12 JOURNAL OF GEOLOGY—SUPPLEMENT 



from this the rocks themselves show that diffusion is quite Hmited 

 in its scope. When centers of crystallization are once estabHshed 

 at the border, diffusion as free as that postulated should lead only 

 to the continued growth of the crystals so initiated. There is no 

 reason why new centers should be established. The result should 

 be the growth inward from the contact of huge crystals of the min- 

 erals of early separation. In reality, of course, a new center of 

 crystallization for these minerals is always established within a 

 few millimeters, a fact which speaks etoquently for the radius 

 through which diffusion acts freely. 



The same objections apply to the supposed maintenance of 

 approximate uniformity in the composition of the liquid through 

 the agency of convection currents, or, better stated, perhaps, the 

 continual supply of new liquid to the border portion. The same 

 huge crystals at the border should result. 



Let us suppose, however, that, in some accidental manner, say, 

 before convection has fairly started, many centers of crystallization 

 of these early minerals are established near the border. If these 

 centers were fed by convection currents coming from distant parts 

 of the magma, none of the crystals except those actually attached 

 to the wall rock would remain in the border portion. They would 

 be carried away and distributed by the same currents that are 

 supposed to feed their growth. If the numerous early crystals 

 estabhshed near the border are supposed to grow by diffusion of 

 material from a distance, there is a similar difficulty. Such a 

 process would require a long period of time and there is no reason 

 to believe that the crystals would remain there when the border 

 concerned is, say, the upper margin of a sill. The crystals would 

 in this case sink out of the border portion.^ 



' Experimental work forbids the belief that diffusion could take place freely 

 through considerable distances in a period of time too brief to permit significant sinking 

 of crystals (cf. Marker, op. cit., p. 322). For example, a liquid consisting of 56 per cent 

 CaMgSi206 and 44 per cent MgSiOi permits quite definite effects in the way of sinking 

 crystals at 1430° in only 15 minutes, but if one tries to make the foregoing liquid by 

 carelessly mixing diopside and MgSiOj and holding the mixture at a temperature 

 considerably above 1430°, httle headway toward a homogeneous melt will be made in 

 15 minutes. The materials must be very carefully mixed in the form of a fine powder 

 (N. L. Bowen, Am. Jour. Sci. [4], XXXIX [1915], 176). 



