62 



Mr Filon, On the variation with the wave-length 



having been differently marked. They all show a steady decrease 

 of the stress-optical coefficient from \ 4500 to \ 6500, the total 

 variation being about 5 per cent. This agrees very well with the 

 results of the previous paper. 



The present observations, however, are sufficiently accurate to 

 allow us to form some estimate of the law of this variation. The 

 observations of the last two groups (Tables II. and III.) lie very 

 close indeed upon a straight line, which has been put in dotted in 

 Fig. 3. Considering the smallness of this residual effect the 



1CS 



2,6.0 



2f.r 



LS-.O 



c>* 



^i 



v* 



at 



X 



^r 



X 



-Qi 



4-S"00 S~QQ0 5T00 6000 6500 



Wave-Lengths in tenth-metres. 

 © Observations of Table I. 



+ „ „ n. 



x „ „ in. 



Fig. 3. 



agreement with the straight line is exceedingly good, except at 

 the blue end where, for reasons explained previously, the obser- 

 vations were far more difficult and the points accordingly more 

 irregular. 



The observations of the first series (Table I.) lie in most cases 

 considerably off this straight line, although the difference is not, 

 perhaps, so great as the large scale of the diagram would lead one 

 to think. This set of observations would be very fairly well 



