For years other than 1910, indirect calculations must be resorted to for domestic trade. I have substantial confidence in the rough accuracy of the figure chosen for 1910 in view of the convergence of two widely different sets of data. My figure for retail deposits in 1909 is $32,000,000,000. King's figure for total income is $30,500,000,000 for 1910. King's figure seems to me a better figure to use for the purpose in hand. I use my own merely as a rough check on his. For years other than 1910, the figure for net income is calculated as a percentage of King's figure for 1910, by means of an "index of variation." It is assumed that the net income of 1905, for example, bears the same relation to the index for 1905 that the absolute figure for net income of 1910 bears to the index for 1910, and net income for 1905 is then computed by "the rule of three." The index of variation chosen is _railway gross receipts_ weighted by _wholesale prices_. I think that railway gross receipts are, on the whole, the most dependable and easily manageable index of physical volume of production that we have, though recognizing difficulties, later to be discussed, in using them for the purpose in hand. Railroads touch virtually every kind of business in the country. Variations in the _pecuniary_ volume of production and consumption, however, if due to rising or falling _prices_, rather than to changing physical volume, would not be indicated by changes in railway gross receipts. The same volume of transportation might represent widely varying pecuniary values of goods transported. Railway rates do not vary from year to year with prices of goods, even though high-priced goods are normally charged higher rates than low-priced goods. The index, therefore, must include _prices_ as well as physical volume of transportation. For 1910, therefore, railway gross receipts and an index of prices are multiplied together, and counted as 100%. The same thing is done for railway gross receipts and prices for other years, and the results reduced to percentages of the result for 1910. The figure for net income in any other year is then readily computed as a percentage of the figure for 1910. The results, for the years 1890-1916, appear in the tables below.[305]