A Contribution to the Life-History of the Brachiopod, Terebratella inconspicua Sowerby.* Determined by Dr. R. S. Allan, to whom the writer is indebted.

Transactions and Proceedings of the Royal Society of New Zealand, Volume 74, 1944-45, Page 1

A Contribution to the Life-History of the Brachiopod, Terebratella inconspicua Sowerby.* Determined by Dr. R. S. Allan, to whom the writer is indebted. By E. Percival, Canterbury University College, Christchurch, New Zealand. [Read before the Canterbury Branch, April 3, 1948; received by the Editor, November 25, 1943; issued separately, June, 1944.] The knowledge of brachiopod development is so incomplete that any addition is still to be regarded as valuable, particularly when it covers a long range of embryological changes and metamorphosis. The present study provides a fairly continuous view of change from early segmentation to young adult life, and fills a gap which was inevitably left after the excellent work of Morse (6 and 7) and Conklin (1). The accounts of these two workers unfortunately do not provide a complete story of the growth to young adult life of Terebratulina septentrionalis. Conklin's report proceeds until the stage of the well-defined larva, but, by analogy with the present findings, stops while yet the mesoderm must undergo further elaboration. The fixed material with which he worked did not show the nature and extent of ciliation, but the details of development were generally very clearly elucidated, especially the origin of the mantle, which was transverse, so making the peduncle posterior in position. There is an uncertain amount of development between the last phase of Conklin's material and the first phase of Morse's. Further, Morse studied his material alive and only externally, so that much of what happened internally was not seen by him. MacBride (5, p. 410) says, with reference to the fate of the blasto-pore (taken from Conklin's account); “Finally, this is closed, but a shallow pit is left, and in this same spot, at a later date, the invagination to form the stomodaeum arises. In this way a valuable landmark is created for the correlation of the region of the larva with reference to the adult organs.” In the following account, it will be evident that it is not possible to foretell the relation between the site of the stomodaeum and that of the closed blastopore: in fact, the orientation of the Testicardines, as far as concerns Terebratulina and Terebratella is quite uncertain when considered in terms of previous work. It is possible to relate the arrangements of the young adult with those of larval and pre-larval life only by the following through of metamorphosis. This has been done in the present work, and results in a reconsideration of previous views on orientation: in short, the current terms of dorsal and ventral, as applied to the adult must be reversed, what was called dorsal is primarily ventral as it is related to the blastopore, and vice versa. Conklin's account of the origin of the mesoderm of Terebratulina as an anterior outgrowth from the archenteron does not apply to Terebratella, where it is posterior. The mode of origin of the mouth and of the lophophore, as described by Morse, does not agree with present findings, or is so inadequately described as to make desirable

further information. It must be kept in mind that Morse and Kowalewski (3) made their studies on whole specimens, alive, and that, having regard to the minute size of the material, their drawings and conclusions must always be considered of high rank. Notes on the Ecology of Terebratella inconspicua. Collections for the present purpose were made between tide-marks in Lyttelton Harbour, New Zealand, on the breastwork of a disused jetty at Governor's Bay, and on the inner face of the retaining wall of the Gladstone Wharf of the Inner Harbour, at Lyttelton. The population at the former site is small, but at the latter is dense and extensive. In both places the brachiopods occur up to about half tide mark, attached to stones and other solid objects, but hidden in such a way as not to be disturbed by currents. It seems that the water which passes over them is always gentle in its action, there being little more movement than that caused by the rise and fall of tide without lateral swirl. At the Inner Harbour, the animals are found on the downward faces of stones which are either heavy, or partly embedded, or deeply situated. The surfaces on which attachment is made are covered with a thin film of organic detritus, and there are indications that a fine suspension bathes the brachiopods and other associated filter feeders during the presence of sea water. This fine deposit indicates quiet conditions in the vicinity and gives support to the conclusion that the attachment of larvae and metamorphosis and early post-metamorphie life could proceed only in quiet circumstances. The very small larvae have been observed in the laboratory to swim chiefly near the substratum but also to rise so much as two inches towards the water surface. In late free life, after emergence from the parental mantle-cavity and before attachment, there was seen a temporary tendency to congregate towards the light, but later, after 20 to 30 hours of freedom, there was a movement towards the substratum over which the animals appeared to run, as though with the aid of legs. This movement was jerky and erratic, with the flat or slightly concave surface ventrally placed, the arched surface being dorsal. Rotation about the long axis ceased in these circumstances. All the early life, prior to the short swimming period, is spent in the mantle cavity of the parent, although on one occasion unfertilised eggs were extruded on the floor of the dish containing the adult. The running about over the substratum had nothing to do with feeding, since there is neither mouth nor gut at this stage, nor will there be for some time. It seems that the whole of embryonic, larval and a significant part of post-metamorphic life is maintained by means of the food reserve stored in the egg. Attachment takes place in nature, in the Lyttelton collecting ground, in rather well-defined conditions, since this species of brachiopod has been rarely found mixed with any other animals of similar size. It occupies the most sheltered portion of a stone and is commonly surrounded by an abundant filibranch mollusc, Hochstetteria meleagrina Bernard* Determined by Mr. A. W. B. Powell, to whom the writer is indebted., which may slightly overlap the margin

of the brachiopod area. The filibranch actively creeps over the substratum in its earlier benthic life, but lives decidedly outside the region of Terebratella. Associated with the brachiopod is commonly found, in autumn, the juvenile form of a species of Sycon as are also several small species of errant polychaetes and of Polyzoa. The breeding season, in Lyttelton Harbour, has been determined as generally from early April to late May, with its greatest intensity during May. Early stages were found in two females taken in the last week of March, but, commonly, material collected at the end of March had not yet spawned. Specimens collected on June 6 were spent or immature. During three years, collections made at different dates in this part of the year established the breeding period. In early June no newly attached young were found on the collecting ground, but the youngest already showed some appreciable growth, addition to the shell margin being discernible. Attachment at first is very weak, but later becomes firmer. How this strengthening comes about is not obvious, but, as those well know who have collected live brachiopods of this kind, there is practical fusion between peduncle and substratum as the animal grows. This, of course, is similar to what takes place with many attached animals, such as barnacles and encrusting Polyzoa. Text-Figure I. Terebratella inconspicua. Distribution of length and breadth in 0.5 mm. groups. Collected 30/3/41, under Gladstone Pier, Lyttelton, Just before commencement of breeding season. Population of 4 adjacent stones covering about 2.3 square decimetres, 711 specimens in whole sample.

