Kienle[3] made experiments at Wilmington, Del., in November, 1912, and obtained a constant flow of gas by means of high- and low-pressure valves; the gas was dissolved in water in an absorption tower and afterwards fed to the water to be treated. Van Loan and Thomas of Philadelphia experimented with liquid chlorine on a large scale at the Belmont Filter Plant in September, 1912. The chlorine was fed into the filtered water basin in the gaseous state and the quantity was regulated by the loss in weight of the containers. The dosage was approximately 0.14 p.p.m. (West[4]). Jackson, of Brooklyn, made similar experiments about the same time at the Ridgewood Reservoir, Brooklyn, and his type of apparatus was shortly afterwards put on the market as the Leavitt-Jackson Liquid Chlorine Machine. The regulation of the flow in this machine was determined by the loss in weight of the gas cylinder which was suspended from a sensitive scale beam. By moving the counterbalancing weight on the beam at a constant rate, a uniform flow of gas was obtained, the area of the orifice being kept constant by the equilibrium in the balance operating controlling valves through a system of levers. This type of apparatus was tried at several places but it was found that the adjustment of the regulating mechanism was too sensitive and produced considerable irregularities in the flow of gas. The type used by Ornstein and Kienle were combined and commercially developed by the Electric Bleaching Gas Co. of New York.[A] In this combined type the gas was collected from one or more cylinders by means of a manifold which delivered it to the regulating mechanism at the pressure indicated by a gauge attached to the inlet pipe. Beyond this gauge were two pressure-regulating devices, the first being used primarily to reduce the initial pressure to about 15 pounds per square inch, and the second for controlling the pressure through a range sufficient to give the desired discharge of gas. The gas from the second regulator passed through an orifice in a plate at a pressure indicated by a suitable gauge which was calibrated in terms of weight of chlorine per unit of time. The gas, on leaving the regulating apparatus, passed up an absorption tower of hard rubber, where it met a descending stream of water. The solution was carried by suitable piping to the point of application. This type was modified in some cases by the substitution of a flow meter of the float type for the inferential pressure meter. [A] This type has recently been withdrawn from the market. [Illustration: FIG. 6.--Manual Control Chlorinator, Solution Feed, Type A.]