ABSTRACT
According to the regularity account of physical law—versions of which have been advocated by Ayer (1963), Braithwaite (1953, ch. 9), Goodman (1983, pp. 17–27), Hempel (1965a, pp. 264ff.), Lewis (1973, pp. 72–77; 1986), Mackie (1962, pp. 71–73), Nagel (1961, pp. 58ff.), and Reichenbach (1947, ch. 8), among others—laws of nature are regularities among events or states of affairs and a law-statement, the linguistic expression of a law, is a description of a regularity that is a law. The familiar challenge faced by this account is to distinguish those descriptions of regularities that are law-statements from those that are accidental generalizations. I wish to consider a more fundamental problem: that many a claim we believe to describe no regularity at all, nomological or accidental, we nevertheless accept as a law-statement. This problem arises from what Hempel (1988) calls “the problem of provisos.” Consider the familiar statement of the law of thermal expansion: “Whenever the temperature of a metal bar of length L 0 changes by ∆T, the bar's length changes by ∆L = k · L 0 · ∆T, where k is a constant characteristic of that metal.” This statement states a relation between L 0, ∆T, and ∆L that does not obtain; it may be violated, for instance, if someone is hammering the rod inward at one end. Since this statement does not describe a regularity, it is not a law-statement, on the regularity account.
