260 Becker — Fractional Crystallization of Rocks. 



of viscosity would preclude the formation of eutectic mother 

 liquors. 



Eutectic mixtures by definition would have no tendency to 

 fractional crystallization however fluid they might be and how- 

 ever strong the convection currents. Where dikes represent 

 the last remnant of magma in a solidifying mass, one would 

 expect to find them of eutectic composition, as has been pointed 

 out by Mr. J. J. H. TealL* Convection being needful to frac- 

 tional' crystallization, it would seem essential that the cooling 

 magma should be surrounded by masses of a lower tempera- 

 ture, f In the case of dikes this condition is ordinarily ful- 

 filled. On the other hand, if laccolites ever form and solidify 

 without ejection at great depths and in contact with rocks of 

 high temperature, it seems improbable that convection and 

 partial crystallization would come in play to a sensible extent. 



It is difficult to see how so simple and natural a process of 

 solidification as fractional crystallization can fail to be carried 

 out in at least some rocks. Dikes ajid laccolites assuredly chill 

 from their external surfaces and (barring either an original 

 eutectic composition or insuperable viscosity) there seems no 

 way of avoiding fractional crystallization. It has often been 

 noticed that there is an accord between the order of consolida- 

 tion of minerals as observed under the microscope and the 

 arrangement of minerals in dikes, the compounds of early 

 secondary crystallization being most abundant near the walls. 

 This is of course what would be expected from a process of 

 fractional crystallization. Observation would no doubt throw 

 further light on the composition of eutectic rock mixtures. 

 Narrow stringers from a so-called "basic" dike would repre- 

 sent the mean composition at the time they were filled ; and 

 unless the composition of the magma changed during flow, the 

 stringers should represent the average dike rock. The middle 

 portion of the dike, on the other hand, should tend to display 

 eutexia. Dikes which are homogeneous ought to be eutectic. 

 Many experiments have already been made on eutectic mix- 

 tures of salt both in the dry and the wet way. It does not 

 seem impossible that experiments on eutectic mixtures of rock 

 components should give results of an approximation sufficient 

 for the purposes of lithology. 



Few, I believe, will maintain that any great progress has 

 been made in explaining the theory of the segregation of 

 magmas into partial magmas. Mr. H. Backstrom 7 J for example, 



* British Petrography, 1888, p. 401. 



f Dr W*. F. Hillebrand reminds me that the changes in density of the mother 

 liquor during- crystallization will of themselves induce convection, though perhaps 

 not powerful currents. 



% Jour, of Geol., vol. i, 1893, p. 773. 



