78 ANNALS NEW YORK ACADEMY OF SCIENCES 



principles of Vogt and Lagorio, he rejected the French synthetic methods 

 as unnatural and studied the deposition of corundum in the melts of a 

 large glass factory, where he could control his temperatures up to 

 2100° C. His melts varied in size up to one hundred pounds or more. 

 They were made up, sometimes of the pure salts and sometimes of pow- 

 dered minerals, to approximate the composition of certain natural erup- 

 tive rocks. They were contained in clay crucibles, whose position in the 

 furnace might be varied according to the degree of heat desired. A point 

 often overlooked in connection with this investigation, and one which he 

 states to be its most serious defect, 65 is that 30 per cent, of his melts 

 were spoiled because they attacked the clay crucibles. Melts high in 

 magnesia with low alumina and alkalis, were especially prone to .this, 

 deriving alumina from the crucible to form spinel, while those rich in 

 lime, alumina and alkalis had no effect at the highest heat. The results 

 of this well-known study prove that the role of alumina in a magma is 

 entirely analogous to that of silica; if it be saturated with respect to 

 alumina, corundum will separate out just as quartz does in a granite. 

 An alumino-silicate magma is saturated when it has the general compo- 

 sition MeO, m A1 2 3 , n Si0 2 (Me = K 2 , Na 2 , Ca, and n = 2 — 13) 

 and m is more than one. If silica is also in excess (over 13), sillimanite 

 is formed. If magnesia and iron are present in excess, they will form 

 spinel; or if silica is also in excess, cordierite will separate. The sepa- 

 ration of corundum from a magma, therefore, is governed by definite 

 laws; and minerals, moreover, which have previously been thought of as 

 due to contact action are thus shown to be pyrogenic under certain 

 conditions. 



Morozewicz in a previous paper 66 had shown that alumina will readily 

 dissolve in a molten magma whose composition approximates that of 

 basic magnesian rocks, and that on cooling, the excess alumina separates 

 out as corundum, and spinel. 



Corundum may, therefore, be formed by the action of "agents mine- 

 ralisateurs" on alumina, or by the fusion of certain aluminous minerals, 

 or by the solution of alumina to supersaturation in molten magmas. 



Origin of Corundum in Other Localities 



Since the emery of the Cortlandt Series is always in igneous rock, 

 only a few similar occurrences of corundum in other localities will be 

 noticed. 67 



86 Op. cit., p. 18. 



68 Zeitsch. fur Krystall., XXIV, 281. 1895. See also Lagorio, idem. p. 285. 



67 For a full discussion, see J. H. Pratt, Bull. 269, U. S. G. S., pp. 71-96. 1906. 



