ABSTRACT
Let us now develop the ideas we have acquired from the case of mechanics by applying them to the problem of introducing the concept of temperature. This time no attempt will be made at a historical analysis, but I will confine myself to giving a hypothetical series of theories and tests which would have enabled a precisely measurable concept of temperature to evolve without circularity. Some details of the actual history may be found in Roller (1957). My suggestion is that we begin by proposing the following law: Tor rods or columns of a large number of different materials θ ∝ l, where θ is the temperature of the rod or column and I its length.’ Now the interesting thing about this law is that ‘θ’ and ‘l’ both stand for new concepts. We are not assuming any prior notion of length, for the measurement of length, as Popper says, ‘needs a (rudimentary) theory of heat’ (Popper, 1962, p. 62. Quoted earlier on p. 44). But if both θ and l are new concepts, how can we test the law? The case seems altogether hopeless.
