Morgan and his informants' interpretation of Australian social categories as 'marriage classes' has survived in Dumont's (and Viveiros de Castro's) distinction of a 'local' (Dravidian systems) and a 'global' (Australian systems) formula. This paper explains that the 'global formula' is neither a necessary nor an applied device in Australian kin category determination, even when genealogical memory is short and when there is a non-limitation of range in the extension of categories. Instead, a heuristic model, which is called the relational triangle, is proposed. This model depicts the procedure through which Australian people pragmatically determine and extend kin categories. Moreover, it also offers a visualisation of the cognitive schema and processes framing discourse and behaviour in relation to kinship and draws some parallels with Greenberg's hypotheses on markedness in kinship classes.