Large crystals are, it is true, very often found in lavas, but these were formed before the ejection of the mass from the volcano. Such crystals frequently contain enclosures of glass, which indicate that the crystallization went on while the surrounding mass was still fluid. The edges and angles of these crystals are often corroded by the action of the melted portion of the lava, and the motion of the stream often cracks them. These facts go to prove that the large crystals were complete when the lava, as a whole, was still fluid and in motion. Stromboli ejects great numbers of perfect crystals of augite, which must have existed in the molten lava of the vent. The lavas which contain large crystals embedded in a fine stony or glassy base are said to be of a porphyritic texture.

It is important to remember that all these various textures may be found in one continuous rock mass, and bear witness to the circumstances under which each part cooled and solidified. These textures also recur again and again in ancient rocks and enable us to determine their volcanic origin. The processes of rock destruction and removal have in many cases laid bare deep-seated masses which were plainly once melted like true lavas, but which have cooled very slowly and under great pressures. In such rocks the texture is usually coarsely crystalline and shows no traces of glass or scoriae. Between the surface lava flows and such deep-seated reservoirs every form of transition may be traced, often in continuous rock masses.

Where several successive lava flows issue from one vent, at intervals which allow one stream to be consolidated before the next is poured out over it, a rough bedding or stratification results, each flow being perfectly distinguishable when seen in section. Deceptive resemblances to the true stratification of sedimentary rocks may thus arise, especially when the exposed section is short. But the wedge-like form of the sheets, the absence of bedding within the limits of each flow, and the nature of the rock itself, always enable us to distinguish these masses from the sediments which have been stratified by the sorting power of water.

A mass of lava, when it cools and solidifies, necessarily contracts, and since the cohesion of the mass is insufficient to allow it to contract as a whole, it must crack into blocks, separated by fine crevices, which are called joints. The mutual relations of the jointing planes, and the consequent shape of the blocks, are determined largely by the grain of the lava and its degree of homogeneity. In fine-grained (and some coarse-grained) homogeneous lavas the jointing is apt to be very regular, and to give rise to prismatic or columnar blocks, which are usually hexagonal. This shape is due to the fact that the formation of hexagons requires less expenditure of work than other figures, and is produced by the intersection of systems of three cracks, radiating from equidistant points at angles of 1200. The long axes of the prisms are at right angles to the cooling surface. Starch and fire-clay, which shrink on drying, joint in the same way. The coarser and more heterogeneous lavas usually break up into blocks of irregular size and shape.

Fig. 25. - Stream gorge, island of Hawaii; displaying modern columnar lava.

(Photograph by Libbey).

It must not be inferred that the joints of all rocks are due to shrinkage on cooling. It will be shown in a subsequent chapter that such is very far from being the case.

Not all the lava produced in and around a volcanic vent can reach the surface. Some of it may be forced horizontally between the beds of the surrounding rocks, thus forming intrusive sheets, which, when exposed in section, may be readily distinguished from surface flows by the fact that they have consolidated under pressure, and hence have no slag or scoriae associated with them. Other portions of the lava will fill up vertical fissures in the volcanic cone or in the underlying rocks, and, solidifying in these fissures, form dykes. Such a fissure, twelve miles in length and filled with molten lava, was observed by Sir Charles Lyell in the neighbourhood of Aetna. In the great eruption of Skaptar Jokul (Iceland) in 1783 lava was poured out at several points along a line two hundred miles long, and doubtless this was a great lava-filled fissure which consolidated into a dyke.

Fig. 26. - Obsidian Cliff, Yellowstone Park. Hexagonal jointing. (U.. S. G. S).

We thus see that the molten masses may not all well up through the crater of a volcano, but will seek egress along the line of least resistance, wherever that happens to be, breaching the walls of the volcanic cone, rising up through vertical fissures, or forcing their way as intrusive sheets between the beds of preexisting rocks. In these various situations the different rates of cooling produce many varieties of rocks, though the original molten mass may have been nearly or quite identical in all of them.

Lavas which flow into the sea from a terrestrial vent, or are poured out from a submarine one, show, as a rule, but little difference from those which solidified on land, because the rapid formation of a cindery crust will protect the hot lava from contact with the water. Sometimes, however, the sudden chill will cause the lava to disintegrate into a mass like black sand.

The lavas which flow from a given vent do not always remain constant in character and composition, but change at successive periods of activity. It has frequently been observed, for example, that a series of lavas, at first intermediate in chemical composition, then acid and finally basic, have been successively ejected from the same volcano. It does not appear, however, that there is any definite law of succession in the kinds of lava emitted.

It should also be noted that neighbouring vents may simultaneously produce lavas of different composition. Thus, in the Lipari Islands, the lava of Stromboli is basic, while that of Vulcano is highly acid.