MARKEY, Chief Judge.
The Laitram Corporation and Intralox, Inc. (Laitram) appeal from a judgment of the United States District Court for the District of Maryland, No. S 83-3126, finding that the Cambridge Wire Cloth Company (CWC) did not infringe Patent Nos. 3,870,141, 4,051,949, and Re.30,341. We reverse and remand.
Background
In 1983, Laitram sued CWC for infringement of U.S. Patents 3,870,141 (’141), 4,051,949 (’949), 4,159,763 (’763) and Re.30,-341 (’341). The accused products were CWC’s modular plastic conveyor belt (CAM-CLEAN I) and a belt drive system. In 1985, the U.S. District Court for the District of Maryland upheld the validity of the patents, found each infringed, and enjoined further infringement.
Laitram Corp. v. Cambridge Wire Cloth Co.,
226 USPQ 289 (D.Md.1985),
aff'd in part, rev’d in part,
785 F.2d 292, 228 USPQ 935 (Fed. Cir.),
cert. denied,
479 U.S. 820, 107 S.Ct. 85, 93 L.Ed.2d 39 (1986) (the first trial). In a thorough and comprehensive opinion, the court outlined CWC’s failed effort to obtain rights to the patented invention through merger, acquisition, or joint venture, followed by CWC’s engagement of a design firm to “incorporate features similar to Laitram patents.”
Id.
at 291. The court correctly interpreted the claims and fully described the inventions and patents in that opinion.
In mid-1986, CWC filed and the court denied a motion to exclude from the injunction a “redesigned” module (CAM-CLEAN II). In November 1986, CWC sought a judgment declaring CAM-CLEAN II non-infringing. The court dismissed that suit. When CWC began manufacture of CAM-CLEAN II, Laitram sued for infringement of claim 1 of the ’141 patent, claim 21 of the ’949 patent, and claim 1 of the ’341 patent. Broadly, the ’141 patent discloses plastic modules having links joinable by plastic rods to create a conveyor belt; the ’949 patent discloses a module with raised ribs; and the ’341 patent discloses a sprocket drive system for conveyor belts.
Laitram moved for a preliminary injunction and for a finding of contempt. The court consolidated all matters before it and conducted a six-hour trial on November 12, 1987.
In an opinion dictated from the bench, the court found that CAM-CLEAN II infringed none of the three asserted claims and denied the contempt motion.
Issues
1. Whether the district court erred in determining noninfringement of the asserted claims of the ’141, ’949, and ’341 patents.
2. Whether the district court properly denied the contempt motion.
OPINION
Introduction
This appeal again illustrates one of the many difficult dichotomies that lurk in the lacunae of patent law. On one side rests the very important, statutorily-created necessity of employing the clearest possible wording in preparing the specification and claims of a patent, one of “the most difficult legal instruments to draw with accuracy.”
On the other lies the equally important, judicially-created necessity of determining infringement without the risk of injustice that may result from a blindered
focus on words alone. The former, set out in 35 U.S.C. § 112, recognizes a competitor’s need for precise wording as an aid in avoiding infringement. The latter is called the “doctrine of equivalents”. While requiring a look at all the words while resisting their tyranny,
and requiring, because the claims measure the invention, a look at all claim limitations, the doctrine, in a proper case, “temper[s] unsparing logic and prevents] an infringer from stealing the benefits of an invention.”
Graver Tank & Mfg. Co. v. Linde Air Prods. Co.,
339 U.S. 605, 608, 70 S.Ct. 854, 856, 94 L.Ed. 1097, 85 USPQ 328 at 330 (1950) (quoting
Royal Typewriter Co. v. Remington Rand,
168 F.2d 691, 692, (2d Cir.1948) (L. Hand, J.)). In that sense, the doctrine recognizes a fact of the real business world: words are not misappropriated; claimed inventions are.
In the present case, the semantic antics of counsel at trial and in their briefs on appeal have obfuscated the true nature of the dispute as it must be perceived in law.
