218 ROBERT B. SOSMAN 



difference is in the strength of the materials. Very considerable 

 stresses may accumulate in a glassy or crystalline rock before rup- 

 ture occurs, and when it does occur, the crack extends suddenly a 

 considerable distance into the mass. A layer of wet mud, on the 

 other hand, accumulates practically no stresses, as the forces of 

 cohesion and liquid surface tension to be overcome are very small. 

 The cracks therefore form much more gradually, and grow little by 

 little as desiccation proceeds. They have even been observed to 

 form under water, 1 probably as a result of freezing and melting. 2 



It is possible that some basalt prisms have been formed in the 

 same way as the slowly formed mud cracks, by the slow shrinkage 

 of a material which is partly solid and partly liquid, for the normal 

 course of crystallization of an igneous rock consists in the separation 

 of certain portions as crystals while the remainder stays liquid until 

 a lower temperature is reached. It has been commonly observed, 

 however, that the boundaries of contraction columns frequently 

 cut across the crystals of the rock, showing that solidification was 

 practically complete before the crack formed. 



An example of prismatic, although not columnar, structure pro- 

 duced in this manner is probably to be found in the " apparent sun- 

 crack structure in diabase," described by Wherry as occurring in 

 the upper surface of the great diabase sill of Pennsylvania, west of 

 Philadelphia. 3 He explains it as due to contraction jointing fol- 

 lowed by the penetration of still liquid material into the cracks from 

 below. At first sight this occurrence has some of the characteristics 

 of prismatic structure due to liquid convection accompanied by 

 segregation, but a re-examination of the structure by Dr. Wherry 

 and the author in May, 191 5, showed that the angles and polygons 



1 Moore, Am. Jour. Sci., XXXVIII (1914), 101-2. 



2 Mud cracks may also belong to the other types of columnar structure. Where 

 the deposit is very fine grained and homogeneous, the walls of the columns may show 

 the feathery patterns characteristic of a fractured solid, resulting from breaks (either 

 sudden or slow-growing) which occurred when the material was nearly dry, and 

 indicating the existence of tensional stresses. On the other hand, a prismatic structure 

 of apparently convectional origin has been observed by Guillaume (Soc. Franc. Pkys., 

 Bull. Seances, 1907, pp. 50-51) in mud flows in sub-Arctic regions. 



3 E.'T. Wherry, "Apparent Sun-Crack Structures and Ringing-Rock Phenomena 

 in the Triassic Diabase of Eastern Pennsylvania," Acad. Nat. Sci., Philadelphia, Proc, 

 LXIV (1912), 169-72. 



