Observations on Feeding and Digestion in Glossobalanus minutus.

There are available at the present time several comprehensive accounts of the morphology of the Enteropneusta, ranging from the classical work of Spengel (1893) to the recent valuable monographs of van tier Horst (1927,1932), but, as the latter remarks, virtually nothing at all is known of the physiology of this group. It has therefore been the object of the present work to examine the feeding and digestion in Glossobalanus minutus Kowalevsky from an essentially functional point of view, in the hope of providing some comparison with Amphioxus and the other lower Chordata. In dealing with the structure of the alimentary canal, no attempt will be made to redescribe in detail the morphology of the several regions, as this has been sufficiently dealt with in the above works ; attention will mainly be confined to the new observations which have emerged from the present study.

The histological material was collected, and the work on the living animal carried out, during a visit to the Stazione Zoológica at Naples, and I wish to acknowledge my indebtedness to Dr. Dohrn and his staff for their unfailing assistance, and particularly for the care taken in collecting complete specimens of this very delicate animal. I am further indebted to the Government Grant Committee of the Royal Society for a grant in aid of expenses, and to the British Association for nominating me to occupy its table at Naples.

The Enteropneusta, with their well-known fragility, are unpromising material for physiological work, and with Glosso-halanus minutus the difficulty is enhanced by the small size which, together with the abundance of the mucus secreted over the surface, renders dissection unprofitable. Against this must be set its habitat, for at Naples it occurs amongst Z o stera detritus, and it is easier to obtain whole specimens than it would be if it had to be removed from sand. Owing to weather conditions I was unable to visit this particular ground, but the collector stated emphatically that the animals were taken from the loose detritus and not from the sand-burrows. For such animals, detritus would thus seem to be the most readily available source of food, a point of some interest in connexion with the nature of their feeding mechanism, as will be explained. Other specimens occur at Naples in Zostera roots, and I was present when some of these were collected. Here they were certainly covered by a thin layer of sand, and had taken some of this into their bodies, where it was easily seen through the body-wall.

Animals required for sectioning had first to be narcotized with cocaine, a process which with some specimens led to histolysis occurring before fixation. In extreme cases this can readily be recognized in sections, but in others there is necessarily a slight uncertainty as to whether or not the sometimes extensive breakdown of cells, involving much nuclear extrusion, is an artefact. Some good fixation was, however, obtained with Susa, Doboscq-Brasil, and Flemming-without-acetic, and it is therefore hoped that the histological account is an accurate record of the natural condition. The material was sectioned at Qp,, and stained with iron-haematoxylin, Mallory’s triple stain, safranin and light-green, mucicarmine, and by Feulgen’s method for chromatin.

For the experiments on enzymes, it was impossible to separate the gut from the body-wall, and the whole body was therefore used. The various regions were weighed, ground up in water with a little silver-sand, and left overnight, the extracts being usually made up to a strength of 2 per cent. The same standard methods were used as have been referred to in a previous paper (Barrington, 1937). The carbohydrases were estimated by the method of Hagedorn and Jensen, except for the maltase for which Barfoed’s reagent was used ; while proteolytic and lipo-clastic activity were estimated respectively by the formoltitration method, and by direct titration with standard caustic soda. The usual controls were set up in each experiment, and the pH was tested before and after incubation (which was at 35° C.) by means of the BDH capillator.

Van der Horst (1935) states that all Enteropneusta, ‘like so many worms’, are sand feeders, and it seems to be implied, at least by some writers, that this involves the taking in of sand as the animal burrows through it. Thus, according to Cori (1902), in Balanoglossus the collar seizes the sand during the burrowing process and transfers it to the mouth, there being no possibility of rejection, while Hanner (1904), quoting Kowalevsky, says that the animal is unable to close its mouth, and that as it burrows the sand which passes into the alimentary canal leaves it in a continuous column through the terminal anus. These views have perhaps been influenced by early beliefs that the Enteropneusta have affinities with worms. Spengel (1893) strongly advocated their Annelid relationships, and the idea finds an echo in the above quotation from van der Horst, although this author recognizes their Chordate affinities. At the present time it is natural to inquire more closely into the feeding mechanism, for, with their Chordate nature accepted, it might be expected that they would exhibit some parallel with the ciliary mechanisms which are a fundamental characteristic of the lower Chordates. It would, indeed, be remarkable if there were no indication of this. Some authors do seem to have suspected that the feeding mechanism was more than an automatic engulfing. Thus Stiasny (1910) describes Balanoglossus as extending a considerable length of its body, or more usually only its proboscis and collar, from its burrow at low tide and ‘sucking ‘the surface layer of sand. He found small Nemertines and Annelids in the gut, and considered that these could have been sucked in by the animal and not merely taken in passively with the sand. Ashheton (1908), in observations extending over a period of six months, never saw the animals come entirely out of the sand, but he noted that the proboscis was frequently protruded, especially during the night. It was often attenuated to a mere thread, sometimes waving and curling in the water, but more’ usually lying along the surface of the sand over which it moved, first in one direction and then in another. Ritter (1902) also found that Dolichoglossus pusillus protruded its proboscis at low tide.

