This is an appeal from the judgment of the United States District Court for the District of Delaware, reported at 594 F.Supp. 1420, holding claims 3-5 and 9 of Moleculon Research Corporation's U.S. Patent No. 3,655,201 ('201 patent) valid and infringed by certain of the well-known Rubik's Cube puzzles. We affirm in part, vacate in part, and remand.

Moleculon, as assignee of the '201 patent which issued to Larry D. Nichols, sued CBS Inc., as successor to the Ideal Toy Corporation, alleging infringement of claims 3, 4, 5, 6, and 9 of the '201 patent.

A puzzle enthusiast since childhood, Nichols, in the summer of 1957, conceived of a three-dimensional puzzle capable of rotational movement. He envisioned an assembly of eight cubes attached in a 2 X 2 X 2 arrangement, with each of the six faces of the composite cube distinguished by a different color and the individual cubes being capable of rotation in sets of four around one of three mutually perpendicular axes.

During the period 1957-1962, while doing graduate work in organic chemistry, Nichols constructed several paper models of his puzzle, making cubes of heavy file-card type paper and affixing small magnets to the inside of the cubes. Although these models confirmed the feasibility of Nichols' conception, they lacked durability. A few close friends, including two roommates and a colleague in the chemistry department, had occasion to see one of these paper models in Nichols' room and Nichols explained its operation to at least one of them.

In 1962, Nichols accepted employment as a research scientist at Moleculon. In 1968, Nichols constructed a working wood block prototype of his puzzle which he usually kept at home but on occasion brought into his office. In January 1969, Dr. Obermayer, the president of Moleculon, entered Nichols' office and happened to see the model sitting on his desk. Obermayer expressed immediate interest in the puzzle and Nichols explained its workings. Obermayer asked whether Nichols intended to commercialize the puzzle. When Nichols said no, Obermayer suggested that Moleculon try to do so. In March 1969, Nichols assigned all his rights in the puzzle invention to Moleculon in return for a share of any proceeds of commercialization. On March 7, 1969, Moleculon sent Parker Brothers an actual model and a description of the cube puzzle. In the next three years, Moleculon contacted between fifty and sixty toy and game manufacturers, including Ideal. Ideal responded to the effect that it did not currently have an interest in marketing the puzzle. Moleculon itself did not succeed in marketing the Nichols cube.

On March 3, 1970, Nichols filed on behalf of Moleculon a patent application covering his invention. The '201 patent issued on April 11, 1972.

The subject matter of the '201 patent, in its preferred embodiment, is a cube puzzle composed of eight smaller cubelets that may be rotated in groups of four adjacent cubes, and a method by which the sets of cubes may be rotated, first to randomize, and then to restore a predetermined pattern on the six faces of the composite cube.

3. A method for restoring a preselected pattern from sets of pieces which pieces have constantly exposed and constantly nonexposed surfaces, the exposed surfaces adapted to be combined to form the preselected pattern, which sets when in random engagement fail to display said preselected pattern which comprises:

a. engaging eight cube pieces as a composite cube;

b. rotating a first set of cube pieces comprising four cubes about a first axis;

c. rotating a second set of four cubes about a second axis; and

d. repeating steps (b) and (c) until the preselected pattern is achieved.

4. The method of claim 3 which includes rotating sets of cubes about one of three mutually perpendicular axes with reference to the composite structure.

5. The method of claim 3 wherein the sets of cubes are rotated through multiples of 90?.

Claim 9 is the only apparatus claim remaining in the suit:2

9. A puzzle comprising eight cubes, visually distinguishable indicia on three faces only of each cubes [sic] with the eight cubes together having six visually distinct indicia,

means associated with each of the remaining faces only of each of the cubes releasably maintaining the cubes in assembled relationship forming a composite cube, said maintaining means enabling three interaffiliated groups of four contiguous cubes each to be rotated respectively about three mutually perpendicular axes, the six distinct indicia being so located on the respective cubes that the cube groups can be rotated to effect the display of a distinct indicia on each of the six faces of the composite cube.

