552 Proceedings of the Royal Society of Edinburgh. [Sess. 
Experiments were made as follows : — The observer stood back in a room 
overlooking a level lawn with a background of trees. In the window was 
the rod B. The window-sill cut off the view of the ground. The rod C was 
set up on the lawn, and the rod D was moved backwards and forwards by 
an assistant according to signals, until the correct position was found. The 
distance C C' was then measured. It will be noticed that the observer had 
no means of knowing the actual positions of C and D until the reading was 
completed. In another set of readings the rod D was fixed on a trolley 
and pulled to and fro by an endless cord round a pulley at the end of 
the lawn. 
The position of C was varied, the greatest distance from B being 110 
feet, for beyond that distance the background came into focus and confused 
the reading. Two or three readings were taken at each distance, D being 
displaced each time, and good agreement was usually obtained when the 
conditions were favourable, as described below, the variation seldom exceed- 
ing three or four inches. 
It was found that readings were most consistent when the distance 
C D was kept small. This of course facilitated the comparison of distance. 
But it was also found that the distance C C' was increased, the largest 
values being obtained when the image of B came close to the edges of C, 
and when D was immediately on the other side of one of the B images, or 
even if B partially overlapped C and D. The apparent distance between 
the two B images did not make any difference to C C r , if it was not much 
larger than the angle subtended by C. But if considerably larger it 
caused irregularity in the reading, and greater difficulty in determining 
the correct position. Probably in this case the eyes were trying to 
see normally, and it is well known that if the eyes are receiving two 
impressions simultaneously, the mental effect is apt to alternate between 
the two competing appearances. 
The judgment was better if C and D were dark or quiet in colour. 
White posts were found to be somewhat difficult to place stereoscopically, 
even without any disturbing images. The posts were the dull greyish 
brown of weather-stained wood, and excellent demonstrations were also 
obtained with the trunks of young trees. 
The distance AB did not affect the results, over a range from 9 to 19 
feet, except that more accurate readings were obtained if A B was chosen 
so as to allow C' and D to be close together. And, as stated above, if A B 
was short when B C was very long, the clear space between the edges of 
the B images and C became too great. This seemed more disturbing on a 
dull day than on a bright sunny day. 