My own observations have shown clearly that the secretions and ciliation of the proboscis and collar are important factors in both the burrowing and feeding processes. The glandular nature of the epidermis is well known. Spengel (1893) identified in it three types of cell, which he described respectively as large cells with an alveolar structure, smaller and thinner pear-shaped bodies with a faintly granular structure, and numerous clear cells in the form of elongated, empty bladders. I find in the epidermis of the proboscis (Text-fig. 1) large cells (muc.¹) with alveolated contents which stain a pale blue colour with aniline blue, and positively with mucicarmine. These, which evidently contain mucus, would correspond with Spengel’s first type. Mucus is also secreted by more slender cells (WIMC.²); these, which are particularly well seen on the collar, are possibly not significantly different from the preceding. - Spengel’s second type seems to be represented by cells containing conspicuous bodies which are sometimes pear-shaped (pb.), although their form and size are, in fact, quite variable. These inclusions have in part a hyaline appearance, but their upper region is often broken into granules (gb.), so that they are evidently a stage in the formation of a granular secretion. Indeed, occasional cells which are seen to contain rather large granules are probably the end-result of this process. In preparations, either the granules or the complete non-granulated bodies may be seen discharged from the cells ; Text-fig. 2 shows this extrusion (gb.) occurring in the epidermis of the trunk region. These bodies vary considerably in their staining reaction with Mallory’s stain, greyish-purple, blue or orange being amongst the possibilities. Some of these colours certainly represent different stages in the elaboration of the material, and this may well account for all the differences in reaction. Often the non-granular part of the body will be orange and the granular part blue or purple, but sometimes this arrangement is reversed. With mucicarmine the bodies stain either not at all or at most give a faim pink reaction, and the same occurs with Feulgen’s reagent, which in these preparations always stains mucus intensely. Probably the faint coloration is given by those bodies which would stain blue with Mallory. It is thus clear that these bodies are not composed of a simple mucus secretion. As for Spengel’s third type, I agree with van der Horst that these are probably the secretory cells which are empty after the discharge of one or other of the above inclusions.

The above description applies especially to the epidermis of the proboscis ; the epithelium is here thick and the gland-cells very abundant, the mucus cells being especially numerous towards the anterior end. The same types are also to be seen on the collar, but they are here differentiated into five zones, as described by Spengel. The second and fourth of these lie in furrows, easily seen in both living specimens (Text-fig. 5, fur.) and in sections (Text-fig. 3, fur.), and it was in these zones that he considered the gland-cells to be best developed. This is in a sense true, but it should be noted that the cells in question are mucus cells of the slender type, which are more closely concentrated in these furrows than anywhere else in the epidermis. The other characteristic secretory inclusions can be identified elsewhere on the collar, but on the whole the production of mucus seems to be its main function, although conditions in different animals are so variable that it is difficult, to generalize. The collar is covered with cilia which in sections are often particularly dense in and immediately posteriorly to the hinder furrow (fourth zone), and around the anterior edge of the collar. This variation in density would not attract much attention in preparations, but living specimens show a very pronounced ciliary beat in the hinder furrow. The functional significance of this regional specialization is not quite clear, but it is a fact that when granules are dropped on to an animal they travel very quickly over the proboscis and collar, and tend to collect at the hind end of the latter (i.e. behind the hinder furrow) before passing down the trunk region, over which movement is much more sluggish and, indeed, results more from the animal crawling out of the mucus tube which it has thus secreted.

The same types of gland-cells are to be seen on the trunk, and the chief characteristic here is that they are collected into rings (Text-fig. 4, gl.) which are separated by non-glandular zones (ngl.) in which the epidermal cells are much less deep. Willey (1894) interpreted this as metameric segmentation, although others, including Spengel, denied this. The matter has recently been discussed by van der Horst (1930), who takes the view that metameric segmentation in the Chordata is fundamentally mesodermal and correlated with the locomotor mechanism, and that therefore the repetition in the Enteropneusta of these rings, as also of the gills and gonads, must be considered as pseudo-metameric. He agrees, however, with the view that metamerism in the Vertebrata has as its distant predecessor the pseudo-metamerism of the Enteropneusta, and I feel that this point has not received sufficient emphasis. In particular, previous writers have ignored the fact that these glandular rings are associated with rings of cilia, which, although not at all conspicuous in sections (Text-fig. 4, cil.), where they are often interrupted by the breaking-down of the epidermis, are very clear in the living animal, particularly in the intestinal region. As may be seen from the figure, a band of dense, short cilia covers part of the glandular ring, while on the non-glandular zones the cilia are only very rarely found. The existence of such ciliated rings, which in a group like the Archiannelida would be considered as evidence of a metameric organization, seems to emphasize the innate tendency of the Enteropneusta towards metamerism, a tendency which fails to reach its full development because the method of locomotion has not conditioned the segmentation of the mesoderm.

