CRYSTALLINE STRUCTURES OF ROCKS. 171 



more or less accurately vertical; because the beds 

 which they divide in a perpendicular manner, are, in 

 the same way, more or less strictly horizontal; but 

 they are also occasionally curved in a variety of modes. 

 They are, further, often divided by transverse joints, of 

 various forms, though sometimes simple. The angles 

 of these columns vary in number, yet so that the pre- 

 valent forms lie between the four and sevensided 

 figures; but it is essentially necessary to remark, that 

 the contact is always perfect; neither vacuity among 

 the angles, nor interval between the approximate sides 

 intervening. The more imperfect forms of this de- 

 scription gradually pass into an irregular prismatic 

 structure ; and that at length becomes so indefinite as 

 to be confounded with a mere tendency in the solid 

 rock to a vertical fracture. 



When these columnar traps are subject to decom- 

 position, it is sometimes observed that they desquamate 

 in successive crusts, so that a spheroidal nucleus at 

 last remains where there was once a prismatic joint. 

 This has been supposed a proof of a peculiar con- 

 cretionary structure giving rise to the prismatic form, 

 the arguments respecting which will be immediately 

 considered. 



As connected with the trap rocks in their general 

 characters, it is proper here to observe that some 

 lavas occasionally assume the same figures. It has 

 sometimes been said that this occurrence took place 

 only where such lavas came into contact with the 

 sea in the course of their progress; and it has been 

 argued that a similar cause may have produced the 

 columnar form in the trap rocks. But the assertion 

 is unfounded, in all respects ; inasmuch as columnar 

 lavas are found where no water can have been present, 

 and amorphous ones occur beneath the sea. 



If the columnar structure is common, in sand- 



