dissenting in part.
I join my colleagues’ thorough opinion in all respects save one, albeit significant, exception. Because the claims lack meaningful limitations on the structure of the erythropoietin-producing cells, I cannot *1359agree that the district court should have abstained from inquiring fully whether the claims were suspect under the enablement and written description provisions of 35 U.S.C. § 112, ¶ 1.
As described by the specifications of the patents in suit, Amgen in 1984 cloned and sequenced the DNA encoding human er-ythropoietin (EPO). Amgen then showed that by introducing the cloned EPO DNA (linked to a promoter sequence) into mammalian cells, those cells could be engineered to express high levels of functional human EPO protein. The parties refer to this as “exogenous DNA” expression of EPO. Amgen obtained several patents that cover the use and manipulation of cloned EPO DNA, and these patents, battle-tested through litigation, have been the foundation of Amgen’s successful business of manufacturing and selling recombinant EPO. But these patents are not in suit here, and TKT’s method for producing EPO does not rely upon manipulation of cloned EPO DNA or “exogenous DNA” expression technology.
The claims in suit here contain no significant limitations as to how the recombinant EPO is expressed, or as to the structure of the EPO-producing cells, so long as the EPO is “non-naturally occurring” or produced in “vertebrate cells.” The central question in this case is therefore whether Amgen’s disclosure of one means of producing synthetic EPO in mammalian cells, namely exogenous DNA expression, entitles it to claim all EPO produced by mammalian cells in culture, or all cultured vertebrate cells that produce EPO. I think this is a question of some importance. Yet it is a question that the district court simply refused to consider. Athough the district court admitted that Amgen’s disclosure was limited to exogenous DNA expression, the district court quite clearly and explicitly refused to decide whether the absence of any exogenous DNA limitations rendered the asserted claims vulnerable to the enablement challenge mounted by TKT under section 112. According to the district court, because the asserted claims were to “compositions” rather than “processes,” “the specification need teach only one mode of making and using a claimed composition.” Amgen, Inc. v. Hoechst Marion Roussel, Inc., 126 F.Supp.2d 69, 160, 57 USPQ2d 1449, 1515 (D.Mass.2001). See also id. at 160, 164 n. 57, 57 USPQ2d at 1516, 1518 n. 57. Likewise, the district court refused to inquire whether the absence of limitations on the means of EPO expression raised questions of compliance with the written description requirement, holding that such an inquiry was irrelevant to composition claims. Id. at 150-51, 57 USPQ2d at 1508.
With respect to the '080 and '422 patents, which claim “non-naturally occurring” EPO and EPO “purified from mammalian cells grown in culture,” the majority, like the district court, essentially passes over the question of whether these limitations — which are essential for patentability of the claims — raise issues of compliance with the enablement and written description requirements of section 112. The majority holds that patentees are free to decorate their composition claims with source and process limitations without any concern for whether the full scope of those limitations is enabled or described, and that these requirements of section 112 are waived so long as the patentee succeeds in characterizing its claims as “product” claims. Competent patent attorneys should be quick to take advantage of the majority’s broad exemption from the disclosure requirements by the appropriate phraseology. Rather than endorse the district court’s elevation of form over substance, I would vacate its decision on these issues regarding the '080 and '422 patents, and remand for further consid*1360eration in light of the vast scope of the claims in suit for which there appears to be insufficient enabling disclosure or written description.
With particular reference to the '349 patent, which claims not EPO polypeptides but the cells that produce them, I think the district court’s abstention from scrutiny under section 112 is even more patent error. The majority focuses on the district court’s findings that the invention could readily be practiced in mammalian or vertebrate cells other than the hamster and monkey cells taught by the specification. I agree that TKT has not shown error in these findings. But, as it did for the EPO claims, the district court simply refused to consider whether the absence of any exogenous DNA limitations raised en-ablement issues, “[bjecause Amgen is only required to enable skilled artisans to make its claimed product by only one method. ...” Id. at 164 n. 57, 57 USPQ2d at 1518 n. 57. For the EPO-secreting cells, the absence of an exogenous DNA limitation is not a failure to limit how the product is made, but a failure to limit the structure of the claimed product itself. A cell, as employed in the patents in suit, is nothing more than a biological machine for making EPO. Even in more predictable arts, one who is first to make a machine is not entitled as a matter of law to claim any or all machines so long as they perform the same function. I would think it uncontroversial that even one who is first to make polymer X or alloy Y cannot obtain a claim as broad as “A machine that makes polymer X,” or “A process that yields alloy Y,” without reciting additional limitations that define the structure of the claimed machine or the steps necessary to carry out the claimed process.
