638 PROCEEDINGS OP SECTION G (l.). 



these soldiers are intended to represent the numbers by linear, square, 

 or cubic measure ; and secondly, if he does know, the complicated 

 form only confuses his eye for measurement. I have taken the trouble 

 to measure some of these comparative pictures, and found that they 

 as often appear to be based on linear and square as on cubic measure, 

 and in some cases on no correct measure of any description. The 

 most natural supposition would seem to be that cubic measure is 

 usually intended. Now it requires a considerable amount of training 

 to gauge approximately the proportions of similar bodies. People 

 will buy oranges all their lives, yet never realise that an orange with 

 a shortest diameter of 2^in. is nearly twice the size of one of similar 

 shape with a shortest diameter of 2in., and more nearly five than four 

 times the size of one measuring l^in. Or take three persons of similar 

 build — A 5ft. high, B 4ft., and C 3ft. Only a person who has had 

 repeated occasion to compare the weights of people will realise, that 

 A weighs approximately twice as heavy as B, and not much less than 

 five times as heavy as C. (See diagram No. 3.) 



To create, then, accurate visual impressions of proportion, impres- 

 sions that approximate the actual facts as expressed in the unerring 

 symbol of figures, linear measure is infinitely preferable to any other 

 graphic method. 



A large proportion of the statistics at present collected relates 

 to separate nationalities, whilst in probably less numerous cases their 

 scope is further confined to localities such as districts, provinces, or 

 municipalities, or even to specific centres of private human activity. 

 For many purposes these limitations are of course desirable, as they 

 usually emphasize natural distinctions. There are, however, subjects 

 on which it would be of still greater value if statistics of more universal 

 scope and application were available. It is, for instance, of interest 

 to every human being living within reach of civilisation that it be known 

 what the world's annual production of certain commodities is, such as 

 gold, silver, and wheat. In the collection of such statistics it would 

 be desirable if international action could be arranged for, so that the 

 work were not left to the unreliable chance of private enterprise, nor 

 hampered by a multitude of varying methods of calculation and ex- 

 pression of results, as is now too frequently the case. For instance, 

 it would be an advantage if all gold or wheat could be reduced to some 

 universally recognised standard, say, the former to fine ounces, the 

 latter to centals or bushels, or preferably some decimal measure. An 

 office which, like the Statistical Department of Western Australia, 

 has for some years been engaged in the endeavor to obtain reliable 

 figures relative to the world's gold output, has thereby gained some 

 striking experience of the difficulties to be overcome in any such effort 

 under the conditions at present prevailing. 



The scope and application of statistics are further determined by 

 the more or less temporary or permanent interest attaching to the 

 figures. Those, for instance, which are of the most immediate value 

 to the financier or commercial man are at the same time often of purely 

 ephemeral importance. Others extend their use over more or less 



