Pebrhaey 11, 1898.] 



SCiEi^GR 



189 



years in seeking out the causes of things, 

 as a necessary preliminary to the intelligent 

 modification of practices and methods in 

 connection with a great corporation, has 

 continually impressed me more and more 

 with the very great help which a properly 

 conceived and executed analysis can give 

 in cases of difficulty. 



But, again, I take pride in the field of 

 analytical work, because of the opportunity 

 which thoughtful analytical work affords 

 for finding new things. The careful, 

 thoughtful, observant analyst is constantly 

 on the verge of either being able to add to 

 his own knowledge or of being able to con- 

 ti'ibute something to the general progress 

 of our science. And here, again, I must be 

 pardoned for using as illustrations cases 

 which have arisen in the laboratory of the 

 Pennsylvania Eailroad Company. 



A few years ago, in our laboratory, we 

 began to get ready to make our analyses of 

 the samples of steel which were designed 

 ultimately to be the international standards 

 for the analysis of iron and steel. Before 

 starting in on these samples, however, it 

 was deemed prudent to do a little prelimi- 

 nary work on some other samples, with the 

 idea in mind of seeing whether apparatus 

 and method were satisfactory. Accordingly, 

 four separate and distinct determinations 

 on the same sample were made for carbon, 

 using the double chloride of copper and 

 ammonium to release the carbon, and burn- 

 ing in oxygen gas. The four determina- 

 tions agreed with each other within 0. 01 or 

 0.02 of a per cent., and were regarded as 

 fairly satisfactory. But as the work was 

 important, and as some parts of the appa- 

 ratus had not worked quite satisfactorily, 

 it was decided to repeat the four determina- 

 tions. Meanwhile a new stock bottle of 

 solution of the double chloride had been 

 made exactly in the manner that had been 

 our custom for some time previous. When 

 the second four determinations were ob- 



tained they differed from the first by more 

 than a tenth of a per cent. I need not 

 weary you with the details of our hunt for 

 the cause of this discrepancy, how every 

 point in the apparatus was tested one after 

 another, how various modifications were 

 tried, how combustions were made on crys- 

 tallized sugar to check ourselves, and how 

 finally we located the difficulty in the double 

 chloride of copper and ammonium solution. 

 These details have all been published.* 

 Suf&ce it to say that, as the result of this 

 work, together with subsequent work by 

 other chemists, it is, we believe, now gener- 

 ally accepted that the commercial am- 

 monium double salt contains carbon in 

 some form, probably pyridine, that its use 

 as a solvent to release the carbon from iron 

 and steel is unreliable, and that the substi- 

 tution of the potassium for the ammonium 

 double salt overcomes these difBculties. 

 The point wliich I especially want to em- 

 phasize is that, in trying to do a little care- 

 ful analytical work, we struck a new and 

 apparently hitherto unsuspected source of 

 error in one of the oldest and best estab- 

 lished methods of iron and steel analysis. 



Another illustration will, perhaps, make 

 this point still more clear. In the regular 

 course of work, at one time a silicon deter- 

 mination was made in a piece of tire steel 

 which had been sent by an ofiicer of another 

 railroad for information. The figures 

 found by our analysis were 0.14 per cent., 

 these figures being sent to the ofBcer above 

 referred to. A little later word was re- 

 ceived that an analysis of a sample from 

 the same tire by another chemist gave 0.28 

 per cent, as the content of silicon. This, of 

 course, led us to look over our work, with 

 the idea of finding where the cause of the 

 discrepancy lay. A careful examination of 

 ' our weights and figures showed that it was 

 not an error of calculation. Accordingly, 

 we decided to duplicate our work, need I 



* Trans. A. I. M. E., 19, 614. 



