The interest attaching to a precise knowledge of the structure and development of the scales of fishes is twofold. Not only may it throw light upon the affinities between groups of existing fishes and their evolution, but it helps to elucidate the problems involved in regard to tooth-genesis; for there can be but little doubt that scales and teeth are homologous structures.

It was from the latter point of view more particularly that I was led to an investigation of the subject, which I commenced during the summer of 1902, in the Gatty Marine Laboratory at St. Andrews, where, through the kindness of Professor McIntosh, F.R.S., I was permitted to work.

In the first instance I confined myself to an examination of the structure and development of the scales in the Gadidæ, especially in G. virens and G. callarias, of which I obtained fairly complete series from Professor McIntosh’s rich stores. I further examined specimens of G. minutus, G. æglefinus, G. merlangus, and G. pollachius. Some of the results obtained were laid before the Zoological Section of the British Association at its meeting at Belfast in the form of a preliminary communication.

Since that time I have further examined material supplied to me from the Marine Laboratory at Millport, as well as a specimen of Gadus argentina kindly given to me for examination by the late Professor G. B. Howes, F.R.S.

From my own experience and from the expressed opinion of other workers in the same field, a succinct account of the earlier literature is a much needed desideratum, and this I have endeavoured to furnish, the more recent and generally known papers being omitted.

The first to note the presence of markings on the surface of the scales of fish was Petrus Borellus in 1656 (2). He thus describes them in ‘Observatio,’ xxxvii (p. 23) : “Squamæ piscium apparent si aspiciantur, lineis orbicularibus multis distinctæ, et in parte qua cuti adhærent radiis ac punctis multis transcurrentibus eas divisa.” In his accompanying illustrations he figures these lines and points over a limited area of the scale only, which is quite in accord with the condition found in some scales, though he makes no reference to this point in the text. This observation is of interest as being one of the earliest results of microscopic investigation. Nine years after this (1665), R. Hooke (10) published still further details upon the same subject. He seems to have investigated the matter from a more comparative standpoint, for though he describes the scales of the sole and dog-fish only, giving illustrations of the former, he says he has examined the scales in “multitudes of others, which it would be too long to enumerate.” This early account of their structure is sufficiently interesting to merit quotation. He writes (p. 162) : “This skin I view’d, was Head from a pretty large Soal and then expanded and dry’d, the inside of it, when dry, to the naked eye, look’d very like a piece of Canvass, but the microscope discovered that texture to be nothing else, but the inner ends of those curious Scolop’d Scales, which on the back side, through an ordinary single magnifying glass, look’d not unlike the Tyles on an house.

“The outside of it to the naked eye, exhibited nothing more of ornament, save the usual order of ranging the Scales into a triagonal form, only the edges seem’d a little to shine, the finger being rubb’d from the tail-wards towards the head, the Scales seem’d to stay and raze it. But through an ordinary magnifying glass, it exhibited a most curiously carved and adorned surface, each of those (formerly almost imperceptible) scales appearing much of the shape [as shown in his figure], that is, they were round and protuberant, and somewhat shap’d like a Scolop, the whole scale being creas’d with curiously wav’d and indented ridges, with proportionable furrow between, each of which was terminated with a very sharp transparent bony substance, which, like so many small Turnpikes, seem’d to arm the edges,” of which he further on says “every other of these are much longer than the interjacent ones.” He noticed that the scales were but partially imbedded in the skin, and remarks that “the texture or form also of the hidden part …. seems to consist of a great number of small quills or pipes, by which, perhaps the whole may be nourished ; and the side parts consist of a more fibrous texture, though, indeed, the whole scale seem’d to be of a very tough grisly substance like the larger’ scales of other Fishes.”

With reference to “the Scales of the skin of a Dog-fish (which is us’d by such as work in Wood, for the smoothing of their work, and consists plainly enough to the naked eye, of a great number of small horny points) through the microscope appear’d each of them curiously ridged and very neatly carved ; and indeed, you can hardly look on the Scales of any Fish, but you may discover abundance of curiosity and beautifying; and not only in these Fishes, but in the Shells and crusts or armour of most sorts of marine animals so invested.”

The next writer upon the subject was Leeuwenhoek in 1685 and again in 1698. He noted the presence of scales and fins in the eel and apprehended that this discovery “is new, at least to persons of the Jewish nation (for to this day they deem this delicate Fish to be unclean, and hold it as an abomination to them),” according to the Mosaic law, which regards as such “whatsoever hath no fins or scales in the waters” (Lev. xi, 12). Consequently he figures the scale in this fish, a figure which is highly creditable. Moreover, he noted the imbrication of the scales and the fact that they varied in size in different regions of the body. He makes the assumption that the scales may be taken as an index of the age of the fish, for he describes the appearances thus : they were principally “composed of a kind of globules or little balls … lying in rows contiguous to each other” which “produced the appearance of divers circles or rings on the face of the scale. And although I did not observe these scales to be exactly alike, yet the circles or rings seemed to me to be of the same number in all of them, whence I was led to conclude, that the scale had been every year augmented by the addition of one circle, and consequently that, as there were seven circles in this scale, this Eel was probably seven years old.” A similar thing to that which “is evident in trees” or “shown in horns whence we gather that as many knots or rings as are found on the cow’s horn, so many years of age is the animal.”

On the authority of Mandl, Leeuwenhoek is stated to have subsequently abandoned this view of the rings on the scales as affording an index of age.

An examination of the scales was made by means of sections and the appearances as seen under the microscope are represented. He concludes that a scale is formed in each year, each succeeding scale being somewhat larger1 than its predecessor and “glued” to its under surface. From the description given, one is led to the conclusion that Leeuwenhoek regarded the rings as the edges of succeeding scales, though he does not expressly state it. This point is of great interest in comparison with Williamson’s work (20) on the ganoid scales in 1842. That this view was not accepted at the time is shown by his remark, that “this assertion of mine is however violently contradicted, because many people think that I cannot by any means prove what I affirm” ; he consequently proceeded to demonstrate his facts by means of sections through the scales.

The paper ends with some very interesting remarks on the longevity and causes of death in fishes.

