324 J. P. Iddings — Columnar stoucture in the 



O'Rourke's quarry are modified near their junction with the 

 smaller ones above in exactly the same manner as the small 

 vertical ones higher up are at their junction with the nearly 

 horizontal columns. In each case the ends of the two sets 

 taper off and curve in one direction, indicating that the forces 

 developing one set of columns were affected by those develop- 

 ing the other set, giving rise to resultant forces which brought 

 the columns finally into parallelism. 



The columnar structure in volcanic lavas is unquestionably 

 a cracking produced by the shrinkage of the mass upon further 

 cooling after it has consolidated into rock, which still retains a 

 great amount of heat. That in every case the mass is solidi- 

 fied before it is cracked is shown by the dividing of crystals 

 and crystal aggregations lying in the path of the crack, which 

 are cut across sharply, the two portions remaining firmly fixed 

 in their respective matrices, whether these be crystalline grains 

 or glass. In the case of volcanic glasses the columnar parting 

 passes through the mass without in the slightest affecting the 

 most delicate indication of flow structure, showing that the 

 glass was rigid before it was fractured. It is also a fact of 

 observation that the direction taken by the columns is perpen- 

 dicular to the plane of cooling, often the plane of contact with 

 another rock. 



Before entering upon a discussion of the nature and develop- 

 ment of columnar structure it will be well to consider the 

 process of cooling and solidification of a molten lava. It is 

 well known that a solid crust quickly covers the fluid rock 

 after its eruption, and increases more and more gradually in 

 thickness as the cooling advances, the fluidity of the mass be- 

 neath in like manner decreasing. Owing to the poor conduct- 

 ivity of rock the rate of cooling of the inner portion becomes 

 less and less as the crust thickens, and the central portion re- 

 mains at a high temperature for a very long period, as has often 

 been observed ; thus the changes in it are far more gradual 

 and uniform than they were near the surface. 



"When the surface layer of a molten mass consolidates rap- 

 idly while the portion beneath is still fluid its contraction will 

 meet with little or no resistance. The more gradual contrac- 

 tion of lower layers will be attended with relatively greater 

 resistance as the fluidity of the mass beneath becomes less. 

 The actual amount of contraction in different parts of the mass 

 must also increase with a decrease in the rate of cooling in 

 consequence of the greater degree of crystallization attained, 

 for the crystalline form of a rock has a higher specific gravity 

 than the more glassy form of the same magma and therefore 

 occupies a smaller volume. 



The resistance to contraction in the portion of the mass near 



