Mathematical .Mktuud.s Ai-i^likd to Statistics. 403 



liuii<liv(l pages, and to look at it one might suppose that little remains 

 to be said about working-class conditions. In truth, it does give a 

 vast amount of interesting fact and information about all the pi-incipal 

 towns in England, Scotland and Ireland : but when 3'ou ask, after much 

 reading of tlisconnected masses of heteri)geneous facts and figures, what 

 exactly you have learned from it, the answer must be— Surpi-isingly 

 little. Jfni/iia iitdii/esfatfuc moles, a glaring sjiecimen of elaborate 

 tritling, a misapplication of industry and ingenuity which throws 

 little, if any, important light upon any social or industrial tjuestion. 

 Statistics are of no use unless they enable us either to support some 

 hypothesis or principle, or to make some induction from them which 

 may supply us with a well-founded {principle or theory. We defy any 

 one, either at the Board of Trade or elsewhere, to say that this report 

 serves either of these purposes. No one broad generalisation of any 

 value can be made from ic, any more than if the painstaking oHicials 

 had compared the lunnber of red-haired with black-haired men and 

 women in the different towns of the United Kingdom." 



It must be admitted that this is ver}' plain speaking, and it is to 

 be hoped it is a beginning of a demand for improvement in the presen- 

 tation of statistical facts. It is frequently said that the mass of 

 statistics published every 3'ear is something immense, and the cry is 

 still for more. But although the statistics are man}", the generalisa- 

 tions are few. Although there is a certain mistrust in figures, and 

 although it is often said that figures can be made to prove anything, 

 still it is generally held that the statement of figures is, after all, the 

 conclusive argument. Yet, as pointed out by Mandello, in the Journal 

 of the Royal Statistical Societ3% it is generally forgotten how carefull}' 

 figures have to be handled, how critically they ought to be weighed 

 and how easy it is to misinterpret and misuse them. 



Mandello's idea is to make the statistical bureaux into what he 

 terms " real offices " for answering statistical questions and inquiries. 

 Why, he asks, should hundreds and thousands of busy people hunt in 

 big volumes for statistical data, compile, calculate, itc, making mis- 

 takes and hai-dly ever finding what they really want, if their requests 

 to know the figures for such and such topics could easily and with 

 authority be answered by the clerks of the offices, or if the}^ could be 

 told that the figures looked for are not known ? He admits that this 

 proposal looks bizarre, and so I think would most people. The 

 answers gi\en by the bureaux would in many cases be (.lisputed, and 

 then there would be nothing for it but to establish a court of appeal, 

 say a High Court of Statistics, with a few professors of Statistics and 

 Economics as .judges I 



Undoubtedly what is wanted is fewer statistics, but greater 

 accuracy in collection and compilation, aiid more uniformity in setting 

 out the final figm-es. There would then be a greater scope iov making 

 generalisations, and here mathematical processes would come into 

 play. 



I close this paper with the following interesting example, of 

 Quetelet's methods, published, it nuist be rememberetl, over sixty 



2 a2 



