August, 19661 RUSSELL 611 Flame-photometric Determination of Sodium and Potassium in Manganese Ores BY B. G. RUSSELL (The lZiational Institute for Metallurgy, Yale Road, Milner Park, Johannesburg) Two procedures are described; in one, the sample is dissolved in hydro- chloric acid, interfering elements are precipitated with 8-hydroxyquinoline in ammoniacal solution, and the precipitate then extracted into chloroform. Sodium and potassium are determined in the aqueous phase by means of a filter flame photometer. The second procedure is more suitable for routine use and involves the dissolution of the sample in hydrochloric acid, followed by the addition of sulphuric acid and aluminium nitrate to suppress interferences, and the direct evaluation of the sodium and potassium contents of the solution by means of either a prism or a filter flame photometer. Comparative results obtained by this alternative procedure on instruments of these two types are given.GROWING importance is being attached to the reliable determination of sodium and potassium contents of South African manganese ores, and this problem is aggravated by the widely discrepant results currently reported by suppliers and customers on the same sample. TABLE I RESULTS OF THE DETERMINATION OF SODIUM AND POTASSIUM IN SOUTH AFRICAN MANGANESE ORES Analysis by suppliers Analysis by customers A 7- r 7 Sample Sodium oxide, Potassium oxide, Sodium oxide, Potassium oxide, No. per cent. per cent. per cent. per cent. 0.56 0.48 0.65 0.44 0.51 0.89 0.76 1-14 0.51 0.62 0.41 0.36 0.46 0-12 0.14 1-08 0.61 0-46 0.61 0.93 The classical gravimetric procedures of J.Lawrence-Smith1 and Berzelius2 for determining sodium and potassium were found to be too time-consuming and inaccurate for present purposes. Grimaldi3 has successfully applied a “standard-addition” method to the determination of sodium and potassium in siliceous rocks, but we were unsuccessful in applying this procedure to the analysis of manganese ores, owing to the non-linear relationship between emission and alkali content of the sample. The choice of a flame-photometric procedure for determining sodium and potassium is largely dictated by the type of instrument available. Instruments involving the use of a prism are invariably more expensive than those in which filters are used. If a simple filter instrument typified by the “EEL” flame photometer (obtainable from Evans Electroselenium Ltd., Harlow, Essex), can be used, it has economic advantages. The “EEL” flame photometer has been described in detail by Collins and Polkinhorne,* and was used in the investigational work described in this paper.To obtain further informa- tion on the performance of this instrument many of the solutions examined on it were also examined on a Beckman Model DU spectrophotometer that had been fitted with a flame attachment. The mutual interferences of alkali metal^,^ calciumJ6 9’ ,8 and some anions4 have already been investigated. Farrow and Hill9 have also studied the effect of cations in the determina- tion of alkali metals, and reported interferences by the chief constituents of manganese ores, i.e., iron and manganese.512 RUSSELL : FLAME-PHOTOMETRIC DETERMINATION OF [Analyst, Vol.91 The relatively cool air - coal gas or propane - butane flame of the “EEL” instrument excites fewer elements than the higher temperature flames of other instruments. Cationic interferences are therefore less with an “EEL” instrument ; however, a filter instrument is not sufficiently selective for accurate determination of sodium and potassium, because emissions due to iron and manganese are known to pass through the filter^.