﻿An 
  Evaluation 
  of 
  the 
  Absolute 
  Scale 
  of 
  Temperature. 
  363 
  

  

  be 
  faint 
  on 
  the 
  lower 
  side 
  and 
  terminate 
  abruptly 
  on 
  the 
  

   upper. 
  Their 
  width 
  diminishes 
  as 
  we 
  ascend 
  the 
  scale 
  of 
  

   frequency. 
  Their 
  lower 
  edges 
  correspond 
  to 
  the 
  proper 
  

   periods 
  of 
  the 
  intermolecular 
  spaces. 
  If 
  we 
  view 
  the 
  light 
  

   reflected 
  we 
  shall 
  see 
  total 
  reflexion 
  corresponding 
  to 
  the 
  

   frequencies 
  of 
  the 
  dark 
  bands 
  of 
  the 
  transmission 
  spectrum; 
  

   for 
  these 
  wave-lengths 
  there 
  will 
  be 
  reversal 
  of 
  phase. 
  

  

  XL. 
  A 
  Numerical 
  Evaluation 
  of 
  the 
  Absolute 
  Scale 
  of 
  

   Temperature. 
  By 
  R. 
  A. 
  Lehfeldt*. 
  

  

  Introduction. 
  

  

  HVTUMEROUS 
  attempts 
  have 
  been 
  made 
  to 
  reduce 
  the 
  

   -i-i 
  readings 
  of 
  thermometers 
  to 
  the 
  absolute 
  scale, 
  since 
  

   that 
  scale 
  was 
  first 
  clearly 
  defined 
  by 
  Thomson 
  and 
  Joule 
  ; 
  

   but 
  though 
  the 
  process 
  of 
  calculation 
  has 
  been 
  varied 
  a 
  good 
  

   deal, 
  the 
  most 
  essential 
  experimental 
  basis 
  of 
  all 
  the 
  reductions 
  

   is 
  the 
  same, 
  viz. 
  Thomson 
  and 
  Joule's 
  own 
  experiments 
  on 
  

   the 
  outflow 
  of 
  gases 
  through 
  a 
  porous 
  plug. 
  It 
  is 
  very 
  

   remarkable, 
  therefore, 
  that 
  no 
  one, 
  so 
  far 
  as 
  I 
  know, 
  has 
  

   attempted 
  to 
  repeat 
  or 
  extend 
  those 
  experiments, 
  except 
  in 
  

   one 
  case 
  studied 
  by 
  E. 
  Natanson, 
  and 
  that 
  notwithstanding 
  the 
  

   great 
  discrepancies 
  in 
  Thomson 
  and 
  Joule's 
  measurements. 
  

   For 
  hydrogen, 
  the 
  substance 
  which 
  is 
  probably 
  the 
  best 
  for 
  

   thermometric 
  purposes, 
  there 
  were 
  twelve 
  experiments 
  carried 
  

   out 
  at 
  about 
  7° 
  and 
  five 
  at 
  about 
  90° 
  ; 
  but 
  these 
  results 
  varied 
  

   from 
  -f-0*9 
  to 
  — 
  0*1 
  ! 
  while 
  the 
  hydrogen 
  was 
  in 
  no 
  case 
  even 
  

   approximately 
  pure 
  ; 
  and 
  although 
  the 
  determination 
  of 
  ab- 
  

   solute 
  temperatures 
  depends 
  upon 
  that 
  experiment, 
  it 
  has 
  

   been 
  left 
  in 
  so 
  unsatisfactory 
  a 
  state 
  for 
  half 
  a 
  century. 
  Of 
  

   the 
  other 
  data 
  required, 
  some, 
  such 
  as 
  the 
  specific 
  heat 
  and 
  

   specific 
  volume 
  of 
  the 
  gases 
  used, 
  are 
  known 
  with 
  sufficient 
  

   accuracy, 
  since 
  they 
  enter 
  only 
  in 
  a 
  small 
  correction 
  term 
  : 
  

   the 
  most 
  important, 
  however, 
  is 
  the 
  coefficient 
  of 
  pressure 
  in 
  

   the 
  case 
  of 
  a 
  gas 
  thermometer 
  at 
  constant 
  volume 
  (or 
  the 
  

   coefficient 
  of 
  expansion 
  in 
  the 
  constant-pressure 
  thermometer). 
  

   On 
  this 
  point, 
  and 
  this 
  only, 
  a 
  distinct 
  advance 
  has 
  been 
  

   made 
  since 
  the 
  time 
  of 
  Regnault. 
  This 
  has 
  been 
  accom- 
  

   plished 
  by 
  Chappuis, 
  working 
  at 
  the 
  Bureau 
  International, 
  

   who 
  has 
  measured 
  the 
  coefficient 
  of 
  pressure 
  between 
  0° 
  and 
  

   100° 
  for 
  hydrogen, 
  nitrogen, 
  and 
  carbon 
  dioxide, 
  with 
  all 
  the 
  

   care 
  and 
  scrupulousness 
  that 
  modern 
  physical 
  methods 
  can 
  

   suggest. 
  My 
  object 
  in 
  writing 
  this 
  paper 
  is 
  partly 
  to 
  take 
  

   advantage 
  of 
  Chappuis's 
  results, 
  and 
  partly 
  to 
  draw 
  attention 
  

   * 
  Communicated 
  bv 
  the 
  Author. 
  

  

  