legal responsibility for the information which this document contains or the use to which this information is subsequently put. Although every step is taken to ensure that the information is as accurate as possible, it is understood that this material is supplied on the basis that there is no legal responsibility for these materials or resulting from the use to which these can or may be put. Note: the telephone and fax numbers given in this guide for addresses outside the United Kingdom are those to be used if you are in that country. If you are telephoning or faxing from another country, we suggest you contact your local telecommunications provider for details of the country code and area code that you should use. Main contents

Forthcoming in Philosophia, vol. 32 (2004). Available online with the Editor's permission.

This paper lies at the edge of the topic of the workshop. We can write down a Π1 1 axiom whose models are precisely the ∈-structures 〈Rα, ∈ ∩R2 α〉 where α> 0 and Rα is the collection of all (pure) sets of rank < α. From this, one can consider the introduction of new axioms concerning the size of α. The question of the grounds for doing so is perhaps the central question of the workshop. But I want to discuss another question which, as I said, arises at the periphery: How do we know that there are structures 〈Rα, ∈ ∩R2 α〉? How do we know that there exist such things as sets and how do we know that, given such things, the axioms we write down are true of them? These seem very primitive questions, but the skepticism implicit in them has deep (and ancient) roots. In particular, they are questions about ideal objects in general, and not just about the actual infinite. I want to explain why I think the questions (as intended) are empty and the skepticism unfounded. 1 I will be expanding the argument of the first part of my paper “Proof and truth: the Platonism of mathematics”[1986a]. 2 The argument in question