34 BULLETIN 1311, V. S. DEPARTMENT OF AGEICULTUEE 
an undissolved residue amounting to about 20 per cent of the origi- 
nal material. According to the general theory of the so-called 
" rational " analysis of clays, this residue would be interpreted as 
the quantity of quartz, feldspars, and other primary minerals pres- 
ent in the colloid. Mellor (27), however, has shown that this inter- 
pretation is not correct. 
COMPOUNDS INDICATED BY STOICHIOMETRICAL CALCULATION 
In the review of previous work it was pointed out that Schloesing 
and Hilgard considered the essential constituent of soil colloids to 
be a hydrous aluminum silicate of the composition of kaolinite. It 
is evident from the analyses given in Tables 1 and 3 that soil col- 
loids are not all of the same composition, and so, of course, can 
not be a definite compound. They may, however, be mixtures of 
two or more different compounds. This possibility could be tested 
for certain compounds by assuming the presence of these compounds 
and seeing if the analytical figures are satisfied. This method has 
been used by Hall (19), by Atterburg (#), and by others. These 
investigators assumed the presence of a hydrous aluminum silicate 
of the composition of kaolinite and calculated the excesses of silica 
and iron oxide as quartz and ferric oxide. 
In our work with oxalic and hydrochloric acids the hypothesis has 
been advanced that part of the iron may be present combined in 
some way with the silica, and in the stoichiometrical calculations 
which follow it is assumed that part of the iron is combined with the 
silica in the same proportion as in nontronite 26 (Fe 2 3 .2Si0 2 .H 2 0) 
when there is an excess of silica over that necessary to combine with 
the alumina. Nontronite is the ferric equivalent of kaolinite. 
In Table 10 all the alumina or silica, as the case may be, is calcu- 
lated to the kaolinite formula. When there is an excess of silica 
over alumina, it is combined with the iron in the proportions that 
occur in nontronite. When there is more than sufficient silica to sat- 
isfy the alumina and iron, it is entered in the column headed " Free 
Si0 2 ," and excesses of alumina and iron oxide are treated in a simi- 
lar manner. The calculated combined water is the sum of the com- 
bined water in the calculated quantities of Al 2 3 .2Si0 2 .2H 2 0, and 
Fe 2 ? .2Si0 2 .2H 2 0. No combined water is assigned to the excesses 
of silica, alumina, or iron; although if these constituents are present, 
some water which would not be driven off at 110° C. is probably 
associated with them. 
26 This is the formula given by F. W. Clark (8). 