Shown below are Figs. 1 and 2b of the patent:

NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLE

NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLE

The accused products are the well-known 3 X 3 X 3 Rubik's Cube puzzle, two 2 X 2 X 2 variations--a Japanese-made Pocket Rubik's Cube and a Taiwanese-made Pocket Rubik's Cube (pocket cubes), and a 4 X 4 X 4 Rubik's Revenge. As shown below, these puzzles externally appear as composite cubes composed of smaller cubes or cubelets (27 for the 3 X 3 X 3, 8 for the 2 X 2 X 2, and 64 for the 4 X 4 X 4):

NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLEInternal inspection reveals that the composite cube is not composed of true six-sided cubelets but rather is composed of an internal mechanism holding together cubelet shells which have one or more external faces and permitting sets of cubelets to be rotated about an axis.

Whether the district court erred in (1) holding that the claimed invention was not in public use nor on sale within the meaning of 35 U.S.C. Sec. 102(b); (2) holding claims 3-5 not invalid for lack of utility and enablement under 35 U.S.C. Secs. 101, 112; (3) holding claims 3-5 and 9 nonobvious under 35 U.S.C. Sec. 103; and (4) finding infringement of claim 9 and induced infringement of claims 3, 4, and 5.

CBS argues that the subject matter of the '201 patent was in "public use" and "on sale" by Nichols, prior to the March 3, 1969 critical date (i.e., one year prior to filing of the patent application), thus rendering the patent invalid under section 102(b).

(a) Public Use

CBS labels as public use Nichols' displaying of the models to other persons (such as his colleagues at school) without any mention of secrecy. CBS ascribes only commercial purpose and intent to Obermeyer's use of the wood model and argues that a conclusion of barring public use under Sec. 102(b) is compelled. We disagree.

This is what the district court had to say:

The essence of "public use" is the free and unrestricted giving over of an invention to a member of the public or the public in general. What I see here, by contrast, is the inventor's private use of his own invention for his own enjoyment. "Private use of one's own invention is permissible."

* * *

While it is true that Nichols explained his puzzle to a few close colleagues who inquired about it and allowed Obermayer to in fact use it, the personal relationships and other surrounding circumstances were such that Nichols at all times retained control over its use as well as over the distribution of information concerning it. He never used the puzzle or permitted it used in a place or at a time when he did not have a legitimate expectation of privacy and of confidentiality. In these respects, I consider the exposure to Obermayer in Nichols' office no different than the exposure of Nichols' close friends in his home.

... Here also the relationship between the participants in the alleged uses evidences a retention of control by Nichols. None of those participants had any basis for inferring that the puzzle was being given over by Nichols for their free and unrestricted use. Holding the public use bar inapplicable in these circumstances will not remove anything from the public domain. Moreover, there is absolutely no evidence in this case of commercially motivated activity by Nichols during the relevant period. Accordingly, the underlying policy against extending the effective term of exclusivity is not offended by a finding that the Nichols invention was not in public use.

594 F.Supp. at 1427 (citations omitted).

CBS correctly recognizes that the district court's conclusion on public use under Sec. 102(b) is subject to review as a question of law while the facts underlying the conclusion on public use are subject to the clearly erroneous standard of review. Hycor Corp. v. Schlueter Co., 740 F.2d 1529, 1537, 222 USPQ 553, 559 (Fed.Cir.1984).

CBS urges that the decision in Egbert v. Lippmann, 104 U.S. 333, 26 L.Ed. 755 (1881), compels a conclusion of public use in the present case. In Egbert, the claimed invention was drawn to improved corset-springs, also called corset-steels. In 1855, the inventor made and presented to his lady friend a pair of corset-steels which embodied the invention. She wore the steels for a long time. The inventor made and gave her another pair three years later which she also wore a long time. On several occasions, when the corset itself wore out, the steels were removed and placed into a new corset. In 1866, an application for patent was filed. It was on those facts that the Court observed that: "If an inventor, having made his device, gives or sells it to another, to be used by the donee or vendee, without limitation or restriction, or injunction of secrecy, and it is so used, such use is public, even though the use and knowledge of the use may be confined to one person." Id. 104 U.S. at 336.

