136 HARVEY AND MURRAY: A SPECTROGRAPHIC METHOD FOR THE [Vol. 83 A Spectrographic Method for the Determination of Rarer Elements in Silicates BY C. 0. HARVEY AND K. L. H. MURRAY (Geological Survey and Museum, Exhibition Road, S. W.7 and Department of the Government Chemist, Clewent’s Inn Passage, Styand, W.C.2) A general method for the determination of rarer elements in silicate rocks and minerals is described. Calcium sulphate is used as the spectrographic buffer and internal standard, the source of energy being a d.c. carbon arc. MANY rarer elements are detectable by spectrographic methods in silicate rocks and minerals. The literature of the subject is extensive: inforrnation concerning their own work and the work of others has been summarised by Mitche1.P and by Ahrem2,3 In the development of the general method described later, the aim has been applicability to specimens that have widely different major constituents and a compromise between simple direct methods and the more accurate methods, in which a selected internal standard is added for the element to be determined. The features of the proposed method are as follows- (i) Its general suitability for silicate rock; and minerals.(ii) The provision for estimations by visua.1 inspection of the photographic plate and for photometric determinations of elements that are sometimes present in relatively large amounts, e.g., chromium, nickel and vanadium. (iii) Some measure of control is achieved by using an internal standard. In order to provide for (i) and (iii), the sample is buffered with calcium sulphate,* which, at the temperature of the arc, will rapidly decompose the minerals in the sample and, being relatively involatile, will persist in the arc gases until the “burn” is completed.This dilution with calcium sulphate helps to stabilise the excitation conditions and to reduce the effects of variations in the major constituents of the samples, and also provides an expedient whereby valid corrections for impurities in the carbon electrodes can be made: as is well known, the detection of impurities in electrodes is sometimes stimulated when the arc contains added mineral matter. To facilitate (ii), the spectra are stepped by using a rotating stepped sector, a device that also greatly widens the ranges of element concentrations determinable by a single exposure.PHOTOMETRIC DETERMINATIONS The calcium-atom line at 3140.78 A is used as the internal-standard control line, com- parison with the analysis lines being made at constant photographic density. At a suitable density, the linear separations of the curves for density against the log of the relative exposure for the calcium and analysis lines are measured and correlated with percentages by means of a working curve, e.g., Fig. 1 (a). A nomogram has been constructed to facilitate the calculation of Seidel densities. The requirements for an ideal internal-standard spectrographic technique include the use of pairs of lines known to respond similarly to changes in excitation conditions, of element pairs of similar volatility, of lines free from self-a.bsorption and of line pairs of similar wave- lengths.Ideal requirements cannot be satisfied in a general method, but the use of the calcium line at 3140.78 A provides some measure of control by internal standard, is practicable for a general method and is preferable to a direct method without an internal standard. Photographic processing conditions must be carefully st andardised in order to minimise the effects of variations of plate gamma with wavelength. * Since the work described in this paper was completed, Turekian, Gast and Kulp4 have shown that, for strontium determinations in silicates, the effects of varying matrix are reduced by adding calcium carbonate. Their figures for the