14 JOURNAL OF GEOLOGY— SUPPLEMENT 



feeding reservoir as crystallization proceeds. Some composite 

 dykes at Cape Ann, on the Massachusetts coast, give distinct 

 evidence of this process. These dykes are more salic in the central 

 portion, and in places there is a continuous gradation in compo- 

 sition toward the more basic margin. Here the variation might 

 be referred to differentiation in place, but that this conclusion 

 would be erroneous is shown by the fact that in other places in the 

 same dyke the more basic phase has been completely frozen, and 

 only after a reopening of the dyke by fracture has the more salic 

 phase entered. This refracturing has in some places been repeated. 

 The evidence is clear that the dyke fissure has served for the passage, 

 throughout a considerable period of time, of material of continually 

 changing composition. 



In general this explanation of basic borders when they are shown 

 by small bodies such as narrow dykes or sills seems preferable to 

 any process of differentiation in place. 



The most promising processes for the production of differentia- 

 tion are those involving crystallization and the relative movement 

 of the crystals with respect to the liquid from which they separated. 

 Of these the most important during the period when the magma is 

 still dominantly liquid is the movement of crystals in the Hquid on 

 account of differences in density, usually the sinking of crystals. 

 During the period when the magma has largely crystallized, "the 

 straining-off or squeezing-out of the residual fluid magma "^ is the 

 most important process. It is to be noted, however, that in the 

 strained-off , reintruded liquid from the latter process, the sinking 

 of crystals may begin again, so that it is difficult to refer either of 

 these processes to any definite period in the complete history of any 

 given magma. 



No one can reasonably question the existence or the effectiveness 

 of a process of squeezing out of residual magma, but many have 

 maintained that there is no evidence that crystals sink in magmas. 

 Numerous examples, however, have now been described of sheetlike 

 bodies with both roof and floor exposed and the differentiates 

 arranged in such a way as to point clearly to the sinking of crystals. 

 An antagonistic view is, therefore, no longer tenable. Those 



' A. Harker, op. cit., p. 323. 



