334 Mr Filon, On the variation with the luave-length 



The case of 103 is more doubtful. A set of good observat- 

 ions (C.) is well fitted by the parabola 



The sets A. and B. are fairly accurately represented by a 

 straight line 



C=,S3-,8 25 (^), 



except towards the ends of the range, where they seem to show a 

 curvature, which may be significant. 



Two of the glasses, 192 and O 154, indicate an approximately 

 linear variation. 



For 192, 



is a good representation of the observations. 



For O 154, the three groups agree very badly with one 

 another. Group C. is fitted tolerably well by the formula 



C= 2-891- -0525 



X - 5400\ 



1000 /' 

 Group A. suggests 



between the green and the red. There are two outlying points, 

 but no great significance should be attached to them. 



Group B. is rather irregular. Altogether the observations of 

 this glass are unsatisfactory. 



11. In order to be able to trace better the effect of the 

 chemical composition upon the optical properties, I calculated, for 

 a number of the glasses which have been examined by Pockels, 

 my coefficient C. I also calculated in each case the mean change 

 in G per tenth-metre, over the range from X 5360 to A, 6730. 



The results, grouped together, are given in Table VIII. below. 



