102 NATURAL HISTORY OF KERGUELEN ISLAND. 



side to side, maldng a semblance to a series of prominent lamellar byp- 

 apophyses along this part of the column. The transverse processes also 

 of all except tlie first dorsal are very thin and broad, projecting as a 

 series of flat tiles above the heads and tubercles of the ribs. The last 

 dorsal lies between the crests of the ilia, beneath and abutting against 

 wh''ch the rib which it bears comes out. The sacro-lumhar vertebrae, i. 

 e., those which are anch^'loseJ with each other and which articulate 

 with the pelvis, appear to be thirteen in number. Viewed from, above, 

 the broad, expanded portion of the sacrum (opposite the acetabula) 

 shows six inter-trabecular spaces, the contour of the exposed surface 

 being approximately diamond-shaped, about f inch broad at the widest 

 part by 1^ inches in length. The median line above is flat, without 

 indication of spinous prbcesses. Viewed from below, the conjoined cen- 

 tra of the sacro-hanhar xeTtehriE are a narrowly fusiform mass, broadest 

 about opposite the middle of the ilia. Inferiorly They are flattened and 

 somewhat excavated, though anteriorly pinched together and deepen- 

 ing to join the articulation with the last dorsal vertebne. The trabeculae 

 are longest and most distinct opposite the acetabula, two of them being 

 especially prominent, while anteriorly four or tive are conspicnous. 

 Then follows an interspace of about the same length, in which they 

 nearly disappear 5 nor are they strongly marked toward the posterior 

 extremity of the column. 



The caudal, i. e., unanchylosed post-sacral vertebraB, are nine in num- 

 ber, considering the pygostyle as one. Pygostyle is simply laminar, 

 with thickened under edge, irregularly quadrilateral in shape; long 

 diameter, J inch. Of the other vertebrae, the transverse processes of 

 the intermediate ones are shorter than those of either extremity. Mod- 

 erate neural spines, with no obvious hypapophyses except on penulti- 

 mate vertebra. The whole series presents no special characters.* 



'Comparison with De Llaiuville's vertebral formula T^'ill show several points to be 

 considered. 



In the first place, De Blaiuville is in expressed doubt as to the number of post-sacral 

 vertebrae, and his fornmlpe, as given at p. 102 and at p. 106, differ with each other, the 

 first being Ifj— 6— 14— 7=4'2, the other being 15—6—14—8=4:3. Accounting for thisdis- 

 crepaucy on the supposition of imperfection of his specimen, we throw the post-sacrals 

 out of further consideration, and turn attention to the remaining elements of his form- 

 ula, which are really less different from ours than appears at first sight, we giving 13 

 — 8—13, and he 15—6—14. 



For we reckon the last costiferous vertebra as dorsal, he as sacral. This leaves the 

 numeration of nou-costiferous auchylosed lumbo-sacrals the same, namely, 15, in each 

 case, adding one to his numeration of dorsals. Wo furthermore reckon as a dorsal 



