336 On a new Species of Stereoscopic Phenomenon. 



of the retinae (since the image on the retina almost corresponds 

 with the object, though of less dimensions, and in an inverted 

 position*). If, on the other hand, a point in the circle be seen 

 stereoscopically, the rays A G and B H, which cut each other in D, 

 fall on parts of the retinae almost corresponding, since in this 

 particular case, indeed, the arcs E G and H are absolutely 

 equal ; and therefore the images on the retinae will almost entirely 

 coincide. In complicated cases the geometrical coincidence is 

 not so approximately exact, but in' any case it is very evident 

 that the rays that unite in a point in the circle fall on points of 

 the retina? of the two eyes far more nearly correspondent than 

 those belonging to a point in the curve at right angles to the 

 circles. And if, in any position of the apparatus and the eyes, 

 the sections are no longer a circle and conic section, a similar 

 observation will always enable us to determine which of the 

 intersections of the two cones we shall see stereoscopically j\ 



* The above observations appear to contradict the views advanced by 

 Mr. W. B. Rogers (American Journal, vols. xx. and xxi. 1855 and 1856), 

 in a memoir which only came to my knowledge after I had written the 

 above paper. In this memoir there is a very complete investigation of the 

 appearance presented by very dissimilar drawings when united by means 

 of a simple stereoscope of suitable construction. In mentioning different 

 stereoscopic drawings, it is stated (vol. xxi. p. 1 /6) that two equal circular 

 arcs which are convex ) ( or concave ( ) to each other, unite themselves 

 stereoscopically into a hyperbolic arc. This, therefore, is analogous to our 

 seeing a parabola (ellipse or hyperbola) in the experiment described in this 

 paper. This is in general correct, if only the arcs are not too great in 

 proportion to the visual angle, since, when the arcs are considerably curved, 

 the experiment requires eyes much practised in the use of the stereoscope, 

 and even then it is effected with difficulty. Two complete circles, more- 

 over, appear, as far as these experiments extend, never to be seen united 

 in this way ; and it is certain that if the mind has the choice to conceive 

 the impressions united in different ways, it prefers that arrangement which 

 unites stereoscopically the impressions on the points of the retinae which 

 most nearly correspond. If, therefore, on account of this fact, the remarks 

 made in this essay are deprived of complete generality, the same circum- 

 stance seems, on the other hand, to confirm the general principle, that it 

 is difficult to unite stereoscopically two images that belong to very different 

 points of the retinae ; especially since the above-mentioned experiment of 

 Mr. Rogers's seems to require a greater exertion than most other stereo- 

 scopic experiments. 



That, moreover, only one stereoscopic image ought in general to be anti- 

 cipated from more complex drawings (if the left-hand drawing were not 

 sometimes observed with the right eye, and conversely) hardly needs to be 

 mentioned, since it obviously must be so. 



f Fig. 7 shows that the rays seen at the same time are necessarily in 

 different planes. While, for instance, the eye B sees the point II, that is 

 to say, while the cylinder passes through the position OH, the eye A sees 

 the point H', which also lies in the line Oil, and it is clear that AH' and 

 BH are not in the same plane. 



