
452 THE AMERICAN NATURALIST. '[Vor. XXXVI. 
cells are found to be more tightly packed, the fine yolk in 
the cell bodies is more abundant, and the nuclei are smaller. 
Hence, to demonstrate cell outlines in the largest (mature) 
gemmules! very thin sections are necessary. (3) Mature 
gemmules are found in which the compact mass of cells is 
broken up into a few large masses, on the surface of which 
are semi-detached single cells, and between and round which 
lie scattered a few similar free cells. The shape and position 
of the partially free cells indicate that they are separating as 
individuals from the subsidiary masses into which the gemmule 
has broken. Gemmules in this stage are larger than in the 
mature undivided condition, the spaces between and round the 
subsidiary masses presumably being filled with absorbed fluid. 
(4) In other gemmules the subsidiary masses (always irregular 
in shape and of varying size) are numerous and small, and the 
free cells are abundant. (5) In yet another stage the constitu- 
ent cells of the gemmule are found to be completely separated 
from one another, save that small multinucleate masses continue 
to be present. The cells of the outermost layer are flattened 
and form an investing membrane for the inner mass, the cells 
and multinucleate masses of which are connected together by 
delicate processes? (6) Transitional stages show that the outer 
layer of cells becomes the surface epithelium of the larva, which 
is ciliated except at the posterior pole, while the inner mass 
becomes histologically differentiated into the several kinds of. 
cells forming the parenchyma of the larva. The larva itself is 
a typical monaxonid larva, essentially identical in structure with 
those that are known to develop from segmenting eggs. 
Thus an unbroken series of stages is found leading from a 
small mass composed of a few parenchymal cells to a ciliated 
larva. The various stages in this gemmule development, 
including the resulting larvze, are scattered in great abundance 
throughout the sponge body, and their presence is associated 
with degeneration in the structure of the parent sponge. 
Nowhere in this continuous series of stages is an egg, either 
undivided or segmenting, interpolated. Very small ova, many 
! Journ. of Morph., vol. ix, No. 3, Pl. XV, Fig. 18. 
* Ibid., Pl. XVI, Fig. 21. 3 Ibid., Pl. XVI, Fig. 22. 
