Reasoning under uncertainty provides plenty of opportunities to err or to be accused of erring even if one is not. One can present a plausible, well-founded hypothesis, which may later be over- turned by new evidence in a totally unexpected way. In this study a drawing as test case was taken by N=259 participants. Our analysis of their solutions showed that deductive partial explanations and eliminations seem to be a good way of trying to find relevant answers to unresolved cases. When an investigation stalls, one should not prematurely be fixed on the full explanation of the story of what presumably happened but make small and coherent advances in the interpretation of the details of the evidence with specific sub- hypotheses. Then again one cannot absolutely trust such deductions, as one does not know how the traces arrived there. Were they caused by the presumed incident or by random events? In a final step we integrated naïve recipes of interpretation into hierarchical levels of epistemology, so as to construct a more sophisticated heuristic.