TWO “DECISIVE” EXPERIMENTS

Given that Maxwell's “evidence” for his molecular assumptions was not decisive, what sort of evidence would be? For proof of the existence of molecules Maxwell was not demanding experiments making molecules “directly observable,” but only experimental results from which their existence and properties could be inferred with certainty.

For an initially skeptical scientist I choose Friedrich Wilhelm Ostwald, Professor of Physical Chemistry at the University of Leipzig, and winner of the Nobel Prize in chemistry in 1909. In 1896 Ostwald published a paper titled “Emancipation from Scientific Materialism,”^[226] in which he lays out his fundamental objections to atomism.^[227] For our purposes here, the most important is his claim that no way had yet been found to experimentally measure, with any degree of certainty, quantities associated with atoms or molecules; this measurement criterion Ostwald took to be necessary and sufficient for proving the existence of a postulated unobserved entity.

However, in 1908 Ostwald was converted by experiments of two physicists: those of Jean Perrin in 1908 on Brownian motion, which Os­twald claims “justify the most cautious scientists in now speaking of the experimental proof of the atomic nature of matter,” and experiments of J.

Perrin's argument contains two stages. In the first, he offers a general qual­itative causal-eliminative argument from experiments on Brownian motion, the haphazard motion exhibited by small microscopic particles suspended in a liquid. In the 1880s various experiments had been performed, principally by Leon Gouy, to determine whether this observed Brownian motion was caused by forces external or internal to the fluid in which the motion occurred. Gouy examined a range of possible external causes. When these were reduced or eliminated, the Brownian motion continued unabated. Perrin concluded that the motion of the observable Brownian particles is caused internally by their bombardment with unobservable molecules comprising the fluid.^[229]

The second stage of Perrin's argument invoked experimental results that completely convinced Ostwald of the existence of molecules. (The first stage by itself Ostwald would probably have regarded as no more con­vincing than Maxwell's “independent warrant” arguments.) In 1908 Perrin conducted a series of experiments in which he prepared tiny (“Brownian”) particles visible through a microscope, each of which had the same mass and density, and inserted them into a cylinder containing a dilute liquid of known density and temperature. He derived an equation containing terms for the number of microscopic Brownian particles per unit volume at the upper and lower levels of the cylinder, the mass and density of the particles, the density of the liquid, the height of the cylinder, the tem­perature of the liquid, and, most important, Avogadro's number N (the number of molecules in a gram molecular weight of a substance).

the same in all cases, approximately 6 10²³. He concluded from this

that molecules exist.^[230]

J. J. Thomson's argument for the existence of electrons (or “corpuscles,” as he called them) is in important respects parallel to Perrin's arguments for molecules.^[231]

In the first stage of their arguments both physicists cite experiments yielding results that purport to show the existence of the object postulated without providing any measurements of the object's properties. Their argu­ments are of a causal-eliminative type. They begin with an observed phe­nomenon: Brownian motion, in the case of Perrin, and cathode rays, in the case of Thomson, which, it is claimed, given the background information, is likely to have one of several different types of causes that are specified. Then it is asserted that experiments make it very probable that all but one of these causes are eliminated, leaving the hypothesis which postulates the unobservable entity in question as the probable cause of the phenomenon. In the second stage of the argument, experiments yielding other effects of the inferred entity are cited, from which certain magnitudes associated with these causes are derived and experimentally measured to be approx­imately the same in various different types of experiments performed. In neither case did the experimenters make the entities inferred “observable.”

It is not my claim that arguments containing both of these stages are in general necessary to decisively establish claims of the existence of unobservable entities, but only that in the cases in question there were arguments of these types, and they did in fact convince not only the ex­perimenters themselves but also at least some initial skeptics.

What is the difference between these cases in which experiments for the postulated entities are decisive, or considered to be so, and Maxwell's empirical arguments for molecules, which he himself did not characterize as decisive? In all three cases, appeals are made to experiments, and in all three cases numerical values associated with the postulated entities are given. The difference is in the strength of the empirical arguments. Maxwell's causal-eliminative argument for the existence of molecules from experiments on heat makes it probable that bodies contain “parts too small to be observed separately,” which Maxwell called molecules. But it didn't make it probable enough to be decisive since it did not de­cisively preclude other possible causes of heat phenomena. Nor was his inductive argument decisive from the success of dynamical principles in other domains to their applicability to the inferred molecules, since, as he notes, such principles have been successfully applied only to macrobodies. And although Maxwell gives some theoretical estimates for various mo­lecular speeds, he had no experimental way to verify these calculations. Accordingly, in the 1870s, although Maxwell had some empirical argu­ments for the existence of molecules and for assumptions he was making about them, even if these arguments provided a reasonable basis for be­lieving the theory true, they were by no means conclusive, or regarded by Maxwell as so.

Had he lived then, what Maxwell might have said in response to Ostwald's late conversion to the molecular theory in 1908 is this: True, in the 1870s “methodized experiments” that provide a “strict demonstra­tion” of molecular theory did not exist. Nevertheless, in accordance with the “method of physical speculation,” epistemic arguments were given that furnished a reasonable basis for believing the central assumptions of the theory, and in addition nonepistemic ones were presented showing how the theory can be developed theoretically. One doesn't need “strict demonstration” to accomplish these purposes.