164 PROF. W. J. SOLLAS ON A PROCESS FOR [May 1902, 
with a well-ground glass stopper to prevent evaporation, is used for 
making the diffusion-column. Methylene-iodide is first introduced 
by means of a pipette, till its surface reaches the first centi- 
metre; fluid of sp. gr. 3°24 is next added, till the second division is. 
reached ; and so on, till the lightest fluid required is finally added. 
A tube to contain the entire column would be inconveniently long: 
it is better therefore to make the column in two or more parts, two 
will usually suffice, using two tubes for the purpose. The column 
may extend from 3°34 to 2°84 in the one and from 2°94 to 2°44 in 
the other. Into each of these the powdered rock is introduced, as 
much being added as will lie on the point of a broad-bladed pocket- 
knife: the tubes are then stoppered, and left to stand. The mineral 
grains soon separate out into layers, according to their specific 
gravity. If each mineral were of a constant composition, and if 
each grain of the powdered rock consisted of but one kind of 
mineral, the layers would always be sharply defined, but this is very 
rarely the case. As a rule the grains arrange themselves in dense 
zones,—thus showing that to this extent they are homogeneous ; 
but the fluid in the spaces intervening does not by any means 
remain clear, mineral grains extending from one zone to another. 
It might at first sight appear as if this fact alone presented an 
insurmountable obstacle to the quantitative determination of the 
mineral composition of a rock, at all events by any process of 
mechanical separation. Fortunately, however, this is not the case. 
The chief reason why the grains extend from zone to zone is that 
they are not all small enough to be homogeneous, some consisting 
partly of one mineral and partly of another. But evidently, on the 
doctrine of chances, in a mixture of two minerals in equal propor- 
tions as much of the heavier mineral associated with the lighter one 
will lie above the mean line between two zones, as of the lighter 
associated with the heavier below it; and thus, if the mixture be 
introduced into a fluid of a specific gravity which is the mean of 
that of the two minerals, the lighter with a certain quantity of 
associated heavier mineral will float above, and the heavier mineral. 
with a corresponding contingent of lighter will sink below. The 
fallen grains, together with those floating in the lower half of the 
separating liquid, may be collected and weighed, and their weight 
will represent that of the heavy mineral taken alone. Similarly 
with the lighter grains. Were the grains needed for the purpose of 
chemical analysis separation in this manner would be futile, but 
for our purpose the method is both logically correct and successful 
in practice. Ifthe two minerals in the case which we have supposed 
should not be present in equal proportions, that circumstance will 
not affect the argument. 
The component minerals having assumed a position in the 
diffusion-column corresponding with their specific gravity, this 
must next be determined: it may be taken as that of the middle of 
each zone into which they have collected. The determination is 
most quickly and conveniently made by means of beads of known 
