36 
undispersed state. If, therefore, the larger part of of the undis- 
persed colloid happens to be mainly of an organic nature having 
a much higher adsorptivity (15 to 35 per cent higher) than that of 
the dispersed colloid, which on the other hand happens to be chiefly 
of an inorganic nature having a much lower adsorptivity, then the 
phenomenon observed in Hawaii soils may be easily explained. 
Other investigators have made similar observations. Bennett and 
Allison (4, p. 25), working with Cuba soils, state that the colloidal 
content, as determined by the water-vapor adsorption method, fre- 
quently runs 10 or 15 per cent in excess of the actual clay content, 
"probably because of exceptional physical properties of the soil 
material." This fact is all the more remarkable because the clay 
fraction of mechanical analysis has an upper limit of 0.005- 
millimeter diameter, whereas the upper limit of colloids, as inter- 
preted in the water-vapor adsorption method, is much less — prob- 
ably below 0.001 millimeter. Therefore, the percentage of colloids 
should be considerably less than that of clay. Furthermore, the 
organic matter of the Cuba soils referred to is very low in compari- 
son with Hawaii soils, most of them ranging between 1 and 5 per 
cent. Much larger discrepancies may logically be expected with 
soils having two or three times this amount of organic matter. 
Steele (S3, p. 31), referring to the colloids of Ohio soil profiles, 
remarks that " there was a fairly good correlation between the 
Robinson, water-absorption, and heat-of-wetting methods, except 
in soils high in organic matter." 
Alway (2, p. 21fi) believes that " it does not appear yet satisfac- 
torily established that the ability of soils to absorb water vapor is a 
reliable measure of their colloid content." 
Joseph and Snow (20, p. 119), working with heavy Sudan soils, 
state that — 
There is no connection between the proportion of very fine material (e. g. 
below 0.5/*) and other important soil properties. The proportion of the very 
fine material determined in this way [viz, sedimentation] would not, therefore, 
afford any indication of the " colloid " properties of the soil. 
The moisture-adsorbing capacity of two different fractions of the 
colloidal phase was determined in the case of two soils and compared 
with that of the soil from which they were isolated. One fraction 
was that part of the colloidal phase which could be isolated from 
suspension by means of supercentrifuging at the rate of 36,000 
revolutions per minute, whereas the other part was that fraction 
which could not be separated from suspension by this treatment, but 
had to be filtered through Pasteur-Chamberland filters and concen- 
trated on the surface of the filtering tubes. This latter fraction is 
sometimes referred to as " ultra-clay." This ultra-clay amounted 
sometimes to 1 to 1.5 per cent of the soil. Table 16 shows the hygro- 
scopicity of these two fractions as compared with that of the parent 
soil in the case of two highly organic, highly hygroscopic soils. 
In both instances the fine fraction of the colloidal phase or ultra- 
clay showed less adsorption than the coarse-colloidal fraction, indi- 
cating the probable presence of the highly adsorptive organic mate- 
rial in larger amounts in the coarse-colloidal fraction and in smaller 
amounts in the fine-colloidal fraction. Moreover, in the case of soil 
No. 12 the adsorptivity of the original soil is higher than that of 
