10 BULLETIN 184, UNITED STATES NATIONAL MUSEUM 



The older analyses (many 75 to 100 years old) should be taken 

 with reserve. Likewise, when an analysis, even though of recent 

 and reputable origin, appears to be wholly inconsistent with the 

 structure of the iron, it is justifiable to suspect some error or acci- 

 dental factor in the determination of the composition. For example, 

 if an analysis of a fine octahedrite showed only 5 percent of nickel, 

 it obviously could not be accepted. 



Errors in nickel content. — In some of the old analyses errors are 

 conspicuous. For example, analyses made in 1866 and 1876 of Pram- 

 banan and Yanhuitlan, fine octahedrites, gave the impossibly low 

 percentages of 2.8G and 1.85 nickel. Later analyses showed, respec- 

 tively, 9.30 and 7.36. Different analyses of the same iron also show 

 great variations ; for example, analyses of three Coahuila irons regarded 

 as identical gave nickel percentages of 2.10 to 7.42. 



In many old analyses the nickel content was greatly understated. 

 Thus, analyses of 25 irons (including many medium and fine octa- 

 hedrites), of Avhich only three were later than 1880 and most of them 

 between 1840 and 1870, show very low percentages, mostly from 

 about 2.5 to 3.5; but 15 of these irons analyzed later showed in 

 only one case less than 6 percent and mostly 7 to 10 percent. Of 30 

 analyses of low-nickel irons (hexahedrites, coarse and coarsest octa- 

 hedrites, and nickel-poor ataxites) made since Farrington's compila- 

 tion (1907), not one shows less than 5.25 percent of nickel; and of 

 58 of the more recent analyses of all types of irons, only two show 

 slightly less than 5 percent, of which one is clearly erroneous. We 

 may therefore fairly conclude that there are very few irons, if any, 

 witli less than 5 percent of nickel. 



At the other end of the series the nickel content in some of the 

 finest octahedrites may have been overstated. Thus 14.82 percent 

 of nickel (15.28 percent of nickel and cobalt) in Tazewell, as deter- 

 mined in 1855, seems clearly to be excessive, as the author knows of 

 no other iron with so high a percentage that does not have a typical 

 ataxite structure. 



In this work analyses are contmually referred to, and it is imprac- 

 tical to give all the dates and references. Where there are two or 

 more analyses the one quoted is either the latest or the one deemed 

 most probably reliable. The figures can never be accepted unre- 

 servedly unless the analysis is recent and was made by an analyst of 

 known competency. 



III. PRIMARY CONSTITUENTS OF METEORIC IRON 



Kamacite. — The principal component of meteoric irons is kamacite 

 (approximately iron 94 percent, nickel 6 percent). Metallographically 

 it is alpha nickel-iron. It is iron-gray, magnetic, with hardness be- 

 tween 3 and 4 on the ordinary Mohrs scale, being easily scratched by 



