The ongoing debate about whether Frege has ever abandoned the metalinguistic solution to Frege’s Puzzle not only is important by itself, but has serious ramifications. In this article, I show that Frege did abandon the metalinguistic solution and his objection to it is as follows. Though the co-reference relation between two signs is mediated by the connection of each of the two signs with the same designated thing, the connection is nevertheless arbitrary because it could be established even if one (...) of the two signs is distinguished from the other only as object not as sign. As a result, if by “a = b” we assert a co-reference relation, we would express no proper knowledge. (shrink)