In his (1980), Hartry Field argues that good explanations of physical phenomena are "intrinsic explanations." Roughly, an intrinsic explanation of some phenomenon is one that invokes objects that are causally relevant to the phenomenon to be explained. For instance, an explanation of the structure of spacetime that appeals to spacetime points and the relations they stand in is an intrinsic explanation, while one that appeals to causally irrelevant entities like numbers is an extrinsic explanation. More carefully, let us say that a predicate F is an intrinsic predicate iff whether F(x1, …, xn) obtains does not depend on anything other than x1, …, xn, and the relations among them. Let us say that a fact is an intrinsic fact iff the predicates it involves are intrinsic predicates. Finally, let us say that an explanation is an intrinsic explanation iff it only involves intrinsic facts and intrinsic predicates. Field argues that his treatment of quantity is able to provide intrinsic explanations of the structure of space, spacetime, and other quantitative properties, as well as intrinsic explanations of why certain numerical representations of quantities (distances, lengths, mass, temperature, etc.) are appropriate or acceptable while others are not.In contrast, Brian Ellis (1960) and(1966) argues that certain quantitative predicates are not intrinsic, 2 and that numerical representations of quantitative features are largely a matter of convention. In a similar vein, Peter Milne (1986) uses arguments like Ellis's to argue that both of Field's claims are false -that Field's account cannot provide intrinsic explanations of either our numerical representations of quantity or the structure of quantity.In this paper, I show where the arguments put forth by Ellis and Milne go wrong, and where they go right. Their arguments that one cannot provide an 1 These notions of "intrinsic" are those employed by Field and Milne. They are not particularly explicit about what they mean by these terms, but the characterizations given above are suggested by the passages in Field (1980, 27-28 and 41-46) and Milne (1986, 344 and 346). (Of course, to make these characterizations completely precise, one needs to spell out the relevant notions of "involves" and "depends.") 2 Ellis does not put it in quite this way. Rather, he suggests that "grounds of convenience" (1966, 82) and "the roles of the various quantities in physical theory" (86) are "the only kind [s] of justification that can be given" (83) for the choice of fundamental scale and fundamental measuring procedure. This entails that certain length relations are, on his view, not intrinsic.