188 N. L. Bowen — Crystallization — 



diopside glass expands at nearly the same rate as diabase glass, 

 so that at high temperatures the ratio of densities is nearly the 

 same as at low temperatures.* It is, therefore, safe to state that 

 the liquid in the experimental case, closely similar to diopside 

 liquid, would still be considerably heavier than diabase liquid 

 at high temperatures, especially since the diabase liquid must 

 be considered somewhat lighter in its natural state owing to 

 the presence of volatile constituents. Nothing is known of 

 the change of density of forsterite with temperature, but it is 

 known that magnesia-iron olivine would be heavier than pure 

 magnesian olivine at any temperature. With the crystals 

 heavier and the liquid lighter in the natural case as compared 

 with the experimental case, the rate of sinking should be 

 greater from this cause. 



The liquid from which the forsterite crystals sank in the 

 experimental example had the composition of a lime-magnesian 

 pyroxene with a little excess silica. Diabase may be considered 

 to be made up of this same material with plagioclase material 

 and certain ferruginous material added. The plagioclase 

 material would undoubtedly render it more viscous, but the 

 ferruginous material would compensate this effect. It is 

 probable, therefore, that even anhydrous diabase liquid is as 

 fluid as the liquid in the experimental example. Day, Sosmjin 

 and Hostetter describe diabase liquid as rising by capillarity 

 in the walls of a porous alundum crucible " like water in filter 

 paper."f 



With the addition of water and other volatile constituents 

 the viscosity of diabase liquid may be considered probably as 

 small as that of the artincal pyroxene melt even though the 

 temperature was considerably lower. Morey has demonstrated 

 the remarkable effect of water in lessening the viscosity of 

 silicate liquids.^ 



If the viscosity of the liquid and the density difference 

 between the olivine crystals and liquid are considered compar- 

 able in the natural case with the same factors in the experi- 

 mental case, a moderate assumption, the larger olivine crystals 

 (0'7 mm diam.) in the olivine-diabase ledge would have settled at 

 the rate of about 1 meter per hour. The thickness of the sill 

 is from 200-300 meters. If the time occupied in cooling 

 through the interval of crystallization of olivine, partly alone 

 and partly in company with other minerals, were as much as 

 200-300 hours, it would suffice for the collection of the olivine 



* Unpublished data furnished by Dr. Sosman. 



f Determination of Mineral and Kock Densities at High. Temperatures, this 

 Journal (4), xxxvii, 34, 1914. 



\ New Crystalline Silicates of Potassium and Sodium, their preparation 

 and general properties, Jour. Am. Chem. Soc, xxxvi, 226, 1914. 



