The invocation of large-scale social unities - states, societies, empires, cultures, civilizations - is a long-established and pervasive practice among sociologists, anthropologists, historians, political scientists and so on. This article examines the treatment of such unities as defined or held together by shared understandings and values, and as independent, boundary-maintaining social systems. We argue that both the ideational and the systemic presumptions at work here are dependent on what Foucault calls the figure of man: the first as an inescapable consequence (...) of that figure, the second as a tempting, but by no means necessary, one. Our first major argument concerns the remarkable persistence of concepts, such as ‘culture’, which designate unities that are ideational in character. We use the case study of anthropology to suggest that this is a consequence of the constitutive role of the figure of man within the human sciences. Human scientists and others critical of the stress on sameness resulting from the concern with ideational unities - cultures, ideologies, discourses and so on - as shared across a population, will find it well-nigh impossible either to modify significantly or to jettison altogether such concepts; as long as they rely on some version of the figure of man, scholars are committed irrevocably to the use of these concepts. Our second major argument concerns the conception of society as a systemic unity, a conception which we see as reflecting the influence of the figure of man in the field of governmental reason. In this part of the article we follow Foucault’s argument that the liberal rationality of government leads to a view of social life as traversed by numerous self-regulating spheres of social interaction. However, we dispute his further suggestion that the concept of society itself, as designating a self-regulating sphere of this kind, can be seen as a product of the liberal critique of police. (shrink)