Kerr  Effects  in  the  Infra-red  Spectrum.  71 
or  minima  as  in  the  present  case,  it  is  not  unlikely  that,  could 
they  be  extended  on  each  side  o£  the  visible  spectrum,  more  of 
a  resemblance  to  the  rotatory  dispersion-curves  would  appear. 
The  preceding  explanation  seeks  to  interpret  the  magnetic 
rotation  o£  the  Kerr  effect  as  a  characteristic  property  of  the 
metal  itself,  having  the  same  physical  significance  as  the 
simpler  case  of  magnetic  rotation  in  transparent  substances. 
Another  point  of  view  may  be  obtained  following  the  sugges- 
tion made  by  Voigt*  some  years  ago,  and  later  developed  by 
du  Bois  f .  The  observations  of  KundtJ,  and  later  of 
Lobach§,  on  rotation  by  transmission  through  thin  films  of 
magnetic  metal  in  a  magnetic  field,  have  suggested  the 
explanation  of  the  Kerr  effect  as  a  special  case  of  rotation 
by  transmission,  with  reflexion  below  the  surface  of  the 
metal.  This  is  the  now  commonly  accepted  explanation,  but 
it  does  not  account  for  the  fact  that  the  rotations  in  the  two 
cases  are  opposite  in  sign;  and  to  explain  this  Voigt  and 
du  Bois  suggest  that  one  of  the  circularly-polarized  com- 
ponents may  in  general  penetrate  deeper  into  the  metal  than 
the  other,  and  hence  introduce  a  difference  of  phase  which 
may  give  any  sort  of  a  rotation  positive  or  negative.  A 
point  in  favour  of  this  argument  is  the  peculiar  reversal 
effect  which  was  found  in  nickel,  and  which  may  most  easily 
be  explained  on  this  basis,  although  as  regards  the  negative 
rotation,  it  is  of  interest  to  note  that  the  Hall  effect  in  nickel 
is  opposite  in  sense  to  what  it  is  in  iron  and  cobalt;  and  on 
the  basis  of  the  Hall  effect  as  an  explanation  of  the  Kerr 
phenomenon,  nickel  should  show  a  negative  rotation. 
But  even  if  this  view  of  Voigt  and  du  Bois  on  the  nature 
of  the  Kerr  effect  be  accepted,  it  does  not  necessarily  mean 
that  the  rotation  dispersion-curves,  as  found,  are  without 
physical  significance,  but  merely  that  certain  effects  which 
have  been  attributed  to  the  property  of  magnetic  rotation 
would  then  be  considered  as  caused  by  this,  which  might  be 
called  the  differential  reilecting-povver  of  the  metal,  which 
causes  the  reflexion  of  the  two  circularly-polarized  com- 
ponents at  different  mean  depths.  For  it  is  not  unreasonable 
to  suppose  that  this,  if  it  exists  at  all,  is  just  as  characteristic 
a  property  of  the  metal,  and  as  intimately  related  to  other 
optical  properties,  as  the  power  of  magnetic  rotation  by 
transmission  itself. 
The  negative  results  given  by  Heusler's  metal  are  quite 
remarkable,  for,  to  the  best  of  the  writer's  knowledge,  this  is 
the  only  case   in    which    this  peculiar  metal  has   failed    to 
*  Wied.  Ann.  xxiii.  p.  49M  (1884).  f  Ibid,  xxxix.  p.  25  (1890). 
|  Ibid,  xxiii.  p.  286  (1884),  §  Ibid,  xxxix.  p.  347  (1890). 
