DIFFUSION IN SILICATE MELTS Raa 
cooling of the marginal layer had proceeded until one-fourth of 
the amphibole contained in that layer (or 5 per cent of the layer) 
had been precipitated. If this condition were maintained the 
precipitation of amphibole at the margin would proceed by diffusion 
from the parts not yet cooled below the temperature of precipitation 
of amphibole. At the end of two-thirds of a year a proportion 
of the excess amphibole would be precipitated equal to the area 
between the one-year curve and the axis of ordinates. Regarding 
this area as a triangle of base sensibly 0.3 m. and assigning any 
arbitrary total thickness « to the mass, then the thickness of the 
amphibole deposit in meters would be 
© 8 Se 
I 
Waa | ico 
That is, a deposit 0.0075 m. or 2 cm. thick would be formed on the 
margin in two-thirds of a year. Its thickness is independent of the 
total thickness of the mass of magma and all of it would be derived 
from a layer of magma 33 cm. thick. After sixteen years the deposit 
would be about 2 cm. thick, all coming from a layer of magma 
about 1.5 m. wide. After two hundred and fifty-six years the 
deposit would be about 8 cm. thick, all from a border portion less 
than 7 m. wide. 
It is apparent that the possible results of diffusion after the 
manner postulated are exceedingly small. A mass of magma 
large enough to remain in the necessary condition for two hundred 
and fifty-six years would have a border phase 8 cm. thick. By 
necessary condition is meant that the margin should be cooled 
within its crystallization range and the main portion of the mass 
be not yet so cooled. The indications are that the mass would 
require to be at least 300 m. thick and be intruded under special 
conditions of temperature of magma and of wall rock, and the border 
phase would then be insignificant. Even by making more liberal 
assumptions as to the amount of chilling at the margin, say, a 
chilling sufficient to precipitate 25 per cent of the magma solution, 
the possible border phase would be increased in magnitude only 
five times. Moreover, as one increases the extent of marginal 
chilling, a stage is soon reached where so much precipitation takes 
