February 12, 1897.] 



SCIENCE. 



245 



purposes, but using much more time and 

 pains. 



Turning now to the determination of 

 combined carbon and graphite, we do not 

 find the state of affairs so satisfactory. As 

 is well known, these two constituents are 

 usually found by first determining total 

 carbon, then dissolving another portion of 

 the sample in hydrochloric acid, filtering 

 and washing with caustic potash, alcohol 

 and ether, and then burning the residue, 

 collecting and weighing the carbon dioxide 

 formed, as in an ordinary combustion. The 

 result is called graphite, and the combined 

 carbon is the difference between the total 

 cai'bon and the graphite. But as Shimer* 

 has so well shown, what we actually get by 

 this procedure is not necessarily the graph- 

 ite and the total combined carbon in the 

 sample, but only the combined carbon 

 which exists in the metal as a carbide 

 soluble in hydrochloric acid. If the sample 

 contains carbides not soluble in that acid, 

 nor in the materials used in washing, the 

 <jarbon of these carbides appears with and 

 is counted as graphite. Shimer shows that 

 titanium, and possibly vanadium carbide, 

 are apparently not infrequently thus 

 counted. The use of sulphuric instead of 

 hydrochloric acid leads to the same error, 

 while the employment of nitric acid as sol- 

 vent apparently gives the graphite much 

 more definitely, but leaves us in doubt as 

 to whether the combined carbon is really 

 the combined carbon which we want, in 

 ■order to have light on the quality of the 

 metal we are dealing with. It is obvious 

 that the difBculty here is in our lack of 

 knowledge as to what carbides actually ex- 

 ist in pig and cast iron, and, if there are 

 several of them, which one or ones do we 

 actually want to know the carbon content 

 of. If we knew positively that the com- 

 bined carbon wanted was that which exists 

 in the metal as carbides of iron and manga- 



* Trans. Am. Inst. Min. Eng., 25, 395. 



nese, and that these carbides were soluble 

 in hydrochloric or sulphuric acid, while all 

 other carbides present were not soluble in 

 these acids, obviously we should use these ' 

 acids when determining combined carbon. 

 On the other hand, if we want to know 

 only graphite, and care little about the 

 combined carbon, apparently nitric acid is 

 the solvent to use. It is clear that much 

 more work is needed on this subject, a 

 state of affairs which, as we progress, we 

 shall find is characteristic of other constit- 

 uents of the metals we are considering. 



Much might be said in regard to the 

 color test for determining carbon in steel. 

 It is diificult to over-estimate the value and 

 importance of this method, especially in 

 the daily operation of steel works, and 

 there seems little doubt but that, if proper 

 precautions are employed, the method, in 

 skillful hands, will give results that are 

 fairly reliable to within three or four hun- 

 dreths of a per cent. It would hardly be 

 possible in this paper to discuss all the pre- 

 cautions which are deemed essential by 

 those best informed. A chemist of wide 

 experience with the method has enumer- 

 ated twenty-four points that must be ob- 

 served if reliable results are to be ex- 

 pected. Let it sufiice for us to say that 

 even approximate accuracy cannot be ex- 

 pected : 



1. If the steel whose carbon is to be deter- 

 mined and the standard steel do not have 

 their carbon in the same condition. For 

 example, if the standard steel has been 

 annealed, and the sample to be tested has 

 been tempered, the results will be worth- 

 less. 



2. If the attempt is made to determine 

 the carbon in any steel by using a standard 

 widely different from it in carbon content. 

 Using a 0.20 per cent, carbon standard, 

 with a steel containing 0.50 or 0.60 per 

 cent., is apt to lead to very fallacious re- 

 sults. 



