THE COLLAR-CELLS OF HETEROCGLA. 27 
was observed in a sponge (S. compressum) which had been 
a month in the circulation of the aquarium, with other 
sponges, &c., allowed to decay in the dish containing it. The 
living section at first sight seemed to be full of embryos; 
these proved, however, to be the remnants of the flagellated 
chambers, some parts still exhibiting perfectly normal collared 
cells with active flagella and cylindrical separated collars; the 
space between the “ Leucon”-like chambers being largely 
filled with parenchym. Paraffin sections showed many wide 
canals, resembling the normal afferent system. Only a few of 
the collar-cells are elongated, and the recognisable collar-cells in 
general are comparatively fewin number; in some places they line 
only part of achamber; in some places the chambers are shorter 
or narrower than in the normal sponge ; in some places they form 
small closed chambers, or pseudo-blastule, consisting of as fewas 
a dozen cells, lying in a plentiful gelatinous parenchym, into 
which appearances suggest that their fellows have migrated. 
The condition appears identical with that recognised as 
common in winter for Spongilla (Lieberkiihn, Metschnikoff, 
Weltner). It becomes a question whether we are not to 
ascribe the metamorphosis of Halisarca as described by 
Metschnikoff (6), and that of S. compressum described by 
Masterman (23), to conditions unfavourable to general vitality, 
rather than to the inception of nutritious sive innutritious 
particles. 
Nutrition. 
Vosmaer and Pekelharing (19) find carmine and milk, after 
one hour’s feeding, in the choanocytes and in the lumen of the 
chamber, especially frequently in the collars themselves. After 
a longer time the particles are chiefly in the cell-bodies, rarely 
free or in the collars; after a still longer time they are found 
in the parenchyme. 
Masterman (28) recently published an account of nutrition 
in S. compressum in which he describes an extraordinarily 
rapid cycle of events. It has been suggested above that he 
may have been deluded by pathological metamorphoses uncon- 