^ Bond and Staces found that the use of narrow wavelength-band filters largely eliminates emission from calcium and strontium and, although the use of such filters might be an advantage in eliminating interference from iron and manganese emissions, these filters were not readily available.METHOD I-SUPPRESSION OF INTERFERENCE BY SOLVENT EXTRACTION EXPERIMENTAL PREPARATION OF A SOLUTION OF THE SAMPLE- Hydrochloric acid was used to dissolve the samples because sodium and potassium are present in manganese ores as cryptomelane and ephesite, both of which are readily soluble in this acid. This was later verified by the good agreement of the sodium and potassium values obtained in repeat tests in which the samples were decomposed with hydrofluoric acid. SEPARATION OF INTERFERING SUBSTANCES- Manganese is the major interfering element in the analysis of manganese ores. It is not readily absorbed on a cation-exchange resin unless it is present in solution as permanganate. This, however, presents difficulties that arise from the marked tendency for manganese in solution to precipitate as manganese dioxide during oxidation.Few reagents, other than 8-hydroxyquinoline, precipitate manganese quantitatively. Many other metals are also quantitatively precipitated by 8-hydroxyquinoline in alkaline solution, but alkalis tend to co-precipitate, and errors are introduced if the precipitate is removed by filtration. If, however, the precipitate is extracted into chloroform, no such loss of alkalis occurs. If calcium and magnesium are present, it is essential to use a large excess of 8-hydroxy- quinoline, but it is claimed that the presence of butyl Cellosolve increases the solubility of the magnesium complex in chloroform,lO and so reduces the amount of chloroform required for the extraction. Butyl Cellosolve also has the advantage of reducing the amount of chloro- form necessary for the extraction of other metal 8-hydroxyquinolinates ; it also aids separation of the organic and aqueous phases.If only a small amount of calcium is present, the calcium precipitate may be removed by filtration after the chloroform extraction. The precipitation of manganese is quantitative only if the pH of the solution is between 7 and 9.5, and a reducing agent, e.g., hydroxylamine,ll is present. When an acetic acid solution of 8-hydroxyquinoline was used, the extraction of manganese was incomplete, and subsequent removal of the large amount of ammonium acetate formed was difficult and time-consuming. When, however, an alcoholic solution of the reagent was used, these problems were not encountered.PREPARATION OF STANDARDS- To minimise error, acidities were adjusted so that each solution contained 3 per cent. v/v of perchloric acid before it was sprayed into the flame. Standard solutions containing known amounts of both sodium and potassium oxides, in the ratio of 1 to 4, were used for calibration purposes, as this was approximately the ratio of sodium to potassium found in the samples investigated. It has been shown5 that wide variations in this ratio are permissible because the mutual interference of sodium and potassium is negligible if the amount of either does not exceed that of the other by a factor of more than 10. METHOD REAGENTS- All reagents should be of the highest purity obtainable.Ammonia solution-Pass ammonia gas into water in a plastic bottle until a saturated This reagent, as supplied, usually has an unacceptably solution of ammonia is obtained. high sodium content.August, 19661 SODIUM AND POTASSIUM I N MANGANESE ORES 513 8-Hydroxyquinoline (8 per cent. w/v)-Dissolve 100 g of the reagent in 70 ml of 96 per cent. ethanol, then add 900 ml of water. Filter the solution through a Biichner funnel, and wash the precipitate with water. Continue to draw air through the precipitate for about 30 minutes, then transfer it to a dark bottle and store it, preferably in a refrigerator. Dissolve 8 g of the purified reagent in 100 ml of ethanol; prepare the reagent solution daily. Standard sodium solution-Dissolve 1.8860 g of sodium chloride (dried at 105" C) in Dilute this solution 10 times for use.water, add 5ml of perchloric acid (spgr. 1-58), and dilute the solution to 1 litre. 