In the summer of 1903 the late Richard B. Dole, chemist of the water-resources branch of the United States Geological Survey, began a systematic investigation of the composition of the river and lake waters of the United States. His plan, which developed gradually, was to have analyses made of the different waters in such a manner as to give the average composition of each one for an entire year. For a few waters, such completeness was impracticable, the analyses covered only part of a year, but even in these waters the data obtained were of much value. As a rule, samples of each water were collected day by day. They were then mixed in sets of ten and analyzed, so that for each river or lake from 34 to 37 analyses were made. For the Mississippi above New Orleans composite analyses were made in sets of seven, giving 52 analyses from which to compute the average. For the Great Lakes, however, only monthly samples were taken, for the reason that their waters vary so little in composition that greater elaboration was not necessary. Some of the larger rivers were treated even more thoroughly; their average composition was determined at more than one point – the Mississippi at six points. For some rivers the analyses cover two years of collection, and for the data, received from a contributor not connected with the Geological Survey, three years.