144 The Food of Birds. 



on which table certain differences of food are clearly shown. 

 Now, if these differences were local and accidental, they 

 would undoubtedly tend to disappear when larger numbers 

 of specimens were examined ; but if they are specific and 

 constant, they should be made the more evident, on the 

 whole, the larger the number of specimens taken. The 

 table on page 147 presents data derived from three hun- 

 dred and fifteen specimens, covering considerably more 

 time and area than the table in the Transactions. If the 

 difference between the food records of the various species 

 are now greater than before, we may conclude that the 

 differences noted are real and not artificial. If they are 

 less, on the other hand, the whole question is still unsettled. 

 The differences apparent in the later table may be specific, 

 but there is no proof of it. In order to apply this crucial 

 test as fully as i)ossible, I have selected twelve food ele- 

 ments in which the differences were most apparent, and, 

 taking the species in pairs, have ascertained the sum of 

 the differences of the ratios of these elements for each pair 

 separately, first from the old table and then from the new. 

 In every case but one the sum of these differences has 

 been much larger by the new table than by the old, 

 thus proving conclusively that the species appear to di- 

 verge in food habits the more widely the greater the number 

 of specimens studied. For example, the differences of 

 the selected elements as shown in the original table of 

 seventy-eight robins and catbirds, amounted to sixty-four 

 per cent. ; and by the new table of one hundred and eighty- 

 four birds, to eighty-two per cent. A similar comparison 

 of the food of the catbird and hermit thrush gives one hun- 

 dred and twenty-five as the sum of the differences of the 

 old table of fifty-five birds, and one hundred and fifty-five 

 as the sum of the differences of the new table of ninety- 

 one birds. Taking the catbird and the brown thrush, we 

 have sixty-four and ninety-nine parts for the old and new 

 tables respectively, the first for sixty-five birds and the 

 second for one hundred and thirty-four ; while the brown 

 thrush and wood thrush give seventy-eight and eighty- 

 eight parts for thirty-nine and eighty-six birds respec- 

 tively, and the catbird and wood thrush give seventy parts 



