440 Walter Keeping — On Columnar Sandstone. 



quarry above Forst, near Durkheirn. 1 It is the " Tabular " structure 

 of Professor Bonney. In these altered prismatic Saxon sandstones 

 this structure is unusually regular, dividing the rock into zones 

 about 4 inches in thickness. 



According to A. von Gutbier, an irregular line of clay ironstone 4 

 to 5 inches thick, and next a zone of "reddish green steinmark," 

 intervened between the sandstone columns and the basalt. 



Next, to inquire into the mode of origin of these prisms and their 

 associated structures. The production of columnar structure by 

 contraction during cooling or drying is now very well understood ; 

 being well illustrated by the structure of dry starch on the one hand, 

 and in the artificial formation of columns in the sandstones of 

 furnaces on the other. The evidence of the former presence of heat 

 in close proximity with the Gorischstein prisms is obvious enough, 

 for the outermost sandstone columns were almost in contact with 

 the liquid lava mass. But whether the heat was also the immediate 

 cause of the prism of the quartz grains, as above described, on the 

 surfaces of the columns, is not quite so clear ; though the fact of its 

 special association with the prismatic structure must not be over- 

 looked. There is no evident reason why the deposition of silica in 

 the wet way should go on over the prismatic joints differently from 

 that over the ordinary rock joints, except during that special period 

 of cooling when the columnar joints were in course of formation. I 

 am therefore disposed to refer this alteration of the superficial layer 

 of the prisms to the later stages of the volcanic period, when such 

 joints would have served for the passage of heated water or of steam 

 to the surface. 



Note by the Rev. Professor Bonney on tlie origin of the curved 

 arrangement of the prisms. 



The rock in order to break requires (1) that its temperature 

 should be considerably elevated ; (2) that this should be again 

 gradually lowered. 



A surface of uniform temperature is a surface of 

 uniform tension. A Ba 



Now if A B G be a heated mass, losing heat prin- , 



cipally from A B, a rock at surface, the surface of C 



uniform tension lies parallel to A B. 



Again, if D EFbe a mass in contact with an igneous rock K at 

 some depth, the main loss of heat will be laterally, 

 and so the surface of uniform tenison be vertical and D 



the columns horizontal. x x 



At the top, then, the loss of heat will be mainly j^ F — ^> 

 from the top ; deep down mainly laterally. Thus x 



the surfaces of equal tension will descend in the x x 

 mass radially, the lateral loss producing no effect E 



at first and vice versa. So that I take it just at 

 first the columns should lie pretty nearly parallel with the basalt, 



1 Quite a distinct thing from the bedding of successive lava flows. See Prof. T. 

 G. Bonney, Quart. Journ. Geol. Soc. vol. xxxii. p. 14$. 