That obfuscation led in turn to a failure to apply the “substantially the same function in substantially the same way to obtain substantially the same result” guidance of
Graver Tank,
339 U.S. at 608, 70 S.Ct. at 856, 85 USPQ at 330.
The basic facts are either undisputed or accepted here as found by the district court. The difficulty with the appealed judgment lies in the district court’s application to the facts of the law governing determinations of infringement.
1. Infringement
The ’141 Patent
(a) Claim Interpretation
Claim 1 of the ’141 patent reads:
1. A module for constructing linked structures, said module comprising in combination:
a first plurality of link ends of substantially identical width, each being formed to circumscribe a pivotal hole, said hole of said first plurality being arranged co-axially;
a second like plurality of link ends of substantially identical width, each being formed to circumscribe a pivotal hole, said holes of said second plurality being arranged coaxially, the axes of respective holes of both pluralities of link ends being substantially parallel; and a plurality of spaced apart elongated members each integrally formed with and joining a pair of corresponding link ends of said first and second pluralities, said members being joined by at least one cross rib also formed integrally therewith and extending parallel to said axes;
said link ends being dimensioned and spaced apart by a distance slightly greater than said width so that said module is end-to-end reversible, so that a plurality of said modules may be engaged with each other at said ends.
It is undisputed that all limitations of the claim, save the last, read directly on CAM-CLEAN II. The district court focused on the words “slightly greater”, correctly viewing that adjectival phrase as a “sloppy term” requiring for its interpretation reference to the specification, the prosecution, and the prior litigation:
You cannot make exactly that determination [whether “slightly greater means .2 but not .3, or .3 but not .06.”]. It is not exact. It is not scientific. It is a sloppy term, but slightly greater, even though it
is a sloppy term, has no meaning for this case outside of the claims of Patent 141, as interpreted by the prosecution of that patent in the patent office and as interpreted by the prior history of this litigation, including Mr. Lapeyre’s testimony, which is what he won on in the first suit.
Having had a witness measure the link end widths and the spacing between link ends of Laitram’s commercial product and of the accused CAM-CLEAN II module, the court found the spacing in the former was 106% of the width and that in the latter was 135%.
Deeming the latter “more than slightly greater”, the court found noninfringement.
As discussed below, application of the law governing infringement by equivalents renders that ultimate finding of noninfringement clearly erroneous.
The phrase “slightly greater” is not presented in a vacuum. The claim contains definitional parameters: “so that said module is end-to-end reversible, so that a plurality of said modules may be engaged with each other at said ends”.
In addition, the district court properly interpreted the claim in light of the prosecution history in the first trial, defining “slightly greater” spacing in terms of its purpose,
i.e.,
spacing that would “minimize bending and maximize shear”.
Assuming, arguendo, the district court’s interpretation of the phrase “slightly greater”, in the second trial, as the 106% width in Laitram’s product, CWC’s 135% width may be seen as avoiding a finding of literal infringement. The question would remain, however, of whether CWC’s 135% width constituted a change sufficient to escape a finding of infringement under the doctrine of equivalents.
(b) Doctrine of Equivalents
In the first trial, the court found that CAM-CLEAN I constituted an “infringement by equivalents of claim 1.” 226 USPQ at 298. In this trial, the court’s entire discussion of equivalence with respect to CAM-CLEAN II, reads:
It is true that all that has been done in the redesign of Cam-Clean II is an alteration of the spacing, relative spacing of the link ends.
That is true, but if the result is that when the patent language, the claim language is viewed in context of its prosecution history, as it has to be under the doctrine of equivalence (sic, equivalents), if the result is that even applying the doctrine of equivalence [sic] there is no longer a spacing between links that is only slightly greater than the width of the links, then there is no infringement. That is all there is to it. It is that simple.
To me the telling proof is made up in two ways. Number one, the tests, in my opinion — although the difference is admittedly small at working loads, at the higher loads, where you would test to ascertain the true strength of the material — in my judgment the tests show that the Cam-Clean II is less resistant to the bending and elongation forces than the original Cam-Clean, but that is not real [sic] really the most important question.
The most important question is, what is the spacing of the redesign Cam-Clean relative to the width of the link ends?
The foregoing application of the doctrine of equivalents is legally flawed. Having determined that all CWC did was to alter the spacing, the court did not apply its own criterion derived from the prosecution history in the earlier trial,
i.e.,
whether the “alteration” left the spacing such that it would still achieve the “minimize bending/maximize shear” purpose the court found as the measure of the “slightly greater” phrase in the limitation. Applied to all the basic facts, the law required a finding that the spacing in CAM-CLEAN II was the equivalent of the “slightly greater” limitation in the claim.
The “tests” relied upon by the district court were non-probative for at least five reasons: (1) they subjected the CAM-CLEAN II to a force more than three times the 1500 pounds to which such belts are subjected in use and for which they are warranted (in the prior trial, the district court correctly rejected such tests because “the belt was subjected to forces way beyond the normal range that would be applied in a working situation” and because “There is nothing in Mr. Lapeyre’s patent, even when considering file wrapper estop-pel, that claims that the rod is immune to bending in his invention under extraordinary conditions”)
; (2) the “admittedly small” difference was an elongation, at the working load, of less than Vs the width of a human hair; (3) there is nothing in the “slightly greater” limitation dealing with any “strength of the material” criterion; (4) whether CAM-CLEAN II is “less” resistant is irrelevant if the test of equivalency is met; and (5) the tests did nothing to show that CAM-CLEAN II had been changed sufficiently to escape the equivalency formulation set out in
Graver Tank.
In applying the doctrine, the “most important question”, is not “what is the spacing,” but whether the spacing in the accused product is equivalent to that in the claim,
i.e.,
whether it continues to meet the district court’s “minimize bending/maximize shear” criterion. If it does, the rest of the claim being fully met, CWC’s CAM-CLEAN II and the claimed invention are substantially the same, used in substantially the same way, to achieve substantially the same result.
The criterion is not “zero bending/unlimited shear.” That may be an ideal condition, but it is not claimed to be achieved by the invention.
See U.S. v. Telectronics, Inc.,
857 F.2d 778, 781 (Fed.Cir.1988) (“avoid fibrous tissue formation” did not mean there could be
no
fibrous tissue). CWC asserts, and cites evidence to show, only that CAM-CLEAN II is less efficient,
i.e.,
that it does not minimize bending and maximize shear
quite as much
as the claimed invention does. Assuming, ar-guendo, the truth of that assertion, it cannot alone prevent a finding of infringement under the doctrine of equivalents.
Perkin-Elmer Corp. v. Computervision Corp.,
732 F.2d 888, 901, 221 USPQ 669, 679 (Fed.Cir.1984) (“exact identity of results is not required in establishing equivalency”). That application of the doctrine arises out of a recognition of a second real-world phenomenon: inefficient infringement is still infringement.
CWC does not dispute its vice president’s testimony that CAM-CLEAN II performs substantially the same function to achieve substantially the same result as CAM-CLEAN I, that CWC sells CAM-CLEAN II modules as interchangeable replacements for CAM-CLEAN I, and that CWC warrants CAM-CLEAN II for use at the same maximum working strength (1550 lbs/ft) as CAM-CLEAN I. Nor does CWC dispute Laitram’s assertion that CWC shaved only “Aooo of an inch off its link ends. CWC does assert, irrelevantly in this case, that it added plastic to beef up the link ends, a change described in a deposition of CWC’s vice-president as “insignificant.” Most significantly, the
Graver Tank
formula for equivalence appears nowhere in CWC’s brief and nowhere does CWC suggest, nor does it show any basis for finding, that CAM-CLEAN II does
not
perform “substantially the same function in substantially the same way to obtain substantially the same result.”
(c) Prosecution History
The district court’s discussion of the prosecution history and its effect on range of equivalents to which claim 1 was entitled was this:
That is, Mr. Lapeyre, although he has [sic] acknowledged as a pioneer inventor and certainly is the first person that came up with the idea of this plastic modular belting, which is a wonderful idea, nonetheless, he could not make a broad claim that would get through the patent office. He tried it. He tried a very broad claim.
