218 
PRESTON. 
the result appears to be, all gravity determinations on conti¬ 
nental mountains, when corrected in this way, show the 
matter under the station to be exceedingly light. Then the 
observers set about finding an explanation, and they hit 
upon the idea that great caverns must exist under the moun¬ 
tain ranges; but when observations were made on island 
mountains, the reverse was found to be true— i. e., that islands 
are generally heavier than one would expect. So, then, we 
are confronted by these two facts: Continents show a defect 
of gravity, and islands, at least in the middle of the oceans, an 
excess of it. The Coast and Geodetic Survey has carried out 
gravity work at the principal island groups of the Atlantic, 
on the coast of Africa, in the Pacific ocean, and on the shores 
of Asia, and the general result has been, as stated, that the 
force of gravity is apparently least where we would expect 
it to be greater, and vice versa. 
It has recently been proposed to omit the correction for 
continental attraction altogether, because by so doing the 
results are much more consistent with Clairant’s law. It 
cannot be admitted for a moment that the visible mountain 
masses exert no accellerating influence on the movement of 
the pendulum, but since no effect is discernible there must 
be compensation from some cause, and we are forced to the 
conclusion that there must be a deficiency of density under 
the continents. 
Let it be put in this way: Every pyramid of matter with 
its apex at the center and its base at the surface of the earth 
contains the same mass. This assumption does no violence 
to preconceived notions. Any contracting sphere will take 
such a form as will most easily relieve its tangential strains, 
and if the ocean bed has sunk and by lateral pressure lifted 
the continent we must admit a diminution of density. 
Summing up the situation, then, for the present status of 
gravitational research, we may say that after many years of 
trial an apparatus has been found that is satisfactory for the 
rapid and economic production of results. It only remains 
to multiply stations. In the theoretical treatment of the data 
