84 BULLETIN 1059, U. S. DEPARTMENT OF AGRICULTURE. 
the test is conducted in such manner as to keep the moisture uniform 
throughout the soil, and hence uniform for the deepest and shallow- 
est-rooted seedlings. 
It is also self-evident that specific differences may be brought out 
by one set of conditions, which would not be apparent in another 
set. particularly conditions which make the requirement for mois- 
ture great or small. If transpiration is very rapid, seedlings of a 
shallow-rooted species may be unable to meet this demand, while 
deeper-rooted seedlings in the same pan may pull through, because 
their supply at this stage is somewhat more readily obtained. For 
an actual test of drought resistance, therefore, it is fundamentally 
necessary that the transpiration and soil-drying process should be 
slow enough to permit equalization of the opportunities before the 
critical test comes. Hence the standard conditions of exposure 
which have alreacjy been suggested. 
The method of recording the death of each seedling in a lot of 
100. together with the pan weight and calculated moisture accom- 
panying such death, has a distinct advantage over the method which 
permits only one determination of the moisture content when all 
of the seedlings, or a majority of them, have succumbed. It gives 
an indication of the possible variation between individuals of the 
same species, and a measure of the probable experimental error 
due both to this variation and to uneven distribution of moisture 
in the pan, which is not wholly avoidable. What it really amounts 
to is practically 100 separate tests on 100 sections of soil. If. on the 
one hand, the first losses occur in sections of the soil which have 
unavoidably become drier than the average, on the other hand, the 
last survivors are undoubtedly in areas which are at the opposite ex- 
treme. These variations should be largely compensated by taking 
the algebraic mean of all the moisture determinations, a figure in 
which a great deal of confidence can be placed. 
INDIKECT METHODS FOE WILTING COEFFICIENTS. 
Inasmuch as the direct determination of the wilting coefficient is 
a process which is likely to require several weeks, at the best is liable 
to rather large experimental errors, and is also, without question, in- 
fluenced by the kind of plant used, various methods have been devel- 
oped by which the affinity of the soil for water may be determined : 
and the amount of water held by it under certain empiric conditions 
of the test may be related to the amount which would be held against 
the pull of plants. 
In addition to furnishing a ready, if only approximate, index to 
the soil conditions which may be encountered in the field, and espe- 
cially an index to the danger of early drought, it seems that the use 
of indirect methods, employing definite physical forces for the crea- 