Very many newly attached larvae are to be found in slight recesses or behind slight ledges on the substratum. In the former case, death may ensue owing to lack of room for growth. It seems that very many young ones become thinned out in the first year, in part through overcrowding, in part through starvation, and in part for no apparent reason. Dead shells of juveniles less than one year old, or less than half a year old especially, were very common in wholesale collections which were picked from stones under the microscope. An attempt was made to determine the structure of a population living in good circumstances—i.e., one showing a wide range of sizes and a high proportion of juveniles. 711 specimens were picked from four adjacent stones covering an area of about 2.3 square decimetres. They were measured along the greatest length and breadth. The measurements, collected in 0.5 mm. groups, are plotted in Text Fig. 1. The collection was made on March 30, 1941, just before the breeding season commenced, so that the youngest specimens would be nine months old at the most and the oldest of the previous year's brood would be hardly 12 months old. The gonads of the adults were nearly mature, but no female showed free eggs. It will be noticed that the first of the peaks of the graph ranges between 0.5 mm. and 3.5 mm. with the maximum length round 2 mm. and the maximum breadth round 1.5 mm. A second broad crest, less definite than the first, lies between 3.5 mm. and 7 mm. long, a third, still less clear, between 7 mm. and 10.5 mm. long, leaving a fourth portion tailing off to a length of 13.5 mm. All this material was brownish, covered more or less densely with an encrustation, not even the smallest specimen showing the sharp, clear, glistening whiteness of new shells, but a sample collected on April 29, 1941, of 150 specimens, contained 18 clear, white small ones, obviously recently produced, of length ranging from 2.3 mm. to 3.8 mm., and three brown, encrusted specimens of lengths, 5.0 mm., 4.4 mm., and 3.3 mm. respectively. These three were, without doubt, of the 1940 year class, and stood out sharply from the other small specimens. The measurements indicate that growth in Lyttelton Harbour is slow, and, if the plottings do in fact show four year classes, the life of the population is short. Nearly two-thirds of the population consists of the first year class, nearly one-quarter of the second year class, nearly one-seventh of the third year class, and nearly one-thirty-fourth of the fourth year class. What the conditions are elsewhere has not been determined. However, while the maximum length in Lyttelton has not been found greater than 13.5 mm., specimens from Foveaux Strait are decidedly longer. Six specimens in a sample of ten ranged in length from 18 mm. to 22 mm., with a similar range of breadth. An attempt was made to examine growth in length by the measurement of growth-lines, but unsatisfactorily. There are many subsidiary lines, well-defined, commonly coincident on both valves and leading to confusion. It might be expected that growth checks would be recorded fairly concurrently by members of a closely settled population, but if the lines do include marks of growth checks, they occur very irregularly. However, it can be said that the results of growth-line measurement agree loosely with the subdivision of the

particular population into four year-classes ranging in length as follows:— First year class   up to 3.5 mm. Second year class 3.5 mm. to 7.0 mm. Third year class 7.5 mm. to 10.5 mm. Fourth year class 10.5 mm. to 13.5 mm. The Embryology of Terebratella. The genital products of the male are probably passed ultimately into the mantle cavity of the female where fertilisation normally takes place after the eggs have collected there from the ovaries. The word “normally” is used because, as has been mentioned, in only one case were eggs extruded from the mantle cavity on to the bottom of a dish, and because material used for study was obtained from the mantle cavity. The close association of individuals on stones would make easily possible the passage of sperms from male to female, and, the sex ratio of several samples totalling about 200 being about 1:1, there is a strong likelihood that the sexes are quite well mixed. The eggs extruded on to the bottom of the dish were not fertilised and resembled unfertilised eggs taken from the mantle cavity. Both eggs and early developmental stages were found applied to the inner surface of each mantle fold, not particularly associated with the lophophore. Although, later, ciliated stages swam freely in the water in a dish, there was no clear indication that free-swimming took place throughout the mantle cavity, except, perhaps, towards the time when the larvae should leave the parent. Later stages were more loosely packed than earlier and more easily passed into the surrounding water when the valves were prised apart. The unfertilised egg, from the mantle cavity, is rather irregular in outline, often polygonal, sometimes ovoid, usually flattened, and generally having the appearance of being recently closely packed. The unreduced nucleus is large, about one-half the diameter of the cell, slightly eccentrically placed, circular in outline, and visible by its relative transparency, the egg contents being otherwise dense and opaque. In this early stage, there is on the surface a coat of closely-fitting flattened cells. After the assumption of the spherical form, no doubt through the absorption of water, the egg is seen to have from 20–30 cells round any meridian (Plate 1, Fig. 1). Turgidity and sphericity of the egg are quickly followed by the disintegration and loss of the investing layer, and the egg wall becomes exposed to the medium. The mixing of teased ripe testes with unfertilised eggs was quickly followed by the disappearance of the translucence marking the position of the large nucleus. This suggests that reduction takes place after impregnation, but polar bodies have not been seen, either in these circumstances or on fertilised eggs taken from the parent. In five minutes after the mentioned mixing, there appeared a narrow clear zone immediately inside the follicular layer, very similar to a fertilisation membrane. In 1 ½ to 1 ¾ hours after the appearance of the “fertilisation membrane,” the follicular layer began to break up and the egg to present a smooth, uniform shape. The diameter of the smooth, spherical egg, with follicle, was about 0.18 mm.

Conklin (op. cit.) described the cleavage in his Terebratulina material, but gave no information about the method of fertilisation or about the conditions in which the eggs developed. He mentioned that segmentaion was variable, from regular to very irregular. In the present study, early segmentation, after mixture of teased ripe testes and unfertilised eggs, was watched and was found to vary very much. Regular cleavage produced two, then four, equal blastomeres by the first two vertical clefts at right angles to each other. The third horizontal cleft gave eight equal blastomeres (Plate 1, Fig. 4), as determined by measurement, which is remarkable, since the unreduced nucleus was nearer one part of the egg surface than to any other. A differentiation into micromeres and macromeres would normally be expected, but this was not seen at any stage where individual cells could be recognised. The fourth cleavage was vertical and radial, producing a blastula of sixteen cells. An eighteen-hour blastula of 24 cells consisted of upper and lower quartettes separated by two octettes, all showing radial arrangement. Irregular segmentation varied from slight size differences of blastomeres in early cleavage, to the budding of micromeres from a single, very large macromere. Regular segmentation proceeded to the blastula stage, when the material was killed and fixed, but the irregular segmentation was followed sooner or later by death. It is reasonable to conclude that regular, radial segmentation is normal in this animal. The heavy death-rate in irregular specimens was not found in nature. Survival in the mantle cavity was found at all times to be high, and observed early segmentation stages were regular, although simultaneous division of blastomeres often did not take place. The irregularity observed in the laboratory was regarded as due to the conditions of the experiment. Conklin (op. cit., p. 46) thought that irregular segmentation might give rise to normal larvae by rapid division of macromeres; nothing comes out of the present study which would lend support to that suggestion. As Conklin well showed in Terebratulilna, in Terebratella gastrulation is by invagination. Late blastulae are immobile and show no recognisable difference between animal and vegetative poles (in the absence of polar bodies). The gastrula is first nearly spherical, then hemispherical with a flat blastoporal face. At first the blastopore is circular with diameter about one-third that of the body, but later undergoes change of shape, as will be noted later. The gastrula early becomes ciliated, at first with no locomotion, then, later, with slow movement, usually blastoporal face downwards, but movement takes place when the abporal surface is down, indicating a general ciliation of the body surface. After about 18 hours from gastrulation, the blastopore becomes narrow and slit-like, with a shallow groove running back from the pore; the body takes on a rather square outline in plan with a blunt, wedge shape seen from the side, the blunter, thicker end being anterior. The groove from the blastopore passes back

