F.— ECONOMIC SCIENCE AND STATISTICS 147 



brass trades the alternative patterns of integration become positively 

 kaleidoscopic and one is grateful for any elementary device to see it 

 through. Indeed, the superimposed circles strikingly reveal how arbitrary 

 is the notion of ' an ' industry. One technical skeleton followed through 

 vertically and laterally from head to tail and finger tip to finger tip, all raw 

 materials to all finished products, may include dozens of industries. And 

 what these industries are is determined largely by the usual course or 

 scope of the enclosing circles, i.e. the policy of integration adopted in the 

 past by the majority of plants or firms. 



Sometimes these circles are firmly drawn and the great majority of firms 

 and plants in one given industry keep within the circle and do not suffer 

 trespassers ; in other cases the frontier is not clearly fixed. The British 

 Production Census of 1930 made, in its Final Report, a manful effort to 

 measure how far firms assigned to an industry really specialised in the 

 principal products characteristic of that industry. Two percentages are 

 given for each industry, one showing the proportion of ' its ' principal 

 products included in the output of ' its ' firms, the other showing the 

 proportion of the output of its firms represented by its principal products. 

 An industry whose firms made all of its principal products and nothing 

 but those products, who were neither trespassed upon nor trespassing, 

 would get 100 per cent, for each percentage. Such full marks for ' defini- 

 tion ' were successfully attained on both counts by only six out of some 

 120 industries ; grain-milling, ice, tobacco, matches, wall-paper and in- 

 candescent mantles. Markedly loose in definition are the industries of 

 building materials, of fertiliser, disinfectant and glue, of tools and imple- 

 ments, of copper and brass, of finished brass, of coke and by-products, 

 of oil and tallow and of manufactured stationery. 



Once the different patterns and scopes of integration are defined between 

 industries or within an industry, research may proceed to compare the 

 prevalence, trends, cost and profits of each of them. Such a comparison 

 of profits on capital investment has been made within the U.S. petroleum 

 industry^' in the years 1922 to 1926. On the whole, ^he most profitable 

 pattern appears to have been a division of the whole process into two 

 ' scopes ' ; refining and marketing integrated (profits on capital varying 

 year by year from 9-8 per cent, to 14-9 per cent.) ; producing isolated, 

 with profits varying from 5-9 per cent, to 18 -6 per cent. Complete 

 integration with producing refining and marketing all in one scope showed 

 profits only from 4 • 7 per cent, to 12 • i per cent. Refining isolated showed 

 losses of o • I per cent, and i • 5 per cent, and profits of 5 • 5 per cent, to 

 13-5 per cent. 



A chance for research into scope — under almost experimental condi- 

 tions — is also offered wherever a given firm or a whole industry deliber- 

 ately changes its policy. I have elsewhere collected several instances of 

 particular standardisation, often effected with the help of branding and 

 advertisement, with its results on cost and profit.^^ One outstanding 

 State experiment in general standardisation lies to hand in the ' Grid ' 

 policy of the Electricity Commissioners. There is some controversy 



'" Recent Economic Changes, Vol. i., p. 194. 



'^ The Logic of Industrial Organisation, pp. 25-29. 



