ESTIMATION OF COLLOIDAL MATERIAL IX SOILS. 15 
the same power of adsorption that it has when present in the un- 
treated soil. If this assumption is not entirely correct in the case of 
all the colloidal materials, if certain colloids after extraction are more 
adsorptive of malachite green and less adsorptive of ammonia than 
they are in the unextracted soil, it might explain the disagreement 
of certain ratios. 
Another general cause for disagreement in certain of the ratios rnav 
lie in the samples of colloidal material. The different kinds of col- 
loids 8 making up the colloidal material extracted from a soil must 
have widely different adsorptive capacities for malachite green, 
also for ammonia, and to a much less extent for water. This seems 
evident from the variations in the adsorptive capacities of the col- 
loidal materials extracted from different soils reported in Table 1 
and from tests made of the adsorptive capacities of artificial inorganic 
gels. The ratio of adsorption of soil to adsorption of colloid would, 
therefore, represent the percentage of colloidal material in the soil 
only when the sample of colloidal matter extracted was truly repre- 
sentative, with respect to adsorptive capacity at least, of all the 
colloidal material in the soil. If the sample of colloidal material 
was not representative, the dye, water, and ammonia adsorption ratios 
would not be expected to agree. Since the results shown in Table 
2 are based on samples which were probably only a tenth to a twen- 
tieth part of all the colloidal material in the soil, it might be ex- 
pected that in many instances the samples were not representative. 
These two most probable sources of error in the adsorption method 
of determining colloids — the lack of representativeness of the sample 
of colloidal material and possible alteration in adsorptive capacity 
of the colloid produced by extraction — are discussed in the following. 
REPRESENTATIVENESS OF A SAMPLE OF COLLOIDAL MATERIAL. 
Plan of investigation. — Since the colloidal material in any one soil 
appears to be a mixture of at least several different colloids, which 
presumably would vary considerably in the stability of their suspen- 
sions and in the readiness with which they are redispersed after 
coagulation, it might be expected that a small sample of colloidal 
material extracted from a soil would be composed only of those col- 
loids which disperse most readily and would not. therefore, be at all 
representative of the remainder of the colloidal material in the soil. 
This is assuming, however, that a mixture of the soil colloids is readily 
separated into its constituents. As a matter of fact, attempts 
at separation of the colloidal constituents by various means indi- 
cated that this separation is not readily brought about. 
A direct comparison between a small sample of colloidal material 
and all the colloidal material in the soil could not be obtained, since 
it was impossible with the means we employed to extract all the 
colloid present. However, a small sample of the extractable colloid 
could be compared with all the extractable colloid. In order to do 
this and at the same time to gain information concerning the uni- 
formity of the extractable colloidal material in soils, small lots of 
colloidal material were successively extracted from the same soil 
and the adsorptive capacities of the different lots compared. 
8 "Different" colloids here is intended to include different states, as of aging or dispersion, of the same 
chemical substance as well as substances of a different chemical composition. 
