Though originally the vesicles and cavernous spaces, blown open by the expansion of the vapours dissolved in molten lava, remained empty on the consolidation of the rock, they have generally been subsequently filled up by the deposit within them of mineral substances carried in aqueous solution. The minerals thus introduced are such as might have been derived from the removal of their constituent ingredients by the solvent action of water on the surrounding rock. And as amygdaloids are generally more decayed than the non-vesicular lavas, it has been generally believed that the abstraction of mineral material and its re-deposit within the steam-vesicles have been due to the influence of meteoric water, which at atmospheric temperatures and pressures has slowly percolated from the surface through the cellular lava, long after the latter had consolidated and cooled, and even after volcanic energy at the locality had entirely ceased. [Illustration: Fig. 2.--Elongation and branching of steam-vesicles in a lava, Kilninian, Isle of Mull, a little less than natural size.] Examples, however, are now accumulating which certainly prove that, in some cases, the vesicles were filled up during the volcanic period. Among the Tertiary basalt-plateaux of the Inner Hebrides, for instance, it can be shown that the lavas were already amygdaloidal before the protrusion of the gabbros and granophyres which mark later stages of the same continuous volcanic history, and even before the outpouring of much of the basalt of these plateaux. Not improbably the mineral secretions were largely due to the influence of hot volcanic vapours during the eruption of the basalts. This subject will be again referred to in the description of the Tertiary volcanic series. Vesicular structure is more commonly and perfectly developed among the lavas which are basic and intermediate in composition than among those which are acid. While the existence of a highly vesicular or scoriaceous structure may generally be taken as proof that the rock displaying it flowed out at the surface as a lava, other evidence pointing to the same conclusion may often be gathered from the rocks with which the supposed lava is associated. Where, for example, a scoriaceous lava is covered with stratified deposits which contain pieces of that lava, we may be confident that the rock is an interstratified or contemporaneous sheet. It has been erupted after the deposition of the strata on which it rests, and before that of the strata which cover it and contain pieces of it. In such a case, the geological date of the eruption could be precisely defined. Illustrations of this reasoning will be given in Chapter iv., and in the account of the volcanic series of Carboniferous age in Central Scotland, where a basic lava can sometimes be proved to be a true flow and not an intrusive sill by the fact that portions of its upper slaggy surface are enclosed in overlying sandstone, shale or limestone.