The Molluscan Radula: Its Chemical Composition, and Some Points in Its Development

THE SIOLLUSCAN BADULA. 115

The Molluscan Radula: its Chemical Composition, and Some Points in its Development.

By Igerna B. J. Sollas.

With Plate 9.

HISTORY. THE molluscan radula, or dental ribbon, has been the subject of research for at least a century and a half. Aristotle (3), though he speaks of teeth in Limax, alludes apparently to the ridges on the jaw, and there is no evidence that he knew of the existence of the radula: but it is interesting to find that the great naturalist was well aware of the fact that whelks bore holes in shells with the proboscis, although he cannot have fully understood the process. Poli made a jest of the tale as a fable, but Osier re-affirmed it in 1832 without knowing of previous work, and is now credited with having been the first to observe this interesting habit. Swammerdam is the discoverer of the radula: he gives a description of both the radula and jaw of the snail (Helix aspersa), in Dutch and Latin, in his 'Biblia Naturae/ Leyden, published posthumously. His death, as we are told by Boei'haave in the Life of the author prefixed to this work, occurred in 1680. The work is now too antiquated to possess more than an historical interest. In 1757 Adanson (1) described radulae from various gastropods of Senegal: the teeth are "infinitely small,hardly 116 IGERNA B. J. SOLLAS. visible, though sometimes perceptible to the touch. Looked at with the microscope . . . the pointed ends of the teeth are turned towards the stomach like those of the tongue of the lion or cat." Adauson observed the regular arrange- ment of the teeth, and in some cases counted them, finding 20,000 teeth in 200 longitudinal rows in a bulimoid land- snail which the uatives call " Kambeul," and 200 in 10 longitudinal rows in Patella. Poli (13), in 1791, was the first to give a clear figure of a radula in his magnificent work ' Testacea utriusque Sicilise.' Troschel (22), in 1836, first established the radula as an organ of great systematic importance. Curiously enough, in the same year van Beneden, in a paper written in 1835, and not quoted in the literature of this subject, points out the possible value of the radula in determining the reality of doubtful species. Troschel's work attracted the interest o£ zoologists to the radula; after an interval, in which Lebert, Allman, and particularly Loven, worked along the new lines, Troschel published his ' Gebiss der Schnecken' (1856-1863) —a general and masterly work now well known. His interest was not restricted to the form of the teeth, but extended to their chemical composition. Though Troschel was the first to make the suggestion—thrown out apparently as a shrewd guess—that growth takes place at the posterior end of the radula to make good the waste going on in front, yet lie did not follow it up by closer study, nor did he investigate the development of the organ. It has been one of the chief problems of later workers, but they have arrived at somewhat conflicting results. By combining a study of the chemical composition with that of the development some of the difficulties which have ai'isen may be removed.

CHEMICAL COMPOSITION. In 1845 Hancock and Embleton (6), in a study of the anatomy of Bolis, state that the radular teeth consist of silica. They base their conclusion on the partly mistaken observation THE MOLLUSCAN RADULA. 117

