IS THIS REALLY A NEW METHOD?

Is Maxwell correct in claiming that there are genuine differences between his method and more standard ones mentioned earlier? It is clearly dif­ferent from the “method of hypothesis,” as formulated by Maxwell, since the latter, unlike the former, requires no independent warrant at all for its hypotheses.

There are two important differences between Maxwell's method and those of Newton, Mill, and Whewell. One pertains to “theoretical devel­opment.” To be sure, the methods of Newton, Mill, and Whewell involve producing derivations of observable phenomena from the basic assump­tions. And Whewell, like Maxwell, emphasizes the idea of developing a theory over time by adding new assumptions in response to phenomena not yet explained. However, Maxwell is also concerned, very importantly, with developing new theoretical assumptions about the unobservable en­tities postulated, whether or not those assumptions are actually employed in explaining observable phenomena or are even capable of being verified at the time. And unlike Whewell's idea of “coherence,” which is an epistemic criterion supposed to guarantee the highest measure of justified belief, Maxwell's “theoretical development” idea does not guarantee any measure of justified belief but nevertheless contributes to a defense of the theory by exhibiting nonepistemic virtues of the theory.

A second difference between Maxwell's method and those of Newton, Mill, and Whewell is that the latter, but not the former, are based on the idea that inference to the truth of a scientific proposition or theory re­quires proof, which these methods are designed by their proponents to enable scientists to provide.

Is there a difference between Maxwell's method and those of Newton, Mill, and Whewell over the types of epistemic arguments that can be employed in defense of a theory? Maxwell is clearly denying Whewell's claim that “consilience” is sufficient for inference; independent warrant— warrant other than the explanatory and predictive success of the theory— is also necessary. This is something with which Newton and Mill would agree. (It is Mill's first step in his “deductive method.”) The relevant differ­ence here is over the strength of the arguments required, not over types. Unlike Newton and Mill, Maxwell has in mind cases in which none of the arguments, individually or collectively, suffice to prove the assumptions. Why do they fail to do so?

Recall just the two empirical arguments Maxwell gives for assuming that molecules exist and that they obey laws of dynamics. The first is a causal-eliminative argument from the theory of heat (in his 1871 book), which starts with the claim that experiments show that heat is a form of energy, not a substance; then it moves to the claim that it must be kinetic energy rather than potential, since observations show that heat radiation doesn't depend on relative positions of bodies (which potential energy does); then, since hot bodies do not necessarily exhibit observable motion, it concludes that the motion must be that of parts of the body too small to observe, parts Maxwell will call molecules.

The second empirical argument Maxwell offers is one for supposing that molecules in motion obey laws of dynamics. The argument is simply an inductive generalization from the fact that such laws have been suc­cessful in astronomy and electrical science. In the absence of conflicting information, although this gives some reason for supposing these laws hold for domains both large and small, it is by no means decisive, since the phenomena in the domains cited are so different; and, of course, its conclusion was abandoned in the twentieth century with the advent of quantum mechanics.

What Maxwell is saying is that despite the lack of certainty in such cases, we provide what empirical and methodological arguments we can. Furthermore, he is saying, we don't need to base our belief in the as­sumptions of a theory on such “independent warrant” arguments alone but on these together with the fact that the assumptions can be used to explain known laws and deviations. And he is saying that a theory can be defended not only on epistemic grounds but also on nonepistemic ones, including the precision and completeness of its theoretical development.

5.