JV. L. Bowen — Crystallization. 175 



Art. X. — Crystallization — Differentiation in Silicate 

 Liquids ; by N. L. Bowen. 



Introduction. 



Ever since Darwin pointed out the possible importance of 

 the sinking of crystals in a fluid magma, this process has 

 received some attention from geologists when considering the 

 causes that have brought about the observed variety of igneous 

 rocks. Though having a few ardent advocates like Schweig, 

 the process lias met on the whole with little favor. The 

 question is not infrequently dismissed with the statement that 

 there is no evidence that crystals sink in magmas. This sum- 

 mary rejection of the process is quite inconclusive. Such a 

 statement, to have any weight, should be accompanied by a fair 

 discussion of the kind of evidence that would be expected and 

 a convincing proof that such evidence is absent. The impor- 

 tance of the process must be judged by the extent to which 

 observed results agree with the results to be expected if such a 

 process were operative. Examined in this way it soon becomes 

 apparent that the subject cannot be dismissed in a sentence, for 

 many igneous bodies showing density stratification forbid it. 

 In the writer's opinion the sinking of crystals will not only 

 escape summary rejection as a result of such examination, but 

 will be accepted as of very fundamental importance. In this 

 paper discussion will be avoided of the whole question of the 

 extent to which observed field-facts indicate the importance of 

 the sinking of crystals in magmas. The purpose of the paper 

 is principally to describe a few experiments which illustrate 

 the operation of the process in mineral solutions from which 

 crystals are forming and to apply the results to one or two 

 occurrences. 



In a recent paper on " The Ternary System : Diopside-Forster- 

 ite-Silica," the writer discussed the results which would follow 

 from the sinking of crystals in some of the liquids of the system.* 

 These liquids are rather more fluid than most artificial anhy- 

 drous melts, and among such melts should be rather favorable 

 for the sinking of crystals. It was determined, therefore, to 

 hold a small crucible of one of these melts at a temperature at 

 which it should consist of liquid and crystals and to observe 

 what effects of this kind could be obtained. The advantages 

 of using artificial melts which had formerly been completely 

 investigated are many. It was known at what temperature the 

 mixture should be held to obtain the desired effect. No com- 



* This Journal (4), xxxviii, 258, 1914. 



