32 CHEMICAL ANALYSIS OF EOCKS. 



There is a vast amount of unconscious humbug in the constant attempts 

 to use chemical analyses for a purpose foreign to their nature ; that is, to 

 determine, as before mentioned, the mineral composition by mathematical 

 calculations founded on rock analj'ses. All these efforts appear to be based 

 on an entire misconception of the nature of rocks. Nothing better illus- 

 trates the inutility of this method alone for the purpose of determining the 

 mineral composition, than a comparison of the speculations of chemists 

 regarding the minerals composing the stony meteorites and the actuality as 

 obtained by microscopic examination. 



The writer holds, as the result of his studies, that the chemical analysis 

 of a normal rock corresponds with its species — that is, certain percentages 

 of the more prominent elements can be laid down, beyond the extremes of 

 Avhich normal rocks belonging to that species will rarely if ever go, and 

 within which normal rocks of other species will rarely if ever come. This, 

 of course, applies especially to the eruptive rocks, for in the case of the 

 sedimentary ones, every degree of composition is to be expected according 

 to the sources from which the materials composing them were derived, and 

 the amount of sorting, chemical replacement, etc., they have undergone. 



The more highly altered or weathei'ed eruptive rocks, especially if 

 chemical constituents have been removed, and either replaced or not by 

 others, would not be normal forms. If the analyses were written in the 

 percentages of the elements, instead of their compounds, it is thought that 

 the chemical relations between the different rock species and varieties 

 would be more clearly apparent than at present, as Nordenskiiild has 

 shown for the meteoric peridotites.* 



From the manner in which. many chemical analyses of rocks have been 

 made (poor work, poor specimens, and no knowledge of the rock) the 

 difficulties in the way of proving these relations are great; but the writer 

 has prepared tables which show them in an approximate manner. 



* Nature, 187S, xviii. 510, oil; Geol. Foren. Forhaudl., 1S7S, iv. 45-61; Zoit. Deiit. gcol. Gesellg., 

 18S1, xxxiii. li-30. 



