390 
DUNCAN S. JOHNSON 
nuclei, without being cut off from the others by a cell wall, might be 
fertilized and so give rise to an embryo which should be nourished 
directly by the surrounding tissue of the parent diploid generation. 
If this happened we should have realized in the seed plants the extreme 
reduction of the diploid generation which is found in animals, and per- 
haps in Fucus, but which is only approximated by the Orchidaceae and 
Podostemaceae. 
7. The relative primitiveness of the 8-nucleate and 16-nucleate em- 
hryo-sacs. — The view that the embryo-sacs with 16 nuclei are primi- 
tive, in that they retain more of the numerous nuclear divisions inter- 
vening between megaspore and ripe egg, which are characteristic of the 
gymnosperms and heterosporous pteridophytes, is one first advanced 
by Campbell (i 900-1901). His conclusion was based on the assump- 
tion that all of the 16 nuclei present in the mature embryo sac result 
from the germination of a single megaspore. The recent work on 
several species of Peperomia (Johnson, 1907, Brown, 1908, Fisher, 
1914), on the Penaeaceae (Stephens, 1909a), on Gunnera (Ernst, 1908, 
Samuels, 1912) and the present study show, as Coulter (1908, pp. 
363-364) has suggested, that Campbell's assumption is clearly incor- 
rect. The individual megaspore in each of these genera gives rise not 
to more prothallial nuclei than in other angiosperms but to just half as 
many, four instead of eight. It is clear that this can hardly be re- 
garded as a very primitive mode of megaspore-germination. It is 
rather a more reduced type than the one usually found in angiosperms. 
From this point of view then Campbell's contention that the 16-nucle- 
ate embryo sac is primitive is clearly untenable. From this point of 
view also it is evident that, since the embryo sac of Peperomia his- 
pidula is shown to be the product of division of 4 individual megaspore 
nuclei, in a common cell cavity, the number of nuclear divisions be- 
tween the definitive archesporial cell and the mature embryo sac is not 
a matter of prime significance. Since then the reduction division and 
megaspore formation in P. hispidula goes on entirely in the normal way 
except for the arrangement of the spores in a tetrad, we cannot agree 
with Ernst (1908, p. 29), that the process of tetrad division here is re- 
duded and that the germination of the megaspore involves one more 
division than usual and that this sac is thereby ''distinguished as an 
older, or at least as an independent, form of angiospermous embryo sac.^' 
It would still, of course, be possible to regard the peculiar type of 
megaspore-germination found in Lilium, Peperomia, Gunnera, etc., as 
