In tropical countries gemmulation occurs as a defence against the ravages caused by the dry season when the waters recede down their banks, exposing all or most of their sponge inhabitants to the direct rays of the sun. The sponges are at once killed, but the contained gemmules being thoroughly dried, become efficient distributing agents of the species; they are light enough to be carried on the wind. It is probable that those individual sponges which escape desiccation survive the dry season without forming gemmules. It has been shown experimentally that gemmules are not injured by drying—Zykoff found that gemmules kept dry for a period of two years had not lost the power of germination. {179}The mature gemmules consist of a more or less spherical mass of cells, which we shall refer to as yolk cells, and of a complex coat. The latter is provided with a pore or pore tube (Fig. 74) which is closed in winter by an organic membrane. There are three layers in the coat: an inner chitinous layer surrounded by an air-chamber layer, which is finely vesicular, showing a structure recalling plant tissue, and containing amphidiscs arranged along radii passing through the centre of the gemmule. One of the discs of each amphidisc lies in the inner chitinous coat, while the other lies in a similar membrane which envelopes the air-chamber layer and is termed the outer chitinous coat. Marshall has suggested that one function of the amphidiscs is to weight the gemmules and thus protect them against the force of the river current; and no doubt the sinking or floating of individual gemmules depends on the relative degree of development of the air-chambers and of the amphidiscs. A study of the development of _Ephydatia_ gemmules vividly illustrates various characters of the inner processes of sponges. Specially noteworthy are the migrations of cells and the slight extent to which division of labour is carried: one and the same cell will be found to perform various functions. [Illustration: FIG. 74.—Part of a longitudinal section of a gemmule of _Ephydatia_ sp. passing through the pore (_a_). (After Potts.)]