1889.] CONVOLUTIONS IN BIRDS. 311 



back, and especially the terminal end of the second resembles some- 

 what a plagioccelous formation. The Pteroclidse have consequently 

 variouspoints in common with several ratherdiversegroups; theyagree 

 with the Rallidce in iiaving the 2nd and 3rd loops left-handed, but 

 differ from them and from the Limicolse in having all the loops closed. 

 They differ from the Rallidpe and agree with the ColumbtTe and lower 

 LimicoliB in the number of loops. All this teuds to indicate that 

 the Pteroclidse have branched off from the common Gralline stock 

 before the separation of the latter into Limicolse and Rallidae had 

 taken effect, and before either typical Columbse or Gallinse were 

 developed. That they have, ia the diagram, to be placed in the 

 isocoelous circle — which really belongs to totally different birds — 

 shows also that they have made an early and special departure. 



The Alcidce are periccelous and strictly orihoccelous ; they agree 

 with the Laro-Limicolse in the configuration of their first three loops, 

 but they differ from them in the number of loops, which is at least 

 six, the last three of which are left-handed. The Alcidse seem to have 

 started from some low Limicoline forms and to have branched off 

 early into a strictly orthocoelous direction. They are, in this respect, 

 further removed from the Laridse (least so from the Terns), and 

 render the term Gavice somewhat vague. They approach the P<j(/o- 

 podes (Colymbida; and Podicipitidte). 



The Colymbidee show unmistakable affinities with what may be 

 called generahzed or low Gralline forms ; their five loops are closed, 

 orthoccelous, and alternating. The Podicipitidcie differ somewhat 

 from the Colymbidse, and besides possessing some sfjecial pecu- 

 liarities, approach the Grallte more closely than do the Colymbidse ; 

 at the same time in the possession of a pyloric dilatation they have a 

 feature in common with certain Fnlicarise and with the Steganopodes 

 and Herodii. A peculiar resemblance also exists between Podiceps 

 and Podica in the widely open and irregularly shaped last intestinal 

 loop. All this assigns a lower position to the Podicipitidse than to 

 the ColymbidEe, and gives them unequal rank, although the validity 

 of the name Pygopodes can be maintained. They connect the large 

 Gralline group with the following congregation, of which the Herodii, 

 Steganopodes, Tubinares, and Spheniscidae are all divergent types. 

 A very close coiuiection exists between the Herodii and the Steyano- 

 podes, and this is supported by numerous other characters. The 

 Tubinares are in more than one respect the most specialized outcome 

 of this great collective order, and reach in the typically mesogyrous 

 Procellariince their highest development ; whilst Pvffinus and 

 Dicmedea are more generalized, and Ossifraga takes up a somewhat 

 intermediate position. There are, in this respect, striking resem- 

 blances, of uncertain value however, with the Laro-Limicolse ; and 

 thus we arrive at the same conclusion as Fuerbringer, who assigns to 

 the Tubinares a position somewhat intermediate between but rather 

 distant from the Laro-Limicolse and Steganopodes. 



The Spheniscidce are now a very specialized group. They possess 

 undeniable characters in common with the Pygopodes, Stegano- 

 podes, and Tubinares ; they are on the whole orthoccelous, but the 



21* 



