MAGMATIC DIFFERENTIATION OF IGNEOUS ROCKS 237 



FeO, MnO, FczOj, and MgO (in the order mentioned) reduce the 

 viscosity of Na2Si03 • SiOz, while CaO and AI2O3 increase the 

 viscosity (AI2O3 most strongly). 



Applied to petrography it will appear from the above that 

 melted (anhydrous) granite, quartz-porphyry, rhyolite, and obsidian 

 at one atmosphere pressure are extremely viscous, even heated to 

 100°, 200°, or 300° above the begiiming of the temperature of 

 crystallization. 



Molten (anhydrous) alkali- syenite, trachyte, and phonolite 

 are still rather viscous, presumed one atmosphere pressure. Molten 

 (anhydrous) gabbros are somewhat more thin, and basalts and 

 related rocks rich in iron are rather thin. 



The statements above are valid for molten anhydrous masses, 

 or generally molten masses without dissolved light volatile com- 

 pounds,^ at one atmosphere pressure. 



As to the direct influence of the pressure, reference may be made 

 to Johnston and Adams in the American Journal of Science, XXXV, 

 226-31. The result is "that the effect of uniform pressure is always 

 to increase the viscosity (water and certain dilute aqueous solutions, 

 both of low temperature, excepted)." 



According to this it must be supposed that the direct effect of 

 the pressure is to increase the viscosity of the silicate melts. How- 

 ever, more important in many cases is the indirect effect of the 

 pressure since the content of the magma of dissolved light volatile 

 compounds is increased with the pressure, and these volatile com- 

 pounds (H2O, etc.) reduce the viscosity to a very considerable 

 degree. 



This applies especially strongly to granitic magmas which 

 according' to geological experience must have been rather thin, 

 and still more to granite-pegmatite magmas which even at so low 

 a temperature as about 7oo°-8oo° must have been especially thin 

 on account of their relatively rich content of H2O, etc. 



The effusives, on the other hand, must as a rule have contained 

 much less H2O, etc., and their degree of thinness, at least in many 



^ I do not here take into consideration that the silicate slags from, e.g., a blast 

 furnace or a matte-smelting contain dissolved a very small quantity of gas, which at 

 least to a substantial extent escapes partly before the begiiming of the crystallization 

 and partly on an early stage during the process of crystallization. 



