Like most people online, I first encountered references to Mithras in the kind of rather crude atheist polemic that goes, “Jesus is really Mithras! Har har!”. A correspondent has written to me about this, and it turns out that he has been reading into the scholarly literature as I have. An interesting paper by a late Cumontian, Geo Widengren, Reflections on the origin of the Mithraic mysteries, has come my way, and I spent some time yesterday reading this.
It’s becoming clear to me why the old Cumontian view has dropped out of favour. It is, simply, insufficiently rigorous in its analysis and classification of sources. Widengren mixes together a stew of references and sources, consisting of snippets of ancient texts, bits of Iranian literature, and even interviews with Caucasian peasants in the present day, in order to tell a narrative story. But the weakness of this approach is evident — the story does not arise from a careful analysis of the data, but instead the data is stuck onto the story.
Widengren’s paper is quite interesting, but he is handicapped by this eclectic approach that all the Cumontians take. It is, simply, confused and confusing, and dreadfully unsystematic. I learn from it that the first query about the Cumontian approach, and the proposal for separation, came from a volume by S. Wikander, Etudes sur les mysteres de Mithras, I, Lund, 1951. His theory involved a proposal that Mithres and Mithras were distinct deities, labelled as such in the sources, one Roman and one Iranian. Widengren correctly indicates that the manuscripts do not support this, and indeed we would not expect it. But it provoked thought, it seems. It is, after all, much easier to analyse the material once you split it down into distinct areas, Iranian Mithra and Roman Mithras, and treat them as distinct but possibly related. Then you can discuss with some precision all the material and the links. This is not possible under the Cumontian approach.
It is still possible that Roman Mithras was derived somehow from Persian Mitra. But I can see that there will be no going back to the rambling style of Cumont. The way forward must be with fewer generalisations, and much more care and precision.