METHOD OF PLOTTING 619 



by Iddings. 6 Silica is the most abundant chemical component and that 

 which controls the development of quartz and the several chief igneous 

 minerals; alkalies determine the nature of the feldspars (or felds- 

 pathoids), which are the preponderant minerals in most igneous rocks; 

 other constituents sustain, to a considerable extent, a reciprocal relation 

 to these two. The arrangement he found distributed rocks so that, unlike 

 rocks are separated. (A possible exception seems to be the placing of 

 anorthosite near basalt and gabbro.) 



Ignoring other and theoretical reasons for the use of alkalies, 7 the last 

 point mentioned assumes an importance beyond any possible theoretical 

 consideration. The fact is that this scheme does give a good distribu- 

 tion; and after a trial of 15 or 20 constituents and combinations, chem- 

 ical, normative, and modal, the writer found no pair that gave a better 

 plot of the main variations in igneous masses. 8 There are a few cases 

 like the rocks of Rougemont, Quebec, 9 in which a change of one-third 

 or more of the mineral content in the series results in little variation in 

 its position in this diagram. On the whole, however, such cases are 

 very rare. 



FORM IN WHICH TO USE THEM 



General statement. — In using alkalies, Iddings calculated the ratio 

 of alkalies to silica, but explained 10 that the use of direct per cent of 

 alkali would give essentially similar results. Osann used a combination 

 of three constituents recalculated to 100 per cent. Various other men 

 may suggest other methods of using the same data. If any special 

 method had been generally followed, it would be inadvisable to start a 

 new plan; but no such plan is discovered. It has seemed to the writer 

 that another hiethod of arrangement of coordinates might have advan- 

 tages. It would seem especially desirable to arrange a plot for which 

 data could be roughly estimated from the hand specimen or from meas- 

 urement by the Rosiwal method. The plot finally adopted shows as 

 abscissas the excess or deficiency of silica in calculation of a norm, and 



6 J. P. Iddings : U. S. Geol. Survey Prof. Paper 18, 1903, p. 19. 



7 L. V. Pirsson : Rocks and rock minerals, p. 144, gives a diagram of the variation in 

 composition of rock families, showing smooth curves of variation when arranged in the 

 order of decreasing alkali content. Such curves would not be possible by any other 

 arrangement. 



8 Potash might be used instead of alkalies, if it proved easier to determine it. Lime 

 (or anorthite) might do as well as potash (or orthoclasej , but in general it increases 

 as potash decreases. Iron and magnesia, in the average analyses for rock families, vary 

 together, but do not separate unrelated rocks as well as silica and alkalies. Soda and 

 Alumina are almost useless for plotting such a series. 



9 J. J. O'Neill : Canada Dept. Mines Mem. 43, 1914, p. 88. 



10 Op. cit., p. 20. 



XLI — Bull. Geol. Soc. Am., Vol. 33, 1921 



