316 



ADAPTATIONS TO SPACE AND MOTION 



of the eyes happen to be, if only the retinal images are of the right kind 

 and in the right places, even if put there by an arrangement of prisms or 

 mirrors in an experimental situation. 



The two retinal images must either be left- and right-eyed views of an 

 actual object or, in the case of drawings which are to be observed in a 

 stereoscope (Fig. 117) they must represent such views of some possible 

 solid object — even if it be an imaginary geometrical figure or a gimmick 

 the like of which the observer has never seen. The two single-eyed views 

 of a solid object can never be identical even if the object is a smooth ball 

 — unless it is so lighted that it has no shadow which can be seen more 



PieTUWE 





-P^-W-I 



, 1/ s 



<a 



M 



Fig. 117 — Optics of the Brewster- 

 Holmes stereoscope. 



A card, c, bearing a left-eyed image //' 

 and a right-eyed image ri of some three- 

 dimensional scene, is observed through 

 the half-lenses hi, whose prism action so 

 bends the light rays that the retinal 

 images are projected to a common place 

 in space for which the eyes are con- 

 verged and accommodated. At this place, 

 a binocular stereoscopic image si is seen. 

 The screen s prevents each eye from 

 seeing the picture not intended for it. 



fully by one eye than by the other. Yet any two pictures placed in a 

 stereoscope must be as nearly identical as right- and left-eyed views are, 

 or they cannot be 'fused' and will be seen doubly or even alternately by 

 the baffled brain, in the phenomenon mis-called retinal rivalry. 



The left eye sees a Httle way around one side of an object, the right 

 eye a little way around the other. Naturally enough, if these two images 

 are fused at all into a single central or cerebral image, the rotundity of 

 the object is perceived. Of course if the object is two-dimensional, it will 

 be perceived as such; but even so it will be seen singly, through the 

 fusion of two one-eyed images. Here, no third dimension is created, not 

 because the object hasn't one but because the two retinal images in this 



