New Zealand Inatifiitf Science Congress. 481 



the collecting, especially by Marshall in the Oaniaru district, has been much 

 more exhaustive for this stage than for any of the others. The difference 

 111 station of the species found in the Ototaran and Hutchinsonian probably 

 accounts for the absence in these stages of iiiany species common to the 

 Waiarekan and Awanioan, and also for the ])resence of other species absent 

 from the two last-named stages. The general result of the analysis is to 

 show that the molluscan fauna did not change greatly during the Oamaruian, 

 and hence distant correlations with individual stag(>s based solely on 

 mollusca are difficult to establish. A closer study of the fossils on 

 evolutionary lines might lead to the discrimination of species now lumped 

 together which might prove to have a zonal value. 



" The Geology of the Middle Clarence and lire Valleys, East Marlborough, 

 New Zealand,"" bv Dr. J. A. Thomson. (Printed in the Transactions, 

 pp. 289-349.) 



Friday, 7th February, 1919. 



Present : Mr. P. G. Morgan, President, in the chair, and a number of 

 others. 



Fapvis. ■" A Quantitative Study of the Silica-saturation of Jgiieous 

 Rocks." by Dr. J. A. Thomson. 



ABSTRACT. 



Variation curves for chemical constituents (mainly oxides) plotted 

 against silica have been given by Harker for Pacific and Atlantic rocks, 

 but trial plottings of a large number of analyses show that the points 

 representing given rocks group themselves together into curved belts (a. 

 sort of Milky Way) and do not tend to lie on lines, as Marker's diagrams 

 might suggest. If these belts could be sufficiently defined by the ])lotting 

 of all reliable analyses, each rock could be then classified as abnormally 

 high, normal, or abnormally low for each chemical constituent compared 

 to silica, and a valuable means of comparing rock-analyses would result. 

 The combination of all such classifications for each important constituent, 

 however, would yield more classes of analyses than there are superior 

 analyses; and would be cumbersome. The attempt has therefore been made 

 to obtain a small number of functions of the analyses combining relation- 

 ships between various oxide molecules such as exist in minerals. A norm 

 somewhat similar to that of the American classification has been adopted, 

 but in order to avoid the difficulty of attempting comparisons of such related 

 molecules as orthoclase and leucite, albite and nepheline, hypersthene and 

 olivine, in Avhicli the second mineral in each of the groups named is under- 

 saturated in silica, it has l)een assumed in calculation that unlimited silica 

 is available. The amount of silica actually used in satisfjdng the various 

 bases on the assumption of complete silica-saturation is then compared with 

 the amount shown to be actually present by the analysis, and a percentage 

 figure is obtained expressing when positive the amount of free quartz 

 present in the norm, and when negative the degree of undersaturation of 

 silica. A further difference from the American norm is also the calcula- 

 tion of excess Al^Og and FcjOg to hydrous silicates Al2O3.SiO2.2H2O 

 and Fe2O3.SiO2.2H2O, since alumina and ferric iron control at least an 

 equivalent of silica in the alferric minerals after possible feldspars and 

 metasilicates have been calculated. By this means a quantitative esti- 

 mate of silica-saturation for any rock-analysis is obtained. This in turn 

 is plotted against total silica, and a variation belt obtained as in the case 



16— Trans. 



