In this study, the reproduction of the mitotic centers in the eggs of a sea urchin, Strongylocentrotus purpuratus and a sand dollar Dendraster excentricus has been studied by means of experimental designs that do not depend on the actual visualization of centrioles. The centers are defined in operational terms as potential poles. Blockage of mitosis by mercaptoethanol, it was found, inhibits the duplication of the centers, but does not inhibit the splitting and separation of centers that have already duplicated and thus potential poles could be realized as actual poles in multipolar divisions. At all times, the center is at least a duplex structure; that is, it contains two potential poles. The actual duplication process is the earliest event in a given mitotic cycle, taking place at very early interphase or in late telophase of the previous division. The splitting of the centers following duplication is a distinct process, dissociable from the duplication as such. Duplication and splitting normally occur at about the same time in the mitotic cycle, with a precession of the former. That is, as the two members of a pair of "old" centers split, each one gives rise to a new one, which remains associated with it until the next phase of splitting and duplication occurs. The results are consistent with what is termed a "generative" model of the self-reproduction of an intracellular body. According to this, the body does not immediately produce a full-fledged copy of itself, with simultaneous fission, but the primary duplication event involves only a part of the parent structure. This gives rise to a "germ" or "seed" which then grows to be equivalent to the parent body, and finally splits from it.