The Wittgensteinian philosophy of language games in the service of "religion" is circular and closed from criticism, and this generates the illusion that "religion" is compellingly universal. Also some historians, while attempting to uncover the specific and localized origins of the modern category "religion" in early modern Europe, simultaneously and unwittingly reconstruct the assumed universality of the category by uncritically using the problematized word in a description of societies in which the modern category, according to their own research findings, did not yet exist. These techniques suggest an unacknowledged ideological field lying behind the usages of the category. I argue that "religion" is one of a pair, the other half being "non-religion" or "the secular". The formation of the modern notion of religion needs to be seen initially in relation to privatized Protestant piety and the separation of church and state, whereby an apparently demystified human nature and fully rational modern civil society was made possible. But the application of the religion-secular distinction to an increasing number of colonies has resulted in "religion" becoming a catch-all container for indigenous institutions and practices that impede progress (i.e., impede realization of secular rationality by the natives). Debates about the logic of the religion-secular category can only become fruitful when we analyze the way in which they each construct their other half. However, the way "religion" is institutionalized and investigated in the academy elides its function in the invention and legitimation of the modern Euro-American secular, for it is treated as though it stands alone, either as a category in its own right or as a natural reality corresponding to that category.