Differentiation in Silicate Liquids. 187 



above the base of the Palisade sill, and is remarkably rich in 

 olivine. He interprets it as due to an accumulation in this 

 layer of sunken olivine crystals. It is to be noted, moreover, 

 that the main body of the diabase, from which the olivine 

 crystals have sunk, contains free silica (quartz), therein pre- 

 senting an analogy to the upper portion of the artificial mix- 

 tures in the experimental work. The observed result appears 

 to be exactly that to be expected if the process pictured by 

 Lewis was operative. It may seem that this occurrence of free 

 silica in the main portion of the sill is merely a superficial 

 analogy with the case of the artificial mixtures, but a little con- 

 sideration should dispel this idea. In the artificial mixtures 

 the olivine would have been redissolved had it not settled to 

 the bottom and become very concentrated there. In the upper 

 layers, as in all the layers, its resolution would have resulted 

 in the formation of the more siliceous pyroxene by reaction 

 with the liquid. The failure of this reaction in the upper 

 layers on account of the removal of the olivine leaves the liquid 

 with an excess of silica. Now in the Palisade diabase there 

 are quickly chilled marginal phases from which the olivine 

 did not settle out. The olivine crystals are partially redis- 

 solved, having reacted with the liquid to form pyroxene. The 

 failure of this reaction in the main mass of the diabase on 

 account of the removal of olivine must, as in the artificial mix- 

 ture, leave the liquid with an excess of silica. The similarity 

 of the two cases is of a fundamental nature and lies in the fact 

 that the olivine would have been partly redissolved. It is clear 

 that the formation of at least some of the quartz of the main 

 mass of the diabase is to be attributed to the process of differ- 

 entiation by the movement of crystals under the influence of 

 gravity. It is not necessary to assume solution of the sur- 

 rounding quartzites or arkoses, a process of which Lewis finds 

 no evidence.;); 



It is likely that the rate of sinking of olivine crystals was 

 many times greater in the diabase liquid than in the artificial 

 melts here treated. The rate of sinking of a crystal increases 

 with its size, with the fluidity of the liquid and with the differ- 

 ence in density between crystal and liquid. Palisade diabase 

 glass has a density at room temperature of 2*763. The liquid 

 inwhich forsterite crystals sank in the experimental work has a 

 density as glass at room temperature of 2*82, considerably greater 

 than that of diabase glass. The artificial liquid is very closely 

 related to diopside in composition and therefore may be assumed 

 to expand at nearly the same rate as diopside liquid. Now 



\ J. V. Lewis : Petrography of the Newark Igneous Bocks of New Jersey, 

 Ann. Eept. State Geologist N. J., 1907, p. 127. 

 JLoc. cit., p. 132. 



