The Contractility and Healing Behaviour of Pieces of Leucosolenia complicata

During an investigation of the mechanism of orientation of calcareous sponge spicules a number of experiments were attempted in which part of the wall of an oscular tube of Leucosolenia variabilis was removed and re-placed in the reverse orientation. Some of the attempts were successful and indicated that the newly forming spicules were not obliged to copy the orientation of their older neighbours, but in other cases the piece fell away, whereupon both the piece and the mutilated tube underwent a process of healing by which the functional tubular form was regained. Since the process invariably involved a curling of the wall, a study was made of the healing behaviour to see whether it would throw light on the contractile properties of the sponge.

The contractility of calcareous sponges, in particular Clathrina coriácea, has been known for many years. The pores and oscula can be closed and the whole sponge can undergo a general process of contraction. Bidder (1898) believed that this was simply a recovery from the stretched condition after the cessation of activity of the collar-cells ; he thought that normally the internal hydrostatic pressure gave rise to a state of tension in both internal and external surfaces of the sponge. However, Minchin (1900) states that ‘the flat epithelium covering the exterior is responsible for the general contraction of the whole body’, and that ‘the closure of the osculum is effected more especially by … cells … which line the oscular rim’. The latter cells often give rise to ‘a special contractile apparatus, such as a ring-like sphincter, or a contractile sieve-membrane’, and, as Minchin points out, this suggests that any contraction in the vicinity is more than an elastic recoil. The sieve-membrane of C. coriácea is certainly contractile, since a tube can become constricted in this region, without either the oscular edge itself, or the rest of the tube, becoming much contracted. Such a sieve-membrane is not to be seen in Leucosolenia complicala, but a diaphragm may be present as a trilaminate shelf around the inner surface of the oscular rim (Jones, 19546). Furthermore, the osculum of this species has been observed to contract in bright light, the area of contraction passing as a wave around the oscular rim. Minchin (1908), however, states that ‘the species of Leucosolenia are very slightly, if at all, contractile’.

The experiments described in this paper also have some relevance to the problem of whether or not there is a nervous system in Leucosolenia. Recently Tuzet, Loubatières, and Pavans de Ceccatty (1952) have affirmed their belief in the existence of a nervous system in sponges, including L. botryoides (Pavans de Ceccatty, 1955), as a result of their studies of stained sections. Such a nervous system, if present, should be capable of narcotization, and for this reason an investigation has been made of the healing behaviour of pieces of L. complicala in a solution containing equal parts of sea-water and isotonic magnesium chloride. This solution narcotizes the neuromuscular system of Metridium senile after about 1 hour (Batham and Pantin, 1951).

The healing behaviour of three species of Leucosolenia has been studied, namely L. complicala, L. botryoides, and L. variabilis (Minchin, 1904), but most of the work has been done with L. complicala. Specimens were kept under running sea-water for several days or weeks and then transferred to the laboratory, where some of the oscular tubes were removed. From these, pieces were derived by bisecting the tubes longitudinally with fine scissors. The pieces were then placed in Petri dishes containing 40 ml of sea-water at room temperature (usually 15-20° C), and kept under low-power observation.

Measurements were made by means of a squared eyepiece-micrometer. The longitudinal bisection of the tube afforded a control half for experimental purposes when necessary.

Some pieces were fixed in osmium tetroxide dissolved in sea-water at different stages of the healing processes, and stained with picrocarmine (Minchin, 1898).

Other methods are mentioned below.

Pieces obtained by the longitudinal bisection of the oscular tubes of L. complicala undergo in sea-water a process of repair and reorganization which can be divided into three main phases. The first phase involves the curling up of the piece in the longitudinal and transverse planes, so that the choanoderm comes to line the cavity enclosed by the resulting canoe-shaped structure (fig. i, B). The second phase begins with the formation at each end of the piece of a ‘healing membrane’, consisting of two cellular layers with a thin layer of mesogloea in between. It exhibits contractile properties, for it draws the edges of the original wall closer together and spreads across the gap, sealing off the cavity until a complete tube has been formed (fig. i c). The final phase then follows with the growth of the tube into the normal form of the species.

