240 Transactions of the Royal Society of South Africa. 
there maj be a tendency for it to take place before (as in G. Rum^phianum) 
or after (as in G. Gnemon) this has commenced. It cannot occur before the 
end of the last free nuclear division in the sac nor after the completion of the 
endosperm, because it is only between these two limits that female gametes 
are available.* It is probable that in every species hitherto investigated 
the conditions of the sac itself are favourable to fertilisation at any time 
between these two limits. 
Certainly this is the case in G. africanum, if we may assume that 
fertilisation immediately follows the introduction of the male gametes into 
the sac. In this species the pollen-tube reaches the sac before septation 
commences — i.e. while all the sac-nuclei are free — or at any stage in the 
formation of endosperm-tissue ; and we have preparations in which the 
whole sac is filled with endosperm while the tips of the longest pollen-tubes 
are at various depths between the bottom of the pollen-chamber and the top 
of the sac, or, in some cases, the ovule is not pollinated. 
When the life-histories of these and other species are more fully known 
it will probably be found that in all of them the formation of endosperm 
beginning at the chalazal end of the sac proceeds gradually and without any 
marked interruption upwards until all the available and accessible f free 
nuclei have been utilised in fusion, and that fertilisation may occur at any 
time so long as there are free gametes in the sac. Griven this condition, the 
time at which fertilisation is effected is probably determined by the accidents 
of pollination and the growth of the pollen-tube. The relations of the 
gametes to the endosperm are not the same in G-netum as in Ephedra and 
the lower Grymnosperms. In the latter the endosperm, a prothallus, produces 
the gametes, and therefore necessarily precedes them in ontogeny. The 
constitution of the gametes is the essential condition which must be realised 
before fertilisation can take place. This implies a pre-existent endosperm 
(prothallus) in the lower gymnosperms, but not in G-netum. In G. africanum^ 
and probably in the other species as well, fertilisation appears to be quite 
independentof endosperm -formation except thatwhenthe latterprocess is once 
completed fertilisation is rendered impossible because the supply of female 
gametes is exhausted. These two processes are equally independent of one 
another in Welwitschia, but their independence is obscured by the extreme 
form of siphonogamy achieved in this genus ; septation, the first step in 
the constitution of the endosperm, being also a necessary antecedent to the 
formation of the embryo-sac-tubes, and therefore to fertilisation. 
The account already given % of the septation of the sac and conversion 
of the multinucleate compartments into uninucleate cells by nuclear fusion 
is fully confirmed for G. africanum, and the same processes have now been 
* But see p. 243. 
t See p. 243. 
X Pearson, 1915 a. 
