The Formation and Properties of the Organic Matrix of Reptilian Tooth Enamel

The nature of reptilian enamel is of interest in that it has now been established that two quite different types of enamel may be found cover-ing the teeth of different vertebrates. Mammalian enamel is prismatic in struc-ture and is produced by the ectodermal cells of the enamel organ, whereas the enamel covering the teeth and scales of fish (Kvam, 1946, 1950; Levi, 1939, 1940a; Kerr, 1955; Poole, 1956a) and that covering the teeth of am-phibia (Kvam, 1946; Levi, 19406) is not prismatic and its matrix is formed by the mesodermal odontoblast cells of the dentine papilla. Although the teeth of a wide range of reptiles have been examined (Erler, 1935; Schmidt, 1947; Poole, 19566), as yet no prisms have been found in the enamel; however, there is some evidence that the enamel matrix of reptiles has certain staining properties in common with mammalian matrix (Kvam, 1946).

Very many accounts of mammalian enamel formation are to be found in dental and zoological literature; summaries of the most important details are given by Kvam (1946) and Marsland (1951, 1952)-It may be briefly stated here that two main stages are recognizable in enamel formation—the production of an organic matrix with only a slight mineral content, followed by a maturation process by means of which this matrix becomes highly calcified.

The organic matrix is formed by the ameloblasts of the enamel organ. The so-called Tomes’s processes seem to be intimately connected with the formation of the matrix. The two most popular views are that either these processes secrete the matrix, or that they are converted into the matrix as they grow out from the basal ends of the ameloblasts. During its development, the properties of the matrix change a number of times. Soon after deposition it is acidophil and is known as pre-enamel; an influx of calcium salts, probably organic or colloidal in form, produces a change to intense basiphil properties. The enamel matrix remains in this condition, with a mineral content not greater than 35% (Weinmann, Wessinger, and Reed, 1942), until it has reached the final thickness of the future enamel. Immediately before the final heavy influx of mineral a return to acidophil properties is shown, but, as calcification proceeds, there is a withdrawal of organic material and water, the residual matrix becoming soluble in acids. As enamel production occurs, certain changes also take place in the ameloblasts including alterations in the size, shape, and contents of the cells, and in the position of the nucleus (Marsland, 1951).

The finished enamel shows incremental lines, the striae of Retzius, which may also be seen in the matrix before its final, heavy calcification. They appear to be related in some way to matrix production (Marsland, 1951) and the direction of calcification is almost at right angles to these lines. Alkaline phosphatase distribution follows a regular pattern in developing mammalian tooth-germs, although different authors have reached slightly different conclusions about distribution in ameloblasts and odontoblasts. A survey of these results is given by Symons (1955) who concludes that alkaline phosphatase may be initially concerned with cell differentiation and growth rather than with calcification.

In a fully calcified reptilian tooth the enamel is made up of incremental layers which follow approximately the shape of the tooth crown. As in mammals, this pattern is due to the presence of striae of Retzius. Mineral crystallites tend to be arranged at right angles to the layers, and whilst there is an irregular variation of crystallite direction in crocodile enamel (Erler, 1935), an extremely regular variation occurs in the enamel of placodonts (Schmidt, 1947b), gorgonopsids, and cynodonts (Poole, 1956). A regular variation produces a prismatic appearance under certain conditions of examination, but no true prisms are found.

Fibres have been described in sections of fully formed crocodile enamel (Kvam, 1946) which are said to cross one another in three planes (Marcus, 1931 ; Schulte, 1930). Erler (1935) was unable to confirm a three-dimensional fibre network, but found air-filled spaces in the enamel impermeable to liquids. Such air spaces occur in the enamel of many reptiles, fossil and recent (Schmidt, 1947a; Poole, 1956b). It is possible that some of these structural features have been confused by different authors; Kvam (1946), for example, states, ‘longitudinally to the tooth there are undulating fibres’; but neither in the text nor in the illustrations does he distinguish between these and the striae of Retzius.

During development the organic matrix of the enamel of lizards and mammals stains red with Heidenhain’s Azan, contrasting with mesodermal enamel of lower vertebrates, which stains blue in the same way as the collagenous dentine (Kvam, 1946). Because of this it is suggested that reptilian enamel has a keratinous nature rather than collagenous and, therefore, is ectodermal in origin rather than mesodermal.

The object of the work to be described here has been to make a more detailed examination of the formation and properties of reptilian enamel matrix, and to see how far such properties compare with the enamel matrix of mammals.

The results given here were obtained from the examination of embryonic jaws of the lizard Agama atricollis. This lizard was chosen because of the relative ease with which eggs are obtained at certain times of the year, but some other lizards including the local gecko (Hemidactylus sp.) and skink (Mabuia varia) were also studied briefly, as well as a few embryos of the crocodile (Crocodilus niloticus). In all these cases the properties of the enamel matrix were the same. As there appears to be no complete account of the structure of a fully formed lizard-tooth, sections of the teeth of the local monitor lizard (Varanus niloticus) were prepared by grinding and examined under the polarizing microscope. The structure of the enamel was so similar to crocodile enamel that a description is not necessary, but the range of reptiles possessing a similar enamel structure (Poole, 19566) must now include modern Lacertilia.

