Classification. —The Nemertines are subdivided into three suborders: — IIoplonemertea, Schizonemertea, and Paleeonemertea (5). The (1) IIoplonemertea embrace all the species with a stylet in the proboscis, and also Malacobdella, which has an unarmed proboscis, but which, by the details of its organization and its development, must certainly be placed here (parasitism may be the cause of its incipient degeneration). The special characters of this suborder may be gathered from the anatomical descriptions hereafter to be given. In those species of which the embryology has been investigated the development was direct. The more common or more important genera are Amphiporus (Μ. pulcher , British coasts, Mediterranean; A. splendidus, Atlantic), which is comparatively short, Nemertes (N. gracilis, Atlantic and Mediterranean; A r . antonina, Mediterranean ; 2V. echinoderma, Mediterranean), which is long and thread-like, Tetra-stemma, Drepanophorus (with more complicate armature in the proboscis), Akrostomum, Malacobdella. (2) In the Schizonemertea all those genera and species are united which have deep, longitudinal, lateral, cephalic fissures. The development of some (Lineus) is characterized by the so-called larva of Desor, of others (C erebratulus) by the curious and characteristic pilidium-larva. The principal genera are Lineus (L. longissimus, Atlantic; L. obscurus), Cerebratulus (C. marginatus, C. bilineatus, both Atlantic and Mediterranean; C. urticans, Mediterranean ; C. fascio-latus and aurantiacus, C. hepaticus and dohrnii, Mediterranean; C. macintosh™, Madeira), Langia (L. formosa), Borlasia (B. elizabethae). (3) Of the Paleeonemertea the most typical and most characteristic genera are Carinella and Cephalotlιrix. They differ considerably both from Hoplo-and from Schizonemertines, and evidently belong to a lower stage of differentiation. The genera Polia (P. delineata and P. curta, Mediterranean) and Valencinia are provisionally arranged in this order because, though less primitive, they are not typical representatives of the other two suborders. The development of these species is not at all, or only very superficially, known. For the further characters of the last two suborders see the anatomical description below.

Another subdivision generally current is that into the Enopla and the Anopla (14). This does not, however, take into sufficient account the primitive and diverging characters disclosed by the very important less highly organized genera.

Anatomy. —(a) Proboscis and Proboscidian Sheath.— The organ most characteristic of a Nemertine is without doubt the proboscis. With very few exceptions {Malacobdella, Akrostomum, where it has fused with the mouth to a single exterior opefiing), there is a terminal opening (subterminal in Valencinia') at the foremost tip of the body, out of which the proboscis is seen shooting backwards and forwards, sometimes with so much force that both its interior attachments are severed and it is entirely expelled from the body. It then often retains its vitality for a long time, apparently crawling about as if it were itself a worm, a phenomenon which is at least partially explained by the extraordinary development of nervous tissue, equally distributed all through the walls of the proboscis, and either united (10) into numerous longitudinal nerve-stems (Drepanophorus, Amphiporus) or spread out into a uniform and comparatively thick layer (Cerebratulus, sp.). This very effective and elaborate innervation, which has been directly traced (6) to the brain, whence strong nerves (generally two) enter the proboscis, renders it exceedingly probable that the most important functions of the proboscis are of a sensi-ferous, tactile nature, a supposition which is again strengthened by the fact that amongst the Rhabdocoel Turbellarians an organ which may be called the forerunner of theNemertean proboscis has been proved (3) to be the morphological equivalent of the foremost tip of the body, which, as an organ of delicate touch, has acquired the property of folding inwards. In Nemertines the everted proboscis is retracted in the same way as the tip of a glove finger would be if it were pulled backwards by a thread situated in the axis and attached to the tip. The comparison may be carried still further. The central thread just alluded to is represented in the Nemertean proboscis by that portion -which is never everted, and the tip of the glove by the boundary between the evertible and non-evertible portion of the proboscis—a boundary which in the Hoplonemertini is marked by the presence of a pointed or serrated stylet. This stylet is thus situated terminally when the proboscis has reached its maximum eversion. It adds a decisively aggressive character to an organ the original significance of which, as ∙we have seen, was tactile. This aggressive character has a different aspect in several genera which are destitute of a central stylet, but in which the surface that is turned outwards upon eversion of the proboscis is largely provided with nematocysts, sending the urticating rods of different sizes in all directions. In others this surface is beset with thick, glandular, adhesive papillae.

