( I'S ) 



All iiuiiortaul and stem lessou, ta\igUt by this tiiglitles.s Cuniini-aiit, a first-rati' 

 swimmer, is of course its analogy with Ilesj/eroriiis, whicli, in spite nf all tiiat lias 

 been said aljout its striictnre and affinities by Fuerbrin<fei' and myself, occasinnally 

 still tignres as a member of the Itutitac. He has no keel, therefore he is a liatite. 

 Fiat justitia, pereat common sense I 



The main feature of the pelvis and of the hindlimbs is that they are relatively 

 stronger and longer than those of Vh. carlo and carboides. So far as mere length 

 is concerned, the jielvis of several well-flying ("ormorants is fully eijual to that of 

 Fli. Iiarrixi ; but the impressions left by the origin of the ilio-trochanterie muscles 

 are stronger than in any other skeleton I have examined. 



It was reasonable to expect that the reduction of the forelimbs would be 

 comi)ensated by the greater develo])meut of the hindlimbs. This is actually the 

 case. Ph. harrisi has actually and relatively by far the largest tibia, decidedly 

 much ■ longer in comparison with his congeners <:arbo and carboiiks. Just as in the 

 arm the ulna has undergone the greatest reduction, so it is the middle segment 

 of the hindlimb, the tibia, which shows a surplus of compensating development. 

 Every one iif the bones is very powerful, with extra strong muscular crests. 



The proportions of the femur to the tarsus differ in the various sjjecies. The 

 femur is longer than the tarsus in Fk. harrisi, melaiwleucas, bicristatus ; ei|ual in 

 liocae-lwlliiiifliae ; shorter in c.arbo, carboides, carius, rristatm. 



Many of the calculations in this i)aper suffer from the want of a reliable 

 standard unit of length. I take the present opportunity of pointing out a fallacy. 

 If we take the length of the trunk as our unit (from the first cervico-dorsal vertebra 

 to the last vertebra which is fused with the postsaerals and with the ilium), the 

 percentages of the total length of the hindlimbs (from acetabulum to the tip of the 

 fourth toe), total length of wing, of tibia, and of ilium are as shown on Table B. 



Assuming the trunk as unit of lUD (cf Table B), it follows that Ph. carboides 

 has a short wing, the shortest leg, and the shortest tibia ; Ph. carbo has a shortish 

 wing, short legs, and short tibia ; Ph. harrisi has the shortest wing and a middling 

 long leg with a long tibia. Further, it follows that the longest tibia is associated 

 with long and longest legs and with middling wings in Ph. bicristatus and varim ; 

 that the longest wing is associated with a long tibia in Ph. mdanoleuciis, and that 

 a decidedly long wing goes with a short tibia in I'h. HOiMe-hollandiae. 



These are obviously contradictory results. It follows that the length of the 

 trunk as a unit of lUU is inapplicable. But if we express the length of the tibia, 

 tirst in per cent, of that of the whole hindlimb, and secondly in per cent, of the sum 

 of femur and tibia and tarso-metatarsus (cf Table D), then we see that the results 

 of the two lists agree with each other, whilst they disagree with the conclusions 

 based upon the trunk-length as unit. 



The Vkktebkal CjOLUMn.— In normal Cormorants the lUth vertebra is the 

 tirst cervico-dorsal, i.e. carries a typical, although non-sternal, rib. The 21st to -i.sth 

 carry complete sternal ribs. The 29th carries the last pre-acetabular transverse 

 buttress. The 31st and :J2nd arc the two primary sacral vertebrae, and both have 

 the sacral rib elements equally developed. But here occurs, considerable individual 

 variation. Either the anterior vertebra (observed by me in <iracali.(s, carbo, novac- 

 liolliindiae), or the posterior vertebra (t'.y. in albiceiitris, bra.'iiliensis), may lose its 

 rib element, so that there remains only one typically developed primary sacral 

 vertebra, either the 32nd or the 31st. The next following seven vertebrae are 

 always fused into the sacrum. Very often, however, eight vertebrae are thus fused, 



