MAY, 1973 Vol. 98, No. 1166 THE ANALYST Precise Coulometric Determination of Acids in Cells Without Liquid Junction Part I.* Introduction and Instrumentationt BY E. BISHOP AND M. RILEY: (Chemistry Dejbartnzent, University of Exeter, Stocker Road, Exeter, E X 4 4QD) The requirements of high-precision amperostatic coulometry in the absolute mode are summarised, and the apparatus and instruments used described. Three constant-current sources are appraised : a 10-mA operational amplifier circuit, a commercial coulornetric titrimeter and a commercial analogue computer high-current source. The first and last have shown noise and drift levels that are acceptable in high-precision work. COULOMETRY, particularly amperostatic coulometry, is capable of producing analytical results of very high precision and accuracy, provided that the following three conditions are fulfilled: firstly, that the desired electrode reaction should proceed at a known high current efficiency; secondly,.that no analyte species should escape from, and no electroactive species should enter, the working compartment of the cell; and thirdly, that a means of locating the point of completion of the reaction, of commensurate precision and accuracy, should be available. The only electrode processes theoretically capable of attaining exactly 100 per cent. current efficiency are those involving electrolysis of the medium (solvent or molten salt) and its ions, and then only if the products of reaction are totally prevented from reaching the other electrode. For such reactions in the amperostatic mode the solvent molecules act as the intermediate, and when the limiting current for, for example, the reduction of hydrogen ion in aqueous solution, is exceeded by the total current, as must happen as the reactionapproachescom- pletion, the potential of the working electrode will change by about 800 mV as reduction of water becomes the dominant process.This change must be taken into account when assessing the influence of potentially electroactive impurities in the solvent or supporting electrolyte. It must further be noted that un-ionised weak acids are directly reducible a t a working cathode and un-ionised weak bases are directly oxidisable at a working anode. Acid - base reactions in a pure supporting electrolyte dissolved in a pure solvent are therefore attractive, and have accounted for more than half of the high-precision coulometric determinations so far made.Most of these have been conducted in multi-compartment cells,l-5 but some simpli- fication has been reported.6 The complexity of the manipulation, the difficulties of main- taining an effective seal between compartments and the limitations imposed by the low currents used make these techniques unattractive for routine work on standards. The alternative to separating the anode and cathode compartments is to change the nature of the auxiliary electrode process to one that proceeds a t very high efficiency at a potential less negative or positive, as required, than the complementary reaction to the main reaction. Both electrodes can then be accommodated in the same vessel, liquid junctions and salt bridges are eliminated, and the cell resistance is greatly reduced.This alternative, together with the use of high generating currents, which reduce background errors, has been explored for high-precision cathodic acidimetry.' The main cathodic reaction in an acidic solution free from reducible impurities at a platinum electrode is8 i c i, .. .. (1) 2H,0+ + 2e + H, + 2H,O .. * For Part I1 of this series, see p. 313. t Presented at the Second SAC Conference, 1968, Nottingham. $ Present address : Electronic Instruments Limited, Hanworth Lane, Chertsey, Surrey. @ SAC and the authors. 