The proper quota of water for irrigation depends on climate and soil and subsoil, as well as on the nature of the crop, and varies indefinitely under diverse conditions. As a rule, the best soils require least water; those which demand most are light sands on one hand and adhesive clays on the other. Where the subsoil is open and dry, more water is needed than where it is moist or impervious. Wherever there is an impervious substratum, the subsoil accumulates moisture and the demand for water diminishes from year to year. These and other considerations so complicate the subject that it is difficult to generalize, and I have found it more practicable to use in my investigations certain limiting quantities than to attempt in every case a diagnosis of the local conditions. By comparing the volumes of certain streams in Utah, that are now used in irrigation to their full capacity, with the quantities of land that they serve, I have found that one hundred acres of dry bench land (_i. e._, land with a deep, dry, open subsoil) will not yield a full crop of grain with less than one cubic foot of water per second, and this under the most favorable climate of the Territory. Where the climate is drier, a greater quantity is required. Where there is a moist subsoil, a less may suffice. In the drier districts, where the streams are small, they are usually employed upon the dry benches, because these are most convenient to their sources; and it is very rarely the case that their utility is increased by the presence of a moist subsoil. But it is also in the drier districts that the extent of agricultural land is ascertained by the measurement of streams; and hence there is little danger of error if we use in all cases the criterion that applies to dry bench land. In the discussion of the lands of northern Utah, I have therefore assigned to each cubic foot per second of perennial flow the reclamation of one hundred acres of land, with the belief that the consequent estimates would never underrate, though they might sometimes exaggerate, the agricultural resources of the districts examined. In the measurement of streams the following method was employed: A place was sought where the channel was straight for a distance equal to several times the width of the stream, and where for some distance there was little change in the dimensions of the cross section. Measurement was then made of the width (in feet), of the mean depth (in feet), and of the maximum surface current (in feet per second). The mean current was assumed to be four-fifths of the maximum current; and four-fifths of the product of the three measured elements was taken to give the flow in cubic feet per second. This method of measurement is confessedly crude, and is liable to considerable error, but with the time at my disposal no better was practicable, and its shortcomings are less to be regretted on account of the variability of the streams themselves.