412 . N. L. BOW EN 



to act as a unit and to sink in virtue of its greater aggregate density. 

 It was evidently suggested by Daly's two-phase convection, in 

 which the two phases involved were liquid and gas that gave a 

 low aggregate density. It is known that masses of foam possess 

 some of the attributes of a rigid body, and it might therefore be 

 expected that localized highly vesicular masses of lava might 

 act as a gravitative unit, but nothing is known to warrant the 

 extension of the idea to a liquid containing suspended crystals. 

 Moreover there appears to be no necessity for appealing to this 

 unlikely process, since ordinary thermal convection would be 

 equally effective and can hardly fail of occurrence in a cooling 

 mass of magma. ^ However this may be, we may now proceed to 

 consider the supposed effects of convection. Grout makes some 

 calculations on the rate of convection which indicate that crystals 

 formed at the cooling portion near the top could be carried to 

 the bottom in half an hour. No reason is offered for this 

 precipitate haste. Undoubtedly many years were available; 

 indeed Daly considers the time available to be embarrassingly 

 long and offers it as an objection to crystal settling that the 

 results are often . very meager when the duration of crystalliza- 

 tion is considered. Moreover, it should be realized that a cur- 

 rent that could carry crystals down in half an hour could carry 

 them up again in half an hour, particularly a current in which 

 a mass of liquid with suspended crystals is supposed to act 

 as a unit. The lagging behind and lodgment of the crystals 

 at the bottom as the currents sweep over the floor, which Grout 

 considers would occur, would undoubtedly be a factor in the case 

 of ordinary thermal convection at any rate (its possibility in the 

 two-phase convection pictured by Grout may be doubted). It 

 should be realized, however, that this occurs in spite of convection 

 and in virtue of the superior density of the crystals rather than in 

 virtue of convection, and the really important question is whether 

 crystals would accumulate at the bottom more effectively by this 

 dragging-behind process than they would by simple settling. In 

 other words, does convection act as an aid or as a deterrent to 



' It should be noted that the writer has never doubted the reality of convection in 

 magmas. He merely doubts its efficacy in producing differentiation. 



