﻿322 
  Dr. 
  C. 
  Chree's 
  Notes 
  on 
  Thermometry. 
  

  

  A 
  direct 
  determination 
  of 
  the 
  error 
  at 
  some 
  one 
  tempera- 
  

   ture 
  in 
  the 
  final 
  state 
  is 
  all 
  the 
  more 
  necessary 
  because 
  the 
  

   operation 
  of 
  knocking 
  off 
  the 
  mercury-column, 
  involving 
  

   severe 
  heating 
  of 
  the 
  bulb, 
  may 
  leave 
  a 
  slight 
  permanent 
  

   effect. 
  

  

  It 
  is 
  important 
  to 
  notice, 
  in 
  the 
  case 
  of 
  any 
  such 
  high- 
  

   range 
  thermometer 
  as 
  that 
  just 
  described, 
  that 
  the 
  method 
  

   assumes 
  the 
  thermometer 
  to 
  be 
  of 
  the 
  same 
  kind 
  of 
  glass 
  as 
  

   the 
  standard 
  with 
  which 
  it 
  was 
  compared 
  in 
  the 
  preliminary 
  

   state, 
  and 
  that 
  the 
  readings 
  of 
  the 
  thermometer 
  in 
  its 
  final 
  

   state 
  — 
  corrected 
  for 
  errors 
  of 
  calibration 
  and 
  of 
  volume 
  of 
  

   mercury 
  thrown 
  off 
  — 
  are 
  temperatures 
  on 
  the 
  natural 
  scale 
  of 
  

   a 
  standard 
  thermometer 
  of 
  this 
  particular 
  kind 
  of 
  glass. 
  The 
  

   readings 
  have 
  no 
  direct 
  relationship 
  to 
  the 
  hydrogen 
  or 
  any 
  

   other 
  gas-thermometer 
  scale, 
  and 
  temperatures 
  measured 
  

   near 
  the 
  top 
  of 
  the 
  range 
  must 
  possess 
  an 
  uncertain 
  value 
  

   unless 
  the 
  natural 
  scale 
  they 
  relate 
  to 
  has 
  been 
  studied. 
  

  

  If 
  the 
  high-range 
  thermometer 
  and 
  the 
  standard 
  it 
  is 
  com- 
  

   pared 
  with 
  are 
  of 
  different 
  kinds 
  of 
  glass, 
  and 
  still 
  more 
  if 
  

   the 
  thermometer 
  has 
  its 
  bulb 
  of 
  one 
  kind 
  of 
  glass 
  and 
  its 
  stem 
  

   of 
  another, 
  the 
  interpretation 
  of 
  its 
  readings 
  would 
  require 
  

   further 
  consideration. 
  

  

  § 
  37. 
  As 
  related 
  to 
  Welsh's 
  method, 
  I 
  would 
  mention 
  an 
  

   ingenious 
  way 
  of 
  obtaining 
  e, 
  the 
  mean 
  coefficient 
  of 
  expansion 
  

   of 
  mercury 
  in 
  glass 
  between 
  0° 
  and 
  100° 
  C, 
  described 
  on 
  

   pp. 
  102-104 
  of 
  the 
  Reichsanstalt's 
  Wiss. 
  Abhandl. 
  vol. 
  i. 
  It 
  

   consists 
  in 
  observing 
  the 
  fundamental 
  interval 
  in 
  a 
  thermo- 
  

   meter's 
  normal 
  state, 
  and 
  also 
  after 
  a 
  series 
  of 
  threads 
  of 
  

   mercury 
  of 
  different 
  lengths 
  have 
  been 
  detached 
  and 
  run 
  into 
  

   a 
  top 
  chamber. 
  For 
  instance, 
  let 
  /j/i,/2j/g 
  represent 
  the 
  

   observed 
  fundamental 
  intervals 
  in 
  the 
  original 
  state 
  and 
  after 
  

   detachment 
  of 
  threads 
  of 
  50, 
  100, 
  and 
  150 
  degrees 
  respec- 
  

   tively. 
  Then, 
  n 
  denoting 
  the 
  number 
  of 
  degree 
  volumes 
  in 
  

   the 
  bulb 
  up 
  to 
  the 
  zero 
  mark, 
  we 
  have 
  

  

  100 
  e 
  =f/n 
  =fj 
  (» 
  - 
  50) 
  =/ 
  2 
  / 
  (n 
  - 
  100) 
  =f 
  3 
  /(n 
  - 
  1 
  50) 
  . 
  

  

  Thence 
  we 
  can 
  deduce 
  e 
  and 
  n 
  by 
  least 
  squares 
  or 
  otherwise. 
  

   Obviously 
  the 
  inethc-d 
  need 
  not 
  be 
  restricted 
  to 
  the 
  range 
  0° 
  

   to 
  100° 
  C. 
  

  

  Comparison 
  of 
  Thermometrie 
  Methods. 
  

  

  § 
  38. 
  It 
  remains 
  for 
  us 
  to 
  consider 
  briefly 
  the 
  relative 
  

   defects 
  and 
  advantages 
  of 
  the 
  ordinary 
  British 
  methods 
  of 
  

   thermometry 
  and 
  the 
  more 
  refined 
  methods 
  of 
  the 
  Bureau 
  

   International. 
  In 
  favour 
  of 
  the 
  latter 
  is 
  the 
  strong 
  argument 
  

   that 
  when 
  a 
  verre 
  dur 
  thermometer 
  verified 
  at 
  the 
  Bureau 
  

  

  