352 BOTANICAL GAZETTE [DECEMBER 
These were equipped with the rainproofing valves described by 
LrvincsTon and THONE (13). Their reservoir bottles were partly 
buried, so that the cups were about 20 cm. above the ground level. 
From April 30 until July 1 readings were made weekly or fort- 
nightly, during the month of July they were made daily, and 
from August 6 until September 20 they were made every two 
days. All readings, after correction, were reduced to mean daily 
rates for ten-day periods. 
TEMPERATURE.—Air temperatures were obtained from a Sixe- 
type minimax thermometer at each station. The thermometers 
were placed at the same level as the atmometers. No artificial 
shelter was used except at station 1, since there was sufficient natural 
cover to protect the instruments from direct insolation at all the 
other stations, and even at the latter place the growth of the 
Impatiens thicket soon made artificial shelter unnecessary. Read- 
ings were made on the same schedule as that used in the atmometer 
observations. The stations were visited each day just before the 
period of maximum temperature, so that the maxima and minima 
for the preceding twenty-four or forty-eight hour period were 
obtained. 
Soil temperatures were obtained by means of test-tubes sunk 
into the soil to a depth of 1ocm. The lower end was kept filled 
with water, and the tube kept stoppered. When an observation 
was to be made, a thermometer was lowered into the tube until the 
b was immersed in the water, and read after sufficient time had 
been allowed for an equilibrium to be reached. Since the readings 
were always made shortly before the period of maximum air tempera- 
ture,it may be assumed that the soil temperatures thus obtained were 
a little below the maximum for the day. Since, however, the total 
diurnal fluctuations in soil temperatures are known to be small, 
and especially since in the present study they seemed in the end to 
have no particular significance, this source of inaccuracy, as well as 
the rather crude method employed for obtaining the data, may be 
overlooked. 
For both air and soil temperatures, it may be remarked that the 
figures up to July 1 can have but little significance, since they apply 
to periods of seven or fourteen days. During July, of course, true 
