Differentiation in Silicate Liquids. 177 



When the time allowed was 45 minutes, about one-half the 

 depth of the crucible was a clear glass ; the lower half milky, 

 to opaque at the bottom. After SO minutes the upper 13 , 5 mm 

 layer was entirely free from olivine crystals, all having collected 

 in a l;5 mm layer at the bottom. In all cases the liquid portion 

 was of uniform composition from top to bottom, as shown by 

 the uniform refractive index of the glass. 



The charges are traversed by innumerable cracks as a result 

 of the quick cooling which must be resorted to. It is, there 

 fore, very difficult to cut a single section from the top to the 

 bottom of the crucible to show the increasing proportion of the 

 olivine crystals. One charge which was held only 15 minutes 

 was broken out of the crucible in G approximately equal layers. 

 Apiece from each of the 2d, 4th, and 6th or bottom layers w T as 

 selected and sections were ground. Photomicrographs of these 

 sections are shown in fig. 1, which indicates, at the same 

 time their positions in the crucible. The very marked effect 

 obtained in only 15 minutes is plain. The manner in which 

 the crystals are packed together at the very bottom is clearly 

 shown. The crystals in this bottom layer are noticeably smaller 

 than those higher up, owing to the fact that they sank only a 

 very short distance before coming to rest on the bottom, and 

 had therefore less opportunity for growth than those which 

 sank a considerable distance. The lowest of the three sec- 

 tions of fig. 1 is shown in greater magnification in fig. la. 

 Close to the middle portion of the right-hand edge, three 

 clusters of crystals are shown, each collected about a bubble. 

 These groups of crystals have weighted the bubble so that it 

 did not escape from the liquid, and the crystals were likewise 

 buoyed up by the bubble. The minimum of motion resulting 

 has left the crystals smaller than their neighbors and compar- 

 able with those near the bottom. 



Fig. 2 shows, in a single section from the lower part of the 

 crucible, the more rapid passage from the portion without 

 crystals to that rich in crystals when the time allowed was 

 45 minutes. 



These trials were repeated a number of times and always 

 with the same result. Other mixtures from which forsterite 

 is the first mineral to crystallize gave similar results. 



The Sinking of Pyroxene. 



In order to observe the effects of increasing viscosity, mix- 

 tures which contain an excess of silica over pyroxene composi- 

 tion (metasilicate ratio) were also used. In the case of the 

 mixtures of pyroxene composition, from which olivine first 

 crystallizes, it was usually possible to cool the crucible quickly 



