16 
therefore it is a boon to the very big firms. 
To return for a moment to the discussion of capital-intensity: 
It is also connected with the question of the "correct” 
measure of size of an establishment or firm. I have shown in 
another place^ that the pattern of relation of size and 
output per man is quite different according to whether size 
is measured by output or employment. (In the first case a 
monotonous increasing relation obtains, whereas in the second 
case the output per head first increases } then falls with size.) 
I also showed that the difference between the two patterns 
is due to the stochastic character of the data - the disper 
sions of the individual values of output per head round the 
average. 
What the "correct" measure of size is can not be decided 
by statistical arguments. We are free to choose our definitions 
on grounds of convenience of language, and simplicity of de 
scription. On these grounds I choose the criterion of output 
capacity; the relations of technical progress, capital-intensi 
ty and scale which were pictured earlier on could not in 
principle be changed by a different terminology but they 
would become unnecessary complicated because the size of 
p^ant^iis^t ori ca?ft.3p C $ir s t increases, and then in the age of 
automation decreases. If one chose men per plant as the 
^ ''Maturity and Stagnation in American Capitalism, Chapter IV, 
Oxford 1952.