AS MOST water works men know, the Treasury standard for safe 1\.. drinking water on interstate carriers provides that not more than one organism of the coliform group shall be present in 100.0 mI. of water. The present standard practice, of course, is to test at least five 1O.0-mI. portions of finished water for control purposes. One positive tube in a series of five, however, yields an index of approximately 2.0 instead of 1.0 per 100.0 mI.Many who are responsible for maintaining a safe drinking water supply, and those who are concerned with standards of control, naturally feel that if 100.0-, 500.0-or perhaps 1,000-mI. samples could be tested, it would increase the chance of detecting contamination and make it possible to evaluate the water with indices of less than 1.0 per 100.0 mI., e.g., 1.0 per 1,000 mI. Experiments have been made in the laboratories of the Chicago Health Department to determine the practical value of such a technique. As less than 35 experiments with 100.0-, 50.0and 1O.0-mI. portions have been made, the work can hardly be considered more than exploratory. Nevertheless, the considerable amount of information afforded by the results seemed to justify making them available at this time.Briefly, the procedure of the experiments was as follows. Believing it would be of statistical advantage to test 3 liters of water in parallel, ten 100.0-mI., twenty 50.0-mI., and one hundred 10.0-mI. volumes of double strength standard lactose broth were inoculated with equal volumes of test water. The latter consisted of raw Lake Michigan water, the coliform content of which was low or zero as determined by overnight direct plate counts with brilliant green lactose bile agar (1). To 3! liters of the water, held in the refrigerator, were