In the Atomization regime, liquid jets breakup either within the nozzle or immediately upon entering the chamber gas and drops much smaller than the jet diameter are formed. The mechanism of Atomization, which is presently unknown, was investigated by the simultaneous use of two photographic techniques. The initial transient was observed with a 106 frames/s camera and the steady state by a technique similar to spark photography. The experiment range was: liquid pressure 500 to 2500 psia; five mixtures of water and glycerol to vary the liquid viscosity; air, nitrogen, helium, and xenon at up to 600 psia as chamber gases to separate gas pressure from gas density effects; and 14 nozzle designs. Not changed were the temperature (room value), the nozzle diameter (340 μ), and the surface tension (70 dyne/cm). It was found that: jet divergence begins progressively closer to the nozzle exit as the gas density increases until it reaches the exit with no evidence of abrupt change; the divergence angle (spray angle) increases with increasing gas density, and sharpness of nozzle inlet and with decreasing liquid viscosity and nozzle length; divergence angle and jet intact length are quasi-steady with respect to upstream pressure changes which occur on time scales greater than 10 to 30 μs; aerodynamic effects, liquid turbulence, jet velocity profile rearrangments, and liquid pressure oscillations, each could not alone be the mechanism of atomization; cavitation or aerodynamic effects, supplemented by cavitation and/or wall boundary layer relaxation processes, could each be the mechanism of atomization; and the criterion Weg > 40.3 for the onset of atomization and a commonly used gas jet expression to predict the spray divergence angle are inadequate. Equations are given for the divergence angle and the onset of atomization which are valid within the tested range.