34 Journal of Comparative Neurology. 



the ventral wall are seen a large number of bodies of about 

 the same size as the nuclei which have a homogeneous grayish 

 brown color. These resemble the nuclei of glia cells. The 

 dorsal wall is likewise composed of a single layer of columnar 

 cells, but these are so high as to meet the ventral wall in most 

 places and so nearly obliterate the cavity. The cells are ex- 

 tremely long and slender, almost like fibers, and have their 

 nuclei in slight enlargements either near the dorsal (basal) ends 

 or near the middle. Between the dorsal and ventral walls, 

 probably connected with the cells of the ventral wall, are a 

 number of darkly staining bodies which appear to be the "rods 

 or cones" described by Studnicka ('oo). The structure of the 

 epiphysis as a whole strikingly reminds one of the lens of the 

 eye at an early stage of development. 



Beneath the epiphysis lies the Zirbelpolster which is also 

 an elongated vesicle. It is connected with the left ganglion 

 habenulae by a band of fibers as previously described. This vesi- 

 cle has a very thin dorsal wall consisting of about two layers of 

 small cells, which passes gradully into a ventral wall consisting 

 of many layers. This wall gives the appearance of a very dense 

 mass of nerve cells such as is sometimes found in the brain of 

 lower vertebrates. The vesicle has a large open cavity. From 

 among the cells of the ventral wall nerve fibers emerge ventrally 

 and form the band which runs to the left ganglion habenulae. The 

 bundle is large and easily followed into the ganglion where it 

 penetrates the cell layer and does not go to the superior com- 

 missure. This apparatus, therefore, is related wholly the left 

 ganglion. 



Retzius ('96) has studied the epiphysis and paraphysis of 

 Ammocoetes by the Golgi method and concludes that the 

 epiphysis contains no nerve elements, while the paraphysis 

 contains bipolar cells in its ventral wall which send fibers into 

 the ganglion habenulae. He concludes: "In Anbetracht der 

 interessanten regelmassigen Gestalt dieser bipolaren Zellenele- 

 mente und ihrer centralwarts ziehenden Fortsatze scheint mir 

 die Paraphysis der Ammocoetes eher als die Epiphysis als ein 

 functionierenden Hirnthiel aufzufassen zu sein." 



