GIBSON— CEPHALOCHORDA : " AMPHIOXIDES." 221 



the adult Branchiostoma, its walls very proliably going to form " lympli-canals." Beyond 

 its limits I am unable to see them, although Goldschmidt figures them right up to the 

 edge of the rostrum. 



The Spaces of the Snout. 



In addition to the " dorsal rostral canal " the rostrum of Amphioxides possesses five 

 other distinct cavities, all probably of true coelomic nature : — 



(1) The subdorsal rostral canal, also present in the adult Branchiostoma and Asym- 

 metron, as a space extending from the anterior end of the brain towards the tip of tlie 

 notochord dorsally. Goldschmidt finds it to connect on both sides with the cavities of 

 the first pair of myotomes, via the lateral rostral cavities. I cannot find these connections 

 and doubt their existence, though it is impossible to be positive upon the point : they are 

 much more probable than the connection last described. This canal is probably to be 

 regarded as the homologue of the dorsal fin-ray spaces *. 



(2) The two lateral rostral canals, simple lateral prolongations of the first pair of 

 myocoelomic cavities. 



(3) The ventral rostral canal, which Goldschmidt again finds to communicate with the 

 first pair of myocoeles : again I cannot follow him f . This space is much lartfer in 

 A. pelagicus than in A. valdivice : in the latter it tapers away before the level of Hatschek's 

 pit is reached posteriorly ; in the former it ends apposed to the front wall of the pit as a 

 square cavity lined with moderately high epithelium. 



It would seem probable that the space below the notochord in the rostrum of the adult 

 Branchiostoma A.e'imheA. by Van Wijhe (1902) is the same as this canal and not the 

 remnant of the rostral cavity. 



(4) The ventral rostral cavity — the well-known " right endoderm-sac " of Hatschek 

 (1881), — the wall of which. \n Amphioxides supplies the muscle of the pre-oral organ: Van 

 Wijhe (1906) has now found the same to be the case in the Branchiostoma larva. It 

 may be worth noting that this muscle takes origin, not below the chorda (as those of the 

 mouth do), but to the left of it, from the septum separating the rostral cavity from that 

 of the first myotome. 



In A. valdivice (not in A. pelag'icus) this cavity sends back a small tubular diverticulum 

 to the left of Hatschek's pit, occujiying exactly the position of the canal regarded by 

 Goldschmidt as a left-side connection between splanchnocoele and first myocoele (shown 

 in his fig. 49, " c.s.k"). I am convinced that the two canals are identical and that no 

 such connection occurs, at any rate in the older animals. The canal has an apparent 

 opening posteriorly into the splanchnocoele, near the anterior termination of the latter 

 immediately behind Hatschek's pit : I cannot be certain that this opening is not artefact ; 

 it appears to be shown in Goldschmidt's fig. 51. 



* The origin of the fin-ray spaces is unknown. Hatschek (iSS8) regarded their epithelium as derived from the 

 " Cutisblatt " of the myocoeles, but did not follow their development. 



t I do not wish to state dogmatically that the communications are absent — particularly in this case, where the 

 walls of the cavities in question are closely united laterally. Possibly they may be visible with certainty in bettor 

 sections than mine. 



SECOND SEKIES. — ZOOLOGY, VOL. XIII. 30 



