DISPERSION IN ARTIFICIAL DOUBLE REFRACTION. 265 



S, in his more recent investigation, observed the artificial double refraction 

 in a number of Jena glasses. His observations, though not primarily intended to test 

 the effect of colour, nevertheless gave exceedingly valuable results in this connection, 

 insomuch as POCKELS experimented with three different kinds of nearly homogeneous 

 light, namely, those of sodium, lithium, and thallium. The results would therefore 

 be far more precise than those obtained with very mixed colours by WERTHEIM. They 

 show that, in certain glasses, the stress-optical coefficient does vary with the wave- 

 length, being numerically greater in the green than in the red ; and in very heavy 

 lead glasses this variation is more rapid as we approach the blue end of the spectrum. 



Some years ago the present author, being at the time unaware of POCK ELS' 

 experiments, devised a method of observing the variation of the coefficient C 

 continuously throughout the spectrum, the object being to test the exactness of 

 WKRTHEIM'S law. 



An account of this method, which was modified and improved from time to time, 

 and of the experiments undertaken to carry it out, will be found in the 

 Camb. Phil. Soc. Proc.,' vol. XL, Part VI. ; vol. XII., Part I. ; vol. XII., Part V. 

 These experiments amply confirmed the results of POCKELS. They also showed that 

 the chief desideratum for obtaining accurate results was that the stress in the glass 

 slab through which the light was passed should be sufficiently uniform. Now the 

 compression apparatus which was used by the author, and by previous experimenters, 

 suffered from the defect that it was practically impossible to adjust it so as to obtain 

 a uniform pressure in the slab of glass under observation. Moreover, what adjustment 

 could be made was long and difficult, and could be attained only by trial ; it apjieared 

 further that this adjustment was disturbed, in a way that could not be calculated and 

 allowed for, when the load was altered. This greatly reduced the accuracy expected. 



An apparatus was then devised, with a view to obtaining a system in which the 

 stresses should l>e known exactly and in which the optical effects should IK- the same 

 as those due to uniform pressure in a slab of constant thickness. For the purposes of 

 this research a Government Grant was kindly placed at the disjHwal of the author by 

 the Royal Society, whereby the necessary apparatus could be constructed and the 

 expensive glasses required for the research purchased. The present paj>er is an 

 account of the experiments carried out with the new apparatus and of the results 

 reached. 



PART I. 



THEORY OF THE EXPERIMENT AND DISCUSSION OF THE VARIOUS ERRORS. 

 2. Simple Thewy of the Experiment. 



l- t N, F (fig. 1) be two rectangular slabs of glass, whose cross-sections are shown 

 in the figure. The slalw are bent in a vertical plane by couples without shear, whose 

 axes are horiaontd and parallel to the plane of. the cross-section. How such couples 

 are obtained will be explained subsequently. 



VOL. OCVII. A. 2 M 