The first phase of curling invariably results in the partial enclosure of the choanoderm and the original spongocoel, and takes place regardless of whether the piece is left undisturbed with the choanoderm facing upwards or downwards. It also occurs with pieces derived from the base of a tube (‘basal pieces’), without a portion of the diaphragm, or inturned porocyte epithelium of the oscular rim. This suggests that the choanoderm is contractile, the curling up being analogous to the curling of a photographic print when the gelatine is shrunk by drying.

The process of curling has been followed quantitatively by measuring the length and the distance from edge to edge at each end and in the middle of the piece. As can be seen from piece A (table i), the length may decrease as the piece curls upwards, but this decrease is at first probably only an apparent one, the ends being drawn closer together with the increase in curvature. In fact in several cases (e.g. piece B, table i) an increase in the apparent length was detected, and also in the true length, measured directly on pieces lying on their sides, which indicates that the process of curling has been accompanied by a longitudinal extension of the supporting mesogloea.

Sometimes the edges may separate in the middle of the piece while at the ends they are moving inwards (piece B, table 1). This partial flattening in the middle is probably a mechanical effect necessitated by the longitudinal curling in such cases. There may also be a slight separation of the edges at the oscular end of ‘oscular pieces’ (having a portion of the oscular rim), or the process of incurling at this end may be delayed. This was observed in 4 out of 7 pieces measured, and in the remaining 3 the oscular end curled inwards relatively less than the basal end. With ‘basal pieces’, on the other hand, the tendency for the oscular end to uncurl was much less evident, only 1 piece out of 6 measured showing this effect. Any tendency for the pieces to uncurl immediately after being isolated, therefore, concerns the part derived from the original oscular rim and is probably to be correlated with the occasional arching of this part in the longitudinal plane towards the dermal side. Both tendencies indicate the presence of tension in the dermal epithelium. Their effects are only temporary, and they confirm that the collar-cells or functional porocytes are responsible for the normal curling behaviour since these cells are absent from the oscular rim.

During the first phase of the healing behaviour the cut edges of the piece become reorganized to form a clear zone which arches between the projecting spicule rays (fig. 2, B). The act of cutting leaves a fairly straight edge, with the collar-cells sometimes reaching close to it (fig. 2, A), but more often pushed back a little, or scraped away from it, by the blade of the scissors used. Here and there along the edge the collar-cells may be concentrated together in small areas by the action of the scissors, but there is no evidence for an immediate retraction of the choanoderm when the wall is cut. The collar-cells must either be firmly attached to their substratum, or else they are not under tension.

As curling proceeds some of the spicules near the edge become jammed against the interlocking rays of adjacent spicules and their rays then project more and more at the edge as the neighbouring spicules are drawn inwards by the process of contraction (fig. 2, B). The numerous quadriradiates are, of course, anchored to the choanoderm by their gastral rays, and it seems that the choanoderm contracts mainly in the transverse direction, drawing these spicules together where possible. This motion is opposed by the mesogloea to some extent, so that the wall curves more in the transverse plane, but some lateral squeezing of the mesogloea takes place and this necessitates the lengthening of the piece, provided the volume and thickness of the mesoglea do not appreciably change. The tendency for the mesogloea to extend against the longitudinal tension in the choanoderm assists in the curling of the piece in the longitudinal plane, and thus the whole process of curling is explicable in terms of the predominantly transverse contraction of the inner epithelium. This contraction indicates that the epithelium responsible is anisotropic, for it can shorten in one direction whilst extending in the direction at right angles. The contraction, however, need not strictly be transverse, and quite possibly the tension is developed on a spiral course, the spirals conforming to the spiral organization of the oscular tube (Jones, 1955).

As some spicule rays project more and more laterally they prop out the dermal epithelium (fig. 2, B). In this figure the dermal cells near the edge can be seen to be stretched in the longitudinal direction, as would be expected from the above explanation concerning the elongation of the piece. The projecting rays disturb the orientation of the cells somewhat and add to the amount by which the cells in the vicinity are stretched.

