176 N. L. Bowen — Crystallization — 



plication due to a chemical action, such as oxidation of ferrous 

 iron, could ensue, as might be the case if a natural rock were 

 used. The results obtained should therefore be capable of 

 straightforward, unequivocal interpretation. That crystals 

 would sink in such a liquid when they are heavier than it 

 seems an obvious necessity. Nevertheless, in the case of many 

 natural occurrences where the conditions are equally favorable 

 the process has been questioned. For this reason the exper- 

 imental results will be described. 



The Sinking of Olivine. 



A homogeneous mixture of the composition of a pyroxene, 

 diopside 56 per cent, MgSi0 3 44 per cent, was first used. A 

 quantity of this mixture was placed in a small platinum cru- 

 cible 2*5 cm deep and l*5 cm in diameter, and the crucible was 

 suspended in a platinum-resistance furnace. The temperature 

 was measured by a Ft : JPt-llh thermoelement whose junction 

 was nearly in contact with the surface of the melt. After 

 raising the temperature until the charge was entirely liquid 

 and filled the crucible to a depth of about l*5 cm , the temper- 

 ature was lowered slowly to 1430°. At approximately 1460° 

 olivine (forsterite) begins to crystallize out. This continues 

 until the temperature 1430° is readied, when the cooling was 

 checked and the temperature maintained at tiiis value for 

 varying periods of time. The charge should consist of olivine 

 crystals 4 per cent, liquid 96 per cent. Sinking of crystals 

 should result in this case merely in their accumulation towards 

 the bottom, the liquid remaining of uniform composition 

 whether crowded with crystals near the bottom or quite free 

 from them near the top. When the temperature had been 

 maintained for the desired length of time the crucible was 

 quickly removed from the furnace and cooled in water in order 

 to prevent further crystallization. 



When this was successfully accomplished it was found that 

 the accumulation of crystals towards the bottom of the crucible 

 was marked, the effect becoming more evident when the 

 time allowed for the sinking of crystals was increased. The 

 effect was plainly visible with the naked eye, even after the 

 brief period of 15 minutes. The glass at the top of the crucible 

 was clear and transparent ; with increasing depth it assumed 

 an increasing milky appearance, and at the bottom it was an 

 opaque white. The milky, opaque appearance was caused by 

 the dispersal of light due to continuedreflection from the faces 

 of the small olivine crystals, and it increased in intensity with 

 the increasing abundance of olivine crystals towards the bottom. 

 The individual olivine crystals were plainly recognizable with 

 the aid of a hand-lens. 



