CRYSTALLIZATION -DIFFERENTIATION IN MAGMAS 417 



We shall begin then with a mass of gabbroid magma wholly 

 liquid and already injected into its present position. When cooling 

 had reached a certain stage crystalHzation began, and the earliest 

 crystals, olivine and perhaps some titaniferous magnetite, were 

 precipitated. As a result of currents in the liquid, whether from 

 thermal convection or wholly mechanical causes, free settling of 

 crystals was much hampered, so that a large proportion of the 

 crystals remained suspended. In spite of this deterrent action, how- 

 ever, some of the crystals of these early minerals did accumulate on 

 the bottom, where they formed masses of peridotite and presumably 

 also of iron ores in some cases. In the meantime, crystallization 

 of both pyroxene and plagioclase had begun. The plagioclase 

 was at first a basic bytownite. It remained suspended because 

 nearly of the same density as the liquid, and, as cooling and 

 crystallization proceeded, it adjusted its composition to approach 

 labradorite. The free sinking of pyroxene was hampered, presum- 

 ably by currents, as was that of olivine, so that no accumulation of 

 pyroxenite was accomplished. A certain amount of accumulation 

 of pyroxene did occur, however, together with that of olivine, and 

 the pyroxene crystals augmented the mass of peridotite at the base. 

 There was also a certain part of the liquid from which both olivine 

 and pyroxene crystals had settled out practically completely and 

 that contained only suspended plagioclase crystals. 



The mass is now entering a stage when in its lower portions, 

 constituting say one-fifth of the total thickness, crystals are in 

 contact with their neighbors and are beginning to grow together at 

 corners and along edges. This action becomes important when the 

 residual liquid and the crystals have each about one-half the volume 

 of the total mass. In the upper portions the crystals are somewhat 

 more widely spaced, the more so as the height above the base 

 increases, but this feature is not so very marked, because crystal 

 settling has been limited by the lack of quiet conditions. The 

 bottom-most layers including peridotite may be regarded as 

 practically entirely crystalline, with only a small proportion of 

 residual liquid. 



Origin of banding. — There is then a tendency toward a layering 

 of the mass on a very large scale but no suggestion whatever of 



