﻿300 
  Scientific 
  Intelligence. 
  

  

  the 
  air 
  by 
  conduction 
  and 
  radiation 
  from 
  the 
  central 
  disc 
  to 
  the 
  

   bottom 
  disc. 
  Since 
  the 
  discs 
  were 
  horizontal, 
  convection 
  cur- 
  

   rents 
  were 
  absent, 
  the 
  density 
  of 
  the 
  gas 
  being 
  constant 
  in 
  every 
  

   horizontal 
  plane 
  and 
  decreasing 
  from 
  the 
  cool 
  bottom 
  disc 
  

   upward 
  to 
  the 
  warm 
  central 
  disc. 
  All 
  the 
  discs 
  consisted 
  of 
  two 
  

   plates 
  0-5 
  cm. 
  thick. 
  Spiral 
  grooves 
  were 
  cut 
  in 
  the 
  upper 
  sur- 
  

   faces 
  of 
  the 
  lower 
  halves 
  of 
  the 
  discs 
  so 
  that 
  the 
  insulated 
  man- 
  

   ganin 
  heating-coils 
  and 
  the 
  constantan 
  thermo-couple 
  wires 
  could 
  

   be 
  embedded 
  in 
  paraffin 
  before 
  the 
  plates 
  were 
  assembled. 
  The 
  

   object 
  in 
  the 
  top 
  disc 
  — 
  which 
  was 
  kept 
  at 
  a 
  temperature 
  slightly 
  

   higher 
  than 
  that 
  of 
  the 
  central 
  disc 
  — 
  was 
  to 
  minimize 
  the 
  net 
  

   passage 
  of 
  heat 
  from 
  the 
  upper 
  surface 
  of 
  the 
  central 
  disc. 
  All 
  

   of 
  the 
  surfaces 
  were 
  silver 
  plated 
  and 
  highly 
  polished. 
  All 
  

   sources 
  of 
  error 
  were 
  thoroughly 
  investigated 
  and 
  the 
  correspond- 
  

   ing 
  corrections 
  were 
  so 
  obtained 
  as 
  to 
  give 
  an 
  accuracy 
  of 
  one 
  

   part 
  in 
  500 
  in 
  the 
  final 
  value 
  of 
  k. 
  The 
  formula 
  representing 
  

   the 
  exchanges 
  of 
  heat 
  may 
  be 
  written 
  

  

  E 
  2 
  /B 
  = 
  kSiO—OJ/x 
  + 
  R' 
  + 
  C{dO/dt) 
  + 
  L, 
  

   where 
  E 
  2 
  /R 
  = 
  energy 
  supplied 
  electrically 
  per 
  second 
  to 
  the 
  

   central 
  disc, 
  8 
  = 
  effective 
  area 
  of 
  central 
  disc, 
  2 
  ±= 
  temperature 
  

   of 
  this 
  disc, 
  6 
  1 
  = 
  temperature 
  of 
  bottom 
  disc, 
  x 
  = 
  distance 
  

   between 
  central 
  and 
  bottom 
  discs, 
  kS(0 
  2 
  — 
  ^ 
  1 
  )/ic 
  = 
  heat 
  passing 
  

   per 
  second 
  by 
  conduction 
  from 
  the 
  central 
  disc 
  to 
  the 
  bottom 
  one, 
  

   R' 
  = 
  heat 
  transferred 
  by 
  radiation 
  from 
  the 
  central 
  disc, 
  

   C(dO/dt) 
  — 
  heat 
  absorbed 
  per 
  unit 
  time 
  by 
  this 
  disc 
  if 
  its 
  tem- 
  

   perature 
  changes 
  with 
  time, 
  and 
  L 
  = 
  heat 
  passing 
  from 
  the 
  cen- 
  

   tral 
  disc 
  to 
  the 
  top 
  disc 
  and 
  to 
  the 
  guard 
  ring. 
  

  

  The 
  final 
  value 
  of 
  k, 
  for 
  22°C, 
  is 
  given 
  as 
  5-74 
  X 
  10~ 
  5 
  C.G.S. 
  

   units. 
  Assuming 
  that 
  the 
  relation 
  kg 
  = 
  k 
  (1 
  + 
  yB) 
  holds, 
  and 
  

   taking 
  y 
  = 
  0-0029 
  (1°C.) 
  _1 
  , 
  the 
  specific 
  thermal 
  conductivity 
  of 
  

   dry 
  air 
  at 
  0° 
  C. 
  is 
  found 
  to 
  be 
  k 
  n 
  = 
  5-40 
  X 
  10 
  -5 
  . 
  By 
  weighting 
  

   the 
  values 
  of 
  k 
  derived 
  from 
  the 
  present 
  and 
  thirteen 
  earlier 
  

   investigations, 
  the 
  mean 
  value 
  comes 
  out 
  as 
  5-22 
  XlO 
  -5 
  cal./(cm. 
  

   sec. 
  cleg.). 
  The 
  average 
  departure 
  from 
  this 
  mean 
  is 
  0-38 
  or 
  

   7 
  per 
  cent., 
  showing 
  that 
  the 
  agreement 
  between 
  different 
  observ- 
  

   ers 
  is 
  very 
  poor. 
  Nevertheless 
  the 
  average 
  of 
  the 
  nine 
  results 
  

   obtained 
  by 
  the 
  "cooling 
  method" 
  is 
  5-24, 
  of 
  the 
  three 
  by 
  the 
  

   "hot-wire 
  method" 
  is 
  5-26, 
  and 
  of 
  the 
  two 
  by 
  Poynting's 
  method 
  

   is 
  5-18 
  ; 
  hence 
  the 
  general 
  weighted 
  mean 
  given 
  above 
  is 
  probably 
  

   fairly 
  close 
  to 
  the 
  true 
  value. 
  

  

  The 
  last 
  part 
  of 
  the 
  paper 
  deals 
  with 
  the 
  constant 
  /. 
  The 
  nec- 
  

   essary 
  numerical 
  data 
  from 
  many 
  sources 
  have 
  been 
  collected, 
  

   properly 
  weighted, 
  and, 
  apparently 
  for 
  the 
  first 
  time, 
  presented 
  

   in 
  a 
  single 
  table. 
  The 
  numbers 
  following 
  the 
  symbols 
  for 
  the 
  

   gases 
  are 
  the 
  newly 
  computed 
  values 
  of 
  f: 
  He 
  2-31, 
  A 
  2-47; 
  H 
  2 
  

   1-76, 
  N 
  2 
  1-76, 
  2 
  1-79, 
  air 
  1-76, 
  NO 
  1-73, 
  CO 
  1-72; 
  C0 
  2 
  1-45, 
  

   N 
  2 
  1-47, 
  H 
  2 
  S 
  1-34, 
  S0 
  2 
  1-35; 
  Cl 
  2 
  1-50, 
  NH 
  3 
  1-23, 
  CH 
  4 
  1-45, 
  

   CoH 
  4 
  1-27. 
  Finally, 
  it 
  was 
  found 
  that 
  / 
  = 
  2-816y— 
  2-2, 
  i. 
  e. 
  / 
  

  

  