of Stereoscopic Phcenomenon. 335 



substantially consistent with an observation made by Wheatstone 

 on the horopter, towards the conclusion of his memoir above- 

 mentioned, and also with the views expressed by Johannes 

 Muller and Meissner on stereoscopic vision. 



8. We can now explain why, in the experiment mentioned 

 towards the conclusion of the third paragraph, we saw only the 

 circle and not the conic section in the plane perpendicular to the 

 ground line, and passing through the axis of f (#=0). Imagine 

 (PL IV. fig. 6) that with the eye A we only saw the rays a 

 and a ; with the eye B the rays b and /3 (so that the points in 

 order are a b, a /3) ; then, on looking with both eyes, should 

 we perceive stereoscopically the points ab and a/3, or the 

 points a fi and b a ? It is clear that if the point b a were 

 seen, the rays a and /3 would impinge on very different points 

 of the retina (according to either Meissner's or Recklinghausen's 

 theory), so that they would not be stereoscopically united, — so 

 much so, indeed, that if there actually were points at a /3 and 

 b a, they could not be seen stereoscopically at the same time. 

 If, however, the point a b be observed, the rays a and /3 fall 

 on parts of the retina nearly corresponding, and their impressions 

 can therefore be easily united. This is the reason why the two 

 latter points are seen and not the former. It will also be easily 

 understood that it is precisely for the same reason that, in the 

 above experiment, the circle was perceived and not the curve in 

 the plane y z, because, when one point of the circle is seen 

 stereoscopically, the images of the other corresponding points 

 are much nearer the identical retina-points in the two eyes than 

 is the case when one point in the curve in y z is so observed ; so 

 that if the curve in y z actually existed, it could not for this very 

 reason be seen stereoscopically, but would always present a 

 double image. 



The foregoing observations will perhaps be rendered clearer 

 by the particular case represented in fig. 7. Here the incident 

 rays are supposed parallel to the axis of rotation z } which bisects 

 the ground-line AB at right angles, the plane of rotation 

 being the plane xy. The eye A sees the circle EGO (drawn in 

 perspective in the figure), the eye B the circle L H 0. These 

 circles meet in 0, and have a common tangent. The two cones 

 generated by the reflected rays intersect, and are, from their 

 position, symmetrical, the lines A E and B E' being perpendicular 

 to the plane yz. The intersections of the two cones are (1) the 

 parabola K N in the plane perpendicular to A B, and (2) the 

 circle CD C in the plane parallel to the plane of rotation (ocy). 

 If, now, the point O, for instance, in the parabola be seen stereo- 

 scopically, in order to observe the point K, the rays A G and 

 B L must be united, which obviously fall on very different points 