The district court distinguished Egbert because here Nichols had not given over the invention for free and unrestricted use by another person. Based on the personal relationships and surrounding circumstances, the court found that Nichols at all times retained control over the puzzle's use and the distribution of information concerning it. The court characterized Nichols' use as private and for his own enjoyment. We see neither legal error in the analysis nor clear error in the findings.

As for Obermayer's brief use of the puzzle, the court found that Nichols retained control even though he and Obermayer had not entered into any express confidentiality agreement. The court held, and we agree, that the presence or absence of such an agreement is not determinative of the public use issue. See TP Laboratories, Inc. v. Professional Positioners, Inc., 724 F.2d 965, 972, 220 USPQ 577, 583, (Fed.Cir.), cert. denied, --- U.S. ----, 105 S.Ct. 108, 83 L.Ed.2d 51 (1984). It is one factor to be considered in assessing all the evidence. There can be no question that the court looked at the totality of evidence, see id. at 971, 220 USPQ at 582; Hycor Corp. v. Schlueter Co., 740 F.2d at 1535, 222 USPQ at 557, and evaluated that evidence in view of time, place, and circumstances.

With regard to the question of control, CBS complains that "[t]he record is devoid of any testimony from the friends, associates and fellow workers who saw Nichols cube and to whom its operation was explained." The simple answer is that CBS had the burden at trial to prove public use with facts supported by clear and convincing evidence. Hycor Corp. v. Schlueter Co., 740 F.2d at 1536-37, 222 USPQ at 558-59. We think the district court's characterization of the evidence of record is entirely apt and we see no ground for reversal. Moreover, we agree with the district court that its conclusion on public use is consistent with the policies underlying the bar. See TP Laboratories, 724 F.2d at 968, 220 USPQ at 580.

CBS further argues in connection with public use that the district court erred when it found no evidence of commercially motivated activity by Nichols prior to the critical date.3 Although CBS attempts to paint a picture of commercialization from the discussions between Obermayer and Nichols, we see only the brush strokes of speculation. The record lacks hard evidence. Discussion between employer and employee does not by itself convert an employee's private pursuit into commercial enterprise with the employer. CBS also makes much of a February 6, 1969 phone call by Obermayer to Parker Brothers to see if the latter was interested in receiving a submission of a puzzle idea from an outside inventor. Nothing concerning the nature or workings of Nichols' puzzle was disclosed. Obermayer simply inquired whether and how an outsider could submit a puzzle for Parker Brothers' consideration. We agree with the district court that those facts do not show commercialization. Thus this case differs from other cases where commercial activity was said to violate the policies of section 102(b). See Kinzenbaw v. Deere & Co., 741 F.2d 383, 222 USPQ 929 (Fed.Cir.1984), cert. denied, --- U.S. ----, 105 S.Ct. 1357, 84 L.Ed.2d 379 (1985); D.L. Auld Co. v. Chroma Graphics Corp., 714 F.2d 1144, 219 USPQ 13 (Fed.Cir.1983); In re Smith, 714 F.2d 1127, 218 USPQ 976 (Fed.Cir.1983).

(b) On Sale

CBS argues that the claimed invention was on sale within the meaning of 35 U.S.C. Sec. 102(b) because Nichols orally agreed prior to the critical date (e.g., during a January 1969 conversation between Nichols and Obermayer) to assign "all his rights in the puzzle invention" to Moleculon. According to CBS, Nichols not only assigned the right to apply for a patent on the invention but also conveyed title in his single wooden model.

