relate the structure to the environment, and develop means of comparing systems. The key to successful synthetic modelling lies in understanding the complex structure of the ecosystem. This structure involves a very large number of interrelated variables and parameters. Hence a good analytic model is a prerequisite for any attempt to create a synthetic model. Our work has been directed toward comparing several quantitative classification methods and combining several classification and ordination techniques into a unified analytic modelling strategy. We have shown how the different techniques complement one another and thus produce a clearer analytic model than any one technique could produce by itself. Another example of the use of complementary techniques is the recent work in the analysis of tropical rainforests (Williams et al. 1969). We have applied the techniques to vegetation data collected during the initial characterization of Walker Branch Watershed. The watershed is a monitored ecosystem on which the secondary effects of forest fertilization are to be studied (Curlin & Nelson 1968). Our goal was to determine whether stands of similar species composition existed which could be used to define vegetation groups. These groups could then be used to make the sampling and study of processes more efficacious.