Agkicultueal Experiment. 



459 



It can only give a true representation of the series when the varia- 

 tion is mathematically normal and not disturbed by any factor. 



An example may illustrate the point more clearly. If a curve 

 of the milk yield of cows obtained in the mature class of the advanced 

 registry scheme of the Friesland Breeders' Association of vSouth 

 Africa up to 1920 is plotted, it is found that the greatest number of 

 cows in this class during 300 days give between 10,000 pounds and 

 12,000 pounds, with a mean yield of 11,340 pounds for the class. In 

 this case, the results being normal, as seen from the curve, the 

 average figure is of value. 



Taking another group of figures, however, viz., the yields per 

 niorgen of maize in the Union, it is seen how little information an 

 average can really provide. According to the census of agricultural 

 production for 1918, the average yield of maize in the three Provinces 

 of Transvaal, Orange Free State, and Natal is 4.6 muids per morgen. 

 If the districts be grouped according to yield, it is seen that the 

 greatest number of districts, however, prodiice between 3.6 and 4 

 bags per niorgen, while there are also more districts producing 

 between 5.1 and 5.6 bags than there are producing 4.6, and the range 

 of yield is from 1.82 to 11.67 bags per morgen. 



In this case the variation of the series is not normal, there being 

 several disturbing factors, such as irrigation, rainfall, method of 

 farming, etc., and a bare statement of the avera|j;'e cannot adequately 

 represent the range of yield or show the yield obtained in the greatest 

 number of districts. 



Even at the risk of labouring this point on the limitations of 

 an average, the following extract from an article on the "Cost of a 