1 ml of solution = 0.1 mg of sodium oxide Standard potassium solution-Dissolve 1.5830 g of potassium chloride (dried at 105" C) in water. Add to this solution 5 ml of perchloric acid (sp.gr. 1-58) and dilute the solution to 1 litre. Dilute this solution 10 times for use. 1 ml of solution = 0.1 mg of potassium oxide Keep the volumes of all reagent solutions to a minimum and ensure accurate com- pensation for the blank by standardising the amount of each reagent added at each stage of the procedure. APPARATUS- funnels. detectable error. Use quartz apparatus wherever possible, and make extractions in Pyrex separating Calibrated flasks made of soda-glass were used, but these did not introduce any Use a filter (e.g., "EEL") flame photometer with an air - coal gas flame.PROCEDURE- Determine a blank on the reagents with each batch of samples. Transfer the sample (see Note 1) to a 150-ml quartz beaker, and add 15 ml of hydrochloric acid (spgr. 1.16). Evaporate the solution to dryness on a hot-plate, then cool it slightly. Add about 2 drops of hydrochloric acid (sp.gr. 1.16) and 20 ml of water. Boil the solution gently to dissolve soluble salts; allow to cool. Transfer the entire contents of the beaker to a calibrated flask and dilute the solution to the mark. Filter the solution through a dry filter-paper into a dry quartz beaker, and transfer a 20-ml aliquot, representing not more than 0.29g of sample, to a 250-ml quartz beaker.Add 2 g of hydroxylammonium chloride, dilute the solution with water to about 150 ml, then add 10 ml of the 8-hydroxyquinoline solution. Adjust the pH of the solution to between 7 and 9, testing with indicator paper; heat it to about 60" C (do not boil), then cool. Transfer the entire contents of the beaker to a 150-ml Pyrex separating funnel, and add to the solution 2 ml of butyl Cellosolve and 15 ml of chloroform. Shake the funnel for about 30 seconds, then draw off and discard the chloroform phase. Repeat the extraction twice, with 5-ml portions of chloroform. Transfer the aqueous solution to a 150-ml quartz beaker; if necessary, filter the solution through a Whatman No. 40 filter-paper ; wash the filter-paper sparingly with water.Add 3 ml of nitric acid (spgr. 1.42), evaporate the solution to dryness, then cool. Add 2 ml of perchloric acid (sp.gr. 1-54), again evaporate the solution to dryness, volatilise the excess of perchloric acid and then cool. Add 1.5 ml of perchloric acid (sp.gr. l a % ) , warm gently to dissolve soluble salts, then transfer the clear solution to a 50-ml calibrated flask and dilute with water to the mark. Determine the sodium and potassium emissions of the solution a t 589 and 766.5 mp, respectively, and calculate the sodium and potassium oxide contents of the sample from calibration graphs prepared by using the appropriate standard sodium and potassium solutions. XOTE 1- When analysing batches of samples containing the same approximate ratios of sodium and potassium, it is advantageous to prepare calibration standards containing both sodium and potassium in similar ratios to the samples.When the ratio of the one element to the other exceeds 10 to 1, this technique becomes necessary to compensate for mutual interference of the two elements.514 RUSSELL FLAME-PHOTOMETRIC DETERMINATION OF [ A nabst, VOl. 91 PRECISION OF THE METHOD- The samples contained about 45 per cent. of manganese, and 13 per cent. of iron, 3.5 per cent. of silica, 1.1 per cent. of barium oxide and 4 per cent. of alumina. This is shown in Table 11. TABLE I1 DETERMINATION OF SODIUM AND POTASSIUM BY PROPOSED METHOD Sodium oxide Potassium oxide - - Sample number . . .. 2 4 2 4 Determinations . . .. 11 7 14 7 Mean value . . .. . . 0.193 per cent. 0.207 per cent. 0.843 per cent. 0.682 per cent. Standard deviation . . . . 0.016 0.014 0.022 0.030 Coefficient of variation . . 