SOUFKIERE, AND ON A VISIT TO MONTAGNE PELEE, IN 1902. 
439 
around the larger pits, and from their arrangement it was impossible to avoid the con¬ 
clusion that they had been emitted from the funnel around which they lay. The sharp 
cut walls of ash surrounding these depressions showed that they had been blown out of 
the deposits of ashes by a large steam explosion. Usually they were situated on the 
course of the stream (we observed none at any distance from the main river or its 
tributaries), and in one place there were three of them in contact with one another, 
planted on the river banks. They were incomjdete, less than half of the whole circle 
remaining in each case, as the side next the river had been cut away as the stream 
deepened its gorge. 
We did not see any of these pits in operation, but the method of their origin is 
probably as follows :—water in some way gets access to the undermost and hottest 
layers of the ash, probably by the river cutting suddenly into its banks at their base 
and undermining them. A large part of the bank then at once subsides into the 
stream, and when the hot ash touches the water large masses of steam are imme¬ 
diately generated, and the explosion lifts the super-incumbent mass and carries it 
upwards into the air. In some cases many tons of material must have been thus 
projected into space. The finest dust floated up to great heights, and was wafted 
away by the wind ; many of the stones must have been shot up obliquely and have 
fallen on the fields of ash which surrounded the focus of eruption. But a very large 
part of the material subsides again into the funnel-shaped cavity through which the 
steam explosion ascended. It is, however, not sufficient to fill it up, and the basin- 
shajied depression is a measure of the amount of material which was borne away by 
the wind or projected beyond the lip of the funnel. The low, Hat cone around the 
crater is formed liy the sand and stones which fell just outside the edge of the cavity. 
There \vas no evidence that more than one explosion had taken place from each funnel, 
but in more than one place a group of pits was seen which suggested that the same 
process was several times repeated at approximately the same spot. 
It is not likely that they were occasioned by water penetrating through cracks on 
the surface and reaching the hot mass below, for cracks were few and small except 
where landslides were taking place on the stream banks, and in no case were fissures 
seen in connection with the pits. Possibly, however, after heavy rains, springs rising 
in the bottom of the valley may introduce water into the basal layers of the deposit, 
and in this way steam explosions would be produced with exactly similar effects. 
In the valley of the Wallibu Dry Piver we found a low, flat cone some 30 yards 
across at its base, and about 10 feet high, with a large flat central crater on its 
summit, perhaps 15 feet across, ch'cular, some 3 feet deep, with a layer of mud in its 
interior, and around the lip of this crater were three others, smaller but almost equally 
perfect. The rains had somewhat destroyed the structure, a little lake had formed in 
the main crater, and had drained out through a notch on one side. It stood about 
15 yards from a deep narrow gully occupied by a small stream, and not far from 
the base of a high bluff’, so that the introduction of water below the surface might 
