ORTGTN TN IGNEOUS ROCKS. 89 
character, which is rare in members of this class ; and, second, because 
of the miusiial position of the olivine in the groundmass, indicating 
that it is here the youngest instead of the oldest constituent, as is 
generally the case." 
As has been stated on page 74, Morozewicz « has described the 
formation of corundum from a number of magmas that Avere super- 
saturated with alumina. The supersaturated alumina-silicate mag- 
mas had the general composition MeO,mAL03,MSi02(Me=Ko, Na.,, 
Ca, and w=2 — 13). In these magmas, on cooling, all the excess of 
alumina (over ?n=l) separates out in the form of corundum crystals, 
this taking place when magnesia and iron Avere not present. 
In these experiments the silicates that were used to form the 
magmas were for the most part those of calcium, sodium, jind potas- 
sium. When alumina w^as present in them in excess of m=l, it 
would separate out in the form of corundum; but when the magma 
was not saturated with alumina or fa was equal to 1 or less, all the 
alumina was used up in the formation of double silicates. WHien 
magnesia was present, the excess of alumina united with the mag- 
nesia to form the mineral spinel, and if not enough magnesia was 
present to unite with all of the ahmiina some corundum was formed. 
When there was an excess of silica, or n was greater than 0, the 
excess of alumina united with the silica to form the mineral silli- 
manite, and if alumina still remained it separated out as corundum. 
Wlien, however, magnesia and iron are present w^ith the excess of 
silica and alumina, cordierite and spinel are the minerals formed. 
From these experiments it would seem that when a magma is com- 
posed of silicates of the alkali and alkali-earth metals — sodium, potas- 
sium, and calcium — the alumina dissolved in this magma will, to a 
certain point, unite to form double silicates of alumina with these 
other bases; but when the ratio of alumina to these bases is greater 
than 1 : 1 the excess of alumina will separate out as corundum, except 
when influenced as just described. Magnesia, it would seem, does 
not have so strong an affinity for the formation of silicates, and 
when present in the magmas influences the separation of alumina as 
a double salt of magnesia and alumina in the formation of the 
mineral spinel, or Avith an excess of both alumina and silica in the 
presence of iron in the formation of the mineral cordierite. 
In some earlier experiments MorozeAvicz '^ has shoAvn that alumina 
will readily dissolve in a molten magma having a composition 
approximately that of the basic magnesian rocks, and that upon 
cooling the alumina separates out as corundum and spinel. 
"Tschermaks mineral, u. petrogr. Mittheil., vol. IS, 189.S, pp. 1-90 and 105-240. 
«>Zeitschr. fiir Krystall., vol. 24, 1895, p. 281. 
