REPORT ON THE TUNICATA. 
391 
different periods. First, not very far from the point C. (see fig. 12), a line of descent 
diverged leading to the remarkable deep-sea genus Pharyngodictyon (see p. 153). The 
ancestors of this form must have diverged from the axis of the Polyclinidge, while 
internal longitudinal bars were still present in the branchial sac, since these vessels 
(according to the interpretation of the structure given above, p. 155) are now present 
in Pharyngodictyon. In this side branch leading to Pharyngodictyon, however, a 
remarkable degeneration of the branchial sac took place, resulting in the total 
suppression of the system of fine interstigmatic vessels. 
The second side branch from the Polyclinidse, that leading to the genus Tylobranchion, 
appears to have left the main axis just about the point where the internal longitudinal 
bars in the branchial sac were disappearing, as Tylobranchion now shows no internal 
longitudinal bars, but possesses a system of papdlas which there is reason to believe are 
rudimentary connecting ducts, and which sometimes give off projections resembling the 
rudiments of internal longitudinal bars found attached to the free ends of the connecting 
ducts in some Simple Ascidians (see p. 161). Pharyngodictyon and Tylobranchion then, 
if the structure of their branchial sacs has been rightly interpreted, furnish us with 
valuable clues to the process of degeneration which went on in the ancestral 
Polyclinidse. 
The genus Atopogaster, which may be placed upon a third short twig given off from 
the axis of the Polyclinidse (see fig. 12), differs very slightly from typical members of the 
family. The only modification of importance which occurs in this genus is in the 
structure of the stomach. The wall of this organ is thrown into a series of more or less 
distinct transverse folds, a condition never found in the other genera of the family. 
The last side branch from the axis previous to its division into two at D. is the line 
leading to the genus Polyclinum, characterised by its smooth-walled stomach, its twisted 
intestine, and its laterally placed post-abdomen (see p. 186). It is a little difficult to 
determine the relations between Atopogaster and Polyclinum. In all probability the 
stomach in the ancestral Polyclinidse was smooth-walled, while in all the higher forms of 
the family it is irregularly thickened either by the formation of a number of short cseca, 
or by being thrown into longitudinal folds. The first of these conditions is probably 
derived from the last, consequently the ancestor occupying the point D. (see fig. 12) had 
a longitudinally folded stomach-wall. Possibly Atopogaster diverged from the axis at 
the period when the tendency towards the formation of a ridged or thickened stomach- 
wall was being developed, and then after its separation the folds became formed more or 
less transversely in place of longitudinally. The fine leading to Polyclinum probably 
diverged from very much the same point, and then the incipient folds in the stomach-wall 
were lost, and the other peculiarities of the genus acquired. Polyclinum cannot well be a 
more primitive form derived from the axis at a point before the formation of a thickened 
stomach-wall, as it shows no traces of internal longitudinal folds in its branchial sac, and 