towards the thinner end in the middle line. During the change in shape is an alteration in the distribution of cilia (Plate 1, Figs. 5, 6, 7). Those over the blunt half remain, and when the wedge shape is clear, the narrower half loses the cilia, earlier from the dorsal, ventral and lateral surfaces, and later from the posterior end. Thus, a body is left with the blunt half only bearing cilia; this blunt half will form the anterior lobe, while the unciliated part will form the peduncle. Soon the right and left sides of the body become slightly concave, with a continued tendency for the posterior part to grow narrower than the anterior part. The blastopore becomes narrower posteriorly, and seems to be carried inward by a slight secondary invagination of the surrounding cells: this recess remains until after closure of the pore. The blastopore, in its recess, shortens and comes to lie in the middle of the ventral side of the anterior lobe, although the recess extends medially to the posterior end of the ventral surface (Plate 1, Fig. 9). It is not possible to decide whether or not the blastopore closes by the fusion of its sides, from behind forwards: there is nothing in sections to support a suggestion that closure is by concrescence, at least, during the earlier part. Final closure is brought about by the complete contraction of the margin, leaving a solid plug of epiblast cells leading to the hypoblast. During the narrowing and closure of the blastopore, an apical tuft of long cilia appears in the middle of the anterior end (Plate 1, Fig. 7). The lack of obvious difference between any two parts of the blastula made impossible the determination of the animal pole, in the absence of polar bodies. Conklin was unable to satisfy himself about the position of this part in the differentiating embryo. He pointed out the fact that the blastoporal surface is ventral in Tere-bratulina, as it is in Terebratella, and we are faced with the difficulty of deciding whether or not the apical tuft of cilia is primarily apical in being produced by cells derived from the animal pole of the egg, as is so very often the case, or is secondarily apical only because it is at the anterior end of the long axis in locomotion, not coinciding with the animal pole which may be somewhere dorsally. Superficially, the apical tuft is derived from a part of the equatorial region of the blastula, if gastrulation proceeds in a manner similar to that—e.g., of Amphioxus, namely, through the invagination of the vegetative region. Nothing comes out of the present work satisfactorily to show the relation between the axes of the fertilised egg, the blastula and the larva. The speed of locomotion gradually increases, the body gliding over the substratum (the bottom of a dish), the blastoporal face being usually downward. The bodily shape markedly changes so that a large downward bulge forms on the anterior end (Plate 2, Fig. 10). This is associated with the formation of a long, broad slope downwards and forwards over the anterior, dorsal half: the apical tuft continues to occupy a central position. Thus, from the side, the body has two fairly well-defined regions, the thickened, ciliated anterior end, with its downward bulge and its antero-dorsal slope, and the unciliated, more slender, posterior part which gives rise to the peduncle.

Posteriorly, bordering the ciliated region, a dorsal constriction marks the front margin of the mantle rudiment which arises as a low dorsal, transverse ridge and extends down each side finally on to the ventral surface (Plate 2, Figs. 11, 12, 13), in the manner described by Conklin in Terebratulina. The two ventral ends come to be one on each side of the hind part of the blastoporal recess, but soon this goes and the mantle fold completely surrounds the body. The fold is slightly higher dorsally and laterally than ventrally, while the blastoporal vestige is present on the ventral face of the anterior lobe. Locomotion is now in the form of a slow, sinistral revolution about the long axis and the body can be raised slightly from the substratum. The mantle rudiment grows backward as a sheath enclosing a continually narrowing peduncular rudiment, until, in the end, only the tip of the peduncle remains visible (Plates 2 and 3; Figs. 15, 16, 17). During the elaboration of the mantle and the stalk, the latter changes from being a thick, squat appendage to a slender, tapering structure. Earlier, the cross section is a rather large, rough parallelogram; later, it is a small circle. The anterior lobe changes in form from being a somewhat broad, transverse body, sloping downwards and forwards, to a more elongated, conical structure still having the tip (with its apical tuft) tending towards the mid-ventral line. In this way the dorsal arch of the body remains, and the orientation can be determined well past the time of the disappearance of the blastopore: indeed, the dorsally arched outline and the ventrally slightly concave, or flat, outline, persist until metamorphosis and serve as valuable data of reference in the work of orientating the adult. When the conical form of the anterior lobe is finally established, the apical tuft of cilia is already lost (Plate 3, Fig. 16). This takes place some time before the larvae leave the parental mantle-cavity, and the later life, up to the normal time of exit from the brood chamber, is spent in the completion of the mantle fold rudiment and in the elaboration of the mesoblast, which, however, can be seen satisfactorily only by means of sections. When larvae (Plate 3, Fig. 17) left the parent in a dish in the laboratory, they swam high, near the water surface, about 1 ¾ inch up, and towards the light. Locomotion at this stage is ciliary and vigorous, with sinistral rotation about the long axis. The anterior lobe is still the only ciliated part, the mantle and peduncle never having been ciliated since they began to differentiate. Although the late larva appears to have cilia of two lengths, a well-defined broad band of long ones round the base of the anterior lobe and apparently much shorter cilia covering the rest of the lobe, all the cilia were of the same length when larvae were freshly killed. The difference in appearance in life may be due to the form of the ciliary beat. The normally free swimming larva is about 0.2 mm. in length and about 0.14 mm. broad at the widest part, through the base of the mantle rudiment. It has about 60 small pigment spots, probably eyespots, round the posterior margin of the anterior lobe, and is somewhat positively phototactie. Four tufts of setae have been for some time on the mantle edge, roughly equidistant from each other, and projecting backward. Conklin (op. cit., Plate 9, Figs. 56 and 57)

figures what he calls setae sacs, indentations on the inner face of the larval mantle, which, if they indeed carried setae, would bring them well on the outer face after reversal. In Terebratella, the pits in which the setae grow, if pits there be, are so small as not to have been seen in sections. The so-called setae sacs of Conklin's figures resemble closely sections of the fold formed when the mantle has already begun to reverse. Metamorphosis: External Changes. The shortest observed period of free-swimming life of the mature larva was 24 hours, when attachment occurred. One batch all became fixed by about 30 hours, but others remained swimming for several days, most of them ultimately disappearing. How long active, free locomotion takes place in nature is not known. As the most successful spatting took place within a short period (24–30 hours) and the least successful in a long one, it is probable that, in nature, the free swimming life is rather short, perhaps about 30 hours. The larvae become loosely gummed to the substratum by means of the tip of the peduncle (Plate 3, Fig. 18). An accumulation of fine detritus stuck to mucus forms round the base—the larger the mass the longer the attachment. Attached larvae were observed for many hours in the laboratory to sit with the mantle slightly dilating and contracting, while the anterior lobe quivered almost continuously, undergoing spasmodic violent contractions and tiltings, all the time tending to become broad and squat. There were generally striking evidences of muscularity and the cilia actively vibrated, particularly those forming the well-marked basal band. The actual process of mantle reversal has not been observed. It must be very quick because numerous specimens kept in the laboratory and collected in the field were found in various stages of rearrangement after reversal, and only one which had a form suggesting the beginning, but not one caught in the act, with the mantle partly reversed (Plates 3 and 4; Figs. 18, 19, 20). Sections and whole mounts indicate that the mantle begins at the base to reverse, so that a circular fold is formed round the base of the anterior lobe, which slides upwards, the lower edge being the last portion to pass up. This kind of reversion is similar to that called pleurecbolic eversion and figured in Pelseneer (8, Fig. 72). The anterior lobe is not immediately totally enclosed at reversal. Many specimens show the lobe unsheathed to varying extents, and as the mantle edge moves upwards it changes from circular outline to the elongated border of an ultimate slit (Plate 4, Figs. 23, 24, 25). The body—i.e., anterior lobe, contained by the mantle, becomes flattened and broadened, the slit-like mantle-opening gradually swallowing the anterior lobe on which, in life, the cilia may still be observed to move. The opening is about one-third the circumference, and bears the larval setae on its edge rather scattered, no longer in tufts (Plate 4, Fig. 22). During the later period of enclosure, when dorsoventral flattening has taken place, the outer surface of the mantle becomes