strontium contents of two standard rock powders are similar to figures obtained by us for the same powders.March, 19583 DETERMINATION OF RARER ELEMENTS I N SILICATES 137 The presence of different amounts of calcium in the rock or mineral specimens will cause only slight variations in the density of the calcium line at 3140.78 A.For specimens con- taining up to 10 per cent. of CaO, errors arising from this source are small, and the provision of a series of working curves for a range of calcium contents is normally not necessary. L 0 I 1 I I I 1 1 I 0.001 I 0.003 1 0010 I 0.030 I 0.1 0 0402 6905 0.020 0.050 0.2 Chromium(Log, scale),% Fig. 1. Calibration curves for determin- ations of chromium by using the Cr 4254 line; (a) photometric method ; ( b ) step-counting method. With (0) synthetic base and (A) rock- powder base CORRECTION FOR BACKGROUND- Corrections for the background produced by continuous radiation are made by using the ratio of the galvanometer deflections for “background” and for “line plus background” to calculate the photographic density of the line freed from background. Although it is not strictly valid, this method of correction appears to serve quite well for a general spectrographic method when a buffer is used.VISUAL ESTIMATIONS Estimations are made by counting the number of steps in which an image of the element line is visible, with fractional estimations of the density of the weakest step, e.g., 3&, 3+, 32 or 4 steps for a line visible in the fourth step, but not in the fifth step. The visual comparison of stepped spectrograms has been used by earlier w o r k e r ~ .~ , ~ , ~ For any chosen element line, a count of the number of steps provides a relative inverse measure of the exposure necessary to produce a visible image, and is related to the intensity of emission. A rectilinear relationship between “number of steps” and “log, yo,” which should apply if reciprocity and intermittency effects are negligible, has been found to be valid for many of the calibratory plots actually obtained, e.g., Fig. 1 (b). The calcium line at 3140.78 A, used as the control line for photometric determinations, can also be used to correct observations made by the visual step-counting method. CORRECTION FOR BLANK DETERMINATIONS- A blank spectrum, obtained by striking an arc in a mixture of calcium sulphate and carbon, is recorded on every plate, so that corrections can be applied as necessary, notably for the small amounts of vanadium sometimes present in the carbon electrodes.The blank spectrum also helps with the avoidance of pitfalls, such as the confusion of a line of an (OH) band with the bismuth line at 3067.7 A. Dingles states that this band often appears in the spectrum of an arc in moist air. Correction for traces of an impurity in the carbon or calcium sulphate is difficult to apply in terms of weight or percentage when a similar correction must also be applied to138 HARVEY AND MURRAY: A SPECTROGRAPHIC METHOD FOR THE [Vol. 83 the standard spectra used for constructing the working curve, i.e., the amount of the impurity that produces the blank response is not easily determinable.For photometric determinations, which are usually made only when the element response is strong, correction for the small blank value should not be necessary. For estimations made by the visual step-counting method, we find that a sufficiently exact correction can be made by deducting the “intensity value” of the blank determination. We define the “intensity value” of the image of a stepped spectrum line visible in 1, 2, 3, 4, 5, 6 or 7 steps, as, respectively, 1, 2, 4, 8, 16, 32 or 64 for a stepped sector having an exposure ratio of 2 to 1. As an example, for an element line visible in, say, 34 steps, with a blank value of 1 step, the corrected “intensity