The beginning of a gemmule is first recognisable[207] as a small cluster of amoeboid archaeocytes in the dermal membrane. These move into the deeper parts of the sponge to form larger groups. They are the essential part of the gemmule, the yolk cells, which, when germination takes place, give rise to a new sponge. They are followed by two distinct troops of actively moving cells. Those forming the first troop arrange themselves round the yolk cells and ultimately assume a columnar form so that they make an epithelioid layer. They then secrete the inner chitinous coat. The cells of the second troop are entrusted with the nutrition of the gemmule. Consequently they pass in among the yolk cells, distribute their food supplies, and make their escape {180}by returning into the tissues of the mother sponge, before the columnar cells have completed the chitinous coat. Yet another migration now occurs, the cells—"scleroblasts"—which have been occupied in secreting amphidiscs at various stations in the sponge, carry the fully formed spicules to the gemmules and place them radially round the yolk cells between the radially lying cells of the columnar layer. The scleroblasts themselves remain with the amphidiscs, and becoming modified, contribute to the formation of the air-chamber layer. The columnar cells now creep out between the amphidiscs till their inner ends rest on the outer ends of these spicules. They then secrete the outer chitinous coat and return to the mother sponge. Carter gives directions[208] for obtaining young sponges from the gemmules. The latter should be removed from the parent, cleaned by rolling in a handkerchief, and then placed in water in a watch-glass, protected with a glass cover and exposed to sunlight. In a few days the contents of the gemmule issue from the foramen and can be seen as a white speck. A few hours later the young sponge is already active and may be watched producing aqueous currents. At this age the sponge is an excellent object for studying in the living condition: being both small and transparent it affords us an opportunity of watching the movements of particles of carmine as they are carried by the current through the chambers. Potts[209] describes how he has followed the transportal of spicules by dermal cells, the end of each spicule multiplying the motion, swaying like an oscillating rod. In _E. fluviatilis_ reproduction also occurs during the warmer months in this climate by means of sexual larvae. These are interesting for certain aberrant features in their metamorphosis.[210] While some of the flagellated chambers are formed in the normal way from the flagellated cells of the larva, others arise each by division of a single archaeocyte. This, it is suggested, is correlated with the acquisition of the method of reproduction by gemmules, the peculiarities (_i.e._ development of organs from archaeocytes) of which are appearing in the larvae.