GILL-FILAMENTS IN SAUROPSIDA 217 



sharply defined. Apparently the limits vary considerably in dif- 

 ferent embryos. This variability may be said to characterize 

 the whole development of the ridges, — a marked contrast to the 

 regularity with which the filaments develop. As each zone shifts 

 its position medially, scattered pycnotic nuclei appear between 

 the epitrichial and basal layers of the epithelium and diffuse 

 tufts of cells arise on its surface. The crests of these evagina- 

 tions form a low ridge which becomes the more conspicuous the 

 nearer the ridge approaches the mid-hne. The filaments, on the 

 other hand, grow out from a narrow strip of epithelium situated 

 on the lateral margin of each operculum, a germinative zone 

 w^hich represents the fusion of the posterior wall of the hyoid arch 

 with the ectoderm of the third and fourth gill arches. The fila- 

 ments thus arise from a specific epithelium, — a portion of the 

 branchial membrane, the whole of which is characterized at an 

 early stage by the presence of degeneration vesicles. The fila- 

 ments grow out of the region where the cysts occur in greatest 

 numbers and are themselves honeycombed with vesicles. Fur- 

 thermore, they do not difTerentiate into diffuse clusters of cells 

 but into a row of more or less distinct evaginations. Again, the 

 iife cycle' of the filaments is staged from two tb three days 

 earlier than that of the ridges. The branchial evaginations first 

 arise in situ and only later become involved in the movements of 

 the ventral body wall, whereas the pectoral ridges are in process 

 of migration when they first appear. Thus one is lead to con- 

 sider w^hether the ridges are not more intimately connected w4th 

 the movements of the body wall than are the filaments, if indeed 

 they are not products of that movement. For as the ridges ap- 

 proximate each other they become heaped up in the midline to 

 form a single structure which persists three days after the fila- 

 ments have disappeared, whereas the latter never meet in the 

 mid-line but maintain their identity as paired rows of individual 

 filaments to the end. It may be that the pectoral ridges repre- 

 sent the survival of some similarly placed outgrowths in a lower 

 vertebrate, but in no other animals, so far as I am aware, do 

 any such structures exist. For the present, then, it seems more 

 reasonable to define the ridges as local manifestations of migra- 



