6 PROCEEDINGS OF THE GEOLOGICAL SOCIETY, [DcC. 1, 



it occurs in Gloucestershire. But what especially attracted my 

 attention was that the surfaces, hoth upper and lower, of some of the 

 sandstone heds, which alternate with laminae of marl, are studded 

 over with small bodies, resembling crystals, of cubical or nearly 

 cubical forais. On examination, these apparent crystals prove not to 

 be cr^^stalline in their interior, but to be wholly composed of white 

 sandstone, passing into, and inseparable from, the stratum from 

 whose surfaces they project. It is also evident that the grains of sand 

 which compose them are not held together by crystalline matter, like 

 those in the well-kno^^^^l sandstone crystals of Fontainebleau, as they 

 present no trace of cleavage planes, and their cubical forms prove 

 them not to be due to carbonate of lime. 



It appears then that these crystal-like bodies belong to the class 

 termed pseudomorphous, and to the lowest subdi\ision of that class. 

 In some cases, where one substance fills by chemical infiltration the 

 ca^-ity which has been formed by another, the product is still a cry- 

 stalline body, though it has assvimed an external form foreign to its 

 real nature. But in the case before us the operation is purely me- 

 chanical ; — the cavities formed by pre-existing crystals being merely 

 filled with sand, poured into them from above, and taking the form 

 of whatever cavities it might meet with. 



The question next arises, what was the nature of the original cry- 

 stals which are now replaced by sand ? On examination we find that 

 the majority of them are cubes or modifications of cubes. In some 

 of them indeed a slight obliquity in their angles is perceptible, but 

 this seems evidently owing to the effect of pressure, which has crushed 

 and distorted the original form of the crystals, after they were re- 

 placed by the sand. We may therefore regard them as having ori- 

 gmally conformed to the cubical type. 



Of substances which crystallize in cubes, the only ones which 

 usually occur in the Triassic formations are sulphuret of iron, or iron 

 pyrites, and chloride of sodium or common salt. It is hardly possi- 

 ble that sulphuret of iron can have supplied the moulds into which 

 the sand was afterwards poured, as it would require a considerable 

 time both for the formation and for the removal of crystals of that 

 mineral, whereas it is e\"ident that the crystals in question must 

 have been formed, and must have afterwards been removed, leaving 

 an empty cavity, in the short interval between the deposition of one 

 bed of sand, and of the one immediately superimposed. All these 

 conditions, however, are supplied in the most satisfactory manner 

 by supposuig chloride of sodium to have been the material which 

 formed the moulds for these pseudomorphous crystals. The ripple- 

 marks, — the cracks formed by desiccation in the argillaceous beds, 

 and afterwards filled with sand poured in from above, — and the not 

 unfrequent impressions of the feet of air-breathing reptiles, all of 

 which pheenomena especially characterize the Keuper sandstones of 

 our English counties, seem to point to a very shallow state of the sea, 

 abounding with sand-banks, and extensive salt-water marshes, often 

 laid bare in the intervals of the tides. If now we suppose that at the 

 locaUty in question, a sandy marsh existed, which at high spring 



