AM) SPHEROIDAL BTB.TTCTTJBE. 153
metry in the surface of rupture ; and thus surfaces of revolution (such as spheroids, hyperboloids, &e.) would be generally produced.
There is thus a certain relation between the spheroids, the curved horizontal joints, and the great curving cross joints, such as are seen at the Koche Tuilliere. Hence the fissile structure so well exhibited both here and at the Eoche Sanadoire is not a cleavac/e structure in the ordinary * sense of the word ; for it is the result of force acting in exactly the opposite direction, being the result of tension ; whilst cleavage is due to compression — somewhat similar results being thus produced by forces with changed algebraic signs.
I consider the cup-and-ball joint structure a special case of the spheroidal ; and the only point on which I feel some doubt is whether one should regard the division as beginning at the centre or at the exterior, or as to some extent independent, — namely, whether as a column cools tolerably uniformly, it tends by symmetry to divide into approximately equal lengths, and so spheroids are formed in the more plastic though warmer parts (the exterior shell being a little more rapidly chilled), and whether the cracks thus formed between two adjacent spheroids are continued toward the exterior, which then, being in a state of strain, is cracked horizontally at these points of weakness ; or whether the exterior, cracking first, caused a line of weakness, which determined the commencement of a spheroid in the inner parts ; or, as a third possibility, whether the two divisional surfaces are to some extent independent — the spheroids forming within in the more plastic part of the column, the cracks opening from without to the more solidified part, and the two surfaces of division running together at last so as to complete the separation.
Some independence in these surfaces of division seems to be indi- cated by the occurrence of distinctly formed spheroids in an unbroken column, and by the fact that the spheroids are occasionally again subdivided into segments which to some extent continue the same structure, two or even more of these being enclosed in a more regular spheroidal shell (see fig. 13). This independence, too, would explain the fact that the ends of columns pointing in the same direction show sometimes cups and sometimes balls. It is quite true that the divisional curves, if due to strain from contraction, should be concave to the surface of cooling, as Mr. Mallet has proved (and this, I have little doubt, is the case in the curvitabular structure) ; but, as Mr. Scrope objects, and as my own experience has shown me, in the case of cup-and-ball structure in columns there is great uncertainty, adjacent columns showing at top, one a cup, the other a ball. If, however, the two fissures were formed to some extent inde- pendently (the curved one beginning at the interior, the plane one at the exterior), this would be likely to happen, though still the one or the other structure might predominate : the outside cracks would
- That it cannot be a true cleavage is shown (1) by the extreme improbability
of this being produced in rocks which are not likely to have been subjected to great earth-movements and must have always been of rather a superficial cha- racter, and (2), in the cases quoted, by the impossibility of explaining the ar- rangement of the divisional planes on any theory of cleavage.
