The doxography for Anaximander’s account of the rings of the sky gives proportions for them that are discrepant. So a widely accepted hypothesis proposes that, since the circles of the celestial bodies are compared to wheels, we should add the thickness of their ‘rims’ to the measurements for the celestial rings. This paper proposes an entirely different hypothesis which avoids this awkward expedient by suggesting that there is a ‘geometrical’ reading of the numerical data. The discrepancies can then be explained (...) by there being two different methods for calculating the circumference of the circles. (shrink)