In Thaliacea the mode of food-collection changes. Except at its extremities, the endostyle of Doliolum is closed by apposition of its lips, and the slime is all driven forwards and upwards along the peripharyngeal bands, from which it is trailed off in fringes by the current which sets towards the gill-slits behind. The fringes of each side are carried dorsally until their bases meet in the "ciliated spire," where they are twisted together into a rotating cord, the free end of which is directed diagonally through the pharyngeal cavity into the oesophageal aperture. The moving fringes thus constitute a conical sieve at the entrance to the pharynx, although the individual threads are constantly being caught into the food-rope dorsally and renewed from the endostyle ventrally (Fol, 1878).
In Salpa the modus operandi is not so clear. The ciliated stripes across the sides of the gill appear to play much the same part as true gill-slits and the food-rope passes backwards along its lower face, instead of freely traversing the pharyngeal cavity. (Fol, /.c.). The fringes of Doliolum are replaced by an almost continuous "curtain" of mucus in Thalia democratica, but the whole process needs re-examination in relation to the great varia tions now shown to exist in the structure of the Salp endostyle, which presents every gradation from a flat symmetrical Amphi oxus-like pad in lasis and Thetys to a closed tube with asym metrical marginal bands in Salpa and Cyclosalpa.
In Copelata Fol's experiments on Oikopleura leave no doubt that the mucus from the small endostyle only supplies the peripharyngeal bands, which function essentially as in Doliolum. The pharynx, however, merely imbibes food-particles already concentrated by previous filtration in the "house," one function of which, at least in forms with "filtering windows," is the ex clusion of all but the minutest pelagic organisms (the Nanno plankton of Lohmann). The meshes of the filtering windows of Oikopleura exclude almost every organism exceeding 3o,u in diameter, and the internal filter-tubes entrap organisms as small as 3/5. Thus Chaetoceras, the elongated Rhizosolenias, Ceratium and most Radiolaria are excluded, while the food captured con sists mainly of small Nitzchia, Chrysomonads, Peridinia, naked Flagellates, Rhizopods and Bacteria between 3p. and a0µ in diameter (Lohmann). Thus the reduction of size and structure in Appendicularians is to be connected with their having tapped a rich reserve of minute food which is wasted by other Tunicates, and the organ elaborated to this end is no longer the pharynx, but their wonderful filtering house, the only clue to the origin of which is provided by Doliolina, which has the habit of period ically exuviating its barrel-like test.
The egg of Ascidians furnishes one of the typical examples of "mosaic" structure and predetermined development, contrasting remarkably in these respects with that of Amphioxus. In Styela (Cynthia) partita destruction of one blastomere in the 2-celled stage results in the formation of a half-gastrula, which may develop even into a half-larva, with notochord, muscles, and mesenchyme all lying on one side of the neural folds. The embryo
possesses little or no power of regenerating the missing parts—a remarkable fact when one notes the great regenerative powers of adult Ascidians.
It follows that the early stages of development throw no par ticular light on pre-Tunicate history; they point forwards to the larval organisation, and not backwards to pre-Chordate ancestors. Neural and neurenteric canals are formed ; the archenteron differ entiates into pharynx in front, into notochord and mesoderm behind ; and the tail in its growth backwards carries with it extensions of the neural tube and of the posterior group of archenteric derivatives. The whole of the intestinal loop arises as an outgrowth from the pharyngeal rudiment. The mesoderm divides at once into somatic and caudal portions; no enterocoeles are produced ; the somatic portion breaks up into blood corpuscles and wandering cells, and the extended caudal portion differenti ates in situ into muscle-cells on either side of the notochord. The whole of the embryonic processes lead directly by the nearest route to the building up of the larva, which is characterised by three features, a locomotive tail, an organ of fixation, and a transitory sensory and nervous mechanism controlling them. Until the close of larval life feeding is entirely suspended, and usually only the rudiments of the pharyngeal mechanism are established prior to fixation and metamorphosis.
The larval brain is developed from an enlargement of the neural canal near its anterior extremity (the cerebral vesicle), in the wall of which two sense-organs normally arise, a stalked unicellular otolith in front, and a multicellular eye behind. As the cerebral vesicle dilates unequally, the otolith cell migrates from a mid-dorsal to a ventral position, while the eye remains postero-lateral. Simultaneously the cerebral vesicle becomes gradually constricted off on the right side, and from before back ward, leaving a narrow canal on its left, which is continuous with the neuroporal extremity in front and the main part of the neural canal behind. According to the extent of the constriction the cerebral (now the sensory) vesicle may open into the neural canal behind, or be completely cut off from it (Botryllus). Its postero-ventral wall becomes greatly thickened as the larval brain ("visceral ganglion"), and gives off nerves to the adhesive papillae in front and, along the neural canal, to the caudal muscles behind, but gives no nerves to the viscera. There is thus a stage of development when the neural canal is forked in front, and into two unequal branches,—a short dilated branch on the right, the sensory vesicle, and a longer narrower canal on the left. The apex of this canal fuses with the pharyngeal wall close to the base of the oral siphon, and becomes converted into the ciliated funnel in front, and the subneural duct and gland behind. This remark able feature of the development is not without many minor modi fications in different types (esp. Botrylloides), but the important point remains that, except in Salps, ciliated funnel, duct and neural gland are invariably developed from the foremost region of the neural canal.