LATER STAGES OF EVOLUTION OF IGNEOUS ROCKS 15 



opposed to the process seem to expect too much of it, especially in 

 the way of a sorting of the individual minerals. Thus it is stated 

 that the iron ores do not accumulate at the bottom of the sill. This 

 statement is not strictly true, for sometimes they do ;^ nevertheless 

 such a phenomenon is to be expected only rarely. Let us assume 

 the most favorable case. The separation of the ore minerals as the 

 first products of crystallization takes place and their separation is 

 complete before other crystals appear, a thoroughly unlikely 

 assumption. The period of crystallization of the ore minerals 

 alone would be even in this case exceedingly brief for bodies of 

 moderate dimensions, and such are most bodies whose floors are 

 visible. It is to be noted, moreover, that the very heavy minerals 

 usually occur as relatively small crystals and the size of crystals 

 is a factor of prime importance in determining their rate of sinking. 

 Very small crystals of magnetite would, as a rule, make little head- 

 way before crystals of other minerals were precipitated.^ Through- 

 out most of the period of crystallization of the magma the ore miner- 

 als are merely one of a number of kinds of crystals and their sinking 

 is seriously interfered with by the presence of other crystals which 

 may tend to sink much more slowly or in rare cases perhaps even to 

 float. In general, we must conclude that there may be a long 

 period in the history of a crystallizing magma during which the 

 crystals tend to sink through the liquid as a swarm, with little 

 tendency to relative movement between the different kinds of 

 crystals, the sinking being determined rather by the mean density 

 of the swarm. The result of this process is, in general, not the 

 sorting of individual minerals, but the downward movement of the 

 heavier minerals of early crystallization as a group and the con- 

 trasting of the minerals of the basic rocks with those of the inter- 

 mediate rocks and with those of the acid rocks. 



^ The lower part of the Duluth gabbro is in places 90 per cent titaniferous magne- 

 tite (Bayley, Jour. Geol., II [1894], 818). FosUe recently described a Norwegian 

 example {Norges. Geol. Unders. Aarbok IV [1913], 66), and many others could be 

 cited. 



= Assuming the densities of magnetite and orthoclase, for example, to have roughly 

 the same relative values at high temperatures as at low temperatures, we may deduce 

 that magnetite crystals o . i mm. in diameter would sink in most granitic magmas no 

 faster than feldspar crystals o . 4 mm. in diameter. 



