306         Prof.  H.  A.  Bumstead  on  the  Heating  Effects 
was  found  to  be  always  in  the  direction  indicating  a  repulsion 
from  the  lead  ;  its  direction  reversed  -with  the  reversal  of  the 
order  of  the  strips.  As  the  aluminium  window  was  the  onlv 
part  much  exposed  to  changes  of  temperature,  the  effect  of 
slio'htlv  warmin o-  it  was  investigated.  When  an  incandescent 
lamp  or  a  warm  soldering-iron  was  held  six  inches  from 
the  aluminium  window,  the  radiometer  was  deflected  away 
from  the  lead  strip — at  first  very  slowly,  then  more  and  more 
rapidly,  until  it  went  off  the  scale  and  usually  bumped  against 
the  zinc  strip.  As  this  was  in  the  same  direction  as  the  effect 
when  both  strips  were  exposed  to  the  Rontgen  rays,  it 
appeared  that  the  whole  effect  might  be  spurious.  By  putting 
the  lamp  about  two  feet  away  from  the  window,  less  violent 
effects  were  produced,  and  it  was  possible  to  follow  the  progress 
of  the  deflexion.  This  was  sluggish  in  comparison  with  that 
produced  by  the  rays,  and  (as  might  have  been  expected  from 
varying  air-currents  in  the  room)  was  irregular  and  some- 
what erratic  in  its  course.  In  one  respect,  however,  the 
behaviour  was  consistent :  if  the  lamp  was  left  glowing  for 
one  or  two  hours,  the  radiometer  always  returned  to  the 
neighbourhood  of  its  zero-point  and  stayed  there  until  the  lamp 
was  turned  off.  It  then  made  another  slow  and  wandering- 
excursion  in  the  opposite  direction  (repulsion  by  zinc),  and 
eventually  again  returned  to  the  middle  of  the  scale.  The 
application  of  ice  to  the  aluminium  window  also  caused 
repulsion  by  the  zinc. 
The  determination  of  the  different  rates  of  heating  and 
cooling  of  the  two  strips  furnished  the  explanation  of  this 
behaviour.  While  the  window  and  adjacent  parts  of  the  case 
are  slowly  warming  up,,  they  radiate  to  the  two  strips  more 
and  more,  and  cause  the  temperature  of  the  latter  to  rise 
steadily;  but  the  lead,  having  a  smaller  time  constant  than  the 
zinc,  has  the  higher  temperature  and  maintains  it  until  the 
amount  of  radiation  reaching  the  strips  no  longer  varies  with 
the  time — that  is,  until  the  window  has  reached  its  stationary 
temperature.  When  this  takes  place  the  zinc  has  time  to 
overtake  the  lead,  both  temperatures  become  the  same,  and 
the  radiometer  returns  to  zero.  When  the  lamp  is  turned  off 
and  the  window  begins  to  cool,  the  lead  cools  more  rapidly 
than  the  zinc;  and  hence  there  is  a  similar  deflexion  in  the 
opposite  direction,  which  again  lasts  only  so  long  as  the 
temperature  of  the  window  is  variable. 
Although  this  explanation  is  reasonable  and,  in  fact, 
inevitable,  the  matter  here  discussed  touches  so  vitally  the 
main  conclusion  of  this  investigation  that  it  was  desirable  to 
test  in   a  decisive  manner  the  behaviour  of  the  instrument 
