Alternating Annual and Biennial Habit. 305 
the distance between the plants was practically the same 
there were about the same number of plants that devel- 
oped stems as there were on the margin of the sand bed, 
in fact a little less, 53 % amongst 348 plants. 
Our main result therefore is that the proportion of 
plants which developed stems in the center of the sand 
bed is 34% as against the 53% and 60% amongst the 
plants on the margin of this bed and in the control bed 
respectively. Equally striking was the sudden change 
in the behavior of the central plants in July. This pointed 
to some special cause. I suspected that it was connected 
with the growth of the roots and that these about this 
time had penetrated the layer of sand and reached the 
fertile earth beneath it. When I dug up the roots at the 
conclusion of the experiment I found that these were, 
as a matter of fact, longer than half a meter and had 
branched freely below the level of the sand. 
In order to find out whether this was the real cause 
of the development of the stems I made an experiment 
in 1891 with a bed in which the layer of sand was much 
deeper (one meter). A part of the original sand bed 
which was only one-half meter deep, and a neighboring 
bed filled with ordinary good garden soil served as con- 
trol. This time the bed was surrounded by boards and, 
consequently, there was no difference in the behavior 
of the central and marginal plants. For this experiment 
I used the seeds of a culture of Oenothcra rubrinervis 
which had been cultivated as an annual for two genera- 
tions (seeds of 1890 of the pedigree of Vol. I, p. 273). 
The sowing took place in May 1891. At the end of July 
I recorded the plants on the three beds, each of which 
was 3 square meters in extent. 
