The cells in the inner mass are classified into (1) undifferentiated cells, recognised by their nucleus, which possesses a nucleolus; these are the archaeocytes; (2) differentiated cells, of which the nucleus contains a chromatin net; these give rise to pinacocytes, collencytes, and scleroblasts. Some of them form a flat epithelium, which covers the hinder pole. Some of the scleroblasts already contain spicules. Fixation occurs very early. The front pole is used for attachment, the pigmented pole becoming the distal end (Fig. 69). The larva flattens out, the margin of the attached end is produced into radiating pseudopodial processes. The flagellated cells retreat to the interior, leaving the inner mass exposed, and some of its cells thereupon form a flat outer epithelium. This is the most important process of the metamorphosis; it is followed by a pause in the outward changes, coinciding in time with rearrangements of the internal cells to give rise to the canal system; that is to say, lacunae arise in the inner mass, pinacocytes pass to the surface of the lacunae, and form their lining; the flagellated cells, which have lain in confusion, become grouped in small clusters. These become flagellated chambers, communications are established between the various portions of the canal system, and its external apertures arise. There is at first only one osculum. The larvae may be obtained by keeping the parent sponge in a dish of sea water, shielded from too bright a light, and surrounded by a second dish of water to keep the temperature constant. They will undergo {174}metamorphosis in sea water which is constantly changed, and will live for some days. We have said that the young sponge has only one osculum. This is the only organ which is present in unit number, and it is natural to ask whether perhaps the osculum may not be taken as a mark of the individual; whether the fistular specimens, for example, of _H. panicea_ may not be solitary individuals, and the cockscomb and other forms colonies in which the individuals are merged to different degrees. Into the metaphysics of such a view we cannot enter here. We must be content to refer to the views of Huxley and of Spencer on Individuality. But it is advisable to avoid speaking of a multi-osculate sponge as a colony of many individuals, even in the sense in which it is usual to speak of a colony of polyps as formed of individuals. The repetition of oscula is probably to be regarded as an example of the phenomenon of repetition of parts, the almost universal occurrence of which has been emphasised by Bateson.[200] Delage[201] has shown that when two sponge larvae fixed side by side fuse together, the resulting product has but one osculum. This, though seeming to bear out our point of view, loses weight in this connexion, when it is recalled that two Echinoderm larvae fused together give rise in a later stage to but one individual.