The further history of the larva has been investigated by Metschnikoff. He has found that the interior of the vesicle becomes gradually filled with mesoblast cells of a peculiar type, called by him rosette-cells, which are probably derived from the walls of the vesicle. When the metamorphosis commences, the larva assumes a flattened form, and cells of a new type, viz. normal amoeboid cells, grow in amongst the rosette cells. The new cells are also derived from the epiblast. The larvæ appear to fix themselves by the hinder extremity. The cilia gradually disappear, and the epiblast cells flatten out and form a kind of cuticle. For some time the larva remains in the two-layered condition, but gradually canals (? ciliated chambers) lined by hypoblast cells become formed. They appear as closed spaces with walls of ciliated cells derived from the amoeboid cells, and the different parts of the system of chambers are established independently. In _H. pontica_ the ciliated chambers are formed before the attachment of the larva. The development was not followed up to the formation of the pores placing the canal system in communication with the exterior. The young sponges at a somewhat later stage have been studied by Schulze and Barrois. They are formed of an external layer of flattened cells, not clearly ciliated as in the adult, within which are a normal mesoblastic tissue, and several spherical chambers lined by ciliated cells exactly like the ciliated chambers of the full-grown sponge. Irregular invaginations of the epiblast give to the young sponge a honeycombed structure. The ciliated chambers in the youngest condition of the sponge are closed; but in slightly older examples they come into communication with the passages lined by hypoblast, and so indirectly with the external medium.