234 NEW Y0RK STATE MUSEUM 



In the caudal vertebrae of Spatula I find the diapophyses to be 

 wide and spreading, while beneath, the ventral apophyses are an- 

 chylosed to the centra upon which they occur and hook forward 

 over the preceding vertebral body. The pygostyle in these and most 

 forms of the group is somewhat elongated, of an irregular quadri- 

 lateral outline, with thickened posterior border. 



Clangula has very wide and spreading transverse processes to 

 its caudal vertebrae, and the chevron bones upon the last two are 

 free and rest mainly upon the intervertebral cartilage. 



In H a r e 1 d a h y e m a 1 i s we find very long haemal spines 

 upon all the dorsal vertebrae, the longest ones being in the middle 

 of the series and they shorten somewhat in either direction, es- 

 pecially as we pass anteriorly, where they become strongly bifid. The 

 neural spines of the dorsum are long and low, while the transverse 

 processes are broad. This duck has seven free caudal vertebrae, the 

 leading one being firmly grasped by the ilia. The pygostyle, mak- 

 ing the eighth segment in the tail here, is much compressed trans- 

 versely, and about twice as long as it is high. Heralda has but 

 one pair of free cervical ribs, and these are lengthened, with well 

 developed uncinate appendages. There is not a little in the form 

 of its dorsal and pelvic ribs to remind us of what we find in some 

 of the auks; they being of slender proportions, with long narrow 

 processes, and gradually lengthen as we pass in the direction of 

 the pelvis. This is well seen in the sternal ribs, the last few pairs 

 of these being greatly elongated, very slender, and sweeping. There 

 are five pairs of true dorsal ribs, and three pairs of pelvic ribs. Two 

 pairs of the latter have their costal ribs reach the sternum, while 

 those of the last pair, though very long, do not. Only the an- 

 terior pair of pelvic ribs supports uncinate processes, and these are 

 very much aborted. 



I have examined the vertebrae and ribs in the genera Anas, 

 Spatula, Dafila, Aix, Netta, Marila, Clangula, Charitonetta, 

 Harelda (just given above), Polysticta, Somateria, Oidemia, and 

 others, and am satisfied that they vary to such an extent that the 

 data obtained therefrom can not be used with quite as much effect 

 in the matter of taxonomy in this group, as can other parts of the 

 skeleton. 



For example, in Netta rufina the first 16 vertebrae of the 

 column bear no free ribs, a well developed pair of these are borne 

 by the 17th vertebra, and they support unciform processes. The 

 true dorsal vertebrae are the 18th to the 22d inclusive, and there 



