PvEPOET OX THE TUXICATA. 
399 
■while in Goodsiria the test is greatly enlarged, thus rendering the colony thick and 
massire. In none of the Polystyelidse have the Ascidiozooids come to be arranged in 
systems, and no common cloacal cavities have been formed ; the atrial apertures of all 
the Ascidiozooids open independently upon the exterior of the colony. In this respect 
the colony is in the same stage of differentiation as that reached by most of the 
DistomidfB, while the other groups of Compound Ascidians {e.g., BotryUidse, Polyclinidm, 
Didemnidse) have advanced a stage further by the formation of systems wnth common 
cloacal cavities. 
It is possible that the family Botr}dlid8e may have a closer relationship with the 
Polystyelidae than I have assigned to it above (p. 396). The primitive Botryllidse, in 
place of arising from the primitive Cynthiidm, may possibly have been derived from the 
Polystyelidae near the point where Synstyela and Goodsiria diverged, as is shown by the 
dotted line in the diagram (fig. 15). In that case the evolution of the primitive 
Botryllidae would consist in the gradual formation of systems in the colony, and the 
complete disappearance of all traces of folds in the branchial sac. 
Probably the two most important conclusions I have arrived at in these j^hylogenetie 
investigations are — (1) as to the relationship of Pyrosoma, and (2) as to the 
polyphyletic origin of the Aseidise Composite. Pyrosoma, although now a pelagic free- 
swimming organism, -was probably derived from the fixed Compound Ascidians. The 
discovery of Ccelocormus huxleyi shows the relationship between Pyrosoma and the 
primitive Didemnidm, and the latter in their turn -were derived from the primitive 
Distomidse ; consequently, Pyrosoma is directly related to the most typical of the 
Compound Ascidians. 
The Ascidiae Composite, or Synascidae, are really an unnatural assemblage of groups, 
as they seem to have been derived from the Simj^le Ascidians or their ancestors at two 
or three distinct points (see fig. 11, p. 388). The result of this is that the so-called 
Compound Ascidians form three ^ groups — (l) the Polystyelidae, (2) the Botryllidae, 
and (3) the remaining families — ■which are more nearly related to particular groups 
of Simple Ascidians than they are to one another. This conclusion renders it even 
more difl&cult than it was before to draw any line of demarcation between Simple and 
Compound Ascidians. 
1 Possibly (1) and (2) may have liad a common origin, as is shown in fig. 15, p. 398. 