During the eighteenth century many other writers made short, but unimportant, references to this subject, among them being Reaumur, Petit, Schaeffer, Ledermüller, and Fontana, an admirable review of the literature of this period being given by Mandi (15). Réaumur gave the name “Argentin” to the translucent material on the surface of scales, and a rough chemical analysis of it was made subsequently by Heinrich Rose, and his results incorporated in a paper by Ehrenberg (5).

Kuntzmann (13) recognised that the variation in scales may be of generic value, and from his observations he was led to classify the scales of fishes into seven groups, which include, but under different names, the Ctenoid, Cycloid, and Clupeoid of to-day.

The first investigator to seriously deal with the subject in a truly scientific manner was Agassiz in his work ‘Recherches sur les Poissons Fossiles,’ in 1834. He here makes use of the scales as a basis for his classification of fishes into the Placoids, Ganoids, Ctenoids, and Cycloids, the latter including the Clupeoids. While mentioning that the scales are contained in pockets of the skin and are not retained in position by means of blood-vessels, as Leeuwenhoek appears to have believed, he nevertheless adopted the idea of the latter, slightly modified, in regarding each newly-formed scale as being successively attached to the lower surface of the preceding scale.

The next paper of importance is that of Mandl, already referred to, in which the histological structure of the scale is examined in a more thorough and detailed manner. He drew attention to the fibrous substratum of the scale, which he accurately figures. The surface is described as composed of “corpuscules,” which Mandl recognised as being quite separate from one another. Radiating from the centrum or “foyer “are longitudinal lines, and connecting these transversely are what he calls the cellular lines. He likens the upper surface of the scale to cartilage, the cells of which he regards as being the “corpuscules.” Many of Mandl’s interpretations will not hold at the present day, but from the point of view adopted in the present paper the isolation of the individual corpuscles is of considerable interest. In addition to this, he came much nearer to a correct interpretation of the peripheral growth of the scale than had been done by any previous observer.

In 1873 Baudelot contributed a lengthy paper to the ‘Archives de Zoologie Experimentale.’ He examined the scales of a great number of fishes, furnishing elaborate statistics and instituting comparisons relating to the various measurements of scales, the number of concentric markings, etc. Beyond this he adds but little to the knowledge of the morphology of the scale. His descriptions of the patterns upon the upper surface agree in the main with that given by Mandl, but using different terminology. Baudelot, however, recognised the calcified nature of the upper layer and its growth by the deposition of calcareous molecules in the membranous zone forming the border of the scale. He also gives a short account of the actions of various chemical reagents, the results of which I have been able to confirm.

Following upon this we have the classical works of Johannes Millier, O. Hertwig, Peters, Williamson, Hofer, Klaatsch, and Nickerson. These are all too well known to need any further reference as to the advances in our knowledge of this subject which each has contributed.

It must not be supposed that the foregoing pretends to afford in any way a complete bibliography, the more important of the earlier works alone having been dealt with in chronological sequence, in order to furnish a short historical sketch of the subject under consideration.

The first trace of the scales in Gradus callarias appears when the animal is between 3 and 4 cm. in length. The skin consists of an epidermis, which is very readily detached, and which, in the majority of preserved specimens, is altogether wanting. This is a fact of some importance which should be borne in mind, since the absence of this layer in much preserved material is doubtless the cause of the discrepancy in the accounts given by writers as to the part played by the epidermis in the formation of the scale. In a freshly caught fish it can be seen, as a soft, thin, almost gelatinous membrane, which tends to become separated when the animal is put into any other than its natural medium. A surface view of this membrane is shown in Fig. 1, taken from an animal 11cm. in length, Fig. 2 representing a section through the same membrane in a somewhat older specimen. It consists of rounded or slightly fusiform cells, crowded together, each cell having a relatively large and well-marked nucleus. Opening on to the free surface are large numbers of spherical glands, the openings of which are proportionately small. Chromatophores are distributed throughout the thickness of the epidermis as well as on both its inner and outer aspects.

On comparison with the epidermis of Lepidosteus, though there is a certain general resemblance, there are points of marked difference. In this animal, Nickerson (16) describes two kinds of glandular structures—one, the larger, oval in shape, with the long axis perpendicular to the surface of the body, and which are irregularly distributed ; the other, considerably smaller, more numerous and spherical, the majority of which occur “near the surface, where many of them open.” Nickerson remarks that the latter “do not appear to have been recognised by previous observers.” He thinks that they are the true mucin secreting glands, the former having “the function of secreting some component of slime other than mucin.”

From a comparison of the epidermis in the two forms, the glands present in the cod correspond to the mucinsecreting, the larger being unrepresented, slime not being present in the cod. When viewing the surface the rounded ends of the glands are visible, and give to the tissue the appearance of ordinary vegetable parenchyma. Lying upon the surface of the cells are numerous chromatophores, distributed with a considerable degree of regularity. Comparison with the gar-pike shows that the presence of the larger glands in the latter cause a more reticulated appearance, the wider meshes representing their cavities.

Beneath this layer is a thin, but well-defined basement membrane, corresponding, I take it, to Huxley’s protoraorphic layer, the readily detached epidermis being consequently ecderonic, the tissue lying subjacent to the basement membrane being enderonic.

The enderon is clearly divisible, even from an early stage, into two layers, an outer composed of elongated, nucleated, closely-set cells, the boundaries of which are distinguishable with difficulty; so numerous are they, that this might almost be designated as the “nuclear layer.” It stains deeply with borax carmine, offering a marked contrast to the next or deeper layer of enderon, in which the tissue is more fibrous, the fibres running longitudinally. The nuclei here are more sparsely scattered, larger, and more elongated than in the previous position. Beneath this and separating it from the muscle is yet another layer of pigment cells, which vary considerably in size and are very unevenly distributed. The conditions are represented in Fig. 3.

Comparison with the dermis of Lepidosteus (16) shows the same general disposition of parts, though differing in details. For example, I have not been able to distinguish the more complicated interlacement of fibres noted by Nickerson.

In the cod the scales arise in the more superficial nucleated layer of the dermis, their first indication being local aggregations of cells, as evidenced by their nuclei. A horizontal slit-like cavity forms in the interior, the floor of some of the larger spaces being raised into a small conical elevation projecting upwards into the cavity. In none of my specimens has this papilla been much more marked than is represented in Fig. 3. In the majority, I have been unable to detect any such elevation. Whether when present they are merely artefacts, or whether they are to be regarded as very diminutive homologues of the dermal scale papillae of Selachians may be open to doubt. If so comparable, it might be objected that more than one such elevation ought to be found in each cavity. It is possible that this may be only a central one, and that more may arise at a little later stage, a stage which so far I have not been fortunate enough to meet with.

