Kerr  Effects  in  the  Infra-red  Spectrum.  G9 
figs.  5  and  6,  by  a  proper  arrangement  of  the  exciting  coils 
of  the  magnet.  In  the  latter  two  sets  of  curves  the  pole- 
coils  only  were  excited,  while  in  the  set  of  fig.  3  one  field-coil 
was  added.  It  was  also  largely  dependent  on  whether  the 
pile  of  plates  or  doubly-refracting  substance  was  used  as 
polarizing  agent.  The  points  in  the  set  of  fig.  3  have  all 
been  corrected  by  tbe  use  of  the  silver  curve,  although  this 
has  not  seemed  necessary  in  the  last  two  sets.  Unfortunately 
no  silver  correction-curve  was  made  for  the  steel  curves  of 
fig.  4,  its  importance  not  being  realized  at  the  time;  hence 
there  is  some  doubt  as  to  whether  they  correctly  represent 
the  position  and  shape  of  the  maximum,  although  it  is  pro- 
bable that  the  correction  in  this  case  would  he  small,  from  the 
fact  that  a  single  determination  at  \=1*34/a  with  the  steel 
covered  with  silvered  glass  surfaces  gave  no  effect.  The 
silver  points  have  been  plotted  by  somewhat  smaller,  and 
the  magnetite  by  larger,  circles  than  the  others,  to  avoid 
confusion. 
As  a  further  check  on  the  accuracy  of  the  results,  some 
direct  visual  observations  were  attempted.  The  method  was 
rather  crude,  for  the  use  of  a  half-shade  apparatus  and 
monochromatic  light  failed  to  give  sufficient  illumination 
wTith  the  existing  arrangement  of  apparatus;  but  it  was  found 
that  fairly  good  settings  for  extinction  could  be  made  on  the 
polarized  image  of  the  glower  itself,  and  the  rotations  so 
determined  have  been  plotted  in  fig.  6  by  the  letter  V.  This 
use  of  white  light  would  of  course  give  the  integral  of  the 
effect  for  the  whole  visible  spectrum;  but  the  range,  save  in 
the  case  of  steel,  is  small,  and  in  any  case  it  is  assumed  to 
correspond  to  the  rotation  for  some  wave-length  near  the 
maximum  of  visibility,  or,  nearly  enough,  for  sodium  light. 
The  agreement  with  more  careful  measurements  by  other 
observers  is  good,  considering  the  necessary  uncertainty  of 
field-strength  in  the  present  case,  for  while  measurements 
with  a  test  flip-coil  indicated  about  10,000  C.G.S.  units,  it 
would  vary  considerably  over  the  surface  of  the  specimen, 
which  had  to  have  an  appreciable  area  to  meet  the  other 
requirements  of  the  problem.  These  points  in  the  visible 
spectrum  have  been  conveniently  used  to  connect  the  du  Bois 
rotatory  dispersion-curves  with  those  of  fig.  6,  the  former 
results  being  plotted  to  such  a  scale  that  the  rotation  for  the 
sodium  line  shall  be  the  same  as  in  the  present  case.  The 
curves  join  together  fairly  well,  save  in  the  case  of  magnetite, 
where,  the  visible  spectrum  measurements  being  indecisive, 
the  scale  is  arbitrarily  chosen. 
Bather   curiously,  Heusler's  metal  gave  entirely  negative 