glistening white and smooth. The shape is no longer plastic, and there is clear evidence of the formation of a hard shell. Acid fixatives lead to the formation of a space between the mantle ectoderm and a thin outer cuticle which presumably lies outside the shell as a periostracum, and the eonclusion is that the shell lay in the space. It seems, therefore, that the mantle produces first the thin cuticular shell followed by the calcareous. During this early period the animal sits upright on the stalk, but soon the body tilts on the joint with the stalk (Plate 4, Figs 24, 25). Even before this there is evidence that the two flat sides are unequal in length, one valve projecting backwards more than the other, and this dissimilarity increasing, brings about a more pronounced tilt towards the other side. At rest, the animals stand as erectly as possible, but quickly bend downwards, closing the valves when disturbed by touch or shock. The area of junction of mantle and stalk is at first circular in outline but becomes transversely elongated through the dorsoventral flattening mentioned above, and tapers narrowly towards the two ends. It later becomes modified in form as the base of the larger valve projects more and more backwards, causing the peduncle to project sideways and backwards instead of backwards. By the time of appearance of the stomodaeal rudiment, the peduncle has taken up almost its final relation with the body. The shell grows obviously by addition to length and width at the mantle edge. In about three weeks after attachment an increase of about half the original length had occurred in laboratory grown specimens. As has been mentioned by other workers, the larval setae disappear and are replaced by others which sit in deep pits on the mantle edge. In the first few days after mantle reversal a great change takes place in the form and arrangement of the anterior lobe. At first, it is a strawberry-shaped mass having a broadly constricted connexion with the mantle (Plate 4, Figs. 23, 24, 25), and the mantle cavity may be described as a thimble-shaped space lying between mantle and lobe. During the period of flattening, a profound change passes over the apical lobe leading to the elaboration of mouth and lophophore, as well as enabling the development of gut, coelom and other adult structures. It will save some description if it is stated here that the large valve is dorsal and the small valve ventral. This is the direct opposite of what has been accepted and the evidence on which the statement is based will be considered later. Thus, we have at first an apical lobe contained in a mantle cavity which may be uniformly deep all round or slightly shallower on the ventral side than on the dorsal. As flattening proceeds, the ventral side of the cavity decreases in depth, more rapidly laterally than in the mid ventral line, so that the ventral surface of the apical lobe shortens until final disappearance as a recognisable layer. The lobe flattens and shortens until it is a layer lying on the inside of the smaller, ventral valve and replacing the ventral mantle lining: the surface faces upwards. This surface is

Text Figure 2. Diagrammatic representation of transformation from larva to young adult, showing elaboration of anterior lobe and gut, and relation between dorsal valve and dorsal adjustor muscle. From left side. Muscles shown as they have been observed in sections. For interpretation of abbreviations see page 23.

formed from the surface of the apical lobe (there is no obvious reason to think otherwise), so it may be regarded as having in its make-up dorsal, lateral and ventral elements, its anterior border immediately inside the ventral valve edge being the advanced, originally innermost, ventral part of the mantle at the junction between it and apical lobe (Text Fig. 2d). On the remodelled, flattened apical lobe appears the rudiment of the stomadaeum (Plate 5, Fig. 26). The earliest position of the stomodaeal invagination which has been observed was dorsally, immediately behind the tip of the apical lobe before the latter was completely enclosed by the reversed mantle. The invagination is at first a fine pit passing backwards and ventrally in the middle line. Its fine opening elongates backwards forming a thin slit and the stomodaeum as a whole is carried backwards with the changing shape of the apical lobe. The first observed appearance of the stomodaeum showed that the ventral portion of the mantle cavity was still large, but was undergoing reduction (Text Fig. 2c). At this period, the thick dense anterior lobe prevents the stomodaeal rudiment from being seen, but later, when the lobe becomes flatter and thinner, the pit becomes obvious in live specimens as a small median, transparent mark. It grows backwards towards a small gastric vesicle which is formed by the opening up of the previously closed, solid endoderm. When the two become confluent, the mouth begins to broaden and its edge proceeds to give rise to the filaments of the lophophore (Text Fig. 2f). The matter of the origin of the mouth provides a problem which does not at present seem soluble. The blastopore closes ultimately without trace on the mid-ventral side of the apical lobe. This latter becomes reshaped into a flattened area on the inside of the ventral mantle wall. The reshaping ultimately brings the ventral face of the apical lobe into apposition with the dorsal mantle wall and in the same general plane as the originally dorsal face. During the reshaping, the stomodaeum early makes its appearance, but it has not been seen on the ventral side. If the stomodaeum arises on the site of the closed blastopore, it must be morphologically ventral and must move round with the reshaping anterior lobe until it assumes a secondary dorsal aspect. If it does not arise on the site of the closed blastopore, there is nothing at present seen which enables a satisfactory conclusion about its first position on the apical lobe. Having regard to the very great rearrangement of material in the apical lobe, there is nothing inherently impossible in the conclusion that the stomodaeal invagination occupies the site of the closed blastopore and migrates from a primary ventral position to a secondary dorsal one. With the appearance of the mouth, stomodaeum and stomach, the main outlines of metamorphosis may be said to have been laid down. A small, very simple, brachiopod is formed, with peduncle, peduncular area, unequal valves and elementary gut. Later changes, as observed, consist chiefly of the elaboration of these structures along with the further development of the coelom. Addition is made to the free edge of each valve, so that it becomes longer and

wider, and the body as a whole, through growth, becomes relatively thinner. As yet there is no sign of the calcified structure known as the loop, supporting the lophophore. The Growth of the Lophophore. When the stomodaeum has established open connexion with the stomach, the posterior margin of the widening mouth proceeds to give rise to two papillae, one on each side of the mid-line, the rudiments of the first two lophophoral filaments (Plate 5, Fig. 27). While these are still blunt and broad, two more appear, right and left anteriorly adjacent to the first two (Plate 5, Fig. 28). All four elongate towards the centre of the mouth opening. The fifth and sixth usually do not appear simultaneously, the sixth being slightly later than the fifth (Plate 5, Fig. 29). This succession has been observed more often than not (Plates 5 and 6, Figs. 30, 31, 32); thus, at six, ten, and twelve filaments, the sixth, tenth, and twelfth were each smaller than the fifth, ninth and eleventh. Addition to the number of filaments and increase in width of the mouth proceed simultaneously, and along with increase in size of the animal: the lengths of the above-mentioned specimens were relatively, six filaments — 1, ten filaments — 1.5, twelve filaments —2.3. In early summer (November and December), at a time when eight pairs of filaments are formed, there is a rapid extension of the anterior border of the mouth, between the latest filaments, without a corresponding increase in the diameter of the opening. The extra margin is accommodated by its bending inward so as partially to close the mouth by forming a crescentic slit with filaments only on the lateral and posterior border: the anterior invaginated border is always devoid of filaments, no matter how complicatedly folded the mouth becomes, and the addition of new filaments is at the end of each horn of the original crescent distally to those already formed and alternately on each side as described (Plate 6, Fig. 32). The change in the shape of the mouth from a circle to an elongated slit makes possible an increase in the number of filaments and in food intake capacity without any other great increase in the size of the animal. Morse (7, Plate 9, Figs. 90 and 91) describes and figures as tentacles or cirri two small bulges placed anterolaterally on the body of what must have been a very young adult of Terebratulina septentrionalis, while a pore between them is labelled as mouth. From the present account, it is difficult to reconcile what happens in Terebratella with what is reported about Terebratulina, especially as Morse's Fig. 94 in the above named paper, agrees more with what is given in this account of the origin of the mouth. Internal Changes. The unfertilised egg shows in section when stained with iron haematoxylin and eosin, or Erlich's haematoxylin, a dense peripheral layer of granules similar to basal granules of cilia. During the formation of the blastula, the granular layer becomes modified in its distribution, concentrated on the outer side of each cell, making ultimately a blastula having a peripheral layer of granules in the cells

