132 GEOLOGY OF THE IIIGH PLATEAUS. 



see if there be anything in the physical properties of the rocks to justify 

 such a hypothesis. We can represent this best by a graphic expression of 

 their physical properties regarded as functions of temperature and acidity. 



Let the axis of abscissas, Plate 4, represent the proportions of silica 

 characteristic of the various groups of volcanic rocks, the figures along that 

 axis representing percentages from 40 to 80. Let the ordinates represent, 

 first, the density of the rocks in the cold state. Considering now any one 

 variety of rock, take the point on the axis of abscissas corresponding to its 

 percentage of silica, and erect an ordinate proportional to its density. For 

 all the varieties of rocks construct ordinates in the same manner and join 

 their upper extremities. On the assumption that the density is rigorously 

 correlated to the percentage of silica, a cm've would be constructed repre- 

 senting the density as a definite function of the silica. This assumption, 

 liowever, is not strictly true, being subject, indeed, to notable variations ; 

 yet in a general way it is more or less an approximation to the truth. The 

 anomalies will be adverted to in the sequel. 



It is known that the rocks of the basic and sub-basic groups are when 

 cold considerably more dense than the average of the foliated rocks, and 

 the same is true of some of the sub-acid rocks, and according to the doc- 

 trine heretofore laid down such rocks could not be erupted at all were it 

 not for the fact that when intensely heated and liquefied, their density is 

 notably diminished and reduced below that of the strata which overlie 

 them. Hence the more basic the rock, the more it must be heated to 

 reach an eruptible density. The ordinates, then, may be used to represent 

 the relative increase of temperature which must supervene in order to ren- 

 der the rocks light enough to reach the surface, and as these increments of 

 temperature are directly proportional to the density of the rock, the same 

 curve may (in the absence of fundamental constants) be used to express 

 the increment? of temperature required by the various rocks to reach an 

 eruptive density. 



Again, let the ordinates represent the relative melting temperatures of 

 the various sub-groups, the assumption still being that the fusibility is a 

 definite function of the proportion of silica. This assumption is probably 

 subject to still wider variations than that which postulates a dependence of 