value’’ is (6 - I), and the corrected number of steps is 3i. NOTES ON THE SPECTRUM LINES Carefully selected spectrum lines, suitable for determinations of the rarer elements in silicate rocks and minerals, are listed in Table 11, together with appropriate comments.Although many of the lines selected appear in Dingle’s* lists of sensitive multiplets, the most sensitive line for any particular element has not always been chosen, either because interference by (CN) bands or by emission from another element is likely, or because the most sensitive line is not located within the chosen wavelength range, i.e., 2750 to 4 6 7 0 ~ . Some elements have therefore been listed that will rarely be detected in silicate rocks, but the inclusion of these elements ensures that their presence in unexpectedly high amounts does not escape notice. The wavelengths quoted are based on those given by Harri~on.~ Many of the lines chosen, being atom lines, call for the use of an atom line as internal- standard control line.Noteworthy exceptions are the zirconium lines at 3392 and 3438~, the beryllium lines at 3130 and 3131 A and the niobium lines at 3163 and 3195 A, all of which are ion lines and none of which responds well to correction by the calcium-atom line at 3140.78 A. SEPARATION OF LINES- The variation of dispersion of a prism spectrograph with wavelength and the resolving power of the photographic emulsion are factors that affect the interference of one spectrum line with another of similar wavelength. TABLE :L SEPARATION OF SPECTKUM LINES Wavelength, 2780 3000 3200 3400 3500 3706 3906 4005 4227 4337 4427 4638 A Plate factors, A per mm 3.7 4.7 5.8 7.1 7.8 9.3 11.0 12.0 13.4 16.1 17.0 18.2 Wavelength differences necessary to eliminate interference by one line with a near line, 0.11 0.14 0-17 0-21 0.23 0.28 0.33 0.36 0.40 0.45 0.5 1 0.55 A TABLE 1 3 SELECTED SPECTRUM LINES Approximate minimum Wavelength Element detectable, of line used, * Comments % A Sb 0.06 2877.92 Ca line may produce slight blank value.Cr may sometimes As 0.1 2780.20 Separable frorn Mn line at 2 7 8 0 . 0 0 ~ interfere 2860-45 Ti line a t 2860.28 A should be separable if present * The lines used for the detection of the minimum percentages are given in italics.March, 19581 DETERMINATION OF RARER ELEMENTS IN SILICATES 139 Approximate minimum Wavelength Element detectable, of line used, * Ba Be Bi Cd Ce Cr C O cu % 0.00 1 0.001 0.01 0.02 0.2 ? 0~0005 0.002 0.0003 Ga 0.002 Ge 0.005 Tn 0.002 Pb Li Mn Mo 0.005 0.005 0.04 (0.02 if Fe absent) 0.0005 0.001 A 3071-59 4554.04 3130.42 3 13 1-07 3321.01 3321.09 3321.34 2897.98 3067-72 3261.06 3 194.83 3234-16 3272-25 2780-70 2843.25 302 1.56 4254.35 3449-17 3449.44 3453-51 2824.37 3247.54 3273.96 2943.64 2944.18 3039.06 3269.49 3039.36 3256.09 3258.56 3245.12 3337.49 2833.07 3232-61 4602.86 2799.84 2801.06 2914.60 2925-57 31 70.35 3193.97 TABLE II-continued Comments Ca line produces slight blank value An ion line.Both are ion lines. Inseparable Cr line will not normally interfere. Avoid confusion with V line a t 3130.27 A Carbon bandhead a t 4553.1 A and Ti line at 3130.80 A Weak Cr line a t 3321-19 A Inseparable weak Mn line should not often interfere Line of (OH) band may produce blank value.Weak Fe line a t 3067.94 A Very weak Ca line just separable. Very weak V line not separable Nb line at 3194.98 A will interfere unless very weak. Very weak Ca line may produce slight blank value Inseparable lines of Pr, Zr, etc., are normally unlikely to inter- fere. Weak Cr line a t 3234.06~. When strong, Ti and Fe lines interfere Inseparable lines of Zr, V, etc., are normally unlikely to interfere. Ti line a t 3272.08 A just separable, unless strong Weak Fe line not separable. Cr line only suitable for strong response An ion line Weak Ti line not