278 DR. TEMPEST ANDERSON ON THE ERUPTIONS OF THE SOUFRlRRE IN 
In 1907 almost the whole of this ash had been washed away, but a fragment 
remained in the shape of a terrace about 60 to 80 feet high, which showed the height 
to which the valley was originally filled (see Plate 10). It is situated on the north 
side of the valley, just below the abrupt turn to the west above mentioned, and is in 
situ except for a few partial landslides; the ash of which it is formed is unstratified 
and contains very few ejected blocks or fragments of any kind. The round hill 
behind it is the same as one shown to the left of the 1902 photograph in the middle 
distance. It was impossible to get a photograph from exactly the same position as 
before, since the ash bank on which the camera stood in 1902 has been all washed 
away, and the Wallibu plateau adjacent, when tried from several directions, proved 
inaccessible. The camera in 1907 was therefore placed on the floor of the valley in a 
similar position, but 60 to 80 feet lower. This floor, or more strictly terrace, as 
mentioned below, is all composed of water-sorted material, chiefly gravel and coarse 
sand, but with a good many blocks as big as a man’s head. They represent ejected 
blocks and fragments of lava derived partly from the ash of 1902, and partly from 
older beds, the fine ash in each case having been washed away. The surface of the 
gravel bed showed marks of quite recent running water, and Mr. Duncan Mac¬ 
donald, who knows the place intimately, stated that during the last winter, 
1906-07, the river ran along the foot of the north bank of the valley. When 
examined in March, 1907, it ran along the south side of the valley, and had already 
in those few months excavated a new channel about 30 feet in depth, as shown in 
Plate 11, which was taken from the top of the valley floor or terrace, the surface of 
which is shown in the previously mentioned Plate 9. The stratification as exposed 
in the side of this new valley is very distinct, and the sorting b}^ water, mentioned 
above, is very evident. This gorge appears to have been excavated when the river 
was in flood, and as the water subsided and the excavating power became less, its 
channel in the bottom of the gorge became more tortuous, and it formed small 
meanders and left small terraces on its sides at different heights. This plate also 
shows the precipitous and inaccessible character of the north side of the valley where 
it is formed of the Wallibu plateau till it gets lower as it approaches the sea, and this 
is a characteristic specimen of the early stages of denudation in these ash beds. 
The south side, on the contrary, is sloping, and its outline is characteristic of a 
more mature stage of denudation. This sloping character has enabled more numerous 
terraces of the new deposits to survive. Some low down by the side of the river 
consist solely of water-sorted material, while others show patches of ash in its 
original condition higher up the slope. In one or two cases it is possible to compare 
the two in juxtaposition (Plate 12, fig. 1), in which case the lower terrace of water- 
sorted material stands almost vertical where it has been undermined by the river in 
two stages ; while behind it a mass of ash in situ shows most beautifully the marks 
of rain-water running over and down the surface. Further away again, the slopes of 
the old valley show denudation in a more mature stage. 
