THE METALLOGRAPHY OF METEORIC IRON 23 



These determinations would seem to establish the existence of an 

 iron-nickel carbide. It is still possible, however, that the nickel 

 thus found was not in chemical combination with carbon. 



The rarity of cohenite may perhaps be ascribed to the graphitiz- 

 ing influence of nickel, which in nickel steels makes it easy to change 

 the carbide (cementite) to graphite by a relatively short heating. 

 Thus in meteoric irons the presence of nickel might tend to prevent 

 the preservation of carbide by causing it to break down into graphite 

 and iron. 



It is noteworthy that cohenite apparently has not been found 

 in the higher nickel (medium and fine) octahedrites or in nickel- 

 rich ataxites, but only in the coarse and coarsest octahedrites in 

 which the nickel content is not more than about 7 percent. (The 

 percentage of 10.01 reported in 1860 in Wichita is probably incorrect.) 



It has already been mentioned that in most irons, notably in 

 Canyon Diablo, the occurrence of cohenite is apt to be localized, it 

 being abundant in some areas and absent in others. It may be sig- 

 nificant that in Canyon Diablo there is also a surprising variation 

 in nickel content. Moissan reported 3.07 and 5.07 percent from 

 different parts of the same specimen, and in another specimen 1.08 

 and 7.05 percent. Other analyses have sho^vn 3.94, 7.94, 5.78, and 

 9.25 percent. It might be instructive if we could know whether 

 the very low percentages were found in areas where cohenite was 

 abundant. 



Identification of cohenite. — It is quite possible that cohenite is 

 commoner than it has been supposed to be, because it resembles 

 schreibersite so strongly that the difference is not easily recognized. 

 Even so great an authority as Vogel (1927) illustrates a sample of 

 Canyon Diablo in which inclusions of cohenite are referred to as 

 schreibersite. 



In hardness and in appearance on a polished surface they are 

 practically identical, and their behavior is the same with any ordinary 

 etching. Their very close resemblance is shown in plate 64, where 

 the two happen to be in juxtaposition. Blowpipe tests, and those 

 of ammonium molybdate and copper ammonium chloride, are not 

 easily applied. 



One indication that inclusions may be cohenite is their arrange- 

 ment parallel with the octahedral structure. In Canyon Diablo 

 and Wichita County, for example, there is a profusion of elongated 

 cohenite bodies in the kamacite bands. Schreibersite inclusions do 

 not show such an arrangement. 



A test easily applied to the macrostructure is the reduction of 

 metallic copper from a salt of that metal. This is most readily done 

 before micropolishing, with a fairly weak solution of copper sul- 