separable, but unlikely to interfere. Weak Cr Inseparable Ag line will not normally interfere Mn line may interfere when Cu response is weak Ca line may produce slight blank value. Cu lines at 3247 and Avoid confusion with Co line a t 2943.48 A.Inseparable Fe line Avoid confusion with Ni line a t 2943.91 A and Fe line a t Very weak Ca line a t 3039.21 A Weak Fe line at 3269.24 A Inseparable Fe line should usually be very faint. Avoid con- Inseparable Mn line often interferes. Avoid confusion with Mn line at 3258.41 A : presence of this line indicates interference An ion line. Inseparable Ce line should not normally interfere An ion line. Inseparable Ce line should not normally interfere. lines a t 3453-33 and 3453.74 A 3273 A are usually detectable in a blank determination will not normally interfere 2944-40 A fusion with Co line a t 3039-57 A Fe line at 3255.89 A of Mn with In line a t 3256-09 A Fe line a t 3337.67 A is separable Sb line a t 3232.50 A will not normally interfere Usable only in absence of Fe Inseparable Zn line will not normally interfere Inseparable weak Fe line will not normally interfere Inseparable V line may sometimes interefere.Weak Fe line a t 3193.80 A * The lines used for the detection of the minimum percentages are given in italics.140 HARVEY AND MURRAY: A SPECTROGRAPHIC METHOD FOR THE TABLE II-continued Approximate minimum Wavelength Element detectable] of line used,* Ni 0-0007 2943-91 3012.00 3380.57 3446.26 3452.89 3163.40 3194.98 3269-90 4246.83 % A Nb sc 0-005 0.002 Sr Ta T1 Th Sn 0.00 1 0.2 0.01 0.03 0-005 W 0-02 U > 0.5 V 0.002 Y 0.00 1 3464-46 4607-33 3311.16 2767.87 2837.30 2870.41 2982.05 2839.99 2863.33 3175.02 2946.98 2882.74 4362.26 2914.93 2924.03 2924.64 3183.42 4379.24 3200.27 3327.88 4643.70 Zn > 0.3 3345.02 Zr 0.0007 3391.98 3438.23 [Vol.83 Comments Inseparable Cu line will not normally interfere Avoid confusion with Ti line a t 3380.28 A Avoid confusion with Co line a t 3446.09 A Weak Fe line at 3453.02 A may sometimes interfere Both are ion lines. Inseparable weak Fe line will not normally interfere An ion line. Inseparable Ce line should not normally interfere. Avoid confusion with Fe line at 4247.43 A. Heavy background An ion line. Yb line a t 3464.37 A will not normally interfere Possible interference by Fe and Mn lines is rarely appreciable. Carbon bandhead a t 4606.1 A Very weak Nb line a t 3311.34 A Inseparable Ce, W and V lines should not normally interfere Inseparable Ce and Zr lines should not normally interfere Inseparable u eak Cr line should not normally interfere.Avoid Fe line at 2981.88 A. Ce line at 2981-91 A Inseparable weak Mn and Cr lines should not normally interfere Sometimes not free from interference by Fe line a t 2863.44 A. Inseparable weak Fe and Ce lines should not normally interfere Avoid confusion with Co line a t 3174.91 A. Inseparable weak Ce and Fe lines should not normally interfere Inseparable Ta line should not normally interfere Avoid confusion with weak Mn line a t 2882.90 A confusion with V line at 2870.55 A An ion line. Inseparable weak Si and Fe lines will not normally interfere An ion line Inseparable Pr line will not normally interfere An ion line. Ti line at 3199.92 A and Fe line a t 3200.48 A inter- fere when strong.Weak Ni line at 3200.42 A will not normally interfere An ion line. Inseparable Ce and Sm lines will not normally interfere Useful only 1:o confirm higher percentages of Y . Avoid con- fusion with Co line a t 4644-32 A and with Eu line a t 4644.24 A. Inseparable from Er, Pr, and weak Fe lines Weak Cr and Ti lines will not normally interfere An ion line. Inseparable weak Fe line should not normally An ion line. Weak Fe line a t 3438.31 A will not normalIy interfere appreciably interfere a:ppreciably * The lines used for the detection of