This paper is concerned with research, the object of which is to discover the mechanism for a particular phenomenon leading to a specific mathematical model. Such investigations are distinguished from those in which the object is merely to estimate the output y of a process over a range of values of the input ∊1, ∊2, …, ∊ k . Frequently, a number of possible mechanisms are suggested from theoretical considerations leading to a number of different mathematical models. To discriminate among these a sequential procedure is developed in which calculations made after each experiment determine the most discriminatory process conditions for use in the next experiment. The method is illustrated with examples.