The limits of the cavity are sharply defined from the surrounding tissues by the still closer arrangement of the nuclei ; this is rather more evident on the floor of the space. In the latter situation there appears to be a very thin layer immediately bounding the cavity, from which nuclei are absent, the layer having a homogeneous aspect. The nuclei in connection with the scale cavities are those of the “Scleroblasten “of Klaatsch, the thin homogeneous layer being the first indication of the scale. Nickerson writes (loc. cit., p. 121) : “The fact that in Selachians the scale is formed over the surface of the papfllæ, while in Ganoids (Lepidosteus) and in Teleosts it arises in the midst of the mass of cells forming the elevation is a fundamental difference not to be overlooked.” This account of the development of the Teleostean scale is evidently drawn from the description given by Klaatsch. The description and figures given by the latter of the early conditions seen in the trout do not correspond in detail with what I have just described in the cod. Here the scale, if I have interpreted the appearance aright, arises as an apparently homogeneous layer covering the surface of what may be regarded as the dermal papillae, precisely comparable to the early process of scale formation in the Selachians. If such be the case, then the “fundamental difference” between Selachians and Teleosteans of which Nickerson speaks does not exist.

It is possible that the difference between the observations of Klaatsch and those just related may be more apparent than real. If the vertical extent of the scale-cavity were reduced, which quite conceivably might be the case in correlation with the varying size of the scale, then the roof and floor of the cavity would be approximated and the scale-anlage lying on the surface of the latter would give the appearance as if arising in the midst of the mass of cells. Such, I believe, from a comparison of the figures, is a very possible explanation of the differences.

Following upon this stage is the laying down of the calcareous material. I am not in a position to say whether this appears upon the surface of the homogeneous layer as a kind of secretion, or whether the homogeneous layer itself becomes impregnated, but I incline to the latter view. One point is, however, certain, namely, that the calcareous material does not form an uninterrupted layer, but is in the form of minute isolated platelets, the external surface of each of which bears a minute pointed and backwardly directed spine; in other words, each is a minute microscopic placoid scale: this is shown in Fig. 4. These are placed upon a basal membrane which stains with borax carmine, and which can be seen extending upwards between the contiguous ends of adjacent denticles.

This condition has been entirely overlooked by previous observers, the reason being not far to seek. The skin has been treated with an acid, either for the purpose of decalcifying, or following the routine method of placing in acid alcohol after staining with borax carmine. I have been careful to avoid all trace of acid in the preparation of the specimens, and thus prevent any decalcification. Naturally, the sections are somewhat liable to be torn, and are difficult to cut, but the results are sufficiently satisfactory to leave no doubt as to the existence and mutual relationships of these minute placoid scales. In order to test this explanation, I took portions of the skin of the same animal and from the same region of the body, and treated it with borax carmine and acid-alcohol. The sections obtained showed precisely the characters figured by Klaatsch. The scales are seen on section to form an elongated continuous plate from which project several small pointed elevations at the caudal and cephalic ends. Owing to having been decalcified, the organic basis of the placoid scale has become stained, like the sub-stratum below and between them, and thus gives the appearance of continuity. Though Klaatsch figures these tooth-like elevations, he does not appear to have recognised their morphological value. Indeed, in his lengthy memoir I have not noticed any particular reference to them, beyond the general statement (p. 196): “Der Spitzentheil der Placoidschuppe ist mit wenigen ausnahmen vôllig reducirt. Die Oberflache der Teleostierschuppen geht nene Komplikationen ein, welche zu mahnigfachen Relief bildungen fiihren.”

The scale continues to increase at its periphery by the addition of more denticles, so that meeting and overlapping the adjoining scales they become imbricated. As a result of this growth, that part of the dermis superficial to the scale cavity becomes stretched and thinned. Accompanying this increase in size of the scale, there is a progressive conversion of the subjacent nucleated layer into fibrous tissue, which thus in course of time presents a laminated appearance, certainly as many as five layers coming to be ultimately formed. I would draw attention in passing to the fact that this conversion into fibrous tissue is from above downwards—that is, it follows the direction of growth which Huxley maintained to be the character of enderonic tissues. The bearing of this point will be discussed subsequently.

As a result then, of the continuous growth of the scales in a closed cavity, there is stretching of the less rigid tissues of the roof, and at the same time the “pull “exerted by it would tend to crowd the denticles together, and also to keep the growing margins of the scale upturned. This mechanical factor is, I believe, of considerable importance in producing the modification in shape which the denticles undergo in their further development.

By the continued growth of the scale, the posterior margin reaches the surface of the derma, lying between it and the superposed epidermis. It is more than probable that this is the cause of the tendency of the epidermis to become separated. As far as I have observed it, there is no corresponding elevation of the epidermis, the scale lying altogether beneath it. It remains throughout in the scale-pocket (Schuppen tasche, Klaatsch), which becomes stretched over the superficial surface of the scale, following all the irregularities. It is this pocket which mainly acts in retaining the scale in position. In larger scales—for example, in the herring—the pocket becomes still more stretched and attenuated, possibly even ruptured, and this would account for the ease with which the scales may be detached in the Clupeoids.

