68  Mr.  L.  R.  Ingersoll  on  the  Faraday  and 
interest;  bat  nickel,  curiously  enough,  appeared  to  reverse 
the  direction  of  rotation  at  about  1*4  fi.  This,  being  such  an 
unexpected  result,  was  for  a  time  regarded  as  spurious,  but  as 
repeated  trials  on  three  different  samples  verified  it  in  all 
cases  where  the  sensibility  was  high  enough^  it  is  now  believed 
to  be  genuine.  The  samples  tested  were  (1)  nickel  plated 
on  polished  brass  (points  plotted  as  crosses)  ;  (2)  cast-nickel 
of  supposedly  exceptional  purity  from  a  magnetic  standpoint, 
although  still  containing  a  trace  of  iron  (plotted  as  circles)  ; 
and  (3)  commercial  rolled  nickel  (plotted  as  triangles).  As 
may  be  seen,  there  is  considerable  discrepancy  between  the 
various  nickel  curves.  Thus,  in  fig.  3,  this  curve,  which  was 
made  with  a  rolled-nickel  specimen,  is  unlike  its  counterparts 
in  the  other  two  sets,  while  the  different  curves  of  fig.  5  seem 
to  show  considerable  separation  after  crossing  the  axis. 
Whether  these  differences  are  real  and  characteristic  of  the 
different  specimens  tested,  or  whether  they  are  spurious,  can- 
not yet  be  stated.  Although  the  measurements  on  nickel 
were  attended  with  more  difficulty  than  in  most  of  the  other 
cases,  and  the  probable  error  of  any  point  is  therefore  large, 
it  can  hardly  account  for  the  consistent  differences  which 
appear  between  the  curves.  In  view  of  this  disagreement  it 
is  rather  odd  that  all  the  curves  should  cross  the  axis  at  nearly 
the  same  point,  as  they  seem  to  at  about  l'-JL  /jl. 
Magnetite  proved  very  difficult  to  test,  and  the  different 
curves  show  poor  agreement.  Because  of  its  low  reflecting- 
power  only  a  single  surface  could  be  advantageously  used, 
the  place  of  the  other  being  taken  by  silver;  so  the  results  in 
the  curves  of  figs.  5  and  6  are  for  this  single  surface  only, 
although  they  have  been  doubled  in  fig.  3,  partly  to  put  them 
on  the  same  basis  as  the  others  for  comparison's  sake,  and 
partly  to  avoid  confusion.  The  most  noticeable  feature  of  the 
curve  for  this  substance  is  that,  for  a  good  share  of  the  infra- 
red at  any  rate,  the  rotation  is  in  the  same  sense  as  for  steel, 
although  the  positive  part  of  the  curve,  which  would  agree 
with  the  positive  results  found  in  the  visible  spectrum  by 
du  Bois,  is  well  indicated.  The  second  reversal,  as  found  in 
fig.  3,  could  not  be  verified  in  later  observations. 
The  points  on  the  silver  curves  were  determined  with  the 
use  of  two  surfaces  of  silver  plated  on  brass,  and  in  this  way 
the  zero  from  which  the  other  rotations  are  to  be  reckoned 
was  obtained.  The  curve  shows  the  characteristic  Faraday 
dispersion  form,  and  in  the  proper,  that  is  positive,  sense.  It 
is  doubtless  due  to  the  effect  of  the  weak  tield  in  the  neigh- 
bourhood of  the  analyser  and  polarizer,  for  its  magnitude 
could  be  altered  and  even  made  to  disappear,  as  is  seen  in 
