This article provides a longitudinal analysis of leadership. In the first section of the article, the evidence reviewed indicates how, historically, leadership has been significant in various societal arrangements in the co-ordination of actions for collective purposes. Such co-ordination may also be facilitated through self-organisation, except that this mechanism struggles in the face of genetic human pre-dispositions to dominate, and the group-level threshold effects of small numbers. The second section discusses the leadership field's longstanding fixation with heroes. It shows how, in the face of the compelling evidence in the previous section of small group reverse dominance hierarchy and dyarchy, for example, the evidential basis for such heroic individualism is questionable. This second section also discusses distributed leadership. It argues for its abandonment and replacement as an analytical unit of analysis by the idea of leadership configurations comprising combinations of role sets. These claims are illustrated with historical and contemporary examples. The final section of the article argues for evidence-based capabilities as an antidote to heroic leadership. Here, a range of problems and possibilities associated with identifying capabilities for leading learning are discussed along with some implications for future research and theory in the field. (Contains 134 footnotes.)