﻿350 
  Prof. 
  A. 
  A. 
  Michelson 
  on 
  

  

  employed 
  in 
  the 
  investigation. 
  S 
  is 
  the 
  source 
  of 
  light, 
  either 
  

   a 
  small 
  hand-blowpipe 
  with 
  a 
  bead 
  of 
  the 
  substance 
  to 
  be 
  

   examined 
  in 
  the 
  flame, 
  or 
  a 
  vacuum-tube, 
  which 
  is 
  usually 
  

   placed 
  in 
  a 
  metal 
  box 
  (for 
  heating) 
  of 
  such 
  form 
  as 
  to 
  permit 
  

   a 
  close 
  approach 
  of 
  the 
  pole-pieces 
  P 
  of 
  an 
  electromagnet. 
  

   One 
  of 
  these 
  is 
  bored 
  out 
  to 
  permit 
  examination 
  of 
  the 
  axial 
  

   ra}\ 
  The 
  light 
  from 
  S 
  undergoes 
  a 
  preliminary 
  analysis 
  by 
  

   the 
  spectroscopic 
  train 
  (two 
  bisulphide 
  prisms), 
  the 
  radiation 
  

   to 
  be 
  investigated 
  being 
  isolated 
  by 
  the 
  slit 
  s. 
  It 
  then 
  enters 
  

   the 
  interferometer, 
  one 
  of 
  the 
  mirrors 
  of 
  which, 
  M, 
  is 
  move- 
  

   able 
  on 
  ways 
  so 
  accurately 
  ground 
  that 
  no 
  readjustment 
  is 
  

   necessary 
  in 
  any 
  part 
  of 
  its 
  path 
  ; 
  that 
  is, 
  the 
  mirror 
  remains 
  

   so 
  nearly 
  parallel 
  with 
  itself 
  that 
  the 
  interference-fringes 
  

   (concentric 
  circles) 
  are 
  always 
  as 
  clear 
  as 
  possible. 
  The 
  

   emergent 
  beam 
  then 
  passes 
  through 
  the 
  analyser 
  N 
  to 
  the 
  

   observing 
  telescope. 
  

  

  The 
  clearness 
  or 
  " 
  visibility 
  " 
  of 
  the 
  interference-fringes 
  is 
  

   estimated 
  at 
  positions 
  of 
  the 
  mirror 
  M 
  corresponding 
  to 
  in- 
  

   crements 
  of 
  the 
  difference 
  of 
  path 
  of 
  1, 
  2, 
  or 
  5 
  mm. 
  according 
  

   to 
  the 
  nature 
  of 
  the 
  curve. 
  This, 
  it 
  must 
  be 
  admitted, 
  leaves 
  

   much 
  to 
  be 
  desired 
  in 
  the 
  way 
  of 
  precision, 
  and 
  in 
  some 
  cases 
  

   there 
  may 
  be 
  corrections 
  of 
  as 
  much 
  as 
  20 
  per 
  cent., 
  to 
  reduce 
  

   the 
  observations 
  to 
  the 
  value 
  they 
  should 
  have, 
  namely, 
  

  

  V 
  = 
  

  

  I1-I2 
  

  

  Ii+V 
  

  

  where 
  Ij 
  is 
  the 
  maximum 
  intensity 
  and 
  I 
  2 
  the 
  minimum 
  

   for 
  adjacent 
  fringes. 
  Doubtless 
  much 
  more 
  accurate 
  readings 
  

   could 
  be 
  obtained 
  by 
  the 
  use 
  of 
  a 
  double 
  quartz 
  lens 
  * 
  for 
  

   comparison 
  ; 
  but 
  the 
  process 
  is 
  so 
  much 
  more 
  tedious 
  and 
  

   troublesome 
  that 
  the 
  form 
  of 
  the 
  curve 
  is 
  liable 
  to 
  alter 
  on 
  

   account 
  of 
  changes 
  in 
  the 
  source 
  during 
  the 
  observations. 
  

   The 
  case 
  is 
  somewhat 
  analogous 
  to 
  making 
  eye-estimates 
  of 
  

   stellar 
  magnitudes, 
  which 
  are 
  but 
  little 
  inferior 
  to 
  photometric 
  

   determinations 
  and 
  much 
  less 
  troublesome. 
  In 
  any 
  case 
  it 
  is 
  

   always 
  easy 
  to 
  distinguish 
  ascending 
  and 
  descending 
  slopes, 
  

   and 
  maxima 
  and 
  minima 
  can 
  be 
  located 
  with 
  very 
  great 
  

   accuracy, 
  and 
  this 
  is 
  usually 
  quite 
  sufficient 
  to 
  permit 
  a 
  fairly 
  

   accurate 
  deduction 
  of 
  the 
  distribution 
  of 
  light 
  in 
  the 
  spectrum. 
  

   It 
  has 
  been 
  shown 
  t 
  that 
  with 
  the 
  definition 
  of 
  visibility 
  just 
  

   given, 
  if 
  y 
  = 
  (f>(x) 
  is 
  the 
  intensity-curve 
  of 
  the 
  spectrum, 
  

  

  in 
  which 
  

  

  py= 
  v^ 
  + 
  s 
  2 
  , 
  

  

  P 
  = 
  \ 
  cj) 
  (<r) 
  cLv, 
  C=\cj) 
  ( 
  e 
  i') 
  cos 
  kx 
  d<v, 
  

   * 
  Phil. 
  Mag., 
  Sept. 
  1892. 
  f 
  Ibid. 
  

  

  