The central purpose of this research has been to examine the possibility of the plasticity of individual saccadic parameters and to identify their patterns of covariation. Experiments using trial-by-trial feedback or continuous on-going feedback methods demonstrated that the saccadic generator can prolong the duration of a saccade above normal levels. However, slowed peak velocities were only evident with continuous feedback. The results showed that although continuous feedback was a more effective method for inducing modifications of saccade trajectories, the effects of trial-by-trial feedback schedules were enhanced by distributing the training over several days. The instructions to produce slow saccades caused a 'staircase' pattern of eye movements in which a continuous sequence of hypometric saccades was manifested. These saccades had latencies 400-600 msecs longer than normal visually controlled eye movements. Detailed measurements (chapter 4) of the latencies of the saccadic components indicated that each component was programmed independently. In chapter 5 the Gurevich claim that a saccade velocity depends only on the spatial magnitude of the saccade was tested by measuring saccades of equivalent amplitudes to targets which varied in movement duration and velocity. The accuracy of saccades but not their peak velocities or durations was sensitive to manipulations in the temporal characteristics of the target. Experiments in chapter 6 showed that the use of spatial signals in the aiming of a saccade can be systematically controlled. When subjects were trained for several days in a visual discrimination task the accuracy of the initial saccadic movement increased over time. The results of these experiments seriously question the Young and Stark (1963b) ballistic model and other formulations (Westheimer, 1954b, 1973; Yarbus, 1967) which assume that the saccadic system operates according to stereotyped mechanisms.