64 Mr Filon, On the variation viith the tvave-length, etc. 



To the first approximation the variation in G is given by 

 a straight line whose slope downwards is '00070 per tenth-metre, 

 from which, using the values of G here obtained, 



G= (29'69 - (-00070) \) x 10~ 8 sq. cms. per kgm. wt. 



where the absolute term is determined with some uncertainty as 

 stated above. 



On examining the points representing the observations still 

 more closely, it will be seen that there are traces of a point of 

 inflexion about X, 5300. The observations come down at first 

 more sharply than they should according to the straight line law, 

 then the rate of decrease is suddenly checked, but recovers itself 

 afterwards. The effect may be seen in all three sets of obser- 

 vations, but is most marked in those of Table I. It appears not 

 unlikely that this effect may be real, and in this case a cubic term 

 should be added to the expression for G. 



Altogether it seems that this method of detecting variations 

 in the stress-optical properties of a substance is more delicate and 

 accurate than any which has yet been tried. It should be re- 

 membered that Pockels ( Wied. Ann. 1902, Ser. IV. Vol. vii. p. 745) 

 has examined light glasses of this density and found no sensible 

 dispersion. The author hopes at no distant date to apply this 

 method to the heavier glasses, which undoubtedly show a large 

 dispersion of artificial double refraction, and it is hoped that the 

 results obtained will yield considerable information. 



The author wishes to thank Professor Trouton of University 

 College, London, for having allowed him to use the Physical 

 Laboratory of University College for this investigation and also 

 for the uniform kindness with which he has assisted him in every 

 way; and also to acknowledge gratefully his indebtedness to 

 Mr A. W. Porter, Assistant-Professor of Physics at University 

 College, London, for much invaluable assistance. 



