202 GEORGE CARROLL CURTIS 



hours of very heavy rain. Two days later I reached the spur of 

 the "St. Pierre ridge" which trends across the "Terre Fendue," 

 or lip of Pelee's crater. It was possible from this position to see 

 directly into the crater. During a minor outbreak, the channel 

 of the gorge began to send out vapor, while its water both visibly 

 and audibly increased. This I interpreted as erupted water. 

 On May 8, according to M. Desere, mayor's deputy at Grande 

 Riviere, at 4 a. m. — four hours before the destruction of St. 

 Pierre — a great flood, bearing bowlders up to six feet or more 

 in diameter, came rushing down and buried the lower part of the 

 town some twelve feet, where the stream formerly debouched 

 into the sea. There had been little rain before this flood. Similar 

 floods took place at this village of Grande Riviere on May 11, on 

 June 6 — the day of perhaps the heaviest eruption of Pelee on 

 record — and on June 22 (when I was about two miles off the 

 delta of the Grande Riviere, in a sloop, though unaware of such 

 occurrence). Little rain fell on this day. The towns to the 

 southward, situated at the mouths of streams heading on Mount 

 Pelee, were visited by similar catastrophes. 



From these accounts it appears that the major eruptions have 

 been accompanied or preceded by "mud flows," and evidence 

 indicates that the waters of the crater lakes have been discharged 

 over the mountain sides. If fragmental material has accompanied 

 these ejections of water, as it has the other eruptions, it is reasona- 

 ble to assume that much of it would have been carried down into 

 the lower valleys. The fine dust which was observed to be accu- 

 mulating within the crater of Pelee would be swept down during 

 such a flood, together with much of the fragmental material lying 

 upon the inner cone. Re-ejected material, both coarse and fine, 

 which had fallen within the crater might be either washed down 

 the inner cone and out through the lip, or in a stronger outburst 

 deluged over the upper slopes of the outer cone to be distributed 

 by the summit-heading valleys. Dr. E. O. Hovey has reported 

 these " mud flows" as due to landslides.' While this may account 

 possibly for some of the minor occurrences, it can explain with 

 difficulty the great volume of water necessary to convert the 



-"Bull. Am. Mus. Nat. Hist., Vol. XVI. 