the minimum percentages are given in italics. Under the conditions prescribed later in this paper, and with the particular spectrograph used, visual or photometric work is possible if tlhe lines are not very strong and are not less than 0.03mm apart.Table I shows the wavelength differences necessary to eliminate interference by one line with a near line. REPRODUCIBILITY OF RESULTS The figures quoted below illustrate the degree of reproducibility attained when the method was applied to a series of silicate rocks, the figures enclosed in each pair of brackets being replicate photometric determinations made on the same sample. The replicate deter- minations are not sufficiently numerous to justify the calculation of coefficients of variation. For comparison, results of chemical determinations of mangancse, chromium and bariumMarch, 19581 DETERMINATION OF RARER ELEMENTS I N SILICATES 141 are printed in italics : the accurate chemical determinations of manganese indicate a fair degree of true accuracy for the spectrographic determinations.determinations of barium and chromium are only approximate. The chemical Manganese, %- Barium, "/d- - Chromium, ?;-- Nickel, yo-- - Strontium, Oh- - Vanadium, %,-- Yttrium, yi- Zirconium, yo- (0.072, 0.065, 0.066, 0*0G9)Q (0.018, 0.015, 0.016, 0.016) (0.072, 0.072. 0.072, 0*0S6)c (0.080, 0.090, 0.090, 0*07'O)d (0.014, 0.017, 0-013, 0.011)O (0.09, 0.10, 0.10, 0-08)d (0.030, 0.030, 0.030, 0.031)g (0.11, 0.12, 0.12, 0 ~ 1 2 ) ~ (0.11, 0.14, 0.12, 0 ~ 1 2 ) ~ (0.13, 0.17, 0.17, 0-17)d (0.014, 0.016, 0.014, 0*019)k (0.11, 0.10, 0.10, 0.08)' (0.060, 0-070, 0.070, 0 0 0 4 ) ~ (0.060, 0.072, 0.070, 0.05)" (0.062, 0.070, 0.073, 0 ~ 0 5 ) ~ (0.048, 0.057, 0.057, U-03)h (0-045, 0.049, 0.047, 0 ~ 0 4 ) ~ (0.020, 0.020, 0-020, 0-02)d (0.085, 0.095, 0.095, O*OG)k (0.020, 0-020, 0.020, 0-02)z (0.015, 0.015, 0.017, 0.01) (0.0069, 0.0074, 0-0067, 0*007)d (0.029, 0.029, 0.030, 0-03) Ir (0.030, 0.033, 0.030, 0.02) (0.032, 0.035, 0.035, 0.02) (0.0060, 0.0050, 0.0060) (0.0026, 0.0034, 0.0032) (0.0070, 0.0070, 0-0070, 0.0060) (0.014, 0.014, 0 ~ 0 1 5 ) ~ (0*0066, 0-0078, 0*0066)" (0.047, 0.048, 0*047)d (0.002, 0-002, 0.002)g (0.082, 0.090, 0.096)' (0.067, 0.072, 0.064)' (0.030, 0.034, 0.032)d (0.017, 0,018, 0 ~ 0 1 7 ) ~ (0-17, 0-19, 0.20)l (0.0058, 0.0033, 0.0050)" (0.0066, 0.0056, 0.0062)C (0.019, 0.018, 0-020)d (0.0070, 0.0080, 0*0070)d (0.0093, 0.0087, O-OIO)i (0.0087, 0*0080, 0-0087)m (0.017, 0.016, 0.015)" (0.021, 0-022, 0.021)' (0.030, 0.030, 0-030)d (0.035, 0.034, 0.033)' (0.0062, 0.0082, 0-0054)n (0.010, 0.011, 0.011, 0.013, 0.011)' (0-0031, 0.0043, 0.0050, 0-0058, 0.0055)c (0.013, 0.013, 0.014)d (0.0038, 0.0041, 0.0039)d Direct determinations, without internal standard.(0.10, 0.09, 0.11)" (0.022, 0.021, 0.022) (0.09, 0.12, 0.ll)C (0.027, 0.027, 0*026)d (0.025, 0.038, 0.023)5 (0.021, 0-020, 0-023)d (0.005, 0-007, 0-007)n (0.032, 0.037, 0~033)~ (0.012, 0-014, 0.012) (0.012, 0.010, O*OIS)m (0.009, 0.008, 0.008)P (0.020, 0.022, 0*021)h 40.016, 0-017, 0*016)i a = gneiss g = granite 1 = diorite b = felsite h = camptonite m = tonalite I: = hornfels i = lamprophyre n = pegmatite d = schist k = granulite o = porphyrite .Is indicated later in the description of the method, each plate carries a spectrum of a control mixture of known element contents..4lthough the original standard mixtures were made entirely from pure chemical compounds, the control mixtures were rock powders (a granite and a felsite) to which additions of compounds of rarer elements were made. The control determinations provide addi tiorial information about the reproducibility of the results: some figures obtained to date are listed below, the figures in italics being the amounts actually present. Manganese, ?