LOCAL CHARACTER OP VOLCANIC PHENOMENA. 115 



than the crust. If the hquid were hghter an eruption would be inevitable, 

 and once started would continue until the lighter liquid had all found its way 

 to the surface. If the liquid were heavier, it could no more be erupted 

 than a frozen lake could erupt its waters and pour them over its icy 

 covering. 



Lest these considerations should seem too purely speculative to author- 

 ize us to conclude that lavas cannot be emanations from a general liquid 

 interior or from vesicles holding primordi.al liquid magma, we may turn to 

 other considerations more concrete and bearing more directly upon the 

 point. Volcanic eruptions are very local phenomena. At any given epoch 

 they are confined to a few localities of very small relative extent. They 

 have no general distribution in the sense of a widely-extended and con- 

 nected system. Each volcano is an independent machine — nay, each vent 

 and monticule is for the time being engaged in its own peculiar business, 

 cooking as it were its special dish, which in due time is to be separately 

 served We have instances of vents within hailing distance of each other 

 pouring out totally different kinds of lava, neither sympathizing with the 

 other in any discernible manner nor influencing the other in any apprecia- 

 ble degree. Again, we find vents at high levels and at low levels in close 

 proximity with each other, and both delivering the same kind of lava. The 

 great craters of the Sandwich Islands are remarkable instances of this kind, 

 and indicate that each crater derives its lavas from a distinct reservoir. It 

 is inconceivable that a liquid from a common reservoir could rise and out- 

 flow from the loftier vent while the lower vent remained open. The same 

 phenomenon is exhibited at JEtna and in Iceland and other active volcanoes. 

 Then, too, we have the outpouring of widely distinct kinds of lava from 

 the same orifice at successive epochs, and as a general rule the grander 

 volcanoes present a succession of eruptions marked by different kinds 

 of lava; and it should be noted that these varieties of ejecta are not 

 intennixed nor formed by the commingling of two or more magmas, nor 

 do they present intermediate and transition types, but each coulee has a 

 well-defined character, which serves to distinguish it and assign it to its 

 proper place in the classification. All these subordinate phenomena, and 

 many others which it is needless to mention here, are apparently incon- 



