146 



NATURE 



[December 13, 1894 



scants to be determined. This is all the more tantalising, 

 as the variation with age (or with selection) of the 

 skewness of the frequency curve for a given organ is 

 verj- probably a most important factor in the mathe- 

 matical treatment of evolution, while it is precisely 

 human material which enables us to record the 

 age, too often an unknown quantity in investiga- 

 tions on the lower animals. It must not be sup- 

 posed, however, that Mr. Porter realises that skew- 

 ness in the percentiles marks want of normality in the 

 frequency curve. On the contrary, he takes the raw 

 material for the height of 2192 girls aged eight, and fits 

 it, not very accurately, to a normal frequency distribution. 

 Although the measurements better fit a skew than a 

 normal distribution, the divergence from the normal is 

 not very marked, and Mr. Porter writes : 



" It is not necessary to make this comparison at more 

 than one age, or in more than one dimension, for it is 

 known that if one series in a group like that with which 

 we have to deal shows this agreement, the other series 

 will be found to do the same." (p. 290.) 



Now the whole point of the measurements, if properly 

 interpreted, is to ascertain the varying degree of skew- 

 ness with age, i.e. the varying amount of divergence 

 from the normal curve. To emphasise this divergence 

 when remarking on his raw material, but to fail to 

 recognise what bearing it has on theory, is typical of the 

 untrained statistician. 



As we have seen, the use of percentiles renders Mr. 

 Porter's material of small value for the problem of selec- 

 tion. The same remark again applies in the case of cor- 

 relation. The ten organs measured would have been 

 most valuable material for the problem of correlation ; 

 the percentile results alone being given, nothing can be 

 done. In Chapter ix. Mr. Porter does develop a theory 

 of correlation, which, however, seems to us absolutely 

 erroneous. As he makes it the basis of a scheme for 

 the school physician to testify whether the pupil's strength 

 is equal to the strain put upon it, this want of accurate 

 theoretical knowledge appears doubly unfortunate. 



We have ventured to point out these failings in Mr. 

 Porter's work, not because they are peculiar to him — 

 they are characteristic of much work of a statistical 

 kind which is now being turned out in both Europe and 

 America — but rather because they point to a new need, 

 which the public has hardly yet recognised. There has been 

 up to the present time — with the honourable exception of 

 courses of lectures by iJr. X'enn in Cambridge — no teach- 

 ing of statistical theory in England. There is no chair 

 of statistics in any English university, and the Newmarch 

 lectureship at University College, except for the year of its 

 tenancy by Prof. Edgeworth, has been held by economists, 

 and not by mathematically trained statisticians. We 

 want a centre, which shall not only contain a statistical 

 museum, but embrace as well a statistical laboratory 

 and workshop. In such a centre students might not only 

 receive a mathematical training in dealing with raw 

 statistics, but also be exercised in the methods of 

 collecting and tabulating, which must precede mathe- 

 matical reduction. To such a centre the biologist and 

 anthropometrist could send their measurements to be 

 "fitted," the physicist his observations to be dealt with, 

 and the economist or sociologist his price or labour 

 NO. 131 I, VOL. 51] 



statistics to be analysed. At the same time, absolute 

 measurements might be made bearing on the problems of 

 evolution, disease and national economy. In this manner 

 a number of efticient young statisticians might be trained, 

 who would not only find a life-work ready for them in 

 craniology, zoology, botany, and economics, but who, 

 passing into government departments, census offices and 

 labour bureaus, might remove from us the reproach of a 

 recent continental writer, that nowhere were statistical 

 dilettanti so rampant as in England. 



Karl Pe.\rson. 



WATER SUPPLY AND WATER-WORKS. 

 The Principles of Water-works Engineering. By J. H. 



Tudsbery Turner and A. W. Brightmore. (London : 



Spon, 1S93.) 

 The Water Supply of Towns. By \\. K. Burton. 



(London: Crosby Lockwood and Son, 1894.) 



WITH the constant growth of population, and the 

 increasing tendency of the population to con- 

 gregate into large towns, the provision of an ample 

 supply of pure water to towns becomes every year of 

 greater importance ; whilst, as the nearer sources of 

 supply become inadequate for the ever-growing demands, 

 a sufticient supply becomes more difficult to obtain, and 

 has to be sought at considerable distances. Accordingly, 

 the subject of water-supply has assumed an enhanced 

 importance within recent years ; and till these books 

 appeared, there was a dearth of comprehensive text- 

 books on this branch of engineering. As the sources of 

 water-supply, and the works for its provision and distri- 

 bution are comprised within very definite limits, both 

 books necessarily traverse much the same ground, and 

 deal with similar classes of works ; but, nevertheless, 

 they exhibit dili'erences in arrangement, and in the 

 method of treating the various subjects. Thus, whereas 

 'Water-works Engineering" extends over 420 pages, 

 separated into only eight chapters, and illustrated by one 

 hundred and twenty figures in the text, " The Water 

 Supply of Towns " is subdivided into twenty-two chapters, 

 occupying only 272 pages, somewhat larger in size, but 

 illustrated by two hundred and fifty-seven drawings, 

 several of which are put into forty-three large folding- 

 plates. The first book, moreover, is subdivided 

 into numbered sections, the numbers of which are 

 merely used for reference, and instead of the pages 

 in the index ; whilst the various subjects in the 

 second book are clearly indicated by black-letter 

 headings. Altogether, " The Water Supply of Towns " 

 is much better arranged for reference ; and though 

 containing slightly less matter, it is much more fully 

 illustrated, is in larger print, and has a somewhat longer 

 index than " Water-works Engineering." Neither book 

 gives a summary of contents at the head of each chapter, 

 which is often serviceable for reference. 



In " Water-works Engineering," the sources, measure- 

 ment, collection, storage, purification, conveyance, and 

 distribution of water, and the maintenance of water- 

 works, are successively dealt with in the eight chapters. 

 The most logical sequence would be to describe in regular 

 order the various processes to which water is subjected, 

 from its source up to its delivery to the consumer. 



