This review appeared in Canadian Philosophical Reviews, vol. VII, no.  9 (Sept. 1987), pp. 381-383.

NORMAN SWARTZ. The Concept of Physical Law. New York: Cambridge University Press 1985. Pp. xii + 220. US$29.95. ISBN 0-521-25838-3.

Among the recent flurry of work on the nature of laws of nature, this book is a welcome addition. It begins with the standard distinction between regularity theories (or what are often called 'Humean' theories) and necessitarian theories, and has as its aim the defense of a regularity theory. The book includes several controversial claims such as the claim that by adopting a regularity theory one can resolve traditional difficulties associated with free will and determinism.

Chapters 1 - 4 provide a general introduction to the subject of laws of nature. They include a helpful discussion that distinguishes between instrumental or scientific laws and the underlying physical laws which scientific laws often approximate, an outline of the necessary conditions for physical lawfulness, and details of the distinction between regularity and necessitarian theories. Prescriptivist theories are also mentioned briefly for the sake of completeness. A quick introduction to the modal issues involved is also provided.

Chapters 5 - 8, 12 & 13 deal with the more substantive theoretical issues. Here Swartz considers the traditional objections to regularity theories, including the most often cited objection that under any regularity theory too many physically possible events will be deemed physically impossible and that too many actual but physically contingent events will be deemed physically necessary. Objections based on statistical laws and counterfactuals are dealt with in chapters 12 and 13 respectively.

Finally, chapters 9 - 11 examine several philosophical consequences of adopting a regularity theory, including the consequences for free will and determinism and the problem of miracles. Swartz concludes that within the context of a regularity theory the problem of determinism cannot arise in its traditional guise since the physical laws which govern human action (like all other physical laws) are merely descriptive, not prescriptive, of such actions (116ff). Hume's problem of miracles is likewise avoided once it is recognized that under a regularity account 'it is logically impossible that a physical law should be "violated" ' (108).

How far is Swartz's defense of regularity successful? Here, the book attempts both to deal convincingly with the traditional objections to any regularity account and to discredit necessitarian theories by means of an epistemological challenge. As Swartz points out, it is commonly agreed by both the regularist and the necessitarian that physical laws are true, contingent, purely descriptive conditionals which express either a universal or a statistical proposition (29, 31). However, disagreement emerges over the question of whether these five characteristics constitute a set of necessary and sufficient conditions for physical lawfulness, or merely a necessary one. According to the regularist such conditions are both necessary and sufficient. However, if so, it follows that every (so-called) accidental universal and statistical generalization will constitute a physical law. Once this is granted the question arises of whether any unrealized physical possibility can ever be said to be consistent with the laws of nature. Kneale's discussion of a non-existent race of white ravens possibly evolving in a permanently snowy region, Popper's example of the extinct New Zealand Moa living beyond its actual lifespan and Molnar's case of a river of Coca-Cola are all examples based on propositions which our intuitions tell us are false yet physically possible. Armstrong gives the problem of all true accidental generalizations becoming laws of nature a new twist by pointing out that, since generalizations of the form (x)(Fx ⊃ Gx) are true whenever there are no Fs, it follows that on the regularity account it is a physical law that centaurs are both adept at philosophy and that they are inept at philosophy simply because, omnitemporally, there are no centaurs.

Swartz's response to such objections is in essence that if such conclusions strike one as unintuitive, then intuitions are simply in need of reform (55). In claiming that certain events are inconsistent with the laws of nature, the regularist is not saying that it is impossible for such events to occur. Rather, 'when we say that it is physically impossible that they do, we mean only that – as a matter of fact – they do not occur' (57). It is still possible to distinguish between failure and doom. It is just that in the crucial cases the regularist and the necessitarian will differ over which projects merely failed and which were doomed from the start (65).

The immediate difficulty with this type of defense is that it merely shifts the ground of dispute. Having appropriated the language of physical lawfulness simply to say that certain events, as a matter of fact, have not occurred, the regularist has reduced the problem to merely a verbal one. After all, Swartz still has to account for the sense in which he claims some events but not others could occur. In reality the question is one of whether science has need of the concept of nomological necessity and whether the distinction which necessitarians emphasize and about which regularists appear to be insensitive (54) has value either in the practice of science or for explaining its predictive success.

Related to the regularist's defense is the epistemic challenge which Swartz raises for the necessitarian (67ff, 79ff). How is it that the necessitarian can claim to distinguish between those regularities which are merely accidental and those which are governed by some form of nomological necessity? Here Swartz is on stronger ground but, even so, it is unlikely that the committed necessitarian will find the challenge compelling. After all, as Swartz emphasizes early on, physical laws (unlike scientific laws) are independent of our beliefs about them and so any epistemic challenge will be weakened accordingly.

Despite these concerns the book is a stimulating and enjoyable one to read. It is unfortunate that none of the recent work done by Armstrong and Tooley is mentioned directly, but when coupled with Armstrong's What Is a Law of Nature? (which gives a spirited defense of a non-regularity theory) the book provides a helpful introduction to the contemporary state of play with respect to laws of nature.

University of Toronto