﻿36 
  JOURNAL 
  OF 
  THB 
  WASHINGTON 
  ACADEMY 
  OF 
  SCIENCES 
  VOL. 
  12, 
  NO. 
  2 
  

  

  The 
  closeness 
  of 
  the 
  fit 
  of 
  the 
  two 
  lines 
  or 
  of 
  the 
  single 
  lines 
  may 
  be 
  

   judged 
  from 
  figures 
  3 
  and 
  4. 
  Figure 
  3 
  shows 
  the 
  results 
  for 
  Tschardjui, 
  

   the 
  means 
  being 
  taken 
  by 
  calendar 
  years. 
  It 
  has 
  been 
  said 
  that 
  this 
  

   station 
  has 
  been 
  subject 
  to 
  great 
  irregularities 
  and 
  this 
  appears 
  plainly 
  

   from 
  the 
  diagram. 
  The 
  only 
  thing 
  that 
  clearly 
  appears 
  is 
  a 
  tendency 
  

   for 
  latitudes 
  to 
  increase 
  very 
  sharply 
  towards 
  the 
  end 
  of 
  the 
  period. 
  

   Figure 
  4 
  shows 
  Mizusawa, 
  perhaps 
  the 
  most 
  regular 
  station, 
  though 
  

   the 
  other 
  stations 
  except 
  Tschardjui 
  and 
  Cincinnati 
  are 
  not 
  greatly 
  

   inferior 
  to 
  it. 
  Even 
  the 
  most 
  regular 
  of 
  stations 
  shows 
  considerable 
  

   departure 
  from 
  a 
  perfectly 
  uniform 
  progressive 
  increase, 
  although 
  the 
  

   irregularity 
  is 
  somewhat 
  exaggerated 
  to 
  the 
  eye 
  by 
  the 
  large 
  vertical 
  

   scale 
  used 
  for 
  the 
  latitudes. 
  The 
  probable 
  error 
  of 
  the 
  mean 
  of 
  a 
  single 
  

   year 
  or 
  of 
  a 
  single 
  14-month 
  period 
  is 
  about 
  ±0".010 
  at 
  Mizusawa 
  as 
  

   determined 
  from 
  the 
  residuals 
  arising 
  from 
  attempting 
  to 
  fit 
  the 
  two 
  

   straight 
  lines 
  to 
  the 
  successive 
  means. 
  The 
  probable 
  error 
  is 
  rather 
  

   larger 
  when 
  only 
  a 
  single 
  line 
  is 
  used, 
  a 
  fact 
  which 
  tends 
  to 
  prove 
  the 
  

   reality 
  of 
  the 
  assumed 
  change 
  in 
  rate. 
  Similar 
  results 
  hold 
  for 
  the 
  

   other 
  stations 
  except 
  Tschardjui 
  and 
  Cincinnati, 
  the 
  probable 
  errors 
  

   being 
  about 
  ±0".015 
  for 
  the 
  two 
  lines 
  and 
  about 
  ±0".018 
  for 
  one 
  line. 
  

  

  The 
  mean 
  rates 
  of 
  increase 
  are 
  shown 
  in 
  table 
  2. 
  They 
  represent 
  

   the 
  mean 
  results 
  of 
  the 
  two 
  methods 
  of 
  procedure 
  already 
  referred 
  to, 
  

   mean 
  latitudes 
  being 
  taken 
  by 
  calendar 
  years 
  and 
  by 
  14-month 
  periods. 
  

   The 
  fitting 
  of 
  the 
  straight 
  lines 
  to 
  the 
  means 
  was 
  done 
  by 
  the 
  method 
  

   of 
  least 
  squares. 
  

  

  TABLE 
  2. 
  — 
  Mean 
  Annual 
  Rates 
  op 
  Change 
  "op 
  Latitude 
  

  

  Observations 
  Annual 
  rates 
  of 
  change 
  

  

  Station 
  Longitude 
  end 
  with 
  year 
  1900-11 
  1912-end 
  1900-end 
  

  

  Mizusawa, 
  Japan 
  141° 
  08' 
  E 
  .. 
  -0".0025 
  +0".0096 
  +0".0008 
  

  

  Tschardjui, 
  Russian 
  Turkes- 
  

   tan 
  63 
  29 
  E 
  1914 
  +0".0036 
  +0".0597 
  -1-0".0115 
  

  

  Carloforte, 
  Sardinia 
  8 
  19 
  E 
  .. 
  +0 
  .0002 
  +0 
  .0182 
  +0 
  .0053 
  

  

  Gaithersburg, 
  Maryland.... 
  77 
  12 
  W 
  1914 
  +0.0087 
  +0.0206 
  +0.0105 
  

  

  Cincinnati, 
  Ohio 
  84 
  25 
  W 
  1915 
  +0 
  .0029 
  +0 
  .0404 
  +0 
  .0099 
  

  

  Ukiah, 
  California 
  123 
  13 
  W 
  .. 
  +0.0075 
  +0.0194 
  +0.0106 
  

  

  The 
  first 
  two 
  columns 
  under 
  the 
  general 
  heading 
  "Annual 
  rates 
  of 
  

   change" 
  give 
  the 
  slopes 
  or 
  rates 
  of 
  change 
  of 
  latitude 
  when 
  two 
  lines 
  

   are 
  used. 
  The 
  last 
  column 
  gives 
  the 
  slope 
  when 
  only 
  one 
  line 
  is 
  used. 
  

   The 
  rate 
  of 
  change 
  may 
  vary 
  with 
  the 
  period 
  of 
  time 
  covered 
  so 
  that 
  

   only 
  in 
  the 
  first 
  column 
  are 
  the 
  rates 
  for 
  all 
  stations 
  strictly 
  comparable 
  

   with 
  one 
  another. 
  In 
  the 
  second 
  and 
  third 
  columns 
  the 
  rates 
  of 
  

   Mizusawa, 
  Carloforte, 
  and 
  Ukiah 
  are 
  comparable 
  in 
  this 
  way. 
  The 
  

  

  