10 SURFACE FAUNA OF THE GULF STREAM. 
with that of a young Velella, and differs radically from that given by 
Pagenstecher for Rataria (as young Velella?), in which the pneumatocyst 
clearly shows the eight primary divisions so characteristic of Porpita with the 
central opening of the disk. Huxley also observed that in young Velellae 
the hepatic mass does not exist, and is only gradually developed under the 
pneumatocyst, and that the canals which cross it are mere subdivisions of 
the somatic cavity produced by the lobes of this organ and their mutual 
anastomoses. My observations on Velella lead me to agree with Kolliker re- 
garding the mode of junction of the liver canals and of the canals on the 
free edge of the mantle and on the upper part of the float. The dendritic- 
like structure of the canals of the crest have also been noticed by Huxley in 
a young Velella measuring somewhat less than half an inch in length. The 
circulation within the canals was most active, and wholly due to ciliary action. 
The figures given by Huxley (page 126, Oceanic Hydrozoa) of the peculiar 
corrugations and lobes of the lower surface of the float, being taken from 
alcoholic specimens, are not quite satisfactory representations of their appear- 
ance in fresh specimens. 
As has already been suggested by Kolliker, Agassiz, and McCrady, the re- 
lationship of the Velellida: and Porpitide, as well as the Physalida,* to the 
Tubularian Hydroids is very great. If we compare this group as a whole 
to the other Siphonophores, the absence of “ Deckstiicke ” and of swimming 
bells seems to distinguish them from all the other Siphonophores except 
Rhizophyza, which perhaps is only a representative of the embryonic stage 
of the Physalidxe, and does not belong into the close association with the 
other Siphonophores, with which it has usually been placed. What is 
known of these different families seems to indicate a far closer relationship 
with the Tubularian Hydroids, such as Hydractinia, which may perhaps be 
the closest ally of the genera named above, and in which the chitinous exten- 
* The chambers of the crest of Physalia can be considered as a sort of girder which stiffens the whole 
float, and to a certain extent takes the place of the chitinous crest of Velella. The structure of the crest 
is seen in section to be a broadly rectangular triangular cell, subdivided by horizontal bars to form the 
smaller trapezoidal cells of a second, third, fourth, and fifth story, adjoining triangles being again con- 
nected longitudinally by similar bars. The float of Physalia remains in fact in the embryonic stage in 
which we find the sail or crest of Velella and young Porpita. In the latter the crest gradually disappears 
to cover the upper part of the float; in the former it continues through to the mature stage, being sup- 
ported by a chitinous vertical projection from the float, which is absent in younger stages, while in 
Physalia we have only the mantle, if I may so call it, of the crest left, the pneumatophore not secreting a 
chitinous float or any structure homologous to the circular chitinous float of Velella or the chitinous disk 
of Porpita. The presence of a net-work of canals at the base of the float of Physalia, similar to that of 
the upper part of the float of Velella Porpita, has been made probable by some observations of Quatre- 
fages. (Ann. Scien. Nat. 1853). 
