dissenting.
With this ease, the court blesses corporate shell games resulting from organizational gerrymandering and willful ignorance by which one can secure the monopoly of a patent while hiding the best mode of practicing the invention the law expects to be made public in return for its protection. Because I believe this is a perverse interpretation of the law, I dissent.
The best mode requirement arises from the first paragraph of 35 U.S.C. § 112 (1988), which provides that the patent specification “shall set forth the best mode contemplated by the inventor of carrying out his invention.” The best mode inquiry is twofold: first, did the inventor know of a preferred mode or embodiment of the invention; and second, did the inventor disclose that mode sufficiently to allow those skilled in the art to practice it. Chemcast Corp. v. Arco Indus. Corp., 913 F.2d 923, 928, 16 USPQ2d 1033, 1036 (Fed.Cir.1990). The first question is largely subjective, looking to the inventor’s knowledge and belief. The second is more objective, focusing on the scope of the invention and the level of skill in the art. Id. Compliance with the best mode requirement is a factual question, which we review for clear error. Engel Indus. Inc. v. Lockformer Co., 946 F.2d 1528, 1531, 20 USPQ2d 1300, 1302 (Fed.Cir.1991). But this assumes a correct understanding of the relevant law, which we review de novo. See Spectra-Physics, Inc. v. Coherent, Inc., 827 F.2d 1524, 1536, 3 USPQ2d 1737, 1745 (Fed.Cir.1987).
Novopharm says Glaxo did not disclose the best mode of practicing the invention — the azeotropic granulation process it used to formulate ranitidine hydrochloride into pharmaceutical compositions. This defense became relevant only late in the game when, on June 2,1993, Glaxo produced some documents that indicated that it had withheld information about the granulation process. Novopharm moved for summary judgment based on the best mode violation, which the district court denied, Glaxo, Inc. v. Novopharm Ltd., 830 F.Supp. 869 (E.D.N.C.1993), but allowed No-vopharm to present the best mode defense at trial. Novopharm tried to take discovery on the issue, including a deposition of Crookes, the named inventor of the ’431 patent, and Collin, his immediate supervisor. After Glaxo resisted and sought a protective order, the district court denied Novopharm any discovery, and the case proceeded to trial on August 9,1993. At the close of Novopharm’s case-in-chief on its best mode defense, the court decided to rule on that question as a matter of law, no factual issues remaining. Glaxo presented no evidence on the issue. The court held that because Crookes had no personal knowledge of the best mode, there was no requirement that it be disclosed.
Glaxo first suggests that the azeotropic granulation process was not an appropriate candidate for disclosure because it fell outside the scope of the claimed invention. Rather, Glaxo says the process is simply a production technique useful in the formulation of ranitidine hydrochloride into pharmaceutical compositions. It says the process is therefore relevant, if at all, only to the claims of the ’431 patent that cover such compositions. No such claims are at issue here, so Glaxo says the best mode should not be considered.
But the statutory language demands that the patent disclose the best mode of “carrying out” the claimed invention. As the district court recognized, this language encompasses not only modes of making the invention, but of using it as well. See Christianson v. Colt Indus. Operating Corp., 822 F.2d 1544, 1563, 3 USPQ2d 1241, 1255 (Fed.Cir.1987), vacated on other grounds, 486 U.S. 800, 108 S.Ct. 2166, 100 L.Ed.2d 811 (1988). The azeotroping process was the best way to formulate raw ranitidine hydrochloride into pharmaceutical compositions suitable for use as a drug in human patients, the only use Glaxo contemplated for the invention. Glaxo admits the process was not generally known to those skilled in the art. Accordingly, the court could have found that disclosure of the process was required by section 112, so long as the other, subjective, elements of the best mode test were met. Cf. Chemcast, 913 F.2d at 930, 16 USPQ2d at 1038 (“Whether char-aeterizable as ‘manufacturing data,’ ‘customer requirements,’ or even ‘trade secrets,’ information necessary to practice the best mode simply must be disclosed.”).
*1054The best mode inquiry begins with what the inventor knew when he filed his application. This subjective part of the inquiry traditionally rests on the factual question whether the inventor actually contemplated a preferred mode. Chemcast, 913 F.2d at 928, 16 USPQ2d at 1036. But the inquiry is not limited to the inventor’s actual knowledge.