I have been unable to find in the literature any adequate account of the fully-formed Teleostean scale. They have been described as “variously sculptured,” “ringed,” “marked with concentric and radiating lines,” etc., but all such descriptions are misleading and fail altogether to interpret their true structure. From the following account it will be seen that the cycloid scale is a complex structure. It consists of a fibrous substratum which preserves the general form of the scale. Upon the upper surface of this are placed a number of calcified plates, or “scalelets “as they may be termed. Covering the entire upper surface is a very delicate membrane distinct from the thickened epidermis which clothes the surface of the body generally, and which is in reality the thinly stretched outer wall of the scale-pocket above mentioned. The fibrous stratum may best be revealed by placing the scale for some minutes in per cent, hydrochloric acid, which dissolves away the greater part of the scalelets, leaving only their outline in organic material, which may easily be removed, by scraping with a scalpel. The same result may be attained merely by scraping without the previous addition of acid, but more force is necessary and there is consequently a greater tendency to a tearing and disarrangement of the underlying fibrous lamellae. This sub-stratum consists of distinct lamellæ of very delicate fibres, having the appearance of ordinary white fibrous tissue, though when torn the individual fibres tend to curl after the manner of the yellow elastic variety. They are extremely resistant to the action of carmine or hæmotoxylin, but are readily stained with eosin or picro-nigrosin. The bundles of the superficial lamella are eccentrically arranged, each bundle corresponding with a ring of the superposed scalelets. Subjacent to this is a layer the bundles of which interlace with each other at right angles, each running diagonally to the long axis of the scale. Between these two lamellae, and also underlying the latter, the fibres appear to form reticulate but much thinner strata. How far these last are to be regarded as definite layers, or merely the results of “teasing out,” I have not been able satisfactorily to determine. The number of lamellæ present varies in different parts of the scale, being most numerous at the centre, while at the periphery the superficial excentric layer alone seems to be present. In the scales of a large common cod about 2| feet in length five layers are to be seen at the centre in transverse section.

The fibrous strata were noted as long ago as 1839 by Mandl (14), and again by Baudelot (1), but seem to have escaped the notice of some recent writers. Mandl, though not describing the arrangement of the lamellæ, appears from his accompanying illustration to have observed them correctly. Baudelot, on the other hand, describes the layers as being very much more complicated than I have detected. He says ; “Le trajet de tous ces faisceaux ou plans fibreux est extrêmement compliqué et des plus difficiles à démêler” (p. 165). The summary of his observations may be given in his own words : “Chaque plan fibreux n’offre pas le même texture dans toute son étendue ; dans sa portion centrale, c’est-à-dire celle qui correspond au centre d’accroissement, il est formé de fibres entre-croissées à angle droit ; dans sa portion périphérique, il se compose de faisceaux fibreux, entrelacés sous divers angles et décrivant, soit des arcades, soit des courbes de diverses natures. Toutes les fibres du bord pérephérique semblent se perdre dans la couche extérieure de l’écaille ou elles prennent une direction plus ou moins parallèle à celle des crêtes concentriques “(p. 174). This description accords in the main with my own observations. It is possible that the scales of the perch, upon which Baudelot chiefly founded his assertions, may present a greater complexity, more particularly, as will be shown subsequently, there is reason to believe that degenerative changes are in progress in at least some species of cod. So far as I have been able to observe no difference exists in the general arrangement of the fibres in any of the species of cod examined.

Sections through the scale merely show the various fibrous lamellae. I have been unable to detect the presence of any corpuscles; if they exist, they must be very minute. Neither have I met with any of the calcareous masses described by Williamson. There are, however, numerous delicate vertical tubes traversing the substance of the scale and which probably subserve a nutritive function. They are particularly in evidence beneath the centrum of the scale. They may possibly represent Williamson’s Lepidine tubules, or they might be regarded as Volkmann’s canals, though the absence of corpuscles would tend to negative the latter interpretation.

Upon the upper surface of the fibrous layer are placed the translucent calcified scalelets, which, in the recent condition, are closely invested by the delicate outer wall of the scale pocket. If the scale be stained while this membrane is in situ the whole surface becomes uniformly tinted. On removal of the membrane the scalelets are quite unstained, though each individual is mapped out by lines of coloration where the membrane persists in the interstices between the more elevated scalelets. The only reagent which I have as yet tried which seems to act readily upon the scalelets is hæmalum, and this is not so marked as in the younger stages.

Scalelets vary in shape and arrangement in different species of cod, more particularly in the lateral fields. The common cod and the coal fish furnish examples of the two most widely separated types. In the common cod (Fig. 6) the scalelets consist of a basal plate quadrilateral in shape. Those in or near the mesial line of the scale at either end are very nearly square, while in the lateral fields they are rhomboidal. Near to the peripheral margin of each scalelet there is a well-marked transverse ridge, which gives to it the appearance of a square envelope with a straight flap not fastened down, while the extreme peripheral and central margins are slightly bent upward, and are in contact with the adjacent margin of the scalelets immediately in front and behind. Towards the growing edge of the scale they are not always in such close contact, and a thin transverse line of colour between them may be seen in stained specimens. The appearances are more readily understood in sections through the scale, and may be represented diagrammatically (Pl. VI, fig. 7).

The same placoid pattern can still be recognised, but each scalelet is considerably expanded laterally. The transverse ridge near the peripheral margin is due to the centrally directed but minute spine ; the anterior and posterior margins of the basal plate are upturned, the lateral being straight. The spines are most marked in the posterior field. The centrum or “foyer” consists of a flattened plate of calcified material, oval in shape, with an irregularly serrated margin. From its appearance in section and on surface view and from the conformation of its margin, I believe it to be formed by the fusion of a number of basal plates, the spines of which have entirely disappeared. There is no indication of any markings such as would suggest that, previous to fusion, there had been any upturning of the central and peripheral margins of the scalelets. This appears to be easily explained. The centrum is the first part of the scale to be formed, while there is comparatively plenty of room in the scale pocket ; with increase in size the walls of the pocket become stretched and exert an upward “pull” all round the margin of the scale; this leaves an impress on the marginal scales in process of formation, and would account for the up-turning at the peripheral margin, while the pressure thus exerted in a centripetal direction would push the central margin against the peripheral border of the scalelets immediately in front and bring about the same result, but, as shown in the diagram, to a lesser degree. At the growing edge of the scale the young scalelets appear to arise from the lateral wall of the scale pocket, for they have at first a direction perpendicular to the surface of the scale.

The type of scale above described, or but slight modifications of it, is that most generally met with in the Gadidæ. In G. virens, however, there are differences which render the scale easily distinguishable (fig. 5). The scalelets have a tendency to become triangular in the lateral fields, while in the anterior and posterior fields, though quadrilateral, they differ markedly from those of the common cod, as will be seen by reference to the figures. Moreover, there is no transverse ridge visible. Under a high power the scalelets are very clearly imbricated. On transverse section, there is no indication of any spine, the attached border is implanted in a kind of socket in the upper surface of the fibrous layer, and I think there can be but little doubt that they represent merely the basal plates of the placoid scale, all trace of spines having disappeared. The centrum appears to be similar to that previously described. One or two other points remain to be mentioned. In individual scales of all species a partial fusion of adjacent scalelets may be seen. In every instance which has come under my notice, this fusion has been a lateral one. While this condition would appear to be the exception in most species, in G. minutus it might almost be said to be the rule, for I have not observed any scale in which it was absent, usually the fusion being considerable. This would seem to be a point of some importance, for if we imagine a lateral fusion of the small scalelets to take place throughout, we then approximate to the typical clupeoid scale, which is composed of excentric imbricated rings.

