Turner et al. PLoS ONE Empirical and Bayesian approaches to fossil-only divergence times: a study across three reptile clades 1. The NEXUS-formatted phylogenetic datasets and BEAST model run parameters were converted into BEAST’s unique XML format using BEASTmasteR. MCMC analyses were run in BEAST v2.4.2 and confirmed as working in the 2.4.4beta version as well. MCMC analyses were run for twenty million generations, sampling the chain every 10000 generations. 2. The BEASTmasteR code we used for the published runs is included here on Dryad. This, in combination with the settings_v1.xlsx included in each BEAST2 run folder per dataset, can be used to replicate the runthis_v1.xml files also included in each folder. 3. While in review, one helpful reviewer noted a potential inconsistency in how BEAST1 and BEAST2 enforce a user-specified ratematrix to generate ordered discrete characters. This is most apparent in multistate characters with more the 3 states. We wish to flag this potential issue here and draw peoples attention to the beast-users thread we have started regarding the potential inconsistency. (https://groups.google.com/forum/#!topic/beast-users/_PM84UATnhM)