page 10

Discussion

ECTODERM. The coelenterates of the Order Hydroida have long been divided into two groups depending on whether or not the hydranths are enclosed within a chitinous periderm (Hyman, 1940). In the calyptoblastic (thecate) hydroids the perisarc of the stems expands at polyp bases to form cup-like structures (hydrothecae) which loosely enclose the hydranths. In the gymnoblastic (athecate) hydroids, the perisarc of stems (if present) ends abruptly at the base of hydranths which are thus naked. The results of the present study indicate, however, that in S. tenella a periderm which is continuous with the perisarc at the hydranth base is present on the hydranth. The periderm is visible only in histological preparations but it is unlikely to be artifact since different fixatives preserve it in the same form and confer upon it similar staining properties, both histological and histochemical. It is much thinner than the perisarc, and is very closely applied to the surface of the ectodermal cells (Pl. 2, Fig. 1, C) except in some places where it projects from the hydranth as a tangled mass (Pl. 1, Fig. 4, C). It is therefore quite different in morphology to the periderm forming the hydrothecae of calyptoblastic hydroids, and in order to avoid a confusion of terms it will for the time being be called a cuticle.

Rudall (1955) states that Protohydra (a naked hydroid) produces a delicate periderm, but he does not elaborate on its nature. Also, the electron microscope reveals a thin cuticle covering the surface of Hydra (Hess, 1961). Manton (1940) shows a cuticle which is very similar to that of S. tenella to be present on the athecate hydroid Myriothela penola. Cowden (1965) studied the cytology of the gymnoblastic hydroid Pennaria tiarella, but makes no mention of the presence of a periderm on the hydranth. However an examination of some of his figures, especially photographs of PAS/AB, and acrolein/Schiff (for protein) stained sections shows a thin, dense line at the free surface of the ectoderm which appears very similar to the cuticle of S. tenella. It would be interesting to know if a similar covering is present on the hydranths of gymnoblastic hydroids in general or if it is confined to those animals cited above. As Rudall has stated, ".... it is a question as to whether the whole mechanism of periderm formation is completely eliminated in naked forms." (Rudall, 1955, p.52).

The cuticle appears to be very flexible, for on contraction of the hydranth it is thrown into folds (Pl. 4, Fig. 4A, C). Such folding must deform the ectodermal cells, especially their distal ends, but to what extent this occurs has not as yet been determined.

Histochemical tests indicate that the cuticle varies in composition in different regions. On the apical and middle regions of the hydranth it appears to be composed of a mucoprotein, or neutral mucopolysaccharide (according to the carbohydrate classification of Pearse (1960)), since it gives a positive reaction in the PAS test and with NYS. However since staining by these methods is abolished by digestion with pepsin, it is likely that the polysaccharide component is firmly bound to protein, and it may be deduced that a mucoprotein is the substance present. At the basal region of the hydranth the cuticle seems to be composed of an inner layer of mucoprotein (which stains less intensely with PAS and NYS than that of the apical and middle regions), together with an outer layer of an page 11 acidic mucopolysaccharide. These two layers are well differentiated in the perisarc in the region of annulations at the polyp base where they are both considerably thicker. Here, the inner layer stains only weakly with the PAS test, and more strongly with NYS. Staining is still abolished by pepsin digestion. The outer layer stains strongly in tests for acidic mucopolysaccharides (Pl. 5, Fig. 2). A double layered perisarc has been described by Berrill (1949) in the stolons of Obelia. As in the present study, the layering is evident in the region of annulations at the hydranth base, and Berrill considers that the inner perisarc layer of closely packed lamellae is laid down subsequent to the outer layer of loose annular lamellae. No histochemical staining was attempted however, so that detailed comparison between the perisarc of Obelia and that of S. tenella is not possible.