One or more agamid eggs from a clutch were opened at weekly intervals and the embryos fixed in either Bouin or Susa. Unfortunately, the series of embryos produced was not regular since the rate of development varied from egg to egg. Nevertheless, enough stages were found to provide the main landmarks in tooth formation. After fixation, no further decalcification was required.

Under the prevailing local climatic conditions, difficulties were experienced with the normal technique of dehydrating and clearing, particularly with bulk tissue about to be embedded. To overcome this, dehydration was carried out with mixtures of ethyl and n-butyl alcohol. The tissue to be embedded was transferred from absolute butyl alcohol into a mixture of butyl alcohol and paraffin wax (m.p. 54^o C) and finally into pure paraffin. The technique proved to be very successful, since no appreciable hardening occurred even when the material was left to embed overnight. Sections on the slide were cleared by passing them from 95% alcohol through terpineol into benzene.

Standard sections of each embryo were prepared and stained with haematoxylin and eosin. Harris’s haematoxylin was preferred to others since it gave more delicate differentiation of tissues than Heidenhain’s, and yet was more stable than Delafield’s. Some sections were also stained with Azan, which, as previously mentioned, is said to give a colour differentiation between keratin and collagen.

Finally, alkaline phosphatase tests were made by the technique of Gomori (1939) as modified by Danielli (1946). Again the dehydration and embedding technique was modified by using butyl alcohol. To reduce the time required for embedding, a vacuum oven was used and the pressure lowered with a water pump. An embedding time of h gave satisfactory results.

Agamid lizards have an acrodont dentition, each tooth being ankylosed to the upper edge of the jaw. The tooth tips are flattened laterally and triangular in shape so that each jaw has the appearance of a continuously serrated cutting edge. The embryonic jaws of Agama atricollis possess nine teeth at hatching compared with about twice this number in the adult jaw. During the posthatching growth period new teeth are produced one by one in a pocket at the back of the jaw and, as the jaw increases in length, new teeth are added at the end of the row. This is also found in acrodont chameleons (Rose, 1892). No evidence has arisen from the study of Agama to contradict the general belief that teeth are not replaced in acrodont forms. Nevertheless, although the teeth are all similar in size at hatching, the second tooth in each adult jaw is considerably enlarged. Presumably, in this particular case, post-hatching growth is maintained for some time.

The first signs of tooth formation were found in an embryo 28 days old. Here the oral epithelium had invaginated along the length of the jaw and soon afterwards this lamina gives rise to the nine germs of the teeth which will be present at hatching. The appearance of an early tooth-germ is shown in figs, i, A, B. At this stage it is possible to distinguish an enamel organ consisting of an inner and outer enamel epithelium separated by stellate cells, and a small number of odontoblasts differentiated from pulp cells. The organization of the enamel organ is rather simpler than that of crocodile, which, according to Rose (1893), is the only reptile in which the condition of the enamel organ approaches that of mammals. Although the structure of this lizard enamel organ is clearly seen over the tip of young germs, only the inner enamel epithelium is clearly defined at later stages.

The tissue separating the inner enamel epithelium from the odontoblasts in fig. i, B is pre-dentine. It gives a faint pink colour with haematoxylin and eosin, whilst its collagenous nature is shown by the fact that it stains blue with Azan. Thus, the first tissue produced is dentine and in this respect the reptilian tooth-germ resembles that of mammals and differs from fish, where the mesodermal enamel matrix is completed before dentine formation begins.

The tooth-germ increases in size by the downward extension of the enamel organ ; new ameloblasts are continually produced at the base where inner and outer enamel epithelia can be distinguished until later stages. Inside the ameloblasts new odontoblasts appear and lay down pre-dentine. Enamel is laid down over the pre-dentine at the tip and eventually, as the downward extension of the enamel organ occurs (fig. 1, c, E), it is also laid down over the pre-dentine at the sides of the tooth. As the tooth-germ enlarges, a bony platform develops below to which the completed tooth will eventually become ankylosed. Occasionally the bony platform is not continuous (fig. 1, E) and blood-vessels pass from the mesodermal tissue surrounded by the bone, through a gap and into the pulp cavity. One large blood-vessel is always seen to enter the pulp cavity, but the origin of the vessel varies from germ to germ.

Finally, some weeks before hatching, a pocket is formed at the back of the jaw. Into this extends the dental lamina and a new tooth-germ begins to form. This is the first of the teeth which will be added to the series after hatching. Ankylosis of the nine teeth in each jaw also occurs after hatching.

