﻿in. a Rectangular Bar by means of Polarized Light. 7 

 § 4. Comparison of Experimental Results with Theory. 



The general appearance of the bands, as soon as the distance 

 from a point of loading exceeded the greatest dimension of 

 the cross-section, corresponded closely with the theoretical 

 results of the last article. 



Quite recently I have been able, through the kindness of 

 Prof. F. T. Trouton and Prof. A. W. Porter, to use the large 

 nicols in the possession of the Physical Laboratory of 

 University College, in order to photograph these bands, as 

 shown by a specimen in a straining-frame ; and although 

 these are not the actual bands whose measurements form tbe 

 subject of the present paper, they should be of identical 

 shape. 



The photographs of these bands are shown in Plate II., 

 and by the side of each is placed a diagram of the corre- 

 sponding theoretical hyperbola as given by (4). Any exact 

 comparison is difficult, as the large thickness of the nicols 

 used introduced an appreciable distortion in the images; this 

 is clearly seen in the pictures from the curved appearance of 

 the (really straight and parallel) edges of the glass. Bat 

 the similarity is sufficiently close to afford, by itself, striking- 

 confirmation of the theoretical results. 



In these pictures the inclination of the nicols increases by 

 differences of 15 c . One corresponds to an inclination of 49°, 

 being the closest to 45° for which the two branches of the 

 hvperbola were still clearly distinguishable, in spite of the 

 dark region joining them. 



In the Cambridge experiments the method of comparison 

 adopted was as follows : a number of measurements were 

 taken on the charts of the bands (mentioned at the end of §2), 

 giving for selected values of y and i a corresponding value 

 of ,v. 



Now if we write K = 3S /A6 2 , we have from (1) and 

 M=S * 



2S = K(& 2 -*/ 2 ), (5) 



P = K..i-/y, ((n 



and also 



2S P. 



rp- = T vtan 2i, 



K lv 



= .ry tan 2i (7) 



2S/K was then calculated by means of formula (7), the mean 

 of a number of measurements on different charts being taken. 

 This was compared with the values of />'—//'. The results 



