﻿G. 
  K. 
  Wead 
  — 
  In 
  ten 
  % 
  ity 
  of 
  So 
  irnd. 
  23 
  

  

  But 
  the 
  numbers 
  lie 
  gives, 
  especially 
  for 
  leakage, 
  show 
  such 
  

   wide 
  variations 
  as 
  to 
  throw 
  great 
  doubt 
  on 
  the 
  accuracy 
  of 
  the 
  

   method. 
  Therefore 
  two 
  modifications 
  were 
  made 
  : 
  first, 
  a 
  

   stop-watch 
  indicating 
  eighths 
  of 
  a 
  second 
  was 
  used 
  ; 
  and 
  

   second, 
  the 
  movement 
  of 
  the 
  wind-indicator 
  above 
  the 
  key- 
  

   board 
  was 
  observed 
  through 
  a 
  space 
  of 
  5<» 
  mm 
  ; 
  to 
  be 
  sure 
  only 
  

   about 
  one-third 
  of 
  the 
  wind 
  was 
  used, 
  but 
  it 
  is 
  absolutely 
  

   necessary 
  to 
  allow 
  10 
  to 
  20 
  seconds 
  to 
  elapse 
  for 
  the 
  subsidence 
  

   of 
  the 
  strong 
  vibrations 
  set 
  up 
  in 
  the 
  top 
  of 
  the 
  bellows 
  by 
  

   the 
  act 
  of 
  pumping. 
  It 
  was 
  sometimes 
  found 
  that 
  though 
  no 
  

   key 
  was 
  pressed 
  the 
  leakage 
  was 
  different 
  according 
  as 
  the 
  stop 
  

   was 
  drawn 
  or 
  closed, 
  especially 
  with 
  one 
  of 
  the 
  pedal 
  stops. 
  

  

  One 
  further 
  modification 
  of 
  method 
  w 
  r 
  as 
  made 
  : 
  since 
  the 
  

   leakage 
  is 
  more 
  than 
  the 
  amount 
  of 
  wind 
  consumed 
  by 
  any 
  

   single 
  pipe, 
  except 
  a 
  few 
  of 
  the 
  largest, 
  the 
  influence 
  of 
  errors 
  

   of 
  observation 
  was 
  diminished 
  by 
  combining 
  several 
  pipes 
  so 
  

   that 
  they 
  might 
  all 
  sound 
  at 
  once 
  ; 
  two 
  ways 
  of 
  doing 
  this 
  

   were 
  tried 
  : 
  

  

  1. 
  A 
  single 
  stop 
  was 
  drawn, 
  and 
  several 
  consecutive 
  white 
  

   keys, 
  usually 
  eight, 
  were 
  held 
  down 
  by 
  a 
  loaded 
  block; 
  thus 
  

   we 
  find 
  the 
  relative 
  consumption 
  of 
  wind 
  by 
  different 
  stops, 
  

   or 
  by 
  pipes 
  of 
  the 
  same 
  stop 
  in 
  different 
  parts 
  of 
  the 
  scale. 
  

   See 
  table 
  I. 
  

  

  2. 
  Several 
  stops 
  were 
  drawn 
  as 
  in 
  ordinary 
  playing, 
  and 
  a 
  

   single 
  key 
  held 
  down 
  by 
  a 
  wedge. 
  In 
  this 
  case 
  each 
  pipe 
  re- 
  

   ceives 
  less 
  wind 
  than 
  when 
  no 
  unison 
  pipe 
  is 
  near, 
  a 
  fact 
  long 
  

   known 
  and 
  further 
  established 
  by 
  these 
  experiments 
  : 
  but 
  we 
  

   may 
  still 
  find 
  the 
  relative 
  wind-supply 
  in 
  different 
  parts 
  of 
  the 
  

   scales. 
  See 
  table 
  II 
  and 
  part 
  of 
  III. 
  

  

  Most 
  of 
  the 
  results 
  of 
  the 
  work 
  can 
  be 
  given 
  best 
  in 
  tabular 
  

   form. 
  In 
  the 
  tables 
  the 
  names 
  of 
  the 
  stops 
  need 
  no 
  explana- 
  

   tion 
  ; 
  where 
  9 
  stops 
  were 
  drawn 
  they 
  comprise 
  all 
  the 
  stops 
  

   drawn 
  by 
  the 
  forte 
  composition 
  pedal, 
  including 
  the 
  7 
  named 
  

   separately, 
  a 
  twelfth 
  and 
  a 
  3-rank 
  mixture. 
  The 
  notation 
  of 
  

   the 
  keys 
  is 
  used 
  consistently 
  always 
  referring 
  to 
  the 
  key, 
  not 
  

   the 
  pitch, 
  c 
  corresponding 
  to 
  middle 
  c(=268 
  d. 
  v.) 
  when 
  an 
  8' 
  

   stop 
  is 
  drawn 
  ; 
  it 
  will 
  therefore 
  be 
  seen 
  that 
  the 
  absolute 
  pitch 
  

   of 
  all 
  the 
  notes 
  in 
  the 
  lower 
  part 
  of 
  table 
  I 
  is 
  the 
  same 
  ; 
  while 
  

   in 
  the 
  next 
  table 
  pipes 
  of 
  five 
  different 
  lengths, 
  besides 
  the 
  

   mixtures 
  will 
  respond 
  to 
  a 
  single 
  key. 
  L 
  of 
  course 
  means 
  

   leakage. 
  The 
  time 
  given 
  is 
  the 
  mean 
  of 
  from 
  3 
  to 
  8 
  observa- 
  

   tions 
  : 
  these 
  agreed 
  so 
  well 
  that 
  the 
  probable 
  error 
  of 
  the 
  

   mean 
  is 
  very 
  rarely 
  1 
  per 
  cent 
  : 
  take 
  two 
  examples 
  at 
  random 
  ; 
  

   Table 
  III 
  C 
  22f 
  , 
  22|. 
  22f, 
  22| 
  ; 
  mean 
  22*66 
  sec. 
  

  

  Q-c 
  6f, 
  6f, 
  6|, 
  6f, 
  6f; 
  mean 
  6-375 
  sec. 
  ^ 
  

   The 
  following 
  columns 
  contain 
  respectively 
  l-.-t, 
  and 
  this 
  

   quantity 
  diminished 
  by 
  the 
  leakage 
  : 
  this 
  remainder 
  represents 
  

  

  