V O L U M E 2 2 , NO. 12, D E C E M B E R 1 9 5 0 used by Hoar ( 3 ) to stabilize the copper rubeanate color. I n the present work it was noted that the intensity of copper color varied with the condition of the gum arabic solution. Freshly prepared solution without heat treatment gave somewhat high color intensities, which became progressively lower with the age of the gum arabic solution until a stabilization point was reached. When the gum arabic solution was stabilized by heating near boiling for 1 hour, the results remained constant and permitted the use of a permanent calibration chart. No investigation was made of this characteristic in relation to different grades of gum arabic. Although iron does not react with rubeanic acid in the above procedure, it does produce some color with ammonium acetate, with an interference equivalent to 0.004% copper for each 10% of iron in the sample. This is rompensated by use of the iron reference sample, which also includes the reagent blank. Xirkel gives an intense color with ruheanic acid (4);but only in the presence of a relatively high concentration of rubeanic acid. When the reagent n as added sIow1~-with stirring, to prevent local contentration, no reaction occurred with nickel. Honever, the greenish color of nickel salts causes an interference equivalent to 0.006% copper for each 10% nickel in the sample. Because this is 0.002% more than the iron interference, the net correction is 0.002% for each 10% nickel. Manganese, molybdenum, titanium, aluminum, vanadium, cobalt, and hexavalent chromium do not interfere in the quantities usually present in steels. Certain other elements such as tungsten and columbium give insoluble residues, which are filtered off. However, the presence of these alloying elements reduces the amount of iron present as rompared nith the reference solution, and a positive correction is required for the iron replaced. Representative results tor copper are shown in Table I. The first four samples show arcuracy obtained for single determinations of copper in stainless steels. The remainder show accuracy and precision for samples of various types. Kith two eucep-
1573 tiohs, good agreement is shown with Sational Bureau of Gtandards values. Results for samples 50b and 153 n-ere someJrhat higher than the certificate values, although separate tests of the alloying elements shoned no interference in the method. Because the nature of the elements present in these samples causes some question of the exact copper values, it would he difficult to make N close evaluation of accuracy at this time. Any doubt as to the arrurary of the method for a complex steel could be removed, however, t)y using a similar sample with knoirn copper content i n preparing the reference solution. Although large amounts of aluminum :ause interference with ropper rubeanate color (j), the small amounts usudly present ill steel have no measurable effect. Slightly high rrsults ma>- br obtained lvith high-aluminum steel, such as siimple 106 in Table I. Correction for nivkel intcrferenc,e is acxwatr. as sho\rrr by results for sample 126 containing 36% nirkel. Except for high-nickel alloys, the method has t w r l n used ewexitially as n-ritten Cor the past 3 years for all types of steels arid various other ferrous alloys. As a result of the above investigation, the method has I ) w n extended to high-nirkrl stwls arid has proved rPliah1r and convc,nient for routine usc~'. L I T E R i T U R E CITEI)
(1) Center, E. J., and Macintosh, H. JZ., 1x1). EN(:. CHEM.,A w r . . ED., 17,239 (1945). (2) Hoar, T . P., .Innlyst, 62,Ci57 (1937). (3) Levine, W. S.,arid Seaman. H., IND. ESG. CHKX., A X I L . ED., 16 80-2 (1944). (4) Willard, H. H., and Dirhl, H . , "Advanced Quantitative Analysis," %-en.York, D. Van Sostrand ('o., 1948. ( 5 ) Willard, H . H . , RIoshrr. E:. SI., and Boylr, -4.,J., .\s.\r,. ('HEM.,21, 598 (1949).
RECEIVED Jiinr 6, 1950. The opinions euprrh-ed iri t h i r arti