BOOK REVIEWS 363 But something more coherent and explicit is required. About that concerning which one can speak, thereof one must not remain silent. Such criticisms do not detract from the value of this book. Wittgenstein lectured with remarkable freshness and openness. The result is that one learns by "sitting in on his classes" many things about his approach that one could not learn otherwise. Wittgenstein's Lectures on the Foundations of Mathematics not only provides an excellent introduction to his later philosophy; on many points it provides a surer grasp of his philosophy of mathematics, and of logic, than do his own manuscripts. It deserves to be widely and carefully read. PETE A. Y. GUNTER North Texas State University Sandra B. Rosenthal. The Pragmatic A Priori: A Study in the Epistemology of C. L Lewis. St. Louis: Warren H. Green, 1976. Pp. 104. Any attempt to present a faithful interpretation and a critical defense of C. I. Lewis's pragmatic theory of the a priori immediately runs into problems. First, there are many prima facie tensions in Lewis's work. For example, how is it possible for a priori truth to be true no matter what and for the same truth to be a priori (analytic by virtue of meaning) in one context and empirical in another? Or, if a priori principles are absolute, how can there be different and incompatible systems of such principles? Second, Lewis's exposition of his theory was often not sufficientlycomplete. For example, to say that the acceptability of a set of a priori truths is tested by its serving adequately some long-term interests and needs seems to require that these needs and interests be characterized independentlyof competing a priori categorizations. But about this and the kind of needs he has in mind little is said. Third, almost all of Lewis's central notions-intension, analyticity, the given, fact versus interpretation, and certainty--have recently been subject to sustained criticism, though not usually explicitly leveled at Lewis. Fourth, Lewis's relation to the pragmatism of Peirce, James, and Dewey remains obscure. He does not unequivocally endorse a pragmatist theory of truth and meaning. The acceptability, but not the truth, of a set of a priori concepts or analytic statements is justified on pragmatic grounds. On the other hand, he does acknowledge important differences between pragmatism and positivism, and his view that experience requires conceptualization is shared with the other pragmatists. Rosenthal acknowledges several of the tensions in Lewis's work and seeks to resolve them. This she does by reviewing his work as a whole in the attempt to find an overall consistency. It is doubtful whether this is the most expedient approach, because An Analysis of Knowledge and Valuation was written nearly twenty years after Mind and the World Order, and the tensions in the earlier work are added to and not resolved by the later one. She does not consider the third and fourth points mentioned above. Beginning with the fact that Lewis was led to his interest in epistemology by his dissatisfaction with the extensionalist logic of Principia Mathematica, Rosenthal acknowledges the strain between Lewis's view that "all . . . logical truths must be intensional" (p. 2)--which seems to rule against extensionalist systems--and his considering these as competitors to intensional logics (pp. 3-4). In "Logic and Pragmatism" Lewis hankers after the view that there is one logical system that expresses the truths of valid inference while accepting that there are genuine alternatives. Rosenthal does not resolve the conflict. Lewis really did think that material implication is inadequate, but his reasons lead to Anderson's and Belnap's systems of entailment rather than to his own modal systems. His intensional sentential modal systems are not theorem rivals but are extensions to classical extensional logics. Rosenthai reports Lewis's view that "inference is analytic of systems, not of propositions in isolation," which says in 364 HISTORY OF PHILOSOPHY effect that the meaning of logical operators may differ according to the systems in which they occur, so that, for example p v -p has a meaning in intuitionist logic different from its meaning in classical logic. Some have contended that this...