352 
Journal of Agricultural Research 
Vol. VII, No. 8 
equilibrium with a partially saturated atmosphere, the amounts of hy¬ 
groscopic water present in the latter case, a 1 and b\ should bear the 
same relation to one another as do the hygroscopic coefficients— 
b' 
= or b=a 
a ' 
b' 
a' 
If the soils can conveniently be brought into equilibrium with the 
partially saturated atmosphere, it would simply be necessary to deter¬ 
mine accurately the hygroscopic coefficients of a few typical soils of 
which large quantities of thoroughly mixed, air-dried samples have been 
prepared and then expose to a partially saturated atmosphere portions of 
some of these along with the samples of which the hygroscopic coeffi¬ 
cients are desired. From the found amounts of hygroscopic moisture the 
hygroscopic coefficients could be calculated by the above formula. This 
would obviate many of the inconveniences connected with the deter¬ 
mination of hygroscopic coefficients, including the difficulty of obtaining 
a fully saturated atmosphere and of preventing dew formation through 
fluctuations of temperature. 
As we had a series of foot samples of loess which had already been 
subjected to careful hygroscopic-coefficient determinations (i, p. 215-216), 
they were the first to be tried. The samples, after being brought from 
the fields in cloth sacks, had been stored for several months in the un¬ 
heated but well-ventilated attic of the Nebraska Experiment Station 
building and later reduced to the desired degree of fineness, thoroughly 
mixed, placed in sealed jars, and again stored in the attic. About a 
year previous the moisture had been determined in some 50 of the field 
samples before preparing the composites and had been found to lie 
between the limits of 2.5 and 4.9, practically the same as found in this 
experiment. In shallow aluminum trays 5 by 7 inches with edges 0.75 
inch high the samples were exposed in triplicate on the shelves of the 
attic storeroom mentioned above. Each tray carried about 10 gm. of 
soil. All the samples were exposed for seven days, those from Wauneta 
and McCook from March 15 to 22 and'the others from March 25 to 
April 1. With the first set the range of temperature in the air of the room 
was i° to 14 0 C.; and that of the humidity at the Lincoln (Nebr.) station 
of the Weather Bureau, 2 miles distant, was 67 to 94 per cent, while 
with the second the corresponding data were 9 0 to 23 0 C. and 42 to 93 
per cent. Three of the samples which had been exposed in the first set 
were exposed again with the second and were found to have the same 
amount of hygroscopic moisture in both cases. Accordingly we may 
assume that all of the former contained the same amount of moisture 
that they would have shown if exposed with the second set, with which 
there were exposed in duplicate two soils, H and S, which had been 
repeatedly used as control soils in the determination of hygroscopic co¬ 
efficients and for which, accordingly, we had a great many concordant 
determinations, the average of which was 5.6 for H and 22 for S. The 