Fig. i, B is an enlargement of the tip of a young germ; a thin layer of predentine is present but, as yet, no enamel has been formed. The ameloblasts are elongated with centrally placed, granular nuclei; the latter are typically basiphil and the cytoplasm acidophil. The odontoblasts producing the predentine are also long, but in this case the nuclei are at the inner ends of the cells adjacent to the pulp.

As soon as the enamel appears it has very precise staining properties. The appearance of a germ where enamel formation is under way is illustrated in fig. i, c. The staining properties of the ameloblasts are unchanged but they are considerably longer and their nuclei are at the outer ends of the cells (fig. i, D). At the base of the germ no enamel is present and the ameloblasts, with centrally placed nuclei, are shorter than those at the tip. A thin layer of pre-dentine is present beneath them so that the condition at the base of this germ is identical with that over the tip of the young germ described above. Enamel formation is thus preceded by the elongation of ameloblasts, the migration of the nuclei to the outer part of the cell, and the deposition of a layer of pre-dentine. These changes proceed gradually as enamel formation extends downwards from the tip of a germ.

The enamel matrix which is present at this stage stains so intensely with haematoxylin that little or no structure may be seen in it. With Azan, enamel becomes a deep red, the same colour as the keratinous scales which develop on the surface of the embryo; this contrasts sharply with the bright blue dentine and bone. These properties do indeed suggest that the matrix is more like keratin than collagen, as was suggested by Kvam (1946).

In sections which have been stained with Azan or lightly with haematoxylin, the matrix is not of a uniform appearance (fig. 1, D). Fibres are seen throughout and there are alternating light and dark stripes running vertically to the tooth surface and parallel with the general fibre direction. The striped effect results from refraction, for it is seen even in unstained preparations and, moreover, the light and dark areas change when the microscope tube is racked up and down. Opposite each ameloblast a slight bunching of fibres occurs and the refractive index along the axis of a group of fibres is probably slightly different from that between groups, where the density of fibres is somewhat less. Thus, alternating zones of slightly different refractive index are found which produce the refractive stripes. Although the cytoplasm is acidophil, the ameloblasts present rather a similar pattern of darker intracellular contents separated by lighter intercellular zones, as may be seen in fig. 1, D. Fibres running in directions other than that described above have not been observed.

The appearance of the inner ends of the ameloblasts is also important. Because of distortion during the various treatments, the ameloblasts are frequently pulled away from the enamel, and in such places the inner surface of this cell-layer is seen to be very irregular (figs. 1, F; 2, B, D). This irregularity is due to the fact that the end of each ameloblast projects through a terminal membrane into the space caused by the distortion. However, where the ameloblasts and enamel are closer together it is possible to see that these projections pass across and are continuous with the fibrous matrix, and it is evident that rupture of the connexion is only produced by severe distortion (fig. 2, B). Similar cell projections are well known in mammalian enamel formation as Tomes’s processes.

If an unstained section is examined with phase contrast, the fibrous properties of the matrix are again apparent and the continuity of the ameloblasts with the matrix in an undistorted area is especially clear. From the base of each ameloblast a zone of the matrix runs out towards the amelo-dentinal junction, gradually fanning out and becoming rather more fibrous as it extends deeper into the zone of enamel.

It is important to note here that in areas where the enamel and ameloblasts have not been pulled apart, Tomes’s processes are not visible and the deeply staining, basiphil matrix extends right up to the bases of the ameloblasts. Where Tomes’s processes do occur, they stain in exactly the same way as the cytoplasm of the ameloblasts (figs. 1, F; 2, B, D). Therefore it seems very likely that Tomes’s processes are artifacts caused by distortion, and really represent the basal parts of the ameloblast cytoplasm which have been pulled out beyond the margins of the cells. However, Tomes’s processes are only apparent during the production of the matrix ; in fig. 1, c, E, they are present over the developing enamel along the sides of the tooth, but are absent over the tip where the matrix has reached its final width. This is illustrated in fig. i, F, G. Thus, when matrix production has ceased, there is no longer continuity between the contents of the ameloblasts and the enamel matrix, and Tomes’s processes are no longer produced.

Fig. 1, E is of an advanced stage of tooth formation. The staining properties of the matrix remain unchanged but it is now uniformly fibrous in appearance, the bunching having disappeared. This has probably arisen from the untwisting of the fibres. Although they cannot be seen in the illustration, the matrix now shows incremental zones which are evident in the fully formed enamel of all reptiles. Wide zones with the typical staining properties of the matrix are separated by lines which do not stain so readily. These lines also have a slightly different refractive index from the rest of the matrix, for even in unstained sections they can be made to appear light or dark by moving the focus of the microscope. The matrix fibres pass uninterruptedly across the striae and, as shown in fig. 2, A, these properties are seen most clearly in the matrix immediately before the final calcification. Fig. 2, A illustrates another important feature for, at this stage, there is a very rigid attachment between the ameloblasts and enamel matrix, with the result that distortion now tears the matrix away from the dentine. Therefore, although continuity between the matrix and the contents of the ameloblasts disappears at the end of matrix formation, there is a later re-attachment of the ameloblastic terminal membrane to the surface of the matrix. Similar features have been described in the formation of mammalian enamel (Marsland, 1951).

