Mr. A. Claudet on the Illusion of Stereoscopic Images. 73 



prisms, bend laterally all the vertical lines of which the photograph 

 pictures are composed, and that these lines in the two pictures pre- 

 sent their concavity to each other, it is eiddent that the coalescence 

 of the two images must give the illusion of a concave image. 



The phsenomenon of the lateral curvature given to vertical lines by 

 the refraction of a prism, which vertical lines, when examined with 

 two prisms, one for each eye, appear by coalescence as one line con- 

 cave in a vertical plane at right angles with the plane of the two 

 separate bent Unes, can be curiously illustrated by the foUovidng ex- 

 periment : — 



If, holding in each hand one prism, the two prisms having their 

 thin edges towards each other, we look at the window from the 

 opposite end of the room, we see first two windows with their vertical 

 lines bent in contrary directious ; but by inclining gradually the optical 

 axes, we can converge them until the two images coalesce, and we 

 see only one Avindow ; as soon as they coincide, the lateral curvature 

 of the vertical lines ceases, and they are bent projectively from back 

 to front : we have then the illusion of a window concave towards the 

 room, such as it would appear reflected by a concave mirror. 



There is another phsenomenon which can be noticed when looking 

 at photographic pictures in the stereoscope ; sometimes the picture 

 appears to project out and sometimes to recede from its mountings. 

 The first effect lessens the illusion, and the second renders it more 

 effectual ; therefore it is desirable to inquire how we can avoid the 

 one and ensure the other. 



We know that the distance of objects is in an inverse ratio vnth 

 the angle of convergence required to see them single ; also that with 

 symmetrical figures or photographic pictures, when the horizontal or 

 lateral distances of the several corresponding points is different, the 

 points less separated will appear nearer, and the more separated will 

 appear further. 



Suppose the two correspondent vertical lines of the openings or 

 frames of the pictures be more distant than the two correspondent 

 points of the furthest plane of the pictures themselves, then the open- 

 ings or frames will appear behind the pictures ; and suppose the cor- 

 respondent vertical lines of the openings be less distant than the two 

 correspondent points of the nearest plane of the pictures themselves, 

 then the openings or frames will appear before the jjicture. 



Therefore, when we wish to have the picture appearing behind the 

 openings or their mountings, we have only to take care that the cor- 

 respondent vertical lines of the mountings should be laterally less 

 distant than the two correspondent points of the first plane of the 

 picture. This can be easily done by taking the measure of the two 

 correspondent points of the first plane by means of a pair of com- 

 passes, and tracing the two ])airs of correspondent vertical lines 

 bounding the openings, after having slightly reduced the angle of 

 the compasses. 



A very 8imj)le experiment may show the cause of the illusion of 

 concavity of flat surfaces when examined through senii-leuses, and 

 further prove that semi-lenses may give alternately the illusion of 



