This article concerns the existence of a segregated order and the functioning of the category of "aborigine" in South-western Australia during the 20th century. Looking beyond the "official" definitions of the aboriginal population, we will analyse the practical modalities of identification, particularly through an analysis of the census and the statistical classifications of the Aborigines. We will show that, in the absence of official documents specifically geared towards the aborigines, attesting to their register in a central file, the techniques concerning their identification resembled pre-industrial forms of recognizing people, based on familiarity and on face-to-face relations. A consequence of this situation is that those people who were locally known to be aborigines, or who had physical traits that were readily identified as being aboriginal, found themselves caught in a coercitive system from which it was difficult to escape, whereas others were able to pass through the identificatory mesh without, in so doing, placing in check a markedly dichotomous system of racial relations.
