Wittgenstein's Tractatus contains a wide range of profound insights into the nature of logic and language – insights which will survive the particular theories of the Tractatus and seem to me to mark definitive and unassailable landmarks in our understanding of some of the deepest questions of philosophy. And yet alongside these insights there is a theory of the nature of the relation between language and reality which appears both to be impossible to work out in detail in a way which is completely satisfactory, and to be bizarre and incredible. I am referring to the so-called logical atomism of the Tractatus. The main outlines of this theory at least are clear and familiar: there are elementary propositions which gain their sense from being models of possible states of affairs; such propositions are configurations of names of simple objects, signifying that those simples are analogously configured; every proposition has its sense through being analysable as a truth-functional compound of elementary propositions, thus deriving its sense from the sense of the elementary propositions when this view is taken in conjunction with the idea that the sense of a proposition is completely specified by specifying its truth-conditions. In this way the Tractatus incorporates in its working out a philosophical system analogous to the classical philosophical systems of Leibniz or Spinoza which are regarded by many people, in a sense rightly, as the prehistoric monsters of philosophy which are not to be studied as living organisms, but studied as the curiosities of human thought. And we may here agree that in the end we must simply reject a philosophy which incorporates such features as its postulation of simple eternal objects, or of a possibility of an analysis of a proposition which was presented as a pre-condition for the propositions that we ordinarily utter to make sense, and yet the specific form of which we are unaware of, and so on.