338 NEW YORK STATE MUSEUM 



The Gastornithidae evidently possessed one, or at least one, very 

 remarkable character. In them the cranial sutures of the skull 

 remained permanent throughout life, whereas in all existing birds, 

 the Anatidae included, these become obliterated in adult life. 



REMARKS ON THE CLASSIFICATION OF THE NORTH AMER- 

 ICAN ANSERES 



So gradual is the merging or shading of one group of anserine 

 fowls of the family Anatidae into another that, by the use of os- 

 teological characters alone, it becomes almost impossible to accur- 

 ately define the limits of the minor divisions, such as the sub- 

 families. The skull can not be relied upon, for, as in the case 

 of Olor columbianus and Branta canadensis, there 

 are no distinguishing characters, apart from the question of size; 

 again, the sternum fails, for we have a fenestrated one in Mer- 

 gus serrator, and also in Clangula islandica, 

 while in most, if not all the others (Charitonetta?), the hinder 

 part of the bone is two-notched. The form of the osseous man- 

 dibles in the mergansers (Merginae) will at once distinguish them 

 from any duck, goose, or swan, and they are well differentiated as 

 a subfamily. 



* There are no strong characters in the skull, in so far as the ducks 

 and geese are concerned, by means of which we can draw with 

 certainty a subfamily line between these two groups. The skull 

 of Branta canadensis presents the same characters, apart 

 from the difference in size, and certain other proportions, as the 

 skull of a Mallard (Anas). Osteologically, I am inclined to be- 

 lieve that the vertebral column offers the best character here, true 

 ducks never having more than 44 vertebrae in their spine (to which 

 the pygostyle is to be added), while true geese never have less 

 than 47 (to which the pygostyle is to be added). This would throw 

 the genus Dendrocygna with the ducks, where it undoubtedly be- 

 longs, and the subfamilies Anatinae and Anserinae can thus be 

 distinguished. Again, probably no true goose has more than 49 

 vertebrae in its spinal column (to which the pygostyle is to be 

 added), while a true swan never possesses less than 52 (to which 

 the pygostyle is to be added), and doubtless in this way the sub- 

 family Cygninae can be osteologically differentiated. 



Of course a number of anything constitutes no valid character, 

 and it is only used here as an indicatory guide to such other dis- 

 tinguishing characters as may be found, as, for example, the pe- 

 culiarities of the furculum, sternum and trachea in swans, and 

 modifications of the latter in certain ducks. 



