Kerr  Effects  in  the  Infra-red  Spectrum. 
67 
they  are  especially  well  brought  out  in  fig.  4,  which  shows 
the  results  of  a  more  detailed  study  of  steel.  Although  the 
magnetic  and  other  conditions  were  considerably  different  for 
the  two  curves,  they  still  agree  very  well.     Those  points  for 
Fig.  6. — Variation  of  Kerr  Effect  with  wave-length.  Constant  field. 
Double-refracting  polarizing  apparatus.  (Magnetite  curve  drawn 
by  comparison  with  fig.  3.) 
-70 
which  the  two  sets  of  readings — which,  as  previously  explained, 
were  always  made  for  each  point — agreed  most  closely,  were 
regarded  as  having  the  least  probable  error  and  have  been 
weighted  accordingly  in  drawing  the  curves.  They  are 
plotted  as  circles,  and  the  others  as  crosses,  while  the  curves 
to  which  certain  ambiguous  points  belong  are  indicated  by 
small  arrows.  To  show  that  the  slope  on  the  short  wave- 
length side  agrees  with  that  of  du  Bois'  dispersion-curve  of 
iron,  made  for  the  Kerr  effect  in  the  visible  spectrum,  the 
latter  curve  has  been  plotted  to  such  scale  that  it  joins  on  to 
the  lower  of  the  given  curves  at  "67  //,,  the  points  being 
represented  by  triangles. 
Cobalt  gave  results  similar  to  steel  and  possesses  no  especial 
F2 
