April, 1919] Subspecies of Passerculus Rostratus 347 



doubtless on their breeding grounds. Mr. Luther J. Goldman 

 obtained typical specimens in very much worn breeding plumage 

 on May 16, 1915, on Montague Island, Lower California. 

 Their abraded condition and their presence here at such a late 

 date in the spring, together with Mr. Goldman's observations 

 on their habits given below are conclusive evidence that the 

 birds are here on their real nesting ground. There is no record 

 in any other localit}^ of the presence of individuals of any form 

 of the species later than a week beyond the middle of April 

 (and even this is very unusual), except on their breeding grounds. 

 Furthermore, the other races are known to breed regularly 

 in April and even as early as March. Mr. Goldman's notes 

 on the habits of the species as observed by him in the Colorado 

 River delta are given below as of interest in this connection : 



Abundant on Montague Island, and with the exception of turkey 

 vultures, the only land bird found. I found them first at the mouth 

 of the Colorado; they range from this point south along the coast as 

 far as my investigations took me, or to a point opposite the south end 

 of Montague Island. The birds were found living near the river only, 

 and frequented the long coarse salt-grass which the Cocopah Indians 

 call "Inpah." I observed them feeding on the seeds of this grass, 

 also working along the mud flats at low tide. 



This discovery of the breeding place of Passerculus rostratus 

 rostratus solves one of the most puzzling problems of North 

 American ornithology. There are several records of the breed- 

 ing of this bird in Lower California and in California, but Dr. 

 Grinnell is probably right in considering them all mistakes 

 of identification.* The same discovery also solves the migration 

 mystery of this species, fully as interesting a result. With the 

 data now at hand it is possible to work out the migration of the 

 large-billed sparrow with some degree of certainty. Part 

 of the individuals of the typical subspecies winter in its breeding 

 area, but the great majority begin, by the middle of August, to 

 leave their breeding ground, whence they move in various 

 directions, some of them southeast along the Sonoran coast, 

 others due south into Lower California, others west to the 

 northern part of Lower California, and still others northwest 

 to the coast of southern California, reaching the extremes of 

 their winter range almost simultaneously, and regularly becom- 

 ing common at Cape San Lucas and in Los Angeles County, 



* Auk, XXII, No. 1, January, 1905, p. 16. 



