Subsequent to the liberation of the ovum there is a gap in the series of stages which I have as yet obtained. As is well-known, the early cleavages of eggs of echinoderms are passed through rapidly. The eggs of Ophiothrix fragilis reach the gastrula stage within thirty-six hours according to MacBride (1907), as also do those of Ophiura brevispina (Grave, 1916). Even in the heavily yolked egg of Kirk's ophiuroid the blastula stage is reached within forty-eight hours (Fell, 1941). As the egg of Amphipholis squamata occupies a position morphologically intermediate between these extremes it is most probable that the time occupied in passing through the early stages up till the formation of the blastula is between thirty-six and forty-eight hours. It is obvious that embryos at such stages must be rare, partly because of their short duration, and partly because only one egg is formed at a time, instead of hundreds or thousands, as is more usual in echinoderms.

Russo (1891) and also Apostolides (1882) have given accounts which claim to describe the process of cleavage of the egg of A. squamata. These two accounts, however, are conflicting. Since Russo could never have actually observed an ovum of Amphipholis squamata—for, as shown above, his supposed description of it can only be applied to an extremely immature oogonium—no particular importance need be attached to his statements in regard to its segmentation. Apostolides' older account, on the other hand, is of some interest, because the peculiar type of cleavage he believed to occur has since been recorded as a variation affecting about thirty per cent. of eggs of Kirk's ophiuroid. He believed that the first cleavage is markedly unequal, resulting in a two-celled embryo in which one blastomere is about thrice the size of the other. In two successive divisions the larger blastomere gave rise to three cells, thus producing a four-celled embryo with equal blastomeres. Thereafter the cleavages are stated to be normal. A very similar plan of cleavage has been recorded and figured for Kirk's ophiuroid (Fell, 1941), and a comparison made with what Apostolides described. However, it is important to note that in Kirk's ophiuroid it affected only a minority of the cases observed, and in more than sixty per cent. of the eggs the first two cleavages did not differ from the usual radial pattern.

Fortunately, there is already available a very satisfactory series of descriptions of the cleavages of eggs of other ophiuroids in which the reproductive cells are produced in far greater abundance. These range from the comparatively non-yolky forms such as Ophiothrix fragilis (MacBride, 1907) and Ophiocomina nigra (Narasimhamurti, 1933), through the intermediate type, such as Ophiura brevispina (Grave, 1916), to the heavily yolked type, illustrated by Kirk's ophiuroid (Fell, 1941). From the facts known about the above species it should be possible to deduce with a reasonable expectation of suceess the nature of the corresponding cleavages of Amphipholis squamata; for all the evidence provided by its development, the greater part of which is now known and described in this paper, indicate that this species is of an intermediate type. As shown elsewhere (Fell, 1945), it is only in the most heavily yolked type that

any alteration from radial cleavage occurs in the earliest stages of segmentation. As Amphipholis squamata is not of this extreme type, it is therefore highly probable that the early cleavages will in normal specimens differ in no important way from those of Ophiothrix, Ophiocomina or Ophiura.

For the reasons given in the preceding paragraph I therefore believe that the type of unequal cleavage described by Apostolides, if it occurs at all, is a variation comparable to what has been described for Kirk's ophiuroid.

Fig. 5.—Early Gastrula.
E.Memb., egg membrane; Blc., blastocoel; Po.B., polar body: Out. Mic., outer micromeres; In.Mic., inner micromeres; Mae., macromeres; Blp., blastopore.

In regard to the form of the blastula, though again an actual example has not been obtained, it can be reconstructed by examination of an early gastrula (Fig. 5), which is the earliest stage subsequent to the ovum that I have been able to obtain as yet. This early gastrula can be recognised at once as showing affinities with the early gastrula described for Kirk's ophiuroid, and is also very similar to the gastrula which Grave (1900) described for Ophiura brevispina. As the form of the gastrula is determined by that of the blastula, it is easy to deduce the latter from the former, particularly when there exist two other species of similar type in which both stages are known. In the specimen figured, a solid mass of macro-meres has been invaginated from the vegetal pole, to constitute a mes-hypoblast which is obliterating the blastocoel. The latter cavity is, in the stage shown, reduced to a narrow lumen, crescentic in vertical section, towards the animal pole. Reference to my previous paper (Fell, 1941) will bring out more clearly the similarity of this gastrula to the corresponding stage of Kirk's ophiuroid. As in the latter species, only a very reduced blastopore is present, and there is no true archenteron. A single cell lying at the animal pole, between the epiblast and the egg (fertilisation) membrane is probably one of the polar bodies. It is obvious that the preceding blastula, as in