that the teeth do not dissolve in either acetic or nitvic acid, while hydrofluoric acid coiTodes them. No particulars of their experiments are given. The same authors investigated the teeth of Buccinum and came to the same conclusion. In 1852 Leuckart (9), being interested in the distribution of chitin in the animal kingdom, examined, among many other objects, the radulse of Gastropoda and Cephalopoda, and pronounced them to be chitin. He emphasised the fact that his identification of chitin rested entirely on two characters—one its resistance to caustic alkali, the other its solubility in boiling nitric acid. He adds: " It is possible that in this sense chitin is a collective conception, and that many special modifications will be discovered later. Perhaps we may conclude this from the varying behaviour of chitin when treated with alkali," and he expresses a wish that chemists would investigate the matter. About the same time Bergh (4), without knowing of Leuckart's paper, con- futed Hancock and Embleton's view, and demonstrated the absence oE silica in three species of Prosobranchiate Gastropods. Bergh's is the first exact investigation; we are indebted to Troschel for a German translation of an extract of his paper, which is written in Danish. Bergh showed that in Buccinum antiquorum (Triton nodif ernm), and in Strombus gibberula most concentrated acids bring about corrosion of the raduia in the cold, and eventually complete solution on boiling, while dilute hydrofluoric acid does not alter the teeth in form, but renders them more transparent. Incinerated ribbons of Marsenia perspicua gave no silica. The raduia of Buccinum antiquorum gave the reactions of iron and calcium phosphate. Troschel was dissatisfied with what he considered the con- tradiction in the results of Leuckai't and Bergh,1 and therefore undertook with Bergemann experiments which combined and reconciled the results of both these investigators. Helix, 1 This remark seems hardly fair to Leuckart, who nowhere states that chitin is the sole constituent of the raduia and is not interested in the ash ; of Bergh's paper I have only read the extract given by Troschel. 118 IGERNA B. J. SOLLAS. Patella, and Doliuin were chosen for study, attention being directed both to the jaw and radula. The radula of these thi'ee forms was found to behave in a similar manner and consists of an organic constituent, chitin, together with the inorganic constituents, iron, calcium, carbonic, and phosphoric acid. It was further shown that the radula of Helix nemoralis contains 5 per cent, of ash, that of Dolium gale a 6 per cent. Koehler's paper, published (7) in the same year as Troschel's fDas Grebiss der Schnecken,' deserves a word of mention, since this observer also affirmed the presence of both an organic and an inorganic constituent, and suspected the occurrence of calcium. Later workers continue to make divergent statements in describing the chemical composition of the radula. Sollas (19) in 1885, when studying the nature of the silica in organisms generally, made use of measurements of the refractive index and specific gravity; he concluded that in the molluscan radula silica was present and that, as in so many organisms, it was in the form of opal (silica hydrate), but he does not mention the species on which his observations were made. Bloch (5) and others speak of the radula as chitin, but their views do not appear to be based on original observation. Similarly some modern text-books refer to this organ as composed of cliitin or conchiolin, others speak of it as siliceous. Huxley and Ray Lankester, with more caution, do not commit themselves on this point. It thus seemed worth while to look once more for definite evidence of the presence or absence of silica and of chitiu in the radula. It will conduce to brevity if I state at once the general results I have obtained. I find that in all the odontophorous Mollusca the radula has an organic basis of chitin; the Docoglossa are unique among Mollusca in the composition of their teeth, of which the most important constituent is silica hydrate or opal. All the other groups, including the Rhipidoglossa, form a second type in which the radular chitin is hardened superficially by deposits containing calcium, iron, and phosphoric acid, which, together THE MOLLUSOAN RADTJLA. 119 perhaps with ati additional organic substance, form that outer covering so long known as the enamel layer but hitherto unexplained. I have not been able to confirm Troschel's statement that carbonic acid is present, and though I have made repeated attempts, I have failed to determine whether magnesium is one of the mineral constituents. These points, therefore, must still be left to chemists. The Chiton- idfe present us with a deviation from the second type and staud alone among the forms I have examined. In this family ferric oxide is the most important mineral constituent and is the cause of the dark colour of the teeth. With the partial exception of Helix aspersa, the ash of the radula preserves the form of the teeth. In the first type of radula or that of the Docoglossa the mineral matter may form as much as 27 per cent, of the whole l'ibbon, this is the case in Patella vulgata; while in the second type it may contribute only 2"4 per cent., as in Helix aspersa, though in this species it sometimes rises to 3"3 per cent.; in Dolium galea it amounts, according to Troschel's analysis, to 6 per cent. It is interesting to find that the Docoglossa, which are so well-defined a group in other respects, differ so widely, not only from the Pectinibranchiata, but also from the Rhipido- glossa in the composition of the radula. We may commence our more detailed account with the Docoglossa, and first with

PATELLA. If the radular ribbon of Patella is boiled in strong nitric acid, the organic parts are completely dissolved, and there sinks to the bottom of the test-tube a coarse-grained insoluble residue. Microscopic examination shows that this consists of the dark red brown cusps or free biting ends of the lateral teeth (fig. 1), together with some thin, colourless, transparent pieces; some of these latter are free, some remain attached to the cusps, showing that they are the skeleton of the basal 120 IGEBNA B. J. SOLLAS. part of the lateral teeth. This last point may be confirmed by performing the solution in the cold, when all the hard parts are left in their natural relative positions. Prolonged boiling with nitro-hydrochloric acid or hydrochloric acid and potassium chlorate in an open test-tube failed completely to remove the dark red-brown colour of the cusps; to accomplish this the teeth must be placed with one of these solvents in a sealed glass tube and heated in a water bath for some days. After this treatment some of the cusps become completely freed from their iron content and appear perfectly colourless and transparent; others, however, after digestion for an entire week retain some of their red-brown colour. The specific gravity of the teeth thus prepared was ascertained by means of a diffusion column,1 in which the majority were found to float in a dense zone at a level corresponding with a specific gravity of 1'98, though numerous examples ranged on each side of this to 1*87 on the one hand and 2-08 on the other. This variation in specific gravity corresponds with a difference in the degree of hydration of the silica : experiments made on colloid silica show that when this is prepared from water glass it possesses a specific gravity of 1"86 and contains 16'3 per cent, of water; when obtained from silicon fluoride, its specific gravity is l-98, and the water amounts to 9"85 per cent.; while sponge spicules with a specific gravity of 2-04 contain 7 per cent, of water. Thus it would appear that the water in the siliceous basis of the Patella teeth varies in amount from 7 to 16 per cent, with a mean of 9-85 per cent. It is interesting to observe in this connection that the association of water and silica in the silica hydrates occurs without any change of volume in either of the constituents, and thus the proportion of water in the hydrate can be directly calculated from the specific gravity. In the following table the results of such a calculation are given for a 1 Tor an account of this method see Sollas, W. J., " Physical Characters of Calcareous and Siliceous Sponge Spicules," 'Proc. Koy. Soc. Dublin,' vol. \v, p. 378, 1885. THE MOLLUSCAN RADULA. 121 small number of cases, the specific gravity of amorphous silicon dioxide being taken as 2-21. SiO3) HSO, water 23 per. cent., sp. gr. 1-73 (SiOs)s, (H2O)2) water 16-6 per cent., sp. gr. l-85. (SiO2)2, H2O, water 13 per cent,, sp. gr. 1'9 ; (SiO2)3 H2O water 9*1 per cent., sp. gr. 2-0.

