OF MONTE SOMMA AND VESUVIUS. 107 
decrease of inclination, not suddenly, of course, but by gradual 
thickening of the soft tufas forming their sides or walls. ‘They, 
naturally having a less inclined base, have been able longer to resist 
the destructive action of flowimg water, and therefore the sum of 
their components often reaches very nearly a maximum of 100 metres, 
no insignificant thickness. The impulse of the torrent collected 
from the rivulets above, and rushing down a slope, first roughly at an 
angle of 35°, and here often 25° to 30°, can be well understood to 
cut its way through these soft beds. As the bottom is excavated, 
the soft sides crumble in, giving the valley a V-shaped section ; and 
when this reaches deep enough to encounter the underlying lavas, a 
new type of erosion results. 
Let us suppose we have to deal with lava-beds regularly interla- 
minated with scorize and other fragmentary materials, and dipping 
in the direction of the flowing water. The torrent will find some 
weak place in the uppermost lava-flow, and having broken up the 
sheet of rock at this point, will produce a miniature cascade. 
The impulse of the water bounding over this rocky step will remove 
the subjacent incoherent scoria, leaving the lower edge of the open- 
ing in the lava unsupported, so that the torrent can act upon it 
below and behind, conditions very favourable to its being broken up. 
At the same time the lip of the cascade recedes by loss of material ; 
but as its inclination is upwards and backwards, it correspondingly 
increases the height of the waterfall, so that profound precipices may 
result, which are often very abundant, forming, as it were, so many 
steps along the valley. in many cases the opening in the upper- 
most lava-stream is very long in the direction of the valley, but 
very narrow, whereas at its bottom it is dilated, forming, as it were, 
half of a pointed dome. 
These cascades often cut through many lava-streams. Lyell de- 
scribes the one near the Fontana Olivello, which shows in section 
as many as six. In another I have counted thirteen of those 
streams. 
The number of cascades that may occur, with their depth, may be 
judged from the measurements taken in the Vallone Cancherone. 
Beginning near where the 450-metre contour-line crosses the valley, 
we have in less than a kilometre no fewer than six of them, giving a 
perpendicular fall inclusively of 894 metres, divided as follows from 
below upwards, 102, 29, 2, 19, 9, 20 metres. Again, in the Vallone 
Sacramento, in the main trunk, before its bifurcation, both arms of 
which are rich in profound precipices and cascades, we have at the 
560-metre contour-line a cascade of 7 metres. 
6038 99 99 99 3 29 
615 29 99 99 Ai 99 
640 32 29 9? 14 33 
655 99 39 29 13 39 
Thus out of a total descent for the water of 95 metres, nearly a 
half, or 41 metres, occurs by perpendicular falls. 
From the peculiar narrowness of the gap in the shedding lava- 
