﻿Ko.n34. 
  I'HILIPriXE 
  OEXITHOUXrY—WOUCESTER 
  AXI> 
  IKfEEXS. 
  609 
  

  

  is 
  any 
  relatiouship 
  between 
  tbe 
  size 
  of 
  an 
  island 
  and 
  the 
  number 
  of 
  

   species 
  of 
  a 
  genus 
  likely 
  to 
  be 
  found 
  upon 
  it. 
  

  

  A 
  glance 
  at 
  Table 
  B 
  will 
  at 
  once 
  show 
  that 
  the 
  actual 
  number 
  of 
  

   geneia 
  with 
  two 
  or 
  more 
  species 
  in 
  an 
  island 
  is 
  far 
  higher 
  in 
  Luzon, 
  

   Mindanao, 
  Palawan, 
  and 
  Samar 
  than 
  in 
  the 
  smaller 
  islands 
  of 
  the 
  

   archipelago, 
  but 
  it 
  will 
  be 
  objected 
  that 
  more 
  species 
  of 
  all 
  kinds, 
  includ- 
  

   ing 
  those 
  distributed 
  according 
  to 
  Steere's 
  law, 
  are 
  known 
  from 
  these 
  

   islands. 
  Manifestly, 
  then, 
  the 
  error 
  arising 
  from 
  the 
  fact 
  that 
  the 
  birds 
  

   of 
  some 
  islands 
  are 
  much 
  better 
  known 
  than 
  those 
  of 
  others 
  must 
  be 
  

   eliminated 
  as 
  far 
  as 
  i 
  lossible 
  if 
  we 
  are 
  to 
  arrive 
  at 
  any 
  conclusive 
  results. 
  

  

  I 
  have 
  first 
  compared 
  the 
  possible 
  with 
  tJie 
  actual 
  exceptions 
  to 
  

   Steere"s 
  law 
  in 
  each 
  island 
  on 
  the 
  basis 
  of 
  our 
  actual 
  knowledge. 
  

  

  If 
  a 
  genus 
  which 
  anywhere 
  in 
  the 
  archipelago 
  has 
  more 
  than 
  one 
  

   species 
  in 
  a 
  place 
  occurs 
  in 
  a 
  given 
  island, 
  we 
  have 
  the 
  possibility 
  of 
  its 
  

   being 
  represented 
  there 
  by 
  more 
  tlian 
  one 
  species. 
  If, 
  then, 
  we 
  take 
  all 
  

   the 
  genera 
  of 
  Table 
  B 
  which 
  occur 
  in 
  any 
  given 
  island, 
  compare 
  the 
  

   number 
  represented 
  by 
  single 
  species 
  with 
  that 
  having 
  two 
  or 
  more 
  

   species, 
  and 
  reduce 
  our 
  results 
  to 
  percentages, 
  we 
  shall 
  have 
  a 
  tolerably 
  

   satisfactory 
  basis 
  for 
  comparing 
  the 
  relative 
  tendencies 
  toward 
  differ- 
  

   entiation 
  of 
  genera 
  into 
  several 
  species 
  in 
  islands 
  of 
  different 
  size, 
  and 
  

   shall 
  have 
  eliminated 
  as 
  far 
  as 
  practicable 
  the 
  error 
  arising 
  from 
  the 
  

   incompleteness 
  of 
  our 
  knowledge 
  in 
  regard 
  to 
  many 
  of 
  the 
  islands, 
  for 
  

   in 
  each 
  case 
  the 
  comparison 
  is 
  between 
  the 
  total 
  genera 
  known 
  from 
  the 
  

   place 
  in 
  question 
  and 
  tlie 
  factors 
  which 
  go 
  to 
  make 
  u]) 
  that 
  total. 
  

  

  Tlie 
  percentages 
  of 
  possible 
  to 
  actual 
  exceptions 
  to 
  Steere's 
  law 
  

   obtained 
  by 
  this 
  method 
  are 
  given 
  as 
  one 
  of 
  the 
  footings 
  of 
  Table 
  B, 
  

   but 
  in 
  order 
  that 
  the 
  facts 
  brought 
  out 
  may 
  be 
  more 
  readily 
  grasped 
  

   I 
  have 
  embodied 
  them 
  in 
  a 
  curve, 
  which 
  I 
  shall 
  refer 
  to 
  as 
  Curve 
  I. 
  It 
  

   is 
  constructed 
  as 
  follows: 
  The 
  jjercentage 
  of 
  genera 
  represented 
  by 
  

   two 
  or 
  more 
  species 
  in 
  an 
  island 
  is 
  in 
  each 
  case 
  indicated 
  by 
  units 
  

   arranged 
  in 
  vertical 
  series, 
  1 
  unit 
  being 
  allowed 
  for 
  1 
  per 
  cent. 
  

   The 
  relative 
  areas 
  of 
  the 
  several 
  islands 
  are 
  shown 
  by 
  units 
  arranged 
  

   horizontally. 
  In 
  order 
  to 
  keep 
  the 
  curve 
  within 
  reasonable 
  limits, 
  and 
  

   still 
  make 
  i)lain 
  its 
  relationships, 
  I 
  have 
  found 
  it 
  necessary 
  to 
  vary 
  the 
  

   scale 
  used 
  in 
  indicating 
  the 
  areas 
  of 
  islands. 
  

  

  In 
  comparing 
  very 
  small 
  islands 
  like 
  Sibutu 
  and 
  Lapac 
  with 
  Minda- 
  

   nao 
  and 
  Luzon 
  it 
  is 
  obvious 
  that 
  the 
  first 
  part 
  of 
  the 
  curve 
  must 
  be 
  

   exi)anded 
  and 
  the 
  last 
  contracted 
  or 
  we 
  should 
  lose 
  the 
  relationships 
  

   at 
  the 
  beginning, 
  and 
  the 
  curve 
  would 
  stretch 
  out 
  at 
  its 
  end 
  to 
  in- 
  

   convenient 
  length. 
  Up 
  to 
  900 
  square 
  kilometers, 
  therefore, 
  I 
  have 
  

   made 
  1 
  unit 
  correspond 
  to 
  10 
  square 
  kilometers. 
  From 
  900 
  to 
  14,900, 
  

   1 
  unit 
  corresponds 
  to 
  100 
  square 
  kilometers, 
  while 
  from 
  14,900 
  to 
  

   114,900 
  I 
  have 
  allowed 
  1 
  unit 
  to 
  each 
  thousand 
  square 
  kilometers. 
  

  

  The 
  relative 
  sizes 
  of 
  the 
  islands 
  determine 
  their 
  positions 
  in 
  the 
  base 
  

   line, 
  while 
  a 
  dot 
  at 
  the 
  proper 
  height 
  over 
  each 
  shows 
  the 
  percentage 
  

   of 
  genera 
  with 
  two 
  or 
  more 
  species 
  found 
  in 
  it. 
  If 
  the 
  dots 
  thus 
  

   Proc. 
  X. 
  M. 
  vol. 
  XX 
  39 
  

  

  