THE CHEMICAL COMPOSITION OF SOIL COLLOIDS 37 
kaolinite but would not agree with the combined water determina- 
tions any better. Allophane (Al 2 3 .Si0 2 .5H 2 0) contains more com- 
bined water than kaolinite and would show much greater excesses 
of silica than shown in the table. Pyrophyllite (Al 2 3 .Si0 2 .H 2 0) 
would take up the excess silica in all but one colloid, but practically 
all the iron would be shown as free iron oxide, even in those soil 
colloids which are grayish white in color. The application of the 
leverrierite 21 formula in a stoichiometrical analysis is unsatisfactory 
because it does* not have a definite composition according to the few 
analyses found in the literature. The compositions of the colloids 
we have analyzed show but little more variation than the composi- 
tions of the minerals included in the leverrierite group. 
From the considerations given above it does not seem that any of 
the formulae of the commoner hydrated aluminum silicates fit the 
composition of soil colloids any better than the kaolinite formula. 
CONCLUSIONS REGARDING COMPOUNDS PRESENT 
It is not possible to draw definite conclusions as to the compounds 
present in the colloidal matter of soils. The constitution of colloidal 
minerals and of artificial silicate gels are still matters of contro- 
versy, and the " make up " of soil colloids will probably not be satis- 
factorily solved till more is known of the constitution of these 
simpler inorganic colloids. The data obtained in this work bearing 
on the " make up " of the colloidal matter may be summarized as 
follows : 
(1) It is probable that there are some finely divided minerals, 
such as make up the larger particles of soil, in soil colloids, but the 
colloidal material is not chiefly made up of such particles. The 
quantity of finely divided minerals probably varies in different soil 
colloids. In general, however, it is probable that only a small part 
of the total colloidal matter can be made up of such material. 
(2) If there are separate and distinct compounds present in the 
colloidal matter of soils, the mixture is very intimate and strongly 
resists separation. Since separate and oppositely charged colloidal 
particles can not form a stable suspension, it seems probable that 
the silica and alumina of soil colloids can not be present in separate 
particles, for the colloidal matter passes through the suspension 
stage during separation. 
(3) The colloidal matter of soils resists the extraction of its con- 
stituents by solvents with considerable tenacity, and the extraction 
of last traces in particular is strongly resisted. Certain constituents 
appear to be more easily removed than others. Thus, soda and lime 
are more easily removed than potash and magnesia, and these last 
two are more easily extracted by water than silica, iron, and alumina. 
(4) There are apparently two forms of iron present in soil col- 
loids — one a hydrous ferric oxide and the other some compound 
without red or yellow color, presumably a silicate. 
(5) Matter brought into solution by treating soil _ colloids with 
dilute alkali or acids was no more definite in composition than the 
original colloid. Likewise the undissolved residue was no simpler in 
composition. 
27 Larsen, and Wherry (26) describe the properties of leverrierite and give several 
analyses. They give the formula A1 2 3 . 2±Si0 2 . 2V 2 ±H 2 0. It contains small amounts 
of Feo0 3 , RO, and R,0. 