(SiOs)4, H2O, water 7 per cent., sp. gr. 2*04. Under the microscope the teeth which have been treated as described above present a faint brown granular appearance when seen by transmitted light and a bright milky white by reflected light: in this behaviour, which results from the abundant presence of minute pores, they resemble common opal. Between crossed Nicols they exhibit faint but evident double refraction with, undulose extinction; the same character is sometimes presented by mineral opal, and is attributed to internal stress, the existence of which in the teeth is suggested by their liability to spring apart along fractures parallel to their length. The refractive index of the siliceous residue of the teeth was determined by Becke's method and found to be closely approximate to 1*45; the fluids used were mixtures, one of 10 parts, the other 20 parts of Price's glycerine, to 1 part of water. Taking the refractive indices of these mixtures to be 1*449 and 1*454, the above result is obtained. Teeth from which the iron has not been completely extracted are distinguished by a relatively high specific gravity • under the microscope the ferric hydrate or oxide presents a blood-red colour by transmitted light and marked double refraction. The ribbon of Patella, when soaked for some hours in strong hydrofluoric acid, becomes colourless or nearly so throughout, but the forms of the teeth are perfectly preserved. If it is next boiled in nitric acid it dissolves completely, as might be expected, since the siliceous matter has already been removed by the hydrofluoric acid. On incineration the ribbon retains its form with a surprising completeness. The ash proves to contain, in addition to silica already 122 IGERNA B. J. SOLLAS. mentioned, a noticeable quantity of calcium, iron, magnesium, and phosphoi'ic acid. The radula after treatment with hydrofluoric acid proves very resistant to prolonged boiling in caustic potash (5 per cent, to 40 per cenb.) ; this suggests that it is composed of chitin. With a view to testing this, radulas were treated with hydrofluoric acid, boiled for a considerable time in a 5 per cent, solution of caustic potash, which was frequently chauged, and were fiually extracted with water, absolute alcohol, and ether. The specific gravity of the organic portion of the radula which remained after these processes was the same as that of chitin obtained from the carapace of Astacus, and subjected to the same treatment—viz. T40. The specific gravity was determined by a diffusion column formed of chloroform and absolute alcohol. I find that chitin from various sources has a specific gravity differing but slightly, if at all, from that of Astacus, and I hope to publish details on this subject shortly. The refractive index of this organic basis of the radula was found to lie between the same limits as that of chitiu, viz. l'55O and 1"557. There is an interest of a general nature attaching to the fact that the radula of molluscs has a chitinous basis, since Bloch has brought forward special arguments (5) to prove that this membi'ane is not produced by the direct metamorphosis of the formative cells, but by secretion. Bloch is able to cite a number of workers who share his view, and he refutes the arguments of Wiren, the only supporter of Trinchese in the suggestion that the radula is formed by direct metamorphosis of cell protoplasm. Chatin, on the other hand, in 1892, maintained that the chitin of Libellulid larvas is formed by direct transformation of the chitinogenous cells. After soaking in hydrofluoric acid the entire ribbon and teeth stain with boi-ax carmine and other preparations, though with differences in intensity in the various parts. When the fresh radula is treated with the ordinary stains this is not the case, as may be seen from fig. 8. The ribbon in this figure has been stained first with Bethe's stain and afterwards THE MOLLUSCIAN RADULA. 123 with borax carmine. Bethe's stain acts readily on those parts which resist borax carmine and remain unstained in ribbons treated with borax carmine only.1 The action of hydrofluoric acid renders it possible to make sections of the radula; these reveal the organic matter as a solid basis having the form of the complete structures. With the ordinary stains the cusps are less deeply coloured than the roots of the lateral teeth (text-fig. 1), with Bethe's stain the bases of the roots of these teeth and parts of the marginals

t.s.m. TEXT-FIGUKE 1.—A longitudinal section of the radula of Patella vulgata after treatment with strong hydrofluoric acid and staining with iron hsematoxylin. I. s. m. Tensor superior muscle. 6. e. Basal epithelium, c. Organic basis of the cusps, r. e. Roofing epithelium.

