45 



points out, however, that there are many intermediate forms ; and that the 

 members of the individual groups exhibit a considerable amount of variability. 



In reviewing this subject from our present standpoint we are led to the 

 conclusion that the variations in composition in the greater number of 

 igneous rocks, although not capable of being reduced to any very precise law, 

 may be expressed in a broad and general way by the method of BuNSEN. 

 Thus, we recognise two extreme types ; the one characterised by a low 

 percentage of silica, high percentages of lime, iron and magnesia ; a low 

 percentage of alkalies, and an excess of soda over potash ; the other by a 

 high percentage of silica, low percentages of lime, iron and magnesia ; a high 

 percentage of alkalies, and an excess of potash over soda. Between these 

 two extreme types we have every possible gradation, and there seems no 

 reason for believing that the extreme types are more abundantly developed 

 than the intermediate forms. As stated in this general manner, and without 

 any attempt to give it numerical precision, the above principle will be 

 referred to as the law of BUXSEX. 



It will be seen, on reflection, that a very interesting relation exists 

 between BUXSEX'S law and the law which expresses the effect of 

 crystallisation in modifying the composition of an igneous magma. 

 If, after crystallisation has progressed to a certain extent in a magma 

 of andesitic composition, we separate the crystals from the part remaining 

 liquid, the former taken together will have the composition of a basic, 

 the latter of an acid rock. It is scarcely possible that this connection 

 can be the result of chance ; and, if not, it seems to suggest that those 

 variations in the composition of igneous rocks which are expressed by 

 BUXSEX'S law may be due to differentiation produced in an originally 

 homogeneous magma in consequence of progressive crystallisation. 



The leucite- hauyn- and nepheline-bearing rocks furnish important 

 exceptions to the law above referred to, and must be regarded rather as 

 forming a group apart from the other igneous rocks, so far as it is concerned. 

 It is to the basalts, andesites, and rhyolites and their plutonic representatives 

 that the law is especially applicable ; but even in these cases it fails when 

 applied to the minute variations which occur in their composition. The exact 

 composition of any igneous rock is doubtless determined by many causes, 

 some of which are general, and others probably local. As the science of 

 petrography advances w r e may reasonably anticipate that these causes will 

 be brought to light, and the enormous mass of disconnected facts, so 

 industriously collected together and admirably expressed in ROTH'S Gesteins- 

 analyseii and his Beitrage zur Petrographie der plutonischen Gesteinen, 

 reduced to something like definite order. At present BUXSEX'S law is the 

 only one which seriously claims our attention. 



So far reference has only been made to the chemical characters of 

 unaltered igneous products. The chemical composition of altered rocks 

 throws a considerable amount of light on the nature and extent of the 

 alterations to which they have been subjected. Many rocks may now be said 

 to be the pseudomorphs of the original rocks which they represent. The 

 chemical alteration of igneous rocks results in the addition and subtraction 



