VIVIPAROUS FISHES OF THE PACIFIC COAST. 
459 
In three cases (78, 148, 149) the dorsal portion is wholly or partly filled with cells. This 
is especially well seen in fig. 78, where the boundaries of the dorsal half of the canal 
are as well marked as those of the ventral half, but in which the dorsal half contains 
cells. The same is true of another larva (fig. 148), but in this instance the walls of the 
dorsal half are not so well defined. In another larva (fig. 149) a narrow but well-defined 
canal extends from the remains of the vesicle upward and then curves forward. This 
is probably only the posterior wall of the neurenteric canal. More or less well-defined 
lines extend from the anterior portion of the dome upward. The same condition can 
be traced in series of transverse sections. There seems then to be but little room for 
doubt about this structure. All the evidence indicates that in part at least it is the 
neurenteric canal.* 
Kupffer’s vesicle in general. — Kupffer’s vesicle is evidently a rudimentary struc- 
ture, without function in the majority of fishes. Several quite distinct views have 
been held as to the significance of this structure. 
Kupffer, who first described it, and recently Henneguy, considered it to be the 
allantois of higher vertebrates. 
Balfour homologized it with the postanal vesicle of elasmobranchs. 
Cunningham, Ziegler, McIntosh and Prince, consider it to represent the invagi- 
nated gastrula (archenteron) of cyclostomes and plagiostomes. 
Kowalewski holds in the main to the same opinion. 
Henneguy formerly considered it the homologue of the primitive intestine of 
cyclostomes and amphibians. Many others who have seen the structure are non- 
committal as to its significance. 
The variety of opinions may in part be due to the variation of this structure in 
different fishes, for there is no doubt about the variability of the vesicle. It is difficult 
to see how a waning structure can represent a condition that has not appeared before 
phylogenetically and does not appear till in much higher vertebrates. I can not see 
how it can represent the allantois. Cunningham’s theory is based on the presence of a 
canal between the vesicle and the blastopore. This canal is certainly not present in 
Cymatogaster , where the vesicle does not appear till long after the blastopore is closed. 
Wilson considers the early stages homologous with the terminal part of the archen- 
teron of amphibians and the later stages homologous with the postanal vesicle of 
elasmobranchs. 
The fact that in the majority of fishes it arises long before the alimentary canal 
and disappears, or at least diminishes, before the alimentary canal is formed argues 
against its homology with the postanal vesicle of elasmobranchs. It must be con- 
ceded, however, that in fishes the alimentary canal is late in making its appearance 
as compared with elasmobranchs. The alimentary tract is retarded in teleosts for 
some reason or other. The caudal vesicle is functionless in elasmobranchs and the 
* I want to point out here the possibility that the neurenteric vesicle alone represents the caudal 
vesicle of other teleosts, and that its connection with the lower oart of the vesicle which is converted 
into the hind gut is after all due to the precocious development of the hind gut, which thus extends 
past the region of the original Kupffer’s vesicle before the growth of the tail has carried the latter 
farther hack. In that case the postanal gut would be represented by the dorsal wall of the hind gut; 
i. e., the region between the hind gut and the neurenteric vesicle. This would account for the fact 
that I have not been able to find any other structure which might be homologized with the postanal 
gut frequently described for teleosts. The lower or main part of the vesicle may then be looked 
upon as the arclientei’ic cavity. 
