68 NEW YORK STATE MUSEUM 



inferior one presents a graceful convex curve. Posteriorly it 

 terminates at the apex of a triangular smooth surface, the outer 

 basal angles of which are opposite the foramina in the xiphoidal 

 extremity [fig. 26]. The line of union between keel and body is 

 rounded, being concave outward. 



Marked differences occur in the manubrium of the Falconidae; 

 here in Circus it is a stumpy process, generally inclined upward, 

 having a sharp median edge below and a triangular anterior surface. 

 Among the Falcons (Falco r. columbarius), it is a narrow 

 spicula of bone, directed forward and upward ; but what is most 

 singular, there exists in these birds a second process that springs in 

 the median line from the border of the body above. These two pro- 

 cesses have the coracoidal grooves between them. 



The grooves for the coracoids decussate in Circus, their inner 

 ends terminating in points ; they decussate still more in Falco, 

 where their inner ends are rounded. Such a decussation of the 

 coracoidal beds is likewise to be seen in the herons, as in the genus 

 Ardea. 



In the specimens of all the Falconidae before me, it is the right 

 coracoidal groove that is the anterior one, and overlaps the superior 

 surface of the base of the manubrium. As well as I can remember 

 such is also the case with the herons. 



Returning now to the spinal column, we find that the 20th verte- 

 bra of Circus becomes anchylosed beneath the ilia. Its broad neural 

 spine has fused into one piece in common with the others that ex- 

 tend back as far as the sacrum; its diapophyses are half covered 

 by the anterior iliac borders, and these with the next vertebra be- 

 hind show the facets for the two pair of ribs already described 

 above, which are here overarched by the ilia [fig. 25]. 



The anterior aspect of the 20th vertebra presents all the require- 

 ments for articulation with the one next beyond, in its prezygapo- 

 physes, and in its centrum. Metapophysial spines, however, are 

 only thrown back by the segment before it, while the locking of the 

 neural spines does not take place. 



This description of the 20th vertebra brings us to a point where 

 we must needs take into consideration the pelvis of Circus [fig. 25, 

 28]. Upon superior view of this bone [fig. 28] we observe that the 

 neural spine or rather its upper surface projects forward as a broad 

 process between the ilia, and is roundly notched anteriorly. The 

 common top of this neural spine for nearly the entire length of the 

 pelvis is smooth and presents little or nothing to indicate where the 



