[95] Bull. Arizona Exp’t Station Nos. 2, 38. These variations in the character of the irrigation water must of course be watched by the farmer who does not receive directly from mountain streams, or from deep artesian wells water known to have a constant content of saline matter. THE DUTY OF IRRIGATION WATER.—The amount of water thought to be needed for the production of satisfactory crops varies widely in different regions, ranging all the way from about two feet to as much as eight annually, within the United States; while in the sugar-cane fields of the Hawaiian Islands as much as three inches per week, or over twelve acre-feet in the course of the year, have been thought to be beneficial, if not absolutely required for the best crop results. As has been stated above (chap. 12, p. 215), the rainfall limit below which irrigation becomes, if not absolutely essential, at least a highly desirable condition for the safety of crops, is usually assumed to lie at about 20 inches (500 milimeters). This general statement is, however, subject to material modification according to the manner in which the rainfall is distributed. Thus in central Montana with 24 inches of rainfall distributed throughout the year, irrigation is indispensable; while in the Santa Clara valley of central California, with an average rainfall of 15 inches falling through the winter and spring, the growth of all ordinary field crops has for fifty years not failed oftener than is commonly the case in the humid region of the North Central states. This is because in California the winter and spring are the growing seasons, while the rainless summers do not stand in the way, for crops are already harvested; and the deep rooting of trees and vines provides these with the needful moisture from the depths of the substrata (see chap. 10, pp. 163 to 173). It would thus seem that twenty inches of irrigation water properly applied ought to be sufficient for all purposes, when added to the natural rainfall, which is rarely entirely absent. Yet in actual practice less than 24 acre-inches is rarely used, and much more is the rule; 72 to 96 ins. being sometimes used in Arizona. Evidently enormous losses occur in practice, and it is of the utmost importance to discover the causes of these. _Causes of Loss._—Since irrigation water is commonly measured at the distributing weirs, loss from seepage and evaporation on the way to the fields is an obvious source of an overestimate of the water actually supplied to the farmer. In sandy districts the loss thus incurred is reliably estimated at nearly 50% in many cases. The apparent duty of the water is thus at once reduced to half its effect, and four instead of two feet of water are supposed to have been used, and are charged for.