GEOLOGY. 265 



rect determinations of the porosity of crystalline rocks, upon which 

 subject I am not aware of any published experiments, the observa- 

 tions upon the fusion of rocks, and the determinations of their densi- 

 ties in the crystalline and vitreous states, admit of an indirect applica- 

 tion to the question before us. The experiments of Charles Ste. 

 Ciaire Deville, in the Comptes Rendus for 1845, and of Delesse in the 

 Bulletin for 1847, agree so closely in this matter that we give them 

 the preference over those of Bischoff, published in 1842. Deville 

 and Delesse found that the fusion of rocks yields glasses whose density 

 is generally inferior to that of the rock in the crystalline state. This 

 diminution for granite is equal to from nine to eleven hundredths, 

 and it is evident that such a glass, passing to a crystalline state and 

 retaining its volume, must present vacant spaces in direct proportion 

 to the augmentation of density, that is to say, equal to about one- 

 tenth of its volume. If we take the mean density of granite at 2.GO, 

 it might with such a degree of porosity imbibe 3.9 parts in 100.0 of 

 its weight of water. This shrinking of one-tenth is no exaggeration, 

 and such a rock would still be a good building material, although 

 containing twenty-five times more vacant space than our calculation 

 requires. 



The vitreous state of a body is nothing more than a fixing of its 

 molecules in the positions which belong to them in the liquid state, 

 and probably represents the liquid in its greatest degree of density. 

 The crystallization of barley sugar, of wrought iron, and of Reau- 

 mur's porcelain, are striking examples of the tendency of molecules 

 to group themselves in crystals even in the midst of solid masses, and 

 we can thus readily understand the absence of vitreous substances 

 among the older crystalline rocks. The great difficulty is to deter- 

 mine with exactness the proportion of the vacant spaces resulting 

 from this change, since these will vary for each body, and probably 

 also with the volume of the mass. Sulphur fused in an open vessel 

 crystallizes slowly, the level of the liquid sinks a little, and after com- 

 plete solidification the surface is covered with hollows resulting from 

 the shrinking, whereas if cooled in a spherical shape these cavities 

 would naturally be formed at the centre. Water and bismuth, as is 

 well known, behave in a very different and remarkable manner, the 

 first dilating eight or ten hundredths at the moment of congelation, 

 and the second one fifty-third. The only conclusion to be drawn 

 from these facts is, that each body in solidification behaves in a 

 different manner, and that for the solution of the question before us 

 we can only take into account the well-known porosity of rocks. 

 The problem, however, appears to me one of great importance in 

 connection with theoretical geology ; if we admit with Deville that 

 at the moment of crystallization the density of rocks is in all cases 

 augmented, we are forced to conclude that all the crystalline masses 

 formed at the surface of the liquid globe must have sunk and accu- 

 mulated at the centre. The effect of a similar action has been 

 shown by physicists, who have demonstrated that the cold of winter 

 would freeze our lakes and rivers from the bottom if the ice sunk at 

 the moment of its formation, as would the solidified parts of a lake 

 of molten sulphur. We should then have, in place of a liquid globe 

 surrounded by a solid shell, a mass solidified to the centre ; a conclu- 

 23 



