VII] OF THE GIANT'S CAUSEWAY 521 



traction, however, does take place, irresistibly, and it may be that 

 long cracks appear; the strain being so far relieved, the next cracks 

 will tend to take place at right angles to the first. But more 

 commonly rupture is delayed until considerable strain-energy has 

 been stored up ; once started, it proceeds explosively from a number 

 of centres, and shatters the whole mass into prismatic fragments. 

 However quickly and explosively the cracks succeed one another 

 each reheves an existing tension, and the next crack will give rehef 

 in a different direction to the first. When one crack meets another 

 it will seldom cross it, for the strain which led to the former fracture 

 does not extend into the new field. In short the cracks will be 

 found to meet one another three, by three, and therefore at angles 

 071 the average of 120°, and the columns will be on the average 

 hexagonal. For the making of a prismatic structure all that is 

 required is more or less uniform tensile strain in the two dimensions 

 of a horizontal plane; uniform tension in three dimensions would 

 have given rise to a cellular structure, of which the hexagonal 

 "causeway"' is the two-dimensional analogue. 



The columnar structure is accompanied by sundry secondary 

 phenomena. The vertical columns tend to break across on further 

 contraction, and exhibit rounded or basin-shaped ends, fitting 

 together in a shallow ball-and-socket; this beautiful configuration 

 has only lately been explained*. When cooling has caused the 

 mass to split into vertical columns, air or it may be water enters 

 the rifts and further cools or quenches the now sohd but still glowing 

 basalt. Each column tends to be chilled all round while still hot 

 within; but the hot unshrunken mass within checks or hinders the 

 contraction of the cooler outer layers. Thus unequal coohng causes 

 vertical as well as horizontal tensions; and just as these last are 

 relieved by the existing cracks, so new rifts appear crosswise to the 

 column, and relieve the vertical tensions. 



If the coohng come downward from above, then, at any given 

 level, the column will always be cooler above it than below; the 



* F \V Preston ball-and-socket jointing in basalt prisms, Proc R S (B), 

 c VI, pp. 87-92, 1930 A study of the rupture of glass Trans Soc of Glass Technology ^ 

 1926 p 263 On the general subject of prismatic structure m igneous rocks, see 

 also Robert Mallet, Phil. Mag. (4), l. pp. 122-135, 201-226, 1875, James Thomson, 

 Trans. Geol. Soc. Glasgow, March, 1877; Coll. Works, p. 422; R. B. Sosman, Journ. 

 Geology, xxiv, p. 215, 1916. 



