"[W]hen we set out to apply... an idealized term, the operative question is not whether actual instances exist in reality to be designated by the new term. Rather, the question is, on what conditions the new idealized concept has any empirical relevance, and how it succeeds in throwing light, both on those situations to which it directly applies, and on those to which it does not. No actual object exemplifies with unlimited precision the mathematician's specification for a 'geometrical point' or 'Euclidean straight line', no real-life material system fully answers to the physical definition of a 'rigid body' or ; no human being conforms unfailingly to the judicial ideal of a 'reasonable man'; nor are there any absolutely perfect instances of the economist's 'free market'. Yet this leaves the explanatory significance of such idealized concepts unaffected and undiminished. Theoretically speaking, indeed, they can be just as revealing in negative as in positive cases. When we consider, for instance, why departs from the physicist's ideal of a 'perfect gas' more strikingly than oxygen, or when we discuss the conditions on which the workings of an can approximate to those of a 'free' market, it is the explanatory fruits of the concepts that matter, not their exemplifications."