305306 BISHOP AND RILEY : PRECISE COULOMETRIC DETERMINATION [Analyst, Vol. 98 until the cathodic current exceeds the diffusion limited value for the hydrogen ion, when reaction (1) is joined by- i C 2H,O + 2e + H, + 20H- .... (2) %a so that the solvent acts as the intermediate in reaching the equivalence point. At a platinum anode in the same solution, reactions (1) and (2) proceed in the reverse direction, and, in addition, are joined by the reverse directions of i C i a 0, + 2H,O + 4e + 40H- .. .. * (3) .. .. (4) 0, + 4H,0+ + 4e + 6H,O in alkaline media and i C .. it3 in neutral or acidic media or when the anodic current exceeds the diffusion limited value for hydroxyl ion. Reactions (3) and (4) will proceed in the forward direction at a clean platinum cathode. The presence of oxygen in the catholyte does not lead to a loss of current efficiency in the determination of acids when the platinum cathode is clean, but when the cathode becomes dirty it no longer catalyses the immediate decomposition of hydrogen peroxide, an intermediate in reactions (3) and (4), and there is then the risk of some hydrogen peroxide diffusing away from the electrode surface and upsetting the acid - base balance of the bulk of the solution.Hydrogen can be removed by purging with nitrogen. The auxiliary reaction must proceed at a working potential much less positive than is required for reactions (3) and (4), while the supporting electrolyte required for it must not affect the hydrogen-ion concentration of the solution and must be suitable for reactions (1) and (2). The electrode reaction must neither introduce nor remove any entity that can affect the acid - base balance of the solution and must not introduce any entity that would be active at the cathode, thereby reducing the current efficiency of reactions (1) and (2).The best reaction is one that gives a solid product that adheres completely to the auxiliary electrode, and an obvious choice is the deposition of halide on a silver anode. This reaction was used in the pioneering work of Szebell6dy and SomogyiQJO and has since been repeatedly used by other workers. It has also been used as the auxiliary system in multi-compartment cells.1 Apart from this last application, methods of determination that involve the use of the principle have not set out to attain particularly high accuracy, 0.3 to 3 per cent. being common. Location of the end-point in high-precision coulometry has mainly been effected by means of zero-current potentiometry.The sensitivity of this process is limited by the effective Q of the reaction under the end-point conditions,l1Pl2 which is rarely high enough for the parts per million precision level. The most sensitive and precise method of end-point location is d.c. differential electrolytic potentiometry, or, even better, time-biassed periodic differential electrolytic potentiometry. The major attenuation of the precision of the results usually depends on the measurement of the generating current. Absolute coulometry, in which the Faraday constant is used as the ultimate standard, is dependent on the following factors: an adequately exact determination of the Faraday constant (the current value13 of 96 486.70 & 0.5 A s mol-1 may be conservatively rated at about 5 p.p.m.) ; the rationalisation of electrode processes, the determination of electrode kinetic parameters and the evaluation of current efficiencies11,12,14~1~ ; and experimental verification by high-precision determina- tions, of which there is a considerable bodyl2J6 to which the present study contributes.The Bishop Report16 proposing the adoption of the Faraday constant covers all these points and has been accepted by I.U.P.A.C. An extensive study has been made17 of cell and electrode design, of constant-current instrumentation and measurement, of auxiliary reactions and the errors introduced thereby, and of the precise location of end-points in high-precision assays of reference standard sulphamic acid of grade C.1839 In this paper the experimental methods and instrumentation are described and the constant-current sources and timing device are evaluated.In further papers, the auxiliary electrode processes, the working electrode processes, including the “silver error,” and the high-precision assays will be discussed. The choice of auxiliary reaction is exacting.May, 19731 OF ACIDS I N CELLS WITHOUT LIQUID JUNCTION. PART I 307 EXPERIMENTAL Volumetric operations were performed with Grade A calibrated glassware ; less critical weighings were carried out on a single-pan, five-place, constant-load balance (Stanton CL3), while critical weighings were carried out on a specially constructed free-swinging balance that had a standard deviation of 1.6 pg for loads up to 100 g.AnalaR reagents were used except for the specially purified sulphamic acid. WATER- throughout the work unless otherwise specified. CARBON DIOXIDE FREE WATER^^--- Carbon dioxide free water was prepared in 2-litre batches by