﻿the 
  Thermoelectromotive 
  Force 
  of 
  Nickel 
  and 
  Iron. 
  187 
  

  

  were 
  heated 
  in 
  hydrogen, 
  which 
  were 
  more 
  systematically 
  

   investigated 
  (see 
  below). 
  

  

  In 
  carbon, 
  observations 
  taken 
  up 
  to 
  700° 
  agreed 
  with 
  the 
  

   hydrogen 
  and 
  vacuum 
  results. 
  Above 
  this 
  temperature, 
  

   carbides 
  of 
  iron 
  are 
  probably 
  formed 
  which 
  give 
  rise 
  to 
  

   the 
  inconsistency 
  in 
  the 
  results, 
  which 
  has 
  already 
  been 
  

   remarked 
  on. 
  It 
  may 
  be 
  mentioned 
  that 
  these 
  observations 
  

   gave 
  values 
  of 
  E.M.F. 
  which 
  were 
  considerably 
  greater 
  than 
  

   those 
  in 
  hydrogen. 
  

  

  It 
  is 
  evident 
  that 
  thermoelectric 
  couples 
  of 
  copper-iron 
  

   and 
  copper-nickel 
  packed 
  in 
  carbon 
  will 
  give 
  consistent 
  

   readings 
  of 
  E.M.F. 
  up 
  to 
  700° 
  or 
  800°, 
  but 
  that 
  above 
  this 
  

   temperature 
  the 
  method 
  fails. 
  

  

  Under 
  every 
  condition 
  tried, 
  copper, 
  nickel, 
  and 
  iron 
  wires 
  

   were 
  always 
  extremely 
  brittle 
  after 
  heating 
  above 
  700°. 
  

   After 
  heating 
  in 
  hydrogen 
  they 
  were 
  invariably 
  quite 
  bright, 
  

   notably 
  the 
  copper. 
  This 
  last 
  metal 
  volatilized 
  considerably 
  

   above 
  900°. 
  

  

  /7F 
  

  

  The 
  curves 
  of 
  thermoelectric 
  power 
  — 
  ■ 
  , 
  shown 
  in 
  fig. 
  6, 
  

  

  were 
  obtained 
  by 
  drawing 
  tangents 
  to 
  the 
  E.M.F. 
  curves 
  

   (figs. 
  3 
  and 
  5). 
  A 
  considerable 
  range 
  

  

  On 
  

  

  / 
  from 
  -200° 
  to 
  +100°\ 
  

   \tmd 
  from 
  +100° 
  to 
  +400°/ 
  

  

  of 
  the 
  copper-iron 
  curve 
  can 
  be 
  represented 
  by 
  straight 
  lines. 
  

   Between 
  + 
  400° 
  and 
  + 
  800°, 
  however, 
  the 
  curve 
  appears 
  to 
  

   be 
  parabolic, 
  and 
  cannot 
  be 
  built 
  up 
  of 
  bits 
  of 
  straight 
  lines. 
  

  

  dE 
  . 
  

   Above 
  800° 
  -j- 
  increases 
  rapidly, 
  being 
  nearly 
  linear 
  be- 
  

   tween 
  900° 
  and 
  1050°. 
  The 
  Cu-Ni 
  power 
  curve 
  can 
  be 
  

   represented 
  by 
  bits 
  of 
  straight 
  lines. 
  Changes 
  in 
  slope 
  occur 
  

   at 
  240° 
  and 
  380°. 
  

  

  /7TT 
  

   The 
  Peltier 
  coefficients 
  T-7- 
  , 
  obtained 
  by 
  multiplying 
  the 
  

  

  thermoelectric 
  power 
  by 
  the 
  corresponding 
  absolute 
  tem- 
  

   perature, 
  are 
  given 
  in 
  the 
  same 
  figure 
  (fig. 
  6). 
  The 
  curve 
  

   for 
  iron-copper 
  is 
  approximately 
  parabolic 
  between 
  —200° 
  

   and 
  +300°. 
  After 
  that 
  it 
  can 
  be 
  made 
  up 
  of 
  straight 
  lines. 
  

   The 
  break 
  just 
  below 
  800° 
  is 
  due 
  to 
  the 
  fact 
  that 
  the 
  E.M.F; 
  

   curve 
  for 
  results 
  above 
  800° 
  did 
  not 
  quite 
  join, 
  but 
  was 
  

   approximately 
  parallel 
  to 
  that 
  for 
  lower 
  temperatures. 
  For 
  

   Cu-Ni 
  the 
  Peltier 
  coefficient-variation 
  can 
  be 
  built 
  up 
  of 
  bits 
  

   of 
  parabolas, 
  if 
  the 
  power-curve 
  is 
  strictly 
  linear 
  ; 
  but 
  

   uncertainty 
  in 
  the 
  values 
  obtained 
  by 
  differentiating 
  an 
  

  

  