ESTIMATION OF COLLOIDAL MATERIAL IN SOILS. 19 
preceding fractions. The chemical composition is discussed in detail 
in an unpublished report (Robinson and Holmes, 27). 
So far as the adsorption method of determining the colloidal 
material in soils is concerned, the important comparison is between 
the adsorptive capacity of the first fraction of colloid and that of the 
total quantity extracted. It will be noted that all the fractions of 
colloid taken together r - total quantity colloid extracted"; had prac- 
tically the same adsorptive capacity as the first fraction. The differ- 
ences are less than 10 per cent, except in the case of the Huntington 
subsoil. It is significant in connection with the cause of the disa- 
greement between the dye, water, and ammonia ratios given in 
Table 2, that the slight variations between the adsorptive capacities 
of the first traction and the combined fractions were practically the 
same for malachite green as for water or ammonia. 
Although this comparison of the samples of colloid successively 
extracted from a soil was not undertaken with a view to determining 
quantitatively the total amount of colloidal material that it was 
possible to isolate by the procedure outlined, nevertheless the total 
quantity of colloidal material recovered is a matter of considerable 
interest. It will be noted that the quantity of colloidal material 
actually extracted from the different soils was 6.19 to 38.5 per cent 
of the weight of the whole soil. This is highly important in bearing 
on the opinion held by Schloesing 40 years ago, and still widely 
current, that soils contain less than 2 per cent of truly colloidal 
material. It is improbable that Schloesing was dealing with a prod- 
uct in a finer state of dispersion than that obtained in this separation 
inasmuch as he'was obliged to rely on time of subsidence for grading 
his material. As mentioned on page 9. suspensions passing through 
the supercentrifuge in this work apparently contained very few 
particles larger than 0.3 micron in diameter. 
The quantities of colloidal material separated from the Sassafras 
subsoil and Marshall soil were doubtless considerably less than the 
quantities of this class of material actually present in the soil. As 
already pointed out,, considerable losses of the fine material of these 
soils occurred in preparing the samples of colloidal material. In the 
cases of the Huntington subsoil, the Sharkey soil, and the Vega Baja 
soil the quantities of supercentrifuge colloid isolated represent prac- 
tically all this class of material that it was possible to separate by 
the means employed. 
To return to the question of the sample of colloidal material as a 
factor affecting the accuracy of colloid determinations by the adsorp- 
tion method, it appears that so far as adsorptive capacity is concerned 
a small sample of extract able colloidal material differs by approxi- 
mately 10 per cent from all the colloidal material which is isolatable 
by the methods described above. Of course the colloid which it was 
impossible to extract with the methods employed might have a 
different adsorptive capacity for dye. water, or ammonia than the 
colloid which was extracted. However, to account for the disa- 
greement in the ratios on the basis of an unfair sample of the colloid 
the degree of difference between the adsorptive capacities of the 
extractable and unextractable colloidal material would have to be 
different with respect to dye. water, and ammonia. Evidence on 
this point is given in the following section. 
Comparison of the extractahle and unextractable colloidal material, — 
The adsorptive capacities of the extracted and the unextracted 
