The work is divisible into three interconnected areas, the first of which is concerned with the empirical modelling of the rubber internal mixing process. Here, the mixing machine variables were changed in a statistical experimental design, and several resulting dispersion dependent material responses monitored. The values of these were next combined into a composite function which related to the goodness of mixing. A response equation in terms of this function and the process variables was obtained by regression analysis. Other response equations derived were for relating mixing time and dumped stock temperature to the input variables. These graduating functions were tested for adequacy of fit; graphical representations of the predictions of the models are shown in the form of contour graphs and isometric projections (both hard copy and colour graphics). These show visual evidence of the influence of each machine variable (and interaction) on the mixing process as a whole. Also by these means, it was possible to establish optimum conditions of mixing. Next nominally identical carbon blacks from different manufacturing locations were mixed into rubber at defined processing conditions. From tests on these mixes the blacks were noted to have different dispersibilities. This was possibly attributable to the varying fines fractions and moisture contents in the test carbon blacks. To check this hypothesis, these factors were varied systematically in two common black types. The compounds resulting from mixing these experimental blacks were subjected to a wide range of tests to establish whether fines or moisture significantly affected dispersibility of blacks in rubbers. Ouring the preceding work, and also in the course of a critical review included here, it became evident that the commonly used methods for establishing dispersion (and thus mixing efficiency) were lacking, from several points of view. Therefore lastly a technique based on the analysis of an image of a cut rubber surface viewed by dark field reflected light microscopy was developed. Specimen preparation is simple and analysis is accomplished using a standard television system and oscilloscope. This method yields results which discriminate between all levels of black dispersion and also correlate admirably with a variety of processing, mechanical and microscopical properties. In the final stages of the work the capabilities of the method were expanded (in cooperation with Dunlop Ltd, Technology Division) by interfacing a computer and peripherals with the original equipment; such that data acquisition, manipulation and parameter output was made easier, faster and thus more effective than in the initial concept.