The court believes Texas Instruments, Inc. v. United States International Trade Commission, 871 F.2d 1054, 10 USPQ2d 1257 (Fed.Cir.1989), expressly limited best mode to the inventor’s actual knowledge, but that case includes no such limitation. We merely agreed with the Commission that Texas Instrument’s computer component patents were not invalid for failure to disclose the alleged best mode, a so-called “word boost” feature, because “ ‘[t]he record does not disclose that the applicant knew of or concealed a better mode than he disclosed.’ ” 871 F.2d at 1061, 10 USPQ2d at 1262 (quoting Commission findings).
The problem arises from reading one sentence of the Texas Instruments opinion out of context. We said that “[t]he fact that Texas Instruments may have manufactured a DRAM containing a different or better form of boosting means is not pertinent to whether the specification disclosed” the best mode known to the inventors. Id. The court reads this to parallel the facts of our case— that the employer of the named inventors developed the alleged best mode wholly without the inventors’ knowledge, and that there was no best mode problem because the inventors did not know of it.
But as our opinion in Texas Instruments shows, the inventors did know of the word boost feature, but did not believe that it was part of the best mode. Id. (“[T]hat the inventors themselves did not consider the ‘word boost’ feature to be part of the best mode of their invention refutes any argument that the inventor ‘knew of and concealed a better mode than he disclosed.’ ”). Accordingly, Texas Instruments stands only for the unremarkable proposition that there is no best mode violation where the inventor knew of the alleged mode and did not consider it to be a part of the preferred embodiment. It says nothing about whether specific knowledge of a best mode by a corporate assign-ee/employer may be imputed to the inventor/employee. Nor does any other precedent of this court.
As the district court recognized, “Glaxo, and not Crookes individually, ... both directed the patent prosecution and has enjoyed the monopoly the issued patent provides.” 830 F.Supp. at 881-82. Glaxo employees acted as agents for inventor Crookes during prosecution of the U.S. application. They were also agents for Glaxo, the assignee of the application and later the owner of the ’431 patent. And that is the crucial point: Glaxo, not Crookes, brought this suit against Novopharm for infringement of the patent. Accordingly, Glaxo’s conduct and knowledge during prosecution is important to the resolution of this case; it is not irrelevant, as the court says.
Glaxo says, and the court agrees, that it did not have to disclose the azeotroping process because Crookes did not know of that method of preparing pharmaceutical compositions of Form 2 ranitidine hydrochloride. On the record before us, one wonders how Glaxo could have been sure of what Crookes knew. Brereton, the Glaxo employee charged with initiating the application for the ’431 patent, testified that he did not consult Crookes, the named inventor, at any time. Instead, in accordance with Glaxo’s standard patent policy, Brereton conferred with Crookes’ superiors to obtain all of the information necessary to secure a patent on the invention. It strikes me as incongruous to rely on the inventor’s actual knowledge here if Glaxo indeed thought that knowledge was so insignificant that it did not even merit consultation during preparation of the application.* At the very least, the district court *1055should have given Novopharm the chance at discovery about just what Crookes in fact knew.
But I believe that even absent further discovery the district court could have found a best mode violation in this case. As the district court stated,
[i]t is undisputed, however, Brereton and other officers within Glaxo believed azeo-troping was the best mode of preparing ranitidine hydrochloride for pharmaceutical use, and Glaxo actually utilized this method in the commercial production of ranitidine hydrochloride. These officials within Glaxo made a deliberate choice not to reveal what they believed to be the best mode of making the patented invention, but instead to protect the knowledge as a trade secret.
880 F.Supp. at 881. This recitation suggests that Glaxo set out to isolate Crookes from any knowledge about the azeotroping technique specifically to avoid the best mode disclosure requirements. If true, these circumstances would justify imputing knowledge to Crookes from Brereton and the other Glaxo employees who acted as agents for Crookes during the application process. I would remand to allow the district court to make the necessary factual findings and decide whether to impute that knowledge.