(,--- Barium, %-- Chromium, ?A- Nickel, %-- Strontium, yo-- Vanadium, yo- Yttrium, yo- Zirconium, "/o-- 0.24, 0.20 0.13, 0.12, 0.12, 0.11, 0-12 0.044, 0.048, 0-046, 0.055, 0.045, 0.050 0.12, 0.14, 0.13, 0.13, 0.14 0.040, 0.044, 0.048, 0.043, 0.055, 0.057 The line used, Ba 4554.4, is an ion line 0.12, 0.10, 0.10, 0.11 0.038, 0.042, 0.044 0.060, 0.060, 0.052, 0.055 0.019, 0.020, 0.018 0.12, 0.10, 0.11, 0.11, 0.12 0.040, 0-034, 0.036, 0.047, 0.047 0.061, 0.051, 0.057, 0-060, 0.060, 0.051 0.030, 0-024, 0.025, 0-028 0.010, 0.006, 0.009, 0.010, 0.010 0.022, 0.020, 0.019, 0.019 0.010, 0.015, 0-011 Direct determinations, without internal standard 0.076, 0-072, 0.096, 0.10, 0.090, 0.10 0.039, 0-036, 0.037, 0.044 0.019, 0.027, 0.018, 0-018 LATITUDE- The use of only one calcium line for internal-standard control of all the other element lines is an empirical device that should eliminate gross errors when operating conditions are carefully standardised.To obtain information about the effects of deliberate departure from the standard operating conditions, some determinations have been made at increased142 HARVEY AND MURRAY: A SPECTROGRAPHIC METHOD FOR THE and decreased current (13 amperes and 8 amperes, respectively), and with a reduced amount of arcing mixture in the anode (35 mg instead of 512 mg).The figures quoted below show the minima and maxima found, the figures in italics being the amounts actually present. Manganese, yo- 0.13, 0.10 t o 0-12 Vanadium, yo- 0.030, 0.022 to 0.028 Nickel, %- 0.060, 0.040 to 0.065 Chromium,%- 0.12, 0.080 to 0.11 Barium, yo- 0.12, 0.070 to 0.17 Strontium, %- 0.12, 0-081 t o 0-15 Zirconium, yo- 0.039, 0.012 to 0.038 (not corrected by internal standard) These experiments demonstrate the effects of major changes in the operating conditions, and indicate that inadvertent fluctuations, which will be relatively minor, are not like&.to be a source of gross errors. METHOD OF RECORDING THE SPECTRA [Vol. $3 An outline of the method that was used for recording the spectra is as follows. Arcing wzixtures-one part by weight of the finely powdered sample is mixed with 3 parts of purified ignited calcium sulphate and 4 parts of powdered carbon. For the blank determinations, the same calcium sulphate and carbon powder are mixed in the ratio of 3 to 5 . The anode (bottom electrode) is drilled with a 3-mm diameter drill (a No. 32 carbon-steel drill was used) to a depth of & inch, and this cavity is completely filled with the arcing mixture (average weight about 50 mg), which is compacted by pressing with a steel rod during filling. The tips of both electrodes are flat, not tapered.(?$tics-A Hilger - Littrow spectrograph, F, 170 cm, with a quartz prism and lens is used. The wavelength range used is 2750 to 4670 A. An image of the arc is focused on a 4-cm horizontal mask attached to the collimating lens to exclude radiation from the electrode tips and to pass radiation from only the central $ of the arc. A six-step rotating sector, of step ratio 2 to 1, is used to produce the stepped spectra. The slit width used is 0.010 mm. PZates-Ilford Thin Film Half Tone ordinary plates (backed) are used and developed for 4$ minutes at 75” F in Ilford I.D.11 developer (M.Q. Borax), the temperature being con trolled thermostatically . Arcing technique-A d.c. arc with a 10-mm arc gap, the gap being maintained at 10mm during the entire arcing period, is used with no pre-arcing.The technique is as follows-- (a) The arc is struck at 3 to 4 amperes, the electrodes being separated gradually to reduce any tendency for mechanical loss during the destruction of minerals. (b) After 30 seconds, the current is increased to 6 to 7 amperes, this current being maintained for 15 seconds. (cj The current is now increased to 104 amperes and arcing is continued “to completion,” and thereafter for 10 seconds. “Completion” is