IMSI'I.KSIOX IN AUTIFICIAl. HMPBLE REFRACTION 27U 



The annealing of the glasses used, which wen- supplied by Messrs. /KISS of Jena, 

 was found to be by no means perfect. In some cases this was revealed even by 

 a cursory inspection between crossed Nicola. In other cases, the glasses being 

 unloaded, a one-wave plate of selenite was introduced between the Nicols, its axes 

 being horizontal and vertical. This showed a black band, on the same principle that 

 the strained glass shows such a band. 



Now if the glasses had MO residual stress the relative retardations should \te the 

 same when the azimuth of the axes of polarization of the selenite plate is altered 

 by 90. 



If there 1 residual stress, however, it will affect the light differently in these two 

 rases and tlif I mud will lie si lilted. In iin.sl cases tin- existence !' MI<-!> a n -si dual tfcnOt 

 was exhibited very plainly by this method. As a rule the l>and due to the selenite 

 plate was straight and vertical, showing that the residual stress was fairly constant. 



If AT, AU, AS be the three components of residual stress in a vertical plane 

 parallel to the axis of the lx;am, then the axes of polarization make an angle <f> with 

 the horizontal, where 



tan 2< = 2AS/(T + AT-AU) 



and the principal stress difference 



P-Q = v/(T + AT-AU) 3 +4(AS) 3 . 



If we neglect squares of 



AT/T, AU/T, AS/T 



it is easy to calculate that the retardation 



2C,T,a, +20,7,0, 



has to be increased by 



20, (AT,a 1 -AU 1 a I ) + 2C s (AT^-AU/t,), 

 or, taking 



c, = c, = c 



in these corrective terms, the retardation must be increased by 



20 [ATiO I -AU l o,+ AT/ij-AU/iJ, 

 and this is equivalent to putting in a constant correction W to W. 



PAKT II. 



EXPERIMENTAL RESULTS. 

 14. Glasses Observed. 



The glasses used in this research were made for me by the firm of Zmss in 

 Jena. The makers bein^ unable to communicate to me the chemical composition 



