1899.] CRANIAL OSTEOLOGY OF THE PARUOTS, 45 



the mandibular fenestra differs in the two genera, being obsolete in 

 Ilyopsittacus, but large in Caica. 



In ComtrMs and Pyrrliura the orbit is sometimes complete, though 

 the suboi-bital bar when complete is slender ; when it is incom- 

 plete the postfrontal process is long, as long or nearly so as the 

 squamosal and very much longer than in any of the Old-World 

 Parrots. The squamosal process is curved below but indistinctly 

 Botched, the suprameatal process is extremely small, the anditory 

 meatus is uncommonly wide and its posterior border is very near 

 to the occipital ridge. The paroccipital processes are prominent 

 and more vertical than in Chrysotis. The nostrils are ot moderate 

 size, rounded, and somewhat distant, being intermediate in all 

 three respects between Chri/sotis and Ili/o^jsittacus. The inter- 

 orbital vacuity is considerably larger than in C'hrysotis or Piomis. 

 The mandibular fenestra is variable, sometimes large and sometimes 

 obsolete. 



There remain to be considered a small number of genera be- 

 tween which I find it harder to draw osteological distinctions or 

 to trace definite resemblances to those already considered. These 

 forms include Palreornis, Tam/gncttJius, Edectus, and Geoffroyus — in 

 other words, the remaining genera of the so-called Palceornithinm 

 after removing from that family Ar/apomis and the Australian 

 genera allied to Aprosmictus ; and also Pacliynus and Brotogerys, at 

 present grouped respectively with the Pionince and the Conurince, 

 Poeocephalus, the African genus grouped with the otherwise exclu- 

 sively South-American Pionince, and Coracopsis, associated by 

 Salvadori with Psittacus. The above genera have all an incomplete 

 orbit and an extremely small postfrontal process in the shape ot: a 

 more or less vertical ridge. The prefrontal process is long, and it 

 extends close to, though it is not united with, the tip of the long 

 straight squamosal process. The auditory meatus is comparatively 

 wide ; the area between it and the occipital ridge is broader than in 

 Conurus ov Clirysotis, but narrower than in Psittacus or the Lories, 

 and a fortiori much narrower than in Agapornis. There is no 

 notch below (except in Coracopsis) nor groove across the base of 

 the squamosal process. 



The skulls of Palceornis and Tanygnathus are extremely alike, 

 the only conspicuous difference being in the nostrils, which in 

 Tanygnatlius are smaller and wider apart. In both genera the 

 interorbital vacuity is extremely small and similar in character. 

 In Edectus the interorbital vacuity is moderately large, the squamosal 

 process is expanded towards its tip, the nostrils are small and oval 

 with the long axis of the oval vertical, and the descending occipital 

 ridge is faintly marked (whereas in the others it was extremely 

 strong) and curves forwards, giving a distinctly different outline 

 to the post-auditory area in front of it. In Geoffroyus, with no 

 very important differences perhaps from Edectus, there is less 

 identity of characters than we should expect to find from the very 

 close association in which it is customary to place the two genera. 



