﻿334 
  Prof. 
  A. 
  A. 
  Michelson 
  

  

  on 
  

  

  to 
  these 
  results 
  vary 
  enormously. 
  Still, 
  in 
  view 
  of 
  the 
  

   limited 
  number, 
  and 
  keeping 
  in 
  mind 
  the 
  criticism 
  of 
  

   M. 
  Cornu, 
  which 
  will 
  be 
  presently 
  referrred 
  to, 
  it 
  may 
  be 
  of 
  

   interest 
  to 
  find 
  the 
  mean 
  of 
  these 
  nine 
  values, 
  giving 
  them 
  

   equal 
  weight. 
  The 
  result 
  is 
  

  

  V 
  = 
  299,664, 
  

  

  with 
  an 
  average 
  difference 
  from 
  the 
  mean 
  of 
  600 
  kilometres. 
  

  

  But 
  the 
  results 
  (1), 
  (2), 
  (6) 
  are 
  admitted 
  to 
  be 
  only 
  first 
  

   approximations, 
  and 
  (5) 
  is 
  undoubtedly 
  affected 
  by 
  some 
  

   serious 
  constant 
  error 
  ; 
  so 
  that 
  a 
  much 
  closer 
  approximation 
  

   to 
  the 
  true 
  value 
  will 
  be 
  obtained 
  by 
  combining 
  (3) 
  or 
  (4), 
  

   the 
  mean 
  of 
  (6) 
  and 
  (7), 
  and 
  (9). 
  

  

  The 
  first 
  combination 
  gives 
  : 
  

  

  Cornu 
  300,400 
  + 
  300 
  

  

  Michelson 
  299,882+ 
  60 
  

  

  Newcomb 
  299,810+ 
  60 
  

  

  Giving 
  these 
  equal 
  weight 
  we 
  find 
  

  

  V 
  = 
  300,030, 
  

  

  with 
  an 
  average 
  difference 
  from 
  the 
  mean 
  of 
  about 
  250. 
  

  

  If, 
  however, 
  we 
  take 
  Listing's 
  value 
  of 
  Corrm's 
  results, 
  we 
  

   get: 
  

  

  Cornu 
  299,990 
  

  

  Michelson 
  299,882 
  

  

  Newcomb 
  299,810 
  

  

  Giving 
  for 
  the 
  mean 
  

  

  Y=299,890, 
  

  

  with 
  an 
  average 
  difference 
  from 
  the 
  mean 
  of 
  about 
  60. 
  

  

  In 
  a 
  paper 
  presented 
  to 
  the 
  International 
  Congress 
  in 
  1900, 
  

   M. 
  Cornu 
  expresses 
  serious 
  doubts 
  concerning 
  the 
  order 
  of 
  

   accuracy 
  of 
  the 
  results 
  obtained 
  by 
  means 
  of 
  the 
  revolving 
  

   mirror 
  ; 
  and, 
  giving 
  equal 
  weights 
  to 
  the 
  two 
  methods 
  

   regardless 
  of 
  the 
  number 
  of 
  determinations, 
  and 
  ignoring 
  

   Listing's 
  discussion 
  of 
  his 
  own 
  results, 
  gives 
  as 
  his 
  estimate 
  

   of 
  the 
  most 
  probable 
  value 
  of 
  the 
  velocity 
  of 
  light 
  

  

  V 
  = 
  300,130, 
  

  

  with 
  average 
  difference 
  from 
  the 
  mean 
  of 
  270. 
  

  

  The 
  criticism 
  of 
  M. 
  Cornu 
  embraces 
  the 
  following 
  points 
  : 
  

  

  1. 
  Can 
  the 
  ordinary 
  laws 
  of 
  reflexion 
  be 
  applied 
  to 
  the 
  

   case 
  of 
  a 
  mirror 
  in 
  rapid 
  rotation 
  ? 
  

  

  2. 
  Can 
  the 
  ordinary 
  laws 
  of 
  reflexion 
  be 
  applied 
  to 
  the 
  

   case 
  of 
  a 
  light-ray 
  whose 
  axis 
  moves 
  with 
  a 
  velocity 
  com- 
  

   parable 
  with 
  the 
  velocity 
  of 
  light 
  ? 
  

  

  