﻿8 
  DE. 
  CHAELES 
  DAVISON 
  ON 
  THE 
  [Feb. 
  1 
  897, 
  

  

  all 
  collected 
  at 
  the 
  centre 
  of 
  the 
  rectangle, 
  and 
  the 
  magnitude 
  of 
  

   the 
  possible 
  error 
  increases 
  with 
  that 
  of 
  the 
  rectangle. 
  Supposing, 
  

   however, 
  the 
  extreme 
  case 
  to 
  occur 
  in 
  which 
  all 
  the 
  epicentres 
  are 
  

   situated 
  at 
  one 
  point 
  close 
  to 
  a 
  side 
  of 
  the 
  rectangle, 
  then 
  the 
  

   maximum 
  deviation 
  of 
  any 
  curve 
  from 
  its 
  correct 
  position 
  would 
  be 
  

   equal 
  to 
  -JL 
  of 
  a 
  degree, 
  i. 
  e. 
  to 
  -£$ 
  of 
  a 
  side 
  of 
  the 
  complete 
  map. 
  

   The 
  probable 
  error 
  is, 
  of 
  course, 
  much 
  less 
  than 
  this. 
  

  

  Another 
  source, 
  less 
  of 
  error 
  than 
  of 
  misconception, 
  should 
  also 
  

   be 
  mentioned. 
  This 
  is 
  due 
  to 
  the 
  form 
  of 
  the 
  rectangles, 
  a 
  degree 
  

   of 
  latitude 
  being 
  always 
  longer 
  than 
  a 
  degree 
  of 
  longitude, 
  except 
  

   at 
  the 
  equator. 
  "When 
  a 
  number 
  of 
  epicentres 
  lie 
  within 
  a 
  single 
  

   isolated 
  rectangle, 
  the 
  resulting 
  curve 
  is 
  a 
  small 
  ellipse 
  whose 
  length 
  

   is 
  less 
  than 
  i 
  of 
  a 
  degree, 
  but 
  the 
  direction 
  of 
  the 
  longer 
  axis 
  

   has 
  no 
  physical 
  meaning, 
  for 
  it 
  is 
  necessarily 
  parallel 
  to 
  the 
  longer 
  

   side 
  of 
  the 
  rectangle. 
  

  

  Again, 
  when 
  a 
  curve 
  is 
  founded 
  on 
  the 
  occurrence 
  of 
  epicentres 
  

   in 
  two 
  adjacent 
  rectangles 
  alone, 
  its 
  axis 
  will 
  generally 
  be 
  parallel 
  

   to 
  the 
  line 
  joining 
  their 
  centres 
  ; 
  and 
  its 
  direction 
  need 
  not, 
  though 
  

   it 
  may, 
  correspond 
  with 
  that 
  of 
  some 
  physical 
  feature, 
  for 
  it 
  is 
  

   possible 
  that 
  all 
  the 
  epicentres 
  might 
  be 
  grouped 
  close 
  to 
  the 
  common 
  

   side 
  of 
  the 
  two 
  rectangles. 
  It 
  is 
  only 
  when 
  the 
  longer 
  axis 
  of 
  the 
  

   curve 
  exceeds 
  in 
  length 
  the 
  sides 
  of 
  several 
  rectangles 
  that 
  we 
  

   can 
  make 
  any 
  probable 
  inference 
  as 
  to 
  the 
  connexion 
  between 
  

   the 
  distribution 
  of 
  after-shocks 
  and 
  the 
  geological 
  structure 
  of 
  the 
  

   district. 
  

  

  6. 
  Turning 
  now 
  to 
  the 
  maps 
  themselves, 
  fig. 
  1 
  (p. 
  4) 
  shows 
  the 
  

   outline 
  of 
  the 
  coast 
  within 
  the 
  Mino-Owari 
  district. 
  The 
  dotted 
  lines 
  

   represent 
  the 
  boundary 
  of 
  the 
  area 
  over 
  which 
  the 
  principal 
  shock 
  

   was 
  felt 
  most 
  severely. 
  The 
  continuous 
  line 
  between 
  them 
  marks 
  

   the 
  course 
  of 
  the 
  great 
  fault-scarp, 
  the 
  northern 
  part 
  of 
  which, 
  not 
  

   having 
  been 
  actually 
  followed 
  on 
  the 
  ground, 
  is 
  indicated 
  by 
  a 
  

   dotted 
  line. 
  1 
  The 
  points 
  A 
  to 
  P 
  denote 
  the 
  centres 
  of 
  the 
  rect- 
  

   angles, 
  2 
  within 
  which 
  the 
  majority 
  of 
  the 
  after-shock 
  epicentres 
  

   lie 
  ; 
  this 
  group 
  of 
  rectangles, 
  as 
  will 
  be 
  seen 
  from 
  the 
  last 
  column 
  

   of 
  Table 
  I., 
  contains 
  sometimes 
  all, 
  and 
  never 
  less 
  than 
  T 
  9 
  ^ 
  of 
  

   the 
  total 
  number 
  of 
  epicentres 
  within 
  the 
  district 
  contemplated. 
  

   In 
  the 
  succeeding 
  maps, 
  the 
  coast-line 
  and 
  fault-scarp 
  are 
  indicated 
  

   by 
  dotted 
  lines, 
  the 
  boundary 
  of 
  the 
  meizoseismal 
  area 
  being 
  omitted 
  

   in 
  order 
  to 
  avoid 
  confusion. 
  

  

  The 
  remaining 
  figures 
  are 
  selected 
  from 
  a 
  large 
  number 
  of 
  maps. 
  

   Fig. 
  2 
  represents 
  the 
  distribution 
  of 
  earthquakes 
  during 
  the 
  pericd 
  

   (January 
  1st, 
  1890, 
  to 
  October 
  27th, 
  1891) 
  immediately 
  preceding 
  

   the 
  great 
  earthquake. 
  The 
  others 
  illustrate 
  the 
  distribution 
  of 
  

   after-shocks 
  for 
  every 
  two 
  months 
  from 
  November 
  1891 
  to 
  October 
  

   1892. 
  

  

  1 
  The 
  course 
  of 
  the 
  fault-scarp 
  and 
  the 
  boundary 
  of 
  the 
  meizoseismal 
  area 
  

   are 
  reduced 
  from 
  plate 
  xxix. 
  in 
  Prof. 
  Koto's 
  memoir. 
  

  

  2 
  They 
  correspond 
  in 
  alphabetical 
  order 
  to 
  the 
  rectangles 
  in 
  Prof. 
  Milne's 
  

   map 
  numbered 
  1237, 
  1238, 
  1352, 
  1353, 
  1401, 
  1402, 
  1403, 
  1404, 
  1457, 
  1459, 
  

   1512, 
  1515, 
  and 
  1567. 
  

  

  