From the above account it may be seen that the epidermis is very glandular, and although previous authors have naturally noted this, sufficient importance has hardly been attached to the variety of secretion produced. It is desirable, then, to emphasize that while some of the secretion is certainly mucus, and is doubtless used in lining the burrow, not all of it can be accounted for in this way, some further function being implied. It is therefore of significance that I find definite evidence for the production of an amylase by the epidermis of the proboscis. At an early stage of this work it was noticed that an extract of the proboscis displayed very strong amyloclastic activity, and further tests (which will be described later, p. 256) showed that the external secretion of the organ contained an amylase. The full significance of this will be appreciated when the feeding mechanism and the action of the pharynx are described (p. 244), but it is referred to here because it does provide some explanation of the variety of inclusions in the epidermal cells.

An examination of a specimen of Glossobalanus in a suspension of carmine shows that the granules which alight upon the proboscis (Text-fig. 5, gra.) are immediately caught up in the abundant secretion and are carried rapidly backwards by ciliary action. Some of this material, consisting of strands of secretion and carmine, is drawn under the edge of the collar (x) and thus into the mouth, for a powerful current sets into the latter as a result of the pharyngeal ciliation, aided, no doubt, by the dense cilia already described on the anterior edge of the collar, although it should be added that it was not possible in living specimens to detect much tendency for movement of material to take place over the internal wall of the collar. This latter organ, however, is by no means inactive. Its anterior end is well supplied with muscle, and in life is very mobile, and it can be seen to respond to the nature of the particles which touch it. On the arrival of large particles it folds up against the neck of the proboscis, thus preventing their passage into the mouth, and after a short pause they are gradually drawn away backwards (y) by the action of the cilia of the collar. It is clear, then, that far from the collar seizing material, as Cori (1902) described, it acts as an important sifting mechanism. At its anterior edge the secretion and granules come under the conflicting influence of the respiratory current and the ciliation of the outer surface of the collar (z), and the factor determining their eventual course will often be the movement of the lip of the collar.

Particles may also enter the mouth in another way, for, as a result of the strong respiratory current set up in the water, material may be drawn in (w) without alighting on the proboscis or being caught in the mucus. This is rendered possible because the mucus which is passing backwards over the animal’s body does not form a continuous tube, but consists merely-of isolated strands, between which free particles can easily pass.

It can thus be seen, and the point will be further discussed below, that Glossobalanus possesses the equipment necessary for ciliary feeding, together with a mechanism for rejecting large particles, and it is certainly incorrect to assume that its nutrition will involve merely the automatic intake of sand as it burrows. Indeed, the ciliary mechanism plays an important part in burrowing, as can be seen by placing a specimen in a dish with a small heap of sand. As soon as its proboscis touches the latter, the sand is at once agitated by the cilia and, mixed with secretion, passed backwards over the animal so that the latter is quickly surrounded by a tube of sand and mucus. Into the space thus formed the proboscis is forced, and it is the expansion and contraction of this organ, together with the displacement of the sand by the cilia, which are responsible for the burrowing. This is in general agreement with Bitter’s (1902) conclusions, according to which the proboscis is much the most vigorous region during locomotion, being driven forwards partly by its cilia and to a less extent by contraction of its circular muscle-fibres, and causing a ‘perfect flood of sand grains’ to pass backwards. Within the burrow it acts as a holdfast, the whole body being drawn after it by contraction of the longitudinal musculature. Some authors have described a swelling of the collar as an important factor, the organ acting as a point of friction. My observations show that this is not so, at least when the sand forms only a small and loosely packed heap; under these circumstances the collar remains more or less passive, moving forwards underneath the stream of mucus and sand which is passing back from the proboscis, and while some sand certainly enters the alimentary canal (see below), being visible through the body-wall, the amount which does so is small relatively to the volume which the animal displaces. At the same time, the collar undoubtedly can expand and contract; violent movements of this type are seen, for example, if a proboscis and collar are amputated from the trunk, and it may well be that such movements become of importance in tightly packed masses of sand. It is not, of course, suggested here that the Enteropneusta do not ‘swallow’ sand, for the well-known existence of their sand-castings is a clear demonstration’ that they do. The point to be emphasized, and which will be returned to below, is that there is the additional method of taking up material by ciliary, action, either by agitating the surface of the sand, or by utilizing suspended material.

This region, which is well ciliated, is a short, broad tube leading from the mouth through the collar, and there is little to add here to previous descriptions. The lining epithelium is very deep, consisting of long, slender cells with little visible contents (Text-fig. 6). Near the free border of the epithelium are to be seen granular nuclei, while others are distributed lower down (nuc.). Secretory cells are quite abundant, mucus being recognizable in goblet-shaped cells (muc.) with slender processes which, according to van der Horst (1927) extend down towards the basement membrane of the epithelium. Also visible are occasional spherical or pear-shaped inclusions (pb.) resembling those already described in the epidermis.

As is well known, this region, which follows immediately upon the buccal cavity, is divided into dorsal and ventral chambers (Text-fig. 7, de., vc.) of which the former (the ‘Kiemendarm ‘of German authors) bears the gill-shts, while the latter is held to be a food canal (‘Nahrungsdarm’), although there seems to be no record of any attempt to determine the physiological validity of this distinction.

