220 PEECY SLADEN TEUST EXPEDITION. 



and in the cellular tliickening which separates them a third, that of the most anterior 

 fin-ray box, makes its appearance. This does not seem to be connected in nny way with 

 the other spaces : the space lying immediately below it ends blindly just above the 

 anterior end of the brain, that lying above tapers away as a long lymjih-canal. In 

 yl. pelagicus the relations are very similar, except that the dorsal rostral canal appears 

 lo be obliterated, save for disconnected spaces, by concrescence of its walls. Above the 

 brain the lamella thus formed tapers away as before, the cavity of the long canal repre- 

 senting the fin-ray spaces appearing at its ventral edge. In neither species could I. see 

 any trace of connection between the dorsal canal and first myoccele. 



These facts, and the non-existence of any such cavity in the adult Branchiostoma or 

 Asymmetron — where in its place we find irregular jiear-shaped spaces in the skeletal 

 tissue — seem to point to the conclusion that the " dorsal rostral canal" is itself no more 

 than a space of that Dature *. 



The occurrence of spaces within the skeletal tissue of the rostrum even at the early 

 stage represented by Amphioxides accords with the fact that they are best developed in 

 the same region in the adult. 



(3) Goldschmidt describes as a probable representative of true connective tissue, 

 occurring in tiie tail, rostrum, and cheeks, " sehr schmale, auf der Kante stehende 

 Lamelleu, deuen in unregelmiissiger Verteilung spindelformiger Kerne eingelagert sind." 

 They show an "Anordnung in Druckkurveu," which suggests at once a supporting 

 function. In the rostrum they are visible dorsnlly only in the region corresponding 

 to the " dorsal canal," ventrally OTcr the whole surface of the fin. So much is seen from 

 whole preparations : in transverse sections the lamellae are said to appear as fine radiating 

 threads, not embedded in any ground-substance, and taking the place of the elsewhere 

 cell-less gt'lalinous layer of the cutis. 



In this account I cannot help thinking tliat two distinct structures are confused. My 

 sections unfortunately are not good enough to admit a proper study of the radiating 

 threads. Goldschmidt, however, figures these (figs. 69 and 47) both at the sides of the 

 notochord, and, dorsally, above the region of the rostral canal. If they occur in these 

 l^ositions, why are they not seen there in whole preparations ? 



The most natural explanation is that they are not seen at all in the latter f , and that 

 the lines of cells which are seen are something quite different. The direction in which 

 the latter run clearly indicates that they are a part of the skeletal system. Dorsally they 

 appear to be directly continuous with the " lymph-canals " opening into the dorsal canal | 

 and are seen only in the region corresponding to the latter ; it seems evident that they 

 situply constitute its walls. If so, they give conclusive evidence that my view of the 

 nature of the canal is correct. Ventrally they are seen to take a corresponding direction 

 in relation to the walls of the ventral rostral cavity, another space which disappears in 



* If t.his is so, the name may conveniently be transferred to Goldschmidt's " subdorsal rostral canal." 

 t Probably the fine radiating threads are simply fibrillae in the gelatinous layer, and do belong to the system of 

 true oonnective tissue the existence oE which Goldschmidt has demonstrated in Amphio.vus, 

 t This is clearly nhown in Goldschmidt's fig. 3-1. 