while the rest of the cells contains none. At gastrulation, the granules disappear from the endoderm cells, there being a sharp discontinuity at the blastopore between endoderm and ectoderm. However, the ectoderm cells show a marked eosinophilous zone distal to the nuclei, and this continues into the endoderm, but not into the cells of the early enterocoelic pouch from which will form the mesoderm (Plate 7, Fig. 33). The blastula shows no obvious differences in cell size between one part and another either in early stages or in late. Conklin's account of blastula formation agrees, in the main, with what has been found here. The gastrula, in section, is similar to that described by Conklin, the blastocoele being obliterated. However, the formation of the enterocoelic pouch, from which the mesoderm arises, is strikingly different from that described by Conklin. He shows the pouch as an anterior outgrowth from the archenteron, formed originally by a downgrowth, from the roof, of a transverse wall one cell thick. In the present case, the enterocoelic pouch is a posterior outgrowth from the archenteron, initiated also by a downgrowth from the roof one cell thick (Plate 7, Fig. 33). Here, the young animal has been oriented by means of its direction of locomotion—there is no doubt about the position of the enterocoelic outgrowth. In Conklin's case, the material was already preserved and he had no opportunity to see live material. Before it is concluded that Terebratella and Terebratulina differ fundamentally in this respect, it would be advantageous to have Terebratulina re-examined. The development of the mesoderm here has been followed step by step from the appearance of the early outgrowth from the archenteron. The larger anterior sac becomes the larval enteron, and opens outward through the blastopore. At the time when the single mesodermal outgrowth is formed, the gastrula has changed in shape to the blunt wedge shape when seen from the side and slightly so from the plan: the tapering part contains the mesodermal rudiment (Plate 1, Fig. 5). The mesodermal sac now gives off laterally two forwardly growing horns which push between the endoderm and the ectoderm. The anterior ends of the horns are solid for some time and the coelom extends forwards a little later. At this time the lateral coeloms are continuous with each other and with the enteron (Plates 6 and 7, Figs. 34, 35). As the wedge shape becomes more sharply defined, the coelomic sacs become broader anteriorly and are cut off from each other and the enteron. The enteron becomes relatively short and separates the coelomic sacs in the blunt, thick part of the body, while posteriorly the sacs lie apposed in the middle line. It seems that the changing form of the body, particularly, for this purpose, from the completed gastrula onwards, is largely the expression of varying relations between coelomic sacs and enteric sac. The closure of the blastopore proceeds while the coelomic sacs are being formed, but its position is observable in sections by means of a plug of epiblast leading into the lower wall of the enteron. When the transverse ridge, which is the mantle rudiment, has ex-

tended to the ventral side, the coelomic sacs have divided each into an anterior and a posterior half (Plate 7, Figs. 36, 37), the former lying in that part forming the mantle rudiment and the anterior lobe, the latter lying in the peduncle. The peduncular sacs are apposed in the mid-line and their lumina are slight and for a time continuous dorsally with each other, while the anterior sacs are separated by the enteron, their cavities are rather large, and they extend quickly laterally and somewhat dorsally into the mantle rudiment. It seems that the mantle ectoderm forms an outward fold which is followed by an enlargement of the anterior coelom, since dorsally, while there is a mantle fold of ectoderm, it contains no mesoderm for the time being. All the coeloms contain a coagulable fluid. When the mantle fold has encircled the body and the coeloms have divided, there is readily seen between the fore end of the enteron and the apical ectoderm a mass of cells, somewhat loosely arranged, agreeing in character with what Conklin described in Terebratulina as mesenchyme It is distinctly outside the coelomic wall, the irregular spaces within having no connexion with the coeloms. With the further growth of the larva, the coelomic spaces become reduced, this being well marked in the anterior lobe, where sacs containing fluid become solid masses of cells. The mesoderm so formed extends into the sides of the mantle fold, but, for the time being, not into the dorso- and ventro-median portions. This may account for the attachment to the basal dorso-lateral mantle fold of the anterior ends of the two adjustor muscles which are the first to become defined and are formed each from a peduncular mesodermal mass (Plate 7, Fig. 39). These two masses have had their antero-dorsal ends in close apposition to the ectoderm. The anterior mesoderm, that is, of the mantle, later surrounds the dorsal ends of these two adjustor muscles. Later, the whole mantle is provided with a middle mesodermal layer, but the manner in which this comes about has not been observed. In the early period, the mesodermal cells extending through the mantle are rather scanty and scattered between the ectodermal layers, but after the metamorphosis the mantle has a well-developed mesoderm throughout. As the larva ages, the anterior mesoderm becomes more solid and homogeneous in appearance, but after metamorphosis there reappear left and right coelomic spaces laterally to the oesophagus and in front of the stomach. This reappearance was observed in sections at the stage of four pairs of lophophoral cirri, when the gastric diverticula were seen as the merest antero-lateral angulations of the stomach. The two anterior coeloms, at this stage of eight cirri, are separated by very broad, dorsal mesenteries which are so low that the endoderm is dorsally and ventrally in contact with the ectoderm. Laterally and posteriorly to the stomach there is yet no clear coelom, but the musculature traverses an extensive space which has no obvious epithelial lining. The disappearance and reappearance of the coelomic spaces are roughly parallel to the changes in the endodermal arrangement.

After the separation of the enterocoelic sacs, the enteron lies in an anterior position, the coeloms, as already stated, lying apposed posteriorly. The enteron decreases in size until, towards the end of free swimming life, its cavity has disappeared and there remains a solid rod of endoderm which anteriorly has a downward curve toward the ventral ectoderm of the anterior lobe. This solid endoderm remains throughout the metamorphosis until, later during the remodelling of the anterior lobe, its cavity reappears and opens out through the stomodaeum. The Gastric Diverticula. The earliest trace of the paired outgrowths, to form the so-called liver or hepatic caeca, was seen when the fourth pair of lophophoral cirri appeared. Two specimens, approximately five months old, reared in the laboratory, showed the earliest traces, when, at the same time the adult coeloms were appearing. There was a slight size difference in their fourth pair of cirri as there was in their diverticular rudiments. Specimens with five pairs of filaments show the outgrowths projecting slightly forwards. With further growth, the caeca extend somewhat in parallel, pushing into the coeloms, while at the same time their forward ends bend slightly inwards. They lie parallel to the oesophagus and are set out sharply from the stomach (Plates 5 and 6, Figs. 31, 32). With the appearance of sixteen pairs of lophophoral filaments the diverticula begin to show signs of branching, three slight prominences appearing at the free end of each (Plate 7, Fig. 38a). These extend as finger-like sacs which then divide and are further increased by other outgrowths. At the same time, there was always seen a single outgrowth, postero-laterally from the angle alongside the place of origin of the primary diverticulum (Plate 7, Fig. 38c). At the stage of three pairs of cirri, the intestinal caecum makes its first apparance as an outgrowth on the dorsal side of the enteron. Its subsequent growth calls for no comment, except that the end of the sac early comes into contact with the dorsal body wall. The Orientation of Pre- and Post-metamorphic Forms. Conklin (op. cit.) showed that the mantle is a dorso-ventral growth, dividing the posterior peduncle from the anterior lobe, and decided that “the valves which are formed by the mantle folds are dorsal and ventral, while the opening of the valves is anterior” (p. 61). Students of Brachiopoda—e.g., Schuchert and Cooper (9, p. 8) name the two valves dorsal and ventral, and seem to assume that the parts so arbitrarily named coincide with larval dorsum and ventrum. Conklin nowhere provides information which satisfactorily links the larval and adult dorsal and ventral surfaces, except that in his Figures 35, 61 and 62 he shows patches of cells which he interprets as rudiments of ventral sense plate and suboesophageal ganglion. This interpretation must have been based on an assumption that the parts did in fact pass over to the young adult and form portions of the nervous system, which could not be known since he was apparently unaware of the process of metamorphosis of Terebratella apart from what was provided by Morse's work. Morse's results give no clue to the relation between the fine structures of larva and adult.

