1899.] OSTEOLOGY OF THE TUBIXARES. 401 



In Majaqueus and Diomedea exulans, for instance, the 2nd phalanx 

 of digit'll. exceeds that of the 1st. Phalanx i. of digit I, is equal 

 in length to that of phalanx i. digit II. In Diomedea again it is 

 rather less instead of equal. In the smaller Petrels the elongation 

 of these phalanges is not so marked. 



ix. The Pelvic Limb, 



The bones of the pelvic limb are non-pneumatic ; the tibio-tarsus 

 is characterized by an enormous flabelliform ectocnemial crest 

 which rises high above the articular surface for the femur : is 

 markedly inflected at its distal end, and provided with an ossified 

 extensor bridge. The fibula does not extend more than | the way 

 down the leg, and is much reduced in thickness distally. The tarso- 

 metatarsus has a well-marked intercondylar tubercle. The 

 hypo-tarsus is complex in the Procellariidae and simple in the 

 Diomedeidie. The outer and middle toes are of equal length. 



In one skeleton of Diomedea exidans I find an ossified tarso- 

 metatarsal extensor bridge on the right foot. The hallux is repre- 

 sented by a metatarsal and an ungual phalanx, the latter often of 

 considerable size. In Pdecanoides it is absent. The femur, as a 

 rule, is about as long as, or less than, the tarso-metatarsus, and is 

 about half as long as the tibio-tarsus ; in Oceaiiites, Pelagodroma, 

 Ci/modroma, and ProceUaria the femur shortens conspicuously, these 

 measurements being about i as long as the tarso-metatarsus and 

 5 as long as the tibio-tarsus. 



X. Eesults. 



Briefly, I think, the outcome of this paper has been to confirm, in 

 a large measure, the conclusious of Porbes as set forth in his most 

 valuable Report on the Petrels collected during the 'Challenger' 

 Expedition (5). The appended diagram (fig. 2, p. 402) is a modi- 

 fication of that published by him in that work. He divided this 

 suborder into two families — Procellabiii).?!; and Oceanitib,?: ; 

 and two subfamilies — Peocellaeiii^^ and DiomedeijSt.e. Pek- 

 canoides he regarded as an abei"rant genus of the first mentioned 

 subfamily. 



I propose to make two Pamilies — the Peocellaeiii)^ and the -4. 

 D10MBDEID.E ; the former being further divided into two sub- 

 families — Procellaeiest^ and Pelecanoidin.e, Thus Porbes's — j- 

 DioMEDEiN^ becomes raised to the rank of a family, his genus 

 Pelecanoides to the rank of a subfamily, whilst his family 

 OcEAJfiTiD^ becomes, in my scheme, reduced to a section of the 

 Peocellaeiin^. The sections in this subfamily are three in 

 number, and can quite conveniently be diagnosed from the 

 characters of the skull alone (see Keys, pp. 403-409). 



Pelecanoides forms the second subfamily. In the great width 

 of the basitemporal region of its skull it differs from every other 

 member of the suborder. The sternum and pectoral girdle are 



