pahoehoe corresponds practically with the Fladen lava of German vulcanologist's, and the aa .with their Schollen lava. Rugged flows

are knoVn in Auvergne as cheires.

The surface of a clinker-field

has often a horribly jagged character, being covered with ragged blocks bristling with sharp points.

In the case of an obsidian-flow

a most dangerous surface is produced by the keen edges and points of the fragmentary volcanic glass.

If, after a stream of lava has become crusted over, the underlying magma should flow away, a long cavern or tunnel may be formed. Should the flow be rapid the roof may collapse and the fragments, falling on to the stream, may be carried forward or become absorbed in the fused mass. The walls and roof of a lava-cave are occasionally adorned with stalactites, whilst the floor may be covered with stalagmitic deposits of lava.

The volcanic stalactites are slender,

tubular bodies, extremely fragile, often knotted and rippled. Beautiful examples of lava stalactites from Hawaii have been described by Professor E. S. Dana. Caverns may also be formed in lava-flows by the presence of large bubbles, or by the union of several bubbles. It may happen, too, that certain monticules thrown up on the surface of the lava are hollow, of which a famous example is furnished by the Cavcrne de Rosemond, at the base of Piton Barry, in the Isle of Rctinion.

It is of great interest to determine whether molten lava contracts or expands on solidification, but the experimental evidence on this subject is rather conflicting. According to some observers a piece of solid lava thrown on to the surface of the same lava in a liquid state will sink, while according to others it floats. It has often been

observed that cakes formed by the natural fracture of the crust on the lava of Kilauca sink in the liquid mass, but it has been suggested that the fragments are drawn down by convection-currents. On the other hand a solid piece, though denser than the corresponding liquid, may be buoyed up for a tinie by the viscous condition of the molten lava. Moreover, the presence of minute vesicles may lighten the mass. Although the minerals of a rock-magma may separately contract on crystallization it does net follow that the magma itself, in which they probably exist in a state of solution, will undergo on crystallization a similar change of volume. On the whole, however, there seems reason to believe that lav? on solidifying almost always diminishes in volume and consequently increases in density. According to the experiments of C. Doelter the specific gravity of molten lava is invariably less than that of the same lava when solid, though in some cases the diff^erence is brit slight. In a vitreous or isotropic condition the lava has a lower density than when crystalline. The differences are illustrated by the following table, where the figures give the specific gravity:-

^