Although the formal written assignment occurred after the critical date, the district court held that even if there were an earlier oral agreement, an assignment or sale of the rights in the invention and potential patent rights is not a sale of "the invention" within the meaning of section 102(b). We agree. The few cases we have found on this issue have uniformly held that such a sale of patent rights does not come within the section 102(b) bar. United States Electric Light Co. v. Consolidated Electric Light Co., 33 F. 869, 870-71 (S.D.N.Y.1888); Scott Paper Co. v. Moore Business Forms, Inc., 594 F.Supp. 1051, 1075 (D.Del.1984); see also Federal Sign & Signal Corp. v. Bangor Punta Operations, Inc., 357 F.Supp. 1222, 1237 (S.D.N.Y.1973). Such a result comports with the policies underlying the on sale bar, see Western Marine Electronics, Inc. v. Furuno Electric Co., 764 F.2d 840, 845, 226 USPQ 334, 337 (Fed.Cir.1985), and with the business realities ordinarily surrounding a corporation's prosecution of patent applications for inventors, see Sun Studs, Inc. v. Applied Theory Associates, Inc., 772 F.2d 1557, 1568, 227 USPQ 81, 89 (Fed.Cir.1985).

As for CBS' contention that Nichols sold or agreed to sell the wood model of the puzzle, we quote the district court's relevant findings:

There is no indication in the subsequent written instrument that the parties contemplated the sale or transfer to Moleculon of the single physical embodiment of the puzzle then in existence. Nor is there any other evidence in the record to that effect.

CBS has shown nothing approaching clear error concerning those findings.

Accordingly, we sustain the district court's determination that the claims are not invalid under section 102(b).

CBS argues that the district court erred in holding the subject matter of claims 3-5 and 9 nonobvious under 35 U.S.C. Sec. 103 in light of U.S. Patent No. 3,081,089 issued to Gustafson on March 12, 1963 (Gustafson patent). The parties and the district court agree that the Gustafson patent is the most pertinent prior art. It teaches a "mechanical puzzle having a plurality of varicolored parts which are movable relative to each other to form various patterns." The district court added:

The puzzle consists basically of a spherical outer shell that rides on a substantially concentric spherical inner core. The outer shell is divided into eight equal parts which are held to the core by tongues that fit into and slide along a series of grooves cut into the core. Groups of four contiguous pieces, viz., hemispheres, may be rotated relative to one another around any of three mutually perpendicular axes.

The court reviewed other references cited by CBS, determined the level of ordinary skill in the art of puzzle design, and thoroughly analyzed the differences between the prior art and the claimed subject matter. CBS does not contest any of the underlying findings but directs its attack on the court's ultimate conclusion of nonobviousness.

As a premise to its argument, CBS says the district court "conceded" that making the puzzle in a cubical embodiment would have been obvious in light of Gustafson's teaching of a spherical design. CBS then asserts that the court improperly relied on unexpected advantages or superiority of the cubical design to overcome obviousness. The premise and argument are flawed. What the court said was that while it would have been obvious to consider changing Gustafson's sphere into a cube or other geometric shape, one of ordinary skill in the art would not have recognized the desirability of a subdivided cube capable of tri-axial rotation. The court relied on expert testimony to the effect that Nichols' concept was a breakthrough and represented "a quantum leap from a sphere." The court noted in addition that Gustafson himself considered other solids which would offer more complexity than the sphere and in his search "dismissed the cube as offering only six faces and went on to work with such larger platonic solids as the octahedron, tetradecahedron, and icosahedron" none of which proved satisfactory.

CBS also argues, citing In re Durden, 763 F.2d 1406, 226 USPQ 359 (Fed.Cir.1985), that method claims 3 through 5 are invalid because they apply an old method (Gustafson) to an obvious object (a cube). It is well worth repeating what we said in Durden: "What we or our predecessors may have said in discussing different fact situations is not to be taken as having universal application." 763 F.2d at 1410, 226 USPQ at 361. The issue in Durden was "whether a chemical process, otherwise obvious, is patentable because either or both the specific starting material employed and the product obtained are novel and unobvious" and that issue was resolved only as a factual matter. Here, the trial court found after reviewing all the evidence of record that nothing in the prior art suggested the combination of a subdivided cube with the rotation of pieces to achieve a preselected pattern. CBS has not shown error in that determination.

