MIGRATION AND BUSINESS CYCLES 
railroad constructed, partly on a calendar year and partly on a 
fiscal year basis, are available throughout the period covered by 
Charts 6 and 7 (1870-1923) ; but the best of the construction indices, 
the volume of building covered by contracts awarded, is available 
only beginning in 1910 and has changed somewhat in scope during 
this period. However, an estimate of the annual total value of 
construction is given in Chart 6 as a rough index of employment 
conditions in the construction industry. 
The Fiscal Year Group. 
To aid in the identification of boom and depression periods when 
data applying to fiscal years ending June 30th are considered, we 
have used annual statistics of the number of miles of railroad con- 
structed, imports of merchandise, the estimated average daily pro- 
duction of pig iron, the estimated average number employed in 
factories, and an index of business conditions compiled by Mr. 
Carl Snyder, of the Federal Reserve Bank of New York. 
The curve for factory employment represents an estimate based 
primarily upon data more fully described later in this chapter in 
connection with monthly estimates of employment. This curve 
presumably underestimates somewhat the size of the fluctuations in 
factory employment, in that it gives no consideration to part-time 
employment and, also, particularly in the earlier years, is based 
primarily upon data for Massachusetts, in which State industrial 
conditions were probably relatively stable. 
The mileage of railroad constructed is significant because it 
reflects general industrial conditions and because immigrant laborers 
in large numbers have been employed as laborers in such construc- 
tion. 
Method of Interpreting the Accompanying Charts (6 and 7). 
The several series discussed in the above paragraphs are plotted 
in Charts 6 and 7, which are so-called ‘‘ratio charts,” or charts with 
the vertical scale so proportioned that equal percentage declines 
between any two years are represented by equal vertical declines on 
the curves involved. If one curve declines ten per cent in 1900, 
and another series ten per cent in 1900 and also in 1904, in each of 
these three cases the vertical drop on the chart would be the same. 
In like manner, equal percentage increases are represented by equal 
vertical rises in the respective curves. Hence, despite the fact that 
the series are expressed in widely different units, it is possible, by 
58