At the hatching stage, no matrix can be seen over the tip of the tooth, although remains of the matrix are found along the sides. Fibres and incremental layer lines may still be recognized in these remains, whose staining properties are somewhat less basiphil than in younger stages. It is suggested that, as calcification proceeds, a large part of the organic matrix is resorbed, but it is unlikely that resorption is complete and, in fact, a slight organic residue can be obtained by extremely slow and careful decalcification of crocodile enamel. This residual organic material becomes soluble in all but the most dilute acids. Finally, the ameloblasts have again altered in shape, being now much shortened with each nucleus appearing to occupy most of the cell. However, at the side of such a tooth as this, development is at an earlier stage and shows all the features of a young germ (fig. 2, c, D); lateral expansion thus continues until the base of a tooth reaches its final size.

From the above description it is evident that developing reptile enamel has a number of features in common with that of mammals. The staining reaction with Azan is similar in the two cases and there is little doubt that the reptilian matrix, like mammalian, is produced by the ameloblasts. The Tomes’s processes of the reptilian ameloblasts are suspected to be artifacts; a similar suggestion has been made in the case of mammals (Kvam, 1946, and others), where the size and shape of the processes is also related to the degree of distortion of the germ. However, the evidence for this is not quite conclusive, with the result that the Tomes’s processes are still regarded by some as a natural feature of the ameloblasts, intimately related to matrix production (Marsland, 1951). In Agama the staining properties of these processes are the same as those of the cytoplasm of the ameloblasts and, therefore, it seems probable that as distortion takes place, the cell contents are pulled out beyond the cell margins. Yet the inner border of the ameloblasts is marked by a distinct membrane which remains in position even after distortion, so that, if Tomes’s processes are produced as suggested here, a means must exist for the cytoplasm of the ameloblasts to pass through the membrane without rupturing it. The terminal membrane in mammals is perforated, spaces alternating with condensations of intercellular material known as the terminal bar apparatus. A similar feature has not yet been recognized in Agama, although the existence of small perforations of some sort might be postulated to account for the production of Tomes’s processes.

Although they differ in staining properties, there is no distinct structural demarcation between the reptilian matrix and Tomes’s processes; the fibres are less distinct in the outermost matrix layers, and gradually merge into the granular processes. The fact that only severe distortion will cause complete rupture illustrates the intimacy of the connexion between fibres and processes. For these reasons it is suggested that the matrix is produced by the basal regions of the ameloblasts which tend to grow out through the terminal membrane and gradually become converted into fibres. Mammalian matrix may also be produced in the same way (Kvam, 1946; Marsland, 1951). In reptiles this process results in an initial bunching of fibres opposite each ameloblast, but eventually the fibres become more uniformly distributed.

Reptilian matrix shows no pre-enamel stage and only a doubtful ‘transitional’ phase, both of which possess acidophil properties in mammals. The layer tentatively described as pre-enamel in the lizard Lacerta vivipara (Kvam, 1946) could be the same as the zone of Tomes’s processes described here in Agama, although, perhaps, there is no reason for making such a distinction. Even in mammals there appears to be no clear-cut distinction between pre-enamel and Tomes’s processes, except that the latter exist as separated units, and it is also possible for the pre-enamel stage to be missing, as in the mouse (Kvam, 1946). In an attempt to account for these varying details, the following developmental plan is proposed.

In both reptiles and mammals, enamel formation begins with the acidophil contents of the ameloblasts growing out through the terminal membrane and becoming transformed into the matrix protein. In reptiles and some mammals, this transformation is rapid and is quickly followed by an influx of mineral converting the matrix into a basiphil form. If, however, the transformation and mineral deposition are delayed, the matrix retains the form and acidophil properties of the cytoplasm and is known as pre-enamel. In either case the matrix remains continuous with the cell contents, so that distortion pulls the cell-walls and terminal membrane away from the matrix, leaving the extenuated cell contents behind as Tomes’s processes. The existence of preenamel will then depend upon the rate of transformation and initial calcification of the matrix, and the length and shape of Tomes’s processes upon the extent to which the ameloblasts are pulled away from the matrix. Thus, until conversion occurs, there will be little structural difference between the pre-enamel, Tomes’s processes, and cytoplasm of the ameloblasts. An attempt to illustrate some of these features has been made in fig. 3.

