The Epicureans held that perceptions serve as a foundation of scientific inference and, further, that all perceptions are true. This is a unique position among ancient philosophers, and it provoked vigorous attacks. Lucretius defended the Epicurean position with an image: just as a building will collapse if the initial measuring rod is crooked, so reason will collapse if it starts out with false perceptions (DRN 4.513-21). This chapter considers perceptions within the context of Epicurus' methodology as a whole. Epicurus proposed two basic rules of investigation: a demand for initial concepts as a means of formulating problems; and a demand for perceptions and feelings as a means of inferring what is not observed. He discussed the rules in a book called Kanōn, literally, a 'straight rod' or 'measuring stick'. The subject, called 'canonic', was considered by the Epicureans to be an adjunct of physics rather than a separate part of philosophy. Canonic is tied to physics in two ways: first, the rules of investigation serve as a preface (or 'approach', DL 10.30) to the physics, and, second, the rules are defended on the basis of the ensuing theory. The Epicurean spokesman in Cicero's De Finibus (1.64) sums up this procedure: 'Unless the nature of things is recognized, we shall not be able in any way to defend the judgments of the senses.' Our sources tend to mingle the pre-theoretical understanding of the rules with their subsequent justification. It is important, however, to keep the two kinds of understanding separate.