168. BOTANICAL GAZETTE [MARCH 
initial of Tumboa. Inasmuch as the archegonium initial of Torreya 
often occupies this beak, the suggestion becomes still more pertinent 
(fig. 21). 
The formation of walls in the endosperm was not observed before 
July 1, and in several instances they did not appear for a month later. 
When wall-formation began, repeated countings showed 256 free 
nuclei, which seems to be a very common limit of free nuclear division 
among gymnosperms. 
Since fertilization was observed August 12, it is evident that the 
archegonium is developed very early in the history of the gameto- 
phyte. In fact, as soon as the very small sac is filled with extremely 
delicate tissue the archegonium initial becomes evident. It does not 
seem possible for archegonia to appear any earlier, for the initials 
are organized as soon as the free nuclear stage has passed. In 
Torreya, therefore, nearly all of the endosperm, which becomes an 
extensive tissue, develops after fertilization. 
The single archegonium initial is always to one side of the central 
axis (fig. 16), often occupying the “beak” referred to above, and so 
projecting above the endosperm (jig. 2z). A neck cell is cut off 
and divides anticlinally, forming a two-celled neck (fig. 22), the 
usual limit of neck-formation among gymnosperms. In fact, it is 
only among Podocarpeae and Abieteae that a more extensive neck 
is usually formed, consisting of more than one tier of cells, unless the 
somewhat anomalous neck of Ephedra be included. The fact that 
there is variation in the number of neck cells in the same form (two to 
twenty-five in Podocarpus), and that as a rule necks are destroyed 
as soon as formed by the growth of the central cell and pollen tube 
(fig. 23), suggests that their extent depends somewhat upon the 
approach of the pollen tube, which usually checks neck formation 
early, but sometimes permits it to become more extensive. In T. 
californica Miss ROBERTSON (g) has found that the archegonia are 
usually three in number, ranging from two to five, and that the necks 
consist of four or six cells. 
The central cell enlarges rapidly, no jacket-layer being evident 
until after fertilization, and even then it is weakly organized (fig- 25). 
The nucleus is spherical and lies near but not against the neck cells, 
more nearly resembling an egg nucleus than in any gymnosper™ 