Striae of Retzius are common to both mammalian and reptilian enamels, yet their significance is uncertain. It has been shown that the final heavy calcification of mammalian enamel advances as a front which is actually perpendicular to the striae (Diamond and Weinmann, 1940) and, moreover, in both reptiles and mammals these zones can be recognized in the matrix a considerable time before the onset of final calcification. In view of these facts, it is now believed that the striae are related to matrix production rather than to calcification. However, there is evidence that the mineral concerned in the initial influx is derived from the ameloblasts (Marsland, 1951). Unless the ameloblasts can produce matrix proteins and assimilate calcium salts simultaneously, these two processes must be separated in time. This would be achieved if, after producing a layer of matrix, the ameloblasts temporarily suspended the production of organic material and began to assimilate and secrete mineral instead. The cells would thus carry on their two separate functions alternately. The periodic change in the rate of matrix production would account for the presence of the striae of Retzius, which, in reptiles, appear as slight periodic differences in staining properties and refractive index along the matrix fibres. In mammals the striae seem to be the result of regular, slight displacements of the prisms (Gustafson, 1945) and this may also be due to the periodic suspension of the normal process of matrix production. On the whole, the zones of normal matrix between the striae of mammalian enamel are much wider than in reptiles, so that the time required to mineralize one zone and convert it into a basiphil form is correspondingly greater. This might account for the fact that an acidophil pre-enamel region exists in the matrix of many mammals, but is missing from reptiles where the thin matrix zones can be converted rapidly.

Results obtained with Agama are only sufficient to indicate that the final maturation of reptilian enamel follows approximately along the same lines as in mammals. Features noted include the firm re-attachment of the terminal membrane to the matrix, the general shortening of the ameloblasts and the residual matrix becoming acid-soluble. A similarity in the final crystallization processes is suggested by the fact that adult reptilian enamel consists of hydroxyapatite with a crystallite size similar to that of mammalian enamel but approximately ten times greater than that of mammalian or reptilian dentine (Little and Poole, unpublished results). This contrasts with the mesodermal enamel of lower vertebrates which has a crystallite size similar to that of dentine (Poole, 1956a).

As shown in fig. 2, E, the enamel organ and pulp cavity are sites of considerable phosphatase activity. However, the ameloblasts remain unstained except for the nucleus and the supranuclear or basal cytoplasm (see Symons, 1955)> which appear grey. It was found that if prolonged periods of incubation are used, up to 24 h for example, the basal cytoplasm and nucleus of each ameloblast stains quite intensely and this could be due to a gradual diffusion of enzyme from the surrounding enamel organ. Such a diffusion has been shown to occur in mammals during both fixation (Lison, 1948) and incubation (Martin and Jacoby, 1949), but it is also possible that the nucleus is behaving as a slowly acting precipitation centre (Johansen and Linderstrom-Lang, 1953), the amount of phosphate produced thereby depending upon the time available for the precipitation to occur. Whatever the correct explanation may be, it seems unlikely that the ameloblasts possess intrinsic alkaline phosphatase at this point in their history.

The inner enamel epithelium at the base of a germ always stains intensely. Here, new epithelial cells are being formed by division and these will eventually differentiate into ameloblasts as the formation of enamel and dentine extends downwards from the tip. Exactly the same condition is found in mammalian germs ; because of this and the absence of phosphatase from the fully formed ameloblast, it has been suggested that, at this stage, phosphatase may be related more to cell differentiation and growth, or to the manufacture of nucleic acids and other materials, rather than to calcification (Symons, 1955)-

Nevertheless, in mammals, phosphatase activity is shown throughout the cytoplasm of the ameloblasts during maturation (Symons, 1955) and, furthermore, phosphatase can be detected in the ameloblasts of fish during the calcification of mesodermal enamel (Kerr, 1955). This evidence has been taken to indicate that alkaline phosphatase is concerned with the process of calcification. However, histological evidence suggests that mammalian ameloblasts are not responsible for the final calcification and that their function is to remove organic material and water from the calcifying matrix (Marsland, 1952). Perhaps alkaline phosphatase is related to the latter function, but further evidence is clearly needed. Unfortunately, because the hardness of the dentine makes section cutting impossible without prior decalcification, it has not yet been possible to discover whether phosphatase is present in the ameloblasts during the maturation of reptilian enamel.

In Agama no general staining of the odontoblasts was ever observed; the inner ends of these cells always stained black, so that it is tempting to suggest that diffusion has also occurred here, in this case from the pulp. However, recent results using azo-dye coupling techniques indicate that mammalian odontoblasts possess intrinsic phosphatase of their own (Symons, 1955) and the significance of the results obtained here is therefore difficult to assess. A feature of interest in the reptilian germ is that the parts of the dentinal tubules passing through calcifying dentine nearly always stain black, whereas the parts of the same tubules passing through the uncalcified pre-dentine remain unstained. Clearly, then, during the calcification of the dentine the contents of the dentinal tubules have the ability to precipitate phosphate either by phosphatase enzyme or by some other means.

