520 
DRS. T. AXDERSOX AND J. S. FLETT OX THE ERUPTIOXS OF THE 
contained a higher percentage of glassy matter than that which was gathered on the 
island. From their freshness, their idiomorphism, and other characters, the crystals 
were clearly formed in a fluid magma, and the glass may be taken to represent the 
still liquid material at the moment of eruption. If so, it seems clear that such a rock 
could never have formed a pumice. It is too highly crystalline, and contains too 
little glass. When the steam separated out in little bubbles, which expanded and 
exjmnded as the retaining pressures diminished, a time arrived when the fluid part of 
tlie magma had j^assed into a spongy froth, in which innumerable crystals occupied 
the walls between the vesicles, and on further expansion the mass could no longer 
hold together, but passed into a mist of droplets, most of which lield a crystal in their 
centre, while some consisted mostly of the glassy material. 
Snell a magma, witli its high percentage of solid inextensible crystals and its 
small proportion of fluid rock, when the steam witliin it began to expand, very soon 
passed beyond its limits of cohesion, and was rent into separate particles, each of 
winch was in most cases a crystal surrounded by a film of glass. The crystals 
themselves v^ere uffl. only incapable of extension, they v'ere also anhydrous, and acted 
as passive ingredients; only the liquid magma contained imprisoned water, and the 
immense amount of steam developed in tlie explosions is all the more remarkable 
when we reflect on the relatively small proportion of the substances which held it in 
solution. 
We may infer from its highly crystalline condition that the magma was also at a 
compai'atively low tenq^erature. The minerals have the characteristics found in those 
of volcanic rocks, and were not formed at very great depths or under such jDressures 
as determine the |)roduction of rocks with plutonic structures. The molten mass was 
lying in the conduits of the volcano, cooling gradually tliere, and the separation of 
the first crop of crystals proceeded no douht for a considerable period and under 
exceptionally favouralfle circumstances. Had not the subterranean pressures 
increased and driven the mass iq^wards, and the temperatures at greater depths 
l)een so high as to render the ascending forces irrepi'essihle, the iqjper part of the 
lava column might in a short time have completely solidified. 
In all probability crystallization was more advanced near the surface than in the 
deeper j^arts, and the lower portion of the magma was more completely fluid than 
that which was ejected as the great l)lack cloud. In this regard it is important to 
remember that the dust avalanche is always the first product of the crowning stage 
of the eruption. As soon as the obstruction in the orifices of the volcano has been 
overcome the cloud wells forth. It is the upper part of the ascending column of 
molten inck. We may almost say that the great hlack cloud is the froth that is 
blown off the surface of the subterranean reservoirs. When it is once over everything, 
so far as we know, goes on in the usual manner. Bubbles of steam arise, burst, and 
* J. S. Htli-EH, ‘ Xational Geographic Magazine,’ vol. xiii., p. 293. 
