12 TRANSPIRATION IN A DESERT PERENNIAL. 
that the plant was replaced by a small atmometer which lost about the 
same amount of water as the plant. This atmometer (D, fig. 2) consisted 
of a tube containing a wick which dipped below into a small reservoir of 
water and ended above in sand, slightly piled. A cover was provided for 
the evaporating surface for use when the atmometer was removed to be 
weighed. This atmometer was standardized to an open pan of water in a 
dark room and its loss reduced to loss per square centimeter of water surface. 
The results of this standardization brought out the fact that the atmometer’s 
rate was a fairly constant one under the same conditions; for the average of 
three readings of 24 hours each it varied less than 1 per cent from the extreme 
readings. The actual losses from the atmometer during the experiments 
were always reduced to unit area and the loss per unit area from the plant 
for the same period was divided by this number, the result being called the 
relative transpiration.* It might seem on a priori grounds that the shape 
of the curve obtained from these results would be parallel to a true curve, 
but that the actual values might be either too high or too low, since the 
conditions under the two bell-jars were seldom absolutely identical; yet 
the values for relative transpiration obtained in 1912, when atmometer and 
plant were under the same bell-jar, are in such close agreement with these 
earlier ones that the results obtained by the use of two bell-jars are shown 
to be as accurate as are those obtained when one jar was used. 
While this last method has the advantage over the earlier one that plant 
and atmometer are under identical conditions, it has the disadvantage that 
a larger beam of light must be let into the jar, with the consequent greater 
rise in temperature within the jar. It will be seen, from the tables of 
September 1912, that the temperatures under the bell-jar rose every day 
as much as 3° above the outside air temperatures. Asin the case of humidi- 
ties, it is true here that the plant is not placed under unusual conditions, 
unless by chance the hottest day of the year was selected for the experiment, 
and this was not the case. Yet, when only parts of a plant are used, it seems 
best that the conditions surrounding the branch under experimentation 
should be as nearly as possible identical with those around the rest of the 
tree. In the subsequent tables, temperatures and humidities are given in 
order that any difference may appear which might affect conclusions. A 
number of experiments were discarded where the meteorological conditions 
within and without the bell-jar differed over 10 per cent; but no experi- 
ments were discarded which might bring out negative evidence for the main 
conclusion. 
Great care was taken to have all joints tight during the experiment and to 
keep the weighing-bottles and atmometer air-tight when not under the bell- 
jar. Before each set of experiments trials were made in which the atmom- 
eter was used as a plant and its losses determined by weight and compared 
with the gain in weight of the calcium chloride plus or minus the correction 
determined from the dew-points. Sensitive chemical balances were used, 
but readings were taken only to 1 mg. and the error taken as plus or minus 
*Livingston, B. E., Carn. Inst. Wash. Pub. 50. 