8.3 per cent. 6.8 per cent. 2-7 per cent. 4-4 per cent. COMPARISON OF RESULTS- The standard deviation was higher for the determination of sodium than it was for the potassium determination. A probable explanation lies in the proportionately higher blank in the sodium determination: the mean of 5 determinations was 0.072 per cent. of sodium oxide; the corresponding blank in the potassium determination was a mean of 0-036 per cent. of potassium oxide for 6 determinations. RECOVERIES OF ADDED SODIUM AND POTASSIUM- To assess the extent of any loss of alkalis in the proposed procedure, additions equivalent to 2.0 mg of sodium oxide and 5.0 mg of potassium oxide were made to a series of 0.5-g samples of purified manganese dioxide.Recoveries of sodium oxide varied between 98-1 and 100.6 per cent. and the blank values between 22 and 65 pg; the standard deviation was 0-035 pg, and the coefficient of variation was 1.7 per cent. Recoveries of potassium oxide varied between 92.0 and 102.0 per cent. and the blank values were between 8 and 49 pg; the standard deviation was 0.167 pg, and the coefficient of variation was 3.5 per cent. These recoveries show that there is no appreciable loss of sodium or potassium in the proposed procedure and emphasise the necessity to make frequent blank determinations, especially with each batch of samples. CONCLUSION The method is satisfactory for determining sodium and potassium in manganese ores.The procedure is, however, time-consuming and not suitable for application on a routine basis. As very few elements interfere, the method is ideally suited for establishing the sodium and potassium content of samples of manganese ores of variable composition. The investi- gation of a batch of 7 samples and a blank can be completed in about 10 working hours. METHOD II-SUPPRESSION OF INTERFERENCES BY THE ADDITION OF ALUMINIUM EXPERIMEKTAL LIMITATIONS OF A FILTER FLAME PHOTOMETER- The main limitation of a filter instrument is due to the transmission of iron and man- ganese emissions through the sodium filter, and iron emission through the potassium filter. Evans Electroselenium Ltd. reported that the presence of aluminium suppresses interference due to and it was decided to investigate the suppression effect of aluminium on iron and manganese emissions.According to Collins and P~lkinhorne,~ hydrochloric acid and chlorides seriously suppress sodium and potassium emissions when the chloride concentration exceeds 0-012 N. Therefore, chlorides introduced during the dissolution of the sample should be removed by evaporating the sample solution with a measured excess of sulphuric acid.August, 19661 SODIUM AND POTASSIUM I N MANGANESE ORES TABLE I11 FLAME PHOTOMETER READINGS SHOWING THE EFFECT OF SULPHURIC ACID ON EMISSION OF SODIUM AND POTASSIUM Sulphuric acid, normality Nil 0.05 0-1 0.4 0.5 0.7 1.0 1.5 2.0 3.0 Sodium oxide, 30 p.p.m. “EEL” Beckman 66.5 60.5 - 61-5 - 62.0 - 62.0 65.0 - - 62.0 63.5 62.0 61.0 60.0 58.5 57.0 - - Potassium oxide, 30 p.p.m.“EEL” 58.0 - 61.0 59.0 59.0 58.0 56.0 52.5 Beckman 63.0 63.0 66.0 64.5 62.5 57.5 53.0 - - - 515 Both instruments were adjusted to give 100 divisions deflection equivalent to sodium oxide and potassium oxide, each at the 50 p.p.m. level. Results in Table I11 show, in general, that sulphuric acid does not have an erratic or pronounced effect on the emissions of sodium and pot4assium, provided that the strength of this acid is about 0-7 N. It was decided, therefore, to standardise the amount of sulphuric acid present at 0.8 N, and avoid an excessive loss of this acid when solutions were evaporated to remove chlorides. The next aim was to establish the effect of added aluminium, and the tests made are summarised in Table IV; the strength of sulphuric acid in these and all subsequent tests was maintained at 0.8 N.TABLE IV FLAME PHOTOMETER READINGS SHOWING THE EFFECT OF ALUMINIUM (AS NITRATE AND