Volcanic Disturbances in West Indies 



297 



the other from Soufriere, on St Vincent. 

 The distinct characteristics are these : 

 Higher silica and potash, with, in gen- 

 eral, lower alumina, total iron, lime, 

 magnesia, and titanium in the ejecta 

 from Pelee. The difference in titanium 

 seems to be particularly marked and 

 characteristic. There are, however, 

 other differences, of which the most 

 striking is the almost complete absence 

 of sulphur in the Pelee ejecta, while in 

 those from the Soufriere it is a very 

 marked constituent, both in the sulphate 

 and sulphide conditions. Another 

 analysis of Soufriere dust from near 

 Georgetown, which I myself have made, 

 but do not feel at liberty to make public 

 yet, emphasizes this last distinction in a 

 much more marked manner than any of 

 those here published. 



Samples 1,2, and 3 represent the ma- 

 terial ejected from Pelee at certainly two 

 different eruptions, and taken from 

 rather widely separated points. The 

 first is tolerably compact, only slightly 

 vesicular pumice, which still retains 

 much of its original water, while the 

 other two are lapilli and dust which 

 have lost much more of their water, pre- 

 sumably because of higher temperature 

 and finer comminution. No. 3, col- 

 lected on the Potomac, at Fort de France, 

 on the 20th of May, about 20 miles 

 away from the seat of eruption, shows 

 the same effects of sifting by transpor- 

 tation through the air that No. 6 does 

 in the case of St Vincent dust — that is, 

 it probably contains more of the origi- 

 nally- molten constituents of the magma, 

 which would be higher in silica and 

 alkalies than the mass as a whole. Mr 

 Steiger's tests on the dust from the 

 steamship Coya show that it not only 

 contains oxidized sulphur, but also 

 sulphur in the state of sulphide. My 

 own as yet unpublished analysis of dust 

 from Georgetown shows the same, and 

 that the sulphide is not pyrite, but one 

 which is readily soluble in hydrochloric 

 acid with evolution of hydrogen sul- 



phide. When separated from other con- 

 stituents its hydrochloric solution reacts 

 for iron. It is therefore doubtless a 

 sulphide of iron, and I regard it as 

 probably pyrrhotite. There is no posi- 

 tive evidence as yet, however, that it 

 may not be FeS. My analyses had to 

 be made in such haste that no oppor- 

 tunity has as yet offered for a nearer in- 

 vestigation of this interesting point, the 

 true composition of the iron sulphide. 



I may here mention an analysis of 

 dust collected at sea off St Vincent, on 

 the S. S. Louisianian, an analysis of 

 which, by Professor Carmody, appears 

 in the Trinidad Mirror oi May r 22. For 

 a copy of this analysis I am indebted, 

 through Dr Diller, to the Weather Bu- 

 reau of the Department of Agriculture. 

 It is not stated in a form which can be 

 compared with those in this Magazine, 

 except as to the water-soluble compo- 

 nents. These are exactly as Mr Steiger 

 found them to be in the dust from the 

 S. S. Coya, calcium and the sulphate 

 ions largely predominating over alkali 

 and chlorine ions. 



Now let us consider the analyses, 

 which I have denominated a, b, and c 

 (page 296), in the light of the evidence 

 thus far accumulated. 



It would seem as if the soda deter- 

 mination of b of this analysis must be 

 faulty, for of all the other analyses it is 

 the only one which shows less than 2 z /i 

 per cent, while six show from 3.23 to 

 3.79 per cent of soda. If the potash 

 fell off correspondingly, this suspicion 

 might not be justified ; but it does not. 

 It is of the same order as the potash of 

 all the other analyses that show rela- 

 tively low silica. To assume that the 

 figure given for soda in b is correct, 

 means that we have here a volcanic pro- 

 duct markedly different in composition 

 from all the others, which, taken as a 

 whole, are essentially alike. This is 

 highly improbable, and I feel little 

 doubt that the value in question is erro- 

 neous. In most other respects the anal- 



