414 E. O. HOVEY— STRIATIOXS AND U-SHAPED VALLEYS 



ble that I have seen being the Larikai and Roseau gorges in the western 

 (leeward) side of the volcano. In the Larikai gorge, about 500 meters 

 from the sea, the bottom of the canyon is crossed by a broad, heavy bed of 

 andesitic lava. In this rock there has been carved the perfectly U-shaped 

 channel that is shown in figures 1 and 2 of plate 40. This is 8 to 10 

 meters wide, 4 or 5 meters deep, and about 50 meters long. 



The explanation of this form of cross-section is hinted at in figure 2, 

 plate 41, which shows an overloaded streamlet in the Larikai gorge depos- 

 iting its excessive burden of sand and gravel in a rock basin at the foot of 

 a fall (see also figure 2, plate 42). This illustrates the tendency of 

 moderate showers to bring down loose material from the steep slopes of 

 the watershed and fill the hollows in the bottom of the gorge (see also 

 figure 2, plate 40, and figure 1, plate 41) and the gorge itself. Another 

 circumstance contributing to the filling of these gorges with loose, angular 

 material is the constant disintegration undergone by the almost vertical 

 bluffs of new ash surmounting the equally steep old ash. During dry 

 weather there is a continual shower of pebbles and sand grains down the 

 faces of the bluffs, building debris cones at their bases, as is shown in 

 figure 1, plate 44. Going back still farther, we know that the eruptions 

 of 1902-1903 threw enormous quantities of lava fragments of all sizes into 

 the gorges and onto the slopes draining into them, and filled more or less 

 completely the radial valleys of the mountain, furnishing vast store of 

 abrasive material for eroding the gorges (figure 1, plate 42, and figure 1, 

 plate 43). 



The bottom of the old gorges being filled to a greater or less extent by 

 one or all of these ways, torrential rains such as are frequent in the tropics, 

 particularly in the rainy season, soak the accumulations of loose ash and 

 gravel past the point of equilibrium and the semi-fluid mass rushes with 

 violence down the slopes and through the gorges into the sea. Figure 2, 

 plate 42, shows such a mud flow falling over a precipice into the sea from 

 the valley next north of the Larikai valley, Saint Vincent. The viscosity 

 of these mud flows and their resultant transporting power was patent to 

 every observer of the Mount Pele and Soufriere eruptions. On my first 

 expedition to Martinique I crossed the gorge of the Riviere Seche, June 

 24, 1902, barely in advance of a torrent of black mud 3 or 4 meters deep 

 that bore along on its surface bowlders 1.5 meters in diameter as if they 

 had been corks. A flood in the Riviere de Basse Pointe, June 9, 1902, 

 brought down from the mountain a huge rounded bowlder 3 meters across, 

 which it left perched on a pier of the railroad bridge near the ITsine 

 Gradis, 4.5 meters above the bed of the stream, after the flood had sub- 

 sided. Such examples of the viscosity of the mud torrents following on 



