I ^1 Snodgrass, Food and Size of Bill in Geosfiza. 377 



James and one from Barrington. Next to these, seeds Nos. lo 

 and 57 were most abundant, each being represented in ten 

 stomachs. These four seeds were also by far the most abundant 

 wherever they were found. The rest of the diet consisted of seed 

 No. 2, found in two Seymour birds and in one Duncan bird ; No. 

 II, found in one Seymour bird; No. 17, found in one Seymour 

 bird; No. 18, found in one James bird; No. 20 found in one 

 Duncan bird ; and Nos. 41 and 59, found in one James bird. 

 These specimens were all collected in April and May. 



The facts just detailed certainly show that the individuals of 

 Geospiza fuliginosa parvula living on Narborough and at Tagus 

 Cove, Albemarle, during December and January, have a different 

 diet from those individuals living at Iguana Cove, Albemarle, in 

 December, and on James, Seymour, Duncan and Barrington in 

 April and May. The proof, from these facts, of proposition I, 

 however, is somewhat invalidated by the consideration that seeds 

 Nos. 14 and 15 may ripen at Tagus Cove and on Narborough 

 later than January. But seeds Nos, 8 and 22 were not found in 

 any stomachs except in those of birds taken at Tagus Cove. We 

 can see, at least, that the diet of the birds depends on the local 

 prevalence of certain seeds ; and that, where the floras differ, the 

 food of a species may differ. 



III. Different specks at the same locality may feed on the same 

 kinds of seeds. 



This proposition is much easier to prove than the last. Com- 

 pare, for example, the food of Geospiza fuliginosa parvula (PI. XI, 

 Fig. 6) and of G. scandens fatigata on Seymour and Barrington 

 Islands. As has already been shown, the food of the former 

 species consists almost wholly of seeds Nos. 10, 14, 15 and 57, 

 Nos. 14 and 15 being in the majority. An examination of the 

 table will show that the food of G. scandens fatigata on the two 

 islands is practically identical with that of G. fuliginosa parvula, 

 consisting mainly of seeds Nos. 14 and 15, with Nos. 10 and 57 

 second in numbers. 



The case of these two species, then, proves that species dif- 

 fering much in the size and shape of the bill (PI. XI, Figs, i and 

 6) may have absolutely the same food habits. We have not, 

 however, the material at hand to justify the statement of this as a 



