In light of the genetic evidence about very long-run jati endogamy, what is the best case for the arguments put forth by Bayly and Dirks? I consider Bayly superior, but they both assume, as a matter of course, that caste identities had always been fluid in the pre-modern period without elaboration