The periderm of hydroids has long been described as being chitinous. Rudall (1955), by means of X-ray diffraction, observed the presence of chitin in Tubidaria colonies and in other hydrozoans. The diffraction patterns obtained were different from the patterns shown by purified arthropod chitin, but that of Tubularia approached the standard arthropod chitin pattern more closely than the other hydrozoans studied. Syncoryne tenella is commonly called a "tubularian hydroid" (Hyman, 1940) and it is reasonable to suppose that it's perisarc is chitinous. This supposition does not go against the histochemical findings of the present study, at least as far as the inner perisarc layer is concerned: Pearse (1960) considers chitin to be usually PAS negative but positive in tests for protein if it is present as protein complexes, and it has been seen that the inner perisarc layer is only weakly PAS positive, and stains with NYS. He further states that chitin does not occur in the pure form in nature but is always mixed with calcium carbonate, or protein, or both. Chitin does not, according to Pearse, stain in tests for acidic mucopolysaccharide, so that the outer perisarc layer in S. tenella may not be chitinous, or may contain only very small amounts of chitin. Similarly, the cuticle of the apical and middle regions of S. tenella, being intensely PAS positive, may not be chitinous or may contain small amounts only. A third possibility is that the outer perisarc layer and the cuticle may contain "non-typical" chitin. It is noteworthy that Rudall (1955) found that the chitin of millipore colonies showed non-typical X-ray diffraction patterns which he considered likely to indicate a second principal constituent.

Berrill (1949) considers the perisarc of Obelia to be secreted by a specialized ectodermal cell and not by the general epidermis since various regions, for example, the tentacles, are free of perisarc. He described ovoid cells containing iron haematoxylin staining granules in the epidermis of the growing stolon tip, and the stem epidermis. In the present study, however, the perisarc (at least the inner perisarc layer) and the hydranth cuticle appear to originate by secretion of substances from the general ectodermal cells. These cells were found to contain fine PAS positive granules (Pl. 1, Fig. 6) in their cytoplasm. As with the cuticle, the PAS positivity was removed by pepsin digestion which suggests that the cuticle and these ectodermal granules have a similar chemical composition consisting of a polysaccharide/protein complex.

The acidic mucopolysaccharide granules seen in the ectoderm at the hydranth base in the present study (Pl. 5, Fig. 1B) quite possibly correspond to the iron haematoxylin staining granules described by Berrill page 12 (1949), since basic dyes such as haematein (the true dye substance of haematoxylin) have a great affinity for acidic tissue substances. However, the use of iron mordants can modify the haematoxylin staining reaction so that virtually all tissue components may be stained, and because of this no great significance can be placed on the above mentioned possibility.

In the present study there remains the question as to whether the general ectodermal cell in the region of perisarc annulations is capable of secreting both the acidic mucopolysaccharide substance (s) of the outer perisarc layer and the mucoprotein substance (s) of the inner layer, or whether specialized cells perform the former function. A further question is whether the 2 layered structure of the perisarc described here is confined to regions of annulations, or whether it is the general pattern of perisarc structure. Until this is known, it would be premature to speculate on the properties which the 2 layered structure might confer on the perisarc, and thus on the hydrocaulus.

The small masses of cuticle which have been seen to project from the hydroid (Pl. 1, Fig. 4, C) are peculiar structures. It is possible that they are artifacts due to separation of the cuticle from the underlying ectoderm during processing. If they are not artifacts, and in the living animal serve some particular function, this function is not at present apparent. The cuticle as a whole can only be supposed, at this time, to perform a protective role.

The other PAS positive, but distase labile ectodermal bodies (Pl. 2, Figs. 4A and 4B) are concluded to be glycogen. They were abundant in both ectoderm and endoderm a short time after feeding, but not often seen in specimens fixed and sectioned after a period of starvation. Hyman (1940) states that excess food coelenterates is stored in the gastrodermis (endoderm) chiefly as fat but that glycogen may be stored without change. In the present study no lipid was detected in either epithelial layer, but the glycogen deposits already mentioned may serve as a food reserve. Both epithelial layers, since they possess muscle fibres, would presumably require reserves of food for muscular contraction as well as other metabolic activities. It is possible that lipid may be stored in other parts of the colony such as hydrocaulus.

The ectodermal cells also contain moderate amounts of RNA (Pl. 2, Figs. 3A, 3B), indicating that protein synthesis occurs within them. This is probably of two main types: Synthesis of proteins to maintain the contractile appartus of the cell, and synthesis of the protein moiety of the secretion granules which form the cuticle. Staining for protein by NYS in ectodermal cells is weak, except for the muscle fibres, but it is possible that proteins within ectodermal cells lack sufficient basic groups to bind NYS; Deitch (1955) considers that NYS is bound by E-basic groups of lysine, arginine, and histidine in tissue sections. There is, however, a distinct increase in the amount of protein in the ectodermal cells immediately surrounding the mouth of the hydranth, an observation also made by Cowden (1965) for Pennaria hydranths.

