ABSORPTION BY SOIL CONSTITUENTS. 7 
It will be noted that the soil fractions containing particles from 
0.001 to O.OoOmillimeterin diameter had absorptive capacities averag- 
ing about 25 per cent as great as the corresponding colloids. The 
coarser soil fractions, made up of particles ranging mostly from about 
0.050 to 2 millimeters, had absorptive values which were for the most 
part much lower than those of the finer fractions. 
The absorptions per gi'am of the noncolloidal soil fractions are thus 
low as compared with those of the colloids. Nevertheless, if we take 
into account the relative amounts of the different fractions present 
in the soil it would appear that from 20 to 70 per cent of the total 
absorptive capacities of the four soils was due to the noncolloidal 
particles in these soils. If these data could be taken as accurately 
representing the absorption values of the noncolloidal part of the soil, 
it would be evident that the colloidal content of the soil could not be 
estimated with even approximate accuracy from the absorption of 
the whole soil. 
Although the above data on the absorptive capacities of the non- 
colloidal soil particles are probably as good as can be obtained by this 
method of investigation, it is believed that they are not reliable and 
that they must be considerably higher than the true values. It 
should be borne in mind that a relatively small contamination with a 
highly absorptive material, such as the colloidal matter, would appre- 
ciably enhance the absorptive capacity of the noncolloidal fractious. 
Furthermore, we have no positive proof that the noncolloidal frac- 
tions were free from colloids. We merely know that all colloid 
possible was extracted with the means we employed. It is very pos- 
sible that by some further treatment still more colloid could have been 
brought into suspension and separated, Just as traces of ammonia or 
gentle rubbing yielded more colloid when mere agitation mth dis- 
tilled water was ineffective.^ 
In the process of separating the coUoidal and noncoUoidal portions 
of the soil the size of the particles in the two products was carefuUy 
checked by microscopical observation. It was evident from this 
control that the noncoUoidal fraction was practically free from par- 
ticles smaller than 1 micron in diameter and the colloidal matter 
contained but very few particles larger than 1 micron. However, if 
our methods of defiocculation were not perfect, aggregates of colloidal 
matter, similar to clumps of dried ferric hydroxide, or aluminum 
silicate gels, or partially decomposed organic matter might well have 
gone into the noncoUoidal fraction. Examination of the dry non- 
colloidal fractions with the petrographic microscope indicated that 
this actually occurred. The fractions from the different soils which 
were made up of particles of 0.001 to 0.050 millimeter in diameter 
contained from 60 to 75 per cent of particles which could readily be 
identified as crystalline minerals; the rest of the material was 
without crystaUine properties and had the same general appearance 
as dried inorganic gels or as the dried soil colloids, which we have 
separated from the soil with a supercentrifuge. 
ESTIMATION OF NONCOLLOIDAL ABSORPTION FROM MINERAL 
PARTICLES. 
Since a complete separation of colloidal from noncolloidal soU 
material is apparently impossible, owing to the difficulty of dis- 
6 Williams (£6) relied largely on boiling to bring about dispersion. However, boiling a fine fraction of 
Huntington subsoil for 18 hours failed to bring about a state of dispersion whereby all the colloid could 
be removed. 
