the keyboard at the remote station, can ask the central storage unit whether seats are available on a particular flight. If seats are available, the invention can make reservations on flights, or it can cancel them, or it will indicate if a flight is full.

The remote stations communicate with the central storage unit through electrical transmission lines. Each remote station has keyboards by which an operator can designate a particular flight about which he desires information and the number of seat reservations needed.

The keys, through appropriate circuitry, indicate to the central station the register which contains information about the desired flight. After setting the keyboard, the operator presses a start switch which operates a selector circuit to insure that only one keyboard at a time can communicate with the central station. This avoids the possibility of simultaneous duplicate requests to a register from more than one remote station.

To make reservations, the operator communicates the number and flight to the central storage unit. There, the total number of seats and confirmed reservations of each particular flight is stored on a corresponding register. When the operator selects the appropriate flight number, the machine selects the corresponding register and adds the number of reservations desired to the previous total. If the sum of these is less than the total of reservations available, the reservation request is confirmed, the number of reservations requested is added to the previous total, and the new total is recorded in the storage register where it is available for similar access in the future. *995 But if the sum exceeds the total of reservations available, the application is rejected, the inquirer is so notified, and the old number remains unchanged.

The central storage unit is a magnetic memory device consisting of several continuously revolving magnetic disks (information disks 2, 3, 4, 5) 1 mounted on a common rotating shaft. An electric motor drives the shaft. Information is stored on the disks in magnetic pulses, which, when arranged in groups around the disk, are called registers, one group being one register. Each register contains information about a particular airline flight such as the number of seats already reserved and the number still available.

Magnetic recording heads (C, D, E, F) are mounted adjacent to the disks. These either read what is on the disk, add to it, or erase from it. Thus, as the disks rotate on the shaft, the recording heads, through appropriate circuitry, can read or erase the magnetic pulses already recorded, or can place new pulses on the disks.

Also located on the common shaft with ¡ the information disks is a master clock disk ' (1). Around the clock disk, two channels of magnetic pulses are recorded. One channel has 160 evenly-spaced pulses; the other, one pulse. Magnetic pickup heads (A, B) are located over each channel. The 160-pulse head is connected to a scale-of-ten counter (b) which divides the information from the disk into sixteen 10-pulse registers. Pulses from the scale-of-ten counter, when applied to a binary counter device (c), produce 16 unique voltage combinations, each one of which corresponds to one of the registers on the information disk. The unique voltage combinations (originating at the various registers) are successively fed into a coincidence circuit (d).

When the remote station operator pushes the key to select a particular register, the key generates a voltage combination corresponding to one of the unique voltage combinations originating on the rotating disk which is also fed into the coincidence circuit (Vi, V2, V8, V4). When the disk rotates to the particular register so that its voltage combination matches the one generated by the operator, the coincidence circuit detects the coincidence and produces an output activating the magnetic recording heads (C, D, E, F) which read, write, or erase.

There is general agreement here (not later binding) that a previous magnetic storage system or systems accurately located the registers as long as operation continued without interruption. When interrupted, however, as by a power failure or merely cutting the control station off, position volatility occurred and the information was lost. In essence, this means that the clock disk and the counters, normally in synchronization, lose synchronization when power is cut off. This happens because the counters, electronic devices, stop operating when power stops; and because of rotational inertia, the disks, being mechanical devices, may rotate slightly after the power is cut off, and thus may move out of synchronization with the counter. On restart, the counters, which have lost count of the physical location of the registers, do not pick up the count where they left off with the corresponding register. On start-up, therefore, it is essential that the counters be reset to synchronize them with the registers.

The Sharpless-Eichert invention does this by the one-pulse-per-revolution channel on the clock disk. The single pulse resets the counters to zero, whether after each revolution of the clock disk or only on start-up is not material here. Even if the power goes off momentarily, causing the counters to start up with arbitrary counts, within one revolution the invention will synchronize the counters to the correct count.

