REVIEWS 3°5 



ture of idiomorphic corundum of pyramidal habit, with anorthite and 

 biotite, and accessory dark green spinel of earlier generation than the 

 corundum, with also apatite and zircon. 



A number of remarkable experiments were made with acid magmas 

 of the general composition of rhyolite or granite. By dry fusion at 

 high temperatures it has frequently been demonstrated that tridymite 

 is a more stable form of crystalline silica than quartz. In the case of 

 the partial fusion of a quartzose block of granite, the quartz became 

 transmuted into an aggregate of shingly tridymite flakes, and the same 

 has been noted in nature in inclusions of granite in a porphyry. The 

 presence of alumina in an acid magma was found to prevent crystalli- 

 zation, where a non-aluminous silicate mixture partially crystallized in 

 the form of tridymite and prismatic silica (the latter of the unusual 

 type described by Fouque and M. Levy). Vogt, in his exhaustive 

 studies of furnace slags, 1 has called attention to the influence of alu- 

 mina in "retarding" the crystallization of a glass or a slag, and this 

 fact is well known to glass workers who add alumina to prevent the 

 development of silicate crystals. With the aid of the theory of solu- 

 tions, this influence is easily explained ; in general, supersaturated 

 solutions give large crystals, a lower degree of saturation gives small 

 crystals, and unsaturated solutions under the same conditions develop 

 no crystals at all. Alkaline silicate magmas are capable of dissolving 

 alumina in large quantities ; alumina possesses for the alkalies and 

 more especially the alkaline earths a very strong chemical affinity, 

 forming with them very stable and widespread natural compounds. 

 Accordingly alumina in small amount dissolved in such a magma has 

 only the effect of uniting with a portion of the bases in potential alu- 

 minosilicate form, and preventing them from crystallizing out as simple 

 silicates which in the absence of alumina would easily saturate the solu- 

 tion. Morozewicz has demonstrated that a very large amount of 

 alumina is required to saturate a solution to the effect of permitting 

 crystallization of the aluminosilicates, as outlined above. When great 

 excess of alumina is present, however, crystallization may be readily 

 induced. Thus the expression, "retarding crystallization," is applica- 

 ble only to access of alumina up to the critical point of saturation, 

 beyond this its effect is that of an accelerator. The effect, in fine, of 

 a small amount of alumina in a glass, is to produce aluminosilicate 



'Vogt, J. H. L.: Beitrage zur Kenntnis der Gesetze der Mineralbildung in 

 Schmelzmassen und in den neovulcanischen Ergussgesteinen, Christiania, 1892. 



