The purpose of the study was to apply a Markov technique to a chemical process that cannot be adequately controlled by the use of conventional laboratory or process control techniques. The case method of study was applied to a phase of the beet-sugar refining process. This process is a beet juice purifying step known in the industry as carbonation or defecation. It was chosen because it is an excellent example of a commercial chemical process that has developed without an adequately verified theory. Input data to the study consisted of actual plant operating data, taken at four hour intervals, for the entire campaign of 1965. The data were reduced to a Markovian format and processed on the 1620 Computer at San Diego State College. Purity of juices entering and leaving the process were categorized into "states" and used as a measure of process performance. This formed a basis for transition matrix construction, that were used as stochastic models of the system. Changes in two major process variables, calcium oxide concentration of in-process juice, and alkalinity of purified juice were studied intensively with respect to their effect on transition matrices. Relative values were assigned to the "states" of the system in order to produce a decision criteria termed "expected immediate relative value". Finally, decision matrices were constructed to aid in selecting courses of action, to be used after technological know-how is exhausted, based upon maximizing the "expected immediate relative value". Reliability of these matrices could not be established within the limitations of the study, but a method of checking their validity, using a in-plant trial period, is outlined. The primary contribution of this work is the development of a general tentative method for applying a Markov analysis to certain types of chemical processes coupled with a detailed actual case application. The author of this work hopes to follow-up on the reliability of this technique should the results of a proposed trial become available.