1 Bethe's method: the object is placed in a 10 per cent, aqueous solution of anilin hydrochloride to which one drop of fuming hydrochloric acid is added for every 10 c.c. of water. After washing thoroughly the object is transferred to a 10 per cent, solution of potassium bichromate. It will be seen that at the young end of the Patella ribbon each row of teeth is uniformly coloured by the carmine stain, the youngest teeth are paler than those immediately in front of them. In front, again, of the darker red teeth the roots of the laterals and centrals (inner laterals) are coloured green by Bethe's stain, their cusps being still red like the marginals. Finally, as we pass forwards, the cusps become golden-yellow and then red-brown owing to the presence of iron oxide, and the innermost marginal bears a band of greeu. The contrast in staining properties between the marginals and remaining teeth is very striking. The basal membrane itself is red, but in the preparation it has been removed as far as possible, as it obscures the differentiation in specimens mounted whole. These differences in staining properties are to be found with greater or less deviations in the radula of other odontophorous molluscs. 124 IOEBNA B. ,T. SOLLAS. alone are coloured; so that in this case there is a curious reversal of the usual relative behaviour of these stains. I have not found any other case in which BetheJs stain will colour in section a structure which has an affinity for the oi'dinary staining reagents, though this is commonly the case in working with material in bulk. The above description refers to Patella vulgata; it is equally applicable to P. pellucida, as well as to a number of other species of Patella sent me by the kindness of Pro- fessor Mitsukuri, and probably to all species of Patella. To Professor Mitsukuri I am also indebted for specimens of Nacella, the teeth of which, isolated by boiling in fuming nitric acid, are shown in Fig. 6. The marginals, as in Patella and all Docoglossa examined, do not contain a siliceous skeleton; they dissolve completely in nitric acid and are consequently not represented in the figure.

OKTPTOBBANCHIA. My thanks are due to Mr. Rathbun, of the United States National Museum, for specimens of Cryptobranchia concentrica. The lateral and central teeth possess a skeleton of siliceous pieces which are fused in each row into a single plate, but it is possible in this plate to detect outlines which seem to mark the limits of once separate teeth. Apparently there were three laterals and one central, the central tooth being much reduced. In the darkly coloured cusps the line of demarcation between the outer and inner laterals is clear, but the two inner laterals of each side are closely fused together (text-fig. 2).

ACMAEA. Specimens of Acmaea virginea were obtained from Plymouth. I owe to the kindness of Dr. Harmef examples of Acmaea saccharina, as well as of many other molluscs. Marginals being absent in this genus the greater part of the radular substance is silica; the solution of the organic matter THE MOLLUSCAN RADULA. 125 sets free from the portion corresponding to each row of teeth a pair of substantial siliceous plates of rectangular outline when seen en face. On each of these are seated the red- brown cusps of the lateral teeth. The cusps are detached from the plates by further boiling. The basal pieces of each

TEXT-FIGURE 2.—A siliceous basal plate and the cusps carried by it, isolated from the radula of Cryptobranchia concentrica by the action of boiling nitric acid, as' Surfaces of attachment of the cusps and plate to one another.

side of the radula, though free from those of the other side, are closely fused among themselves, leaving no trace of the outlines of separate pieces. The teeth, after treatment with nitric acid, have a refractive index which closely approaches that of chloroform; 1*4 being a little lower in the case of A. virginea, and a little higher in that of A. saccharina, there is a corresponding difference in the value of the specific gravity, which was found to be 2'1 in the latter and 2'0 in the former species. It is open to -doubt, however, whether this difference is constant, for the examples investigated were far from numerous. 126 IGEENA B. J. SOLLAS.

LEPETA. I take this opportunity of thanking the Rev. Professor Gvvatkin and Dr. 0. Nordgaard for examples of this genus. Lepeta coeca.—The siliceous basal pieces of each row are united into a single piece as in C. concentrica, and as in that species the basal plate is divided by longitudinal lines into six areas, whereas the coloured cusps appear to be formed by the union of four pieces. In this case, however, the double basal piece belongs to the outer cusps.

