G05 
to the mutual action between particles a few hundred yards or a few 
inches asunder, seems to give only very small, scarcely appreciable, 
forces between bodies of such masses as those we experiment on in 
our laboratories, everywhere placed as close as possible to one 
another, — that is to say, in contact, and does not seem to provide 
for any considerable increase of attraction when the area of contact 
is increased, whether by pressing the bodies together, or by shaping 
them to fit over a large area. 
But if we take into account the heterogeneous distribution of 
density essential to any molecular theory of matter, we readily see 
that it aione is sufficient to intensify the force of gravitation be- 
tween two bodies placed extremely close to one another, or between 
two parts of one body, and therefore that cohesion may be accounted 
for without assuming any other force than that of gravitation, or 
any other law than the Newtonian. To prove this, let two homo- 
geneous cubes be placed with one side of each in perfect contact 
with one side of the other ; and let one-third of the matter of each 
cube be condensed into a very great number, i, of square bars per- 
pendicular to the common face of the two ; and let the other two- 
thirds of the matter be removed for the present. The mass of each 
bar will be — of the whole mass originally given in each cube. 
Let us farther suppose that the two groups of bars are placed so 
that each bar of one group has an end in complete contact with an 
end of a bar of the other. The attraction between each two such 
conterminous bars, however small their masses are, may be in- 
creased without limit, by diminishing the area of its section, and 
keeping its mass constant. But the whole mutual attraction be- 
tween the two groups exceeds i times the attraction between each 
of the conterminous pairs, and may therefore be made to have any 
value, however great, merely by condensing each bar in its trans- 
verse section, and keeping their number and the mass of each con- 
stant. 
We may now suppose another third of the whole mass to be con- 
densed into bars parallel to another side of the cube, and the re- 
maining third into bars parallel to the remaining side. If, then, 
either of these cubes be placed with any side in contact with any 
side of ti*e other, and allowed to take the relative position to which 
it will obviously tend — that in which the bars perpendicular' to the 