Finally, mention must be made of the appearance of enamel and dentine after the phosphatase tests have been made. As shown in fig. 2, F, enamel is intensely black whilst the dentine is only light grey. In controls which have been subjected to the last part of the test only—the precipitation of phosphate as cobalt sulphide—exactly the same pattern of black enamel and grey dentine is produced, indicating that such phosphate is that laid down by the germ naturally. Despite the fact that it stains only lightly, the dentine must possess a considerable amount of mineral, for it is already very hard and brittle, and readily cracks on sectioning. Moreover, it can be made to turn black by prolonged exposure to a cobalt solution. Probably, therefore, the dentine mineral is in the fully formed hydroxyapatite condition and since the exchange of Ca⁺⁺ ions and Co⁺⁺ ions in this mineral is sluggish (Johansen and Linder-strom-Lang, 1953), complete ionic exchange can only occur after a relatively long period of exposure to a cobalt solution. On the other hand, enamel becomes black after only a few minutes’ exposure, suggesting that the mineral here is in a simpler form and is possibly related to the mineral produced under the influence of enzyme in the phosphatase technique. These results confirm the suggestion made earlier that the young basiphil enamel matrix possesses calcium phosphate in some elementary form.

Thus, the behaviour of the mineral involved in the first influx is different from that involved in the final calcification. In the latter process, apparently under the influence of the matrix, mineral crystallizes into orientated hydroxyapatite units of a characteristic size; yet the mineral concerned in the initial influx remains in an elementary condition. Presumably, therefore, during its basiphil stage the matrix undergoes some sort of modification, as a result of which the matrix fibres are rendered capable of exerting an influence over the final calcification of mineral. The striae of Retzius have been shown here to become increasingly apparent as the age of the matrix advances ; this might also be taken to indicate the occurrence of some sort of progressive change in the matrix before final calcification begins.

The properties described below were found in enamel matrix at all stages, irrespective of the fixative used. When a tooth section is examined in water between crossed niçois, both enamel and dentine light up as the surface of the tooth is rotated into the 45° position. If a sensitive tint (first order red) quartz plate is introduced with its slow axis parallel with the tooth surface, enamel appears yellow and dentine blue-green. The birefringence of the enamel is, therefore, negative with respect to the tooth surface, or positive with respect to the axis enamel fibres which are perpendicular to the surface. The positive birefringence of the dentine with respect to the surface indicates that, as in many vertebrates (Poole, 1956a, b) the positive collagen fibres are orientated parallel with the surface of the tooth.

A significant difference in the behaviour of the enamel and dentine matrices was found when sections were mounted in aqueous phenol solution and reexamined as above. Under these conditions both the enamel and dentine are yellow, indicating that the sign of birefringence of the dentine has been reversed. This reversal in phenol solution is a property of collagen not shared by other skeletal proteins (Frey-Wyssling, 1953), and is to be expected in dentine but not in enamel if the latter is keratinous. However, the optical properties of the enamel matrix were found to depend very much upon the refractive index of the solution in which sections were mounted. A summary of the properties in various mounting liquids is given in table 1.

By means of the Becke line test the refractive index of the enamel matrix was found to be about 1-57; young matrix was very slightly less than this, and mature matrix slightly greater. It may be seen from the table that the fibrous matrix has no biréfringent properties when it is mounted in a liquid with a similar refractive index of 1-57. Birefringence is only produced by mounting in liquids whose refractive indices are either significantly greater or smaller than i*57, with maximum birefringence produced at the extremes (1-33 and 1-76). All this points to the fact that the matrix fibres have no intrinsic birefringence ; but since they are parallel with each other they form, together with the mounting liquid, a Weiner mixed body. Such a system produces a form-birefringence positive with respect to the direction of orientation of the micelles, the magnitude depending upon the diflferencè between the refractive indices of the two components. If these two indices are the same, the system is optically isotropic.

The behaviour of the enamel matrix in paraffin wax is very interesting. During the examination of newly prepared sections before the wax had been removed, it was found, even under these conditions, that the matrix showed not only distinct activity but also, rather surprisingly, a birefringence which was negative with respect to the fibre axis; that is, opposite in sign to the form-birefringence described above. As the wax is dissolved away in a solvent, the negative birefringence is replaced by the positive form-birefringence. These properties were investigated further by taking wax sections and heating the slide until the wax became molten. In the molten medium the fibrous matrix showed, as in any other liquid, the typical positive form-birefringence ; but as soon as the wax solidified the negative birefringence reappeared. It was possible to repeat these reversals indefinitely by alternately warming and cooling the slide.

