160 



EEPORT 1889. 



counting from tbe highest, are 12 and 4, both corresponding to the 

 ratio 89. The simple median is thas 89. 



127 



1,G 



It was pointed oat in the former Memorandum that there is a plau- 

 sible hypothesis on which, even for the present purpose, it is proper to 

 •attach some importance to the values of the commodities, though not 

 mecessarily that degree of importance which is prescribed for the standard 

 based on national consumption. The simplest method of attaching im- 

 portance to the values is to take the simple median of the ratios on 

 the supposition that each of them occurs as often as the number which 

 indicates the corresponding value, or the sum of such numbers where 

 there are more than one of them. Upon this understanding there are in 

 all G66 constructive observations — as near as may be, half above and 

 half below 88. That figure then is the weighted median. 



It is pretty certain that this complex median assigns too much im- 

 •portance to the values. And it is probable that the simple median 

 assigns too little. Accordingly a good solution is afforded by combining 

 or comparing the two results, in the example before us taking 88' 5 for 

 the answer. Should the two results be markedly different, inquiry may 

 be made as to the cause of the difference, and a preference should be 

 •given in general to the simpler combination, 



A more elaborate method of weighting the median by taking the 

 square roots of the values was recommended in the former Memorandum. 

 But on second thoughts it appears that the special advantages which this 

 plan may confer hardly compensate for the additional trouble which it 

 involves. 



' Comparative price of stocMngs per dozen ; not explicitly given by Mr. Bourne 

 but inferrible from the entries in his valve and volume columns. 



- Not explicitly given by Mr. Bourne, but inferrible from his data. 



