XXX11 
THE VOYAGE OF H.M.S. CHALLENGER. 
and places the interior of the umbrella in free communication with the surrounding 
water. 
This account of the development of the Medusa-bud agrees in almost all points with 
that previously given by Agassiz, who regarded the radial canals of the umbrella as the 
result of the non-adhesion of the two layers of the endodermal cup along definite longi- 
tudinal lines, while it differs from the view formerly urged by myself, in which I 
regarded the radial canals as growing out in the form of tubular processes from the basis 
of the manubrium. 
The theory of the formation of the canals by arrested adhesion of the two endo- 
dermal layers along the lines of these canals will satisfactorily explain the phenomena, 
and the truth of this explanation seems to be placed beyond doubt by the discovery of 
the endoderm lamella shown by the Hertwigs to exist as a persistent structure in the 
umbrella of the Medusa. 
It must not be supposed, however, that radial canals are never formed by an out- 
growth of tubes from the base of the manubrium into the walls of the umbrella. Though 
the primary canals may be always formed as described above, there are instances 
(JEquorea, &c.) in which the canals increase in number after the detachment of the 
Medusa, and in such cases it is certain that the newly formed canals extend themselves 
in the umbrella walls from the base of the manubrium in order to open into the circular 
canal round the margin . 1 
The formation of the medusoid also takes place through the medium of an endocodon, 
and is similar to that just described for the Medusa, except in the fact that the radiating 
canals are in most cases entirely suppressed, and that in those rare instances in which 
they may be seen in a rudimental or even completely developed form the umbrella does 
not present the wide codonostome of the Medusa, and is not developed so as to be employed 
as an organ of natation. 
In the sporosac, on the other hand, the development does not take place through the 
formation of an endocodon, and no endoderm lamella being formed, the walls of the sac 
retain the condition presented in the early stages of the bud, being simply composed of 
two layers, an internal endodermal layer, and an external ectodermal layer, these two 
being subsequently separated by the presence between them of the generative products. 
Between the medusoids and those gonophores, which as planoblasts have attained 
the condition of true Craspedote Medusae, there is a close parallelism, the various parts 
in the one having their representatives in a more or less modified form in the other. 
In the medusoid, passing from without inwards, we meet with the following layers : — 
Most externally a layer of ectoderm ; then a thin lamina of endoderm in which gastro- 
vascular canals may or may not be developed, and which though present in the early 
stages of the bud is not always demonstrable in the mature gonophore. To this 
1 Gymnoblastic Hydroids, p. 79, fig. 35. 
