This paper explores some ways in which a neo-Reichenbachian time-relational framework can be applied to diachronic data. The target language of this study is Vedic Sanskrit, the language of the sacred texts of Hinduism. The main focus of the paper concerns the evolution of the Vedic past tense system, which at the beginning of the Vedic tradition is aspect-based and later develops into a system where temporal remoteness and evidentiality distinctions determine the distribution of the past tense categories. This language therefore offers a particularly intriguing data set for exploring the diachronic relationship between aspect, proximal tense and evidentiality, a field of diachronic semantics which has only received limited attention in the research literature.