In order to simplify this appeal, Techni-trol has limited its argument to representa *996 tive claim 19. 2 It takes the position that the reset feature appears there. Because we think it is for the purpose of this motion, we do not address the arguments regarding the other patent claims. Although Control Data apparently takes the position that claim 19 is not representative, it fails to state why, only offering the conclusion. Since the case will be remanded for further proceedings, the district court may take appropriate action to any extent it may properly determine that a claim is not representative when considered in the light of this opinion.

We emphasize that the case comes to us as an appeal from the grant of summary judgment holding the patent claims invalid. While summary judgment may be proper in patent infringement cases, it is only where, under FRCP 56(c), there is no genuine issue as to any material fact and the moving party is entitled to a judgment as a matter of law. Smith v. General Foundry Machine Co., 174 F.2d 147, 151 (4th Cir. 1949), cert. den. 338 U.S. 869 (1949). It should be employed with great caution, Morpul, Inc. v. Glen Raven, 357 F.2d 732 (4th Cir. 1966), and is not ordinarily appropriate for the disposition of a patent case. Long v. Arkansas Foundry Co., 247 F.2d 366 (8th Cir. 1957). As the moving party, Control Data has the burden of showing the absence of a genuine issue of material fact. Adickes v. Kress & Co., 398 U.S. 144, 157, 90 S.Ct. 1598, 26 L.Ed.2d 142 (1969).

In support of its motion, Control Data submitted an expert’s affidavit which concluded that the claims did not adequately describe the automatic reset feature. The district court referred to the affidavit as uncontroverted. But in our opinion it was not, and in all events had to be construed in the light most favorable to the plaintiff. Adickes, p. 157, 90 S.Ct. 1598. The court also had before it the report of a Court of Claims Commissioner, 3 the opinion *997 of the Court of Claims, and the record in the Court of Claims which included expert testimony. We note the district court in its opinion referred to all of this additional information.

It seems to us that what the district judge did was to adopt the opinion of the expert for Control Data and to disregard the testimony and other proof favoring Technitrol. Since letters patent are contracts, they should be construed with the interest of the parties in mind to give effect to their legitimate expectations. Del Francia v. Stanthony Corp., 278 F.2d 745, 747 (9th Cir. 1960). And, while matters of construction are ordinarily for the court, “[t]he claim of a patent must always be explained by and read in connection with the specification.” American Fruit Growers, Inc. v. Brogdex Co., 283 U.S. 1, 6, 51 S.Ct. 328, 75 L.Ed. 801 (1931). Although in a simple and clear-cut case, it may be proper to disregard one party’s evidence, see Ethyl v. Borden, Inc., 427 F.2d 206,208 (3rd Cir. 1970), where there is conflicting evidence before the court, as here, disposing of the issue on motion for summary judgment is not proper.

It may well be, as Technitrol argues and as the commissioner found in his report to the Court of Claims, that the term “representative” in clause 2 of claim 19 points out the automatic reset feature. Shaw’s testimony before the commissioner can be construed that way when read in the light of the rules of construction we have recited just above. And, perhaps the patentees could have been more thorough in their patent claims. One part of the opinion of the district court suggests this may have been their only error, p. 517. Nevertheless, they have “the right to use such words as to [them] best describe [their] intention, and they will be so construed as to effectuate that result.” Bianchi v. Bari-li, 168 F.2d 793 (9th Cir. 1948) (quoting H. J. Wheeler Salvage Co. v. Rinelli & Guardino, 295 F. 717, 727 (D.C.N.Y.1924). The rule is that patentees are allowed much latitude in terminology, and their language will be accorded the meaning intended if it can be ascertained from the context. Strong-Scott Mfg. Co. v. Weeler, 112 F.2d 389 (8th Cir. 1940) (citing Smith v. Goodyear Dental Vul-canite Co., 93 U.S. 486, 494-95, 23 L.Ed. 952 (1876)). See Topliff v. Topliff, 145 U.S. 156, 171, 12 S.Ct. 825, 36 L.Ed. 658 (1892).

