242 
Transactions of the Royal Society of South Africa. 
nuclei of the fertile (i. e. micropylar) part of the embryo-sac surround 
themselves with a denser protoplasm and a membrane and thus form cells. 
I will call these retarded prothallium cells of the fertile part. They may 
form regularly in the neighbourhood of the young zygotes,* or they may be 
found rather irregularly distributed,! or they maybe limited to a very small 
number, frequently but one or two.j After these changes have taken 
place the role of the fertile part of the embryo-sac has been played, nothing 
more occurs in it, it is gradually obliterated owing to the growth of the 
sterile part ; its retarded prothallium cells play no role whatever." It has 
been clear for a long time that " this account may require some amend- 
ment," § but it is only lately that we have been able to obtain convincing 
evidence. 
Lotsy's " retarded prothallium cells " are multinucleate endosperm cells 
of precisely the same kind and origin as those which are formed in the 
process of endosperm-formation in the chalazal region. The nuclei they 
contain are, as in the former case, unfertilised sac-nuclei. When there 
is little or no disturbance caused by the entrance of the pollen-tubes or the 
activity of the oospores, septation and nuclear fusion proceeds without any 
sign of interruption to the very summit of the sac. Two of a series yielded 
by our preparations are shown in Plate XL VIII, fig. 21 b ((?. Gnemon), and 
Plate XLYIII, fig. 22 a, b {G. africamim). In the former (Plate XLYIII, 
fig. 21 b) a small part of the micropylar end is still unseptate ; the endo- 
sperm below this is clearly of uniform origin (c/. Plate XLYIII, fig. 21 a). 
In the latter (Plate XLYIII, fig. 22) septation has proceeded to the 
summit of the sac ; three of the uppermost compartments contain groups 
of free nuclei whose constituents are in process of fusion (Plate XLYIII, 
fig. 22 b). In Plate XLIX, fig. 23, which shows at least two oospores, 
numerous groups of free sac-nuclei are seen, though the preparation does not 
yield proof that cell- walls enclosing them have yet appeared. Plate XLIX, 
fig. 24, represents an interesting and not uncommon case in which the 
septation of the micropylar region is much less regular. This is almost 
certainly due to the influence of oospores, of which one (o) appears in the 
section. Septation at a particular point may be prevented, or multinucleate 
compartments may be destroyed, just as uninucleate endosperm cells are 
destroyed, by the oospores and proembryos. It is quite possible that a 
multinucleate compartment may sometimes suffer from the activities of a 
uninucleate endosperm cell in its neighbourhood. But there can be no doubt 
that compartments which do not succumb to the growth of the proembryo, 
or perhaps occasionally to the greater vigour of more advanced endosperm 
^ Loc. ext., figs. 44, 45. 
t Loc. cit., fig-. 60. 
I Loc. cit., figs. 53, 55, 58, 
§ Pearson, 1915 a, p. 322. 
