88 o 
Vol. XXIV, No. lo 
Journal of Agricultural Research 
inclosed minute mineral particles. The clean appearance of the obviously 
mineral particles was, of course, a strong indication that such was not 
the case, but the element of doubt remained nevertheless. 
Under the petrographic microscope in transmitted light these colloidal 
^^^^g 3 .tes were found to be almost universally transparent and, in the 
rare cases where this was not so, to be highly translucent. Therefore 
any mineral grain included within the colloidal aggregate would be 
readily visible provided that either its color or index of refraction was 
different from the color or index of the colloid. Observation of bire¬ 
fringence between crossed nicol prisms would, of course, readily differ¬ 
entiate between minerals and colloids regardless of similarity of color or 
refractive index. Such observation, however, showed that practically all 
of the colloidal aggregates were free from mineral inclusions. Presumably, 
aggregates containing such had been broken up by the repeated rubbing 
and washing process. 
Therefore, since the residues consist of easily determinable minerals 
and colloidal aggregates, one readily distinguishable from the other, a 
microscopic estimation of the relative quantity of each in a given sample 
was easily obtainable. Owing to the extreme variation of size of parti¬ 
cles a straight count would have been subject to large errors. Therefore, 
the estimation was made by means of a checker-work eyepiece micrometer, 
relative space occupied by the minerals and colloids being the basis of 
the calculation. It is believed, on the basis of results obtained ^vith 
synthetic samples, that errors due to the var5dng thickness of the parti¬ 
cles, both minerals and aggregated colloids, are fairly well balanced and 
do not appreciably affect the results. Quartz is the predominant mineral 
in the fractions, and since the specific gravity of this mineral (2.66) and 
that of the colloids (2.53-2.68) extracted from the soils are very near 
together, it was believed that differences in specific gravity .would not 
give rise to any serious error. 
In order to test this method of microscopic estimation, four synthetic 
samples were made up, each containing a known amount of colloidal 
material. Air dried Marshall soil colloid, which had been extracted 
mechanically and graded by means of the supercentrifuge, was mixed 
with quartz in the proportion of 1.2 gm. colloid to 0.8 gm. quartz and 0.8 
gm. colloid to 1.2 gm. quartz. The first of these mixtures was ground 
dry to pass a 200-mesh sieve. In order to facilitate any tendency of 
the colloid to form coatings on the mineral particles, the second sample 
was ground wet, dried and subsequently rubbed lightly with a pestle to 
pass a 200-mesh sieve. 
Since quartz is practically without cleavage, the particles formed by 
grinding have a tendency to assume shapes in which the dimensions vary 
little in different directions within the same particle, although the dimen¬ 
sions of different particles may vary widely. This statement also holds 
good for the colloidal aggregates. Such a similarity of fracture facilitates 
a microscopic estimation of the relative quantities of the different con¬ 
stituents present, and therefore does not afford a very difficult test of the 
method. But, since quartz is the predominant mineral constituent of 
soils, the results obtained with quartz are probably very near those 
obtained on the soil separates themselves. 
For the purpose of testing the method under more unfavorable condi¬ 
tions, Orangeburg subsoil colloid, extracted mechanically and graded 
by the supercentrifuge, was mixed with hornblende which has a tendency 
