66o 



HENRY S. WASHINGTON 



oxide, the ordinal value for each successive tenth of an inch, and 

 taking the mean. The result of this process is given in II 

 below, that given by the previous process being given in I. 



The two agree fairly well, and are of the same general char- 

 acter, though there are marked discrepancies, II being decidedly 

 more basic in all respects than I. What may be the explanation 

 of this, I am not mathematician enough to say. But the general 

 agreement would indicate that one of the two, or their mean, 

 cannot be far from the truth, i. e., as near as the data at hand 

 permit of approximation. 



It is of interest to note that I have been unable to find the 

 analysis of any rock which agrees at all closely with either of 

 these two results. Those which are as hiof-h in alkalis being 

 lower in bivalent oxides, while those which agree in this respect 

 are lower in alkalis and alumina. Whether this indicates that 

 there are serious sources of error in the method employed, or 

 else that some undifferentiated magmas may possess chemical 

 compositions not corresponding to those of rocks as yet known, 

 is a question which cannot be decided here. It would seem as 

 if there were nothing a priori contrary to the latter hypothesis. 



In this connection Harker's^ remark may be cited: " Given 

 a series such that its diagram has markedly curved lines, the 

 result of the admixture of two members may be something not 

 only foreign to the series, but highly peculiar by comparison 

 with igneous rocks in general." It is true that Harker was dis- 

 cussing the case of the mixture of two members of a series, but 



' Marker, op. cit., p. 395. 



