Following such reasoning further, we find: (1) A rending explosion raises some very coarse and heavy material, (2) this material falls soonest and nearest the crater, (3) such falling material would have very high velocity, but would come very quickly to rest on cone slopes of 17°. Its fall would certainly deflect upward discharges of lesser intensity, and its rush down the mountain with such deflected vomitings would have the aspect of an avalanche, sending forward an avalanche blast capable of short-lived but very fierce destruction. A deflected discharge from the crater would be propelled by the same force as the up-blast, and thus become both an avalanche and a down-blast with the energy of the volcano behind it. It seems probable that such a deflected mass of incandescent gravel and bowlders formed the avalanche seen by Anderson and Flett. On this hypothesis, both gravity and deep-seated explosion contribute to the tornado violence of the black cloud, and the displacement of the air before the avalanche resembles that frequently seen in high mountain districts, where snowslips project an air blast capable of leveling forests. Such snow-slides have a tendency to hug the slope, to move very rapidly above and to come to rest quickly below, and such movement was noticed in the eruption which destroyed St. Pierre and in the eruption of Bandaisan. In the latter also, the discharge changed from upward to horizontal, the change being gradual as the upper air became overcharged with solid matter. The other eruptions of Pelée and Soufrière showed the same general characters, and that, of July 9 may be taken as an example of an eruption of first magnitude. Its direction was more to the westward than on May 8, hence the escape of Carbet and St. Pierre. The materials ejected in the fiery avalanche were believed by Anderson and Flett to be the products of exploded molten lava, which welled over the lip of the crater and was blown to shreds by expansion of gases. This view I can not accept, for I have as yet seen no glass fibers or sherds, nor any evidence of new molten lavas in any way connected with these eruptions. The angular bits of andesite and coarser rocks found on the mountain after eruption are pieces of country-rock. In some cases 'bombs' have been formed from old andesites with half-molten surfaces. All the evidence so far goes to show that these volcanoes are great steam crushers, comminuting and distributing ancient volcanic materials.
In conclusion, I wish to record my admiration of the excellent work done by the Royal Society's commission, recalling with pleasure association with Dr. Anderson and Dr. Flett in Barbados and Martinique. Their memoir is by far the most thorough and scientific work that has been published as yet on the West Indian eruptions.
