cxx 
THE VOYAGE OF H.M.S. CHALLEHGEE. 
desmas of Macandrewia. In both, further, the chamber-system is aphodal, and the 
choanosomal mesoderm sarcenchymatous. 
No dichotrisenes, however, are present in Pachastrella, as there are in Pleroma 
among the Pleromidse, in most of the Corallistidse, and in some Tetracladidae ; and though 
these spicules do not occur in Macandrewia, they are represented by the closely similar 
phyllotrisenes, which have been evolved from them. This difference however is not of 
sufficient importance to shift the connection from the Pachastrellidse to the Theneidse (a 
connection however which by another line of reasoning I admit to be possible), it merely 
suggests that the Pachastrellid ancestor of the Tetracladidse possessed dichotrisenes which 
the genus Pachastrella has lost. 
So far as can be judged from resemblances between existing Sponges, the Tetra- 
cladidse have descended from the Pachastrellidse, and are the oldest family of the demus. 
In some respects however the Pleromidse and Corallistidse are least removed from what 
we should conclude was the ancestral type, and it is possible that these families should be 
regarded not as derived from the Tetracladidse, but from an ancestor common to them 
both. Thus in Pachastrella some of the calthrops are replaced by triods and oxeas ; and 
in Macandrewia some of the tetracrepid desmas by triacrepid and rhabdocrepid ones ; if 
now the ancestral Lithistid possessed a similar mixture of forms, the Tetracladidse might 
be regarded as having originated by the selection of the tetracrepid desmas exclusively 
(the selection being incomplete in the case of Macandrewia), and the Corallistidse and 
Pleromidse by a selection of the rhabdocrepid forms. This view is in accordance with 
the fact that the flagellated chambers in Pleroma are larger, and one would therefore 
presume more primitive, than in any other Lithistid, and indeed than in Pacha- 
strella itself. It would also accord with the fact that dichotrisenes are more common in 
the Corallistidse and Pleromidse, and phyllotrisenes and discotrisenes in the Tetracladidse. 
The modification thus introduced into our earlier results will be carried still further when 
we reflect that a Lithistid ancestor with dichotrisenes and large flagellated chambers, 
involves a Pachastrellid ancestor with dichotrisenes and large flagellated chambers, and 
such a Pachastrellid would approach more nearly the Theneidse than any existing Pacha- 
strellid of which the soft parts have been examined. This is a further indication of the pro- 
bability that the Pachastrellidse have been derived from the Theneidse. On the whole then 
it would appear that the Lithistida have descended from an extinct pro-Lithistid ancestor, 
which in turn had descended from an extinct pro-PachastreUid ancestor, possessing closer 
affinities with the Theneidse than do any of the existing forms of the Pachastrellidse. 
From the Trisenosa to the Anomocladidse we are presented with an almost continuous 
series of gradually simpler forms, simpler, that is, as regards the characters of the skeleton. 
Thus in the Ehabdosa trisenes have disappeared, and while in one family (Cladopeltida) 
microscleres are absent, in another they are sigmaspires {Scleritoderma)} 
1 In this case at all events the sigmaspire is to be regarded as a degenerate and not as a primitive spicule ; of course, 
one is left with a last appeal to “ reversion.” 
