of the double refraction in strained glass. 
487 
Special attention had to be given to the polarizing Nicol. The 
latter required to be displaced in order that the whole spectrum 
might be examined. Care had to be taken always to keep it 
parallel to itself. This was done by keeping the edge of the 
square stand of the Nicol flat against a fixed straight edge. In 
this way the motion of the Nicol could introduce no shift of the 
lines of the spectrum. 
The vertical distances between the black bands were measured 
by means of a second glass scale which carried a vertical milli- 
metre scale. This was placed in contact with the other scale, but 
could be shifted freely to the right and left so as to bring the 
vertical scale into any required horizontal position. 
6. For the two beams actually used 
Tj = 2*905 cms,, r 2 = 2*88.5 cms., 
a = 152 cms., 6 = 6 cms. 
The bending moments, after all the corrections had been 
applied, were found to be 
71^ = 676 kgms. weight cms., 
M 2 = 672 kgms. weight cms. 
The heights of the beams were given by 
A 1 = 3 , 67 cms., 
h 2 = 3 - 65 cms. 
Whence c lf c 2 were calculated to be 
c x = 164T kgms. weight per sq. cm., 
c 2 = 165 - 9 kgms. weight per sq. cm. 
and 
p = 340'8 kgms. weight per sq. cm. 
Accordingly, \ being known and y 0 (n+1) ~yf l) being measured in 
cms., the absolute value of G can be found in sq. cms. per kgm. 
weight. 
As what is required, however, is the relative value of C for 
different colours, and as p remained the same throughout the 
whole series of experiments, the values of (y 0 {n+1) — 2/o w )/^ have 
not been reduced to absolute values by multiplying by p in each 
case, but the quantity studied has been kept in the form 1/Gp. 
The fringes in the field numbered 2 in the red, increasing to 3 
in the violet, on either side of the central one. The distances 
y (x) _ y(o) } 2 / (2) — y (o) , y {3) — y (0) were measured for a number of 
sections corresponding to divisions of the horizontal scale. A 