The comparison with the glove-finger is in so far insufficient as the greater portion of the non-evertible half of the proboscis is also hollow and clothed by glandular walls. Only at the very hindermost end does it pass into the so-oalled retractor-muscle (fig. 2), which is attached to the wall of the space (proboscidian sheath) in which the proboscis moves about. This retractor-muscle, indeed, serves to pull back with great rapidity the extruded proboscis, and is aided in its action by the musculature of the head. The extrusion itself depends entirely upon contraction of the muscular walls of the space just mentioned (proboscidian sheath). As it is (1) closed on all sides, and (2) filled with a corpuscular fluid, the contractions alluded to send this fluid to impinge against the anterior portion, where the proboscis, floating in its sheath, is attached with it to the muscular tissue of the head (fig. 3). Partial extrusion lessening the resistance in this region inevitably follows, and when further contractions of the walls of the sheath ensue total extrusion is the consequence. It is worthy of notice that in those Nemertines which make a very free use of their proboscis, and in which it is seen to be continually protruded and retracted, the walls of the proboscidian sheath are enormously muscular. On the other hand, they are much less considerably or even insignificantly so in the genera that are known to make a rather sparing use of their proboscis.

The proboscis, which is thus an eminently muscular organ, is composed of two or three, sometimes powerful, layers of muscles—one of longitudinal and one or two of circular fibres. In the posterior retractor the longitudinal fibres become united into one bundle, which, as noticed above, is inserted in the wall of the sheath. At the circular insertion of the proboscis in front of the brain the muscular fibres belonging to the anterior extremity of the body and those connected with the proboscis are very intimately interwoven, forming a strong attachment.

The proboscis broken off and expelled is generally reproduced, the posterior ribbon-like end of this reproduced portion again fusing with the walls of the sheath (11). There is reason to suppose that, when a wound is inflicted by the central stylet, it is envenomed by the fluid secreted in the posterior proboscidian region being at the same time expelled. A reservoir, a duct, and a muscular bulb in the region (fig. 4) where the stylet is attached serve for this purpose. The significance of two or more (in Drepanophorus very numerous) small sacs containing so-called “reserve” stylets resembling in shape that of the central dart is insufficiently known.

The proboscidian sheath, which by its transverse contractions serves to bring about eversion of the proboscis in the way above traced, and the muscular walls of which were similarly noticed, is attached to the musculature of the head just in front of the ganglionic commissures (fig. 3). In nearly all Nemertines it extends backwards as far as the posterior extremity, just above the anus; in Carinella it is limited to the anterior body-region. The corpuscles floating in the fluid it contains are of definite shape, and in Cerebratulus urticans they are deep red from the presence of haemoglobin. Internally the muscular layers are lined by an epithelium. In the posterior portion this epithelium in certain Schizonemertea has a more glandular appearance, and sometimes the interior cavity is obliterated by cell-proliferation in this region. Superiorly the sheath either closely adheres to the muscular body-wall, with which it may even be partly interwoven, or it hangs freely in the connective tissue which fills the space between the intestine and the muscular body-wall.

(δ) Cutaneous System.— Externally in all species a layer of ciliated cells forms the outer investment. In it are, moreover, enclosed unicellular glands pouring their highly refracting contents, of a more or less rod-like shape, directly to the exterior. They appear to be the principal source of the mucus these animals secrete. In Schizonemertines these elements are separated by a thin homogeneous basement membrane (fig. 8) from the following,—that is, from a layer in which longitudinal muscular fibres are largely intermixed with tortuous glands, which by reason of their deeper situation communicate with the exterior by a much longer and generally very narrow duct. The pigment is also principally localized in this layer, although sometimes it is present even deeper down within the musculature. The passage from this tegumentary layer to the subjacent longitudinal muscular one is gradual, no membrane separating them. In Carinella, Cephalothrix, Polia, and the Hoplonemertines the two tegumentary layers with their different glandular elements are fused into one; a thick layer of connective tissue is situated beneath them (instead of between them) and keeps the entire cutaneous system more definitely separate from the muscular (figs. 7 , 8).

