﻿the 
  Absolute 
  Scale 
  of 
  Temperature. 
  

  

  373 
  

  

  result 
  as 
  to 
  variation 
  with 
  the 
  temperature 
  ; 
  but 
  it 
  is 
  not 
  so 
  

   well 
  supported 
  by 
  the 
  experimental 
  numbers. 
  The 
  observa- 
  

   tions 
  recorded 
  in 
  their 
  second 
  memoir, 
  excluding 
  those 
  in 
  

   which 
  the 
  gas 
  contained 
  more 
  than 
  10 
  per 
  cent, 
  of 
  impurity, 
  

   are 
  as 
  follows 
  : 
  — 
  

  

  Mean 
  

   temperature. 
  

  

  No. 
  of 
  

   experiments. 
  

  

  e 
  (obs.). 
  

  

  e 
  (calc). 
  

  

  o 
  

  

  8 
  

   36 
  

   54 
  

   96 
  

  

  2 
  

   1 
  

   1 
  

   2 
  

  

  -l°-233 
  

  

  1022 
  

  

  •885 
  

  

  •645 
  

  

  -1°233^ 
  

   1020 
  

   •911 
  

   •715 
  j 
  

  

  per 
  

  

  atmo. 
  

  

  The 
  numbers 
  in 
  the 
  last 
  column 
  are 
  calculated 
  from 
  the 
  

   inverse 
  square 
  rule, 
  taking 
  e 
  as 
  — 
  1 
  0, 
  293. 
  The 
  cooling 
  

   effect 
  seems 
  to 
  fall 
  off 
  more 
  rapidly 
  ; 
  I 
  have 
  therefore 
  

   preferred 
  to 
  make 
  a 
  diagram 
  of 
  the 
  observed 
  results, 
  together 
  

   with 
  those 
  published 
  in 
  the 
  earlier 
  memoir, 
  which 
  agree 
  with 
  

   the 
  above, 
  and 
  read 
  off 
  the 
  value 
  of 
  e 
  for 
  each 
  temperature 
  

   from 
  the 
  curve. 
  With 
  the 
  value 
  of 
  e 
  assumed 
  in 
  the 
  above 
  

   table 
  and 
  the 
  inverse 
  square 
  rule, 
  the 
  mean 
  over 
  the 
  range 
  

   0° 
  and 
  100° 
  (which 
  is 
  0-732 
  of 
  the 
  value 
  at 
  0°) 
  is 
  -0°'956 
  ; 
  

   whilst 
  the 
  value 
  taken 
  from 
  the 
  curve 
  is 
  — 
  o, 
  930. 
  Lord 
  

   Kelvin, 
  in 
  the 
  encyclopaedia 
  article, 
  gives 
  — 
  1°*105, 
  without 
  

   explaining 
  the 
  process 
  of 
  reduction 
  used. 
  1 
  have 
  not 
  been 
  

   able 
  to 
  see 
  how 
  this 
  number 
  is 
  arrived 
  at. 
  

  

  E. 
  Natanson* 
  has 
  repeated 
  Thomson 
  and 
  Joule's 
  expe- 
  

   riment 
  on 
  carbon 
  dioxide, 
  using 
  the 
  commercial 
  liquefied 
  gas 
  

   from 
  a 
  cylinder 
  : 
  he 
  made 
  all 
  his 
  experiments 
  at 
  one 
  tempe- 
  

   rature, 
  — 
  20°, 
  and 
  found 
  that 
  the 
  cooling 
  is 
  not 
  quite 
  pro- 
  

   portional 
  to 
  the 
  pressure, 
  but 
  may 
  be 
  represented 
  by 
  

  

  -€=1-18 
  + 
  00126>, 
  

  

  p 
  being 
  in 
  atmospheres. 
  This, 
  for 
  low 
  pressures, 
  is 
  in 
  accord- 
  

   ance 
  with 
  Joule 
  and 
  Thomson's 
  result 
  at 
  the 
  same 
  temperature. 
  

   We 
  have, 
  however, 
  seen 
  that 
  there 
  is 
  an 
  indirect 
  means 
  of 
  

   estimating 
  the 
  cooling 
  effect, 
  by 
  means 
  of 
  the 
  density-varia- 
  

   tion 
  of 
  the 
  coefficient 
  of 
  pressure, 
  the 
  relation 
  being 
  

  

  e 
  = 
  

  

  AK 
  T 
  V 
  

  

  * 
  E. 
  Natansou, 
  Wied. 
  Ann. 
  vol. 
  xxxi. 
  pp. 
  502-526 
  (1887). 
  

  

  