WHY ARGUE FOR MOLECULES IN 1908?
Suppose we agree that Perrin's reasoning has been adequately represented and that it is not circular. Why in 1908, eleven years after the discovery of the electron, let alone in 1913, when his book appeared, should Perrin have thought it necessary or even useful to present an argument for the existence of molecules?
To begin with, extending into the first decade of the twentieth century, serious opposition to any atomic-molecular theory had been expressed by some physicists and chemists.
French and German positivists, including Mach, Duhem, and Poincare, for whom unobservable entities underlying the observed phenomena were anathema, rejected any realist interpretation of atomic theory. At best, atoms and molecules, if invoked at all, were to be construed simply as instrumental, conceptual devices. The German physical chemist Friedrich Wilhelm Ostwald until at least 1908 rejected atomic theory in favor of the doctrine of “energetics”. His grounds for doing so were partly philosophical (a repudiation of any form of unverifiable materialism) and partly based on scientific reasons (including the belief that atomic theories, being purely mechanical, should always entail reversible processes, something incompatible with observed thermodynamic phenomena). Indeed, in the preface to his book Atoms, Perrin explicitly mentions Ostwald's rejection of hypotheses about unobserved atomic structure. He describes Ostwald as advocating the “inductive method,” which he (Perrin) takes to be concerned only with inferring what is observable from what is observed. By contrast, Perrin is employing what he calls the “intuitive method,” which attempts “to explain the complications of the visible in terms of invisible simplicity" (1990, p. vii; his italics).Although Ostwald rejected atomic theories well into the first decade of the twentieth century, he changed his views by 1909, as a result of the work of Thomson and Perrin.[163] At the end of his 1913 book Perrin refers to the recent controversy over atomic theory and boldly claims that as a result of his (and other) determinations of Avogadro's number “the atomic theory has triumphed.
Its opponents, which until recently were numerous, have been convinced and have abandoned one after the other the sceptical position that was for a long time legitimate and no doubt useful” (Perrin 1990, p. 216). Perrin was very conscious of the controversies over atoms extending into the twentieth century and felt the need to settle the issue on the side of the atomists.[164]As noted earlier, independently of his 1908 experiments, Perrin believed that available information provided at least some support for the existence of atoms and molecules. Yet he regarded his own experimental results as particularly important in this connection. Why?
Although he considered previous arguments to be supportive, he believed that they did not provide sufficiently direct evidence for molecules. Interestingly, in 1901 he regarded the evidence for the existence of electrons—evidence he himself had helped to develop (see Achinstein 1991, essay 10)—to be more direct than that for molecules: “It is remarkable that the existence of these corpuscles [following Thomson, Perrin used this term for electrons], thanks to the strong electric charges which they carry, is demonstrated in a more direct manner than that of atoms or molecules, which are much larger” (Perrin 1901, p. 460; quoted in Nye 1972, pp. 83-84). In 1901, although neither molecules nor electrons were visible as discrete particles observable with a microscope, the effects of electrons were more directly observable than those of molecules. Cathode rays, that is, streams of negatively charged electrons produced in cathode tubes, were observed to produce fluorescence in the glass of the tube as well as on zinc sulfide screens and to be deflected by magnetic and electric fields. Neutral molecules and atoms were not known to have these or analogous observable effects. Prior to his experiments, or at least prior to the study of Brownian motion, Perrin regarded the evidence for molecules to be less direct, based as it was in chemistry on the regularities of chemical composition and proportion, and in physics, especially in the kinetic theory of gases developed by Maxwell, on phenomena of heat transfer, on the success of mechanical theories in general, and on the ability of chemical theory as well as the kinetic theory to explain a range of observable phenomena.[165] For Perrin, Brownian motion was for molecules what cathode rays were for electrons. Both phenomena provided a relatively direct link between the postulated entities and their observable effects.
Second, Perrin regarded his evidence for molecules as providing more precise and certain quantitative information about molecules than was previously available. He considered his determination of Avogadro’s number and of the masses and diameters of molecules and atoms to be more accurate than previous estimates. He writes, “This same equation [one corresponding to eq. (9)] affords a means for determining the constant N, and the constants depending on it, which is, it appears, capable of an unlimited precision. The preparation of a uniform emulsion and the determination of the magnitudes other than N which enter into the equation can in reality be pushed to whatever degree of perfection [is] desired. It is simply a question of patience and time; nothing limits a priori the accuracy of the results, and the mass of the atom can be obtained, if desired, with the same precision as the mass of Earth” (Perrin 1984, pp. 555-556).
Finally, Perrin regarded his experimental results on Brownian motion as important in the confirmation of atomic theory for another reason as well. These results included not only a determination of Avogadro's number from law-of-atmosphere experiments—which has been the focus of attention in this article. They also included such a determination from Einstein's theory of Brownian motion, which from kinetic theory assumptions generates a formula relating the displacement of Brownian particles to N. Perrin considered Einstein's theory crucial in providing what he called a “mechanism” by which an equilibrium is reached in molecular situations such as those governed by the law of atmospheres. Prior to his 1908 experiments Perrin considered Einstein's theory to be experimentally unverified.
Let us return now to the probabilistic reconstruction of Perrin's reasoning from his law-of-atmosphere experiments. We formulated his major experimental result as proposition C; a theoretical claim for which C is supposed to provide evidence is theoretical assumption T (see section 4).
Perrin believed that T's probability is increased by establishing C; that is,p(T/C&b) > p(T/b). (iv)
Indeed, because of two of the facts noted above concerning the evidence reported in C—its directness and precision—by contrast to other evidence for molecules contained in b, Perrin believed that C gave a substantial boost to the probability of T If this is right, then for those who adopt account (11) of evidence, according to which increase in probability is sufficient for evidence, the bigger the increase, the stronger the evidence, Perrin's experimental result C provided substantial evidence for the theoretical claim T More precisely, on this account of evidence, C and b together count as stronger evidence for T than b by itself if and only if claim (iv) obtains.[166] So on this view of evidence, we can understand at least one reason why Perrin regarded his experimental result C as important. Not only did it, together with b, provide evidence for T, but also it provided stronger evidence for T than b alone, that is, than information available before Perrin's experiments.
Matters are not so simple on the account of evidence represented in (PE). Although the latter sanctions the conclusion that, given b, Perrin's experimental result C is evidence for theoretical claim T, the question of the strength of that evidence is not settled by (PE). Nevertheless, at least this much can be said. If, in accordance with (PE) both e1 and e2 are evidence for h, given b, and if e1 reports a higher frequency of the property in question than does e2, or if e1 contains a larger, more varied, or more precisely described sample than e2, or if it describes items more directly associated with those in h than does e2, and so on, then e1, is stronger evidence for h than e2 in one or more of these respects. In Perrin's case, I have argued, definition (PE) is satisfied. His experimental result C counts as evidence for T, given the background information b in question. It can also be argued that b itself contains evidence for T But C is stronger evidence for T than b in several respects, including precision and directness. This is among the reasons why Perrin believed his evidential claim was worth making in 1908.
6.