The theory of language articulated in Plato's Cratylus anticipates many of the points of the contemporary language theories in the work of Ferdinand de Saussure and Jacques Derrida. The Cratylus is the only dialogue in the canon of Plato's work devoted exclusively to the study of language signs, called by him "names," and I will argue that it contains Plato's primary theory of language. I will also argue that passages of the Cratylus are misread by Derrida. My argument is divided into four parts. The first part is devoted to tracing out the traditionally read metaphysics of language in Plato's Phaedo and placing it in context as the basis for the metaphysics to which and against which Saussure and Derrida respond in their work, Course in General Linguistics and Speech and Phenomena, respectively. Chapter II elucidates Plato's theory of the sign in the synthesis of the Cratylus. My point in this chapter is that the theory is not metaphysical and that it has several features in common with the theories of Saussure and Derrida. Also in this chapter I cite specific passages from the Cratylus which I believe Derrida has misconstrued. The third chapter consists of the thesis and antithesis of the dialogue. The thesis of the dialogue, Cratylus' argument, is that the signifier is bound naturally to what it signifies. The antithesis, Hermogenes' counterargument, is that language signs are radically arbitrary and that each individual speaker can make up his own names for things. The last chapter is an analysis of the praxis of the dialogue in which Socrates tests Cratylus' thesis by means of etymological analysis of words, names. In this section, Socrates demonstrates the instability of language signs and the unreliability of meaning inferred from signs. This is the last persuasion that language signs are conventional, differential, ambigous traces of other signs, not things.