The explanation of this phenomenon is probably to be found in the properties of the crystallized paraffin. If paraffin molecules are orientated parallel with each other (e.g. by flow or crystallization), the system shows a birefringence which is positive with respect to the direction of orientation. Paraffin wax crystals are flat plates with the paraffin molecules perpendicular to the plane of the plate. These plates may be joined together, edge to edge, to form elongated flat lamellae which appear needle-like when seen in side view. In such lamellae the wax molecules are again perpendicular to the plane of the lamella and also perpendicular to the axis of elongation of the lamella. Thus, in side view, the lamella exhibits a negative birefringence with respect to the longitudinal axis and because of this is known as a ‘negative streak’ (Frey-Wyssling, 1953). These facts can now be applied to the behaviour of wax in enamel matrix. In molten form the paraffin wax molecules will be randomly scattered and so the medium behaves normally as a liquid to produce a positive form-birefringence. As the wax crystallizes, negative streaks are formed which become orientated parallel with the enamel fibres. Thus the positive form-birefringence becomes replaced by the negative intrinsic birefringence of the paraffin streaks as crystallization occurs. These properties are illustrated in fig. 4, A, B.

The method by which the negative streaks become orientated is a matter of speculation. It may simply be a mechanical process with the paraffin streaks tending to slip into the spaces between the fibres, but other properties of paraffin crystallites suggest that the mechanism of orientation may be more complex than this. In strips of cold-drawn polyethylene it is found that the molecules become orientated approximately parallel with the direction of drawing. If paraffin wax is allowed to crystallize on the surface of such strips the paraffin molecules become aligned parallel with the polyethylene molecules with the result that the crystal plates of paraffin are arranged with the plane of the plate perpendicular to the direction in which the polyethylene strips are drawn (Richards, 1951). Similarly, in spherulites of polyethylene where molecules are arranged tangentially, molecules of crystallizing paraffin again become parallel with those of the polyethylene with the result that the paraffin streaks are arranged radially across the concentric layers (Wilems and Wilems, 1956). From this it is evident that the arrangement of paraffin streaks is secondary since it depends upon the initial direction in which the constituent molecules become orientated.

A similar mechanism may determine the arrangement of paraffin streaks in enamel matrix. If so, the matrix must possess some sort of cross structures running at right angles to the fibres. Such structures would determine the primary orientation of the paraffin molecules perpendicular to the fibres axes and would result in the secondary arrangement of negative streaks parallel with the fibres, as suggested in fig. 4, c. The coarse fibrous appearance of the enamel matrix is, perhaps, only the visible manifestation of a more delicate submicroscopic pattern. The latter would need to possess two main components; first, fibrils lying parallel with each other and at right angles to the tooth surface which set up the positive form-birefringence; secondly, side groups lying perpendicular to and possibly connecting the fibrils together. The whole system thus forms a lattice grid, the cross members of which determine the orientation of paraffin molecules. It has been shown that the crystal structure of polyethylene is very closely similar to that of paraffin wax (Bunn, 1939), and is therefore ideal for the orientated overgrowth of crystals (Wilems and Wilems, 1956). If the above suggestions are correct, the side groups running between the fibrils in the organic matrix may also possess a crystal structure related in some way to that of paraffin wax.

The orientation of crystals by organic fibres is of special interest in the study of calcified tissues, for in bone, dentine, and mesodermal enamel (Schmidt, 1938, 1940; Poole, 1956a), and in reptilian enamel as described here, crystallites of hydroxyapatite are arranged with their crystal c axes parallel with the organic fibres axes. Recent electron microscope studies (Jackson and Randal, 1956; Fernández-Morán and Engstrom, 1956) have made it possible to ‘see’ the intimate relationship between mineral crystallites and organic fibres, but the precise mechanism of mineral orientation, whether or not it involves crystal overgrowth, is still obscure.

The evidence presented here indicates that the process of enamel formation in reptiles is essentially the same as the corresponding process in mammals. The main difference is that whereas developing reptilian matrix is continuously fibrous, mammalian matrix is broken up into units, the prisms, each one having been produced by one ameloblast and isolated from its neighbours by interprismatic material. Nevertheless, the fibres of young reptilian matrix have been shown to be arranged in groups which correspond with the ameloblasts; the production of new material to separate the groups from each other would result in a condition very similar to that found in mammals and, presumably, this is the most important change which took place in the evolution of the mammalian type of enamel from reptilian.

It has been suggested that the terminal bar apparatus in mammals may be associated with the production of interprismatic material (Marsland, 1951), so that it is of special interest to note that these heavy intercellular condensations at the bases of the ameloblasts have not been found in the reptiles where the enamel lacks prisms. The transition of prismatic enamel from the homogeneous reptilian type, therefore, may have been achieved by the development of the terminal bar apparatus, which, by the production of a material differing somewhat from the matrix proper, separates the products of the ameloblasts into prisms.

