﻿W. 
  P. 
  White 
  — 
  Specific 
  Heat 
  Determination. 
  55 
  

  

  approached 
  1 
  per 
  thousand, 
  and 
  were, 
  on 
  the 
  whole, 
  the 
  

   most 
  serious 
  errors 
  encountered. 
  A 
  knowledge 
  of 
  the 
  

   cause 
  of 
  these 
  errors 
  and 
  of 
  possible 
  methods 
  of 
  pre- 
  

   vention 
  in 
  other 
  cases 
  would 
  therefore 
  be 
  a 
  useful 
  result 
  

   of 
  the 
  present 
  work. 
  The 
  variety 
  of 
  the 
  observations 
  

   was 
  such 
  that 
  it 
  seems 
  possible 
  to 
  disentangle, 
  to 
  a 
  con- 
  

   siderable 
  extent, 
  the 
  causes 
  of 
  these 
  discrepancies. 
  

  

  It 
  seems 
  reasonably 
  certain 
  that 
  the 
  process 
  of 
  drop- 
  

   ping 
  the 
  crucible 
  was 
  mainly 
  responsible 
  for 
  them. 
  

  

  Their 
  possible 
  sources 
  seem 
  to 
  be 
  fully 
  covered 
  by 
  the 
  

   following 
  five: 
  (1) 
  The 
  temperature 
  determination 
  in 
  

   the 
  furnace; 
  (2) 
  the 
  loss 
  of 
  heat 
  in 
  the 
  drop; 
  (3) 
  the 
  

   exposure 
  of 
  the 
  calorimeter 
  to 
  the 
  outer 
  air 
  during 
  the 
  

   drop; 
  (4) 
  the 
  effect 
  of 
  external 
  conditions 
  (jacket 
  or 
  

   electrical 
  measuring 
  system) 
  on 
  the 
  calorimeter 
  after 
  

   the 
  drop; 
  (5) 
  internal 
  effects 
  (temperature 
  change) 
  in 
  

   the 
  calorimeter 
  after 
  the 
  drop. 
  

  

  Of 
  these 
  sources 
  of 
  error, 
  the 
  external 
  ones, 
  (3) 
  and 
  

   (4), 
  will 
  be 
  independent 
  of 
  the 
  heat 
  quantity 
  measured, 
  

   and 
  will 
  therefore 
  be 
  revealed 
  when 
  that 
  is 
  made 
  very 
  

   small; 
  in 
  the 
  second 
  place, 
  error 
  from 
  temperature 
  

   change 
  in 
  the 
  calorimeter, 
  (5), 
  will 
  be 
  proportional 
  at 
  

   least 
  roughly 
  to 
  the 
  heat 
  quantity, 
  and 
  so 
  will 
  error 
  in 
  

   the 
  temperature 
  measurement 
  in 
  the 
  furnace, 
  (1), 
  since 
  

   that 
  affects 
  the 
  whole 
  charge 
  ; 
  such 
  errors 
  will 
  therefore 
  

   be 
  revealed 
  by 
  varying 
  the 
  total 
  heat 
  ; 
  and 
  finally, 
  error 
  

   coming 
  from 
  the 
  dropping 
  process, 
  (2), 
  will 
  presumably 
  

   vary 
  with 
  the 
  temperature, 
  but 
  only 
  on 
  account 
  of 
  the 
  

   container, 
  and 
  will 
  be 
  independent 
  of 
  the 
  charge, 
  since 
  

   that 
  loses 
  no 
  heat 
  in 
  the 
  drop. 
  The 
  average 
  discrepan- 
  

   cies, 
  in 
  calories, 
  of 
  about 
  160 
  determinations 
  are 
  given 
  in 
  

   Table 
  I. 
  From 
  them 
  it 
  appears: 
  (a) 
  The 
  discrepancies 
  

   at 
  100°, 
  which 
  include 
  as 
  many 
  external 
  sources 
  of 
  error 
  

   as 
  any 
  of 
  the 
  determinations, 
  are 
  relatively 
  small, 
  and 
  

   thus 
  show 
  that 
  such 
  errors 
  are 
  probably 
  only 
  a 
  minor 
  

   factor 
  at 
  all 
  higher 
  temperatures. 
  17 
  (b) 
  Above 
  100° 
  

  

  17 
  The 
  discrepancies 
  in 
  dropping 
  the 
  container 
  alone 
  at 
  100° 
  are 
  much 
  

   less 
  than 
  with 
  the 
  charges. 
  It 
  is 
  probable 
  that 
  the 
  better 
  results 
  with 
  the 
  

   container 
  show 
  the 
  precision 
  reached 
  with 
  the 
  calorimeter, 
  itself, 
  since 
  the 
  

   lower 
  temperatures 
  of 
  the 
  charges 
  were 
  determined, 
  by 
  means 
  of 
  a 
  small 
  

   auxiliary 
  mercury 
  thermometer, 
  with 
  a 
  precision 
  which 
  is 
  now 
  seen 
  to 
  be 
  

   less 
  than 
  corresponds 
  to 
  that 
  of 
  the 
  container 
  results, 
  though 
  enough 
  for 
  

   the 
  argument 
  above 
  and 
  for 
  the 
  main 
  objects 
  of 
  the 
  investigation. 
  At 
  

   any 
  rate, 
  for 
  the 
  8 
  container 
  determinations 
  the 
  exposures 
  to 
  the 
  air 
  were 
  

   so 
  uniform 
  that 
  over 
  0-8 
  of 
  the 
  exposure 
  effect 
  canceled 
  out, 
  though 
  the 
  

   necessary 
  conditions 
  for 
  this 
  to 
  occur 
  again 
  can 
  not 
  be 
  exactly 
  stated. 
  

  

  