* * * * * *
It is true that the belt is not, at rated operating load, significantly weaker than the original Cam-Clean belt was; and if, in fact, Mr. Lapeyre were entitled to a broad claim here that covered all modular plastic belting with multiple links, then certainly the Cam-Clean II, as redesigned, would still infringe and I would so hold. But he is not entitled to a broad patent.
The “very broad claim” referred to is necessarily the broadest,
i.e.,
claim 1 as originally filed. The amendment added “by a distance slightly greater than said width” after “apart”. The court went on to cite file history arguments and the inventor’s testimony in the earlier trial, all in support of its perfectly correct determination that the purpose of spacing no more than slightly greater was to minimize bending and maximize shear.
It is clear that a module in which the spacing was so great as to have no or very little effect in minimizing bending and maximizing shear would not accomplish the purpose attributed to the “slightly greater” limitation, but that is not this case. Nor did the district court find, as it could not on this record, that CAM-CLEAN II’s spacing had a minimize bending/maximize shear effect substantially different from that intended by the limitation, or from that of CAM-CLEAN I, or from that of Laitram’s product.
The court erred in treating the amendment of claim 1 and arguments in the file
history as sufficient to preclude a finding of infringement under the doctrine of equivalents. It is not necessary, of course, that an inventor be entitled to a broad claim covering all possible products in a line of products before a court may award an invention pioneer status,
or a range of equivalents sufficient to encompass a particular accused product. It is commonplace for claims to inventions, pioneer and non-pioneer, to be amended during prosecution. Though the “area of equivalence” may vary, the doctrine of equivalents is applicable to both,
Graver Tank,
339 U.S. at 608, 70 S.Ct. at 856, 85 USPQ at 330. Thus it was not necessary that the present inventor be entitled to a broad claim covering “all modular plastic belting with multiple links” before it could be determined that “CAM-CLEAN II, as redesigned, would still infringe.”
What counts is the thrust of an amendment and remarks in the prosecution history. Here the amendment added “by a distance slightly greater than said width” and the inventor’s remarks described the purpose of such limited spacing (necessarily in cooperation with the rest of elements in the claim) as being to minimize bending and maximize shear. From the absence of any finding that CAM-CLEAN II did
not
minimize bending and maximize shear, and from the district court’s uncontested finding that the “difference is admittedly small at working loads”, and from the district court’s uncontested finding that CAM-CLEAN II, even at extraordinary loads, is only “less resistant to bending and elongation forces”, and from the district court’s effectively uncontested finding that “the belt is not, at operating load significantly weaker,” and from CWC’s admission that CAM-CLEAN I and II are interchangeable and identically warranted, it can only be concluded that in CAM-CLEAN II bending is minimized and shear is maximized. Nothing in the prosecution history of the ’141 patent conflicts in any manner with that conclusion.
(d) Conclusion on the '141 Patent
Every limitation in claim 1, or its equivalent (the spacing in CAM-CLEAN II being equivalent to “slightly greater”), is thus found in CAM-CLEAN II.
Compare Pennwalt,
833 F.2d at 939, 4 USPQ2d at 1743 (means-plus-function claim is not infringed under the doctrine of equivalents when accused device includes no component performing one of functions listed in claim or an equivalent function). The simple fact that the spacing, measured in lOOOths of an inch, is greater in CAM-CLEAN II than in Laitram’s commercial product, or greater than in CAM-CLEAN I, or greater than some generalized notion of “slightly”, is irrelevant when the spacing is insufficient to move CAM-CLEAN II outside the guidance of
Graver Tank.
Application of the law governing determination of equivalents to the established facts requires a finding of infringement.