' Yet that is exactly what the district court and the majority allow the '349 patent to achieve. It claims any or all cultured vertebrate cells that can secrete a defined amount of EPO, with only a single limitation on their structure: that they “comprisfe] non-human DNA sequences which control transcription of DNA encoding human erythropoietin,” or that they “comprise transcription control DNA sequences, other than human erythropoietin transcription control sequences, for production of human erythropoietin.” This is little more precise than a recitation of “A machine that makes polymer X, wherein the machine comprises means for controlling how much polymer X is made.” The specification teaches only a single means by which the use of a transcription control sequence can coax a vertebrate cell to secrete EPO: by transforming that cell with an exogenous expression vector on which the transcription control sequence is linked to cloned EPO DNA. Yet the claims leave this essential aspect of the invention undefined. It is black-letter law that claims failing to recite a necessary element of the invention fail for lack of an enabling disclosure, In re Mayhew, 527 F.2d 1229, 1233, 188 USPQ 356, 358 (CCPA 1976), and that disclosure of one or two species may not enable a broad genus under these circumstances. In re Vaeck, 947 F.2d 488, 496, 20 USPQ2d 1438, 1444-45 (Fed.Cir.1991). At the very least, the absence of structural limitations in the '349 patent raises questions of its enablement, and I cannot agree that the district court chose correctly by ignoring those questions altogether. We should vacate the district court’s judgment that the '349 patent passes enablement muster, and require the court to apply the correct law to the plain facts.
I must also disagree with the majority that the district court’s approach was faithful to this court’s articulation of the written description requirement of section 112, as expressed in Regents of the University of California v. Eli Lilly & Co., 119 F.3d 1559, 43 USPQ2d 1398 (Fed.Cir.1997) *1361and Gentry Gallery, Inc. v. Berkline Corp., 134 F.3d 1473, 45 USPQ2d 1498 (Fed.Cir.1998). Eli Lilly articulated two principles of the written description requirement: that in haec verba description of broadly described generic subject matter may not suffice to describe the subject matter of that particular claim, 119 F.3d at 1567, 43 USPQ2d at 1404-05, and that disclosure of a species may not suffice to describe a genus, id. at 1568-69, 43 USPQ2d at 1405-06. The district court followed neither of these principles here, and the majority, dismissing Eli Lilly on the grounds that no undisclosed DNA molecule appears in this case, verges on confining Eli Lilly to its facts.
Nor am I convinced that the district court’s approach was faithful to Gentry Gallery. In Gentry, only those claims that included limitations such as “wherein the control means are located on the console” satisfied the written description requirement. Because the specification failed to disclose any location for the controls other than on the console, those claims that lacked such limitations were invalid under § 112, ¶1. 134 F.3d at 1479-80, 45 USPQ2d at 1503-04. The question here is similar: whether the claims fail the written description requirement for lack of “exogenous DNA” limitations, because the specification discloses only the exogenous DNA technology that was state of the art in 1984.
Even if we ignore the patents’ statement that the claimed EPO molecules are “uniquely characterized by being the product of ... expression ... of exogenous DNA sequences” (which of course we cannot), I think the parallels between this case and Gentry Gallery are inescapable. The invalid claims in Gentry recited elements that could readily be found in the text of the specification (a couch, controls, a console), but those claims nonetheless failed the written description requirement because they included no limitations on how those elements were arranged. Likewise, the '349 claims' — for which I think it must be conceded that structure of the EPO-secreting cell is a relevant question— recite particular elements found in the specification (cells, non-human control sequences, EPO-coding DNA), but do not include limitations on the arrangement of those elements, e.g. that the non-human control sequences and coding DNA are present on an exogenous expression vector in the cell. I agree that as a matter of claim interpretation there is no justification for importing an “exogenous DNA” limitation into the claims. But the absence of such limitations must weigh heavily in the section 112 inquiry, else we hold that claims become more resistant to written description challenges the more broadly drafted they are.
While I share my colleagues’ admiration for the considerable efforts of the district court in this complicated case, I cannot share their faith that the district court properly and conscientiously applied Eli Lilly and Gentry Gallery, when the district court’s opinion is completely devoid of reference either to those cases or to the principles they espouse. If the district court did not focus on the correct law to be applied, then its factual findings merit no deference, and the correct remedy for this omission is to vacate the district court’s judgment on this point and remand for further consideration. Our precedent has little value if the district courts may overlook its certain pertinence, if not its plain applicability.