George L. Streeter 103 



fusion of the latter with au overlying patch of thickened epidermis, and 

 an apparent absence of such a fusion in case of the ganglia of the roots. 



The relation existing between the ganglia of the seventh, ninth, and 

 tenth nerves and the overlying epidermis has-been described in mammals 

 by Froriep, 85, and it is regarded by him as the anlage of rudiments of 

 the phylogenetically lost branchial sense organs of Beard, 85, and 

 van Wijhe, 82. In elasmobranchs Froriep, 91, describes later a double 

 "ne of fusion between ganglia and epidermis, forming the lateral and 

 epibranchial sense organs, which may perhaps be considered as com- 

 parable to the ganglia of the roots and ganglia of the trunks. In mam- 

 mals, however, he had found only a single line of epidermal fusion, that 

 existing over the ganglia of the trunks. In our series of embryos the 

 ganglia of the roots do not seem to take part in the formation of epi- 

 dermal sense organs, and show no sign of fusion. The condition here 

 resembles that' described by Fronep, 85, in his earlier paper. The 

 adherence between the epidermis and the ganglion petrosum and nodosum 

 is indicated in Fig. 6. It is found in all the human embryos studied 

 from 4.5 to 7.0 mm., after which it disappears. This is a somewhat 

 earlier and briefer period than given by Froriep for other mammals. No 

 indication of interchange of cells between ganglia and epidermis could be 

 made out. 



The ganglion petrosum in embryos 7.0 mm. long has become connected 

 with the ganglion of the root by a definite strand of mixed fibres and cells, 

 the fibre elements more and more predominating as the embryo becomes 

 older. At the same time a tapering bundle of fibres sprouts from the 

 distal end of the ganglion petrosum, and forms the main trunk of the 

 nerve, the ramus lingualis, and supplies the third arch. Another branch 

 appears in 14.0 mm. embryos, the ramus tympanicus, and extends for- 

 ward into the second arch. The ganglion thus gives- off a branch both 

 oral and caudal to the second gill cleft, and this completes the glosso- 

 pharyngeus as a typical visceral arch nerve. This was pointed out in 

 mammals by Froriep, 85, who regards the r. lingualis as the post-trematic 

 and the r. tympanicus as the pre-trematic branch. 



Communications exist between the ganglion petrosum and the ganglion 

 nodosum in an embryo of 7.0 mm. (Figs. 4 and 5), where they seem 

 almost as a continuous structure; in other embryos of this stage, and 

 younger, they are completely separated. Later (Figs. 7, 8, and 9), fol- 

 lowing the relative change in position of the adjacent parts which 

 succeeds their unequal growth, these structures are gradually brought 

 close together, and secondary communications are established between 

 them. 



