SOUFRIEEE, AND ON A VISIT TO MONTACINE PELEE, IN 1902. 
429 
the level gTound beneath. The trees where not uprooted and cast down were 
beginning to put out fresh green leaves. The wet and sticky black mud hung in 
festoons on the slope, separated by sharp narrow little rivulets cut by the water (see 
Plate 28, fig. 1). 
Along the sea cliffs and on the banks of the streams excellent sections were 
exposed, both of the layer of new hot ash, and of the older tuffs on which it 
rested (see Plate 25, fig. 1). The deposits of the recent eruption varied from 5 feet 
to about 40 feet in depth, as seen in these sections. They formed irregular hog- 
backed rounded mounds, the long axes of which were roughly parallel to the 
direction of the valley, resembling in many ways the drumlins of boulder clay which 
one sees on the low grounds of Scotland (see Plate 25, fig. 1), These rounded 
mounds of sand bore no very close relationship in their disposition to the pre-existing 
features of the topography; the thickest deposits were not in every case in the old 
stream valleys, but the sand seemed to have been irregularly heaped and not sjjread 
out into a sheet with level upper surface. 
It was not difficult to see that originally the deposits had had a smootli and 
rolling character, though the rains had scored them deeply with a converging 
pattern of stream channels. We saw no evidence, however, of wind-rippled surfaces, 
or of dunes; it looked rather as if a vast mass had been dumped down suddenly in 
great irregular mounds, and then the rain of ashes and of stojies had smoothed 
the surface over. 
In the upper part of the valley of the Wallibu Dry Piver the rolling surface 
of the fields of ash, with rain sculpture everywhere, was very like that seen 
along the shore, but it was only in the main stream channels that the undeilyiug 
tuffs were visible, and the thickness of the recent accumulation could be ascertained. 
It was mostly from 5 to 12 feet deep, and here also it presented little uniformity 
in this respect. 
But the channel of the AVallil)u Ptiver was on the whole more narrow and steep¬ 
sided than that of the Wallibu Dry Biver. It was from 400 to 600 yards wide 
and from 200 to 400 feet deep. Here the deposit was of greater thickness and 
had suffered more severe erosion. Its original depth could not be made out with 
certainty, but at the time we saw it, it was from 60 to 80 feet thick in some places 
(see Plate 29, fig. 2). 
The river flowing tlirough the masses of sand which had been heaped up in its 
channel had cut a deep narrow gorge, sometimes with perpendicular walls. More 
frequently on one or both sides of the stream there was a series of terraces wliich 
varied in breadth from a few feet to 20 or 30 yards. As many as six of these could 
sometimes be seen one above the other (see Plate 29, fig. 1). They had all the marks 
of river terraces cut by water, the highest were always farthest from the stream, and 
their surfaces had a gentle slope down the valley. Where sections of these terraces 
were exposed it was evident that they were eroded out of the thick deposit of new 