Fig. 1.—Optical section through part of equator of egg showing follicle cells. Figs. 2, 3, 4.—First, second and third cleavages. Figs. 5, 6.—Late gastrula, showing disappearance of posterior cilia, differentiation into regions of anterior lobe and mantle and stalk, and appearance of ventral groove from blastopore. Live material. Figs. 7, 8.—Complete loss of posterior cilia; appearance of apical cilia. Live material. Fig. 9.—Differentiation of anterior lobe, from blastoporal face. Live material. All figures based on camera lucida drawings.

Fig. 10.—Differentiation of anterior lobe, from left side. Live material. Figs. 11, 12.—Closure of blastopore, appearance of mantle rudiment and of rudiment of peduncle. Fig. 11 from dorsal side (blastoporal site seen by transparency), Fig. 12 from left side. Based on live material and whole mounts. Fig. 13.—As Fig. 11, from ventral side. Live material. Fig. 14.—Later stage with enlarging mantle fold, and more slender peduncle. Live material, lateral view. Fig. 15.—As Fig. 14, dorsal view. All figures based on camera lucida drawings.

Fig. 16.—Older larva, with longer mantle fold and more slender peduncle. Apical tutt absent. Live material. Fig. 17.—Mature larva on emergence from parental mantle cavity, with eyespots and setae. Based on live material and whole mounts. Fig. 18.—After two hours' attachment. Live material. Fig. 19.—After attachment; mantle beginning to reverse. Live material. All figures based on camera lucida drawings.

Fig. 20.—Shortly after mantle reversal, pai tial enclosure of anterior lobe. Live material. Figs. 21, 22.—Enclosure of anterior lobe, Fig. 22 with early calcified shell. Live material. Figs. 23, 24, 25.—As 20, 21. 22. Fig. 23 showing left pedicle adjuster muscle. Fig. 24 vential side, Fig. 25 dorsal side. Whole mounts. All figures based on camera lucida drawings.

Fig. 26.—Early appearance of stomodaeal invagination, before its union with rudimentary enteron. Ventral side. Whole mount. Fig. 27.—Rudiments of first pair of lophophoral cirri. Ventral side. Whole mount. Fig. 28.—Rudiments of first two pairs of lophophoral cirri. Ventral side. Whole mount. Fig. 29.—Three pairs of lophophoral cirri, fifth and sixth appearing in succession. Ventral side. Whole mount. Fig. 30.—Five pairs of lophophoral cirri. Whole mount. Fig. 31.—Six pairs of lophophoral cirri. Gastrie caeca well formed. Whole mount. All figures based on camera lucida drawings.

Fig. 32.—Nine pairs of large lophophoral cirri, and a nineteenth cirrus appearing. Inflection of anterior border of mouth producing a crescent-shaped slit. Whole mount. Fig. 34.—Nearly horizontal section of slightly later stage than Fig. 33, showing mesoderm lying laterally to enterior, and posterior connexion between endodermal and mesodermal cavities. All figures based on camera lucida drawings.

Fig. 33.—Sagittal section of late gastrula, wedge-shaped, showing posterior enterocoelic pouch opening from archenteron, and showing absence of granular layer from endodermal and mesodermal cells. Fig. 35.—Optical section of same stage as Fig. 34, where anterior lobe is marked off and blastopore closed, showing posterior continuity between enteron and coelomic cavity. Whole mount. Fig. 36.—Dorsal view at stage of appearance of mantle rudiment, showing right and left anterior and posterior coelomic sacs and restricted anterior position of enteric cavity. Whole mount. Fig. 37.—Ventral view of same stage as Fig. 36. Whole mount. Fig. 38 a, b, c.—Stages in elaboration of right gastric diverticulum. Fig. 39.—Transverse section of base of mantle of late larva after emergence, showing dorsal (pedicle) adjustor muscles connected to dorsal surface. All figures based on camera lucida drawings.

In the present study, it has been possible to follow one batch of material in development from early gastrula to fixation, but only one specimen survived to fix itself and it died before any further change took place. Two lots were killed and preserved daily and sketches were made similarly, so that a reliable set of records exists of that particular material. Another brood of larvae emerged naturally from a female and proceeded to fix, as has already been described. It was not possible in the circumstances to decide by direct external observation how the orientation of the newly-fixed larva was related to that of the young adult. There was no certainty that rotation did not take place, and the striking changes in the form of the anterior lobe precluded any following of the parts from one condition to another during a period of 24 hours or more. Thus, from that standpoint, there was little more hope of connecting larval and adult orientation than there was of linking Morse's and Conklin's observations. The clue to the connexion lies, however, particularly in the presence in the late larva, while still free swimming, of a pair of muscles, right and left, passing from the peduncle forwards and upwards, each being attached dorso-laterally in the base of the mantle, the place of attachment being marked externally by a small indentation (Plate 7, Fig. 39). After mantle reversal, and for a short time later, these are the only muscles distinguishable. They are then connected with the larger mantle flap which produces the larger valve and are the so-called ventral adjustor muscles: they must now be renamed. In view of the fact that the adjectives “dorsal” and “ventral” have already been used on the understanding that the smaller and larger valves are dorsal and ventral, confusion would arise merely by changing their application. It seems therefore advisable to cease the usage of the words “dorsal” and “ventral” and to maintain the use of other terms such as are given by Thomson (11, p. 4). It seems that the words “pedicle” and “brachial” would well satisfy the need of the conchologist and palaeontologist as far as these are concerned with the Testicardines. Lingula, for reasons to be later considered, falls into another category. The “ventral sense plate” of Conklin is, in Terebratella, represented also by a well-marked thickening along the mid-ventral line of the mantle of the late larva, and, in section, is densely nucleated. In the present study, it is not seen to form anything significant in the newly reversed animal, and cannot yet be said to contribute anything important to the structure of the adult nervous system. It becomes obvious that only by examination of material in various stages of transformation the relations of the parts of pre- and post-metamorphic can be decided. Quite truly, we must accept the view that the ventral side of the late larva is the same as that of the early one, or we must reject it. If the former, then the identification of the dorsum and the ventrum of the sedentary animal is easily possible by means of the musculature: if the latter, then at present there is nothing which has been found to persist throughout early life which might conveniently be used as a datum of reference. There is no reason to doubt that the greater dorsal curvature of the early larva remains until metamorphosis, and this determination,

based on a number of recorded steps, refers in the beginning to the position of the blastopore. Conklin's orientation, as far as it goes, agrees with the present one, except in the case of the gastrula when the enterocoelic sac is in formation. The Appearance of Muscles. It has been noted that the dorsal (pedicle) adjustor muscles appear before fixation, and that they enable the determination of dorsal and ventral surfaces after metamorphosis. When the apical lobe begins to show the earliest signs of remodelling, about the time when the valves show a slight size difference, four pairs of muscles are already present, the adductors, the protractors, the divaricators and the dorsal (pedicle) adjustors. At the first stage after mantle reversion, the mantle cavity is extensive, but undergoes reduction in depth, antero-posteriorly, when the body flattens, particularly along the sides. As a result, the mantle cavity becomes saddle-shaped with the ventral limb continually shortening as the anterior lobe is remodelled. Antero-laterally to the median portion of the cavity lie the right and left adductor muscles, well forward, relatively much more so than in the adult. The divaricators lie obliquely running backwards and upwards immediately behind the mantle cavity which also for a time projects backwards between them dorsally and ventrally. The pedicle adjustors converge obtusely towards the peduncle from sites closely associated with the dorsal ends of the divaricators. The protractors lie nearest the middle line between the adjustors and pass almost vertically between the valves. Text Figure 3. Diagrammatic ventral view of same stage as Text Fig. 2b, showing distribution of   muscles and mantle-cavity. When the muscles have appeared, the mesoderm is with difficulty recognisable as a distinct layer, the apical mass consisting, apart from the muscles and small endodermal body, of a uniformity of cells. Later, when the anterior lobe begins to reform, cell layers become distinct while forming stomodaeum and gut, but for some time there is little in the main mass to distinguish one germ layer from another. As has been seen, addition to the length and width of the shell takes place chiefly at the edge of the valve, so that the mantle cavity increases in capacity. This makes possible the expansion of the anterior lobe during remodelling, and also brings about the location of the musculature in the posterior region. The elaboration of the mouth to form the lophophore proceeds concurrently with the growth of the shell.

