of the parent; (b) in any case, there is no mouth opening, nor oesophageal passage (save in the inaccurate diagrams of Metschnikoff and Russo himself); (c) at this early stage there are no muscle fibres developed about the oesophagus to perform such contractions; and (d) no other observer has seen the phenomena. It is therefore necessary to reject this bizarre notion of a cannibal embryo.

In regard to the actual mode of nutrition of the embryo, it can readily be seen that the embryonic attachment cannot be essential, for two chief reasons. Firstly, it disappears soon after the metamorphosis of the embryo, whereas the greater part of the growth in size takes place after the atrophy of the attachment. Secondly, there never occur any traces of vascular organs or sinuses in the structure. The embryo lies closely invested by the wall of the bursa. Now, it will be remembered that during the course of pregnancy numerous sinuses appear in the wall of the bursa. These are absent from it at other times. This significant fact suggests that the wall of the bursa may itself be a nutritive organ, by supplying a secretion which is poured into the lumen and there absorbed directly by the tissues of the developing embryo. However, there still remains the possibility that the embryo, like that of Kirk's ophiuroid, may develop solely upon the food provided by its own yolk material—though the yolk in Amphipholis is much smaller in amount than that of Kirk's ophiuroid.

In order to test whether or not the yolk is sufficient for development to continue, experimental culture of excised embryos was carried out. As recorded (Fell, 1940b) it was proved that the embryo is unable to develop for more than five days in vitro unless certain substances (contained in Erdschreiber culture medium) be added to the culture fluid. In the light of this evidence it is reasonable to assume that secretions are in fact poured from the sinuses in the wall of a pregnant bursa, since experimental investigation confirmed the results of previous anatomical study.

It follows that the function of the embryonic attachment can be no more than that of an anchoring organ while the embryo is still minute. With increase in size the embryo no longer requires such an attachment, for it is safely held within the bursa by contact with its sides. The attachment then atrophies. It is for this reason that the term “umbilical cord” has been abandoned, and the name “embryonic attachment” used for the organ throughout this paper.

The appearance of the sinuses in the wall of a pregnant bursa invites comparison with the analogous changes in the vascular supply of the uterus of viviparous vertebrates. In vertebrates the changes in the uterus or oviduct are known to be brought about by the liberation of a hormone. The question arises whether the same holds good for Amphipholis squamata. It can be answered quite safely, I think, that hormones cannot be responsible—at least, not in the way that they are in vertebrates. For, whereas in vertebrates if one uterus be pregnant and not the other, both uteri respond to the hormone, in Amphipholis, on the other hand, it is only the pregnant bursae which become enlarged and develop sinuses, the others remaining unchanged. Therefore the anatomical and physiological changes