This investigation into the effects of configural properties (properties determined by the interrelationships existing between component parts) used a selective attention task to determine whether intersection is a primary encoding feature or is constructed after slopes and heights are perceived. The method for encoding feature identification, primary encoding features, and the question of whether or not configural superiority can be modified are discussed. The subjects (N=12) were adults with no prior experience in card sorting tasks. The task required subjects to classify stimuli that vary in two of their properties - speed and accuracy. A previous task with stimuli varying in only one dimension served as a baseline. Results indicate that the configural property of intersection is perceived directly rather than constructed from the stimulus components. Modifications to configurations, the experimental methodology and statistical analysis including error analysis are described. The geometric concept of intersection is described as analogous to the statistical concept of crossed interaction and to its mathematical interpretation as a departure from parallelism. The results of the experiment are discussed relative to the usefulness of the findings to graph design. A list of 13 references, 3 figures and 3 tables are attached. (JM)