rapidly boiling water in a Pyrex glass vessel. The lower part of the vessel was then cooled, without stirring, so that the top 30 to 50-mm layer remained close to the boiling-point. The contents were then rapidly transferred to a conditioned polythene bottle under a stream of white-spot nitrogen. The bottle was closed with a soda-lime guard-tube and was furnished with a polythene outlet tap. Such water was stored for a maximum of 12 hours, any that remained unused then being discarded. This water, blanketed by nitrogen or oxygen, was used for the preparation of electrolyte solutions and for washing electrodes and cells.ELECTRODES- Platin~m-Platinum-wire electrodes were made by welding 35 mm lengths of 22 s.w.g. grade C platinum to tinned copper connecting leads, sealing them into sheaths of soda-glass tubing and trimming them to a length of 25 mm. Gauze electrodes were of the cylindrical Fischer type (Johnson Matthey). The larger electrode (72020) weighed 31 g, the cylinder having a diameter of about 45 mm and an apparent surface area of about 125 cm2; the smaller (72050) weighed 19 g, was 32 mm in diameter and of 70 cm2 surface area. When necessary, the electrodes were cleaned by immersion for 2 to 3 minutes in freshly prepared aqua regia, washed in several batches of boiling water, and stored in water.Pre-treatment will be dealt with in Part 111. Silver-Silver-wire electrodes were made from 22 s.w.g. mint silver. A thread of blue glass was wound round the heated metal to form a sealed bead, which was then sealed into soda-glass tubing and annealed. The exposed wire was trimmed to a length of 25 mm. Large silver electrodes were made by manually winding 10 s.w.g. mint-silver rod on a mandrel mounted in a lathe so as to give cylindrical, helical coils with a stem for electrical connection. The coils were wound as tightly as possible so that, on removal from the lathe, their inside diameters were only slightly larger than the outside diameter of the mandrels, while adjacent turns were slightly separated, leaving space for liquid to circulate between them.Two sizes, designated A and B, were made, having the characteristics shown in Table I. Water was distilled in an automatic still with quartz condensing surfaces2* and was used TABLE I SILVER COULOMETRIC ANODES Size A Size B Over-all length/mm . . .. .. .. 145 135 Cylinder length/mm . . .. .. .. 48to53 30 to 35 Cylinder internal diameterlmm . . .. 25 to28 50 to 55 Mass/g .. .. .. .. .. 115to 120 155 to 165 Apparent surface area/cm* . . .. .. 125to135 170 to 180 Number of turns in cylinder . . .. 13to 14 9 to 10 Apparent surface areas (neglecting the roughness factor) were calculated from the known length of wire originally used, corrected for the length .of stem not immersed in the solution, the density of silver and the mass at the time, the mass giving the amount of silver consumed during use.Electrodes of area greater than 300 cm2 were made by mounting a size A electrode inside and concentrically with a size B electrode and making a common electrical connection. The large platinum-gauze cathode was mounted concentrically with and between the two silver helices. Silver electrodes were cleaned by immersion for 30 s in 1 + 1 nitric acid solution, followed by thorough washing with water. Silver bromide coatings were removed308 BISHOP AND RILEY : PRECISE COULOMETRIC DETERMINATION [Analyst, Vol. 98 by immersion in 1.0 M potassium cyanide solution, and the electrodes then thoroughly washed with water. An amount of silver black, representing the excess of silver present in the bromide film, remained after the cyanide treatment; most of this deposit was removed during washing, and the residue dissolved in the nitric acid.After a final washing, the electrodes were stored in water. Antimony-Antimony-rod electrodes with glass sheaths were prepared by a modification of the casting technique.lg Suitable lengths of 4-mm outside diameter Pyrex tubing of 0.75-mm wall thickness were cleaned with chromic acid, thoroughly washed and pre-heated to 110 to 120 "C in an air-oven. Specpure antimony was heated in a crucible to just above the melting- point, a tube removed from the oven, clamped vertically with one end dipping in the molten antimony, and a 40 to 60-mm column of antimony drawn into the tube by suction from a pipette filler, all operations being carried