The gill-slits are separated by a regular alternation of primary and secondary arches (Text-fig. 8, pa., sa.) which differ appreciably in their structure. Most of the anterior and posterior face of each primary arch is covered by a deep epithelium which is densely ciliated. These cilia may be called, in accordance with the now usual nomenclature, the lateral cilia (Ze.) and are no doubt responsible for the maintenance of the current of water. The inner edge (in.) of each primary arch is covered by a shallow epithelium on which cilia are often not distinguishable at all ; occasionally, a few short cilia may be made out on the inner face of the arch, and these may be called the frontal cilia, although they are too sporadic to be of any functional significance. The secondary arches are much larger than the primaries ; they have, in addition to a coelomic canal (coe.), a double gill-bar (gib.) instead of the single one present in the primaries. The anterior and posterior face of each arch bears a deep epithelium with dense lateral cilia and on the inner side of this there is a non-ciliated area covered by a shallow epithelium, corresponding to the inner edge of the primary arch. In addition, each secondary arch extends farther into the lumen of the pharynx than do the primaries, and this extension is covered by a deep epithelium. This bears a covering of short but clearly recognizable frontal cilia (/c.), which are associated with mucus cells. It may be added that forming the mid-dorsal line of the roof of the pharynx, between the two rows of gill-slits, is a median strip of ciliated tissue (the ‘Epibranchialstreifen’) in which slender mucus cells are abundant (Text-fig. 7, es.).

It has not been possible to examine directly the action of the cilia of the pharynx, for if attempts are made to expose this region any ciliary currents are at once obscured not only by the distortion of the very delicate tissues but especially by the action of the cilia and mucus of the epidermis. It does seem, however, that if the pharynx were merely concerned with the maintenance of a respiratory current the lateral cilia would be adequate, while the presence of frontal cilia on the secondary arches, and of the abundant mucus cells, recalls the situation in Amphioxus, where, of course, the gill-arches are concerned also with the trapping of food particles. In other words, the distinction generally drawn in the Enteropneusta, apparently on purely morphological grounds, between a dorsal respiratory chamber and a ventral food chamber is not quite convincing, and observations upon specimens placed in a carmine suspension very much strengthen this doubt.

Since, as has been shown, carmine particles are drawn into the mouth by the respiratory current, they might be expected to enter the dorsal chamber, and it is easy to see that they actually do so. The body-wall is not sufficiently transparent for it to be possible to follow all that occurs inside, but strands of particles and secretion can clearly be seen through the open gill-pores to be passing rapidly backwards down the chamber. Many free particles are also swept out through the gill-pores with the ejected water. At the same time some material passes into the ventral chamber ; progress here is much more sluggish than in the dorsal chamber, for the respiratory current is absent and movement depends on the relatively weak ciliation of the wall, aided, perhaps, by some degree of muscular action. Thus the material in this chamber collects into a bulky mass, which moves very slowly backwards. Rough tests suggest that the weight of the particles is a factor in determining which course they shall take inside the pharynx, for sand tends to accumulate to a much greater extent in the ventral chamber than does carmine. Quantity may also be expected to be a factor in increasing the amount which enters the ventral chamber, and sand certainly accumulates there during the burrowing process.

From the above observations it is probably still open to argument that the passage of food into the respiratory chamber is an accidental by-product of the respiratory mechanism and is of no physiological significance, but further facts rather contradict this. If an animal is examined after it has been for some fifteen minutes in a carmine suspension, it can be seen that some of the carmine which has passed down the pharynx has become compacted into a small solid mass (Text-fig. 9, m.) situated dorsally just behind the last gill-pore. This is only readily observed when the animal is fully extended, the mass at other times being concealed by the greater thickness of the body-wall. It can be further seen that this mass is in regular rotation, evidently under ciliary action, the axis of rotation varying between the vertical and the horizontal according to the degree of extension or folding of this region. Now it seems always to have been known that in the Ptychoderidae, to which family Glossobalanus belongs, the dorsal chamber is prolonged backwards behind the last gill-slit to form a blind pouch, which can be seen from the outside as a dorsal prominence (Text-fig. 9, pbc.). This ‘Postbranchiale Kiemendarm’ or post-branchial chamber, as it may be called, is shown in transverse section in Text-fig. 10, in which it appears as a vertically elongated region (pbc.) with very thick walls, by the occlusion of which its lumen can be virtually cut off from the ventral chamber (vc.). It is also seen in sagittal section in Text-fig. 11 (pbc.), which shows its relationships with the dorsal chamber (de.), ventral chamber (vc.), and oesophagus (oes.). It is of small size, extending in one specimen over a length of about 270p, although it must be remembered that this is after the very considerable contraction which takes place on fixation ; it is probably better developed in other species, for Spengel (1893) states that it is little developed in Glossobalanus minutus. Its epithelium, which is strongly ciliated, contains many secretory cells, some of which seem to be identical with those described above as occurring in the epidermis. Thus, in addition to mucus cells, there may be distinguished many pear-shaped bodies, staining usually blue with Mallory’s stain, and not staining at all with mucicarmine, and also occasional groups of orange-staining granules. In addition, there are a number of very slender cells charged with fine granules which stain either blue or orange-red with Mallory’s stain, and which also take up haematoxylin ; cells of this type have not been noted in the epidermis. There is thus no doubt of the secretory activity of this post-branchial chamber, and it may be assumed that mucus and digestive secretions are produced here; the resemblances which some of the inclusions bear to the epidermal ones are, of course, particularly significant in view of the secretion of an amylase by the latter tissue.

