How can the modalities and whereabouts of power – and specifically the spatio-temporal contingencies and reach of power relations - be more effectively studied? This paper shows how issues of validity and reflexivity restrict existing empirical work’s ability to advance understandings of power, and demonstrates how such issues can be overcome through the refined use of methods and analytical techniques that tease out the double contingencies of the social relations that underlie power. Refinements are shown to be possible by learning, in particular, from approaches to analysing power elsewhere in the social sciences, and particularly from management studies and linguistics.