The ’9It9 Patent
Claim 21 reads:
21. A linked belt adapted to move through a circular arc for use in conjunction with a conveyor comb having a plurality of spaced apart parallel teeth, said belt comprising:
a plurality of like modules, each of said modules including a first and second like pluralities of link ends of substantially
identical width, each such end being formed to circumscribe a pivotal hole through said width;
means for joining said pluralities to one another so that the axes of said holes of said first plurality are arranged coaxially, the axes of said holes of said second plurality are arranged coaxially and the axes of respective holes of both pluralities of link ends are substantially parallel,
a plurality of spaced apart elongated upstanding vanes mounted on said modules so as to extend in a substantially parallel direction to one another in a direction transversely to the axes of said holes of said both pluralities and protrude substantially perpendicularly from said module;
said link ends being dimensioned and spaced apart by a distance slightly greater than said width.
said link ends of each of said modules being releasably engaged between the link ends of an adjacent module except for individual link ends disposed at the extreme sides of said belt; means for pivotally connecting said modules at engaged link ends; said elongated vanes of each said modules being arranged in staggered relation with respect to the vanes of adjacent modules, said vanes of each module being spaced from one another sufficiently to form a multiplicity of channels lying beneath the upper most surface of said belt when staggered, said channels being adapted to receive said teeth.
It being undisputed that CAM-CLEAN II met each limitation save the fourth after “comprising”, the district court focused again on the “slightly greater” limitation in determining non-infringement of claim 21. Recognizing that the prosecution history of the ’949 patent contained nothing about spacing, the court cited the prosecution history of the ’141 patent
and the inventor’s testimony relating thereto in the first trial, to the effect that the purpose of the spacing was to minimize bending and maximize shear. The court found noninfringement, again without reference to the guidance in
Graver Tank,
concluding:
So, I think you are estopped to reclaim any broader sense of the claim language in Claim 21, 949 that is readable in claim 1 of 141. Therefore, I find no infringement of Claim 21, 949 because in my judgment the Cam-Clean II as redesigned does not have spacing between adjacent link ends then [sic, that] is only slightly greater than the width of the link ends; and therefore I find no infringement of Claim 21 of 949.
Thus, the district court found non-infringement of the asserted claims of the ’141 and ’949 patents on the same basis. For the reasons outlined above in respect of claim 1 of the ’141 patent, a finding of infringement of claim 21 of the ’949 patent is required.
The ’341 Patent
(a) The Claim
Claim 1 reads:
A sprocket and drive shaft assembly for a chain link conveyor belt subject to variations in width comprising, in combination:
a drive shaft with a substantially uniform, noncircular cross section; and a plurality of sprockets mounted in driven engagement on said shaft for supporting and driving a conveyor belt; one of said sprockets being releasably locked against axial motion to provide for proper belt tracking, all other of said sprockets being free to move axially.
There was in this second trial no question of any “redesign” of CWC’s sprockets.
The district court stated the ’341 patent issue before it:
The question also is presented with regard to whether the sale of square drive sprockets by CWC and the sale of round bore keyed drive sprockets by CWC constitutes an infringement of Claims 1 and 3 of reissued Patent 341.
(b) Claim Interpretation
In the first trial, CWC admitted that its sale of sprockets constituted contributory infringement and limited its defense to an assertion of invalidity. In holding that CWC had failed to carry its burden of sustaining that defense, the district court had this to say about the invention claimed in the ’341 patent:
The essence of patent ’341 is the need to maintain a positive drive relationship between the sprocket and the belt while taking into account the lateral expansion in the belt caused by fluctuations in temperature when the belt is put to applications where the product is conveyed from hot to cold or cold to hot areas.
This was a problem in practice for Mr. Lapeyre which he satisfied by coming up with the idea embodied in ’341, that is fixing the central drive sprocket and allowing the other two or more lateral drive sprockets to float. Additionally, to avoid the problem of loss of a key or shearing of a key (I take judicial notice that keyed devices are frequently subject to having the key sheared and thus losing the drive connection) he came up with the idea of making the shaft on which the sprockets are mounted square.
I do find that it was a known principal [sic] at the time this invention was made that a square shaft could be substituted for a keyed round shaft.
This was a principle of mechanical design. I also find that Claghorn and Comstock each showed sprocket drive mechanisms that were not all fixed in position along a drive shaft. However,
none of them showed a combination of one fixed and several free floating sprockets, whether mounted on a keyed round shaft or on a square shaft.