Previous workers have dismissed this chamber as of little importance, and apart from Maser’s suggestion (1913) that it might serve as a safety-valve in the event of a back pressure being exerted upon the food-material in the intestine, have paid no attention to its possible function. It is clear, however, that it is in the post-branchial chamber that the rotating material is situated, although only occasionally is it seen there in sections (Text-fig. 11, m.) as it is usually extruded into the oesophagus during narcotization.

It is a little difficult to estimate the importance of this rotation in digestion, but it is of interest to compare the situation with that in Amphioxus, for in that animal I have shown (1937) that a rotation of the food mass in the mid-gut is the central feature of the digestive mechanism, serving to mix the food with the secretion. It is not, of course, suggested that a rotation of this sort is peculiar to the Chordates, and in any event the conditions in Glossohalanus are hardly the same, for here the rotation takes place immediately behind the pharynx and in front of the ‘hepatic ‘region which, as will be shown, must be of importance in secretion. It may be suggested, however, that in Glossohalanus the post-branchial chamber provides a region in which food can be retained while much sand is being passed back ventrally, and there is the possibility of secretions being added to the food from the epithelium.

There is, however, a point of further interest. It has already been explained that the proboscis secretes an amylase, and it can now be seen that this will be carried with the food into the mouth and on into the respiratory chamber. Some of this material will collect in the post-branchial chamber, and it is thus apparent that the rotation may well serve to compact the food more closely with this enzyme, and thus facilitate digestion. To this extent, then, there is a parallel with Amphioxus, but again it is impossible to say how important this rotation is for Glossohalanus. A few tests gave no evidence for the secretion of other enzymes by the proboscis, and on the other hand, the general similarity of the gland-cells in the different regions of the epidermis does not suggest that the production of the amylase is confined to that organ. It is certain that the existence of a post-branchial chamber is not essential for digestion in the Enteropneusta, for this region is absent in the Spengelidae and the Harrimanidae (van der Horst, 1927). It is significant that in these latter forms the next region of the alimentary canal, the oesophagus, is much more elaborate in structure than it is in the Ptychoderidae, but unfortunately nothing is yet known of the physiology of digestion in them.

There is little to be said concerning the ventral chamber of the pharynx. The epithelium, which is well ciliated, is here very shallow, and apart from the production of some mucus there is little sign of secretion, although occasional rounded bodies can be seen resembling those described above in the buccal cavity and in the epidermis. From its structure, this region would seem to be concerned mainly with the transport of the material rather than with the addition of secretions to it.

The above observations thus seem to show that the dorsal and ventral chambers of the pharynx cannot on physiological grounds be divided into respiratory and food chambers respectively. On the contrary, the dorsal chamber serves both for respiration and the passage of food, while it might be argued that the ventral chamber is primarily a channel for the passage of excess sand, although no doubt organic material in this can be utilized in its further passage through the gut. It is of interest to compare the situation with that found in the other lower Chordata. Microphagy is clearly primitive in the Chordata, ranging from the lophophoral mechanism of C e p h a 1 o -discus to the pharyngeal mechanism of the Tunicata, and Garstang (1929) has explained how the Enteropneusta and Amphioxus could be regarded as illustrating the assumption of free life and burrowing habits at two different evolutionary levels. In Amphioxus microphagous feeding is, of course, retained, and the present work shows that the same is broadly true of Glossohalanus except that here the situation is complicated by the habit of excavating sand, which must lead to the utilization of the organic matter swallowed with it. The important point is, however, that the ciliation of the proboscis and pharynx, combined with the action of the collar, confer on Glossohalanus the additional capacity for utilizing organic matter in suspension, or taking it up from over the surface of the sand, this doubtless constituting the physiological basis of the ‘sucking’ action referred to by Stiasny (see above, p. 229), and it seems safe to assume that this is a retention, with modification, of the primitive microphagous mechanism of the Chordata. As for the structure of the pharynx, and particularly the presence of frontal cilia, this seems to foreshadow the elaboration found in Amphioxus. Indeed, the function of these cilia deserves examination in a larger and more convenient species. I have no observations to record on the feeding habits of Glossohalanus under natural conditions, although it is reasonable to assume that when living amongst loose detritus it would utilize suspended or deposited material; detritus particles are certainly taken into the gut under laboratory conditions. In this connexion, however, the observations of three authors, referred to above (p. 229), on the protrusion of the proboscis and collar, strongly suggest the use of the ciliary mechanism for feeding, essentially analogous to the protrusion by Amphioxus of the oral hood from the gravel in which it burrows. At the same time, it might be argued that this protrusion would facilitate respiration, and it is clear that more observations on the feeding habits of other genera from this point of view are very desirable.