Everyone admits that without the reset feature the system was position volatile. And the record supports the conclusion that Sharpless and Eichert intended their system to be one unhampered by position volatility. To be so, it was essential that the disks and the counters stay in synchronization. Otherwise-, the counters might not produce signals representative of each register on restart.

The grant of a patent is some evidence that the patented device is operative. Dashiell v. Grosvenor, 162 U.S. 425, 432, 16 S.Ct. 805, 40 L.Ed. 1025 (1895). To be patentable, the device must be useful. 35 U.S.C. § 101. Certainly it would have to work in order to be useful. H. Brinton Co. v. Mishcon, 93 F.2d 445, 448 (2d Cir. 1937).

Without the automatic reset feature to insure that the counters produce signals representative of each register, whether or not following a loss of power, the Sharpless-Eichert patented invention would not work as intended. We are of opinion that the *998 claims at issue should not be construed so rigidly as to defeat the utility of the device as the inventors intended it. We think the use of the term “representative” without qualification could easily have meant that the signals would always represent the proper register. That is how their system worked, and based on this record, that is what we think they intended to claim. 4

In any event, it is an established principle of patent law that when a claim is fairly susceptible of two constructions, one will be adopted which will preserve to the patentee his actual invention. Coupe v. Royer, 155 U.S. 565, 577, 15 S.Ct. 199, 39 L.Ed. 263 (1894). In this case, at least two electronic experts who gave opinions did not agree on the effect of the inventor’s claim descriptions. We think, for the purpose of summary judgment, that this indicates at the least the claim may be susceptible of two constructions.

We are supported in our conclusion by the third paragraph of 35 U.S.C. § 112, which is:

“An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.” 5

We begin with the proposition that the claims measure the invention. Continental Paper Bag Company v. Eastern Paper Bag Company, 210 U.S. 405, 419, 28 S.Ct. 748, 52 L.Ed. 1122 (1908). (Continental Paper Bag is referred to also in footnote 5).

Remembering that the district court held against the patent in question because in its opinion the claims did not particularly point out and distinctly claim the subject matter which the applicant regarded as his invention, 35 U.S.C. § 112, paragraph 2, we also are of opinion that a valid rule of construction with respect to paragraph 2 is correctly expressed in Application of Lundberg, 244 F.2d 543, 44 C.C.P.A. 909 (1957), that the requirements of the second paragraph of § 112 just above mentioned are not diminished by the addition of the third paragraph, also above mentioned.

We think a correct construction of the third paragraph of § 112 has been stated in Application of Knowlton, 481 F.2d 1357 (Cust.Pat.App.1973). There, the paragraph was described as dealing with permissible forms of claiming (italics from Knowlton), and, after reciting some of the language from the statute, the court stated the rule adopted was that “If the applicant chooses to use such language, the statute instructs the interpreter of the claims, e. g., the Patent Office or the courts, as to how such language shall be interpreted.” p. 1366. Thus, the third paragraph is held to deal with permissive claiming; and, if language coming within the statute is used in the claim, directs how the courts shall construe that language. Knowlton also acknowledges that it does not dispense with *999 the definiteness requirement found in the second paragraph of § 112, as we have just above recited from Lundberg.

We also think a correct construction of the third paragraph of § 112 is found in Stearns v. Tinker & Rasor, 252 F.2d 589 (9th Cir. 1957), in which the court stated: “We construe the section to mean that while an element in a claim for a combination may be expressed as a means or step for performing a function without recital of structure, material or acts in support thereof; the structure, material, or acts must be described in the specification, and if so described, the claim will be construed to cover that which is described and the equivalents thereof. But the structure need not as well be recited in the claim.” pp. 597-598.