In sum, we have considered each of CBS' arguments and the evidence relied upon by the district court, and conclude that CBS has shown neither legal error in the court's determination on nonobviousness nor clear error in the court's probative findings underlying that determination. Atlas Powder Co. v. E.I. DuPont De Nemours & Co., 750 F.2d 1569, 1573, 224 USPQ 409, 411 (Fed.Cir.1984).

CBS raises related issues of utility under 35 U.S.C. Sec. 101 and enablement under 35 U.S.C. Sec. 112. We review utility as a question of fact and enablement as a question of law. Raytheon Co. v. Roper Corp., 724 F.2d 951, 956, 960 n. 6, 220 USPQ 592, 596, 599 n. 6 (Fed.Cir.1983), cert. denied, --- U.S. ----, 105 S.Ct. 127, 83 L.Ed.2d 69 (1984); Cross v. Iizuka, 753 F.2d 1040, 1044 n. 7, 224 USPQ 739, 742 n. 7 (Fed.Cir.1985).

CBS contends that the district court erred in refusing to hold claims 3 through 5 invalid because what the claims define "does not work--a cube puzzle randomized by rotation about three axes cannot be solved by rotation about just one or two axes." CBS' allegation of inoperativeness relies on a "plain" reading of the claims, i.e., that claim 3 defines a method involving rotation about only two axes (step (b)'s "rotating a first set of cube pieces ... about a first axis", step (c)'s "rotating a second set ... about a second axis", and step (d)'s "repeating steps (b) and (c) until the preselected pattern is achieved"); that dependent claim 4 defines rotation about only "one" of three axes; and that dependent claim 5 defines rotation in "multiples of 90?." CBS' argument effectively rests on whether the district court erred in construing the claims. Thus, we must first determine whether the district court erred in interpreting claim 3 as not limited to rotation about only two axes. The court said:

It is plain from the specifications of the '201 patent that the user is to rotate sets of cube pieces about three axes in order to solve the puzzle. Not only are three axes an inherent feature of the 2 X 2 X 2 cube described in the specifications, but there are explicit references to these three axes or the three corresponding coordinate planes in every part of the patent, from the abstract to the drawings....

Claim 3 clearly can be read to contemplate rotation about three axes. By using the indefinite article "a" (i.e. "a first axis," "a second axis"), it is left open to the user, when he comes to the third step, and is directed to repeat the first two steps, to go back to the first axis, or alternatively to select another axis instead, depending upon his solution strategy. This seems to be a reasonable interpretation of the words of the claim, and is clearly consonant with the implications of the rest of the patent and the intent of the patentee.

594 F.Supp. at 1430 (footnote omitted).

The reasoning is sound. This is not a case of a trial court improperly "reading" a limitation found in the specification into the claim. Rather, the trial court correctly ascertained the true meaning of the claim by interpreting the claim language (e.g., "a first axis," "a second axis") in light of the specification and the patent as a whole.4 Loctite Corp. v. Ultraseal Ltd., 781 F.2d 861, 867, 228 USPQ 90, 93 (Fed.Cir.1986); Fromson v. Advance Offset Plate, Inc., 720 F.2d 1565, 1569-70, 219 USPQ 1137, 1140 (Fed.Cir.1983). Accordingly, the inoperativeness argument must fail.

CBS also argues that the subject matter of the claims are not "useful" because the claims do not "teach" anyone the complicated method of solving Nichols' or Rubik's puzzle. The argument misperceives the purpose of a claim. The claims are directed to "a method for restoring a preselected pattern." They claim a general approach for solving the puzzle. As the district court correctly observed, neither the claims nor the disclosure need set forth a particular series of moves to solve the puzzle. Not only do the series of moves for restoring depend on how the preselected pattern was randomized, but there may be more than just one sequence of steps that will restore the preselected pattern. Manifestly, CBS did not come close to meeting the burden of a non-utility defense by proving total incapacity with facts supported by clear and convincing evidence. See Envirotech Corp. v. Al George, Inc., 730 F.2d 753, 762, 221 USPQ 473, 480 (Fed.Cir.1984). Nor has CBS shown that the claims fail for lack of an enabling disclosure, i.e., one that enables a person of ordinary skill in the art to use the claimed methods. See Atlas Powder Co. v. E.I. DuPont De Nemours & Co., 750 F.2d 1569, 1576, 224 USPQ 409, 413 (Fed.Cir.1984).

