CRYSTALLIZATION-DIFFERENTIATION IN MAGMAS 409 



as is introduced by crystal . settling. As a result of this more 

 advanced consolidation of outer layers an upper layer may become 

 more or less definitely a part of the upper competent member of 

 the stratified series and may act in conjunction with it during the 

 process of up-arching. The later differentiate would not normally, 

 therefore, be the uppermost. It would, indeed, be located more 

 nearly toward the central part of the mass according to the degree 

 of this freezing inward from the roof. 



Sheet with thin central late differentiate. — In certain cases where 

 the freezing has been nearly symmetrical at the top and bottom of 

 a sheetlike mass a horizontal thrust, occurring at a time when there 

 was still a thin layer mostly liquid centrally placed, might cause a 

 general shortening of the lateral dimensions of the sheet, with 

 consequent thickening vertically. These would take place con- 

 comitantly and would be accomplished by closer packing of the 

 crystals in the outer layers, where crystallization was well advanced, 

 and a thickening of the originally thin central Hquid layer by inflow 

 of the interstitial liquid from the outer layers. There would thus 

 be formed a mass having a thin central later differentiate with an 

 intrusive to transitional relation to its surroundings, and with a 

 distinct tendency toward a streaky development, for the liquid 

 squeezed in from the layers where crystallization was well advanced 

 would differ from the liquid already present in the narrow central 

 band when this action began. 



Thick central late dijferentiate. — There is also a very simple 

 method whereby a sheetlike mass, whether dyke or sill, can develop 

 locally a very thick, centrally placed layer of this late differentiate. 

 It is the warping of the containing walls of the sheet, with con- 

 sequent thickening of the sheet in a certain place and thinning of it 

 elsewhere. If this takes place at any time except at a late stage of 

 crystallization it will normally have no significant result in the way 

 of differentiation. If it takes place at a time when the crystal 

 mesh has considerable strength it may have important conse- 

 quences. We shall consider a segment of the sheet including a part 

 that has been thickened. Figure 3a represents part of the original 

 sheet and Fig. 36 the same after warping of the walls. In the section 

 where the sheet is thinned the crystal mesh will be broken down and 