This name is applied to the length of the gut which connects the pharynx with the ‘hepatic’ region; it is little developed in immature specimens, where the first ‘liver-sac’ follows closely upon the last gill-pore (see, for example, Text-fig. 9, Is.). It is abruptly marked off from the ventral chamber of the pharynx by the form of its epithelium. The cells, which, as in most parts of the gut, have a prominent striated border (Text-fig. 12, sb.), are much deeper and, being more slender, are more closely massed together, so that the nuclear region appears darker. These cells contain a number of small spherical inclusions (inc.) staining blue with aniline blue; tests with mucicarmine show that these are not mucus, which is produced in typical goblet-cells (muc.) quite distinct from the columnar cells. It seems likely that these inclusions would represent a digestive secretion ; the material is extruded in cytoplasmic spheres (ext.) from the cells, and often a body (nuc.) staining intensely with acid fuchsin is included in these. This body is shown by its staining reactions to be a nucleus, and a similar extrusion of nuclei has been noted in one specimen or another in many parts of the gut, being, therefore, commonly associated with secretion in this animal. These densely staining nuclei here seem to be derived from the more typical granular nuclei by a condensation of chromatin, intermediate stages, with part of the nucleus staining more densely than the rest, being visible in Text-fig. 12.

In the oesophagus the cilia are much less well developed than further forward. This applies throughout the gut until near the anus, where a strong ciliation again develops. A strongly ciliated groove, described by earlier writers, begins in the hinder region of the oesophagus and extends backwards throughout most of the gut, running on the left wall. I have nothing to add with regard to its special function, although it will clearly be of value in maintaining a through passage when the gut is filled with material. It is, however, of interest to note here a parallel with Amphioxus, for I have shown that in that animal the ciliation through much of the intestine (hind-gut) is weak, except for a strongly ciliated groove, and that strong ciliation develops in the neighbourhood of the anus.

It may be concluded from the above that the oesophagus is of physiological importance in producing a characteristic secretion.

As van der Horst rightly points out, there is no reason to suppose that this region has any close affinity with the liver of the Vertebrata, and the term ‘hepatic’ is altogether unfortunate, as it is also in Amphioxus (Barrington, 1937), a point which is sufficiently demonstrated by the histological structure of this region. In the Ptychoderidae and in Schizocardium (Spengelidae) this part of the gut bears conspicuous diverticula, the ‘liver-sacs’, which create corresponding diverticula of the epidermis. In Glossobalanus minutus these are regularly paired (Text-fig. 9, Is.), wide laterally, and compressed antero-posteriorly, and the anterior ones exhibit, a brown colour. A similar or green colour distinguishes this region even in those forms (most of the Spengelidae, and the Harrimanidae) in which sacs are not found.

The epithelium of the sacs is deeper than that of the intestine from which they arise. Previous investigators have said little about the histology of this region, except to refer to green or golden-brown secretory granules which are found in the cells of the diverticula but not in those of the gut itself (e.g. Spengel, 1873). I find, however, that the ‘hepatic’ region can be divided rather clearly into three sections. The most anterior of these, extending over perhaps the first twelve pairs of diverticula (the number naturally varies with the size of the animal), is characterized by the brown colour already referred to, this resulting from the presence within the cells of numerous golden-brown vesicular inclusions (Text-fig. 13, bin.), which may be distributed throughout the length of the cells, even down to below the level of the nuclei ; sometimes as many as three or four are visible within a single vacuole (vac.).

These inclusions take up haematoxylin, but not very readily, often only the outer rim of each being stained. They are extruded from the cells, several at a time, in spherules of cytoplasm. There is also to be seen in these anterior sacs a second type of inclusion (Text-fig. 14, cun.) in the form of granules which are mainly confined to the upper half of the cells and which lack the brown colour, but which are also extruded in spherules of cytoplasm (ext.) ; nuclei are sometimes also extruded (Text-fig. 15, nuc.) either with or without the inclusions. In addition to the lack oi colour, the second type is distinguishable from the first by its staining reactions; it is stained by alum-carmine, and is stained green-blue by pyronine-methylgreen, the first type taking up neither of these stains, while it also is intensely stained by haematoxylin. Further, this second type is especially characteristic of the third (hindmost) section of the ‘hepatic’ region, occurring there in abundance, while the brown inclusions are confined to the first section.

It seems safe to assume that these colourless inclusions represent zymogen granules, and the same may possibly be true of the brown inclusions, but this colour must have some physiological significance, and at the moment there is no evidence as to what this is.