(Emphasis added).
Therefore, the notion of the free float of several sprockets, while fixing one central sprocket to maintain positive drive is not disclosed in any of the patents. So, I find that the use of one fixed and severally completely floating unrestrained sprockets was not obvious from the prior art. Furthermore, applying the test of obviousness, that is, that something in the prior art must itself suggest the combination of all the ingredients of the patent
I do not find anything
in Comstock, Cremona, Claghorn and the prior art cited by the Patent Office
that suggests putting together that “one fixed and several free floating” arrangement, together with a non-round shaft. Variations on the non-round shaft have been around for years in one application or another.
(Emphasis added).
¡}s sj: jjs j{s s}« sj«
With regard to infringement of the ’341 patent, it is not contested. The sprocket as manufactured by Cambridge and as taught in their engineering devices, has one fixed and several free floating it’s [sic] on a square shaft and it clearly, literally infringes.
226 USPQ at 298-99.
In opposing Laitram’s motion for contempt, CWC described the invention claimed in the ’341 patent:
The Patent’s claims in issue, Claims 1-3, are directed to a sprocket and drive shaft assembly for a chain link conveyor subject to variations in width with the elements comprising a drive shaft of substantially uniform non-circular cross section and a plurality of sprockets mounted
in driving engagement on the shaft to drive the conveyor belts, one of the sprockets being releasably locked against axial movement to provide belt tracking and all other sprockets being free to move axially.
Thus, the Patent is not directed to square bore sprockets per se, or even the combination of square bore sprockets and square bore shafts. Instead, the Patent covers the specific combination of non-circular bore sprockets and non-circular shaft, one of the sprockets being fixed and the others free to move axially on the shaft.
(Emphasis added).
(c) Infringement
CWC sells two types of sprockets, referred to by the district court and CWC as “round bored” and “square bored”. A cylindrical keyed shaft was referred to as “round” and “circular”. For convenience, we employ the district court’s terminology.
In finding that CWC’s sales of its “round bored” sprockets did not constitute contributory infringement, the district court said:
Now, the next question pertinent to the reissue of 341 patent is whether the plaintiff Laitram has proved by a preponderance of the evidence that the sale of sprockets that are to be fitted into a shaft that is essentially round in cross-section but that has rectangular key ways installed — in particular, the application shows that the — not the patent application but the practical application— shows that there are two key ways horizontally opposed around the periphery of the shaft, the round shaft, and that that is the mechanism that fixes the sprocket into position on the shaft, that transmits the motion of the shaft to the sprocket, and I do find that the plaintiff has not proved by a preponderance of the evidence that this device infringes on reissued Patent 341, even when it is sold in an installation where one of the sprockets is fixed and the other is floating axially along the shaft.
The reason I so hold is that I find that that Laitram is estopped from its position taken and testimony made in a prior trial, as well as my prior findings with regard to this to say that the round shaft with key ways is the functional equivalent of a square board [sic, shaft].
It is evident from Claims 1 and 3 of reissued 341 and also the narrative in column 4 of that patent, especially lines 39 to 43, that the benefit claimed from the square drive is the better transmission of force, which is accomplished by the use of a regular polygon.
Now, certainly there is nothing in the patent that explicitly says that there has to be a regular polygon; and here I am talking about something that does not have circular sides but something with straight sides, like a hexagon or a square. There is nothing in the patent that limits it to an irregular polygon. The language, in fact, says, substantially uniform in cross-section and noncircular, or words to that effect.
Now, certainly, the round shaft with keys is substantially uniform in cross-section because as you go up and down it, the cross-section is the same. The question is whether it is noncircular.
It is true that if you trace a line around the periphery, it is uncircular,
but under any practical construction of the patent
in light of the testimony of Mr. Lepeyre
at the prior trial — and here I am talking about a
transcript, page 59,
with regard to keys in a circular shaft or a round shaft, that
his idea of a circular shaft includes a circular shaft that has key ways
in it or key ways in the shaft to hold keys that transmit the force to the sprocket. That is what he said, and in his judgment that was essentially — or that, not essentially, that was a circular shaft. His idea is noncircular.
