MOVEMENTS IN CRYSTALLIZING MAGMA 257 



in currents. To this there is no question at all. But, in addition 

 to banding, the Duluth rocks show an orientation. The combina- 

 tion is explained only by circulation. 



Convection, as a general stirring, may be supposed to interfere 

 with crystal settling. Probably in one sense this is true, for any 

 liquid containing sediments clears more rapidly if left quiet. But 

 the arguments for a circulation at Duluth are conclusive and have 

 not been answered. Settling, if it occurred at all (and it probably 

 did), occurred during circulation. Since the calculated rates of 

 motion indicated that convection was more rapid than settling,^ 

 it is clear that convection might aid in bringing a crystal formed at 

 the top of a magma chamber to some point near the bottom. Then 

 as the current moved along the floor there might be settling enough 

 to cause the growth of a rock layer at the bottom. Settling, how- 

 ever, is not the only process by which rock may grow. Simple 

 cooling makes the shell of solidified wall or floor increase in thick- 

 ness. The bottom layer of the moving current may become part 

 of the floor by cooling and increasing in viscosity as well as by 

 settling. Thus it may happen that even when circulation tends 

 to prevent settling some settling may occur, and certainly some 

 rock may accumulate along the floor. 



In order to "make a case" for convection certain complications 

 in the process were omitted in the original paper.- No doubt 

 convection is far from a simple motion in large sheetlike chambers. 

 There are the irregularities in the banding of most banded igneous 

 rocks to prove the complexity of motion. There may even have 

 been an approach to the theoretical hexagonal cells^ of circula- 

 tion. These, however, would have little effect on the result. No 

 crystals would deposit on the sides of a cell — there is no support 

 on the walls of such cells. Deposition would occur in bands 

 on the bottom, almost as if the circulation was the simple case 

 described. 



' Frank F. Grout, op. cit., p. 494. 



= Unfortunately an error occurs in the copy of the calculation for gas-phase 

 convection, op. cit., p. 491 : "lbs. per sq. inch " should be atmospheres. It seems correct 

 values were used in calculation, but stated wrongly in copying. 



3 See Jour. Geol., XXIV, pp. 219 ff. 



