I argue that fidelity to the context principle requires us to construe reference as a theoretical relation. This point helps us understand the bearing of Putnam's permutation argument on the idea of a systematic theory of meaning. Notwithstanding objections that have been made against Putnam's deployment of that argument, it shows the reference relation to be indeterminate. But since the indeterminacy of reference arises from a metalinguistic perspective, our ability, as object-language speakers, to talk about the ordinary features of our lives is unaffected.