A 'research model' can be used to investigate cause and effect relations and generally improve the understanding of extractive metallurgical processes. As such, it is a useful research technique which complements the traditional methods of laboratory and plant measurement. The procedure of model development is iterative and involves observation of the process, hypothesis of phenomena, prediction using mathematical descriptions of the phenomena and comparison of predictions with real-world measurements. Lack of agreement of predictions with real-world measurements usually indicates an inadequacy in the hypotheses and necessitates revision. Experimentation has an integral role in the process of model development but its emphasis is different from that when experimentation alone is used to study processes. Since a process as a whole is investigated by using a model, experimentation is restricted to the study of specific phenomena to aid the description of these in the model, to the characterisation of the raw materials (for initial values of the model parameters) and to the measurement of physical and chemical constants required in the model equations. Similarly, plant measurements are made for the purpose of comparison with the model predictions not to establish cause and effect relations. This makes plant experimentation much simpler and cheaper'since most measurements-can be taken during normal operation.