his pattern has clear implications for the dissemination of research. If industry sponsored materials and methods are shared more selectively, others are less likely to pursue their programs and expand on their experiments. Due to increased ambiguity, experimental duplication not only might be attempted less often, but when attempted it might less likely produce similar results.7 If so, ideas from industry-sponsored articles will spread less frequently beyond the scientists, research organizations, regions, and scientific areas that spawned them, and their resulting networks of scientific influence will be smaller (Latour, 1987). This is consistent with research on patenting and the dissemination of invention. Economists have found that patents initially are cited most within their geographic region (Jaffe et al., 1993) and sponsoring firm (Adams and Jaffe, 1996), although this oncentration lessens over time. Sorensen and Fleming (2004) demonstrate how patents that do not cite academic literature - presumably produced by non-academics - are cited less rapidly than those that do, and the inventors who cite them are less geographically dispersed. Following these findings, I predict that industry collaboration and sponsorship will slow the social, geographic, and scientific spread of ideas.