Quine’s underdetermination thesis is roughly that, for any theory T, there is an empirically equivalent theory T′ such that T and T′ are jointly inconsistent. Thus formulated, however, it’s easy to see that the thesis is false. But we can define a weaker version of the thesis to the effect that, for any theory T, there are theories Tand Tempirically equivalent to T such that Tand Tare jointly inconsistent. The weaker thesis is still interesting because it refutes realism about scientific theories, understood as the thesis that we should have a high degree of belief in any empirically adequate theory. This raises our first question: what notions of empirical equivalence support the weak underdetermination thesis?

Suppose that, on some reasonable notion of empirical equivalence, there is an empirically adequate theory T not satisfying it. Let’s also stipulate that two theories have the same empirical content iff they are empirically equivalent. There’s still the question whether one should believe in more than the empirical content of T. In particular, whether one should believe in the existence of any theoretical entity that postulates. It’s sometimes argued that one should do that because the theoretical terms of T are indispensable for a practical formulation of the empirical content of T.

This claim would be refuted if one could establish a dispensability thesis to the effect that, for any theory T with a practical formulation, there’s an empirically equivalent theory T′ with a practical formulation in purely empirical terms. This raises our second question: what notions of empirical equivalence and practicality support the dispensability thesis? The paper presents partial answers to both questions.