119 MIGRATION AND BUSINESS CYCLES 
data available makes a precise answer to this question impossible, 
particularly for the earlier years. However, the available estimates 
are probably accurate enough to establish within reasonable limits 
of error the volume relation between migration and unemployment. 
Professor Hornell Hart has made an estimate of the numbers 
unemployed in non-agricultural occupations by months from1902 
to 1917, which will serve for a preliminary comparison of the volume 
of immigration and the estimated volume of unemployment. 
There is, as suggested above, necessarily a considerable margin of 
error in these estimates and consequently they should be looked 
upon, not as giving a refined measure of the volume of unemploy- 
ment, but as an approximation probably sufficiently close to the 
truth to permit rough comparisons to be made with a reasonable 
degree of accuracy. 
Upon what basis should the volume of immigration and employ- 
ment be compared? 
Immigration is Appropriately,Compared with Changes in the Volume 
of Employment. 
Immigration represents an addition to the supply of labor over 
a period of time, and, as to numerical volume, is logically comparable, 
not so much with the amount of unemployment existing at a given 
time, as with the change in the number unemployed over the same 
period of time. This principle may be illustrated by making the 
assumption that the domestic labor supply is kept regularly em- 
ployed, with no seasonal or cyclical unemployment. Under such 
conditions, any changes in the volume of employment would rep- 
resent additions to or subtractions from the labor supply by migra- 
tion. Thus, with migratory workers as the sole fluctuating element 
in employment, there would be perfect correlation between the 
fluctuations in migration and those in employment. An increase of 
100,000 in the number employed would be accompanied by a net 
immigration of 100,000; a decrease of 50,000 in employment, by a 
net emigration of 50,000. 
But under conditions as they actually exist, employment changes 
do not correlate perfectly in numbers with migration, and the dis- 
crepancy represents either a failure of the immigrants to obtain 
employment or a change in the number of domestic workers em- 
ployed. If the net migration is always less than the employment 
change, but in the same direction—for example, if an excess of 
5For a somewhat fuller description of this estimate, see Chapter III. 
il