IN recent years it has been suggested that the laws of Nature can be reached by deductive methods instead of by inductions from observation. We may, with Eddington, assert that the chain of deduction starts from epistemological premises or, with E. A. Milne, that it begins with axiomatic statements defining the character of the system to be studied. Opinions differ widely regarding the value of the deductive method. If it is to be adopted, however, we may well ask that the author of a deductive theory should fulfil the following requirements. First, he must himself be aware of all the axioms or premises which he employs. If he is not, there can be no guarantee that inductive premises have not unwittingly been incorporated into the theory. Secondly, the axioms should be clearly recognized for what they are. They should not be introduced piecemeal in the course of the argument under the guise of deductions from the theory previously developed. In this respect, the model is Euclidean geometry, which begins with a statement of all the definitions and axioms that are afterwards used. Thirdly, the theory must be self-consistent and, naturally, free from mathematical mistakes. Under this heading we may also ask that consequences deduced from the theory should either be capable of check by observation or, if this is impossible, that they should at least be plausible. Time and the Universe A New Basis for Cosmology. By Dr. F. L. Arnot. Pp. 76. (Sydney: Australasian Medical Publishing Co., Ltd., 1941.) n. p.