﻿W. 
  P. 
  White 
  — 
  Silicate 
  Specific 
  Heats. 
  7 
  

  

  better 
  than 
  4 
  per 
  mille 
  and 
  are 
  really 
  preliminary 
  results. 
  

   These 
  were 
  being 
  repeated 
  when 
  the 
  furnace 
  was 
  disabled 
  

   by 
  a 
  short 
  circuit 
  and 
  it 
  seemed 
  best 
  to 
  discontinue 
  the 
  

   work. 
  The 
  results 
  as 
  they 
  stand 
  are 
  sufficient 
  to 
  estab- 
  

   lish 
  all 
  the 
  conclusions 
  which 
  seem 
  possible 
  at 
  present; 
  

   and 
  experience 
  in 
  the 
  present 
  work 
  had 
  shown 
  that 
  when 
  

   better 
  results 
  seem 
  to 
  be 
  needed 
  they 
  can 
  be 
  obtained 
  

   more 
  rapidly 
  and 
  accurately 
  by 
  a 
  somewhat 
  different 
  

   procedure. 
  

  

  The 
  Heat 
  Determination 
  in 
  the 
  Calorimeter. 
  

   The 
  calorimeters 
  had 
  shown 
  themselves 
  capable 
  of 
  a 
  

   precision 
  approaching 
  0-2 
  per 
  mille 
  ; 
  8 
  the 
  thermoelectric 
  

   sensitiveness 
  was 
  further 
  increased 
  to 
  0-0001°, 
  which 
  was 
  

   far 
  more 
  than 
  was 
  necessary 
  for 
  most 
  of 
  the 
  determina- 
  

   tions 
  ; 
  the 
  unusual 
  practice 
  was 
  followed 
  of 
  computing 
  for 
  

   a 
  single 
  specific 
  heat 
  determination 
  3 
  or 
  more 
  cool- 
  

   ing 
  corrections, 
  from 
  data 
  mostly 
  different; 
  all 
  deter- 
  

   minations 
  were 
  made 
  in 
  duplicate. 
  (These 
  last 
  two 
  

   precautions 
  were 
  of 
  course 
  more 
  effective 
  in 
  avoiding 
  

   accidental 
  errors 
  and 
  mistakes 
  in 
  computation 
  than 
  sys- 
  

   tematic 
  errors.) 
  The 
  critical 
  calculations 
  which 
  were 
  not 
  

   controlled 
  by 
  duplication, 
  including 
  those 
  in 
  the 
  original 
  

   calibration 
  of 
  1910, 
  were 
  carefully 
  revised 
  and 
  for 
  the 
  

   most 
  part 
  done 
  and 
  written 
  out 
  anew 
  at 
  the 
  close 
  of 
  the 
  

   work 
  in 
  1916. 
  

  

  Special 
  Measures 
  disclosing 
  Systematic 
  Errors. 
  

   The 
  simultaneous 
  use 
  of 
  2 
  furnace 
  thermoelements, 
  and 
  

   the 
  employment 
  of 
  2 
  radically 
  different 
  methods 
  of 
  calo- 
  

   rimeter 
  calibration 
  have 
  already 
  been 
  mentioned. 
  The 
  2 
  

   furnaces 
  could 
  also 
  be 
  compared 
  by 
  using 
  both 
  for 
  the 
  

   same 
  substances 
  at 
  500°. 
  There 
  were, 
  besides: 
  2 
  sepa- 
  

   rately 
  calibrated 
  calorimeters, 
  which 
  were 
  intercompared 
  

   with 
  2 
  silicates 
  at 
  900° 
  ; 
  2 
  calorimetric 
  thermoelements, 
  

   of 
  different 
  type 
  and 
  sensitiveness, 
  also 
  thoroughly 
  inter- 
  

   compared 
  in 
  work 
  on 
  regular 
  charges 
  ; 
  2 
  very 
  different 
  

   steam 
  heaters 
  ; 
  and 
  2 
  platinum 
  containers. 
  Every 
  part 
  

   of 
  the 
  measuring 
  system 
  (but 
  two) 
  was 
  thus 
  replaced 
  by 
  

   another 
  at 
  some 
  time. 
  9 
  Again, 
  while 
  the 
  112 
  points 
  

   observed 
  called 
  for 
  224 
  duplicate 
  determinations, 
  over 
  

  

  8 
  A 
  Test 
  of 
  Calorimetric 
  Accuracy, 
  loc. 
  cit., 
  p. 
  700. 
  

  

  9 
  These 
  two 
  parts 
  were: 
  (1) 
  the 
  thermoelement 
  used 
  above 
  600°, 
  and 
  

   this 
  was 
  frequently 
  standardized; 
  (2) 
  the 
  potentiometer, 
  which 
  certainly 
  

   did 
  not 
  cause 
  appreciable 
  error. 
  

  

  