Vol. XIXl Sy!ODGRASS, Food a/id Si'ze of Bi'll in Geospiza. 38 I 



1902 J 



29. G./ortis (PL XI, Fig. 3) eats such moderately large seeds 

 as Nos. 35, 40, and 48 together with larger ones such as Nos. 28 

 and 57. An examination of the table will show, however, that 

 the larger-billed species by no means confine themselves to large 

 seeds. It appears most probable that they eat the larger seeds 

 simply because their large bills makes it possible for them to eat 

 a greater variety of seeds. There is no evidence that they show 

 a special preference for large seeds. 



The foregoing six propositions are about the only conclusions 

 that we can deduce from a study of the material in hand. It is 

 evident that if these propositions were stated severally as general 

 facts they would be mutually conflicting. Each is true only m 



some cases. 



If it be assumed that the various sizes and shapes of bills 

 amongst the Geos/>ms have been developed as adaptations to 

 differences in food habit, then it must be shown that the different 

 species of the genus feed on different species of seeds. This 

 cannot be done. We can prove definitely that some species 

 with very different bills feed on exactly the same kinds of seeds. 

 On the other hand some of the evidence seems to indicate that 

 some of the species and subspecies do have different food habits. 

 We cannot say, however, that these differences of diet are not 

 forced upon different species as a result of their living in differ- 

 ent localities. Especially is this probable since, in some cases, 

 we find that individuals of the same species living at different 

 localities feed on different seeds. This is due evidently to flora 

 differences between the two regions. 



The evidence, then, seems to be in favor of the general con- 

 clusion that t/iere is no correlation between the food and the size and 

 shape of the bill If this is true, then we must look elsewhere for 

 an explanation of the variation of the Geospiza bill. 



