This paper offers an analysis of three historical science novels, John Banville 's Kepler (1981), Tracy Chevalier's Remarkable Creatures (2009), and Clare Dudman's Wegener 's Jigsaw (2003), to address a number of interrelated questions that cover, for example, specific features of the genre, the narrative mode employed as well as its epistemic advantage, and the fictional integration of an emergent scientific discourse. While Banville's Kepler establishes a close affinity between the mathematical, the aesthetic and the narrative form, Chevalier's neoRomantic novel Remarkable Creatures makes full use of alternating first-person narratives to highlight the exceptional friendship of the historical fossil hunters Mary Anning and Elizabeth Philpot at a time when palaeontology was about to become a scientific discipline. The third novel, Clare Dudman's imagined autofiction Wegener's Jigsaw, uses an autodiegetic narrator and a device that I refer to as 'mentor-mentee constellation' to facilitate a narrative transmission which guarantees that Alfred Wegener and his continental drift theory will find favour with the readership when he falls victim to oppressive scientific discourse.