Imputing knowledge to an inventor may be necessary under appropriate circumstances, to protect the public’s “paramount interest in seeing that patent monopolies spring from backgrounds free from fraud or other inequitable conduct.” Precision Instrument Mfg. Co. v. Automotive Maintenance Mach. Co., 324 U.S. 806, 816, 65 S.Ct. 993, 998, 89 L.Ed. 1381 (1945) (endorsing equitable doctrine of unclean hands in patent suits). “The possession and assertion of patent rights are ‘issues of great moment to the public.’ ” Id. at 815, 65 S.Ct. at 998. And the best mode requirement lies at the heart of this public interest. It is a vital part of the statutory quid pro quo that justifies a patent. Amgen, Inc. v. Chugai Pharmaceutical Co., 927 F.2d 1200, 1210, 18 USPQ2d 1016, 1024 (Fed.Cir.1991). In return for a seventeen year monopoly the patentee must disclose his invention to the public. But he must go beyond simply informing the public of the bare outlines of the invention. He must also tell what he believes to be the best embodiment of the invention, and he must do so in a way that allows the public to practice that embodiment. This prevents the inventor from obtaining patent protection with a minimal disclosure that reveals only inferior forms of the invention, while retaining for himself the most advantageous modes of carrying the invention into practice. In re Gay, 309 F.2d 769, 772, 135 USPQ 311, 315 (CCPA 1962). The court’s pinched reading of the best mode requirement surely violates at least the spirit of this rule at the public’s expense.
Imputing an agent’s knowledge to the principal has sound roots in law and equity. An agent’s acts and knowledge can be imputed to the principal when necessary to protect the interests of .others, so long as the acts or knowledge in question fall within the scope of the agent’s authority. Restatement (Second) of Agency, § 261 (principal liable for agent’s fraud within scope of agency), § 274 (knowledge of agent acquiring property for principal imputed to principal); see also American Soc. of Mechanical Engineers, Inc. v. Hydrolevel Corp., 456 U.S. 556, 566, 102 S.Ct. 1935, 1942, 72 L.Ed.2d 330 (1982) (principal liable for antitrust violation based on agent’s fraud within apparent authority). .This precept is firmly rooted in patent law as well, in the traditional doctrine of inequitable conduct, whereby the inventor’s duty to disclose material information to the Patent Office is extended to all those involved in the filing and prosecution of a patent application. See 37 C.F.R. § 1.56 (1994); Fox Indus, v. Structural Preservation Sys., 922 F.2d 801, 804, 17 USPQ2d 1579, 1581 (Fed.Cir.1990).
The fact that Crookes’ agents knew about the process does not by itself justify imputing that knowledge to him. If he really was unfamiliar with the azeotroping process, that unfamiliarity may simply have resulted from the normal division of labor necessary within a large corporate enterprise. But the district court appears to have inferred a darker subtext — that Glaxo may have deliberately *1056set out to screen this inventor from the azeo-troping technique to conceal the process for itself.
The problem is that Glaxo’s version of best mode, which the court now adopts, would allow, if not encourage, employers to isolate their employee/inventors from research directed to finding the most advantageous applications for their inventions, knowledge that the inventors would probably have had but for the employer’s efforts to keep the work secret. As a result, inventors may have only limited perspective on the real value of their inventions, and can accordingly share only this limited perspective with the public. All the while, the employer/assignee will have a view of the big picture, fully aware, through its other employees, of superior modes of practicing the invention. But the assignee will be under no obligation to disclose those modes to the public. This deliberate subversion of the statutory disclosure would deprive the public of the benefits of the best mode of practicing the invention. There is no reason why this court should condone such abuse of the public trust. Cf. Precision Instrument Co., 324 U.S. at 815, 65 S.Ct. at 997-98.
I would hold that if there truly was such a pattern of deliberate concealment of information that would otherwise have been known to the inventor, the knowledge of those who sought to conceal that information and who now attempt to enforce the patent may be imputed to the inventor. The district court can refuse to enforce the patent and should be given the opportunity to do so with a correct understanding of its powers.
The court finds comfort in the regulations requiring that the inventor sign an oath attesting that he has reviewed the application, 37 C.F.R. §§ 1.41(a), 1.51(a)(2) & 1.63(b)(1), (2), reasoning that Glaxo must have at least let Crookes review the application before it was filed. But none of these regulations speaks to the best mode requirement. Nor are we told how Crookes could sign such an oath if he was never consulted before the application was filed. Perhaps the court has hit upon grounds for a charge of ineq*1055uitable conduct against Glaxo that everyone else missed.