It is obvious that in a wide fissure the central portion may remain molten for some time after the sides have consolidated. If the fissure served as a channel for the ascent of lava to the surface, it is conceivable that the central still fluid part might be driven out and be replaced by other material from below, and that this later material might differ considerably in composition from that which first filled the opening. Such, according to Mr. Iddings, has been the probable history of some of the dykes at the old volcano of Electric Peak.[31] But we can hardly suppose that this explanation of compound dykes can have any wide application. It could only hold good of broad fissures having an outlet, and is probably inadmissible in the case of the numerous compound dykes not more than 10 or 15 feet in diameter, where the several bands of rock are sharply marked off from each other. The abrupt demarcation of the materials in these dykes, their closer texture along their mutual boundaries, the indications of solution of the older parts of the group by the younger, and of injection of the latter into the former, show that they belong to separate and unconnected intrusions. These questions will be again referred to in the account of the British Tertiary dykes (Chapter xxxv. vol. ii. p. 159). [Footnote 31: _12th Ann. Rep. U.S. Geol. Survey_ (1890-91), p. 587.] Another kind of compound dyke has arisen from the manner in which the original fissure has been produced. While, in general, the dislocation has taken the form of a single rectilinear rent, which on opening has left two clean-cut walls, cases occur where the rupture has followed several parallel lines, and the magma on rising into the rents appears as two or more vertical sheets or dykes, separated by intervening partitions of the surrounding rock. Examples of this structure are not infrequent among the Tertiary dykes of Scotland. One of these may be noticed rising through the cliffs of Lewisian gneiss on the east coast of the island of Lewis, south of Stornoway. One of the most extraordinary instances of the same structure yet observed is that described by Professor A. C. Lawson from the Laurentian rocks at the mouth of White Gravel River, on the N.E. coast of Lake Superior. In a breadth of only about 14 feet no less than 28 vertically intrusive sheets or dykes of diabase, from 1 inch to 6½ inches broad, rise through the granite, which is thus split into 27 thin sheets. The diabase undoubtedly cuts the granite, some of the sheets actually anastomosing and sending veins into the older rock.[32] [Footnote 32: _American Geologist_ (1894), p. 293.]