Literal infringement, i.e., whether properly interpreted claims read on the accused product or method, is a question of fact and reviewable under a "clearly erroneous" standard. ACS Hospital Systems, Inc. v. Montefiore Hospital, 732 F.2d 1572, 1582, 221 USPQ 929, 936 (Fed.Cir.1984). Claim interpretation, a threshold inquiry when resolving infringement, is a question of law. Loctite Corp. v. Ultraseal Ltd., 781 F.2d 861, 866, 228 USPQ 90, 93 (Fed.Cir.1986).

First, CBS argues that this court should determine the issue of infringement de novo because Moleculon's only witness to testify in regard to the claims disclaimed any expertise in interpreting claims and because Moleculon represented that its experts were not offered for the purpose of applying the claims to the Rubik puzzles. The argument borders on the frivolous. We have never required a party to proffer expert testimony on claim interpretation or on application of claim language to accused devices.

Next, CBS faults the district court for not mentioning in its opinion a Citation of Prior Art filed by Moleculon in the U.S. Patent and Trademark Office (PTO). Filed pursuant to 35 U.S.C. Sec. 3015 one day before Moleculon filed the present suit, the Citation identified the Gustafson patent and distinguished it from the claimed invention. CBS urges that those statements distinguishing Gustafson from the claimed invention also distinguish the Rubik puzzles from the claims in suit. Moleculon answers that the district court did not refer to the Citation because CBS did not argue it to the trial court. In any event, says Moleculon, the statements in the Citation have no limiting effect on the scope of the claims because prosecution history estoppel "has absolutely no application to the facts here" since the statements were not made in order to obtain issuance of claims, nor to amend claims, nor submitted as part of a reexamination proceeding.

Moleculon's contention that prosecution history estoppel has no application in this case does not solve the problem. While it is true that the effect of prosecution history arises as an estoppel when applying infringement analysis under the doctrine of equivalents, Loctite Corp. v. Ultraseal Ltd., 781 F.2d at 870, 228 USPQ at 96, the prosecution history can and should, where relevant, be assessed (along with, e.g., claim language and specification) in properly interpreting claim language. Id. at 867, 228 USPQ at 93; Lemelson v. United States, 752 F.2d 1538, 1550, 224 USPQ 526, 532-33 (Fed.Cir.1985). Thus the question is whether a Citation of Prior Art made by the patentee under section 301 may be considered by the court in interpreting claim language.

The statute plainly says: "[T]he citation of such prior art and the explanation thereof will become a part of the official file of the patent." A citation may be made at "any time" either during prosecution or, as here, after the patent has issued. If made during prosecution, it is clear that the statements may be considered for claim interpretation purposes, just as any other document submitted during prosecution. If submitted after issuance, the answer, again, is it may be considered. To say that it may be considered is not to say what weight statements in the Citation are to be accorded. For example, a Citation filed during litigation might very well contain merely self-serving statements which likely would be accorded no more weight than testimony of an interested witness or argument of counsel. Issues of evidentiary weight are resolved on the circumstances of each case.

Having considered the statements appearing in Moleculon's Citation, we conclude that they give no additional light to our task of interpreting claims 3-5 and 9 and therefore they will be ignored.

The district court, in construing method claims 3, 4, and 5 said (emphasis added):

Claim 3, and its dependent claims, 4 and 5, read on the 2 X 2 X 2 form of the puzzle, as well as on such larger cubic embodiments as the 3 X 3 X 3 and the 4 X 4 X 4, because it employs the open transitional phrase "which comprises." The term "comprising" denotes a patent claim as being "open," meaning that the recitation of structure in the claim is open to additional structural elements not explicitly mentioned.

