Differentiation in Silicate Liquids. 189 



crystals in a layer near the bottom.* This does not seem to be 

 an unreasonable expectation as to the rate of cooling of such a 

 body of magma in its interior portions. Towards the border 

 it would no doubt cool much faster, and this fact is clearly 

 expressed in the contact phase which bears olivine. Near the 

 contact the cooling was too fast to allow time for the sinking- 

 out of the olivine crystals, which are therefore retained. 

 There is no reason for appealing to the Soret action, an 

 accumulation of certain constituents towards the cold border 

 while the magma is still entirely liquid. f Neither is there any 

 reason to accept an explanation involving a diffusion of 

 material towards the cold border during crystallization, Har- 

 ker's preferred explanation of such phenomena.;): It is a plain 

 case of restriction of differentiation at the contact on account 

 of quick cooling and continuance of gravitative differentiation 

 by sinking of crystals in the more slowly cooled part removed 

 from the contact. This is substantially the process advocated 

 by Daly to explain ' basic contacts '§ in general, though he con- 

 siders immiscible liquid ^portions as the probable units of 

 gravitative differentiation. To this phase of Daly's explana- 

 tion many objections can be raised, but of the truth of the 

 general idea the Palisade sill furnishes an excellent example. 



In stating that the accumulation of olivine crystals might 

 have been accomplished in a period of 200-300 hours the influ- 

 ence of convection currents has not been mentioned. The 

 deduction was, however, made on the basis of experimental 

 work in which the opportunity for convection was favorable 

 and possibly comparable with that in the main interior portion 

 of the sill, though this is a question difficult to decide. That 

 convection has played no important part seems, however, to be 

 shown by the sill itself. It was cooled both at the top and at 

 the bottom surfaces, the temperature therefore rising inward 

 from both to a layer near the center. In the lower half the 

 temperature gradient was in the proper direction to eliminate 

 convection currents, whereas in the upper half they were 

 favored. It might be considered therefore that sinking of 

 crystals was free to take place in the lower half, but seriously 

 interfered with or prevented in the upper half. Thus would 



*In assigning a viscosity to the diabase magma comparable with that of 

 the artificial melt from which olivine settled (4, or about 400 times that of 

 water) Becker's estimate of the viscosity of Hawaiian basalt has been ex- 

 ceeded 8 times. If the Palisade diabase magma had a viscosity as low as 50 

 times that of water, and this is quite possible, the collection of olivine 

 crystals might have taken place in from 25-40 hours. 



f J. V. Lewis, Petrography of the Newark Igneous Rocks of New Jersey, 

 Ann. Rep't State Geologist of N. J., 1907, p. 131. 



tThe Natural History of Igneous Rocks, pp. 317-320. 



§ Igneous Rocks and their Origin, pp. 243-242. 



