184 Jounsron-Lavis—The Eruption of Vesuvius in April, 1906. 
others it will rise up for 500 or 1000 m., and then, meeting with a strong current, 
be carried away as a horizontal band in the sky. ‘There is no doubt, I think, 
that such a strong south-west wind was blowing, although not so near the ground 
as to be within the reach of the meteorologist’s observations, and, therefore, not 
the imaginary one of De Lorenzo and other writers. This wind swept the ejecta 
away to the north-east over the whole of Italy in that direction and across the 
Adriatic. As the stones fell from the upper south-west current into that of the 
north-east, they must have returned over part of their route. When they 
reached Ottajano, they struck the window-panes directed to a north-east aspect. 
Unquestionably, there must have been a tendency for a vorticose inrush of air 
all round the voleano; but this would obviously be less marked on the side of 
Somma than on the other unobstructed sides of Vesuvius. 
To sum up, we may attribute the peculiar distribution of the scoria to 
different concurrent influences—(a) Heterogeneous structure of cone, unsymme- 
trical crater, and oblique projection; (2) two strata of air-currents in opposite 
directions. 
Immediately after the eruption the edges of the crater were so dangerous, 
and the sides so plastered with the last expiring puffs of dust and sand, that very 
little could be made out of the structure of the cone. In November the gradual 
peeling of the walls, helped by some heavy rains, left the surface clean. Many 
dykes could be distinguished, mostly in the vertical, or nearly vertical, radial 
orientation; but a few, obviously not lava-streams, were horizontal. I tried, as 
on former occasions, to photograph with a telephotic lens, but the unstable 
and dusty state of the atmosphere within the crater prevented the operation 
from being successful. 
In fig. B(a) was a dyke on the north wall of the crater, perhaps some 30 m. from 
the crater-edge. Its upper limit was hidden by a large volume of steam escaping 
from its upper canal. It is a most interesting example of the hollow dykes, the 
nature of which I was the first to describe in referring to the eruption of 1885. The 
two canals are divided by a floor, and the lower one is nearly filled by loose scoria. 
This dyke may possibly be that of 1891, and records somewhat the following 
sequence of events. The dyke was formed and reached the outer surface of the 
cone; lava flowed through the rift. This went on until most of the fissure was 
filled by the solidifying rock, only a channel (4’, A’, a®*) remaining, where the flow 
was greatest, and probably just above the level of the outlet on the cone. The lava 
in the chimney went on draining away, till its surface was reduced below the lateral 
outlet, and it then stopped, and what remained in the lower part of the rift cooled 
a crust on its surface between a' and a*, A fresh outlet was formed on the cone 
at a lower level, and drained the lava to a much lower level, emptying the part of 
the dyke at a’, except its congealed walls. Then came closure of the outlet on 
