INTRODUCTION 7 



tinguished by their long necks, which actually have more vertebrae than the necks 

 of geese, and by other characters of less importance. Some swans show most re- 

 markable convolutions of the trachea. The Semipalmated Goose, Anseranas semi- 

 palmata, is usually placed next to the swans in a subfamily of its own (Anseranatinae). 



The third subfamily consists of the Spur-wings, Plectropterince, a very heterogene- 

 ous group, which is usually placed next to the swans. Its members are included in 

 this work because it contains various species, such as the Muscovy, the Comb Ducks 

 and the Goose Teal (Nettapus), which ordinarily come under the category of ducks. 

 The constitution of this group is open to much speculation. There is no doubt that 

 some of its genera, as arranged by Salvadori, should be removed and placed nearer 

 the true ducks. I have considered it as composed of the same genera as Salvadori in 

 his Catalogue of Birds, except that I have taken out the Mandarin and Carolina 

 Ducks and placed them at the end of the true ducks, as they now appear in the 

 Check-List of the American Ornithologists' Union, 



The Spur-wing subfamily are birds with rather long tails, the feathers of which 

 are broad and rounded at the tips. The upper parts are mostly glossy and the hind 

 toe is long. These ducks inhabit the tropics of both the Old and the New Worlds 

 and have arboreal habits. 



The subfamily Cereopsince, represented by only one living species, the Cape Bar- 

 ren Goose of Australia, also includes the great extinct New Zealand Goose, Cnemi- 

 ornis calcitrans, mentioned above. 



We may place the true geese, Anserinae, next in order, and among them should 

 perhaps be included the Swan Goose (Coscoroba), although its exact position is still 

 doubtful, and it may be nothing more than a huge duck related to the Sheldrakes, 

 and possibly to the Tree Ducks. These Anserinae differ from the swans in having 

 the neck always shorter than the body, and in having the front of the tarsus covered 

 with small hexagonal, instead of narrow, scales. They differ from the last subfamily 

 by the absence of a cere. 



The Chenonettinoe are another subfamily typified by the brant geese of the pampas 

 of South America. This subfamily has not been considered among the ducks and is 

 probably nearer to the Brant than to the true ducks. It has received various posi- 

 tions in schemes of classification. 



The Tree Ducks, Dendrocygnince, are often placed among the true ducks, but I 

 believe they well deserve to rank as a subfamily by themselves, a position which 

 Shufeldt assigned to them in his anatomy of the group. They are denned by that au- 

 thority as arboricole ducks, with very long legs, the tibio-tarsi more or less denuded, 

 the middle toe considerably more than one third the length of the tarsus, and the 

 thecse of the leg and tarso-metatarsus reticulated. The orbits of the cranium are 

 completely surrounded by bone, and there are but seventeen cervical vertebrae. 

 Morphologically they are nearer to the ducks than to the swans and geese, and they 