On the other hand, if the spines present in the scalelets of the G. callarias be more pronounced and slightly more perpendicular in direction, we have the spines of a ctenoid scale. In the latter the spines are present only in the posterior field, in the same position as they are most evident in the cod. The earlier tendency to disappearance in the anterior field is probably due to the imbrication.

One further point : the radiating lines of scalelets at a certain distance from the circumference become replaced by double lines. This is probably in adaptation to the increased circumference of the scale. Still nearer the margin one of the four again bifurcates (if the scale be placed with the anterior pole towards the observer, it seems to be always the line to the right which thus becomes divided). This is only marked in the posterior field, though elsewhere the same takes place, but is neither so striking nor does it appear to be of such regularity.

Clupeoid Scale.—The scales of the adult herring, sprat, and pilchard have been examined, but, with the exception of some unimportant differences, such as relative size, there is but little to distinguish between them ; consequently, the description of the herring scale here given applies equally to the sprat and pilchard.

The excentric markings seen on the clupeoid scale is limited to the anterior portion, the posterior uncovered part being quite smooth, each excentric line roughly describing a semicircle. Neither in stained nor unstained specimens are any radiating lines visible such as are seen in the scales of the cod. Further, the excentric series of markings presents a considerable degree of uniformity in passing from the centre of the scale to the periphery. In all the scales examined there have been lines of intermission which become more visible on staining. This is due to the deficiency of the calcified upper portion of the scale allowing the underlying fibrous (staining) stratum to be visible.

Though these lines of intermission occur with a certain amount of regularity, I do not think they can be regarded as delimiting the different periods of seasonal growth of the scale, for two reasons. In the first place, they are not always of equal number in scales removed from the same situation in the same fish ; and secondly, they do not always extend around the anterior moiety of the scale, but are often interrupted so that the excentric (calcified) lines of one area pass into those of the area next to it. I believe these lines of intermission have little, if any, morphological value.

All markings are absent from the posterior uncovered portion of the scale, the calcified layer being smooth and of extreme thinness.

Examination of a section through the scale shows that the excentric lines in the anterior part are due to minute ridges arising from a continuous basal calcified layer, each ridge having an inclination towards the centre of the scale.

The fibrous portion of the scale has an arrangement similar to that previously described for the cycloid scale, but in one section the perpendicular tubules are slightly larger.

Sections through the long axis of the scale in a young herring show the same placoid condition as in the young cod, the individual scalelets being much more minute. The projecting spines in the median line of the scales are backwardly directed, as in the Elasmobranchs ; that is, they have the same inclination as that of the excentric markings in the adult scale. These placoid scales are limited to the anterior portion of the scale, the scalelets of the posterior portion being merely elongated flattened plates without any spines.

I cannot say whether they have been present and have disappeared, but I have never seen them, though I have examined fish of various ages, and I am inclined to think that that portion of their philogenetic history has disappeared altogether from their ontogeny.

As development proceeds, the basal portions of all the scalelets fuse to form a continuous plate, except at the lines of intermission. The spines appeal to fuse laterally with their neighbours, thus giving rise to the excentric markings which have been seen to be in reality projecting ridges.

Comparison with the scales of the cod show certain points of marked difference. In the latter individual scalelets remain distinct in certain species—for example, in G. virens and G. callarias; though in G. minutus, as has been said, there is a considerable amount of lateral fusion ; while in all species there is an entire fusion, with a total absence of spines in the region of the centrum. The same influences seem to have been at work in the clupeoid scale, and to have carried their results still further. The condition found in the centrum of the cycloid scale is identical with that seen in the posterior uncovered portion of the clupeoid, while in the anterior part fusion has taken place to a greater extent, the basal portions of the scalelets fusing throughout to form a continuous sheet (except at the lines of intermission), while a lateral fusion involving the spines has given rise to the continuous excentric ridges.

There is one point of marked difference in comparison with the scales of such a form as G. virens. In the latter the appearance of excentric lamination is due to an imbrication of the scalelets, the spines having more or less disappeared, whereas in the clupeoids there is no such imbrication, the excentric markings being produced by the spinous portions of the scalelets.

What have been the determining causes of these differences in the various scales, or for the retention and disappearance of the spines in the anterior and posterior portions respectively of the individual clupeoid scale, I am unable to say. The disappearance on the exposed portions of the latter cannot be due to friction, or it would equally have affected the exposed part of the cycloid scale. I am inclined to think that the retention of the spines in the covered areas may have the effect of retaining the scales in position, since in the clupeoids, where the spines are but very feebly developed, the scales are readily removed, whereas ctenoid scales are the most difficult to separate, and in them the spines are most marked. In this respect the cycloid pattern occupies an intermediate position.

Amongst the species of cod whose scales have been examined, those of G. callarias retain the most primitive condition, while those of G. minutus seem to be converging towards the clupeoid pattern. But both cycloid and clupeoid type agree in this fundamental particular—that they both arise as a number of minute scalelets of the placoid vari ety.

It now remains to discuss briefly the morphological bearings of this interpretation of the scale structure. Hitherto the Teleostean scale has always been regarded as a morphological entity. This will be evident from the following quotation from the important memoir of Klaatsch (12). On page 174 he writes : “Die Teleostierschuppe entspricht der Basalplatte der Placoidschuppe “; and again : “Jede Teliostierschuppe einer Placoidschuppe entspricht.”

From a consideration of the facts set forth in this paper one is led to the opposite conclusion, namely, that the scale of the Teleostei is a compound structure, the morphological unit being the individual scalelet, each of which is the homologue of a single placoid scale of the Selachians. When I first enunciated this view at Belfast I was unaware that a similar interpretation had been placed upon the Ganoid scale by O. Hertwig (6), an interpretation which both Klaatsch and Nickerson refuse to accept. This view is, I believe, quite new in regard to the Teleosteans, and the fact that a similar interpretation has been placed upon the ganoid scale by so high an authority as Hertwig is a fact of the utmost value in its support.

