Davidson’s paratactic account of indirect speech has it that a natural-language report of an utterance such as Galileo’s supposed one of ‘The Earth moves’ should be understood as analyzable into two separate, and semantically independent, utterances, the first of which points to the second, with the latter meaning in the reporter’s mouth what Galileo’s meant in his. The account rests on the assumption—shared by most writers on the subject, including critics of the account—that the correct natural-language report of Galileo’s utterance is ‘Galileo said that the Earth moves.’ I show that on that assumption the paratactic analysis misfires: the two utterances—Galileo’s and the reporter’s—do not samesay one another. However, this is also the case if the verb in the demonstrated sentence is changed to respect the tense-sequencing rule as does ‘Galileo said that the Earth moved.’ Since the latter does correctly report Galileo, that must be because, contrary to the central claim of the paratactic analysis, its two clauses are not semantically independent.