311 
THE ERUPTIONS OF MAY, 1902, AT THE SOUFRIERE IN ST. VINCENT. 
augite, hypersthene, plagioclase and magnetite, very often consisting of granular or 
imperfect crystals closely packed together. 
Dark greenish-brown hornblende is not often seen in sections of these bombs. It 
is always irregular in shape and corroded; the other ingredients are iron oxides 
(titaniferous magnetite), pyrites and apatite. 
The groundmass is of the hyalopilitic type, consisting of felspar and augite 
microlites, with magnetite embedded in a dark brown glass. The felspars are mainly 
andesine, as stated above. The pyroxene is often in long narrow prisms. Hypersthene 
has been recorded in the groundmass of many of the Martinique rocks, but in 
St. Vincent, where the microlites are large enough to react clearly on polarised light, 
they have mostly an oblique extinction, and cross-sections may show twinning on a 
pinacoidal face ; hence they belong to augite. The iron oxides form small octahedra. 
The glassy base varies somewhat in abundance, and is occasionally nearly free from 
crystals of the second generation. These rocks are always vesicular, and their steam 
cavities are empty. 
A number of excellent analyses of the dust and bombs of the eruption of the 
Soufriere in May, 1902, have appeared. Of these we select the following as the most 
complete. The great similarity is at once apparent. The first analysis probably 
represents most closely the composition of the magma in the later stages of the 
eruption. The second analysis contains an abnormally high percentage of water, 
which leads us to suspect that the specimen was not a recent bomb, but an old and 
somewhat weathered one. No other analysis of the recent products of the Soufriere 
or Pelee shows so much combined water, except one which Professor Lacroix* suspects 
for precisely the same reason. Analysis IV. of the dust that fell at Chelston, Barbados, 
has been criticised by Dr. Hillebrand,! who doubts whether nickel and cobalt are 
present in these rocks. Repeated tests, however, both in the laboratory of the 
Geological Survey of Great Britain and by other analysts to whom samples were 
sent, confirm the substantial accuracy of the analysis in this respect, though the 
amount of cobalt and nickel may be very slightly over-estimated. The curious 
feature of this analysis is that it proves the Barbados dust to contain less silica and 
alkalies and more magnesia, lime and iron oxide than the average magma of the 
eruption. In other words, the glass and acid felspars were not deposited in Barbados 
in normal amount, but were swept past it by the anti-trade wind. Analyses V. and VI. 
of ash collected 150 miles further away show, as might be expected, exactly the 
converse. This brings out clearly the sifting action of air currents on clouds of 
volcanic dust. 
Professor Lacroix| has computed that Pisani’s analysis shows that the rock, it 
completely crystallised, would have yielded :— 
* ‘ La Montagne Pelee,’ p. 527. 
t ‘National Geog. Mag.,’ XIII., p. 299, 1902, 
1 ‘ La Montagne Pelee,’ p. 599. 