The union of a number of placoid scales upon a single fibrous basis would lead to the formation of a teleostean scale in which the spines might persist only in the anterior covered field, as in many scales of the ctenoid type; a further suppression of the same would lead to the cycloid pattern. The tendency to lateral fusion of the scalelets so frequently seen in the latter would give rise to the scales of the clupeoids, the degree of fusion being seemingly correlated with the relative increase in size of the scale. A still further stage of suppression is seen in the herring, where not only all vestiges of the spines are wanting, but the scalelets themselves have almost disappeared from the uncovered posterior field.

The phylogenetic order of the scales would, therefore, appear to be placoid, ctenoid (?), cycloid, and lastly, clupeoid, and as far as the Gadidæ are concerned ontogeny recapitulates the phylogeny. I hope shortly to have completed researches into the development of the ctenoid type.

Within the limits of the Gadidæ one meets with unmistakable signs of fusion of the scalelets in G. minutus and G. æglefinus, which, in the earlier stages, are quite distinct. A further advance in the same direction gives the clupeoid scale. One reason for regarding the cycloid scale as a ctenoid in which the spines have almost (G. callarias) or entirely (G. virens) disappeared has already been given. There are, however, certain other facts which tend in the same direction—for example, the fact that the scales on the blind side of Arnoglossus are cycloid, while on the ocular side they are ctenoid. Further, the opercular bones in the turbot and many other Pleuronectids are heavily armed in the young, whereas they are smooth in the adult. Again, the ctenoid scales in the dwarf variety of the plaice (Pl. pseudoflesus) maybe taken as another indication in the same direction.

On the other hand, Dunker (4) states that in the male plaice after maturity the cycloid may develop into ctenoid scales, and that the same may happen in flounders only 2-3 cm. in length. I do not wish to pass lightly over any point adverse to the views here expressed, but I may be permitted to refer to the opinion of an independent reviewer, Mr. Stead (17), who throws considerable discredit upon the conclusions, and upon Dunker’s want of care in the establishment of his facts.

Cunningham, in dealing with the ontogeny of the plaice (3), seems to imply that the cycloid precedes the ctenoid condition, but it is not evident upon what grounds he rests the implication. Holt has shown (9) that the scale tubercle of the adult turbot is developed from a simpler and more cycloid form in the young. It is, however, by no means certain that this scale tubercle is the homologue of the ctenoid spine, and not a special development.

Wiedersheim (18) states that the cycloid scale is the more primitive, but without stating the grounds upon which the assertion is made. Hofer (7) is of the same opinion, basing his conclusion both on developmental history and on palaeontology. With respect to the latter evidence this writer takes into consideration only the Physostomes of the Jurassic, whereas the earliest Teleostomes date back certainly to the Lower Devonian. The embryological evidence put forward by Hofer appears to me to be inconclusive, based on supposed histological homologies, and from the point of view taken in this paper is open to the objection that the scale and not the scalelet is taken as the morphological unit.

Klaatsch (loc. cit.) claims a polyphyletic origin for the cycloid scale. It is beyond the limits of the present paper to enter into a general discussion as to all the ganoid, dipnoan, and other scales, but I may say in passing that such a complication appears to me to be unnecessary, and that the various scales, so far as I have knowledge of them, can be reduced to modifications of a single primitive placoid pattern.

In consideration of the eminence of the authorities just quoted, one cannot but speak with the greatest diffidence and regard this question as at least “non proven,” though as yet I have been unable to find any direct evidence against the view of the ctenoid ancestry for the cycloid scale.

If the interpretation, that the individual scalelet and not the whole scale is the ultimate morphological unit be accepted, it follows that many of the deductions of previous authors from the evidence of the scales alone will necessarily have to be modified.

As above stated, Hertwig has made a similar suggestion as to the morphology of the ganoid scale. To this Klaatsch objects owing to the great number and indefinite arrangement of the spines in the Selachians. Neither of these objections appears of much weight, since the number of scales must have been very considerably greater in the larger and more primitive Selachians, and with a reduction in body size thé spines would tend to be more crowded, but not of necessity in the first instance to have undergone any very great numerical reduction. Secondly, the tendency to a fixation of the scales in situ upon a fibrous basis would impede the movements of the body. In order that this might be prevented as much as possible without impairing their other exoskeletal functions the scales have assumed a definite shape which has consequently interfered with the regular linear arrangement of the scalelets. If the disposition of the latter be examined, not from the point of view of excentrically arranged lines, it will be seen that those on the lateral fields still preserve a very fairly definite antero-posterior series, while in the anterior and posterior areas this plan is but little disturbed.

I pass to a consideration of the points involved in the account of the early stages of the development. It has been seen that the teleostean scale arises and remains throughout its existence as a dermal structure. This at first sight appears to present a difficulty in homologising the scalelet with a placoid scale or with the scales of Lepido steus. In the development of the two last-mentioned the greater part of the spine is dermal; it is only the enamel tip which is epidermal in origin. This in Lepidosteus is only subsequently added when the dermal spine has reached the lower boundary of the epidermis. It is also at this time and from the same source that the scale receives its layer of ganoin. In the Teleosteans the spines are so much reduced in size that they do not, during the period of their growth and development, attain that superficial position. Consequently, neither enamel nor ganoin is formed upon them. This does not offer to my mind any serious obstacle in establishing their homology ; it is simply that they have not reached to the same level, and as a consequence the epidermis has had no opportunity of sharing in their formation. This view finds support in the opinion to which Nickerson inclines, that the enamel is not secreted in Lepidosteus until at least a part of the spine has been formed. The development of the teleostean scale shows that it is entirely a dermal structure ; and regarding the line of demarcation between the dermis and epidermis as Huxley’s protomorphic layer, it follows that the entire scale must be regarded as of enderonic origin, and that the scale is covered throughout its existence by the superficial layer of the scale-pocket, which is also enderonic. The successively formed layers of the fibrous substratum are added to the deeper surface, the direction of growth being from without inwards, precisely in the direction which Huxley regarded as being indicative of the growth of enderonic tissues. The calcified layer, the first part of the scale to be formed, does not increase in thickness after being once formed, there being merely an increase in size and alteration in pattern of the individual scalelet. As to the presence of the true ganoin in the teleostean scale I cannot but agree with Nickerson in thinking that Klaatsch is in error, and that in the scales of the Gadidæ no such material is present.

