Page:The Journal of geology (IA journalofgeology11893univers).pdf/355

compounds forming the original rock magma are completely soluble in one another? I think not.

We are told by Vogt^[1] that silicates can be melted together in all proportions. This may be true, but it does not prove that this mixture would not separate into layers of different composition, or at least become heterogeneous, if it were kept molten for a sufficient time. The viscosity of molten glasses is very great and consequently the separation must take time. Still evidences of such separation—or liquation as we may call it, following Durocher—in the manufacture of glass are not wanting. It is well known to be very difficult to produce large pieces of homogeneous glass, for example for optical purposes. According to Wagner's Handbuch der chemischen Technologie^[2] this comes from the fact, "either that the individual compounds formed during the melting process have not dissolved one another or that they have separated from the mixture by a lowering of the temperature"; and further, "One will seldom find large pieces of glass, which are completely free from this fault."^[3] But it is not necessary to leave the field of geology in order to decide the question whether magmatic differentiation is a diffusion, or a liquation, process. Let us select some examples of differentiation, and examine them in the light of both theories. I have chosen two, one on a small scale, the basic inclusions, and one on a large scale, the great petrographical province of Iceland.

By diffusion, according to "Soret's principle," the basic inclusions could never be thought to have been formed in situ or approximately so—for, between them and the surrounding magma there would be no difference in temperature, or at least no difference sufficient to alter the osmotic pressure, which is proportional to the absolute temperature, or enough to produce