The particular embodiment of it in Claim 3 is square
Claim 1 is broader but still noncircular,
[sic, and] in my judgment
does not cover the round shaft with keys, which is a technology that has been known for donkeys’years and, in fact, would not itself be a patentable device
and that is quite obvious; and, further, if you look at my oral opinion in the other case at 299,
I indicated in there that the square shaft was adopted
for the reason of avoiding the problem of loss of a key or shearing of a key.
And this is consistent with Mr. Le-peyre’s testimony and I find that a round shaft with keys is essentially a circular shaft. It is not a noncircular shaft, as that is used in 341, as the inventor himself views what is a circular and noncir-cular shaft; and I think that the resolution of this is clear and
the inventiveness in 341 was putting together the square shaft or other regular polygonal shaft, which gives you a very positive drive mechanism, putting that together with fixing one sprocket and letting others float.
And frankly, I think that it would be a result that would be inequitable and not in conformity with the evidence that has already been laid before the Court and therefore
would not be a sustainable result for me to say now, that, oh, yes, well, the circular shaft with keys is the functional equivalent of a square shaft and therefore the circular shaft with keys infringes.
That would be an inequitable result and one that is not allowed, given the fact that I think it’s fair to estop the Plaintiff from adopting that position in light of the history of the 3126 [the first trial] case, the testimony of Mr. Lepeyre in that case, and my findings with regard to this issue. (Emphasis added).
The inventor’s testimony described by the court as found in the “transcript, page 59” was this:
Q. And could you explain the problems that relate to the use of round shafts and tell the Court why you went to a square shaft and the rest?
A. Well in a round shaft, with a sprocket having a round bore, the sprocket in some way must be fastened, prevent it from rotating on the shaft. Otherwise it would be useless as a driving member. Therefore, key seats are cut into the shaft, key seats are cut into the sprockets, keys are put into those key seats. Set screws are put in to hold the keys in position. If any of those parts get loose, or if they fall into the machine you have a devil of a time getting them out, so I didn’t like that. Number two, whereas you can make a key sloppy fit in a shaft, it’s a poor, it tends to be a poor guide for axial guidance of the sprocket, as the sprocket wants to move along the shaft.
As Laitram correctly points out, the foregoing testimony dealt only with the inventor’s reasons for selecting a square shaft and square bored sprockets as his preferred, commercial embodiment. It could not, therefore, estop the inventor from asserting that a keyed cylindrical shaft is the functional equivalent of a square shaft. Similarly, Laitram correctly points out that “column 4 of that patent, especially lines 39 to 43”, cited by the court, merely described the preferred embodiment. The entire sentence runs from line
36
to 43 and reads:
Although other noncircular cross sections such as hexagonal may be employed, the square cross section is preferred because it provides maximum strength coupled with maximum driving force — imparting surfaces and corners, is symmetrical with respect to the rotational axis and lends itself to fabrication simply and easily of a variety or combination of materials.
That the inventor preferred and adopted commercially a square shaft, because keys tend to fall out of keyed shafts, is not a basis for limiting claim 1 to square or polygonal shafts. References to a preferred embodiment, such as those often present in a specification, are not claim limitations.
SRI Int’l. v. Matsushita Elec. Corp.,
775 F.2d 1107, 1121, 227 USPQ 577, 589 (Fed.Cir.1985) (in banc). Such reference is thus not a basis here for limiting “noncircular” to square or regular polygonal shafts.
CWC points to a single statement in the inventor’s testimony at the first trial (to which the district court made no reference in this case). That statement was that any engineering professor would view a section of the prior art Claghorn shaft as a circular cross section because a keyway has been milled in. The next question in that testimony, however, was directed to whether Glaghorn’s shaft met the “noncircular” limitation, to which the inventor answered, “It’s up in the air, I guess. I would say to bracket that if it’s noncircular cross section.” Nothing in that testimony would estop the inventor from reliance on the functional equivalency of square and keyed cylindrical shafts or from asserting that the sale of keyed cylindrical shafts and round bored sprockets for such shafts, for use in an installation in which one sprocket was fixed on the shaft and the other sprockets were free to float axially, constituted contributory infringement under the doctrine of equivalents.