SULPHATE) ON EMISSIOXS OF SODIUM AND POTASSIUM Aluminium added, p.p.m. 0 25 50 75 100 150 200 300 500 1000 1500 2000 3000 4500 5000 6000 30 p.p.m. of sodium oxide with aluminium added as- A f > Nitrate Sulphate 3 E - z G z GzBeckman’ 66-5 66.5 66.5 66.5 - - 67.0 - - 66.5 - 65.0 67.0 66.5 67-0 66.5 - 67.0 67.5 69.0 67-0 __ - - 67.0 68.0 66.6 68-0 66.0 66.0 - - 62.0 64.0 66.5 66.0 59.0 62.0 66.5 63.0 - 66.5 63.0 - 66.5 63.0 - - - - - - - - - - - - - - - - - - 30 p.p.m. of potassium oxide with aluminium added as- Nitrate 5zF-ZGz 58.0 63.0 - - - - 58.5 - - - - - 58.5 65.0 58.0 65-0 58-0 60.0 58.0 - 54.5 55.0 53.0 55.0 - 55.0 - - - - Instrumental settings as for Table 111.Sulphate Gr-kzBeckmaI: 58.0 63.0 - 62.0 59.0 - 59-0 64.0 59.5 - - 66.0 60.0 - 59-5 68.0 59.0 66.0 58.6 66.0 56.5 63.5 - - - - The results contained in Table IV show that the sodium emission is constant when the concentration of aluminium (as nitrate) exceeds 1000 p.p.m. and that the potassium emission is reasonably steady for concentrations of aluminium (as nitrate) up to 3000 p.p.m. If the aluminium is added as sulphate, the emissions of both sodium and potassium vary. The sulphate concentration must therefore be kept constant, and this requirement precludes the use of aluminium sulphate in place of aluminium nitrate to suppress inter- ferences arising from the presence of iron and manganese.516 RUSSELL : FLAME-PHOTOMETRIC DETERMINATION OF [AnaZySi!, VOl.91 TABLE V FLAME PHOTOMETER READINGS SHOWING THE EFFECT OF IRON AND MANGANESE ON EMISSIONS OF SODIUM AND POTASSIUM Iron or manganese added, p.p.m. 0 1000 2000 4000 5000 8000 10,000 30 p.p.m. of sodium oxide, with added- 30 p.p.m. of potassium oxide, with added- A r > I h Iron Manganese Iron Manganese e- 5 B e c k m z - n 66.5 66-5 66.5 66.5 61.5 63.0 61.5 63.0 75.5 - 70.5 63.0 - 63.0 83.0 65.0 74.5 71.0 66.0 61.0 61.5 63.5 - 66.0 - 65-5 79.0 - - 62.5 61.5 - - 66.5 - - 73.0 - - 68.0 92.0 75.5 - 64.0 70.0 67.0 - - > 100 - - 71.5 83.0 - - - Concentration and instrumental settings as for Table VI. Table V shows that the potassium emission is unaffected by up to 5000 p.p.m.of man- ganese] although the presence of iron at the 1000 p.p.m. level causes high readings to be obtained. Interference due to the presence of iron is satisfactorily suppressed in the presence of between 2000 and 5000 p.p.m. of aluminium (as nitrate)] as shown in Table VI. TABLE VI FLAME PHOTOMETER READINGS SHOWING THE EFFECT OF ALUMINIUM NITRATE ON THE COMBINED INTERFERENCE OF IRON AND MANGANESE ON THE EMISSIONS O F SODIUM AND POTASSIUM Aluminium Sodium oxide, added 30 p.p.m. (as nitrate), p.p.m. 1 d e n 0 - 79.5 150 - 81.0 300 - 78.0 500 - 76.5 600 - 76.0 1000 92.0 75.0 2000 88-0 - 3000 86.0 75.0 5000 84.0 75.0 10,000 84.0 82.5 12,000 84.0 - The solutions contained 1400 p.p.m. of iron and Instrumental settings were as for Table IV. Potassium oxide, 30 p.p.m.- “EEL” Beckman 76.0 72.0 76.5 71.0 74.0 - 72.5 69.5 66.5 65.5 66.5 61.5 - 61.5 66.5 61.5 66.5 < 60.0 58.0 - - - 4880 p.p.m. of manganese. Interference effect on the sodium emission by both iron and manganese (Table V) is probably a result of the light from the respective emission lines at 586.8 mp and 586.0 mp passing through the sodium filter. The presence of aluminium (as nitrate) in concentrations TABLE VII DETERMINATION OF SODIUM AND POTASSIUM IN THE PRESENCE OF MANGANESE AND IRON Sodium oxide, Potassium oxide, Manganese p.p.m.* p.p.m.