May 7, 1891] 



NATURE 



part of the work before us— namely, the investigation of 

 the causes of disease. This medical portion of the volume 

 may, as the author fears, " prove too technical for many 

 readers"; and, perhaps we should add, critics. The 

 student of such statistics must bring much knowledge in 

 order to carry away much. The need of this requisite 

 may be illustrated by one of Mr. Longstafif's examples. 

 Certain of the curves which he traces show a remarkable 

 correspondence between the outbursts of diphtheria and 

 a group of other diseases, amongst which are croup and 

 cynanche maligna. And yet between the two latterdiseases 

 and diphtheria the correspondence at some dates is not 

 so close as the suggested theory desiderates. Diphtheria 

 in 1859 rose enormously, while the other diseases did not 

 rise simultaneously, or even feil. But, as we understand 

 the matter, the theory is saved by the. surmise that many 

 cases previously ascribed to croup and cynanche maligna, 

 were put down to diphtheria in 1859 and afterwards, when 

 the stir created by letters in the newspapers had excited 

 the attention of observers to the " new disease." This is 

 one of those explanations of figures which an outsider 

 would probably not even have thought of, and the import- 

 ance of which he is little qualified to estimate. 



The " aetiology " of the subject must be left to the ex- 

 pert. The general reader, if he cannot penetrate to the 

 laws of causation, may at least admire the uniformity of 

 results which the author's diagrams exhibit. The nature 

 of some of his observations, and the labour and care 

 which they required, are indicated in the following 

 quotation : — 



" The object of my investigation was . . . [principally] 

 to see whether any, and if so what, relations subsist 

 between diseases believed to be distinct. ... I accord- 

 ingly traced eighty-nine curves representing the death- 

 rates per million in England and Wales from as many 

 'alleged causes.' ... By a simple application of the law 

 of combinations, it will be found that to compare all these 

 eighty-nine curves two and two together, would involve 

 3916 operations. Of these I have as yet actually made 

 only 1425." 



This comparison of curves representing the fluctuation 

 of death-rates for different diseases forms some of the 

 most beautiful pieces of statistics which we have ever 

 seen. We may allude in particular to the comparison 

 of erysipelas, scarlatina, rheuniatism of the heart, and 

 certain other diseases with each other and the variations 

 in the rainfall (Plate xix.). The death-rates are shown 

 to be parallel to each other, not only for different times, 

 but also, in the case of three of the diseases, for different 

 places in all the eleven registration counties of England 

 and Wales. The splendid diagram which exhibits this 

 manifold comparison (Plate xxi.) affords, as the author 

 points out, a good illustration of the value of large 

 numbers in statistical inquiries. 



" The curves for England and Wales exhibit smaller 

 fluctuations than those for sections of the country, and 

 the correspondences between them [between the rise 

 and fall of death-rates for three specified diseases] are 

 in nearly all cases much closer." 



Among investigations of which the interest appeals to 

 the mere statistician as distinguished from the medical 

 expert, we may mention the calculation of the frequency 

 with which coincidences between the deaths of both 

 husband and wife from phthisis "might be expected to 

 occur as a pure matter of chance, on the hypothesis that 

 NO. I 123, VOL. 44] 



phthisis is not a communicable disease." By a beautiful 

 application of the calculus of probabilities, the following 

 conclusion is reached :— 



" It is plain, therefore, that, to show any substantial 

 argument for the existence of infection, it would require 

 a much larger collection of cases than has yet been 

 published." 



Another inquiry which the general reader will follow 

 with peculiar interest relates to hydrophobia. The 

 statistics suggest laws very different from popular beliefs. 

 The paucity of the observations, however, necessitates 

 caution ; which Mr. Longstaff does not fail to inculcate. 

 It is not his least merit that he instils what may be called 

 the logic of statistics by occasional precept, as well as by 

 repeated examples. 



OUR BOOK SHELF. 



The Best Books: A Contribution towards Systematic 

 Bibliography. By William Swan Sonnenschein. 

 Second Edition. (London : Swan Sonnenschein and 

 Co., 1891.) 

 The idea of this " contribution towards systematic biblio- 

 graphy " is excellent, and has been excellently carried 

 out. When interest in a subject has been excited, the first 

 question of the student, of course, is, Who are the best 

 and most recent authorities on the matter .? The question 

 is by no means always easily answered, for as yet there 

 are few good subject-indexes, and the most valuable of 

 them are not within the reach of everyone. The present 

 volume may almost be said, for ordinary practical pur- 

 poses, to have solved the problem. Mr. Sonnenschein 

 has not attempted anything so ambitious as a philosophic 

 classification of the sciences. He has worked out his 

 scheme on what he properly calls "a common-sense 

 plan," grouping books first into large classes, then break- 

 ing them up into sections, sub-sections, and paragraphs — 

 " with the result of obtaining all the literature of one 

 subject in one list, and that of outlying subjects close at 

 hand." He begins with theology, next takes mythology 

 and folk-lore, then philosophy, society (including many 

 different branches), geography, history, archaeology, and 

 so on, until all important departments of knowledge have 

 been included. No one who has occasion to use the 

 book will have the slightest difficulty in understanding 

 the principle, or in finding the particular subdivision pre- 

 senting the facts of which he is in search. The new 

 edition contains the titles of twice as many books as 

 the first edition (50,000 as against 25,000) ; and, so far as 

 we have been able to examine them, they seem to have 

 been admirably selected. Here we have to do only with 

 the scientific part of the work ; and, considering how 

 vast is the material from which Mr. Sonnenschein had to 

 choose his lists of scientific treatises, he may be con- 

 gratulated on the manner in which his task has been 

 accomplished. For the most part, he refers only to books 

 that are in print, and easily obtainable. The very best 

 books he has " asterisked," and in every case he gives the 

 dates of the first and last editions, with the price, size, and 

 publisher's name. Two separate indexes— one, a list of 

 authors, with the titles of their works ; the other, a list 

 of subjects— add greatly to the value of the compilation. 



The Fairyland Tales of Sdence. By the Rev. J. G. 



McPherson. Second Edition. (London : Simpkin, 



Marshall, and Co., 1891.) 

 This volume consists of a number of papers which 

 appeared originally in various periodicals. The author 

 does not profess to embody in them the results of inde- 

 pendent research. His object is to give to readers who 

 may not have access to recent scientific authorities "an 

 accurate and at the same time interesting account of the 



