A number of issues remain unresolved in genre theory today, but one of the most central is the question of how best to describe generic norms and variability. While there is clear agreement today that a genre norm should not be characterized simply as an average number of uses, nor as a single type of use, significant discussion remains over whether genre norms are best characterized as ‘sufficiently recurrent’ or as ‘definitively typified’ (Bhatia 2004, Coe et al. 2003, Devitt 2004, Gillaerts & Gotti 2008, Matsuda 2001, Swales 2004). In effect, while the genre regularities presented as ‘typical’ are useful for teaching purposes, they do not authentically represent actual genre use. Actual genre use is instead characterized by substantial, but equally viable, variation from perceived norms. Two questions which arise from these observations are, how much variation is permitted for a norm to still be perceived as such, and how might we measure it ? Addressing these questions, this paper describes the results of a genre-based study of linguistic variation, using a methdology new to genre analysis (Dressen-Hammouda forthcoming). The study examines how five francophone researcher-writers use a disciplinary genre’s primary linguistic features over a period of ten years. A measure of standard deviation is used to compare the authors’ use of the features with norms established in a genre corpus. Measuring standard deviation is seen to be a valuable tool, not only for examining individual shifts along a continuum of expertise over the course of academic careers, but also for feeding collective reflections about the nature of genre norms. Writers with greater disciplinary experience are quantitatively seen to rely more often on the affordances genres provide, surpassing the genre features’ normal spread range and thereby adhering less strictly to the more common conventions that may characterize their junior colleagues’ writing strategies.