100 BOTANICAL GAZETTE | FEBRUARY 
because they interfered with the rapid penetration of the fixing 
fluid, but more especially because their brittleness very quickly 
spoiled the edge of the microtome knife. The older akenes, con- 
taining embryo-sacs or embryos, were picked off separately, and 
cut one at a time. 
For fixing the early stages of development, as well as the 
mature embryo-sac structures, the best results were obtained with 
a I per cent. solution each of chromic acid and acetic acid. A 
weaker solution, containing 0.7 per cent. chromic acid and 0.3 per 
cent. acetic acid, worked well, especially when the specimens were 
first dipped for a moment in alcohol. Flemming’s weaker solu- 
tion was tried for killing embryo-sac stages, but was abandoned 
as being unnecessary. A saturated solution of picric acid in 70 
per cent. alcohol, with a small addition of acetic acid, was fre- 
quently used, and was found to be by far the best for all stages 
of embryos, as it penetrated the hard ovary wall more quickly. 
After dehydration the objects were brought through xylol into 
paraffin, and cut in serial sections. 
Delafield’s haematoxylin, either alone or followed by ery- 
throsin, was found to be the best stain for floral development 
stages and for embryos. For embryo-sacs, safranin-gentian- 
violet, cyanin-erythrosin, and Heidenhain’s iron-alum-haema- 
toxylin were used ; also a combination of fuchsin and iodin-green. 
For the nuclei within the pollen grains, safranin-gentian- 
violet, and especially the iron-alum-haematoxylin, gave good 
results. 
The following five species of Silphium, growing near Chi- 
ia were examined, a more or less complete series in each 
being made : S. integrifolium Michx., S. trifoliatum L., S. terebinthi- 
naceum L., S. laciniatum L., and S. perfoliatum L. Most of the 
statements apply to all these species, hence no attempt will be 
thade to separate them in the account, excepting in cases where 
points of specific difference are under discussion. 
My acknowledgments are due to Professor John M. Coulter 
and to Dr. Charles J. Chamberlain for great assistance rendered 
during the course of the work. 
