Plate 64 



1. Canyon Diablo, Arizona; coarsest octahedrite. Portions of four areas of cohenite 

 (white) surrounded by kamacite, with an irregular taenite lamella (V-shaped, lower right). 

 The dark area is plessite. The cohenite is indistinguishable from schreibersite. Picral 

 80 seconds; X 30. 



2. Canyon Diablo. The area shown in figure 1 with additional 10 seconds electro- 

 lytic etching with alkaline sodium picrate, darkening the cohenite. Two small schreibersite 

 bodies (uncolored) are revealed at the left edge of the larger cohenite area. The V-shaped 

 lamella is shown to be taenite. 



Owing to diversities of structure in different masses, and even in parts of the same mass, 

 analyses of Canyon Diablo vary widely. Thus Moissan (1893) found in different parts 

 of the same sample 3.07 percent and 5.07 percent of nickel, while Booth, Garrett, and Blair 

 (in Barringer, 1905) found 7.94 percent. The latter analysts found the large proportion 

 of 0.417 percent of carbon, though Moissan reported none. Moissan found 0.20 percent 

 phosphorus in one sample, in another none. Carbon actually occurs abundantly, both 

 in occasional nodules of graphite and in combined form as cohenite, the latter profuse 

 in some samples and absent in others. 



3. Cosby Creek, Tennessee; coarse octahedrite; Ni-Co 7.42 percent. A large area of 

 cohenite enclosing several small inclusions the nature of which is not apparent, surrounded 

 by coarsely granular kamacite. Picral 60 seconds; X 30. 



4. Cosby Creek. The same area as figure 3 with additional 10 seconds electrolytic 

 etching with alkaline sodium picrate. The small inclusions, along the left side of the 

 cohenite area, are shown to be schreibersite. The rounded islands are kamacite, identical 

 with the groundmass. 



Several analyses of Cosby Creek have been made, of which the latest by Fahrenhorst 

 (in Cohen, 1900) gives Ni 6.91 percent, Co 0.51 percent, P 0.37 percent. Cohenite appar- 

 ently is not mentioned in the literature of this iron, nor does carbon appear in any of the 

 analyses except the first made by Troost in 1840, who reported 0.50 percent. 



184 



