This theory paper examines an ex post rationale for the patenting of scientific discoveries. Scientists do not know which firms can make use of their discoveries, and firms do not know which scientific discoveries might be useful to them. To bridge this gap, either or both sides need to engage in costly search activities. Patents increase (decrease) dissemination when the search intensity of firms is sufficiently inelastic (elastic), relative to that of scientists. Patents also facilitate the delegation of search activities to the universities' technology transfer offices, which enables efficient specialization. Patenting may be thus become a complement to doing research. (c) 2007 Elsevier B. V. All rights reserved.