94 F. LOEWINSON-LESSING 



admit that the oHvine, which is formed at the begimiing of crystal- 

 lization, really sinks, that this sinking is not counterbalanced by cur- 

 rents and movements of the magma, that the sunken crystals are 

 not dissolved, that the deeper horizons of the magma have not a 

 higher temperature, that the crystals sink to the bottom, that 

 there is a bottom, that the crystallization is going on very fast, 

 and so on. 



And even if all these suppositions can be confirmed, the result 

 of such a differentiation would be very small, as will be shown 

 below by a series of calculations. Those minerals which can be 

 fixed by chilling do not correspond to the final equihbrium, as is 

 illustrated by certain alloys of metals, which when chilled show 

 components and structures not formed in these alloys when normally 

 crystallized. Rankin has also found in the system CaO-Al203- 

 SiOa combinations which do not occur in natural rocks, but no one 

 would therefore conclude that these combinations must also occur 

 in the igneous rocks. The conditions of crystallization, the rate of 

 cooling, and the admixtures are evidently different in the two cases. 

 Since the relatively simple mixtures of chemically pure components 

 used in the laboratory and the complex natural magmas are essen- 

 tially different, deductions from laboratory experiments with simple 

 mixtures should be applied very cautiously and with certain restric- 

 tions to the very complex natural magmas. So, for instance Bowen 

 states several times that spinels must crystalhze from a basaltic 

 magma, but we know that basalts do not contain spinels, neither 

 is there clinoenstatite in the igneous rocks. The conditions of 

 crystallization of a simplified ' 'haplobasalt " and a natural basalt 

 are different. Not only is the composition of the latter much more 

 complex, but the gaseous components of the magma, the mineral- 

 izers, those ingredients whose importance is now so often noted, 

 must be considered. Experimental laboratory work is certainly 

 very important for the problems of petrogenesis, but we must not 

 forget that our experiments are performed with simplified mixtures 

 and in simplified conditions. Let us not repeat the mistake of Mohr 

 who affirmed that a simple experiment in a crucible was sufficient 

 to refute all the theories of the plutonists. Let us continue the 

 experiments with simple mixtures of cheniically pure compounds, 



