262 BOTANICAL GAZETTE [OCTOBER 
breaking strength of the controls in this experiment, 5.8 per cent., 
was much less than in experiment 2, since the breaking strength was 
so much greater. 
The tension fastenings were made just below the fourth pair of 
leaves. Similar fastenings were made on the control plants, and 
from these the plants were fastened to upright supports to keep them 
from being swayed by the wind. 
Experiment 4. Sinapis alba, field culture.—These plants grew 
approximately 15°™ apart in the row. Every other plant was used 
as a control. The tension fastening was placed just below the 
seventh node above ground. Similar fastenings were placed on the 
control plants. From these fastenings, the plants were attached to 
strong cords, stretched one on each side of the row and parallel to it. 
A weight of 250%™ was attached to the experimental plants on July 
16, 1907, and gradually increased to 1150%™ on July 24. The expen- 
ment ended two days later. The details of this experiment are giv 
in table IV. The average increase of the breaking strength of the 
experimental over the control plants was 4.8*% or 32 per cent. 
The anatomical structure of these stems was determined by 
methods 2, 3, 4, and 5. Free-hand sections were made of all the 
stems about 5™™ below the point where they broke. These sections 
were stained in anilin safranin and mounted in glycerin. All the slides 
were then taken by a second party who labeled them with a secret 
label. The measurements given in table IV were then made, after 
which the second party gave me the key to the labels. The measure 
ments of the experimental plants were now separated from those of 
the controls. An average of each set shows a decrease of 10 per cent. 
in the total xylem area and an increase in the thickness of the xylem 
cell walls of 5 percent. The increase in hard bast in 5 stems which 
were selected wholly by chance was 52 per cent. The increase © 
hard bast elements in the same stems was 38 per cent. 
The total xylem area for the five stems in which the hard bast 
determinations were made was also measured by method 4 < 
results that agreed within less than one-half of 1 per cent with the 
measurements made by method 3. 
The hard bast in the controls was lignified in all but a few ste" 
where it was only partly lignified. Stem 18 showed the least 