The Vascular System. The lophophoral cirri are solid until three pairs have appeared, when they and their successors become hollow with a single tube. The cavities later become continuous with a series of spaces round the gullet, and, as the lophophore becomes more intricate in its folding, so do the related spaces become more involved. The cirrial canals are early lined by an endothelium which continues throughout the perioesophageal sinuses. No sign of a heart-like organ has been found connected with this system of spaces, and no connexion has been found between these vascular spaces and the coelom. The absence of any sign of a cardiac organ leads to the question of the causation of blood flow. The cirri are provided with longitudinal muscle fibres and a skeletal rod consisting of an inverted grooved strand which contains the vascular space. The groove is closed by the band of muscle fibres and the vascular endothelium lines the tube so formed. If the skeleton be suitably elastic, it would tend to resume its original straight position after being bent by the muscle band. These bending and straightening movements are to be seen not commonly as in polyzoan tentacles, but they occur, and could serve in a crude way to help in blood circulation. Gonocoele and Coelomoducts. The adult has commonly four gonadial cords in the dorsal mantle, the two inner being unbranched, the two outer being branched on the outer sides. Two cords lie in the ventral mantle, with outer branches. The dorsal cords are in two pairs, right and left, each pair having a common base on the wall of the common coelom. The ventral cords lie separately, posteriorly, each on its own side of the common coelomic wall. No visible connexion occurs between the dorsal and ventral cords. The gonadial masses arise as proliferations from the walls of the gonocoeles, and ultimately fill these spaces at sexual maturity. At a size of 1.5 mm. long and 1 mm. broad, about eleven months of age, the gonocoeles appear as divertieula of the common coelom. The ventral sacs at first project into the wall immediately behind the angle formed by the stomach and the blind gut. They pass forward a short distance and then emerge as thin walled sacs projecting into the common coelom one on each side of the gut. At this stage they are quite short and project no further toward the ventral mantle. The dorsal gonocoeles are a pair of dorsal projections of the common coelom into the tissue of the dorsal mantle and may show indications of primordial germ cells. No subdivision has yet taken place. The coelomoducts are two, right and left, and, at the end of the first year, lie horizontally as blind diverticula of the coelom terminating forward at each end of a transverse fold a little behind the mouth. The internal ends of the coelomoducts are yet plain funnels and the walls distinct and one cell thick. Later, at earliest sexual maturity about the end of the second year (length 6 mm., breadth 4 mm.) the ducts have acquired external openings. Their position has changed in accordance with a general rearrangement

of parts through growth in the previous year. They lie on the anterior wall of the common coelom, their course being somewhat vertically upwards. The internal openings are now plicated funnels and the wall is clearly ciliated. An immature specimen of this size showed no sign of external opening. The coelomoducts may be primarily gonoducts, since they have been found closed until sexual maturity. Up to the stage of ten pairs of cirri, no sign of either gonocoele or gonoduct has been found. Excretion may, therefore, take place through the general ectoderm, but later also through the open gonoduct. Comparative Embryology. Thomson (op. cit., pp. 32–4) summarises the, then, knowledge of brachiopod development. It will serve today as a satisfactory source of modern information. It appears that the close studies of embryology are those of Yatsu (12) on Lingula and Conklin (op. cit.) on Terebratulina. Kowalewski (3) contributed valuable information and figures about the development of Argiope and Thecidium. These developments fall into two well marked groups, one containing Lingula and the other containing the rest. These two groups naturally coincide with Ecardines and Testicardines. Those in the second group, Terebratella, Terebratulina, Argiope, Thecidium, are uniform in that the larva consists of apical lobe, sheath-like rudiment of mantle and peduncle, and that metamorphosis proceeds by reversion of the mantle. Gastrulation is commonly by invagination and, where the origin of the mesoblast is known, there is usually found an enterocoelic diverticulum. Segmentation of the fertilised egg, where known, is generally similar, with the gradual growth of a spherical blastula. Dawydoff (2, p. 330, Fig. 146) figures, after Kowalewski, a pair of lateral enterocoelic pouches from which the mesoderm of Argiope is said to arise. Oehlert and Deniker's analysis of Kowalewski's work (op. cit.) shows the mesoblastic sacs of Argiope as coming from the end of the enteron in somewhat the same way as they do in Terebratella and Terebratulina. Sedgwick (10, p. 582) says that “the last remnant between the two” (i.e., mesoblast sacs and enteron) “is at the front end of the body.” This statement is quite reasonably based on the Fig. 3 of Oehlert and Deniker's analysis. In Figs. 4 and 6 of this analysis, there is seen a set of relations between coelomic sacs and enteron very similar to that in Terebratella. The question arises, therefore, is the primary enterocoelic pouch of Terebratulina and of Argiope anterior or posterior in origin? Kowalewski had not available to him the technique of modern microtomy and Conklin was unable to study his material alive. In these cases, it would be possible to mis-orientate small objects through inability, due to force of circumstances, to relate a given external form of live larva with a given internal arrangement. We need, therefore, further information about these matters. Morse concluded that the lophophore of Terebratulina arose outside of the mouth, Kowalewski thought that the cirri of Argiope were formed from mantle and concluded that in Thecidium the anterior

lobe made a contribution. The gut of Terebratella passes backwards and upwards, without an anus. Lingula stands in sharp distinction from those above-mentioned in its early development. It has a relatively long embryonic life, emerging from the egg membrane about the time of appearance of the first pair of cirri. Segmentation gives rise to a blastula consisting of two apposed layers each of 16 cells, similar to the condition in phylactolaematous Polyzoa, as shown by Yatsu. The mesoderm arises as two lateral proliferations from the endoderm, and the coelomic spaces are schizocoeles. The mouth arises on the site of the closed blastopore and is ventral in position, the lophophore is definitely extra-oral in origin, the gut continues backwards, downwards, and forwards to open by an anus. The mantle begins as a posterior circular flange which later becomes bilobed, the peduncle appears late, having a quite different kind of origin and constitution from that in Terebratella and Terebratulina and, by development, the small value is dorsal and the large one ventral. The peduncle is related to the large value. Thus, the orientation of Lingula is essentially that of the embryo, without any remodellings or reversals to make possible drastic rearrangements, the adult being achieved without metamorphosis. It becomes clear that between Lingula and Terebratella (along with the other Testicardines) there are profound differences in development. Many of the striking facts have already been reviewed immediately above. Included in them must be the nature of the shell of the adult. Thomson (op. cit., p. 115) mentions that the shell of his Gastrocaulia (approximately equal to the Ecardines) is of chitinophosphatic material (tricalcium phosphate and chitin or keratin, see Thomson, p. 96), or of something derived therefrom. The shell of his Pygocaulia (approximately equal to the Testicardines) is calcareous (i.e., of calcium carbonate) and for the most part with fibrous prismatic structure (op. cit., p. 116). These shell differences indicate very sharp physiological differences, which are properly taken into account in the systematic arrangement of Brachiopoda. It seems, therefore, that Brachiopoda consist of two very sharply differentiated groups, from the standpoints of embryology and physiology. Indeed, the adult features of Ecardines (Gastrocaulia) and Testicardines (Pygocaulia), which are regarded as common, may be quite properly looked at as convergent and as in no way indicating a common origin. Sedgwick (10, p. 572) defines the phylum Brachiopoda as “Fixed, solitary, apparently unsegmented Coelomata, with a tentaculated buccal groove often prolonged into arms, and a bivalve shell.” This statement says the least about the phylum, yet enables the distinction of a brachiopod from any other animal, serving until it is possible to inquire more specifically into the constitution of the phylum. The accepted orientation of the brachiopod body seems to have been based on the assumption of the homology of the structures in the group as a whole. The dissimilar valves, the lophophore, the position of the mouth, the bent gut, the nervous system, the peduncle could all be conceived as homologous without any knowledge of early development, but as soon as this is known it becomes clear that the assumption of homology is to a considerable extent, if not entirely,

