THE PRE-WAR QUARTER CENTURY 
and the marked boom in 1913 find no close counterparts on the 
employment curve. Also the depression of 1911 is more clearly 
defined on the immigration curve. Various minor irregularities 
may also be detected, but none of these differences appear sufficient 
to overcome the presumption raised by the general similarity in 
movement, that fluctuations in the two series are to a large extent 
cause and effect or are dominated by common causes. 
When the curves are closely examined to determine the extent 
to which the major turns in the two curves coincide, it appears that 
they agree most closely when it is assumed that the fluctuations in 
immigration lag from two to four months after the corresponding 
fluctuations in the employment curve. This statement is not, of 
course, to be interpreted as meaning that the lag is always from two 
to four months. For example, the high points in 1893 and the low 
points in 1908 appear to be approximately simultaneous. 
Indexes of Industrial and Commercial Activity. 
The employment curve used in the above comparison is cons- 
tructed from limited material and, as noted, there are some relatively 
large fluctuations in migration, notably in 1892, 1905, 1909, 1911, 
and 1913, for which the factory employment curve does not afford 
adequate explanation. 
Pig Iron Production. 
For additional evidence we turn first to a comparison with an 
index of pig iron production corrected for computed trend and 
seasonal variation by methods described in Chapter III. In Chart 
14 this series is plotted together with the male immigration curve 
for the quarter century 1890-1914. As in the comparison with the 
factory employment curve, in most of their major movements the 
two curves agree, especially in the second half of the period. The 
immigration curve appears to lag at the major turn from one to 
four months, although there are a number of instances, particularly 
at the turning points of moderate booms or depressions, where the 
immigration curve turns first.: 
An examination of the pig iron curve affords some explanation 
of certain immigration fluctuations which we noted as contrary to 
"The conclusion reached from visual inspection of the charts is supported by the 
mathematical computation of coefficients of correlation with various intervals of lag. 
When it is assumed that there is no lag, a coefficient of -+.66 is obtained; when im- 
migration is assumed to lag two months, the coefficient is 4.76; four months, +.72; 
and six months, 4.55. 
*With no lag assumed, the correlation of the monthly indices is +.63 + .02. 
03