Quine, taking the molecular constitution of matter as a paradigmatic example, offers an account of the relation between theory confirmation and ontology. Elsewhere, he deploys a similar ontological methodology to argue for the existence of mathematical objects. Penelope Maddy considers the atomic/molecular theory in more historical detail. She argues that the actual ontological practices of science display a positivistic demand for “direct observation,” and that fulfillment of this demand allows us to distinguish molecules and other physical objects from mathematical abstracta. However, the confirmation of the atomic/molecular theory and the development of scientists’ ontological attitudes towards atoms was more complicated and subtle than even Maddy supposes. The present paper argues that the history of the theory in fact supports neither Quine’s and Maddy’s accounts of scientific ontology. There was no general demand from scientists to “see” atoms before they were reckoned to be real; but neither did the indispensable appearance of atoms in the best theory of chemical combination suffice to convince scientists of their reality.