false. The brief review, in this section, of the chief developmental and physiological features of Lingula and the Testicardines, taken with a knowledge of the function of the adult structures, particularly the lophophore, justifies the conclusion that Lingula and, for instance, Terebratella show a high degree of convergence, and, from the standpoint of Brachiopoda, show nothing that is acceptable as divergence—i.e., from a common ancestor. It appears not unreasonable, in the light of the foregoing, to consider the separation of Lingula and its allies completely from Terebratella and its allies, even to the extent of placing them in separate phyla, as has been done with Polyzoa Ectoprocta, and Polyzoa Endoprocta. Lowenstein (4) has placed these under the phyletic names of Bryozoa and Calyssozoa respectively, thus resolving a difficulty which has been long apparent. Text Fig. 4.—Lingula anatina with 10 pairs of cirri (after Yatsu). a., anus; a.l.l., antero-dorsal lobe of liver; c., cirrus; coel, coelom; ep., epistome; m., mouth; m.d., dorsal mesentery; m.g., mid-gut; mv., ventral mesentery; n.f., nephridial funnel; n.t., nephridial tube; p., peduncle; p.l.l., postero-dorsal lobe of liver; p.o.m., posterior occlusor muscle; s., statocyst; v.g., ventral ganglion; v.l.l., ventral lobe of liver. To whatever lengths the students of Brachiopoda may go in arranging their material, it seems clear that Pygocaulia and Gastrocaulia are inadequate as sub-classes. Too much similarity is implied, when the fact is that so much difference exists. Further knowledge about the development of the relatives of Lingula will throw light on this very important problem in Brachiopod systematics. Summary. 1. The development of Terebratula inconspioua has been studied from material gathered in Lyttelton Harbour, New Zealand. 2. The breeding season is in April and May. 3. Segmentation is equal, producing a blastula which gastrulates by invagination. The gastrula is ciliated. 4. The blastopore closes completely and all trace vanishes. 5. An apical tuft of long cilia is present for part of the development which proceeds considerably in the female mantle cavity. 6. The mesoderm originates as a posterior enterocoelic pouch which ultimately separates into right and left sacs. The sacs divide into anterior and posterior segments. 7. The embryo differentiates first into a ciliated anterior lobe with apical tuft and an unciliated posterior peduncular rudiment. 8. The mantle rudiment arises as a transverse dorsal fold bordering posteriorly the apical lobe, and passes downward laterally, the two sides ultimately joining to complete the fold ventrally. 9. The mantle rudiment extends backwards to ensheath the narrowing peduncle. It ultimately bears four tufts of marginal setae. 10. The larva on leaving the mantle cavity bears a ring of pigment spots on the posterior border of the apical lobe.

11. On attachment by the peduncle, the mantle sheath reverses and encloses the apical lobe which flattens dorso-ventrally, leaving a slit-like opening to the mantle cavity now formed. 12. The apical lobe is reshaped and is extended on the inside of the ventral valve, the valves by now having been calcified. 13. The stomodaeal invagination is a dorsal invagination on the reshaping anterior lobe and grows in to join the appearing gastric cavity which arises in a solid endodermal mass, the original enteron having disappeared. 14. A pair of dorsal adjustor muscles appears before metamorphosis, in the late larva, formed from the posterior coelomic sacs, and shows the pedicle valve, formerly called ventral, to be dorsal. The other muscles appear early after reversal. 15. The mesoderm early becomes solid and the adult coelomic cavities appear later about the time when the adult gastric diverticula appear. The germinal cavities are extensions of the adult coeloms. 16. The lophophoral cirri are formed from the edge of the mouth. 17. An attempt has been made to determine the life history of a population. 18. Terebratella and Lingula, in development, are compared and contrasted. It is proposed that the differences are so great that the adults are convergent in structure. List of References. 1. Conklin, E. G., 1902. The Embryology of a Brachiopod, Terebratulina septentrionalis, Cout. Proc. Am. Phil. Soc., vol xl, pp. 41–76. 2. Dawydoff, C., 1928. Traité d'Embryologie comparée des Invertébrés. Paris. 3. Kowalewski, M., 1883. Observations sur le Development des Brachiopodes. Analysis by Oehlert and Deniker. Arch. de Zool. Expérim., Ser. 2, vol. i, pp. 57–76. 4. Löwenstein, O., 1940. A Text Book of Zoology, by Parker and Haswell, 6th ed., vol. i, London. 5. Macbride, E. W., 1914. Textbook of Embryology. Invertebrata. London. 6. Morse, E. S., 1817. On the Early Stages of Terebratulina septentrionalis. Mem. Boston Soc. Nat. Hist., vol. ii, pp. 29–39. 7. —– 1873. On the Embryology of Terebratulina, Ibid, vol. ii, pp. 249–264. 8. Pelseneer, P., 1906. A Treatise on Zoology. Part V. Mollusca. Lankester. 9. Schuchert, C., and G. A. Cooper, 1932. Brachiopod Genera of the Suborders Orthoidea and Pentameroidea. Mem. Peabody Mus. Nat. Hist., IV (1). 10. Sedgwick, A., 1898. A Student's Textbook of Zoology, vol. i, London. 11. Thomson, J. A., 1927. Brachiopod Morphology and Genera (Recent and Tertiary). Wellington, New Zealand. 12. Yatsu, N., 1901–3. On the Development of Lingula anatina. Jour. Coll. Sci. Imp. Univ., Tokyo, vol. xvii, Art. 4, pp. 1–112. Abbreviations. a.l. anterior lobe. a.s. adult seta. a.t. apical tuft. bp. blastopore. bp g. blastoporal groove. bp.s. blastoporal site. c.s.a. anterior coelomic sac. c.s.p. posterior coelomic sac. d. dorsal side. e. enteron. ect. ectoderm. e.d. enterocoelic diverticulum. e.l. solid larval endoderm. e.m. edge of mouth. end. endoderm. e.r. rudiment of adult enteron. e.s. pigment spot. f. follicle cell. g.d. gastric diverticulum. i.b.m. inflected border of mouth. int. intestine. l.c. lophophoral cirrus. m. mantle before reversal. m1. mantle after reversal. m.a.l. left adductor muscle. m.adj.l. left dorsal adjuster muscle. m.adj.l1. left ventral adjustor muscle. m.c. mantle cavity. m.div.l. left divaricator muscle. mes. mesoderm. m.pro.l. left protractor muscle. m.r. mantle rudiment. mth. mouth. oes. oesophagus. p. peduncle. p.r. peduncular rudiment. s. seta. s.r. stomodaeal rudiment. st. stomodaeum. v. ventral side. v.d. dorsal or pedicle valve. v.v. ventral or brachial valve. z.e. eosinophil zone. z.g. zone of granules. * The writer is indebted to the Council of Canterbury University College for a monetary grant in aid of publication of this paper.