At the time when the abnormal results were obtained with ammonium hypochlorite and mixtures of bleach and ammonia, the phenomenon appeared to be of scientific interest only and especially so as Rideal had attributed the obnoxious tastes and odours, sometimes produced by chlorination, to the formation of chloramines. During the winter of 1915-1916 the price of bleach, however, advanced to extraordinary heights and the author then determined to try out the process on a practical scale for the purification of water. A subsidiary plant pumping about 200,000 Imperial gallons per day (240,000 U. S. A. gallons) was found to be available for this purpose and the chloramine process was substituted for the bleach method previously in operation. The process was commenced by the addition of pure ammonia fort, in the amount required to give a chlorine to ammonia ratio of 2:1, to the bleach solutions in the barrels. The results were not in accordance with those obtained in the laboratory and it was found that the samples of bleach solutions received for analysis were far below the strength calculated from the amount of dry bleach used. This experience was repeated on subsequent days and the deficiency was found to increase on increasing the ammonia dosage. Solutions of similar concentration were then used in the laboratory with similar losses, and it was observed that on the addition of ammonia a copious evolution of gas occurred. An investigation showed that the ammonia and bleach must be mixed as dilute solutions and prolonged contact avoided (_vide_ p. 127). Alterations were accordingly made in the plant and the bleach and ammonia were prepared as dilute solutions in separate vessels and allowed to mix for only a few seconds before delivery to the suction of the pumps. This method of application was instantaneously successful and results equal to those obtained in the laboratory were at once secured. The dosage was reduced until the bacteriological results were adversely affected and continued at values slightly in excess of this figure (0.15 p.p.m.) for a short period to prove that the process was reliable. From a consideration of the work of Raschig and Rideal, it appeared that the most efficient proportions of available chlorine and ammonia would be two parts by weight of the former to one part of the latter and this ratio was maintained during the run on the experimental plant. Lower ratios of chlorine to ammonia were contra-indicated by the laboratory experiments, which showed that the efficiency was not increased thereby whilst higher ratios were left for future consideration.