Seduced by logic: Émilie Du Châtelet, Mary Somerville and the Newtonian revolution, by Robyn Arianrhod

discussion is offered and no strong opinions voiced, but, given the extent to which these matters have provided the material for notorious disputes, this is perhaps wise. I have a couple of minor quibbles. The lives of the scientists in the vignettes are, of course, sometimes right-censored, since they are still alive, but many are also leftcensored, as if their dates of birth were unimportant. Also the title promises us Permutation Methods but what is largely offered is Permutation Tests. This, given the nature of the field, is perhaps inevitable, but it also provides one explanation as to why not all statisticians will share the enthusiasm of the authors as regards permutation tests. They provide logical rigour within a rather limited framework. As soon as one asks for more it becomes less obvious that they are a good place to start. A major omission is that there is no reference to John Nelder’s theory of General Balance (Senn 2004) which is at the heart of the GenStat approach to analysis of variance, and which categorizes experiments in terms of block and treatment structure. This is related to deep ideas of symmetries in designed experiments and provides a natural link between analysis of variance and permutation methods. However, the ideas have made little impact in the USA and perhaps the one charge that can be laid against the authors is that their book is a little Americocentric. These are minor criticisms, however. The authors are to be congratulated on this very fine collaboration. In their acknowledgements they mention table 20 at the Rainbow Restaurant, Fort Collins, Colorado at which most of the book was written. This surely deserves a plaque. It conjures up a delightful vision of three scientists first ordering a random permutation of one starter, one main and one pudding, and then settling down to the serious business of the day. In short my recommendation to the reader is clear: buy this book! My recommendation to those invited to review using Springer’s online reviewing ‘tool’ is also clear: don’t!