320 Mi- Filon, On the variation with the wave-length 



7. In order to obviate as much as possible this source of 

 error, in the later sets of experiments the slit was placed with its 

 edge coincident with each of the two vertical edges of the block, 

 in succession. This adjustment could be made with considerable 

 exactness. The readings of the black band for these two extreme 

 positions were noted, and the slit moved until the reading of the 

 black band was the mean of the extreme readings. When 

 possible, the position of the black band was measured without 

 narrowing the beam by a diaphragm. This could be done if the 

 stress was sufficiently uniform, and this should have given a result 

 free from error. As a rule, however, this was only possible over a 

 limited part of the range. The diaphragm was then introduced, 

 and adjusted so that its effect was to increase the definition of the 

 band without shifting its position. 



In spite of these precautions, the results are not so consistent 

 as could be wished. The method adopted has been to take 

 several series of observations for each glass, at different times 

 and under different conditions. Characteristic variations which 

 are reproduced in such different sets may safely be taken as 

 significant. 



8. The results of the observations are given in Tables II. — 

 VII. below. 



In what follows C is expressed throughout in a unit equal to 

 10 -7 sq. cms. per kilogramme weight. 