It is notable that at the base of the polyp, where the inner proteinaceous perisarc layer is much thicker than is the cuticle, the underlying ectodermal cells contain more RNA (as judged by RNase reversible pyroninophilia), and they stain more densely with NYS.

Results indicate that the cnidoblasts present in the tentacles of S. tenella arrive there by migration through the ectoderm of the hydrocaulus page 13 and hydranth. Cnidoblasts bearing mature nematocysts are found in the head and shaft of tentacles, in the hydranth and in the hydrocaulus. Cnidoblasts with immature nemotocysts are present only in the hydranth and hydrocaulus, not the tentacles. Further, the nematocysts in tentacle shaft and hydrocaulus, and often those in the hydranth, are orientated with their long axis parallel to the surface of the ectoderm. This suggests that they are in transit rather than in the sites in which they will be used. There is no indication of a "nematocyst replacement zone" in the tentacle ectoderm immediately beneath the terminal cap as was found by Cowden (1965) for Pennaria capitate tentacles.

Cnidoblast migration has been described in Hydra (Lenhoff, 1959; cited in Picken & Skaer, 1966). Also the early observations of Hadzi (1907, in Picken and Skaer, 1966) on the migration of ectodermal cnidoblasts in Tubularia from perisarc covered regions of the hydrocaulus have been confirmed by more recent studies (Tardent, 1962, and Tardent & Eymann, 1959; both cited in Picken & Skaer, 1966). In Tubularia, however, the supply of cnidoblasts to the hydranth is discontinuous and the renewal of the nematocyst stock coincides with regeneration of a new hydranth (Picken & Skaer, 1966). During this process large numbers of cnidoblasts accumulate at the primordial area, so that the replacement of cnidoblasts is here linked with normal, regular autotomy of the hydranth. However, Mackie (1966) showed that in fully grown Tubularia crocea many weeks old, there exists a capacity for large scale cellular proliferation (as evidenced by, for example, regrowth of amputated tentacles), and for production of nematocysts and digestive enzymes. Mackie points out that such hydranths had been using up nematocysts and digestive products in the daily process of feeding throughout their lives. The "normal, regular autotomy" of hydranths did not occur in Mackie's cultures of Tubularia, and he states that in gymnoblastic hydroids such autotomy has only been observed in the laboratory under wholly unnatural conditions. He concludes that it is very probable that the hydranths of Tubularia, and perhaps of gymnoblastic hydroids in general, can live indefinitely.

In the present study sections of S. tenella buds (hydranth primordia) show clusters of cnidoblasts with developing nematocysts in the ectoderm, and it is noteworthy that in the mature hydranth no clusters of cnidoblasts are visible except in the capitate tentacles. It seems probable, therefore, that nematocysts are initially provided in S. tenella hydranths by a similar process to that described above for Tubularia, and that in the mature hydranth other migratory mechanisms are involved. Further work on this problem is desirable, especially the examination of sections of hydranths at different stages of development. One of the nematocysts seen in this study (Pl. 4, Fig. 6B) presents a curious internal structure. No explanation of the function of such a structure can be given at present. It seems unlikely to be artifact, because of its clarity, and contrasts markedly with the nematocyst shown in Pl. 4, Fig. 6A. This latter nematocyst capsule appears to have been flattened, or compressed during preparation and does not show the internal structure of the first. It resembles a stenotele which has lost its barbs.

The hydroids characteristically possess undifferentiated cells (interstitial cells) which are able to differentiate into any of the cell types when required (Hyman, 1940). The clusters of interstitial cells seen in page 14 Hydra have long served as a "model" for the way in which these cells are organized with respect to the epithelial layers. However, nothing resembling the clusters of interstitial cells which occur in Hydra are found in S. tenella, and if any kind of cell in S. tenella is to be called an interstitial cell there seem to be only two contenders. These are the small cells which occur singly at the base of the ectoderm, and the cells also occurring singly which have already been termed nerve cells. Several factors indicate that the former may be interstitial cells. Firstly, the cells are small, and round in sections, as are the interstitial cells of Hydra; the cells have no cytoplasmic extensions as are seen in nerve cells. Secondly, the moderate RNase labile pyroninophilia of the cytoplasm of the cells indicates the presence of RNA which Singer (1952) and Slautterback (1961) show is characteristic of interstitial cells. Thirdly, the morphology of the nucleus in the cells is very similar to that of cnidoblasts with mature or developing nematocysts. The cells found singly at the base of the ectoderm in Pennaria (Cowden, 1965) are similar in morphology to bipolar nerve cells when seen in section. Cowden found these cells to have a high level of nucleolar and cytoplasmic RNA, and on this basis identified them as interstitial cells. However it is characteristic of nerve cells to have high levels of RNA also. He states that the "interstitial" cells were not numerous, and that no epidermal concentrations of these cells was ever observed. He found difficulty in identifying nerve cells, and did not describe them.