In the use of means clauses in patent claims, a “means for” clause such as those used here “ . . .in effect calls for structure — more precisely — for apparatus or, indeed, for any physical body or bodies having the capacity to perform the function recited after the words ‘means for’. Such a clause is completely devoid of the details of apparatus capable of satisfying the recitation of function. Indeed, a ‘means clause’ (consisting of the words means for and a statement of the function which such means is supposed to perform) is to be construed as calling for ANY means capable of performing the indicated function.” Rosenberg, Patent Law Fundamentals (1975), p. 48. 6

With these principles in mind, we turn to the question at hand. Technitrol claims the reset feature is disclosed in clauses 1 and 2 of claim 19, which are:

“Claim 19 [repeated for convenience] “In an information storage system,

“1. magnetic recording means having a plurality of information recording sections constituting registers and also having a register-selection section on which are recorded pulses coordinated with said registers,

“2. means for producing from said recorded pulses different successively-occurring register-selection voltage combinations representative respectively of said registers,”

We are unable to ascertain precisely why the district court held that there was insufficient disclosure under paragraph 2 of § 112. It may have been because it thought there was no causal relationship between the specified function and the “apparatus purportedly included” in the means clause. 394 F.Supp. at 516. Or it may have been because of the mere omission of the words “in conjunction with the automatic reset function,” which is more strongly suggested on p. 517. 7

In either event, we think the court was in error.

Both clauses 1 and 2 of claim 19, for example refer to a register-selection section, and clause 2 of the claim provides that the different successively occurring register selection voltage combinations are to be “representative respectively of said registers.” Clause 2 also provides a “means for” producing such “voltage combinations representative respectively of said registers.”

Referring back to the first part of this opinion in which the problem to be solved, as acknowledged by all, was shown to be position volatility, we think the invention would not work in the manner conceived by the inventor if it were position volatile. This is so because the voltage combinations would not be representative respectively of *1000 the registers at each necessary place in the system in an apparatus which was position volatile. The voltage combinations in such a position volatile apparatus might well be not so representative after power failure or cutoff as we have before described.

Considering that clause 2 of claim 19 is in one of the permissible forms of claiming under paragraph 3 of § 112, since it expresses a “means or step for performing a specified function,” we think the specified function is the production and transmission to necessary places in the system of voltage combinations “representative respectively of said registers” and not of something else. That being true, it is not necessary that the claim recite a “structure, material, or acts in support thereof,” for it “shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.” § 112, paragraph 3.

The clock disk and reset feature is mentioned at various places throughout the specifications, but for our present purpose we need mention only one. Column 4, line 33, contains the following language:

“The one pulse per revolution output supplied from the clock disk 1 through amplifier al is used to initially set the counters so that the registers on the disks will always remain the same relation with the pulses on the clock disk as checked by counters b and c, even though the power be shut off and later turned on with the counters coming up containing arbitrary counts.”

Thus, by reference to the specifications in accordance with paragraph 3 of § 112, we think the clock disk and its function, which amounts to the reset feature, is covered as a corresponding structure or act as a part of the claim and helps serve to prevent invalidity for indefiniteness. See opinion of the Patent Commissioner, 164 USPQ at 55-56. We think our construction of the language used is consistent with Continental Paper Bag Company (the claims measure the invention); with Lundberg (we have not diminished the effect of the second paragraph by reference to the third paragraph of § 112); with Knowlton (in permissible claiming, we have construed the language of the claim as directed by the third paragraph of § 112); and with Stearns (the structure is recited in the specifications, although it need not as well be in the claim).

We are therefore of opinion that the district court erroneously construed the claims for the reasons set forth just above, as well as those in Part I of this opinion.

Ill

The order of the district court adjudging claims 1-24 inclusive of the ’813 patent to be invalid must be vacated and the case remanded for further proceedings not inconsistent with this opinion.

VACATED AND REMANDED.

*1001

. Fig. 4 appended hereto is used for numbered or lettered references.

. 19. In an information storage system, magnetic recording means having a plurality of information recording sections constituting registers and also having a register-selection section on which are recorded pulses coordinated with said registers, means for producing from said recorded pulses different successively-occurring register-selection voltage combinations representative respectively of said registers, means under control of an operator at a remote position for producing a group of pulses indicative of a particular register and also containing numerical information which it is desired to store in that register, means responsive to some of said pulses for producing a pattern of voltages, means responsive coincidently to said voltage and said voltage pattern for selecting said register, and means for storing said numerical information in said register.