RHIPIDOGLOSSA. Classified according to the chemical composition of the radula, these forms belong to the second type mentioned above (p. 118), in which the proportion of mineral matter in the radula is small and does not include silica. In all the forms belonging to the second type which I have investigated the teeth can be isolated from the membrane by cold nitric or hydrochloric acid. Teeth of Trochus ziziphinus freed in this way and washed were found to have a refractive index lower than 1-557, and higher than l-550. The same is true of all the radula) of this type. The refractive index of cliitiu lies between the same limits. Fig. 12 shows the result of staining the radula of Trochus ssiaiphinus with heematoxylin ; it is necessary to isolate the teeth by teasing, as the membrane stains darkly and the teeth are so close-set that it is impossible to make out details in the radnla stained intact. All the teeth take the stain at their roots; in the series of marginals the length which takes the stain increases as we pass outwards along a transverse row, while, finally, on the outer side of the marginals is a paddle- shaped piece made up of five flat pieces, placed edge to edge (the flabelliform teeth). This piece stains completely, and contrasts with the teeth lying to its inner side. This contrast is most marked in the younger parts of the radula. The THE MOLLUSCAN RA.DULA. 127 marginals pass by many gradations of form into the lateral teeth. These outermost pieces, on the other band, present a sudden change in the series of marginals, being longer than their neighbours, and agreeing in their staining properties with the marginals of Patella. In fact, the reactions to stains almost tempt us to suggest that the teeth generally regarded as marginals are multiplied laterals, and that the marginals are represented by these deeply staining teeth at the outermost edge of each row. Haliotis and Emarginula give somewhat similar results.

T.ENIOGLOSS.A. Littorina littorea furnished the chief and most con- venient material for the investigation of this group. The radulee were dissected out from a thousand individuals of the species, and dried at 100° C.; they weighed 0'430 grin., and afforded, on incineration, 0-0158 grm. of ash, or 3-7 per ceut. This was found to contain iron, calcium, and magnesium. A second experiment was made on the ribbons of several thousand individuals, from which 0'0508 ash was obtained; this yielded 0"0083 of phosphoric acid, or rather, of P2O6, corresponding to 16*3 per cent, or to 35'6 of calcium phosphate. A contrast between the staining reactions of the marginals and the central and lateral teeth exists here as in the case of Patella, though it is less marked (fig. 9), and the teeth run through the same stages as regards staining capacity during their growth as in that genus, but in the long radula sheath it is noticeable that the teeth retain for a long time their affinity for the common stains, and only quite near to the mouth-cavity become green under the action of Bethe's stain. Their great hardness is discovered when an attempt is made to cut sections of the buccal mass. Previous treatment with nitric acid is necessary for success in this process. 128 IGERNA E. J. SOLLAS.

RHACHIGLOSSA. In Buccinum undatum, as there are no marginal teeth, the ribbon, doubly stained by the above method, does not become marked out, as in previous cases, into a green median region, bordered on each side by a red marginal band; none of the teeth in the fully developed parts of the radula are coloured by protoplasmic stains, but all take Bethe's stain. It is noticeable that the teeth are very rapidly matured—a necessary property in a tongue which must be quickly worn out. In a specimen of Purpura lapillus I found every one of the teeth in the mouth-cavity with cusps broken off. It would seem as though the molluscan radula had become adapted to hard work in two distinct ways: in the one case it becomes very resistant through a long-continued process of development—witness the long radula sac—in the other a less strong radula is rapidly worn out and as speedily renewed.

POLMONATA.

Helix aspersa.—A remarkable discrepancy was found to occur in the behaviour of the l'adula of this species when incinerated on different occasions. Several lots containing from twenty to one hundred and fifty radulae were heated, in some cases with an ordinary Bunsen burner in a platinum crucible, in other cases with a Herapath. These experiments were all made in winter-time, and all gave the following result: the radula at first preserved its form in ash, but soon fused with continued heating: it gave an exceedingly minute fusible drop, which solidified to an enamel-like glass, yellow when hot, white when cold. In winter, also, the ash was always found to contain some silica: the ash from some 150 ribbons was boiled in fuming nitric acid, in this reagent it proved only partially soluble; the insoluble residue was THE JIOLLUSCAN BADULA. 1.29 washed in water and returned to the platinum spoon and again heated : it was then infusible and was presumably silica. In a quantitative analysis the amount; of silica was found to be about one third of the total ash. In one case, in the mouth of April, 500 radulse were incinerated with a Bunsen and the ash afterwards ignited in a Herapath with a dissimilar result from the preceding ; no fusion of the ash occurred, and in the analysis which followed no silica was found. On the other hand, the presence of a large quantity of phosphoric acid was indicated: the ash, which weighed O'OlOl grm., yielded 0-0036 grm. of P2O5, or no less than 35-6 per cent.

DEVELOPMENT.