out as quickly as possible.The tube was removed from the still molten antimony, and electrical connection made by dropping several short lengths of resin-cored solder into the tube, melting them by gently heating with a flame and inserting a length of tinned copper wire. Finally, the bottom 5 to 10 mm of the tubing was cut off square with a diamond wheel so as to expose a cross-section of antimony rod of area about 6 mm2. Temperatures are critical and the method as described gave leak-free electrodes in which the antimony rod was tight-fitting and bonded to an antimony mirror on the inner wall of the tubing. Higher or lower temperatures gave either a rod capable of sliding in the tube, or an antimony mirror together with a sliding core; leakage in such instances was obvious.The electrodes were stored in water and were never allowed to dry out. Prior to use, the antimony surface was wiped with moist paper tissue and rinsed with water. When necessary, a fresh antimony surface was exposed by cutting off another small slice of the electrode. Reference electrodes-Saturated calomel (S.C.E.) or saturated mercury(1) sulphate (S.M.S.E.) electrodes of high capacity, the surface area of the mercury being 30 cm2, were used (the S.C.E. had a potential of +245 mV and the S.M.S.E. +640 mV versus a standard hydrogen electrode at 20 "C). In each instance a double remote junction was used, and the salt bridge was terminated by a low-leakage ceramic plug that dipped into the electrolyte solution. COULOMETRIC CELLS- The vessels most often used were amber-glass reagent bottles of about 400-ml capacity that had their necks sawn off and the rims ground flat; these bottles were subjected to a prolonged leaching and conditioning process before being brought into use.They were then cleaned with chromic acid, very thoroughly washed with hot water and re-conditioned by storing them filled with water for 1 week, the water being changed twice a day. Subsequently, until further cleaning was required, cells were washed with hot tap water, well rinsed with water and stored filled with water when not in use. Close-fitting lids were machined from 11-mm thick Perspex sheet and drilled with suitable holes to accommodate rubber bungs carrying the various electrodes and other equipment. In some experiments, cells made from sawn-off 400-ml beakers were used.Magnetic stirring was employed, the follower being coated with PTFE. CONSTANT-CURRENT SOURCES- Separate sources were used for high currents up to 2 A and for low currents up to 10 mA. When in use, the sources were run continuously, dummy load resistors being switched into the circuits in place of the coulometric cell in the intervals between increments or deter- minations. The 2-A soztrce-The 2-A source was a mains-operated solid-state power unit (Solartron AS 141 1) that incorporated a silicon controlled rectifier (SCR or thyristor) circuit designed to operate either as a constant-voltage supply in the range 0 to 40 V, or as a constant-current supply in the range 0 to 2.2 A. In the latter mode, the output voltage is directly proportional to the external load resistance, up to the value pre-set on the voltage limit control.If the load resistance rises to too high a value, the unit switches over to the constant-voltage mode, supplying a current inversely proportional to the resistance. The maximum cell load for 2 A at the maximum voltage setting of 40 is therefore 20 0. A small coil of resistance wire, having a resistance of 0.6 Q, was used as the dummy load.May, 19731 OF ACIDS IN CELLS WITHOUT LIQUID JUNCTION. PART I 309 The 10-mA sowce-The 10-mA source was constructed from chopper-stabilised analogue computer valve amplifiers (Solartron AA 1023) of d.c. gain lo6 (120 dB) and maximum outputs of 100 V and 12 mA. The circuit is shown in Fig. 1. Amplifier A,, in the constant-voltage configuration, gives a reference output voltage without imposing a significant current drain on the standard cell VR (Mallory cell, Type 303114,l-35 V).The output voltage ( E ) is given by and can readily be set by means of the 15-turn precision Helipot, R,. This produces a constant current in Rs in the input ts amplifier A, in the constant-current configuration, which maintains a constant current in its feedback loop through the coulometric cell. The current, I , is obtained from and with the