PROCEEDINGS OF SECTION G (l.). 635 



may be considered as coming within his province and that which he 

 should reject ; and, as a matter of fact, he has nothing more definite 

 to guide him than precedent and his own judgment. It is evident, 

 therefore, that an international statistical society may do excellent 

 work in fixing more definitely the limits of the science of statistics. 



Statistical information is intended to serve two purposes, separate 

 and different in their immediate bearing, though undoubtedly both 

 aiming at the welfare of humanity. The one is that of scientific in- 

 vestigation, the other that of direct practical use. 



In collecting statistics for purposes of scientific investigation 

 it is obviously necessary to have a fairly clear conception as to what 

 data are likely to have any scientific value. Supposing, for instance, 

 that we msh to compare the respective degrees of fertility of women 

 of different nationalities among the Australian popidation ; it will 

 then, of course, be necessary to obtain the average number of children 

 born to married women of each nationality. Yet if this were done 

 without differentiating between women in various age groups, the 

 information would not only be useless, but absolutely misleading, as 

 the proportion of women in each age group will naturally vary very 

 much in the different nationalities, the Australian-born women being, 

 of course, far more numerous at the younger ages, when families are 

 consequently as yet small. 



It is, therefore, indispensable before collecting statistical data to 

 study every factor that may contribute to the result they represent, 

 so that enough detail may be collected for purposes of adequate analysis. 

 Only then shall we be enabled to make reliable deductions from the 

 information obtained. 



A point of the greatest importance, in drawing conclusions from 

 the particulars available, is the length of the period to which they relate. 

 The annual recurrence of certain phenomena for a considerable number 

 of years warrants the assumption that a more or less permanent cause 

 underHes their nature, and that consequently the future is likely to 

 present similar features. The popular call for " the latest figures " 

 is therefore somewhat vainly concerned about statistics of this nature, 

 which, like good wine, only derive their value from the length of time 

 for which they have been kept. 



Having most carefully weighed all the circumstances, and observed 

 a constant recurrence of the same phenomenon, the statistician 

 cautiously predicts that in all probability the same feature will again 

 present itself in coming years at the appointed time ; and thus his 

 prognostications may be eminently useful to the legislator, the states- 

 man, the physician, and also to those whose personal interests are 

 bound up with commercial or industrial enterprise. The danger, 

 however, of drawing conclusions, and attempting to prognosticate 

 from insufficient data, has found ample illustration in some of the 

 striking failures on the part of celebrated statisticians to account for 

 certain social phenomena or foreshadow events. 



The practical use of statistics that naturally appeals most to the 

 commercial and industrial classes is that which conveys information 



