60 Washington — Statement of Rock Analyses. 



of rocks belonging to any given group or kind. Scattered as 

 they are throngh the literature, it is often a matter of some 

 practical difficulty, or at least inconvenience, to bring several 

 together under the eye so as to observe readily the features 

 which it is desired to see. This inconvenience is increased by 

 the "fact that there is no uniformly adopted sequence in which 

 the various constituents are stated. 



In some cases, as in the publications of the U. S. Geological 

 Survey, the analytical order is followed. The constituents are 

 put roughly in the order in which they are determined in the 

 course of the analysis. 



Again a roughly chemical order is followed, either by put- 

 ting all the acid radicals at the top of the column and the basic 

 radicals below, as is usually done in Tschermak's Mittheil- 

 ungen, or by putting Si0 2 first, followed by Ti0 2 , then the 

 metallic oxides in the order R 2 3 , RO, R 2 0, with H 2 0, P 2 5 , CI, 

 etc. last of all, as is the custom of Roth, Zirkel, Rosen busch, 

 and many others, including the writer. But even here there 

 are variations in the order followed, as in some cases CaO is 

 put before MgO, and K 2 before ]STa 2 0. 



A third method is that adopted of late years by Pirsson. 

 This, which may be called the petrographic order, consists in 

 putting the nine most important oxides first, beginning with 

 Si0 2 and ending with H 2 0, while the subordinate and more 

 rarely occurring constituents follow after in the same general 

 order. 



A last method shows a lack of any system, the constituents 

 being put down almost at haphazard, and with little apparent 

 attempt at order or natural association. This method is fortu- 

 nately rarely met with, and is to be dismissed at once as quite 

 unworthy of consideration. 



In consequence of this confusion not only is the inconveni- 

 ence of and time required for copying and comparing analyses 

 greatly increased, but there is introduced a positive danger of 

 error due to unavoidable slips, which, unless the results are 

 carefully and laboriously checked, is apt to lead to wrong 

 conclusions. 



The benefits then of a uniform system are evident, and it is 

 probable that every petrographer has realized this to a greater 

 or less extent. After some experience in the copying and use 

 of rock analyses and full consideration of the subject, as well 

 as discussion of the matter with other petrographers, I would 

 propose that the third method mentioned above, which was, I 

 think, inaugurated by Prof. Pirsson, be generally adopted, this 

 presenting the greatest advantages to the petrographer, with a 

 minimum of disadvantage. 