The fact that interstitial cells in S. tenella occur singly and not in clusters indicates that they may migrate from a region of proliferation, and in the present study the only mitotic figures observed occurred in the ectoderm at the base of the hydranth. The increased RNA content of cells in this region may indicate increased metabolic activity and perhaps support the view that it is a region of cell proliferation. However in this region the small interstitial cells do not appear more numerous than in other regions. Haynes & Burnett (1963) have demonstrated that in Hydra the mucous cells of the gastroderm have the capacity to dedifferentiate into interstitial cells and redifferentiate into other cell types including ectodermal cells. Also Slautterback (in Lenhoff & Loomis, 1961, p.314) states that when the pedal disc of Hydra is amputated, the secretory cells are soon replaced by partial dedifferentiation of cnidoblasts followed by differentiation into secretory cells. It seems that "totipotency" may be a property of more cells than just interstitial cells. A similar mechanism of dedifferentiation, if it occurred in either ectoderm or endoderm of S. tenella, could also explain the sparse distribution of interstitial cells. Cowden (1965) thinks it possible that in the nematocyst replacement zone of Pennaria capitate tentacles, ectodermal cells may differentiate into interstitial cells which subsequently redifferentiate into cnidoblasts.

MESOGLOEA. The mesogloea of hydroids, especially that of the polyps (as opposed to hydromedusae) continues to be a very difficult subject for study. In polyps the mesogloea is thin and difficult, if not impossible, to isolate, so that its chemical composition must be inferred from its staining reactions. Even in Hydra, a hydroid which has been very extensively studied (Lenhoff and Loomis, 1961), the structure of the mesogloea is still not well known. Very little work has been done on the mesogloea of other hydroid polyps.

page 15

The mesogloea of S. tenella stains blue in the Mallory and Mallory/Azan methods. Hess, Cohen, and Robson (1957) noted a similar result for Hydra mesogloea, and believed it to indicate an affinity with the mesogloea of other coelenterates and with collagens. Rudall (1955) however examined whole freeze dried Hydra specimens by X-rays but could not detect any collagen. He concluded that collagen is either absent or at most minutely present in hydrozoan polyps.

The fibrous nature of the mesogloea evident from this study is very similar to that found by Cowden (1965) for Pennaria. Chapman (1953, 1966) showed the mesogloea of hydromedusae to be fibrous, and Hess, Cohen and Robson (1957) found that the mesogloea of freeze dried formalin fixed Hydra sometimes exhibited fibres visible in the light microscope. In the electron microscope, however, they found the mesogloea to be finely granular, with some evidence of very fine fibrils which certainly could not be seen with the light microscope. An electron microscope study of S. tenella is planned by the present author, and it will be interesting to discover if the mesogloeal fibres seen with the light microscope are visible in electron micrographs.

The staining reaction of the mesogloea in the PAS/AB/NYS test suggests that it is composed predominantly of PAS positive but pepsin labile mucoprotein, together with acidic mucopolysaccharide. Further, the sites of PAS positivity following pepsin digestion indicate a nonprotein bound constituent, perhaps neutral mucopolysaccharide. The failure to exhibit metachromasia after toluidine blue staining (when examined in DPX) could indicate a low degree of sulphation of the acidic mucopolysaccharides present (Pearse, 1960; Bergeron & Singer 1958) or that the positions of the binding sites (carboxyl, sulphate) do not favour the expression of metachromasy (Bergeron & Singer, 1958).