The history of the study of the development of the radula has been so excellently set forth by Rossler that on this point I cannot do better than refer to his paper (15). It has already been stated that there is a certain amount of disagreement among those who have studied the growth of the radula. The radula passes backwards into the radular sac, which is the seat of its renewal; it exhibits a continuous forward movement due to growth from behind takiag place simultaneously with wear of the front end which is in use. The radular sac is an evagination of the wall of the foregut; it consists of a basal epithelium and a roofing epithelium; these two epithelia pass into each other by way of a heap of cells at the posterior end of the sac. In front of this cell aggregate, immediately behind and below the last formed tooth, are a set of cells, the odontoblasts, which may be few and large or more numerous and small. The roofing epithelium extends between the young teeth at the posterior end of the sac, fitting closely into the spaces between them; it, in many cases, secretes a cuticular substance in the anterior portion of the sac which likewise fits into the spaces between the teeth, so that the teeth lie in pits of this secretion. All

VOL. 51, PART 1. NEW SKKIES. 9 130 IGEBNA B. J. SOLLAS. are agreed that the odontoblasts give origin to the teeth, but some would have it that the roofing epithelium secretes a kind of enamel or hard outer layer which is spread over the surface of the tooth originally laid down by the odontoblast. Among those who take this view areRossler (15), Riicker (14), Sharp (18), and Bloch (5). Bloch describes the enamel layer as consisting of specially hard cuticular substance. Two writers, more recently—Rottmann and Schnabel—deny the presence of any such enamel, and maintain that the teeth are laid down from the first in their definitive form and size, and that the roofing epithelium contributes no substance whatever to the radula, and is not to be regarded as secretory. There is further difference of opinion as to the exact method by which the forward movement is brought about, and on other points. The facts which have been already stated in dealing with the Docoglossan and other radulas place beyond all doubt the importance of the roofing cells of the radular sac in all cases; in the case of Patella the lateral teeth as first formed are soft and colourless, and it is only those situated at a considerable distance from the odontoblasts which betray by their yellow colour the presence of iron oxide, and this must have come from the roofing cells. The other changes in the maturing teeth all point to the secretory nature of the roofing epithelium, a function which is strongly suggested a priori by the conspicuous accuracy with which the cells of this epithelium fit in between the teeth, leaving not the minutest portion oE tooth or membrane surface untouched. Schnabel and Rottmann were working with material other than Docoglossa, namely with Gastropoda and Cephalopoda. Therefore it will be worth while to consider at any rate one of these cases in greater detail. That some considerable changes do take place in the teeth after their first formation is already clear1, but the changes as seen in microscopic sections are interesting and afford some further light. The young teeth, as we know from dissection are soft; they take protoplasmic stains but slightly, and agree in this with the THE MOLLUSOAN RADULA. 131 part of the membrane on which they are seated. They show lamination, the laminte running parallel to the matrix cell (Rossler's]. Soon, as we pass forwards along the ribbon, there is a sharp contrast in staining properties between the young teeth and the underlying part of the basal membrane (text-fig. 3), but this contrast is not maintained, the increased

TEXT-FIGURE 3.—Longitudinal sections of the radula of Helix aspersa, stained with hffimatoxylin, showing four stages in the development of the teeth, a, the youngest tooth of the radula; d, a tooth from the mouth-cavity; b and c, intermediate stages. (Drawn with Zeiss camera lucida, Zeiss obj. D., eyepiece 4.)

