Kerr  Effects  in  the  Infra-red  Spectrum.  47 
the  sacrifice  of  purity  to  the  extent  of  causing  a  dispropor- 
tionately large  error  in  the  measurement  of  wave-length. 
As  to  the  actual  intensity  o£  the  spectrum,  it  may  be  stated 
that  with  the  arrangement  used,  galvanometer-deflexions  were 
not  uncommon,  measured  with  reduced  sensibility,  which 
would  correspond,  on  the  basis  o£  full  sensibility,  to  1500 
centimetres  on  the  scale. 
The  general  arrangement  of  apparatus  maybe  seen  from  fig.  I 
(p.  48) .  Light  from  the  Nernst  glower  G-j  was  polarized  by 
reflexion  at  the  pile  of  thin  glass  plates  P,  and  after  reflexion 
at  Mi  formed  a  real  image  of  the  glower  between  the  poles 
of  the  magnet,  and  again  on  the  slit,  after  reflexion  at  M2 
and  from  the  pile  of  analyser-plates  A.  This  was  drawn  out 
into  a  spectrum  at  the  bolometer-strip  B  by  a  60°  rock-salt 
prism.  The  latter  had  the  customary  Wads  worth  minimum 
deviation  mounting,  the  plane  mirror  being  a  Brashear  flat 
4  x  10  cm.  The  prism  itself  had  faces  4x5  cm.,  polished  by 
Brashear,  and  was  protected  by  an  enclosure,  not  shown  in 
the  diagram,  mounted  on  the  spectrometer-table  and  con- 
taining sulphuric  acid.  The  table  was  rotated  by  a  micro- 
meter-screw connected  with  a,  steel  strap  pulling  tangentially 
over  a  lever  arm.  One  division  of  the  micrometer-head 
corresponded  to  a  rotation  of  0*06  minute  of  arc.  The 
mirrors  M1?  M2,  M3  were  each  40  cms.  in  local  length,  while 
M4  was  30  cms.  They  were  mounted,  together  with  the 
other  parts  of  the  spectrometer,  on  heavy  slabs  of  slate. 
The  broken  lines  represent  various  adjustments  which,  by 
means  of  cords,  could  be  conveniently  made  by  the  observer 
without  moving  from  his  position  in  front  of  the  galvano- 
meter scale.  (1)  Could  be  used  to  rotate  the  polarizer  ; 
(2)  allowed  the  carbon -bisulphide  tube  to  be  thrown  in  or 
out  of  the  path  of  the  beam  ;  (3)  operated  a  shutter  in  front 
of  the  slit ;  (4)  rotated  the  mirror  M2  and  hence  shifted  the 
image  of  the  glower  on  the  slit. 
!N either  bolometer  nor  galvanometer  present  any  especial 
novelty  of  construction,  save  that  the  latter  was  made  as 
compact  as  possible  to  allow  effectual  shielding.  It  was  of 
the  type  described  by  Mendenhall  and  Waidner  *,  and  had 
four  coils  of  twenty  ohms  each,  measuring  lb*  nuns,  outside 
and  2  mms.  inside  diameter.  The  system,  which  had  of  course 
quartz-fibre  suspension,  weighed  something  less  than  2  mg. 
and  had  a  mirror  1  mm.  square.  Deflexions  of  the  image 
of  a  lint1  wire  in  front  of  a  Nernst  glower  Gra  were  observed 
on  a  scale  a.  metre  and  a  hall'  distant.  Fairly  good  magnetic 
shielding  was  secured  by  surrounding  it  with  six  concentric 
!   A.mer.  Journ.  Sri.  xii.  v.  249  (Oct.  L901). 