CBS argues that claim 3 is limited to a method for solving a composite cube of eight cube pieces, i.e., a 2 X 2 X 2 puzzle.

Neither party challenges the general proposition that an accused method does not avoid literally infringing a method claim having the transitional phrase "which comprises" (or "comprising") simply because it employs additional steps. See A.B. Dick Co. v. Burroughs Corp., 713 F.2d 700, 703, 218 USPQ 965, 967 (Fed.Cir.1983), cert. denied, 464 U.S. 1042, 104 S.Ct. 707, 79 L.Ed.2d 171 (1984). CBS contends, however, that infringement is avoided when a method claim's step is "missing" from the accused method. E.g., Amstar Corp. v. Envirotech Corp. 730 F.2d 1476 (Fed.Cir.), cert. denied, --- U.S. ----, 105 S.Ct. 306, 83 L.Ed.2d 240 (1984). But here the step is "missing" only in the sense that it is different. CBS says that in solving the 3 X 3 X 3 Rubik's Cube or the 4 X 4 X 4 Rubik's Revenge there is no step of "engaging eight cube pieces as a composite cube" as found in claim 3. Moleculon counters that using the transitional phrase "which comprises" not only opens the claim to additional steps, but also opens the claim and its individual method steps to additional structural elements.6 Moleculon's statement is far too broad, as an examination of step (a) of claim 3 shows.7 That step ("engaging eight cube pieces as a composite cube") contains a structural recitation ("eight cube pieces"). The question before us is not whether, because of the transitional phrase "which comprises," the step reads on a step of an accused method which recites additional structure (hypothetically, "eight cube pieces and six metal plates"), but rather whether this step which recites engaging "eight cube pieces as a composite cube" reads on a step which engages more than eight cube pieces as in the 3 X 3 X 3 Rubik's Cube or the 4 X 4 X 4 Rubik's Revenge.

The transitional phrase, which joins the preamble of a claim with the body of a claim, is a term of art and as such affects the legal scope of a claim. While a transitional term such as "comprising" or, as in the present case, "which comprises," does not exclude additional unrecited elements, or steps (in the case of a method claim), we conclude that the transitional phrase does not, in the present case, affect the scope of the particular structure recited within the method claim's step.

We note decisions where structural recitation in a method claim step was construed as a limitation on the claim. See Austin Powder Co. v. Atlas Powder Co., 568 F.Supp. 1294, 1316 (D.Del.1983); Laminex, Inc. v. Fritz, 389 F.Supp. 369, 373-74 (N.D.Ill.1974). Whether structural recitation limits a claim depends on the language of the claim, the specification, prosecution history, and other claims. The district court erred, therefore, in using the transitional phrase "which comprises" to expand the scope of the recited "eight cube pieces".

Based on our review of the patent, we hold that claim 3 is limited, at least for purposes of literal infringement, to a method for restoring a 2 X 2 X 2 composite cube. That interpretation follows not only from the language of step (a), but also from the way the eight cube piece composite cube is manipulated in steps (b), (c), and (d). These steps are limited to manipulating sets of four cubes only.8

The district court's erroneous interpretation of claims 3-5 renders clearly erroneous its finding that the method of solving the 3 X 3 X 3 Rubik's Cube and 4 X 4 X 4 Rubik's Revenge literally infringe those claims. Accordingly, we vacate the finding. The district court made no finding on whether solving the 3 X 3 X 3 Rubik's Cube or the 4 X 4 X 4 Rubik's Revenge would infringe claims 3-5 under the doctrine of equivalents. The question there is whether the methods for restoring the Rubik's puzzles with their structural differences and the claimed invention perform substantially the same function in substantially the same way to give substantially the same result. Atlas Powder Co. v. E.I. DuPont de Nemours & Co., 750 F.2d 1569, 1579, 224 USPQ 409, 416 (Fed.Cir.1984). On remand the court should address this question. See Loctite Corp. v. Ultraseal Ltd., 781 F.2d 861, 869-70, 228 USPQ 90, 95-96 (Fed.Cir.1986).

