occur on the south (sic, query west) and the north sides are remarkable neither for beauty nor for magnitude." The caves are sufficiently numerous to furnish an argument. There are very few hollows worn by the sea in the Scotch coast. The islet, which contains a dozen, has not the 1100000 part of the indented line of the mainland, and bears an infinitesimal ratio to the sea-board, including the islands. Its parent, Mull, within whose bosom rests this irregularly oval rock, "measuring about one and a half mile in circumference," has in the dimension of length one hundred and fifty times better right to a "museum of wonders." The "Isle of Columns" is a speck too tiny to show on any ordinary map. The chance that it would contain, as a legitimate yet exceptional result of normal contact between igneous rock and seawater agitated by wind, "the most remarkable cave in Europe," is less than 0. It is the
The uneven table-land is formed of "three distinct beds of rock of unequal thickness, inclined toward the east at an angle of about nine degrees. The lowest is a rude trap tufa, the middle one is divided into columns placed vertically to the plane of the bed, and the uppermost is an irregular mixture of small columns and shapeless rock." The columnar bed-is never more than GO feet thick. The island itself attains a maximum height of 129 feet. It has no peak from which rain-water might descend in a considerable quantity. There is no series of fissures corresponding to the perforations. There is nothing on the flat top to suggest the tunnels beneath. Proceeding toward the south from the landing-place, there are six cases of alleged erosion, each presenting its own seemingly insuperable difficulty, and cumulatively requiring a more thoughtful and serious consideration than the fantastic phrases in which 'stupendous (!) columns, three feet thick and thirty feet high, rise from a dark-red or violet-colored rock over which the ocean rolls, and reflects from its white bottom a variety of crimson and yellow'.
It appears now to be well established that the peculiar structure of columnar basalt is due to contraction and splitting consequent upon cooling. The analogy is rather to the splitting often seen in the mud bottom of a dried-up pool than to ordinary crystallization. The various conditions point to the contractile origin of the structure, at the same time that the result suggests a curious mimicry of imperfect crystallization. If the cooling mass of basalt be in one of its vertical sections of such a form that successive isothermal couches, taken in descending order, are not parallel to the original cooling surface, as they are in all cases of straight and parallel prisms, but divergent gradually from the cooling surface and from each other, then the lines of the splitting of the prisms, always true to these couches, must be curved in one direction. This will be true, whether the isothermal couches be plane surfaces divergent from a thinner to a thicker part of the mass, or whether they be curved surfaces arising from the mass reposing on
