788  Fluorescence  and  Lambert's  Law. 
If  the  luminous  particles  are  so  close  together  as  to  be 
nearly  or  quite  in  contact,  the  radiation  from  the  surface 
obeys  Lambert's  law,  which  in  this  case  follows  as  a  result 
of  the  screening  action  of  the  particles  upon  the  radiation 
Fig:.  3. 
coming  from  their  neighbours.  Suppose  now  that  the 
luminous  particles  fuse  together  into  a  continuous  surface. 
The  irregular  refractions  and  reflexions,  which  resulted  in  a 
screening  action,  are  now  replaced  by  regular  refraction  at 
the  surface,  all  rays  incident  upon  the  surface  at  an  angle 
greater  than  the  critical  being  totally  reflected  back  into  the 
medium.  It  is  probable  that  no  other  action  than  that  of 
refraction  is  required  to  explain  Lambert's  law  in  the  case 
of  transparent  media,  a  conclusion  which  has  also  been  reached 
by  Uljanin  (Wied.  Ann.  lxii.  p.  528). 
In  the  case  of  an  incandescent  metal  surface  it  is  also 
probable  that  refraction  plays  some  part,  since  the  emitted 
light  is  partially  polarized  at  certain  angles.  The  law  of 
emission  can  be  ascribed  to  the  absorption  of  radiations 
coming  from  points  beneath  the  surface,  the  relation  having 
been  worked  out  by  Lommel. 
So  little  is  known,  however,  about  the  nature  of  the  vibra- 
tions of  incandescent  solid  and  liquid  surfaces,  that  it  is 
difficult  to  formulate  any  very  definite  conception  of  the 
processes  which  are  directly  responsible  for  Lambert's  law. 
It  is  perhaps  worthy  of  remark  that  complete  irregularity 
of  the  phases  of  the  vibrations  is  necessary.  If  the  vibrating 
particles  on  the  surface  of  a  white-hot  metal  plate  could  be 
brought  into  the  same  phase,  the  plate  would  emit  a  plane 
wave,  and  would  cease  to  be  visible.  No  light  would  be 
perceived  at  all  unless  the  eye  were  located  on  some  normal 
to  the  surface,  in  which  case  a  very  brilliant  point  source 
wTould  be  seen  in  the  direction  of  the  normal. 
