﻿320 
  Dr. 
  C. 
  Chree's 
  Notes 
  on 
  Thermometry. 
  

  

  100° 
  C. 
  on 
  a 
  centigrade 
  thermometer 
  is 
  equivalent 
  to 
  throwing 
  

   over 
  100° 
  F. 
  on 
  a 
  Fahrenheit 
  thermometer 
  is 
  one 
  against 
  

   which 
  experience 
  shows 
  me 
  a 
  warning 
  is 
  necessary. 
  The 
  

   contraction 
  of 
  scale 
  required 
  in 
  the 
  former 
  case 
  is 
  of 
  course 
  

   9/5 
  times 
  that 
  required 
  in 
  the 
  latter. 
  

  

  Welsh's 
  figures, 
  as 
  a 
  matter 
  of 
  fact, 
  differ 
  slightly 
  from 
  

   those 
  given 
  above, 
  in 
  a 
  systematic 
  way 
  ; 
  he 
  has 
  201*87 
  for 
  

   example 
  in 
  place 
  of 
  201*89, 
  and 
  171*48 
  in 
  place 
  of 
  171*55. 
  

   The 
  cause 
  of 
  this 
  discrepancy 
  was, 
  I 
  suspect, 
  his 
  neglecting 
  

   any 
  expansion 
  in 
  the 
  glass, 
  for 
  his 
  figures 
  are 
  deducible 
  from 
  

   (31) 
  by 
  taking 
  '000180 
  as 
  the 
  value 
  of 
  e. 
  The 
  B.A. 
  figures 
  

   in 
  fact 
  accord 
  with 
  the 
  rale 
  " 
  contract 
  the 
  scale 
  by 
  1 
  per 
  

   cent, 
  for 
  each 
  100 
  degrees 
  (Fahrenheit) 
  of 
  mercury 
  to 
  be 
  

   thrown 
  over,''' 
  which 
  I 
  know 
  Welsh 
  to 
  have 
  actually 
  laid 
  

   down. 
  

  

  § 
  35. 
  In 
  discussing 
  the 
  theory 
  of 
  Welsh's 
  method 
  we 
  sup- 
  

   posed 
  the 
  surplus 
  mercury 
  to 
  be 
  measured 
  at 
  0° 
  C, 
  as 
  this 
  

   shows 
  most 
  simply 
  its 
  relation 
  to 
  what 
  remains. 
  In 
  the 
  

   actual 
  operation 
  it 
  is 
  hardly 
  possible 
  to 
  have 
  mercury 
  and 
  

   glass 
  at 
  0° 
  C, 
  and 
  it 
  is 
  usually 
  inconvenient 
  to 
  keep 
  the 
  

   detached 
  column 
  in 
  the 
  tube 
  and 
  measure 
  it 
  at 
  0° 
  C. 
  before 
  

   throwing 
  it 
  off. 
  Welsh's 
  own 
  idea 
  for 
  boiling-point 
  thermo- 
  

   meters 
  was 
  simply 
  to 
  put 
  the 
  thermometers 
  in 
  steam 
  and 
  

   throw 
  off 
  the 
  quantity 
  of 
  mercury 
  required 
  to 
  make 
  them 
  

   give 
  the 
  correct 
  boiling-point 
  reading. 
  Supposing 
  this 
  not 
  

   exactly 
  accomplished, 
  he 
  directed 
  that 
  the 
  error 
  observed 
  at 
  

   212° 
  F. 
  in 
  the 
  final 
  state 
  should 
  be 
  regarded 
  as 
  a 
  constant 
  

   zero 
  error, 
  being 
  combined 
  of 
  course 
  with 
  the 
  calibration 
  

   errors 
  detected 
  by 
  the 
  comparison 
  with 
  the 
  standard 
  thermo- 
  

   meter 
  in 
  the 
  preliminary 
  state. 
  

  

  This 
  is 
  certainly 
  satisfactory 
  in 
  short-range 
  thermometers 
  

   such 
  as 
  Welsh 
  had 
  in 
  view. 
  For 
  example, 
  to 
  introduce 
  an 
  

   error 
  varying 
  by 
  0°*01 
  O. 
  throughout 
  a 
  range 
  of 
  25° 
  C. 
  there 
  

   would 
  require 
  to 
  be 
  a 
  mistake 
  of 
  about 
  2J 
  scale-divisions 
  in 
  

   the 
  length 
  of 
  the 
  mercury-column 
  thrown 
  over, 
  and 
  this 
  is 
  at 
  

   least 
  ten 
  times 
  as 
  great 
  as 
  is 
  at 
  all 
  likely 
  in 
  the 
  hands 
  of 
  a 
  

   competent 
  workman. 
  

  

  § 
  36. 
  The 
  modern 
  extension 
  of 
  Welsh's 
  method 
  to 
  high- 
  

   range 
  thermometers 
  is 
  no 
  doubt 
  exposed 
  to 
  greater 
  uncer- 
  

   tainty. 
  Suppose, 
  for 
  instance, 
  we 
  wish 
  the 
  thermometer 
  in 
  

   its 
  final 
  state 
  to 
  cover 
  the 
  range 
  200° 
  C. 
  to 
  350° 
  C. 
  approxi- 
  

   mately. 
  In 
  the 
  preliminary 
  state 
  the 
  thermometer 
  may 
  

   cover 
  say 
  from 
  0° 
  C. 
  to 
  150° 
  C. 
  — 
  the 
  accuracy 
  of 
  the 
  scale 
  

   throughout 
  the 
  upper 
  50° 
  being 
  determined 
  by 
  calibration 
  

   with 
  a 
  mercury-column 
  whose 
  value 
  is 
  observed 
  at 
  the 
  lower 
  

   part 
  of 
  the 
  scale, 
  where 
  direct 
  comparison 
  with 
  a 
  standard 
  

  

  