26 BULLETIN 1311, U. S. DEPARTMENT OF AGRICULTURE 
geologists, states that " the physical process of weathering, abrasion 
by running water, can not reduce rock materials below a certain 
size." He considers that the diameters of the particles in the finest 
separates of the soil are below this minimum size and that therefore 
the finest soil fractions do not contain undecomposed rock- forming 
minerals. While this argument is against the presence of undecom- 
posed mineral fragments in soil colloids produced by mechanical 
forces,, there is nothing in it to disprove the presence of undecom- 
posed soil minerals of colloidal dimensions formed by the chemical 
disintegration of the larger particles. 
It was shown in Table 4 that the average composition of the 
colloidal matter is quite different from the average composition of 
the mixture of minerals constituting the coarser mineral particles 
and is different from any probable combinations of such minerals. 
It is therefore evident that the colloidal matter is not made up pre- 
dominantly of the common soil-forming minerals. However, it 
might be assumed that the colloidal matter contains considerable 
quantities of either the mixed common soil-forming minerals or 
particular minerals, together with the gels of iron and aluminum 
oxide and some adsorbed monovalent and divalent bases. Although 
such an assumption is possible, judging only from the ultimate chem- 
ical composition, it leads to rather improbable conclusions. 
The usual mixture of soil-forming minerals in the coarser par- 
ticles of the soil contains a high proportion of quartz and therefore 
has a high silica content — about 84 per cent — while the colloid con- 
tains approximately half as much. It might be assumed, then, that 
the colloidal matter in many cases contains up to about 50 per cent 
of the mixed minerals as they occur in the coarser soil fractions. 
According to this assumption a considerable part or^all of the silica 
of the colloidal matter would necessarily be present as a constituent 
of the soil-forming minerals, and the chief or only products of 
chemical decomposition of the parent soil minerals would be hydrous 
aluminum and iron oxides. This is hardly reasonable. It is much 
more probable that a large part of the silica in the colloid is also a 
decomposition product of the parent minerals and is present in the 
same state as the alumina and iron — either free or combined with 
them. If this is true, the colloidal matter could contain only a 
relatively small proportion of common mixtures of soil-forming 
minerals. 
Since the feldspars and micas are frequently concentrated in the 
fine sand and silt fractions of the soil, it might be inferred that 
these particular minerals — which, like the colloid, contain compara- 
tively low percentages of silica — make up a large proportion of 
the colloidal matter. 20 In a few cases the colloidal matter contains 
a sufficiently high total of the monovalent and divalent bases to 
permit the presence of approximately 45 per cent of feldspar or 
mica. In these cases, however, the monovalent and divalent bases 
are not present in the same proportions as in either the commoner 
feldspars or in the commoner micas, but certain mixtures of these 
minerals could be assumed. 
20 Biotite, hornblende, pyroxene, and garnet can not be present in appreciable quan- 
tities in soil colloids because soil colloids contain little, if any, ferrous iron. Nine col- 
loids averaged 0.69 per cent ferrous iron by Cooke's method, and this figure is probably 
greater than the true content on account of the presence of organic matter. 