Huxley (11), following Williamson, regards fish-scales as “essentially tegumentary teeth,” and of this there can be but little doubt, but the further conclusion of these writers, that the scales are formed by a calcareous deposit in a deep layer of the ecderon, I think erroneous. Presumably what they interpret as the deep layer of the ecderon I have regarded as the superficial nuclear layer of the enderon. It is difficult to establish a difference now that the importance of the basement membrane is no longer upheld ; but from the marked difference in the histological characters, and judged by Huxley’s own criterion of exogenous or endogenous mode of growth, I think the entirely enderonic origin of the whole teleostean scale must be admitted. It would seem that a difficulty had presented itself to Huxley, since he regards the nature of the deeper layers of the scale as uncertain ; for he writes (11, p. 386) that it is “an open question whether the deep layers of all scales are produced by a continuation of the process” (i. e. of a calcification of the ecderon), “or whether in some cases a deep truly enderonic structure may be added to this superficial ecderonic constituent to constitute the perfect scale. A process of the latter kind would, at any rate, find its parallel in the eventful union of the teeth of many fishes with their jaws, and in that of the plates of the Chelonia with the vertebral elements.”

The view which I have advanced seems in no way to invalidate the conclusion as to the homology between teeth and scales. In the former there are both dermal and epidermal constituents as represented by the dentine and enamel cap ; the same is seen in the scales of Selachians and Lepidosteus. In the latter the dermal portion grows upwards to reach the epidermis and there receives its enamel addition. In the teeth of the higher vertebrates the process has become more specialised, and the enamel germ grows inwards to meet the uprising dermal papilla. In the Teleosteans conditions are similar to those obtaining in the Selachians, but owing to the smallness of the spines and their failure to reach the level of the epidermis, no enamel tip is consequently formed.

There remains yet one further point to which I would refer, namely, that raised by the recent work of Hoffbauer and Stuart Thomson (17). These writers believe that there is an annual growth of the scale in rings, which therefore furnish an index of the age of the fish. Thomson likens this annual growth to the annual rings in the stem of a Dicotyledon. Neither the idea nor the simile are by any means new; both were, as has been shown, originally suggested by Leeuwenhoek in 1685.

The same idea has also from time to time been referred to by other writers. During the course of my observations I have been led to note carefully the appearance of the scales endeavouring to estimate the ages of the fish with whose scales I was dealing, and the length of which only was known to me. In young fish the rings are quite easily to be recognised, which Thomson regards as the rings of growth of the first summer and first winter. Such a scale he figures in his paper.

It may be noticed in passing that the point of the first bifurcation of the lines of scalelets at the one pole marks what Thomson regards as the period of first summer growth, and that the further bifurcation previously mentioned as being present in some scales corresponds with the limit of first winter growth. This is about the farthest point to which I am able to follow. With the increasing size of the scale, I confess I am quite unable to detect that regular series of alternating broad and narrow bands such as Thomson describes. From the scale of a G-. Callarias 24-feet in length, one of which is represented in fig. 6, I cannot make any estimate as to the age. If this be so in the cod, I find still greater difficulty in the case of the herring and other Clupeoids, concentric markings upon which are not merely irregular, but to many the term “labyrinthine” might almost be applied. As a method of practical value, it seems to be of but little value, at any rate in my hands.

Apart from this, however, there seems to be many theoretical objections. Klaatsch and others have pointed out that the scales do not appear simultaneously all over the body. They commence to appear just behind the pectoral fins in the neighbourhood of the lateral line. From this point the appearance of the scales radiates, those in the tail region being last to appear. It appears, therefore, that in instituting comparisons of age, the scales should always be taken from the same region, and that region stated. Of course, if the extension of scales over the whole surface of the body is rapid, then the limit of error will be so small as to be practically negligible. Until we have some data as to the rate of extension in different fishes, it would appear that scales taken from different regions of the same fish, might yield different results; this according to Baudelot (1) is precisely the case. He states that many fish present scales of different varieties : “L’existence simultanée d’écailles cténoïdes et d’écailles cycloïdes surdes points du corps différents, a été constatée par moi dans les Trigla Lineata, Sargus Rondeletii; Perca Fl uviatilis, sur divers pleuronectes (Pl. sola, PI. flesus, etc.) chez plusiers scorpions. Le même fait a été observé chez le Pelamys sarda par Péters, et sur l’apron par L. Vaillant. Le Thon possède aussi deux sortes d’écailles distinctes. Les écailles de la carène ventrale de l’Alose et du Hareng, les écailles de la ligne latéral d’un grand nombre de poissons (Trigle),présentent aussi une conformation particulière “(p. 432). With regard to the number of concentric rings, the same author writes : “En comptant les crêtes concentriques dans chacun des champs de l’écaille, on constate que le nombre de ces reliefs n’est pas le même pour chacun d’eux.” Again : “Le nombre des crêtes est susceptible d’offrir les plus grandes variations dans les écailles d’un même poisson,” citing as examples the perch and the pike (p. 436). In speaking of the increase in size of the scale by the addition of peripheral rings, he says : “Ajoutons enfin que l’accroissement n’a pas lieu au même degre pour tousles écailles d’un même poisson.” A large number of tables are furnished by this writer giving the number of concentric lines in the different fields; many vary as much as six or eight, and in some instances ten.

From these considerations, number would seem to be very unreliable as an index of age, and from personal observations I am of opinion that their relationship to one another is equally untenable.1 A further point must not be lost sight of in this connection; it is the possibility of the shedding and replacement of the scales. I have been assured by practical fishermen of considerable experience on the east coast of Scotland that it is well known to them that such is the case, more particularly in fishes after spawning, analogous to the moulting of birds after the breeding season. 1 am told that it is very evident in the herring. As yet I am not in a position to speak definitely, but the examination of sections of the skin in a fully grown herring certainly suggests such a possibility. That a replacement should occur seems probable, for in such fish as the herring, in which the scales are so easily removed, it would be but a very natural provision.

It might be urged that even should it be proved to occur, it would not entirely overthrow the view of age-index, since the new scales might begin with the same number of rings as that at which the older ones ceased, after the manner of the antlers in deer. Such a view appears to me improbable, but it can only be settled by accurate investigation.

