﻿276 
  On 
  Temperature 
  Compensation 
  of 
  Standard 
  Cells. 
  

  

  copper. 
  This 
  also 
  simplifies 
  the 
  calculation 
  of 
  the 
  ratios 
  n 
  

   and 
  p, 
  and 
  still 
  gives 
  sufficient 
  variation 
  to 
  balance 
  that 
  of 
  a 
  

   Clark 
  cell. 
  It 
  is 
  convenient 
  to 
  place 
  the 
  cell 
  inside 
  the 
  wide 
  

   brass 
  tube 
  on 
  which 
  the 
  coils 
  are 
  wound. 
  

  

  In 
  the 
  other 
  arrangement 
  exhibited 
  three 
  resistance-coils 
  

   are 
  connected 
  as 
  in 
  fig. 
  2. 
  

  

  Fig. 
  2. 
  

  

  

  /WWW 
  COPPER 
  

  

  MANCANIN. 
  

  

  The 
  resistances 
  r 
  and 
  s 
  are 
  of 
  manganin, 
  and 
  ^is 
  of 
  copper. 
  

   A 
  current 
  C 
  is 
  sent 
  from 
  D 
  to 
  G 
  from 
  an 
  outside 
  source. 
  

   The 
  values 
  of 
  g, 
  r, 
  and 
  s 
  are 
  so 
  chosen 
  that 
  a 
  fixed 
  value 
  of 
  

   C 
  produces 
  at 
  all 
  temperatures 
  a 
  potential-difference 
  from 
  F 
  

   to 
  G 
  which 
  would 
  balance 
  the 
  voltage 
  of 
  the 
  standard. 
  

  

  In 
  the 
  specimen 
  shown 
  3*75 
  q 
  = 
  r 
  + 
  s 
  and 
  r 
  = 
  s. 
  The 
  

   actual 
  values 
  at 
  15° 
  C. 
  are 
  

  

  9 
  , 
  = 
  s 
  = 
  3 
  , 
  6385 
  international 
  ohms. 
  

   g= 
  1-9382 
  

  

  A 
  current 
  of 
  1*000 
  ampere 
  gives 
  a 
  balance 
  with 
  a 
  Clark 
  cell 
  

   at 
  F 
  and 
  G 
  at 
  all 
  ordinary 
  temperatures 
  (to 
  within 
  1 
  in 
  2000). 
  

   This 
  definite 
  current, 
  obtainable 
  at 
  all 
  temperatures, 
  can 
  be 
  

   passed 
  through 
  known 
  resistances 
  and 
  so 
  used 
  to 
  measure 
  

   unknown 
  voltages 
  by 
  potentiometer 
  methods. 
  

  

  If 
  the 
  resistance 
  r 
  be 
  subdivided 
  other 
  exact 
  currents 
  such 
  

   as 
  2, 
  3, 
  4 
  . 
  . 
  . 
  amperes 
  can 
  be 
  got 
  directly. 
  

  

  It 
  should 
  be 
  noted 
  that 
  although 
  1 
  have 
  made 
  special 
  

   reference 
  to 
  the 
  Clark 
  cell, 
  either 
  of 
  the 
  two 
  forms 
  of 
  com- 
  

   pensator 
  may 
  be 
  arranged 
  for 
  any 
  other 
  type 
  of 
  standard 
  cell 
  

   so 
  as 
  to 
  wipe 
  out 
  the 
  greater 
  part 
  of 
  the 
  temperature 
  varia- 
  

   tion. 
  For 
  instance, 
  if 
  a 
  Carhart 
  1 
  volt 
  cell 
  has 
  a 
  coefficient 
  

   of 
  0*0097 
  per 
  cent, 
  per 
  degree 
  C, 
  used 
  with 
  a 
  compensator 
  it 
  

   would 
  not 
  vary 
  by 
  more 
  than 
  0*001 
  per 
  cent, 
  per 
  degree. 
  It 
  

   will 
  thus 
  be 
  seen 
  that 
  in 
  some 
  cases 
  a 
  compensating 
  arrange- 
  

   ment 
  such 
  as 
  I 
  have 
  described 
  above 
  would 
  prove 
  of 
  value 
  

   not 
  only 
  in 
  ordinary 
  testing 
  work 
  but 
  also 
  in 
  measurements 
  

   where 
  much 
  higher 
  accuracy 
  is 
  required. 
  

  

  