162 
and fill it full of boiled water, and cork it. By this means 
the water, being deprived of air, when cold, dissolves any 
that may be amongst the powder. It is then weighed in 
water, and after that dried in the bulb, from which the weight 
in air is told. It is then burned to ashes, and their amount, 
and also their weight in water, ascertained ; and these, being 
deducted from the weight of the substance in air and water, 
we are able to calculate its specific gravity, free from the 
efifects of the ashes. In this manner I find the specific 
gravity of the different varieties of carbon to be thus, which 
are mean results of several experiments, the details of which 
it is needless to particularize here: — Graphite 2.177, hard 
coke 1.891, lamp-black 1.774, anthracite 1.760, and charcoal 
1.784. I have not examined the specific gravity of diamond, 
but the mean of many statements is 3.521. 
Now I have ascertained the following remarkable fact with 
regard to these specific gravities. If we assume the ultimate 
atoms to be spheres in the regular system and the hexagonal 
prism, and spheroids in the square prismatic, and calculate 
mathematically the relative specific gravities that we ought to 
have, assuming them to be arranged in the manner indicated 
by the crystalline form and cleavage, we find that the whole 
of the four species of carbon are brought about by its existing 
in ultimate atoms, having the relative volumes of i, J, and 
In the first place, beginning with the square prismatic 
system, the mean specific gravity is 1.773, and this, multi- 
plied by 2, gives 3.546 for the specific gravity of diamond; 
assuming the atoms to have their axes of the same relative 
magnitude as those of the two crystalline forms, and arranged 
similarly, that is to say, in the case of diamond, spheres, and 
in the other spheroids, with axes in the ratio of 3 and 5, as 
shown in Figs. 1, 5, 4, and 8. The mean of many state- 
ments of the specific gravity of diamond, which vary from 
3.48 to 3.6, is 3.521, which differs from the calculated spe- 
