REVIEWS 757 The other groups deal with statistics of more immediate service to economists. After warning us that the main requisite of a statistician is not arithmetic but logic, (p. 2), Dr. Longstaff explains the terms death-rate and ' corrected death-rate (ch. ii.); a high birth-rate is shown to be in no proper sense the cause of a high death-rate; and we are led to see the great difficulty of tracing any exact connection between the three different factors of an ?tat civil. The birth-rate tends on the whole to follow the marriage-rate (ch. iii. p. 14); but since 1872 it has fallen by itself (cf. 15). The marriage-rate varies with the commercial prosperity of the country (and not with the price of wheat), while the death-rate varies irregularly, and (for example) is not, necessarily low after an epidemic (p. 18). The statistical proof of these and the multitude of other propositions is brought out by coloured curves in diagrams exceptionally agreeable to the average eye and exceptionally intelligible to the average intellect. The diagrams are lighted up by timely interpretation. show, for example, what everybody knew, that there are The figures more wolnen than men in England; but not every one would have observed that the surplus consists of widows and is really a sign of the greater longevity of the gentler sex (pp. 8, 9, 249 seq.). The chapters (ch. iv. and v.) on the growth of population in England and on migrations form a transition to the second theme of the book, the Growth of New Nations (ch. vi. to ix.). Our space forbids analysis of these chapters; but readers with the faintest passion for statistics will discover their charm. They bring out, for example, such striking facts as the unique frequency of census- taking in Canada, and the extraordinarily heterogeneous composition of the people of the United States, a country so often conceived by us as an Anglo-Saxon settlement. In the tenth and eleventh chapters Dr. Long- staff returns to English demography, dealing with our Census, the results of which he had shrewdly forecasted, ? and with our Food Supplies. We should not all of us join him in the hope (expressed on p. 189) that in consequence of the completed settlement of the Western States of America the rents in our own country will begin to rise, though we may agree that this doubtfully desirable result is very likely to happen. It cannot be denied that a sceptical view of the credibility of modern statistics will find some support from the pages of this candid author. Medical statistics for any considerable periods of years are made uncertain by the changes in the classification and nomenclature of diseases (see e.g. pp. 316, 39.4, 39.9, 374, 394 7). The yearly death- rates, &c., are misleading, especially in the later years of the inter- censal periods, unless they have been "ground out" again by Thomas's machine to suit the actual or told out populations as compared with the merely estimated ones (pp. 369, 378). The English Census has often put questions that were either ambiguous, or certain for other reasons to be ? See Economic Journal, June, 1891, pp. 382 seq., Sept., 1891, p. 547, and compare Stmiies in Statistics, pp. 191, 208, 204, 369. To the desiderata of our Census Reports, we may add a general index. What accident removed it in 1881 ?