The clear zone comprises two layers of cells with a thin layer of mesogloea in between. Each epithelium consists of flattened cells with polygonal outlines that can be faintly seen on living specimens. The cells of the inner surface may be derived from the outermost choanocytes by a process of flattening and spreading, such as has been described by Duboscq and Tuzet (1939), but there is little evidence for this ; almost everywhere along the edge the collarcells are sharply demarcated from the flattened epithelium and very few cells that could be regarded as intermediate in form can be seen. Thus it is more probable that, as the spicule rays are left projecting by the contraction of the choanoderm, part of the dermal epithelium becomes pulled over the edge in between these rays. Alternatively, the inner layer of the clear zone is produced from amoebocytes in the mesogloea.

The first phase of healing may be regarded as ending with the appearance of healing membranes at the oscular and basal ends of the piece. A membrane is usually seen first at the oscular rim and this seems to be identical with the diaphragm sometimes present within the rim of complete, but possibly not fully expanded, oscular tubes. The membrane may arise at the extreme edge of the piece, or from the surface of the porocyte epithelium, or from the junction between this epithelium and the choanoderm. In the pieces observed it became conspicuous at times ranging from to 19 h (fig. 2, c). A similar membrane is formed at the basal end, although usually a little later (fig. 3, A). Both membranes are capable of drawing the edges of the piece closer together. They advance towards each other, merging on each side with the lateral membranes that arch between the projecting spicule rays, until only a circular or oval hole is left in the mid-line (fig. 2, c). This hole usually lies nearer the oscular end owing to the more rapid advance of the basal membrane. It may become completely obliterated (sometimes within 24 h after excision) after serving as an osculum for a time. Before this currents can be detected leaving through the hole, confirming that the flagella remain active. The pores on the original wall stay open throughout.

The basal healing membrane arises close to the basal end of the piece as a shelf projecting inwards from the inner surface. As with the diaphragm, its formation appears to be the result of tension developing in a transverse band or line of cells across the inner surface, and its spread is presumably due to the continued maintenance of tension at the free edge. The cells certainly are stretched along the edge, as seen in optical section, except perhaps when the gap is practically sealed (fig. 3, c) ; but what happens to them as the gap closes and the edge diminishes in extent is still a problem. According to Minchin (1900) the dermal cells become mushroom-shaped when they contract, the cell-bodies moving into the mesogloea. Maas (1910) agrees with this observation. Cells of this shape, however, have not been seen at the edges of healing membranes on fixed pieces, but possibly their withdrawal from the edge is a relatively rapid affair, so that few examples would be present at the moment of fixation. On the other hand, since the area of the original piece continues to shrink in both length and width as the membranes spread across the gap, there may be no need for an alteration in the total number of cells in the epithelia. This would imply that the cells are capable of losing contact with the adjacent cells and making contact with their new neighbours without leaving the epithelium, since the pattern of distribution of the cells must change as the membrane obliterates the gap.

While it would appear that tension is developed at the free edge, the membrane as a whole is tending to contract and draw the edges of the piece closer together. This process results in the collar-cells being dragged beyond the limits of the original wall, and spicule fragments may accompany them (figs. 2, c; 3, B). The spicules themselves, however, remain and become crowded together as the piece shrinks. No doubt the sharpness of the angle between the membrane and the original wall precludes the spicules from being drawn across with the collar-cells. The interlocking of the spicule-rays gives support to the original wall and prevents the inturning of the edges, so that as the healing membrane spreads across the gap, the inner and outer epithelia of the original wall are drawn beyond the limits of the shrinking area of spicules. Presumably the cells of the epithelia part and rejoin to allow movement past the gastral and other projecting rays.

The result of the healing process is usually a small, thin tube, which is often crooked, particularly if the original tube from which the piece was obtained was bent, or not bisected symmetrically. Sometimes one end of the piece becomes bent over the other end, producing a spiral, while at other times the two ends meet to form a ring. The healing membranes then spread across the gaps bounded by the lateral edges, forming spheroidal objects or coiled tubes.

Newly forming spicules become visible in the healing membrane after about 20 or more hours. Slender monaxons appear first, lying in between the two surface epithelia and often in contact with the inner one. They tend to be oriented at right angles to the free edge of the membrane, but later (after 3 days), when the piece has healed completely, they are pushed through the outer dermal layer and project from the surface. Young triradiates, which may also continue to grow on the original wall, make their appearance within the membrane after about 2 or more days, developing on the inner epithelium. Their orientation quite often appears to be directed towards the free edge of the membrane, but the arrangement is more often than not confused, which is to be expected from the time taken for these spicules to develop (over 1 day), for the direction of the free edge in relation to the growing spicule may be constantly changing as the membrane spreads across the gap. Young spicules may also be dragged on to the membrane from the original wall as the piece shrinks, and their orientation will be disturbed if, as seems likely (Jones, 1952), it is caused by the operation of mechanical factors. Abnormal conditions of development are demonstrated by the large number of aberrant spicules that are produced under these conditions. Some arise in formative complexes with an unusual arrangement or number of calcoblasts. For example, tiny triradiates have been seen in groups of 4 and 5 cells, instead of the normal 6, while sometimes no calcoblasts may be present around a small primordium, the cells presumably having departed or been torn apart. In some sextets the calcoblasts are arranged in the form of a rosette, whereas in others the outer 3 cells are displaced with respect to the inner 3 as in some of Minchin’s drawings (see Jones, 1954a). Since abnormal spicules develop in the membranes, the latter are not so suitable for studying the normal development of the sextets, although some growing spicules do show a perfect bilateral symmetry, with the 3 thickener cells in the angles between the rays and the 3 founder cells at the ray-tips (Minchin, 1908).

Histological examination of fixed membranes confirms that the membrane consists of two epithelia with a thin layer of mesogloea in between. In L. botryoides the nuclei of the outer dermal layer may be larger than those of the inner. No conclusive signs of cell-division have been observed in the dermal epithelia. The collar-cells form a continuous sheet apart from the porocytes, which may be open or in the process of opening. Collars and flagella are clearly visible on the preserved choanocytes. There is little evidence of a transformation of collar-cells into flattened cells, although at the border of the choanoderm one may see choanocytes with their granular cytoplasm more spread out. Generally, however, the area of the collar-cells is sharply demarcated from the rest of the membrane.

The mesogloea contains amoebocytes and calcoblasts in contact with the inner epithelium. Spherical cells with a peripheral zone of large granules (the ‘excretory cells’ of Minchin, 1908) are quite common on or in the membrane.

Examination of the healing membrane under the polarizing microscope reveals no birefringence when the membrane is seen in surface view, but in optical section the membrane glows and darkens four times in a complete 360° rotation. This is best seen when the membrane has completely healed over the gap and the piece is resting on one side. No birefringence is detectable on membranes fixed and mounted in balsam. The birefringence seen in optical section when the medium is sea-water probably arises from the layering of the material in the cell-walls of the flattened and closely set epithelia.

The third phase of the regenerative behaviour, involving the opening of an osculum and the growth of the olynthus, has not been followed far, but there can be no doubt that, provided the piece is supplied with nourishment, the normal specific form will be attained.

It should be noted that the first two phases of the healing process are essentially distinct. In the first, the piece curls owing to the predominantly transverse contraction of the internal epithelium, with little or no reduction in the mesogloeal volume. In the second, both internal and external epithelia spread across the gap mainly as a result of the contraction of the free edge, while the original wall, as represented by the area of fully-grown spicules, shrinks considerably. The importance of the spicules to both processes is, however, obvious. The gastral rays assist in the curling because they anchor the quadriradiates to the internal epithelium; and the other rays, interlocked together, provide the necessary rigidity to enable the healing membranes to spread across the gap without the gastric cavity becoming obliterated.

Mutilated tubes that are not removed from the specimen show the same kind of healing behaviour as the isolated pieces. For example, when distal longitudinal slits are made at diametrically opposite positions in the wall of an oscular tube and half of the cylinder is then excised, the remaining half usually arches at first towards the dermal side, either as a result of longitudinal tension in the dermal layer, or owing to the pressure of the outflowing current of water ; but then invariably a phase of inward curling takes place. The part bends right across the outflowing current and the free corners close together until they have met and fused, making the gap in the tube triangular. Healing membranes then spread across the gap from the three angles and the tube constricts and remoulds itself distally, until the cut edges have been brought as close together as possible. Elsewhere spicules form in the healing membrane. Thereafter, with the normal growth of the tube, the distal kink becomes somewhat smoothed out as the tube lengthens and increases in girth.

