Harold Garfinkel wrote a series of highly detailed and lengthy 'memos' during his time (1951-53) at Princeton, where remarkable developments in information theory were taking place. These very substantial manuscripts have been edited by Anne Warfield Rawls in Toward a Sociological Theory of Information (Garfinkel 2008). This paper explores some of the implications of these memos, which we suggest are still relevant for the study of 'information' and information theory. Definitional privilege of 'information' as a technical term has been arrogated (...) by information science, which thereby excludes the interactional occasions of use of 'information'. The authors examine some 'professional' and 'laic' determinations of 'information'. Looking at in situ uses of 'information' shows how dealing with 'information' is characterized by ad hoc practices, such as specifications, 'authorization' and 'particularization' procedures. The authors report on a series of workplace studies in academic libraries, looking at how librarians account for 'information' through practices of classification. Classifying 'information' is a member's local accomplishment, and explicating practices of classifying 'information' undermines the formal-analytic project of the 'Philosophy of Information,' as formulated, for instance, by Luciano Floridi. Implications of Garfinkel's work must remain beyond the purview of information science if it is to maintain its status as the recognized field dealing with 'information'. However, such omission risks 'losing the phenomenon' of 'information': to adapt an argument from Dorothy Smith (Catalyst, 8, pp 39—54, 1974), it trades upon decontextualized uses and recontextualizes 'information' for the practical purposes of formal analysis. (shrink)