50  Mr.  L.  E.  Ingersoll  on  the  Faraday  and 
thick.  Reflexion  from  thin  glass  plates,  while  by  no  means 
free  from  objection,  seemed  to  be  the  only  practical  scheme 
of  obtaining  polarized  radiations  for  this  work  ;  for  the  use 
of  a  polarizing  grating  would  involve  too  much  loss  of  energy, 
and  the  absorption  of  any  doubly-refracting  substance  makes 
its  use  prohibitive  beyond  about  \=2'5  /x.  The  polarizer 
carried  a  divided  circle  and  was  so  mounted  that  it  could  be 
rotated  about  the  axis  of  the  beam.  The  glower  was  arranged 
to  rotate  with  the  polarizer,  and  at  the  same  time  to  turn 
about  an  axis  of  its  own,  so  as  to  keep  its  image  on  the  slit 
always  vertical.  The  analyser  also  had  a  number  of  possible 
adjustments,  and  carried  with  and  parallel  to  it  a  plane  silver 
surface,  by  reflexion  at  which,  after  passing  the  polarizer, 
the  beam  would  be  brought  out  parallel  to  its  original 
direction. 
Because  of  its  rather  exceptional  performance  as  regards 
brilliancy  and  constancy,  the  glower  and  its  mounting  are 
worthy  of  special  mention.  When  unprotected  and  in  the 
open  air,  a  Bernst  glower  may  suffer  variations  of  intensity 
of  upwards  of  one  or  two  per  cent.,  even  when  supplied 
with  a  perfectly  constant  E.M.F.  This  is  doubtless  due  to 
irregularities  of  convection  ;  and  at  the  suggestion  of  Prof. 
Mendenhall  it  was  tried  inside  an  enclosure  made  of  plaster- 
of -Paris  or  cement.  This  gave  a  much  greater  constancy, 
and  as  the  objections  to  the  plaster  enclosure,  which  was  soon 
destroyed  by  the  heat,  have  since  been  removed  by  making  it 
of  firebrick,  it  leaves  little  to  be  desired  as  a  source  for  infra- 
red w7ork.  Specimen  curves  have  been  obtained,  using 
96-volt  direct-current  glowers  made  by  the  bernst  Lamp 
Company  of  Pittsburg,  and  furnishing  the  current  with  a 
storage  battery,  which  show  a  maximum  variation  of  inten- 
sity of  less  than  y\_  per  cent,  for  several  minutes  at  a  time. 
Some  of  the  special  adjustments  may  be  mentioned.  The 
analyser  and  polarizer  piles  of  plates  were  set  with  the  aid  of 
a  nicol  at  the  angle  of  incidence  wdiich  gave  most  completely 
polarized  reflected  light.  It  might  be  supposed  that  this 
setting  would  have  to  be  changed,  to  suit  the  different  wave- 
lengths used;  but  as  a  matter  of  fact,  as  will  be  pointed  out  in 
a  discussion  of  the  sources  of  error,  it  wras  not  a  sensitive 
adjustment,  and  might  be  made  once  for  all  with  sufficient 
accuracy  for  all  wave-lengths  within  the  range  over  which  it 
was  possible  to  work.  The  angle  between  the  principal 
planes  of  the  polarizer  and  analyser  was  determined  by 
placing  a  large  nicol  with  divided  circle  at  N,  and  setting  so 
.as  to  extinguish,  first,  by  crossing  with  the  principal  plane  of 
the    polarizer,  and,   second,  with  that  of  the  analyser.     In 
