Kerr  Effects  in  the  Infra-red  Spectrum.  49 
sections  o£  annealed  steel  pipe.  This  arrangement  could  be 
made  to  give  a  sensibility  of  4  x  10-11  for  5  ohms  resistance, 
but  considerations  of  steadiness  and  of  proportionality  of 
deflexions  led  to  reducing  this  *to  about  2  x  10-10  amp.  per 
mm.  with  10  sec.  period,  for  ordinary  usage. 
The  bolometer  used  in  most  of  the  work  had  strips  \  mm. 
wide  and  8  mms.  long.  Heavy  copper  leads  connected  it 
with  the  balancing-bridge,  and  switches,  Sw,  which  permitted 
reversal  of  galvanometer  connexions  and  also  changes  of 
sensibility  in  the  approximate  ratios — 3*5,  9,  and  28  times 
respectively.  At  highest  sensibility  a  candle  at  a  metre's 
distance  gave  a  deflexion  corresponding  to  200  centimetres 
on  the  galvanometer- scale. 
The  magnet  was  designed  to  give  the  greatest  field-strength 
over  ranges  of  several  centimetres,  with  the  smallest  external 
field  possible.  The  pole-pieces  were  of  Swedish  iron  9  cms. 
in  diameter,  and  were  bored  with  conical  holes  narrowing 
from  5  cms.  diam.  at  the  outside  ends,  to  a  slit  5  mms.  wide 
and  16  mms.  high  at  the  conical  ends.  The  advantages  of  a 
rectangular  opening  over  a  circular  one  are  obvious,  con- 
sidering the  shape  of  the  glower  used  as  a  source,  When 
excited  with  about  100,000  ampere-turns,  it  gave  fields  as 
high  as  26,000  lines  per  cm.  over  an  area  of  a  few  square 
millimetres,  using  pointed  pole-pieces.  It  will  be  noted  that 
this  is  not  as  great  as  the  fields  which  may  be  obtained  with 
the  du  Bois  ring  type  of  magnet  ;  but  it  may  be  remarked 
that  in  the  present  work  the  air-gap  was  never  in  practice 
much  less  than  3  cms.,  with,  of  course,  a  necessarily  great 
loss  of  field-strength  :  therefore,  since  the  ring  type  depends 
for  its  powerful  fields,  not  on  an  excessive  number  of  ampere- 
turns,  but  on  the  low  reluctance  of  its  magnetic  circuit,  it  is 
doubtful  whether  its  performance  under  the  same  circum- 
stances would  excel  that  of  the  present  type.  Moreover,  the 
varied  adjustments  of  which  it  must  admit  would  be  impossible 
in  a  magnet  of  the  ring  type, — at  least  without  very  groat 
modification. 
The  external  field  was  small  and  its  effect  at  the  galvano- 
meter was  compensated  for  by  a  secondary  coil  of  high 
resistance,  placed  near  the  latter  instrument  and  connected 
in  shunt  with  one  of  the  main  magnet-coils.  When  run  at 
full  power  there  was  necessarily  considerable  heating  about 
the  magnet,  and  to  protect  the  neighbouring  apparatus  from 
this,  the  whole  magnet  was  enclosed  in  a  large  wooden  box 
(not  shown  in  diagram)  whose  sides  were  water-jacketed. 
Analyser  ami  polarizer  were  each  made  ap  o\'  a  Dumber  o\' 
plates  of  microscopic  cover-glass  3*5x5  cms.  and  0*2  mm. 
Phil.  Man.  S.  6.  Vol.  11.  No.  61.  Jan.  L906.  E 
