In the backdrop of a long-standing debate between Russellians (or direct-reference theorists) and Fregeans (or sense-based theorists) over the semantics of language, Brian Loar (1976) presented a case meant to score a victory for the Fregeans. Recently, this conclusion has met resistance. Epistemic Russellianism (ER) has emerged as a strong new contender. Claiming that the significance of Loar’s puzzle is due to an element of luck (as in Gettier cases), ER concludes that communication requires knowledge of co-reference -and not a (...) match in sense. My aim is two-fold. Firstly, I deploy some defensive moves against recent attacks on Fregeanism on the basis of ER's interpretation of Loar-like puzzles. Secondly, as an offensive manoeuvre, I argue that ER’s knowledge-based view faces the problem posed by cases of lucky communication. These cases show that drawing an epistemic condition on communication is unwarranted. The upshot of my discussion is that the Fregean approach is still the stronger contender as an account of communication. (shrink)