The Little-Mirrlees Manual of Industrial Project Analysis in- Developing Countries is divided into two parts; the first of which is addressed "to the senior administrator or politician, who should understand the broad lines of what is implied by operating a system of social cost-benefit analysis"; and the second "to those who will actually make project evaluations, and teach others how to make them". In fact, the two parts make an integral whole, since the first part of the book raises several broad issues the answers to which are given in the second part. For instance, on page 44 the "senior administrator or politician" is told of the dilemma of the choice between employment-generating and rein¬ vestment-generating projects; a project which employs a lot of labour will get higher marks because it results in a lot of consumption by the poor now. But the incomes generated by such a project will be almost entirely spent. There will, therefore, be little savings generated and so such a project will contribute little to further investment, which, in turn, yields future consumption", while a capital-intensive project with a high reinvestment rate makes a greater contribution to the future, but a smaller one to present welfare. Unless the "senior administrator or politician" is willing to read difficult and often obscure discussions in Partll (Ch. 13) he will have no idea how to go about resolving the dilemma. There is little fear, however, that the nonprofessional reader will get through Part I, not to speak of Part II. To be sure, the authors give a warning that "some of the chapters of Part I may be a little academic for the senior man who has become familiar with economics by practical exposure and does not want to feel that he is going back to school", but they grossly underestimate the gap between the layman's ability to cope with abstract concepts and their "own ability to use plain English. I wonder what the intelligent layman is to make of the following passage which occurs on page 41 and which I chose almost at random: