INTRODUCTION

In 1908 Jean Perrin conducted a series of experiments on Brownian motion from which he drew two conclusions of particular importance: (1) that molecules exist, and (2) that Avogadro’s number, N, the number of molecules in a substance whose weight in grams equals its molecular weight, is approximately 6 10²³.

clusions were set forth in a series of articles published in 1908 and 1909, the most famous of which is his “Brownian Movement and Molecular Re­ality” (Perrin [1909] 1984). An expanded version of his results appeared four years later in his book Atoms (Perrin [1913] 1990).

In 1926 Perrin received the Nobel Prize in physics primarily for his work on Brownian motion. Despite his considerable success, philosophers and historians of science who read his articles and book should find some of his key arguments puzzling. For one thing, why in 1908, after nineteenth­century successes in the kinetic theory of gases and after the discovery of the electron in 1897 by J. J. Thomson, should Perrin have thought it necessary to argue that molecules exist?^{[CXLV] [CXLVI] Yet argue for this he did.}

A second puzzling fact is that Perrin's argument for the reality of mole­cules seems circular. In brief, from assumptions in kinetic theory involving the existence of molecules, Perrin derives a formula, the “law of atmo­spheres,” that governs a volume of gas. The law relates the number of mol­ecules per unit volume of a gas at a height above some reference plane to Avogadro's number. He then assumes that a slightly modified version of this same law can be applied to the distribution of much larger, micro­scopic particles (Brownian particles) suspended not in a gas but in a fluid. With this assumption he proposes a formula that relates the number of suspended Brownian particles per unit volume at a height above a refer­ence plane to Avogadro's number and to various experimentally measur­able quantities of the visible particles, including their mass, density, and numbers at different heights.

In this article I will examine Perrin's reasoning to see whether it is in fact circular. I believe that it is not, and indeed that it conforms with a valid pat­tern of reasoning frequently used by scientists to infer the existence of “un­observables.” I will show why, even in 1908, it was reasonable for Perrin to employ this pattern of reasoning in arguing for the existence of molecules. Finally, I will discuss the relationship between Perrin's reasoning and the debate between realists and antirealists regarding unobservable entities.

2.