LARVAL DEVELOPMENT. 159 



to the head-cavities. Moreover, as was pointed out above, 

 the dorsal groove of the archenteron, which gives rise to 

 the notochord, remains open into the archenteric cavity 

 in the region of the first myotome, and even somewhat 

 behind the level of the neuropore, for some time after 

 its walls have approximated to form the solid notochord 

 behind this region. 



The forward extension of the notochord in Amphioxus 

 is, therefore, dc facto, to a large extent an ontogenetic 

 phenomenon, although, from the very beginning, it shows 

 what may be described as a precocious tendency to extend 

 beyond the nerve-tube. We shall also find that there is 

 every reason to suppose that it is a cenogenetic, and not a 

 palingenetic, feature. 12 



Since we know for an actual fact that the primary gill- 

 slits of the larva belong ancestrally to the left side, it fol- 

 lows as an absolute topographical necessity that the mouth 

 has been brought to one side from an originally median 

 dorsal position, by the same semi-rotation of the pharynx 

 (in the sense explained above) which has demonstrably 

 carried the primitive left-side gill-slits under the pharynx 

 up to the right side of the larva. But this is not the only 

 criterion by which we can judge of the ancestral position 

 of the mouth. 



In the larvae of the Ascidians, the nearest existing rel- 

 atives of Amphioxus, there is a proeoral lobe and a neuro- 

 pore, which opens at first to the exterior in the mid-dorsal 

 line, just as in Amphioxus. But in contrast to the latter 

 form the notochord does not extend forwards into the re- 

 gion of the praeoral lobe, but it stops short behind the 

 cerebral vesicle. 



Immediately in front of the neuropore, in the Ascidian 

 larva, the wall of the pharynx comes into contact with the 



