﻿362 
  Fenner 
  — 
  Stability 
  Relations 
  of 
  Silica 
  Minerals. 
  

  

  merit 
  B 
  in 
  the 
  ice-bath. 
  Terminals 
  b 
  and 
  c 
  are 
  connected 
  

   to 
  the 
  galvanometer 
  and 
  give 
  the 
  temperature 
  of 
  A. 
  C 
  is 
  

   placed 
  within 
  the 
  neutral 
  charge 
  alongside 
  of 
  A, 
  and 
  the 
  con- 
  

   nections 
  a 
  and 
  b 
  give 
  the 
  difference 
  in 
  temperature 
  between 
  A 
  

   and 
  C. 
  By 
  means 
  of 
  a 
  switch 
  one 
  may 
  read 
  first 
  the 
  current 
  

   in 
  a-b 
  and 
  then 
  the 
  current 
  in 
  b-c 
  on 
  the 
  same 
  galvanometer. 
  

   As 
  ordinarily 
  carried 
  out 
  the 
  temperature 
  of 
  A 
  was 
  read 
  and 
  

   noted 
  every 
  second 
  minute 
  during 
  heating 
  or 
  cooling, 
  and 
  the 
  

   difference 
  A-C 
  every 
  half 
  or 
  quarter 
  minute 
  except 
  at 
  the 
  two- 
  

   minute 
  points. 
  

  

  Under 
  ideal 
  conditions 
  the 
  neutral 
  body 
  should 
  be 
  of 
  such 
  a 
  

   nature 
  and 
  so 
  placed 
  in 
  the 
  furnace 
  that 
  its 
  temperature 
  should 
  

   be 
  exactly 
  equal 
  to 
  that 
  of 
  the 
  other, 
  and 
  hence 
  the 
  temperature- 
  

   reading 
  between 
  them 
  should 
  be 
  zero 
  up 
  to 
  the 
  point 
  at 
  which 
  

   the 
  substance 
  under 
  observation 
  begins 
  to 
  undergo 
  transfor- 
  

   mation, 
  which 
  would 
  cause 
  its 
  temperature 
  to 
  rise 
  less 
  rapidly 
  

   than 
  that 
  of 
  the 
  neutral 
  body. 
  In 
  practise 
  this 
  could 
  not, 
  of 
  

   course, 
  be 
  realized 
  absolutely, 
  but 
  it 
  was 
  not 
  difficult 
  to 
  approach 
  

   the 
  ideal 
  condition 
  with 
  satisfactory 
  closeness. 
  

  

  Ordinarily 
  the 
  powdered 
  cristobalite 
  was 
  placed 
  in 
  a 
  thin- 
  

   walled 
  test-tube 
  of 
  about 
  10 
  mm 
  diameter, 
  and 
  the 
  junction 
  A 
  

   was 
  imbedded 
  within 
  the 
  powder. 
  In 
  a 
  similar 
  tube 
  was 
  placed 
  

   a 
  like 
  amount 
  of 
  powdered 
  quartz 
  or 
  feldspar 
  inclosing 
  B. 
  The 
  

   two 
  tubes, 
  placed 
  close 
  together 
  but 
  separated 
  by 
  an 
  air-space, 
  

   were 
  symmetrically 
  disposed 
  in 
  the 
  furnace 
  and 
  the 
  heating 
  

   current 
  turned 
  on. 
  As 
  the 
  temperature 
  rose 
  the 
  differential 
  

   reading 
  showed 
  some 
  variation, 
  which, 
  however, 
  was 
  small 
  and 
  

   followed 
  a 
  smooth 
  curve 
  up 
  to 
  a 
  certain 
  point, 
  after 
  which 
  the 
  

   progress 
  of 
  transformation 
  could 
  be 
  plainly 
  perceived 
  by 
  an 
  

   increasing 
  magnitude 
  of 
  deflection. 
  This 
  rapidly 
  reached 
  a 
  

   maximum 
  and 
  then 
  fell 
  back 
  with 
  equal 
  rapidity. 
  The 
  graph 
  of 
  

   a 
  typical 
  heating 
  and 
  cooling 
  curve 
  has 
  been 
  plotted 
  in 
  fig. 
  6. 
  

   The 
  temperature 
  at 
  which 
  the 
  peak 
  of 
  the 
  differential 
  curve 
  

   was 
  attained 
  was 
  considered 
  the 
  temperature 
  of 
  inversion 
  and 
  

   its 
  value 
  was 
  calculated 
  by 
  interpolation 
  between 
  the 
  nearest 
  

   temperature 
  readings 
  on 
  either 
  side. 
  A 
  linear 
  interpolation 
  

   under 
  such 
  conditions 
  is 
  not 
  strictly 
  correct, 
  but 
  the 
  magni- 
  

   tude 
  of 
  the 
  error, 
  due 
  to 
  this 
  source 
  and 
  to 
  one 
  or 
  two 
  others, 
  

   which 
  might 
  be 
  mentioned, 
  is 
  believed 
  to 
  be 
  so 
  small 
  as 
  to 
  be 
  

   negligible. 
  In 
  practice 
  a 
  repetition 
  of 
  determinations 
  was 
  fre- 
  

   quently 
  made 
  under 
  variation 
  of 
  conditions 
  and 
  gave 
  closely 
  

   concordant 
  results. 
  An 
  agreement 
  within 
  2° 
  was 
  common, 
  

   and 
  a 
  variation 
  of 
  4° 
  was 
  quite 
  unusual. 
  As 
  the 
  errors 
  are 
  not 
  

   of 
  such 
  a 
  nature 
  as 
  to 
  be 
  systematic, 
  the 
  limits 
  found 
  undoubt- 
  

   edly 
  show 
  the 
  degree 
  of 
  approximation 
  to 
  the 
  correct 
  value. 
  

   In 
  the 
  figures 
  which 
  are 
  given 
  later 
  the 
  calculated 
  tempera- 
  

   tures 
  are 
  expressed 
  to 
  the 
  nearest 
  tenth 
  of 
  a 
  degree, 
  but 
  only 
  

   the 
  integers 
  should 
  be 
  considered 
  of 
  real 
  significance. 
  

  

  