REPORT ON THE TUNIC ATA. 131 



differs from the Ascidiitlai in having no internal longitudinal l)ar.s in the branchial sue ; 

 while it difi'ers from all the Clavelinidas, and agrees with all the Ascidiida3, in not 

 reproducing l)y gemmation. It prol)ably separated from the main l^ranch earlier than 

 Ecteinascidia, sis it has not acquired internal longitudinal bars ; and it lost the 

 property of gemmation after its origin, since that property was found in the ancestral 

 forms occupying the main branch as far along as the position of Ecteinascidia. 



Following now the important branch given off at F. (see table, p. 120), which leads 

 to most of the groups of Compound Ascidiaus, it is found that at an early period two 

 main lines of descent were formed (G. in table), the one leading to the Pol)'- 

 clinidse, and the other, I believe, through the Distomidas to the Didemnidse, the 

 Diplosomidse, Ccelocornuis, and finally to Pyrosoma. In both of these lines, and in 

 their common ancestors occupying the line from F. to G., the power of reproducing 

 by gemmation was retained and even increased, and the members of the resulting 

 colonies became more closcl}- united with one another than is the case in the 

 Clavelinid?e. 



The line which leads from G. to the existing Polyclinidre (on the right hand side of 

 G. in the table, p. 120) must have Ijeen occupied by a series of forms in which the 

 body became gradually more and more elongated antero-posteriorly, and finally divided 

 more or less distinctly into three regions — (1) an anterior part, the thorax, containing 

 the branchial sac and other organs; (2) a middle portion, the abdomen, containing the 

 stomach and the intestinal loop ; and (3) a posterior region, the post- abdomen, 

 containing mainly the reproductive organs and the heart (see Fig. 21 on next page). 



Large colonies now l)ccame formed, and the Ascidiozooids (produced by gemmation 

 from the post-abdomen) composing the colony were so closely placed that their tests 

 became united to form a continuous investing mass. On account of several 3'oung 

 Ascidiozooids being usually produced from a single older Ascidiozooid in the colon)', a 

 more or less regular grouping into systems naturally took place, and then the atrial 

 apertures of the various Ascidiozooids in a system coalesced to form a centrally placed 

 common cloacal aperture (Fig. 21, cl., and Fig. 26, p. 141). The branchial sac also 

 went a gradual degeneration, resulting in the complete disappearance of the system of 

 internal longitudinal bars inherited from the ancestral forms at F. 



The central axis of the Polyclinida), extending from G. to li. in the table (p. 120), 

 was composed of a series of ancestral forms in wdiicli these and some other less 

 important changes were gradually being affected, liut from this axis a few short side 

 branches were given off at different periods. First, not very far from G., a line of 

 descent diverged leading to Pharijngodictyon, the most remarkable of the deep-sea 

 Compound Ascidians ' o]>tained during the Challenger Expedition. The ancestors of this 

 form must have diverged from the axis of the Polyclinida) while internal longitudinal 



> This Report, Part 11. p. 152. 