indicated by a sudden change in the character of the arc accompanied by a fall in current of about 2 amperes. The total time of arcing is about 4 to 44 minutes. The current is controlled by means of a tapped resistance, sections of which are shorted by switches to effect the desired increases in current.The maximum current is adjusted before making the exposures by setting a variable resistance to pass a current of 8.6 amperes with a 10-mm arc between plain 5-mm carbon rods. Each plate carries spectra in triplicate of the sample, a spectrum of a control mixture of known element contents and a blank spectrum. Preparation of pur<fied calcium suZ$hate-The pure calcium sulphate normally obtainable often contains traces of strontium and barium. Calcium sulphate free from these metals has been prepared by the following method. Suspend 100 g of AnalaR calcium carbonate in about 500 ml of water and dissolve by adding 250 ml of concentrated hydrochloric acid. Dilute to about 1300 ml and precipitate about 90 per cent. of the calcium by adding to the boiling solution a hot aqueous solution containing 130 g of ammonium oxalate.After an interval of 15 minutes, add a little filter- paper pulp and filter through a large Buchner funnel, washing the precipitate once with hot water. Electrodes-Carbon rods of diameter 5 mm are used. Exposure, &.-The plate is exposed during the entire period of arcing.March, 19581 DETERMINATION OF RARER ELEMENTS IN SILICATES 143 Disperse the cake of calcium oxalate in about 500ml of hot water, heat the beaker on a steam-bath, and add concentrated hydrochloric acid gradually until dissolution is complete. Re-precipitate the calcium oxalate from the near-boiling solution by adding slowly, with constant stirring, a hot 9 N solution of ammonia. When the neutral point to methyl orange is reached, slightly acidify the solution by adding 20 drops of concentrated hydrochloric acid.Filter and wash the precipitate as before. Again re-precipitate in a similar manner. Convert the calcium oxalate to oxide and destroy organic matter by ignition at a tem- perature of about 900” C in a platinum basin, slake by the cautious addition of water, and convert to sulphate by adding a slight excess of diluted sulphuric acid (1 + 1 ) . Evaporate on a steam-bath and drive oft‘ the excess of sulphuric acid by heating at the fuming-point. Ignite the calcium sulphate at 900” C for about 1 hour and mix thoroughly. This calcium sulphate does not hydrate when exposed to a moist atmosphere.1° We thank Dr. G. 34. Bennett, C.K., F.R.S., the Government Chemist, and Sir William Pugh, O.B.E., F.R.S., Director of the Geological Survey and Museum, for permission to publish this paper. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. REFERENCES Mitchell, R. L., “The Spectrographic Analysis of Soils, Plants and Related Materials,” Tech. Ahrens, L. H., “Spectrochemical Analysis,” Addison - Wesley Press Inc., Cambridge, hlassachu- _- , “Quantitative Spectrochemical Analysis of Silicates,” Pergamon Press Ltd., London, 1954. Turekian, I<. K., Gast, P. W., and Kulp, J. L., Spectrochim. Acta, 1957, 9, 40. Breckpot, R., Spectrochirn. Acta, 1939, 1, 137; J . Inst. Metals, 1939, 64, 409. Strock, L. W., “Spectrum Analysis with the Carbon Arc Cathode Layer,” Adam Hilger, London, Eeckhout, J., Nature, 1945, 156, 175; Ver. Vlaan?. Acad. Wetem. Belg., 1045, 7, 5. Dingle, H., “Practical Applications of Spectrum Analysis,” Chapman & Hall Ltd., London, Harrison, G. R., “Massachusetts Institute of Technology Wavelength Tables,” John Wiley & Sons Harvey, C. O., “Notes on Spectrographic Analysis,” Bull. Geol. Szdrv. Gt. Britain No. 9, 1955, p. 52. Comm. Bur. Soil Sci. No. 44, 1948. setts, 1950. 1936, p. 43. 1950, p. 99. Inc., New York, 1939. Received June 7th. 1957