Birefringent Action of Strained Glass. 331 



earlier measurements being left out as loss accurate, the 

 average of the remaining eiglity determinations, carried to 4 

 figures, was 2 '004. 



Upon the whole, from these measurements, and from others 

 taken recently by another method, I conclude that the strain- 

 generated retardation of the pencil H, in glass vertically 

 compressed, is sensibly equal to half the corresponding 

 retardation of the pencil V, its value in any case being 

 probably as accurate, when so inferred from a well measured 

 retardation of V, as when found itself by actual measurement. 



Proposition V. 



Strain-generated retardations, absolute as well as relative, are 

 sensibly proportioiial to the strain. 



11. The effect of directional strain which is observed in the 

 polariscope, that is, the difference of strain-generated absolute 

 retardations of the two component rays, varies directly as 

 the strain. This was roughly proved by the experiments of 

 Brewster, and has been already quoted as a general conclusion 

 drawn by Wertheim from a large number of accurate measure- 

 ments. On this part of the subject, and for illustration rather 

 than proof, I may present a set of measurements of my own, 

 which I took at an early stage in my experiments. 



A fine rectangular bar of plate glass, with a small con- 

 necting piece clamped upon it at each end, was suspended 

 vertically in front of a Jamin's quartz compensator (between 

 a pair of crossed Nicols each at 45° to the vertical), and was 

 strained by a suspended weight ; and the effect of the strain 

 so impressed was compensated and measured. The dimen- 

 sions of the bar in millimetres were — length about 150, thick- 

 ness (along the ray) 6*57, breadth 12*95 ; and the only 

 weights applied were one of 56 pounds (total), and another 

 of 26 pounds (total). One hundred fairly concordant 

 measurements were taken with each of the weights, and the 

 average effects of the strains in the two cases (56 and 28) 

 were found to be, in wave-lengths, 



•0965 and '0483. 



In Wertheim's experiments, the closely approximate pro- 

 portionality of optical effect to intensity of strain was found 

 to hold good, through the whole of the large range of effect 

 examined, with exceptions easily understood in the case of 

 very feeble strains. 



Knowing then that the relative retardation of V is pro- 

 portional to the strain, and that the absolute retardations of 

 V and H and the relative are as the numbers 2, 1, 1, we infer 



