The plug, like the pipe up which it has risen, is in general irregularly circular in ground-plan. It may be conceived to be a column of rock, descending to an unknown depth into the interior, with a casing of pyroclastic debris surrounding it. It may vary considerably in the proportion which its cross-section bears to that of the surrounding fragmental material. Sometimes it does not occupy more than a small part of the whole, often appearing in the centre. In other cases, it more than equals all the rest of the material in the vent, while instances may be noted where only occasional patches of tuff or agglomerate are visible between the lava-plug and the wall of the pipe. From these we naturally pass to the second type of vent, where no fragmentary material is to be seen, but where the chimney is now entirely filled with some massive once-molten rock. A neck with a lava-plug probably contains the records of two stages in volcanic progress, the first of which, indicated by the tuff or agglomerate, was confined to the discharge of fragmentary materials; while the second, shown by the lava-plug, belonged to the time when, after the earlier explosions, lava ascended in the vent and solidified there, thus bringing the eruptions from that particular orifice to an end. Where a small central column of lava rises through the tuff, we may suppose that the funnel had been mainly choked up by the accumulation in it of ejected detritus, which was compacted to a solid mass adhering to the wall of the funnel, but leaving a central orifice to be kept open by the gradually waning energy of the volcano. By a final effort that impelled molten rock up that duct and allowed it to consolidate there, the operations of the vent were brought to a close. Where, on the other hand, only occasional strips of tuff or agglomerate are to be found between the lava-plug and the wall of the pipe, the last uprise of lava may be supposed to have been preceded by more vigorous explosions which cleared the throat of the volcano, driving out the accumulated detritus and leaving only scattered patches adhering to the sides of the funnel. There is, no doubt, some downward limit to the production of fragmentary material, and if we could lay bare successive levels in the chimney of a volcano we should find the agglomerate eventually replaced entirely by lava. The materials of the lava-plugs vary widely in composition. Sometimes they are remarkably basic, and present rocks of the picrite or limburgite type; in other cases they are thoroughly acid rocks such as felsite and granophyre. Many intermediate varieties may be found between these extremes. It is noteworthy that, in districts where the lavas erupted to the surface have been andesitic or basaltic, the material which has finally solidified in the vents is often more acid in composition, trachytic rocks being specially frequent. [Illustration: Fig. 27.--Section of agglomerate neck (_a_ _a_) with dykes and veins (_b_ _b_).]