﻿Hence 



in a Rectangular Bar by means of Polarized Light. 23 



line the observations. The fit is strikingly close for small 

 values of W, bat the curves diverge as y increases. This is 

 no more than we should expect, since the terms neglected in 

 (22) then rise in importance. But where the formula is 

 really valid the agreement is remarkable. 



§ 9. Observations of the Points where the Jsoclinic Lines 

 cross the Horizontal Axis of the Block. 



The same formulae (22) allow us to bring yet another test 

 to bear on the theory. To find where the band crosses the 

 horizontal for neutral) axis put in (22) y = 0, r = x } <j) = ir/2, 

 and neglect terms of fourth and higher orders in xjb. We 

 find 



2W/ G 3 ^\ 



2W/ x <3tia*\ 



Gr.bV 6G 2 b 2 )/ \ 2G 1 fcV 



_G 2 ^r /G3 G 4 \^ 2 ~| 

 "G^L + \2G 1 ~6G2/6 2 J' 



if xjb be small. Putting in numerical values from p. 99 of 

 the memoir, we have 



3 



tan 2t = 3-072 y +5-115^, . . . (24) 



b (r 



where i, x are to remain small. 



When i = the band is vertical. Small positive values 

 or! i deflect it to the right; small negative values of i to the 

 left. 



These results were also verified experimentally, a set of 

 simultaneous values of i and xjb being obtained. These are 

 shown in fig. 1.1, tan 2i being plotted to x/b, and the full 

 curve shows the graph of equation (24). 



The agreement is here by no means so good as in fig. 10. 

 Nevertheless the general nature of the two curves is the 

 same. It should be noticed that a slight change in the vertical 

 scale would <»'ive much closer agreement. This suggests an 



tan 



J-,- . ~ ~~ * K.V.. W , 



