An experimental investigation was conducted in which distilled water under normal atmospheric conditions was atomised by a high velocity air stream. The atomising unit (which also could be used for solid-injection atomisation) consisted of a glass convergent air nozzle with a short straight section and a concentric hypodermic tube for the purpose of water injection. The purpose of the investigation was to determine the effect upon the efficacy and comparative degree of atomization by using various methods of water injection, and by varying the injection position along the axis of air flow, for use in possible engineering applications. The sprayed droplets were collected on magnesium oxide coated glass slides, housed in a specially constructed sampling apparatus which exposed the slides for an appropriate time. Three samples were taken for each test at the horizontal axis of the spray, and under a calibrated microscope courted until no appreciable difference in droplet distribution was noted. From the droplet distribution the volume-surface or Sauter mean diameter, do, in microns was evaluated. Precautions were taken against drop evaporation and possible settling out of drops before reaching the sampling apparatus;The air velocity, va, was varied from 106 to 316 m. per sec. The air to water volume ratio, Qa/Qw was varied from 1,000 to 20,000. The water rates used were small, and, consequently, the water velocity never exceeded one m. per sec. Four different methods of infection were used: one utilized rater injection parallel to the ambient air stream and the others with various typos of orifices used injection normal to the air stream;In light of this investigation, the following conclusions can be made. Liquid injection parallel to the air stream results in a better degree of atomisation than liquid injection normal to the ambient air stream. The former method does not seem to impair the normal process of attenuation of the liquid surface, ligament formation, and drop formation as does the latter method. Differences in the degree of atomisation in the latter methods depend principally upon the amount of liquid surface that ceases in contact with the velocity air;With liquid injection parallel to the ambient air stream, for the best degree of atomisation the mass of liquid should enter the region of maximum velocity. For the convergent nozzle used, the maximum va attainable is acoustic. (Abstract shortened by UMI.)