— 319 — 
a notch on the inner coracoid margin, and undoubtedly, as shown 
in the young Stereosternum, the coracoid developes from a single 
centre of ossification. From the relation of clavicles to scapulae, and 
width of chest, as indicated by the anterior thoracic ribs, I believe the 
inner borders of the coracoid were probably nearly apposed, or possibly 
slightly separated by narrow epicoracoidal rims of cartilage. The over- 
lapping of the coracoids in Gervais, specimen of the South African Me- 
sosaurus (1867-69, Pl. XLII), I believe to be due to post-mortem displa- 
cement. 
The scapular portion of the complex is low, broad, and fan-shaped, 
its form suggesting that of the Mesosaurus, Platecarpus, and, to a 
less degree, that of the ichthyosaurian genus Mixosaurus, is proba- 
bly an adaptation to aquatic habit. Plate Il., fig. 2, and Plate III., 
fig. 4, show very clearly the inner face of the scapula. 
The interclavicle is a slender rod, expanded anteriorly for articula- 
tion with theclavicles. The largest specimen (Pl. III., fig. 5) measures 
36 mm. in length and 9 mm. in witdth in the expanded portion, 
while the narrow median part is only 2 mm. wide. The anterior 
portion shows a narrow median groove. The interclavicle is almost 
identical with that of Stereosterum as figured by Osborn (1903), and 
resembles closely that of many diapsid types. The clavicles are some 
4 mm. in width at their inner ends, where they partly overlap ven- 
trally the anterior expansion of the interclavicle and approach each 
other within some 2mm. Toward the scapular end the clavicle tapers 
toa point. In the specimen shown in Plate III, fig. 3, the right clavicle 
is 16 mm. long and barely reaches the border of the scapula. It is 
not unlikely that one or two millimetres of the outer end have 
been lost. 
In Gervais, figure (1867-69), Pl. LXII.) the coracoids are shown 
overlapping, due I believe to displacement, butif the figure given by 
Seeley (1892, Pl. XVIII.) is correctly drawn there is no escape from 
theconclusion that the two scapulo-coracoids have been displaced, so 
that the scapulae lie ventral to the vertebral column, and the inner bor- 
ders of the coracoids are directed outwards. The apparent glenoid cavi- 
ties are probably artefacts due to crushing of coracoids over the humeri. 
Tadmit that the symmetry of the two sides in Seeley’s plate seems op- 
posed to such an interpretation, but comparison with Gervais’ plate and 
with the very complete material at hand of the two Brasilian types, 
leaves no doubt that Seely in his restoration (1902, p. 601) has re- 
