40 THE SALAMANDER 



passes dorsally and the other ventrally across the apex of the triangle. 

 Which side is ventral and which dorsal depends on the relation of 

 the coracoids to one another, since the grooves in the sternum occupy 

 the corresponding relative positions. The sternum is attached to 

 the coracoids by a membrane, memhrana sterno-coracoidea (Engler), 

 which arises from the linea alha^ and is also closely attached to the 

 pericardium dorsally, its dorsal surface being slightly concave so as 

 to fit this structure. Although the exact shape is subject to consider- 

 able fluctuation, the present writer has never seen a sternum with the 

 long anterior process such as is figured by Perrin, and it would seem 

 that his figure must have been prepared from a very unusual speci- 

 men. The sternum serves for the attachment of the pectoralis muscle 

 ventrally, while superficial rectus muscles of the abdomen and neck 

 are attached to its edges, dorsal to the coracoids. A very good and 

 detailed account of this structure and of its development is given by 

 Engler. 



As may be judged from the above synonyms, the several parts of 

 the pectoral girdle have been homologized with widely different 

 structures. It is perhaps unnecessary to review the somewhat 

 involved discussions here, since the identities of the elements as 

 expressed above are now generally accepted. The most noteworthy 

 papers are those of Gegenbaur (1865), and Parker (1868), while 

 Engler (1929) has recently worked out the ontogenetic history in 

 considerable detail, particularly with regard to the sternum. 



2. The Fore-limb (figs. 23—8 and 39—42). 



This comprises the usual elements found in a tetrapod fore-limb, 

 namely, the humerus in the arm, the radius and ulna in the fore-arm, 

 carpals and metacarpals in the wrist, ?in6. phalanges in the digits. 



The chief points of interest are: (i) the peculiar shape of the 

 humerus, and particularly of its proximal end, (ii) the morphology 

 of the carpus and the homology of the digits. Each of these features 

 will be dealt with in the appropriate place. The epiphyses of all the 

 long bones, and all articular surfaces, are cartilaginous. 



{a) The Arm (Figs. 23, 24, and 42). The long axis of the humerus 

 is almost straight. The head is expanded dorso-ventrally in a peculiar 

 manner and merges into the crista ventralis humeri (cr.v.h.) The 

 cartilaginous epiphysis at this end when seen in side view is therefore 

 shaped like a comma (fig. 42). The crista ventralis is short, occupy 

 ing only about a third the total length of the bone, but it is of con- 

 siderable expansion and rises steeply from the shaft. It fits into the 

 notch, already referred to, formed in the rim of the glenoid cavity 



