When smooth matt test surfaces are suspended by fine threads in a strong illumination gradient and in isolation from any ordinary background, lightness constancy fails. It may be restored by introducing other objects into the field in the same plane as the test surfaces. Using a large number of subjects making independent judgements by a comparison scale method, the present investigation attempts to discover the principles underlying the restoration of lightness constancy by the introduction of inducing surfaces of varied size and reflectance at varied distances from the test surface.

This is a descriptive study of the various judgments of shape which are spontaneously made when the inclination of an object is varied. The full range of inclinations is examined, using independent and unsophisticated subjects at each inclination, and recording subjects' descriptions.The chief findings are: (1) T w o main judgments are possible: judgments of real shape (R), which approximate to the true shape at all inclinations; and judgments contrasted with these (not-real or N), which show typically a compromise between true and retinal shape. (2) Some subjects (group RN) are spontaneously aware of both possibilities. They describe the N judg ment as an attempt to equate on the basis of a direct, instantaneous, visual impression; whereas the R judgment involves 'allowing for the slope'. Those subjects predominate at greater inclinations. (3) The remaining subjects (group RO) are spontaneously aware of the R possibility only. For these subjects, the R judgment seems to approximate closely to that quality of direct visual impression which for group RN characterizes the N judgment.It is suggested that judgments of real shape involve a skill. Group RN are in an 'investigating' frame of mind, and offer an analysis of this skill. Group RO, in an 'unreflecting' frame of mind, take it for granted. Determining factors are also noted in the experimental situation and the instructions. It is suggested that previous studies fail to take adequate notice of these factors, and so give partial accounts.

A number of studies have led to the generalization that there is a tendency to “overconstancy” in the perception of size. The experiments reported here suggest that over-estimation is limited to objects subtending angles of approximately 2° or less, and that objects subtending greater angles are judged fairly accurately. This raises the possibility that foveal diameter (approximately 2°) is involved.

No abstract

No abstract