The joints that constitute the fractured rock masses normally change dramatically. After measuring the orientation of numerous joints in a given area

it is imperative to categorize the joints. Unfortunately

there is no widely accepted method to categorize joints at present. Joints are categorized into several clusters by using quick cluster analysis. The parameters of Fisher distribution model are estimated by maximum likelihood

and the model is finally verified by using Pearson test. The results demonstrate that the categorized clusters must be suitable

otherwise cluster centroids would change dramatically and the data around the centroids are not continuous

which leads to the failure of Pearson test. By using the proposed methods

it is effective to categorize the fissures in practice.