At Theaetetus 201d-202c5 Socrates claims to have dreamt that we can have no knowledge of the simple elements of which we and all other things are composed, since no logos or ‘rational account’ can be given of them. We can, however, give a logos, and thereby acquire knowledge of a complex entity, a logos being understood in a minimal way as a combination of names. I argue that in his presentation and criticism of his ‘dream theory’ Plato systematically employs two (...) different words for knowledge, gnôsis and epistêmê. Whenever Socrates expresses agreement with the theory, he affirms that there can be neither a logos nor any epistêmê or ‘scientific knowledge’ of the simple elements, although we can give a logos and thereby have epistêmê of the larger complexes. But when he expresses disagreement with the theory, he challenges its claim that there can be no gnôsis, no ‘acquaintance with’ or ‘recognition’ of any individual element. Taking as his test case our knowledge of the letters of the alphabet, he concludes that elements in general give us gnôsis that is much clearer than that of the complexes’. I conclude that through the presentation and refutation of Socrates’ dream theory Plato was pointing up the need for perception-based acquaintance with the world, in addition to having propositional knowledge about it. (shrink)