Quine in his paper "Two Dogmas of Empiricism"¹ has propounded a radical conventionalist thesis, arguing that only science as a whole, including the laws of logic, is empirically testable. Grünbaum, on the other hand, has in various places including Philosophical Problems of Space and Time ² been critical of an even moderate Duhemian conventionalism, and in particular attempts to show that the geometry of space is testable independently of other physical theory. Between these two extremes lies the Duhemian thesis. We will attempt to describe in this physics, indicating the manner in which it is a semantical conventionalism while not being trivially so.