The intersections of dykes prove that the process of fissuring in the earth's crust took place at more than one period, and prepare us for the reception of evidence that the same line of fissure might be again re-opened, even after it had been filled with molten material. Numerous instances have now been accumulated in which dykes are not single or simple intrusions, but where the original dyke-fissure has been re-opened and has been invaded by successive uprisings of lava.[197] Compound dykes have thus been formed, consisting of two or more parallel bands of similar or dissimilar rock. [Footnote 197: See an example figured by Macculloch, _Western Isles_, plate xviii. Fig. 1.] While it is not difficult to conceive of the re-opening of a vertical fissure during terrestrial strain, and the injection into it of later intrusions of a volcanic magma, it is not so easy to understand the mechanism where the line of weakness has been slightly inclined or horizontal, and where, consequently, there has been the enormous superincumbent pressure of the overlying part of the earth's crust to overcome. Yet gently inclined compound dykes exhibit their parallel bands with hardly less regularity than do those that are vertical. The difficulty of explanation is felt most strongly in the attempt to realize the origin of the compound sills described in Chapter xlviii. In the re-opening of dyke-fissures the later intrusions have generally taken place along the walls, or where the dykes were already compound, between some of the component bands. Less frequently the first dyke has been split open along the middle, and a second injection has forced its way along the rent. Of the first of these two types, numerous instances have now been observed in the West of Scotland. If the portion of a compound dyke exposed at the surface be limited in extent, we may be unable to determine which is the older of two parallel bands of igneous rock, though the fact that they present to each other the usual fine-grained edge due to more rapid cooling, shows that they are not one but two dykes, belonging to distinct eruptions. So far as I have noticed, where one of the dykes can be continuously traced for a considerable distance, the other is comparatively short. I infer that the shorter one is the younger of the two.