VOLCANIC EKUPTIONS ON MARTINIQUE AND ST. VINCENT. 341 



land controlled, in a nieaisure, the direction taken hy the xolcanie 

 wind^;. The presence of 'a partially encircling- ridge or soninia, on 

 the northeast side of the volcano, seemingly accounts for the escape 

 from destruction of a narrow fringe aT)Out the northeast l)orderof the 

 island. The outward direction that the ])last took from the mountain, 

 its decreasing intensit}' with increase in the distance it traveled, and 

 the absence of even hypothetical subcratei's. all Ix^ar witness that Ihe 

 heavily dust and stone charged steam from the old crater near the 

 sunnnit swept downward and outward with hurricane force, in a 

 similar way to the more localized blast from Mont Price which 

 destroyed St. Pierre. 



The one conspicuous feature of Mont Pelee which diliers from any- 

 thing on La Soufriere is the presence in the southwest })ortion of its 

 active crater of a deep notch — the Fente or Terre Fendue which, as 

 stated by lleilprin. has been a conspicuous feature of the mountain 

 since the eru})tion of ls5I. and may have existed pre^'ious to that 

 event. This cleft is in plain \iew from St. Piin're, and during my 

 visit to the dead city one could look into it and plaiidy sec^ tlie ruddy 

 cone of eruption with its ascending steam column that was being built 

 within the crater. The area rendered desolati^ l>y the hot blast from 

 Mont P(dee on May 8, and again swept ov<m' by a similar blast on May 

 -!0, is fan-shaped, the apex of tln^ triangle being essentially at the 

 sunnnit of the mountain. The coincidence Ix'tween the position and 

 direction of the Fente and the apex of the expanding Aolcaiiic ])lasts 

 may well be considered signiticant. From the various accounts of the 

 eruptions of Mont Pelee a\'aila]>le, it now seems e^•ident that the 

 blasts which destroyed so nuich of the vegetation of Marlini(|ue and 

 wrought ha\'oc in St. Pierre ctune fi'om the crater with a (h^'ply 

 notched rim, and that the direction taken by the blasts, at least on 

 May 8 and May i?»», was determined by that rift in the crat(>i"s rim. 



As stated by T. A. Jaggar, the downward blasts from xolcanoes do 

 not require a horizontal iiozzle to project them. "'They are simpl}- 

 the result of the down ])last after the heavy gravel has begun to fall, 

 acting against the upl)last from the throat of the volcano, and both 

 together deflected and thrown into terrific whirls or tornadoes." This 

 explanation, although briefly stated, ma}' seemingly ]h\ taken as the 

 leading cause of the downward sweep of the steam chai'ged with rock 

 fragments on both Mont Pelee and La Soufriei-e. It does not seem 

 clear, however, that the down blasts occur only after a towering 

 colunm of debris-charged steam has reached a great height and the 

 fall of the hea^Mer material within it has l)egun. Then, again, it may 

 be ask(Hl why it is that every strong ei'uption is not followed l)\' a 

 down blast. 



Variations in 'the character of ^•olcani(■ eruptions of the ty]>e under 

 consideration occur on account of variations in the eneri-'v of the 



