The archesporium of the species of Peperomia studied arises as a single hypodermal cell, which cuts off a single parietal cell and then forms the embryo sac directly. The first division of the embryo sac nucleus is heterotypic. The nucleus goes into synapsis. This stage is followed by an apparently continuous spirem. This splits longitudinally but later the two halves come together again. The chromosomes are formed from loops in the spirem. When these divide, they seem to separate two parts of the spirem which were originally placed end to end. The second division may divide the chromosomes along the longitudinal split seen in the prophase of the first division. In the third division of the embryo sac nucleus of P. Sintensii, no cell plates were seen on the spindles, but in the last division cell walls are formed on all the spindles. These walls cut off, against the embryo sac walls, one of each of the eight pairs of nuclei, and leave the other eight free in the cytoplasm. These free nuclei fuse to form the endosperm nucleus. The egg and a nucleus with the position of a synergid are cousins. The other six nuclei which are cut off against the embryo sac wall finally degenerate. The mature sac contains sixteen-nuclei, which are apparently derived from four megaspores. That the first four nuclei of the embryo sac are megaspore nuclei is indicated by the fact that the first division of the embryo sac mother cell nucleus is heterotypic and reducing, and that in P. pellucida cell plates are formed on the spindles of the first two divisions, while in P. Sintensii and P. arifolia these plates grow into evanescent walls which extend across the embryo sac and separate the nuclei. We are not justified, however, in extending the conception of four megaspores in an embryo sac to all angiosperms in which a row of megaspores is not formed, because we do not know that the division of the mother cell to megaspores may not be omitted and the place of the heterotypic division be changed. In the fertilization of P. Sintensii some cytoplasm appears to be taken into the nucleus. Note.-Since the above was written, there have appeared two papers dealing with sixteen-nucleate embryo sacs. One is by Coulter, in which he expands a suggestion offered by Lloyd in 1902, that when four megaspores are not formed the first four nuclei of the sac are spore nuclei so far as development is concerned, and says that the formation of megaspore nuclei cannot be omitted. In a paper on the phylogeny of the angiosperm embryo sac, Ernst describes a sixteen-nucleate embryo sac in Gunnera. Here he finds an egg, two synergids, six antipodals often in two groups of three, and seven nuclei which fuse to form the endosperm nucleus. He then attempts to fit the archegonial theory of Porsch to Gunnera, and concludes that the egg group represents an archegonium, while two more are represented by the six antipodals together with two of the nuclei which fuse to form the endosperm. He thinks that the other four nuclei fail to form an archegonium, and that the explanation which he gives of the embryo sac of Gunnera may apply to Peperomia pellucida. In the case of the embryo sac of P. Sintensii, which resembles very closely that of P. pellucida, it has already been shown that if we should apply the theory of Porsch there would be four and not three archegonia. Ernst has not worked out the relation between the nuclei in the embryo sac of Gunnera, and therefore it would seem premature to speculate as to the conditions there, but there seems to me to be no sufficient reason for thinking that the nuclei would represent three rather than four of the archegonia of Porsch.