component values used (R, = Rs = lo3 Q, 2 W, 1 per cent. high stability; R, = lo4 a, 15-turn; R, = 8.2 x lo3 Q, 2 W, 1 per cent. high stability) gives a range of currents from 1.5 to 15 mA, the voltage being limited by R, to within the saturation voltage of the amplifiers.The power supply to the amplifiers was a Solartron AS 853.3 unit. .. .. - * (5) E = VR (1 + R,IR,) .. .. - * (6) I = E/Rs = (V/Rs) (1 + R,/R,) .. Fig. 1. Operational amplifier 10-mA constant-current source (component values are given in the text) MAINS POWER SUPPLY- The sources already described, the AS 853.3 power unit, the pH meters, the crystal clock, the recorders and all other mains-driven instruments were supplied from a very low distortion, saturable reactor, constant-voltage transformer (Volstat CVN 500A, Advance Electronics), which was loaded to within 1 per cent. of its full output by adding resistance wire elements to the load. This technique ensures that the transformer is operating under optimum conditions. TIME MEASUREMENT- Electrolysis times were measured on a solid-state quartz crystal controlled clock (Venner TSA 3314) that had an operating frequency of lo4 Hz and made use of frequency divider circuitry.The over-all measurement range was 0.1 ms to 99 990 s in six ranges with a carry pulse at overflow, which was used to operate a solid-state driven electromechanical counter (Venner TS 8). The clock was calibrated, via a Marconi CR 100 communications receiver, against the standard frequency broadcast of station WWVH operated by the US. National Bureau of Standards in Hawaii. Errors of h 0 . l s in lo4 s were largely due to human reaction time in triggering, but over a period of 10 days the error was less than 1 s, corresponding to an accuracy of 1 p.p.m. In use, the clock was triggered by the action of switching sources from the dummy load to the electrolysis cell, and vice versa.310 BISHOP AND RILEY : PRECISE COULOMETRIC DETERMINATION [Analyst, vol. 98 ELECTRICAL MEASURING INSTRUMENTS- For fiotentials-All cell potentials and working or indicator electrode or differential electrolytic potentiometric potentials were measured on a vibrating capacitor input electro- meter (E.I.L.39A pH meter) that had a measured input impedance greater than 1014 $2. Com- prehensive backing-off circuitry and a recorder output to which back-off and band-spread could be applied permitted the expansion of 10 mV to 28 cm of recorder chart width. Voltage outputs from the sources were monitored on Sangamo-Weston S82 multi-range 1000 V a-1 voltmeters. For currents-Currents were monitored on multi-range Sangamo-Weston S82 meters.Accurate measurements were made by means of the voltage drop across standard, calibrated four-terminal resistors (Croydon Precision RS. 1) immersed in transformer oil. Mercury-in- glass thermometers immersed in the oil were used to determine the running temperature for reference to the resistance - temperature calibration. The voltage drop across the standard resistors was measured with a Croydon Precision P3 potentiometer, on which potentials could be measured to &5pV after calibration of the galvanometer, which showed a maximum sensitivity of 0.3 mm pV-l. Recorders-Honeywell-Brown 153 X 17, 10 mV, 2 s, strip-chart recorders with a range of gear boxes, supplemented by a 0-25-s high-speed version when required, were used.The X - Y recorder was a Houston EHR 921, with a 280 x 215 mm platen. Currents were con- verted into potential inputs by the use of calibrated resistors. Potentiostat-A solid-state potentiostat (Southern Analytical Wadsworth A.1654) was occasionally used. Coulometric titrator-A commercial coulometric titrator (Thorn Electrical TE 110) was evaluated together with the constant-current sources. This titrator had a maximum output of 200 mA at 14 V, derived from a classical series-regulated circuit, and incorporated a machine integrator adjustable to give a read-out in micrograms of determinand. APPRAISAL OF CURRENT SOURCES- The current was allowed to flow through a chain of standard resistors, with or without the inclusion of the coulometric cell, and the voltage drop across a part of the chain was measured by using the 39A meter as a transducer.The major portion of the voltage drop was backed off, and the remaining small portion recorded on the 10-mV recorder. The calibrated backing-off supply of