( I'O 



allliougU in tUi'SO cases the .^th postsacral (4(.)th of the whole series) is sometimes 

 still imperfectly assimilated. Then follow five to seven bona-Jule candal vertebrae, 

 and lastly, of course, the pygostyle. The tirst of the caudal vertebrae is often more 

 or less firmly connected, although not anchylnsed, with the posterior process of 

 the ilium. Obviously amalgamation of tliis vertebra with the sacrum reduces the 

 number of free caudal vertebrae by one. The pygostyle itself is not always 

 composed of the same number of vertebral elements, but there is no specific 

 ditl'erence between the species which have 12, or those which, like I'h. Iiiirrhi, 

 carboiiles, carlo, JilnmcHtosus, lucid tis, capensis, and gaimardi, have l-l rectrices. 

 The pygostyle begins with the 46tli or 47tli vertebra. 



Tiie three specimens of Ph. harri.'ti examined by me possess, like other 

 Cormorants, l<s cervical and 2 cervico-dorsal vertebrae, and the 2Uth forms the 

 last pre-acetabular buttress. But, instead of five, there arc only four comjiletc 

 sternal ribs, the fifth pair not reaching the sternum by about half an inch. In one 

 specimen the distal end of the right sternal rib still possesses the articulating 

 knob, although so far removed from the foot of the sternum. On the left side tiie 

 corresponding bone is tapering into a point as usual. This little abnormality, 

 seemingly quite without meaning and not worth mentioning, is after all most 

 suggestive. The same process of the conversion of sternal into floating or lumbar 

 ribs was still going on in this full-grown I'ormorant. The right fifth rib was 

 doing what had long been finished on the left side, and the fifth pair was in a 

 condition not yet dreamed of in other Cormorants. The next following rib, 

 belonging to the 26th vertebra, is rather remarkable, since its " sternal " or ventral 

 portion has quite severed its connection with the dorsal piece, and now lies loosely 

 and separately imbedded in the flesh, tucked on to the sternal portion of the fifth 

 pair. Such conditions are not very rare amongst other Cormorants. 



The two sacral vertebrae of Ph. harri.si seem to be the 32nd and ',i'.M, so that 

 this region of the whole sacrum contains one more vertebra than other Cormorants. 

 The following postsacral, caudal, and pygostyle vertebrae exhibit no more than, or 

 rather as much, individual variation in numbers and connections as other species. 

 The pygostyle begins with the 48th vertebra, in one specimen apparently with the 

 4'.)th, but it is unfortunately not possible to determine how many vertebrae liave 

 been fused into the pygostyle. It is possible that the additional vertebra contained 

 in the presacral complex of the sacrum, and thus lengthening the whole pelvic 

 region, accounts for the unsatisfactory results which we arrive at by comparing 

 the proportions of length of pelvis to the bones of the hindlimbs of Pk. harrisi 

 with other Cormorants. 



Table B. 



