[ 6o 5 3 
ginal dimenfions by the heat. As this will be brought 
about gradually, whilft the matter fpreads itfelf^or 
grows hotter, the parts over the fire will be gradually 
raifed and bent ; and this bending will, for fome 
time, go on without any other confequence ; but, as 
the fire continues to increafe, the earth will at laft 
begin to be raifed fomewhat beyond the limits of it. 
By this means, an annular fpace will be formed at the 
edges next to the fire, and furrounding it, a vertical 
ledtion of which fpace, through a diameter of the 
fire, will be two long triangles, the fiiortefi: fide or 
bafe of each lying next the fire, and the two longer 
fides being formed by the upper and lower firata, 
which will be feparated for a confiderable extent, 
proportionably to the diftance through which they 
are raifed from each other * *. This fpace will be 
gradually 
likewife; but the matter of fubterraneous fires is yet much more 
extended, than thofe bodies which are only capable of being melted 
into a folid glafs, if we may judge of it from what we fee of vol- 
canos ; for the lavas, fciari, and pumice ftones, thrown out from 
thence, even after they are cold, are commonly of much lefs fpe- 
cific gravity, on account of their porous fpongy texture, than the 
generality of earth, ftones, &c. and they frequently are evert 
lighter than water, which is itfelf lighter than any known foflil 
bp dies, that compofe ftrata in their natural ftate. 
* In Fig. 4, A is fuppofed to reprefent a vertical fe&ion of the 
matter on fire ; B B, parts of the fame ftratum yet unkindled ; 
C C, the two fedlions of the annular fpace, (furrounding the fire) 
which is fuppofed to be filled with water, as far as the ftrata are 
feparated ; D, the feveral fijts of earth, ftones, bV. lying over 
the fire, which are raifed a little, and bent, by the expanlion of 
the matter at A. As it is not eafy to reprefent the things above 
defcribed in their due proportions, it may not be amifs, in order to 
prevent the figure here given from mifleading the reader, to give 
fome random meal^res of the feveral parts, fucb as may probably 
4 1 2 approach 
