9. Nature and relation of pyroxenes in several series of rocks. For explanation see text. divisions along the margins. The limits drawn by Hess (1941, p. 518) for the fields of diopside, endiopside, salite, ferrosalite and heden-bergite are omitted. The small crosses indicate the optically determined compositions of phenocrysts in Japanese non-alkaline basalts and andesites, showing their normally augitic to subcalcic nature and lack of continuous variation towards the rare pigeonite phenocrysts indicated by the two larger asterisks marked by p. Triangles indicate the optically determined compositions of groundmass pyroxenes in the same Japanese rocks and demonstrate the continuous variation from the lower calcic range of augite through subcalcic augite to pigeonite and rarely ferropigeonite, for these rocks are abnormally rich in iron. (Data from Tsuya, 1937.) Black circles show the average normative compositions of the representative rock series considered in Table VIII as follows:—A = Antarctic dolerites; B = Plateau basalts; G = German tholeiites; K = Karroo dolerites; O = Otago dolerites, basalts and tholeiites; P = Palisadan dolerites; S = Scottish and North of England dolerites and tholeiites; Sp = Spitz-bergen dolerites; T = Tasmanian dolerites; W = Watchung basalts, the effusive products of the Palisadan magma. The black rings indicate the chemically determined composition of:—D = the pyroxene from the Deccan basalt submitted to X-ray spectrographic examination (see below) and M = the ferro-pigeonite in the inninmorite of Mull. I, II, III and IV show the fields of composition-range for successive generations of pyroxene in the Karroo dolerites as determined optically by Walker and Poldervaart (1941, Fig. 5) being respectively I = initial hypersthene, II = early magnesian pigeonite, III = “hypersthene augite”—i.e., subcalcic augite (with a little augite poor in lime), IV = late-formed pigeonite. It will be noted that the average composition of the modal pyroxene in the Karroo dolerites is less ferruginous than the average normative composition K. It will further be seen that the sequence I-II-III is almost exactly that found by Kuno (1933) in a Japanese cristobalite-bearing olivine, augite-hypersthene basalt and plotted by Alling (1936, p. 218), whose “Pigeonite No. 2″ (with 2V = 48°–50°) in Kuno's and our view [and indeed on his own classification (Alling, 1936, p. 102)] is an augite poor in lime. It lies on Alling's (op. cit., pp. 100, 218) co-tectic line (CT—CT on our Fig. 9) near the turning point in the