468 



ALBERT V. G. JAMES 



of the liquid or plastic lava and thereby ot the isotherms; 

 (ii) rise of steam from an overwhelmed stream; (12) chemical 

 variation in the lava with corresponding change in thermal 



conductivity. 



It is not likely that lava 

 columns bend after they are 

 formed, for if lateral pressure 

 were sufficiently strong, it 

 would fracture the brittle col- 

 umns but not bend them. The 

 plastic state, when the lava 

 would yield to lateral pressure, 

 was lost when the molten mass 

 crystallized. 



8. Convection as a factor on 

 column production. — R. B. 

 Sosman^ in 19 16, stated that 

 many, if not 

 most, lava col- 

 umns were due 

 to convection 

 currents in the 

 molten flow. 

 He said con- 

 V e c t i o n col- 

 umns are char- 

 acterized by the fact that hexagonal columns and 

 angles of 120° predominate, while the majority of 

 contraction columns are pentagonal.^ After testing 

 some hundreds of columns in the Sydenham area I 

 am unable to agree with him in several respects. 

 The attitude of the columns there stamps them 

 as being definitely due to contraction, not convec- 

 tion, and yet 40 per cent are hexagonal, 22 per 

 cent heptagonal, 22 per cent pentagonal, 13 per cent octagonal and 

 3 per cent quadrilateral, and the great majority of the angles are 



I R. B. Sosman, Jour. Geol., XXIV (1916), pp. 224, 228, 233. 

 ^ Ibid., pp. 225-26. 



Fig. 14. — Bent columns. (Jackson's 

 Creek, Sydenham, Australia.) 



Fig. 15. — Bent 

 columns. Dia- 

 gram showing 

 the relation of 

 the attitude 

 of the columns to 

 the isothermal 

 planes. 



