268 REPORT— 1887. 



modities we ought to take account of all articles, without regarcliug whether they 

 are consumed in the same proportions, or in different proportions by different 

 persons. But if we proceed by way of sample, then we ought to assign special 

 weight to those articles which, as Engel's law and the American labour statistics 

 have established, are consumed in nearly equal proportions by each household 

 throughout a large class of the community. Less weight should attach to those 

 articles, the ' sundries ' of the statistics referred to, which appear more fitfully in 

 the household budgets. How far in England we have to proceed by way of sam- 

 ples afforded by certain markets and certain commodities is a question not to be 

 decided in this memorandum. The difference upon which these distinctions turn 

 is that which the writer, in treating of the theory of errors, has drawn between 

 simple induction and inverse probability (see Observations and Statistics, ' Camb. 

 Phil. Trans.' 1885). 



(3) A more obvious ground of selection is that some articles (however large 

 their money value) interest only a comparatively few (rich) persons. Accordingly, 

 in constructing a standard adapted to the general requirements of the community, 

 we ought upon utilitarian principles to treat the variations in the price of that 

 class of articles as of comparatively little account.^ 



It may be doubted whether the practical worth of these subordinate modifi- 

 cations corresponds to their theoretic interest. For to assign less importance to 

 some of the data on the ground of a deficiency of weight which is not susceptible 

 of numerical evaluation is a practice which, though countenanced by the example 

 of physicists in their reduction of observations, is apt to diminish confidence in 

 sociological calculations. For the sake of a little additional accuracy it may not 

 be worth while incurring the suspicion of cookery : — 



Denique sit quidvis, simplex dumtaxat et unum. 



2. We come now to the case where, the interval between the com- 

 pared ejDochs being considerable, the quantities consumed at the two 

 epochs are materially different, and the ratio of the quantity consumed 

 at one epoch to the quantity consumed at the other is no longer even 

 approximately the same for the different commodities. The difficulties 

 presented by this case, which seemed to defy science, have been 

 triumphed over by Professor Marshall.^ The incommensurable propor- 

 tions of the dissimilar expenditures he manages to compare by means 

 of a series of the intercalated intermediate forms presented by the 

 changing national inventory. Equating each term of this series to its 



' Another modification which might be suggested is that less weight should be 

 attached to those commodities of which the price-variations alfect the general 

 public and a particular class in different senses — a fall, for instance, benefiting the 

 consumer, but ruining the producer. It will be found, however, a difficult and 

 endless task to carry out this principle. For what commodities would be excepted 

 from it 1 Imports perhaps, in so far as it is tlie foreign producer chiefly who is 

 damaged by the fall and benefited by the rise of tliose prices. But with regard to 

 the home industries, in order that the interest of the producer and the consumer 

 should vary in opposite directions, we must suppose an equilibrium of profits to be 

 transmitted from trade to trade, according to Eicardian principles, with a rapidity 

 that is not supposable. 



But not only is the working of the proposed principle difficult, but also it is 

 incorrect ; here, in this section, where our object is that the unit should afford a 

 constant quantity of valuables to the average consumer, without reference to the 

 number of units which the different classes of consumers have to spend (see below, 

 p. 272, note 2). To tamper with certain items of expenditure, such as wages of Domestic 

 Service, on the ground that these transactions belong to distriJmtion, as distinguished 

 from exclianqc, is virtually to introduce the princi^Dle of the sliding scale, to substitute 

 the attribute c for C. 



The exclusion of ' unproductive ' labour has been maintained on other grounds 

 considered in note to p. 263. 



' Contemporary Review, March 18S7. 



