154 BOTANICAL GAZETTE [FEBRUARY 
It is necessary also that the experiment be continued long enough 
(that is, a month or longer at about 23°C.) to allow the shoots to 
reach a sufficiently large size, since if the shoots are too small the 
error in measuring their masses prevents exact results. On the 
other hand, the experiment must not last too long, for if the shoots 
become too large they produce themselves too considerable a share 
of the material needed for their own growth. The leaves were 
generally kept on wet filter paper in flat dishes with a loose glass 
cover. One of the greatest sources of error or variation in the 
results was probably the differences in the absorption of water by 
the roots of different leaves or pieces of leaves. Furthermore, light 
is an important factor in determining the masses of shoots produced, 
and when leaves are suspended in an aquarium and able to shade 
each other, inequality of illumination of sister leaves also forms a 
source of error. The new shoots can be cut off from the leaf com- 
paratively neatly, although slight variations or errors are unavoid- 
able in this operation. The shoots were freed from water droplets 
on their surface and weighed fresh, on the assumption that the 
dry weight under the conditions of the experiment is a fairly con- 
stant fraction of the fresh weight, which has been found to be 
approximately correct. The leaves were usually but not always 
weighed without their petioles. 
TABLE I 
' Number of sh Mem. of sh - 
Sister leaves sentacod Tian teal eet és nog ying 
I Leal #0006338 3 350 
BOM 2 ces = 345 
Tl eat Piedweus bs I 290 
pa oe Vaan en 2 3006 
Laat f.0550655 2 375. 
TH. ‘Laat aa 4 385 
Fs ae eee a 5 504 
IV. Leaf iu coe 4 607 
V Leet toc 4 457 
© ALOAR oho ees 5 455 
Table I gives the weight of the shoots produced by 5 pairs of 
sister leaves in 33 days (February 15—March 20). The two sister 
