Roots of Some Rosaceous Genera. 63 
beautiful golden pigment, while the cell walls were not 
pigmented. I have not studied sections of the root of this 
species during different times of the year. In some species a 
difference can be traced in the amount of pigment according to 
the season of the year in which the sections are studied. As 
examples of this fact I would quote the conditions observed in 
Agrimonia parvifiora and Spirea Filipendula. In the former 
species, sections taken from roots collected in September 
showed very little pigment in the cell walls or cavities of the 
uniseriate layers, sections from those collected in December 
contained no pigment, while sections from roots collected in 
July had pigment alike in the cell walls and in the cell cavities 
of this layer. In the latter species, sections taken from roots 
collected in December and April showed cells almost lacking 
pigment in the cell walls and cell cavities, while sections taken 
from roots collected in July showed pigment in the cell walls 
of all the cells, as well as pigment in the cell cavities of the 
uniseriate layers. While I have attempted to determine the 
chemical nature of this pigment, I have not as yet obtained 
sufficiently definite results to draw conclusions. 
(f) Comparison of Root and Shoot.—Not very many obser- 
vations have been made along this line, but I will note the 
following: Intercellular spaces have been observed in the cork 
region of the rhizome of Geum album, Rosa, Alchemilla 
pyrenaica and Rubus villosus stem. In stems of Pyrus Malus 
and P. communis there are no intercellular spaces. In general 
it may be said that in those species studied, the cork region of 
the stem was more strongly suberized than that of the root as 
was shown by the use of the cork tests. 
(g) Evidence as te the Relative Evolutionary Position of the 
Rosaceous Groups.—In treating the Rosacee as an order Dr. 
Macfarlane informs me that he regards the only workable 
scheme of evolutionary relationship in this order to be one that 
follows closely the following lines: Starting with the Poten- 
allege as the most primitive group, we pass by tolerably direct 