The same occurs when the oscular rim has been excised before removal of the rectangular piece: the presence of the porocyte epithelium is thus not essential for the inward curling. Also with other types of mutilation the behaviour is essentially the same ; there is a phase of inward curling, followed by the spread of the healing membrane and the shrinkage of the original wall.

Pieces of the wall that have been immersed for 5 min in distilled water (or derived from a tube treated in this way) and then replaced in sea-water, sag and eventually flatten against the bottom of the dish. Subsequently they shrink in both transverse and longitudinal dimensions and continue to do so for several days, becoming more flexible at the same time. In one specimen the mesogloea became noticeably thicker. There was some variation in the total shrinkage (10-20% in 5 days) and also in the time at which shrinkage first became apparent (1-12 h), but the rate of shrinkage eventually decreases with time, probably as a result of the crowding together of the spicules. The cause of the shrinkage has not been ascertained, but it appears to involve the softening of the mesogloeal substance, possibly by bacterial activity.

Pieces that are left in distilled water also sag, though occasionally the oscular end may incurl as the basal edges flatten. However, this effect is not sustained. A very slight shrinkage occurs and the spicules corrode and have disappeared after about 3 days. .

Thus the changes in shape which the healthy pieces undergo in sea-water depend on the presence of the epithelia. The cells are destroyed by the 5 minutes’ treatment with distilled water, whereas the colloidal properties of the mesogloea are not appreciably modified (Jones, 1956).

The facts given above suggest that the curling is caused by a contraction of the inner layer working against the relatively firm mesogloea. In order to test this, longitudinal halves of oscular tubes were held at one end by means of a needle and brushed repeatedly on their inner surface. The end damaged by the needle was then cut away and the behaviour of the brushed pieces was compared with that of the control halves. The latter underwent the normal process of curling and healing, even after their dermal surface had been brushed for the same number of times as had the inner surface of the other halves. The pieces without their choanoderm, however, after partially recovering their initial curvature (for they are flattened by the brushing), rapidly shrank and became saddle-shaped (fig. 4, c). The resulting appearance indicates that the pieces had curved towards the dermal side in the longitudinal plane, while decreasing their transverse curvature at each end. Table 2 shows the typical changes in size of a piece brushed on the choanodermal side. The changes in width are due to shrinkage and not to curling inwards in this case, while the changes in apparent length are due partly to the arching towards the dermal side and partly to shrinkage. The rate of shrinkage decreases as the spicules become closely packed together. At 43I h the piece had lost its longitudinal curvature, which accounts for the apparent increase in length.

As the brushed pieces shrink, a few large circular gaps form in the mesogloea between the spicule-rays. They may be derived from cavities originally occupied by porocytes, or possibly from small holes made by the bristles of the brush. These have been seen on pieces fixed immediately after the brushing. The enlargement of these cavities (which may reach 30 p, or more in diameter) and the very rapid shrinkage suggest that the mesogloeal substance has been softened and possibly reduced in volume.

The saddle-shape appears to be the result of a contraction of the dermal surface, possibly at the edges only, but in any case opposed by the interlocking rays of the spicules. The paired rays of the triradiates do not lie in the same plane (Jones, 1954b), but are inclined so as to embrace partially the oscular tube. The overlapping of the rays of a number of adjacent spicules would hence hinder any process tending to curve the wall transversely towards the dermal side, especially after the wall has shrunk and the spicules have become crowded together. Thus while a limited amount of longitudinal curling towards the dermal side is possible from the original straight condition, the contraction of the dermal layer merely results in a partial uncurling transversely at the ends of the piece. In the middle the piece remains markedly curled in the transverse plane, thereby accommodating the longitudinal curvature towards the dermal side of the lateral edges of the piece.

It is necessary to brush the pieces really well (over 50 strokes with a camelhaired brush) for the above to occur, for if the brushing is not sufficient to remove most of the collar-cells and porocytes, the pieces still tend to curl in both transverse and longitudinal planes to the original choanodermal side, although the curling is not so striking as with the controls. Such pieces usually have small islands of collar-cells remaining, and open pores and tiny spicules may be seen in the regions from which the collar-cells have been removed. The collar-cells tend to spread themselves out somewhat, and the internal epithelium becomes reorganized and causes the curling behaviour. In such pieces healing membranes develop at the ends and laterally, but they appear usually some distance from the original edges of the wall and at a later time than in the unbrushed pieces. Such membranes have even been obtained with well-brushed pieces, and by their contraction and spread they can induce a reversal of the longitudinal curvature and form a complete tube, roughly elliptical in cross-section, along the centre of the piece.

