THE PROBLEM OF THE ANORTHOSITES 237 



such extremes of composition. There is no promising method of 

 doing so. Another difficulty presents itself in the very rapid 

 changes from one type to another. Even granting some method of 

 obtaining a heterogeneous liquid, one encounters the problem of 

 maintaining these sharp contrasts in adjacent liquids, for diffusion, 

 though unquestionably a slow process, would nevertheless accom- 

 plish much through moderate distances in the time required for 

 the cooling of such masses. On the other hand, it seems reasonably 

 possible both to obtain and to maintain almost any degree of 

 heterogeneity as a result of the accumulation of crystals. On this 

 assumption it is necessary to imagine the source of the olivine-rich 

 types in a portion of the magma reservoir where olivine crystals 

 had accumulated and of the feldspar-rich types where feldspar 

 crystals had accumulated. These partly crystalline masses were 

 thrust into the position where found. The greater the approach to 

 monomineralic composition, the less liquid there could have been. 

 In accordance with this conception it is found that in the allivalite 

 the feldspar crystals are arranged with their elongation in the 

 direction of flow of the sheets, and that this becomes more marked 

 the richer the rock is in feldspar. In the case of bands consisting 

 almost entirely of one mineral, which should have had very little 

 liquid to lubricate their flow, it is found that characters consequent 

 upon this are developed. Thus the nearly pure feldspar rock is 

 described by Harker as strongly fissile and the pure olivine rock as 

 foliated. 1 Possibly connected with the nearly solid condition of 

 these rocks as injected is the fact that their intrusion apparently 

 involved overthrusting, at least it is intimately connected with a 

 line of overthrusting along which earlier, later, and possibly con- 

 temporaneous movements took place. 



In correspondence with the unusual conditions of formation 

 and intrusion of these ultra-basic rocks we find them to be scarcely 

 duplicated elsewhere. The Russian ultra-basic rocks described by 

 Duparc and Pearce seem to be their nearest relatives. They show a 

 not dissimilar banding of closely related types and possibly may be 

 explained in a like manner. The peridotite dikes are described as 



1 Op. cit., pp. 72 and 87. 



