F— ECONOMIC SCIENCE AND STATISTICS 135 



Census of Production for 1930.1° The situation is shown to be surpris- 

 ingly similar to that already known for the United States of America, 

 through its long series of decennial Censuses of Manufactures. Even 

 when due allowance is made for differences of definitional the average 

 size of the British appears extraordinarily similar to that of the con- 

 temporary American factory. But the interest does not lie in average 

 size for industry as a whole so much as in the typical differences between 

 industries. These differences, again, seem extraordinarily similar in Great 

 Britain and the United States, and indeed in Germany, too. 



Some years ago, before the publication of the 1929 American and the 

 1930 British Census, I presented a table showing the distribution of the 

 wage-earning population of the U.S.A. in 1909 among plants of eight 

 ranges of size in fourteen leading manufacturing industries.!^ I have 

 now been able to draw up a similar table for the U.S.A. in 1929, for 

 Great Britain in 1930, and Germany in 1925, adding three more indus- 

 tries, furniture, hosiery and chemicals that have since become equally 

 leading. Only a brief summary can be given here, by presenting in a 

 table the percentage of all employees that are found in plants of over 

 500 employees.!^ As the footnotes to the table indicate, a certain caution 

 must be used in comparing these percentages too literally as between 

 different countries ; yet the conclusion is justified from the general 

 distribution of workers that in spite of difference in country, the same 

 size of plant tends to prevail in the same industries. This conclusion 

 would indicate that in determining size of factories there are technical 

 or economic factors at work common to all industrialised countries, 

 rather than factors particular to any one country. To be of use to policy 

 research must discover these underlying factors. 



At Birmingham we are now engaged in tracing the relation of size of 

 plant with the degree of mechanisation. Mechanisation is difficult to 

 measure directly, but some cross between horse-power and overhead costs 

 per worker is perhaps the most sensitive index. Though it is easiest to 

 obtain, horse-power per head fails to indicate the presence of many hand- 

 operated machines, as in shoe-making, and must be supplemented by some 

 measure of overhead costs. This is obtainable in the British Census by 

 subtracting for each industry average earnings per operative from net 

 output per head.i* Admittedly, overheads per head may measure other 

 elements besides mechanisation, such as specialised salaried staff and 

 marketing cost, yet the machine costs such as maintenance, repair, depre- 

 ciation and obsolescence are probably predominant. On the whole the 



'" Where he owns more than one factory it is officially left to the employer to 

 decide whether to return the size of each single factory or the whole firm. Com- 

 parison with statistics of the distribution of sizes provided by H.M. Chief In- 

 spector of Factories point to the size of single factories as that normally returned. 



11 See Footnotes to the Table. 



'= Logic of Industrial Organisation, Table II, p. 30. The reason for the choice 

 of industries and for omitting 1919, is given in a footnote, p. 29. 



" Gas has been omitted from the original fourteen industries as being a public 

 utility rather than private manufacture. 



*■* See table pages 102-103. Pinal Report on the Fourth Census of Production 

 (1930). 



