John Stuart Mill on Induction and Hypotheses STRUANJACOBS SINCE ITS APPEARANCE in 1843, John Stuart Mill's great work of epistemology and metascience, A System of Logic, has been closely studied as a philosophical text, but its composition and the revisions Mill made to its several editions have been largely overlooked. Historians of ideas have been derelict in detail- ing changes in Mill's conception of scientific methods. The following study is a contribution to redressing that neglect. According to received opinion, Mill's theory of science is fundamentally inductivist, envisaging basic laws of science as produced and proven by inductive methods? It will here be contended that this was not his mature, settled view. He came to believe that most scientific laws actually originate as hypotheses, and on account of this he had to narrow his conception of induction and the purpose of inductive methods. An interest- ing, but hitherto inevitably obscured, effect of Mill's espousal of the hypotheti- For helpful comments on earlier versions of this text the author is grateful to Mr. Derek Crabtree, Dr. Bruce Langtry, Dr. Robert Orr, Dr. Simon Schaffer, Professor C. L. Ten, and two anonymous readers for this journal. ' As anyone with at least a nodding acquaintance with A System of Logic knows, Mill attributed a hypothetical method to science, besides an inductive method. His hypothetical method is usually seen as an unhappy appendage to the inductive. (Mill also recognized a deductive method, but being parasitic on the inductive, it highlights for readers the depth of his allegiance to in- ductivism.) What is not appreciated is that only after strategic metascientific parts of the Logic had been written did Mill become a proponent of the hypothetical method, which he then reaffirmed through successive editions. For earlier thoughts of mine on this subject, see Struan Jacobs, "From Logic to Liberty," The Canadian Journal of Philosophy 16 (December 1986): 751-63. Some of the many studies which read Mill's metascience synchronically and fail to note his later adoption of the hypothetical method are: Michael St. John Packe, The Life of John Stuart Mill (London: Secker and Warburg, 1954), 256; William Thomas, Mill (Oxford: Oxford University Press, 1985), 52ff.; Edward H. Madden, "John Stuart Mill's System of Logic," in Theories of Scientific Method, ed. R. M. Blake, C. J. Ducasse, and E. H. Madden (Seattle: University of Washington Press, 1966), 218ff.; Alan Ryan,John Stuart Mill (New York: Pantheon Books, x 97o), xi; R. P. Anschutz, The Philosophy of J. S. Mill (Oxford: Clarendon Press, 1969), 97 ft. [69] 70 JOURNAL OF THE HISTORY OF PHILOSOPHY 29:1 JANUARY 1991 cal method -- to be described later in the paper -- is that his methodology and that of William Whewell, though commonly interpreted as disjoint, actually share important ground. I . Before examining Mill's account of the methods of science we should briefly describe the aim he sees them serving. He rejects, in the first place, the ideal of certainty that Whewell and other intuitionists affirmed. There are, Whewell said, instructive (synthetic) necessary truths for scientists to discover, proposi- tions that can be known with complete certainty. Whewell argued that these propositions must be true because counterinstances are inconceivable. Mill's response is to treat inconceivability as indicative of, not necessity, but our having never experienced the subjects of the negations? Convictions about necessity arise from repeatedly associated experiences. Mill knew of the Humean objection against inductive reasoning-- consistently with Victorian philosophy's tendency to disregard him, ~ Hume is seldom referred to in the Logic -- and made use of this clear statement of Whewell: " 'Experience must always consist of a limited number of observa- tions; and, however numerous these may be, they can show nothing with regard to the infinite number of cases in which the experiment has not been made' " (L 2.5.6). 4 One might have expected this argument to dispose Mill to some form of scepticism (whether absolute or what Richard Popkin terms "mitigated"5). On the contrary: through most editions of the Logic Mill un- hesitatingly accepts fundamental laws of science as proven verities. When laws of causation are proven by the methods of science, their proof is, by...