In the second (middle) of the three sections of the ‘hepatic’ region, again extending over perhaps some twelve pairs of sacs, there are found conspicuous intracellular bodies of a rather peculiar nature. The general appearance of the epithelium is shown in Text-fig. 16, and, as may be judged, these bodies (ov.) form one of the most striking features of the gut epithelium, although they seem not to have been previously described. At their maximum development (Text-fig. 17, ov.) they are ovoid in form, with a boundary staining deeply with the aniline blue of Mallory’s stain. Their contents are best seen in material fixed in Flemming-without-acetic, and stained in safranin and light-green. In the centre of each is a small compact mass consisting of a large granule (gra.) with often some smaller granules, and other small granules can be distinguished elsewhere. These granules stain intensely with safranin, as also with acid fuchsin and haematoxylin, while the general ground substance, apart from a central light area, takes on a rose-pink colour with safranin, and has a rather reticulate appearance. The bodies are situated inside the epithelial cells, and they can be traced back to very small vacuoles (vac.) which appear at the distal end of the cells. The contents of each vacuole are grey in haematoxylin preparations, and at the centre of each is a minute granule surrounded by a clear area. These vacuoles gradually enlarge (ovd.) to form the fully developed structures. During this enlargement they acquire the dense v all, staining blue with aniline blue and faintly red with mucicarmine, and therefore presumably incorporating some mucus, while the central granule enlarges considerably. In the safranin preparations the central granule stains red, while the rest of the original vacuole, outside the clear area, takes up the light-green. During the enlargement this reaction changes and the mature body acquires the rose-pink tinge, the granules continuing to stain red while the central unstained area persists. The acquisition of the pink colour suggests some chemical change accompanying the enlargement. To some extent the staining reaction of the granules resembles that of chromatin, but it seems clear, nevertheless, that they are not chromatinic, for they do not stain with alumcarmine as do nuclei, and, more precisely, they are negative to the Feulgen reaction ; indeed, their whole behaviour within the cell is suggestive rather of some material being laid down within it. They are finally extruded from the cells, and many can be seen in the lumen of the gut, together also with the smaller developmental stages. It is not clear how far the extrusion of the latter is an artefact, but it would certainly seem to be encouraged by the curious way in which the bodies develop at the apex of the cells. It is possible that the very small vacuoles with their central granule arise from the rather vague granulation of the cytoplasm which can often be seen in these cells, but this does not alter the fact that most of the development of the bodies takes place at the distal end.

In considering the possible significance of these bodies, one point of interest is that after extrusion from the cells they do not break up, but on the contrary can be identified in sections, apparently unchanged, in the lumen of the intestine even at its extreme hind end. It may be concluded, therefore, that they are passed to the outside, and this certainly suggests the possibility that they may have an excretory function. Excretion in the Enteropneusta is commonly attributed on rather scanty evidence to the glomerulus, to certain granular cells of the coelomic epithelium of the proboscis, and to other cells which arise in the proboscis and fall into the coelom, the products in all cases passing out of the body through the proboscis pore. The physiology of excretion in this group seems never to have been thoroughly investigated, however, and the existence of some additional excretory mechanism would not be surprising, while the co-operation of the intestinal epithelium in excretion is well established for other groups of animals. At present, unfortunately, I have no evidence for the accumulation of material such as indian ink within the body, but this is inevitable since the specimens were primarily fed with such material in order to provide evidence for absorption, and, as will be seen, this proved unsatisfactory. An excretory function for these bodies cannot, then, at present be proved, and it is desirable in the meantime that a search for these intracellular bodies should be made in other members of the group, and particularly in those which do not possess ‘liver-sacs ‘. It might also be possible to investigate their contents by histochemical methods.

These are the three main types of inclusion found in the ‘liver-sacs’, but certain others require mention. The first of these, an intracellular body (Text-fig. 18, int.), occurring in a very limited area at the hind end of the ‘liver’ region, seems at first sight to resemble somewhat the ovoid bodies just described, but differs in that the wall does not stain a conspicuous blue colour with Mallory’s stain, and in the fact that the contents, in addition to including some conspicuous non-chromatin granules, are of a denser nature and include two vacuole-like spaces. Moreover, these bodies occur mostly at the base of the cells in which they are contained (Text-fig. 18), instead of at their distal end, and it is not possible to see from what they take their origin.

There are also to he seen in the epithelium large bodies with a number of regular spherical inclusions arranged mostly around the periphery, these latter inclusions staining with Feulgen, at least over their surface (Text-fig. 19, bg.). This reaction implies that they are composed of chromatin but is not quite convincing, for, since the Feulgen reagent regularly stains mucus in these preparations as well as chromatin, a negative reaction with it is more significant than a positive one. These bodies occur in the same limited area as the preceding, but may also be seen farther back in the intestinal epithelium. The significance of both of these types of body is quite obscure, although it is conceivable that they may be stages of intestinal parasites.

Finally, there are found in the same region numerous inclusions of an irregular shape, commonly of a yellow colour in Mallory preparations (Text-fig. 19, ir.). Similar masses of material can be identified in the cells of the coelom and also in the epidermis ; they seem to represent some form of accumulation, possibly associated with excretion, although there is no definite evidence for this.

The ‘liver-sacs ‘decrease in size behind, and merge gradually into the intestine of the abdominal region without any sharp demarcation. The intestinal epithelium here, and in the region of the sacs, is relatively shallow, and becomes increasingly so towards the anus. Mucus is produced by it, particularly in the ‘hepatic’ region, where numerous goblet-cells are a conspicuous feature. Some cells contain secretory granules which resemble in their staining reactions the colourless granules of the sacs which have been described above, and which are extruded in cytoplasm. This latter process is essentially like that described above for the oesophagus, except that the extruded secretion differs in its appearance from the blue-staining vesicular inclusions characteristic of the latter region.

