Existing characterizations of ‘trace’ in the philosophy of the historical sciences agree that traces need to be downstream of the long-past event under investigation. I argue that this misses an important type of trace used in historical reconstructions. Existing characterizations of traces focus on what I propose to call direct traces. What I call circumstantial traces (i) share a common cause with a past event and (ii) allow an inference to said event via an intermediate step. I illustrate the significance of checking the alignment between direct and circumstantial traces in historical reconstructions through a case study from (micro-)palaeontology.