As above indicated, the district court held in its opinion after the first trial that none of the prior art “showed a combination of one fixed and several free floating sprockets, whether mounted on a keyed round shaft or on a square shaft,” and that “it was a known principle ... that a square shaft could be substituted for a keyed round shaft,” and that “variations on the non-round shaft have been around for years in one application or another,” and that “[T]he essence of patent ’341 is the need to maintain a positive drive relationship between the sprocket and the belt” in light of expansion caused by temperature fluctuations.
Thus, the court, in considering the validity issue, interpreted the claim as drawn to fixed and free floating sprockets on a non-round shaft, without regard to the particular cross sectional shape of the non-round shaft. “Having construed the claims one way for determining their validity, it is axiomatic that the claims must be construed in the same way for infringement.”
W.L. Gore & Assoc., Inc. v. Garlock, Inc.,
842 F.2d 1275, 1279, 6 USPQ2d 1277, 1280 (Fed.Cir.1988). Here, the district court committed legal error in interpreting claim 1 differently for infringement,
i.e.,
as though it were limited to use of a square or other regular polygonal shaft (“The inventiveness in 341 was putting together the square shaft or other regular polygonal shaft, which gives you a very positive drive mechanism, putting that together with fixing one sprocket and letting others float.”).
Claim 3 is limited to a square shaft and claim 6 is limited to a regular polygonal shaft. The doctrine of claim differentiation precludes a reading of those limitations into claim 1 in determining literal infringement.
D.M.I., Inc. v. Deere & Co.,
755 F.2d 1570, 1574, 225 USPQ 236, 239 (Fed.Cir.1985). Apart from literal infringement, the district court recognized that a keyed cylindrical shaft is the functional equivalent of a square shaft.
It declined, however, to find infringement under the doctrine of equivalents because it felt that that result would be “inequitable”. The basis for that view was that it was “fair to estop” Laitram from reliance on the doctrine in light of the history of the prior trial including the inventor’s prior testimony.
If
the inventor had led the court to sustain
claim 1 of the patent in the first trial on the basis that claim 1 must be limited to noncir-cular shafts that are square or otherwise polygonal in cross section, a judicial estop-pel might have precluded any change from that position.
Colonial Refrigerated Transp., Inc. v. Mitchell,
403 F.2d 541, 550 (5th Cir.1968). As above indicated, however, we can find no indication in the record that that is what happened, or could have happened, here.
(d) Conclusion on the ’341 Patent
The district court erred in applying the law of contributory infringement when it declined to find that sale of CWC’s keyed cylindrical shafts and corresponding sprockets, for use in installations in which one sprocket is fixed and the others float, constituted contributory infringement of claim 1 of the ’341 patent.
2. Contempt
Absent infringement, there cannot be contempt for violating an injunction.
KSM Fastening Sys. v. H.A. Jones Co.,
776 F.2d 1522, 1528, 227 USPQ 676, 680 (Fed.Cir.1985).
As the district court correctly stated, the standard for proof of infringement is “by a preponderance of the evidence” and the standard for contempt is “by clear and convincing evidence.” Having found that Laitram failed its burden under the lesser standard, the court necessarily denied Laitram’s motion for contempt in view of the court’s noninfringement findings. Those findings having been reversed, remand becomes necessary to enable the district court to exercise its discretion in determining whether CWC’s conduct amounted to contempt.
See MAC Corp. v. Williams Patent Crusher & Pulverizer Co.,
767 F.2d 882, 226 USPQ 515. (Fed.Cir.1985).
CONCLUSION
The portions of the judgment based on findings that claim 1 of the ’141 patent and claim 21 of the ’949 patent were not infringed by sales of CAM-CLEAN II, and that claim 1 of the ’341 patent was not contributorily infringed by sales of keyed round shafts and round bored sprockets, are reversed. The case is remanded for consideration of the motion for contempt.
REVERSED and REMANDED.