Experimental tools for the investigation of the mechanism of limited coalescence (LC) processes have been developed, and a theoretical analysis, based on Monte Carlo methods, of the particle size distributions resulting from LC, has been carried out. The method involves performing the LC process in a stepwise fashion in one of two ways. In the first, a fine emulsion of an oil phase (dodecyl phthalate in our studies) in water is made using a soap (sodium laurate) as an emulsifying agent. This emulsion is then acidified in the presence of an appropriate quantity of silica (Ludox TM) and a promoter, under which conditions a normal LC process occurs. In the second, a first LC process is performed using colloidal cupric oxide as the LC stabilizer. The CuO is then dissolved in an acidic solution containing silica and a promoter, at which time a second LC process occurs. The second method allows the initial particle size distribution to be well characterized and also allows the kinetics of the LC process to be followed. From a consideration of the results obtained from the stepwise LC studies, we conclude that practical LC procedures involve a mechanism in which reversible, shear-dependent flocculation is followed by rate-determining coalescence. Particle collisions are driven by the stirring, and not by Brownian motion. The theoretical analysis, assuming either diffusional or turbulence-driven collision, predicts particle size distributions that are in close accord with experimental results and that are much narrower than those resulting from the so-called "self-preserving distribution" resulting from simple Brownian coalescence and the even broader distributions expected for hydrodynamically driven coalescence processes.