170 
surface (fig. 6, 2’), and then rapidly diminishes in thickness, curving inward towards its 
fellow, which it does not quite meet, above the neural canal (Pl. LXII. fig. 5,n). The 
anterior end of the centrum of the atlas occupies the notch (c) between the hypapophysis 
and neurapophysis, completes the occipital cup, and gives attachment to the ligament 
answering to the ‘odontoid’ in anthropotomy. The figures of the atlas in Pl. LXII. 
are of the natural size: the specimen was obtained with the incomplete skull from the 
fissure at ‘ Timaru.’ 
§ 5. Scapulo-coracoid Arch of Dinornis robustus. 
The existence of such arch in the skeleton of Dinornis was inferred, in the Memoir 
on Palapteryx (p. 124), from the articular depressions in the sternum (Pl. XXXY. 
fig. 2,c,c); and, by the peculiarly small size, shallowness, and shape of these depres- 
sions, I recognized the convex extremity of the bone (PI. LXIV. figs. 2, 3, 4, x), forming 
part of the skeleton of the Dinornis robustus from Manuherikia, as being the sternal end 
of the coracoid. It presents a rather irregular convexity, of an oval shape, 10 lines by 
6 lines in the two diameters, with a rough surface indicative of ligamentous union with 
the sternal fossa, not of articulation by a synovial joint, as in birds of flight. From 
the tuberosity («) the bone (52) rises straight, decreasing in thickness and increasing in 
breadth at its upper end, which is confluent with a much longer and thinner bone (51), 
forming with the coracoid a widely open angle, and slightly curved in its course. This 
bone I take.to be the ‘scapula’ confluent with the coracoid, partly from characters of 
proportion and shape and partly from the analogy of the scapulo-coracoid arch in the 
Apteryx'. In this bird the coracoid and scapula are confluent, and present relative pro- 
portions as to length like those in Dinornis. But the coracoid is relatively much broader 
in the Apterywx; its sternal end is adapted to a long groove, as in most other birds ; it 
also shows a perforation near its scapular end, and a more important difference in the 
presence of the glenoid cavity for the humerus on the posterior margin of the scapulo- 
coracoid confluence. ‘There is no trace of such articular cavity in the scapulo-coracoid 
arch of Dinornis, but in place thereof a rough, slightly produced ridge (fig. 4, r), to 
which, if any rudiment of humerus existed, it must have been suspended by ligament. 
I, however, infer that such appendage of the scapular arch did not exist in the living 
bird; that the Dinornis offers the previously unknown and unique exception to the 
tetrapodal type in Birds ; that the anterior members, like the posterior ones in Cetacea, 
were represented only by their supporting arch, and that this arch was limbless, as it 
is in Anguis among the Lacertian Reptiles. 
The scapula (PI. LXIV. figs. 2, 3, 4, 51) soon decreases in breadth, from 11 lines at 
the confluent part (m) to 7 lines within an inch from that part, beyond which it more 
gradually narrows to a breadth of 5 lines at the extremity ; the thickness of the bone 
gradually decreases also from the coracoid confluence (fig. 4, m), viz. from 4 lines to 
* Ante, p. 34, pl. 1x. fig. 4, e, f, g. 