* added, & 7-- p.p.m. “EEL” Beckman “EEL” Beckman 1000 33.0 30.0 30.4 30.8 2000 33.6 30.3 30.4 30.2 3000 35.0 36.5 30.7 30.7 4000 36.8 30.0 30.2 30.6 Iron added, p.p.m. 1000 36.0 30.0 30.6 30.4 2000 37.6 30.1 29.9 30.0 3000 39.9 30.4 30.4 30.8 4000 41.8 31.4 30.5 30.6 * 30 p.p.m. added: aluminium, 5000 p.p.m.August, 19661 SODIUhl AND POTASSIUM I N MANGANESE ORES 517 above about 3000 p.p.m.partly suppresses this interference (Table VI), but, as the results in Table VII show, errors are incurred if the amounts of iron and manganese are variable, therefore, iron and manganese equivalent to the amounts of these elements present in the sample must be added to the standards used for calibrating the instrument. This is not necessary for potassium determinations as the interference of iron and manganese is com- pletely suppressed by the addition of between 2000 and 5000 p.p.m. of aluminium (as nitrate). COMPARISON OF A FILTEK INSTRUMENT WITH A PRISM INSTRUMEKT- Experiments conducted on the “EEL” instrument (Table V) were repeated on the Beckman instrument and, for convenience, the two series of results are placed side by side in Table V.A comparison of these results shows that interference arising from the presence of manganese and iron is less pronounced with a prism instrument than it is with a filter instrument. At this stage, it was thought that valuable information could be obtained by repeating the experiments, the results of which are shown in Tables 111, I V and VI; on this occasion a prism instrument was used, and, for convenience, the two series of results are placed side by side in these tables. The monochromator of a prism instrument enables a narrow band-pass to be obtained that largely eliminates emission bands at 586.8 and 586-0 mp.Conclusions reached from these additional experiments with a prism instrument make it possible for both sodium and potassium to be determined in the presence of manganese and iron without compensating for these elements in the standards used to prepare the calibration graph. COMPARISON OF THE TWO METHODS WITH OTHER METHODS- Each of the 5 manganese ore samples examined (Table VIII) contained about 3.5 per cent. of silica, 1.1 per cent. of barium oxide and 4-7 per cent. of alumina, The manganese contents ranged from 40 to 45 per cent. and the iron contents from 12 to 17 per cent. Results obtained by procedures given in the references and by the recommended methods referred to in this paper are shown in Table VIII. The samples were South African man- ganese ores.METHOD REAGENTS- All reagents should be of the highest purity obtainable. Aluminium nitrate solution-Dissolve 713 g of hydrated aluminium nitrate (A1(NO,),.9H2O) in water, and dilute the solution to 1 litre. This solution contains 50,000 p.p.m. of aluminium. Standard sodium and potassium solutions-See Method I. APPARATUS- with a flame attachment and an oxy - hydrogen flame. Use an “EEL” flame photometer with an air - coal gas flame, and a Beckman Model DU PROCEDURE- Determine a blank on the reagents with each batch of samples. Transfer the sample (see Kote 1 under Method I) to a 250-ml quartz beaker, add 20 ml of hydrochloric acid (sp.gr. 1.16) and boil the solution for about 15 minutes. Add 10 ml of 8 N sulphuric acid and evaporate the solution until fumes of sulphuric acid begin to appear (avoiding any significant loss of this acid), then cool the solution.Dilute the solution with water to about 30 ml, then filter it quantitatively through a Whatman No. 40 filter-paper into a 100-ml calibrated flask containing 10 ml of the aluminium nitrate solution (see Xote 2 ) . Dilute the solution with water to the mark. Determine the sodium and potassium emissions of the solution a t 589 and 766.5 mp, respectively, on either of the two types of flame photometer, and calculate the sodium and potassium oxide contents of the sample from appropriate calibration graphs prepared by using the standard sodium and potassium solutions. If a filter instrument is used for the determination of sodium oxide, add, as nearly as possible, the same amount of iron and manganese as that present in the sample solution, to the standard sodium solution before preparing the calibration graph.518 RUSSELL FLAME-PHOTOMETRIC DETERMINATION OF [ A PZUlySt, VOl.91 XOTE 2- After taking into account the aluminium content of the sample, this addition should provide a final 100-ml solution containing between 3000 and 5000 p.p,m. of aluminium, if a prism flame photometer is used; otherwise, the range should be between 3000 and 6000 p.p.m. of aluminium. PRECISION OF THE METHOD- This is shown in Table IX. The samples were similar in composition to those analysed in Table 11. Sample number 1 2 3 4 5 1 2 3 4 5 r Method given in Ref. 1. 