Mr.  J.  Walker  on  Talbot's  Lines.  533 
but  those  (b)  of  an  odd  order  will  be  shifted,  and  as  the 
amount  of  the  displacement  varies  with  the  wave-length,  will 
be  made  to  slant  across  the  spectrum.  Thus  the  right  half 
being  that  which  is  covered,  the  appearance  will  be  that 
represented  diagrammatically  in  figure  (2). 
Now  the  effect  of  a  dispersion  of  the  light  will  be  to  move 
each  horizontal  line  of  the  spectrum  in  its  own  direction  by 
an  amount  depending  upon  its  distance  from  the  end — in 
fact  to  give  the  figure  a  kind  of  shear.  If  this  motion  be  to 
the  right,  the  result  will  be  that  both  sets  of  bands  will  slant 
across  the  spectrum,  but  if  it  be  to  the  left,  the  bands  (a) 
will  be  made  to  slant,  while  the  others  (b)  will  be  placed 
more  nearly  along  the  spectrum  and,  the  shift  being  properly 
adjusted,  will  become  exactly  parallel  to  the  sides. 
Hence  in  the  former  case  no  bands  will  be  seen  when  the 
focal  plane  is  viewed  directly  with  an  eyepiece,  but  in  the 
latter  case  the  bands  (b)  will  become  visible. 
3.  The  best  thickness  of  the  retarding  plate  for  a  given 
part  of  the  spectrum  may  now  be  determined. 
It  is  obvious  that  the  resultant  disturbances  in  any  direction 
from  the  two  halves  of  the  aperture  have  the  same  amplitude, 
and  that  their  phases  are  those  of  the  secondary  waves 
emanating  from  the  central  elements.  Since  the  source  of 
light  is  a  line  parallel  to  the  sides  of  the  aperture,  directions 
perpendicular  to  these  sides  alone  need  be  considered,  and  if 
