1906.] OF THK TRACHEOPIJONTE PASSERES. 145 



form a series of separate upstandizig plates, or may, on the other 

 hand, interlock one with another by means of a bifurcating process 

 from the anterior and posterior angles of the doi'sal border. In a 

 skeleton of Siptornis sp. inc. three of these vertebrte (1-3) have 

 become anchylosed, but whether this is an individu.al peculiarity 

 or is a feature peculiar to the genus I am at present unable to say. 



The last thoracic, in all the species of the groups here dealt with, 

 has fused with the synsacrum, and in the Dendrocolaptinje two 

 thoracics appear to be generally fused therewith, each of which 

 bears long ribs (see p. 146). Hypapophyses in the Pittidse are 

 wanting ; in no case is there more than two of these processes. 

 In the Hylactinte and Dendrocolaptinfe they are fairly well 

 developed, but never so long as those of the two cervico-thoracic 

 vertebrae immediately preceding. In the Pipridse and Philepittidte 

 they are degenerate. Philepitta, by the way, is remarkable for 

 the small size of the centra. 



Twelve vertebrje enter into the composition of the synsacrum, 

 but the series from which these are drawn is not always the same, 

 as may be seen by the following tables: — 



Gonopophagida?, 

 Fitta. Pipi-idas. e, g-,, Hylactes. Philepittidse. 



Thoracic 1111 



Lumbar 2 2 '2 .3 



Lumbo-sacral ... 3 3 3 3 



Sacral 2 [caudal. 2 2 2 



Caudal 44-7 free 3 + 8 3 + 8 3 + 8 



12 11 11 12 



Total 19. Totall9. Total 19, Total 20, 



Formicariinte. Deiidrocolaptiiias. HynallaxiiiiB. 



Thoracic 1 2 1 



Lumbar 3 2 3 



Lumbo-sacral ... 2 2 2 



Sacral 2 2 2 



Caudal 4 + 8 4 + 7 4 + 8 



12 12 12 



Total 20. Total 19. Total 20. 



The analyses here given ai'e undoubtedly worth publishing, if 

 only as a basis for further work ; but before they can be of 

 any great value a comparison of a much larger series of skeletons 

 than I have been able to make is necessary. The ventri-lateral 

 processes of the second lumbar are in all cases well developed, 

 but those of the sacrals are more feeble, so that some difficulty is 

 experienced in distinguishing, superj&cially, between sacral and 

 postsacral vertebrfe. The dorsi-lateral processes of the saoral 

 and postsacrals are long, keeping the innominates wide apart, 



Piioc. ZooL. Soc— 1906, YoL. I. No. X. 10 



