Chemistry and Physics. 407 



according to the preceding formulas, we fail to find it, and often are 

 surprised to see that the specific weight of a mineral presents but slight 

 oscillations which scarcely accord with the widely varying composition 

 attributed to its different varieties. For example, it is not easy to un- 

 derstand why titaniferous iron has always a specific weight, varying 

 only from 4*745 to 4*78, while it is composed sometimes of equal equiv- 

 alents of ferric oxyd and ferrous titanate, and sometimes of one equiva- 

 lent of ferric oxyd and six equivalents of ferrous titanate. 



These anomalies disappear entirely if we write all these oxyds after 

 one formula, OM 2 similar to that of water 0H 2 , and in which M can 

 be replaced by different metals in indefinite proportions, provided that 

 the sum of the equivalents of these metals are equal to 2.* 



See the oxyds thus represented with their atomic volume, which 

 equals the product of the atomic weight divided by the density. 



Type oxyd, 0M 2 



Oligist iron, OFe^ 



Mean. 



no 



11-4 

 112 



Braunite,t OMn£„ 



Magnetic iron, 0{Fe$Fe|), 11-4 



Gahnite, O(Al^ZnJ), 10-9 



Spinelle, OfAMJMni)*", 10-6 



Ceylanite, 0(AI^F ft yMo^_ 10-6 



Chlorospinelle, 0(AI^FejfrMg z ) 2 10-6 



Chromic iron, 0(Al^Fe^Cr^MgyFe z ) 2 11-2 



Titaniferous iron, 0(Ti«*Fef?yFe z )„ 10-9 



Franklinite, 0(Fe^Mn^Zn>Fe z ) 2 11-1 



Perovvskite, 0(Ti«£Ca§) 2 112 



Periclase, 0(Fef?>Mg*) 2 10-9 



I am obliged to indicate by letters the most of those fractional num- 

 bers, of which the sum is equal to two equivalents. It is easy to re- 

 place these letters by their numerical values, which are derived from the 

 formulas we have previously given. In the same manner I suppress 

 the details of the calculations relative to the atomic volume. 



This volume, it is to be remarked, is essentially the same for all the 

 °xyds mentioned ; there are but slight differences apparent, which are 

 due to the fact that in calculating these formulas, Ave have not always 

 taken into account those oxyds which are contained in small quantities 

 in the minerals. It is very rare that a mineral is chemically pure, and 

 a s the presence of a very small quantity of any foreign substance al- 

 ways modifies the specific weight, it is evident that we cannot obtain a 

 number rigorously exact in dividing the specific weight by the atomic 

 height of the substance supposed to be chemically pure. 



* It is necessary to recollect that Fe§, AI|,Cr|, Mn§=Fep(ferricum), AIB(alu- 

 D 'inicum), Crj3 (chromicum), Mnp (manganicum), are the equivalent ol 11, re 

 (terrosum), Mn (Mangaoosum), Mg (magnesium), etc., because an oxyd M 4 <J 3 re- 

 acts with six equivalents of hydrogen toform3H 2 0, M 4 resting in the place ot H 6 . 

 If we represent the oxyd of titanium by Ti 2 0,(Ti0 2 , Berzelius), Ti 2 becomes 

 e q»al to H 4 : then Tii=Tia equals H. , 



[It will be remembered that M. Gerhardt divides the ordinarily received equiva- 

 lents of hydrogen and the metals. The protoxyds MO then become M 2 0, and the 

 sesquioxyds M 2 3 , M 4 3 . See this Journal, p. 171, this volume.] 



* The'Hausmannite D=422 gives an atomic volume of 12 2. 