Although it seems that the mesogloea must originate from the cell layers, Chapman (1966) believes that in hydrozoans there is no evidence of any special intracellular precursors in ectoderm or endoderm. However the present study reveals mucoprotein granules in many ectodermal epitheliomuscular cells in S. tenella. It is possible that these mucoprotein granules are mesogloeal precursors, as well as cuticle precursors, but the difficulty of chemical analysis of the delicate cuticle and mesogloea probably will not allow the testing of such a hypothesis. In any case, the present results certainly lend weight to Chapman's (1966, p.165) statement that "The secretion, on the outside of the ectoderm of various Hydrozoa, of the aminopolysaccharide chitin and on the inside of the mucoprotein and collagenous mesogloea are perhaps not as far removed from one another chemically as was first considered."

As Chapman (1966) says, there has never been much doubt that the function of the mesogloea is to stick the cells and cell layers together and to provide a flexible bed to which muscle fibres can be attached. The folding of the mesogloeal layer in contact with the endodermal cells on contraction of the ectodermal muscles (Pl. 4, Fig. 5, M) is very similar to the process of buckling which occurs in sea anemones (Batham & Pantin, 1951), and may indicate similar mechanical properties of the mesogloea in the two groups.

ENDODERM: There is a marked histological and histochemical specialization of the endoderm in the basal, middle, and apical hydranth page 16 regions. This has also been observed in Pennaria (Cowden, 1965). Such specialization is in full accord with the concept of digestion in hydroids. In these animals, and in many other hydrozoans, after food has been ingested, digestion occurs in two phases (Hyman, 1940). The first phase is one of extracellular digestion, in which ingested food is reduced to a kind of "broth" containing fragments and liquid in the coelenteric cavity. The second phase is one of intracellular digestion following the active ingestion by certain cells lining the coelenteron of the broth produced by the first digestive phase.

In the apical region of S. tenella histochemical tests indicate the presence of three types of gland cells. Type 1 appears to have a mucoprotein secretion, type 2 a predominantly acidic mucopolysaccharide secretion, and type 3 a proteinaceous secretion. In addition, the type 2 gland cell may contain a neutral mucopolysaccharide, or mucoprotein, as it displays moderate PAS positivity, a feature not characteristic of acidic mucopolysaccharides (Pearse, 1960). Cowden (1965) recognises two types of gland in the pharynx region of Pennaria, one which is both PAS and AB positive, and another which stains for protein.

The function of each of these different secretions would be difficult to determine. They possibly act independently, or perhaps together, probably to lubricate the pharynx region during the swallowing of stunned prey. The villated appearance of the pharynx (that is, the endoderm of the apical region) in transverse section has been reported also by Cowden (1965) and Wineera (1968) in other athecate hydroids.

The middle endoderm region is one of both extracellular and intracellular digestion. The gland cells here are probably cells which secrete the enzymes necessary for extracellular digestion. The cytoplasm of these cells is rich in RNA (Pl. 2, Figs. 3A, 3B) and they contain large protein granules which stain in the Millon test for tyrosine and in the DMAB-nitrite test for tryptophan. In fact their morphology and staining reactions are strikingly similar to those of vertebrate pancreas acinar cells, an observation made by Cowden (1965) for similar secretory cells in the middle hydranth region of Pennaria. While Cowden observed the "digestive cell" granules to be both PAS and acrolein/Schiff (protein) positive, in the present study the granules were PAS negative. The timing of the secretory cycle, or rather, the cycle of secretory granule synthesis, also suggests that these cells secrete the enzymes for extracellular digestion. In starved animals the cells are full of protein staining granules while epitheliomuscular cells are very vacuolated. Shortly after feeding (12-18 hrs.) the secretory cells are empty of granules, or nearly so, while adjacent epitheliomuscular cells contain many vacuoles with protein and polysaccharide staining contents.

These results also show that the endodermal epitheliomuscular cells are the site of intracellular digestion. The "food vacuoles" resemble those of protozoans, and it has been stated (Hyman, 1940) that intracellular digestion proceeds in regular protozoan fashion. Hyman also remarks that extracellular digestion is purely proteolytic, while in the food vacuoles the digestion of protein, fats, and in some cases carbohydrates occurs. The endoderm of the basal hydranth region, since it consists wholly of epitheliomuscular cells, is capable only of intracellular digestion. The products of extracellular digestion appear to be transported to other parts of the colony also, since 12-18 hrs. after feeding, the hydrocaulus contains page 17 food vacuoles in the endodermal cells. During periods of starvation the vacuolar contents of endodermal cells probably represent residual food wastes to be excreted. Burnett (1961) describes similar bodies in Hydra which persist during starvation. She concludes that they probably represent some type of excretory crystal.

