556 Trow. — On Fertilization in the Safrolegnieae. 
degeneration. At one end of such a series the nuclei would with difficulty 
be recognized as degenerate at all — at the other end it would in most cases 
be impossible to feel sure that they really were nuclei. Notwithstanding 
much study it has proved impossible to give any definite account of the 
various stages in degeneration which the nuclei pass through. Degenera- 
tion may apparently commence either when the nuclei are in the resting 
condition, or when in the prophases or even the metaphases of mitosis. 
The nuclear membrane seems to disappear, the surrounding protoplasm to 
suffer a change which causes it to take up stains more readily, and the 
central mass — the doubtful nucleolus — to break up into smaller portions or 
remain intact, according to circumstances (Fig. i8, m, n, o,p). A blurred 
astrosphere-like arrangement of rather thick but indefinite protoplasmic 
filaments is sometimes associated with these degenerate nuclei, as seen in 
Fig. 1 8, q , r , I would suggest as a provisional explanation of this 
rather curious feature that they are the daughter-nuclei of the second 
mitosis undergoing degeneration along with their associated astrospheres. 
This observation is difficult to reconcile with Davis’s contention that his 
‘ coenocentra’ select and nourish the functional nuclei. 
In the last phases of degeneration, after the supernumerary nuclei have 
entirely vanished as such, large deeply-stained granules are to be found in 
the ooplasm, as shown in Fig. 19. These may best be regarded as certain 
by-products of the degeneration process. They correspond somewhat to 
the similar structures figured by Davis, but are much fewer and larger. 
Development of the astrospheres and ovocentra. Before all the super- 
fluous nuclei are quite destroyed, the process of the formation of the 
oospheres (balling) begins. The astrospheres grow rapidly in size. The 
protoplasm associated with them loses its larger vacuoles and becomes fine 
grained. The nucleus remains in close contact with the centrosome — 
Fig. 19, a — and its chromatin seems to concentrate itself in the immediate 
neighbourhood of that body. As Davis has already pointed out, it would 
be very easy to mistake this dual structure for a single one, even in fine 
preparations. The centrosome absorbs nuclear stains with the same facility 
as the chromatin of the nucleus. In good preparations the centrosome, 
like the nucleus itself, is readily demonstrable by its structure. It always 
appears, when well developed, as a single granule, from which there radi- 
ates outwards in all directions a relatively small number of protoplasmic 
filaments. These filaments are traceable right up to the centrosome and 
are continuous with it. Centrosome and astrosphere correspond very 
closely with those in Pellia , figured and described by Farmer and Reeves 
(’ 94 ). The nucleus can be recognized with certainty even when it lies im- 
mediately under the centrosome. Nuclei occupying this position have not 
been inserted in the drawings for obvious reasons. The nuclei are not 
shown consequently in Fig. 19, b, c } d. In one case — Fig. 19, d — I 
