86 The American Geologist. Feb. \m 
Indeed the upper surface of the lowest descends as shown above, 
three times as far as the shrinkage in any one direction re- 
quires. But the upper surface of the second shell must sink 
by about the same amount in consequence of its own contrac- 
tion and must also settle down on the sunken surface of the 
first. Were their contractions equal, which is not the case, the 
settlement of the second shell would be double that of the first 
but as the cooling increases upward its actual descent is rather 
more than this. 
Evidently this settlement steadily and incessantly increas- 
es upward, its sum rising faster and faster until the layer of 
maximum cooling is reached and then more and more slowly 
to the surface. The increase does not in this case as in the 
former pass into a decrease at any point. Consequently its 
total amount is greatest at the surface though its increase is 
very small indeed through the superficial layers. That is to 
say, in other words, that the fall of each shell in consequence 
of the contraction of those beneath it commences practically 
at the depth of 400 miles and becomes greater up to the super- 
ficial shell which falls most, the maximum increase being at 
or near the shell of greatest cooling. 
It is almost needless to point out that in consequence of the 
spherical form the fall of the different shells is accompanied 
by a narrowing of the space which they must occupy. The 
nearer to the centre they fall the narrower this space will be. 
did every shell descend equally the change of space might be 
represented by a straight line diverging from the axis to the 
surface, every step of fall being represented by an equal dimi- 
nution of space. But as shown above their descent is not 
equal. Consequently the successive positions must be repre- 
sented by a curve, similar in geiieral form to that in fig. 1. 
Starting with the other from zero, at a depth of 400 miles, it 
diverges from the axis more rapidly as it rises until the level 
of maximum cooling is passed, when the previous rate of in- 
crease as shown above, becomes less and less as the surface is 
neared. 
Investigation shows that in the deeper part of that portion 
of the sphere with which we are now concerned the shrinkage 
from cooling exceeds the loss of space due to de.'^cent. The 
curve representing the latter will therefore lie nearer the axis 
than the other. 
