The way we record knowledge, and the web of technical, formal, and social practices that surrounds it, inevitably affects the knowledge that we record. The ways we hold knowledge about the past - in handwritten manuscripts, in printed books, in file folders, in databases - shape the kind of stories we tell about that past. In this lively and erudite look at the relation of our information infrastructures to our information, Geoffrey Bowker examines how, over the past two hundred years, information technology has converged with the nature and production of scientific knowledge. His story weaves a path between the social and political work of creating an explicit, indexical memory for science - the making of infrastructures - and the variety of ways we continually reconfigure, lose, and regain the past. At a time when memory is so cheap and its recording is so protean, Bowker reminds us of the centrality of what and how we choose to forget. In 'Memory practices in the sciences' he looks at three 'memory epochs' of the nineteenth, twentieth, and twenty-first centuries and their particular reconstructions and reconfigurations of scientific knowledge.