114 ECONOMIC ESSAYS IN HONOR OF JOHN BATES CLARK 
a reward for a positive contribution by a fourth factor of 
production. 
But natural resources, at least, are a third factor and the ques- 
tion naturally arises how they may be fitted into the analysis? 
A method which naturally suggests itself is to compare labor 
with a combination of land and capital. Since the elasticity 
of the supply of natural resources, if not precisely zero, is cer- 
tainly very close to it, the combination of land with capital 
will (if the supply curve of the latter is positive) make the com- 
posite elasticity of the two less than that for capital alone. 
In securing the composite elasticity for these two factors, the 
elasticity of each factor should, of course, be weighted by the 
percentage of the national income originally enjoyed by each. 
The comparison of how labor fared as compared with the com- 
posite fortunes of the owners of land and capital would afford 
a basis for judging the effect of given changes upon service 
income as compared with property income, and hence would 
be valuable in itself. 
The relative effects produced upon rent as compared with (1) 
wages and (2) interest, could then be studied in turn and their 
results isolated. Since labor and capital (and hence wages and 
interest) have previously been compared for the purpose of isolat- 
ing the effects, labor and natural resources could also be merged 
together and compared with capital. It would be possible 
then to disentangle the approximate effects produced on each of 
the factors and to frame a general conclusion for each according 
to its relative coefficient of elasticity and the relative share which 
it originally received of the total product. 
2. Real difficulties are encountered when we move to a con- 
sideration of several commodities. Hitherto we have been dealing 
with only one and consequently have taken into account only one 
general productivity surface, composed as it was of (a) the rate 
of increase of the total product with equal proportional changes 
in the factors, (b) the rate of slope of the product as the propor- 
tion of X to a constant quantity Y was altered, and (c) the rate 
of slope of the product as the ratio of Y to a constant quantity of 
X was altered. 
But as we deal with several commodities, we encounter diverg- 
ing slopes of marginal productivity as measured in terms of 
physical units, and the question naturally arises how these