This paper is a methodological exploration of history theses that analyses intra-disciplinary variations using a statistical technique, correspondence analysis. Many previous studies concerning academic discourse variations had to rely on pre-set externally classified genres, disciplines, fields, and so on, without questioning the validity of such pre-existing text-type boundaries. However, a valid method for drawing text-type boundaries in a corpus has not yet been established in academic discourse studies. In particular, intra-disciplinary studies face difficulties due to the fuzziness of official boundaries (i.e., fields and sub-fields) within disciplines. This paper is a methodological attempt at quantifying the variations of move components, without presetting boundaries, in an intra-disciplinary corpus. It explores the potential of the quantification of the co-occurrence patterns of move components to describe variations. The correspondence analysis in this study brought to light patterns of co-occurrence, and visualises trends (patterns associated with moves) in a history thesis corpus. The paper concludes that a corpus itself can show evidence for genres and text types without assuming pre-existing boundaries in the corpus, and that a correspondence analysis enables a statistically-evidenced corpus-driven description of move componential profiles (trends) of a corpus. The paper suggests that this methodology can be applied to analyse the same academic genres of other disciplines and expanded to diachronic research. The findings of the paper suggest that quantifying the variations of move components may reveal many undetected features of genres and text types. The paper concludes that, despite the somewhat widespread belief that corpus linguistics falls short of giving a full account of the dynamism of genre, it can provide a snapshot of a phase of evolving genres and text types, with a refinement in corpus processing and an appropriate utilisation of statistical techniques.