A great deal has been said in recent years about patent disclosure. But to say that there is a disclosure function in the patent system implies that there is non-disclosure functioning in the patent system as well. For some information to be disclosed in a patent, other information must go undisclosed; for some things to be included, other things must be excluded. In this article I review the surprising number of doctrines that allow and encourage patent applicants to remain silent about aspects of their inventions. I find that some silences in patents are inadvertent, while some are deliberate; some are necessary, while some are strategic. I conclude that a combination of such explicit and tacit silences allows patents to function as boundary objects, that is, as artifacts that have sufficiently definite meaning to be useful in disparate social worlds, but which simultaneously are sufficiently ambiguous to become objects of collaboration between disparate social worlds. Because innovation is known to occur when localized knowledge is transferred across social boundaries, this function of the patent document is critical to its stated purpose, and occurs largely because of its open rhetorical spaces. Thus, rather than fixating on enhanced disclosure, I argue that much of the critical work of the patent system can and should occur in the open rhetorical spaces where patents are silent.