the 39A meter is electronic and is less stable than the mercury battery driven buffer controls. External Mallory RM 42 R batteries were used to supply most of the backing-off potential and the meter's buffer controls were used for fine control. The recorder was set to a centre zero and 4-45 mV of the total 1.35 V was recorded. Noise is defined as excursions from the mean trace of duration less than 1 minute, and drift is defined as a long-term departure from the zero setting. The inherent instability of the current measuring and recording system produced a noise of about -J=0.05mV over 24 hours.The supply noise and drift detectable above the background noise therefore corre- sponded to *5 and &lo p.p.m., respectively. The time constant of the measuring system was about 1 s, so that fast transients were not recorded. These, and the noise, were examined on a calibrated measuring X - Y oscilloscope (IIewlett-Packard 130 C), which, although designed for a maximum of 500 kHz, maintained a reasonable response to beyond 60 MHz. RESULTS AND DISCUSSION THORN TE 110 COULOMETRIC TITRATOR- Tests were made at three current levels, 10pA, 10mA and 200mA, over periods of 2 hours, during which the laboratory temperature was constant to within 0.5 "C. Noise at all levels was usually within h 0 .l per cent. of the current, and the drift was unidirectional upwards at the two higher currents and downwards at 10 PA, and fluctuated between 0.05 and 0.35 per cent. overall. The performance of the complete unit was also assessed by carrying out replicate determinations of ceriurn(1V) with electrogenerated iron( 11) and of arsenic(II1) with electrogenerated bromine ; the current efficiencies of the generation reactions were known to be essentially 100 per cent. For amounts of sample of the order of lO-*mol the relative standard deviation was not better than -+0*3 per cent. It was concluded that this instrument would be suitable only for uncritical routine use when a reproducibility of 0.5 to 1 per cent. would suffice.May, 19731 OF ACIDS IN CELLS WITHOUT LIQUID JUNCTION.PART I 311 THE OPERATIONAL AMPLIFIER SOURCE- A single-amplifier unit was first constructed, in which A, was replaced by either a 2-V accumulator or the Solartron AS 1411 power supply in the constant-voltage mode. The performance was not satisfactory, and the instability observed is attributed almost entirely to the instability of the reference voltage source under the current drain of 10 mA. The operational amplifier reference voltage source A, was constructed and showed an excellent performance at a current output of 10 mA. Over periods of 2 to 5 hours the noise was less than 5 p.p.m. and the drift less than 10 p.p.m. The two circuits, A, and A,, were then combined as in Fig, 1, and the over-all performance was found to be scarcely inferior to that of the reference voltage source, A,.Tests were carried out at currents of 5 and 10mA under ordinary open laboratory conditions without taking any precautions about temperature or draughts. Noise and drift levels were about &5 and &lo p.p.m., respectively, that is, hardly detectable above the background. This source is therefore entirely adequate for high-precision coulometry. THE SOLARTRON AS 1411 SOLID-STATE 2 A SUPPLY- Preliminary tests at a current of 1 to 2 A showed noise levels of less than &20 p.p.m. and drifts of less than &lo0 p.p.m. over periods of 1 hour. These tests showed that the instrument was highly sensitive to temperature changes (50 p.p.m. “C-l) and particularly sensitive to airflow caused by draughts, which altered the temperature of the heat sinks of the power transistors. An overnight “constant draught” test showed a typical bow-shaped curve, indicative of the slow drift with temperature changes.At a current of 2 A, noise levels were -+lo p.p.m.; drifts over 2 to 5 hours were &40 p.p.m., and over 1 to 2 hours were &25 p.p,m. The above test suggested that good performance could be expected if the temperature and airflow in the vicinity of the instrument could be maintained constant. However, when run in a large fan-circulation oven at ambient temperature, or in a large cupboard, the heat generated was sufficient to raise the air temperature to 40 “C, when, although the specification named an upper ambient limit of 50 “C, malfunctioning was apparent. A refrigerated constant-temperature