CELL MEASUREMENT 195 
Each epicotyl was divided into lo equal segments, and in each 
segment loo primary epidermal cells, taken at random, were measured. 
Thus for each epicotyl i,ooo primary cells were measured, uniformly 
distributed along its length. Many of these primary cells were divided 
by transverse walls into secondary cells, which were also measured, 
bringing the number of measurements up to 2 or 3 thousand for each 
epicotyl. Kraus had observed in his work that the ends of many 
epidermal cells were not even approximately perpendicular to the 
longitudinal axis, but were pointed. He wrote: "It is a common 
phenomenon that the cells, at any rate those of the epidermis, do not 
remain parenchymatous during elongation in etiolation, but like the 
elongating wood-forming cambial cells, become completely prosenchy- 
matous, the long, sharp ends dove-tailing in between one another." 
This condition occurs in the normal as well as in the etiolated epider- 
mis, but a further fact is that many cells do have perpendicular end 
walls, and appear, as we have already indicated, to have been formed 
at a late stage in the development of the internode. With regard to 
the latter point, MacDougal says of epidermal cells of etiolated stems 
in general: ''The epidermis, in common with many other tissues, does 
not advance beyond a certain primary stage of development, and retains 
the power of growth and division in the cells during a much longer 
period than in the normal plant ; consequently it can respond to stresses 
and other factors, which may cause it to undergo increase in size, 
alterations in form, or multiplication of the cells by division." In the 
epidermis of Phaseolus multiflorus it proved to be a simple matter to 
distinguish by the character of the end walls between the primary cells, 
formed by division of the primary meristem, and secondary cells, 
formed by subsequent divisions. The prosenchymatous shape of 
the former is obvious, even after they have divided into secondary 
cells. The latter have at least one transverse end wall, as in any 
typical parenchymatous cell. (See figs. I and 2, reduced to scale 
from camera lucida drawings.) Measurements were made of cells 
taken at random, but were so recorded so as to distinguish undivided 
primary from secondary cells. All the secondary cells derived from a 
single primary were measured, and their lengths added to obtain the 
total length of the epidermis derived from it. 
It is obvious that in any discussion of cell length as a factor in 
internode length, one must distinguish carefully between primary 
and secondary cells. Our tables give statistical data (i) for primary 
