Quine says that in radical translation we have two stages, one in which there is real evidence and so in which translation is determinate, and the other in which we cannot translate without depending on the analytical hypotheses and therefore in which translation is indeterminate. I argue that we can do nothing but use our linguistic habits in every stage of translation. Because we must rely on our hypotheses as to what behavior to count as the native's assent or dissent, question, and sentences, translation is in every stage indeterminate. Further because there is indeterminacy as to whether the native linguistically behaves with mind or not, translation is completely indeterminate. In consequence we find that synonymy is, without exception such as stimulus synonymy, semantic synonymy, that reference is inscrutable, that truth is indeterminate, and that belief is indeterminate.