210 



GEORGE CARROLL CURTIS 



denly found vent by eruption through the impounding beds; 

 thus a chimney or vent was formed through which copious sup- 

 plies of water were poured during the intervals of explosion, to 

 be ejected again from the same pipe, 



TJie instigation of eruptioji. — On reaching the hot, covered ash 

 beds in sufficient quantities, the water must bring about geyser- 

 like explosions until the underground heat is dissipated. There 



Fig. 7. — An ash geyser eruption in the bed of the Wallibou river. Soufriere 

 crater in the distance. 



are a number of distinct processes by which this water manages to 

 collect in sufiPicient quantities to instigate eruption. One is the 

 flooding of the stream channel by a sudden tropical downpour, as 

 observed on May 30 ; another is that of an eruption flood from 

 the primary crater — as that from Mount Pelee on June 24. Still 

 a third process, one of gradual accumulation, is that which 

 obtains its water supply by the damming of the stream through 

 landslides. I observed this in the Wallibou river at 8 o'clock 

 on the morning of May 30. A portion of the undercut riparian 

 ash banks had fallen, slipping across the channel, and by thus 

 damming the stream caused a lake to form behind this barrier. 

 When its water had reached and covered a certain place near 



