of Edinburgh, Session 1865-66. 513 
values for the whole upper crust of the earth, the quantity con- 
hK 
ducted out across the whole earth’s surface per annum will be • 
The bulk of a sphere being its surface multiplied by J of its radius, 
the thermal capacity of a mass of rock equal in bulk to the earth, 
and of specific heat s per unit of bulk is J Ars. Hence is the 
elevation of temperature which a quantity of heat equal to that 
lost from the earth in a year, would produce in a mass of rock 
equal in bulk to the whole earth. The laboratory experiments of 
Peclet ; Observations on Underground Temperature in three kinds 
of rock in and near Edinburgh, by Forbes ; in two Swedish strata, 
O 
by Angstrom, and at the Eoyal Observatory, Greenwich, give values 
of the conductivity in gramme-water units of heat per square 
centimetre, per 1° per centimetre of variation of temperature, per 
second, from *002 (marble, Peclet) to *0107 (sandstone of Craigleith 
quarry, Forbes); and *005 may be taken as a rough average. 
Hence, as there are 31,537,600 seconds in a year, we have A; = ’005 
X 31,537,600, or approximately 15 x 10k The thermal capacity of 
surface rock is somewhere about half that of equal bulk of water ; so 
that we may take s = *5. And the increase of temperature down- 
wards may be taken as roughly averaging 1° Cent, per 30 metres ; 
so, that, D = 3000 centimetres. Lastly, the earth’s quadrant being 
according to the first foundation of the French metrical system, 
about 109 centimetres, we may take, in a rough estimate such as 
the present, r=6 x 10^ centimetres. Hence, 
3/b 3 X 15 X 10^ 30 X 10^ 5 
D7^“3000x 6x10^x*5“10^^x6 “10^ 
This, multiplied by 20,000 x 10®, amounts to 10,000, or to 100 times 
as much heat as would warm 100 times the earth’s bulk of surface 
rock by 1° Cent. 
3. Note on the Atomicity of Sulphur. By Dr Alexander 
Crum Brown. 
We now know a considerable number of elements which exhibit, 
in their compounds, two or more distinct degrees of atomicity. 
Thus we have N triatomic in ammonia and its analogues, in 