As regards structural properties, it must also be noted here that although mammalian matrix shows no fibres even when examined with the electron microscope (Little, 1956), it has, nevertheless, a faint but definite positive birefringence with respect to the axes of the prisms (Schmidt, 1934; Keil, 1937). This property has been taken as an indication of an organized fine structure within the matrix which might be responsible for the orientation of mineral crystallites parallel with the prisms. However, it is clear that the orientated, possibly molecular, units producing the birefringence must be of very narrow width if they are beyond the resolution of the electron microscope. Consequently, the second important change in the evolution of prismatic enamel was the production of a more refined and delicate matrix instead of the coarse fibrous type characteristic of reptiles. As has already been pointed out, the fibrous appearance of reptilian matrix may only be the outward manifestation of a more intricate submicroscopic structure.

The evolution of prismatic enamel from reptilian seems, therefore, to have been a comparatively simple step, and tubular enamel, found in most marsupials and a few placentals, must be considered to be a later specialization of the simpler prismatic type. Tubular enamel possesses both prisms and tubules, the latter arising at the amelodentinal junction in the so-called ‘clumsy’ joint. The origin of these tubules is still not clear, but they are known to be interprismatic and may be occluded by the deposition of calcium salts in older enamel (Sprawson, 1930). Although X-ray analysis shows the presence of hydroxyapatite crystallites orientated approximately parallel with the prism axes, as in normal prismatic enamel, marsupial enamel exhibits most unusual optical properties when treated with dehydrating and clearing agents (Poole, 1952). These properties may be in some part due to the high organic content of the enamel. The possession of prisms shows that tubular enamel must be related to the more normal mammalian type, but the tubules, the peculiar optical properties, and the relatively high organic content all point to a specialized type of prismatic enamel.

In higher vertebrates, therefore, three distinct types of enamel matrix exist, all of which result from the calcification of a matrix derived from the ectodermal ameloblasts. Similarly, in lower vertebrates, different types of enamel are to be found with the common feature of being formed in a matrix derived from the mesodermal odontoblasts (Poole, 1956a). The phylogenetic relationships between mesodermal and ectodermal enamels have been discussed by Kvam (1946, 1953), who considers the first step in the transition from one to the other to have been the production of a tissue jointly by mesoderm and ectoderm. In general, it has been tacitly assumed that both the production and the calcification of mammalian matrix are performed by the ameloblasts, and the suggestion has repeatedly been made that even in fish the ameloblasts may furnish the mineral required for the calcification of the mesodermal matrix (Tomes, 1898; Kvam, 1946, 1953; Kerr, 1955). If this is so, the only further requirement for the evolution of an ectodermal enamel is the ability of the ameloblasts to produce an organic matrix.

However, evidence that has gradually accumulated over a number of years suggests strongly that much of the mineral involved in the calcification of mammalian enamel is derived from the dental pulp (see Marsland, 1952). Marsland himself came to the conclusion that, although the initial mineral influx originates from the ameloblasts, none of the calcium salts required for the final calcification comes from these cells and their function during maturation is to withdraw organic material and water from the calcifying matrix. Thus, there is a double source of mineral salts for the calcification of mammalian enamel and the phylogenetic transition suggested above is not quite so easy to accept. In fact, no hypothesis can be regarded as satisfactory until the source of the mineral used in the calcification of the mesodermal enamel has been clearly established.

Finally, brief reference may be made to the problem of the organizing activities of tooth-tissues. In mammals it is generally believed that as soon as the enamel organ and dentine papilla are completed, the ameloblasts produce an organizer which induces the odontoblasts to begin dentine formation; later, the ameloblasts themselves are induced by the odontoblasts to produce enamel. In lower vertebrates also it is possible that the ameloblasts organize the activities of the odontoblasts, but the induction of the ameloblasts by the odontoblasts is lacking and, presumably, this is a necessary condition for the production of a wholly ectodermal enamel. There appears to be no direct evidence for such an organizer from the dentine papilla even in mammals; nevertheless, in the somewhat similar process of feather production, where mesodermal and ectodermal tissues also work in close association, there is evidence that the mesodermal papilla induces activity in the ectodermal cap, culminating in the formation of a feather (Wang, 1943). This is regarded as an example of a ‘secondary’ organizer acting at a late stage in the life history of an animal (Waddington, 1956), and it is tempting to suggest that the induction of ameloblastic activity by the dentine papilla, which also occurs at various stages in life, may be a parallel process.

It is hoped that, as a result of this account of the formation and properties of reptilian tooth-tissues, some of the relationships between the various types of vertebrate enamel are a little clearer. Much of the discussion has necessarily been speculative, but further research may solve many of the outstanding problems, such as the points of origin of the first type of ectodermal enamel and of the later prismatic type. The study of the teeth of primitive mammals and their possible reptilian ancestors would be of great value, and it is hoped that such material may sometime become available.

In conclusion, the author would like to express his gratitude to the Research Grants Committee, Makerere College, for providing funds for the purchase of a polarizing microscope; to Professor A. I. Darling, University of Bristol Dental School, for a number of helpful comments on the work undertaken; and to Professor L. C. Beadle, Zoology Department, Makerere College, for reading and criticizing the manuscript.