204 WILBER BROTHERTON, JR., AND H. H. BARTLETT 
tion of all etiolated internodes grown under like conditions. Although 
it is unlikely that this assumption is wrong, it requires proof ; we have 
not yet undertaken the labor of making enough measurements to 
place it beyond criticism. It must be admitted, however, that relative 
position in the frequency distribution affords a logical basis for the 
determination of comparability. 
In order to establish a relationship between length of the primary 
cells and their division into secondary cells, the divided and undivided 
primary cells were separately enumerated, with the striking results 
shown in Tables II, III, VII, and VIII. In both normal and etiolated 
internodes there is a high correlation between length and condition 
with regard to division, the shorter cells in each cases being the undi- 
vided ones, which are but little more than half as long as the divided 
ones. There is a pronounced tendency for the short cells at the base 
of the internode to remain undivided, and this is particularly so in 
the case of the markedly shorter cells of the normal internode. 
In comparing the data, one is impelled to speculate as to whether 
light directly retards division of the primary cells, or acts indirectly 
by retarding the extension of the cells, so that relatively few of them 
attain the specific size for division. The latter supposition is the more 
simple. Bearing it in mind, we observe that in the plant grown in the 
light, 59 percent of the cells are undivided, 41 percent divided. In 
the dark 15 percent are undivided, 85 percent divided. From the 
frequency distributions in Tables I and VI we find that all below 59 
percent of the illuminated cells or 15 per cent of the etiolated cells 
would include all below a length of about 0.140 mm. Some divided 
cells are shorter, and some undivided ones are longer, the two classes 
approximately balancing one another. 
Turning to Tables II and VII, we find that the length 0.240 mm. 
is exceeded only by a trivial number of undivided cells, whether 
primary or secondary. Conversely (Tables III and VIII) we find 
that in only a trivial number of divided cells does the sum of the lengths 
of the derived secondary cells fall below 0.120 mm. The conclusion 
appears reasonably justified that the range of length within which 
division generally takes place is 0.120 to 0.240 mm., and that the 
greater part of the divisions occur at a length not far from 0.140 mm. 
The expression "specific size," used without quaHfication, refers, 
of course, to specific volume, and since our measurements concern 
only one dimension, length, they are insufficient to determine a specific 
