408 K T. Allen, F. E. Wright and J. K. Clement— 



raised considerably above the melting point, and Lehmann* 

 states that benzyl-phenyl-nitrosamine also yields an unstable 

 form when strongly heated and very rapidly cooled. 



"We have proved that (pp. 402 and 411), all other forms of 

 magnesium silicate change into the monoclinic pyroxene before 

 the melting point is reached. If the crystal structure depends 

 on crystal particles or units which consist of regularly arranged 

 chemical molecules, it is not impossible that some of these 

 groups might preserve their identity some time after fusion of 

 the substance, and become completely disintegrated only after 

 some lapse of time or at a higher temperature. The favorable 

 influence of a high initial temperature on the formation of 

 amphibole would then be due to the complete breaking down 

 of nuclei which condition the crystallization of the monoclinic 

 form.f 



We tried the effect of dropping small portions of the solid 

 amphibole into the undercooled liquid in order that they 

 might serve as nuclei for crystallization, but the experiment 

 did not succeed. The silicate liquids are so viscous that the 

 nuclei do not seem to exert any influence on the crystalliza- 

 tion. The negative effect of nuclei is very strikingly brought 

 out in the following way : Magnesium silicate can be chilled 

 in such a manner as to yield radial spherulites of monoclinic 

 pyroxene embedded in glass. Now when this cake is heated 

 again at 900°, not only do the existing nuclei fail to grow, but 

 they exercise no orienting influence on the new crystals which 

 form. 



As this substance has not been obtained before, some further 

 details regarding its crystallization may be of interest. It 

 usually begins on the upper surface of the melt next to the 

 walls of the crucible, forming a fringe of converging fibers 

 which grow toward the center, though sometimes a number of 

 rather widely separated nuclei start at points on the upper or 

 lower surface. This crystal form is characterized by a con- 

 siderably slower growth than the monoclinic. If, in a crucible 

 in which the amphibole has already begun to crystallize, the 

 still liquid portion be touched by a wire, the disturbance is 

 instantly followed by the appearance of a monoclinic nucleus. 

 The difference in the rate of growth of the two crystal forms 

 is then very striking. We early noticed a very marked differ- 

 ence in the rise of temperature during the crystallization of 

 these two forms, the monoclinic raising the temperature to a 

 dazzling white; and though the difference must be partly due 



*0. Lehmann, Molekularphysik, vol. i, p. 211. 



f A similar conception has been advanced in a paper by Day and Allen to 

 explain the superheating of the alkaline feldspars. This Journal, series (4), 

 xix, p. 124, 1905. 



