Dr Searle, Experiments with a prism of small angle 159 
will then make an angle 2: with its former direction. Hence 
this motion of the prism will entail a change of direction of 
27 in the parallel beam of rays which falls normally upon the 
face AC. 
The prism is held in a simple geometrical clamp (Fig. 5). 
Three bicycle balls are soldered to a brass plate and one face 
of the prism (AB in Fig. 3) is kept in contact with the spheres by 
two steel sprmgs. This spring clamp is attached to a horizontal 
rod carried by a heavy stand. The screw at the centre of the 
brass plate allows the plate to be adjusted so that the ledge at the 
bottom of the plate is horizontal. If the ends of the prism have 
been made perpendicular to the refracting edge (which is not 
always the case), this will secure that the prism 1s properly placed. 
The goniometer is fixed at such a height above the table that 
light from a sodium flame passes properly through its reflecting 
prism. The clamp holding the prism to be tested is then set to 
the corresponding height, and the clamp is placed close to the 
lens of the goniometer and is turned about the horizontal rod 
supporting it so that, when the arm of the goniometer is suitably 
directed, an image of the illuminated slit 1s seen just above the 
totally reflecting prism. There are two positions of the arm in 
which an image is seen. In one position the light is reflected at 
the front face of the prism, in the second it is reflected at the 
back face; the first position corresponds to PN in Fig. 3 and the 
second to PQ. A reference to Fig. 3 will enable the observer to 
distinguish between the two images. [The image formed by 
reflexion at the back face of the prism can be destroyed by 
smearing vaseline over that face. | 
It may be found that there is slight parallax between the 
image of the wire and the wire itself when the prism is used. If 
the setting of the wire is perfect as judged by a good plane 
mirror, this indicates that one or both faces of the prism is not 
quite plane. 
If the two images are not at the same height above the bar 
carrying the totally reflecting prism, the prism to be tested is not 
quite accurately adjusted. The error may be corrected by turning 