It has been stated by earlier workers that the food material does not enter the ‘liver-sacs’, and I have found nothing to conflict with this. Absorption might therefore be expected to occur in the intestinal epithelium, but unfortunately although a number of specimens were ‘fed’ with such material as Indian ink, carmine, iron saccharate and colloidal gold, substances which had been found to give good results with Amphioxus, no certain evidence of absorption could be obtained. The animals were all treated in sea-water containing a suspension of the material, without the presence of sand, and it may well be that in the absence of the latter around their body the animals will not react satisfactorily; Morgan (1894), however, found it impossible to obtain evidence for the absorption of carmine, even when this substance was mixed with the sand on which the animals were resting.

From the above account it can be seen that in the post-oesophageal region of the alimentary canal the ‘liver-sacs ‘constitute a centre of great physiological importance, their cells being charged with inclusions of three predominant types. The function of these is to some extent obscure, but at least one type may be considered as a digestive secretion, and the sacs may possibly be concerned also, in excretion. The misleading use of the terms ‘liver ‘and ‘hepatic ‘in this connexion hardly requires emphasis; at the same time, the peculiar features of the epithelium of these sacs makes it unwise at this stage to draw any comparison between them and the well-known diverticulum of Amphioxus and certain Tunicata, if, indeed, such a comparison could be justified between two rather different levels of organization. As for the intestinal epithelium, this is concerned in secretion, although never to the marked extent of the epithelium of the sacs, and it becomes progressively reduced in importance towards the hind end; further work may show, however, that it is important in absorption.

As has been explained earlier, all the enzyme experiments had to be carried out on extracts of the whole body, and not of the alimentary canal alone ; they were restricted to a determination of the main types of enzymes present and of their pH range. For most of the experiments, extracts of the ‘hepatic ‘region were used, taken often from incomplete specimens, most of the complete ones being required for histological purposes.

An amylase was readily identified in extracts of all parts of the body, and some pH curves are shown in Text-fig. 20. The optimum point is probably a little above pH 6 · 5, although there was some variation between the different curves obtained, and often they were too flat to indicate the optimum with any precision.

An early experiment in this series showed that extracts of the proboscis possessed strong amyloclastic activity, and in view of the rotation of material which was later discovered to occur in the post-branchial chamber (p. 242) this property was investigated further. Some sand was thoroughly boiled and then dropped gently upon the proboscis of an animal at rest in clear water ; it was at once caught up in the secretion produced by that organ and passed back over the collar. As soon as it reached the latter it was picked off with forceps, entangled with secretion, and the material shaken with a little distilled water. In this way a solution of the external secretion of the proboscis was obtained, and this was found by the usual tests to have a very potent amyloclastic activity, as may be seen from Textfig. 20, in which the pH range is shown. (The solution would, of course, contain also some secretion from the collar, but this does not affect the argument, as the preliminary tests had shown that the proboscis by itself was active.) The significance of this external secretion of an amylase in the feeding process has been discussed above (p. 244) ; it means, of course, that the fact that extracts of all parts of the body will digest starch does not necessarily imply that the enzyme is produced by all parts of the gut, since the activity may result from the epidermal secretion either from that part of the body, or carried backwards from the anterior end.

Further tests with extracts of the ‘fiver-sac’ region showed that such extracts would readily digest glycogen and sucrose, but not raffinose or inulin, while a weak maltase was also present (Table 1). The other parts of the body were not tested with these substrates.

Digestion of carbohydrates by extracts of the ‘hepatic’ region.

A weak protease was identified in extracts of the ‘hepatic’ region, using casein as substrate. As may be seen from Text-fig. 21, the optimum activity is probably between pH 7·5 and 8, a result similar to that obtained for the protease of Amphioxus (Barrington, 1937). A few tests of proboscis extracts at this pH failed to give any indication of proteolytic activity.

Using triacetin as substrate, lipoclastic activity could be identified in extracts of the ‘hepatic’ region, the optimum activity lying perhaps near pH 6-5 (Text-fig. 22), which again agrees closely with the lipase of Amphioxus, although, as may be judged from the two curves shown, no very precise point could, be established. Extracts of the proboscis showed no significant lipoclastic activity.

bg., body with granular inclusions; bin., brown inclusions; cil., cilia; cun., colourless inclusions; coe., coelom; col., collar; epb., epithelium of buccal cavity; epc., epidermis of collar; es., ‘Epi-branchialstreifen’; ext., extruded material; fc., frontal cilia; fur., furrow; gb., granules derived from intracellular body; gib., gill-bar ; gl., glandular zone of trunk epidermis ; gp., gill-pore ; gra., granules; in., inner edge of primary gill-arch; inc., inclusions; ini., intracellular body; ir., irregular inclusions; lc., lateral cilia; Is., ‘liver-sac’; m., material in post-branchial chamber; mit., mitosis figure ; muc., muc.¹, muc?, mucus cell; ngl., non-glandular zone of trunk epidermis; nuc., nucleus; oes., oesophagus; ov., ovoid intracellular bodies ; ovd., developmental stage of the ovoid bodies; pa., primary gill-arch; pb., ‘pear-shaped’ intracellular bodies; pbc., post-branchial chamber; prob., proboscis; sa., secondary gill-arch; sb., striated border; sec., secretion; vac., vacuole; vc., ventral chamber of pharynx; w, x, y, z, reference points.