Thus, provided the brushing has been so severe that the porocytes and other cells associated with the collar-cells are largely removed, the normal process of curling does not take place. One must not therefore overlook the possibility that the porocytes are responsible for the contractility of the inner surface. Furthermore, the production of the saddle-shape indicates that the outer dermal epithelium also is contractile.

When pieces of L. complicata are left in a mixture containing equal parts of sea-water and isotonic magnesium chloride (7!%), no significant difference in behaviour can be detected from that occurring when ordinary sea-water is used. In both media the pieces always curl up with the choanoderm on the inside, often elongating as they do so. Some unimportant differences were noticed, however. Thus in the MgCl₂ sea-water the formation of a healing membrane was sometimes delayed, although never prevented (4 pieces out of 7). Also the spicules showed signs of corrosion after 24 h and new spicules did not grow in the healing membrane. The flagella also seemed to beat more actively (judging by eye), and ‘blisters’ developed on 4 pieces in the mixture compared with only 1 on the controls in sea-water. The development of blisters is interesting, since Maas (1910) believed that they were caused by starvation, which would be more likely in the artificial medium. However, the contraction of the internal epithelium and the formation and spread of the healing membranes are certainly unimpaired in the MgCl_a sea-water.

Now the presence of a nervous system in L. botryoides and other sponges has been claimed by Pavans de Ceccatty (1955) as a result of his study of mesogloeal cells that resemble the neurones of higher animals. These cells are dissimilar, however, in being interconnected asynaptically, but they are described as making synaptic connexions with other types of cell. One would expect that such cells, if they are nervous in function, would play a part in the behaviour of the sponge, and also that their action would be modified in the presence of MgCl₂ sea-water, for the magnesium ion is believed to act on cellsurfaces (Danielli, 1950), reducing their excitability. For example, MgCl₂ sea-water completely inhibits the neuromuscular system of the coelenterates, and yet the curling and healing behaviour of Leucosolenia takes place in this medium. There is no reason to suppose that the ions of magnesium chloride do not penetrate into the mesogloea, particularly as tubes left in isotonic magnesium chloride become softened after only 11 h (Jones, 1956). One can only assume, therefore, that either the ‘neuromuscular system’ of Leucosolenia is unaffected by the magnesium chloride, or the ‘nervous system’ does not play a significant part in the behaviour of the sponge ; but neither alternative enables one to accept with confidence the claim that a nervous system is present in Leucosolenia.

The process of curling of the isolated pieces indicates that the internal surface is contractile, and since this surface directly opposes the excess internal hydrostatic pressure one can assume that it is normally in a state of tension. Whether this tension is the result of a tonic contraction of the epithelium, as Dr. C. F. A. Pantin has suggested (private communication), or whether it is due to the elastic stretching of the material, as Bidder (1898) believed, cannot be decided on the evidence presented. However, the existence of tension of one form or another is shown by two observations. Bidder (1898) noticed that flagellated chambers often turned themselves inside out when teased, and concluded from this that there was elastic matter just beneath the collar-cells, while Huxley (1912) likewise noticed the tendency for blocks of collar-cells to change the direction of curvature, and attributed this to tension existing in the interstitial substance at the base of the cells, or to tension in the epithelium as a whole. He did not notice any other types of cell associated with these blocks. The blocks of collar-cells eventually rounded themselves off into spheres, and Huxley explained this by the tendency of the cells at the periphery to draw themselves into closer contact with their neighbours. This, if correct, would indicate that the collar-cells are capable of developing tension, but other evidence favours the porocytes as the elements responsible for the contractility of the internal surface.

