THE EVOLUTION OF THE CEREBRAL HEMISPHERES. 305 



membranous roof has been displaced in selachians by the growing 

 up, folding over and fusion of the lateral walls. In other fishes 

 the corresponding part of the lateral wall is much smaller and the 

 membranous roof is more extensive than in selachians. In 

 some bony fishes (Fig. 151 C) there seems to be a sort of inversion 

 of the lateral wall, so that the striatum and epistriatum bulge 

 high up in the ventricle and the pallium passes over them and is 

 attached to the lateral wall on the ventro-lateral aspect of the brain. 

 A membranous roof over the median ventricle is universally 

 present in the forebrain of vertebrates. In the more speciahzed 

 selachians the greater part of it is displaced by the upgrowth from 

 the lateral walls ; in bony fishes it is extraordinarily broad because 

 of the reduction and receding of the lateral walls. 



The nervus terminalis which is characteristic of selachian brains 

 is found also in Amia, in Protopterus and in embryonic stages of 

 Ceratodus, but nothing is known of its central connections. 



The FOREBRAIN OF AMPHIBIA presents two external peculiarities : 

 first, the relatively great size and independence of the lateral lobes, 

 and second, the f?ct that they extend forward far beyond the 

 lamina terminalis. A glance at Figures 8 and 9 will show that the 

 real difference between the amphibian and selachian brains in 

 this regard is sHght. In front of the lamina terminalis the lobes 

 are separated by an open space known as the sagittal 'fissure. In 

 the frog this fissure is bridged across by the fusion of the two 

 olfactory bulbs, but remains open behind the bulbs. The internal 

 structure of the amphibian forebrain has been puzzling chiefly 

 because its fiber tracts and commissures are difficult to compare 

 with those of other vertebrates. An understanding of the gross 

 relations in one of the lower tailed forms, such as the hellbender 

 (Cryptobranchus) or mud-puppy (Necturus), will help to make 

 the finer structure intelligible. The median ventricle is relatively 

 short (Fig. 150). Its floor in front of the optic chiasma is formed 

 by the nucleus praeopticus and farther forward by the lamina 

 terminalis, which is very greatly thickened by two commissures 

 crossing one above the other in nearly the same plane. In front 

 of the commissures the lamina terminalis is nearly vertical in posi- 

 tion and becomes continuous with the membranous roof. This is 