Kirk's ophiuroid, must have comprised a thick-walled sphere of cells surrounding a somewhat reduced blastocoel, and the cells would be differentiated into micromeres at the animal pole, and macromeres at the vegetal pole. The gastrula shows that it is from the latter elements that the mes-hypoblast is derived; no evidence has been found of any process of inturning of micromeres to form part of the hypoblast as has been described for Kirk's ophiuroid. In Ophiura and Kirk's ophiuroid the archenteron is formed much later by a secondary hollowing out of the central inturned mass of macromeres. The next stage obtained in this species shows that a central cavity has arisen in the hypoblast, and there can be no doubt that this, as in the two other species, arises by a similar secondary splitting; for there can be no possibility of an archenteron arising in the way in which it does in the non-yolky forms. As pointed out in previous papers (Fell, 1940a, 1941) this reduced mode of gastrulation is attributable to the reduction of the blastocoel in relation to the large, yolk-laden mass of the blastomeres, making typical invagination an impossibility for physical reasons.

Two important differences are to be observed from the conditions seen in Kirk's ophiuroid, firstly, the blastocoel, though reduced, is not so much so as in Kirk's ophiuroid—resembling more that described for Ophiura (Grave, 1916). Thus, in gastrulation all the mes-hypoblast can be inturned in one mass, rendering unnecessary the long process of epibolic inwandering of micromeres which follows in Kirk's ophiuroid. This distinction is consistent with the smaller amount of yolk present in Amphipholis squamata, and the consequent lesser degree of modification which the development has undergone.

A second difference from Kirk's ophiuroid is provided by the varied extent of the pigmentation of the mes-hypoblast and epiblast in Amphipholis. The epiblast cells are paler and clearer than the inturned macromeres, the latter having an opaque reddish tint. It can be seen that in this respect Amphipholis shows some degree of similarity to Paracentrotus, for in both cases the pigmentation becomes restricted to the cells that are to give rise to the endoderm. Why changes in the distribution of pigments should be found in these species and not in others is an interesting problem. It may be that it is to be understood in terms of whether or not the yolk—and consequently the pigment associated with the yolk granules—is sufficient in quantity to be present not only in the hypoblast cells, but also in the epiblast. In Kirk's ophiuroid, where there is such an abundance of yolk, all germ layers contain yolk material in quantity, and all, as would be expected, are pigmented. In Amphipholis, on the other hand, there is less yolk, and this is mostly aggregated in the mes-hypoblast cells, the pigmentation at the same time being denser in that region. The conditions in this regard in Paracentrotus do not appear to have been described, so that it remains to be seen whether this explanation can be applied to it too.

The next stage which has been observed is shown in section in Fig. 6. Here the hypoblast has developed a central cavtiy, to form the archenteron—evidently through an internal splitting, as in Ophiura brevispina and Kirk's ophiuroid. The archenteron lies

Fig. 6.—Late Gastrula.
Mesch., mesenchyme; Post. P., posterior pole; Arch., archenteron; Ect., ectoderm; Oes.Rud., oesophageal rudiment; Ant.P., anterior pole.

centrally within the outer shell of ectoderm cells, and between the two is a small amount of mesenchyme, probably derived as in Kirk's ophiuroid from the outermost cells of the inturned mes-hypoblast mass. At the future anterior end a solid ingrowth of ectoderm cells as a stomodaeum is commencing, to form the rudiment of the oesophageal sac. Thus, here as in other echinoderms, the oesophageal cells are originally ectodermal. As will be seen below, it is a notable feature of the development of Amphipholis that future hollow organs frequently arise first as solid masses of cells, in which cavities later form by splitting. This is a feature to which I have already drawn attention in previous papers (Fell, 1940a, 1941).

The series of stages from now onward is fairly complete, but before proceeding to describe them it is necessary to refer briefly to some statements made by earlier workers on the mode of formation of the endoderm. Metschnikoff (1869) studied only some late stages which he had obtained, and so never had occasion to suspect that there might be anything unusual about gastrulation. Consequently, he assumed that invagination occurs, no echinoderm at that time being known to form its endoderm by any other method. Apostolides (1882), however, stated that the endoderm arises by delamination, though he gave no description of the process, nor did he give any figures, despite the very surprising nature of his claim, seeing that at that period it was believed that invagination is universal in echinoderms. Russo (1891) repeated Apostolides' statement, and gave some highly diagrammatic figures, supposed to illustrate the process—which he compared to that found in Hydra. Thus the matter remained till 1914, when MacBride dismissed the statement as highly improbable, and no further attention was paid to the matter till I called attention to Russo's work (1940a), showing