We present a new analysis of the positions of holes beneath the calendar ring of the Antikythera mechanism, as measured by Budiselic et al. (2020). We significantly refine their estimate for the number of holes that were present in the full ring. Our $68\%$-credible estimate for this number, taking account of all the data, is $355.24^{ +1.39 }_{ -1.36 }$. If holes adjacent to fractures are removed from the analysis, our estimate becomes $354.08^{ +1.47}_{-1.41}$. A ring of 360 holes is strongly disfavoured, and one of 365 holes is not plausible, given our model assumptions.