room was not available at the time.Attempts were made to compensate for noise and drift by sensing the change in current level and by using an operational amplifier source to feed sufficient additional current into the circuit to give a constant summation current. Insoluble problems of interaction between sources were encountered, and it was concluded that the successful application of the additive technique required the interposition of some non-electrical current sensing system between them. Optical methods that make use of a mirror galvanometer and two photocells might be suitable, but the oscillation period of a galvanometer suspension is too long to sense noise. An attempt to improve the regulation of the 2-A source was made by using operational amplifiers to increase the gain of the automatic current control circuit, but the results were not encouraging.The best performance was obtained with the source in a corner of a seldom used laboratory, unenclosed but partially screened by other equipment. Drifts within &20 p.p.m. over periods of 90 to 150 minutes were attained. Thyristor circuits have inherently high noise levels, generated by the large switching pulses, and so the higher frequency noise was examined on a measuring oscilloscope at a sensitivity of 20 pV mm-l, the d.c. being backed off. With a current of 2 A flowing through a 1 Q standard resistor as the input, a 50-Hz ripple of rather distorted sine wave form and amplitude of 200 pV was observed. In each complete cycle at 50 Hz, two large positive- going and two large negative-going spikes of amplitude 50 mV were observed.Closer examination revealed that the spikes comprised bursts of oscillation at 1 MHz, the amplitude of which died to zero in 10 ps. The true noise was therefore greater than that indicated by measurements with sensors of long time constant. This, however, is not random “white” noise, but tends to average out close to zero over periods of 1 s or more; nevertheless, it is possible that potentiometric measurement of current may be subject to some bias. Overall, the unit performed satisfactorily over the range 20mA to 2 A, and it was decided to use it in acid assays, the benefit of the large current being that electrolysis times for 0-1-mol samples are reduced to about 40 minutes.The operational amplifier source, covering the current range up to 12 mA, would be used for the final 0-1 per cent. of the reaction, and differential electrolytic potentiometry for location of the end-point.312 BISHOP AND RILEY One of us (M.R.) is greatly indebted to the Charitable and Educational Trust of the Worshipful Company of Instrument Makers for financial support in the form of a Research S tuden tship. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. REFERENCES Taylor, J. K., and Smith, S. W., J. Res. Natn. Bur. Stand., 1959, 63A, 153. Marinenko, G., and Taylor, J. K., AnaZyt. Chem., 1968, 40, 1645. Marinenko, G., and Champion, C. E., Ibid., 1969, 41, 1208. Quayle, J. C., and Cooper, F. A., AnaZyst, 1966, 91, 355. Cooper, F. A., and Quayle, J. C., Ibid., 1966, 91, 363. Eckfeldt, E. L., and Shaffer, E. W., Analyt. Chem., 1965, 37, 1534 and 1581. Riley, M., and Bishop, E., Proc. Soc. Analyt. Chem., 1966, 3, 143. Bishop, E., Chemia Analit., 1972, 17, 511. Szebellkdy, L., and Somogyi, Z., 2. analyt. Chem., 1938, 112, 323. , Ibid., 1938, 112, 332. Bishop, E., in Shallis, P. W., Editor, ‘‘Proceedings of the SAC Conference, 1965, Nottingham,” W. Heffer & Sons Ltd., Cambridge, 1965, p. 291. -, “Coulometric Analysis,” Volume IID of Wilson, C. L., and Wilson, D. W., Editors, “Compre- hensive Analytical Chemistry, ” Elsevier Publishing Company, Amsterdam, 1973. Taylor, B. N., Parker, W. H., and Langenberg, D. N., Rev. Mod. Phys., 1969, 41, 375. Bishop, E., Analyst, 1972, 97, 761. -, “Report on the Status of the Faraday Constant as an Analytical Standard,” International Union of Pure and Applied Chemistry, Commission,,V-5, Pure A$@. Chem., in the press. Riley, M., “Some Studies in High Precision Coulometry, Ph.D. Thesis, University of Exeter, 1969. Analytical Methods Committee, Analyst, 1965, 90, 251. Bishop, E., and Sutton, J. R. B., Analytica Chim. Acta, 1960, 22, 590. Short, G. D., Ph.D. Thesis, University of Exeter, 1962. , -- -, Ibid., 1972, 97, 772. -, Ibid., 1967, 92, 587. Received December lath, 1972 Accepted January lst, 1973