A contact lens optical lever was used to measure two-dimensional fixation characteristics when experienced Ss fixated at various positions within simple forms small enough to fall entirely on the foveal floor ( < 80 min). Fixation stability and the average direction of the line of sight were not markedly or systematically affected by the shape of the fixation stimulus. Similar results were obtained when all saccades were suppressed and the line of sight maintained exclusively by means of slow control. These results cast doubt on motor theories of form perception.Eye movements or tendencies toward eye movements have been proposed as the explanatory principle in motor theories of visual illusions and form perception for more than a century (Lotze, 1852; Wundt, 1910;Hebb, 1949;Festinger, 1971). According to these theories, visual perception of form is based on the pattern of oculomotor responses or, more recently, the programming of these responses. The particular eye-movement pattern is determined by the shape of the retinal stimulus. Similarly, illusions are explained by reference to tendencies to eye movements that produce an incongruity between the pattern of oculomotor responses made or programmed and the pattern of retinal stimulation. Despite theorizing and much speculation, very little is actually known about the relationship between tine-grain characteristics of the human oculomotor pattern and the configuration of a visually presented stimulus. Detailed study of this relationship is essential not only for an evaluation of motor theories of form processing but also for complete descriptions of human oculomotor characteristics. The present experiments examine this relationship. METHOD Eye-Movement Recording and AnalysisAn electronic contact-lens optical lever was used to make continuous two-dimensional recordings of eye position. A detailed description of the apparatus has been reported by Haddad and Steinman (\973). As used in the present experiments, the recording limits of this instrument were I deg of arc on both meridians. permitting resolution of eye position to approximately 3 sec of arc. For each trial, the mean and standard deviation of eye position voltages on the horizontal and vertical meridians were calculated on-line by a special-purpose analog device. The time constants of the integrating circuits in this device were chosen empirically such that means and standard deviations estimated by the device from a random sample of fixation trials for each S matched. to within 1%, means and standard deviations calculated from a random sample of brief eye-position voltages taken from the same trials. The output of thc analog device was digitized and printed out at the end of each trial. StimuliThe stimuli were high-contrast (>900/0) photographic transparencies of computer-generated and drawn forms. The transparencies were shown in rear projection on a diffusing screen (Callier's coefficient = 4.6) that had no visible texture at the viewing distance of 2 m. A pellicle was placed in the S's line of sight...