SEASONAL FLUCTUATIONS 211 
subsequent years, the war, the abnormal situation in transport 
immediately following the war, and the influence of the quota act, 
have distorted the seasonal movement from any close resemblance 
to that exhibited prior to the war. Consequently, it has seemed 
expedient to base the computation of the normal seasonal, in most 
instances, upon the pre-war years, despite the fact that for many 
series data are available for only five pre-war years (beginning 
Jan. 1, 1909). 
Method. 
For the longer series, such as male immigrants, beginning in 1893, 
the typical seasonal has been obtained by adding and averaging 
like months (e. g. all the Januarys) and adjusting the results for 
any upward or downward bias ascribable to a trend in the data. 
The adjusted results were then translated into percentages of their 
mean, giving twelve seasonal indices or type numbers. 
In some cases, particularly for shorter series where the seasonal 
indices were to be used in isolating the cyclical movement, they have 
been computed by somewhat more refined methods, principally 
by the link-relative method developed by Professor Warren M. 
Persons or by finding the typical percentage deviation from a trend- 
cycle curve obtained by computing a twelve-month moving average 
and adjusting this average to make it represent our best estimate 
of the course of the cycle and trend. 
Quota-Period Seasonals. 
With the exception of certain classes of arrivals who are not 
counted against the quotas, the immigration law of 1921 limited 
the number of aliens of any nationality who might be admitted in 
any one year to three per cent of the foreign-born persons of such 
nationality resident in the United States as shown by the Census 
of 1910, and permitted a maximum of twenty per cent of the annual 
quota for any nationality to be admitted in any one month. The 
new quotas begin to be available on July 1st of each year, hence 
this law has tended to concentrate arrivals in July and the four 
following months. It was, therefore, necessary to make a special 
computation of the typical seasonal variation for the period since 
the quota law went into effect. This computation was based upon 
immigration data for the period from July, 1921, to June, 1924, 
inclusive. Such a short period, of course, does not give a clearly 
adequate basis for estimating the typical seasonal movement under