FIG. 141.—Three diagrams to illustrate the sequence of events within the crater of a composite cone during the cone-building and crater-producing periods. _a_ and _b_, two successive stages of the cone building or Strombolian period; _c_, enlargement of the crater, truncation of the cone, and destruction of the upper chimney during the relatively brief crater-producing or Vulcanian period.] Both the older and newer sections of this plug or chimney are furnished some support against the outward pressure of the contained lava by the surrounding wall of tuff; and they are, therefore, in a condition not unlike that of the inner barrel of a great gun over which sleeves of metal have been shrunk so as to give support against bursting pressures. On the other hand, when not sustaining the hydrostatic pressure of the liquid lava within, the chimney would tend to be crushed in by the pressure of the surrounding tuff. Its strength to withstand bursting pressures is dependent not alone upon the thickness of its rock walls, but also upon its internal diameter or caliber. A steam cylinder of given thickness of wall, as is well known, can resist bursting pressures in proportion as its internal diameter is small. So in the volcanic chimney, any tendency to remelt from within the chimney walls must weaken them in a twofold ratio. We are yet without accurate temperature observations upon the lava in volcanic chimneys, but it seems almost certain that these temperatures rise as the Vulcanian stage is approaching, and such elevation of temperature must be followed by a greater or less re-fusion of the chimney walls. The sequence of events during the late Vesuvian eruption is, then, naturally explained by progressive re-fusion and consequent weakening of the chimney walls, thus permitting a radial fissure to open near the top and gradually extend downwards. Thus at first small and high outlets were opened insufficient to drain the chimney, but later, on April 7, after this fissure had been much extended and a new and larger one had opened at a lower level, the draining began and the surface of lava commenced rapidly to sink. [Illustration: FIG. 142.—The spine of Pelé rising above the chimney of the volcano after the eruption of 1902 (after Hovey).] When the rapid sinking of the lava surface occurred, the lower lava layers were almost immediately relieved of pressure, thus causing a sudden expansion of the contained steam and resulting in grand crater explosions. The partially refused and fissured upper chimney, now unable to withstand the inward pressure of the surrounding tuff walls, since outward pressures no longer existed, crushed in and contributed its materials and those of the surrounding tuff to the fragments of fresh lava rising in volume in the grand explosions (Fig. 141 _c_). In outline, then, these seem to be the conditions which are indicated by the sequence of observed events in connection with the late Vesuvian outbreak. [Illustration: