GEOPHYSICAL LABORATORY. 153 



lengthened, and the expected effect of increased viscosity (decreased 

 fluidity) in slowing up the rate of sinking was easily demonstrated by 

 adding sihca to the mixtures. In one case, too, of a mineral with 

 exceptionally low density, not common in nature, a floating upwards 

 was demonstrated. 



In applying to the case of natural rocks the facts developed experi- 

 mentally, it is necessary to make due allowance for a difference of 

 conditions. The molten rock is continually cooling, sometimes slowly, 

 sometimes more quickly. The mineral which crystallizes first is 

 shortly joined by crystals of another kind, these two kinds by yet 

 another, and so on as the temperature falls until the time when the 

 whole mass has solidified. During most of the period of crystallization 

 many different kinds of crystals occur in the liquid, the heavier ones 

 tending to sink rapidly and the lighter ones less rapidly, or possibly 

 even tending to float in rare instances. The results should be that they 

 would seriously interfere with one another as far as relative motion 

 is concerned and that the whole mass of crystals would tend to sink 

 rather as a swarm. It is precisely this association of minerals of differ- 

 ent densities in the same rock which has led some geologists to conclude 

 that the sinking of crystals can not be an important matter. They 

 reason that the action of gravitj^, if it is to be regarded as a factor of 

 any considerable importance, should cause a complete sorting of the 

 heavier and the lighter minerals; but it is plain from the above con- 

 siderations that this is by no means a necessary conclusion. However, 

 in the very earliest stages of the crystallization of a slowly-cooled mass 

 of molten rock, one would expect a considerable degree of sorting of 

 the different kinds of crystals and a greater degree of fluidity than in 

 the later stages. It is noteworthy, therefore, that whenever rocks do 

 consist entirely of crystals of a single mineral, this mineral is always 

 among the few belonging to the generally accepted minerals of earUest 

 crystallization, viz, pyroxene, olivine, the calcic plagioclases, and the 

 iron ores. Not infrequently, too, the heaviest of these minerals have 

 been found accumulated at the bottom of sheet-Uke masses of igneous 

 rock in a manner which leaves no room for doubt as to their collection 

 under the influence of gravity. 



Even in a case where no sorting of individual minerals is accom- 

 plished the sinking of minerals of various kinds as a swarm should 

 nevertheless be a potent factor in producing igneous rock types. It 

 should bring about the more or less perfect separation of the minerals 

 of early crystallization as a group from the minerals of the middle 

 period as a group, and both of these from the minerals of the later 

 period. This result is actually found in nature in the common gabbro- 

 diorite-granite sequence. 



With an emphatic affirmative, then, experimental investigation 

 answers the query as to whether crystals can sink in bodies of molten 