The conclusions arrived at in this paper further establish the view of a true selachian ancestry for the Teleosteans, as well as affording indications of the affinities existing between the Teleostei themselves. On the other hand, I think it points in the direction of breaking down the separation of the Anacanthini as defined by Johannes Müller, and of which other evidence is not entirely wanting.

1.
Baudelot
,
IS
. —“
Recherches sur la Structure et le Développement des Écailles des Poissons Osseux
.”
‘Arch, de Zool. Expér. et Générale,’
t.
ii
,
1873
, pp.
87
-
244
et 427-480.
2.
Borbllus
,
Petrus
. —
‘Observationum Microscopicarum Centuria,’ Hagoe comáis
,
1656
.
3.
Cunningham
,
J. T.
—“
On the Peculiarities of Plaice from different Pishing Grounds
.”
‘Journ. Marine Biolog. Assoc., N. S
., vol.
iv
, pp.
315
359
.
4.
Dunkbb
,
G.
—“
Variation und Verwandschaft von Pl. flesus (L.) und Pl. Platcssa (L.) untersucht mittelst der Heincke scheu Methode
.”
‘Wiss. Micros, untersuch. Neraus-gegeben von d. Komm. zur Wiss. untersucb. d. deutschen Meere in Kiel,’ Heft 2
,
1896
. (Vide Stead.)
5.
Ehrenberg
,
C. G.
—“
Notliwendigheit einer feineren mecbanischen zerlegung des Gehinis und der Nerven vor der Cbemiscben
.”
‘Poggendorff’s Ann. der Physik u. Cliemie,’ Leipzig
,
1833
, Bd.
xxviii
, pp.
449
473
.
6.
Hertwig
,
O.
—“
Ueber das Hautskelet der Fishe
.”
‘Morpb. Jabr.,’
Bd.
ii
,
1876
, Bd.v, 1879, Bd. vii, 1881.
7.
Hoyer
,
Bruno
. —“
Ueber den Bau uud die Entwicklung der Cycloid und Ctenoidschuppen
.”
‘Sitzungsber. d. Gesellscb. f. Morpb. und Physiol. München,’
Bd.
vi
, pp.
103
118
.
8.
Holt
,
Ebnest W. L.
North Sea Investigations, VII. “On a Dwarf Variety of the Plaice (Pleuronectesplatessa, Linn.), with some Remarks on the Occasional Cilialion of the Scales in that Species.’ “
Journ. Marine Biolog. Assoc.,’ N. S
., vol.
iii
, pp.
194
200
.
9.
Holt
,
Ernest W. L.
—“
Note on some supposed Hybrids between the Turbot and the Brill
.”
‘Journ. Marine Biolog. Assoc.,’ N. 8
., vol.
iii
, pp.
292
299
.
10.
Hooke
,
R.
“Micrographia, or some Physiological Descriptions of Minute Bodies made by Magnifying Glasses, with Observations and Inquiries thereon.”
London
,
1665
.
11.
Huxley
,
T. H.
—“
Tegumentary Organs
.”
Todd’s ‘Cyclopedia Anat. and Phys.,’
1855-56
.
12.
Klaatsch
,
H.
—“
Zur Morphologie der Fischschuppen und zur Geschichte der Hartsubstanz gewebe
.”
‘Morpb. Jahrb.,’
Bd.
xvi
,
1890
, pp.
97
202
und 208-258..
13.
Kuntzmann
. —
‘Verhnnd. d. Gessellsch. naturforsch. d. Ereunde in Berlin,’
1824
, p.
269
, and 1829, p. 369.
14.
Van Leeuwenhoek
,
Antony
. —“
Select Works, containing his Microscopical Discoveries in many of the Works of Nature
.”
Translated from the Dutch and Latin editions, published by the author. By Samuel Hoole. London
,
1798
.
15.
Mande
,
L.
—“
Recherches sur la Structure intime des Écailles des Poissons
.”
‘Ann. des Sci. Naturelles,’ second series
, t.
ii
,
1839
, pp.
337
370
.
16.
Nickerson
,
W. S.
“The Development of the Scales of Lepidosteus.’
‘Bull. Mus. Comp. Zool.’ (Harvard)
, vol.
xxiv
,
1893
, pp.
115
138
.
17.
Stead
,
F. B.
—“
Variations and Relationships of the Flounder and the Plaice
.”
‘Journ. Marine Biolog. Assoc.,’ N. S
., vol.
iv
, pp.
293
-
299
. (Vide Dunker.)
18.
Wiedeksheim
. —
‘Comparative Anatomy of Vertebrates.’
19.
Thomson
,
J. Stuakt
. —“
The Periodic Growth of Scales in Gadidæ and Pleuronectidæ as an Index of Age
.”
‘Journ. Marine Biolog. Assoc.,’
vol.
vi
, No.
3
,
1902
, pp.
373
375
.
20.
Williamson
,
W. C.
—“
On the Microscopic Structure of the Scales and Dermal Teeth in some Ganoid and Placoid Tish
.”
‘Phil. Trans. Loud.,’
1849
,
pt. 2
, pp.
435
476
.
21.
Williamson
,
W. C.
—“
Investigations into the Structure and Development of the Scales and Bones of Pishes
.”
‘Phil. Trans. Lond.,’
1851
pp.
643
702
.

Illustrating Mr. H. W. Marett Tims’s paper on “The Development, Structure, and Morphology of the Scales in some Teleostean Fish.”

Fig. 1.—Surface view of epidermis of G. virens, showing the chromatophores and the openings of the glands.

Fig. 2.—Section through the same.

Fig. 3.—Section through the dermis of G. virens (4’6 cm.) after treatment with acid alcohol.

Fig. 4.—Section through a scale of G. virens (5 6 cm.). Prom a specimen to which no acid had been applied.

Fig. 5.—View of the external surface of a scale of G. virens (9 cm.).

Fig. 6.—Ditto, G. callarias ( feet).

Fig. 7.— Section through scale of G. callarias.

1

Since writing the above Mr. Stuart Thomson’s more detailed paper has been published (‘Journ. Marine Biolog. Assoc.,’ vol. vii, 1904). I see no reason, however, for modifying the opinion which I have already expressed.— H.W.M.T.