dogmatic that the coelom of echinoderms is always formed enterocoelously from pouches nipped off from the archenteron. Russo (1891) again was the first to state that the perivisceral coelom arose in Amphipholis squamata by a process of splitting in mesenchyme. It is instructive to note the way in which his results came to be ignored and forgotten. MacBride (1892) in his own paper on the very late stages of development of the same species makes no mention of the earlier development whatever. He does, however, mention that “. the coelom in Amphiura squamata is represented at first by a mass of mesenchyme; and as this condition of things is certainly not primitive, I do not think that even if reliable results as to the development of the cavities originating before the coelom clears were obtainable, they would be of much phylogenetic importance.” The reasoning here is very confused. First, there is no proof that the mode of origin of the coelom in Amphipholis is not primitive—for, as shown in my analysis at the beginning of this paper, there are as many directly developing echinoderms known to science as ones with indirect development. Secondly, MacBride expresses his opinion that any reliable results as to the development of the cavities originating before the coelom clears (italics mine) would not be of much phylogenetic importance. Here is a complete reversal in his attitude, for the only cavities which originate in the embryo before the coelom “clears” are the vestigial enterocoelous pouches! If, then, MacBride considers these vestigial enterocoels of little phylogenetic importance, why does he also regard the schizocoelous coelom as “certainly not primitive”?

However, it is clear from MacBride's next paper (1907) that he once more reversed his position, for he now proceeds to quote Metschnikoff's account of the enterocoels in Amphipholis, and uses this account to disprove Russo's later one. Still no actual reinvestigation has taken place of the development of Amphipholis to confirm or disprove Russo's account.

Finally, in MacBride's “Text-book of Embryology” (1914) Russo's results are rejected as “improbable in the highest degree”—though still no reinvestigation of Russo's work had been done. Russo's conclusions in regard to the origin of the coelom in Amphipholis squamata remained forgotten till, in the couse of my work on Kirk's ophiuroid, I was amazed to find the coelom originating as a schizocoel in a mass of mesenchyme. Such a mode of origin was of course foreign to embryological theory as taught in the text-books. Eventually I was able to obtain photostats of Russo's paper from the British Museum, and then the full similarity between his account and the conditions independently observed by myself became evident. My paper on the development of Amphipholis squamata contains the results of a reinvestigation of his work, and confirmation of his description of the origin of the schizocoel in mesenchyme (Fell, 1945).

Having thus described the rather belated recognition of a schizocoel in ophiuroid development, we can now pass to a comparison of the modes of origin of the coelomic cavities in the various types of development (see Figs. 25–27).

Figs. 25 and 26.—The enterocoels in Ophiothrix and vesicles of Amphipholis. In Ophiothrix (25) all parts of the coelom arise as enterocoelous vesicles. In Amphipholis (26) the number of vesicles is reduced, and none of them is an enterocoelous pouch. Only the hydrocoel survives, the other vesicles degenerating and contributing to the general mesenchyme, in which the remainder of the coelomic structures is later formed schizocoelously. Oes., oesophageal sac; Stom., stomach; Mesench., mesenchyme; Lft., Rt., Hydr., left and right hydrocoels; Lft., Rt., A. Coel., left and right anterior coelomic vesicles; Lft., Rt., P. Coel., left and right posterior coelomic vesicles.

In Ophiothrix, representing the non-yolky type, MacBride's description records that both right and left enterocoels form. The left divides into anterior and posterior parts, and soon afterwards the right does so also. From the posterior end of the left anterior coelom arises the hydrocoel. The wall of the left posterior coelom forms the arms, and its lumen the general coelom. The right coeloms become vestigial. MacBride also claimed that the right hydrocoel rudiments sometimes developed the five-lobe pattern, as does the left, but Narasimhamurti has since shown that this was a misinterpretation.

In Amphipholis, representing the moderately yolky type with vestigial larva, right and left enterocoels form. The right member disappears and the left gives rise only to the hydrocoel, and occasionally to a small posterior pouch which disappears later. The definitive general coelom arises by a process of splitting in an extensive mass of mesenchyme.

In Kirk's ophiuroid, where the embryo is heavily yolked and the larval stage completely omitted, both hydrocoel and general coelom

and its derivatives arise by splitting in mesenchyme. The hydrocoel is the first to form, and the embryo is free-living for some days before any general coelom commences to form.

In Ophiomyxa brevirima a closely similar condition holds, the coelom being preceded by an extensive zone of mesenchyme (Fell, 1940 b).

Fig. 27.—Schizocoelous origin of the coelom in a directly developing ophiuroid (Kirk's ophiuroid). A, the “rosette” stage, a solid gastrula assuming radial symmetry, and with budding podia; B, later stage, with the hydrocoel developing as a system of splits in the mesenchyme; C, general coelom developing by splitting in mesenchyme at a later stage than B. All represented in vertical section. Simplified from Fell (1941). Ect., ectoderm; Mes-end., mes-endoderm; Pod., podium; End., endoderm; Mo., mouth; Hydr., hydrocoel; Coel., general coelom; Mesench., mesenchyme.

To sum up, we note that as the yolk mass increases so does the tendency to form an enterocoel decrease. The hydrocoel lingers on longest as an enterocoel, but in the heavily yolked types it, too, arises by splitting, and is thus a schizocoel.

As noted in my paper on Amphipholis, the structures termed for convenience “enterocoels”—as they are clearly homologous with the enterocoels of Ophiothrix—are in fact formed by a splitting in small, solid masses of tissue formed on either side of the archenteron. Thus we have a series leading from the enteric pouches with preformed internal cavity, through enteric tissue-masses which later acquire an internal cavity, to finally the forms where the whole definitive coelom arises by an extensive process of splitting in a mesenchyme mass.