414 N. L. BOWEN 



When we come to a consideration of the more intimate details of 

 crystallization we find this same tendency to urge an advantage of 

 convection over settling, an advantage that has no real existence. 

 In the case of the bodies of magnetite, for example, it is pointed 

 out that the best concentrations are centrally placed in the gabbro 

 and not at the bottom, and to explain this it is pointed out that 

 magnetite is, "at least partly, later in time of crystallization" 

 and ''would remain liquid until the lower parts of the chamber 

 were filled with rock." Is the explanation not equally applicable 

 to the process of crystal settling ? Is there anything in that process ■ 

 that demands the settling of a crystal before it exists ? It is true 

 that the writer referred to the accumulations of iron ore as evidence 

 of settling of crystals because they have been described by others as 

 occurring at the base of the Duluth mass, and it is safe to say that in 

 many cases some of the iron ore crystallizes very early, but the 

 writer has always been opposed to the unrestricted acceptance of 

 the notion that all the ore minerals separate early. ^ 



We come now to that very special feature of the near-uniformity 

 in the nature of the plagioclase in all parts of the gabbroid portion 

 of the mass, a feature that has been considered to favor especially 

 the idea of convection. Vigorous convection would of course keep 

 plagioclase crystals suspended and in a slowly cooled mass would 

 permit perfect adjustment of their composition to that of the 

 surrounding liquid, so that there would be no necessity of the 

 appearance in the final product of any very basic plagioclase, by- 

 townite, or anorthite. But the role assigned to convection by 

 Grout is not so much that of keeping crystals suspended as that of 

 aiding in their accumulation in layers by dragging along the bottom, 

 and, at a time when the rhythm of crystallization was in a phase 

 such that the crystallization of plagioclase dominated, the layer 

 formed is supposed to show a corresponding dominance of plagio- 

 clase. These plagioclases, dragging behind and caught in the 

 viscous border, have according to Grout "there slowly, maintaining 

 equilibrium, adjusted their composition to that of the surrounding 

 magma." It should be realized, however, that in so far as basic 



^ N. L. Bowen, "The Order of Crystallization in Igneous Rocks," Jour. GeoL, 

 XX (1912), 457. 