CBS appears to argue, as it did below, that even its 2 X 2 X 2 pocket cubes do not literally infringe the method claims because the pocket cubes are not composite cubes consisting of "eight cube pieces" but instead are partial cubes whose internal faces are cut away to accommodate an internal engaging mechanism. We reject the argument. The Nichols patent as a whole shows that the word "cube" is not limited to geometrically true cubes but refers to the cubelet shape as perceived by the puzzle user.

One more issue remains. Method claims 3-5 can be infringed only by a puzzle user. Thus, Moleculon's claim is one for inducing infringement under 35 U.S.C. Sec. 271(b).9 CBS argues that it cannot be liable for inducing infringement of claims 3-5 because there is no evidence of direct infringement of the method claims. The district court held that Moleculon had met its burden of showing infringement under section 271(b) with circumstantial evidence of extensive puzzle sales, dissemination of an instruction sheet teaching the method of restoring the preselected pattern with each puzzle, and the availability of a solution booklet on how to solve the puzzle.

If CBS is arguing that proof of inducing infringement or direct infringement requires direct, as opposed to circumstantial evidence, we must disagree. It is hornbook law that direct evidence of a fact is not necessary. "Circumstantial evidence is not only sufficient, but may also be more certain, satisfying and persuasive than direct evidence." Michalic v. Cleveland Tankers, Inc., 364 U.S. 325, 330, 81 S.Ct. 6, 10, 5 L.Ed.2d 20 (1960). We see no clear error in the court's finding of infringement under section 271(b) with regard to the 2 X 2 X 2 pocket cube versions.

Moleculon acknowledged during trial and the district court held that claim 9 was limited to a 2 X 2 X 2 puzzle form. The question presented to the court was whether Rubik's 2 X 2 X 2 pocket cubes infringed the claim. The district court found that each of the pocket cubes literally infringed claim 9. Of the three claim interpretation arguments presented to and rejected by the lower court, only one is raised on appeal--namely, that neither of Rubik's pocket cubes infringes claim 9 because the pocket cubes' pieces are not "releasably" maintained in a composite cube. According to CBS, "releasably" maintained means capable of disengagement or separation during normal puzzle manipulation, as would be possible for Nichols' embodiment using magnets to hold the cubelets together.

The district court determined that "in the context of the patent as a whole, it is far more reasonable to read the phrase 'releasably maintaining' as referring to a means of holding the puzzle pieces in assembled relationship while permitting relative rotation of groups of pieces." We agree. As noted by the court, the '201 patent's specification states that "the invention includes such devices wherein the engagement is provided by mechanical or magnetic means providing structural integrity without restriction of rotational freedom...." Further, in the patent's "Description of Preferred Embodiment" it is pointed out that while disengagement is an inherent feature of magnetically-engaged cubes "it is also possible to achieve engagement by mechanical rather than magnetic means, as for example by using a pop-in snap linkage, or a tongue-in-groove arrangement allowing rotation without disengagement."

We see no error in the court's claim interpretation. Accordingly, we affirm the finding of infringement as to claim 9.

Those parts of the district court's judgment holding that the claimed invention was not in public use or on sale, that claims 3-5 are not invalid for lack of utility or enablement, and that claims 3-5 and 9 are not invalid for obviousness, are affirmed.

That part of the judgment holding the '201 patent infringed by the 3 X 3 X 3 Rubik's Cube and 4 X 4 X 4 Rubik's Revenge is vacated and remanded for proceedings consistent with this opinion.

That part of the judgment holding claims 3-5 and 9 infringed by the 2 X 2 X 2 Rubik's pocket cubes is affirmed.

AFFIRMED IN PART, VACATED IN PART, AND REMANDED.