(_b_) More definite information, and often from a wider vertical range, is to be gathered on coast-cliffs and in hilly districts, where the same dyke can be followed through a vertical range of many hundred feet. But so far as my own observations go, no general rule can be established that dykes sensibly vary in width as they are traced upward. Every one who has visited the basalt-precipices of Antrim or the Inner Hebrides, where dykes are so numerous, will remember how uniform is their breadth as they run like ribbons up the faces of the escarpments.[183] Now and then one of them may be observed to die out, but in such cases (which are far from common) the normal width is usually maintained up to within a few feet of the termination. [Footnote 183: This point did not escape the attention of that excellent observer, Berger, in his examination of the dykes in the North of Ireland. We find him expressing himself thus:--"The depth to which the dykes descend is unknown; and after having observed the sections of a great many along the coast in cliffs from 50 to 400 feet in height, I have not been able to ascertain (except in one or two cases) that their sides converge or have a wedgeform tendency" (_Trans. Geol. Soc._ iii. p. 227).] All over the southern half of Scotland, where the dykes run along the crests of the hills and also cross the valleys, a difference of level amounting to several hundred feet may often be obtained between adjacent parts of the same dyke. But the breadth of igneous rock is not perceptibly greater in the valleys than on the ridges. The depth of boulder clay and other superficial deposits on the valley bottoms, however, too frequently conceals the dykes at their lowest levels. Perhaps the best sections in the country for the study of this interesting part of dyke-structure are to be found among the higher hills of the Inner Hebrides, such as the quartzites of Jura and the granophyres and gabbros of Skye. On these bare rocky declivities, numerous dykes may be followed from almost the sea-level up to the rugged and splintered crests, a vertical distance of between 2000 and 3000 feet. The dykes are certainly not as a rule sensibly less in width on the hill-tops than in the glens. So far, therefore, as I have been able to gather the evidence, there does not appear to me to be, as a general rule, any appreciable variation in the width of dykes for at least 2000 or 3000 feet of their descent. The fissures which they filled must obviously have had nearly parallel walls for a long way down. 6. INTERRUPTIONS OF LATERAL CONTINUITY