(c) Musculature and Connective Tissue.— The muscular layers by which the body-wall is constituted have been very differently and to some extent confusingly described by the successive authors on Nemertean anatomy. There is sufficient reason for this confusion. The fact is that not only have the larger subdivisions a different arrangement and even number of the muscular layers, but even within the same genus, nay, in the same species, well-marked differences occur. Increase in size appears sometimes to be accompanied by the development of a new layer of fibres, whereas a difference in the method of preparation may give to a layer which appeared homogeneous in one specimen a decidedly fibrous aspect in another. Nevertheless there are three principal types under which the different modifications can be arranged. One of them is found in the two most primitively organized genera, Carinella and Cephalothrix, i.e., an outer circular, a longitudinal, and an inner circular layer of muscular fibres

(fig. 7). The second is common to all the Schizonemertines as well as to Polia and Valencinia, and also comprehends three layers, of which, however, two are longitudinal, viz., the external and the internal one, there being a strong circular layer between tnem (fig. 9). To the third type all the Hoplonemertea correspond; their muscular layers are only two, an external circular and an internal longitudinal one (fig. 8).

The Schizonemertea thus appear to have developed an extra layer of longitudinal fibres internally to those which they inherited from more primitive ancestors, whereas the Hoplonemertea are no longer in possession of the internal circular layer, but have on the contrary largely developed the external circular one, which has dwindled away in the Schizonemertea. In only one instance has the present writer met with a thin exterior circular layer in a very large specimen of Cerebratulus ; younger specimens of the same species did not show it. It is noticeable that Keferstein (9) also observed four layers similarly arranged in one of the specimens of Cerebratulus which he investigated. The situation of the lateral nerve-stems in the. different genera with respect to the muscular layers lends definite support to the interpretation of their homologies here given.

In Carinella, Cephalothrix, and Polia, as well as in all Hoplonemertines, the basement membrane of the skin already above alluded to is particularly strong and immediately applied upon the muscular layers. In the Schizonemertines there is a layer in which the cutaneous elements are largely represented below the thin basement membrane (fig. 8), between it and the bulk of the outer longitudinal muscles. The difference in the appearance of the basement membrane—sometimes wholly homogeneous, sometimes eminently fibrillar—can more especially be observed in differently preserved specimens of the genus Polia.

The connective tissue of the integument and basement membrane imperceptibly merges into that which surrounds the muscular bundles as they are united into denser and definite layers, and this is especially marked in those forms (Akrostomum) where the density of the muscular body-λvall has considerably diminished, and the connective tissue has thus become much more prominent. It can then at the same time be observed, too, that the compact mass of connective tissue (“reticulum,” Barrois) which lies between the muscular body-wall and the intestine (1) is directly continuous with that in which the muscular layers are imbedded. Nuclei are everywhere present. The omnipresence of this connective tissue excludes the idea of any true body cavity in Nemertines.

In Polia the connective tissue enclosed in the external muscular layer is eminently vacuolar,—all the intermediate stages between such cells in which the vacuole predominates and the nucleus is peripheral and those in which the granular protoplasm still entirely fills them being; moreover present.

In addition to the musculature of the proboscis and proboscidian sheath, longitudinal muscular fibres are found in the walls of the oesophagus, whilst transverse ones are numerous and united into vertical dissepiments· between the successive intestinal caeca, thus bringing about a very regular internal metamerization (4). The genitaî products develop in intermediate spaces similarly limited by these dissepiments and alternating with the digestiver caeca.

