﻿W. 
  P. 
  White 
  — 
  Silicate 
  Specific 
  Heats. 
  3 
  

  

  errors 
  is 
  a 
  rather 
  imperfect 
  guide. 
  Dr. 
  M. 
  D. 
  Hersey 
  has 
  

   informally 
  proposed 
  a 
  suggestive 
  classification 
  into 
  

   "visible" 
  and 
  "concealed" 
  errors, 
  where 
  the 
  visible 
  

   errors 
  are 
  those 
  ordinarily 
  called 
  accidental. 
  But 
  the 
  

   concealed 
  errors 
  are 
  of 
  several 
  different 
  kinds. 
  The 
  

   most 
  useful 
  basis 
  of 
  classification 
  is 
  perhaps 
  according 
  

   to 
  the 
  constancy 
  of 
  the 
  causes 
  of 
  error. 
  Following 
  it, 
  we 
  

   may 
  distinguish 
  3 
  sorts 
  of 
  error 
  : 
  

  

  1. 
  The 
  visible 
  or 
  "accidental" 
  errors, 
  that 
  is, 
  errors 
  

   due 
  to 
  conditions 
  which 
  are 
  almost 
  continuously 
  varying. 
  

  

  2. 
  Errors 
  from 
  constant 
  causes 
  of 
  deviation 
  whose 
  

   presence 
  will 
  not 
  appear 
  at 
  all 
  unless 
  as 
  a 
  result 
  of 
  special 
  

   investigation 
  (the 
  systematic 
  errors 
  par 
  excellence) 
  . 
  

  

  3. 
  Errors 
  from 
  slowly 
  or 
  occasionally 
  varying 
  causes, 
  

   often 
  the 
  hardest 
  to 
  detect 
  of 
  all. 
  

  

  It 
  is 
  always 
  likely 
  that 
  the 
  rapidly 
  varying 
  deviations 
  

   do 
  not 
  vary 
  impartially 
  about 
  zero 
  ; 
  hence 
  there 
  is 
  always 
  

   reason 
  to 
  suspect 
  concealed 
  errors 
  due 
  to 
  the 
  same 
  causes 
  

   as 
  the 
  visible 
  errors. 
  It 
  is, 
  therefore, 
  especially 
  impor- 
  

   tant 
  to 
  inquire 
  how 
  far 
  the 
  visible 
  errors 
  can 
  be 
  accounted 
  

   for 
  by 
  causes 
  which 
  would 
  not 
  produce 
  concealed 
  error. 
  

  

  These 
  various 
  causes 
  of 
  error 
  can 
  be 
  partly 
  estimated 
  

   by 
  considering 
  the 
  methods 
  and 
  precautions 
  used, 
  and 
  

   also 
  by 
  examining 
  the 
  results. 
  

  

  Methods 
  and 
  Precautions. 
  

   The 
  High 
  Temperature 
  Standard. 
  

  

  The 
  furnace 
  temperatures 
  were 
  based 
  on 
  the 
  identical 
  

   "Element 
  G" 
  used 
  by 
  Day 
  and 
  Sosman 
  in 
  establishing 
  

   their 
  high 
  temperature 
  scale, 
  checked 
  by 
  six 
  concordant 
  

   determinations 
  of 
  the 
  silver 
  and 
  gold 
  melting 
  points, 
  and 
  

   by 
  comparisons 
  with 
  2 
  other 
  standards, 
  standardized 
  by 
  

   silver 
  and 
  gold, 
  and 
  directly 
  intercompared 
  at 
  4 
  tempera- 
  

   tures 
  from 
  500° 
  to 
  1400°. 
  

  

  The 
  maximum 
  discrepancy 
  in 
  these 
  various 
  observa- 
  

   tions 
  was 
  0-7°. 
  Since 
  even 
  these 
  small 
  discrepancies 
  are 
  

   largely 
  accidental 
  their 
  final 
  effect 
  is 
  quite 
  negligible 
  ; 
  of 
  

   far 
  more 
  consequence 
  is 
  the 
  question 
  whether 
  the 
  thermo- 
  

   elements 
  read 
  the 
  same 
  in 
  actual 
  work 
  as 
  they 
  did 
  in 
  the 
  

   calibrations. 
  "Whether 
  they 
  did 
  or 
  not 
  depends, 
  of 
  

   course, 
  on 
  the 
  homogeneity 
  of 
  the 
  thermoelements 
  and 
  on 
  

   the 
  constancy 
  of 
  the 
  position 
  of 
  the 
  temperature 
  gradi- 
  

   ent. 
  3 
  The 
  homogeneity 
  of 
  the 
  standards 
  was 
  thoroughly 
  

  

  3 
  Cf 
  . 
  Walter 
  P. 
  White, 
  The 
  Thermoelement 
  as 
  a 
  Precision 
  Thermometer, 
  

   Phys. 
  Eev. 
  31, 
  137, 
  1910; 
  or, 
  Thermoelements 
  of 
  Precision, 
  especially 
  for 
  

   Calorimetry, 
  J. 
  Am. 
  Chem. 
  Soc, 
  36, 
  2293, 
  1914. 
  

  

  