Hyman (1940) describes two types of solid tentacle in hydroid polyps. One type (present in the Tubulariidae) has a core formed by several rows of endodermal cells. This type is quite different from the tentacles of S. tenella, and will not be discussed further. The other type has a core formed by a single row of highly vacuolated, stiff cylindrical cells. Cowden (1965) also describes the solid core of Pennaria capitate and filiform tentacles as being composed of a single row of endodermal epitheliomuscular cells. To support these statements Hyman has published illustrations, and Cowden photographs, depicting longitudinal sections of solid tentacles. When these are compared with longitudinal sections of the tentacles present in S. tenella (Pl. 4, Fig. 1) the resemblance to the endodermal cores described by Hyman is striking. However the present author disagrees with the concept of the tentacle core cells as advocated by Hyman and Cowden, for three reasons. Firstly, regarding the formation of "cylindrical cells' it is possible that such cells could be formed at the tentacle base by coalescence of cells around the perimeter. But this process would call for massive cellular and subcellular reorganization. Secondly, assuming cylindrical cells could be formed, these cells would possess a base, (that is, the surface in contact with the mesogloea) but no apex; the "base" of the cell would completely envelop it. Now all of the other endodermal epitheliomuscular cells in the hydranth possess basal muscle fibres, and there seems no reason why the muscle fibres of the tentacle endodermal cells also should not be basal in position. Accepting this, the muscle fibre could then possibly be circular (because the cell base is circular), or perhaps there could be a series of fibres (one for each cell used in the formation of the cylindrical cell) each running for a small distance around the cell base. Both alternatives seem far to complex as solutions. The third objection to the "cylindrical cell" concept is that in the polyp the epitheliomuscular cells of the endoderm are also digestive cells and actively engulf particles. It is difficult to understand how the tentacle endodermal cells would accomplish this. If they, however, have lost their digestive function, then it is reasonable to suppose that they are specifically a distinct type of endodermal cell. In this case nutrition supposedly is sustained by diffusion from other cells. However the tentacles are often very long, and it is difficult to see the process of diffusion as being efficient enough to serve the cells in the distal tentacle regions, especially since the tentacles are such active structures in the important process of feeding.

Results from the present study suggest a much simpler arrangement of endodermal cells in solid tentacles which is obtained without the need for gross mophological changes in the cells, and which is still consistent with the appearance of these tentacles in longitudinal sections as described by Hyman (1940) and Cowden (1965): It is evident from Pl. 4, Fig. 1, which shows a longitudinal section of a tentacle base, that the polyp endoderm immediately above and below the tentacle base curves into the tentacle, so that in longitudinal section the tentacle endoderm is seen to be 2 layers thick. It has already been noted above that the endodermal page 18 epitheliomuscular cells have cytoplasm and nucleus distally placed and they possess basal vacuoles (see RESULTS: ENDODERM). In longitudinal sections of tentacles the contact of the two endodermal layers results in a mass of cytoplasm and nuclei in the centre of the tentacle with vacuolated areas between this cytoplasm and the mesogloea (Pl. 4, Fig. 1). It can be seen that this arrangement gives the appearance figured by Hyman (1940) and Cowden (1965). If this explanation for the structure of the tentacle core is correct, it could be expected that the process which causes it, occurs around the whole perimeter of a tentacle base, and not just in the plane of the longitudinal section. It follows from this that a transverse section of a tentacle, particularly near the base, should show a radial arrangement of endodermal cells. This has been observed in the present study (Pl. 4, Fig. 2). The only modification necessary to the epitheliomuscular cells would be the loss of flagella. In the present study also, food vacuoles have been observed in the distal endodermal cells of a tentacle (Pl. 4, Fig. 3). It seems probable that food particles could reach such distally placed cells more easily with the arrangement of tentacle endoderm proposed here, perhaps by flattening of the endodermal cells to form a temporary narrow passage down the centre of the tentacle. If this did occur, the tentacles could be described as not being strictly solid all of the time. Further work is necessary on this problem of food transport in tentacles, but the present evidence strongly suggests that the tentacle core of solid tentacles is formed by ordinary columnar endodermal cells and not by cylindrical cells as described by Hyman and Cowden.