gases, finally rush down the mountain slopes in whatever directions they are guided by vale and ridge and wind. They were guided into the cliff-hollow of St. Pierre at Martinique and Walliabou at St. Vincent, both to leeward of the craters. In St. Vincent there was also an outrush symmetrical to the great crater in all directions, windward as well as leeward; hence the destruction of the windward estates. That there was tornado action is proved by the frightful demolition of masonry and the bending of trees; in St. Pierre this is away from the mountain and curving from south-southwest to southwest as we go southward from the volcano, these directions being shown by the downbent trees.
A common question is 'How were the people killed? Was it some strange gas? Were many killed by lightning?' There is no need for calling in any unusual gas; no doubt there were several gases present and the combustion of tropical vegetation made others that were highly explosive if mixed with air. This may account for some flame explosions reported as coming in from the sea-front at St. Vincent. The people of St. Pierre were killed by steam, hot dust, falling stones, falling buildings, drowning, burial alive and burning. The heat of a burning city fanned into a 'whirlwind of fire,' killed all who were left over from the other sources of death. The little city was as an anthill to a forest fire, in the presence of the terrible earth forces at work.
In conclusion let me say a word about the place of these eruptions in the geological history of the islands. They are not at all different from what the old cliffs, carved by the sea along the shores, show to have been the source of the heaping up of all the rocks that make the foundation stones the hills of these islands are carved upon. For these old sections are what geologists call agglomerates—masses of volcanic fragments, large and small, bedded in gravel and mud. Just such agglomerates are made anew by the great banks of dirt the volcano has vomited into the Walliabou at St. Vincent or Pelée into the gulch of the Factory Guerin. There are also old lavas interbedded with the agglomerates, showing that frequently the eruptions have been concluded in the past by the flowing out of molten incandescent rock. Perhaps this may come now, in the near future; even as I write there are cable despatches describing glowing lavas at Mont Pelee. The bared slopes of the Soufriere, devoid of vegetation, reveal a topography and geological structure so exactly like many things in our own Rocky Mountains that it is difficult to believe that one is in the tropics. There, too, in the tuff beds of the Yellowstone Park, is a fossilized tropical verdure which gives evidence that when those volcanoes of Wyoming were active, their slopes were covered by the vegetation of a warm climate near the level of the sea.