3 o8 SCIENCE PROGRESS 



and quartz; which is inexplicable, except on the supposition that 

 it was actually a mixture of felspars and quartz" (pp. 166-7). 



Mr. Harker's last assumption goes, however, somewhat 

 farther than is warranted by the facts. Potash, soda and lime 

 are able to form, without the assistance of silica, definite com- 

 pounds with alumina in the same proportion as in felspars and 

 other silicates of similar composition, such as nepheline, 

 analcime, leucite and most of the zeolites, that is to say, with 

 one molecule alkali or lime for each molecule of alumina. The 

 potassium and sodium aluminates are freely soluble in water 

 at ordinary temperature, and the former yields crystals with the 

 composition A1 2 3 . K 2 . 3H2O. 1 The corresponding lime com- 

 pound is insoluble. The possibility that it is these aluminates 

 and not the felspars or other allied silicates that are present 

 in the magma must not be overlooked. If this be the case 

 the silica will no doubt occur in combination with water, 

 for which it possesses a great affinity at high temperatures, 

 and the excess of alumina will probably be present in a 

 similar hydrated form. In exceptional cases, where there is 

 an excess of alkalies over alumina, the trisodic aluminate 

 Al 2 3 .3Na 2 0, which has also been isolated, may be present and 

 give rise to minerals of the sodalite group. It is, however, 

 usually convenient in discussing the physical principles that 

 govern the consolidation of an igneous rock to assume, for 

 the purposes of exposition, that the minerals that crystallise 

 out were present with the volatile fluxes in a state of mutual 

 solution in the magma. 



The crystallisation of a magma is, as a rule, partly the 



result of cooling and partly a consequence of the escape of 



the volatile fluxes. The latter is, however, in some cases the 



result and not the cause of the crystallisation. The laws that 



regulate the crystallisation of silicate magmas on cooling in 



the absence of the volatile fluxes have been worked out by 



Vogt, Doelter and others from the examination of artificial 



slags and a comparison of the results thus obtained with the 



microscopic characters of igneous rocks. It has thus been 



shown that the order in which the constituents of a slag or 



igneous magma crystallise out is largely dependent on the 



proportions in which they are actually or potentially present. 



1 These analytical formulae are adopted for practical convenience without 

 any assumption that they represent the structural constitution of the substance. 



