COLUMNAR STRUCTURE. 71 



perpendicular to the isotherms for which the tensions reach the limit of 

 cohesion. 



Each of the columns cools as a separate body, and if the following 

 figure represents a vertical section of one of them, the dotted lines approxi- 

 mately represent the position of the isotherms. In the separate column 

 the tension will be greatest at the edges, because these, exposing a great 

 surface per unit volume, will chill most rapidly. When the column lias 

 reached a sufficient independent length, the tension on the edges will be 

 so great that they will rupture perpendicularly to the isotherms. These 

 ruptures will cut inward from the sharp edges and divide the columns 

 laterally by cup-shaped surfaces. The interval between these vertical 

 subdivisions of the column depends on the amount of tension which the 

 mass can stand without rupture, and will evidently be of the same order 

 as the diameters of the columns themselves. 



Figure 15.— Cooling of Columns. 



The foregoing deductions all correspond very closely to the phenomena 

 of columnar structure in massive rocks. Basalts, diabases, and the like, 

 however, are far from being homogeneous, and it is surprising that the 

 surfaces of the columns should be so smooth. If one were to cut a 

 suitable bar of diabase and" break it by tension in a testing machine the 

 fracture would certainly lie much rougher than the side of a diabase 

 column. This seems to be accounted for, at least in part, by the fact that 

 rupture probably takes place immediately after solidification of the 

 groundmass and before any difference in rate of cooling between the 

 embedded crystals and the groundmass has locally weakened the cohe- 

 sion of the latter. When masses of mud in drying out split into columnar 

 fragments, the torn surfaces are less smooth and the divisions less regular* 



Review of Theories of slaty Cleavage. 



}\liy Needful. — So little attention lias been paid by geologists to systems 

 of faulted fissures that the field may be said to be a new one. I know 



♦ There is an intimate connection between the problem of columnar structure and that of the 

 division of space with minimum partitiona) area. See an investigation of tic latter subjeel by Sir 

 William Thomson (Lord Kelvin), Mittag-Lefflers Acta Math . vol. 11, 1887-88, p. 121, and Plateau, 

 Statique des Liquids, vol. 1. 



XI— Bull. Gkoi,. Soc. Am., Vol. 4, 1892. 



