72 EXPERIMENT'S on WIIINSTONE and LAFJ. 



to have flowed in crevices, we were miftaJcen as to their direc- 

 tion ; for inflead of flowing downwards, I am convinced they 

 have flowed upwards, and that the crevices have performed the 

 oflace of pipes, through which lateral explofions have found a 

 vent. This will appear in the higheft degree probable, when 

 we attend to the known hiftory of volcanic eruptions. It gene- 

 rally happens, that the lava begins to flow from the fummit, in 

 confequence of the crater being filled with liquid matter up to 

 the brim. At that moment the bafis of the mountain murt: 

 be prefl'ed outwards by a very great hydroflatical force, eqvial 

 to the weight of a column of liquid lava as high as the moun- 

 tain itfelf. It is natural then to expedl, that this preflure, 

 aflifted by ftrong percuflions of exploflon, fliould lacerate the 

 body of the hill, and form great rents. The lava, urged up- 

 wards by the fame preflure, would flow through thefe rents, 

 and emerge at the furface with violence. The difcharge would 

 continue through this channel till the propelling force had cea- 

 fed, when the rents would be left full of lava ; which, cooling 

 in that pofition, would produce vertical lavas, fuch as thofe 

 of Somma. This fuppofition is confirmed by various pheno- 

 mena : The lava ceafes to flow from the crater as foon as a la- 

 teral eruption has begun ; when it ruflies with fuch violence 

 from the fide of the mountain as to fly to a great height into 

 the air, like a jet d^iiu ; and it often makes its appearance, 

 in the fame inftant, at various mouths, which are not fcat- 

 tered at random, but placed in one continued line, indicating 

 the difcharge from a rent. Some circumftances likewife, which 

 I obferved on a clofe examination of the vertical lavas, indi- 

 cate that the crevices had performed the office of pipes. Fi^e- 

 quently the fubfl;ance at the middle differs from that on 

 each fide, whilfl the fides i-efemble each other exadly. I explain 

 this, by fuppofing that the lava, which had firft flowed through 

 the pipe, and had coated its fides with folid matter, had been fol- 

 lowed 



