Historical scientists appear to work under conditions of evidential scarcity: the traces of important past events are frequently lost to deep time. This leads some philosophers and scientists to a kind of epistemic skepticism about historical science. I explore three ways that relations between explanations can themselves count as evidence, focusing on narrative explanation (i.e. explanations that target particular events, entities or processes, as opposed to regularities). These ways are (1) common cause explanations, where preference is given to the hypothesis which unites the most traces in a common history; (2) common process explanations, which unite a group of narratives as analogues, as instances of the same process; (3) interdependent explanations, wherein different explanations of hypotheses are confirmationally coupled (evidence for one boosts the other, and vice-versa). My conclusion is two-pronged.  First, ‘common cause’ explanations, which many people think are essential to ‘palaeoepistemology’, do not play as prominent a role as previously thought, and only do so in collaboration with background theory. Second, the explanatory structures of narratives contain both common process and interdependent explanations, which undermines epistemic skepticism.