AND SPHEROIDAL STRUCTURE.
151
lenticular space. In this, two internal spheroids were exposed by the exfoliation of the face, which were not connected with any cross joints. In the adjoining mass four large, rude and very irregular spheroids were shown (the bounding surfaces being incomplete or with more than one centre). Similar instances of spheroids discon- nected from the main joints may be seen in a quarry called Turner's Pit in the Rowley-Eegis basalt, where also, in some cases, the nuclei of the spheroids are further subdivided, so that imperfect spheroids are enclosed by a spheroidal shell, like the twin kernels of a nut (fig. 13). It is therefore clear that, though the spheroids often do corre-
Fig. 12. — Spheroids in an 2in- jointed column near Le Puy.
Fig. 13. — Complicated spheroidal, structure {Rowley-Regis bascdt).
( }— f , , -1
spond with the spaces between cross joints in the column, there is no necessary connexion ; both maybe due to a somewhat similar cause; but the spheroid can exist without the cross joint, and vice versa.
Is it, then, possible to find any one cause which will explain these various divisional structures from the fissile to the spheroidal ? I think that which has already been suggested (I may say proved) for some of them, viz. the contraction of a cooling mass, is capable of explaining all. Mr. Mallet has shown that if a mass of molten rock be cooling uniformly from a surface, it will, in consequence of the mathematical principle of least action, break into hexagonal prisms at right angles to the surface of cooling. By the same principle, if a cube were contracting in consequence of a uniform loss of heat from each of its sides, it would be more likely to rupture internally (supposing that the solidification of the exterior prevented diminution of volume) in spheroidal shells ; for the more rapid loss of heat from the angles would tend to bring the isothermal surfaces within into a rudely spherical form ; and then, when the strain