. The suit in the Court of Claims included an issue arising because Sharpless and Eichert, the inventors of the automatic reset feature, had been employed by the University of Pennsylvania under a U. S. Navy contract to do research on the ENIAC and EDVAC government contracts. Under this contract, research associates were to grant to the United States a royalty-free right and license to all discoveries and inventions growing out of the research. When the government manufactured and used the inventions of the Sharpless-Eichert patent, Technitrol brought suit in the Court of Claims to recover compensation for the unauthorized use.

The commissioner recommended to the court that it find the government not licensed under the patent, but the court did not wholly adopt his recommendations. It found, instead, that the government was licensed “except to the extent that those other claims may be limited to the system’s automatic reset feature.” Technitrol, Inc. v. United States, 440 F.2d 1362, 1364, 194 Ct.Cl. 596 (1971). That decision rested on the fact that the reset feature was the only portion of the patent disclosure invented after the inventors left the employ of the government’s contractor.

But the court declined to pass on the question of whether that feature was adequately described in the claims, stating:

. In determining validity, as the Dominion case [Dominion Magnesium Ltd. v. United States, 320 F.2d 388, 162 Ct.Cl. 240 (1963)] and many others illustrate, courts must frequently choose between a narrow construction of the claims that upholds the patent and a broad construction that strikes it down. We do not wish to make that choice at this stage, believing that it would be both unwise for the court and unfair to the parties to affix a meaning to the claims in the absence of further proceedings, including the making of a record, on validity. 440 F.2d at 1369.

*997 Control Data urges us to disregard the findings of the commissioner who had construed representative claims, including claim 19, so as to include the reset feature, because his report was not followed by the Court of Claims. While it is true that the court did not wholly follow his recommendations, it decided the case on other grounds and specifically did not decide the issue of whether the automatic reset feature was included in the claims of the patent as just above stated. We neither accept nor reject finally the commissioner’s construction of claims although his reasoning is persuasive.

The opinion of the district court is reported at 394 F.Supp. 511 (D.Md.1975), and the report of the commissioner in the Court of Claims at 164 U.S.P.Q. 51 (Ct.Cl.Commr.1969).

. Technitrol takes the position clauses 1 and 2 of claim 19 disclose the reset feature.

. The third paragraph of § 112 was added in 1952. There seems to be general agreement that it was intended to ameliorate the effect of Halliburton Co. v. Walker, 329 U.S. 1, 67 S.Ct. 6, 91 L.Ed. 3 (1946), and to restore the authority of Continental Paper Bag Company v. Eastern Paper Bag Company, 210 U.S. 405, 28 S.Ct. 748, 52 L.Ed. 1122 (1908). Halliburton and Continental narrowed and broadened, respectively, functional claiming. The following articles to some extent or other address the addition, and we have considered them in the preparation of this opinion. Smith, Functional Claims and the Patent Act of 1952, Journal of the Patent Office Society, July 1966, Vol. XXVIII, No. 7; from the same Journal, Vol. XXXVII, Oct. 1955, No. 10, Riesenfeld, The New American Patent Act in the Light of Comparative Law, Part II; Wachsner, Commenta-ria, Functional Claims.

The case before us has been commented upon in Patent Law Perspectives, 1969-70 Annual Review, § E.T, and in 1975 Developments, § A.5. The editors, in general, would seem to be in agreement with our conclusion.

. We are aware that means and function claiming results in quite broad claims which are to greater or lesser extent limited by the specifications. We do not have before us at this time the question of the extent to which the claims here should be limited. See Deller’s Walker on Patents, 2nd ed., e. g. §§ 248 and 255.

. If it be thought that the reason the district court held against the claim was because it thought the claim did not include a function, we think that it did as explained in the body of the opinion.

Needless to say, the decision of the district court on claims 5 and 23 will have to be reconsidered in the light of this opinion.