0.50 0.48 0.47 0.41 0.5 1 TABLE VIII COMPARISON OF METHODS Sodium oxide, per cent.L > Method I1 Method given Method I, in Ref. 3 "EEL" Beckman 0.28 0.22 0.22 0.22 0.39 0.19 0-17 0.17 0.42 0-20 0.20 0.20 0.50 0.2 1 0.2 1 0.2 1 0.41 0-23 0.20 0.19 Potassium oxide, per cent. f h \ 1-12 1.18 1.01 1.01 1.00 0.96 0.93 0.84 0.79 0-80 1-02 1-14 1.11 0.95 0.94 0.74 0.68 0-68 0.66 0.62 1-58 1.22 0.98 0.92 0.91 TABLE IX DETERMINATION OF SODIUM AND POTASSIUM BY PROPOSED METHOD 11 Sodium oxide Potassium oxide Beckman Beckman Determinations . . .. . . . . 12 12 12 12 Mean value, per cent. . . . . . . 0*212 0.2 13 0.670 0.632 Standard deviation . . . . . . 0.007 0.006 0.007 0.00 1 Coefficient of variation, per cent. . . 3.3 2.6 1.1 1.7 Table X shows the results obtained by another laboratory on six international standard samples with a Zeiss PMQ I1 spectrophotometer with flame attachment. TABLE X DETERMINATION OF SODIUM AND POTASSIUM IN INTERNATIONAL STAXDARDS BY METHOD I1 Sodium oxide, per cent.f A 7 Value by Sample Method 11 Accepted mean value T-112 . . .. . . 4.46 4-39 SY-113 .. . . 3.40 3.24 G-114 . . . . . . 3.37 3.39 N.B.S. NO. 9115 . . 8.48 8.48 STD-GH" . . . . 3.82 3.75 STD-GR" . . . . 3.90 3.80 Potassium oxide, per cent. r 7 Value by A Method I1 Accepted mean value 1-28 1.23 2.64 2.75 5.60 5.52 3.22 3.25 4.62 4-70 4.48 4.50 CONCLUSIONS The direct determination of both sodium and potassium by Method I1 is reliable if a prism instrument is used. With a filter instrument, iron and manganese interfere in the determination of sodium only, but this interference can be overcome by adding iron and manganese to the calibrationAugust, 19661 SODIURf AND POTASSIUM I N MANGANESE ORES 519 solutions.For the determination of potassium only, a filter instrument can be used with advantages in speed and simplicity and without any serious loss of precision. Results obtained by Method I1 (and Method I) are more reliable than those obtained by alternative procedures examined.l j3 I thank Dr. R. E. Robinson, Director of the National Institute for Metallurgy, for valuable advice and for permission to publish this paper, Mr. T. W. Steele for his helpful suggestions throughout this project, and Mr. J. Ferguson of the Geology Department, Witwatersrand University, for the results quoted in Table X. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. REFERENCES Lawrence-Smith, J., Amer. J . Sci., 1871, 50, 269. Berzelius, J. J., A n n l n Phys., 1824, 1, 169. Grimaldi, F. B., Prof. Pap. U.S. Geol. Surv., 1965, No. 400B, 226. Collins, G. C., and Polkinhorne, H., Analyst, 1952, 77, 430. Bond, R. D., and Stace, H. C. T., Ibid., 1958, 83, 679. Leaflet Reference No. 1704.16 “EEL” 1162, Evans Electroselenium Ltd., Harlow, Essex. Bond, R. D., and Hutton, J . T., Analyst, 1958, 83, 684. Leaflet Reference No. 1704.8, Evans Electroselenium Ltd., Harlow, Essex. Farrow, R. N. P., and Hill, A. G., Talanta, 1961, 8, 116. Umland, F., and Hoffman, W., Analytica chim. Acta, 1954, 11, 120. TVelcher, F. J., Editor, “Organic Analytical Reagents,” Vol. I, D. van Sostrand Co. Inc., New Tanganyika Geological Survey, Standard Geochemical Sample T-1, Supplement No. 1 ( 1961). Webber, G. R., Geochim. cosmochim. Acta, 1965, 29, 229. Ingamells, C . O., and Suhr, N. H., Ibid., 1963, 27, 897. Natn. Bur. of Stand., IVashington, D.C., Standard Sample KO. 91, “Opal Glass.” Roubault, H., and Govindaraju, H. de la R., Sciences Tewe, 1963, 9 , No. 4, 339. York, 1947, p. 264. Received September Oth, 1963 Amended, April 26th, 1966