V O L U M E 26, NO. 3, M A R C H 1 9 5 4 enough to obtain an accurate end point. Attempts to titrate a-mercaptopalmitic acid were unsuccessful. DISCUSSIOV
Application of the potentiometric nonaqueous titration method to mixed wool wax acids has made possible the determination of neutralization equivalents, which usually were unattainable or grossly inaccurate. The relatively pure and colorless model compounds liqted in Table I demonstrate the comparable accuracy of the nonaqueous methods and the accepted method foi the determination of neutralization equivalents. The mixed wool wax acid sample whirh gave the best agreement by the two methods is given in Table I. Other mixed wool wax acid samples gave either poorer agreement between the two methods or could be titrated only by the nonaqueous method. The compounds used demonstrate the feasibility of titrating fatty acidq containing substituted bromine or an amino, epoxy, dihydrosy, a-sulfonic acid, and ammonium a-sulfonate group. This has indicated that lactone, lactam, and epoxy functional gioups do not interfere in the determination of the neutralization equivalent. This method has also been used to analyze samples derived from wool wax which have an acid number as low as 17 and are so highly colored that they could not be analyzed by the indicator method. Determination of aminostearic acid (mixed position isomers) dissolved in 95% ethyl alcohol and titrated potentionietrically
59L with 0. liV aqueous sodium hydroxide was unsuccessful unless the amino group was first allowed to react with formaldehyde. In the method here employed, the methoxide ion is capable of titrating the aminostearic acid directly. Similarly, the ammonium a-sulfopalmitic acid which resulted in appreciable precipitate and a gradual color change when titrated with 0.LV aqueous sodium hydroxide in 95% ethyl alcohol could bereadily titratedpotentiometrically with sodium methoxide in the benzene-methanol solvent. 4CKNOWLEDGMENT
The samples of a-broinopalmitic, a-sulfopalmitic, and amnionium a-sulfopalmitic acids were kindly supplied by J. K. Weil. The 9, 10-dihydroxystearic, 9,10-epoxystearic, 12-ketostearic acids; and mixed aminostearic acids were supplied by Daniel Swern and the ystearolactone by Ahner Eisner, all members of the lahoratory staff. LITERATURE CITED
Folin, O., and Flanders, F. F., J . A m . C'hern. SOC.,34, 774 (1912). Folin, O., and Wentworth, -4.H., J . Biol. Chem., 7, 421 (1910). Fritz, ,J, S., and Lisicki, N. >I., ANAL.CHEM.,23, 589 (1951). Ogg, C. L., Porter, W. L., and Willits, C. O., ISD. ENO.C H m r . , ~ A L ED., . 17, 394 (1945). (5) Rescorla, A. R., Carnahan, F. L., and Fenske, 11. R., Ibid., 9, 505 (1937).
RECEIVED for review October 7, 19.53. Accepted November 23, 1953.
Possible loss of Iron during Sodium Carbonate Fusion of Silicates and Rocks HASKIEL R. SHELL Analytical Laboratory o f the Electrotochnical Laboratory, HE first step generally employed in the analyEis of silicates T a n d rocks containing much silica is a basic fusion with .qodium carbonate to prepare the sample for acidic separation of silica. This fusion is usually done in a platinum crucible with a gas burner as a source of heat. That the loss of iron during this st.ep may be serious is known, but' in practice is frequently disregarded. Duparc ( 2 ) said that the loss of iron to the platinum crucible \\-as caused by the reducing effect of flame gases that diffuse through t,he platinum and avoided loss by fusion in a muffle. Jacob (4)discussed losses of iron to the containing platinum rrucible during sodium carbonat'e fusions, pointing out t,hat !ossep occurred even if all iron was present initially as iron(II1) oxide. Also, he suggested that some (hut not too much) potas$ium chlorate be added to the podium carbonate to prevent this loss of iron. S o mention was made of possible loss of iron as a volatile chloride by such additions. Potassium chlorate, which decomposes a t 400" C., would be largely gone before t'he sodium caarbonate fusion started; therefore any effect noted by Jacob may have been from the residual potassium chloride. Dittler ( 1 ) also noted that such losses occurred and suggested that potassium nitrate be added to iron( 11)oxide-enriched samples to prevent such loss. I n analyses in this laboratory, potassium nitrate did not prevent the loss of iron to the containing platinum crucible. Hillebrand and Lundell ( 3 )objected to the use of R muffle because of a strong attack of the crucible around its rim by the molten sodium carbonate, with subsequent introduction of platinum into the solutions. Some procedures may state that "oxidizing conditions should be maintained" but do not give :my concrete data on how to obtain them or point out the serious errors involved when oxidizing conditions are not obtained.
U. S.
Bureau o f Mines, Norris, Tenn.
Usually, however, the Puhject is ignored entirely and each analyst must discover the details for himself or falsely assume that all iron of the sample, and only the iron of the sample, is contained in his later solutions. That this false assumption is frequently made is exemplified by the analysis of a kyanite mineral by well-known and respected laboratories both in the United Stateq and abroad. On identical samples of mineral the following rewlts were obtained by the various laboratories for percentage of iron(II1) oxide: 0.57, 0.66, 0.44, 0.24, 0.75, 0.39, 0.36, 0.51, 0.73, and 0.58. Estimation of the true value for iron in this sample \ \ a