Manders' analysis of Euclid's proofs in the first four books provides a model for understanding how proofs work in practice, as actual mathematical practice diverges from the ideal of a formal proof being an unbroken chain of logical inferences from explicit axioms. The plausibility that inspecting and manipulating mental representations can reliably convey mathematical information depends on drawings sharing the structure of mathematical objects in an inexact way and being systematically related to formal notation and inference through labeling.