According to the diffusion-reaction model developed in Part II, the product distribution from two competitive, consecutive reactions will be a function of the stoichiometric and volumetric ratios of the reagents, the ratio of the rate constants, a time parameter and a Thiele-like modulus. The time parameter is large for fast reactions, signifying that yield and conversion are independent of it (i.e. reaction then goes to completion, whereby the limiting reagent is used up). The diazo coupling reactions of Part I have sufficiently well-defined properties that the models of Part II and [2] for CSTR and batch reactor respectively can be evaluated. In a first model-experiment comparison the Thiele modulus was changed by varying the viscosity and the turbine speed of a CSTR. Despite significant scatter the agreement was good; the size of the reaction zone was about half the Kolmogoroff turbulence microscale. In a second comparison using two geometrically similar reactors the principal variables were mode of operation (semi-continuous and continuous), turbine speed, feed point (3 locations) and scale (0.0025 and 0.063 m 3) . The sizes of the reaction zone ranged from 10 to 40% of the Kolmogoroff scale, the ranking being consistent with known flow patterns. Laminar shear of the reaction zone during reaction might explain this result. Various scale-up rules were evaluated. The unexpected and potentially dangerous phenomenon of back-mixing into the feed pipe, with consequent pre-reaction, was observed and studied empirically.