102 MEMOIRS NATIONAL ACADEMY OF SCIENCES, VOL. XIII. 



the solvent, by which more of the original content is recovered, but it is also evident that this mineral, particularly 

 in the acicular form of rhabdite, is more abundant than in the generality of the meteoric mass. 

 An analysis of tsenite which was dissolved in copper-ammonium chloride gave: 



Fe Ni Co Cu P 



66.46 30.28 0.68 0.32 0.30 =99.69 



In composition as well as in physical aspect (thin, tin-white, flexible lamellae) this agrees very closely with the 

 group rich in nickel and free from carbon of Toluca, Wichita, etc., as given by Cohen. With Toluca also it agrees in 

 the presence of a determinable amount of copper. The phosphorus of the above analysis indicates, as Professor Cohen 

 has remarked, that schreibersite is not wholly insoluble in copper-ammonium chloride. 



Two analyses of cohenite were made, No. 1 being the free grains from the general solution of the mass and No. 2 

 the vein matter. Both were dissolved in copper-ammonium chloride and the percentages calculated for the difference 

 in weight after deducting the considerable residue of undissolved schreibersite, separated from the carbon, which was 

 determined by burning in a stream of hydrogen and weighing as carbonic acid : 



Fe Ni+Co P C 



1. 92.88 1.33 0.48 5.33 =100.02 



2. 91.67 2.43 0.09 6.07 =100.26 



The phosphorus is undoubtedly due to a slight action of the solvent on the schreibersite, which is not wholly 

 insoluble in the copper-ammonium chloride. This result agrees very well with the analyses of cohenite given by 

 Weinschenk and Cohen, and with an unpublished analysis of that of Bendego by Dafert. In appearance the cohenite 

 grains agree with those of Bendego although they are richer in inclusions of tabular schreibersite. Owing to the general 

 distortion of the crystals and the rounded character of the faces no measurements could be made, but the forms are 

 undoubtedly identical with those of Bendego, on which Hussak succeeded in demonstrating that they belong to the 

 cubic system. 



Three distinct forms of iron and nickel phosphide occur, which, although differing greatly in appearance and some- 

 what also in chemical composition, are probably different phases of a single mineral species. The most abundant 

 individually are the acicular forms known as rhabdite, though, owing to their minute size, they do not equal in weight 

 the granular and tubular forms known as schreibersite. Both are generally distributed throughout the mass, the 

 echreibersite form being particularly abundant, included in, or adherent to, the surface of the cohenite grains, while 

 the rhabdite needles are especially concentrated in the less soluble metallic portions and in the spongy, coal-like 

 particles. Both are distinctly crystalline and may occur in the same individual. As Cohen has already shown the 

 chemical identity of the two types, no further proof seems necessary that schreibersite and rhabdite belong to the same 

 mineral species, for which the former name, being the older, should be retained. Diligent search was made without 

 success for crystals that would admit of measurement, the rhabdite individuals being too minute and those of the 

 schreibersite type too much distorted and with strongly rounded faces. The general appearance of the latter type is 

 strongly suggestive of distorted crystals of the cubic system. On crystals which will be described later, separated from 

 the Sao Francisco do Sul mass, Dr. Hussak succeeded in proving that the crystalline form of schreibersite is really 

 tetragonal. 



The third form of phosphide occupies the center of the vein mass, being inclosed between walls of cohenite. This 

 is massive and extremely brittle, breaking with a conchoidal fracture and in color and general appearance strongly 

 resembling arsenopyrite. The cohenite of the walls of the vein also forms a massive crust covered, however, with 

 crystalline faces on its outer surface. As shown by the analysis below, No. 3, the composition differs from that of the 

 typical schreibersite of the same meteorite in the relative proportions of the iron and nickel. The phosphorus is also 

 higher in the complete analysis and approaches more nearly to what Professor Cohen considers as the normal proportion, 

 but in a separate determination (No. 4) the proportion is nearly the same as in the normal granular schreibersite with 

 a slight admixture of rhabdite needles, Nos. 3 and 4 of the vein matter. In all the material was freed from tsonite and 

 cohenite by treatment with copper-ammonium chloride, and in No. 2 special care was taken in the washing to make 

 sure that the copper found in 1 and 3 previously executed, really belonged to the substance and did not come from 

 this solvent: 



P Fe Ni Co Cu Sn 



1. 12.82 54.34 31.48 0. G7 0.20 0.00 =99.45 



2. 13.17 51.25 33.68 0.17 1.18 =99.45 



3. 14.58 66.72 17.54 .... 0.13 trace =98.97 



4. 12. 98 



The most reliable published analyses of meteoric phosphide, or phosphides, show very variable relative propor- 

 tions of iron and nickel and cobalt even in the same meteoric mass and as regards phosphorus, a larger group with about 

 15 to 16 per cent and a small group with about 12 to 13 per cent. The above analyses place Canon Diablo in the latter 

 group. Copper is only reported in two, Schwetz and Seelasgen, both of which have been reanalyzed by Cohen with very 

 different results and without copper, which possibly, however, was not looked for. Tin has not been reported, possibly 

 because the solution has usually been made in aqua regia in which it would only appear through a special research. In 

 the present case the solution was made in plain nitric acid and the tin appeared as oxide and was verified by blowpipe 

 tests. The proportion in No. 3 was certainly as great as thaj in No. 2, but was not determined for fear of losing the 