staining power spreading downwards through the thickness of the membrane but always leaving a thin layer which preserves the original resistance to stains. At the same time that the membrane darkens the teeth, which nearer the origin were uniformly darkly stained, acquire a lighter peri- phery, the core remaining darkly stained; finally the surface layers become quite colourless and only some scattered small spheres of darkly staining matter remain in the interior. These spheres are arranged with a certain definiteness in rows which converge towards the posterior basal part of the tooth. Such are the appearances seen in sections treated with a protoplasmic stain only, whether haematoxylin, safranin, borax carmine, or carmalum. If we first apply Bethe's stain and then safranin or carmalum, the whole section will be coloured pink with the exception of the cores of the adult teeth, which are green, and the surface layers of the same, which are colourless. These outer layers are Sharp and Rossler's enamel. Eossler made the interesting note that they are not doubly refracting, while the other parts of the 132 IGEBNA B. J. SOLLAS. radula are so. But they are evidently not formed, as Rossler suggested, as a special secretion apposed to the outer surface of the tooth as first formed, but rather by intussusception. With regard to other doubtful points, Rossler assumed that the same odontoblasts secrete all the teeth that are produced during life; but later writers have thought that these cells must be periodically replaced, for, as Bloch points out, "unless we suppose the odontogenous cells to be replaced from time to time we cannot understand how larger teeth are formed with the growth of the animal, for it is unlikely that cells continually engaged in active secretion can also grow" (p. 381). The new odontoblasts, according to Bloch, are formed out of the same cell complex from which the upper epithelium is renovated. This assumption, he thinks, contains no improbability " since the upper epithelium and the odontoblasts have the same task, namely, the secretion of chitin." But from what has preceded it is clear that the functions of the roofing epithelium and of the odontoblasts are very different; this need uot, however, impair Bloch's main argument that the odontoblasts are replaced; there is no reason why a single cell complex should not give off chitm-secreting cells on the one hand and cells secreting mineral matter on the other. The single apical cell of the stems and roots of various plants does more than this. Again, Kossler considered that the radula must glide forwaz-d to a certain extent over the basal epithelum, which he said is delayed by the action of the retractor muscle (tensor superior muscle of Amaudrut). Though I do not agree with his view, Rossler, in alluding to the action of the muscle, seems to me to have touched on an important problem which has been quite neglected by some authors. To this I shall return presently. Bloch objects to Rb'ssler's view: " Ich kann mir nicht denken wie die Basalplatte, die die Ziihnchen tragt, und nach den Praparaten init ihrem Bpithel in innigen Verbindung zu sein scheint, iiber diesen Zellschicht hinweggleite. Da ist nur eine Moglichkeit niimlich die Zelle bewegen sich mit der Basalplatte nach vorn." In the paragraph which follows he THE MOLLUSCAN BADtJLA. 133 seems to say that the two structures keep pace in the younger end of the radular sheath, but that later the epithelium may be delayed. This admission amounts to granting that the epi- tiielium throughout its length cannot keep pace with the membrane except by stretching—if this word can be permitted to stand for the conversion of the high columuar cells of young basal epithelium into lower cells. For if equal increments of epithelium and of membrane are added in the radular sac, then clearly there will be no gliding of one structure over the other in any part of their length : they will keep pace with each other. If, however, the increments of epithelium are less, then cells of the epithelium must "stretch" or increase in volume in order to keep pace with the overlying membrane. It is highly probable that the increments of epithelium are less than those of the membrane, for in the mouth-cavity the epithelium is generally lower than in the sac; in Helix aspersa one cell in the basal epithelium in the mouth-cavity covers as great a length of membrane as four cells in the sac. Consequently if we were to assume that equal lengths of epithelium and of basal membrane were added in the sac, this would involve a considerable relative movement of the epithelium and membrane in the mouth-cavity, the epithelium moving more quickly than the membrane, and this is wholly unlikely. That the odontoblasts are replaced by fresh cells derived from the cell aggregate at the extremity of the radular sac seems most probable and is, I think, the view which has gained acceptance; at the same time, investigators differ among themselves as to whether the replacement is gradual, so that each group of odontoblasts secretes several teeth before it passes on into the basal epithelium, or, more sudden, each odontoblast group only secreting once before it is relieved by recruits. On full consideration this view will be found to involve a somewhat remarkable life-history for the odontoblasts. Starting from the indifferent cell mass from which they arise by cell-division, they become elongated and form a set of cells which possess as a whole a definite and constant shape. They secrete chitin first for the teeth, next for the basal membrane 134 IGERNA B. J. SOLLAS. and are then described as exhausted. But they now pass on and become the youngest cells of the basal epithelium, shortening till they are of a uniform height with their neighbours. They then travel forward, adhering to the basal membrane and gradually shorten still further. As they continue their course they encounter, as they leave the radular sheath to enter the mouth-cavity, the superior tensor muscle of Amaudrut (2). Now they have to play anew and arduous r61e: they must adhere to the radular membrane on the one hand and make connection with the tensor muscle on the other, and their tensile strength must be at least as great as that of the pull from the muscle. Sti-ange to sa,y, they next become liberated from the muscle again and pass forwards, now as a low epithelium, until they encounter another muscle, the inferior tensor, when some of them become connected with it. After this they once more move forwards to form part of the walls of a groove (the sublingual groove of Bossier) which is the natural outcome of the mode of growth of the radula and which allows of the free play of the buccal cartilages in eating. In conclusion I wish to express my indebtedness to Professor Sollas for much suggestion and help in carrying out the work contained in this paper, particularly in connection with the observations on the relation between the specific gravity of silica hydrate and its degree of hydration.

BIBLIOGRAPHY.