(d) Nervous System and Sense Organs.— The nervous system of Nemertines presents several interesting peculiarities. As central organs we have to note the brain-lobes and the longitudinal lateral cords which form one continuous unsegmented mass of fibrous and cellular nerve-tissue. The fibrous nerve-tissue is more dense in the higher differentiated, more loose and spongy in the lower organized forms; the cellular nerve-tissue is similarly less compact in the forms that are at the base of the scale. No ganglionic swellings whatever occur in the course of the longitudinal cords. The brain must be looked upon as the anterior thickening of these cords, and at the same time as the spot where the two halves of the central nerve system intercommunicate. This is brought about by a double commissure, of which the ventral portion is considerably thicker than the dorsal, and which, together with the brain-lobes, constitutes a ring through which both proboscis and proboscidian sheath, pass. The brain-lobes are generally four in number, a ventral and a dorsal pair, respectively united together by the above-mentioned commissures, and moreover anteriorly interfusing with each other, right and left. In Carinella this separation into lobes of the anterior thickenings of the cords has not yet commenced, the ventral commissure at the same time being extremely bulky. There is great probability that the central stems, together with the brain, must be looked upon as local longitudinal accumulations of nervous tissue in what was in more primitive ancestors a less highly differentiated nervous plexus, situated in the body-wall in a similar way to that which still is found in the less highly organized Ccelenterates. Such a nervous plexus indeed occurs in the bodywall of all Schizonemertines (7), sometimes even as a comparatively thick layer, situated, as are the nerve stems, between the external longitudinal and the circular muscles (fig. 9). In Carinella, where the longitudinal nerve-stems are situated exteriorly to the muscular layers, this plexus, although present, is much less dense, and can more fitly be compared to a network with wide meshes. In both cases it can be shown to be in immediate continuity with the coating of nerve-cells forming part of the longitudinal cords.

It stretches forward as far as the brain, and in Carinella is again continued in front of it, whereas in the Schizouemer-tines the innervation of the anterior extremity of the head, in front of the brain, takes the form of more definite and less numerous branching stems. The presence of this plexus in connexion vitlι the central stems, sending out nervous filaments amongst the muscles, explains the absence, both in Palaeo-and Schizonemertines, of separate and distinct peripheral nerve stems springing from the central stems innervating the different organs and body-regions, the only exceptions being the nerves for the proboscis, those for the sense organs in the head, and the strong nerve pair (n. vagus) for the oesophagus. At the same time it renders more intelligible the extreme sensitiveness of the bodywall of the Nemertines, a local and instantaneous irritation often resulting in spasmodic rupture of the animal at the point touched.

In the Hoplonemcrtea, where the longitudinal stems lie inside the muscular body-wall, definite and metamerically placed nerve branches spring from them and divide dichotomously in the different tissues they innervate. A definite plexus can here no longer be traced. In certain Hoplonemertines the lateral stems have been noticed to unite posteriorly by a terminal commissure, situated above the anus, the whole of the central nervous system being in this way virtually situated above the intestine. In others there is an approximation of the lateral stems towards the median ventral line {Drepanophorus}; in a genus of Schizonemertines (Langia), on the other hand, an arrangement occurs by which the longitudinal stems are no longer lateral, but have more or less approached each other dorsally (6).

In addition to the nerves starting from the brain-lobes just now especially mentioned, there is a double apparatus which can hardly be treated of in conjunction with the sense organs, because its sensory functions have not been sufficiently made out, and which will therefore rather be considered along with the brain and central nervous system. This apparatus is usually known under the name of the lateral organs. To it belong (a) superficial grooves or deeper slits situated on the integument near the tip of the head, (⅛) nerve lobes in immediate connexion with the nervous tissue of the brain, and (c) ciliated ducts penetrating into the latter and communicating with the former. Embryology slιows that originally these different parts are separately started, and only ultimately become united into one. Two lateral outgrowths of the foremost portion of the (esophagus, afterwards becoming constricted off, as well as two ingrowths from the epiblast, contribute towards its formation, at least as far as both Hoplo-and Schizonemertines are concerned. As to the Palaeonemcrtea, their embryology has not yet been studied, and in the most primitive genus, Carinella, we do not find any lateral organs answering to the description above given. What we do find is a slight transverse furrow on each side of the head, close to the tip, but the most careful examination of sections made through the tissues of the head and brain shows the absence of any further apparatus comparable to that described above. Only in one species, Carinella inexpectata, a step in advance has been made, in so far as in connexion with the furrow just mentioned, which is here also somewhat more complicated in its arrangement, a ciliated tube leads into the brain, there to end blindly amidst the nervecells. No other intermediate stages have as yet been noticed between this arrangement and that of the Schizonemertca, in which a separate posterior brain-lobe receives a similar ciliated canal, and in which the oesophageal outgrowths...