1916] FORSAITH—ON AGRACEAE 473 
swell to their original size. The pollen thus treated was examined 
microscopically, and the relative amount of abortive grains (those 
which were shrunken, small, or empty) was determined by a more 
or less careful count of several fields of view. The difference 
between the fertile and the sterile microspores is perfectly distinct. 
The fertile ones appear as fully developed uniform spores filled 
with protoplasm; while those which have become degenerate are 
small, shrunken, and empty. Those buds, however, which showed 
less than 5 per cent of infertile pollen were considered as repre- 
sentative of genetically pure descent. This figure was chosen by 
Dorsey (9) after a careful investigation of Vitis. In this genus 
species of unquestionable ancestry were shown by a careful count 
never to contain more than this ratio as a result of physiological 
conditions. 
From this general discussion of the relation of pollen sterility 
and variation in respect to hybrid formation, attention may be 
turned profitably to a more detailed discussion of the Chamaenerion 
group of Epilobium. It is of interest to note that DeVries (8) 
has stated that the pollen in Epilobium is “wholly fertile,” but the 
writer’s observations have led to different conclusions in regard 
to microspore conditions in this genus. Epilobium angustifolium 
L., with few exceptions, presents good pollen grains in the anthers 
chosen from plants which were collected in the more southerly part 
of its range. This condition is significant, since this region repre- 
sents that part of its zone where it is outside the influence of E. 
latifolium L. This uniformly pure condition of the male generative 
cells is, without doubt, due to the fact that in the more southern 
Stations E. angustifolium is practically a monotypic species, owing 
to the absence of an ally with which it is able to cross. On the 
other hand, where the “great willow herb” grows within the range 
of E. latifolium, the pollen shows a tendency to abortion in a vary- 
ing degree. 
A still more comprehensive understanding of the apparent 
relation of pollen sterility to geographical distribution can be 
obtained by reference to figs. 1-8, text fig. 1, and list I. Fig. 1 
Shows the pollen of E. angustifolium L. as it appears in plants from 
Auburn, New Hampshire. It will be observed that all the grains 