Dendy (1890) noticed that in Grantla labyrinthica the inner layer of some of the radial tubes had shrunk, the collar-cells forming a multilayered mass at the centre of the tube and the mesodermal cells being pulled out into strands forming a radially-disposed network serving to suspend the central mass. This observation has been confirmed and extended by Duboscq and Tuzet (1939). They point out that the collar-cells of Sycon raphanus rest on a very thin membrane of collencytes and that when contraction occurs the collar-cells become vespiform in shape. From their bases rooting processes join up with the processes of the underlying mesodermal collencytes, some of which are radially disposed. The latter spread out distally at the level where the vespiform bodies of the collar-cells appear to make contact together, and at this level eosinophil granules which were present in the cytoplasm of these collencytes could be seen between the collar-cells. This implies that it is the collencytes forming the membrane on which the choanocytes rest that are the contractile elements, and that their contraction results in the membrane moving inwards, with the consequent stretching of the radially-disposed collencytes and the pulling of the anchored bases of the choanocytes through this membrane. Dendy’s observation of the piling up of the collar-cells to form several layers is consistent with the view that the collar-cells are not the contractile elements.

Now Minchin (1900) has described the inward migration of the porocytes when Clathrina coriácea contracts completely, while Maas (1910) gives a similar account for Leucosolenia lieberkühnii. The collar-cells first become narrow and elongated, and then pile up on one another, while the porocytes pass between them and come to form ‘an epithelium lining the now greatly reduced gastral cavity’ (Minchin, 1900, p. 30). Such a process could be explained if one assumed first that the collar-cells are anchored to the parts beneath the choanoderm, and secondly that the porocytes are interconnected by cell-processes which pass between the bases of the collar-cells. Contraction of the porocytes together would then result in the choanocytes piling up, slipping through the gaps, and eventually coming to lie outside the porocyte mass. The porocytes of Clathrina and Leucosolenia would then be analogous to the collencytes which form the membrane beneath the collar-cells of Sycon, and would constitute the tense, contractile system of the internal layer of the normal oscular tube.

The assumption that the porocytes of Leucosolenia are responsible for the contractility of the inner surface is an attractive one. The porocytes are obviously contractile, since they can occlude their pores, as Minchin (1900) has pointed out. Furthermore, they are derived from the inner edge of the porocyte epithelium lining the oscular rim (Minchin, 1898), from which also the obviously contractile diaphragm is produced. This has been fully confirmed by means of photographic records of the growth of oscular tubes, of L. variabilis (Jones, 1952). These demonstrate the formation of pores at the junction between the porocyte epithelium and the choanoderm as the limit of the collar-cells advances, keeping pace with the longitudinal growth of the tube. The choanocytes apparently move around the newly-opening pores (Minchin, 1898) and this suggests that they are only loosely contiguous and that they are freely mobile either on the mesogloeal surface, or on the extensions which probably persist between the porocytes after the pores have opened. The movement of the collar-cells does not suggest an epithelium under tension ; more likely the collar-cells mutually compress each other, since the division of the collar-cells is presumably causing the limit of the choanoderm to advance.

Histological evidence for interconnexions between the porocytes is rather scanty. Prenant (1925) observed that in the fully-expanded specimens of Clathrina coriácea the porocytes were insinuated between the choanocytes situated around their border, which suggests that there might be processes from porocyte to porocyte between the bases of the collar-cells. Minchin and Reid (1908) observed a honeycomb network after removing the collar-cells and staining a piece of wall with picronigrosine ; they believed that this was the interstitial substance between the bases of the choanocytes, but it may well have consisted of porocyte processes.

Thus there are grounds for believing that the porocytes are directly interconnected by a perforated membrane on which the choanocytes can move and through which they can pass; and it seems probable that this membrane, if its existence can be established, will be the part responsible for the contractility of the internal surface. Otherwise the contractility is dependent on the ability of the collar-cells to develop tension and to cohere together.

A preliminary study of the healing behaviour of Leucosolenia variabilis and L. botryoides was made while I was occupying the Cambridge University Table at the Marine Biological Association Laboratory, Plymouth, where Mr. F. S. Russell and the staff gave me friendly assistance. I am also greatly indebted to Dr. D. J. Crisp for permission to cultivate specimens of L. com-plicata in the aquarium room of the Marine Biology Station, Menai Bridge, Anglesey, and to Dr. C. F. A. Pantin for kindly reading the original manuscript and making many helpful suggestions.