1. ADANSON.—' Histoire Naturelle du Senegal,' 1757. 2. AMAUDBTJT.—'Ann. Sci. nat.,' (8), vii, 1898, pp. 1-291. 3. ARISTOTLE.—' De Animalibus Historia,' iv, 4, 7, 8, 9. 4. BERGH.—Extract in Troscliel (23) from ' Konigl.Danske Videnskabernes Selkabs Skrifur,' 5th Raekke, 3 Bind. 6. BLOCH.—' Jena Zeitschr.,' xxx, 1896, p. 350. 6. HANCOCK AND EMBLETON.—'Ann. Nat. Hist.,' 1845, xv, p. 9. 7. KOISHLER.—' Zeitscbr. Naturw.,' 1856, viii, p. 106. 8. LEBBRT.—' Muller's Archiv,' 1846. 9. LEUCKART.—' Arcb. Naturg.,' 1851, i, p. 25. TSE MOLLUSCAN RADULA. 135

10. LOVEN.—' Ofv. Ak. Forh.,' 1847. 11.v .MiDDENDOnr.—'Mem. Ac. St. Petersb.,' 1847. 12. W. OSLEK.—' Phil. Trans.,' 1832, p. 97. 13. POLT.—' Testacea utriusque Siciliee,' 1791. 14. RUCKER.—'Ber. Oberhess. Ges.,' 1883. 15. ROSSLER.—' Zeitschr. wiss. Zoo!.,' xli, 1885, p. 447. 16. ROTTMANN.—Ibid., lxx, 1901, p. 651. 17. SCHNABEL.—Ibid., lxxiv, 1903, p. 651. 18. SHARP.—'Inaug. Dissert. Wiirzburg,' 1883. 19. SOLLAS.—'Proc. Dublin Soc.,' 1885, n.s., vol. iv, p. 374. 20. STEEKI.—'Proc. Amer. Phil. Soc.,' 1893. 21. SWAMMEKDAM.—' Biblia Naturae,' Leyden, 1737; 'Bibel der Natur Leipzig, 1752; 'Book of Nature,' 1757. 22. TBOSCHEL.—' Arch. Naturg.,' i, 1836. 23. TBOSCHEL.—' Das Gebiss der Sclmecken,' 1856-1863. 24. VAN BENEDEN.—'Ann. Sci. nat.,' 1835.

EXPLANATION OF PLATE 9, Illustrating Miss Igerna B. J. Sollas' paper on "The Molluscan Radula: its Chemical Composition, and Some Points in its Development."

I'll}. 1.—Lateral teeth of Patella vulgata, isolated by nitric acid. 1, Two inner laterals (with broken cusps), showing the overlap of the basal pieces ; 2, an outer lateral; 3, an inner lateral; 4 and 5, basal pieces from which the cusps have become detached. EIG. 2.—Lateral teeth of Patella vulgata which have been subjected to the action of nitro-hydiochloric acid in a sealed tube. FIG, 3.—A portion of the ash of an incinerated radula of Patella vulgata. FIG. 4.—a, A siliceous basal plate from the radula of Acmaea virginea bearing three cusps, isolated by nitric acid, b, c, The same less magnified and showing the three cusps separated from the basal piece, by further action of nitric acid, and adhering together. FIG. 5.—A siliceous basal plate from the radula of Acmaea saccharina bearing cusps. 136 IGERNA B. J. SOLLAS.

FIG. G.—Lateral teeth of Nacella. sp. a, Outer'; b, inner lateral tooth. The portion a" becomes isolated (by boiling acid) from a, leaving a'. FIG. 7.—A lateral tootli of Chiton sp. isolated from the radula by the action of strong cold nitric acid, showing the hollow chitinous basal portion and the brown cusp which contains iron oxide. PIG. 8.—Portions of the radula of Patella vulgata in order of succession (a—d) from the radula sac to the mouth cavity. Bethe's stain and carmalum. FIG. 9.—Portions of the radula of Li ttorina littorea in order of succession (a—d) from the radula sac to the mouth-cavil.y. Bethe's stain and eosiu. FIG. 10.—Portions of t.he radula of Buccinum u.ndatum in order of succession (a—d) from the radula sac to the mouth-cavity. Bethe's stain and eosin. EIG, 11.—A rough sketch of a strip of the radula of Trochus ziziphinus including the marginals and laterals of three rows of one side. Stained with carmalum. PIG. 12.—Isolated teeth teased out. of a radula of Trochus ziziphinus which had been previously staiired with heematoxylin. TIG. 13.—Eight teeth from two rows of the radula of Helix aspersa, including two centrals. Bethe's stain and eosin ; surface view. TIG. 14.—Two teeth and the underlying basal membrane of Helix aspersa in longitudinal section. Bethe's stain and safranin. ElG. 15.—Three teeth and the underlying basal membrane of Helix aspersa in longitudinal section. Stained with safranin only. "n )

3. '

c/..

111 , i , I

(-~-

14'.

15. 1Q

Uut.li i.s.W I... HE M0LLU5CAN RAOULA