FACTORS IN COLUMNAR STRUCTURE IN LAVAS 



461 





m^- 



.Kr- 



^ 



3. The fact has frequently been recorded' that the upper layer 

 of rock is not columnar but fragmental, platy, ropy, or scoriaceo 

 This fact shows that if coohng is rapid, columns are not formed, for 

 the surface layer is cooled with very great rapidity. If coohng is 

 very rapid contraction is Hkely to cause fragmental partings since 

 the isothermal planes would probably be irregularly spaced through 

 the flow. 



4. In the rock sections of columnar basalt examined very Httle 

 glass was noted, whereas the surface rock contained a considerable 

 amount of glass. The holocrystalline 

 nature of the columnar basalt strongly 

 suggests slow rather than rapid cooling.^ 



2. Ball-and-socket structure. — The cross 

 joints are secondary, i.e., they are formed 

 after the columns. This is shown by the 

 fact that they never continue across from 

 one column to another. The vertical col- 

 umns continue to shrink as they cool. 

 Horizontal contraction simply causes a 

 widening of the vertical joints, but tension 

 in other directions can only be relieved by 

 cracking. 



There are necessarily centers of con- 

 traction distributed equally along each 

 column (if lava is homogeneous) and the 

 particles will concentrically be drawn 

 toward these points. This causes not only 

 concentric weakenings and cracking but 

 also a well-defined horizontal joint midway 

 between any two centers of contraction 

 (Fig. 2). Mallet^ says that the concave surfaces face the direction 

 of coohng. This is not true in the areas I have studied, for adjacent 

 columns have the cups facing directions regardless of the coohng 

 surface. 



Fig. 2 Fig. 3 



Figs. 2 and 3. — Onion 

 structure and spheroidal 

 weathering. Centers of 

 contraction along the col- 

 umns cause concentric 

 planes of weakness which, 

 on weathering, develop into 

 the well-known "onion 

 structure." The solid 

 angles a and b, which tend 

 to break off, are due to 

 these concentric planes of 

 contraction. Ball-and- 

 socket Joints are formed 

 between the adjacent 

 spheroids. 



^ J. P. Iddings, Amer. Jour. Sci., XXXI (1886), p. 325. ^ Op. cit., p. 331. 



3 R. Mallet, Phil. Mag., L (1875), p. 204; J. P. Iddmgs, Amer. Jour. Sci., XXXI 

 (1886), p. 329; R. B. Sosman, Jour. GeoL, XXIV (1916), p. 229. 



