To summarize a post above, his math indeed wasn't very good (or at least very advanced) for a research-level mathematician because he was a physicist, not a mathematician.